HARVARD UNIVERSITY

Library of the

Museum of

Comparative Zoology

Bulletin of the Museum of Comparative Zoology
AT HARVARD COLLEGE.

Vol. XL. No. 1.

CHANGES ACCOMPANYING THE MIGRATION OF THE EYE

AND OBSERVATIONS ON THE TRACTUS OPTICUS AND

TECTUM OPTICUM IN PSEUDOPLEURONECTES

AMERICANUS.

By Stephen R. Williams.

With Five Plates.

CAMBRIDGE, MASS., U.S.A.:

PRINTED FOR THE MUSEUM.

Mat, 1902.

MhY

1902

No. 1 — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY
OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD
COLLEGE, UNDER THE DIRECTION OF E. L. MARK, No. 130.

Changes accompanying the Migration of the Eye and Observations
on the Tractus o'pticus and Tectum opticum in Pseudopleuro-
nectes america7ius. By Stephen E. Williams.

TABLE OF CONTENTS.

I. Introduction 1

II. Material 2

III. Methods 6

IV. Migration of the eye and

changes in the cartilagi-
nous skull G

1. Summary of previous studies

on the migration of the eye 6

2. Description of stages ... 9

3. Homologies of the anterior

bones of the skull ... 11

4. Changes in the cartilaginous

skull 15

a. Stage I. 15

b. " II 16

c. " III a 19

d. " III ^ 22

e. "IV 25

PAGE

f. Comparisonof Bothus with
Pseudopleuronectes ameri-
canus 28

g. Discussion of Pfeffer's
work oO

h. Resume 32

V. The optic portion of the cen-
tral nervous system ... 33

1. General condition in the adult 33

2. The optic nerves .... 35

3. The chiasma and tracts with

related ganglia 37

4. The tectum opticum . . 40
VI. Theoretical considerations . 47

VII. Summary 49

Bibliography 51

Explanation of Plates 56

I. Introduction.

The strarige want of symmetry iii, tlic head region of flounders
has attracted much attention especially because in adults both eyes
occupy the same side of the head. The peculiarity is the more re-
markable because, for some time after hatching, the eyes and all otlier
parts of the head are as symmetrical as in any other fish, and conse-
quently this asymmetrical condition is brought about afresh in the
individuals of each generation, instead of once for all, as is the case
with most variations.

Regarding the migration of the eye, with a single exception (Pfeffer,
'86, '94), only such phenomena liave been recorded as can be observed
from surface study or dissections. It has seemed desirable therefore to

VOL. XL. — NO. 1 1

2 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

learn from careful preparations of specimens in, transition stages whether
there was merely a mechanical twisting of the facial region in an other-
wise normal fish, or a more elaborate rearrangement of the parts with
reference to each other, and especially whether any histological clianges
accompany the more obvious external modifications.

II. Material.

The most of my work has been on the so-called winter flounder
(Pscudopleuronectes americanus Walbaum), a dextral ilatlish, but I
have also used for the sake of comparison a sinistral species, the
sand-dab (Bothus maculatus Mitchill).

My material was all collected at Wood's Hole, Mass., during the years
1898 and 1899. I obtained a series of developing eggs and young
Pscudopleuronectes from the hatchery of the United States Fish Com-
mission in April, 1898. Adult fishes can be taken by nets at any time
through the year. The larval stages at or about the time of the
migration of the eye are to be obtained during the month of Juno
only. Early in the month only a few are at the point of assuming
the adult position, and after June 20th, all the fish of this species taken
were already metamorphosed.

These larvae were caught by surface towing with a coarse scrim tow-
net near the wall of the "outer basin" of the U. S. F. C. wharf during
the rising tide. They are most abundant on clear days when the wind
is on shore and the tide comes in from the east. On very calm or very
rough days they are not plentiful. My most successful skimmings
were made early in June, and twice I obtained as many as 100 young
fish during the inward flow of tlie current (3-4 hours). I was able to
save a few of the young fish alive by frequently emptying the tow-net
and placing the uninjured specimens in as pure water as possible.

In the summer of 1898 the sand-dab larvae were taken more abun-
dantly than the winter flounders, while in 1899 the winter flounders
were about ten times as niunerous as the sand-dabs.

I kept the young fish in the " outer basin " ^ in large lamp chimneys,

1 The granite inclosure for the protectioiv of smaller boats belonging to tlie
United States Fish Commission is divided l)y projecting parts of tlie dock into the
" inner " and " outer " basin. Tiiere are numerous openings in the stone walls to
allow the free circulation of the water, and near one of these the float was
moored, thus securing as nearly normal conditions of water and food as consistent
with protection from violent wave action.

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 3

which were made into separate aquaria by tying netting over the ends
and were supported by a floating frame. After they had remained here
for a time they were removed to the laboratory and kept under
observation in running water.

The period at which the eye turns is one of great mortahty among the
young fish captured, so that most of those in this stage died before re-
moval from the net. Since there is as yet no bony orbit, the eyes are
absohitely unprotected. As the eye which is to change its relative
position must for a time be on the dorsal side of the head, held in
position merely by the skin and a limited amount of connective tissue,
it is not strange that in a number of instances young fish were taken
alive which had lost the migrating eye some time before their
capture.

The actual turning is a comparatively rapid process in the species I
have observed, though, as will be seen later, a long preparation is made
for it. For instance, those fishes taken in which the migrating eye had
reached the sagittal plane of the head swam in an upright position,
though they came to rest more often on the future eyeless side.
Within three days after the capture of a fisli in this stage both the
orientation in swimming and the position of the eyes became essen-
tially that of the adult.

The growth of the fish after turning is rapid. A sand-dab measuring
10 mm. in length and 5 mm. in depth (i.e., the measurement taken along
the dorso-ventral axis) was confined in a lamp-chimney aquarium for 11
days and then was found to measure 22 mm. in length and 12 mm.
in depth. If the third dimension, the breadth or thickness of the
fish, be assumed to increase in the same proportion, which is a reason-
able assumption, the volume of this individual increased more than ten-
fold during the 11 days. The winter flounder of corresponding stages,
according to my obsei'vations, does not grow quite so rapidly. It
reaches a lengtli of about 75 mm. by the end of August, when it
is at most 7 months old.

There are six species of flatfishes comparatively common at Wood's
Hole, according to Smith ('98). Three of these, Pseudopleuronectes
americanus, Limanda ferruginea, and Achirus fasciatus, are dextral (i. e.,
the fish lies normally with the right side uppermost), and three, Paral-
ichthys dentatus, Paralichthys oblongus, and Bothus maculatus are
sinistral.

Of these six species, Paralichthys dentatus probably breeds in the open
sea, as small fish are not found. Paralichthys oblongus and Bothus

4 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

breed in May and the sole about the end of June. I can find no ac-
count of the breeding time of Limanda. P. americanus breeds from
the middle of February to the first week of April.

In the summer of 1899, when P. americanus was especially plenty,
metamorphosed fish of two different lengths were taken in the tow.
These were about equally abundant. The smaller measured not over
8-9 mm, at the end of metamorphosis. Tlie larger was a moi-e bulky
fish with slightly more pigment and it was found swimming upright
until it reached a length of 13-14 mm., when it also turned left side
down. I found no specimen intermediate between the two lengths. The
larger, more pigmented specimens may have been either the larva8 of
the black-bellied variety or possibly the young of Limanda. The more
important specific differences between Limanda and Pseudopleuronectes
are the following : Tlie anterior part of the lateral line of Limanda is
more arched and this species has more fin-rays in both dorsal and ventral
fins. But it is difficult in the young fishes to establish a satisfactory
division on the basis of the number of fin-rays. According to Jjumpus
('98), P. americanus at Wood's Hole averages 66.1 fin-rays to the
dorsal and 49.6 to the ventral fin. Jordan and Evermann ('96-00) give
for Limanda 85 dorsal and 62 ventral fin-rays. The specimens of Li-
manda I have counted at Wood's Hole vary from 81 to 78 in the dorsal
and 61 to 47 in the ventral. I counted the fin-rays in six small fishes,
three of each type, and found that in two of these — they belonged to
the 14 mm. type — the rays corresponded to the formula for Limanda,
and that in one (9 mm. long) they agreed with P. americanus, there
being 64 dorsal and 47 ventral rays. The number of rays in the other
three were absolutely intermediate, two (8.5 mm. long) having resj^ec-
tively 71-54 and 76-51 rays, the remaining one 75-56 rays.

The work of Kyle ('98) at the St. Andrews laboratory is valuable for
comparison at this point. There are five dextral flounders on the Scotch
coast which may be confused with one another. Tlie ones most like
our species are Pleuronectes flesus, the flounder, P. platessa, the plaice,
and P. limanda, the dab. Of these, when metamorphosis is completed,
the flounder is the shortest (about 8 mm., according to Petersen), the
plaice next and the dab the longest. The plaice may vary in length
from 13 to 16 mm. ; the dab from 16 to 19 mm. at metamorphosis.
In Danish waters (Petersen, '94, p. 14) the metamorphoses of these two
species are complete when the fish is from 4 to 6 mm. shorter.

As the plaice and dab overlap each other in length, their fin formula)
were ascertained by Kyle in the hope of finding there a distinctive

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 5

character. These also overlap, the dorsals varying in both forms from
68 to 77 and the anals from 50 to 61, the dab usually presenting the
higher number. The flounder has from 58 to 64 dorsal rays and
from 38 to 46 anal rays.

Pseudopleuronectes is intermediate in the number of fin rays between
P. flesus and P. platessa. It also turns at an intermediate length.
Taking Petersen's figures for Denmark, P. flesus turns at 8 mm. and P.
platessa at from 10 to 11 mm. The length at which my shorter larvae
turned was from 8 to 9 mm. No individuals longer than this were
found metamorphosing until the length of about 14 mm. was reached.

Limanda ferruginea has more fin-rays than P. limanda. If I am cor-
rect in the assumption that the larger, more bulky fish, which turns at
a length of 14 to 15 mm., is the young of Limanda, its length at meta-
morphosis would be intermediate between those found for P. limanda
by Kyle and by Petersen.

If this fish is the young of Limanda, another problem would be
solved. How is it that, with two such distinct sizes at metamorphosis,
the small flatfishes seined a month later are about uniform in size 1
Limanda is a comparatively deep-water fish, being found in the deepest
parts only of Vineyard Sound ; the young may have returned by the last
of July to the region where the adults live, so that there would be left ■
only the young of the on-shore species, P. americanus.

That I took only a few specimens of these problematical coarser larvae
in June, 1898, and that half the larvae taken in the same month of the
next year were of this kind, leads me to believe that the breeding sea-
sons of P. americanus and Limanda may not always exactly coincide.
This question can very easily be settled by breeding the fish, and satis-
factorily only in that way. It may be that the phenomena we have to
deal with here are explainable in another way. Looss ('89) found that
tadpoles metamorphosed in " waves," a part only of a brood changing
at a time. There might be something of this sort here, metamorphosis
at the one length or at the other depending on the advancement of
development.

I wish to thank Mr. Alexander Agassiz for the privilege of occupying
one of the Museum tables at the U. S. F. C. laboratory during parts of
the summers of 1898 and 1899, and Mr. W. A. Willard for a number of
brains of adult fishes. The work on the nervous anatomy was done, in
part, under the direction of Dr. G. H. Parker. I am deeply indebted
to Dr. E. L. Mark, at whose suggestion the work was undertaken, for
useful advice and the supervision of the whole work.

bulletin: museum of comparative zoology.

III. Methods.

The killing fluids used were (1) 10% forniol, (2) Flcmming's stronger
fluid, (3) Vom Rath's picro-sublimate mixture, (4) bichromate of po-
tassium, (5) Gilson's fluid, arranged in the order of their value. I failed
to get successful preparations with Vom Rath's platinic chloride mix-
ture. Where decalcification was necessary Flcmming's mixture gave
very good results. The usual methods of further procedure for sections
by the parafiiu process were used. Heidcuhain's iron hematoxylin gave
the best stain, though Delafield's and Ehrlich's hajmatoxylins also gave
successful preparations. These were followed by Congo red or acid
fuchsin to differentiate fibre tracts. The acid fuchsin has the further
advantage that it stains developing bone and fibrous connective tissue.
The Weigert stain with copper and the Weigert-Pal method were both
used in nerve study. Both adult brains and the larva) proved to be
refractory material for the Golgi method. The rapid method was used,
but not more than 5 per cent of the specimens gave any impregnation
whatever. A sojourn of three days in the Golgi fluid and more than
two in the silver bath were found to give the most successful prepara-
tions. Material was left in the silver until wanted for sectioning,
though much of it was sectioned after an exposure of two days to the
silver nitrate.

IV- Migration of the Eye and Changes in the
Cartilaginous Skull.

Before proceeding to describe the conditions which I have found in
Pscudopleuronectes americanus, I shall give a brief account of the main
results reached by previous observers, omitting for the present those of
Pfcffer.

1. Summary of Previous Studies on the Migration

OF THE Eye.

It was suggested about the middle of the last century, that the Plcu-
roncctidaj, though unsymmetrical as adults, are, in their young stages,
bilateral animals like other fish. The brief accounts of Van Beneden
('53) and Malm ('54), who found young fish quite similar in markings
to adult flatfishes, but with eyes in a different position, seemed to indi-
cate the possibility that one of the eyes migrated around the head from
one side to the other.

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEUEONECTES. 7

The first paper which really describes a method of transition of the
eye in flatfishes is that of Steenstrup ('63). According to Wyville
Thomson ('65), on whose abstract of Steenstrup's paper I have relied
(see also Steenstrup, '64), this author contends that the final posi-
tion of the eyes cannot be explained as simply the result of a torsion of
the front part of the head ; and there is, in his (S.'s) opinion, a pene-
tration of the tissues of the head by one of the eyes. This process
Steenstrup described carefully from alcoholic specimens of different sizes
of the young forms which he provisionally termed Plagusise. In this
species development resulted in a sinistral flounder, i. e., one in which
the left side during adult life is uppermost. The right eye was slightly
in advance of, as well as dorsal to, the left eye. The mouth became
oblique toward the blind side, and the posterior part of the face, where
the normal eye is located, seemed pressed " upward " toward the future
eye-side. The right eye no longer projected from its own side of the
head in a large orbit, but was deeply imbedded in the tissues, so that it
had only a small orbit-opening on the right side. Later, an opening was
made on the left side and for a time the eye had two orbits. The orig-
inal orbit soon closed, and as the eye reached the surface level on the
left side of the head the new orbit increased in size. This second orbit-
was described by Thomson as a bony one in the adult fish, being formed,
so Thomson contended, by the frontal and prefrontal of both sides.

Schiodte ('68), working on other species, showed that the passage of
the eye around the 'head is a normal method of development. The
penetration of the eye through the tissues of the head is restricted to a
few fishes whose larval forms were once considered adults, and given the
name Plagusia.

He observed a Pleuronectes platessa — a dextral flounder — 10 milli-
metres long, of which he says, " The right eye stands over the beginning
of the lower third of the maxillary bone. The left eye stands at the top
of the head, so much inclined to the right that from the left side only
slightly more than one-third of the pupil can be seen ; it stands in front
of the dorsal fin, so that the latter is just behind the end of the left and
[the] beginning of the middle thirds of the eye." In a 14 mm. speci-
men the pupil of the left eye had become invisible from the left side
and the dorsal fin touched the left margin of this eye, the foremost ray
being a little in advance of the extreme posterior margin of the eye. In
a 40 mm. fish the right eye had moved so that it stood over the lower
end of its maxillary bone and the left eye had followed it, so that they
were almost as close to each other as in the last stage, the left eye being

8 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

a little farther back than the riglit. In this specimen the dorsal fin
reached as far forward as the middle of the left eye.

Schiodte held from these observations that the dorsal fin kept its po-
sition and that the left eye migrated forward around it and then passed
backward to its final position. His implied argument, if I understand him
riglitly, is, that the right eye moves backward from a position over the
lower (posterior) third of the maxillary bone to one over its lower (pos-
terior) extremity, and that the left eye moves backward still further
proportionally, because in the end (the 40 mm. specimen) it is not only
above but "a little behind " the right eye. This conclusion was in his
opinion confirmed by the observation that the rays in the dorsal fin of
young specimens corresponded in number with those of the adult.

He described under the name Bascanius taadifer, n. s., a peculiar
flounder (evidently sinistral), which had a semilunar depression between
the right eye and dorsal fin. Here the body was so thin that, if
incautiously handled, it broke in pieces or separated itself from the
dorsal fin. In that case a part of the right eye appeared through the
hole, giving the animal the appearance of possessing two eyes and a
half.

Agassiz ('78) described definitely for the first time the two methods
of development by which the eyes of flatfishes change position. His
description of the method by migration around the head is briefly as
follows (p. 5) : " The first change — and the process is identical,
whether we take a dextral or sinistral flounder — is the slight advance
toward the snout of the eye about to be transferred. . . . This move-
ment of translation is soon followed by a slight movement of rotation ; so
that, when the young fish is seen in profile, the eyes of the two sides no
longer appear in the same plane, — that on the blind side being slightly
above and in advance of that on the [future] colored side. With increas-
ing age, the eye on the blind side rises higiier and higher toward the
median longitudinal line of the head ; a larger and larger part of this
eye becoming visible from the colored side where the embryo is seen in
profile, until the eye of the blind side has, for all practical purposes,
passed over to the colored side."

Later the dorsal fin finds its way forward toward the nose, dorsal to
the transposed eye.

Agassiz also well described the method by penetration discovered by
Steenstrup in Plagusia. The change was followed day by day in fishes
kept captive in his Newport laboratory. He pointed out that these two
methods are merely two extremes of the same process ; probably the

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 9

peculiar fish described by Schiodte was an example of an intermediate
method .

Only two other descriptions of intermediate methods of eye-transition
need be noticed. Ehreubaum ('96) has discussed, among other points,
metamorphosis in the flatfishes of the German Ocean. Stages of the
larvae of the commoner species in which the eye passes around the head
are given. In the larva of Arnoglossus laterna, which strongly resembles
the so-called Plasrusise, the dorsal fin extends to the nostril while the
fish is yet symmetrical, so that the eye must pass under the dorsal fin as
in Plagusia. Tlie prolongation of the dorsal fin to the nasal pit and the
position of the right eye close to the lower margin of the fin (after
migration) prove, in Ehrenbaum's opinion, that the right eye is
shoved through imder the dorsal fin from the right to the left side.

Recently a Japanese zoologist, T. Nishikawa ('9'), found a case
where the dorsal fin extended along the head as far as the end of the
snout in close contact with, but not fused to, the skin. There were no
fin rays located in the eye region. The right eye passed through a slit
between the fin and the head in one day, passing thus from one side
completely to the other. Unfortunately the fish died, so that it is not
known whether the fin would have fused later to the dorsal part of
the head or not.

2. Description of Stages.

For convenience of description four stages of development may be
recognized in Pseudopleuronectes americanus.

Stage I., the recently hatched fish, is represented (Plato 1, Fig. 1) by
a specimen 3.5 mm. long and 12 days old. Owing to its wide dorsal
and ventral fins being so transparent as to be scarcely visible, the
living animal resembles, in its general appearance, a very minute
pin with an elongated head. It is" essentially symmetrical. I have
sectioned the eggs as well as the young fish and find a close resem-
blance to the figures given by Fullarton ('91) in his work on the develop-
ment of the plaice, Pleuronectes platessa, which is the nearest European
representative of our flatfish. His drawings, too, show the eyes to be
symmetrical in position. There are few pigment cells in the body of
an animal of this stage and they are arranged in much broken
longitudinal lines.

The largest of the recently hatched fishes are nearly as long as the
smallest of the pelagic larvre (Stage II., Plate 1, Fig. 3), which were
taken the first of June ; but between the two there is a great diflTerence

10 bulletin: museum of comparative zoology.

in depth and* bulk. To this stage are assigned all those fishes which, in
a strictly lateral view from either side, exhibit only one eye. The shorter,
proportionately deeper, larv;« metamorphose when they reach 8 or 9 mm.
in length. The degree of symmetry can better be seen in a front view
(Fig. 4) of a fish 4 mm. long, the only trace of asymmetry at this stage
being the slight elevation of the left nasal pit and the lack of absolute
bilateral symmetry in the shape of the mouth. The upper lip is slightly
drawn upward on the right side directly opposite the right nasal pit
{fv. olf.).

Stage III. (Fig. 2) has been made to include those fishes in which the
eye of the blind side had so far migrated as to be visible when tlie fish
was viewed in profile from the ocular side. At this stage the eye lies in
the median plane in a depression immediately in front of the dorsal fin,
which has grown forward since the preceding stage. There is also a
noticeable change in the direction of the urostyle ^ (iir'stl.).

In the last stage, IV., the eye has completed its migration, and, so far
as regards the distortion of the head, the fish is essentially in the adult
condition. Changes after this are merely accentuations of what is
found here. Figure 6 shows the dorsal tin {pin. d.) at this stage
extending as far forward as the middle of the eye. On the body are
to be seen the beginnings of the pigment areas which later color the
right side of the fish.

The sinistral fish, Bothus, is at first symmetrically pigmented. The
lower side does not become colorless until the disappearance of the first
color pattern and the establishment of the much lighter adolescent
color, which comes after the turning. P. amcricanus, on the contrary,
is essentially non-pigmented until it is ready to become a bottom feeder.

The front view of P. araericanus at this stage (Fig. 5) — the com-
pletely turned fish — is most instructive in bringing out the want of
symmetry. The left eye has moved through an arc of about 115
degrees, as may be seen by comparing this view Avith that of Stage II.
(Fig. 4). The left nostril has moved dextrad and dorsad, as if in the
passage of the eye it, too, had become involved. The angle of the
mouth on the right side bends sharply ventrad ; and the upper lip
of the right side is apparently drawn dorsad toward the right nasal
pit. From this point the mouth opening has the form of a long slit
which extends to the left and ventrad in a nearly straight line.

In Paralichthys oblongus and in Bothus the mouth remains nearly
horizontal and symmetrical.

1 For the development of the caudal fin of the flounder, see Agassi/, ('78).

WILLIAMS : MIGKATION OF EYE IN PSEUIjOPLEURONECTES. 11

3. Homologies of the Anterior Bones of the Skull.

The changes in the cartilaginous facial skeleton will be more easily set
before the reader, if the homologies of the bones of the face as explained
by the more recent writers be first made clear.

The papers of PfefFer ('86, '94), which deal with the cartilaginous
skeleton, are also reviewed here.

Traquair ('65) has given a careful account of the adult skulls of
flounders of both dextral and sinistral types. The greatest changes, as
compared with a symmetrical fish, the cod, he finds in the facial region ;
the brain case remaining nearly symmetrical, except with regard to the
position of the ridges and wings on the bodies of the bones for the at-
tachment of muscles.

The adult skulls of (1) the halibut, (2) the pole flounder, and (3) the
plaice (Platessa vulgaris) form a series, in which he shows that there is
a progressive modification, especially of the frontal bones. In the hali-
but, though the main part of the frontal of the " eyeless " side is back
of the migrating eye, a thin curved process from it extends between the
two eyes and with the corresponding interocular process of the frontal
of the ocular side (to which it is closely applied) forms a part of the
orbit of the migrating eye. In the case of the pole flounder this process
from the frontal of the eyeless side is reduced to an exceedingly thin
curved strip. Finally, in the common flounder even this thin strip has
entirely disappeared, so that the frontal of the eyeless side is now joined
with the front of the head exclusively by means of the great externa]
connection, since called by German writers the "Brtlcke."^

Steenstrup ('63), according to Thomson ('65), considered the " Brtlcke "
the principal frontal of the eyeless side.

Thomson himself thought that it represented the prefrontal of the
eyeless side, and that the partition between the eyes was the frontal of
the ocular side.

Malm ('68) at first held the " Brticke " to be infraorbital, but later
adopted Steenstrup's view.

Reichert ('74), disregarding the beliefs of previous authors, decided
that the frontal formed two infraorbital processes, which then fused with
the latent " Brticke " to form the orbital ring. The parts between the
eyes he thought were normal.

* This is a new and peculiar bridge or bar fpseudomesial) of bone wbich has no
(single) equivalent in the crania of synunetrical fishes.

12 bulletin: museum of comparative zoology.

Klein ('68) called the outer edge of the " Brlicke " prefrontal, and the
inner and huider part of the same, principal frontal,

Traquair (-65, pp. 27G, 277) summarizes the changes from the condi-
tion of the symmetrical type of skull as follows :

" (1) The mesial vertical plane of the cranium has become inclined over to
the now binocular side, very slightly in the posterior part of the cranium, very
much in the region of the eyes (so that the original vertical interorbital septum
becomes now nearly horizontal), returning in the nasal region nearly to its
original vertical position in the turbot, but never doing so in the halibut or
plaice.

" (2) In consequence of this, the middle line of the base of the skull remains
still comparatively straight; while the middle line of the upper surface, diverg-
ing from the apparent or pseudomesial line, curves round between the eyes, . . .
and returns to the middle in front. Having got in front of the eyes and nasal
fossae in the turbot, it again coincides, or nearly so, with the apparent middle
line ; but in the halibut, and still more in the plaice, the apparent and mor-
phological middle lines, if produced, would cross each other.

" (3) In the anterior part of the cranium, the parts on the eyeless side of the
middle line of the base are, in all the Pleuronectidas, more developed than on
the ocular side. . . .

" (4) On the top of the head the interocular parts of the frontal and pre-
frontal bones are more developed on the ocular side. The interocular process
of the frontal of the ocular side is always much stouter than that of the other
[eyeless side] bone, and always articulates with a corresponding process sent
back from the prefrontal. But the prefrontal of the eyeless side sends back
no process to articulate with the frontal of the same side, whose interocular
part, if examined in a series of flatfishes, gets smaller and smaller, till in the
plaice it seems almost gone. The same condition affects the morphologically
mesial plate of cartilage fonning the anterior part of the interocular st-ptum,
•which cartilage we have already seen to be chiefly developed on the ocular
side.

" (5) To accommodate the two eyes, now both on one side of the head, the an-
terior parts of the frontal bones remain as a narrow bar, never widening out into
a broad arch as in the cod and other fishes. Accordingly, to maintain the
requisite stability of the cranium, a new bar or bridge of bone is formed (pseudo-
mesial) by the union of a process sent forwards from the anterior external
angle of the frontal of the eyeless side with one sent back from the correspond-
ing prefrontal. By means of this bar the uppei; eye becomes closed round by
a bony orbit, whose boundaries in the turbot consist of the interocular process
of the frontal of the eyeless side, the external angular process of the same bone,
the external angular process of the corresponding prefrontal, and a small por-
tion of cartilage in front. In the halibut and plaice, however, the nasal bone
comes to take part in the boundary of the orbit principally by a development
from its eyeless side; and in the latter fish, owing to the atruuhy of the inter-

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEURONECTES. 13

ocular portion of the frontal of the eyeless side, the corresponding part of the
other frontal forms almost the entire external boundary of the orbit.

" (6) The olfactory foramen and the place of suspension of the anterior sub-
orbital bone are further forward on the ocular side. . . . The articulation of the
epitympanic bone to the cranium, in the halibut and plaice, likewise extends
further forward on the ocular side.

"(7) The axis of the keel of the cranium . . . points . . . to the eyeless side."

PfefFer in a preliminary paper ('86) without illustrations, has described
the larval stages of development in one of the Pleuronectidae. As he is
the only writer who speaks of the conditions iu the interior of the bead,
his conclusions are given in some detail.

The young fish has an entirely cartilaginous cranium, in which the
eye sockets are separated below by the sphenoid, and above by the inter-
orbital roof (Zwischenaugen-Decke) ; but between these the sockets com-
municate freely with each other. The ethmoid, constituting the anterior
part of the cranium, develops a wing on each side, the place where the
wings join the body of the ethmoid being marked by the presence of the
nasal openings. In very young animals the bulbi olfactorii are embraced
by the ethmoidal roof; but later they are forced backward behind it.

Over the interorbital and ethmoidal regions runs a ridge-like dermal
bone, which is triangular in cross section, and stands vertically ; it sup-
ports the dorsal fin, and is at first free from the cranium. It is the
" principal frontal " of authors.

In the second stage examined by Pfeffer, the migratory eye has risen
so that half of it is above the level of the interorbital roof. The brain
capsule remains unchanged, except that it has received the bulbus olfac-
torius, which has been forced backward by the migration of the eye.
The interorbital roof is bent outward toward the eye side and soniewhat
twisted on its long axis. At the same time the frontal, now grown fast
to the interorbital, makes with it a gfeat bend. However, only a broad
band — its basal portion — remains, while the greater, vertical part of it
is for the most part resorbed by the migrating eye. There now remains
between the migrating eye and the eye side only the translucent, thin
outer skin which previously covered the dermal bone. The front part
of the ethmoidal region is symmetrical ; but the upper part of the wing
of the eye side has fused to the fron to-orbital and is now continuous
with the developing supraorbital cartilage [bone?], while the whole rim
of the wing of the blind side remains free.

The transposed eye at a later stage occupies a pit which opens up-
ward and toward the eye side and is surrounded by a high rim of thin

14 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

dermal bones. The previously upper side of the eye now lies on the in-
terorbital sej^tum, therefore most ventral; whereas the previously lower
side of the eye is now near the dorsal fin, therefore highest. The eye
has thus rotated 180 degrees. The side of the migrating eye that is
turned toward the blind side of the head is now closed in by the forma-
tion of new dermal'bones. The socket is completely open in the region
of the optic nerve. By the migration of the eye, the anterior oblique
eye nmscles, which arise from the hinder border of the ethmoid, are laid
bare ; a thin covering of dermal bone grows over these also. The wing
of the ethmoid on the eyeless side, is fused to a part homologous with
the supraorbital cartilages ; these grow upward and inward, the latter
helps in forming the anterior wall of the new orbit.

PfefFer says that, though the ossification is a continuous process, one
may distinguish, if he will, three stages in the development of the paro-
stotic cranial bones of fishes, characterized by —

(1) The first delicate osseous investment of the cartilage ;

(2) The dermal ossification which establishes approximately the per-
manent forms of the bone ;

(3) The ridges, crests, wings, and the like, — entirely superficial addi-
tions, — which are probably always connected with muscular action.

In the flounder the rotation begins while the frontal region of the
young fish is in the first of these stages. Soon the frontal (cartilaginous)
is in quite another place, under quite another region of the skin. When
it has changed its position, there is dermal bone produced over it in its
new position ; but there is not the least reason why the skin under which
it would normally have lain should suddenly lose the power of producing
bone, — and in fact it does not, for it produces the bridge. The bony
bridge, then, is the parostotic ossification of a precise region of the cutis,
and if the cranium had remained symmetrical, it would have fused to
the frontal ; but inasmuch as there is a displacement of the region of the
(cartilaginous) skull, this dermal ossification has become attached to
those bones which took a position directly beneath this bone-producing
region of the cutis after the displacement of the (cartilaginous) skull.

Pfeffer's final paper, so far as I know, has not yet appeared ; but in a
short note ('94) the author states again that the interorbital septum
twists on its long axis, and adds: (1) that the migrating eye, when it
reaches the mid-line, loses the thin patch of skin which has separated the
cornea from the outer world, and (2) that the dorsal fin, the muscles
and the bones develop along the physiological axis of the body, the con-
tiniuition of the sftinal column.

WILLIAMS: MIGKATION OF EYE IN PSEUD0PLEUR0NECTE3. 15

4. Changes in the Cartilaginous Skull.

In order to have freshly in mind the normal condition of the cartilagi-
nous skull in fishes with which to compare the youngest flounder skulls, I
give a brief statement of the essential parts of Pai'ker's ('73) paper on
the skull of the salmon :

In a salmon of the second week, according to Parker, the cartilaginous
skeleton is fully formed. There is a large fossa on the top of the head
ever the mid-braiu. In front, the skull is roofed over with a thin carti-
laginous plate, the ethmoidal " tentorium," or tegmen cranii. Anteriorly
this is directly continuous with the ethmoid ; its posterior lateral cor-
ners are connected with the cartilage of the auditory region by the supra-
orbital bars, which curve upward and outward. The ethmoid is contin-
uous with the trabeculse cranii, — now fused together in front, but
diverging behind, — which run backward forming a partial floor to the
skull cavity. The superior and inferior oblique eye muscles liave their
origin on the posterior face of the ethmoid. The recti originate from a
lamina on tlie hinder part of the parasphenoid.

I have projected upon the frontal plane the cartilages of the facial
region of Pseudopleuronectes in each of the four stages. But because
of the great length of the dorso-ventral axis of the older stages, this
method needs to be supplemented either by projections upon the sagit-
tal plane or by some other process. The most satisfactory recon-
struction is, of course, the model. Accordingly with the aid of sections
I have modelled in wax by Bern's method the facial region of Stages
II., III., and IV., and cuts made from photographs of these models are
given in the text.

a. Stage I.

A dorsal view of the cartilages of the facial region in Stage I. is shown
in Figure 7 (Plate 1) as they appear in frontal projection. As in the
salmon (Parker, '73), the first cartilages to form are the trabeculee cranii
and Meckel's cartilage. The slight want of uniformity in the shape of
Meckel's cartilage on the two sides may be merely an individual varia-
tion. Certainly this cartilage is essentially symmetrical. The line
passing through the middle (third) brain ventricle and between the
lobes of the tectum and cerebrum I have assumed to lie in the sagittal
plane in a normal fish of this stage. This plane, represented in projec-
tion in the figure by the two ends of a fine line, cuts lengthwise the
fused trabeculee, dividing the mass at the anterior end, which is to be

16 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY,

the future etlnuoid, nearly into halves. The line falls midway between
the two arms of the trabecuUe, where they diverge to allow space
for the pituitary body. In front the ethmoidal mass overlaps slightly,
on either side, Meckel's cartilage a little behind its points of sharpest
curvature.

lu tlie flatfishes there is no distinct " tentorium," or tegmen cranii,
extending backward from the ethmoid to roof over the front part of the
brain case, as there is in the salmon.

b. S(a(je 11. .

Between Stages I. and II. there is an interval of six weeks and the
manner of differentiation of the many cartilages and projections found

^

trh. su^orb. dx.

>.

trb. s7i'orb. .1. p.

- trb. .tu'orb. s. a.
. - . ms'elfi.
- - ■ Vcis. eth. s.

^ . _ ec^el/i.

' . . crl. orb. a.

\pl-pal. dx.
hn-hy.

Kj crt. ink.

Fig .4.

Oblique view of the facial cartilages of P. americanus. Stage II. Pliotographed
from a wax model (Bonrs method) seen from a point midway between sagit-
tal and transverse planes and about .30' above the horizontal i>lane. X 75.

For meaning of lettering, see Abbreviations under Explanation of Plates.

in Stage II. (Fig. A and Plate 2, Fig. -10) cannot be traced here.
Figure 10 is a dorsal view of the facial cartilages of this stage. But,
as it gives a less complete view than the model of the same specimen
(Fig. A), I call attention to the two supraorbital bars only — the com-
plete one on the right (trb. su'orb. dx.), fastened to the right ethmoid
wing, and the two parts (a. and p.) of the left one, between which is

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEURONECTES. 17

the space through -which hater the eye must pass. Figure A is from
a photograph of the model of the front part of the cartilaginous cra-
nium of a 3.5 mm. fish, viewed obliquely from the front, the right side,
and above. The line of vision makes an angle of about 30 degrees with
the horizontal plane. Meckel's cartilage no longer forms a simple bow
lying in the horizontal plane. The anterior end is curved slightly ven-
trad, and the bar of either side in passing backwards bends sharply
ventrad to join, nearly at right angles, a series of cartilaginous masses
(Fig. A hy-md.) representing the future quadrate, articular, symplectic,
and hyomandibular bones. In cross section these cartilaginous masses
have, in general, the form of an elongated oval, the axis of which in-
clines dorsad and mesiad ; the ventral margin is slightly thicker than
the upper. The space occupied by each separate cartilage in this series
is not indicated in the models, though in the sections the boundaries
can be determined by the presence of the connective-tissue sheaths which
limit the cartilages.

The pteryffo-pcdatine bars (p(-pal.) extend ventrad and caudad from
each side of the ethmoid to the quadrate region (compare also F'ig. 10).
At this stage the fish has a very small gape. The hyoid and gill-arch
cartilages are present in their general shape, occupying most of the space
between the right and left hyomandibular-quadrate masses, and ending
in front just beneath the body of the ethmoid in the basi-hyal (ba-hy).

From the ethmoid mass arise also the supraorbital bars. These, in
the salmon, extend backward from the ethmoid, curving upward and
outward above the eyes, to the heavy cartilaginous mass of the otic cap-
sules. In the flatfish of this stage, as shown in the reconstruction,
there is but one complete supraorbital bar (the riglit), the left being
represented by two remnants, an anterior and a posterior ; the anterior
(trb. siCorb. s. a.) is a process extending backward from the dorsal left-
hand corner of the ethmoid ; the posterior {trb. suorb. s. p.) extends
forward from the left otic capsule. It is through the space between
these two projections that the left eye migrates. While, as yet, there
is no external sign of an asymmetrical position of the eyes, internally
preparations for such a condition are clearly established, for the middle
portion of the left supraorbital bar has disappeared.

I have sectioned only a few individuals of P. americanus in which the
left supraorbital bar is still continuous, and even in them at the region
corresponding to a transverse plane passing through the middle of the
two eyes the bar is so reduced in thickness as to show in cross section
only one or two cartilage cells.

VOL. XL. — NO. 12 *

18 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Since Bothus spawns in May, I was able to get specimens which were
certainly not more than one month old. The one shown in frontal sec-
tion in Figure 14 (Plate 3) was 2 mm. long. However, as P. america-
uns grows much more slowly than Bothus, it is not jiossible to compare
ages on the basis of relative lengths. In Bothus at this stage both
supraorbital bars are present and there is as yet no sign of reduction in
either of them. In the sinistral flounder (Bothus) it is, of course, the
riglit supraorbital bar which disappears to give passage for the eye,
whereas in P. amcricanus it is the left. Since in the middle of the bar
its jilane slants inward and downward, and since the bar in its course
from ear capsule to ethmoid is also slightly convex dorsally, it is evident
that no one section in any plane could show the whole bar. Both bars
extend over the eyes, as can be seen from the position of the dotted
lines shown in the figure (Plate 3, Fig. 14), which represent the location
of the eyes, as seen in a more ventral section, accurately projected upon
the plane of this section.

Appearances of degeneration in P. americanus taken after June 1
are rare. The youngest fish must be at least six weeks old at that time,
and only the most nearly symmetrical of the smallest fishes sectioned
show any trace of the left supraorbital bar, either normal or degenerat-
ing. Figure 15 (Plate 3) sliows the appearance, in frontal section, of
the anterior degenerating end of the posterior remnant in P. americanus
at Stage III. a, extending forward from the region of the ear capsule.
The whole section of the bar has been drawn, so as to show the difterence
in appearances at the two ends. The cell bodies {cl. crt.) at the anterior
end of the bar are much shrunken and the intercellular ground sub-
stance has for the most part disappeared. The nuclei are much crowded,
have lost the characteristic form seen in most normal nuclei, and are
angular and dense in appearance.

The degenerating portion of the cartilage is darker than the un-
changed cartilage cells next to it. The connective-tissue sheath {tu.
cont. lis.) around the cartilage is, however, persistent and can be
traced to the ethmoid.

In this specimen there is a coagulum filling the space in which the
degenerated portion of the cartilage bar formerly lay. The presence of
this coagulum is easily accounted for on the assumption that the sheath
has retained the material resulting from the degeneration of the carti-
lage cells, and that the killing fluid has caused it to be precipitated.
This condition is similar to that observed by Looss ('89) in the resorp-

WILLIAMS : MIGEATION OF EYE IN PSEUDOPLEURONECTES. 19

tion of cartilage in the tail of the tadpole. In that case, according to
Looss's interpretation, it was the chorda sheath which restricted the
diffusion of some of the products of the degenerating cells. He, too,
found that the intercellular substance was the first to disappear in
resorption.

Whether the cartilage nuclei, when set free by the disintegration of
the intercellular substance, degenerate completely, or join the nuclei of
the connective tissue, I cannot determine. There is much resemblance be-
tween the compact nuclei of degenerating cells and those of the sheath.

Since the bar disappears first in the middle region, there are, for a
short time, two degenerating regions, one which will end at the ethmoid
and the other at the persistent stub in front of the ear capsule. The
location of these will be evident by reference to Plate 2, Figure 10 {trb.
su'orb. s. a. and p.).

When in P. americanus the frontal of the eyeless side is formed, its
main body takes the position of this posterior stump of the left supra-
orbital bar. It is significant that there is no more space provided by this
degeneration than is barely necessary for the ready passage of the eye.

The body of the ethmoid is very irregular in shape. Besides the two
wings with which the supraoi'bitals are connected, there is a median-
elevation in the sagittal plane of the fish {ms'eth., Fig. A), and a forward
knob-like projection {crt. orb. a.) in the same plane. The two olfactory
pits lie just in front of the wings of the ethmoid, and the olfactory nerves
pass to them through the two deep notches {i'cis. eth. dx. and s.) seen
on the dorsal surface of the cartilage. The right nerve passes between
the supraorbital bar of the right side and the median elevation ; the
left nerve between the left supraorbital stub and the median elevation.
In this left notch the superior oblique muscle of the left eye takes its
origin, and in some cases the superior oblique muscle of the right eye
has its origin also close to that of the feft eye, therefore at the left of the
sagittal plane.

c. Stage HI a.

Figure B is photographed from the model of the cartilages of a fish of
Stage HI. (Plate 1, Fig. 2), where the left eye could be barely seen pro-
jecting over the top of the head as the fish lay on its left side. The left
wing of the ethmoid cartilage (ec'eth. s.) has no longer any trace of the
projection repi-esenting the anterior portion of the left supraorbital bar.
The posterior portion of the bar (trb. su'orb. s. p. ) projects forward from

20

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

tlio ear capsule substantially as in Stage II., there being just room for
the eye — now, of course, increased in size — to pass between the front
end of it and the ethmoid. The right supraorbital becomes a little
more arched as the fish increases in depth. The wings of the ethmoid
extend out from the mid-line farther proportionally and are more flat-
tened antero-posteriorly. Upon the surface of these wings of the eth-
moid cartilage the ect-ethmoid bones, or pre-frontals, are later formed.

irb. su^orb. s. p. iW.». eth.jlz.

t)b. sit'orb. <lj-.f-

'}

crt. mk. dx.

ms-eth.

'•c'elfi. s.

. . /or. olj s.
_ '•;■/. orb. a.
...pt-pal. s.

ha-hy.

Fig. D.

Oblique view of the facial cartilages of P. americanus, Stage III. Photographed from a

model, as m the case of Fig. .-l. X circa 75.
For meanmg of lettering, see Abbreviations under Explanation of Plates.

The gape has been greatly increased by the growth in length of all
the facial cartilages, but these have not increased in diameter propor-
tionately. The pterygo-palatine bars, which from the first support the
upper jaw, in lengthening have come to lie nearly parallel to Meckel's
cartilage, and their articulation with the fjuadrates is so far posterior
that the one of the left side alone falls within the region modelled. At
this stage these cartilages are in some instances so reduced in diameter
toward their posterior ends, as to show in cross sections only one cortilage
cell. A process from the left wing of the ethmoid has fused with the

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 21

median region of the ethmoid, thus bridging over the left ethmoid notch
and leaving between the mes-ethmoid and the region of the anterior end
of the right supraorbital cartilage an orifice {for. olf. s.), which corre-
sponds to the notch on the right. In other specimens I find that both
wings of the ethmoid have sent out processes to fuse with the mes-
ethmoid, thus converting both notches into foramina for the passage of
the olfactory nerves to their capsules on the front of the ethmoid.

In this model a bent wire is inserted into the mes-ethraoid in the median
plane to aid in locating the position of that plane, — the plane in which
the future interorbital septum is to develop. There is as yet no trace
of this septum in the specimen modelled ; but Figure 18 (Plate 4) shows
a cross section of the head of a fish (P. americanus) of this stage, which
does indicate the position of the future interorbital septum. The fine
vertical lines outside the figure represent the projection of the sagittal
plane of the fish. A small bar of cartilage (arc. eth. m.) is seen in cross
section above the mes-ethmoid. Traced anteriorly a few sections, this
fuses with the ethmoid. Traced posteriorly it soon unites with the
thin fused trabecular cranii not far from where they pass over into the
ethmoid. It is, then, a slanting bar, or arch, from near the anterior
end of the trabeculcC cranii to the posterior face of the ethmoid. In an-,
other specimen (Figure C", p. 24) this arch has become larger and ap-
pears as the forwai'd prolongation of the trabeculise (trb.). In the space
beneath this arch lie the oblique eye muscles, two of which (the right
and left inferior oblique) appear in Figure 18. The same figure shows
that the migrating eye may exert pressure directly on the cartilage, for
the left eye-ball is indented by the left wing of the ethmoid.

In another specimen of this stage, which had lost the migrating eye
in the process of turning, there were certain peculiarities worthy of con-
sideration. This fish, too, had a well-developed median arched cartilage
on the posterior f;%ce of the ethmoid. "The right superior oblique muscle
had its origin at the angle produced by the junction of the arch and the
body of the ethmoid. The inferior oblique was attached lower, at the
angle made by the union of the ethmoid and the trabeculse. The pos-
terior face of the ethmoid is the usual place of attachment for these
muscles, though a specimen of B. maculatus had both the inferior and
superior oblique muscles attached on the median arched bar. The most
noticeable peculiarity of this specimen was shown in the origin of the
supraorbitals. As I have said, there was no eye present on the left
side. The anterior end of the left supraorbital bar still persisted in this
specimen in the form of a stub projecting backward and slightly upward

22 bulletin: museum of comparative zoology.

from the left wing of the ethmoid, though unmaimed individuals whose
cartilages were otherwise in a like stage of advancement showed no
traces of it. Furthermore, the stub, instead of disappearing by a grad-
ual reduction of its diameter iu the region midway between the ethmoid
and the ear-capsule, through which the eye normally passes, preserved
the bar-like shape — the flat side being directed towards the top of the
head — until its abrupt disappearance behind the middle region of what
should have been the path of the migratory eye. Both supraorbitals,
instead of being backward extensions of the wings of the ethmoid, as in
most other specimens examined, took their origin from a mes-ethmoid
enlargement which extended backward directly above the median arch
that indicates the position of the future interorbital septum. In this
specimen there was, therefore, a suggestion of a tegmon cranii, such as
has been described by Parker for the salmon. This, instead of being a
complete roof, however, was a comparatively narrow plate of cartilage
which extended backward toward the brain region.

In describing tlie model of Stage II., a prominence (Figure A, crt.
orb. a.) on the front face of the etinnoid was mentioned. Tliis prom-
inence is really a separate cartilaginous mass, resting iu a socket of the
ethmoid. There is also a pair of small labial cartilages in front of and
below this plate ; but owing to their small size and the difficulty of pre-
serving small detached processes on the wax plates, they have been
omitted from the models. In Stage III. this large cartilaginous mass
has become rounded and projects further forward from the body of the
ethmoid. Its future history will be given in connection witli the de-
scription of the most advanced stage modelled (Figure D).

d. Stage III b.

The forms of the cartilages change very rapidly at this stage of
development, and it is with some difficulty that one finds a cranium
exhibiting a condition intermediate between Stage III a (Fig. B) and
Stage IV. (Fig. D), which shows the completely twisted head. How-
ever, I found one fish, larger than many of the recently metamorphosed
specimens, whicli I have designated as Stage III b, to distinguish it
from the more common condition just described as Stage III a.

In this specimen (Figs. C and C) the left eye lies in the sagittal
plane, even though the fish is 15.5 mm. long, the eye usually being
transformed when the fish reaches a length of 13.5 to 14 mm. There
is i]0 trace of the left supraorbital bar. The right supraorbital {trb.

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEUllONECTES. 23

su'orb. dx.), as but now described for the specimen that had lost the left
eye, is the backward extension of a plate of cartilage which connects the
right ect-ethnioid with the naedian mes-ethmoid arch. This flattened
anterior portion of the right supraorbital cartilage corresponds to the
tegmen cranii of the right side of the head in the salmon. The median
mes-ethmoid arch is, at its anterior end, fused to this plate or partial

irb. su''orb. dx. trb-

ec'eth, s.

for. olj. s.

cri. orb. a.
pt-pi:l. a.

bn-hy.

vrl. mk.

Fig. C.

From photograph of wax model of the fackil cartilages of a large specimen of
P. americnnus intermediate between the stages shown in Fig. B. and
Fig. />. Viewed from a point nearly in front, only a little to the right of
the sagittal and a little above the horizontal plane. X 45.

For meaning of lettering, see Abbreviations under Explanation of Plates.

tegmen, but from the short region of fusion backward for some distance
the two cartilages are merely crowded closely together, a distinct line
of perichondria! connective tissue being found between them. The car-
tilages then diverge, as may be seen in Figure C, and the median mass
continues backward as the fused trabeculae cranii, while the higher, lateral
portion, the right supraorbital bar (trb. su'orb. dx., Figs. C and C),
passes upward ami backward to the ear capsule.

24

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Ill older specimens this right supraorbital begins now to disappear,
the disappearance progressing from behind forward as the ensheathing
ocular-frontal takes its place and function. The remnant of this carti-
lage {ham. eth.), as it appears at a later stage, when it has been forced
into the horizontal position (vertical as the fish lies on its side), is shown
There is no longer a region of close appression without

iu Figure D

trh. .'fn''orb. dr.

ecV">. .».

Fig. C.

Same model as that shown in Fig. C, viewed obliquely from right side and
behind. A probe is thrust through the right olfactory foramen. X 45.
. For meaning of lettering, see Abbreviations under Explanation of Plates.

fusion between it and the median arch, but the hook arises directly from
the arch.

In Figure Ca bristle is shown passing through the left olfactory for-
amen, to indicate the axis of the opening, ^'hich now is not parallel to
the longitudinal axis of the fish, — as the right olfactory foramen still
is, — but makes with it an angle of about 45 degrees, being directed
caudad, mediad, and dorsad. In Figure C a white probe marks the
position and direction of the right opening.

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 25

There is also indicated at this stage a beginning of the forward rotation
of the dorsal margin of the ect-ethnioid cartilages about a transverse axis
passing through them. The end of the bristle (Fig. G) over the trabec-
ule cranii is, therefore, not greatly posterior to the outer end, which is
seen against the left pterygo-palatine as a background. The final result
of this rotation of the ect-ethmoids about the axis connecting them is to
make the axes of both foramina transverse instead of longitudinal. Con-
sequently in an oblique view from the right side, as in Figure D, one is
looking at the olfactory foramina from that face of the ect-etiimoids
which at an earlier stage (Figs. A, B) was directed posteriad. Instead,
therefore, of seeing the ends of the olfactory nerves wliich are distal to
the foramina, as would be the case if the cartilages were viewed from
the same direction at an earlier stage (Figs. A, B, and C) one would
now see the'iv proximal ends.

A twisting of the ethmoids (in a clockwise direction when viewed
from behind) about the antero-posterior axis of the fish, greater than is
indicated in Figure C, results in the further elevation of the ect-ethmoid,
olfactory foramen, and pterygo-palatine of the left side, while the supra-
orbital, the ect-ethmoid, the olfactory foramen and the pterygo-palatine
of the right side are correspondingly depressed.

e. Stage IV.

The oldest facial region modelled (Fig. D) — that of a small fish
(Plate 1, Figs. 5, G) having the eyes in the adult position — represents
my Stage IV.

The eyes are located one on each side of the flat hook-like plate of
cartilage (Fig. D, ham. eth.) which, with the previously mentioned
median arch {arc. eth. m.), runs back along the morphologically median
plane (the plane between the eyes). Tlie interorbital septum of con-
nective tissue is continuous with these^two cartilaginous processes, filling
the space between them and extending thence backward. That this
occupies the morphologically median plane, is proven by the position of
the olfactory nerves, which lie one on each side of this septum. Ante-
riorly the left nerve passes through the opening (for. olf. s.) seen in the
left (now upper) wing of the ethmoid and ends in the nasal capsule,
which lies immediately in front of it. The right nerve comes from be-
low the hook-shaped cartilage and passes through a foramen (for.
olf. dx.) in the anterior part of the ethmoid to the right nasal capsule,
which is located somewhat in front of the ethmoid and near the anterior
end of the right pterygo-palatine.

26

BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

The external opening of the left nasal pit is about 30° higher in
Stage IV. (Fig. 5) than in Stage II. (Fig. 4).

The superior oblique muscles of the eyes have their origins at or near
the junction of the median arch with the mes-ethmoid. The inferior
oblique of the right eye is attached to the ethmoid on the dorsal (mor-
phologically left) side of this median arch and that of the left eye im-

Irb. arc. elh. m. ec''eih. "

hnm. till

pt pal. dx.

rl. orb. n.

pt-pal. <!z.

for. olf. dx.
■ 'ec''eth, dx.

crt. mk

ba-hy.

Fig. D.

Oblique view of the facial cartilages of P. americanus. Stage IV. Viewed from tlie same

direction as in Figs. .1. and B. X 70.
For meaning of lettermg, see Abbreviations under Explanation of Plates.

mediately behind that of the right. The large passage ^ between the
ethmoid in front, the median arch at the right (morphologically dorsal),
and the trabeculsc cranii at the left (ventral) shown in Figures C and D
has therefore in the growth of the cartilage been loft to accommodate
the oblique eye muscles, just as the olfactory foramina in the ethmoid
were left because of the presence of the olfaQtory nerves.

The now ventrally projecting right ect-ethmoid partially hides in a

1 Tliis passage is seen in Figure C" directly above the pointed end of the probe
inserted through the right olfactory foramen ; it is indicated in Figure D by a
triangular area at the right of the dotted line, arc. eth. m.

WILLIAMS: MIGRATION OF EYE IN rSEUDOPLEUEONECTES. 27

lateral view the pterygo-palatine of its own side. The pterygo-palatine
(Fig. D) ends abruptly at its posterior end, since the membrane bones
which are to supersede it in supporting the upper jaw are already de-
veloped there.

The left pterygo-palatine {pt-pal. s.) is visible in Figure D only in
the region between the left ect-ethmoid and the cartilage sphere (crt.
orb. a.) in front of the ethmoid. This terminal spherical mass of car-
tilage {crt. orb. a.) can be traced to its position in the adult skull. In
a fish two inches long the ethmoid cartilage had pushed its way under
this spherical cartilage, which had elongated in antero-posterior direc-
tion, but was still located between the nasal pits. I regard it, there-
fore, as the cartilage which forms in the adult the median anterior por-
tion of the single orbit in which the left eye is to be found. The nasal
bones lie on either side of it, and the rest of the orbit is made up of the
right frontal, the left frontal and the left pre-frontal, or ect-ethmoid,
liones.

By comparing the position of the olfactory openings in Figures, B, G,
and D, it is plain that there has been a twisting of the ethmoid region
from left to right, through an arc of 90 degrees. The line joining the
centres of the ect-ethmoids in Figure B is horizontal, whereas in Figure
C it makes with the horizon an angle of more than 30 degrees, and in
Figure D is vertical. But with this twisting about the longitudinal
axis the plane of the ethmoids has also revolved from a transverse
position into one nearly coinciding with the sagittal plane, — possibly
due to the pressure caused by the increase in the size of the eyes, — so
that the axes of the olfactory foramina, which at first were parallel to
the long axis of the fish, now pass from right to left. Accompanying
these torsions, there has been a shifting in the relative positions of the
olfactory foramina and surrounding cartilages till those of the right side
are considerably in advance of those of the left. It is, however, the twist
about the longitudinal axis which makes the migration of the eye seem
rapid. This occupies in my experience not over three days, and accord-
ing to Nishikawa ('97) it was completed in the fish which he observed
in twenty-four hours.

The wliole of the cartilaginous system of the facial region has been
supported up to this time by two cartilage rods, the fused trabeculse
cranii {trb., Figures A-D ; Plate 1, Fig. 7 ; Plate 2, Fig. 10; Plate 3,
Fig. 17) and the right supraorbital bar {trb. su'orb. dx., Figures A-D ;
Plate 2, Fig. 10 ; Plate 4, Fig. 18).

The twisting is greatest in the optic region, the brain case showing

28 bulletin: museum of comparative zoology.

little of it, and the anterior part of the ethmoid, as seen by the final
position of the anterior ends of the pterygo-palatines, having turned
not more than 45 degrees.

In the turbot, according to Traquair ('65, p. 276), the nasal region is
nearly normal in position, the sagittal plane of the anterior part of the
head nearly coinciding with that of the body.

f. Comparison of Bothus with Pseudopleuronectes americanus.

The nearest representative in American waters of the sinistral turbot
is Bothus, the sand-dab, and I shall now compare briefly its turning
with that of P. americanus. The sand-dab is much deeper than the
flounder, but being thinner, though of the same length, it weighs about
the same as that fish. Its translucency has gained for it the name of
window-pane.

Traquair's statement that the turbot is less unsymmetrical than the
plaice holds as truly here, the sand-dab being less distorted than the
winter flounder. The mouth is straight and the length of the jaw on
the ocular and eyeless sides is more nearly equal. The mouth is much
larger and the gape greater than that of the winter flounder. The nasal
pits are very nearly symmetrical, that of the right side being, however,
a little the higher (Plate 3, Fig. 13). The transposed eye is not at all
posterior to its mate, as is the case in P. americanus. The dorsal fin in
this species reaches forwai-d entirely past the riglit eye (Plate 3, Figs. 13,
16, crt. pin. (1.). After the passage of the eye, the bases of the fin rays
arise nearly over the right wing of the ethmoid.

The ethmoid is relatively a much more slender cartilage in Bothus
than in P. americanus. The cross section of its anterior end (Plate 3,
Fig. 13) has the shape of an inverted letter T, and its dorsal margin is
turned not more than 20 degrees to the left from the sagittal plane. In
the posterior region (Fig. 16) the ethmoid is turned about 45 degrees.
The relation of the cartilage marked trb. su'orb. s. to the ethmoid mass
in Figure 16 indicates the angle, though the median bar itself is farther
forward. The wings of the ethmoid fuse to the median bar in a peculiar
way. The right wing (ec'etk. dx. Fig. 13) points toward the rays of the
dorsal fin which lie next it. It does not connect with the basal part of
the ethmoid directly, but merely with the median upright part. The
left wing has a process running anteriorly into the region of the lip
at the level of the basal part of the ethmoid, with which this wing is
fused. It then passes around the olfactory nerve of its own side, be-

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 29

coming much thinner as it does so, and unites with the upright bar.
Thus the foramen for the left nerve (/. s., Fig. 16) has a very thin outer
wall, while for the right olfactory nerve (/. dx., Fig. 16) there is no
foramen. The olfactory nerves pass under the wings of the ethmoid to
the capsules, which are located on the front faces of the wings.

Since the head of Bothus is less unsymmetrical than that of P. ameri-
canus, there is a corresponding difference in the conditions of the supra-
orbitals. The right supraorbital (Fig. 16, trb. sii'orb. dx.) is crowded over
until it comes to lie directly over the median bar of the ethmoid, which is
continued backward into the interorbital septum. There it persists for
a distance equal to nearly one-half the diamete rof the eye in all the
specimens of Stage IV. (Bothus) which I have sectioned. It should be
said that Bothus reaches this turned stage at a much earlier age than
does P. americanus.

The left supraorbital is proportionately of larger diameter than the
persisting supraorbital in P. americanus, and it also lies nearer the mesial
arch, with which it is often connected. Such a connection sometimes
occurs in the winter flounder, the condition of which has been previously
described.

In the older specimens there is no separate supraorbital, but the
upper end of the upright mesial cartilage bears a wedge-shaped enlarge-
ment on the side toward the left eye (Plate 3, Fig. 16, trb. siCorb. s.).
When, in the more posterior sections, the mesial cartilage ends, this
enlargement persists, and can be followed iintil it reaches the ear region,
thus showing that it is the supraorbital cartilage. The cartilage form-
ing the mesial arch is heavier and extends farther back between the eyes
than in P. americanus. The result is as if some of the space between the
hook and the trabecular cartilage in Stage IV. of P. americanus {ham.
eth., Fig. D) were filled out solid, and the whole plate were thickened.

In the transformation of the cartilaginous skull into the typical
condition of the adult teleost, the skull bones, as is well known, may be
formed (1) by ossification in the subcutaneous fibrous tissue (paros-
tosis), or (2) by ossification between perichondrium and superficial
cartilage cells, gradually replacing both by bone (ectostosis). There are
no dermostoses, and, as in the case of the salmon (Parker,' 73), I saw no
indications of endostosis. Of the bones directly involved in the turn-
ing, the frontals originate as parostoses and the pterygo-palatines and
pre-frontals as ectostoses.

30 bulletin: museum of comparative zoology.

g. Discussion of Pfeffer's Work.

I have purposely omitted, up to this point, any comparisons with
Pfetfer's work. He is the only author I have found who deals with the
twisting in the larval Pleuroncctidoe from other than the external point
of view. Unfortunately, he does not give the name of tlie species on
wiiich his statements are based, nor are his papers illustrated.

In his earlier article ('86, p. 4) he describes the general conditions to
be found in very young Pleuronectidae. The general topography is
that of other young lish. The eye sockets — separated below by the
sphenoid [trabeculse cranii ?J, above by the " Zwischenaugen-Decke " —
communicate freely with each other in the intervening region. In the
interorbital and ethmoid regions there is a vertical ridge-like dermal
bone, having in cross-section the form of an elongated triangle, and sup-
porting the dorsal fin, which, in Pfeffer's specimens, reaches to the eth-
moid. This bone is still free from the cranium, and is the frontale
principale of authors.

The bulbils olfactorius, which at first is lodged in the " Zwischen-
augen-Decke," becomes crowded backward into the brain capsule. The
" Interorbital-Decke " [supraorbital bar X] is bent out toward the eye
side and twisted somewhat on its long axis, so that its transverse axis,
previously horizontal, now becomes oblique, slanting downward and out-
ward toward the ocular side, while the chief part, which was vertical, is
mostly resorbed by the migrating eye. As a consequence there now re-
mains between the migrating eye and the surface of the head on the
ocular side only the thin, glass-like, scarcely perceptible outer skin
which previously covered the dermal bones. At the same time the der-
mal bone known as the froutale principale has grown fast to the inter-
orbital roof-piece, and its course, at first straight from the median crest
of the brain capsule to the ethmoid, now makes a great bend. Onlj' its
basal part, in the form of a broad band remains, while the vertical (and
at first the larger) part has been resorbed. The upper part of the wing
of the etlmioid on the ocular side has fused with the fronto-orbital, and
the upper part of its outer margin is continuous with the now develop-
ing supraorbital cartilage or bone, while the wing of the eyeless side
remains free on all sides, not forming any.connection with the supra-
orbital of its own side.

This description of the relations of the wings of the ethmoid to the
supraorbitals resembles the condition which I have found in Stage
III a of P. americanus (Figure B, pp. 19, 20) ; but in P. americanusand

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 31

in Both us the dermal frontal is not yet present in the region through
which the eye passes, and therefore cannot be resorbed. At Stage IV.,
t. e., after the migration is practically completed, there is to be found in
P. americanus under the surface of the skin behind the eye region a thin
plate of bone, which I take to represent the left frontal. The supra-
orbital cartilage of the side from which the migrating eye comes lies in
the region to which Pfeffer assigns the degenerating frontal in his
species, and we have seen that this bar is resorbed. Perhaps in his
species the dermal bone (frontal) is formed relatively earlier than in
P. americanus.

PfeflFer's statement that the transposition of the eye is accompanied
by a rotation on its own axis through an arc of 180 degrees is not quite
correct for our species. The arc in P. americanus varies slightly in dif-
ferent individuals, but is approximately 120 degrees.

Neither will his theory of the formation of the " Ivnochenbrtlcke " fit
the facts in Pseudopleuronectes. His argument (p. 8) is that when the
frontal bone of the blind side changes its position, dermal bone is pro-
duced, not only over it in its new position, but also in the region of the
integument beneath which the frontal was originally located, the latter
dermostosis being known as the "Briicke." In our species at least, the
trontal, when once formed, does not change its position. So its onto-
genetic location does not explain the formation of the " Briicke."

In PfefFer's more recent paper ('94) he states, as before, that very
young symmetrical Pleuronectidte have cartilaginous crania. The " In-
terorbitalbalken " [Interorbital-Decke 1] twists on its long axis, its dorsal
edge toward the future ocular side. One eye moves downward while the
other comes to lie upon the " Interorbitalbalken." If any sheathing
bone is already formed on the " Interorbitalbalken," the elevated eye
resorbs the part of the bone which is in its way. Then, on the side of
the upper eye corresponding to the blind side of the adult fish there
is formed a bony orbit, which fuses with the gradually developing dermal
bones, so that the skull of such an individual leaves the false impression
that the eye has traversed some of the bones of the skull.

The upper eye does not, according to Pfeffer, travel around to the
other side of the skull, but ascends only a little, until on a level with the
part of the skull between the eyes ; however, from this time forward it
looks in the direction of the ocular side. At the same time the thin
piece of skin (" Korperhaut ") now separating the cornea from the outer
world, disappears.

In regard to the last point, I may say that in both species I find a

32 bulletin: museum of comparative zoology.

layer of epidermis over the corneas of both eyes in the oldest fishes which
I have sectioned, as indeed one would expect ; so that Pl'effer's statement
apparently would have been more accurate if he had said " Lederhaut "
instead of " Korperhaut."

Unless the conditions in the species described by Pfeffer are totally
different from those found in P. americanus and Bothus, Pfeffer has not
distinguished between the cartilaginous supraorbital bar, which may be
in direct connection with the cartilaginous wings of the ethmoid, and
the dermal frontal bone, wliich fuses with ectoatotic bone-tissue formed
on the wings of the ethmoid.

h. Resume.

The twisting which takes place in the ethmoid region of the skull of
Pleuronectidae can best be explained by reference to the three mutually
perpendicular axes of the head of the symmetrical young. Tliere are
two important torsions of about 90 degrees each. The most evident
change (incidentally described by those who have discussed the migra-
tion of the eye) is that twisting of the ethmoids which can be rep-
resented by the revolution of the horizontal transverse axis until it
approximately coincides with the original dorso-ventral axis.

The second change (limited to the upper part of the ethmoid mass)
results in carrying the dorsal end of the dorso-ventral axis forward,
so that it coincides with the longitudinal axis of the head. This change
is probably due to growth along the anterior face of the ethmoids and
resorption of the posterior dorsal margin, which is pressed upon by the
eyes, or to a gradual displacement of the cartilage, due to the pressure
referred to, without absorption.

In Pseudopleuronectes there is a further complication due to a slight
retrocession of the parts on the eyeless side, amounting to about 30 de-
grees. This obliquity does not exist in Bothus.

The changes which have been described in the head of the flounder
all take place in the cartilaginous skull, ossification occurring only after
the shifting is complete. Therefore I cannot accept Pfeffer's view that
a portion of the " froutale principale " lying in the path of the migrating
eye is resorbed. The history of the two .supraorbital cartilages links to-
gether to some extent the cartilaginous and bony conditions. The
supraorbital cartilage bar next the migrating eye (the left in P. ameri-
canus, the riglit in Bothus) degenerates in its middle region, and the
eye is carried through the gap thus made by the unequal growth of the
facial cartilages of the two sides.

WILLIAMS: MIGRATION OF EYE IX PSEUDOPLEURONECTES. 33

Later the ect-ethmoid of the " blind " side is formed as an ectostosis
around the cartilage of that wing of the ethmoid and sends back a
process along the line which the supraorbital cartilage had occupied.
This meets and fuses with a forward process of the frontal of that side,
thus forming the "Brlicke," which becomes in the adult fish the most
voluminous bony support of the nasal region.

The supraorbital of the other side keeps its connection with the ear-
capsule much longer. Since the non-migrating eye moves downward to
only a slight degree, the sspraorbital has small space for movement to
evade the pressure of the tissues in front of the migrating eye. So we
find, in the latest stages in which this supraorbital appears at all, that
the structures of the median plane have been crowded over upon the
supraorbital and that this now appears as the cartilage " hook " (ham.
eth., Fig. JD), which extends backward between the eyes and is at this
time the chief tissue separating them.

In Bothus each frontal bone, when formed, sends forward a slender
process between the eyes, but in P. americanus the process arises from
the frontal of the ocular (right) side only.

V. The Optic Portion of the Central Nervous System.

1. General Condition in the Adult.

If the brain of the cod be taken for comparison, the axis of the cerebro-
spinal part of the nervous system of P. americanus shows bendings that
seem not to exist in the cod. There is in the spinal cord a bend which
is convex upward (dorsad) and is apparently induced by the size of the
digestive organs. In front of this, in the region of the medulla, occurs
a bend which is convex ventrad (Plate 1, Fig. 6). Finally there is
also a decided bend whicli is convex towards the eyeless side (Plate 2,
Fig. 11). The muscles of the eyeless side being less developed, that side
is more nearly flat than the oculai- side, which is convex.

Figure 8 (Plate 2) is a dorsal view of the brain of a fish (P. ameri-
canus) three inches long. The curves mentioned are not yet empha-
sized. An evident sign of asymmetry is seen in the inequality in the
size of the olfactory lobes, that of the right side being much the larger.
This lobe may, in the adult, have six times the volume of that of the
left side (compare Fig. 11). The relative sizes of the lobes of the cere-
brum is different in different individuals. In the specimens shown in
Figures 8 and 9 (Plate 2) and in Figure F (p. 36) the left lobe is the
larger ; but in a number of adult fishes the right lobe was the larger.

VOL. XL. NO. 1 3

34 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

The optic lobe of the loft side is usually cut first in cross-sections, when
one begins the cutting at the anterior end of the animal, as is plain from
the relative positions of the two in tiiis specimen (Fig. 8). The course
of the optic nerve to the transposed (left) eye is shown by dotted lines
(//. s.) in the figure. Its slack condition allows the eyes to be thrust
upward when the fish is buried in the mud or sand. One or two move-
ments of the fins will cover a fish with loose sand ; except for the pro-
jecting eyes, the animal is then entirely concealed. This protrusion of
the eyes is done by means of the so-called orbital heart. This organ,
mentioned by Agassiz in his description of tlie developing flounder, is
described as the recessus orbitalis by Holt ('94). It is shown in cross
section at rec. orb. in Figure 18 (Plate 4).

A side view of the same brain as that shown in Figure 8 (Plate 2) is
seen in Figure 9, which makes clearer the position of the brain with
reference to the eyes ; but in the dissection the left eye has been raised
somewhat from its normal position in order to show the eye muscles
and the location of the optic nerves, which are purposely shaded some-
what darker than the surrounding muscles.

In all the flatfishes which I have examined, the optic nerve from the
transposed eye is dorsal (anterior) in the chiasma. In P. americanus
the right optic tract and the left optic nerve are anterior (dorsal) to the
corresponding parts of the opposite sides (Fig. 12), whereas in Bothus
the left tract and the right nerve are anterior (dorsal).

Figure 11 is drawn froni a dissection of the adult fish. The oculo-
motor nerve (HI.) supplying the transposed eye passes toward the eye-
less side before it divides into the four customary branches. The fourth
cranial nerve (IV.) is still more noticeably changed in its direction. In
the cod this nerve lies near the mediaii plane, at a distance from and
above the eyeball ; but in the flounder the fourth nerve of the migrat-
ing eye lies in contact with the eyeball and rests on the dorsal rectus
muscle. The optic nerve (Figs. 8, 11) also shows before reaching the
eyeball a bending in the same direction as that which the eye-muscle
nerves exhibit. These alterations in the directions of the nerves in the
adult indicate the nature and the place of the transposition which we
have followed in the larvae, and show that nerves retain throughout life,
as far as possible, their phylogenetically nT)rmal position. I was unable
to find from my dissections that the flounder, P. americanus, has a cuta-
neous branch of the fifth nerve. If it has, the nerve must be small. The
fifth has a mandibular, a maxillary and a superior ophthalmic branch.
The large ophthalmicus profundus of the cod is represented in the flounder

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEUEONECTES. 35

by a few twigs only (V. opt. j^fiid., Fig. 11). The left superior ophthal-
mic of the flatfish {V. opt. su.), after emerging from the skull with the
rest of the fifth nerve, as in the cod, runs from left to right (Fig. 11)
through the passage formed by the "Brucke," which results from the
fusion of the posterior angle of the pre-frontal and the corresponding
anterior angle of the left frontal. It then takes the regular median path
between the eyes to its distribution on the snout. The bone is formed
around the nerve in its new position after the migration of the eye.

The seventh nerve in both the cod and the flounder emerges from the
skull with the fifth. The ninth in the cod lies between the two chief
roots of the tenth, with which it passes out. In the flounder the ninth
nerve lies in front of the tenth and passes through the ear capsule to its
distribution on the hyoid and first gill arch.

2. The Optic Nerves.

In the cross-section of a fish in Stage I. (Plate 3, Fig. 1 7), one section,
lOyu, thick, contained the whole length of both optic nerves from the
blind spot to the chiasma. The blind spot is very near the outer ventral

ch. dx.

id. opt.

_ oc. mig.

A precisely front view of tlie fore part of the brain, the optic nerves and
a portion of each of the optic cups, modelled in wax (Born's method)
from a specimen in Stage III. X 50.

For explanation of lettering, see Abbreviations under Explanation of
Plates.

edge of the retina and in about the middle of the eye antero-posteriorly.
Therefore the chiasma is in the transverse plane which passes through
the middle of the eyes. There is, as yet, scarcely any want of symmetry,
the left eye being only slightly higher than the right.

36

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

I have no corresponding illustration of the condition of the optic
nerves in Stage II., but a model of the anterior part of the brain and
the optic nerves of a specimen in Stage III. a is shown in Figure E {\\\q
anterior portion of the left optic cup has been omitted in the model ; the
cut surface being indicated by horizontal lines). The left eye is higher
than the forebrain ; its ventral edge is at the same level as the dorsal

tcl. opt. s.

oc. m%g.

cb, dx.

.. cb. s.

n. opt. s.

efts. opt.

^

Fig. F.

Front view of the fore part of the brain, the optic nerves and portions of the
optic cups in Stage IV. From a model (Born's method). X 50.
Compare Fifj. E.

For meaning of lettering, see Abbreviations under Explanation of Plates.

side of the right eye, and the transverse plane tangent to its posterior
surface would cut the right eye about midway between its anterior and
posterior faces. The right eye may have moved slightly ventrad from
the position which it occupied in Stage I", The slackness of the nerves
is shown by the curve that they take as "they ])ass forward and out-
ward. The whole of the midbrain and most of the forebrain have lost
their earlier position between the eyes, owing to the growth in length
of the facial cartilages. Figure 9 (Plate 2), a side view of the brain of
a fish three inches long, shows this antero-j)OSterior separation between

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONEGTES. 37

braiii and eyes farther advanced, and Figure 1 1 (from an adult) shows
it completed.

In the essentially adult condition of Stage IV., as shown in a front view
of the modelled brain and optic nerves (Figure F), the left eye has passed
so far to the right side that, taking into consideration the high degree of
mobility of the eye its field of vision almost coincides with that of
the right eye. The optic nerves curve still more in their passage
from chiasma to eye, and the distance is proportionately greater.
The right cerebral lobe (c6. dx.) is seen in the figure between the eyes,
and the left cerebral lobe {cb. s.) is seen on the right, behind the left
eye, and below the tectum. The left olfactory lobe is covered by the
left eye, but the right olfactory lobe — modelled as a continuation for-
ward of tlie right cerebral lobe — is seen between the two eyes. The left
optic lobe {tct. opt. s.) in both these instances (Figures £ and F) extends
farther anteriorly than the right. This is seen in the dorsal view of
the brain (Fig. 8). This figure also shows why in making cross-sec-
tions the left lobe of the cerebrum is cut before its olfactory lobe in
case one begins at the anterior end.

The optic nerve — round in cross-section in the larvse — becomes
thrown into folds in the adult (Plate 5, Fig. 24). This condition is also
figured by Studnicka ('97) for one of the Pleuronectidse. The cross-sec-
tion may show as many as six or seven folds closely pressed together.
Small neuroglia nuclei are scattered throughout the length of the nerve.

3. The Chiasma and Tracts with related Ganglia.

The optic crossing is complete as in all teleosts. There is no inter-
lacing of fibres, as can be seen in Figure 19 (Plate 4), which is from a
fish in Stage IV. This is an approximately transverse section, which,
however, cut the left side of the fish sdmewhat farther caudad than it
did the right side. The plane of the section also inclines a little back-
ward and upward, so that it coincides with the plane of the anterior part
of the left optic tract, which slants in Figure 19 backward and upward
on its way to the tectum. The right tract is cut crosswise, nearly at
right angles to its course. (This is by mistake lettered ?«. opt. s. in
Figure 19. Of course, as it is posterior to the chiasma, it should have
been labeled trt. opt. dx. For the second section anterior to this the
label n. opt. s. would be correct.) The median, dorsal portion of the
tract (trt. opt. d.) passes upward through the nidulus corticalis (to be
described later) on its way to the median portion of the tectum. The

38 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

external, ventral portion (trt. ojot. v.) passes outward and around to its
distribution on the posterior, lateral, and ventral tectal surfaces.

The geniculate body (Figs. 20, 21, cp. gyiic.) lies in the angle be-
tween the two portions of the Y-shaped tract, but almost entirely in
front of their plane. There is some indication of a division of the corpus
geniculatum into anterior and posterior parts.

In both Weigert and Congo-red preparations it could be seen that a few
optic fibres entered the geniculate bodies (Plate 4, Fig. 21). C. L. Her-
rick ('92, p. 430) found no ending of optic fibres before reaching the
tectum. This ending has been demonstrated, however, by Mayser ('81)
in Cyprinoids, by Auerbach ('88) in the trout, by Haller ('98) in Salmo,
and by Krause ('98), who used Marchi's method for degenerate nerves,
in Cyprinus auratus. Edinger ('96, p. 126), makes the following state-
ment for vertebrates. " Im Geniculatum [laterale] endet ein Theil
des Sehnerven mit mJichtiger Anfsplitterung, und mitten in diese Faser-
ung tauchen die Dendriten langgcstreckter Doppelpyramiden. Das
mediale Ende dieser Pyramidenzellen splittert auf in eiuem Zuge, der
wahrscheinlich auch dem optischen System angehort."

I have no Golgi preparations which show optic fibres actually fibril-
lating in these bodies. There was, however, in the geniculate bodies
but one type of cell impregnated with the chrome-silver. This was a
small unipolar cell (Plate 5, Fig. 22) with a short process ending in
very thick short fibrillations directed towards the end of the geniculate
body into which the optic fibres enter. In a single exceptional instance,
a cell, otherwise like the ones described, had another short but un-
branched process extending in the opposite direction (see diagram of
tectum, Plate 5, Fig. 22, cp. gnic).

Fusari ('87), after a study of Carassius, Macropodus, Anguilla, and
Lopodogaster, stated that in his opinion fibres from the tractus pass
through the corpus geniculatum and unite again with tlie tract to fibril-
late in the tectum. No preparations of P. americanus indicated such a
possibility.

No other bundle of fibres could be found to leave the tract before it
reached the tectum itself. Mayser ('81) describes a small bundle pass-
ing into the thalamus at about the point of origin of the paraphysis.
Auerbach ('88), Mirto ('96), and Haller ('98) also indicate a thalamus
bundle, and Haller describes a small bundle running to the fore-brain.
In my opinion Mayser, Auerbach, Mirto, and Haller have mistaken a
portion of the ventral division of the tract, which bends outward sharply
in its course to the ventral posterior part of the optic lobes, for a thala-

WILLIAMS: MIGKATION OF EYE IN PSEUDOPLEURONECTES. 39

mus bundle. In parasagittal sections the cut ends of this portion of
the tract appear to be pointing into the thalamus. But no one of these
authors has described fibrillations or cell endings for this thalamus
bundle, and the absence of degeneration in Krause's experiment would
indicate that Mayser's thalamus root was non-optic.

A frontal section (Plate 4, Fig. 20) shows the relation of the thalamus
ganglia to the tectum. The geniculate bodies lie anterior to the lobes
of the tectum, and between them are the ganglia habenulse (jjn. hah.),
which bound the third ventricle, and are separated from each other by
the pineal-gland region. A few sections dorsal to the one shown in this
figure the habenular commissure appears.

As Haller ('98) has found in the case of Salmo, the habenulse are
symmetrical, in the young fish at least. Because of the want of sym-
metry in older brains it is impossible to obtain single sections in which
one is certain that the habenulse are cut in like planes. In a cross sec-
tion which passes through both ganglia the left ganglion has a greater
dorso-ventral diameter than has the right, while the right ganglion
measures moi-e from side to side than the left.

In Figure 20 the fibres of the two pai'ts of the optic tracts are shown
in cross-section behind the edges of the geniculate bodies. Also behind
the geniculate bodies lie large cells which belong to the nidulus corti-
calis of Fritsch, the " Dachkern " of Edinger and others.

Since fibres from this nidulus enter the tectum, I will describe its loca-
tion more particularly in the two Pleuronectidaj studied. There are two
symmetrically placed groups of very large ganglionic cells lying at the
front part of the tectum ; they extend anteriorly from the angle of the
optic ventricles, where the lobe of the tectum and the axial portion of
the midbrain meet, to the outer surface of the brain above and outside
the geniculate bodies. There is no difl&culty in identifying the cells
of the nidulus {nid. ctx., Plate 5, Fig. 23), as they are pear-shaped aud
many times larger than those of the gray layer of the tectum, into which
the posterior portion of the nidulus extends.

The nucleus lies in the blunt end of the pear-shaped cell, at the end
opposite the coarse cell process. Since these processes gather into
bundles in the middle layers of the tectum, the nucleated ends of the
cells are directed towards the surface when the cells are more super-
ficial, but toward the optic ventricles if they are deep (compare
Fig. 22).

There is a similar nidulus, consisting of a few (20-30) even larger
cells, which lies ventral and exterior to the nidulus corticalis ; it lies

40 bulletin: museum of comparative zoology.

posterior to, but in contact with the optic tract. This possibly is the
nidulus anterior of Edinger, though I have traced no fibres from it. A
few cells of this nidulus are shown between the two portions of the tract
in Figure 1 9 (Plate 4) .

In one instance I found a cell of the nidulus corticalis which sent a
fine process, probably a neurite, ventrad with the other fibres of the
optic tract (Plate 5, Fig. 22), This could be followed nearly to the
chiasma, but whether it continued to the eye or bent backwards into
one of the post-optic commissures, I cannot say.

I can confirm C. L. Herrick ('91-'92) in his statement that the com-
missura horizontalis (corns, hz., Plate 5, Fig. 22) arises from the nidulus
corticalis. The fibres forming this bundle were fine and took the same
quality of Golgi impregnation as the single fibre just described from
one of the cells of the same nidulus which passed downward through the
tractus opticus. The fibres composing this bundle can be followed in
two or three parasagittal sections to the nucleus rotundum of the same
side ; they pass through this nucleus, and then turn forward and cross
to the opposite side behind the chiasma as the horizontal commissure.

4. The Tectum Opticum.

Since the tectum is that portion of the bi-ain in which the optic
tracts terminate, it should be the place in which the transition from
sensory to association or motor neurons takes place.

There are certain points of interest which can be shown fi'om a sur-
face view. At the anterior ends of the tectal lobes, in P. americanus,
but not in Bothus, there is an exterior furrow or sulcus (sul. tct. opt.,
Plate 2, Fig. 11), much like one that is found in the cerebrum of simple
type — in that of a turtle, for example. This gradually disappears toward
the posterior region of the tectum. Cross-sections in the anterior region
show that this sulcus is due to a lateral horizontal depression in each
optic lobe, which divides it into almost equal dorsal and ventral parts.
The ventral portion of the tractus supplies the ventral half of the lobe
and the dorsal portion the dorsal half. The geniculate bodies lie in the
region of greatest constriction of the tectum.

For convenience, I divide the tectum into seven layers, indicated by
the numerals 1-7 (Plate 5, Figs. 22, 23), in addition to the membranes
of the brain, wliich are the vascular connective-tissue layer (the arach-
noid, mb. ach.) and, beneath this, a very thin membrane, the pia, to
which the endings of the ependymal cells reach, and along which is
found here and there a nucleus.

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEURONECTES. 41

Passina: from without inward, the tectal layers are as follows :

(1) A thiu outer layer, composed principally of nerve fibrillations
with a few nerve cells. In this layer the ependymal fibrillations end.
A corresponding layer is recognized by writers on the finer anatomy of
the tectum in the bony fishes, from Stieda ('67) onwards, except by
Fusari ('87, '96) and Van Gehuchten ('95). Eusari ('87) described a
layer of vascular connective tissue beneath the pia, and later ('96) his
first layer of the tectum was made to embrace this vascular layer and
the optic-fibre layer.

(2) The layer of the medullated optic fibres. This is the continua-
tion of the optic tract and is recognized as a separate layer by all writers
on the tectum. '

(3) A layer of optic fibrillations. This is not made a distinct layer
by Stieda ("67), but Mayser ('81) and nearly all writers since his time
have emphasized its presence.

(4) A spindle-cell layer.

(5) The fillet layer,' composed of longitudinal fibres and cross com-
missural fibres. Stieda considered the fibres, which here run in two
directions, as two layers. C. L. Herrick ('91-92) describes a layer of
commissural fibres beneath the fillet connecting the two optic lobes.

(6) The " gray " layer.

(7) The reticulate and ependymal layer. Some authors consider
that this is composed of two distinct layers. The reticulate portion
is not described at all by Neumayer ('95), Van Gehuchteu ('95) nor
Edinger ('96).

Mirto ('96) based his division of the tectum into layers on the shapes
of the cells which he was able to demonstrate by the Golgi method.
Following Cajal's work on the tectum of birds, he describes fourteen
layers.

The degeneration methods did not yield much of importance in my
hands, although the flounder, owing to its habit of protruding the eyes,
is a favorable fish on which to operate. The animals, even the very
small metamoi'phosed fishes, stand the shock of the removal of the eye
well and bleed very little from the operation. The specimens tried by
the Marchi method were very brittle, and demonstrated but one point
clearly, that the sixth (nerve-cell) layer was reduced. Fusari ('96), who
used the "Weigert-Pal staining method on a Cyprinoid, concluded that
all the tractus fibres degenerated when the eye was removed. Krause
■ ('98), after the Marchi treatment of fish from which the eyes had been
removed, found that about one-tenth of the tract — mostly distributed

42 bulletin: museum of comparative zoology.

in the dorsal root, wliich spreads on the roof of the tectum — did not
degenerate. In a very old one-eyed fish both the geniculate ganglion
and the torus longitudinalis were, he found, much atrophied and the
fillet was reduced. The spindle-cell layer contained fewer cells than
were found in fishes more recently operated on.

Turning next to the finer anatomy of the tectum a diagrammatic rep-
resentation of a parasagittal section is shown in Figure 22 (Plate 5).
This exhibits the types of cells found in the tectum by the aid of the
silver method.

In layer 1 few cells were impregnated. Of these the more common
type (Fig. 22, a) vvas oval and bipolar, its two processes running parallel
to the fibres of layer 2. In some instances, however, the cell had a
third and even a fourth process. Similar cells have been described by
Fusari, except that the cell bodies described by him were spherical.
Neumayer ('95) has shown elongated bipolar cells with })rocesses parallel
to layer 2, and also rounded cells whose neurites fibrillated in the layer
of optic fibres. Mirto ('96) indicated cells in corresponding positions,
but with triangular bodies. I, also, have found a few peai'-shaped cells
(Fig. 22, j8) in this layer. These lay near the surface and sent off their
processes from their deeper, smaller ends. Some of these processes
passed through the optic layer (2) into layer 3, while others turned at
right angles and ran in layer 1 parallel to the surface.

Layer 2 is composed of the medullated fibres which enter the tectum
as the optic ti'act. At the beginning of the tectal region the fibres of
the tract, after having passed beneath the geniculate body, bend toward
the surface of the brain to form this second layer. Some of the cells
of the nidulus corticalis {nid. ctx.) lie in this layer, since the nidulus ex-
tends from ventricle to surface. The bulk of the dorsal bundle of fibres
from the tractus passes too near the sagittal plane to touch the nidulus
corticalis, and the ventral division does not reach as far dorsally as the
nidulus. So tliere is little disturbance in the course of the fibres of the
tractus in passing these very large cells. The diminution in the thick-
ness of the optic-fibre layer in passing from before backwards, which is
due to the fibres continually spreading out over more of the surface of
the optic lobe, and to the termination of many of them in anterior
regions, is shown in Figure 25 (Plate 5).

Here and there other cells, besides those of the nidulus corticalis,
which lie at the anterior end of the tectum, are seen in the optic layer;
these have fibres, some of which extend inward, others outward. The
cell-body of one of these (Fig. 22, y) was pear-shaped, the smaller end

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEUEONEGTES. 43

beiu» directed outward. From this smaller end processes ran both au-
teriad and posteriad, the most of them parallel to the surface ; one,
however, took an oblique direction, running forward and inward, and
reached layer 3. Neumayer represents in this optic layer spindle-
shaped cells, the upper ends of which fibrillate in layer 1, and the
lower in layer 3.

The third layer contains cells of many shapes, (a) Short spindle-
shaped cells (Fig. 22, 8) with one process directed outward and fibrillat-
iug in layer 1, and one or more processes directed inward. Cells lilie
these are described by Fusari, Neumayer, and Mirto, and the last two
authors say that the neurites are directed inward and reach the fillet
layer. Fusari also describes a type of cell which is spindle-shaped with
processes extending downwards and fibrillating just above the fillet
layer. A neurite of one of these cells is figured running through the
corona radiata of Gottsche ^ into the torus semi-circularis. (b) Pyri-
form cells (Fig. 22, e) with all the processes directed inward and the
ends of the fibrillations reaching into layer 4. (f) Eouuded cells
(Fig. 22, 0 with rather long sparsely branched processes, the outward
process having been followed in one case into the optic-fibre layer.
(d) Cells (Fig. 22, t]) the reverse of those denominated e in this layer,
with fibrillations having the opposite direction and reaching to, or even
through, the optic layer into layer 1. (e) Lying near the boundary
between this (3) and the next deeper (4) layer were found a few cells
(Fig. 22, 0) flattened in a direction perpendicular to the surface of the
optic lobes. Each of these possessed a process running from either end
parallel to the surface of the tectum and sometimes a third one passing
out towards the surface. At or near this transitional region between
layers 3 and 4 the fibres from most cells send oft" short branches parallel
to the surface.

I have separated layers 2 and 3 because in the anterior portion of the
tectum some fibres from the optic tract take a direct course into layer
3 without first bending outward into layer 2. In the posterior portion
of the tectum, however, it is not possible to distinguish these two layers.

Bundles of large processes from the nidulua corticalis (nid. ctx.) enter
the anterior portions of these two layers and form a prominent fibrilla-
tion, traceable for some distance backward. These coarse, wavy processes
are much larger than the fine fibres, which I have shown (p. 40) to be
the neurites which make up the horizontal commissure, and there may
be two or three of them from one cell. These coarse processes can be
1 This is the " Stabkranz," the descending fillet fibres.

44 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY,

followed backward for some little distance along distinct paths in layers
3 and 4, and the general appearance of the fibrillations farther back
indicates that these processes, branching continually, pass backward
through the tectum much farther than continuity can be directly traced.
A dendrite may branch and follow the fibrillar paths in eacii of the two
layers.

A large system of fibres also enters the same general region of the
tectum from the axial part of the mid-brain ; some of these cross from
the opposite side of the brain in the lower part of the posterior commis-
sure. These fibres may constitute the most anterior portion of the cora-
missura mesencephali (Herrick's sylvian commissure) or, as I think
more likely, they may come from the motor regions, possibly Ilaller's
anterior connective. I have not succeeded in tracing these fibres to
any cells.

In layer 4 appear the cells which are most characteristic of the tectum
(Fig. 22, i). They were impregnated in most of the Golgi preparations.
They are spindle-shaped, being much elongated in a radial direction,
and have fibrillations which extend outward as far as layer 2. Some-
times there is an impregnated process which goes from the deeper end
of the cell into layer 5, and sometimes there is not. Neumayer and
Mirto each state tiiat the neurites of these spindle cells are traceable to
the fillet layer and the fibrillations to the optic layer. ^lirto describes
cells with the same processes but witli much more slender bodies. The
spindle-shaped bodies are shown by my hematoxylin preparations to
be very abundant indeed in this layer, only a few taking tlie Golgi
impregnation in a single specimen. In this layer (4) there were also
found sparingly cells (Fig. 22, k) with rounded bodies and processes
which fibrillate inwards and extend into the fillet layer (5). A very few
pyriform cells lie near the deep surface of this layer (4) and send their
processes outward (Fig. 22, A). Fusari shows irregular, large-bodied
cells with many processes and neurites, when such are present, extending
into layer 5. A bifurcate cell is figured by Mirto with its telodendrites
in layer 3. My flounder impregnations produced neither of these types.

I have spoken of layer 5 as the fillet layer because it is composed
chiefly of fibres which pass backward and medianward, forming the so-
called corona radiata of Gottsche, the lemniscus or fillet system.

This layer is composed of cross and longitudinal fibres which, seen in
tangential section, form a meshwork over the whole of the dorsal part
of the tectum. In front of the optic ventricles bundles of fibres
(Plate 5, Fig. 22, Imn.) can be followed from the axial part of the mid

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEURONECTES. 45

brain through the region of the nidulas corticalis into the longitudinal
fibre layer. Most of the cross-lying fibre bundles, which form the com-
missura mesencephali, lie below the longitudinal layer. Some of these
cross bundles seem to turn longitudinally after crossing the mid-line.
It may be that the uncrossed fibres of the fillet are a continuation of
these. The longitudinal fibres, at any rate, pass back in bundles to the
reo-ion of the anterior peduncles of the cerebellum. In any section
which cuts through the whole thickness of the tectum, whether cross or
parasagittal, some bundles will be shown (Plate 5, Fig. 25, Imn.). As
the tectum is dome-shaped, the more nearly median parasagittal
sections will cut the fibre bundles at the anterior and posterior ends of
the tectum, whereas the more lateral sections will show the fibres of the
middle of the tectum cut longitudinally. There is a rather distinct
portion of the fillet which arises from the anterior ventral part of the
tectum and, slanting upwards and inwards, passes through the nidulus-
corticalis region back towards the cerebellum, beneath and behind the
median boundary of the optic ventricles. The fillet fibres may be
roughly likened to the slightly curved fingei's of an open hand, palm
inward, wrist beneath the cerebellum, grasping the most of the gray
laj'er of the tectum. The gray of the posterior portion of the tectum
seems, however, to be outside the region surrounded by the fillet- fibre
bundles.

The fibres of the commissura mesencephali cross just above the gray
layer in the anterior part of the tectum in the region of the torus longi-
tudinalis. According to Herrick they form a continuation of the series
found in the posterior commissure.

Besides these fibres, there are in layer 5 a number of different forms
of cells : (a) Cells with rounded bodies (Plate 5, Fig. 22, ju) of the same
size as those (Fig. 22, p) in the next deeper layer (6) — the gray layer
— and with processes which may fibrillate into any one or all of the more
superficial layers (1-4) of the tectum, (b) Spindle-shaped cells (Fig. 22,
v) like those (t) characteristic of layer 4. When an axonic process can
be followed from the deep end of such a cell, it finds its way into the
fillet layer, but whether into the cross or longitudinal system I cannot
determine, (c) Long triangular cells (Fig. 22, o) with a single process
extending toward the periphery, and from each of the corners of the
deep end a process I'unning parallel to the fillet layer. (c?) Rounded
cells (Fig. 22, ir) with fibres which turn immediately into the fillet
layer and with very short dendritic processes.

The next layer (6) is the gray molecular or granular layer. This is

46 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the most noticeable portion of the tectum, especially in young animals.
The nuclei are closely crowded together, with a definite arrangement
due to the radially directed processes of the ependymal cells, which pass
through all the layers from the ventricle to the pia. Only one type of
cell body (Fig. 22, p) is evident, that being the small and rounded
form ; in Golgi preparations, it is slightly pear-shaped, and resembles
much the ependymal cell. But since the cells of this layer have pro-
cesses of a number of types, they cannot all be, as Fusari ('96) main-
tained, ependymal cells. They may fibrillate in any or all of the layers
outside the sixth. In Golgi preparations a very few spindle cells, like
those in layers 4 and 5, appear. Some of the peripheral cells (Fig. 22, a)
of this layer, as well as the very deep ones, may send to the surface a
process which ends in branching fibrillations beneath the pia. The
fibres from other cells were found to break up in layers 3, 4, and 5.
These fibres are often impregnated when none of their processes take the
silver, or vice versa. The cells next to adjacent layers, whether the
deeper or those nearer the periphery, are more likely to become impreg-
nated than those in the middle of the layer.

The innermost layer (7), less dense than any of the preceding, is
composed of the bodies of the ependymal cells and the basal portions of
their processes. A reticulate portion of this layer (next to layer 6) is
not apparent in young specimens, and so I have not i*ecognized it as
a separate layer, but have included in layer 7 all that lies between the
gray layer (6) and the ventricle.

In the adult brain there are scattered through this loose layer a few
large-bodied very irregular cells (Fig. 22, t), each having a multitude
of long beaded processes. I was unable to discover any neurite con-
nected with these cells.

In order to simplify the diagram (Fig. 22\ I have omitted in all cases
the free fibrilLations. Inmost impregnations where there are any at all,
there are so many that only a few can be traced to any definite
medullated layer. Layer 3, however, certainly contains, among other
fibrillations, free branches from the optic layer (2). In layers 3 and 4
free fibrillations of fibres from cells in layer 5 are doubtful, because
any one of the many cells in the granular-layer (6) may have its fibre
impregnated though itself remaining clear..

Between the fillet layer (5) and the optic layer (2) there are two
especially dense fibrillar regions corresponding in gene/al to the two
bundles of dividing processes which arise from the cells of the nidulus
corticalis.

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 47

For the purpose of comparing the impregnation of the tectal region in
these Pleuronectidse with that of the same region' in a symmetrical fish,
in order to ascertain whether there are any noticeable histological dif-
ferences, I have applied the Golgi method to the brain of Fundulus
heteroclitus, the mud minnow. These were found to take the stain very
much more easily than do flounders; but there was also more of the
silver precipitate carried inward from the surface. I conclude, there-
fore, that the tissue in Fundulus must be more open. Except as to the
size of certain cells and the relative thickness of some fibre bundles,
the two brains correspond closely. The cells of the nidulus corticalis
in the minnow are much smaller proportionately, though their tectal
processes can be followed in layers 3 and 4 as far as in the Pleuro-
nectidse. The spindle-shaped cell found most abundantly in layer 4
was again in the minnow the most noticeable cell impregnated, and was
found most often. A triangular cell in layer 5, very similar to the
cell o found in the corresponding layer of the flatfish, had its outward
process extended to layer 1, where it fibrillated like an ependymal cell.

Most of the cells of layers 3, 4, and 5 in Fundulus had neurites
traceable into layer 5, the fillet layer.

VI. Theoretical Considerations.

The conditions in the tectum are the same as those found in the optic
lobes of typical Teleostei. The division of the tectum into layers is of
importance as a means of more precise description. There must be a
place where the fibres of the optic tract, which come in as layer 2, end ;
that region is layer 3. There must be an association system connecting
with the posterior motor regions, and the fibres of this system are either
a part or the whole of layer 5. If only a part, then the purpose of the
commissura mesencephali is to put the two optic lobes in communica-
tion with each other. The cells in layers 3, 4, and 6, especially the
spindle cells in layers 3 and 4, probably serve to receive and transmit
optic stimuli.

The nidulus corticalis, developing early, as it does, is probably one of
the most effective association centres of the brain.' Lying at the entrance
to the tectum, with a strong bundle of neurites running through the two
niduli rotundi in the ventral part of the brain, and with its numerous
large dendrites passing into layers 3 and 4 of the tectum, it should be
able to connect the optic sensory region with the motor areas quickly,
and thus account for the extreme rapidity of movement of these larvae.

48 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

The " why " of the peculiar metamorphosis of the Pleuronectidso is an
unsolved problem. The presence or absence of a swim bladder can have
nothing to do with the change of habit of the young flatfish, for P.
americanus must lose its air-bladder before metamorphosis begins, since
sections showed no evidence of it, whereas in Bothus the air-sac can
often be seen by the naked eye up to the time when the fish assumes
the adult coloration, and long after it has assumed the adult form.

Cunningham ('92-97) has suggested that the weight of the fish acting
upon the lower eye after the turning would press it towards the upper
side out of the way. But in all probability the planktonic larva rests
on the sea bottom little if at all before metamorphosing. Those taken
by me into the laboratory showed in resting no preference for either side
until the eye was near the mid-line.

That the change in all species is repeated during the development of
each individual fish, has been used to support the proposition that the
flatfishes as a family are a comparatively recent product. They are, on
the other hand, comparatively ancient. According to Zittel ('87-90, pp.
315-316) flatfishes of species referable to genera living at present,
Rhombus and Solea, are found in the Eocene deposits. These two
genera are notable in that Rhombus is the least and Solea the most
unsymmetrical of the Pleuronectida).

The degree of asymmetry can be correlated with the habit of the ani-
mal. Those fishes, such as the sole and the shore-dwelling flounders,
wliich keep to the bottom, are the most twisted representatives of the
family, while the more freely swimming forms, like the sand-dab, summer
flounder and halibut, are more nearly symmetrical. Asymmetry must
be of more advantage to those fishes which grub in the mud for their
food than to those which capture other fishes ; of the latter, those that
move with the greatest freedom are the most symmetrical.

This deviation from the bilateral condition must have come about
either as a "sport," or by gradual modification of the adults. If liy the
latter method, — the change proving to be advantageous, — selection
favored its appearing earlier and earlier in ontogeny, until it occurred in
the stages of planktonic life. Metamorphosis at an age younger than this
would be a distinct disadvantage, because of the lack of the customary
planktonic food at the sea-bottom. At present some forms of selection
are probably continually at work fixing tlie limit of the period of meta-
morphosis by the removal of those individuals which attempt the trans-
formation at unsuitable epochs, — for instance, at the time of hatching.
That there are such individuals is shown by FuUarton ('91), who figures

WILLIAMS: MIGKATION OF EYE IN PSEUDOPLEUKONECTES. 49

a fish just hatched "anticipating the twisting and subsequent unequal
development exhibited by the head of Pleuronectids." Those larvae
which remain pelagic until better able to compete at the sea bottom
become the adults which fix the time of metamorphosis on their progeny.

VII. Summary.

1. The young of Limanda ferruginea are (probably) in the larval stage
at the same time as those of Pseudopleuronectes americanus.

2. The recently hatched fish, both P. americanus and Bothus, are
symmetrical, except for the relative positions of the two optic nerves.

3. The first observed occurrence in preparation for metamorphosis in
P. americanus is the rapid resorption of the part of the supraorbital
cartilage bar which lies in the path of the eye. This is probably due to
pressure from the migrating eye.

4. Correlated with this is an increase in the distancb between the eyes
and the brain, caused by the growth of the facial cartilages.

5. The migrating eye moves through an arc of about 120 degrees.

6. The greater part of this rotation (three-fourths of it in P. ameri-
canus) is a rapid process, taking not more than three days.

7. The anterior ethmoidal region is not so strongly influenced by this
twisting as the ocular region.

8. The location of the olfactory nerves shows that the morphological
mid-line follows the inter-orbital septum.

9. The cartilage mass lying in the front part of the orbit of the adult
eye is a separate anterior structure in the larva.

10. With unimportant differences, the process of metamorphosis in
the sinistral fish is parallel to that in the dextral fish.

11. The original location of the eye is indicated in the adult by the
direction first taken, as they leave the brain, by those cranial nerves
having to do with the transposed eye. ""

12. The only well-marked asymmetry in the adult brain is due to
the much larger size of the olfactory nerve and lobe of the ocular side.

13. There is a perfect chiasma.

1 4. The optic nerve of the migrating eye is always anterior to that of
the other eye.

15. The optic tract is divided into dorsal and ventral portions.

16. There are fibres from the tract which enter the geniculnte body.
No other bundles of fibres leave the tract before it reaches the tectum.

17. The ganglia habenulae are symmetrical, at least in the larva
before metamorphosis.

VOL. XL. — NO. 1 4

50 bulletin: museum of compakative zoology.

18. There is a notable sulcus on the lateral side of the adult optic
lobe, which increases the surface area of the tectum.

19. The nidulus corticalis is the origin of the horizontal commissure
and of a large bundle of nerve fibres which pass into layers 3 and 4 of
the tectum.

20. The most important receiving cells for the fillet layer are proba-
bly the large spindle cells in layer 4.

WILLIAMS : MIGRATION OF EYE IN PSEUDOPLEURONECTES. 51

BIBLIOGRAPHY,

Agassiz, A.

'78. On the Young Stages of Bony Fishes. 11. Development of the Floun-
ders. Proc. Amer. Acad. Arts and Sci., Vol. 14, pp. 1-25, 8 pis. (3-10).

Auerbach, L.

'88. Die Lobi optici der Teleostier und die Vierhiigel der hoher organisierten
Gehirne. Morph. Jahrb., Bd. 14, pp. 373-393, Taf. 16.

Bellonci, J.

'81. Ueber den Ursprung des Nervus opticus und den feineren Ban des
Tectum opticum des Knocheniische. Zeit. f. wiss. Zool., Bd. 35, pp. 23-
29, Taf. 1, 2.

Bellonci, J.
'88. Ueber die centrale Endigung des Nervus opticus bei den Vertebraten.
Zeit. f. wiss. Zool., Bd. 67, pp. 1-46, Taf. 1-8, und 4 Holzsch.

Beneden, J. P. van. See Van Beneden.

Bumpus, H. C.

'98. On the Identification of Fish Artificially Hatched. Am. Nat., Vol. 32,
No. 388, pp. 407-412, 2 figs.

Cunningham, J. T.

'92. The Evolution of Flatfishes. Nat. Sci., Vol. 1, No. 3, pp. 191-199.

Cunningham, J. T.

'97. Recapitulation. Science Progress, Vol. 6, No. 5, pp. 483-510.

Edinger, L.

'96. Vorlesungen iiber den Bau der nervosen Centralorgane des Mcnschen
und der Thiere. 5te. Aufl. xii + 386 pp., 258 Abbild. Leipzig, F. C. W.
Vogel.

Ehrenbaum, E.

'96. Eier und Larven vou Fischeu der deutschen Bucht. Wiss. Meeres-
unters., N. F., Bd. 2, Heft 1, Abt. 1, pp. 253-324, Taf. 3-6.

Fritsch, G.
'78. Untersuchungen iiber den feineren Bau des Fischgehirns mit besonderer
Bcriicksichtigung der llomologien bei anderen Wirbclthierklassen. 94 +
XV pp., 13 Taf., 16 Textfig. Berlin.

52 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Fullarton, J. H.
'91. On the Devclopmeut of the Plaice (PIcuronectes platcssa). (Prelimi-
nary Report.) 9th Aim. Rep. Fish. Board for Scotland, Part III.,
pp. 311-316, pi. 7-9.

Fusari, R.

'87. Untcrsuchungcn iiber die feinere Anatomic des Gehirncs dcr Tele-
osticr. Internat. Monatsschr. fiir Anat. uud llistol. Bd. 4, pp. 275-300,
Taf. 9-11.

Fusari, R.

'96. La terminazione centrale del nervo ottico nei Teleostei. Riv. patol.
nervosa e mentale. Anno 1, Vol. 1, fasc 8, pp. 300-302.

Gehuchten, A. See Van Gehuchten. '

Goode, G.

'84. The Flatfishes and Flounders. The Fisheries and Fishery Industries
of the United States, Sect. 1, Natural History of Useful Aquatic Animals,
pp. 175-199, pis. 41-57.

Haller, B.

'98. Vom Bau des Wirbelthiergehirns. I. Theil. Salmo uud Scyllium.
Morph. Jahrb., Bd. 26, Hefte 3 u. 4, pp. 345-641, Taf. 12-22, 23 TextCg.

Herrick, C. J.

'91. Siluridffi. Jour. Comp. Neur., Vol. 1, pp. 211-228, pi. 17. In : Herrick,
C. L. and C. J., '91-92.

Herrick, C. L.

'91. The Commissures and Histology of the Teleost Brain. Anat. Anz.,
Bd. 6, pp. 676-681, 3 figs.

Herrick, C. L.

'91-92. Studies on the Brain of some American Fresh-water Fishes. Journ.
Comp. Neur., Vol. 1, pp. 228-215, pi. 19-21 ; pp. 333—358, pi. 24-25 ;
Vol. 2, pp. 21-72, pi. 4-12. In : Herrick, C. L. and C J., '91-92.

Herrick, C. L.

'92. Additional Notes on the Teleost Brain. Anat. Anz., Bd. 7, Nr. 13 u.
14, pp. 422-431, 10 figs.

Herrick, C. L. and C. J.

'91-92. Contributions to the Morphology of the Brain of Bony Fishes.
Journ. Comp. Neur., Vol. 1, pp. 2il-245, pi. 19-21 ; pp. 333-358,
pi. 24-25 ; Vol. 2, pp. 21-72, pi. 4-12.~

Holt, E. W. L.

'94. Studies in Teleost Morphology from the Marine Laboratory at Clce-
thorpes. Proc Zobl. Soc. Loud., pp. 413-446, pi. 28-30. [II. On the
Recessus Orbitalis, an Accessory Visual Organ in Pleuroncctid Fishes.]

WILLIAMS: MIGRATION OF EYE IN PSEUDOPLEURONECTES. 53

Jordan, D. S., and B. W. Evermann.

'96-00. The Fishes of North and Middle America, etc. Bull. United States
Nat. Mus., No. 47, lx + xxx + xxiv + ci + 3312 pp., 392 pi. [Floun-
ders, Part 3, pp. 2602-2692.]

Klein.

'68. Der Kopf der Pleuronecta;. Jahreshefte des Vereins f. vat. Natur-
kunde in Wiirttemberg, Jahrg. 24, pp. 271-308, Taf. 6. (Stuttgart.)

Krause, K.

'98. Experimentelle Untersuchungen iiber die Sehbahnen des Goldkarpfen
(Cyprinus auratus). Arch. f. mikr. Auat., Bd. 51, pp. 820-839,
Taf. 27.

Kyle, H. M.

'98. The Post-Larval Stages of the Plaice, Dab, Flounder, Long Rough
Dab, and Lemon Dab. 16th Ann. Rep. Fish. Board for Scotland (for
1897), Part 3, pp. 225-247, pis. 10, 11.

Looss, A.

'89. Ueber Degenerations-Erscheinungen im Thierreich, besonders iiber die
Reduction des Froschlarvenschwanzes und die im Verlaufe derselben auf-
tretenden histolytischen Processe. Preisschriften der Jablonowski'schen
Gesellschaft zu Leipzig, [No.] 27, 116 pp., 4 Taf. Leipzig.

Malm, A. W.

'54. De flundre-artade fiskarnas kroppsbyggnad ar raera skenbart an verk-
ligt osymmetrisk. Ofvers. Kgl. Sv. Vetcnsk. Acad. Fdrhandl., 1854,
pp. 173-183.

Malm, A. V/.

'68. Bidrag till kannedom af pleuronektoidernas utvekling och byggnad.
Kgl. Sv. Vetensk. Acad. Handl., N. F., Bd. 7, Nr. 4, pp. 28, Taf. 2.

Mayser, P.

'81. Vers-leichend anatomische Studien iiber das Gehirn der Knochenfische
mit besonderer Beriicksichtigung der Cyprinoideu. Zeit. f . wiss. Zool., Bd.
36, Heft 2, pp. 259-364, Taf. 14-23. .

Mcintosh, W. C, and A. T. Masterman.

'97. The Life Histories of the British Marine Food Fishes. 467 pp., 20 pis.
and Frontispiece. London, C J. Clay & Sons.

Mirto, D.

'95. Sulla fina anatomia del tetto ottico dei posci teleostei e sull' origine
reale del uervo ottico. Riv. sper. di frenatria e di med. legale. Vol.
21, fasc. 1, pp. 136-148, pi. 2, 3.

Mirto, D.

'96. La terminazioue centrale del nervo ottico nei Teleosti (In risponta
alia nota d. R. Fusari). Riv. patol. nervosa c mentale. Vol. 1, fasc. 10,
pp. 394-390. (= Arch. Ital. per le Malattie nervose e mentale, Anno 32.)

54 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Neumayer, L.

'95. llistologischc Uutersuclmngcn iibcr den feineren Ba\i des Central-
nervensystems von Esox lucius mit Beriicksichtigung verglcicbeud-analo-
mischer Vcrbiiltuisse. Arch. f. mikr. Anat., Bd. 44, Heft 3, pp. 345-3G5,
Tui". 23.

Nishikawa, T.

'97. On a Mode of the Passage of the Eye in a FlatGsh. Annot. Zool.
Japon., Vol. 1, Pars 3, pp. 7fi-96, 2 figs.

Parker, W. K.

'73. On the Structure and Development of the Skull in the Salmon (Salmo
salar, L.). Phil. Trans. Roy. Soc. London, Vol. 103, pp. 95-145, pi. 1-8.

Pfeffer, G.

'86. Ucbcr die Schiefheit der Plcuroncctidcn. Referat iibcr ein Vortrag,
u. s. w. Abhandl. Naturw. Vercin, Hamburg. Bd. 9, Heft 1, pp. 41—48.

Pfeffer, G.

'94. Ucber die Wauderung des Auges bei den Plattfischen. Verb. Deutsch.
Zool. Gesell. 3tcn Jahresversam. zu Gottingen, 1S93, Nr. 3, p. 83.

Petersen, C. G. J.

'94. Rei)ort of the Danish Biological Station to the Home Department,
IV. [For] 1893. v + 146 pp., 2 Tav., 18 Tabcl. Reprinted from Fiskeri-
Beretningen, 1893-94.

Raffaele, F.

'88. Lc uova galleggianti e Ic larve dci Teleosti nel golfo di Napoli. Mitth.
Zool. Stat. Neapcl, Bd. 8, Heft 1, pp. 1-84, Taf. 1-5.

Reichert, C. B.

'74. Ueber den asymmetrischen Ban des Kopfes der Pleuroncctiden. Arch,
f. Anat. Physiol, u. wiss. Med., Jahrg. 1874, pp. 19G-216, Taf. 5, 6.

Sanders, A.

'78. Contributions to the Anatomy of the Central Nervous System in Verte-
brate Animals. Iclithyopsida. Pisces. Tcleostei. Phil. Trans. Roy. Soc,
London, Vol. 169, pp. 735-776, pi. 58-65.

Schiodte, J. C.

'68. On the Development of the Position of the Eyes in Plcuronectidffi.
Ann. Mag. Nat. Hist., Ser. 4, Vol. 1, pp. 378-383.

Smith, H. M.

'98. The Fishes found in the Vicinity of Woods Hole. Bull. U. S. Fish.
Com., Vol. 17, for 1897, pp. 86-111, pi. 3.

Steenstrup, J. J. S.

'63. Bidrag til en rigtigere Opfattelse af Skjaevheden hos Flyndcrne (Plcu-
ronectides) og til Forklaring af bcgge Oines Fremkomst paa samme Side af
Kroppen. Oversigt Kgl. dauskc Vidcnskab. Selsk. Forh., 1803, pp. 145-
193, 16 wdcts. and 1 pi.

WILLIAMS: MIGKATION OF EYE IN PSEUDOPLEURONECTES. 55

Also separate: Om Skjaevheden hos Ilynderne og navnlig om Van-
dringen af det ovre 6ie fra Bliudsideu til Oiesideii tvers igjennem Hovedet,
m. m. Kjobenbavn, 1864. Saerskilt Aftryk af Oversigt over d. Kgl.
danske Videnskab. Selsk. Forbandl. i Nov. 1863. 52 pp.

Extract in: Arcb. sc. phys. et nat. Geneve. Nouv. periode, T. 25,
1866, p. 175-179.

See also Thomson, W., '65.

Steenstrup, Q. J. S.]

'64. Observations sur le developpement des Pleuronectes, extraites d'uno
lettre adressee a M. Milne-Edwards. Ann. Sci. Nat., Ser. 5, Tom. 2,
pp. 253-258, pi. 19, B.

Stieda, L.

'67. Studien iiber das centrale Nervensystem der Knocbenfische. Zeit. f.
wiss. Zool., Bd. 18, Heft 1, pp. 1-70, Taf. 1, 2.

Studnicka, F. K.

'97. Untersuchungen iiber den Ban des Sebnerven der Wirbeltieren. Jena
Zeitscbr. Bd. 31, Heft 1, pp. 1-28, Taf. 1, 2.

Thomson, W.

'65. Notes on Prof. Steenstrup's Views on the Obliquity of Flounders.
Ann. Mag. Nat. Hist., Ser. 3, Vol 15, pp. 361-371, pi. 18. (Contains
abstract of Steenstrup, '63.)

Traquair, R. H.

'65. On Asymmetry of the Pleuronectidse, as elucidated by an Examination
of the Skeleton of the Turbot, Halibut, and Plaice. Trans. Linn. Soc.
London, Vol. 25, Part 2, pp. 263-296, pi. 29-32.

Van Beneden, P. J.

'S3. Note sur la symmetrie des poissons Pleuronectes, dans leur jeune age
Bull. Acad. Roy. Belgique, Tom. 20, Part 3, pp. 205-210, 1 pi.

Van Gehuchten, A.

'95. Contribution a I'etude du systeme nerveux des teleosteens. La Cellule,
Tom. 10, pp. 255-295, 3 pi.

\A^inslow, G. M.

'98. The Chondrocranium in the Ichthyopsida. Bull. Essex Inst., Salem,

Mass., Vol. 28, pp. 87-141, 4 pi.

Also, '98«, Tufts Coll. Studies, No. 5, pp. 147-200, 4 pi.

Zittel, K. A. von.

'87-90. Handbuch der Palseontologie, Bd. 3,Vertebrata, xii + 900 pp. 719

Abbild.

56

bulletin: museum of comparative zoology.

EXPLANATION OF PLATES.

Figures 13, 14, and 16 are of Bothus maculatus. All others are of Pseudopleu
ronectes americanus. All except Figure 11 were outlined with the camera lucida.

ABBREVIATIONS.

a. ...

. Anterior.

gl. pin

Pineal gland.

an. . . .

. Anus.

gn. hub. . . .

Ganglion habenula.

arc. eth, m.

Mesial cartilage arch

ham. eth. . . .

Ethmoid hook in

of the ethmoid.

mid-line over me-

ba-hy. . .

. , Basi-hyal.

sial cartilage arch.

can, smi'crc.

. Semicircular canals.

hy-md

Ilyomandibular.

cb. (clx., s.)

. Cerebrum (right

i'cis. eth. (dr., s.)

Ethmoid notch

lobe, left lobe).

(right, left).

cbl. . . .

. Cerebellum.

Imn

Lemniscus (fillet).

chs. opt.

. Optic chiasma.

loh. olf ....

Olfactory lobe.

cl. crt. . .

. Degenerating carti-

lob. opt. (dx., s.) .

Optic lobe (right,

lage cells.

left).

corns hz.

. . Commissura hori-

mb. ach. . . .

Arachnoid mem-

zontalis.

brane.

cp. gnic.

. Geniculate body.

ms'eth

Mesethmoid.

crt. ink. (dx., s

) . Meckel's cartilage

nid. ctx. . . .

Nidulus corticalis

(right, left).

(Fritsch).

crt. orb. a. .

. Antorbital cartilage.

nid. rot. . . .

Nidulus rotundus.

crt. pin. d.

Rays of dorsal fin.

n. opt. (dx., s.) .

Optic nerve, right,

ec'eth. {dx., s.]

. Ect-ethmoid or pre-

left).

frontal (right.

oh. inf. ....

See obi. inf.

left).

oM. inf. (dx., s.) .

Inferior oblique

eth. . . .

. Ethmoid.

muscle (right.

eth-f. . . .

. Diagrammatic rep-

left).

resentation of the

obi. SU

Superior oblique

psGudomesial bar

muscle.

formed by the

ob. sv

See 06/ sii.

union of ect-eth-

oc. mig. . . .

Migrating eye.

moid and pre-

P

Posterior.

frontal.

pall. . \ . . .

Pallium.

fir. olf. {dx., s

.) . Foramen for olfac-

pa'sph

Parasphcnoid.

tory nerve (right,

pia

Pia mater.

left).

pin. an. . .

Anal or ventral fin.

fv. olf. {dx., s.

Olf.actory pit (right,

pin. d

Dorsal fin.

left).

pin. plv. . . .

Pelvic fin.

WILLIAMS: MIGKATION OF EYE IN PSEUDOPLEUEONECTES. 57

pt-pal. {dx., s.)

rec. orb.

rt. a.

rt. d

rt. p. .

rt. V. . .

sul. tct. opt.
tct. opt. (tij:

.,s.) .

tct. opt. 1
2
3
4
6

6

7

trb. . .

trb. su'orb.

(dx.,s.}

trb. su'orb. s. a.

Pterygopalatine

cartilage (right,

left).
Eecessus orbitalis.
Anterior rectus

muscle.
Dorsal rectus.
Posterior rectus.
Ventral rectus.
Sulcus of tct. opt.
Optic tectum (right,

left).
Outer layer.
Optic fibre layer.
Optic fibrillar layer.
Granular layer.
Fillet, longitudinal

and cross layers.
Gray layer.
Reticulate andepen-

dymal layer.
Trabecule cranii.
Supraorbital bar

(right, left).
Anterior part of left

supraorbital bar.

trb. su'orb. s. p. . Posteriorpart of left
supraorbital bar.

trt. opt. (d., V.) . Optic tract (dorsal,
ventral part).

tu. co'nt. tis. . . Connective tissue
sheath.

ur'stl Urostyle.

vnt. opt. . . . Optic ventricle.

I, ... X . . . First, . . . tenth cra-

nial nerves.
I.(dx.,s.) . . Olfactory nerve
(right, left).

II. {dx., s.) . . Optic nerve (right,

left).

//. d Dorsal portion of

optic tract.

II. V Ventral portion of

optic tract.
V. opt. su. . . . Superior ophthal-
mic branch of
nerve V.

V. opt. pfnd. . . Deep ophthalmic

branch of nerve V.

For explanation of Greek letters,

see text.

Williams. — Eye of Flounder.

PLATE 1.

{Pseudopleuronectes americanus.)

Fig. 1. Recently hatched fish (12 days old) from right side. X 30.

Note. — The line indicating the length of this specimen is J millimetre
too long. The length of the fish was 3.5 miUimetres.
Fish of Stage III. X 10.
Fish of Stage II. X 10.
Fish of Stage II, face view. X 35.
Fish of Stage IV, face view. X 8.
Fish of Stage IV, from right side. X 8.

Facial portion of the cartilaginous cranium of a recently hatched fish.
Stage I, projected on the frontal plane. X 200.

Fig.

2.

Fig.

3.

Fig.

4.

Fig.

5.

Fig.

6.

Fig.

7.

u.r's

II.

a re.

^*^

'^^.

\

I

pin.d.

\^

fy.oif.

crt. TnJc.s.—;^. f

/t H

iS

l-

^

Jirh.

pin.d.

pLn.a/i.

pin..plv.

Williams. — Eye of Flounder.

PLATE 2. I

{Pseudopleuronectes americanus.)

Fig. 8. Brain offish 75 millimetres long, dorsal view. X 8.

NoTK. — ob. inf. should have been oil. inf.
Fig. 9. Same brain viewed from right side. X 8.

Note. — ob. sv. should have been obi. xu. •

Fig. 10. Facial cartilages of fish of Stage II. as seen from above. X 100.

Note — Meckel's cartilage does not extend as far caudad as the letter-
ing, crt. mk. dx., which is placed opposite the quadrate-hyomandibular
mass.
Fig. 11. Dorsal view of brain, transposed eye and cranial nerves of adult. From

a dissection. X 2.
Fig. 12. Chiasma of a fish at Stage I. seen from in front. X 760.

.lAIvISr

ob. inf. ..

/V

X. viir.

1

if<

lob. oir.

Lob. opt. dx.

cbl.
IX.

vn.

|. ob.su.

\/

Ls 5k

r opt. SIC.

i.dx.

a

10

^^ ec'eih.

W

crt.mk.dx.

V- opt.p' fnd.
i Irb.sft^orb.dx. ^'

///

trb.

0

trb.sii' orb.s.

etk-f.

rn. '

vui.
IX.

J'

IV.

J

VI.

U.S.

h

siil.tct.opt.

n.s.

chs.opt

WiLLiAUs. — Eye of Flounder.

PLATE 3.

Fig. 13. Botkus. Anterior face of a cross-section through the nasal pits of a fish

in Stage IV. X 40.
Fig. 14. Bothus. Dorsal aspect of a frontal section through a fish of Stage II.

X 100.
Fig. 15. Pseudoplcuronectcs. Supraorbital bar cut in frontal section showing signs

of resorption. X 760.
Fig. 16. Bothus. Anterior face of a cross-section from the same individual as in

Fig. 13. X 40.
Fig. 17. Pseudopleuronectes. Anterior face of a cross-section of the head of a fish

in Stage I. X 200.

3-EXE

TE 3.

14

cri.pin.

pa sp

crt. mk.
ean.smi'erc.'

\ irb.su'orb.dx-

:y.

W

cl. crt.

crt. pin.. d.

inxo'nt.tia.

t'c'etliJS.

17

i.dx.

oU.su.

■tr6.su.*ori.dx.
ir6.su.'orbs.

lct.opt.s

c6.

/

.r

%

ba-hy.

/r.

crt.mk.

fr6.

Williams. — Eye of Flounder.

PLATE 4.

( Pseudopleuronectes americanus.)

Fig. 18. Anterior face of a cross section through the head of a fish of Stage III.
XIOO.
Note. — 06. inf. s. should have been obi. inf. s.
Fig. 19. Portion of a slanting cross section through cerebral lobes and dienceph-
alon. X 100.
Note. — The letters n. opt. s. in this figure should be changed to tr. opt. clx.
Fig. 20. Frontal section through habenulas and geniculate bodies. X 100.
Fig. 21. Parasagittal section through geniculate body and optic tract. X 100.

18
trb.siC orh.cLx.

ec'elhs.

19

/

(b-

- I 9 l%g

v*»

rec.OrS.

'^■:i^

pt-paL.

crt. mlc.

20
gl. pin. ,

rS.su'ori.s..

trt. opt. V

V

t. opt d.

"%.' oS.inf.s,

I

n.opt.s.
ifrt.hab.

tHMptd.
\

wd-ctx.

\

trtoptv

21

tct.opi. J. 2. 3. ^t. S. 6.
cp.gnic.

dd.ctx.

Imn.

I'.

H

/

/

trt. opt.

Williams. — Ej-e of Floimder.

TLATE 5.
( Pseudoplcuronectes americanns.)

Fig. 22. Diagram of parasagittal section of tectum. X 67.

Fig. 23. Portion of a parasagittal section of tectum from the anterior p;irt of tlic

optic ventricle to tlie surface. X 223.
Fig, 24. Cross section of optic nerve. X 50.
Fig. 25. Parasagittal section of diencephalon and part of metcnceplialon. X 18.

liutetx.

cpgnic.

22

\

Imn.
ta.co'nttis *

coJns.Az.

25

Ict.opl. J.
2.

Imn.

n^ nfii-

^•«^

vnt.opt.

^

nid.cljr.

UcLrot.

Bulletin of the Museum of Comparative Zoology
AT HARVARD COLLEGIA.

Vol. XL. No. 2.

THE EARLY DEVELOPMENT OF LEPAS. A STUDY OF
CELL-LINEAGE AND GERM-LAYERS.

By Maurice A. Bigelow.

With Twelve Plates.

CAMBRIDGE, MASS., U. S. A. :

PRINTED FOR THE MUSEUM.

July, 1902.

i

No. '2. — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY

OF THE MUSEUM OF COMPARATIVE ZOOLOGV AT HARVARD

COLLEGE. E. L. MARK, Direltor. No. 132.

JUl

Oil'

The Early Development of Lepas. A Study of Cell-Lineage

and Germ-Layers.

By Maurice A. Bigelow.

TABLE OF CONTENTS.

I. Introduction

II. Historical

III. Materials and methods . .

I V. Maturation and fertilization .

The unsegmented ovum.
Review of literature on
maturation and fertil-
ization

V. General sketch of cleavage

and germ-layers . . .

VI. Nomenclature of cleavage .

VII. Cleavage

1. Introductory ....

2. First cleavage. Two

cells

3. Review of the litera-

ture on first cleavage

4. Second cleavage. Four

cells

5. Review of the litera-

ture on the second
and succeeding cleav-
ages .......

3. Third cleavage. Eight
cells

7. Fourth cleavage. Six-

teen cells . . . .

8. Fifth cleavage. Thirty-

two cells . . . .

9. Sixth cleavage. Sixty-

two cells. Closing
of the blastopore.
The germ-layers
10. Seventh cleavage.
The mesoblast . .

VOL. XL. — NO. 2

PAGE

62
63
64

68

71

73
74

77
77

77

85

89

91

98

100

102

104

111

PAQE

11. Review of literature

on late stagesof cleav-
age, on closing of
blastopore and on
differentiation of the
germ-layers . . . 113

a. Late stages of

cleavage . . . 113

b. Closing of blasto-

pore .... 114

c. Differentiation of

the germ-layers 114

12. Determinate cleavage 116

13. Notes on cleavage and

germ-layers in L.
fascicularis . . . 117
VIII. Extension of the mesoblast
and entoblast. Later de-
velopment of the germ-
layers 119

IX. Formation of the append-
ages of the Nauplius, and
development of the or-
gans 121

X. General considerations on

cleavage and cell-lineage. 122
XI. Comparison of the germ-
layers of Lepas with those
of other Crustacea . . 127
XII. General summary, with
table of cell-lineage of

Lepas 133

Addendum 136

Bibliography 138

Explanation of Plates 143

62 bulletin: museum of comparative zoology.

I. Introduction.

In the inception of this work on the barnacles of the genus Lepas it
was planned to make a careful investigation of the early development
with reference to the origin and fate of the germ-layers. With this
object in view the methods of workers on cell-lineage were adopted,
because detailed studies seemed necessary in order to determine accu-
rately the origin of the germ-layers. These studies were not undertaken
with any expectation of extending or testing the accuracy of the generali-
zations which have come from the epoch-making investigations on cell-
lineage in the eggs of annelids, mollusks, and other animals. Whatever
opinion may be held regarding the fundamental importance of the gen-
eralizations growing out of such studies, it is usually conceded that the
tracing of cell-lineage gives a basis for accurate description of the details
of embryological development. Such accuracy in itself seems to furnish
sufficient present justification for studies in cell-lineage, for no one can
predict what interpretations may in the future grow out of any recorded
facts of to-day.

A study of Lepas fascicularis was begun by me in June 1894. Late
in that year there appeared an elaborate and important paper by T. T.
Groom on the development of several Cirripedia. As stated in a prelim-
inary note (Bigelow, '96), my independent studies of Lepas fascicularis
partly confirmed Groom's results in the case of other species of this
genus, but evidence in hand at the time of the publication of Groom's
paper indicated that, so far as accurate description of cleavage and the
formation of germ-layers is concerned, his account did not agree with
the development as observed in L. fascicularis. The studies already
begun by me were, therefore, continued and extended to Lepas anatifera
and other species which Groom had described. The account given in
this paper is based primarily upon studies of L. anatifera, and L. fasci-
cularis.

I take this opportunity to express my great indebtedness to my former
teacher, Prof. E. G. Conklin of the University of Pennsylvania, under
whose guidance the general outlines of the work were developed.

The completion of the observational work was carried out during the
year 1898-99 in the Zoological Laboratory of the Museum of Compara-
tive Zoology at Harvard College. To all the instructors of the depart-
ment I am greatly indebted for stimulating interest, but especially do
I owe acknowledgment to Dr. W. E. Castle, who continuously followed
my work and gave me the benefit of his advice and criticism, and to

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 63

Prof. E. L. Mark, who has carefully examined and criticised all my re-
sults and given me many helpful suggestions during the arrangement of
the results for publication.

Durin'Tf several summers the work has been carried on in the Marine
Biological Laboratory and in the United States Fish Commission Station
at Wood's Hole, jNIass. I wish to express my appreciation of the assist-
ance, in the line of facilities for work, which was extended to me by the
officials of these two laboratories, particularly by their respective direc-
tors, Prof. C. 0. Whitman and Prof. H. C. Bumpus.

II. Historical.

The history of the development of our knowledge of the Cirripedia
has been so often written that for the purpose of this paper it is suili-
cient to give a mere outline. The now classical monograph of Darwin
('51, '54) reviewed so exhaustively the knowledge obtained by earlier
observers, and added such a mass of original information on structure,
metamorphosis, relationships, and natural history, that in these respects
the Cirripedia have since ranked among well known groups of inverte-
brate animals. Since Darwin's time much of the investigation on the
animals of the group has been concerned with embryological develop-
ment, to which very little of Darwin's work was devoted. In the
" Challenger " Reports Hoek ('83, '84) made important additions to our
knowledge of the anatomy and relationships of many cirripedes, and gave
a good historical sketch of the group. Gerstacker's historical review in
Eronn's Klassen u. Ordnungen is exhaustive.

The papers of Van Beneden ('70), Willemoes-Suhm ('76), Hoek ('76),
Lang ('78), Nassonow ('85, '87), Nussbaum ('90), and Groom ('94) deal
in more or less detail with embryonic development, and these papers
include the most important existing conti'ibutions to our knowledge of
cirripede embx-yology. Muller ('64), Filippi ('65), Mllnter und Buchholz
('69) and Bovallius ('75) have made contributions regarding certain
points in the early development.

Our knowledge of the early development of species of Balanus is due
principally to the studies of Miinter und Buchholz ('69), Hoek ('76),
Lang ('78), Nassonow ('85, '87), and Groom ('94).

The early development of species of Lepas is known through the in-
vestigations of Willemoes-Suhm ('76), Groom ('94), and Bigelow ('96).

The only recorded observations on the early development of Lepas
fascicularis earlier than those of the present writer are the published

64 bulletin: museum of comparative zoology.

notes of Willemoes-Siihtn ('76), who died during the voyage of the
" Challenger " before his studies were completed. His paper gives a
very complete account of the history of the above mentioned cirripcde
from the Nauplius to the sessile adult, but only a short Jind fragmentary
description of embryonic development. In some of the later embryonic
stages the observations are quite correct, but the few descriptions and
figures of cleavage stages are very inaccurate.

The embryology of Pollicipes has been studied by Nussbaum ('90),
but his account is somewhat fragmentary.

Among the Rhizocephalan Cirripedia the only description of a complete
scries of embryonic stages is Van Benedeu's ('70) account of Sacculina.

Further, one or more of the investigators already mentioned has
studied the early development of species of the following genera of Cirri-
pedia : — Conchoderma, Scalpellum, Tetraclita, Dichelaspis, Chthamalus.
However, much of this embryological work has been fragmentary, and
often superficial.

The last, and by far the most important, paper on the early embry-
ology of the Cirripedia was published by Groom in 1894. This contains
a good resumd of the previous work on the subject, reviewing the con-
tributions of the various investigators mentioned in the preceding para-
graphs. Groom studied the embryology of five species, namely, Balanus
perforatus, Lepas anatifera, L. pectinata, Chthamalus stellatus, and Con-
choderma virgata. His observations on the later stages of embryonic
development and on the larval stages were exhaustive. The study of the
cleavage was undertaken secondarily, and was not investigated as accu-
rately as were the later stages.

The accounts of the early embryology of cirripides which were given
by observers before Groom do not as a ride contain records of detailed
observation, which alone could be used comparatively in a paper from
the standpoint of cell-lineage. Groom reviewed well the general accounts
of previous investigators, and brought their results into line with his own
observations. In reviewing the literature I must necessarily deal pri-
marily with Groom's account, because he is the only investigator who
lias attempted detailed description of the early stages of cirripede
development.

III. Materials and Methods.

The material upon which this paper is based was collected at Wood's
Hole, Mass., in the summers of 1894, 1895, 1898, and 1899. Prof.
Harold Heath of Stanford University, Cal., has collected and preserved

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 65

for me the eggs of Lepas hillii, Pollicipes polyraerus and Sacculina, which
have been used for comparative study.

In Vineyard Sound and Buzzard's Bay, groups of Lepas fascicularis,
L. anatifera and L. pectinata have been found at various times between
June and September. Any of these forms may appear at times when
the prolonged south-east winds have carried the drifting material of the
Gulf Stream in the direction of the Elizabeth Islands. So many elements
of chance are involved in getting the animals that it has been found
difficult to collect complete developmental series, and the work has been
often delayed.

A very large majority of the animals of all species cany eggs in ad-
vanced stages of development when they arrive in the waters near Wood's
Hole. This has been found especially true of the numerous specimens
of L. fascicularis, hundreds of which have been found carrying eggs ready
to hatch, but only a few dozen with eggs in early cleavage stages. In
two different summers a few animals of this species have been found
early in June with eggs in stages of maturation, but when large numbers
of animals arrived in July, few cleavage stages could be found and in
many cases Nauplii were escaping from the brood-lamellae.

Much drifting timber carrying L. anatifera was obtained about the
middle of August, 1898. The adult animals all carried eggs which were
in advanced stages of development and were hatching rapidly. Many
animals which were about half the adult size were laying eggs. The
timbers were anchored in the harbor, and for several weeks it was possible
to obtain an abundance of material in maturation and cleavage stages.
The stages of living and preserved material thus secured for study rep-
resented the important phases of every mitotic division in the early
development.

As is well known, the development from egg to Nauplius takes place
in the mantle chamber. The eggs, each enclosed in a vitelline membrane,
lie in the cavities of the egg-plates, or ovigerous lamellae, which lie be-
tween the body and the mantle. In studying living ova it is easy to tear
the lamellae and thus free large numbers of eggs, but in preserving mate-
rial it is more convenient to fix the lamellae in large pieces.

Maturation and cleavage were studied first in the living eggs. It was
found impossible to keep eggs developing normally under artificial con-
ditions outside the mantle cavity longer than from five to ten hours.
Other workers on Cirripedia have had the same experience. It was
rarely possible to follow a single egg through the maturation phases to
the close of the second cleavage, and fresh material, which had under-

66 bulletin: museum of comparative zoology.

gone the early cleavage while in the brood-lamelke, was necessarily used
for the study of later cleavages.

Many of the fixing reagents ordinarily employed in embryological work
have been tried, but only solutions contaiuiug picric acid have proven
entirely satisfactory. Kleinenberg's stronger fluid and a saturated solu-
tion of picric acid in 35% alcohol both gave excellent fixation, but a
saturated solution of picric acid in 5% acetic acid gave results which were
far superior to those obtained by any other fixing solution. This fluid
penetrated rapidly, and eggs thus prepared were very transparent wlicn
stained and mounted entire. This transparency was a very important
feature in the study of all cleavage stages. The picro-acetic mixture also
gave the best results for material which was to be sectioned. It should
be remarked that solutions with less acetic acid lack penetrating power.

Strong solutions of mercuric chloride in distilled water, in sea water,
in alcohol, or combined with picric acid, gave some good results in the
study of maturation and early cleavage stages by means of sections, but
material thus fixed proved too opaque for preparations of entire eggs.
Material fixed in the mercuric chloride solutions was especially valuable
in determining the distribution of the yolk, which readily stained differ-
entially after such fixation. In the study of all stages of development use
was made both of sections and of entire eggs viewed as transparent ob-
jects. The method of preparing the latter will be described first. Small
pieces of egg-lamellae which had been fixed in the picro-acetic mixture
were stained from one to three hours in a concentrated solution of borax-
carmine in 35% alcohol (Grenacher's formula). They were then wasiied
in alcohol and rapidly decolorized in 70% alcohol containing 0.3% hydro-
chloric acid. The decolorizing was watched with a compound microscope,
and quickly checked when nuclei and cell-boundaries began to appear.
The piece of egg-lamella was then dehydrated and, within two or three
hours after staining, cleared.

All the ordinary clearing oils were tried, but no other one gave results
comparable in excellence with those obtained by the use of clove oil.
This oil renders the egg-lamellaj brittle, so that the eggs can easily be
isolated by the use of needles. In practiqe the stained pieces of egg-
lamellse were placed in a drop of clove oil on a glass slide. Then, using
a dissecting microscope, the lameUte were cut with fine needles and the
eggs set free, but they were still surrounded by the vitelline membrane.
All attempts at removing this membrane proved unsuccessful. After
the greater part of the clove oil had been drained away, the eggs were
mounted in xylol-balsam.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 67

Elmo's prepared by the above method were so transparent that even in
later stages the outlines of cells on either side of the embryo could be
clearly seen by appropriate focussing. It was, therefore, easy to study
and draw optical sections in any plane. The refractive index of clove
oil ^ is such that the vitelline membrane becomes almost invisible.

By carefully moving the cover glass it is possible to roll eggs into any
desired position, and for this purpose the balsam was for months kept
semi-fluid by occasionally applying a drop of xylol to the edge of the
cover glass.

It was found practicable, and in some cases profitable, after studying
an egg in balsam, to remove the cover glass, dissolve the surrounding
balsam with xylol, lift the egg by means of a capillary tube, transfer it
to paraffine, imbed by the watch-glass method and section it. When
imbedded near the surface of the block of paraffine, the long axis of the
egg can be distinguished by the use of a lens, and hence sections can be
cut longitudinally or transversely as desired. This method of sectioning
single eggs was employed only for the purpose of gaining an idea of the
appearance of sections of particular stages in known planes. As a rule,
pieces of the egg-lamellte rather than single eggs were imbedded and
sectioned, the sections being stained on the slide. Since the eggs have
no definite arrangement in the lamellae, sections in all planes were thus
obtained. By comparison with sections of single ova in which the orien-
tation had been definitely established, it was possible to choose with
certainty the sections representing any desired plane in any stage of
development.

For staining sections on the slide Delafield's haematoxylin diluted
with four or five times its volume of distilled water gave the best results.
In the later cleavage stages and in embryonic stages orange G or eosin
were used after the hsematoxylin. By this means the entoblastic yolk-
cells were sharply differentiated.

In the study of preparations of the entire eggs a sub-stage condenser
with iris diaphragm was absolutely necessary. A ^ inch homogeneous
immersion objective with long working distance was of great service.

Most of the preparations upon which this paper is based are yet in
good condition, and are therefore available as evidence in support of the
following account of the development of Lepas.

1 Since this paper was written I have found that oil of cassia for clearing gives
results even superior to those obtained by the use of clove oil. It has also proved
to be an excellent mounting medium, but probably the preparations will not retain
stains permanently.

68 bulletin: museum of comparative zoology.

The methods employed have been given at length, because it is be-
lieved that the results obtained, which difter widely from those of earlier
workers, are due largely to the successful making of transparent prepa-
rations of entire eggs. In examining the figures given by previous
workers it is evident that none of them had the advantage of such
preparations, and consequently none of them were able to follow accu-
rately the history of the nuclei, which is very important for the determi-
nation of cell-lineage.

IV. Maturation and Fertilization. The Unsegmented Ovum.

In agreement with the observations of Weismann und Ischikawa ('88),
ef^'^s taken from the oviducts were found to contain the first maturation
spindle. Owing to mutual pressure, there is great distortion of the eggs
in the oviducts, but when artificially liberated into sea water they quickly
assume a spherical form. The separation of the first polar cell takes
place at about the time when the eggs leave the oviducts. Soon after
this the formation of the vitelline membrane begins, so that it occupies
a position between the first polar cell and the egg (Plate 11, Fig. 95,
mh.vL). This is followed by the development of a second polar cell
(Plate 2, Fig. 17), which lies within the vitelline membrane (Plate 11,
Fig. 95, cl.pol}). From the time of assuming the spherical shape, soon
after leaving the oviduct, the eggs retain this form, except when pressure
of surrounding eggs in the egg-lamellte distorts them. The egg repre-
sented in Figure 17 is an example of the influence of pressure in the egg-
lamcllse ; such a form at this stage has not been seen among eggs kept
isolated in watch glasses. It should be noted here that the uniform
distribution of yolk serves to distinguish such eggs, which are pressed
into an elongated shape, from later stages in which the eggs are normally
ellipsoidal even when isolated, but in which the yolk is collected at the
vegetative pole.

Eggs which are isolated soon after oviposition retain the spherical con-
dition and the uniform distribution of the yolk until about the time when
the second polar cell is formed. Then the egg begins to elongate in the
direction of the chief axis, and the protoplasmic materials begin to con-
centrate at the animal pole, where the polar cells are located ; at the
same time the yolk is removed to the lower half of the egg, being con-
centrated around the vegetative pole. This movement of protoplasm
and yolk, towards animal and vegetative poles respectively, continues

BIGELOW: EARLY DEVELOPMENT OF LEPaS. 69

and finally results in a telolecithal arrangement of the materials of the

egg-
Eggs taken from the egg-lamellse at all phases of the maturation have
been carefully compared with the corresponding stages of isolated eggs
which were kept in watch glasses. The distortions in form produced by
pressure apparently do not disturb the normal course of cytological
changes in the egg.

Figures 1-6 represent a series of camera sketches made from a living
egg at intervals within a period of three hours. In Figure 1 the egg is
represented just at the completion of the separation of the second polar
cell. The egg is approximately spherical and closely surrounded by the
vitelline membrane (mb.vt.). The yolk with its oil globules is in general
uniformly distributed, but already some of the globules have been seen
to move towards the vegetative pole. Figure 2 shows the well-marked
beginning of elongation ; the yolk is collecting at the vegetative pole and
a mass of protoplasm, concentrating into the animal half of the egg, is
dark and granular. Figure 3 represents a stage some minutes later. A
circular depression has appeared around the egg at the equator constrict-
ing the egg into nearly equal lobes. The upper, protoplasmic lobe is dark
and granular, especially near its centre, whereas the lower or yolk-lobe is
relatively clear and transparent, as represented in Figure 18 (Plate 2),
The constriction now moves toward the vegetative pole of the egg, where
the yolk is collecting (Fig. 4). Gradually the constricting furrow dis-
appears (Fig. 5), and the egg becomes ellipsoidal, as shown in Figure 6.
At the animal pole the egg continues to be bluntly rounded, while at the
vegetative pole it becomes more pointed. The vitelline membrane, hav-
ing taken on this shape, retains it throughout the development, and
appears to be quite rigid from this stage onward. At the close of the
elongation the upper, animal portion x>f the egg is largely composed
of dark granular protoplasm containing some small granules of yolk,
but no oil globules (Plate 2, Figs. 19, 20). The lower vegetative
part of the egg is more transparent and contains the mass of yolk gran-
ules. The oil globules are concentrated at the pointed end of the egg
and for a time are arranged in strict radial symmetry with respect to the
long (chief) axis of the egg. Protoplasmic strands extend throughout
the vegetative half of the egg.

The elongation of the egg and the separation of yolk and protoplasm,
which result in the telolecithal condition and the establishment of visible
polarity, are entirely distinct from the first cleavage processes, with which
Groom ('94) has confused them (see review of the literature on first

70 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

cleavage). They beloug more properly to the maturation j)hases, and
have many characteristics known for ova of other groups of animals.
The polar axis thus established in the cirripede ovum has the same rela-
tion to polar cells, maturation spindles, and first segmentation spindle,
as is found ordinarily in telolecithal ova.

The phenomena occurring during the elongation and distribution of
the materials of the cirripede egg, especially the formation of a constric-
tion which marks off a yolk-lobe at the vegetative pole, are apparently
similar to conditions which obtain in some molluscan eggs ; for example,
in the gasteropods Nassa (Bobrctzky, '76) and Ilyanassa (Crampton,
'96). In these cases the formation of the yolk-lobe closely resembles
that process in Lepas, but its later history is widely different. At one
stage of the maturation, the eggs of Nassa and Ilyanassa have a form
similar to that of the egg of Lepas as represented in Figure 3, a
constriction marking off a yolk-lobe. Whereas in the cirripede the con-
striction disappears before the first cleavage, in the gasteropods the first
cleavage plane forms so that in the unequal division a smaller cell (ah)
is separated from a larger one (cc?), which still retains the yolk-lobe.
After cleavage the yolk-lobe gradually disappears and the cell cd becomes
spheroidal in form. In Lepas, as in Nassa and Ilyanassa, the materials
composing the yolk-lobe are after the first cleavage contained in the cell
cd.

In my attempts to determine the precise time of penetration of the
spermatozoon I have failed, as have all earlier investigators ; but we may
infer that it enters before the formation of the vitelline membrane,
probably about the time when the first polar cell is separated. In sec-
tions similar to that represented in Plate 2, Figure 17 (formation of
second polar coll) I have noted a darkly staining body near the vegeta-
tive pole of the egg. I am not certain of having identified the male
pronucleus in a stage earlier than one corresponding in external form to
Figures 3 and 18, in which, however, the pronuclei were widely separated,
as shown in Figure 19. A further comparison of Figures 18 and 19
shows that there is not a constant relation between the relative posi-
tions of the pronuclei and the telolecithal .distribution of the yolk and
protoplasm. In external outline and in the, presence of tlie constriction
marking off the yolk-lobe, the egg represented in Figure 18, correspond-
ing to Figure 3, is earlier than that shown in Figure 19, which cor-
responds to Figure G. But in Figure 18 the size and contact of the
pronuclei indicate an older stage than that of Figure 19.

After the disappearance of the yolk-lobe the pronuclei are usually

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 71

found in contact, as shown in Plate 2, Figure 20, which suggests that
there is retardation in the approach of the pronuclei in cases similar to
Figure 19. All ray observations point to the conclusion that the pro-
nuclei usually come into contact during the time when the yolk-lobe is
disappearing, and the egg is assuming the ellipsoidal form, that is, in
stages corresponding to Figures 4-6.

Review of Literature on Maturation and Fertilization.

A general review of the literature on these phases of cirripede devel-
opment is given by Groom ('94), consequently reference will not be made
in this connection to writings unless they have direct bearing upon
observations recorded in this paper.

.The formation of polar bodies and vitelline membrane have been ob-
served and described by Weismann und Ischikawa ('88), Xussbaum ('89),
Solger ('90), Groom ('94), and others. My observations on the forma-
tion of these structures are merely confirmatory of these earlier writers,
and have been recorded simply to complete my account of associated
phenomena.

The contractions of the egg during elongation and the segregation of
protoplasm and yolk have been observed by Groom and others ; but the
process has, apparently, not been followed continuously, and has been
confused with the first cleavage, as will be shown in the review of litera-
ture bearing on that stage.

Groom ('94, p. 133) states that in the unfertilized ovum of Lepas
anatifera no difference can be distinguished between the two poles, and
suggests that the ovum may become oriented only upon fertilization.
Opposed to such conclusion is the fact that in eggs taken from the ovi-
ducts the first maturation spindle marks the chief axis of the eg'g, which
thus seems to be determined long before fertilization. Nussbaum ('90)
correctly observed that the axes of the embryo are established with the
formation of the polar bodies.

Groom ('94, p. 136) states that "the axis of the spindle of the seg-
mentation-nucleus is not at right angles to that of the second directive
spindle." In the account of the first cleavage it will be shown that, in
opposition to this view, the first cleavage spindle is formed in a plane
perpendicular to the chief axis of the eg^^, with which the second matu-
ration spindle coincides at the moment when the polar cell is separated.
There is, therefore, in Lepas complete agreement with the usual condi-
tion in the eggs of other animals.

72 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

With regard to the male pronucleus Groom ('94, p. 134) states :
*' Sections made of ova of Lepas anatifei'a before or shortly after the
formation of the first polar body show the first directive spindle or a
small round nucleus with several chromatin elements." Having failed
to find the male pronucleus, he concluded that it " must be exceedingly
small and easily overlooked, otherwise it would be necessary to concludo
that the fusion of the two pronuclei takes place immediately after the
first polar body is formed (in which case it would bo very rarely detected
in ova which had given off the first polar body) ; but this seems improb-
able, though traces of a male pronucleus were never found in sections at
any later phase even in ova where the second polar body was being or
had just been given off."

Some of these observations by Groom are in accord with my statement
that the male pronucleus has not been certainly identified in sections
corresponding to a stage earlier than that represented in my Figure 3>
although the spermatozoon is probably present at a stage earlier than
that represented in Figure 1, in which the second polar cell has just
been separated. Groom's supposition that the pronuclei fuse soon after
the formation of the first polar cell is opposed by the evidence afforded
by my Figures 17-21. It will be shown later that Gi-oom probably saw
the male pronucleus in these later stages, but misinterpreted it as one
of the daughter nuclei resulting from the first division of the egg.

Groom says (p. 135), "The nucleus, which, during the period at which
the ovum was undergoing contraction [yolk-lobe stages], was small and
situated peripherally and anteriorly [at animal pole], and was invisible
without special preparation, now becomes larger, and appears as a defi-
nite clear spot." He further states (p. 137) that, "the clear spot
appearing with the separation of the protoplasm is almost certainly the
segmentation-nucleus." I have seen this " clear spot," and sections show
that it is the female pronucleus, or sometimes the two pronuclei so ap-
proximated that viewed through the opaque substance of the living egg
the appearance is that of one transparent area. Groom's statements
regarding these stages were apparently based upon studies of living eggs,
which are so opaque as to render observation difficult and uncertain.

In a stage which Groom interpreted as that of the first cleavage, he
found " two nuclei in the newly-formed [first] blastomere " ; these were
regarded as the daughter nuclei of the first segmentation nucleus
(pp. 137, 142, 145). In the review of literature on first cleavage it will
be pointed out that Groom apparently has mistaken for the first segmen-
tation of the ovum a maturation phase, such as that represented in my

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 73

Figures 3 and 18 ; the two nuclei which he describes being evidently the
pronuclei and not daughter nuclei sprung from the first segmentation
nucleus. The figures in the present paper show that a segmentation
nucleus does not exist during the separation of yolk and protoplasm.
Two pronuclei are in the egg, but they do not appear to fuse completely
until the nuclear membranes fade away at the beginning of division.
My figures of the first cleavage show, as opposed to Groom's description,
that the nuclei resulting from the first division are not at first both
located in the upper half of the egg, where the protoplasm is more
concentrated.

Nussbaum ('90) observed the two nuclei in Pollicipes as the waves of
constriction passed over the egg during the separation of yolk and proto-
plasm, and interpreted them as pronuclei. He figured and described
the pronuclei as approaching along a line nearly coinciding with the long
axis of the egg ; and he assumed that the plane of the first cleavage is
perpendicular to the contact surface of the pronuclei. My Figures 18-
20 confirm his observations on Pollicipes, for it is certain that there are
two pronuclei in the protoplasmic mass at the animal pole of the egg in
L. anatifera and L. fascicularis as the separation of yolk and protoplasm
progresses. I have studied sections of Pollicipes which show similar
conditions. Nussbaura's interpretation of these nuclei as pronuclei is
certainly correct, as is likewise his description of their approach and
contact.

V. General Sketch of Cleavage and Germ-Layers.

The cleavage of Lepas is total, unequal, and regular. Stages of 2, 4,
8, 16, 32 and 62 cells are normally formed. Cells of a given generation
may anticipate their sister cells in division, but no second division of
such cells takes place before all other cells have completed corresponding
cleavages and reached the same generation.

The first cleavage plane is nearly parallel to the long axis of the ellip-
soidal egg, which divides into a small anterior cell (micromere) and a
large posterior yolk-bearing cell (macromere). The plane of the second
cleavage is perpendicular to that of the first, a second micromere being
cut off" from the yolk-bearing macromere, while the first micromere divides
into two of equal size. The plane of the third cleavage is essentially
perpendicular to both the preceding ones. A third micromere is sepa-
rated at this cleavage from the yolk-macromere, which is now purely
mes-entoblastic. Thus by the first, second, and third cleavages three

74 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

micromeres are separated from the yolk-bearing macromere. These three
cells contain all the ectoblast, and by their repeated division form the
blastoderm. Certain cells of the blastoderm, which are derived from the
first two micromeres, give rise to a portion of the mesoblast, hence these
two micromeres are not purely ectoblastic. The third contains only
ectoblast. In the fourth cleavage a mesoblast cell is separated from the
yolk-macromere, which now represents entoblast alone.

The sixteen-cell stage, therefore, is composed of fourteen derivatives
of the three micromeres, one mesoblast cell, and one entoblast cell (yolk-
macromere). The entoblastic yolk-macromere is nearly enveloped by
the fourteen smaller cells composing the blastoderm, only a small part
of the entoblast cell being exposed at the blastopore. The single meso-
blast cell lies at the posterior edge of the blastopore, and were its history
not known would certainly be regarded as a cell of the blastoderm. At
the fifth cleavage each of the sixteen cells divides, the two resulting
mesoblastic cells still remaining at the surface. At the sixth cleavage
all the cells except the two entoblast cells divide, thus producing a sixty-
two-cell stage. Dui-ing the sixth cleavage the two mesoblastic cells,
before dividing, sink beneath the blastoderm, as this closes over the ento-
blast and obliterates the blastopore. At the same time four cells of the
blastoderm, lying at the anterior and lateral edges of the blastopore,
divide parallel to the surface. The four deep cells thus formed beneath
the blastoderm constitute a part of the mesoblast. The mesoblast, then,
is derived in part from one cell which is separated from the entoblast in
the fourth cleavage (sixteen-cell stage) and in part from four other cells
which are detached from the blastoderm during the sixth cleavage.

Gastrulation is of the cpibolic type, and is the result of the extension
of the blastoderm over the entoblastic yolk-macromere. During the
sixth cleavage, which leads to the formation of a sixty-two-cell stage, the
blastoderm usually closes over the blastopore, which marks the ventral
and posterior part of the future embryo.

In the general features of the late development of the embryo the
results of this investi<ration confirm those of some earlier workers.

'o^'

VI. Nomenclatiire of Cleavage.

For convenience in describing the cell-lineage of Lepas and in making
comparisons with the development of other forms, it is desirable that
some system of cell-nomenclature should be applied.

The common systems, which have been developed with special refer-

BIGELOW : EARLY DEVELOPMENT OF LEPAS. 75

ence to the conditions in the developing eggs of annelids and mollusks,
are dominated by the conception of cells cleaving in sets of fours or quar-
tets. The system of Blochmann ('81) and its successors have, with few
exceptions, been applied to eggs in which a quartet of macromeres (in a
morphological sense) is formed by the first two cleavages, and by later
cleavages these give rise to successive quartets of micromeres. In all
the annelids and mollusks in which the cell-lineage has been determined
with certainty, the cells of the four quadrants (a, b, c, d) formed by the
first two cleavages are equivalent, in that each cell contains a portion of
the two primary germ-layers, ectoblast and entoblast. The mesoblast
is not so distributed with reference to the quadrants. It will be shown
in this paper that the four-cell stage of Lepas is not a quartet of equiva-
lent cells so far as the two primary germ-layers are concerned. Whereas
in the annelidan and molluskan eggs each cell of the four-cell stage con-
tains both ectoblast and entoblast, in Lepas three of these cells (a, b, c)
contain ectoblast but no entoblast ; and the fourth cell {d) contains both
ectoblast and all the entoblast. In the annelids and mollusks the cells
of the first quartet of micromeres (eight-cell stage) contain the ectoblast
which is first separated from the entoblastic macromeres ; but in Lepas
one of the cells of the two-cell stage is the first ectoblast to be separated
from entoblast.

Enough has been said, in anticipation of the account of the cleavage,
to make it evident that the well-known quartet systems of nomenclature
would not have their usual significance as indexes of homologies, if
applied to the cleavage of Lepas, for the cells of the four-cell stage in
annelids and mollusks are apparently not comparable with the cells of
the same stage of Lepas, which would be given the same designations.
However, a quartet system has been employed for the purposes of this
paper, for the reason that it is convenient and familiar. The above
statements will show that the system has not been used here with a view
to indicating by it homologies with which it has become associated in its
application to the spiral cleavage of annelids and mollusks. As far as
regards the cirripede egg, the known facts do not seem to me to warrant
the interpretation that cleavage occurs in cells grouped as quartets in the
sense in which the term is applied to spiral cleavage ; and while the
notation of a quartet system has been adapted to the purposes of this
paper, the term " quartet " has not been applied in description as desig-
nating groups of cells in the cleaving egg of Lepas. ^

1 See Addendum by E. L. M. and W. E. C. (p. 136) following the General
Summary.

76 bulletin: museum of comparative zoology.

The systom devised by Kofoid ('94) — which Castle applied to the
bilateral cleavage ot tunicates, where the conditions of cleavage resemble
those of Lepas — has with some necessary modifications been followed.
The cells of the four-cell stage are designated a, b, c and d iu the usual
order, a being the left anterior cell. An exponent iudicatQS the number
of the generation, starting Avith the ovum as the first, e. g. a', 5^ etc.
A second exponent is used to distinguish a cell from other cells of the
same generation and derivation, e. g. a*'\ a*-', a*-^, etc. In assigning
the second exponent I have followed in part suggestions made by Kofoid
('94) and put into practice by Castle ('96). In cases of equatorial
division the odd numbers have been applied to the cells nearer the ve(/e-
tative pole, and the even to those nearer the animal pole. Thus of the
cells in the four-cell stage a^ divides, forming a*-' which is nearer the
vegetative, and a*-^ which is nearer the animal pole, while its sister cell,
b^, forms 5*-^ and i**^ (see Plate 4, Figs. 34-38). In later stages, where
cells do not divide equatorially, but parallel to the sagittal plane, the odd
exponent has been applied to the cell lying nearer that plane. In cases
where a cell lies in the sagittal plane and undergoes division in the same
plane, the daughter cell on the i-ight side of that plane is designated by
the odd exponent. Whenever cells divide transversely to the chief axis
of the embryo, the anterior cell is designated by the odd exponent.

In determining the designation of cells, the rules given by Kofoid are
here applied to Lepas. The designation of any derivative of cells a, b,
c, d being given, the designation of mother cell or daughter cells can be
quickly determined. The first exponent indicating the generation of the
mother cell will, of course, be one less than that of the daughter cell.
The second exponent of the mother cell will be one-half of that of the
daughter cell, if that be an even number, and one-half the sum of the
second exponent plus one, if that be an odd number. Thus a*'^ and a*-'
are daughter cells of a'-\ Likewise, to determine the first exponent of
the daughter cells, add one to the first exponent of the mother cell ; to
determine the second exponent, multiply the second exponent of the
mother cell by two and the product is the designation to be applied to
the cell bearing the even number as exponent, while that product less one
designates the sister cell. Thus a*-*^ dividing forms a*-^^ and a^-".

A summary of the important points in the cell-lineage of Lepas is given
in a table in connection with the general summary.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 77

VII. Cleavage.

1. Introductory.

The following description of the cleavage of the egg of Lepas applies
particularly to L. anatifera, of which I obtained abundant material of all
stages in 1898, being thus able to study the early development in con-
siderable detail. An extensive series of the eggs of L. fascicularis was
later obtained and its development has been carefully compared with
that of L. anatifera. There is such close parallelism in the development
of the two species that the following account will apply in all important
respects to L. fascicularis as well as to L. anatifera. Figures 95—126
(Plates 11, 12) of L. fascicularis when compared with those of L. anati-
fera show how close is the similarity between the two species. At the
close of this chapter (p. 117) there are some notes on the early develop-
ment of L. fascicularis which supplement and correct a preliminary
account of this species published by me in 1896.

The principal stages in the development of L. pectinata and L. hillii
have also been examined, but their development does not appear to differ
in any important respects from that of L. anatifera and L. fascicularis.

2. First Cleavage. Two Cells.

The first cleavage of the egg of all Lepadidse and Balanidse whose
development has been heretofore described results in the formation of
two unlike cells. The smaller cell, rich in protoplasm, is situated at the
rounded end of the vitelline membrane ; the other, laden with yolk, at
its pointed end (Plate 1, Fig. 16). In previous accounts the first cleav-
age plane has usually been described as being formed perpendicularly
to the long axis (chief axis) of the egg. The first cleavage plane has,
accordingly, been characterized as equatorial, and the long axis of the
two-cell stage has been regarded as identical with the long axis (chief
axis) of the unsegmented egg.

In the following account ^ it will be shown that the first cleavage fur-
row appears approximately in the long axis (chief axis) of the egg ; and
that, therefore, the first cleavage is meridional, not equatorial as was
hitherto supposed. It will be shown, further, that the position of the
cleavage plane in the two-cell stage is due to a rotation of the dividing

^ Some notes on the first cleavage of L. anatifera have already been published
(Bigelow, '99).

78 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

e*'" as a whole through an arc of 90° -within the vitellhic membrane.
The long axis of the two-cell stage is, therefore, at right angles to the
chief axis, which has rotated 90° from its original position of coincidence
with the long axis of the vitelline membrane. The chief axis, which is
the longer axis of the unscgmented egg, becomes the shorter axis of the
two-cell stage. An examination of Figures 1-1 G, which represent a
series of camera lucida drawings made at intervals during cleavage, will
make clear the changes in form and position which the egg of Lepas
undergoes in the course of the first cleavage.

In a px'eceding chapter it has been shown that, after the formation
of the second polar cell and at about the time of tlie union of the pro-
nuclei, the yolk becomes partially separated from the protoplasm and
becomes aggregated at the vegetative pole of the egg (Figs. 2-G, 18-20).
Shortly afterwards it is shifted to one side of the polar area (Figs, 7, 8) ;
this is the first indication that the egg is rapidly approaching cleavage.
Soon a wide shallow groove appears, passing obliquely around the ovum
from the animal pole (Fig. 8). The furrow rapidly deepens and the
forming cells become spheroidal, causing the ovum to elongate perpen-
dicularly to the plane of cleavage (Figs. 9, 10). The ovum as a whole
at the same time gradually rotates within the vitelline membrane (Figs.
10-15) ; consequently the plane of cleavage rotates until, at the comple-
tion of cleavage, the furrow is usually transverse to the long axis of the
vitelline membrane, still unchanged in form; that is, the cleavage furrow
occupies a plane almost at right angles to that in which it at first ap-
peared relative to the vitelline membrane (compare Figs. 8 and 15).
These facts explain the conflict between the conclusions of earlier obser-
vers and the generally accepted idea that the first cleavage is meridional
in the ova of nearly all animals.

The figures show that the second polar cell continues to lie in the
cleavage furrow, and consequently has retained a fixed position with
reference to the egg during its rotation within the vitelline membrane.

In some ova the rotation is through less than a quadrant, so that at
the close of the first cleavage the plane of division is more or less oblique
to the long axis of the vitelline membrane. In examining living ova
taken at random, many oblique cleavage furrows are noticed, but con-
tinuous observation usually shows that the obliquity is the result of
preparation for the second cleavage. Accordingly, it may be stated as
a general rule that at the close of the first cleavage of the ova of Lepas
the cleavage plane is transverse to the long axis of the vitelline mem-
brane, and that only in comparatively few cases is it markedly oblique.

BIGELOW : EARLY DEVELOPMENT OF LEPAS. 79

In those eggs in which it is obUque at the close of the first cleavage,
the vitelline membrane appears relatively broader, and the divided
ovum is easily adjusted to an oblique position within the membrane.

Fifteen or twenty minutes usually elapse between the first external
appearances of division and the complete separation of the cells. From
the cases which I followed continuously it appears that the cleavage
beo-ins within two to three hours after the formation of the second polar
cell.

During this cleavage the ova are seen to undergo a series of marked
contractions, as shown in Figures 11 and 14. Immediately following
each contraction the cleavage furrow deepens and the ovum rotates
tiirough several degrees. These phenomena are probably due to the
action of the astral fibres, which, as will be shown later, are a well-
marked feature of the cleaving ovum. The external appearances would
lead one to think that the internal contractions occur spasmodically
rather than continuously. Similar appearances were many times noted
also in the later cleavages.

Additional evidence in support of this observation concerning rotation
of the dividing egg has been obtained from living eggs of L. fascicularis
and a species of Balanus. In L. fascicularis (Plate 11, Figs. 95-97) the
first polar cell has been observed to remain attached to the vitelline
membrane at its blunter pole until after the close of the first cleavage,
when the second polar cell, attached to the egg, has moved 90'' from the
blunt pole of the vitelline membrane. This observation is conclusive
confirmation of my earlier observations on L. anatifera.

While no observations have as yet been made on the living ova of
species of Cirripedia other than those already mentioned, the study of
preserved material of other species indicates that in these the first cleav-
age takes place as in L. anatifera and in L. fascicularis. In L. hillii, L.
pectinata, Pollicipes, and Balanus the chief axis coincides with the long
axis of the unsegmented ovum and of the vitelline membrane. After the
first cleavage, I find the polar cell in the cleavage furrow, which approx-
imately coincides with a transverse plane of the vitelline membrane.

So far as known similar relations exist between the ovum and the
vitelline membrane before and after cleavage in the ova of all Eucirri-
pedia ; therefore, it is very probable that cleavage takes place in the
entire group as in L. anatifera. Van Beneden's ('70) figures of Saccu-
lina suggest that the same may also be true for the ova of Rhizocephalan
Cirripedia.

The internal phenomena connected with the cleavage could not be

80 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

accurately interpreted from observations on the opaque living egg, but
sections of ova killed at various stages in the cleavage show some in-
teresting conditions. About the time when the pronuclei come into
contact, two clear areas are often seen near the pronuclei, as shown in
Figure 20 (Plate 2), but frequently in a plane more nearly transverse
than that in which they are shown in the figure cited. In the same
positions well-defined asters later make their appearance, and the first
cleavage spindle begins to form with its axis oblique to that of the vitel-
line membrane (Fig. 21). In many cases the spindle begins to form in
a plane almost perpendicular to the long axis of the ovum. This is true
particularly in L. fascicularis (compare Plate 11, Fig. 98).

In the metaphase of the mitosis the spindle is usually oblique to the
long axis of the ovum (Fig. 22) ; sometimes it is almost transverse
(Fig. 98), but never parallel to the long axis. In L. fascicularis it is
most frequently perpendicular to the chief axis, as shown in Figure
98. In L. anatifera the spindle is usually almost as long as the trans-
verse axis of the ovum. The astral radiations are very distinct, and
appear to be continuous with the general protoplasmic reticulum of the
cell (Fig. 22). In the stage of the living ovum corresponding to this
the yolk has taken an eccentric position at the vegetative pole (Fig. 7).
The relation seen to exist between the yolk and the aster nearest the
vegetative pole (Fig. 22) suggests that the movement of the yolk to the
eccentric position has some relation to the formation of the aster, for
it is during the development of that structure that the yolk moves to
the eccentric position.

In the next stage figured, an early anaphase (Plate 3, Fig. 23), the
spindle is still oblique and the cleavage furrow has not begun to form.
The chromosomes have separated along a plane which is usually inclined
to the plane in which the cleavage furrow later appears. This stage
corresponds to a stage of the living ovum which is slightly later than
that represented in Figure 7.

Figure 24 represents a stage in the anaphase after the cleavage fur-
row has become well developed, and the dividing ovum has begun to
rotate. This is the condition in stages of the living egg corresponding
to those shown in Figures 10-13. The central part of the spindle is
almost perpendicular to the plane of cleavage, but there is a distinct
bend in the spindle near either end. These bends may be regarded as
evidence of torsion. Comparing Figures 23 and 24, it appears that
during division there has been some shifting of the egg substance with
reference to the spindle, which is at first somewhat oblicpie to the plane

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 81

in which the cleavage furrow will appear ; but later, when the furrow
begins to form, the spindle becomes perpendicular to the plane of
cleavage. In L. fascicularis the spindle is usually from the very begin-
ning of cleavage perpendicular to the chief axis, in which the cleavage
furrow later appears. I have noticed the same conditions in the eggs
of a species of Balanus. In living eggs of Lepas I have observed move-
ments of the egg substances which lend support to the evidence afforded
by sections. Figures 8-11 represent conditions between the stages cor-
responding to Figures 23 and 24, and they show that the egg under-
goes great changes in form before rotation begins. It is probable that
the turning of the spindle takes place at the time of contractions of the
egg such as those represented in Figures 9-11.

The astrospheres are well-marked features of the anaphase (Fig. 24),
and are distinctly visible as clearer regions in the living egg.

In a late anaphase the spindle has become straight again and is per-
pendicular to the cleavage plane (Fig. 26). The rotation of the ovum
is now completed. In this stage the cells are still connected in the
centre by a mass of cell-substance, surrounding the spindle (Fig. 26).

Finally, in the telophase the chromosomes swell into vesicles, and
then fuse together to form the nuclei of the two daughter cells in a
manner well known for other ova (Figs. 25-27). The cell plate is next
completed, and then the separation of the cells (ab\ cd}) is accomplished.
Remnants of the spindle may persist for some time, and a well-marked
" Zwischenkorper " is often seen.

Figure 25 represents the condition in the comparatively rare cases
in which the cleavage plane remains oblique in an early telophase.

In observing the living egg it was noted that at the close of the
anaphase the protoplasm of the yolk-cell {cd?') is centrally located and
that the yolk remains in its original position in the vicinity of the pointed
end of the vitelline membrane (Figs. 15, 26). The chief axis of the egg
now coincides with the transverse axis of the oval vitelline membrane,
the animal pole being marked by the second polar cell, which lies in the
cleavage furrow. The formative and nutritive materials of the yolk-cell
are not as yet arranged with reference to the chief axis, as they naturally
would be if they kept their original relations to the chief axis during
the rotation of the dividing ovum. It has been observed that in
the living egg the yolk and the central mass of protoplasm move
to their respective poles in from twenty to fifty minutes after the com-
plete separation of the cells (Figs. 15, 16). It will be seen later that
this can have nothing to do with the processes of the second cleavage,

82 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

which occur two to three hours later. Sections of ova which were fixed
at intervals during the first hour after the close of the first cleavage
show that the above mentioned movement of protoplasm and yolk occurs
at about the time when the spindle and asters have disappeared (Fig.
27). These facts suggest that the spindle and asters may have in some
way inhibited the movement of the yolk in its return to its orig-
inal position at the vegetative pole of the chief axis, out of which it
appears to have been forced during the rotation of the dividing egg.
The relative positions of spindle, protoplasmic mass and yolk, as shown
in Figures 22-27, seem to lend support to this suggestion. The spindle
and astral radiations appear to be arranged so as to hold the cell-sub-
stances in the same relative positions which they occupied before the
cleavage (Figs. 7, 22) ; with the disappearance of the spindle and asters
the mass of protoplasm apparently became free to move toward the
animal pole, while the yolk was moved to the vegetative pole (Plate 1,
Fig. 16 ; Plate 3, Fig. 27). It seems that the formative and nutritive
materials after having been displaced return to their respective poles
of the egg as soon as the displacing and inhibiting cause is removed. In
this case the tendency to return to the original polar relations seems
to be related to the phenomenon of cell-polarity, the causes of which
are thus far hidden.

Throughout cleavage the mass of protoplasm in the yolk-cell re-
mains at the animal pole of the egg, which is marked by the second
polar cell, and the successive blastomeres formed by the unequal division
of the yolk-cell are cut off as near the animal pole as is consistent with
the position of previously formed cells.

Conkliu ('97) has pohited out for the egg of the gasteropod Crepidula
a tendency of the protoplasmic mass in the macromeres to remain near
the animal pole, while successive ectomeres are cut off as near that pole
as the position of previously formed cells will allow. The condition in
the egg of Lepas furnishes a parallel case, and the return of the pro-
toplasmic mass to the polar position after displacement in the first cleav-
age indicates a strong tendency towards adherence to the original
polarity of the unsegmented ovum.

The rotation of the dividing ovum appears to be dependent upon the
cleavage processes, and capable of an explanation along mechanical
lines. The cleavage furrow arises in an almost longitudinal position,
passing through the animal pole (Plate 1, Fig. 8). As the furrow
deepens, the forming cells tend to become spheroidal and hence to
lengthen the axis of the ovum perpendicular to the plane of cleavage

BIGELOW: EAIILY DEVELOPMENT OF LEPAS. 83

(Figs. 9-11). If no firm envelope confined the ovum, interfering with
change in its form, the long axis of the two-cell stage would be per-
pendicular to the plane in which the cleavage begins ; but the vitelline
membrane evidently does interfere with extension in a direction per-
pendicular to that plane. Therefore, as the cleavage progresses and the
resulting cells become more and more spheroidal (Figs. 10-13), a rota-
tion of the ovum becomes necessary, for evidently the long axis of the
two-cell stage must approximately coincide with the long axis of the
vitelline membrane. An examination of the figures makes it appear
that, as the forming blastomeres become more spheroidal and conse-
quently increase the length of the axis of the ovum perpendicular to
the plane of cleavage, pressure is obliquely applied to the vitelline mem-
brane with the result that the ovum as a tvhole rotates, and gradually
the dividing ovum adjusts itself to the form of the vitelline membrane.
The cleavage plane becomes transverse or oblique, depending npon the
amount of rotation necessary to meet adjustment. With a relatively
wide vitelline membrane the rotation is less than 90°, for the divided
ovum can then become adjusted to an oblique axis of the membrane,
and the cleavage plane consequently remains oblique.

A rotation of the ovum as a result of cleavage has also been shown in
the case of the rotifer Callidina, described by Zelinka ('91). Like that
of Lepas, the ovum of Callidina is ellipsoidal and surrounded by a rigid
membrane. The polar body is situated at one end of the ovum, and
the cleavage plane passes through this point. Zelinka figures an ob-
lique spindle, but no sections showing the relations in the various stages
of mitosis. According to Zelinka the rotation of the ovum occurs after
division, but the extent of the cleavage plane at the time of rotation
was not determined by study of sections. It seems probable that, as
in the cirripede ovum, the rotation may be found to take place during
the division.

Jennings ('96, p. 20), commenting upon the rotation in Callidina,.
writes: — "It thus appears that in Callidina the direction of division
itself is determined neither by the principle of Berthold [surface ten-
sion] nor that of Hertwig [spindle in long axis of protoplasmic mass],
hut that the later arrangement of the cells might be held to be due to
the action of Berthold's principle." The conditions in Lepas appear to
be similar to those in Callidina, and Jennings' conclusion is applicable
in the case of the cirripede.

In the eggs of some nematodes there are conditions at the time of
fertilization very similar to those existing in Lepas. The contiguous

84 bulletin: museum of comparative zoology.

surfaces of the pronuclei are in a plane which is perpendicular, or slightly
oblique, to the long axis of the ellipsoidal egg, and the spindle often
bc<nns to form with its long axis in the same transverse plane. Several
investigators, among wliom may be cited Auerbach ('74, p- 212, Taf. 4) and
Ziegler ('95, pp. 379-387), have observed that there occurs a turning
of the pronuclei around each other so that their contiguous surfaces and
the spindle axis come to coincide with the chief axis of the egg. This
turning of the pronuclei and spindle appears to be brought about by
streaming movements of the substances of the egg. In addition to these
observations on the nuclei during their rotation, there is evidence in the
two-cell stage of the nematode that the egg as a whole has not rotated,
for the polar cell remains in the long axis of that stage 90° from the
equatorial cleavage plane.

As a result of the turning of the pronuclei and the consequent longi-
tudinal position of the spindle, the nematode egg divides in such a plane
that the two-cell stage does not require readjustment in order to ac-
commodate its long axis to that of the surrounding egg envelope. Thus
the turning of the pronuclei and spindle in the nematode eggs affects
the orientation of the two-cell stage as completely as does the rotation
of the dividing egg as a ivhole in the case of Lepas, My observation
that in L. anatifera the spindle often appears to begin its formation in
a transverse plane and then becomes oblique, suggests that there is a
tendency towards coincidence of the spindle axis with the long axis
of the egg. If such a tendency really exists, it is inhibited by some
unknown conditions, possibly the yolk-mass influencing the streaming
of the protoplasm, and as a result the cleavage plane is formed in
such a position that the two-cell stage must become readjusted to
the vitelline membrane.

Summary of the First Cleavage.

It has been shown that in L. anatifera, L. fascicularis, and a species
of Balanus, the cleavage plane lies at the beginning of cleavage approxi-
mately in the long axis of the unsegmented ovum as well as that of the
vitelline membrane, and passes through the animal pole. During the
division a rotation of the ovum as a lohole through an arc of 90° takes
place, so that at the close of the division the plane of cleavage coincides
with the transverse axis of the vitelline membrane.

The evidence afforded by preserved material and published figures
makes it probable that a rotation of the dividing ovum occurs in all

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 85

Cirripedia which have ellipsoidal eggs surrounded by a rigid vitelline
membrane.

The rotation appears to be due to the mechanical relations existing
between the dividing ovum and the vitelline membrane.

The first cleavage is a typical case of unequal cell division ; this is
widely at variance with the account given by Groom (see the following
review of the literature).

3. Review op the Literature on the First Cleavage.

According to the accounts or figures of Fillippi ('65), Mlinter und
Buchholz ('69), Hoek ('76), Lang ('78), Nassonow ('87), and Groom
('94), the first cleavage plane in all the species of Lepadidse and Balan-
idfe, which have been studied by them, is generally transverse to the
chief axis ; but it has been sometimes described as occasionally more or
less oblique owing to variation. These investigators noticed that the
long axis (chief axis) of the unsegmented ovum coincides with the long
axis of the vitelline membrane, and that in the two-cell stage the plane
of separation is transverse to that axis. These positions of the egg
with reference to the vitelline membrane before and after cleavage led
to the view that the first cleavage plane is formed at right angles to the
chief axis of the egg, i. e., that cleavage is equatorial. Had the position
of the polar cell during and after cleavage been carefully observed, this
view would not have gained acceptance. Of the above named authors
Groom and Nassonow have figured the polar cell in the two-cell stage,
and they represent it as situated in the original position near the
rounded end of the vitelline membrane, 90° from the cleavage plane.

Nussbaum ('87, '90) observed in some ova of Pollicipes cleavage
planes in various degrees of obliquity with reference to the vitelline
membrane, from nearly longitudinal to transverse. He is the only
autiior who has figured or described a polar cell as lying in the cleavage
furrow of the two-cell stage of a cirripede egg. Nussbaum explained
tliese varying positions of the cleavage plane and polar cell with refer-
ence to the long axis of the vitelline membrane by assuming that the
ovum divides almost longitudinally, and that after division the egg
turns within the vitelline membrane. The various positions of the first
cleavage plane, which were observed by Nussbaum in different eggs,
were assumed to represent phases in the turning of the egg as it rotated
from the position in which the forming cleavage plane is nearly longitu-
dinal to the final position, in which it is transverse. Nussbaum sug-

86 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

gested that the turn of the egg might be explained on the principle of
least resistance, since the long axis of the divided egg can only be ad-
justed to the long axis of the vitelline membrane. He failed to study
sections of stages in the first division and to follow continuously the
cleavage of a living ovum. Groom ('94) expressed doubt concerning
Nussbaum's identification of the body in the cleavage furrow as the
polar cell, for it had not been followed continuously from its formation.
Nussbaum's figures of three different ova with cleavage planes respec-
tively in almost longitudinal, in oblique, and in transverse positions do
not give conclusive evidence in support of his assumption that the egg
rotates after cleavage. Groom has remarked that, if a rotation occurs,
an ovum with oblique cleavage plane should show a correspondingly
situated polar cell, and Nussbaum's figure of such a stage does not
show this. So far as the evidence offered by Nussbaum is concerned,
one might well accept Groom's view, that the various positions of the
first cleavage plane in different ova indicate merely variation of the posi-
tion in which it forms.

Although Nussbaum failed to support his assumption with conclusive
evidence, he was certainly in the main correct, as the evidence offered in
this paper proves. Studies of the preserved material have convinced me
that the relations in Pollicipes agrees with those in Lcpas. Nussbaum's
assumption that the rotation takes place after division does not agree
with the facts in the case of Lepas. I have shown that the rotation
takes place not after, but during division, and have suggested that the
forces concerned in cleavage, reacting upon the rigid vitelline membrane,
are apparently the cause of the rotation of the dividing ovum.

Groom's account of the first cleavage is so involved with his descrip-
tion of the separation of the protoplasm from the yolk during matura-
tion that no sharp line is drawn by him between the two processes. I
quote from his paper ('94, pp. 135-13G) the following description: —

" The polar bodies become pale and disintegrated, and the external one often
gets washed away. The protoplasm is at last mainly collected at the anterior
pole of the egg, and the yolk at the other (Figs. 6, 7)- . . . The surface separ-
ating the protoplasmic half from the yolk commonly intersects the ovum in a
perfect circle, and marks off what will form the first blastomere. . . . Very gene-
rally the line of separation of the protoplasm and yolk is almost accurately
transverse, ... I have frequently seen cases when the wall was accurately
transverse, and the polar body situated apically (Figs. 6, 7). Lastly I have
been able to watch the gradual formation of the protoplasmic half in a single
ovum ; the line of junction in these cases was transverse from the first."

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 87

It is evident that this account refers to the processes which I have
described in the chapter on maturation of the ovum. They are phenom-
ena concerned with the establishment of visible polarity in the egg,
and not with the cleavage process, as Groom's account leads us to infer.
The surface marking the boundary of yolk and protoplasm, as shown in
Groom's Figures 6 and 7 (in this paper Figs. 3 and 18), does not
*' mark off what will be the first blastomere." Groom evidently mis-
took the constriction which I have described in the account of matura-
tion (Fig. 3) for the forming cleavage plane ; but I have shown the
cleavage plane to be almost perpendicular to this transverse constriction,
which merely marks off the yolk-lobe (see Figs. 3 and 18). Groom's
misinterpretation explains the cases described by him, in which the
cleavage plane appeared transverse and the polar cell apical in position ;
see his Figures 6 and 7, which evidently correspond to my Figures 3
and 18. Groom has interpreted his Figures 6, 7 and 8 (L. anatifera),
and 45, 46 and 47 (L. pectinata) as representing successive stages in the
formation of the first cleavage plane. As a matter of fact there inter-
vene between the last two stages of each of these series all the stages
which are shown in this paper by Figures 4-15. The identification by
Groom of the transverse constricting furrow of the maturation period as
the forming cleavage furrow has probably led to his erroneous interpre-
tation of the position of the polar cell with reference to the first cleavage
plane. It was natural that Groom, considering the three figures men-
tioned above (Figs. 6, 7, 8) as a continuous series, should expect to find
the polar cell at the place of its formation, and should overlook it in the
first cleavage farrow. The best of observers could easily have been mis-
led, unless an opportunity came for following a single ovum uninterrupt-
edly through the maturation and first cleavage stage. The polar cell lies
deep in the cleavage furrow, and is easily overlooked in the living ovum,
unless one's attention has been attracted to it in prepared ova, where it
is clearly shown in the majority of cases. The rare cases observed by
Groom of ova in which the polar cell retained its original position in
undoubted two-cell stages are explained by my observation that the
polar cell sometimes, but very rarely, fails to rotate with the ovum.
That the polar cell is not soon lost, as Groom believed, is evident from
many of my figures of later stages. In preparations it is as often seen
i-n later stages of cleavage as in the unsegmented ovum.

Groom's Figure 101 (L. anatifera), showing a longitudinal position
of the spindle, is certainly from a section taken in a plane oblique
to the chief axis so as to show the spindle in the long axis of the sec-

88 bulletin: museum of comparative zoology.

tion. A spindle parallel with the chief axis would be in harmony with
Groom's view that the first cleavage furrow is perpendicular to tliat
axis. Numerous transparent preparations of entire eggs have convinced
nie that such is never the case.

In the review of literature on maturation and fertilization I have
already referred to Groom's mistake in identifying the pronuclei as the
daughter-nuclei of the segmentation nucleus. He speaks (p. 145) of two
nuclei seen in "the first blastomere " (cell ah'^ of this paper). One of
the two nuclei which he regards as the daughter-nuclei of the segmenta-
tion nucleus remains as the nucleus of " the first blastomere," the other
j)as8e3 into the "yolk hemisphere" (yolk-cell cd? in this account) just
before the cell-plate is formed. This is certainly erroneous, and is ap-
parently the result of his interpretation of the transverse furrow accom-
panying maturation as tiie cleavage furrow. In Groom's Figure 8 two
distinct nuclei are represented in the "protoplasmic" part of the egg,
which he considered " the first blastomere." It is evident from my
fiirures that the dauerhter-nuclei of the segmentation nucleus could not
normally get into such a position ; but the pronuclei are often seen on
one side of the constriction during maturation phases (see my Figure
18). I interpret Groom's Figure 8 as representing the pre-cleavage
stage corresponding to my Figures 3 and 18, and the lower half of the
egg as the yolk-lobe, not the yolk-cell cd^. I have already stated that,
unless eggs are kept under continuous observation, it is easy to confuse
this stage with the two-cell stage, when only living eggs are examined.
My series of figures shows that no such interpi-etation as that above
quoted fits the facts. There are two nuclei (pronuclei) in the proto-
plasmic hemisphere during the later maturation phases (Figs. 18, 20) ;
but in the " first blastomere " (cell ah^ in my Figs. 26, 27) there are never
two, one of which is destined to pass into the yolk. Groom's description
of the " yolk " (cell cd}) as at first without a nucleus, but receiving one
from the " first formed blastomere " (first micromere ab'''), is erroneous.
Neither cell can be said to receive a nucleus from the other, fur the
division of the segmentation nucleus, and the formation of the first
cleavage plane is such as ordinarily takes place in unequal cell division.

The last statement applies also to allthe later cleavages. The micro-
meres rich in protoplasm, which are later cut off from the yolk-macro-
mere, cannot be said to give rise to a nucleus which migrates into the
yolk before complete separation of the " protoplasmic " cell.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 89

4. Second Cleavage. Four Cells.

The first cleavage results in the division of the ovum into two cells of
unequal size ; the smaller cell (first micromere ah^), which is anterior in
position, is largely protoplasmic, whereas the larger, posterior cell {c(F)
contains the yolk, and will be designated as " yolk-cell." For conve-
nience in description this cell is regarded in the following account of
cleavage as a macromere ; it retains its individuality during three suc-
cessive unequal cleavages, giving rise to three "protoplasmic " micromeres,
the yolk after each cleavage remaining in the larger daughter-cell, which
in each stage will be designated as " yolkcell." The addition of the ex-
ponent indicating the cell generation will pi-event the confusion which
would arise from the use of the term *' yolk-cell " alone, when applied to
the cell d^, d^-^ or d^-^, which are the yolk-bearing derivatives of the cell
cd^ of the two-cell stage. The micromeres are numbered in the order of
their separation from the yolk-cell, aV^ being the first and c^ the second.

The nearly synchronous successive divisions of the first two cells {ab\
cd^), and afterwards of their derivatives, result in " resting " stages of
the egg, which normally consist of 2, 4, 8, 16 and 32 cells, and it be-
comes easy to classify the successive cleavages of the egg as second, third,
fourth and fifth. It will be noticed, however, that in the second and
following cleavages the yolk-bearing cell tends to divide after the other
cells, and that its division becomes more retarded at each successive
generation. This seems to be correlated with the fact that at each divi-
sion the protoplasm in the yolk-cell is diminished in proportion to the
amount of yolk. In the fourth and fifth cleavages the yolk-cell usually
completes its division just as the other cells prepare for the next cleav-
age. However, it is not until after the fifth cleavage (thirty-two cells)
that it lags a full generation behind the other cells. The cleavages can,
therefore, be classified naturally according to the resting stages, each
stage containing twice as many cells as the preceding.

Tlie second cleavage may take place in the cells aW- and cd'^ simulta-
neously (Fig. 28), but either cell may complete the cleavage slightly in
advance of the other. In the majority of cases division of the anterior
cell iaW) precedes (Fig. 99), but usually the differences in the phases of
mitosis in the two cells are very slight.

In both cells the mitotic spindles for the second cleavage are formed
perpendicularly both to the first cleavage spindle (compare Figs. 26 and
28) and to the chief axis of the egg. In the first micromere {ah^) the
spindle is centrally situated ; the cleavage plane is formed at right angles

90 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

to the first cleavage plane, and passes through the animal pole of the egg
(Figs. 29, 30).

The spindle in the yolk-cell cd- is eccentric in position, lying nearer
the animal pole of the egg, and near the centre of the protoplasmic
mass; it is nearly perpendicular to the chief axis (Fig. 29). As cleavage
progresses the spindle becomes inclined so that one end dips into the
yolk-mass, which lies at the vegetative pole of the yolk-cell (Figs. 31 and
99). From the point of view of a miniature observer occupying the
chief axis of the ovum with his head directed toward the animal pole, the
left end of the spindle is the one that is nearer the animal pole, that is,
the spindle is laeotropically oblique. Usually the spindle makes an
angle of about 30° or 40° with tiie chief axis.

The yolk-cell cd'^ cleaves unequally, and the cleavage plane may be
considered a modified meridional one. The cleavage planes of the "pro-
toplasmic " cell ab^ and of the yolk-cell meet in a line which passes
through the animal pole, but does not coincide with tlie chief axis ; it
makes with this axis an angle of about 45°. To our imaginary observer
the resulting smaller cell (c^) lies to the left of and above the larger or
yolk-cell d^ (Fig. 31), and also this cell lies above the anterior cell b^.
The cell c* is the second micromere which is separated from the yolk.

At tlie close of the second cleavage a general tendency towards a Ino-
tropic arrangement of the cells is noticed (Figs. 32-34, 100-102). This
arrangement in the case of the posterior cells (c", d^) is apparently the
result of the oblique position of the spindle in the yolk-cell cd'^. When-
ever the anterior cell ab^ (first micromere) divides in advance of the yolk-
cell cd', there is no suggestion of a lipotropic arrangement either in its
spindle or in the position of the resulting cells (a^, J/, Fig. 99) ; but
after cleavage of the yolk-cell, the right anterior cell i* is depressed by
the higher lying cell c'. This change can be seen in the living ovum as
the cleavage of the yolk-cell cd^ progresses.

Soon after the completion of the second cleavage the four cells tend
to become rounded, and adjustments of position occur. Figures 32-35
and 102, 103 represent the arrangements which are usually seen, and
in all of them a definite plan can be recognized. The axis of tlie future
eiul)ryo can now be described as passing through the nuclei of the an-
terior cell, i', and of the yolk-cell, rf* (Fig. 31). The anterior cell, b^,
always comes to lie nearer tlie vegetative pole than the cells a* and c',
and it is usually more or less covered on the animal side by one or both
of these cells (Figs. 34, 35). After examining the eight-ccU stage, in
which the bilateral symmetry is distinctly marked, it will be seen that

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 91

the arrangement of the cells in the four-cell stage and of the spindles
for the next cleavage are such that the daughter cells invariably assume
definite and constant positions in the eight-cell stage.

Summary of the Second Cleavage.

Both cells of the two-cell stage divide nearly or quite simultaneously.
The second cleavage plane is meridional and perpendicular to that uf the
first cleavage. The first micromere {ab'^) divides equally, whereas the
yolk-cell cd^ divides unequally, giving rise to the second micromere, c*.

After the second cleavage the four cells (a*, b^, c^, d^) become adjusted
in a laeotropic arrangement.

In the four-cell stage a plane passing through the second polar cell
and the nuclei of cells h^ and d^ is apparently near the sagittal plane of
the future embryo. In this stage, then, there is a suggestion of bilateral
arrangement of the cells.

The yolk-cell undergoes ordinary unequal cleavage (see the following
review of the literature).

5. Eeview of Literature on Second and Succeeding Cleavages.

In this connection it is necessary to give a general review of the litera-
ture bearing on all early cleavages after the first, because no previous
worker has recognized definite stages into which the cleavages of the
cirripede ovum can be grouped. It is therefore impossible to make any
comparison of my account with that of others, except in a general way.

The division of the " protoplasmic " cell {ab-) of the two-cell stage of
the cirripede egg has been correctly described by most authors. The
plane of cleavage has been generally described as perpendicular to the
first cleavage plane, but Nussbaum ('90) has recognized that in Polli-
cipes it intersects the first cleavage plane at the polar cell and is, there-
fore, meridional.

No investigator of the early development of Cirripedia, except Groom,
has shown that the yolk-cell, cd"^, of the two-cell stage divides and adds
new cells to the blastoderm. All other observers, Buchholz ('69), Hoek
('76), Lang ('78), Nassonow ('87), and Nussbaum ('90), have described
the yolk-cell cd"^ as remaining undivided while the other cell (ab'^ re-
peatedly divides and its products grow around the yolk-cell, forming the
blastoderm. After completion of the blastoderm, and closing of the
blastopore, the yolk-cell cd^ was said to divide, separating the mesoblast
from the entoblast. According to this view the cell ab', which forms

VOL. XL. NO. 2 3

92 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

the blastoderm, contains only ectoblastic material. An exception is to
be noted in the case of Nussbaum, who saw the mesoblast apparently
proliferating from the edge of the blastoderm. The cell ab'^ according
to his interpretation, then, contains all the ectoblast and the mesoblast.

The erroneous interpretations of the earlier observers are largely
explained by the fact that their observations were almost exclusively
confined to living eggs, in which the nuclear conditions are hidden.
Without sections or transparent preparations divisions of the yolk-cell
might be easily overlooked. Lang ('78) and Nassonow ('87) figured for
Balanus, and Nussbaum ('90) for Pollicipes, distinct protoplasmic radi-
ations in the yolk-cell, but failed to see their significance as indicating
division. I am convinced that the structures seen were asters or archo-
plasmic radiations, Korschelt und Heider ('90) made the suggestion,
based on Nassonow's figures, that the yolk-cell ccP divides and contributes
cells to the blastoderm.

Groom ('94) described the yolk-cell cd"^ in the case of all cirripedes
whose development he observed, as a macromere giving rise in succession
to a number of " blastomeres," which are added to the blastoderm. He
proved conclusively that the "protoplasmic" cell ab^ (his "first blasto-
mere," my " first micromere ") does not give rise to all of the ectoblast,
as supposed by all previous observers. According to his account several
cells (estimated at nine or ten) are cut off from the yolk-cell after the
first cleavage, and with the derivatives of the " first blastoraere " form
the blastoderm.

Several years ago, without knowledge of Groom's results, owing to the
inaccessibility of the literature, I ('96) found that in Lepas fascicularis
the yolk-macromere divides several times, practically synchronously with
the divisions of the other cells, thus contributing to the formation of
stages of 2, 4, 8, 16 and 32 cells. This confirmed Groom's results in
general ; but as to the order, method, and number of the divisions I was
forced to dissent from his account.

According to Groom's description there is great variation in the num-
ber, order, and position of cleavages both in the yolk-cell and in the
other cells of the cleaving egg. He concluded that the cleavage of the
cirripede egg is decidedly irregular. Ho \yrite3 (p. 140), "there is no
constancy in the mode of growth of the blastoderm over the yolk ;" and
mentions (pp. 139-140) many of the variations which occur.

Many of these supposed variations are certainly misinterpretations
due to errors in orientation, and others are apparently based upon ab-
normal eggs. Mention may be made of several cases. Groom states

BIGELOW: EAKLY DEVELOPMENT OF LEPAS. 93

that the " second blastomere " (cell c^, second micromere, in my figures)
may be formed on either side of the yoik-cell d^, and illustrates such
conditions by his Figures 10 and 12 (L. anatifera). There is nothing
in either his text or figures to prove that these are not entirely similar
eggs viewed from almost opposite poles. They were certainly drawn
from different points of view, and the apparently different positions
occupied by the " second blastomere " are thus easily explained. Like-
wise, the "third blastomere" {d*'^, third micromere, in this paper) is
said to arise on either the right or left of the second. Groom's Figures
15 and 16 (L. anatifera), which illustrate this, are certainly views of two
similar eggs, and apparently the cell considered the *' second blastomere "
is not the same in both cases. The position of the " third blastomere "
shown as " emerging from the yolk," in one figure on the right and in
the other on the left, I interpret as being near the animal pole of the egg.
A number of other cases of such results based upon uncertain orientation
of the egg might be drawn from Groom's paper ; but enough has been
said to show that his evidence is far from convincing, tliat there is much
variation even in the earliest stages, and that the assumed variability of
the later stages rests upon a very uncertain basis. In opposition to this
view of the cleavage of the cirripede egg as variable and irregular, I shall
give evidence supporting my interpretation of the cleavage of Lepas as
normally regular and constant.

In this connection I wish to consider Groom's account of the method
in which the yolk-cell divides. The discussion will apply to the second
or any later cleavage by which blastoderm cells are cut off from the
yolk-cell, for the method of division is the same in all.

The following quotations from Groom's paper give his interpretation
of the method by which new cells are formed from the yolk-cell. On
page 197 he writes : "As the first blastomere becomes cut off from the
yolk the nucleus divides and one daughter-nucleus passes into the yolk
half, and soon emerges accompanied by protoplasm to form a second
blastomere and generally situated close to the first. As this becomes
cut off from the yolk it gives off into the yolk a nucleus, which behav-
ing similarly to the daughter-nucleus of the germinal vesicle, forms new
protoplasm and emerges as a third blastomere. At each successive
stage the yolk is in communication with one merocyte or newly-forming
blastomere, and this, before becoming shut off as a blastomei'e, gives off
a single nucleus into the yolk." A similar statement on page 145 of
Groom's paper contains some other points to which it will be necessary
to refer. One daughter-nucleus of the segmentation nucleus is said to

94 bulletin: museum of comparative zoology.

" pass into the yolk hemisphere, where it transforms yolk material into
protoplasm ; the second merocyte, formed partly in tnis way and partly
from previously existing protoplasm, issues as the second blastomere,
while the first becomes simultaneously cut off from the yolk . . . the
nucleus of the third merocyte is derived from that of the second ; the
latter becomes spindle-shaped, and gives off a nucleus, which, accom-
panied by little or by no appreciable quantity of protoplasm, passes into
the yolk. . . . The third merocyte, in similar manner, while emerging
as a blastomere, divides and gives off a nucleus to the yolk, which in a
similar manner gives rise to new merocytes and blastomeres."

It is evident, as indeed Groom distinctly states in another place, that
he regards the yolk as non-nucleated and receiving nuclei from the suc-
cessively formed blastomeres. In the discussion of the first cleavage I
have pointed out that a nucleus from "tiie first blastomere" (the coll
ab^ in this paper) does not pass into the yolk cell just before the separa-
tion of the two cells. This also applies to all succeeding cleavages. The
yolk-cell does not derive its nucleus from successively formed ''proto-
plasmic " cells (" blastomeres ") — such a description is inaccurate and
misleading. In no case can either " blastomere " or the yolk-cell be
said to derive its nucleus from the other, for the micromeres are merely
the result of ordinary unequal division, which differs frum the division
of cell ab^ in the inequality of the products, but not in the method by
which it is brought about.

The term "merocyte" conveys the idea that the protoplasm is more
or less sharply distinct from the yolk, as in the case of eggs whicli un-
dergo superficial cleavage. This is evidently the idea intended to be
expressed in the above quotations from Groom. Neither living eggs
nor stained sections support such an interpretation. A considerable
part of the yolk-cell rcP is protoplasmic, the yolk and protoplasm being
80 mingled that there is no justification for the use of the term " mero-
cyte." I cannot agree with Groom's statement that throughout the
main portion of its mass the yolk-cell contains little protoplasm. Pro-
toplasmic processes extend even among the oil droplets which lie near
the periphery at the vegetative pole of the egg (Fig. 27). I cannot
confirm the statement (p. 198) that there is little protoplasm left in
the yolk-cell immediately after the separation of a new blastomere, and
that the nucleus rapidly transforms yolk into protoplasm to form the
new blastomere. The amount of yolk is not very much diminished
before the sixth cleavage. This is in accord with the fiicts known
in the case of the development of other animals, for rapid transforma-

BIGELOW: EAKLY DEVELOPMENT OF LEPAS. 95

tion of yolk during cleavage has rarely been described. The mass of
protoplasm in the yolk-cell after the first cleavage is certainly nearly
equal in volume to the next cell (second micromere c*) which will be
cut off (see Fig. 27). The same is true for the later cleavages. All
these facts, together with those relating to the nucleus which were
mentioned in the preceding paragraph, are opposed to the idea of an
" emergence of merocytes from the yolk," and support the interpreta-
tion which I have given, viz., that all divisions of the yolk-cell are cases
of unequal total cleavage. There is nothing to warrant the phrase
" emergence of merocytes."

In concluding this general discussion of the method of cleavage of the
yolk-cell, I wish to emphasize the statement that there appears to be
no reason for regarding that cell in any of the cleavage stages as essen-
tially different in its nature or in its method of division from such well-
known examples of yolk-macromeres as are found in gasteropod eggs.
So far as I have found, the division of such macromeres is described as
differing essentially from that of other cells more rich in protoplasm
only in the inequality of the products. Furthermore, I can see no
essential difference between the process of cleavage in the yolk-cell of
L. anatifera, where there is much yolk, and in that of L. fascicularis, in
which there is relatively little yolk, and in which the division is clearly
of the ordinary unequal type.

According to Groom's account ('94, p. 137) a forming or " emerging
blastomere " is characterized by a radial arrangement of granules around
a clear central space situated near the periphery of the yolk-cell.
Groom's Figures 50, 86 and 88 represent this condition. He speaks of
the nucleus of the forming blastomere as the centre of the radiation (see
his Fig. 14). The clear area seen in a living egg at this stage is certainly
not the nucleus, but the astrosphere, tmd the radiations represent an
aster. Groom's description of the development of these structures
(p. 137) is good. During the division well-marked protoplasmic move-
ments give visible evidence of the differential distribution of the cell-
substances. The nucleus itself is not easily seen in the living egg at
any stage, and certainly is not vesicular at the time when the astro-
sphere is clearly defined. Figures 25, 26, and 30 represent sections of
eggs in which, when living, the centres of the radiations presented much
the appearance shown in Groom's Figures 10-15. The centres of the
radiations are seen to be the astrospheres, and the nuclei are repre-
sented by the chromatin vesicles, which are certainly invisible in the
living egg.

96 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Groom correctly described the radial arrangement of the protoplasm
as persisting for some time after cleavage. In my Figure 27 there is
represented a radial arrangement of granules which is a persistence of
the condition shown in Figure 2G as occurring at the close of the first
cleavage. The astrospheres have disappeared, and the nuclei lie near
the centres of the persisting radiations. This radial arrangement dis-
appears as soon as the second cleavage spindle forms (Fig. 28), but the
new radiations then formed may in turn persist after the cleavage until
the formation of the spindles for the third cleavage (Fig. 30).

Groom ('94) states that two or more blastomeres may arise simul-
taneously from the yolk-cell ! " Similar cells [blastomeres from the
yolk-cell] are seen to arise in quite different positions at later stages,
.sometimes two or more at a time," (p. 138). Again, on page 140 he
writes : " In the early as in the later stages the merocyte before emerg-
ing from the yolk may not uncommonly be seen to give rise by division
to a second merocyte." Such conditions are represented in Groom's
Figures 17a (L. anatifera), and also in his Figures 53 and 57 (Balanus).
Certainly none of these figures really represents two blastomeres arising
at once. The two sets of radiations (asters) which Groom wrongly
interpreted as two " emerging merocytes" probably represent cases in
which the spindle was in such a position that both asters were visible
at the surface. Usually, however, only one aster is to be seen in the
living egg, the other being closely connected with the yolk. Sometimes
the spindle is long, so that the two asters are visible on opposite sides
of the egg. I have frequently seen the two sets of radiations in the
living egg, and sections show that the interpretation which I have just
given is the correct one.

Sometimes multipolar spindles, which are probably the result of
abnormal conditions, are seen in sections of the yolk-cell, and these may
possibly result in a multiple cleavage.

Rarely the cell c^ (Groom's "second blastomere ") may be formed
near the posterior end of the yolk-cell, as shown by Groom in his
Figure 13.

Many other deviations from the regular course of cleavage have been
seen, but they are comparatively rare, and are to be regarded as abnor-
malities. Certainly they should not be interpreted as showing great
variability in the cleavage, as was done by Groom. I have noticed that
such cases are much more common when the animals have been kept
for some time in aquaria, but are rarely seen in eggs taken from ani-
mals whicli were recently removed from the open sea. 1 have attributed

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 97

these abnormalities to the action of chemical impurities and to lack of
oxygen. The respiratory movements of the animals are more sluggish
when they have been kept several hours in aquaria, and hence the eggs
in the mantle chamber may fail to get a sufficient amount of oxygen.
It is well known that such abnormal conditions may affect great modi-
fications in otherwise regular cleavage.

Orientation of the Embryo.

It has already been stated that in the four-cell stage a line drawn
through the nuclei of the cells b^ and cP coincides with the longitudinal
(antero-posterior) axis of the future embryo, the cell d^ being posterior.
This relation is shown in the orientation on the plate of Figure 31, from
which it also appears that the first cleavage plane is oblique to the same
axis. The chief axis of the egg coincides with the dorso-ventral axis of
the future embryo, the second polar cell at the animal pole being dor-
sal. The spherules of yolk are at the opposite pole of the yolk-bearing
cell, thus marking the vegetative pole and the ventral side of the em-
bryo. The blastopore later appears on this surface near the posterior
end of the egg.

The anterior end of the embryo lies, as several investigators have
noted, at the rounded end of the vitelline membrane. In the four-cell
and later stages the long axis of the vitelline membrane and that of the
future embryo apparently coincide, but in the two-cell stage the long
axis of the future embryo is oblique to that of the vitelline membrane.
The long axis of the embryo is brought into coincidence with that of
the vitelline membrane when the cells adjust themselves after the com-
pletion of the second cleavage (compare Figs. 31 and 32).

The animal and vegetative poles, which are marked respectively by
the second polar cell and the mass of yolk spherules, have a constant
relation to the blastomeres and to the planes of cleavage, and I have
made use of them as a basis for orientation. Previous investigators of
tlie cleavage of cirripede ova have recognized no definite and constant
points of orientation. In 1896 I pointed them out in the cleaving
ovum of L. fascicularis ; since then I have found that the polar cell
has exactly the same relations to the embryonic cells in all the stages of
cleavage in four species of Lepas and in Pollicipes polymerus.

98 bulletin: museum of COMrARATIVE ZOOLOGY.

G. TuiiiD Cleavage. Eight Cells.

The third cleavage is essentially equatorial. The spindle figures
arrange themselves approximately parallel with the cliief axis, and
therefore nearly perpendicular to the spindles of the preceding cleav-
ages. The spindle in the median anterior cell (l/^) is somewhat excep-
tional, in that it is more or less inclined toward the horizontal plane
. (Plate 4, Fig. 3G). The spindle in the yolk-cell d^ is generally more
nearly parallel to the chief axis. The cells a^, b^ and c^ often complete
their division in advance of the yolk cell (Plate 11, Fig. 103). Some-
times the spindle in the yolk-cell is just forming as the other cells
divide, but the yolk-cell completes the cleavage while the other cells
remain in the " resting " condition. Stages with five, six, or seven
cells are seen when examining living ova, but after preparation of such
ova the nuclei of some cells are found to be retarded in the third divis-
ion. Such variations in the rhythm of cleavage are not uncommon in tlie
synchronously cleaving ova of other animals. The normal " resting "
stage following the third cleavage in Lepas is composed of eight cells as
invariably as if the cleavage were perfectly synchronous in all of the
cells.

The positions of the cells which result from the third cleavage arc
shown in Figures 37-40 (Plates 4, 5), and 104-106 (Plate 11). The
three " protoplasmic " cells («', b^, c') have divided equally, the yolk-
cell unequally. The cell {d^'') which is cut off from the yolk-cell lies
in the median plane near the animal pole (Fig. 37). This is the
third micromere. The cells resulting from the division of a^ occupy the
left side, and are symmetrical with those derived from e^, which occupy
the right side of the egg (Fig. 37). The cell b^ has given rise to two
cells lying in the median plane, one {b*-^) near the yolk-cell at the
vegetative pole, the other {b'^''^) at the anterior end of the egg (Figs.
38, 40).

The seven " protoplasmic " cells have now begun to form the blasto-
derm (Plate 8, Fig. 66), which will later enclose the yolk-entoblast.
A very small space, Avhich is the cleavage cavity {cav. sq., Fig. 66),
is often seen in sections, but it soon becomes filled with yolk, by the
ingrowth of the yolk-cell.

The bilaterality in the arrangement of cells was indicated in the stage
with four cells ; it is well marked in tlie stage with eight. The charac-
teristic arrangement of tlic cells, as shown in Figures 37-40, is visible
in the great majority of living or prepared ova, if they are properly

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 99

oriented. The bilateral arrangement of cells when the egg is viewed
from the animal pole and the position of the yolk near the vegetative
pole (Figs. 38, 66) are features which aid in quickly identifying the
individual cells when the egg is rolled into proper positions.

During the third cleavage the polar cell is usually crowded beneath
the blastoderm, and comes to occupy in the cleavage cavity the position
indicated in Figure 66 — a condition which has been described as occur-
ring in the eggs of several other Eutomostraca. Sometimes at the close
of this cleavage it is found lodged between cells. Occasionally it be-
comes shifted in the earlier stages so that it no longer lies deep in the
cleavage furrow ; in such an event it is not forced beneath the blasto-
derm during the third cleavage, but may be found on the surface in
later stages. I have noticed it on the outside of the embryo in stages
as late as those of about five hundred cells. In such cases it is some-
times far from its normal position at the anterior dorsal side (animal
pole) of the embryo. In its usual position beneath the blastoderm the
polar cell is quite definitely situated until very late stages. In the
eight-cell stage it is almost equidistant from the two poles of the chief
axis of the egg; but it usually lies much nearer the animal pole after
the fourth cleavage, and is a very nseful *' landmark " for orientation of
the later stages. In good transparent preparations of entire eggs of any
cleavage stage the polar cell is clearly visible, and it is often seen lying
beneath the blastoderm in stages with over five hundred cells.

The yolk-cell of the eight-cell stage (d*-\ Plate 5, Fig. 40 ; Plate 8,
Fig. 66) contains onl}' future mesoblast and entoblast, and will be re-
ferred to as mes-entoblast. The third micromere (d*-^), separated
from the yolk-cell in the third cleavage, is purely ectoblastic, and is the
last cell containing ectoblast which is given off* from the yolk-macro-
mere. The ectoblast is, therefore, separated from the yolk-laden ento-
blast in the fii'st three cleavages, being contained in the derivatives of
tlie three micromeres, aP, c^ and d*-"^, which are separated from the yolk-
l)earing macromere in the first, second and third cleavages respect-
ively. A study of the cell-lineage through the later stages of cleavage
shows that the cells ab^ and c^ are not purely ectoblastic, but contain
a portion of the future mesoblast ; they may, therefore, be called mes-
cctoblasts. Of their descendants in the eight-cell stage, the cells at the
animal pole (a^*^ b*-% c*"^) are purely ectoblastic, while the lower cells
around the vegetative pole (a*'^ i**S c*'^) contain future "secondary
mesoblast" (ectoblastic mesoblast).

100 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Sumynary of the Third Cleavage.

The spindles for the third cleavage are essentially perpendicular to
those of the tirst two cleavages, the cleavage being practically equatorial.
The three cells a*, l^ and c^ divide equally and synchronously. The
yolk-cell d^, which is often slightly retarded, divides unequally, the
smaller, more protoplasmic, product (c?'*-^) of this division, being the
third and last micromere containing ectoblast which is separated from
the yolk-macromere.

The yolk-cell (c?**^) is now mes-entoblastic, and bilaterality in cleav-
age is well marked.

The arrangement of the cells of this stage is definite and constant.

The second polar cell is crowded into the cleavage cavity during the
third cleavasre.

*&^

7. Fourth Cleavage. Sixteen Cells.

The mitotic spindles for the fourth cleavage, shown in Figures 39, 40
(Plate 5), and 104-lOG (Plate 11), have a well-marked bilateral arrange-
ment. The cell b^'"^, at the anterior end of the egg, and also the cell
d*'^ have their spindles perpendicular to the sagittal plane of the future
embryo, and their cleavage planes coincide with that plane. In the
yolk-cell d*'^ the mitotic spindle approaches parallelism with the chief
axis, as in the third cleavage. In all the other cells the spindles are
parallel with the long axis of the egg.

The seven "protoplasmic" cells divide as a rule equally and quite
synchronously. Division of the yolk-cell d*-^ is dela^-ed more than in
the preceding cleavage, but is completed while the fourteen "protoplas-
mic" cells are in the "resting" phase following division (Plate 5,
Fig. 41; Plate 8, Fig. 67; Plate 11, Fig. 108). The stage with all
cells in the " resting " phase is composed of sixteen cells (Figs. 42, 43).
Tiie yolk-cell, as in the preceding divisions, has divided unequalh^ and
the smaller, "protoplasmic" cell (c?^*^) thus formed lies in the median
plane on the dorsal side of the embryo (animal pole) and immediately
posterior to the cells d^* and d^-^, which have resulted fi'om the division
of <;<-2, the third micromere (Figs. 42,- 44, 45, G8). This cell (rf^-^),
formed by division of the yolk-cell d*-^ m the fourth cleavage, is the
primary mesohlast, as will appear from the subsequent history of its
descendants, which sink beneath the blastoderm in a later stage. The
yolk-cell d^'^ is now })urely entohlastic. The cells a^-"^, U"-"^^ and c^-^, which,
touch the yolk-cell on the anterior and lateral boundaries of its uncov-

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 101

ered ventral portion (Fig. 43) are mes-ectoblasts, and the remaining
eleven dorsally-lyiug cells contain only ectoblast.

Figures 42-46 (Plate 5), and 107-113 (Plates 11, 12), show the
positions of the cells in the sixteen-cell stage, regarding which it will be
sufficient to call attention to their bilateral arrangement. All the cells
of the eight-cell stage, with the exception of the cell b*-^, which lies at
the vegetative pole (Fig. 40), divide so that their daughter cells both
lie either on the right or on the left of the median plane of the embryo.
The exceptional cell, 6*-\ divides in a plane parallel to the plane of the
preceding cleavage, and, consequently, the daughter cells {b^-'^ and b^-'^)
are not separated by a plane coinciding with the median plane of the
embryo (see Figs. 40 and 43).

The regular and definite arrangement of the cells represented in the
figures of the sixteen-cell stage is quite noticeable. This first suggested
to me that the arrangement had arisen from an equally definite one in
the earlier stages. Figures of a similar stage accompany the accounts
of other investigators, who seem to have observed a constant arrange-
ment of the cells in this stage.

At the sixteen-cell stage the " protoplasmic " cells have become ex-
tended far over the yolk-cell (compare Plate 5, Fig. 40 with Fig. 45, and
Plates, Fig. 66 with Fig. 68). This extension is due in part to the
addition of a new cell (the primary mesoblast) from the yolk-cell, but
more especially to the spreading of the blastoderm, which is caused by
division of the derivatives of the three micromeres {ab% c*, d*'^).

The blastopore is marked by that portion of the entoblast cell (d^'^),
which is still exposed to the exterior (Figs. 45, 46, 68), and it is widely
open. Eggs with a relatively small amount of yolk have the blastopore
more nearly closed ; but, as will be shown later, the number and order
of cleavages are constant whether an egg contains a large or a small
amount of yolk.

Summary of the Fourth Cleavage.

A sixteen-cell stage is regularly formed with cells of partictilar origins
occupying definite and constant positions in relation to other cells.

The derivatives of the three micromeres {ah'^, c^, d*-^) divide synchron-
ously. The yolk-cell d*-^ (mes-entoblast) is delayed in cleavage.

The primary mesoblast (d^-^) is separated from the yolk-cell c?*-^,
which is now entoblast.

The blastoderm is greatly extended during the fourth cleavage.

102 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

8. Fifth Cleavage. Thirty-two Cells.

All of the sixteen cells of the previous stage are involved in the fifth
cleavage, but the primary mesoblast cell (c?^-*) and the yolk-entoblast
(d^-^) arc greatly retarded in division (Plate 5, Figs. 44-4C). The four-
teen cells of the blastoderm divide about synchronously, but occasion-
ally some of the anterior cells slightly precede in the cleavage (Plate 5,
Figs. 44, 45 ; Plate 6, Fig. 47 ; Plate 12, Figs. 112, 113). The nuclear
spindles for this cleavage are arranged perpendicularly to those of the
preceding cleavage, with the exception of those in the three mcs-ecto-
blast cells (a^-^, b^-^, c^'^), which touch the yolk-cell at the blastopore
(Fig. 46). The spindles in the cells a^'^ and <^'^ are always somewhat
obli(iue to those of the preceding cleavage (compare Figs. 40, 45, 46).
They appear to be arranged more or less at right angles to the lines
along wliich the greatest pressure would.be exerted by the contiguous
cells of the blastoderm (see Figs. 45, 46), and the arrangement tlicrefore
seems to be in accord with the principle that spindles tend to become
arranged in the line of least resistance.

The spindle in the median cell U"-^ is sometimes placed almost longi-
tudinally (Figure 113), in which case the resulting cells {b^'^, 5*-*, Fig.
46) are arranged as in Figures 48, 52 and 116. Sometimes the spin-
dle in J^"^ is almost transverse (Fig. 112) and the resulting arrange-
ment of the daughter cells is shown in Figure 51. Many intermediate
oblique positions of spindle and cleavage plane have been noted. This,
too, is apparently a case of adjustment to least resistance. In the next
stage these two cells (6'-*, 6®-*) become so shifted in position that they
lie one to the right and the other to the left pf the sagittal plane, but
usually one is more or less in front of its companion. In the sixty-two-
cell stage their derivatives always form the anterior boundary of tlie
blastopore, although in the thirty-two-cell stage one of the cells (6®-^)
may not be in immediate contact with the yolk-entoblast, a condition
shown in Figures 48 and 52.

In Figure 70 (Plate 8) it is noticeable that the cleavage planes which
separate the mes-ectoblasts a^-', and c**-* from their sister cells (a®-*, c^-*)
are markedly oblique, so that the latter overlap the former. Attention
is here called to the tendency of cells around the blastopore to divide in
this manner, for in the succeeding stage there is a similar oblique divis-
ion of a®-' and c®.^, and the inner derivatives are overgrown by the outer
overlapping cells.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 103

About the time that the fourteen blastoderm cells have completed
their division, the primary mesoblast cell (d^-'^) prepares to divide, its
spindle being transverse to the long axis of the egg (Plate 5, Fig. 48).
The cleavage plane coincides with the sagittal plane of the embryo, and
the resulting cells form the posterior boundary of the blastopore (Fig.
52). The constant and definite position of these two mesoblast cells,
their retarded division, which gives them distinctive nuclear phases,
their tendency to stain less intensely than other cells, the definiteuess
of the position and cleavage direction of the surrounding cells — all
these features make it possible to identify positively the derivatives of
the primary mesoblast cell (d^-^) in this and the following stages.

The yolk-cell (entoblast, d^-^) is the last cell to undergo the fifth
cleavage ; it commonly divides about the time that the blastoderm cells
prepare for the next (sixth) cleavage ; but at times the cleavage of the
entoblast is so delayed as to be nearly simultaneous with the sixth
cleavage of the blastoderm cells. The nuclear spindle is usually almost
perpendicular to the sagittal plane (Figs. 52, 116, 117). A cleavage
plane, dividing the yolk nearly equally makes its appearance at this
stage, but it becomes more clearly visible about the time that the next
division takes place in the blastoderm cells, and it may therefore be
described later, in connection with the figures w:hich illustrate the
account of the sixth cleavage.

The blastoderm has been greatly extended since the last stage., owing
to the multiplication of its cells by division, and to the accompanying
increase of surface produced by the flattening of the cells. The blasto-
pore has become less extensive as the yolk-cell (entoblast) has become
more completely covered (Plate 6, Figs. 51, 54 ; Plate 8, Fig. 69). It
is filled by the protoplasmic portion of the yolk-entoblast, and is bounded
posteriorly by the two primary mesoblast cells {d^'^, d^-*), anteriorly
and laterally by the four mes-ectoblast cells (a'•^ 6®'^ 6®*, c^-^). With
the exception of these four cells, which are in contact with the yolk-
entoblast at the blastopore, all other cells of the blastoderm are purely
ectoblastic.

Figures 47-55 (Plate 6), 69, 70 (Plate 8), and 114-117 (Plate 12),
show the details of cell arrangement in the thirty-two-cell stage. There
is slight variability in the adjustment of the cells to one another, but
examination of the figures shows that the relative positions of the cells
are the same in all cases. In good transparent preparations I have
seen hundreds of eggs in the thirty-two-cell stage conforming to the
conditions shown in the figures, very few in which the arrangement of

104 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the cells could not have been harmonized with the general plan indi-
cated by the direction of the spindles of the lifth cleavage as represented
in Figures 44-47.

Summary of Fifth Cleavage.

The blastoderm cells of the sixteen-cell stage divide synchronously.
The primary mesoblast {d^'"^) and yolk-entoblast {d^-^) are greatly
delayed in cleavage.

The blastoderm has extended far over the yolk-entoblast.

Regular arrangement of cells of definite origin is as characteristic of
this as of preceding stages.

9. Sixth Cleavage. Sixty-two Cells. Closing of the Blastopore.

The Germ-Layers.

The twenty-eight cells of the blastoderm of the thirty-two-cell stage
are the first ones to undergo the sixth cleavage. Cases are often seen
in which all of the blastoderm cells have spiiidles arranged approxi-
mately perpendicular to those of the preceding cleavage. About the
time that the resulting fifty-six cells pass into the " resting " phase the
two daughter cells of the primary mesoblast (c^^ ■^ c^®"*) are found to be
in division. The two entoblast nuclei (t/'*, d^-^) remain undivided
until a much later stage. The sixth cleavage, therefore, results in the
formation of a sixty-two-cell stage.

A preliminary description of the sixty-two-cell stage resulting from
the sixth cleavage will aid in the discussion of the details of that
cleavage. Figure 5^ (Plate 7) represents an optical sagittal section of
an egg with closed blastopore. All of the twenty-eight blastoderm
cells of the preceding stage have divided. The two yolk-entoblasts
((/''•*, d^'"^) have not divided. The two mesoblast cells (t^**, (/'•*) are in
the sixth cleavage. Two cells {b'''^ and c''^) are represented between
these mesoblasts and the blastoderm in the region of the closed blastopore.
These two cells contribute to the mesoblast of the embryo, and for pur-
poses of description they may be called the "secondary mesoblasts," to
distinguish them from the mesoblasts, fZ*-"and </'•*, which are derived from
the primary mesoblast d^'"^ (Plate 5, Figs. 44, 45), which was separated
from the yolk-entoblast in the fourth cleavage. Referring to Figures 72
and 73 (Plate 8), which represent transverse sections, it will be seen that
there are two pairs of "secondary mesoblasts" (ms'bl'.), an anterior pair,
6'''*and b^-'' (compare Plate 7, Fig. 62), and a posterior pair, a'** and c'-^.
The series of sections represented by Figures 74-77 (Plate 9) shows con-

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 105

clusively that there are, besides the four " secondary mesoblasts," two
eutoblasts and two dividing primary mesoblasts in the egg of this stage.
The cells of the anterior pair of " secondary mesoblavSts " {b''-^, b''-'') are
always hemispherical in form (Fig. 73), while those of the posterior pair
are flattened between the primary mesoblast cells {d^-^, d^-*) and the
blastoderm (Fig. 72). It also appears from the figures that the two
derivatives of the primary mesoblast (c/^'^), the two pairs of " secondary
mesoblasts," and the two entoblasts, are arranged according to a plan
of bilateral symmetry. The division plane in the yolk (Fig. 73) is the
cleavage plane formed between the entoblast cells during the fifth cleav-
age. With this brief description of the sixty-two-cell stage we may now
turn to a more detailed consideration of the sixth cleavage, which formed
the stage.

The large number of small cells and the absence of " landmarks "
makes rapid and certain identification of individual cells of the blasto-
derm on the dorsal surface impossible in the sixty-two-cell and later
stages. By carefully comparing drawings of stages in which the cells
of the blastoderm are in early and late stages of mitosis, it is often
possible to identify all the individual blastoderm cells in the sixty-two-
cell stage. But since it is impossible to follow the blastoderm cells
to their fate in organs of the Nauplius, I have not attempted to give iu
this account the lineage of all cells after the thirty-two-cell stage.
After that stage the most important cells concerned with the gei'm-
layers are near the blastopore. These are followed easily and with
certainty.

During the fourth and fifth cleavages the blastoderm was greatly
extended by the flattening of its cells and by the increase of surface
associated with cell-division. This is repeated during the sixth cleavage,
and the result is that the blastoderm in the majority of cases is com-
pleted, the yolk-eutoblast cells being no longer exposed to the exterior
at the blastopore (see Plate 7, Fig. 56, and Plate 8, Fig. 71).

In most cases a very small opening between the blastoderm cells
represents the remnant of the blastopore. In fact the cells bounding
the blastopore rarely come so closely together in this stage as to com-
pletely obliterate the opening (see Plate 7, Figs. 57, 60, 62 ; Plate 8,
Fig. 71 ; Plate 2, Fig. 76). This persistence of the blastopore has
been of great service in determining the origin of the " secondary mes-
oblasts " and in the orientation of succeeding stages.

Along with the growth of the blastoderm over the blastopore during

106 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the sixth cleavage, the two primary mesoblast cells (d^-^, <£«••*) are
crowded into the yolk beneath the blastoderm, pushing the two eu-
toblast nuclei deeper into the yolk (Plate 7, Fig. 59). The primary
mesoblast cells thus come to lie beneath the blastoderm at the posterior
end of the embryo. As in the two preceding stages, they are easily
identified by their distinguishing features, and furthermore the divisions
of all surrounding cells are accounted for, so that there can be no doubt
of the lineage of the primary mesoblast cells. In series of eggs in
various phases of the sixth cleavage the primary mesoblast cells have
been seen in their successive positions, from that of the thirty-two-cell
stage to that of the sixty-two-cell stage. At a time when some ecto-
blastic cells are undivided and the blastoderm is not completed, the
two primary mesoblast cells are seen filling the blastopore and in part
exposed to the exterior, but as the blastopore becomes closed they sink
into the yolk, and the blastoderm closes over them.

The primary mesoblast cells (d^-^, d^-*), before the sixth cleavage
takes place in them, may be symmetrically placed with reference to
the sagittal plane (Plate 7, Fig. 64 ; Plate 8, Fig. 72 ; Plate 12,
Fig. 120) ; but more often one (d^-^) is found in a position dorsal
or anterior to the other (Figs. 56, 59, 60, 71). In tlie majority of
eggs the two cells appear to have undergone torsion as the blasto-
derm closed around and over them. In the thirty-two-cell statue
they are usually symmetrically placed side by side, but even in
this stage there may be some shifting, as shown in Figure 52 (Plate
6). Figures 62 and 63 (Plate 7) show a very common condition,
in which they have been so turned that the cleavage plane between
them no longer coincides with the sagittal plane. In all such cases
they appear to retain their original positions with reference to the
right and left sides of the embryo. The various positions occupied
by these cells may be the result of shiftings in adjustment to least
resistance at the time when the overgrowing blastoderm crowds them
inwards.

The spindles concerned with the sixth cleavage of the two derivatives
(t/*-^ d^-*) of the primary mesoblast cell are more often about perpen-
dicular to the long axis of the egg (Plate 7, Fig. 56), but sometimes
almost parallel to that axis ; all intermediate conditions are seen. In
Figures 65 (Plate 7) and 121 (Plate 12) the two cells are represented
as having completed the sixth cleavage, so that there exists a stage
with sixty-two cells. Immediately after division the four resulting
cells (d ''•^-^) are rounded, as shown in Figure 65, but soon afterwards

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 107

they become flattened and massed together at the extreme posterior
end of the egg (Fig. 121).

The amount of yolk in the entoblast cells is in some eggs so great
that the blastoderm cannot completely close over the blastopore during
the sixth cleavage. Eggs are sometimes seen in which all the blasto-
derm cells have undergone the sixth cleavage and the two primary
mesoblasts, still in division, are seen lying in the blastopore, and pro-
jecting far into the yolk (Plate 7, Figs. GO, 61). The anterior pair of
" secondary mesoblasts " (h''-^, i^-') are seen in their usual place beneath
the blastoderm immediately in front of the anterior edge of the blasto-
pore ; but the posterior pair {a'-^, c'-^), which originates from cells lying
at the sides of the blastopore, are seen at the surface at the side of the
primary mesoblasts (Fig. 60). As these primary mesoblast cells com-
plete the sixth cleavage they move farther into the yolk. Their posi-
tions with reference to the surrounding blastoderm cells (Fig. 61)
suggests that the change of form during cleavage results in a movement
of the dividing cells into the yolk, in which direction there is, appar-
ently, the least resistance. The posterior pair of " secondary meso-
blasts " (a'-^, c''^) sink below the level of the surface as the blastoderm
closes over the blastopore. In many cases this closing is evidently
brought about by the next (seventh) cleavage of the blastoderm cells.
Certainly the blastopore is always closed and both the ^jrimary and
" secondary mesoblasts " are completely covered by the blastoderm
after the seventh cleavage.

The origin of the two pairs of the " secondary mesoblasts " now re-
mains to be described. Careful study of the cleavage in numerous eggs
gives evidence that these are the result of the sixth cleavage in the four
blastoderm cells, a^-^, b^-^, b^-\ c*-', which form the lateral and anterior
boundaries of the blastopore in the thirty-two-cell stage (Plate 6, Figs.
51, 52). These four blastoderm cells have their spindles for the sixth
cleavage arranged more or less perpendicular to the surface, as shown in
Figures 58 and 59 (Plate 7). The anterior pair of " secondary meso-
blasts " {f-^ b''-'') lies in front of the anterior edge of the blastopore, as
is shown in Figure 57, which represents a section through an egg with
incompletely closed blastopore. This is exactly the position of the cells
^"■8 and ¥■* in the thirty-two-cell stage (Fig. 51). In Figures 58 and 59
(Plate 7) these cells are shown with spindles (sixth cleavage) somewhat
inclined from a perpendicular to the surface. Their relation to the blas-
topore leaves no doubt that they are the cells b^-' and b^* of the thirty-
two-cell stage.

VOL. XL. — 2 4

108 BULLETIN: MUSEUM OF COMrAKATIVE ZOOLOGY.

It has been stated iu the account of the preceding cleavage that the
cell b^-^ does not always touch the anterior edge of the blastopore (see
Plate 6, Figs. 48 and 52), for the reason that tlie cleavage piano between
b^-^ and b^-* may vary in position from perpendicular to the long axis of
the egg to coincidence with the sagittal plane of the embryo. In any
event it seems certain that these two cells always form the anterior pair
of " secondary mesoblasts." In cases like that represented in Figures
48 and 52, the cells become shifted during the sixth cleavage, so that
the plane between them approaches coincidence with the sagittal plane
of the embryo — the common position of these cells in the thirty-two-
cell stage.

The position of the posterior pair of "secondary mesoblasts" with
reference to the anterior pair and also to the blastopore leads to the
unavoidable conclusion that they are cut off from the cells a^"' and c®'^,
Avhich are at the sides of blastopore in the thirty-two-cell stage (Figs. 51,
52). These cells are represented in Figures 58 and 59 (Plate 7) as
dividing. From their position later, I infer that as division progresses
the extension of the blastoderm causes these cells to approach the median
plane, where the}"^ meet and complete the closing of the blastopore. At
the same time the primary mesoblasts d^'^, cP-^ are overgrown by the
blastoderm, and the cells a^-^ and c^'^ complete their division into the
outer cells (a^•^ c'-^), which remain in the blastoderm, and the inner cells
{d''^, c'-^), which constitute the posterior pair of " secondary mesoblasts,"
lie between tlie blastoderm and the primary mesoblasts (see Plate 7,
Fig. 62 ; Plate 8, Fig. 72).

Cases like those illustrated by Figures GO and Gl (Plate 7) give addi-
tional evidence in support of the above interpretation of the origin of tlio
" secondary mesoblasts." In the egg represented in Figure GO a rem-
nant of the blastopore is present and at its anterior edge are the two
blastoderm cells V-^, W-^. Immediately beneath them are the derivatives
b''-^ and b''"'', the anterior pair of " secondary mesoblasts." In the egg
represented in Figure 71 (Plate 8) the primary mesoblasts (c?"-', d'^-*)
have sunk beneath the blastoderm. The same relations exist between
blastopore and anterior " secondary mesoblasts." Similarly iu Figure
62 the posterior "secondary mesoblasts" lie beneath the cells a'-^ and
c'-*, which bound the sides of the blastopore. These cells are contigu-
ous to V-^ and 5''•^ The same relations hold in Figure 60 and in Figures
58 and 59 (Plate 7), which represent the divisions forming the " secon-
dary mesoblasts." Comparison of the arrangement of the cells around
the blastopore iu the thirty-two-cell stage (Plate 6, Figs. 51, 52) with

BIGELOW: EARLY DEVELOPMENT OF LEPAS.

109

the cell arrangement and spindles as shown in Figure 58, 60 and 62
gives evidence entirely in favor of the explanation given of the cell-
lineage of the "secondary mesoblasts." T^ey are certainly derived
from the ectoblastic cells of the blastoderm, and the evidence com-
pletely supports the interpretation that they are derived directly from
the cells bounding the blastopore laterally and anteriorly in the thirty-
two-cell stage.

The cell-lineage of the " secondary mesoblasts " is, then, as shown in
the following table (see also complete table of the cell-lineage on page
135).

16:

b^-i:

^6■S

,6-3;

i(7-5 right anterior "secondary mesoblast" call.
b~-^ blastoderm cell (ectoblast).

b''-'' left anterior " secondary mesoblast " cell.
6^-8 blastoderm cell (ectoblast).

a^-5 left posterior " secondary mesoblast " cell.
a'-^ blastoderm cell (ectoblast).

c~i-5 right posterior " secondary mesoblast " cell.
c''S blastoderm cell (ectoblast).

It will be noticed that " secondary mesoblast " originates from the
quadrants a, b, and c. One cell each is contributed by a and c but two
cells come from b. Tracing the lineage to the three micromeres which
are separated from the yolk-macromere in the first three cleavages, it is
found that only the first (ab^) and the second (c') contain " secondary
mesoblast " ; the third (d*-^) is purely ectoblastic.

After the sixty-two-cell stage the derivatives of the " secondary meso-
blasts" have not been distinguished from those cells which were derived
from the primary mesoblast. The cells of the two origins become
mingled together and there appear to be in Lepas no distinguishing
characteristics. Hereafter the term mesoblast (nis'bl. in the figures) will
be used in the description as including the mesoblast cells of the two

origms.

The entoblast nuclei (d^-\ ^'^) are always near the primary mesoblast
cells, but, as shown in the figures, they occupy no constant position in
relation to particular cells. They stain more intensely than the nuclei

110 bulletin: museum of COMrARATIVE ZOOLOGY.

of the mesoblast cells, and iu good transparent preparations of the entire
egg are easily recognizable. The cleavage plane separating the yolk-
entoblast cells may occupy various positions at this stage. If the pri-
mary mesoblasts are symmetrically placed with reference to the median
plane (Plate 7, Fig. Gl), the cleavage plane in the yolk coincides ap-
proximately with the sagittal plane of the embryo ; but when one of the
primary mesoblasts is anterior or dorsal to its sister cell, the plane of
separation between the entoblasts is inclined towards the horizontal, or,
if vertical, is oblique to the long axis, as in Figure 63. In all cases it
appears to extend from near the plane separating the right and left pri-
mary mesoblasts towards the antero-dorsal side of the embryo (Figs. 63,
64, 65, 73). This relation suggests tliat the horizontal and oblique
positions are secondary and due to movement of the yolk when the pri-
mary mesoblast cells are forced beneath the blastoderm and adjusted to
unsymmetrical positions. The fact that when the primary mesoblasts
retain their original symmetrical relation, the cleavage plane in the yolk
is found apparently coinciding with the sagittal plane, lends support to
this view.

It may be of interest to notice that the cleavages involved in the seg-
regation of the germ-layers are always the same, no matter whether the
blastoderm is completed in the sixth or seventh cleavages. The cleav-
ages separating from the yolk-cell the micromeres which form the blas-
toderm are not variable in number, but definite (three) ; and there is no
variation in regard to the number of micromeres which produce the
variable numbers of blastoderm cells required to cover the yolk. This
conclusion is opposed to that of Groom ('94, p. 141). (See review of
literature on late cleavage.) This relation is exactly what has been found
in the case of the eggs of gasteropods and annelids, in which it has been
shown (Conklin, '97, pp. 61-63) that the number of micromeres (ecto-
blasts) separated from the macromeres (mes-entoblast) is constant for all
species which have been studied, although the macromeres in some cases
are very large and require a large number of ectoblastic cells to complete
the blastoderm ; in such cases — precisely as in Lepas anatifera and L.
fascicularis — there is more subdivision of the micromeres before the
blastoderm is completed. It appears that the same relation exists in
the case of the other species of Lepas.

Summary of Sixth Cleavage.

All derivatives of the three micromeres (aP, c^ and d*-'^ and of the
two primary mesoblasts {d'^'^, d^*) undergo division. The two entoblast

BIGELOW : EARLY DEVELOPMENT OF LEPAS. Ill

cells remain undivided. The '' resting " stage following the sixth cleav-
age normally consists of sixty-two cells.

By the extension of the blastoderm during the sixth cleavage the blas-
topore is usually closed. As to the method of closing the blastopore,
this account completely disagrees with Groom ('94 ; see also review of
literature on the closing of the blastopore).

During this cleavage the two primary mesoblasts sink beneath the
blastoderm as it closes over the blastopore.

Four blastoderm cells, derived from cells a^, b^ and c' (the first and
the second micromeres, aW' and c^), are divided parallel with the surface,
thus cutting off four cells which lie in the yolk beneath the blastoderm.
These are designated " secondary mesoblasts."

The mesoblast is, then, derived from each of the four quadrants of the
four-cell stage. In the cells a^, b^ and c^ there is mesoblast in connec-
tion with ectoblast (ectoblastic mesoblast), whereas in the d quadrant
the mesoblast arises directly from entoblast, and may be designated
entoblastic mesoblast. The origin of the mesoblast in Cirripedia has not
heretofore been traced accurately (see review of the literature on the
germ-layers).

All cells sharing in the formation of the lip of the blastopore in the
thirty-two-cell stage, as represented in Figure 51, contribute to the
mesoblast.

The blastoderm is composed of derivatives of three, and only three,
micromeres (ab'\ <?, rf^*^), even when the size of the yolk-mass does not
permit of the blastopore being closed until the following cleavage.

10. Seventh Cleavage. The Mesoblast.

The sixty-two-cell stage has been described as embracing fifty-two
ectoblastic cells composing the blastoderm, which has usually grown
over the blastopore ; eight mesoblast cells, of which four have been
designated as " secondary " ; and two entoblast cells, resulting from the
division of the yolk-macromere. All these, excepting the two entoblast
cells, divide more or less synchronously and form a stage which may bo
estimated to consist of about one hundred and twenty-two cells. The
planes of cleavage appear in most cases to be perpendicular to those of
the sixth cleavage. For convenience in description this may be desig-
nated the seventh cleavage.

Figures 78-80 (Plate 9) represent a series of parasagittal sections
through an egg of the 122-cell stage, but some of the cells have not
completed the seventh cleavage. Figures 81-86 represent a series of

112 bulletin: museum of comparative zoology.

transverse sections of the same stage, of which 81 is the most posterior.
In the blastoderm at this stage there is nothing wortl)y of note except
the indentation which marks the former position of the blastopore. The
cells in this region are rarely as closely arranged as in the other parts of
the blastoderm.

The mesoblast cells are crowded together, and it is impossible to dis-
tinguish in all cases between those derived from the primary mesoblast
and those from the " secondary mesoblast." As used in the description
of later stages, the term mesoblast includes both the primary and
" secondary mesoblast."

The possibility of origin of mesoblast cells from the blastoderm after
the sixth cleavage has been kept in mind during the observations, but
there is no evidence of such an origin. The cleavage spindles in all parts
of the embryo have been seen, but not one perpendicular to the surface
has been detected. Moreover, the mesoblast cells have been repeatedly
counted in sections and their nuclei have also been counted in transpa-
rent preparations of the entire egg, and there have never been seen more
cells than could be accounted for by the division of the eight mesoblast
cells described in the sixty-two-cell stage.

It should be mentioned that by rapid decolorization of specimens
stained in borax carmine it has often been found possible to draw the
color from the nuclei of the blastoderm cells and stop the reaction while
the mesoblast nuclei were still brilliantly stained. With such prepara-
tions it is easy to count the nuclei of the mesoblast cells in the entire
egg. This method has been employed in all the stages with mesoblast.

The entoblast nuclei are stained brightly by this carmine method, and
are easily identified in transparent preparations of entire eggs, as well as
in sections. In all stages between that of thirty-two cells and that with
about one hundred and twenty cells there is no evidence of division of
these nuclei. In these stages only two " resting " nuclei are to be found
in the yolk, as shown in Figures 78-80 and 81-86 (Plate 9). Usually
in the 120-cell stage the two nuclei are enlarged, while the chromosomes
are distinct. Evidently the nuclei are preparing for division, but the
spindles are rarely seen until after the blastoderm cells have divided
again. In the resulting stage, with about two hundred and fifty cells,
four entoblast nuclei are often seen. It does not seem possible that
there can have been an overlooked division of these nuclei. Moreover,
the origin of the mesoblast cells has been determined to be independent
of the two entoblast cells, which are seen in this and in the preceding
stage.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 113

Summary of the Seventh Cleavage.

All cells, except the two entoblasts, divide.

Derivatives of the two kinds of mesoblast have not been distinguished
after the cells are crowded together at the posterior end.

There is no evidence that mesoblast originates otherwise than as de-
scribed in the preceding account of the sixth cleavage. The entoblast
nuclei have been traced from the sixteen-cell stage and there has been
but one division. Hence, contrary to the assumption of earlier investi-
gators, the entoblast nuclei cannot contribute to the mesoblast (see the
following review of the literature).

11. Review of Literature on Late Stages of Cleavage, ox Clos-
ing OF the Blastopore, and on Differentiation op the Germ-
Layers.

a. Late Cleavage. — Groom ('94) did not follow the later cleavages in
detail, because his results showed so great variation in the early stages.
He describes the later growth of the blastoderm over the yolk as " tak-
ing place in precisely the same manner as iu the earlier stages, i. e., by
the emergence of merocytes from the yolk and the division of blasto-
derm cells. . . . The variation is so great tliat the process may be said
to be irregular. ... I am unable to say how many merocytes take part
in the formation of the blastoderm ; but in all probability the number
is variable, V)ut not large. As the ovum is often half covered when
four or five have emerged, some such number as nine or ten may not be
far from the mark" (Groom, '94, pp. 140, 141).

The supposed variation in early stages of cleavage has already been
discussed in the reviews of the literature on those stages. The later
cleavage and growth of the blastoderm have been shown in this paper
to be very regular, and the variations upon which Groom has placed
much stress are comparatively rare. These variations can usually be
ascribed with strong probability to unfavorable conditions in the en-
vironment of the developing egg. The number of " protoplasmic " cells
(micromeres) formed from the yolk-cell has been shown to be not varia-
ble (nine or ten), as Groom supposed, but constant, viz. four, of wliich
the first three — containing all the ectoblast and " secondary mesoblast "
— are separated from the yolk by the first three cleavages, while the
fourth cleavage differentiates the primary mesoblast from the yolk-en to-
blast. Groom's statement (p. 198) that epiblastic cells continue to be

114 bulletin: museum of comparative zoology.

formed at tho expense of the yolk-cell until the hlastopore closes, is
completely disproved by the facts of cell-lineage.

h. Closing of Blastopore. — Groom did not see the closing of the blas-
topore in L. anatifera, but he ('94, p. 141) described it for other species
as follows : " The end of the yolk projects out at one point as a small
rounded elevation. ... A merocyte appears in the centre of this, and
fills the gap between the surrounding cells, and finally emerges from
the yolk as the blastomere."

This description is far from being in harmony with the fiicts in the
case of L. anatifera. The closing of the blastopore has been shown in
this paper to be due to the repeated divisions of the ectoblastic deriva-
tives of the three micromeres {a¥, c^, d*'"^) which are separated from
the yolk-macromere in the first three cleavages. The " merocyte "
which Groom saw in the blastopore (see his Fig. 127) is represented by
the protoplasmic mass concentrated around the nucleus of the entoblast
cell, which is situated as shown in my Figure 54 (Plate 6). I have
shown by tracing the cell-lineage that this cell divides (Fig. 52, fifth
cleavage), usually before the closing of the blastopore, sometimes during
the sixth cleavage of the ectoblastic cells, and that the resulting cnto-
Ijlast nuclei are later found deeper in the yolk. Nussbauiu observed in
Pollicipes a division of the yolk before the blastopore closed. Groom
('94, p. 147) states that this may rarely occur, a condition which is
completely at variance with his account of the closing of the blastopore.

The evidence presented in the present account of the cell-lineage leads
to the conclusion that no cell is cut off directly from the yolk to fill tho
blastopore. It has been shown that at the time of closing there are two
nuclei in the yolk, not as Groom stated, a single one. Hence Groom's
conclusion, that tho " merocyte " which fills that blastopore " before be-
coming shut off as a blastomere, gives off a single nucleus into the yolk "
('94, p. 198), cannot be accepted. The evidence is completely opposed
to such a view. It appears that in Groom's account of the closing of
the blastopore, his view of " emerging merocytes " has led, as in the
early stages, to an erroneous interpretation.

c. Differentiation of the Germ-Layers. — Groom's account of the
" meso-hypoblast " agrees in general with" the descriptions of all the
earlier authors, who regarded this as represented by the yolk-cell, or
cells, after the closing of the blastopore. Groom ('94, p. 146) writes:
"The closing of the blastopore is almost immediately followed by the
division of the yolk into two pyramids or segments ; the formation of
the mesoblast immediately commences by the successive cutting off and

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 115

sub-division of nucleated segments from the two yolk segments." Ac-
cording to Groom these yolk-segments after separation of the mesoblast
divide and form endoderm cells.

In opposition to this it has been shown in the present paper that the
mesoblast clearly does not originate directly from the yolk-cells after the
closing of the blastopore ; but from certain cells which have been desig-
nated in this account as primary and secondary mesoblasts. The origin
of all these cells has been definitely traced. Moreover, evidence has
been presented to show that the two yolk-entoblasts do not begin to
divide after the thirty-two-cell stage until at least one hundred and
twenty cells are present, of which more than a dozen are mesoblastic.
Since the entoblast cells do not divide during these stages, they cannot
be the direct pi'ogenitors of any of the mesoblast cells. All the evi-
dence given seems conclusive and opposed to Groom's interpretation.

The figures of Groom fail to establish his conclusions regarding the
origin of mesoblasts from yolk-entoblasts, for in no case are nuclear spin-
dles, the only unimpeachable evidence of such origin, shown. His inter-
pretation of the origin of mesoblast cells seems to be based upon their
position. In numerous preparations I have seen all the conditions
which Groom figures, but I have found no evidence opposed to my in-
terpretation of the origin of the mesoblast. Groom did not have trans-
parent preparations of entire eggs, and his account of the mesoblast is
based entirely upon sections. His figures represent isolated sections,
when in many cases only complete series of sections would be convinc-
ing. His eiToneous conclusion, that the mesoblast is cut off in a series
of divisions occurring in a pair of yolk-cells (" meso-hypoblast "), may
have resulted from certain conditions which I have frequently noted.
Sometimes in stained sections the cell-boundaries of the mesoblast cells
are invisible, they appearing to be continuous with the yolk. Under such
conditions the mitotic spindles of the mesoblast cells might easily be mis-
taken for division of the yolk-cells to form new mesoblast cells, I have
seen many such cases which exactly simulated some of Groom's figures,
but after removal of the cover glass and restaining, the cell-boundaries
of the mesoblast cells and the nuclei of the yolk-entoblasts appeared as
usual.

Nussbaum ('90) described the mesoblast in Pollicipes as formed by the
division of blastoderm cells surrounding the blastopore before it closes.
The mesoblast was said to grow inwards and anteriorly over the yolk.
The account of the origin of mesoblast given in the present paper makes it
probable that Nussbaum's description is in a general way correct. Had

116 bulletin: museum of comparative zoology.

not the details of the cell-lineage been traced in Lepas, I should be led
to describe in similar general terms the origin of the mesoblast. I infer
from Nussbaum's description that in Pollicipes the blastopore does not
become closed as early as in Lepas. It seems probable that in Polli-
cipes the primary and secondary mesoblast cells may undergo some
divisions before they are forced beneath the overgrowing blastoderm.
Such a process would have the appearance of the production of meso-
blast from the blastoderm cells at the edge of the blastopore.

In stages preceding gastrulation Nussbaura saw two large cells at the
posterior pole, but he lacked material for following out their history. It
seems probable that he saw the two primary mesoblasts which I have
seen in the thirty-two-cell stage of Lepas.

12. Determinate Cleavage.

The small size and large number of cells make it impossible to de-
termine the lineage of the individual cells of the embryo beyond the
sixty -two-cell stage, and they cannot therefore be traced directly to
particular organs of the Nauplius. However, the great regularity and
constancy of preceding stages renders it extremely probable that the
cells are destined for definite organs. Cells of definite origin have been
traced to definite positions in the later cleavage stages. Careful ob-
servation has given no evidence of changes in position of cells taking
place after the completed segregation of the germ-layers. Indeed the
beginning of irregularity is scarcely to be expected in such late and well
differentiated stages of development. The regions of the embryo from
which particular organs arise have been definitely traced to groups of
cells of known lineage. There seems to be no reasonable doubt that the
cells of the late cleavage stages are destined to enter into the formation
of particular organs. The cleavage of Lepas is, then, an example of
what Conklin ('98) has termed "determinate cleavage."

The conclusions in the preceding paragraph on " determinate cleav-
age " are widely at variance with those of all previous writers on cirri-
pede development. The early development of the ova of cirripedes has
always been regarded as irregular and indeterminate. Great variations
have been said to occur.

Groom ('94, p. 199) summarizes his study of the cleavage of various
Cirripedia as follows : — " In describing the details of division of the
cells of the l)lastoderm and yolk-endoderm much variation has been
shown to occur, so much indeed that the process may be termed irregu-

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 117

lar. Such differences show well the morphological insignincance of the
details of cell division in the present case, for the Nauplii vary pro-
portionately much less ; every one of the numerous, simple, or com-
pound bristles or spines of the Nauplius has its definite character and
position, which are maintained with surprising constancy throughout,
although they must have been produced by epiblast cells having very
different modes of origin and arrangement."

In the preceding account of the various stages of cleavage this sup-
posed great variation in development has been discussed. It has been
shown that the development is extremely regular, and that there is nut
the slightest foundation for views such as those above quoted.

In a preliminary paper on L. fascicularis (Bigelow, '96) the results
were summarized as follows : — "In all important respects the cleavage
of L. fascicularis is as regular as is ordinarily found in other Metazoa.
All previous observers have failed to recognize any definite order in the
cleavage of cirripede ova. It has always been described as exceedingly
variable, irregular and sui generis. There is undoubtedly some irregu-
larity and variation in the cleavage of the ova of those cirripedes where
a great amount of yolk is present. However, as will be pointed out in
a future paper, the cleavage of these forms, when interpreted by tlio .
cleavage of L. fascicularis, is seen to follow a much more regular order
than has been supposed."

Later studies have completely supported this interpretation, and even
the irregularity of development which I formerly believed to exist in
the case of those cirripedes whose ova have much yolk, appears not to
exist in the course of normal development. More extended study has
shown that L. anatifera, one of the forms which I at first interpreted as
somewhat variable in its development, is extremely regular. Studies
now in progress on other genera support the conclusion which I have
drawn from L. fascicularis and L. anatifera, namely, that the evidence
derived from a study of cell-lineage indicates that the development
of Lepas is as regular as the well known cases among gasteropods and
annelids.

13. Notes on Cleavage and Germ-Layers in L. Fascicularis.

The early development of Lepas fascicularis is so closely like tliat
already described in the case of L. anatifera that extensive special
description is unnecessary, but some remarks are needed in order to
correct and supplement a preliminary note on this species which I
published in 189G.

118 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Figures 95-121 (Plates 11, 12) show how close is the resemblance
to the cleavage of L. aiiatifera. Except in size and some unimportant
details, the various stages of the two species are indistinguishable, and
the description of the figures of L. anatifcra may be applied to those of
L. fascicularis.

A renewed study of the few old preparations, supplemented by many
new ones, shows that I ('96) was wrong in the conclusion that the
ectoblast is detached from the yolk-macromere by means of four succes-
sive divisions ('96, cctomeres A, B, C, and D). The supposed fourth ecto-
mere ('96, Figs. G and 7 D) is the primary mesoblast cell. In origin and
position it corresponds exactly with the mesoblast cell (d^-^) seen in the
sixteen-cell stage of L. anatifera. I now interpret the spindle seen in
the yolk during the fifth cleavage ('96, Fig. 7), which was then supposed
to represent the separation of the mesoblast and the entoblast, as a
rare case of precocious division of the entoblast. Study of the complete
series, with all mitotic phases represented, shows that in L. fascicularis,
as in L. anatifera, the first, second, and third cleavages form micro-
meres containing the ectoblast and " secondary mesoblast," while the
fourth cleavage separates mesoblast and entoblast from each other.

With regard to the planes of cleavage and orientation, I find no
important disagreement with L. anatifera. The descriptions of the
first and second cleavages in the preliminary note were similar to those
of L. anatifera given in this paper. The rotation during the first
cleavage was not then known. The equatorial nature of the third
cleavage was not clearly shown by the figure of a four-cell stage with
inclined spindles in the preliminary note ; Figures 100-103 (Plate
11) in this pai)cr better represent the four-cell stage and the third
cleavage. The figure of the eight-cell stage ('96, Fig. 6) was drawn
from an egg which is now .known to have been incorrectly oriented.
Eggs which give exactly such camera tracings will, when properly
oriented by moving the cover glass, always show the same arrangement
of cells as that seen in Figures 104-lOG in this paper.

Figure 6 of the preliminary paper represented a separation of
mesoblast and entoblast (fourth cleavage), and not as was incorrectly
assumed, the formation of a "fourth ectomere." Figures' 108-110 are
the corresponding figures in this paper.

Tlie primary mesoblast cell, shown in Figure 8 of the preliminary
paper as filling the blastopore, represented the delayed fifth cleavage,
which was in progress. The single entoblast nucleus was not yet under-
going the fifth cleavage. The inferred connection between the spindle

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 119

in the yolk-cell, in the sixteen-cell stage, and the separation of a
mesoblast cell is now known to have been an erroneous interpretation.
The series of stages is now so complete as to leave no doubt that the
mesoblast cell is separated from the yolk-entoblast in the fourth and
not in the fifth cleavage.

In the sixty-two-cell stage the origin and position of cells is certainly
the same as in L. anatifera. The " secondary mesoblasts " were observed
and figured during my earlier studies, but were interpreted as deriva-
tives of the primary mesoblast, which seemed to divide more rapidly
than did the other cells. It now appears from a study of all phases of
the sixth cleavage that there are eight mesoblast cells in the sixty-two-
cell stage, only four of which are derived directly from the ectoblast.
Up to this stage the divisions of the primary mesoblast are the same
as have been described in detail in the case of L. anatifera. In living
eggs recently studied, and also in preparations of favorably preserved
material, I have observed the cell-wall between the two entoblast
nuclei of this stage, and it follows that — contrary to my former sup-
position — there is no exception to the rule that every nuclear division
during the cleavage is associated with total cell division.

VIII. Extension of the Mesoblast and Entoblast. Later
Development of the Germ-Layers.

The mesoblast in the 122-cell stage consists of a mass of cells at the
posterior end of the embryo, near the former position of the blastopore
(Plate 9, Figs. 78-86). The arrangement of the cells leaves no doubt
about the position of the blastopore, but orientation of the succeeding
stage is more difficult and uncertain. During the next division the
embryo begins to elongate posteriorly. A comparison of the blastoderm
cells on the ventral surface of the 122-cell and 2.50-cell (estimated num-
bers) stages leads to the suggestion that the elongation is due to flat-
tening of the ventral blastoderm cells, while those on the dorsal surface
remain columnar in form. At any rate, this elongation appears to be
confined mostly to the ventral region of the blastoderm, anterior to the
former position of the blastopore. The result is that the cells which
closed the blastopore and the adjoining mesoblast cells are moved from
the ventral surface towards the extreme posterior end, whei'e for a time
the mesoblast consists of a conical mass of cells (compare Plate 9, Fig.
80 with Plate 10, Fig. 87). The rapid division of the mesoblast cells
produces a plate, which grows forward on the dorsal side of the embryo

120 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

(Fig. 87). That this plate of mesoblast is on the side of the embryo
opposite that on wliich the blastopore was situated, is supported to
some extent by the facts above mentioned concerning the posterior
growth of the blastoderm. Further evidence of this is found in the
columnar shape of the cells, which is characteristic for those on the
dorsal side ; moreover many embryos long retain a slight depression
marking the place of the blastopore, and the blastoderm (ectoblast)
cells in this region are often delayed in division in late stages, as well
as in the earlier stages, as may be seen when the position of the blasto-
])ore is definitely known. It should also be mentioned that the second
])olar cell, which lies dorsally (animal pole) in the yolk at the anterior
end, is often visible near the anterior extension of the mesoblast both
in sections and in transparent preparations of entire embryos corre-
sponding to Figures 87 and 88 (Plate 10), These facts all seem to
favor the conclusion that the forward growing band of mesoblast (Figs.
87, 88) is on the side opposite that occupied by the blastopore in
earlier stages, and consequently opposite that on which the mesoblast ex-
tends farthest forward at the time of the closing of the blastopore (Plate
8, Fig. 71 ; Plate 9, Fig. 80).

Examination of Figures 88, 89 and 90 (Plate 10), representing long-
itudinal and transverse sections, will give some idea of the direction
and extent of growth in the mesoblast. A solid, conical mass of cells
lies at the extreme posterior end and extends anteriorly as a broad band
on the dorsal side (Fig. 88) ; this grows laterally towards the ventral
side (Fig. 90). The mesoblast at first consists of a single layer of cells,
which divide rapidly; the layer becomes many cells in thickness on
the dorsal side, but gradually thinner towards the ventral edges of the
band (Figs. 90, 92). At the same time that the extension of the meso-
blast has been in progress, the entoblast cells have been dividing.
Their cell-boundaries are often well defined, and the nuclei do not
migrate far from the positions where they are formed by division (Figs.
91, 92).

The blastoderm has remained a single cell in thickness, as shown in
the Figures 87-94.

As shown in the preceding chapter. Groom's (94) view of the origin
of the mesoblast is erroneous, but the accolint which I have given of
the extension of the mesoblast is, in essentials, entirely confirmatory of
Groom's description of the same process. Groom has given many good
figures of entire eggs, showing the appearance of the entoblast yolk-cells
in living eggs of Lepas and Balanus. All my observations on these

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 121

stages agree essentially with his account. His figures showing the
extension of the mesoblast closely correspond with those which 1 have
given and described, not with an idea of contributing new facts, but in
order to connect these stages with my account of the early development.
Groom interpreted the anterior growth of the mesoblast as taking
place on the dorsal side, and I shall later give confirmation of this
opinion, which rests on an orientation that I have used thus far without
adequate proof.

IX. Formation of the Appendages of the Nauplius, and De-
velopment of the Organs-

With regard to these phases of the development, my observations are
quite in harmony with the account by Groom ('94, pp. 151-154). A
few figures have been placed in this paper in order to show relations to
the early stages, but since there is such close agreement with Groom, it
is unnecessary to give a detailed description and numerous figures.

Groom's important observation, that the appendages first appear on
the side which has the band of mesoblast, and that this is dorsal, is
supported by my Figures 91-94 (Plate 10) and 122-126 (Plate 12). ■
All earlier writers on cirripede development had considei'ed the mesoblast
band as ventral (see i-eview of literature in Groom's paper).

Figures 91 and 122 represent the first indication of the segmentation
of the embryo. Two transverse fui-rows {1, 2) appear on the dorsal side,
and extend around towards, but do not reach, the ventral surface. The
limit of extension of the transverse furrows corresponds closely with that
of the underlying mesoblast. The body is divided by the two furrows
into three regions, corresponding to the three segments of the Nauplius.

Soon after the appearance of the transverse furrows there appears a
median longitudinal furrow on the same side (dorsal) of the embryo.
This is shown in transvei'se section in Figure 92 and in dorsal view in
Figure 125. This furrow intersects the two transverse furrows, but
does not extend to the extreme end of the embryo. Two new transverse
furrows now appear (3, 4, Figs. 93, 123-125), superficially dividing the
antei'ior and posterior segments of the Nauplius. Earlier writers have
published many drawings of these stages, and it seems unnecessary to
insert similar ones in this paper.

The transverse furrows and the median longitudinal one deepen
rapidly, and cut off the three pairs of appendages, as has been correctly
described by Groom and earlier workers. The extension of the floor of

122 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the longitudinal furrow laterally and ventrally is shown in Figure 94,
which also shows the cctoblast and mcsoblast composing the appendages.
The deepening of tlic furrows progresses and the appendages are folded
off commencing at their dorsal distal ends until linally their attachment
is to the ventral side of the embx'yo, as determined by the position of the
mouth and labrum (Figs. 124, 126). It will be seen that my account
contirras Groom in that the mesoblast band and the furrows are dorsal,
and that the appendages are fo*lded off from dorsal to ventral, the free
ends of the appendages remaining directed dorsally until about the time
of hatching. Investigators before Groom gave good descriptions and
figures of the formation of apj)endages, but considered that the meso-
blastic band and the furrows were ventral instead of dorsal.

Many of my preparations and unpublished figures of later stages con-
firm Groom's account regarding the formation of the stomodseum and
proctodseum, and the development of the mesenteron from the yolk-
entoblast cells.

It is to be noted that many of Groom's minor observations on later
stages were confirmatory of earlier writers, whose work he has reviewed,
and it has, therefore, for my purposes been sufficient to refer directly to
Groom's paper. For the details of late development of organs of the
Nauplius, reference must be made to Groom and earlier workers, for this
paper is concerned, primarily, with cleavage and germ-layer formation.

The fate of the germ-layers, which were identified in the sixty-two-cell
stage, may be sunmaarized as follows : — The ectoblast forms the outer
covering of the body and appendages, the stomodasum, proctodicum, and
the nervous system. The yolk-entoblast forms the mesenteron. The
mesoblast forms the muscles and connective tissues of the appendages,
and of the body of the Nauplius.

So far it has not been possible to distinguish between the fate of the
primary and secondary mesoblasts. It can only be stated that at least
a part of the muscular and mesenchymatous tissues of the Nauplius come
from the ecto-mesoblast ("secondary mesoblast "). In other genera of
Cirripedia an attempt is now being made at tracing the two kinds of
mesoblast farther than has been possible in Lepas.

X. General Considerations on Cleavage and Cell-Lineage.

Korschelt vind Heider ('90-91) have classed the cleavage of the cirri-
pede ovum with their typfe II of crustacean cleavage — a type beginning
with total cleavage, but soon changing to superficial. This classification

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 123

was evidently based upon Nassonow's figures of Balanus ; but is shown
to be erroneous by subsequent investigations. It is controverted in the
case of Balanus, by the account of Groom, as well as by unpublished
observations of my own ; and in the case of Lepas it is clearly inappli-
cable. In both these genera cleavage is total and unequal.

Knipowitsch ('92) described the cleavage of the Ascothoracidan genus
Laura as superficial from the very beginning of development. His figures
do not warrant such a conclusion, for cell -boundaries appear to form after
every nuclear division. The few figures of segmentating eggs in Knipo-
witsch's paper resemble the figures which other authors have drawn from
the eggs of parasitic copepods ; for example, Pedaschenko's ('93) figures
of Lernsea. The latter is evidently a case of total, but very unequal,
cleavage, and the cleavage of Laura is apparently to be interpreted in
the same way.

Van Beneden's ('70) figures illusti-ating his account of the develop-
ment of Sacculina indicate to my mind that the cleavage of Ehizoce-
phalan Cirripedia is also of the unequal total type. Even the fact that
in late stages the four yolk-macromeres appear to fuse does not support
the interpretation that the cleavage is in later stages superficial. In no
stage of the development is there nuclear division which is not associated
with total cell division, and we are led to the conclusion that the cleavage
of Sacculina cannot be correctly characterized as superficial in any stage.

Eegarding the type of cleavage of cirripede ova, the conclusion is that,
so far as present knowledge extends, the eggs undergo unequal total
cleavage, and with respect to the cleavage processes there is no close
resemblance to the superficial cleavage of the higher' Crustacea ; rather
is the resemblance to that of the yolk-laden eggs of gasteropods.

In the order of the cleavages involved in the establishment of the
germ-layers there are in Lepas some interesting resemblances to the
annelids and mollusks. As is well known, studies of the cell-lineage of
annelids, gasteropods, lamellibranchs, and chitons have shown that in
all of these forms the ectoblast is separated from the raes-entoblast by
three successive cleavages, while a fourth cleavage separates the primary
mesoblast from the entoblast. Moreover, it has been shown in the cases
of some gasteropods and lamellibranch mollusks, that the mesoblast is
derived from both primary germ-layers; in addition to the primary
mesoblast (entoblastic mesoblast) there are mesoblast cells which come
from the ectoblast (ectohlastic mesoblast). This has been designated
" secondary mesoblast " or " larval mesenchyme " (Lillie, '95, p. 24 ;
Conklin, '97, p. 150).

VOL. XL. — 2 5

124 BULLETIN : MUSEUM OF COMPArvATIVE ZOOLOGY.

So far it has not been shown conchisively that the mesoblast of anne-
lids has a hke double origin, but the studies of Wilson ('98) make it
appear probable that in tlie annelid egg there is mesoblast of ectoblastic
origin, which is comparable to the " secondary mesoblast " or " larval
mesenchyme " of mollusks.^

It must be understood that, in offering the following suggestions of
some resemblances between the cleavage of Lepas and the forms above
mentioned, it is not here claimed that any cell homologies exist. Our
knowledge of this subject is not as yet sufficiently extensive to warrant
any decision for or against such a conclusion.

The fact that in Lepas the ectoblast is separated from the mes-ento-
blast by three successive cleavages, while the fourth separates the pri-
mary mesoblast from the entoblast is, at least, an interesting coincidence.
The double origin of mesoblast is another point of resemblance, for in
Lepas, as in gasteropods, lamellibranchs and probably annelids also, the
ectoblast is a second source of mesoblastic cells.

In one important respect there seems to be a wide difference between
the cleavage of Lepas and that of annelids and mollusks ; for in these
latter groups there are three quartets of ectoblastic micromeres formed
by as many successive cleavages of four macromeres, whereas in Lepas
there are not three quartets of cells but three cells formed in the same
order of cleavage. In the annelids and mollusks the first segregation of
ectoblast from entoblast is represented by the upper four cells (first
quartet of micromeres) of the eight-cell stage, formed by the third cleav-
age, whereas in Lepas the first segregated ectoblast is one of the two
cells formed by the first cleavage. Stated in other terms, in annelids
and mollusks, unlike Lepas, the first and second cleavages are not
directly concerned with the segregation of ectoblast from entoblast, but
they divide the egg into a quartet of macromeres, each containing ento-
blast, from which in succession three quartets of ectoblastic micromeres
are separated. In Lepas the segregation of ectoblast begins, as it were
precociously, without the previous division of the entoblast into a quar-
tet of cells. As a result of this there is in Lepas one entoblastic macro-
mere instead of four, as in annelids and mollusks, and single micromeres
appear to represent quartets. So far as the order of cleavage involved
in the segregation of the primary germ-layers is concerned, the first
micromere {air) of Lepas apparently corresponds to the first quartet of

^ Since this paragraph was written, several investigators have given support to
the suggestion that tliere is a double origin of the mesoblast in annelids. See
Treadwell (: 01, p. 427), Wilson (:01, p. 801) and Torrey (: 02, p. 576).

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 125

ectoblastic micromeres seen in the eight-cell stage of such eggs as have
four macromeres resulting from the quartet-forming (first and second)
cleavages. The micromeres of Lepas are, then, according to this view,
to be regarded as equivalent to quartets of micromeres, while the single
yolk-macromere equals a quartet of macromeres. It must be recognized
that there are great, perhaps irreconcilable, differences between the de-
velopment of the cirripedes and that of annelids and mollusks, and that
consequently, the above comparisons might be extreme, if they were to
be used as evidence of the existence of cell-homologies. At present it is
possible simply to compare the order of cleavages involved in segregating
the germ-layers.

A similar relation in cleavage occurs within the group of the Cirripe-
dia. Van Beneden ('70) showed that in the Rhizocephalan genus Saccu-
lina, the first and second cleavages divide the egg into a quartet of
yolk-bearing macromeres, all containing entoblast, from which a quartet
of ectoblastic micromeres is separated by the third cleavage in the
formation of the eight-cell stage. This is exactly the order of cleavages
in the eggs of annelids and mollusks. In Sacculina, then, the first
segregation of ectoblast occurs two cleavages later than in Lepas, in
which there is precocious segregation of ectoblast. In Sacculina the
first and second cleavages divide the egg into four yolk-bearing macro-
meres, each containing entoblast and ectoblast, and the segregation of
the primary germ-layers begins at the third cleavage ; but in Lepas
the segregation begins at the first cleavage without subdivision of the
egg into four quadrants. Comparing the four-cell stage of the two
genera, the entoblast in Lepas is all concentrated into one of the
four cells each of which in Sacculina contains entoblast. According to
this view the first cleavage of Lepas corresponds to the third of Saccu-
lina so far as the first segregation of ectoblast is concerned. Whether
the first micromere of Lepas is homologous with the quartet of micro-
meres in Sacculina cannot be determined until the fate of those cells is
traced in the latter genus. There is reason for inferring that in Saccu-
lina other quartets of ectomeres are cut off" from the yolk-macromeres
and added to the ectoblast. This must be settled before any further
conclusions can be drawn. The final result of the development — the
Nauplius — is similar in Lepas and in Sacculina. A comparison of the
cell-lineage of the two genera may be expected to yield some results
bearing on the suggestion that possibly the micromeres (ab^, &, d'^-^) of
Lepas may be equivalent to quartets of ectoblastic micromeres in Saccu-
lina, and possibly to those in more distantly related forms. These are

126 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

merely suggestions which have grown out of comparison of the order of
the cleavages involved in segregating the germ-lajers.

The segregation of the ectoblast as three micromeres is apparently
not peculiar to Lepas among Entomostraca. The cleavage of certain
parasitic Copepoda has close resemblances to that of Lepas as regards
number of cleavages involved in the segregation of the germ-layers.
In Lerncca, according to Pedaschenko ('93), the ectoblast and mesoblast
are separated from the yolk-macromere. (entoblast) by means of four
cleavages. It will appear in the discussion of the germ-layers in the
following section of this paper, that in the instance just cited the first
three micromeres probably contain all the ectoblast with the " second-
ary mesoblast," while the fourth is the primary mesoblast ; in this case,
then, the number and order of cleavages involved in germ-laj'er segrega-
tion would agree with my observations on Lepas.

In the figures and accounts of the cleavages of various phyllopods and
copepods, in whicli the germ-layers appear to be established as early as
the thirty-two-cell stage, there are found many suggestions that further
investigations may show a close resemblance to the cell-lineage of Lepas.
Some examples of such suggestive papers are those of Grobben ('79, '81)
on Moina and Cetochilus, Urbanowicz ('86) and Hacker ('92, '97) on
Cyclops, and Pedaschenko ('93) on Lerneea ; but in none of these genera
are the facts as yet sufliciently well known to warrant close comparison
with Lepas, especially since there is much disagreement between the
observations of these investigators. At present this mention of a possi-
ble resemblance to the cleavage of Lepas can have only the value of a
suggestion, which may possibly stimulate compai'ative study of the
cleavage of those Entomostraca in which the early segregation of the
germ-layers makes it possible to trace the lineage of the cells to the com-
plete separation of the germ-layers.

The cleavage of Lepas has some general resemblances to that of the
nematodes. Particularly is there resemblance in the early segregation
of the germ-layers ; but, as to the order of cleavage involved in this
process, there are great and at present irreconcilable differences. The
first cleavage in Nematoda begins the. separation of the germ-layers.
Thus the cell aW- contains ectoblast in tJie nematodes as in the cirri-
pede, and cdP contains ectoblast and mes-entoblast. The second cleav-
age in the nematodes completes the segregation of the mes-entoblast
from ectoblast, whereas this is accom])lished by the third cleavage in
Lepas. It is obviously impossible to make any comparison of the
details of the early development.

BIGELOW : EARLY DEVELOPMENT OF LEPAS. 127

In certain respects the cell-lineage of Lepas recalls that of some roti-
fers, as described by Zeliuka ('91) and especially by Jennings ('96).
In the rotifers, as in Lepas, the separation of the primary germ-layers
begins with the first cleavage, the cell ab^ being ectoblastic, and cd?
containing ectoblast in addition to all the entoblast. Still more remark-
able is the resemblance in that the entoblast is derived from the cell
d?'^ both in Asplanchna and in Lepas. This cell is purely entoblastic in
Lepas, and probably so in Asplanchna; its two minute derivatives d^-"^
and d"^-"^ are regarded by Jennings as belonging to this germ-layer. The
macromere d^ in this rotifer, as in Lepas, gives rise to d^'"^ in the third
cleavage and d^'^ in the fourth. In both d*-' is purely ectoblastic. In
Lepas d^'"^ is the primary mesoblast, but in Asplanchna it is ectoblast.
However, the exact origin of the mesoblast in the rotifers is unknown.
It is evident that the number and order of cleavages which are involved
in the segregation of the entoblast from the ectoblast are the same in
the rotifer as in the cirripede.

XI. Comparisons of the Germ-Layers of Lepas with those

of other Crustacea.

The account here given of the development of Lepas agrees with the
published descriptions of the development of the majority of Crustacea,
in that the blastopore is posterior and ventral, and apparently near the
position of the future anal aperture. This similarity in the relation of
the blastopore appears at first to be without significance, if one com-
pares the embryo of Lepas, which has the mesoblastic band on its dorsal
side, with crustacean embryos containing much yolk and having the
mesoblastic plate ventral in position, as it is in decapods. However,
the facts appear to allow of the following interpretations : In crustacean
eggs which are heavily laden with yolk, the embryonic disk is at first
confined to the ventral surface, but gradually extends dorsally over the
yolk-mass. The mesoblast is formed while the embryonic disk is ven-
tral. In Lepas, and some other Crustacea in which there is a relatively
small amount of yolk, the embryonic disk is not confined to the ventral
surface, but from the close of cleavage it is extensive enough to sur-
round the yolk completely. In consequence of this the mesoblast,
which in higher Crustacea forms bands on either side of the median
ventral line, in Lepas extends along the dorsal line. If one imagines an
ordinary decapod egg deprived of the greater part of its yolk until, at
the close of cleavage, the edges of the embryonic disk meet on the dor-

128 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

sal surfiicc, the conditions in Lepas would be closely imitated. The
mcsobUist bands would in such a case come to lie more and more dor-
sally, in proportion as the loss of yolk allowed the enibryouic disk to
cover the whole surface. In Lepas these bands in their position near the
median-dorsal line, where the distal ends of the appendages later appear,
may be considered as representing the outer edge of the embryonic disk
of eggs having so much yolk that the disk is spread out over the ventral
surface only, not being folded completely around the yolk as in the case
of Lepas. It appears, then, that, though the mesoblast of Lepas is dor-
sal and that of yolk-laden eggs of higher Crustacea ventral, the two
may be regarded as having homologous positions. In comparing Lepas
with most other Crustacea the blastopore may be considered as having
the same relative position, and the germ-layers may be compared with
reference to their method of formation at the blastopore and their
extension from that region.

Groom ('94, p. 199), who regarded the mesoblast and entoblast as
originating from a single yolk-cell after the blastopore is closed, was
necessarily led to the conclusion that " with respect to the origin of the
mesoblast and hypoblast of the Nauplius, the cirripedes occupy an iso-
lated position among Crustacea." This statement is based upon his
view that the yolk-cells after the closing of the blastopore constitute the
mes-entoblast. This view is at variance with the conditions in other
Crustacea, for the mesoblast commonly originates from the blastoderm
and not from yolk-cells lying beneath that structure. In this paper it
has been shown that, in general terms, the mesoblast in Lepas origin-
ates from the blastoderm, and that, consequently, Groom's view is
incorrect.

The accounts of most earlier workers on cirripede embryology lead to
conclusions practically the same as Groom's. In opposition to such
conclusions it will be pointed out in the following discussion that in the
formation of the germ-layers there are many fundamental resemblances
between Lepas and other Crustacea.

Among all Crustacea whose embryology is at present known, the
closest resemblance to the development of Cirripedia appears to be
found among the Phyllopoda and Copcpoda, especially the latter. In
the preceding chapter reference has been made to similarity of cleavage
in these three groups of Entomostraca, but here the comparison between
the germ-layers is to be emphasized.

Urbanowicz ('86) has studied the germ-layers of the copepod Cyclops
and has found only one entoblast cell, over which the ectoblast grows

BIGELOW : EARLY DEVELOPMENT OF LEPAS. 129

closing the blastopore. Ectoblastic cells around the blastopore give rise
to mesenchyme (" secondary mesoblast "), which forms most of the
mesoblastic structures of the Nauplius. The mesoblast proper probably
originates from the entoblast, as does the primary mesoblast of Lepas.
It is evident that there is in Cyclops, .according to Urbanowicz, a condi-
tion closely resembling that of Lepas.

In close agreement with Urbanowicz's account of Cyclops and ray
own of Lepas, is Pedaschenko's ('93) description of the formation of
the germ-layers of the parisitic copepod Lerusea. In this genus the
mesoblast and ectoblast are separated from the yolk-entoblast in the
first four divisions, as in Lepas. The four micromeres thus produced
subdivide and form the blastoderm, which grows over the entoblast.
At the margin of the growing blastoderm (blastopore) some cells (ap-
parently ectoblastic) divide parallel to the surface and form migrating
mesenchyme cells. These apparently correspond to the " secondary
mesoblast " of Lepas. On the ventral side four of the cells sink beneath
the ectoblast and constitute the primitive mesoblast cells. The lineage
of these cells has not been definitely traced, but from their position I
infer that they are probably the direct descendants of the fourth micro-
mere, in which case the primary mesoblast originates directly from the
entoblast, as in Lepas.

Hacker's ('92, '97) studies of Cyclops led to results widely different
from those of Urbanowicz. According to Hacker, a cell lying in the
blastopore divides into a genital cell and a primitive mesoderm cell.
The cells surrounding the blastopore divide, giving rise to the primitive
endoderm cells ; this is in line with Grobbeu's account of Cetochilus,
to which reference will be made later, and opposed to Urbanowicz, who
found mesenchyme cells originating from cells bounding the blastopore.

Grobben's ('81) views of the formation of the germ-layers in the
copepod Cetochilus do not agree with the account of Cyclops given by
Urbanowicz, and only in part is there agreement with Hackei-'s account
of Cyclops. His description of the thirty-two-cell stage of Cetochilus
forms the best starting-point for purposes of comparison. In this stage,
viewed from the vegetative pole, there is noticed a distinct bilateral
symmetry in arrangement of the cells. A *' central entoderm " cell and
one small " anterior entoderm " cell lie in the median plane. Four
cells placed symmetrically on either side of the " central entoderm "
cell will by the next division form "entoderm" and ectoderm. The
cell in the median line and posterior to the " central entoderm " cell
forms in later division four cells, of which the two nearer the " central

130 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

entoderm " are said to be the primitive raesoblast cells, and the two
posterior products ectodermal.

It appears that the " central entoderm " cell of Grobben is probably
the single entoblast cell to which Urbanowicz refers. The blastoderm
cells lying laterally and anterior to the entoderm cell in Cyclops are
said by Urbanowicz to give rise to mesenchyme, while Grobben in
Cetochilus and Hacker in Cyclops find entoderm originating from cells
in corresponding positions. It is probable that this contradiction arose
from failure to follow the germ-layers into the ultimate organs. The
figures of Cetochilus by Grobben and those of Cyclops by Hacker do
not give conclusive proof regarding the fate of the cells which they con-
sider endoderm. I have not seen the original figures by Urbanowicz.
The differences between these authors will probably be adjusted when
the later history of the mesoblast and entoblast is more accurately
traced.

The cell posterior to the " central endoderm " cell in the thirty-two-
cell stage of Cetochilus is said by Grobben to form the mesoblast and
also to contain some ectoblast. This latter point must still be regarded
as problematical, for Grobben's figures do not give convincing proof.
It is possible that the cell in question may be wholly mesoblastic, in-
stead of only partly so. However, the important point is that this cell
appears to originate in connection with the " central endoderm " cell.
Accordingly mesoblast in Cetochilus originates from etitoblast ; a con-
dition certainly existing in the case of the barnacle Lepas, and the
studies of Urbanowicz make it appear probable that such is also the
case in Cyclops.

Grobben's ('79) account of the development of the phyllopod Moina
agrees with Urbanowicz's account of Cyclops and my own account of
Lepas as to the formation of ectoblastic mesoblast from blastoderm cells
bounding the blastopore laterally and anteriorly. But in a position
corresponding to that of the entoblast cell of Lepas and Cyclops there
is in Moina a " primitive genital cell," and the entoblast is said to be
developed from a cell lying immediately posterior to it. It should be
mentioned here that Samassa ('93), while agreeing essentially with
Grobben's description of cleavage stages, failed to find evidence of such
early differentiation. With respect to this result it must be considered
improbable that the visible peculiarities of the cells in the region of the
blastopore in cleavage stages are without significance. It seems more
probable that the peculiar features of certain cells do represent early
differentiations, as Grobben claimed. The results of Samassa and

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 131

others render doubtful the early differentiation of a genital cell in
Moina ; but Hiicker ('92, '97) has contributed some important cyto-
losical evidence favorable to Grobben's conclusions.

To summarize the comparison of Lepas with the Copepoda and Phyl-
lopoda, it has been pointed out that —

1. In Lepas, in Moina (Grobben), in Cyclops (Urbanowicz), and
probably in the parasitic copepod Lernsea (Pedaschenko) mesoblast
originates from ectoblastic cells of the blastoderm around the blasto-
pore. In Cetochilus (Grobben) and in Cyclops (Hacker) there is a
disagreement with Lepas, in that the entoblast cells are said to originate
from cells whose origin and position is similar to those which in the
above mentioned forms produce mesoblast.

2. In Lepas, Cyclops (Urbanowicz) and Lernaea a single entoblast
cell, in Cetochilus (Grobben) the " central entoblast " cell, at first lies
in the blastopore and it, or its derivatives, are overgrown by the
blastoderm.

3. In Lepas, Cyclops (Urbanowicz), Cetochilus (Grobben) and
Lernffia (1) (Pedaschenko) some mesoblast originates directly from
the entoblast cell which lies in the blastopore, that is to say, the
yolk-macromere is mes-entoblastic. In all of these except Cetochilus
(Grobben) mesoblast also originates from ectoblastic cells around the
blastopore.

The foregoing comparisons of the gerra-Iayer formation in Lepas and
other Entomostraca in which early differentiation takes place, brings
out many points of resemblance. But in some cases there are differ-
ences apparently irreconcilable. One can scarcely believe that such
contradictory statements as have been summarized in the preceding
paragraphs are based upon observations all equally reliable. Renewed
investigation of the uncertain points is much needed. The numerous
resemblances even from the beginning "of development, make it very
desirable that the cell-lineage should in these cases be carefully studied
so as to give a basis for accurate comparisons. Until such data are
accessible it is unsafe to draw conclusions respecting homologies of cells
or even of the germ-layers,

In many Crustacea there is at the blastopore an immigration of
many cells into the cleavage cavity. In some of these cases the cavity
.is up to that time filled with yolk. The cell-mass thus formed by
immigration into the cleavage cavity is mes-entoblastic, and the meso-
blast and entoblast are at first indistinguishable, or at any rate inves-
tigators have failed to find distinguishing marks. As examples of

132 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

such conditions may be cited Daphuia, according to Lebedinsky ('9l) ;
!Moiua and Daphnia, according to Samassa ('93) ; and many higher
Crustacea.

Such au origin of mesobhist and entoblast is not necessarily opposed
to the account which I have given of the gerra-layer formation of Lepas,
for differentiation, though not observable, may yet occur in the cases
mentioned. Were there not in Lepas peculiarities by which the cells
can be distinguished at au early stage, the inmiigratiug mass of cells,
composed of entoblast, and of primary and secondary mesoblast, would
be correctly described as mes-entoblast, out of which the two layers
become later visibly differentiated. If the entoblast cells of Lepas were
completely separated from the yolk-mass, as is the case in many other
Crustacea, it would perhaps be impossible, in the absence of the easily
recognized yolk-laden entoblast, to trace the lineage of the mesoblast
independently of the entoblast, and in such conditions it would be nec-
essary to consider the immigrating mass of cells as mes-entoblastic. It
is probable that some such conditions obtain in some of the Crustacea
in which a mes-entoblastic immigration is said to occur. At any rate,
germ-layer formation in such cases agrees in essentials with that
observed in Lepas. Grobben's ('79) study of Moina suggests that in
this genus, at least, the immigrating mass of mes-entoblast may be not
entirely undifferentiated as Samassa ('93) supposed.

There is some evidence that the comparison between Lepas and cer-
tain higher Crustacea may be carried still farther than the suggestions
offered in the preceding paragraph. In Astacus, according to Reichen-
bach ('86), the mesoblast originates at the anterior margin of the
blastopore, where the ectoblast joins the entoblast. Keichenbach dis-
tinguished in the invagination both yolk-absorbing cells (vitellophags),
which enter into the yolk-pyramids, and also the cells forming the
entoderm plate. All these cells are said to enter into the mesenteron
and liver lobes, and hence the invagination is entoblastic. However,
McMurrich ('95, pp. 135, 136) reviews the evidence and suggests that
the yolk-pyraraids give rise to some mesoblast. If this proves true, the
invagination is to be regarded as mes-entoblastic ; but, in addition to
mesoblast so formed from entoblast, other mesoblast cells certainly
originate from the blastoderm in front of -the invagination. It follows
that there are, as regards origin, two kinds of mesoblast — ectoblastic
and entoblastic.

In other accounts of development of the higher Crustacea there are
suggestions of such a double origin of mesoblast, but there is as yet

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 133

lack of a definiteness of statement sufficient to afford basis for com-
parisons of any value.

Comparing the development of Astacus with that of Lepas, the ecto-
blastic mesoblast at the anterior edge of the blastopore appears to be
equivalent to the " secondary mesoblast " of Lepas. If the suggestion,
that the invagination is mes-entoblastic, proves true, it may be possible
to regard the mes-eutoblastic cell c?^*^ of Lepas as representing the invagi-
nated cells of the higher Crustacea ; the primary mesoblast and ento-
blast of Lepas would then be comparable with the germ-layers derived
from the invagination in the higher forms. In such a case there would
be further agreement with Lepas in that the mesoblast originates from
both ectoblast and entoblast.

Summary.

1. Lepas resembles most other Crustacea (a) in respect to position of
the blastopore, which is ventral and posterior, (b) in extension of the
entoblast and mesoblast from the blastopore as a starting-point, (c) in
the mode of formation of the organs of the larva.

2. In Lepas, as in most other Crustacea, the mesoblast and entoblast
originate in the region of the blastopore from cells which, speaking in
general terms, at first lie in the blastoderm and later migrate into the
cleavage cavity.

3. Among the migrating mes-entoblastic cells one can distinguish in
Lepas the individual cells of entoblast and of two varieties of meso-
blast. Representatives, if not precise homologues, of these kinds of
cells are probably present both in other Entomostraca and in the higher
Crustacea.

XII. General Summary with Table of Cell-Lineage of Lepas.

The results which are of special interest in relation to the develop-
ment of Cirripedia have already been summarized in connection with
the accounts of the several stages of development. Only results of
more general interest are again summarized here.

The cleavage of Lepas is throughout total and unequal.

Stages with 2, 4, 8, 16, 32, and 62 "resting" cells are regularly
formed.

In the eight-cell stage and thereafter there is a well-marked bilateral
arrangement of the cells.

In the first three cleavages thi-ee " protoplasmic " micromerea are

134 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

separated from the yolk-bcaring macromcrc, and the fourth cleavage
separates the primary mesoblast from the yolk-eutoblast. Thus, in the
sixteen-cell stage the entobhxst is completely separated from the other
germ-layers.

Mesoblast originates both from entoblast (fourth cleavage) and from
ectoblast (sixth cleavage). The mesoblast derived from ectoblast
C secondary mesoblast ") forms a large part at least of the mesen-
chyme of the Nauplius. The fate of the primary mesoblast (entoblastic
mesoblast) has not been distinguished from that of the " secondary
mesoblast" (ectoblastic mesoblast).

The blastoderm grows over the yolk-bearing entoblast, usually closing
the blastopore after the sixth cleavage. In cases where the yolk-mass is
very large, the closing of the blastopore may not occur until the suc-
ceeding cleavage. But in all cases the blastoderm is formed from de-
rivatives of three and only three micromeres {aP, c', d*'^), which are
cut off in the first three cleavages.

The yolk-macromere of the sixteen-cell stage has been traced to the
mesonteron. All the evidence supports entirely the interpretation that
after the fourth cleavage the yolk-macromere is purely entoblastic.

The irregularity and variability which authors have ascribed to the
cleavage of cirripedes do not normally exist in the case of Lepas. The
origin, relative position, and fate of all cells of all cleavage stages have
been shown to be constant, definite, and " determinate " so far as the
formation of germ-layers is concerned. In later stages specific areas of
cells, known to be of definite origin, enter into the formation of particu-
lar organs. It is therefore probable that the cells in cleavage stages
bear a definite and constant relation to future organs.

The chief points in the cell-lineage and their relation to the formation
of the germ-layers are summarized in the accompanying table.

Describing the formation of the germ-layers of Lepas in general terms,
there is no conflict with most existing accounts of the development of
other Crustacea ; in the absence of complete records of the cell-lineage
in other Crustacea, it is not possible to compare the details with cer-
tainty (see Summary, p. 133).

BIGELOW : EARLY DEVELOPMENT OF LEPAS.

135

TABLE OF THE CELL-LINEAGE OF LEPAS.

1 ceU.

Fertilized
Ovum

2 ceUs.

(1)

b^

(2)

4 cells. 8 cells.

ai-^(ec'bl.)

,4.1

M-^ (ec'bl.)

c*-2 (ec'bl.

rfi-2 [ec'bl.)
(3)

IG cells.

flS-l (ec'W.)

65-1 (ec'6/.)

^5.2'

c^-i (ec'bl.)
d^-"^ (ms'bl.)

(/■^•i (en'W.)

32 cells.

flG-* (ec'6;.)

i6-3

h'i-i

lfi.3

c6-4 (ec'W.)

6-3

C°"'

62 cells.

a'-6 (ec'W.)
a^-^(ms'bl.')
b'^-^ (ec'bl.)
b'-' (ms'bl.')

b'-^ (ms'bl.')

c"-6 (ec'6/.)
cT-5 (ms'bl.')

y", y^, y*, y^ designate the yolk-bearing raacromere; (1), (2), (3), the three
micromeres containing ectoblast ; ec'W., ectoblast; e/j'W., entoblast ; ms'bl., 'pvima.ry
mesoblast , ms'bl.', " secondary mesoblast."

136 bulletin: museum of compakative zoology.

ADDENDUM.
By E. L. Mark and W. E. Castle.

To avoid any misunderstanding we wish to state that the opinions
expressed by Dr. Bigelow regarding "quartet" cleavage are not wholly
shared by us. Lepas seems to us a good example of modified " quartet "
cleavage, and for that reason we think the quartet nomenclature has
more than mere convenience in its favor. To be sure, the quadrants in
Lepas are not symmetrical, but perfect symmetry is rarely met with in
quartet cleavage. So far as we recall, complete symmetry of the quad-
rants is found only in platodes. The condition there realized may be
considered primitive, all four quadrants sharing equally in the produc-
tion of ectoblast, mesoblast, and endoblast (see Wilson, '98). One
modification of this primitive symmetry is found in annelids and
moUusks, another in rotifers and cirripedes.

In the first-named groups the mesoblast is segregated, more or less
completely, in quadrant d, while the endoblast remains distributed
among all four quadrants. In the rotifers (see Jennings, '96) the en-
doblast is segregated in quadrant d, precisely as in Lepas, yet the cleav-
age progresses in perfect quadrant symmetry through at least the first
eight cell-generations, even though, to realize this symmetry, so-called
" mechanical laws of cleavage " are repeatedl}' transgressed. The origin
of the mesoblast in rotifers remains uncertain, but in Lepas, as Dr.
Bigelow clearly shows, the mesoblast arises from all four quadrants.
An examination of his table of cell-lineage (p. 135) shows other un-
mistakable evidences of quadrant symmetry in Lepas.

L The first-formed definitive ectomeres — which are also the first
cells to be differentiated for a particular germ-layer — arise sym-
metrically and synchronously from all four quadrants. They are the
four dorsal cells of the eiglit-cell stage, namely, a*'^, 6* ^ c*"^, and d'^-'^.
They correspond with what in polyclads, annelids, and mollusks have
been called the " first quartet of micromeres," which in these forms, as
in Lepas, are always the first ectomeres to be differentiated.

2. At the sixteen-cell stage, in Lepas, the mesoblast is included in
corresponding blastomercs (a■^•^ b''-'^, c^'^, d^'^) in all four quadrants.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 137

The only essential difference among the quadrants in the mode of sep-
aration of the mesoblast is this : In quadrant d, cell d^'- is purely meso-
blastic ; but the corresponding cells in each of the other quadrants
contain mesoblast associated as yet with ectoblast, and the two are
not separated until the second later generation, that is, in the sixty-
four-cell stage. The earlier separation of the mesoblast in quadrant d,
as compared with the other quadrants, may be due to the relatively
greater bulk of the mesoblast in quadrant d. The mesoblast is really
partially segregated in quadrant d, — since that quadrant contains a
greater portion of mesoblast than any of the three remaining quadrants,
— while the endoblast is completely segregated in that quadrant. The
segregation of the mesoblast in quadrant d finds a parallel repeatedly in
mollusks and annelids ; that of the endoblast in the same quadrant is
paralleled in rotifers.

Notwithstanding these coenogenetic modifications, the primitive quad-
rant-symmetry finds frequent expression in the cleavage of Lepas, a fact
to which the quadrant nomenclature clearly directs attention.

It is true that in Lepas radial symmetry is replaced by bilateral sym-
metry considerably earlier than is the case in most annelids and mol-
lusks, and much earlier than in rotifers, but the difference is one of
degree rather than of kind. Cleavage in Lepas, as truly as in the other
forms mentioned, is at first radial, and only gradually becomes bilateral.

138 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

BIBLIOGRAPHY.!

Auerbach, L.

'74. Organologiscbe Studieu. Brcslau. 262 pp., 4 Taf.

Beneden, E. van.

'70. llccherclics sur rcmbryogeuie dcs Crustaces. III. Developpcnient
de I'cKuf et dc rembryou dcs Sacculiues (Sacculini carciiii, Thoinps.).
Bull. Acad. Roy. Belgique, ser. 2, Tom. 29, pp. 99-112, 1 pi.

Bigelow, M. A.

"96. ()u the early Development of Lepas fascicularis. (A preliminary note.)
Auat. Anz., Bd. 12, pp. 263-269, 9 figs.

Bigelow, M. A.

'99. Notes on the First Cleavage of Lepas. Zool. Bull., Vol. 2, pp. 173-
177, 7 figs.

Blochmann, F.

'81. U bcr die Entwickelung der Neritina fluviatilis MiiU. Zcit. f. wiss. Zool.,
Bd. 36, pp. 125-174, Taf. 6-8.

Bobretzky, N.

'76. Studien iiber die embryouale Entwickelung der Gastropodcn. Areli. f.
mikr. Anat., Bd. 13, pp. 95-169, Taf. 8-13.

Bovallius, C.

'75. Einbryologiska Studier. I. Om Balanidcrnas Utvcekling. Stockholm.
44 pp., 5 pis.

Castle, W. E.

'96. The Early Embryology of Ciona intestinalis. Bull. Mus. Comp. Zool.,
Harvard Coll., Vol. 27. pp. 201-280, 13 pis.

Conklin, E. G.

'97. The Embryology of Crcpidula, a Contribution to tlic Cell Lineage and
early Development of some Marine Gasferopods. Jour. Morph., Vol. 13,
pp. 1-226, pis. 1-9.

1 No attempt has been made to cite a complete bibllograpliy on cirripede embry-
ology. Groom ('94) has given a tolerably complete list of the literature on the
structure and development of the Cirripedia.

BIGELOW: EARLY DEVELOPMENT OF LEPAS. 139

Conklin, E. G.

'98. Cleavage aud Differentiation. Biol. Lect., Wood's Holl, 1896-97,
pp. 17-4:5.

Crampton, H. E.

'96. Experimental Studies on Gasteropod Development. Arch. f. Entwickel-
ungsmech. d. Organisnien, Bd. 3, pp. 1-19, Taf. 1-4.

Darwin, C.

'51. A Monograph on the Sub-class Cirripedia. Lepadidee. Ray Society,
London. xi-|- 400 pp., 10 pis.

Darwin, C.

'54. A Monograph on the Sub-class Cirripedia. (Balanidae, Verrucidae,
etc.). Ray Society, London, viii -|- 684 pp., 30 pis.

Filippi, F.

'65. Ueber die Entwicklung von Dichelaspis Darwinii. Untersuchungen
zur Naturlehre (Moleschott), Bd. 9, pp. 113-120, 2 Taf.

Grobben, C.

'79. Die Entwicklungsgeschichte der Moina rectirostris, u. s. w. Arb. Zool.
Inst. Wien, Tom. 2, pp. 203-268, Taf. 11-17.

Grobben, C.

'81. Die Entwicklungsgeschichte von Cetochilus septeutrionalis. Arb. Zool.
Inst. Wien, Tom. 3, pp. 243-282, Taf. 19-22.

Groom, T. T.

'92. On the Early Development of Cirripedia. (Abstract.) Proc. Roy. Soc
London, Vol. 52, pp. 158-162.

Groom, T. T.

'94. On the Early Development of Cirripedia. Phil. Trans. Roy. Soc. Lon-
don, Vol. 185, B, pp. 119-232, pis. 14-28.

Hacker, V.

'92. Die Kerntheilungsvorgange bei der Mesoderm- und Entodermbildung
von Cyclops. Arch. f. mikr. Anat., Bd. 39, pp. 556-581, Taf. 24, 25.

Hacker, V.

'97. Die Keimbahn von Cyclops, u. s. w. Arch. f. mikr. Anat., Bd. 49, pp.
35-91, Taf. 4, 5.

Hoek, P. P. C.

'76. Zur Entwickelungsgeschichte der Entomostraken. (I. Embryologie
von Balanus.) Niederl. Arch. f. Zool., Bd. 3, Heft 1, pp. 47-82, Taf. 3, 4.
Hoek, P. P. C.

'83. Report on the Cirripedia collected by H. M. S. Challenger. (Historical
and Systematic.) Vol. 8, Part 25, Zool. Series, 169 pp., 13 pis.
Hoek, P. P. C.

'84. Report on tlie Cirripedia collected by H. M. S. Challenger. (Anatom-
ical.) Vol. 10, Part 2S, Zool. Series, 47 pp., 6 pis.

VOL. XL. — NO. 2 6

140 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Jennings, H. S.

'96. The Early Devdoi)rucnt of Asplanclina Ilcrrickii de Gucnie. A Con-
tribution to Developmental Mechanics. Bull. Mus. Comp. Zocil., Harvard
Coll., Vol. 30, pp. 1-116, 10 pis.

Knipowitsch, N.

'92. Bcitrage zur Kenntniss der Gruppe Ascotlioracida. Trav. de la Soc.
Nat. St. Petersburg, Tom. 23, pp. 82-155, 3 Taf. (Russian with abstract
in German.)

Kofoid, C. A.

'94. On some Laws of Cleavage in liimax. A Preliminary Notice. Proc.
Am. Acad. Arts and Sci., Vol. 29, pp. 180-203, 2 pis.

Kofoid, C. A.

'95. On the Early Development of Limax. Bull. Mus. Comp. Zool., Har-
vard Coll., Vol. 27, pp. 35-118, 8 pis.

Korschelt, E., und Heider, K.

'90-91. Lehrbuch der vergleichcndeu Entwicklungsgeschichte der wirbel-
loscn Thiere. Jena, xii + 908 pp.

Lang, A.

'78. Die Dotterfurcliung von Balauus. Jena. Zcit. f. Naturw., Bd. 12,
pp. 671-674, Taf. 20, 21.

Lebedinsky, J.

'91. Die Entwicklung der Daphnia aus dem Soramereie. Zool. Anz., Jahrg.
14, pp. 149-152.

Lillie, F. R.

'95. The Embrvology of the Unionidse. A Study in Cell-Lineage. Jour.
Morph., Vol. 10, pp. 1-100, pis. 1-6.

McMurrich, J. P.

'95. Embryohjgy of the Isopod Crustacea. Jour. Morph., Vol. 11, pp. 63-
154, pis. 5-9.

Miiller, F.

'64. Eiir Darwin. Leipzig. 91 pp., 67 figs.

Mvinter, J., und Buchholz, R.

'69. tJber Balanus improvisus Darw., var. Grypliicus Miint., Beitrag zur
carcinologischen Fauna Dcutschlands. Mittli. aus dem Naturw. Verein
von Neu-Vorpommcrn und Riigen. Bd. 1, pp. 1-40, Taf. 1, 2.

lieview by Gerstdcker in Arch. f. Naturg. Jahrg. 1871, Bd. 2, pp. 364,
365.

Nassonow, [W.] N.

'85. Zur embryonalen Entwicklung von Balanus. Zool. Anz., Jahrg. S,
pp. 44-47.

BIGELOW; EARLY DEVELOPMENT OF LEPAS. 141

Nassonow, [W.] N.

'87. Oil the Ontogeny of the Crustaceans Balauus and Artemia. Izvyest.
imp. Obshch. Ljubit. Estestv. Autrop. i Ethuog. Moscow. Tom. 52,
pp. 1-14, 35 figs. (Russian.)

Nussbaum, M.

'87. Vorlaufige Bericlit iiber die Ergebnisse einer mit Unterstiitzung dcr
Koniglichen Akademie ausgetuhrte lleise uach Califoniien. Sitzungsb.
d. k. preuss. Akad. der Wiss. zu Berlin, pp. 1051-1055.

Nussbaum, M.

'89. Bildiing und Anzahl der Richtungskorper bei Cirripedieu. Zool. Anz.,
Jahrg. 12, p. 122.

Nussbaum, M.

'90. Anatomische Studieu an Califoruisclien Cirripedien. Bonn. 97 pp.,
12 Taf.

Pedaschenko, D.

'93. Sur la segmentation de I'oeuf et la formation des feuillets embryonnaires
chez la Lerusea branchialis L. Rev. des. Sc Nat. St. Petersburg, Tom.
4, pp. 186-199, 11 figs. (Russian with abstract in French.)

Reichenbach, H.

'86. Studien zur Entwicklnngsgeschiehte des Elusskrebses. Abhandl.
Senckenberg. Naturf. Gesellsch. Frankfurt. Bd. 14, pp. 1-137, Taf.
1-14.

Samassa, P.

'93. Die Keimblatterbildung bei den Cladocereu. I. Moina rectirostris
Baird. [II. Daphuella und Daphnia.] Arch. f. mikr. Anat., Bd. 41, pp.
339-366, Taf. 20-22; pp. 650-688, Taf. 36-39.

Solger, B.

'90. Die Richtungskorperchen von Balauus. Zool. Anz., Jahrg. 13, pp.
607-609.

Torrey, J. C.

:02. The Cell-Lineage of the Mesoblast-l3ands and Mesenchyme in Thalas-
sema. Science, n. s., Vol. 15, No. 380, pp. 576, 577-

Treadwell, A. L.

:01. The Cytogeny of Podarke obscura (Verrill). Jour. Morph., Vol. 17,
No. 3, pp. 399-486, pis. 36-40.

Urbanowicz, F.

'84. Zur Entwicklnngsgeschiehte der Cyclopiden. (Vorl. Mitth.) Zool.
Anz., Jahrg. 7, pp. 615-619.

Urbanowicz, F.

'86. Contributions to the Embryology of Copepods. Kosmos, Lembcrg,
Jahrg. 10, pp. 239-259, 300-314. (Polish.)

142 bulletin: museum of COMrAUATlVE ZOOLOGY,

Urbanowicz, F.

'86". Contributions a renibryologie dcs Copepodcs. Arch. Slav. Biol.,
Tom. 1, t'asc. 3, pp. 663-G67 ; et Mem. Uiiivers. Varsovic. (Abstract of
preceding paper.)

Van Beneden, E. See Beneden, E. van.

Weismann [A.] und Ischikawa [C]

'88. Weitere Untersuchuiigen zum Zalilengesetz der Richtungskorpcr. Zool.
Jabrb., Abth. f. Anat. u. Ontog., Bd. 3, pp. 575-610, Taf. 25-28.

Willemoes-Suhm, R. von.

'76. On the Development of Lepas fascicularis and the " Arcliizoea " of
Cirripcdia. PhU. Trans. Roy. Soc London, Vol. 166, pt. 1, pp. 131-
154, pis. 10-15.

Wilson, E. B.

'98. Considerations ou Cell-Iiincage and Ancestral Reminiscence, etc. An-
nals N. Y. Acad. Sci., Vol. 11, pp. 1-27, 7 ligs.

Wilson, E. B.

:01. Studicn iiber den Korperbau der Auneliden, V. By Eduard Meyer.
(Review.) Science, u. s., Vol. U, No. 362, pp. 389-91.

Zelinka, C.

'91. Studien iiber Riiderthiere. III. Zur Entwicklungsgeschichte der Riider-
thiere nebst Bemerkungcn iiber ihre Anatomic und Biologic. Zeit. f. wiss.
Zool., Bd. 53, pp. 1-159, Taf. 1-6.
Jlso in : Arbeiten a. d. Zool. Inst. Graz. Bd. 4, pp. 323-481, Taf. 1-6.

Ziegler, H. E.
__ '95. Untcrsuchungeu iiber die erstcn Entwicklungsvorgiinge der Nematoden.
Zugleich ein Beitrag zur Zellenlehre. Zeit. f. wiss. Zool., Bd. 60, pp. 351-
410, Taf. 17-19.

BIGELOW : EARLY DEVELOPMENT OF LEPAS. 143

EXPLANATION OF PLATES.

The figures of Plates 1-10 were drawn from the eggs of Lepas anatifera, and
those of Plates 11 and 12 from L.fascicu/aris.

An Abbe' camera lucida was in every case used in sketching the eggs. The
figures of Plate 1, and Figures 57, 63-05, 74-77 were drawn at a magnification of
about 220 diameters ; all others in Plates 1-10 at about 365 diameters. The fig-
ures of Plates 11 and 12 are magnified about 210 diameters.

All figures, except those of transverse sections, are so arranged that the posterior
end of the embryo, or the more pointed end of tlie vitelline membrane, is directed
toward the bottom of the Plate; in transverse sections the ventral side is toward
the bottom.

Double-headed arrows are used in some of the figures to connect two cells of
common origin.

The vitelline membrane has not been represented, except in Figures 1-17 and
94-97.

Figures 1-30 and 95-99 are oriented by the axis of the vitelline membrane; all
others by the axis of the embryo.

The small circles without stippling indicate the positions of the oil spherules in
the yolk. Nuclei are distinguished by wavy lines, or by stippling, to represent
chromosomes.

In Plates 2 and 3 a pale yellowish buff tint has been used to represent the more
finely granular and more "protoplasmic " portfon of the egg and blastomeres.

Plates 1, 4, 11, and 12 have been printed without tint. To aid in quickly dis-
tinguishing between the derivatives of quadrants a, h, and c, all the blastomeres
of quadrant h in Figures 38-59, 61 are printed in stipple without tint, and in Figures
60 and 65 (Plate 7) the same method of designation has been employed to indicate
the cells (b'^-^-b''-^) of tiiis quadrant concerned in the formation of the secondary
mesoblast.

In Plates 5-10 the pale yellowish bnff tint has been employed to indicate the
blastomeres derived from quadrant d, the primari/ mesoblast (#2 and its descend-
ants) being distinguished from the other derivatives by receiving a stipplinq in
addition to the tint. In Plates 8-10 the tint has been restricted to d^-^ (entoblast)
and its derivatives.

144 BULLETIN : iMUSEUM OF COMPARATIVE ZOOLOGY.

ABBREVIATIONS.

I

For explanation of tlie letters and exponents designating blastoracrcs, see
explanation of tlie nomenclature of cleavage (pp. 74-70).

ec'bl. Ectoblast.

en'W. Entoblast.

Ibr. Labrum.

mb.vt. Vitelline membrane.

md. Mandible.

7ns'b/. Mesoblast of double origin.

ms'bl'. " Secondary mesoblast " (ecto-

blastic mesoblast).

pr'nl. $ Male pronucleus.

pr'nl. 9 Female pronucleus.

The Roman numerals I. II. (Figs. 28, oO) indicate the position of the first and
second cleavage planes, respectively; the Arabic numerals 1-4 (Figs. Ul, 9o, ll*2-
12(3), the sequence in which the transverse furrows marking off tiie Nauplius
appendages make their appearance.

Plate 7. Figs. 56-65.

Plate 8. Figs. 6G-73.

Plate 9. Figs. 74-86.

Plate 10. Figs. 87-94.

Plate 11. Figs. 05-110.

Plate 12. Figs. 111-120.

ast'cal.

Astroccel.

npp.

Appendage.

„tK

First antenna.

af^.

Second antenna.

Jil'po.

Blastopore.

bl'clnn.

Blastoderm.

cacsg.

Cleavage cavity.

cl.pol 1.

First polar cell.

c/.pol-\

Second polar cell

d.

Dorsal.

Plate

1.

Figs.

1-lfi.

Plate

2.

Figs.

17-22.

Plate

3.

Figs.

23-30.

Plate

4.

Figs.

31-38.

Plate

5.

Figs.

30-46.

Plate

6.

Figs.

47-55.

BioELOw. — Development of Lcpos.

PLATE 1.

Figures in this plate are all from living eggs, and represent stages between
oviposltion and the close of the first cleavage. The small circles represent the
oil splierules which are embedded in the yolk.

Fig. 1. Egg about thirty minutes after oviposition. Vitelline membrane and
second polar cell have appeared. Yolk uniformly distributed in the
egg.

Figs. 2-5. Egg elongating. Protoplasm concentrating in upper half of the egg.
Yolk becomes aggregated at the vegetative pole. Development of yolk-
lobe.

Fig. 6. Yolk-lobe has disappeared. Yolk radially symmetrical with reference to
chief axis of egg. Vitelline membrane has assumed its definitive form.

Fig. 7. Yolk moves to eccentric position with reference to the chief axis.

Figs. 8-15. First cleavage. Time thirty minutes. Drawings made at intervals of
about four minutes. Rotation of the dividing egg within the vitelline
membrane.

Fig. 16. One hour after close of first cleavage (Fig. 15). Yolk has returned some-
what toward the vegetative pole.

BiGELow.- Development ofLepas.

Plate 1

MAB del

B y;e'.5ei.:iih.BjSDf.

BioBLOw. — Development of Lepas.

PLATE 2.

Sections of eggs representing stages shown in Plate 1.
The vitelline membrane is represented in Figure 17 only.

Fig. 17. Formation of second polar cell. Yolk uniformly distributed in tlie ecg,

which is somewhat distorted into a form more than normally elongated,

owing to pressure in the egg-lamella.
Fig. 18. Same stage as that represented in Plate 1, Figure 4. Male and female

pronuclei in contact. Yolk collecting at the vegetative pole. The

pronuclei in this stage, which is characterized by the presence of a

yolk-lobe, are often separated as in Figure 19.
Same stage as that shown in Figure G. Pronuclei approaching; tliey

are usually in contact in tliis stage, as in Figure 20.
From an egg fixed in mercuric chloride, showing the distribution and

relative amount of the yolk. Early api)earance of the asters {?).

Pronuclei in contact. Same stage as tliat shown in Figure 0.
Formation of first cleavage spindle. Yolk becomes eccentric, as shown

in Figure 7.
Beginning of metaphase of first cleavage.

Fig.

10.

Fig.

20.

Fig.

21.

Fig.

22.

BiGELowr Development of Lepas.

Plate 2.

d .poU

/

i^^:^>^€)iW

5c-^^^ya

2, e-

/

\

■/

MAB.del,

B.Meisel.lilh.Bosiori

BiOBLOw. — Development of Lepas.

PLATE 3.

All Figures drawn from sections.

Fig. 23. Early anaphase of first cleavage.

Fig. 24. Late anaphase. Dividing egg in rotation. Second polar cell in cleavage

furrow.
Fig. 25. Telophase of egg, which has not yet rotated tlirough a complete quadrant.
Fig. 26. Rotation completed. Cleavage plane developing. Spindle disappearing.

Chromosomes vesicular.
Fig. 27. Two-cell stage. Vesicular chromosomes unite to form the nuclei. Yolk

has approached the vegetative pole, as in Figure 16.
Fig. 28. Second cleavage at beginning of metaphase, viewed from animal pole.
Fig. 29. Equatorial-plate stage of second cleavage; same egg as Figure 28.

Lateral view.
Fig. 30. Second cleavage in late anaphase, viewed from animal pole. /, /, indi-
cate first cleavage plane, //, //, second cleavage plane.

The long arrow falls in the projection of the sagittal plane of the

embryo.

BiGELow- Development ofLepas.

Plate 3.

23.

^5.

ousVcoel.

^4.

^.

o&?

\

y

«

X

^^.

<

^F.

ab'

clpol'-

■J

.cti^

29.

-\.-ab'

28.

ri-.cd^

JO.

Ipoi'

.cd-

«•>....

^E

*

«^^?-i

X

d^..

MAB.del

B.Meisel.liili.Bosffiii.

BiOELow. — Developmeut of Lepas.

PLATE 4.

Figures drawn from transparent preparations of entire eggs. Vegetative pole
at tlie left in lateral views.

Fig. 31. Egg viewed from animal pole. Late anaphase of second cleavage.

Fig. 32. Four-cell stage. Nuclei in " resting " phase. Egg viewed from animal

pole.
Fig. 33. Same egg viewed laterally. Yolk at vegetative pole of cell d"^.
Fig. 34. Four-cell stage during tliird cleavage. Viewed from animal pole.
Fig. 35. Same egg from vegetative pole. Oil spherules of the 3 oik near the

surface.
Fig. 36. Same egg in lateral view.
Fig. 37. Eight-cell stage from animal pole. All nuclei are in " resting " phase.

Second polar cell covered in by the meeting of a*'^ and c*"^.
Fig. 38. Same egg from vegetative pole. Oil spiierules near lower surface of j'olk-

cell. Cells of quadrant I (b^-^, b'^-'^) stippled.

BiGELOw.- Development ofLepas.

Plate 4.

>lAB.del

BiORLOW. — Development of Lepas.

PLATE 5.

Figures from transparent preparations of entire eggs. Vegetative pole at the
left in figures which represent lateral views.

Fig. 39. Kight-cell stage from animal pole. The seven " protoplasmic " cells are
in tiie fourth cleavage; the nucleus of yolk-cell (d^'^) is preparing for
division.

Fig. 40. Same egg in lateral view. Yolk at vegetative pole of cell c/*i.

Fig. 41. Fifteen "protoplasmic" cells; the yolk-cell (</^"S mes-entoblast) divid-
ing. Lateral view.

Fig. 42. Sixteen-cell stage from animal pole. Nuclei of all cells are in " resting "
phase. Primary mcsoblast (#-) separated from entoblast (ci°'^).

Fig. 43. Same egg viewed from vegetative pole. Oil spherules near lower surface
of the yolk-cell.

Fig. 44. Sixteen-cell stage from animal pole. All cells, except yolk-cell (entoblast
d^i) and the primary raesoblast cell (#'2), are undergoing the fifth
cleavage.

Fig. 45. Same egg in lateral view.

Fig. 4G. Same stage from vegetative pole. The three mcs-ectoblasts (compare
Fig. 43, w>'-, b^'^, c^ 2j contiguous to yoik-cell.

Note. — Cell ct^- is represented as divided, and its derivatives should have been
labelled aO-3, aS*.

BiGELOw.- Development ofLepas.

Plate 5.

''U-.S.del

B,Me!5el,liir..B(is(cn.

BloELOW. — Developmeut of Lepas.

PLATE 6.

Figures from transparent preparations of entire eggs. Vegetative pole at tlie
riijlit in figures representing lateral views.

Fig. 47. Si.\teen-cell stage with all colls of the blastoderm in fifth cleavage.
Primary mcsoblast {d^'-) and entoblast (d'"-'^) with enlarging nuclei.
Lateral view.

Figs. 48 and 51. Eggs with thirty cells, but the primary mesoblast cell (d^'-) lias
not yet completed the fifth cleavage. Nucleus of entoblast cell (r/^-i)
still in " resting " phase, but chromosomes preparing for fifth cleavage.
Entoblast (blastopore) bounded anteriorly and laterally by mes-ecto-
blasts (a^-^, ¥"^, b^"^, c^-'^). Viewed from vegetative pole.

Figs. 49, 50 and 53. Same stage seen in lateral view. In Figure 53 more of the
dorsal than of the ventral side is seen. Comparison shows that the cells
have essentially the same positions in the three eggs.

Fig. 52. Egg with thirty-two cells, reckoning the dividing yolk-entoblast as two
cells. Derivatives {d'^'\ t/*^ -') of the primary mesoblast at the posterior
edge of entoblast (blastopore). Viewed from vegetative pole.

Fig. 54. Optical section in sagittal plane of egg similar to one represented in
Figure 50. Cleavage cavity occupied by the yolk-entoblast, which is
uncovered at the blastopore only.

Fig. 55. View from animal pole of egg represented in lateral view in Figure 53.

BiGELOw- Development ofLepas.

Plate 5.

' „:t>, \\.«,'=

../^ © -r @ ^"

cipol- -7

,f-

B.Keisel, lift. Boston

BiOELOW. — Development of Lepas.

PLATE 7.

Figures drawn from transparent preparations of entire eggs. Vegetative pole
and blastopore at the rlrjlit side in figures seen in lateral view.

Fig. 5G. Optical section in sagittal plane. Si.xty-two cells, counting the dividing
primary mesoblasts (#■', f/**'^) as four cells.

Fig. 57. Same stage. Actual section. Blastopore not completely closed.

Fig. 68. View from vegetative pole. The mes-ectoblasts [a^-'^, Ifi-^, 6*'-*, c^-^) in
sixth cleavage, which results in forming the " secondary mesoblasts."
Blastopore slightly open.

Fig. 59. Same egg in optical section in parasagittal plane. The' primary meso-
blasts (d^-^, (/*>•*) not yet in sixth cleavage. Two entoblastic nuclei
((/<*-i, d^'^). Mes-ectoblast cells W'-^ and (••'■■■' dividing parallel to the
surface of blastoderm, to form " secondary mesoblasts."

Fig. 60. View from vegetative pole of egg in which the primary mesoblasts
(#•3, £^6.4) have not been overgrown by the blastoderm during the sixth
cleavage. These cells nearly fill the blastopore ; the posterior pair of
"secondary mesoblasts" («"'5, c^-^) lie at the sides of the primary
mesoblasts.

Fig. 61. Optical section near sagittal plane of same egg, showing anterior pair of
" secondary mesoblasts" (/*"'' and h'-') and two entoblast nuclei.

Fig. 62. View from vegetative pole of egg with fifty-six blastoderm cells, four
"secondary mesoblasts" (a'-5, h''-^, h'-^, c'-^, represented by broken
lines), two dividing primary mesoblasts (f/*'-^, d*^*, outlines shown by
fine continuous line), and two entoblast nuclei (seen at deeper level
but not figured).

Figs. 63, 64. Optical sections in horizontal plane of different eggs, viewed from
vegetative pole. Same stage as Figure 5G. Figure G3 represents a
common condition in which mesoblasts and entoblasts are not separated
by the sagittal plane.

Fig. 65. Optical section in sagittal plane of egg with sixty-two cells. The
primary mesoblasts have completed the sixth cleavage, forming d'-^-^.

BiGE LOW.- Development ofLepas.

Plate 7.

■Aiidel

BiOELOW. — Development of Lepaa.

PLATE 8.

All figures drawn from sections ten micra thick. Vegetative (ventral) pole and
blastopore at the left in views of sagittal sections.

Fig. 66. Parasagittal section of eight-cell stage, a little to the left of the sagittal
plane, and corresponding to the stage shown in Figure 40 (Plate 5).

Fig. 67. Section, in same plane, of stage with fifteen blastoderm cells ; the yolk-
cell still in the stage of fourth cleavage. This stage corresponds to
that of Figure 41.

Fig. 68. Parasagittal section of sixteen-cell stage, corresponding to that shown in
Figure 45.

Fig. 69. Sagittal section of egg with twenty-eight cells in blastoderm ; primary
mesoblast cell (d^-'^) in division ; entoblast nucleus preparing to divide.
Compare with Figures 49, 50 (Tlate 6).

Fig. 70. Horizontal section of same stage, seen from vegetative pole.

Fig. 71. Sagittal section of sixty-two-cell stage, counting two dividing primary
mesoblasts {d^'^, d'''-*) as four cells. Same age as i]igure 56 (Plate 7).

Fig. 72. Transverse section of egg in similar stage cut througli the primary mes-
oblasts and tlie posterior pair of " secondary mesoblasts " {a''-^, c'-^).

Fig. 73. Section immediately anterior to the one represented in the preceding
figure. The anterior " secondary mesoblasts " {b~'^, I?-') and the two
entoblast cells (d^'^, d^-"^) are represented.

BiGELow.- Development ofLepas.

Plate 6.

V \-cLpol.'

,/j .

cLpoL- ■

cLpoL- ■■■/.. '>

MA3.de!.

B.Kei5e;,lilli.8tston

BlOELOW. — Development of Lepas.

PLATE 9.

Figures from three sets of consecutive serial sections. Vegetative (ventral) pole
and blastopore are at the left in Figures Ti-SO and at the loiver side in Figures
81-86. Blastoderm one cell in thickness.

Figs. 74-77. Series of consecutive sections parallel to sagittal plane from an egg
in sixty-two-cell stage, counting two dividing primary mesobhists
as four cells. The first and sixth sections of this series contained
only blastoderm cells and have not been figured.

Figs. 78-80. Series of consecutive sections parallel to sagittal plane through egg
in a stage with about one hundred and twenty cells. The first and
last sections of the series are not figured.

Figs. 81-86. Series of consecutive transverse sections (viewed from their posterior
faces) from an egg in same stage as that of last series. Figure 81
shows the most posterior of the sections represented. The first and
last sections of the series, containing only blastoderm cells, and
three anterior to and similar to Figure 86 have not been figured.

BiGELow.- Development of Lepas.

Plate 9.

t !^ rr' v„.

l:iS- 01.

76

'df-3

76'

bCpo \ . ^

-d62

ms'bi

■dOJ

SO

■' n®

rl.poL

) o_a=P

C KJ

MAB.(i£l

B.KeiseUilli.BosBi;

BiGELow. — Development of Lepas.

PLATE 10.

Figures from sections. Ventral sitle (blastopore) at the left in figures of sagit-
tal sections, and at the lower side in figures of transverse sections. Blastoderm
one cell in thickness.

Fig. 87. Sagittal section of a stage with two hundred and fifty cells (estimated).
The mesoblast band (ws'W.) is extending anteriorly along the dorsal
side.

Fig. 88. Sagittal section of a later succeeding stage. Egg has elongated posteri-
orly. Continued extension of the mesoblast.

Figs. 89, 90. Transverse sections through an egg similar to the one represented
in Figure 88 and made at the levels indicated in that figure by the
numbers S9 and 90. Mesoblast dorsal in Figure 90.

Fig. 91. Sagittal section of later stage. Two transverse dorsal furrows (/, 2)
mark off the three metameres. Compare with Figure 122.

Fig. 92. Transverse section of egg in same stage as that of Figure 91, showing
the median dorsal longitudinal furrow. Tlie mesoblast has greatly
thickened and extended ventrally on either side of the entoblast. Com-
pare with Figure 90.

Fig. 93. Sagittal section of still later stage. Two new transverse furrows (S,4)
partially subdivide the first and third metameres of the previous stage.
Compare with Figures 123-125.

Fig. 94. Transverse section of stage similar to that shown in Figure 93. Longi-
tudinal furrow extending laterally and ventrally folding off the
appendages, in which process the transverse furrows 1-4 share

BiGELow.- Development ofLepas.

Plate 10.

MAB.(iel

S.KesseUiiti-Bositr,.

BiOELOW. — Development of Lepas.

PLATE 11.

Lepas fascicularis.

Tlie figures in parenthesis following the descriptions refer to corresponding
stages of L. anatifera.

Fig. 95-97. Outlines of a living egg, showing its rotation within tiie vitelline

membrane during tlie first cleavage. (Figs. G-IC.)
Figs. 98-110. Drawn from transparent preparations of entire eggs.
Fig. 98. First cleavage, spindle arranged transversely to chief a.xis of egg. (Figs.

21-23.)
Fig. 99. Second cleavage. View from animal pole. (Fig. 31.)
Figs. 100, 101. Four-cell stage from animal pole. (Figs. 32, 34.)
Fig. 102. Same from vegetative pole. (Fig. 35.)
Fig. 103. Same seen from the lejl side. " Protoplasmic " cells already in third

cleavage. (Fig. 36.)
Fig. 104. Eiglit cells. View from animal pole. Seven "protoplasmic" cells in

fourth cleavage. Yolk-cell (d^'^) retarded in division. (Fig. 39.)
Fig. 105. Same stage from left side. (Fig. 40.)
Fig. 106. Same stage viewed from vegetative pole.
Fig. 107. The divisions shown in Figure 104 as beginning are now completed.

View from animal pole. (Compare with Figs. 41, 42.)
Fig. 108. Same stage viewed from left side. Yolk-cell ((/^'^ nies-entoblast) in

fourth cleavage. (Fig. 41.)
Fig 109. Optical sagittal section of egg in same stage viewed from left side.

(Fig. G7.)
Fig. 110. Optical sagittal section of sixteen-ccll stage. Left lateral view. (Fig.

68)

BiGELow.- Development of Lepas.

Plate 11

d.poll

• i.ijoi'-:' ■

!)3

/'

III h (•/

*•' \...d.pol^ /

T^OL

10^

,/■'

/ #

l(/^

l-tAB-del.

B.Meisehliifi.BBsim.

BioELow. — Development of Lepas.

PLATE 12.

Lepas fascicularis.

Tlie figures in parentliesis following the descriptions refer to corresponding
stages of L. anatij'era.

Fig. 111. Horizontal section of sixteen-cell stage. (Compare with Fig. 43.)

Fig. 112. Sixteen-cell stage viewed from vegetative pole. Fifth cleavage. (Fig.

46.)
Fig. 113. Same stage, seen from kft side. (Fig. 45.)
Fig. 114. Thirty-two-cell stage viewed from animal pole. (Fig. 55.)
Fig. 115. Same stage seen from Icjl side. (Fig. 53.)
Fig. 116. Same stage viewed from the vegetative pole. Primary mesoblast

(rf5-2) and entoblast (ri^i) in fifth cleavage. (Fig. 48.)
Fig. 117. Egg in same stage, looking upon the posterior pole.
Fig. 118. Sixty-two-cell stage seen from left side.
Fig. 119. Same stage. Sagittal optical section seen from left side. Primary

mesoblasts still in sixth cleavage. (Fig. 56.)
Fig. 120. Same stage. Horizontal optical section seen from animal pole. (Fig.

64.)
Fig. 121. Sixty-two cells. Primary mesoblasts have completed sixth cleavage,

being now four in number (d'''^-d'-^). Two entoblasts.
Fig. 122. Profile of late stage. P'ormation of dorsal transverse furrows [1, S),

which mark off the three metameres. Seen from lejl side. (Fig. 01.)
Fig. 123. Somewhat later stage seen from left side. Appearance of a third fur-
row superficially subdividing the posterior (mandibular) metamere.
Fig. 124. Still later stage seen from left side. Another furrow subdivides the

anterior (first antennary) metamere. (Fig. 93.)
Fig. 125. Dorsal view of same stage showing the longitudinal and transverse

furrows, which, growing ventrally, fold off the appendages.
Fig. 126. Nauplius after development of paired appendages and beginning of the

labrum. Seen from the left side, ventral being up.

Bigelow.-Development of Lepa^

Plate 12.

///

//.-'

//:-'

\
\

^

J

d'^

hl'po}

,/.IV

/.

'd'pu

V

117

as

^-^'

W)

0

■ I I

- ?--\ /FO

121

f.:

,MU. r'-i

!1AB.4el

Bifesei.liifiBiisio'!

Bulletin of the Museum of Comparative Zoology
AT HARVARD COLLEGE.

Vol. XL. No. 3.

THE DEVELOPMENT OF THE DEFINITIVE FEATHER.

By R. M. Strong.

With Nine Plates.

CAMBRIDGE, MASS., U.S.A.:

PRINTED FOR THE MUSEUM.

October, 1902.

OG

1902

The Development of Color in the Definitive Feather. By R. M.

Strong.

TABLE OF CONTENTS.

I. Introduction . . .
II. Metliods and material
III. The development of the
feather . . .

A. The feather germ

B. The differentiation of the

feather . .

1. The barbules

2. The barbicels

3. The barb . .

4. The rhachis .

5. The residual cells

6. Cornification and with-

drawal of the feather .

PAGE

147

148

151
151

156
156
157

158
160
160

161

PAGE

IV.

V.

The production of color in

the feather 161

The' pigmentation of the

feather 163

Tiie chemical nature of
feather pigments . . .
The origin of pigment . .
The distribution of pig-
ment in feathers . . .
Change of color without

molt

VII. Summary 176

Bibliography 179

A.

B.

C.

VI.

163

164

168
172

I. Introduction.

The more or less striking variations in color exhibited by many
species of birds at different seasons of the year have been a fruitful
theme for discussions and speculation among ornithologists. Numerous
cases of change of color not apparently connected with the ordinary
process of molt have been reported from time to time. A theory of
change of color without molt was the subject of a rather warm con-
troversy about the middle of the nineteenth century, and there has been
something of a revival of the discussion in the last few years.

It has seemed to me that a solution of the problem could not bo
attained without a thorough consideration of the causes of color and its
development.

The present work was begun in the fall of 1899 under the direction
of Professor E. L. Mark in the Zoological Laboratory at Harvard
University. I wish here to acknowledge my great indebtedness to

VOL. XL. — NO. 3 1

148 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Professor Mark for the encouraging interest he has shown in my inves-
tigations, for helpful suggestions, and for invaluable training in precision
of method.

In the course of my histological studies on the developing feather I
have naturally examined the literature of the subject, and believe tliat
a more elaborate analysis and description of the various stages in tlie
development of the complex structure of the feather, especially of those
elements producing color, is highly desirable. This work therefore deals
mainly with the histological side of the subject of color in the definitive
feather with some contributions to the general knowledge of the
development of the feather.

II. Methods and Material.

My principal material has been obtained from the remiges of Sterna
hirundo Linn. During the summer of 1899 while occupying a table in
the laboratory of the United States Fish Commission Station at Wood's
Hole, Mass., I obtained two young birds of S. hirundo with feather
germs (" pin feathers "), some of which had begun to expose fully corni-
fied portions at their ruptured distal ends.

Immediately after killing the birds, the wings and strips of skin
bearing feathers were placed either in Kleinenberg's picro-sulphuric
mixture, or saturated aqueous solution of corrosive sublimate.

In the summer of 1900 I put up some more material of S. hirundo,
this time using Kleinenberg's picro-sulphuric fluid and the fixing
mixtures of both Hermann and Flemmi ng. T found that better pene-
tration was secured when the feather was simply pulled from the
feather follicle and dropped into the fluid, without the superfluous
tissue of the follicle and the connective tissue below the inferior
umbilicus. One soon learns to perform this operation easily and
without injury to the tissues, in spite of the fact that the latter are
very delicate at the proximal end of the feather germ.

I have found Kleinenberg's picro-sulphuric mixture and Hermann's
fluid the most satisfactory fixing agents; the latter gives by far tlie
best preservation. Kleinenberg's picro-sulphuric is especially advanta-
geous for the study of developing pigment cells, in that it leaves no stain
after proper washing, whereas osmic-acid fluids produce- a blackening of
the cytoplasm that is very objectionable in the study of early stages
of the pigment cell.

STKONG : DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 149

Material was kept in the picro-sulphuric solution for about five hours
and then transferred to 70% alcohol followed by 90%. It usuallj^ took
one to two weeks with several changes of alcohol to remove all traces
of picric acid. A fixation of three hours was found sufficient for
Hermann's fluid and the usual methods of washing and hardening
followed.

Dehydration was accomplished by immersion in absolute alcohol for
at least twenty-four hours.

For clearing and infiltration with paraffin, I have found the chloro-
form method especially satisfactory ; it was the only successful medium
for coruified portions of the feather when anything like complete series
w^ere desired. I have found it particularly good in preparing material
for sections of dry feathers. I have often secured almost perfectly
complete sei'ies with it, whereas with xylol, or cedar oil, only occasion-
ally would a section i-emain in the paraffin I'ibbon.

Feather germs were left in melted paraffiii two to five days and were
then imbedded in hard paraffin (135° F.).^ Dry feathers were, in
ordinary cases, dropped into chloroform for a few hours and then
transferred to melted paraffin for about twelve hours.

Serial sections were cut with a Minot-Zimmermann microtome 3^ to .
10 micra thick, mostly 3^ or 6§ micra. Also a few sections at the proximal
end of the feather germ were cut 2 micra thick by means of the Minot
microtome having Zimmermann's improved feeding attachment. I
found it necessary to have the temperature as low as 60° F., and each
section was cut with a very slow motion of the object carrier. For
almost all purposes, however, sections 3-^ micra thick are thin enough.

Sections of the cornified portions of the feather germ are very elastic
and tend to curl and spring from the paraffin ribbon, especially when
the sections are as much as ten micra thick, but with the methods
described above fairly complete series -were obtained.

Mayer's albumen fixative was used successfully for affixing sections to
the slide ; but with osmic-acid material it was found necessary to
spread, in addition, a thin film of celloidin over the sections, immediately
after the immersion in alcohol which followed the removal of paraffin
with xylol. This celloidin film held the sections securely in position and
did not interfere with subsequent work.

A number of stains were tried, but l)y far the most satisfactory
were (1) for material fixed in picro-sulphuric a double stain, viz.

^ A mixture of hard paraffin with about 5% of resin was suggested by Professor
G. H. Parker and was used with some success for dry feathers.

150 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Klcincnberg's 70% alcohol hacmatoxj'liu followed by eosin, and (2) for
osmic material, the iron haematoxylin as used by Heidenhain.^

Slides bearing sections of picro-sulpliuric material were placed in
the haematoxylin solution for three or four minutes only ; it was found
advisable in some cases to dilute the stain with an equal amount of
70% alcohol. The superfluous haematoxylin was removed with 70%
alcohol and then the slide was simply dipped into a jar containing
70% alcohol with a fevv drops of a sat. solution of eosin in 70% alcohol.
Cornifying tissues are stained by the eosin bright red, which stands out
in beautiful contrast with the light blue of other tissues. By this
method pigment cells and their granules are finely demonstrated. I
found, however, with material fixed in the })icrc)-sulphuric mixture a
slight tendency to shrinkage, which made it inferior to Hermann's fluid
for general histological purposes.

Material fixed with Hermann's fluid for three hours only was blackened
superficially ; this was corrected by Weigcrt's decolorizer. The iron-
haematoxylin stain was used in the usual way.

Feather germs were sectioned transversely, longitudinally, and
obliquely, and were mounted in Canada balsam. Glycerine was used in
most cases for mounting sections of dry feathers.

Teased preparations were also found very instructive, material fixed
in Hermann's fluid being especially favorable for such treatment. For
this purpose a feather germ was first split longitudinally into strips and
the epidermal portions removed from the pulp. These strips, after be-
ing stained in toto in haematoxylin followed by eosin, were teased on
the slide in balsam or xylol. Fully cornified portions were unstained
by the haematoxylin and eosin, but they retained a light brown stain
from the fixing fluid. Elements in process of cornification took an eosin
stain, which was deepest in the more advanced stages, though not ap-
pearing in the completely cornified elements. Stages preceding cornifi-
cation took the haematoxylin, as did also nuclei in cornifying portions
of the feather.

Dry feathers have also been studied in toto, and control observations
have been made on them to guard against the possibility of overlooking
a pigment that might be dissolved by the histological reagents used.
This matter will be brought up later in a cliscussion of the chemical
characteristics of feather pigments.

Besides Sterna liirundo, feather germs from Passerina ciris Linn.,

1 Picrocarminate of lithium lias been used for difTcrentiating cornifying tissues,
but I have found it inferior to the stains mentioned above.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 151

Passerina cyanea Linn,, Munia atricapilla Hume, and the common dove
have been studied ; and dry feathers from the following birds have also
been used : Cyanocitta cristata Linn., Sialia sialis Linn., Pitta sordida
Sharpe, Pitta moluccensis Swinh., Cotinga cayana Bp., and Megascops
asio Linn.

I wish here to expi-ess my thanks to Messrs. Outram Bangs and J. D.
Sornborger for aid in procuring material.

III. The Development of the Feather.
A. The Feather Germ.

Of the many accounts of the structure and development of the feather,
by far the most accurate and thorough is that of Davies ('89), who
also gave an extended review of the literature up to the time of his
writing. He studied the feather witli particular reference to its homol-
ogies with other integumentary structures, but did not consider the
question of color.

According to Davies the definitive feather is always preceded by a
down feather, — though in some cases the latter is represented by only
a rudimentary structure, — and it has the same follicle and the same
dermal papilla or pulp as the down feather. The epidermal fundament
of the future definitive feather has the same cell layers as the down
feather, except tliat the epitrichial layer is absent. In a longitudinal sec-
tion of the feather germ, it is easily seen that the cylinder-cell layer,
the intermediate cells, and the layer of cornifying cells are continuous
with corresponding layers in the epidermis of the skin.

A description of the development of color in the feather can be better
appreciated if it is preceded by an account of the various steps in the
differentiation of the barbs and barbules. The formation of the latter,
especially, is complicated, and must be explained before giving a de-
scription of the process of pigmentation.

Davies gave a good description of the differentiation of the various
parts of the feather, but his account of the formation of the barbs and
barbules, especially of the latter, is incomplete. Moreover, his prepa-
rations had evident defects in preservation, which led him into some
errors in his description of the conditions connected with the differen-
tiation of the feather fundament, which I hope to correct.

Since the portions of the feather germ near the inferior umbilicus
constantly present conditions which are younger than those of portions

152 bulletin: museum of comparative zoology.

more distal in position, a single feather presents at successive levels con-
ditions which are identical with those of a given region of a feather in
successive stages of its growth. The conditions shown in Figures 12-23
were taken from sections marked in the diagram, Figure 1, by the num-
bers 12-23, which are successively more and more distal in position.
They correspond to successively older stages in the development of a
feather germ. I begin my account of the conditions presented by the
remiges of Sterna hirundo with a description of the conditions nearer
the inferior umbilicus (12, Fig. 1).

In Figure 12 (Plate 2) is shown a portion of a cross-section just above
the umbilicus. A peripheral portion of the pulp (drm.) is shown at the
bottom of the figure. It consists of closely packed connective-tissue cells,
whose long axes are cut at right angles. Blood vessels are especially
numerous at the periphery of the pulp.

Between the pulp and the epidermis lies the so-called basal mem-
brane. This is seen most favoi'ably in preparations where decolorization
was not carried very far. I have also recognized this structure in picro-
sulphuric material, but far less clearly. Studer ('73) described as
structureless a membrane lying between the dermis and epidermis of
the feather, but later ('78, p. 425) noticed that it was cellular. Davies
('89) noted Studer's observations of a basal membrane in liis review of
Studer's work, but, in his own account, does not mention the basal
membi-ane as a separate structure. He treats of it as a part of the
connective-tissue pulp, without, however, discussing the subject.

That this structure is cellular in Sterna hirundo, is evident from the
presence of the nuclei which ai-e inclosed in it (Plate 2, Fig. 14, nl.).
There can be no doubt, moreover, that it is of dermal origin, for the
nuclei have the characteristic smaller size of dermal nuclei ; besides, a
sharper line of demarcation exists between the membrane and the cylin-
der-cell lavcr than between it and the dermal cells. The nuclei are not
abundant, but where they do occur they leave no doubt as to the cellu-
lar nature of the structure.

Proceeding distally along the fundament of the feather, the basal
membrane becomes thinner and therefore less conspicuous (Figs.
15-21).

The epidermis of the feather germ, including the feather sheath,
comprises four ftxirly well marked layers : The deepest layer, that next
the pulp, consists of a single row of spindle-shaped cells (d. ci/l.) elon-
gated in the direction of the radii of the cylindrical germ, and called
cylinder cells. Except for their blunt deep ends and their weaker stain-

STRONG: DEVELOPMENT OF COLOK IN DEFINITIVE FEATHER. 153

ing properties, these cells are in no way distinguishable from the adjacent
cells in the deeper portion of the intermediate cell layer at this level.

In his description of the cylindei'-cell layer, Davies ('89, p. 574) re-
marked tluit the typical cylindrical form is seldom seen in cells of this
layer. On the contrary, as will be seen in Figures 12-14 (Plate 2) and
21-24 (Plates 4, 5), I have found the cylindrical form a very common
characteristic of tliese cells in Sterna; however, it must be admitted
that in the region from 15 to 20, Figure 1, the cylindrical form is lost
(Plate 3, Fig. 15 ; Plate 4, Fig. 20).

The intermediate cells (cl. i'm.) occupy about owe third of the thick-
ness of the epidermis. They ai"e undergoing active proliferation, which,
as far as I have observed, is always accomplished by mitotic division.
Their nuclei, like those of the cylinder cells, are elongated in tlie direc-
tion of the long axes of the cells.

Outside the intermediate cells comes the layer of inner-sheath cells
(cl. tu. ?'.), which occupies about one half the thickness of the epider-
mis. The deeper cells of this layer are easily distinguishable from the
intermediate cells by their larger and more sphei'ical nuclei, their more
sharply defined cell boundaries, and their more or less polygonal form.
The more superficial inner-sheath cells are flattened, with their long
axes at right angles to those of the intermediate cells. Those most
superficial are cornifying to form the sheath, which at this point has
not attained to the full thickness shown in Figure 14. It is also not
separable from the follicular sheath at the level of this section.

The sheath (tn.') consists of flattened cornified cells more or less fused
together. Its finer structure has been described by Lwoff ('84). All
layers appear thicker and the cells more elongated than they would in a
section strictly perpendicular to the epidermal walls (cf. 12, Fig. 1).
At the level of the section from which "Figure 13 was made some changes
are to be noticed. The intermediate-cell layer is now easily distinguish-
able from the cylinder-cell layer and the inner-sheath cells. Though it
was possible to demonstrate cell boundaries at the stage shown in Figure
12, this could not be done for the intermediate cells at this later stage.
The nuclei are larger and more spherical. They are also more numer-
ous. The whole thickness of the epidermis is much reduced from that
of the first stage described.

A very short distance above this level we have, as seen in Figure 14,
the first evidence of the differentiation of ridges, in the form of exten-
sions of tlie basal membrane. The intermediate cells are in great con-
fusion and their nuclei are still larger than they appeared in Figure 13.

154 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

The cylinder cells are less elongated and their nuclei are also larger.
Their boundaries are not easily determined.

At the stage shown in Figure 16 (Plate 3), the cylinder cells and the
intermediate cells are completely divided into ridges by the extensions
of the basal membrane. These ridges are destined to give rise to the
barbs and their barbules.

Davies left undecided the question whether the formation of ridges
was brought about by the cylinder-cell layer invading the mass of inter-
mediate cells and dividing it up into ridges, or whether the intermediate
cells grouped themselves into ridges and thus made room for the
cylinder-cell layer to enter between successive ridges ; but he con-
sidered the latter view the more probable.

I, too, believe that the initiative in the process of ridge formation is
taken by the intermediate cells (cl. i^m.), and for the following reasons :
(1) they are evidently changing position, as may be seen in Plate 2,
Figures 12-14; (2) a tendency to group themselves is manifested in
the formation of lateral plates, which are represented in cross-section
by rows of cells (Plate 3, Fig. 16, ser. cl.).

Maurer ('95) has pointed out that there must be a very great pres-
sure upon the central pulp by the growing epidermal region with its
increasing need of space, and that this seems to result in the formation
of numerous small elevations and depressions (Plate 2, Fig. 12, crs".)
varying in size with the resistance at difteretit points. I agree with
him in considering this a factor also in the formation of ridges (Plate 2,
Fig. 14, crs.), especially in producing extensions of the basal membrane
into the epidermis of the feather germ.

As was observed by Davies, the ridges do not arise simultaneously
at any given level, but are first seen on the sides of the feather germ.
The distal portion of a ridge is formed before the proximal part, where
it joins the shaft or rhachis ; the differentiation of the barb and its
barbules therefore begins at the distal tip of the ridge and gradually
approaches the proximal insertion on the rhachis. In a single cross-
section, there will be ridges cut at various distances from their point
of union with the shaft. The sections of the ridges most distant from
the I'hachis, i. e. of those on the ventral side of the feather germ, pass
through the distal ends of ridges which will appear successively nearer
to the shaft in sections taken at more proximal points in the germ.
These relations may be more easily understood by reference to Figure 4
where ridges (crs.) in various stages of differentiation are represented by
rows of pigment cells.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 155

The common condition of asymmetry in the vane, with the barbs on
one side of the rhacliis longer than those of the other side, causes the
point where the distal ends of the ridges meet to be more or less ;it
one side of the median plane of the feather-germ (Plate 9, Fig. 41, dst^.).
A conspicuous out-curving of the two sides of the feather funda-
ment at this point is seen in a wing-feather from the dove (Plate 9,
Fig. 42, dsL).

The cylinder-cell layer, which forms a continuous sheet of cells
covering the ridge completely on the pulp side and between adjacent
ridges, takes no direct part in the formation of barb or barbule. These
are formed exclusively from the " intermediate cells," which constitute
the greater portion of the ridge. These intermediate cells become
differentiated into three parallel structures, an axial plate, longer in a
radial than in a tangential direction, and two lateral plates. A large
portion of the cells forming the axial plate are ultimately metamorphosed,
or fused together, t-o form the barb; the cells wliich compose the lateral
plates of the ridge, and which are separated from the furrows by the
cylinder-cells, are to be connected into barbules, whose attachment to
the barb will be near the inner or pulp margin of the axial plate. In
each I'idge one lateral plate will form the distal barbules and the other
the proximal barbules of a single barb.

Davies ('89, Taf. 24, Fig. 19) described and figured clefts or spaces,
which he found occurring between the plates of barbule cells and the
cells forming the axial plate. He called these spaces " Langsfurchen,"
a term which seems inappropriate for a fissure-like space, and especially
so in this case, because he uses the same word for the spaces that he
found between successive ridges. The latter could with some reason
be called furrows, but the spaces between the barbule rows and the
axial plate are nothing but artificial clefts. I have never found them
except in preparations that had experienced shrinkage in fixation. In
osmic material these clefts are altogether wanting, as are also the wide
V-shaped furrows whicli he described and figured as occurring between
ridges (Davies, '89, pp. 574-5 ; Figs. 17-19).

The growth of the cells comprising the feather fundament and the
proliferation of cells at its l)asal, or proximal, end brings about a lon-
gitudinal growth of the feather germ, the sheath preventing lateral
expansion.

Davies described this extension of the feather germ as due exclusively
to cell pi'oliferation at the base, ignoring the growth of the cells as a
factor. This is partly explained by his conception that there were

156 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

clefts (Langsfurchen) between the lateral plates and the axial plates,
lie described these clefts as being filled ultimately by tlie growth of the
cells of tlie barbule fundaments. They would thus provide room for
the expansion.

B. The Differentiation of the Feather.

1. The Barbules.

Each barbule is composed of a single series of "intermediate cells"
placed end to end, thus forming a column of cells (Plate 7, Fig. 38,
col. cL), which comes to lie nearly parallel to the feather germ, with its
own axis forming a feeble spiral. The columns of cells are so closely
arranged as to be in contact with each other by their edges. Accord-
ingly, in cross-sections of the germ many columns are cut cross-
wise, each being represented by a single cell. Tliose cells form,
in any given series, a row (Plate 3, Figs. 16, 18, ser. cL) ; those nearest
the pulp in the row are also nearest the cells destined to form the
barb. They are cut nearer the base, or attached end, of the prospective
barbules than cells which lie farther from the pulp in the row. Those
at the extreme periphery, next to the inner-sheath cells, are the ones
which are destined to form the tips of the barbules. A single row
of tliese cells in a cross-section (Figs. 16-21, set', cl.) therefore shows
conditions of development for various portions of different barbules.

By a comparison of the stages shown in Figures 16-21 and 24, it may
be seen that the deeper cells in a row undergo a great metamorphosis in
shape and size to form the broad flattened portion of tlie future barbule
(Plate 5, Figs. 25 and 26). The more superficial, and tlierefore more
distal, barbule cells become elongated to form the attenuated portion
of tlie barbule. They appear, consequently, much smaller in cross-
section than the proximal cells.

In the broad flattened cells the nuclei come to occupy a ventral
position (Plate 5, Figs. 23, 27). The boundaries between contiguous
proximal cells of a single barbule run obliquely forward from the dorsal
n)argin to a point near the ventral margin just proximal to the nuclei,
where they turn slightly backwards towards the proximal end of the
barbule (Plate 5, Figs. 26 and 27). In the region of transition from the
bi-oad flattened form to the slender distal portion (Fig. 27), the outline
of these inter-cell boundaries changes to a form presenting a convexity
in an opposite direction, ^. e. towards tlie proximal end of the barbules;
the sides of the convexity being likewise more symmetrical.

strong: development of color in definitive feather. 157

The broad cells of the proximal barbules {brh. prx., Plate 5, Fig. 23)
undergo a special metamorphosis, in which their dorsal margins are
bent over and inwards towards the axial plate to form the well-known
recurved margin (Fig. 25, marg.) to which the booklets of the distal
barbules are ultimately to secure attachment.

It should be noticed here that the barbule fundaments are not cut
exactly at right angles by cross-sections, but somewhat obliquely,
especially in their broad proximal portions.

At a very early stage in the differentiation of the barbules, the barbule
columns lie in the plane of a radius of the feather germ (Plate 3,
Fig. 16, ser. cL). They also make an angle of over 60° with the
long axis of the feather germ. With the growth of the cells composing
the barbule fundaments, this angle becomes smaller and smaller, while
the distal, attenuated portion comes to lie nearly parallel with the axis
of the feather germ.

The surface made by the barbule fundaments collectively undergoes
a bending, which is clearly seen to increase steadily from the stage
shown in Figure 16 to that of Figure 20, ser. cl. This, I think, is
brought about partly by the great increase in the size of the ridges near
their attachment to the rhachis, at the expense of tkeir distal ends,
wiiich lie farther away from the rhachis. It results from the fact that
the barbules will be largest at the proximal ends of the barbs and will
gradually decrease in size towards the distal ends of the latter. A
cross-section at a point where the ridges are first differentiated does not
show so great a contrast in size between sections of ridges near the shaft
and those on the ventral side. This increase in size must be accom-
panied by lateral displacement, which would account for the gradual in-
crease in the curvature of the rows of cells representing the barbules.

2. The Barblcels.

The barbicels arise as one or two processes of single barbule cells at a
comparatively late stage in the development of the barbule. The bar-
bicel appears first as a thick blunt projection of the cell (Plate 5, Fig,
27, brbc); its final form is not attained until the end of cornification.

The cells of the distal halves of the distal barbules arc, except for a
few of the most proximal, each provided with two distinct barbicels, —
one ventral and one dorsal (Plate 5, Figs. 26, 27, brbc). Of these the
ventral is the longer. Towards the middle of the barbule the ventral
barbicels are of considerable size, and they are more or less recurved at
their distal ends to form the so-called " booklets" or "hamuli " (haml.).

158 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

The two most proximal of the ventral barbicels (Plate 5, Fig. 27) are
smaller and without hooks.

The bai'bicels of the proximal barbules (Fig. 25, hrhc.^ are rudimen-
tary except for the two most proximal on the ventral side, which are
similar in form and size to the corresponding barbicels of the distal bar-
bules. They may be absent altogether from both sets of barbules, as is
frequently the case in the more distal portions of body coverts.

In a cross-section of the feather germ at the level of 21, Figure l,the
barbicels appear as loose irregular fragments. I have found teased prep-
arations most favorable for studying their origin.

3. The Barb.

^Between the two rows of barbule cells for each ridge, as seen in cross-
section, there is a group of cells which I have called the axial plate (la.
ax., Plate 3, Fig. IG). The cells of this plate never acquire a regular
arrangement like those of the lateral rows. At the same time it is to
be noticed that the rows of barbule cells do not extend quite to the apex
of the ridge, the apex being occupied by a group of cells (Plate 4, Fig.
20f fnd. brb.) wdiich is continuous with the axial plate. Differentiation
begins at a rather late stage.

The cells in the deeper portions of the axial plate, near the cylinder-
cell layer, become large and conspicuous and have a more or less polyg-
onal form (Plate 4, Fig. 21, vied.). They are destined to form the
medulla of the future barb.

The number of cells entering into the formation of the medulla at any
given place depends on the size of the barb at that region. Around
these medullary cells, as around an axis, other cells become applied and
flattened, so that, in cross-section, they appear spindle-shaped. These
form the cortex of the barb. In a region where the barb is large, i. e.,
near its proximal end, almost all of the axial-plate cells enter into its for-
mation.

With this differentiation the ridge experiences an extension in the
direction of a radius of the feather germ, and the diameter of the cen-
tral pulp decreases correspondingly. Before this differentiation began,
the region corresponding to the prospective barb occupied a compara-
tively small area in the cross-section (Plate 4, Fig. 19); but after the
differentiation, it occupies a large portion of the ridge (Plate 5, Fig.
23). The barbules are thereby pushed farther and farther away from
the pulp.

The structure of the medulla and cortex was early studied by

STRONG : DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 159

Schwann ('39), who gave a very good general description of them.
Since then they have been considered by various writers on the struc-
ture of the feather. I have nothing to add to the more recent accounts,
except to call attention to the venti-al ridge (crs'.) of the cortex of the
barb, which is shown in transverse section for several birds (Plate 1,
Figs. 7, 8, 9 ; Plate 5, Fig. 24), and also to the structure of the
dorsal thickened portion of the cortex (Plate 5, Fig. 23, ctx. d. ; Fig. 24,
ctx?). I find the ventral ridge, or keel, a frequent and important feature
of the ventral cortex. It furnishes a convenient "ear niarlc"for the
orientation of barb sections ; its apex in transverse sections alwaj's
points towai'ds the shaft. During the process of cornification, it be-
comes much reduced fi'om the conspicuous size which it has in stages
corresponding with that shown in Figure 23, but it still retains the same
characteristic want of symmetry (Fig. 24, c?V.).

The dorsal portion of the cortex is made up of cells which fuse at a
comparatively late date in the feathers I have studied.

Haecker ('90) described thick-walled medullary cells which he found
in the barbs of certain birds, designating them by the term " Schirm-
zellen." I have examined sections of the barbs from two of the species
of birds which he studied (Cotinga cayana and Pitta nioluccensis), and
also from Pitta sordida, and have identified his so-called " Schirmzellen"
(Plate 2, Figs. 10 and 11, d. med.)} I regret not having been able to
get material for the study of their development ; but there seems little
reason to doubt that they are modified medullary cells, as Haecker him-
self leaves one to infer.

They were observed and figured by Krukenberg ('82) in Irene puella;
he called them thickened medullary cells (" Markzellen "). Gadow
('82) saw them in Pitta nioluccensis, but his figures and descriptions
are incorrect. He described them as prismatic columns with minute
parallel ridges on their surfaces; but neither Haecker nor I have found
any ridges. Gadow seems to have depended solely on observations from
the exterior, having apparently worked without the aid of sections.

The " Schirmzellen," as described by Haecker, occur mostly on the
dorsal side of the barb immediately underneath the cortex ; but they
are also represented by two or three typical thick-walled cells on the
ventral side in Pitta moluccensis.

1 As this paper goes to press and since the printing of the plates, an article ap-
pears by Haecker und Georg Meyer ( : 01) in which the Schirmzellen are recog-
nized as modified medullary cells and are re-named " Kastchenzellen," a much more
appropriate term.

160 bulletin: museum of COMrARATIVE ZOOLOGY.

Haecker also mentioned an outer cpitrichiura covering the cortex. I
have not been able to satisfy myself that such a layer actually exists.
There are appearances suggesting an epitrichium, but these I regard as
purely optical effects.

Haecker's figures of transverse sections of barbs are, with few excep-
tions, the only ones that I have found approaching accuracy in detail,
and even his are sometimes confusing. I have therefore prepared figures
showing in detail cross-sections of barbs from different birds, though
several of them have been figured before. The figures given by Jeffries
('83) for transverse sections of barbs are almost worthless, but their
crudity is probably largely explained by the lack of a suitable technique.

The cortex in a cross-section of a barb from Megascops asio, which
appeared in an otherwise beautiful plate published by Chadbourne ('97),
is wholly erroneous.

4. The RhacMs.

The shaft, or rhachis, arises on the dorsal side of the feather germ
and represents two or more combined ridges (Plate 1, Fig. 2 ; Plate 9,
Fig. 42, rch.) ; its structure is, in general, like that of a barb with a
central medulla of polygonal cells and an outer thickened cortex. It
also bears barbules like those of the barb, between the points of inser-
tion of the latter, on its sides. The development of the rhachis was
carefully studied by Davies, to whose account I have nothing to add.

5. The Residual Cells.

As has already been stated, not all the cells of the ridge are employed
in the formation of the barbules and barb. With the growth of the
ridges, the layer of cylinder cells is pushed closely against the corre-
sponding layer of the neighboring ridges, and these cells (Plate 3, Fig.
16, cl. cyl.) still continue to be so crowded in the layer that their nuclei
appear almost to touch each other ; but with the great longitudinal ex-
tension of the germ, due to the growth of the barbs and barbules, in
which the lateral cylinder cells do not share, the cylinder cells become
more and more spread out (Plate 4, Fig. 19, cl. cyL, Figs. 20-21).
The inner-sheath cells also experience a contraction during the growth
of the feather. In Figure 23, Plate 5, "the elements of the feather
proper have been shaded. Residual cells are scattered through the
more superficial spaces not occupied by the barbules. Tiieir nuclei are
shrivelled. The deeper cells, including the cylinder cells, retain their
regular form and size until a later stage.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 161

6. Cornijication and Withdrawal of the Feather.

With cornification, the barb coi'tex differentiates from the surround-
ius; tissue and the outhnes of individual cells become less and less evi-
dent, until, finally, in the fully cornified barb there is little or no
evidence of its former cellular nature. The nuclei of the barbule cells
shrink, and the last seen of them is a small glistening mass of shrivelled
chromatic substance, which finally disappears along witli all traces of
cell boundaries. Nevertheless the former position of tlie nucleus can
frequently be distinguished, through the different refractive properties
of this region. The barbule thus becomes a horny, almost homogeneous
body with no evidence of its original cellular structure, except such as is
furnished by the position of the barbicels, the nuclear region, and the
presence of pigment patches, to be discussed later.

Toward the end of tlie process of cornification the feather elements
withdraw or shrink away from the non-differentiated cells, which them-
selves become more or less shrivelled and cornified (Fig. 21, Plate 5).
After the completion of cornification, the feather begins to break forth
from the distal end of the feather sheath, a process that begins and con-
tinues some time before the formation of the calamus takes place. The
barbules, on escaping from the confining sheath, swing about by their
own elasticity from the position shown in Plate 1, Figure 6, to tliat
seen in Figure 3.

The process by which the pulp atrophies, having been well described
by Davics, will not be discussed here. In the completed feather, as is
"well known, all that remains of the dermal pulp is tlie series of dry
horny caps found in the quill and a small functional papilla, whicU pro-
jects slightly up into the quill through the inferior umbilicus. At the
time of molt, this papilla is destined to become active again in the
formation of a new feather.

The cornification of the feather elements has been described by Wald-
eyer ('82) and Lwoff ('84).

IV. The Production of Color in the Feather.

The researches of Altuin ('54, '54"), Bogdanow ('58), P>rucke ('61),
Gadow ('82), Krukenberg ('84), and Haecker ('90) have shown tliat the
colors of birds may in general be divided into two classes, (1) those due
simply to the presence of a pigment, and (2) the so-called structural
colors. Under simple pigment colors they have placed rod, yellow,
orange, black, and brown; whereas white, gray, blue, the so-called metal-

VOL. XL. NO. 3 2

162 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

lie colors, iridescent phenomena, find lustre are called structural coloi-s.
According to Haecker, green is a structural color except for tlio single
case of turacoverdin, a pigment described by Krukenberg ('82).

Tiie production of structural colors has been variously explained as
due to either (1) light-interference phenomena or (2) diffraction or dis-
persion of light-rays. Except for white, however, a dark granular pig-
ment (melanin) has always been found associated with such effects.

Peculiar modifications in structure are associated with blue colors.
Altum ('54'^ ) observed that feathers giving bright blues have the barbs
isolated, i. e., not connected with each other by barbules.

Haecker ('90) considered as necessary for tlie production of blue : (1)
a thickened unpigmented cortex, (2) a deposit of brown pigment in the
medullary cells of the bai'b, and (3) the occurrence of more or less poly-
gonal, porous-walled " Schirmzellen."

I have examined blue feathers from the indigo bird (Passerina
cyanea), the blue-bird (Sialia sialis), Pitta sordida, Pitta moluccensis,
Cotinga cayana, and tlie blue-jay (Cyanocitta cristata). The brilliant
blue feathers furnished by Pitta and Cotinga have the barbules rudi-
mentary or of insignificant size where the color is most intense. The
lateral diameter of the barb is also greater than in the more proximal
and less brilliant portion. Such feathers never appear blue except
when seen from above. Their ventral surface gives a dull brown color.
The "Schirmzellen" are conspicuously developed (Plate 2, Figs. 10-11,
cl' . vied.).

The cavities of the ordinary medullary cells have a thick peripheral
layer of dark brown pigment. In Cotinga I found no ordinary medul-
lary cells, but the ventral cortex was thickened and appeared black from
a rich supply of pigment.

Blue feathers from the blue-jay, blue-bird, and indigo bird show no
" Schirmzellen," but there is a pigmentation of the central medullary
cells (Plate 1, Figs. 7-8, med.) similar to that observed in the Pittas
(Plate 2, Fig. 11).

The distal portions of blue feathers from the blue-bird which I exam-
ined gave a much more brilliant blue than the proximal portions. The
transition from bright to dull blue was abrupt. "With the aid of a mi-
croscope, it could be seen that a light blue color of uniform intensity
was given by the barbs in both proximal and distal portions. Where
the feather appeared hright blue, the barbules were absent. A similar
relation between brightness of color and the absence of barbules has
been noticed by other writers f:)r otlier birds.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 163

A variation from the conditions described by Haecker for the pro-
duction of blue is found in the blue feathers of the indigo bird. I have
never seen any pigment in the medullary cells, but heavily pigmented
barbules occur and they are not reduced in size (Plate 5, Fig. 29).

A section of a barb from the dark brown tertiaries of the " homer "
pigeon shows little, if any, more pigment than is found in gray
feathers of Sterna (cf. Plate 1, Fig. 9, and Plate 5, Fig. 24). The
distal as well as the proximal barbules are libei'ally sup;)lied with
brown pigment, however; whereas in Sterna, only the more proximal
portions of the distal barbules have an appreciable amount of pigment.
The wing feathers of the juvenal plumage vary from plain gray to
brownish gray. When the latter color occurs, there is a noticeable
pigmentation of the proximal barbules.

V- The Pigmentation of the Feather.
A. The Chemical K'ature of Feather Pigments.

The researches of Bogdanow ('56, '57) and Krukenberg ('81-'84)
have shown that the pigments of birds' feathers may be divided into
two groups: (1) those soluble in alcohol and ether, — yellow, orange,
and red pigments (also a single green pigment, turacoverdin) ; and
(2) those soluble in acids and alkalies, — the dark brown to black
pigments.

Krukenberg ('8^) designated the first group under the general terra
of lipochromes or fat pigments. The second group is included among
the widely distributed dark brown animal pigments known as melanins.

The solubility of tlie lipochromes in alcohol and ether renders the
study of their origin in the feather by-ordinary histological technique
impracticable. I have found, for instance, that yellow feather germs
from the canary and from the nonpareil (Passerina ciris), though re-
taining their color after fixation, lose it in all except the cornified
portions during the process of hardening in alcohol. Various writers
■who have alluded to tlie origin of pigment in feathers have described a
melanin pigment, but they usually fail to recognize that the melanins
are not the only pigments present in feathers.

The dissolving action of chemical re-agents on the melanins of differ-
ent animals has been described difterently by various authors, but, in
general, a great resistance to acids and alkalies has been found.
Alcohol, ether, chloroform, xylol, etc., seem to have no action whatever

164 bulletin: museum of comparative zoology.

on them. I have had material in alcohol for months without any
apparent effect on melanin granules. It is not inconceivable that
histological re-agents may produce chemical changes in the developing
melanin granules, but I have had no positive evidence of any such
alterations.

Especially to bo noticed is the red pigment turacin, which was
described by Church (*69, '93) as containing 7.1% of copper. Feathers
containing this pigment are said to give a red color to water in which
they may be placed. At the same time, there is more or less of a
tendency for such feathers to exchange their normal red color for blue ;
but the red returns when the feather is dried. Church found turacin
easily soluble in water, especially if the latter was slightly alkaline.

B. The Origin of Pigment.

The many writers on the origin of pigment in epidermal structures
may be divided into two groups : (1) those believing in an exogenoi^s
formation of pigment, and (2) those who argue for an endogenous or
autocthonous development of pigment in the epidermis.

The theories ascribing an exogenous origin to pigment all involve a
more or less direct relation of pigment to the blood. Most prominent
is that which derives the melanins from the haematin of the red blood
corpuscles. Certain writers have argued that pigment originates in
internal organs, from which it is transported to the integument either
in solution in the blood plasma or as a colorless mother substance in
the blood-cells. Closely allied to this is the excretion- (or waste-)
product theory advocated by Eisig ('87) and others for invertebrates.
Finally, there is the leucocyte theory, which makes leucocytes the
bearers of pigment from the blood to the epidermis.

The writers who have argued for an endogenous formation of pigment
in the epidermis believe that pigment results from the metabolic
activity of either the nucleus or the cytoplasm of epitlielial cells.

Among those who have advocated an exogenous origin of the pigment
of epidermal structures are Langhans ('70), Gussenbauer ('75), Kerbert
("76), Riehl ('84;, Aeby ('85), Quincke (,'85), Ehrmann ("83, "91, '92),
Kolliker ('87), Karg ('88), Phillipson ('90), Kaposi ('91), and Bloch ('97).

The following have supported tlie endogenous origin : Demii^ville ('80),
Krukenberg ('84), Mertsching ('89), Jarisch ('91, '92), Kabl ('94),
Post ('94), Rosenstadt ('97), Loeb ('98), and Prowazek (:00).

Pigment may be present either, (1) in the dermis only, (2) in the

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 165

epidermis only, or (3) in both. Most writers who advocate origin from
the blood have described pigment as being formed in the dermis, either
in ordinary connective-tissue cells, or in special cells differentiated for
the purpose, which in the case of epidermal pigmentation wandered
from the dermis into the epidermis or sent amceboid processes up be-
tween the cells of the cylinder-cell layer.

I have found the remiges of the tern (Sterna hirundo) especially
favorable material for studying the formation of epidermal pigments.
Their pigment cells attain a large size, are comparatively regular in
contour, and very abundant.

The first signs of pigment formation appear in certain of the " inter-
mediate cells '.' of the fundament of the feather immediately before the
differentiation of the ridges. The pigment arises in the form of grayish
or light yellowish corpuscles, of exceedingly small size, arranged along
delicate protoplasmic strands, which radiate from the nucleus and
sometimes anastomose more or less with one another. These corpuscles
increase rapidly in size and are soon large enough to be recognized with
a -^ inch oil immersion lens as definite rod-shaped granules (Plate 6,
Figs. 30, 31). At the same time they become deeper in color and
more and more numerous until finally they form a complete ball,
Plate 3, Fig. 16 ; Plate 6, Fig. 35, cl. pig-), which was often taken
by the earlier writers to be a homogeneous mass.

In the course of development these rods are easily seen to be radially
distributed about the nxacleus, an arrangement which has been described
for the pigment cells and chromatophores of other animals.

The nuclei of these pigment cells are entirely destitute of the pig-
ment granules, a condition which Solger ('89, "90, '91) also noted in
the pigment cells of fishes and mammals.

Kromayer ('97), too, observed in the developing chromatophores of
frog skin that the first appearance of pigment granules was along proto-
plasmic strands ; the granules were at first light in color, but gradually
grew darker.

Post ('94, pp. 4:91, 492) found that melanin pigment granules have
characteristic variations in shape and size for different animals. " Die
Pigmenttheilchen in den Oberhautgebilden verschiedener Thierarten
sind ebenfalls sehr verschieden, z. B. bei der Katze lang nnd ziemlich
dick, beim Hunde wetzsteinforming in der Mitte verdickt, beim Meer-
schweinchen und Kaninchen kurz und dick, beira Rinde ziemlich lang
und schlank. Auch das Pijrment der Taubenfedern besteht aus Stabchen
von massicjer Grosse." I have also found variations in size for the birds

166 bulletin: museum of COMrAUATIVE ZOOLOGY.

I luive studied, but pigment rods when fully formed, i. c, at the stage
indicated in Figure 36 (Plate 6) ai"e of uniform size for each species.
The peculiar rod-like appearance and also the size ai'e indicated in Figure
36 (Plate 6), which was drawn with a magnification of 1500 diameters.
I have found the pigment rods of Sterna invarial)ly as near to 2 niicra
long as I could measure, and about one-third of a micron in diameter.
The shape does not seem to vary noticeably in different snecies.

In the following species the rods are of practically the same size as in
Sterna: Passerina ciris, P. cyanea, and the "homer" pigeon. In the
common dove (reddish-brown feather) the length is onlj'^ 0.9 p..

I iiud myself in entire agreement with Post ('94) as to the origin of
melanin in feathers. At no time have I found pigment in the pulp.
The pigment cells, moreover, have alwaj's been separated from the pulp
by the cylinder-cell layer and the basal membrane, so that there could
be no question of misinterpretation as to the place of the pigment
granules. Habl ('9*) has made the same observation on the down
feathers of the chick.

I have examined many preparations, at stages both preceding and
accompanying the formation of pigment cells, for evidence that leuco-
cytes enter the epidermis. Although leucocytes are to be found in the
blood capillaries close to the basal membrane, I have not seen a single
case suggesting actual invasion of the epithelium by them or by any
other form of cell. It may be objected that because my preparations
did not catch wandering cells at the moment of their entering the
epithelium, I have not sufficient ground for denying that they ever pen-
etrate. Even granting the force of this contention, we still should
have a right to expect transition stages in the form of the nuclei from
that of typical leucocytes to that of pigment cells, but such intermediate
stages I have never been able to find. Furthermore, if there were an
immigration of prospective pigment cells, or melanoblasts, from the pulp,
it is reasonable to suppose that at the earlier stages of the development
of pigment the cell would be comparatively near to the cjdinder-cell
layer ; but there is no evidence that such is at any time the condition.
In order to have something more definite than a general im{)ression on
this point, I have noted the distances of pigment cells from the pulp at
various stages in their development, and for this purpose have divided
the cells into four groups. The following table gives the results of
these measurements. Group A includes the youngest stages, those
represented in Figures 30-32 (Plate 6) ; B, those shown in Figure 33 ;
C, those in Figure 3-t ; and D, those in Figure 35. The table gives

strong: development of color IX DEFINITIVE FEATHER. 167

the number of cells of each group found at the indicated distances
from the basement membrane.

10 m

15 m

20 m

25 m

30 m

35 m

40 m

45 m

Total

A

3

4

11

3

8

8

6

2

45

D

2

0

4

2

9

G

3

26

C

2

4

9

8

14

18

3

1

54

D

1

5

12

9

22

12

5

66

8

13

36

22

53

39

17

3

191

The measurements given in this table show that there is no no-
ticeable correlation between the position of pigment cells and their stages
of development. Moreover in stages later than those of Group Z), the
pigment cells come to occupy a position very close to the pulp, seeming
in some cases to migrate towards rather than away from it.

It would be absurd to deny all physiological relation whatever of the
melanins to the blood, since the whole feather germ is of course depend-
ent on the blood for nourishment.

I have observed that the nuclei of pigment cells lose stainable chro-
matin, as described by Jarisch ('92), and it is only reasonable to sup-
pose that the nucleus must sliare to some extent in the profound
changes that take place in the pigment cell. The first visible pigment
elements appear, however, in the cytoplasm, and it seems probable that
the pigment rods are formed from cytoplasmic material.

Against the hypothesis that pigment is an excretion product, may be
urged the striking variations in amount of pigmentation for djfferent
animals, where there is no reason to believe that corresponding differ-
ences in excretion occur. Albinos lack entirely melanin pigmentation
in integumentary structures, yet no one would deny that they have
normal excretory processes. Then, too, such a theory requires, as Kru-
kenberg ('84) has said, a marvellous selective power on the part of the
pigment cells, and it is more difficult to conceive of this than it is to
imagine that certain cells manufacture from a common nourishing
material the pigment granules that are to be supplied to neighboring
cells.

168 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

C. TuE Distribution of Pigment in Feathers.

When the pigment cells or chromatophores have reached the stage
represented in Figure 35 (Plate 6), they send out processes (Plate 3, Fi(^
18, pre.) which take a sinuous course among the cells of the axial plates
and at length approach the cells of the future barbules wiiich are to be
pigmented and in some way distribute pigment to tliem. The form of
these processes varies in the feather germs of different species. In Ster-
na hii-undo they are especially regular and well defined. These pig-
ment-cell processes usually branch one or more times, and they are
frequently swollen or beaded at the points of branching (see Plate 7,
Figure 38, cl. pig.).

I have studied many preparations to ascertain whether the cell
wall of the pigment cells grows out in the form of a process the exist-
ence of which can be shown by any other evidence than these rays
of pigment granules. I have also endeavored to see whether there is
a flow of pigment granules inside the process. In preparations fixed
in Hermann's fluid and stained in iron haematoxylin there ai'e fre-
quently appearances suggesting the existence of regions in the processes
which are not completely filled with pigment. In Figure 18 pre'.
(Plate 3), I have shown such a condition, the process seeming to lack
pigment granules for a short distance near its proximal end. This sup-
position is further strengthened by the presence of a loose arrangement
of the pigment rods at each end of the region apparently free from pig-
ment, as though there were here a transition to the closely packed con-
dition. Ordinarily the pigment process appears as a sinuous limb of
the cell which contains pigment rods packed together so closely as to
be indistinguishable from one another and gives no evidence of possess-
ing an enclosing membrane.

Post, ('94, p. 497) gave the following mechanical explanation for the
production of these ramifications of feather pigment-cells. "Bis diese
Zellen [Barbule cells] zu verhornen beginnen, bleibt jenes vorrathige
Pigment in den verzwcigten Zellen aufgcspeicliert und wird erst all-
mahlich dorthin iibergefiihrt, ein Vorgang, der durch mechanische Mittel
wie den Wachstumsdruck der umgebenden Zellen, die wechselnde Blut-
fiille der Pulpa, Zugwirkung der Musculatur des Federbalges hinreichend
erkliirt werden kann."

In the case of the dove, the pigment-cell processes are so irregular in
form that it is easy to see how Post was led to such a conclusion. In
Sterna and Cyanea, however, we have processes whose contour does not

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 169

suggest a simple mechanical cause (Plate 3, Figs. 17, 18, and Plate 7,
Fig. 38). They are more uniform iu diameter than those of any
dove which I have observed, and they frequently branch in a manner
that is very characteristic of chromatophores, whose processes are un-
questionably the result of cell outgrowths.

The transfer of the pigment granules contained in the processes of the
pigment cells to the barbule cells is even more diflBcult to explain. Ac-
cording to Post it does not take place until after coruification has
begun.

Riehl ('84) thought that in the case of the pigmentation of hair, the
cornifying cortex cells of the hair might take up the pigment granules
brought to them by the pigmeut-cell processes in much the same way
that an amoeba engulfs particles of foreign substance. Against this hy-
pothesis Mertsching ('89) objected that the hair cells are motionless
and show no amoeboid movements. I have found that the form of the
barbule cells when they receive pigment is conspicuously uniform and
constant (Figs. 17, 18, and 19, ser. cL), with no suggestion of amoeboid
movements.

Another explanation was suggested by Post ('94, p. 494), — that the
barbule cells of the feather fundament might i-eceive pigment by a pro-
cess of osmosis, which would sweep the pigment I'ods iu through pores
in the cell walls. " Auf diesen Befunden darf man schliessen, dass die
grossen Pigmentzellen ihr Pigment allmahlich in jene Nebenstrahlen-
zellen tiberfiihren, und dass diese letzteren erst auf einer gewissen Stufe
im Verhornungsprozesse das Pigment aufuehmen. Dieser Vorgang
dtirfte am einfachsten erklart werden durch die Annahme, dass die Ober-
flilche der verhornenden Zellen porose werde. Die Pigmentstabchen
werden vermoge des osmotischen Austausches in die Zellen eiugesch-
wemmt und in den Maschen des Protoplasmas festgehalten."

In Sterna, the pigment-cell processes come in contact with the bar-
bule cells (Figs. 17, 18, 19, and 36) on their dorsal margins; at such
points pigment rods are found in the cytoplasm of the barbule cells,
mostly dorsal to the nucleus, where they i-emain permanently. The
barbule cells of other birds, so far as I have observed, are supplied with
melanin in a similar way, but they may have their cytoplasm packed
witli pigment on all sides of the nucleus. The pigment-cell processes
may branch so as to supply a group of barbule cells, as is shown in Fig-
ure 38 (Plate 7) for the Indigo bird, Passerina cyanea.

A question naturally arises as to the factors which determine the
direction taken by the pigment-cell processes and cause them to go to the

170 bulletin: museum of comparative zoology.

particular cells which are to be permaneutly pigmented. It seems not
impossible that a condition of chemotaxis exists between the cells which
are to receive pigment and the pigment-cell processes.

A unique theory has been advanced by Kromayer ('97) for the cliro-
matophores of the frog's epidermis. He considers the chromatophore to
be something more than a simple cell ; it has a cell at its centi-e, but it
includes parts of numerous other ei^ithelial cells lying near it. It may
be that in the case of the feather we have an actual connection between
the pigment-producing cell and the cells which receive pigment. These
united cells might, for the time being, be considered an organ in the
sense of Kromayer's hypothesis. However, the short duration of such a
condition for any particular cell makes such an explanation improbable,
even if connection actually occurs.

The pigmentation of the differeut cells in a barbule is accomplished
by a distribution of pigment rods, accompanying the growth of tlie pig-
ment cell processes, such that the more peripheral barbule cells receive
pigment later than those nearer the pulp. In the case of Sterna the
pigment found in the barb is the last to be distributed.

As we have already seen, the barb develops much later than its bar-
bules, and with its differentiation the undifferentiated epithelial cells
near the basal membrane are shoved farther and farther inwards and
away from the barbule fundaments, as can be seen in transverse sections
(Plate 4, Figs. 19, 20, and 21). This separation breaks the continuity
of the pigment-cell process, and the main mass of the cell becomes
widely separated from the pigmented barbule cells. The pigment seen
in the dorsal cortex of the barb in Sterna (Plate 5, Fig. 24, ctx.) seems
to come from the more proximal portion of the pigment-cell process,
which is now some distance away from its original position.

I have tried to determine whether all of the ' pigment borne in the
processes is taken up by cells of the feather germ, but though this is
probable, I am unable to state it positively. Neither can I deny that
there is a free formation of pigment in barbule cells independently of
that supplied by the pigment cells, as was supposed by Klee ('86).
However, I have not been able to discover any evidence of such a con-
dition, and the fact that there is a copious supply of pigment by the
pigment cells makes Klee's supposition improbable.

It is interesting to note that the amount of melanin produced is not
always correlated with the darkness of the feather, even in the case of
simple pigment colors. If a preparation such as is shown in Figure 4
be examined under low magnification, we see, in the case of Sterna, a

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 171

field of numerous dark bodies a short distance above the inferior um-
bilicus ; these are developing pigment cells. They soon become more
conspicuous and pass abruptly into regularly arranged massive black
rows, corresponding to the differentiating ridges. The whole inner sur-
face from tiiis point to the distal end appears almost continuously black,
except for very narrow spaces between the ridges and the sparsely pig-
mented region in the ventral side of the feather germ. If, however, we
take a similar preparation from a dark brown feather of a dove, we find,
instead of dense rows of pigment cells, a comparatively sparse and
inconspicuous distribution of the latter along the ridges. A cross-
section of a stage when the barbs are differentiated shows that the
pigment cell has given up all of its pigment to the feather funda-
ment and that nothing remains of it except the nucleus (Plate 9,
Fig. 42).

In tlie nonpareil (Passerina ciris) there are enormous pigment cells
which also give up all of their pigment contents to the barbules (cf.
Fig. 40, Plate 8 and Fig. 41, Plate 9). Here is seen a heavy pigmen-
tation of long barbules, which requires a large supply of pigment.
Likewise, in the indigo bird (Passerina cyanea) all of the pigment
formed is used by the feather.

The persistence of a surplus of pigment in the main body of the
pigment cell, which I have described for Sterna, seems to have been
observed by Haecker ('90) in the feather germ of Scolopax major. I
have found the distal portions of barbs, with their barbules, which are
developed on the ventral side of the feather germ to be unpigmented.
Pigment cells occur in this region, however, making an almost complete
circle of pigment cells about the pulp, as seen in cross-section. By
this arrangement the series of pigment cells (Plate 1, Fig. 4, crs.)
belonging to each ridge is continued to the distal end of the ridge
on the ventral side of the feather germ. The pigment cells in the
distal portions of the ridges, where the feather is not to be pigmented,
are smaller, however, and less numerous ; and they do not branch nor
give up any of their pigment.

This development of pigment in excess of what is used by the feather
fiuidament I am inclined to consider as of some phylogenetic importance,
for it may indicate ancestors whose feathers were much more heavily
pigmented.

I have examined white feathers from the dove, and, like Post, have
found no pigment.

In the barbules of the completed feather, the rods of melanin are

172 BULLETIN : MUSEUM OF COMPAIUTIVE ZOOLOGY.

arranged parallel with the axis of the barbule (Plate 5, Figs. 26, 27),
a condition for which I luive no explanation.

The variations in pattern exhibited by a single feather, in the form
of bars, spots, etc., are easily correlated with variations in the distri-
bution of pigment in the corresponding regions of the feather germ.

That the distribution of lipochrome pigments to the feather funda-
ment takes place at about the same stages in the development of the
feather as that of the melanins, seems certain. Tiie germs of yellow
feathers from the canary and the nonpareil show a yellow color which
corresponds in position to the dark color of feather germs pigmented
with melanin.

VI. Change of Color without Molt.

The changes in color claimed by many writers to occur without molt
may be grouped under two heads : (1) the destructive, and (2) the con-
structive. Under destructive changes are included the results of
abrasion and physical disintegration. Constructive changes include
supposed regeneration and rearrangement of pigment.

For a review of the general literature of change of color without molt,
the reader is referred to Allen ('96). More recently Meerwarth ('98)
has claimed that change of color without molt occurs in the tail-
feathers of cei'tain Brazilian Raptores. He describes variations in color
pattern that he has observed in material consisting mostly of skins.
His paper gives no satisfying evidence that the changes alleged may
not have taken place through irregular molting. Furthermore, he does
not offer any explanation of the process of change.

Descriptions of repigmentation have been mostly pure speculation.
Within a few years the following remarkable explanation of the pig-
mentation of the feather has been given by Keeler ('93) : " Pigment is
a definite chemical substance which travels through the various l)r:inches
of the feather, advancing farthest and most rapidly along the lines of
least resistance and accumulating in masses where the resistance is
greatest. Now the pigment cells must reach the various parts of the
feather by way of the shaft, and we should a priori expect to find tliat
the resistance would be least down the shaft. It might spread out a
very short distance on the barbs, but the main tendency would be
towards the tip. This would produce a streaked feather as the most
primitive form."

Still more recently Birtwell (:00), in arguing for change of color with-

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 173

out raolt in Passerina cyanea, described a process of rearrangement of
melanin granules as follows: " The rhachis appeared, centrallj^, to be
cellular in construction with an enveloping sheath thickly supplied with
the black pigment matter, the granules arranged in an order suggestive
of a streaming movement towards the tip of the feather. The stream-
ing movement of the color granules is now especially prominent in an
actively changing feather, and it readily appears that the rhachis gives
up a part of its matter to the barbs, which in turn supply it to the
barbules. A positive change of pigment is manifested macroscopically,
for a fall feather held to the light or crushed remains yellowish in its
yellow-colored parts, while a spring feather, appearing entirely blue, so
treated, shows darkly, due to the addition of black pigment."

This idea of a streaming movement was probably suggested by the
regular longitudinal arrangement of pigment rods in the cortex.

An anomalous case is that of the pigment turacin which was described
by both Church and Krukenberg as leaving the feather when the latter
is placed in water. Krukenberg mentioned a regeneration following the
drying of the feather.

Fatio ('66) attempted to prove that pigment may dissolve and spread
in the feather. He placed a feather so that the proximal portion of the
calamus was immersed in a carmine solution and observed an ascent of
the latter in the feather structure as far as the first few barbs. He also
noticed that when a feather is immersed in ether, the latter may pene-
trate to the medulla of the barbs.

Chadbourne ('97) argues for a so-called vital connection of the feather
with the organism, " The mature feather (z. e., one which has reached
full functional development) is fir from being ' dead and dry,' a for-
eign body no longer connected witli the vital processes of the rest of the
organism, as has sometimes been asserted ; for during its life it receives
a constantly renewed supply of fluid from the parts around it. In strong
contrast to this is the really dead feather, in which the fluid matter is
deficient, as, for example, the majority of cast-off feathers. Some of the
evidence in support of these flicts maybe of vital interest: — (a) The
fatty or oil-like droplets on the surface of the feather can be shown by
micro-chemical tests (staining, etc.) to be some of them identical witli
the oil from the so-called 'oil-gland;' while others are totally unlike
that secretion ; and these latter are alone found extruding from the
pores on the surface of the rami, radii, and shaft. The poi'es, some with
drops of varying size issuing from them, show best at the distal ends of
the segments of the downy rays, (b) In the living bird the imported

174 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

fluid can be colored, its progress noted, and the feather stained intra
vitam. Soon after death this becomes no longer possible. To see the
stain the microscope is usually necessary. Call this ' osmosis,' ' capil-
larity,' or what you please, it is none the less a vital process in that it
ceases soon after death, and must be studied in the fresh featlier.
(c) The broken tips of the rays forming the vanes are, when ficsh,
capped by a mass of the fluid, which has escaped, leaving tlic part
immediately below the stump pale from the loss of the fluid pigmented
matter, (d) In museum skins this fluid matter gradually dries and by
its consequent increase in density, and that of the feather tissue, tlie
colors darken : while the freshness and gloss of life disappear, (e) 'I'lie
evanescent tints of some species, — notably the fading of the rosy
' blush ' of some of the Terns, soon after life is extinct, is due to the
drying up or escape of this fluid, while the lost tint was due to the
physical effect of structure, the shrivelling and change of form would
act on the light rays and the former colors would be lost in conse-
quence. Comparisons of specimens of Sterna paradisea, S. dougalli,
and other Terns in my collection, showed that examples having the
'blush ' most marked are those in which the feathers are least drj'."

Cliadbourne ('97 °) has described the case of a canary ^ which was sup-
posed to have changed under the influence of being fed with red pepper
to the reddish yellow color which, as is well known, may be pro-
duced at the time of molting. It was clearly demonstrated by Sauer-
mann ('89), however, that in the birds experimented on by him the
color is not altered unless the special feeding is carried on while the
feathers are in process of development. This I have found to be also
the testimony of bird fanciers.

Though it is probable that the oil supplied by the uropygeal gland is
a factor in the production of color effbcts^ especially in giving gloss or
lustre, it is unreasonable to suppose that the feather itself produces or
gives forth any of the oil found upon it. Although the feather struc-
ture is slightly permeable by liquids, as Fatio observed, it does not fol-
low that the pigment imbedded or diff'used in its horny substance is able
to flow about.

There is no satisfactory evidence of the occurrence of repigmentation.

1 Dr. Chaflbourne has explained to me tliat tliere was a misunderstanding in
the case of the canaries he mentioned. They were not kept by him, but were in
the possession of tlie janitor of the Harvard Medical School, wlio tells me that the
changes mentioned by Dr. Cliadbourne were produced only by feeding at the time
when the feathers were developing.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 175

The number of supposed cases was greatly reduced when it was discov-
ered that more than one molt may take place in a year, and the recent
researches of Chapman ('96), Dwight (:00, :00 '), and Stone ('96 and
-.00), which I can corroborate from my own observations on caged birds,
have shown that partial molts may take place at various times during
the vear. Changes due to such partial molts seem sufficient to account
for all forms of color change hitherto attributed to a process of repig-
mentation.

I iiave found no good record of actual solution by natural causes of
pigments contained in the feather except in the case of the pigment
turacin. In the great majority of cases, artificial solution is accom-
plished by chemical reagents with great difficulty. Even if pigments
were dissolved in the feather, it is inconceivable that they should be re-
distributed to form the exceedingly constant and often complex patterns
characteristic of bird feathers.

Pigmentation takes place, as has been shown, at a very early stage in
the differentiation of the feather, when the cells composing its funda-
ment are in an active condition and in intimate relation with sources of
nutrition. In the case of melanin pigments, there are bj-anched pig-
ment cells which supply pigment in the form of rod-shaped granules
directly to the feather fundament. The contention for a flow of pig-
ment from the barbs into the barbules, etc. (Keeler), is at once made
absurd by the fact that the barbules are pigmented before the barbs are
differentiated.

Variations in color patterns are easily correlated with variations in
the distribution of pigment in the early stages of the feather's develop-
ment. When completed, the feather is composed of cells which have
been entirely metamorphosed into a firm horny substance and its
pigment is imbedded in that lifeless matter. The cells composing a bar-
bale are fused into a solid, more or less homogeneous structure. The
pigment of one portion of the barbule is as effectually isolated from that
of another as is the coloring of various parts of a piece of agate. Like-
wise in the barb and rhachis, pigment is definitely and permanently
located either in the solid cortex or in effectually separated cells of the
medulla; and there are no pores large enough to admit the passage of
melanin granules. The characteristic longitudinal arrangement of
melanin granules, which one finds at the close of cornification of the
feather, is permanent.

The case cited by Krukenberg of a regeneration of the pigment tura-
cin was unfortunately not described. It seems to me probable that the

176 bulletin: museum of comparative zoology.

reappearance of the normal color after drying was not due to any true
regeneration, Init to the fact that upon drying a pliysical change had
taken place in the pigment and that it had not been dissolved.

When the feather is completed, the dermal pulp possesses no func-
tional connection with it; tl»e barbs and barbules are tlien practically
isolated from the vital processes of the organism and have no further
power of growth.

The arguments against change of color Avithout molt through repig-
mentation or regeneration of pigment may be summed up as follows :

1. Most feather pigments are too resistant to chemical reagents to
warrant belief in their solution and redistribution.

2. Pigmentation of the featlier has been observed to take place only
in the younger stages of the feather germ.

3. At the end of cornification melanin granules have a detinite ar-
rangement, which is permanent,

4. When cornification has ensued, the various elements of the feather
are hard, more or less solid, structures and their pigment contents are
effectually isolated from one another.

5. There is no satisfactory evidence of the occurrence of repigmenta-
tion, and all the histological conditions render such an event highly im-
probable.

VII. Summary.

1. The intermediate cells at the base of the feather germ multiply by
mitosis, not all of them being derived from the cylinder-cell layer directly.

2. The barbules are formed each from a single column of cells
placed end to end. These columns are arranged parallel to each other
and form the two lateral plates in each ridge of the feather fundament.
The lateral plates correspond respectively to distal and proximal sets of
barbules. The final form of the barbule results from a change in the
shape of its component cells.

3. Each of tlie cells composing the distal half of a distal barbule may
send out one or two processes, the barbicels.

4. The barbs are differentiated from cells making up the axial plate,
and appear later (Figs. 20, 21) than the, barbules. On tlie ventral
cortex of the barb is often found an asymmetrical ridge, which lias its
apex pointing towards the rhachis, as may be seen in a cross-section of
the feather germ. The epitrichium described by Haecker as covering
the cortex, I consider to be only an optical effect.

5. A basal membrane composed of flattened dermal cells separates the

STKONG: development of color IX DEFINITIVE FEATHEK. 177

epidermis of the feather germ from the pulp. This was seen by
Studer, but apparently overlooked by Davies.

6. The cylinder-cell layer comprises cells having the characteristic
cylindrical form, except in the region where there is an extensive
growth of the intermediate cells which go to form the barbules.

7. The initiative in the differentiation of " ridges " is taken by the
intermediate cells, not by the cylinder-cell layer, nor by the dermis.

.8. The condition of asymmetry with reference to the rhachis in
the vane of the completed feather is represented in a cross-section of the
feather germ by an unequal number of ridges on the two sides of the
rhachis.

9. The " Langsfurchen " described by Davies as occurring between
successive ridges, and also within the ridges themselves, are artificial
clefts due to imperfect fixation.

■ 10. The longitudinal extension of the feather germ is accomplished
by proliferation of cells at its base and also by the growth of the cells
composing the feather fundament.

11. The columns of cells composing barbules experience bendings in
two directions, resulting in a slightly spiral course. (1) By the growth
of its component cells the barbule column increases greatly in length.
Lateral extension in the feather germ being prevented by the confining
sheath, its more distal portions are bent inwards until they come to
lie nearly parallel with the long axis of the feather germ. (2) During
the development of the feather the ridges become larger near their
attachment to the rhachis. At a given level, as may be seen in cross-
sections, this results in a crowding or lateral displacement of ridges
towards the ventral side of the feather germ. The lateral plates (com-
posed of barbule columns) are bent so that they present a concave face
towards the rhachis. This condition is represented in a cross-section by
the curving of the roivs of barbule cells.

12. While a deposit of melanin pigment in the more central of the
medullary cells of the barb is usually associated with the production of
blue, as described by Haecker, the pigment may occur in the barbules
and not in the barbs. This is the case in the indigo bunting (Passerina
cyanea).

13. The melanins are supplied to the feather by branching pig-
ment cells, which distribute their pigment rods to cei'tain cells of the
feather fundament during, or immediately preceding, early stages of
cornificatiou.

14. The granules of melanin found in feathers are formed in the cyto-

VOL. XL. — NO. 3. 3

178 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

])lasm of so-called pigment cells. These are differentiated exclusively
from epidermal cells which lie in the intermediate cell layer of the epi-
dermis of the feather near the apices of the epidermal ridges.

15. Before cornification has ceased, all the pigment wliich the feather
is ever to receive has been supplied to the cells composing its fundament.

16. Changes in the color of plumage may take place either (1) by a
molt, during which the new feathers may have the same pigmentation
as tlieir predecessors or a different one ; (2) by a loss of certain portions
of the feather ; or (3) by physical disintegration in the cortex of the
feather as the residt of exposure. There is no satisfactory evidence of a
process of repigmentation, and the histological conditions of the feather
render such a process highly improbable.

1 wish to express my sincere gratitude to Professors IMark and G, H.
Parker for helpful criticism and revision of the manuscript.

strong: development of color in definitive feather. 179

BIBLIOGRAPHY.

Aeby, C.

'85. Die Herkunft des Pignient.es im Epithel. Centrallbl. f. d. med. Wiss.,
Jahrg. 23 (1885), No. 16, pp. 273-275.

Allen, J. A.

'96. Alleged Changes of Color in the Feathers of Birds without Molting.
. Bull. Amer. Mus. Nat. Hist., Vol. 8, Art. 3, pp. 13-44.

Altum, B.

'54. Ueber die Farben der Vogelfedern im Allgemeinen, iiber das Schillern
iiisbesondere. Naumannia. Arch. f. d. Ornith., Jahrg. 1854, pp. 293-
304.

Altum, B.

'54*. Ueber den Bau der Federn als Grund ihreFarbung. Journ. f. Ornith.,
Jahrg. 2, pp. xix-xxxv.

Birtwell, F. J.

:00. The Occurrence of Aptosochromatism in Passerina cyanea. Science,
u. s., Vol. 11, No. 269, pp. 292-299.

Bloch, A.

'97. Le pigment du systeme pileux et son origine. Bull. Soc. d'Anthrop.
Paris, ser. 4, Tome 8, No. 6, pp. 573-587.

Bogdanow, A.

'56. Note sur le pigment des plumes d'oiseaux. Bull. Soc. Nat. Moscou.
Tome 29, No. 2, pp. 459-462.

Bogdanow, A.

'57. Note sur le pigment rouge des plumes du Calurus auriceps, Gould.
Rev. et Mag. d. Zool., ser. 2, Tome 9, pp. 511-514. See also Compt.
Rend. (Paris), Toine 45, pp. 688-690.

Bogdanow, A.

'58. Etudes sur les causes de la coloration des Oiseaux. Rev. et Mag. d.
Zool., ser. 2, Tome 10, pp. 180-181.

Briicke, E.

'61. Ueber den Metallglanz. Sitzungsb. Akad. Wiss. Wien., raath.-naturw.
CI., Bd. 43, Abth. 2, pp. 177-192.

180 BULLETIN : MUSEUM OF COMPATxATIVE ZOOLOGY.

Chadbourne, A. P.

'97. Tlie Spring Plumage of the Bobolink, with Remarks on ' Color-
Change ' and 'Moulting.' The Auk, Vol. 14, No. 2, pp. 137-149,
PI. 1 a.

Chapman, F. M.

'96. On the Changes of Plumage in the Snowflake (Plectrophenax nivalis).
Bull. Amer. Mus. Nat. Hist., Vol. 8, Art. II., pp. 9-12, 2 text figures.

Church, A. H.

'69. Researches on Turacine, an Animal Pigment containing Copper.
Chem. News, Amer. Reprint, Vol. 5, No. 2, p. Gl.

Church, A. H.

'93. Turaein : A Remarkable Animal Pigment containing Copper. Nature,
Vol. 48, pp. 209-211.

Davies, H. R.

'89. Die Eutwicklung der Feder und ihre Bczichungcn zu andern Integu-
mentgebilden. Morph. Jahrb., Bd. 15, pp. 5G0-645, Taf. 23-26.

Demi^ville, P.

'80. Ueber die Pigmentflecke der Haut. Arch. f. path. Auat. u. Physiol.,
Bd. 81, Heft 2, pp. 333-354, Taf. 9.

Dwight, J., Jr.

:00. The Plumages and Moults of the Indigo Bunting (Passeriua ejanea).
Science, n. s., Vol. 11, i)p. G27-G30.

Dwight, J., Jr.

:00'. The Sequence of Plumages and Moults of the Passerine Birds of New
York. Ann. New York Acad. Sc, Vol. 13, No. 1, pp. 73-3G0, PI. 1-7.

Ehrmann, S.

'85. Untersuchungen iiber die Physiologic und Pathologic des Hautpig-
mcntes. x\rch. f. Derm. u. Syph., Jahrg. 12, pp. 507-532, Taf. 22-25.

Ehrmann, S.

'91. Ueber Hautentfarbung durch sekundar sypliilitische Exantheme.
Arch. f. Derm. u. Syph., Jahrg. 23, Erganzungsheft 2, pp. 75-102,
Taf. 5.

Ehrmann, S.

'92. Zur Kcnntniss von der Eutwicklung und Wandcrung des Pigments bci
den Amphibien. Arch. f. Derm. u. Syph., Jahrg. 24, pp. 195-222, Taf. 8.

Eisig, H.

'87. Monographie der Capitelliden, u. s. w. Fauna u. Flora des Golfes von
Neapel. Bd. XVI, xxvi + 906 pp., 37 Taf., 20 Holzschn.

Fatio, V.

'66. Des diverses modifications dans les Formes et la Coloration des Plumes.
Mem. Soc. Phys. et Hist. Nat. Geneve, Tome 18, Part 2, pp. 249-308,
3 pis.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 181

Gadow, H.

'82. Oil the Colour of Feathers as affected by their Structure. Proc
Zool. Soc., London, pp. 409-4.21, Pis. 27, 28. 3 text figures.

Gussenbauer, C.

'75. Ueber die Pigmentbildung iu melanotischeu Sarcomen und eiufacheu
Melanomeu der Haut. Arch. f. path. Anat. u. Pliysiol., Bd. 63, pp. 322-
355, Taf. 8.

Haecker, V.

'90. Ueber die Farben der Vogelfedern. Arcli. f. mikr, Anat., Bd. 35,
Heft 1, pp. G8-S7, Taf. 4.

Haecker, V., und G. Meyer.

■■01. Die blaue Farbe der Vogelfedern. ZooL Jahrb., Abth. f. Syst., Geog.,
- u. Biol. d. Thiere, Bd. 15, Heft 2, pp. 267-294, Taf. 14.

Jarisch, [A.J

'91. Ueber die Anatomic und Eutwicklung des Oberhautpigmentes beini
Frosche. Arch. f. Derm. u. Syph., Jalirg. 23, pp. 559-590, Taf. 10.

Jarisch, [A.]

'91a. lar Anatomic und Herkunft des Oberliaut-und Haarpigmentes beim
Meiisciien und den Siiugethieren. Arch. f. Derm. u. Syph., Jalirg. 23,
Ei-ga.izungsheft 2, pp. 35-55, Taf. 1.

Jarisch, [A.]

'92. Ueber die Bildung des Pigmentes in den Oberhautzellen. Arch. f.
Derm. u. Syph., Jalirg. 24, pp. 223-234, Taf. 9.

Kaposi, M.

"91. Ueber Pathogenese der Pigmentirungen und Entfarbungen der Haut.
Arch. f. Derm. u. Syph., Jahrg. 23, pp. 191-205.

Karg.

'88. Studieniiber transplantirte Haut. Arch. f. Anat. u. Physiologic, Jahrg.
1888, Anat. Abth., pp. 369-406, Taf. 20-22.

Keeler, C. A.

'93. Evolution of the Colors of North American Land Birds. Calif. Acad.
Sci., Occasional Papers, [No.] 3, xii + 361 pp., 19 pis.

Kerbert, C.

'76. Ueber die Haut der Reptilien und anderer Wirbelthiere. Arch. f.
mikr. Anat., Bd. 13, pp. 205-262, Taf. 18-20.

Klee, R.

'86. Bau uiul Eutwicklung der Feder. Hallersche Zeitschr. f. Naturwiss.,
Bd. 59, Heft 2, pp. 110-156, Taf. 3/4.

KoUiker, A.

'87. Ueber die Entstehung des Pigmentes iu den Oberliautg(;bildeu. Zeit-
schr. f. wiss. Zool., Bd. 45, Heft 4, pp. 713-720, Taf. 37-38.

182 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Kromayer, E.

'97. Eiiiigc cpitlieliale Gebildc in iieucr Anfiassung. Bcitriige zur Pig-
mcutfragc. Derm. Zeitsclir., Bd. 4, Ilei'f, 3, pp. 335-400, Taf. 1-4, 13
Textligureu.

Krukenberg, C. F. W.

'81. Die Farbstott'e der Federn. Erstc Mittheiluiig. Vergleichend-physiol.
Studien. (Heidelberg.) Keibe I., Abth. 5, pp. 72-99, Tuf. 3.

Krukenberg, C. F. W.

'82. Die Farbstotl'e der Federn. Zweite Mittheilung. Vergl.-physiol.
Studieu. (Heidelberg.) Reilie XL, Abth. 1, pp. 151-171.

Krukenberg, C. F. W.

'82*. Die Farbstoffe der Federn. Vierte Mittheilung. Vergl.-piiysiol.
Studieu. (Heidelberg.) Keihe II., Abth. 3, pp. 128-137. Taf. 9.

Krukenberg, C. F. W.

'84. Grundziige einer vergleichenden Pliysiologie der Farbstoffe und der
Furben. Vergl.-physiol. Vortrage. Bd. 1 [Theil] 3, pp. 83-184.
Heidelberg.

Langhans, T.

'70. Beobachtungen iiber Resorption der Extravasate und Pignientbildung
in denselben. Arch. 1". path. Auat. u. Physiol., Bd. 49, pp. 66-116, Taf.
3, 4.

Loeb, L.

'97. Ueber Transplantation von weisser Haut auf einen Defekt in scluvarzer
Haut und unigekehrt am Ohr des Meerschweiuchens. Arch. f. Entwick.-
Mech., Bd. 6, Heft 1, pp. 1-14, Taf. 1-3, 2 Textfiguren.

Loeb, L.

'98. Ueber Regeneration dcs Epitliels. Arch. f. Entwick.-Mech., Bd. 6,
Heft 3, pp. 297-364, Taf. 15-22, 9 Texlfiguren.

Lwoff, W.

'84. Beitrage zur Histologic des Haares, der Borste, des Staehels und der
Feder. Bull. Soc. imp. Moscou, Tome 59, Part 1, pp. 141-174, Taf. 5-8.

Maurer, F.

'95. Die Epidermis und ihre Abkommlinge. Leipzig. ix + 352 pp., 9
Taf., 28 Textfiguren.

Meerwarth, H.

'98. Beobachtungen iiber Verfiirbung (ohne' Mauser) der Schwanzfedern bra-
silianischer Raubvogel, nebst eiuem Beitrag zur Phylogcuese der Raubvu-
gelzeichnung. Zool. Jahrb., Abth. f. Syst., Geog., u. Biol. d. Thiere,
Bd. 11, Heft 2, pp. 65-88, Taf. 8-10.

Mertsching.

'89. llistologische Studien iiber Keratohyalin und Pigment. Arch. f. path.
Auat. u. Physiol., Bd. 116, Heft 3, pp.' 484-516, Taf. 9.

STRONG : DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 183

Meves, W.

'55. Ueber die Farbenveranderung der Vogel durch und ohne Mauser.
Jour. f. Oruith., Bd. 3, pp. 230-238, Taf. 2, ^. Translfrom Oefversigt
at" k. Veteuskaps Akadem. Forhaudl, 1854, Nr. 8 (Stockholm).

Phillipson, A.

'90. Ueber Hautpignient. Fortschr. d. Med., Jahrg. 8, pp. 216-221.

Post, H.

'94. Ueber normale und pathologiscbe Pigmentirung der Oberhautgebilde.
Arch. f. path. Auat. u. Physiol., Bd. 135, Heft 3, pp. 479-513, Taf. 13.

Prowazek, S.

:00. Beitragzur Pigmeutfrage. Zool. Anz., Bd. 23, No. 623, pp. 477-480.

Quincke, H.

'84. Beitrage zur Lehre vom Icterus. Arch. f. path. Anat. u. Physiol., Bd.
95, Heft 1, pp. 125-139.

Rabl, H.

'94. Ueber die Eutwicklung des Pigmentes in der Dunenfeder des Hiihn-
cheus. Centralbl. f. Physiol., Bd. 8, p. 256.

Rabl, H.

:00. Ueber Bau und Entwickelung der Chromatophoren der Cephalopoden,
nebst allgemeiuen Bemerkungeu iiber die Haut dieser Thiere. Sitzungsb.
Akad. Wiss. Wien, niath.-uaturw. CL, Bd. 109, Abth. 3, Heft 7, pp. 341-
404, 4 Taf.

Abstract in Centralbl. f. Physiol., Bd. 14, No.' 24, p. 615, 1901.

Riehl, G.

'84. Zur Kenntniss des Pigmentes im menschliclien Haar. Arch. f. Derm.
u. Syph., Bd. 11. pp. 33-39, Taf. 5.

Rosenstadt, B.

'97. Studien iiber die Abstammung und die Bildung des Hautpigments.
Arch. f. mikr. Anat., Bd. 50, Heft 2, pp. 350-384.

Sauermann, C.

'89. Ueber die Wirkung organischer Farbstoflfe auf das Gefieder der Vogel
bei stomachaler Darreichung. Arch. f. Anat. u. Physiol. Jahrg. 1889,
Physiol. Abth., pp. 543-549.

Schwann, T.

'39. Mikroskopische Untersuchungen, u. s. w. Berlin, xviii + 270 pp.,
4 Tab.

Solger, B.

'89. Zur Structur der Pigmentzelle. Zool. Anz., Jahrg. 12, No. 324, pp.
671-673, 1 Textfigur.

Solger, B.

'90. Nachtrag zu dem Artikel: "Zur Structur der Pigmentzelle.'' Zool.
Anz., Jahrg. 13, No. 328, pp. 93-94.

184 bulletin: museum of comparative zoology.

Solger, B.

'91. Zur Kcnntnis dcr Pigmentzellen. Anat. Anz., Jalirg. 0, No. G, pp.
162-165, 2 Abbilduugeii.

Stone, W.

'96. The Molting of Birds with Special Reference to the Plumages of the
Smaller Land Birds of Eastern North America. Proc Acad. Nat. Sci.
Phila., 1896, pp. lOS-167, Pis. 4-5, 2 text figures.

Stone, W.

:00. Report on the Birds and Mammals collected by the Mcllhenny Expe-
dition to Pt. Barrow, Alaska. Proc. Acad.. Nat. Sci. Pliila., 1900, pp.
4-49.

Studer, T.

'73. Die Entwicklung der Federn. Inaug -Dissert. [Pliilos.] Facult. Bern.
Bern. 29 p., 2 Taf.

Studer, T.

'78. Beitrjigc zur Entwicklungsgcschichte dcr Fedcr. Zeitschr. f. wi?s.
Zool., Bd. 30, i)p. 421-436, Taf. 25, 26.

Waldeyer, W.

'82. Untersuchungen iiber die Ilistogenese der Ilorngebilde, iusbesondere der
Haare und Federn. Beitrage z. Anat. u. Embry. als Festgabe. J. Hcnle.
Bona. pp. 141-163, Taf. 9, B.

STRONG: DEVELOPMENT OF COLOR IN DEFINITIVE FEATHER. 185

EXPLANATION OF PLATES.

Figures 12-21 and 23 are from sections of a featlier germ (secondary) of Sterna
hirundo which was fixed with Hermann's fluid and stained in iron haematoxylin.
They represent corresponding regions, indicated in Figure "2 by an asterisk (*),—
but taken at diflferent levels. The levels of the sections are indicated in Figure 1
by the horizontal lines 12, 13, 14, etc. Figures 3, 35, 36, and 37 are also from
material fixed in Hermann's fluid and stained with iron haematoxylin. Figures
22, 24, 38, 39, 40, 41, 42 were made from material fixed with Kleinenberg's picro-
sulphuric mixture and stained in Kleinenberg's haematoxylin followed by eosin.
All drawings were made with the aid of a camera lucida.

ABBREVIATIONS.

brb.

Barb.

dst.

Distal.

brbc.

Barbicel.

e'th.

Epithelium.

brbt.

Barbule.

fnd.

Fundament.

cal.

Calamus.

gran. pig.

Pigment granule.

cl. cyl .

Cylinder-cell layer.

haml.

Hamuli or booklets.

cl. i'm.

Intermediate cells.

la. (IX.

Axial plate.

cl. med.

Medullary cells.

mac pig.

Pigment patches.

cl. pig.

Pigment cells.

marg.

Recurved margin of prox-

cl. tlt.l.

Inner sheath cells.

imal barbule.

coll. cl.

Column of cells forming a

mb. ba.

Basal membrane.

single barbule.

med.

Medulla.

cpl. sng.

Red blood corpuscles.

nl.

Nucleus.

crs.

Ridge of epithelium marked

nil.

Nucleolus.

oS by 7«6. Int.

pre.

Process of pigment cell.

crs'.

Ventral ridge of barb.

prjr.

Proximal.

crs".

Irregular ridges of epithe-

rrh.

Rhachis.

lium.

ser. cl.

Row of barbule cells seen

ctx.

Cortex.

in transverse section.

cyV pi.

Cytoplasm.

tu.

Feather sheath.

d.

Dorsal.

uinb. inf.

Inferior umbilicus.

drm.

Derma.

V.

Ventral.

Strono. — Development of Color in Feathars.

PLATE 1.

All Figures except 7-9 are of Sterna hirundo.

Fig. 1. Diagrammatic longitudinal section. X15. Figures 12-21 and 23 were

drawn from sections taken at the points indicated by the dotted

lines 12, 13, 14, etc.
Fig. 2. Semi-diagrammatic cross-section, indicating by an asterisk (*) the region

chosen for illustration in Figures 12-21 and 23.
A portion of a barb and its barbules seen from the dorsal side. Xll7.
A " primary " feather having been split dorso-ventrally and the pulp

removed, the inner or pulp, surface of the proximal portion of one

half of the feather fundament is here shown. X16.
External view of definitive feather germ. The dotted line 23 corresponds

in position to the line 23 in Fig. 1.
Diagram, to show position of barbules with reference to the barb, while

still enclosed in the feather sheath.
Transverse section of barb from blue body-covert of Sialia sialis. X495.

as'. Ventral ridge of cortex of barb.
Transverse section of barb from blue wing-covert of Cyanocitta cristata.

X495.
Transverse section of barb from brown wing-covert of the "homer"

pigeon. X4y5.

Fig.
Fig.

3.
4.

Fig.

5

Fig.

6.

Fig.

7.

Fig.

8,

Fig.

9.

Strong.— Development of Color in Feathers.

d. d.

Pi^TE 1.

/

I

\

\
\

V

23

20

18

I

llUti

/

hrbl.

Qt

1

CT'-S".

r/s.

N^-

-Jv

1 /'

9
*-e-rs'.

17

16
-15

'm^-^-

>'<!y

V ... '4

/ |- - - 13

/

mnh inf.

fiM.prx

5 U

■MS a.-^

Strong. — Development of Color in Feathera.

PLATE 2.

All Figures magnified 495 diameters.

Fig. 10. Transverse section of barb from blue featlier of Cotinga cayana.

Fig. 11. Transverse section of l)arb from l>lue wing-featiier of Pitta molnccensis.

Figures 12-14 are portions of transverse sections of wing-featliers from Sterna
liirundo.

Fig. 12. Section at level of 12 in Fig. 1. The position of the part of the section
here siiown is indicated in Figure 2 by tiie asterisk (*). crs". Small
ridge in epithelium preceding formation of barb ridges.

Fig. 13. Section at tiie level 13, in Figure 1. cl'. Dividing cell.

Fig. 14. Section at the level 14 in Figure 1.

Strong — Development of Color tn Feathers.

Plate 2.

RM.S. del.

Strong. — Development of Color in Feathers.

PLATE 3.

Figs. 15-18. Transverse sections of feather germs of Sterna liirundo. x49o.
Fig. 15. Section at level 15 in Figure 1.
Fig. 16. Section at level 16, Figure 1.
Fig. 17. Section at level 17, Figure 1.

Fig. 18. Section at level 18, Figure 1. pre' A pigment-cell process apparently
not entirely filled with pigment granules.

Strong- Development of Color tm Feathers.

PLATE 3.

tv.

d.l*f-

mbM.

15

5 - CTdC^^-^"®- <*.'

III.

17

tv.

Stl.'Cl.

(J inn. ply.

I>ir,

rl.pi(J.

/"*■•

■/f^^

^V

18

RMS. del.

Strong. — Development of Color in Feathers.

PLATE 4.

Figs. 19-21. Transverse sections of feather germ of Sterna liirundo. X 495.

Fig. 19. Section at level 19, Figure 1.

Fig. 20. Section at level 20, Figure 1.

Fig. 21. Section at level 21, Fig. 1. cl. pig. Unused pigment.

Fig. 22. Section of feather germ of body covert of Passerina cyanea, showing
pigmentation of blue portion of feather and also the witlidrawal of
the feather elements from the surrounding tissue. X 496.

Strong— Development of Color in Feathers.

Plate 4.

Stbono. — Development of Color in Feathers.

PLATE 5.

All Figures are from feathers of Sterna hirundo except Fig. 29.

Fig. 23. Transverse section of feather germ at level 23 in Fig. 1. X 495.
Note, — By an oversight tlie proximal and distal barbules are lettered brb. instead
of hrbl.

Fig. 24. Transverse section of wing-covert, showing withdrawal of barbs from the
surrounding tissue preceding the unfolding of the feather. X -lOS-

Fig. 25. A proximal barbule from wing-feather. X 117.

Fig. 26. A distal barbule from wing-feather. X 117.

Fig. 27. Middle portion of a barbule from wing-feather showing distribution of
pigment, the form of the cells composing the barbule, and the forma-
tion of barbicels. Cornification is not yet complete. X 495.

Fig. 28. Distal portion of barbule siiown in Figure 27. X 495.

Fig. 29. Transverse section of barb from blue portion of a body-covert of Pas-
serina cyanea with portions of barbules on either side. X 495.

Strong— Development of Color in Feathers.

Plate 5.

ummm^ r'1

cs>^

K^h^^^^>^Mr^

' — biba

\'V 'TO

sil

''''*''""""*ii'tf

km

28

■.:M;i '

27

RMS. del.

Stbono. — Development of Color in Feathers.

PLATE 6.

All Figures are from feather germs of Sterna hirundo.

Fig. 30. Transverse section showing first appearance of pigment granules in the

cytoplasm of the pigment cell. X 1500.
Figs. 31-34. Successive stages in development of pigment cells. Figures 31 and

32 represent about the same stage. X 1500.
Fig. 35. Pulp edge, or apex, of a ridge of the fi-atlier fundament, showing three

pigment cells with granules crowded into an opaque mass and with

processes beginning to be formed. X 1500.
Fig. 36. A somewhat later stage, showing pigment granules or rods entering

barbule cells (compare Plate 3, Fig. 17). X 1500.

Strong — Developivient of Color tn Feathers.

Jt'l/.te; 6.

^^'•

k'

• ^

?

X

ynm.p(fj.

rvl'pl.

iiL

\V

tnrm.])ifj.

r^

30

*!,•*«.

y

rial.

35

36

MS. del.

Strono. — Development of Color in Feathers.

PLATE 7.

Photomicrograplis.

Fig. 37. Portion of transverse section of feather germ from Sterna hirundo. X 300.
Fig. o8. Portion of longitiulinal section of blue-featiier germ from Passerinji
cyanea. X 4bO.

Strong.-Coloration Of Feathers.

Plate 7.

brl. dxt.

hrl. prx.

I. med.

nh. b(t.

Fig, 37.

tu.

* *■

1

/

drm.

cl. pig.

Fig. 38.

Strong. — Development of Color in Feathers.

PLATE 8.

Photomicrographs.

Fig. 39. Transverse section of blue-feather germ from Passerina cyanea. X 250.
Fig. 40. Transverse section of green-featiier germ from l^asserina oiris, showing
process of pigmentation of the barbules. X 157.

I

Strong. -Coloration of Feathers.

Plate 8.

Fig, 39,

, c'. pi'./-

Btbono. — Development of Color iu Feathers.

PLATE 9.

Photomicrographs.

Fig. 41. Transverse section of green-feather germ from Passerina ciris, showing
pigmentation completed and cornification nearly so. X 157.

Fig. 42. Transverse section of wing-feather from the " homer " pigeon, showing
differentiation and cornification completed. X 09.

Strong. -Coloration of Feathers.

Plate 9.

Will

iA

rrh.

• •• • (• • •,

• ••_•♦••.• ,

• •>* t ^ • V

• • •• . . \

Fig. 41 .

Fig. 42.

Bulletin of the Museum of Comparative Zoology

AT HARVARD COLLEGE,

Vol. XL. No. 4.

THE HEREDITY OF SEX.

By W. E. Castle.

CAMBRIDGE, MASS., U.S.A.:

PRINTED B^OU THE MUSEUM.

January, 1903.

\K\\

24

1903

No. 4. — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY
OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD
COLLEGE. E. L. MARK, DIRECTOR. No. 138.

TJie Heredity of Sex. By W. E. Castle.

CONTENTS.

1. Introduction 189

II. Sex an attribute of eacli ga-
mete, and hereditary 190

III. Principles of heredity appli-

cable to sex . . . 191

1. Mendel's law 191

(a) The principle of domi-

nance 191

(b) The principle of segre-

gation 192

2. Jlosaic inheritance . . . 192

IV. Application of the principles

stated 193

1. Dioecious and hermaphro-

dite organisms . . 193

2. Parthenogenetic organisms 198
(a) General application . . 198

PAGE

201
201
202
203

(6) Special cases . . .

A. Rhodites rosae

B. Hydatina senta .

C. Artemia salina

D. Exceptional parthe
nogenesis in Bombyx
mori, etc 205

V. Abnormal sex proportions

among hybrids . . 205

1. Relative infertility of cer-

tain combinations of
gametes 206

2. Coupling of certain sex and

somatic characters in

the germ-cells . . . 208

VI. Summary 214

Bibliography 216

I, Introduction.

A NEW theory of sex is advanced in this paper, yet a theory which in
its elements is not new. It is an attempt to correlate three ideas, the
correctness of which, separately considered, is generally recognized :
(1) the idea of Darwin ('76), that in animals and plants of either sex
the characters of the opposite sex are latent ; (2) the idea of Mendel
('66), that in the formation of the gametes of hybrids a segregation of
the parental characters takes place, and when in fertilization different
segregated characters meet, one will dominate, the other become latent
or recessive ; (3) the idea of Weismann ('93) that in the maturation of
egg and spermatozoon, a segregation of ancestral characters takes place,
and that this segregation' is attended by a visible reduction in the num-
ber of chromosomes in the germinal nuclei.

VOL. XL. — NO. 4

190 bulletin: museum of comparative zoology.

II. Sex an Attribute of each Gamete, and Hereditary.

The last forty years have seen the rise, cuhui nation, and at least
incipient decline of a plausible but fundamentally erroneous idea about
sex, — the idea that it is subject to control through the environment of
the developing organism. The latest manifestation of this idea is found
in Schenk's (:02, :02'' ) theory of sex-control in man through regulation
of the nutrition of the mother. One or the otlier, or both, of two fal-
lacies are involved in all such theories of sex-control. (1) It is known
that in animals which reproduce sometimes by parthenogenesis, some-
times by fertilized eggs, good nutrition favors the former process, poor
nutrition the latter. But in the former process, when it proceeds with-
out interruption, the offspring arc all of the female sex, whereas the
lirst eftect of poor nutrition is the production of mak's, and tliis is fol-
lowed by the production of fertilized eggs. The conclusion is drawn
that good nutrition favors the production of females among animals gen-
eraliy, and that poor nutrition results in general in tlie production of
males. As a matter of fact the primary effect of good nutrition, in the
case described, is not female production, hut ixtrthenogenesis, and the
effect of poor nutrition is, not jiTiniarily male production, but reproduction
bi/ fertilized eggs, in wliicli process the production of males is necessarily
involved. The determination of parthenogenesis instead of sexual re-
production is one thing, determination of sex in animals not parthe-
nogenetic is quite another thing. (2) The other fallacy mentioned
relates solely to the case of animals not parthenogenetic. Its true
nature has been repeatedly pointed out, but apparently none too often,
for Schenk seems to rest his theory upon it. Feeding experiments,
especially with Lepiiloptera, often lead to the production of an excess of
males when tlie nutrition is scanty, simply because the female requires
a greater amount of food to complete her development. Excess of males
because of a greater mortality among female individuals is wrongly
interpreted as a production of male individuals by a scanty diet.

On the other hand, evidence has been steadily accumulating in recent
years to show that sex is inherent in the germ, and is not subject to
control in the slightest degree by environment. A masterly summary
of this evidence has been made in the case of animals by Cucnot ('99),
and in the case of plants by Strasburger (:00).

If it be true that sex is inherent in the germ, and is independent of
environment, it must be contained in one or the other or both of the

CASTLE : THE HEREDITY OF SEX. 191

sexual gametes, and the appropriate subject for investigation is the law
or laws of its inheritance, rather tlian the visionary external causes of
sex.

That sex is borne by the egg is shown clearly by the case of partheno-
genetic animals, which without the intervention of a male produce young
of both sexes. That the spermatozoon also bears sex is manifest in the
case of animals lilce the honey-bee, for the egg of the bee, if unfertilized,
invariably develops into a male, but if fertilized, into a female. We
have, therefore, specific reasons, iu addition to the general ground of the
equivalency of egg and spermatozoon, for supposing that sex is a char-
acter possessed by every egg and spermatozoon.

In the following pages I liave attempted to formulate certain of the
laws of sex-heredity, an attempt which is greatly aided by recent devel-
opments in our knowledge of heredity in general.

III. Principles of Heredity Applicable to Sex.

1. Mendel's Law.

Perhaps the greatest discovery ever made in the study of heredity is
what is commonly known as Mendel's Law. Eateson and Saunders (: 02)
in a recent paper suggest that sex may be inherited in accordance with
that law. In the light of this suggestion certain phenomena of sex are
in this paper examined, and found to have their almost perfect parallels
in recognized Mendolian phenomena. In consequence we get a new
point of view from which to study the phenomena of sex, and many of
its long-time mysteries find ready explanation. The basic principles
of Mendel's law are two, the principle of dominance and the principle
of sejrrerration.

(a) The Principle of Dnniiiance. When there unite in fertilization
two gametes, one of wdiich bears one of a pair of alternative characters,
while the other gamete bears the other character, it often happens that
the zygote formed manifests only one of tlie two characters. This char-
acter may be called the dominant one. The other character becomes
latent, or r-ecessive, and is first seen in the next genei'ation of offspring.
For example, when white mice are crossed with wild gray mice, all the
offspring ai-e gra}', that character being dominant, white recessive.
White mice are never obtained in the first hybrid generation, but upon
breeding of the primary hybrids inter se, both white and gray offspring
are obtained approximately in the ratio, 1 : 3.

192 bulletin: museum of compaeative zoology.

(J)) The Principle of Ser/regation. The appearance of white mice, as
just described, in the second hybrid generation, follows from the prin-
ciple of segregation. Tlie primitive germ-cells of the primary hybrid
contain both parental characters. D (dominant) and R (recessive), but
in the maturation of the germ-cells the two are separated, so that the
ripe gerni-cell (or gamete) contains either D or R, but not both. This
is demonstrably true in both sexes. Accordingly there are ova, D and R,
and spermatozoa, I) and R. If dominants and recessives are produced
by each parent in equal abundance, and they unite at random, the sorts
of zygotes resulting and their relative frequencies of occurrence will be
expressed by the product, —

D-\- R (ova)

D -\- R (spermatozoa)

DJ) -{-2 D (R)* -\-RR (zygotes).

One individual in four will be a pure dominant, DB (gray in the
case of mice) ; likewise one in four will be a pure recessive, RR (white
in mice) ; while two in four will be hybrids, D (R), like their parents,
the primary hybrids, though indistinguishable in appearance from the
pure dominant, I)D.

2. Mosaic Inheritance.

An important exception to the two principles just stated needs to be
noted. In cases otherwise conforming to Mendel's law, tliere sometimes
occur exceptional hybrid individuals in which the normal dominance of
one character is not realized, but the two alternative characters coexist
in a patchwork or mosaic arrangement. Such a condition is illustrated
in the case of piebald, or spotted, mice.

Segregation of characters does not commonly occur in the formation of
the gametes pi'oduced by mosaic individuals. The gametes, as well as the
parents, are mosaic, DR. For when two mosaic individuals are mated,
they commonly produce only mosaic offspring ; and when a mosaic is
mated with a pui'e recessive, RR., no recessive offspring are as a rule
produced. These facts show clearly that the ordinary mosaic individual
forms no ])ure recessive gametes; in other words, that segregation does

* Tlie parenthesis is used to indicate tliat the recessive character, though
present, is not visible. Wlienever the recessive cliaracter alone is present in an
individual [as iu {ltR)'\, it will of course be visible; but whenever the recessive
character is present together with the dominant [as in the two individuals Z> (/?)],
the recessive character will not be visible.

castle: the heredity of sex. 193

not occur at the formation of its gametes. Nevertheless a mosaic indi-
vidual does occasionally occur which produces a certain proportion of
segregated (that is, pure) gametes. Exceptionall}' a spotted mouse
when paired with a recessive mate produces pure recessive (white)
offspring as well as hybrid (dark) offspring. The peculiarity is inherent
in the parent and is manifested with uniformity by certain individuals,
but not at all by others.

IV. Application of the Principles Stated.

1. Dioecious and Hermaphrodite Organisms.

Sex in dioecious animals and plants is inherited in accordance with
Mendel's law; that is, in accordance with the principles of dominance
and segregation. The ordinary dioecious individual is a sex-hybrid or
" heterozygote " (Bateson), in which the characters of both sexes are
present, one dominant, the other recessive. In the male, the female
character is recessive, and conversely in the female, the male character ;
but each sex transmits the characters of both.

The existence of each sex (in a latent condition) in the other is
shown by the occurrence in each sex of rudimentar}^ organs peculiar
to the other. This evidence is supported by numerous observations
brought forward by Darwin ('76) to show that an animal in its old age,
or when its genital organs become diseased, often manifests characters of
plumage or of voice, or even instincts, which are characteristic of the
opposite sex.

But perhaps the strongest evidence of the latency of each sex in the
other is afforded by the transmission through one sex of the characters
of the other. Thus, as Darwin states, when the domestic cock is crossed
with the hen pheasant, the male offspring have the secondary sexual
characters of the viale pheasant ; these, manifestly, must have been
inherited through the female pheasant.

Again, in many animals which reproduce by parthenogenesis, the
female bears (without fertilization) both male and female offspring,
showing that she really possesses both sex-characters.

Experimental evidence of the latency of one sex in the other in plants
has been produced by Bordage ('98). He cut back the apex of young
male plants of Carica papaya, just before the appearance of the first
male flowers. Lateral branches, two on each plant, then arose immedi-
ately below the cut, and these produced female flowers and fruit.

194 bulletin: museum of comparative zoology.

A somewhat similar case is described by Strasburger (: 00), in wliich a
smut, Ustilago violacea, when present as a parasite in the female plant of
Melandryum album, causes the female organ of the latter, the pistil, to
remain undeveloped, while the anthers, normally mere rudiments, grow
to a large size and actually form pollen-mother cells, which the fungus
then attacks and destroys. In this case it is the male character which,
though normally recessive, is made to appear upon destruction of the
genital fundament of the opposite sex ; in the case of Carica papaya, it
is the female character which behaves in a similar way.

Tlie objection may be offered that certain of the examples cited really
belong in the category of imperfect hermaphroditism, or at any rate of
potential hermaphroditism. This I freely grant ; I would even go
farther and say that all animals and plants are potential hermaphrodites,
for the;/ contain the characters of both sexes, but ordinarily the characters
of one sex only are developed, those of the other sex being latent or else
imperfectly developed.

In true hermaphrodites, however, the characters of both sexes exist
fully developed side by side, as do the gray and the white coat-colors in
spotted mice. The true hermaphrodite, then, is a sex-mosaic ; to the
heredity of sex, in its case, we may expect to find applicable the
general principles of mosaic inheritance.

The difference between a hermaphrodite and a dioecious animal is
precisely parallel to that which exists between a spotted and a normal
hybrid mouse. In the hermaphrodite, as in the spotted mouse, two
characters ordinarily alternative exist as co-ordinates, side by side ; in
dioecious animals, as in ordinary hybrid mice, the same two characters
exist in their more usual relationsliip of dominant and recessive. The
only difference between the two classes of cases is this. In coat-color
among mice gray is invariably dominant over, or balanced with white,
but never recessive toward it. But in dioecious animals the male char-
acter is sometimes dominant over the female, sometimes balanced with
it, and sometimes recessive toward it. This condition, though not paral-
leled in the illustration chosen (coat-color of mice), is not without a
parallel among other Mendelian cases. For,-Tschermak (:00) finds that
in certain crosses among peas, one charactev may be, with reference to
another, sometimes dominant, sometimes recessive.

We have seen that spotted (hybrid) mice commonly produce gametes
which are, like themselves, mosaic, DR, whereas ordinary (gray) hybrids,
in which white is recessive, produce '* pure " gametes, either D or i?, in
accordance with the principle of segregation. Similarly the sea'-mosaic,

CASTLE : THE HEREDITY OF SEX, 195

the normal hermaphrodite, probably produces mosaic gametes, ^ 9 , for
when in fertilization these unite in pairs, they invariably form hermaph-
rodite individuals, ^ 9 • K segregation occurred in the production of
the gametes, we should expect the occurrence also of its counterpart,
dominance, in fertilization. Since in hermaphrodites the latter does not
occur, it is probable that the former does not occur either.

But in dioecious species sexual dominance almost invariably occurs ;
it is probable, therefore, that in such species segregation of sex-char-
actex's takes place in the formation of the gametes. If so, and if, as in
color heredity among mice, all possible combinations of gametes are
formed in fertilization, and in the frequencies demanded by the law of
chance, the sex of the oflfspring should be indicated by the product, —

(? + 9 (ova)

(? + 9 (spermatozoa)

SS + -2^9 + 99 (zygotes).

According to this, half the offspring, it will be observed, must be pxirely
of one sex or the other ; that is, must contain and transmit the characters ■
of one sex only. But we have no reason to think that such sexually
"pure" individuals exist. On the contrary, when, as in the case of the
honey-bee, the individual apparently transmits uniformly the character
of one sex, that sex is invariably the opposite to its own. It is highly
probable, therefore, that an egg bearing the character of one sex can
unite in fertilization only with a spermatozoon bearing the character of
the opposite sex. Our present knowledge of the process of fertilization
indicates that in it a union is accomplished between elements strictly
equivalent to those which were separated in the formation of the
gametes. But there exist, as we have'seen, strong reasons for believing
tliat in the formation of the gametes, opposite sex-characters are sepa-
rated. Consequently, on a prio7-i grounds, we should expect only
opposite sex-characters to unite in fertilization.

But, some one may object, if a ripe egg of one sex can be fertilized
only by a spermatozoon of the opposite sex, it follows that half the eggs
produced are infertile toward half the spermatozoa. This, however, is
not so serious an objection as it may at first thought seem to be. It
does not involve impotency of half the eggs and spermatozoa, nor of any
portion of them. All the eggs of one sex will be fertile toward all the
spermatozoa of the opposite sex ; the remaining eggs will be fertile
toward the remaining spermatozoa. The infertility which exists is only

196 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

a relative one, and relative infertility much greater than this is a well-
established fiict in other cases. Thus, the writer (Castle, '96) showed
some years ago that more than 90% of the eggs produced by the
hermaphrodite tunicate, Ciona intestinalis, are wholly infertile toward
sperm produced by the same individual ; yet toward the sperm of
another individual the fertility is almost perfect. This instance is only
one of many which might be cited as indications that successful fertili-
zation depends upon iinlikeness between the gametes uniting. In the
case of the tunicate, which is hermaphrodite, sexual unlikeness between
gametes probably does not occur, hence it is some other unlikeness
which brings egg and sperm together, and it is not surprising to hud a
degree of gametic differentiation between the eggs and sperm of the
same individual which is insufficient, in most cases, for successful
fertilization.

On the hypothesis advanced, the zygote must, in all cases, bear both
the male and the female characters. In the zygote of a hermaphrodite
species, these two characters will exist in the balanced relationship in
which they were received from the parents, a relationship which has
not been disturbed by segregation, and which accordingly is stable.
But in a dioecious species the male and female characters meet anew
in a struggle for supremacy at each fertilization. Sometimes one, some-
times the other, dominates in the zygote, the vanquished character
becoming recessive. Exceptionally, as in the occasional or the mixed
hermaphrodite of a dioecious species, the fight is indecisive, and neither
combatant is supreme.

In parthenogenetic species, the female character appears to be uni-
formly the stronger of the two, so that it dominates in every contest,
for the fertilized egg in such species develops invariably into a female.
In dioecious species, on the other hand, neither character, apparently,
has any uniform advantage over the other. Males and females are
produced in a[)proximatcly equal numbers. In hybridization the con-
test between gametes may often be an unequal one, and it will not be
surprising to find the gametes of one species uniformly dominant over
those of another hi sex as well as in somatic characters. This is a
matter to which further attention will presently be given.

But, it may be objected, the hypothesis presented is improbable
because in i)arthenogenetic animals like the honey-bee, each sex uni-
formly transmits the opposite. INIay it not be so in dioecious animals
also? (See Wedekind, :02.) This suggestion is negatived by the follow-
ing considerations : (1) Most parthenogenetic animals, like Daphuia,

castle: the heredity of sex. 197

for example, produce both male and female offspring from unfertilized
eggs! (2) The eggs of Dinophilus, laid by the same mother, are
of two distinct sizes, one about three times as large as the other.
From the larger sort develop females, from the smaller, males (see
Korschelt, '87). (3) Similar morphological differences, though less
obvious ones, exist between the male and female eggs of the gypsy-moth,
Ocneria dispar, according to Joseph ('7l) and Cuenot ('99), and of
the silk-moth, Bombyx mori, according to Brocadello as quoted by
Cuenot. This case is supported by the observations of von Siebold
('56) and others, which show that eggs of the two species mentioned
occasionally develop ivithout fertilization, and that in such cases normal
individuals of hotli sexes are produced.

On the other hand, dimorphic spermatozoa exist in the case of
Paludina and some other animals, bat there is no adequate reason at
present for supposing that this dimorphism is related to sex. The
consensus of opinion on the part of those who have studied these cases
is that the more usual form of spermatozoon alone is functional, the
other being pathological. Nevertheless, the subject is one meriting
further investigation.

The occasional occurrence of cases of true hermaphroditism, in species
normally dioecious, may be cited as evidence in favor of the hypothesis
of sex presented in this paper. Each dioecious individual, we have sup-
posed, is a potential hermaphrodite, but has tlie characters of one sex re-
cessive. The true hermaphrodite (I'are in dioecious species) is an animal
in which neither sex is recessive, but the characters of both sexes are devel-
oped together. Unilateral and mixed hermaphrodites are an exceptional
form of sex-mosaic : they may in some cases be animals in whose devel-
opment fusion of the pronuclei has not occurred, one side or region of
the body containing only nuclei derived from the male, the other from
the female gamete. A similar result might follow, if, even after fusion
of the pronuclei in the egg, segregation of sex-characters should occur in
cleavage, instead of the normal equation divisions. Or, thirdly, a mosaic
sex-character may exceptionally be possessed by the gametes themselves,
comparable with the mosaic character as to color possessed by the
gametes of spotted mice.

Gynandromorphic individuals, not rare among arthropods, clearly
result from imperfect dominance of the characters of one sex over those
of the other. It is significant that such individuals are especially com-
mon among hybrids, which represent abnormal combinations of gametes
untried and uncertain as to their relative strength. One of the most

198 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

interesting and instructive recorded cases of this sort was reported by
von Siebold ('64). A hive of bees possessed by a certain Herr Eugster
of Constance contained a queen of pure Italian race, which had been
mated with a drone of the common German race. During a period of
four years this liivo produced hundreds of hermaphroditic bees, and it is
important to observe, always from fertilized eggs. For the drones pro-
duced in this hive were of pure Italian race, like the mother ; whereas
the hermaphrodites showed the characters of both parents, though more
often with a })redominance of maternal characters.

The peculiarity, apparently, lay not solely in tlie gametes of the
mother, for in that case the hermaphrodites should have been of pure
Italian race, but rather in the combination of the (male) gametes of
the Italian queen with tlie (female) gametes of the German drone. The
dominance, normal among bees, of tlie female character (borne by the
spermatozoon) was not niiJized in these hybrid hermajjlirodites.

Siebold obtained some two hundred of the hybrid bees and dissected
many of them. They included about all conceivable sorts and degrees of
hermaphroditism. There were true unilateral and antero-posterior her-
maphrodites, as well as others with intermediate or mixed characters, as
in size of eyes, number of joints in antennce, etc. Internal organs were
usually not closely correlated with external in character, but animals
male posteriorly possessed both testes and male copulatory organs, yet
sometimes had an imperfect sting (a female character), or a certain num-
ber of egg tubes fused with the testis, or even an ovary in place of a
testis.

The hermaphrodite character clearly resulted in the case of these Vjees
from imperfect realization of the normal dominance of the female sex
character.

2. Parthenogenetic Organisms.
(rt) General Application.

A study of sex-heredity in parthenogenetic animals shows (1) that in
such animals the female character uniformly dominates over the male
whenever the two are present together, precisely as in the case of hybrid
mice gray coat-color dominates over white ; (2) that when a segregation
of sex-characters occurs in the formation of the gametes, it does so at the
second maturation division of tlie egg (in all but one or two exceptional
cases), and probably at the corresponding stage in spermatogenesis.

In a few species of animals parthenogenesis is the only known method
of reproduction, males never having been observed. But in a far greater

CASTLE: THE HEREDITY OF SEX. 199

number of cases, sexual reproduction (by fertilized eggs) occurs in the
same species with parthenogenesis, the two processes either alternating
with each other, or occurring under different external conditions. Favor-
able conditions in such cases result in parthenogenesis ; unfavorable con-
ditions of any sort may result in sexual reproduction.

1. With a single exception to be discussed presently, we know that in
uninterrupted parthenogenetic reproduction, as it occurs, for example, in
the Daphnidse and Rotifera at certain seasons of the year, the partlieno-
genetic egg forms only one polar cell, and the animal developing from
such an egg is invariably female, or more correctly 9 ((?), the male
character being recessive. In other words, the daughter produced by
parthenogenesis is exactly like her mother. No segregation of sex-char-
acters has taken place in her development. That the male character is
still present in the agamic female is known from the fact that such a
female retains the capacity to produce males under appropriate external
conditions.

2. At the return to sexual reproduction, the parthenogenetic mother
produces eggs which form a second polar cell, and from such eggs (if
unfertilized) only males develop. It is clear, then, that in the second
maturation division the female character has been eliminated from the
egg, for were it still present there, it must from its nature dominate.

In the honey-bee, all the eggs without exception form two polar
bodies, and the unfertilized egg invariably develops into a male. Ac-
cordingly a queen-bee which has not copulated can produce only male
offspring. But one which has copulated produces both male and female
offspring, the former, however, only from unfertilized eggs, the latter
always from fertilized eggs.

In parthenogenetic Rotifera and Crustacea, under optimum external
conditions, the egg develops straightv\^ay after the formation of a single
polar cell, usually while still within the body of the niDtlier, and without
awaiting the occurrence of a second maturation division. No segrega-
tion of sex-characters has yet occurred within the egg, whicli develops,
without the necessity of fertilization, into an agamic female like the
mother. If, however, external conditions are unfavorable, the egg will
not proceed to develop until it has undergone a second maturation divi-
sion. Tlie egg is then capable of development either with or without
fertilization. If it is not fertilized, as must necessarily be the case unless
the mother has copulated, development takes place at once within the
body of tlic mother, and a male is produced. But if the egg is fertilized,
it takes up yolk and acquires a resistant shell, which ordinarily prevents

200 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

its development until the following season ; that is, it becomes a "winter
egg." From such eggs there hatch invariably agamic females.

These facts support the view already advanced, that in parthenogenetic
animals a segregation of sex-characters takes place at the formation of the
second polar cell. The female character passes into the second polar cell,
leaving only the male character in the egg. Hence, if the egg which has
formed two polar cells develops without fertilization, it must develop into
a male. But if such an egg is fertilized, it invariably forms a parthenoge-
netic female, 9 {$), that is, an individual in which the male character is
recessive. Accordingly the functional spermatozoon must in such cases
invariably bear the female character, and this is as invariably dominant
over the male character when the two meet in fertilization.

But we are now confronted with a serious difficulty. The egg, which
has formed two polar cells, we have supposed, is purely male, yet the
animal which develops from it by parthenogenesis produces only gametes
purely female.

The studies of Petrunkewitsch (:0l) on tlie iioney-bee give us a clue
to the solution of this difficulty. The genital gland of the male bee
probably develops, not from any part of the mature egg, but from the
second polar cell, after the union of that body with one of the two prod-
ucts of division of the first polar cell. But the second polar cell con-
tains, according to our hypothesis, only the female character ; the same
is probably true of one of the products of division of the first polar cell,
perhaps of that one which fuses with the second polar cell. If so, the
genital gland of the male bee will contain onli/ the female character, and
in the spermatogenesis of the bee, no segregation of sex-characters will
be found to occur. On the other hand, if the male character is borne by
that derivative of the first polar cell which fuses with the second polar
cell, the body formed by their union will contain both the male and
female characters, and will be homologous with the cleavage nucleus of
a fertilized e^Q. In that case we shall expect to find the occurrence of
a normal process of spermatogenesis with segregation of sex-characters.
If this is so, there doubtless are produced male as well as female sper-
matozoa in the honey-bee, but the latter sort alone can be functional
because the fecundable egg, as we have seen^ invai'iably bears the male
character.

In support of the important observation of Petrunkewitsch may be
cited the earlier observation of Henking ('93). Ho finds that, as a rule,
in insects generally no polar cells arc formed at maturation, but merely
polar nuclei which remain imbedded in the cytoplasm of the egg. The

CASTLE : THE HEREDITY OF SEX. 201

first of these polar nuclei commonly divides about at the time of forma-
tion of the second polar nucleus. There are thus formed three polar
nuclei (or cells), which all lie imbedded in the cytoplasm of the
egg. There regularly takes place a fusion of the inner derivative
of the first polar cell with the second polar cell, exactly as observed
by Petrunkewitsch in the case of the honey-bee. Further develop-
ment of this body was not observed in most of the cases studied by
Henking, though he mentions certain apparently abortive " attempts" at
division by this body. The outer product of division of the first polar
cell was observed regularly to undergo disintegration without further
change, except in a few cases, such as that of the parthenogenetic gall-
wasp, Rhodites rosae, in which all three polar nuclei fuse into a single
body. Henking seems to regard ultimate disintegration as the normal
fate of all the polar nuclei, whether or not conjugation has occurred
among them. This is precisely what the observations of Petrunke-
witsch would lead us to expect in the case of all fertilized eggs, as well
as of parthenogenetic eggs which form but one polar cell. We have no
reason to suppose that Henking ever studied the development of a male
parthenogenetic egg, in which sort alone (in addition possibly to Rhodites).
we should expect to find the genital gland of the embryo developing out
of the conjugated polar nuclei.

If, contrary to the opinion of Petrunkewitsch, it shall be found that in
the male honey-bee the testis develops, not from polar cells, but from a
blastoraere, we may well look for evidence of segregation of the testis fund-
ament early in cleavage. For, if our assumption be correct, that in par-
thenogenetic animals the female character is uniformly dominant over the
male, it will be impossible for the male character to find expression in
the soma of the individual, until the female character has been elimi-
nated from it.

(J) Special Cases.

The explanations offered of sex-heredity in the honey-bee and rotifer
are applicable to all cases known to the writer of normally parthenogenetic
animals, except two. These are the gall-wasp Rhodites rosae, and the
rotifer Hydatina senta.

A. Rhodites rosae

In Rhodites males are very rare, and parthenogenesis is' the normal
method of reproduction. According to Henking, the unfertilized egg in
this species undergoes two maturation divisions, yet the ofispriug devel-

202 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

oping from such eggs must be almost invariably female, because males, as
already stated, are extremely rare. Yet for the very reason that males
are occasionally produced, we are forced to the conclusion that the male
character is present, recessive, in the ordinary female of Rhodites. If so.
the egg does not eliminate the character of that sex at the formation of
the second polar cell, but retains the characters of both sexes, and so has a
formula, (J 9, a supposition for which we have warrant in the mosaic
gametes of spotted mice. In further support of this idea may be men-
tioued the observation of Henking, that in the maturation of the egg of
Khodites no reduction diviaion occurs ; the nucleus of the ovarian egg, the
three polar nuclei, and the nucleus of the mature egg, all alike contain
nine chromosomes each. It is probable, therefore, that normally the
second maturation division in Rhodites is qualitatively like the first, an
equation division, in whicli no segregation of sex characters takes place.
But the occasional production of a male Rhodites indicates tliat the
egg still retains a capacity to eliminate the dominant female character in
maturation, and so to become male, as do the eggs of other partheno-
genetic animals under appropriate conditions.

B. IItdatina senta.

Hydatina senta differs from other iiarthenogenetic animals in the fol-
lowing respects. Its female summer eggs, instead of forming one polar cell,
form none. Its male summer eggs and fecundable (winter) eggs (doubt-
less at the outset one and the same sort), instead of forming tivo polar
cells, form one. It is evident that one of the normal maturation
divisions has in this species been omitted. Clearly it is not the normal
second division, for the single one which occurs is a segregation (or
reduction) division. Manifestly, then, tlie maturation division which
is suppressed in Hydatina is the normal first maturation division of
fecundable eggs, the sole maturation division of eggs not fecundable.

Corroborative evidence of the correctness of this interpretation comes
from an unexpected source, the mammals. Sobotta ('99) finds that in
the egg of the mouse there occurs usually oidy a single maturation
division. Tiiis is the homologue of the setond maturation division of
other animals. When two maturation divisions occur in the same egg,
the second is always of the same type as the single maturation division of
other eggs, and it occurs in a like stage of matuinty of the Graafian
follicle. The single maturation division of one type of egg, and the
second maturation division of the other type, are apparently alike
reduction divisions, for the mitotic spindle, according to Sobotta's figures,

castle: the heredity of sex. 203

bears in these cases about half as many chromosomes as it does in
tlie case of the first maturation division of e<,'<j;s of the less usual
type.

In the mouse, then, and perhaps in other mammals also, the first, or
equation, maturation division is usually, but not always, omitted ; in
Hydatina, however, it appears to be regularly omitted.

C. Artemia salina.

Weismann und Ischikawa ('88) observed the formation of only one
polar cell in the parth^nogenetic eggs of about a dozen different species
of Crustacea as well as in two species of Rotifera. Presumably their
observations were made exclusively on the commoner form of partheno-
genetic egg, the " female summer egg." In the fertilized eggs of three
of the same species of Crustacea (namely, Daphnia longispina, Moina
rectirostris, and M, paradoxa) the same authors found that tivo polar
cells are regularly formed. In the case of the remaining species, includ-
ing Artemia salina, no fertilized eggs were examined.

Maturation of the eggs of Artemia salina has since been studied
by Brauer ('94) and Petrunkewitsch (: 01). Both agree that the.
ovarian egg contains regularly 84 chromosomes, and Petrunkewitsch
finds that the chromosomes are clearly double! Both observers like-
wise are in substantial agreement as to the method and result of
the first maturation division. The first polar cell and the egg con-
tain each 84 double (Petrunkewitscli) chromosomes. No reduction
division has occurred. But from this point on, the two observers differ
in their accounts of what happens. Petrunkewitsch stoutly maintains
that no second maturation division occurs; this is in accord with the
observations of Brauer as to a large majority of the eggs studied by
him, but in a certain number of eggs" he observed the occurrence of a
second maturation division. However, a second polar cell was in no
case extruded. Two nuclei wei'e formed, one peripheral, the other cen-
tral in ])ositioii, and these later came together and fused, exactly as male
and female pronuclei do in the fertilized eggs of other species, thus form-
ing a cleavage nucleus. Each of the two nuclei was found to contain
84 small chromosomes, indicating that at the second maturation division
a separation had taken place between the two parts of the originally
double chromosomes ; in other words, that the second maturation divi-
sion is a reduction division. Moreover, these small or part chromosomes
were observed to remain distinct even after the union of the two nuclei,
the cleavage cells containing 168 small chromosomes, whereas in eggs

VOL. XL. — NO. 4 2

204 BULLETIN : MUSEUM OF COMrARATIVE ZOOLOGY.

which had formed only one polar cell, the cleavage cells contained 84
double chromosomes.

As the eggs of the second type were rare and sometimes showed multi-
polar spindles, Brauer is uncertain whether they were really capable of
normal development or not. Petrunkewitsch is certain that they must
have been purely pathological, for he never observed evidence of any
such second method of maturation in his own preparations, though this
was the especial object of his search, and he worlicd with material from
the same locality, Triest, that had furnished Brauer's material, and in
addition with material from a second locality, Odessa, where male Arte-
mias not infrequently occur.

But a moment's reflection will show that the apparently discordant
results of Brauer and Petrunkewitsch are readilv reconcilable. Brauer's
second type of maturation may have been observed in the rare male (or
fecundable) eggs.

But why, then, it may be asked, did not Petrunkewitsch encounter
this second type of g^^, the especial object of his search, for he exam-
ined material from Odessa, where males frequently occur. Probably be-
cai;se he, as he explicitly states, worked exclusively xoith winter eggs
("Dauereier"), whereas Brauer worked both with summer eggs ("Subi-
taneier ") and with winter eggs. Though Brauer makes no statement
concerning the matter, I confidently hazai'd the conjecture that the
second type of maturation was observed by him only among the summer
eggs, for in no species, so far as I know, in which parthenogenesis
occurs, has the development of a male animal from a winter egg ever
been observed. In parthenogenetic Crustacea, llotifera, and Platodes
alike, there invariably hatches from the winter egg a parthenogenetic
female. Should Petrunkewitsch study the parthenogenetic summer eggs,
instead of the winter eggs, produced by Artemias of the Odessa race, I
venture to predict tliat his search for the second type of maturation
will be abundantly rewarded, at least to this extent, that he will
find the occurrence of two maturation divisions in the male summer
eggs.

It is doubtful whether the other process Qbserved by Brauer, a fusion
of the nucleus of the second polar cell with the egg nucleus, takes place
in the development of the male Artemia. More probably the result of
this process would be the same as that of fertilization, or of an entire
suppression of the second maturation division ; namely, the production
of a female in which the male character is recessive. This view is
quite in harmony with Brauer's own interpretation of his observations.

castle: the heredity of sex. 205

D. Exceptional Parthenogenesis in Bombtx mori, etc.

Occasional parthenogenesis is known to occui' in certain Lepidoptera,
when the mother is forcibly prevented from copulating. The cases
which have been most carefully studied are those of the silk moth,
Bombyx mori, and the gypsy moth, Ocneria dispar. The unfertilized
as well as the fertilized eggs of these species are known, through the in-
vestigations of Platner ('88) and Henking ('92), to undergo tivo matu-
ration divisions. But only an occasional unfertilized egg develops to the
larval stage, — only one in several hundred, or even one in thousands.
A still smaller proportion attain the condition of imagos. These few,
however, are of both sexes, and are capable of reproduction when bred to
ordinai'y individuals (von Siebold, '56).

But it is entirely possible that in the very exceptional eg^ which de-
velops normally, a second maturation division has for some reason failed
to take place, or after it has taken place, a reunion has occurred of the
second polar nucleus with the egg nucleus, as sometimes in the egg of
Artemia, according to Brauer. Such a reunion would bring together
again the sex-characters segregated in maturation, and would produce
the physiological and morphological equivalent of the cleavage nucleus
of a fertilized egg. A similar result would follow the complete sup-
pression of a second maturation division.

The occurrence of individuals of both sexes among the partheno-
genetic offspring of the silk moth and gypsy moth shows that in these
species, as in other normally dioecious animals, there is no uniform
dominance of one sex over the other, such as we find occurrins: amone:
normally parthenogeuetic animals, where the female character regularly
dominates.

V. Abnormal Sex Proportions among Hybrids.

Bateson and Saunders (: 02, p. 139) consider it as "on the whole
against the hypothesis that sex depends chiefly on gametic differentiation
that the statistical distribution of sex among first crosses shows great
departure from the normal proportions." The writer does not share this
opinion, for on the hypothesis of sex advanced in this paper departures of
the sort indicated are capable of ready explanation.

It should be stated, however, that the known cases of this sort ai-e
comparatively rare, whereas the statement of Bateson and Saunders
might lead one to expect their frequent occurrence. The writer knows
of but two cases about which our information is full enough to warrant
statistical examination.

206

bulletin: museum of comparative zoology.

1. Relative Infertility of Certain Combinations of Gametes.

Tutt ('98) reports that in crosses between two nearly related species
of Lepidoptcra, Tcphrosia bistorta and T. crcpuscularia, it has been
found that when l)istorta is the male parent, the hybrid offspring show
a normal distribution as to sex, a slight excess of males. See crosses
[1] and [2] in Table I. r)at in the reciprocal cross, with crepuscularia
(or its dark aberration, delamerensis) as the male parent, the olfspring
are practically all males. See Table I., crosses [3] and [4].

TABLE L

Sex-proportions among hco gcnerntionx of hybrid offsprinrj of Tephrosia lmtorta{V>)
X T. crepusndaria (C) or the dark aberration of the latter, delamerensis (D).
[Statistics of Tutt ('98).]

a «

2i-*
t-> .

x> a
>.«
WO

Wo

Hybrid female offspring of bistorta ^ X delamerensis 9 (cross [2],
Table I.) when crossed with crepuscularia ^ gave (cross [6], Table I.)
a large excess of males, as we should expect on the Mendelian hypothesis
that tlie hybrid furnishes in equal numbers gametes haviug the pure
character of either parent race. For we should exj)ect the combination
of pure delamerensis with crepuscularia gametes, wliich would occur in
half the total cases, to yield offspring having the normal sex-proportion,
a slight excess of males (compare cross [1],, Table I.) ; but pure bistorta
ova fertilized by crepuscularia sperm should yield only male offspring
(compare cross [3], Table 1.). Accordingly the result to be expected is
3+^:19; the observed result is 38 ^J : 11 9 .

To explain the peculiar sex-distribution observed in these crosses, we
may make two simple hypotheses, which, I believe, are warranted by
the facts observed. (1) 7'he sex-character borne by a bistorta (B) gamete

CASTLE : THE HEREDITY OF SEX.

207

dominates in all unions with a crepuscularia (C) or a delamerensis (D)
gamete. Tutt states that the species bistorta " predominates" in crosses
with crepuscularia. It would not be surprising, accordingly, to find
that the sex-character borne by the " predominant " gamete likewise
dominates in the zygote. (2) 0/ the four possible combinations of
gametes, one is sterile ; namely, the combination, ovum B 9 + sperm C
(or D) ^. The three fertile combinations are, —

ovum B S + sperm C (or D) 9,

" C (or D) 9 + " B i,
" " <? + " B 9.

A sufficient justification of this hypothesis is that it explains satisfac-
torily the results observed. Those results are, indeed, peculiar, but
there is no reason to question their accuracy, for they represent the com-
bined and harmonious observations of two independent and competent
experimenters. Calculating the sex-proportion in the various crosses on
the basis of the two hypotheses stated, we obtain the results shown in
Table II. For convenience in comparison, the observed ratios are placed
opposite the calculated ones.

TABLE II.
Sex-'proportions among hybrid offspring of Tephrosia. {Compare Tabic I.)

Cross
(Table I.)

Calculated
Ratio.

Observed
Ratio.

[1] + [2]
[3] +[4]

[5]

[6]

[7]

d-

?

d"

?

1
1
4
2
4

1

a
3
1
3

1 +
158
4-
3+

5+

1
1

0

1
1

The calculation has been made on the basis of a normal equality
between the sexes. As a matter of fact, males are normally slightly in
excess of females, so that it is not surprising to find the calculated num-
ber of males a little too low in nearly all cases. Not improbably the
normal excess of males results from greater mortality among female
larvae; and since the mortality is especially high among hybrid broods.

208

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the normal disparity between the sexes is naturally accentuated. Never-
theless, the differences between calculated and observed ratios are small
in all the crosses except [6] and [7]. Even iu these two cases calculated
and observed results are qualitatirely harmonious. Both indicate a
large excess of males; but the observed excess is larger than tlie
expected one, especially in cross [7].

2. Coupling of Certain Sex and Somatic Characters in the

Germ-cells.

In certain other crosses among Lepidoptera, males and females occur
in their normal proportions, approximate equality, but there is a ten-
dency for the offspring which resemble one parent to be pi'edouiinantly
of one sex, those which resemble the other parent being predominantly
of the other sex. In the following crosses between a species and its
inelanistic aberration, Standfuss ('96) notes the predominance of males
among the offspring having the aberrant form, while females predomi-
nate among those which have the species form.

Psilura monacha X ab. zatima

Aglia tau b X ab. lugens

Grammcsia trigrammica X ab. bilinoa . . .
Angerona jiriinaria X ab. sordiata . ....
Boarmia repandata X ab. conversaria . . .

Offspring like
aberration.

Offspring like
species.

cf

?

d

9

18

186

14

24
4

5
118
14
18
2

2

43

13

3

10

20
89
20
10

18

In these cases, there is clearly an imperfect correlation between the male
sex-character and the aberrant form-character. Is such correlation con-
sistent with the doctrine of gametic difierentiation 1 It is ; correlation,
or "coupling," between members of different ])airs of characters is a
recognized Mendelian phenomenon. Tlius, Correns OOO) has shown that
in crossing Mathiola incana with M. glabra, those hybrid plants which
have villous leaves always bear pink flowers, and tliose which have
glabrous leaves bear white flowers. Leaf character and flower color are
in this case perfectly correlatod, or " coupled," so that they cannot be sepa-
rated in heredity. Similarly, though less perfectly, in the butterfly crosses

CASTLE : THE HEREDITY OF SEX.

209

already cited, the male character is coupled with the aberrant form, and
those gametes of the hybrid which bear the aberrant character bear also
the male sex-character in a majority of cases. This can, I believe, be

TABLE III.

Sex-distribution among offspring of Aglia tau (T) crossed with its dark aberration
lugens (L). [Statistics of Standfuss ('96).]

Generation

Tcf (wild)

9 d- ?
11 13 25

conclusively shown from the statistics of Standfuss. The cross on which
lie made the most extensive observations is that between Aglia tau and
its aberration lugens. The various matings obtained and their outcome,
so far as recorded, are shown in Table III.

210 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Inspection of this table indieiites that lugens is dominant over tau,
for when the two forms are crossed, in Generation III., the offspring
are apparently all of the lugens form, at least Standfuss does not
mention the occurrence of any tans. The resulting fourth genera-
tion hybrids, L in the table, but really L (T), when bred mter se, or
wheu crossed with normal tau, produce, as we should expect, both lugens
and tau forms. See Table III., crosses [IJ, [2], [3]. Likewise the lifth
generation lugens, obtained by intercrossing lugens of the fourth genera-
tion (cross [2]), produce when bred inter se both lugens and tau forms.
See Table III., crosses [4j, |]5]. We have, then, convincing evidence that
tau may be recessive (or latent) in lugens, but lugens is in no case shown
to be latent in tau. Accordingly we have here a case of simple domi-
nance of lugens over tau. The numerical proportions of lugens and tau
in the crosses between those two forms are close to those demanded by
the Mcndelian principles of dominance and segregation. See Table IV.,
Generations III., lY. [1], and IV. [3]. But when hybrid lugens indi-
viduals are bred i?iter se ([2], [4], ami [•>]), considerable discrepancies
occur between calculated and observed results. These discrepancies, I
believe, arise from coupling — in tlie gametes produced by the h^'brids —
of the male character with the lugens character, and of the female char-
acter with the tau character. This explanation acco\ints at the same time
for the peculiar sex-distribution between lugens and tau forms observed
in all the crosses.

Suppose that in the germ-cells of every hybrid individual, D (R), the
segregation of characters occurs in such a way that the male sex-
character passes into the same gamete as the dominant (lugens) form-charac-
ter. Then there will be produced only gametes D $ and II 9 • I <lo
not say that this is invariably so ; indeed, it clearly is not so for any of
the crosses in all cases. It occurs only in a certain number of cases in
each cross, but this number is large enough materially to affect tlie
result. The calculation, however, will be simplified if, for the time
being, we suppose the segregation to occur in all possible cases among
the gametes of hybrids. See Table IV.

In Generation IV., crosses [1] and [3], a hybrid lugens^ I) (R), is
mated with a recessive wild tan, R. The two crosses are reciprocals,
but the outcome is substantially the same in both, so that evidently
whatever peculiarity is possessed by hybrid ova belongs also to hybrid
spermatozoa. Suppose, as suggested, that it be coupling of the male
character with the lugens character. Then we shall have gametes D ^
and R 9 furnished by the hybrid parent, and gametes li $ and K 9

CASTLE : THE HEREDITY OF SEX.

211

furnished by the recessive parent. If gametes of opposite sex always
unite in fertilization, and the sex-character borne by the hybrid gamete
always dominates, the resulting zygotes will be D (R) ^ and E 9 •
See Table IV. But if dominance attaches to the gametes of one
joarent as often as to those of the other, the result will be D(R) ^ +
D (R) 9 + R (? + R 9. Manifestly neither of these results agrees
closely with the one observed, which lies between the two. It seems
probable, then, that if coupling does occur, it occurs not in all possible
cases, but only in a part of them.

TABLE IV.

Sex-distribution among offspring of Aglia tail (R) crossed with its aberration

lugens (D). Compare Table III.

Nature of
Cross.

ti

a

'u

p.

SO

o

"3

&

■?

86
86
75
102
87

Calculated ratio,

D : R, without

coupling.

Total R.

Sex of
Offspring.

Calculated

distribution,

coupling iu all

cases.

Calculated

distribution

coupling iu \ of

cases.

-a

"3
"3

0

43

21.5

37.5
25.5

21.7

a
t
m

O

0?
42
31
38
11
10

45
44
39
52

49

?

41
42
36
50

38

Dor
D(R)

R

D or

D(K)

R

45
44

39
52
49

?

42

50

38

cf

?

41
36

d"

30

31
40
3!^
26
26
46
45
44
/^6

9

13
13
28
21
12
11
33
42
25
31

d"

15

U
5

10
13
13
6
3
5
3

?

28
2S
14
21
24
25
17

8
13

7

in.

IV., [1]

IV., [2]
IV., [3]
v., [4]

v.. [5]

DXR
D(R)XR
D(R)XD(R)
D(R) XR
D (R) X D (R)

allD
1:1
3:1
1:1
3:1
3:1

NoTK. Numerals in italics indicating the observed disiribution are, for conve-
nience in comparison, inserted immediately belovv the calculated numbers.

Suppose that it occurs in only one-third of them ; then the gametes
of the hybrid will be 2 D ^ + D 9 + R c? +' 2 R 9 . If such gametes
meet others all of which are R, as in a cross with a recessive individual,
and if sexual dominance is 2wssessed hi all cases hy the gamete of the hybrid
parent, we get the following distribution of zygotes, 2 D (R) J + D
(R) 9 + R (? + 2 R 9 , whicli, as we have seen, is close to that ob-
served. Compare Table IV., Generation IV. [1] and [3]. On the other
hand, the assumption that sexual dominance is possessed as often by
the gamete of one parent as the other would lead to the result normal

212 bulletin: museum of comparative zoology.

iu the case of other crosses of a hybrid with a recessive form, namely,
D (R) c? + L> (H) 9 +i.i S + I'^ ? > '^vhich is not the result obtained
iu this case.

Hence to explain the exceptional results before us we must assume
two exceptional occurrences, (1) a partial coupling, among the gametes
of the hybrids, of the male sex-character with the dominant (lugeus)
form-character, (2) possession of sexual dominance by the gametes of
the hybrid parent, when that pai'ent is crossed Avith a recessive. But
when two hybrids are intercrossed, as in Generation IV. [2] and Gen-
eration V. [4] and [5], we should not expect to find sexual dominance
possessed uniformly by the gametes of either parent, since both are
hybrids. If, on the other hand, coupling occurs among all the gametes
of both hybrid parents, only hybrid offspring will be produced and in
the normal sex-i)roportion, approximately an equality. See Table IV.
For each parent will produce only gametes D ^ and R 9? f^"d when
opposite sex-characters meet, the zygote formed must always be D R.
The result will be the same whether sexual dominance is possessed ex-
clusively by the gametes of one parent, or is shared equally by those
of both. The fact that in all of the three matings indicated a certain
number of recessive offspring occurs, shows conclusively that coupling
between the male character and the lugens character does not occur in
all possible cases. In Generation IV. [2], the total number of recessive
offspriug is even greater than it should be if no coupling occurred, and
I am at a loss for an explanation of the discrepancy, unless one parent
furnished considerably more than the theoretical number (one-half) of
recessive gametes. But in the two similar crosses of Generation V., the
total number of recessive offspring, on the supposition that no coupling
occurs, is less than half the theoretical. In all three cases the sex-
projwrtion among the offspring, both dominants and recessives, ap-
proximates that which would result from chance combinations of gametes
of two hybrid parents on the suppositions: (1) that there occurs a
coupling of the male character with the lugens character and of the
female with the tau character in approximately one-third of all cases, and
(2) that when coupled gametes meet uncoupled ones in fertilization, the
sex of the former always dominates in the zygote. On these two hypoth-
eses, each hybrid parent will furnish gametes in the proportions 2 D ^J
+ D 9 -fR^ + 2R9, of which one of the two D <? s and one of the
two R 9 s will be coupled. If all possible matings occur and the coupled
gametes are sexually dominant over uncoupled ones, the distribution of
the offspring will be 8 D J : G D 9 : R ^J : 3 R 9- Ou this basis

CASTLE: THE HEREDITY OF SEX. 213

are calculated the numbers inserted in the last four columns of Table IV.,
resfard beiner had for the observed ratio of males to females in each
cross. Thus the males in each cross between hybrid parents are dis-
tributed between D and R in the ratio, 8:1; and the females in the
ratio, 6 : 3.

To sum up, an examination of Table IV. shows in three of the six
crosses considerable discrepancies between the calculated Mendelian
ratios of D to R and those actually observed. In two of the three
crosses mentioned, the discrepancies are satisfactorily accounted for on
the assumption that coupling occurs in about one out of three cases
among the gametes produced by hybrids, on the one hand between the
male sex-character and the aberrant form-character, and on the other
hand between the female sex-character and the species form -character.
The same assumption explains satisfactorily the peculiar sexual distribu-
tion of dominant and recessive forms in all five broods, if we suppose
further that coupled gametes are sexually dominant over uncoupled
ones, and the gametes of hybrids over those of recessive individuals.

The principles of coupling involved in this case may serve to explain
other apparent exceptions to Mendel's law. We have seen how devia-
tions from the expected ratios of dominants to recessives may result
from partial coupling of each with a different sex-character. Complete
coupling of this sort must necessarily result in the production of a
stable or self-perpetuating hybrid form. In case the hybrid form is indis-
tinguishable from a pure dominant, its real nature may be unsuspected,
until a cross with a third form may serve to break the coupling and
bring to light a series of new combinations. How many of our suppos-
edly pure species may be sexually coupled hybrids.'* May it not be that
many aberrant variations (mutations, de Aeries) result from resolution
of these couplings 1

Furthermore, the principle of coupling affords an explanation of the
inheritance of sexual dimorphism in general. There is one set of form-
characters coupled with the male sex-character, another with the female.
Dominance in the zygote of one sex-character necessitates dominance
also of the form-characters which are coupled with it, while the other
sex-character and the form-characters coupled with it together become
recessive.

The author desires to thank Professor E. L. Mark for valuable assist-
ance in the revision of his manuscript and proofs.

214 bulletin: museum of comparative zoology.

VI. Summary.

1. Sex is an attribute of every gamete, whether egg or spermatozoon,
and is not subject to control through environment. It is inherited in
accordance either witli Mendel's law of heredity or with the principle of
mosaic heredity.

2. Mendel's law includes two principles, (1) the principle of domi-
nance in hei'edity of one of two alternative characters over the other, and
(2) the principle of segregation of those characters at the formation of
the gametes.

3. Mosaic inheritance is an important exception to both these prin-
ciples. In this process alternative characters coexist without domi-
nance of either, and pass together (without segregation) into the
gametes.

4. The Mendelian principles of dominance and segregation apply to
the heredity of sex among dioecious animals and plants, but among
hermaphroditic animals and plants mosaic inheritance of sex takes
place.

5. Latency of one sex in the other, among dioecious animals and
plants, is shown by evidence both anatomical and experimental.

6. Segregation of sex, among the gametes of dioecious animals and
plants, is accompixnled l)y morphological differences between the male
and female eggs in Dinophilus and certain Lepidoptera, and possibly
also by dimorphism among the spermatozoa of Paludina,

7. Among dioecious animals, a gamete of one sex can unite, in fertili-
zation, only with one of the opposite sex ; consequently no individuals
are produced from fertilized eggs, which are purely of one sex or the
other.

8. Dominance, in dioecious species, is possessed sometimes by the
male character, sometimes by the female.

9. In parthenogenetic species, the female character invariably domi-
nates, when the characters of both sexes are present together. Accord-
ingly in such species : (a) All fertilized eggs are female, {h) Unfertilized
eggs Avhich are produced without segregation of the sex-characters are
female, (c) ]\Iales develop only from unfertilized eggs from which the
female character has been eliminated.

10. The female character, eliminated from the male partlienogenetic
egg, passes into the testis ; accordingly the spermatozoa bear the female
character, though the individual producing them is in soma purely
male.

CASTLE : THE HEREDITY OF SEX. 215

11. Possibly the testis, in males of partheuogenetic species, contains
the male character as well as tlie female. If so, these are doubtless
segreo-ated in spermatogenesis, but only the female spermatozoa can be
functional, because only male fecundable eggs are produced by such
species.

12. The segregation of sex-characters takes place in most partheuo-
genetic animals, and doubtless in dioecious animals also, at the second
maturation division (the " reduction division ") of the egg, and probably
at a corresponding stage in spermatogenesis. For (1) eggs which de-
velop without fertilization and without undergoing a second maturation
division contain both the male and the female characters, the former
recessive, the latter dominant; but (2) in normally partheuogenetic
species, eggs which, after undergoing a second maturation division,
develop without fertilization, are always male (except in lUiodites). In
such species the female character regularly passes into the second polar
cell, the male character remaining in the egg. In dioecious animals,
on the other hand, either sex character may remain in the egg after
maturation.

13. In Hydatina senta there is no maturation division homologous
with the first maturation division of the eggs of other animals. A single
maturation division occurs in the male (or fecundable) eggs, but this is
clearly homologous with the second maturation division of other parthe-
uogenetic animals, for in it a segregation of sex-characters takes place.
In the female partheuogenetic egg, no maturation division occurs.

14. The partheuogenetic egg of Rhodites rosae undergoes two matura-
tion divisions, but appai-ently without the occurrence of segregation in
eitlier of them. If segregation does occur in one of the two maturation
divisions, the character retained in the egg must be regularly the female,
because the offspring are uniformly of that sex. In that case, the geni-
tal gland of Ehodites probably develops, as does the testis of the honey-
bee according to Petrunkewitsch, from the fused polar cells.

15. Abnormal sex-proportions among hybrids are capable of explana-
tion, in some cases, on the ground that certain combinations of gametes
are infertile.

16. Sexual dimorphism, in a species, is the result of coupling, in
the zygote and in the gametes, of certain form-characters with one or the
other sex-character. A similar explanation accounts satisfactorily for
abnormal sex-distribution of the offspring, in the case of certain crosses,
between the two parent forms.

216 bulletin: museum of COMrAKATIVE ZOOLOGY.

BIBLIOGRAPHY.

Bateson, W., and Saunders, E. R.

:02. Experimental Studies in tlic Physiology of Heredity. Reports to the
Evolution Committee of the Royal Society. Report I., 160 pp. London.

Bordage, E.

'98. Variation sexuelle consecutive a une mutilation chcz le Papayer commun.
Conipt. Rendu. Soc. de Biol., ser. 10, torn. .5, pp. 708-710.

Brauer, A.

'94. Zur Kenntniss der Reifung des partlienogenetisch sich entwickelnden
Eies von Artemia salina. Arch. f. mikr. Anat., Bd. 43, Heft 1, 19. Febr.,
pp. 162-222, Taf. 8-11.

Castle, W. E.

'96. The Early Embryology of Ciona intestinalis, Elcmming (L.). Bull.
Mus. Comp. Zool., Vol. 27, no. 7, pp. 201-280, 13 pi.

Correns, C.

:00. Uebcr Levkojenbastarde. Zur Kenntniss der Grcuzcn der Mendcl'schcn
Regeln. Bot. Centralbl., Bd. 84, pp. 97-113.

Cuenot, L.

'99. Sur la determination du sexe cbez les animaux. Bull. Sci. Prance ct
Belg., tom. 32, pp. 462-535.

Darwin, C.

'76. The Variation of Animals and Plants under Domestication. Second
Edition, revised. N. Y., D. Appletou and Co., 2 Vol., xiv + 473 and x +
495 pp.

Henking, H.

'92. Untersuchuugen iiber die ersten Entwicklungsvorgange in den Eieru
der Insekten. III. Specielles und Allgemeines. Zeit. f. wiss. Zool., Bd.
54, pp. 1-274, Taf. 1-12.

Joseph, G.

'71. Ueber die Zeit der Geschlechtsdifferenzirung in den Eiern einigcr
Liparidinen. 48. Jabresber. d. Schlcs. Gesell. fiir vaterl. Cultur (1870),
pp. 143-146.

castle: the heredity of sex. 217

Korschelt, E.

'87. Die Gattung Diuophilus und derbei ilir auftretende Geschlechtsdiinor-
pbismus. Zool. Jahrb., Bd. 2, pp. 955-967, 1 fig.

Mendel, G.

'66. Versuclie iiber Pflauzen-Hybrideu. Verb, uaturf. Vereines in Briiun,
Bd. 4, Abhaiidl., pp. 3-47.

Petrunkewitsch, A.

:01. Die Ricbtuugskorper und ibr Scliicksal im befrucbteteu und unbe-
frucbteten Bienenei. Zool. Jahrb., Abtb. f. Anat. u. Outog., Bd. 14, Heft
4, 22. Juli, pp. 573-608, Taf. 43-46.

Petrunkewitsch, A.

:02. Die Reifung der parthenogenetiscbeu Eier von Artemia salina. Anat.
Auz., Bd. 21, No. 9, 27- Mai, pp. 256-263, 4 fig.

Platner, G.

'88. Die erste Entwickhmg befrucbteter und partbenogenetiscber Eier von
Liparis dispar. Biol. Centralbl., Bd. 8, No. 17, 1. Nov., pp. 521-524.

Schenk, L.

:02. Maine Metbode der Gescblecbtsbestimmung. Verb. V. luternat. Zool.-
Congresses zu Berlin, 12-16 Aug. 1901, pp. 363-374.

Schenk, L.

:02? Zusammengefasste Antworteu zur Diskussion iiber seinen Vortrag.
Verb. V. Internat. Zool.-Congresses zu Berlin, 12-16 Aug. 1901, pp.
379-402.

Siebold, C. T. von

'56. Wabre Partbeuogenesis bei Scbmetterliugen und Bienen. Ein Beitrag
zur Fortpflanzungsgescbicbte der Tbiere. Leipzig, vi + 144 pp., 1 Taf.

Siebold, C. T. von

'64. Ueber Zwitterbienen. Zeit. f. wiss. Zool., Bd. 14, Heft 1, pp. 73-80.

Sobotta, J.

'99. Ueber die Bedeutung der mitotiscben Figuren in den Eierstockseieru
der Saugetiere. Festscbr. pbys.-med. Gesell. Wiirzburg, 1899, pp. 185-
192, 1 Taf.

Standfuss, M.

'96. llaudbucb der palaarktiscben Gross-Scbmetterliuge fiir Forscher und
Sammlcr. Jena, G. Fiscbcr. xii + 392 pp., 8 Taf.

Strasburger, E.

:00. Versucbe mit diociscben Pflanzcn in Riicksicbt auf Gescblocbt-sverteil-
ung. Biol Centralbl., Bd. 20, No. 20-24, pp. 657-665, 689-698, 721-
731, 753-785.

218 bulletin: museum of compakative zoology.

Tschermak, E.

:00. Ucber Kiiustliclic Kreuzung bei Pisuni sativum. Zeit. f. landwirths.
Versucbswesen in Oester., Bd. 3, pp. 465-555.

Tutt, J. W.

'98. Some Results of Recent Experiments in Hybridising Tepbrosia bistor-
tata and Tepbrosia crepuscnlaria. Trans. Ent. Soc. Lend. I'or tbc Year
1S9S, pt. 1, Apr. 20, pp. 17-4-;i.

Wedekind, W.

:02. Die Partlieuogcnese uud das Scxuali:cesctz. Verb. V. luternat. Zool.-
Congresses zu Berlin, 12-16 Aug. 1901, pp. 403-409.

Weismann, A.

'93. The Germ-pksm, A Theory of Heredity. Translated by W. N. Parker,
xxii + 477 pp., 24 fig. New York.

Weismann, A., und Ischikawa, C.

'88. Wciterc UntcrsucbungCMi zum Zablengesetz der Ricbtungskorper.
Zool. Jabrb., Abtb. f. Anal u. Oniog., Bd. 3, Heft 3, 30. Nov., pp. 575-
610, Taf. 25-28.

Bulletin of the Museum of Comparative Zoology
AT HARVARD COLLEGE.

Vol. XL. No. 6.

THE OPTIC CHIASMA IN TELEOSTS AND ITS BEARING
ON THE ASYMMETRY OF THE HETEROSTOMATA

(FLATFISHES).

Br G. II. Parker.

With One Plate.

CAMBRIDGE, MASS., U. S. A. :

PRINTED FOR THE MUSEUM.

January, 1903.

JAN 28 1905

No. 5. — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY
OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD
COLLEGE. E. L. MARK, DIRECTOR. No. 138.

The Optic Chiasma in Teleosts and its Bearing on the Asymmetry
of the Heterosomata {Flatfishes).

By G. H. Parker.

TABLE OF CONTENTS.

PAGE

I. Introduction 221

II. Positions of the nerves in the
chiasmata of symmetrical

teleosts 222

III. Positions of the nerves in the

PAGE

chiasmata of the Hetero-
somata 224

IV. The asymmetry of the Hetero-
somata 233

V. Summary 238

Bibliography 239

I. Introduction.

The optic chiasma in the great majority of teleosts is formed by a
crossing of the optic nerves without an intermingUng of their fibres ;
hence these vertebrates are pecuHar in that the two optic nerves can
be readily dissected apart even at the chiasma. Since the organs con-
nected by these nerves — tlie eyes and the optic lobes — are, as a rule,
symmetrically disposed, it would seem a matter of indifference whether
an optic nerve in its course from the eye to the optic lobe should pass
in the chiasma dorsally or ventrally to" the other optic nerve. Appar-
ently very little attention has been given to this relation, for a search
through the papers on the cranial nerves of fishes has yielded only a
few scattered observations and general statements unsupported by much
evidence. Stannius ('49, p. 12) declared that for the most part the
nerve from the left side of the brain, that is, the right nerve,^ is dorsal at

1 There has been some confusion in the use of the terms right and lejl as applied
to the optic nerves. Some authors, particularly tlie older ones, designate the nerve
right or left depending upon the side of tlie brain from which it arises ; others use
these terms in accordance with- the eye to which the nerve is attached. In this
paper the nerves are termed right or left depending upon their attachment to the
right or to the left eye.

VOL. XL. — NO. 5 1

222 bulletin: museum of comparative zoology.

the chiasma, but he farther remarked that this relation is not constant,
and that individual difi'erences occur. Owen ('66, p. 300) observed that
the nerves cross each other without interchange of fibres, and that some-
times the nerve of the right eye is dorsal, as in the hake, and sometimes
that "of the left, as in the halibut. He added in a note that both con-
ditions had been seen in different individuals of the cod. Gegenbaur
('98, p. 796), in his recent comparative anatomy, reiterates the chief
statement made by Stannius ; namely, that the right nerve is usually
dorsal, but he cites no examples supporting this opinion. C. J. Ilerrick
('99, p. 394), in his work on Menidia, remarks that in this fish the
left nerve is dorsal, as " is typical for teleostomes," and in this state-
ment I understand him to mean the nerve connected with the left eye,
an interpretation already put on this passage by Cole and Johnstone
(:01, p. 116). Finally Greeff (: GO, p. 25), in the new edition of the
Graefe-Saemisch Handbuch der Augenheilkunde, reaffirms the statement
originally made by Stannius that the right nerve is dorsal. Thus there
is a difference of opinion as to which nerve usually is dorsal, — a con-
dition of affairs that can be cleared up only by reinvestigation.

Much of the material upon which the following studies were made,
was either from the collections of the Museum of Comparative Zoology
or from those of the United States Fish Commission. To the officers of
both these institutions I express my grateful thanks. The materials
obtained from each of the two sources are indicated by foot-notes in con-
nection with the Tables ; material not otherwise designated was obtained
by myself.

II. Positions of the Nerves in the Chiasmata of Symmetrical

Teleosts.

To ascertain whether the right nerves or the left nerves are more
usually dorsal at the chiasmata of symmetrical teleosts, I examined a
hundred specimens each of ten common species. The results of this
examination are given in Table I., in which the columns opposite the
name of the fish show the number of instances of right nerves dorsal
and of left nerves dorsal in a total of one Jbundred cases. These two
conditions, as Owen ('66, p. 300) long ago observed, are well shown in
the cod (Figs. 1 and 2).

This table shows that in six of the ten fishes examined (Fundulus,
Rhombus, Stenotomus, Tautoga, Prionotus, and Melanogrammus) the
left nerve was dorsal about as frequently as the right, the greatest dif-

PAKKER: OPTIC CHIASMA IN TELEOSTS.

223

ference being never more than ten per cent, and that in the remaining
four (Menidia, Pomatomus, Tautogolabrus, and Gadus) this difference
does not exceed in any instance twenty per cent. The differences, more-
over, are not all in favor of one side ; in four species the excess is in left
nerves dorsal, and in six in right nerves. Summing all together, it
appears that in a total of one thousand the right nerve was dorsal 514
times, the left 486. Since in each of the ten species both conditions
are so abundantly represented and are often so nearly equal, one is
justified in concluding that neither nerve is characteristically dorsal,

TABLE I.

"3
o 2

U3 o

O oj
w >

.h1
US

iFundulus majalis (Walbaum). Woods Hole, Mass. . . .
^ Menidia notata (Mitoiiill). Martha's Vineyard, Mass. . .

Rhombus triacanthus (Peck). Boston Markets ....

Pomatomus saltatrix (Linnaeus). Boston Markets . . .
1 Stenotomus chrysops (Linnaeus). Woods Hole, Mass,
1 Tautogolabrus adspersus (Walbaum). Woods Hole, Mass.
1 Tautoga onitis (Linnaeus). Martha's Vineyard, Mass.
^Prionotus carolinus (Linnaeus). Woods Hole, Mass. . .

Gadus morrhua Linnaeus. Boston Markets

Melanogrammus aeglefinus (Linnaeus). Boston Markets .

51
61
53
43

49
43
45
53
40
48

49
39
47
57
51
57
55
47
60
52

Total

486

514

though there is a slight difference in favor of the right. This difference
is so slight, however, that it is probable that a larger numV)er of observa-
tions would give a still closer agreement in numbers, a state indicative
of the unimportance from a physiological standpoint of the dorsal or the
ventral position of a nerve at tlie chiasma.^

Since both types of nerve crossing were abundantly represented in

1 Material supplied from the Biological Laboratory of the United States Fish
Commission, Woods Hole, Mass.

2 A condition of approximate equality, essentially like that just pointed out,
has been. observed by F. H. Herrick ('96, p. 143) in the right or left occurrence of
the crushing claw of the common lobster and by Yerkes (:01, p. 424) in the enlarged
claw of tlie male fiddler crab.

224 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

each of the ten species examined, these species may be said to be
dimorphic in this respect, and one might naturally ask whether this
dimorphism is correlated with other characters such as sex, race, etc.
To the question. Is the dimorphism of the chiasma correlated with sex ?
a conclusive answer can be given, for two of the ten species examined.
In Fundulus of the 51 specimens with left nerves dorsal 29 were females
and 22 males, and of the 49 with right nerves dorsal, 29 were females,
and 20 males. Of the 43 specimens of Tautogolabrus with the left
nerves dorsal 26 were females, and 17 males; and of the 57 with right
nerves dorsal, 26 were females, and 31 were males. These figures show
clearly that there is no close correspondence between the crossing of the
optic nerves and sex.

Whether or not the two types of nerve crossing represent racial differ-
ences,^ cannot at present be decided. In Fundulus, Menidia, Tautogo-
labrus, Tautoga, and Prionotus the whole material came in each instance
from a very restricted area, presumably from a single colony, and yet
both conditions were abundantly present. But evidence of this kind
is obviously very inconclusive, and a satisfactory answer to this question
can probably be obtained only by experiments in breeding.

It thus appears that symmetrical teleosts are from the standpoint of
their optic chiasmata dimorphic, and that their optic nerves cross with-
out either nerves being preponderantly dorsal, a condition of approxi-
mate equality not previously recognized.

III. Positions of the Nerves in the Chiasmata of the

Heterosomata.

From the symmetrical teleosts one naturally turns to the flatfishes as
a group whose lack of symmetry, particularly in the positions of the eyes,
invites study. In the older classifications these fishes constituted one
family, the Pleuronectidae ; in more recent taxonomic works, such as
that by Jordan and Evermann ('96-00), the group is raised to a sub-
order, Heterosomata, and divided into two families, the Pleuronectidae,
or flounders, and the Soleidae, or soles.- This separation agrees well
with the facts to be given in the subsequent part of this paper and will,

1 For a good instance of this kind among the Crustacea, we are indebted to
F. II. Herrick ('95, p. 143), who states tliat " in Alpheus saulcyi, wlicre tlie large
crushing chela can be recognized even before the animal is hatched, the members
of a brood are either right-handed or left-handed ; that is, have the crushing claw
on the same side of the body."

PARKER: OPTIC CHIASMA IN TELEOSTS.

225

therefore, be adopted here. I shall begin with a consideration of the

soles.

The Soleidae, according to Jordan and Evermann ('96-00, p. 2692),

may be divided into three subfamilies : the Achirinae, or American

soles ; the Soleinae, or European soles ; and the Cynoglossinae, or

tongue fishes. The Achirinae and Soleinae have their eyes on the right

side, that is, they are dextral ; the Cynoglossinae are sinistral. I have had

the opportunity of studying representatives of all three subfamilies,

and the positions of their optic nerves at the chiasmata are given in

Table II.

TABLE II.

Family Soleidab (Soles).

Sinistral
individuals.

Dextral
individuals.

Subfamily Achirinae (American Soles).
Species dextral.

Left
nerve
dorsal.

Right
nerve
dorsal.

Left
nerve
dorsal.

Right
nerve
dorsal.

1 Achirus lineatus (Linnaeus). Tampa Bay, Fla.

lAchirus fasciatus Lacepede. Wareham River,

Mass

0
13

1
4

6
3

1
14

8

3

0
14

Subfamily Soleinae (European Soles).
Species dextral.

^Solea solea (Linnaeus). Mersey River, Eng.

Plymouth, Eng.

Subfamily Cynoglossinae (Tongue Fishes).
Species sinistral.

2 Symphurus plagusia (Bloch et Schneider). Rio

Janeiro.
1 Symphurus plagiusa (Linnaeus). Tampa Bay, Fla.

Of the American soles two species were examined, Achirus lineatus
and A. fasciatus. All specimens were dextral, as is typical for this sub-
family, and in both species individuals with the left nerve dorsal, and
others with the right nerve dorsal were found. The numbers given in
the Table indicate an approximate equality in the occurrence of these

1 Material supplied by the United States Commission of Fish and Fisheries.

2 Material from the collections of the Museum of Comparative Zoology.

226 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

two types of chiasmata. The Aiuericau soles may, tlierefore, be said to
be diinorpliic in the same sense that symmetrical toleosts are.

The only representative of the European soles that was studied was
the common sole, Solea solea (Linn.), or, as it is often called, S. vulgaris
Quens. All the specimens at hand were dextral. As the Table shows,
about half had the right nerve dorsal and half the left one dorsal.
Cunningham ('90, p. 68) states that in this species the left nerve is
dorsal, but he makes no mention of the number of specimens examined.
Doubtless his information was based on the inspection of too few
individuals.

Of the tongue fishes, which are typically sinistral, observations were
made on two species, but only in Symphurus plagiusa was the material
sufficient to yield significant results. Here, as in the American and
the European soles, both types of crossing were observed, but specimens
with the left nerve dorsal were much more numerous than those with
the right nerve dorsal.

One may conclude from these facts tliat the species of Soleidae, both
dextral and sinistral, are characterized, like the symmetrical teleosts, by
dimorphism in the structure of their optic chiasmata.

The dimorphism of the Soleidae, since it is accompanied by asymmetry,
gives rise to rather unusual conditions in the optic nerves, and these con-
ditions are characteristic for each of the two types of nerve crossing.
Thus, in a dextral species the individuals with the left nerve (that is,
the nerve connected with the migrating eye) dorsal have in a measure
begun to uncross the optic nerves, since the migration of the left eye
tends to draw the nerve connected with it into a course more nearly
parallel with the right nerve (cf. Fig. 8); whereas individuals witli the
left nerve ventral have emphasized the crossing of the nerves by having
the left nerve drawn around the right one by the migration of the hift
eye. Thus, though the Soleidae are like symmetrical teleosts in hav-
ing two types of optic nerve crossings, their chiasmata are more or
less pronounced, according as the nerve connected with the migrating eye
is ventral or dorsal.

The Pleuronectidae, or flounders, are -divisible into some six sub-
families, three of which are abundantly represented in American waters ;
these are the Hippoglossinae or halibuts, of which some species are
dextral and some sinistral, the Pleuronectinae, or flounders proper, which
with very few exceptions are dextral, and the Psettinae, or turbots,
which are as a rule sinistral. I have had the opportunity of examining
in all twenty-eight species of Pleuronectidae. Of these, three were

PAKKER: OPTIC CHIASMA. IN TELEOSTS.

227

represented each by both dextral and sinistral individuals and their
consideration will be reserved till later. The conditions found in the
remaining twenty-five, each of which was represented by specimens
either exclusively dextral or sinistral, are recorded in Table III.

TABLE III.

Familt Pleitkonbctidae (Floundeks).

Subfamily Hippoglossinae (Halibuts).
Species dextral or sinistral.

^Atheresthes stomias (Jordan and Gilbert). San
Francisco Markets

lEopsetta jordani (Loclcington). San Francisco
Markets

2 Hippoglossoides platessoides (Fabricius). Salem,
Mass

1 Psettichthys melanostictus Girard. San Fran-
cisco Markets

2 Paralichthys brasiliensis (Ranzani). Callao,
Peru

1 Paralichthys dentatus (Linnaeus). Woods Hole,
Mass

1 Paralichthys albiguttus Jordan and Gilbert.
Anclote, Fla

Subfamily Pleuronectinae (Flounders).
Species dextral.

2 Hypsopsetta guttulata (Girard). San Diego,-- Cal.

1 Parophrys vetulus Girard. San Francisco

Markets

1 Isopsetta isolepis (Lockington). San Francisco

Markets

2 Oncopterus darwini Steindachner. East Pata-

gonia

Limandaferruginea (Storer). Massachusetts Bay.
^ Fseudopleuronectes americanus (Walbaum).

Martha's Vineyard, Mass

2 Pleuronectes platessa Linnaeus. Triest, Austria.
2 Liopsetta putnaiui (Gill). Salem, Mass. . . .
1 Glyptooephalus zachirus Lockington. San

Francisco Markets .

Sinistral
individuals.

Dextral
individuals.

Left
nerve
dorsal.

Right
nerve
dorsal.

Left

nerve

dorsal.

Right
nerve
dorsal.

1

0

11

0

1

0

23

0

0

1

0

17

0

11

0

11

0
0
0

51

0

100

0
0
0

6

0

228

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

TABLE III. {continued).

Family Pleuronectidae (Flounders).

Subfamily Psettinae (Turbots).
Species siuistral.

Lophopsetta maculata (Mitchill). Massachu-
setts Bay

1 Platophrys spinosus (Poey). Tampa Bay, Fla.
1 riatophrys pavo Bleeker. Kingsmill Isl. . .

1 Syacium papillosum (Linnaeus). Tampa Bay,

Fla

2 Syacium micrurum Ranzani. Rio Janeiro . .
2 Azevia panamensis (Steindachner). West Pan-
ama

1 Citliarichthys sordidus (Girard). San Francisco

Markets

2 Citliarichthys spilopterus Giinther. Rio Janeiro.
1 Etropus rimosus Goode and Bean. Tampa Bay,

Fla

Sinistral
individuals.

0
0
0

0
0

0
0

34
1

II
1

10

Dextral

individuals.

« SO

An inspection of Table III. will show at onCe that the conditions of
the optic chiasmata in the Pleuronectidae are radically different from
those in the Soleidae and the symmetrical teleosts. In tlie Hippoglos-
sinao the first four species in the table are dextral, and in every one of
their thirty-six representatives the left nerve was dorsal. The three
remaining species are sinistral, and in all of their representatives the right
nerve was dorsal. In like manner the nine species of Pleuronectinae,
all typically dextral, invariably had the left nerve dorsal, and the nine
species of P.settinae, all sinistral, regularly had the right nerve dorsal.
Summarizing the whole table, it may be stated tliat in all the dextral
Pleuronectidae examined the left nerve was dorsal and in all sinistral
ones the right nerve was dorsal. These results agree perfectly Avith the
observations of those few investigators who have recorded the positions
of the optic nerves in flounders. Thus in -the two dextral species, Pleu-
ronectes platessa, studied by Cole and JoJinstone (: 01, p. 116), and
Pseudopleuronectes americanus, studied by Williams (: 02, p. 34), tlie
left nerves are said to be dorsal ; and in the sinistral species, Lophop-
setta maculata, the right nerve is reported by Williams (: 02, p. 34) to

1 Material supplied by the United States Commission of Fish and Fisheries.

2 Material from the collections of the Museum of Comparative Zoology.

PARKER: OPTIC CHIASMA IN TELEOSTS. 229

be dorsal. It is thus evident'that the Pleuronectidae, unlike all other fishes,
do not have a dimorphic condition of the chiasma, but a monomorphic one,
in that destral species, have the left nerve dorsal (Fig. 4) and sinistral
species the right nerve dorsal (Fig. 3). This monomorphic condition
sets the Pleuronectidae in strong contrast not only with the symmet-
rical teleosts, but also with the Soleidae, and justifies the recent tenden-
cies in the taxonomy of fishes to separate these two groups.

So far as the species of Pleuronectidae thus far examined are con-
cerned the generalization reached in the preceding paragraph may be
put in a still simpler way. In the sinistral species the right eye is the
one that migrates and its nerve, as we have seen, is always dorsal ; in
the dextral species the left eye migrates and its nerve is likewise dorsal.
Hence in all Pleuronectidae thus far considered the nerve of the mi-
grating eye is dorsal. This conclusion was reached by Williams (:02,
p. 34) for the two species studied by him, and, as the preceding account
shows, it probably applies generally to such species of the Pleuro-
nectidae as are exclusively dextral or sinistral.

There is a certain mechanical advantage in the dorsal position of the
nerve of the migrating eye. Since this eye moves through the dorsal '
part of the head, its nerve is in a more advantageous position to move
with the eye if dorsal at the chiasma than if ventral. With the
nerve dorsal the effect of the migration, as already pointed out, would
be to bring the two optic nerves into more nearly parallel positions, that
is, to make the chiasma less emphasized than in a symmetrical fish, as
Cole and Johnstone (:01, p. 117) have already observed it to be m
Pleuronectes platessa. Were the nerve ventral, the effect of the migra-
tion would be to wrap it around its fellow so as to accentuate the chiasma.
While this latter condition is not impossible, for, as we have seen, it
exists in many of the Soleidae, it is certainly less advantageous mechani-
cally than the other. One may, therefore, say that the monomorphic
condition of tlie Pleuronectidae is of such a kind as to give a mechanical
advantage to the migrating eye.

The crossing of the optic nerves in young Pleuronectidae is established
in the eggs long before the young fishes hatch and is, I believe, as
uniformly monomorphic there as in the adults. It is well known to all
who have had any experience in rearing young flounders that their
period of greatest mortality is during the migration of the eyes. It
might be supposed that those which die at this stage are flounders whose
migrating eyes had ventral nerves; that, in other words, the flounders
hatched from eggs included animals with the nerve of the migrating eye

230 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

ventral as well as those with that nerve dorsal, and that, when
metamorphosis sets in, only those whose migrating eyes had dorsal nerves
survived. Unfortunately there is no evidence in favor of this view and
much against it. Williams, whose paper (:02) I iiave already quoted,
informs me that in the two species of Pleuronectidae studied by him all
the symmetrical young had the same type of optic nerve crossing that
the metamorphosed individuals had. I have myself determined the
positions of the nerves in the chiasmata of ten newly hatched but un-
metamorphosed Pseudopleuronectes americanus, and in all, the left
nerve was dorsal, as was characteristic of the adult. I therefore believe
that the young Pleuronectidae are hatched with tlie type of optic nerve
crossing characteristic of the adult, and that this may be looked upon
as an adaptation preparatory to the migration of the eye.

Writers in the past, and even recent writers, such as Cunningham
('90, p. 51) ; and Williams ('02, p. 1), often refer to the newly hatched
Pleuronectidae as "perfectly symmetrical" and with "eyes and all
other parts of the head ... as symmetrical as in any other fish." But
the way in which the optic nerves cross sets this question in a somewhat
different Hght. The soles, so far as their optic chiasmata are concerned,
doubtless are hatched in a condition like ordinary fishes, but those
Pleuronectidae that turn in one direction only come from the egg witli
a monomorphic type of nerve crossing that conforms in a mechanically
advantageous way to the ultimate direction of their turning. It is doubt-
ful whether the term symmetrical should be applied to the conditions of
the optic chiasmata of ordinary teleosts, but if it is so applied, the young
Pleuronectidae are not in that sense symmetrical, for of the two kinds of
chiasmata found in each species of ordinary teleosts only one occurs in
each species of Pleuronectidae, and this condition is established some
time before hatching.

It might be inferred from what has gone before that the factors that
determine which eye in the Pleuronectidae will migrate are to be sought
for, not, as is usually done, in the environment when the young fisli
undergoes its metamorphosis, but in the egg at the time when the optic
chiasma is established, or even earlier. ~ But this assumption would
imply that the manner of the crossing of the optic nerves and the mi-
gration of the eye are mutually dependent phenomena. That they are
not invariably so can be shown by the following observations.

A few species of Pleuronectidae are represented by both sinistral and
dextral individuals. Thus Pleuronectes platessa, a dextral species, may,
according to Duucker ('96, p. 83) be occasionally represented by a

PARKER: OPTIC CHIASMA IN TELEOSTS,

231

sinistral specimen, and Pleuronectes flesus, also dextral, has been re-
ported by the same authority (:00, p. 339) as represented in different
localities by from five to thirty-six per cent of sinistral individuals. In
American waters three such species are known : the halibut of the
Atlantic and Pacific coasts, and the bastard halibut and starry flounder
of the California coast. The halibut is typically a dextral species and,
like Pleuronectes platessa, is only rarely represented by sinistral in-
dividuals. The bastard halibut, according to Jordan and Evermami
('96-00, p. 2625), is almost as frequently dextral as sinistral, and the
starry flounder, a dextral species, is said by the same authorities

TABLE IV.

Family PLEtrEONECTiDAB.

Sinistral
individuals.

Dextral
individuals.

>

a> I—'

J3 O

s

> .

1-1

>

So

s

Subfamily Hippoglossinae.

Halibut, Hippoglossus hippoglossus (Linnaeus).
Grand Banks

0
50

0

11

0

12

0

50

0

15

0

2 Bastard halibut, Paraliclithys californicus
(Ayres). San Francisco Markets ....

Subfamily Pleuronectinae.

2 Starry flounder, Platichthys stellatus (Pallas).
San Francisco Markets

{'96-00, p. 2G07) to be frequently sinistral. If now the determina-
tions as to which optic nerve shall be dorsal at the chiasma and as to
which eye shall subsequently migrate are dependent phenomena, it
follows that in those species in which the left eye migrates in some
individuals and the right one in others, there should be found two
corresponding types of nerve crossings. In ascertaining whether such
is the case or not, I examined specimens of the three American species
mentioned ; the results of this examination are given in Table IV.

1 Atypical individuals are indicated by italic numerals.

2 Material supplied in part by the United States Commission of Fish and
Fisheries.

232 bulletin: museum of comparative zoology.

Of the halibut, liippoglossus hippoglossus, thirteen specimens were
examined, twelve dcxtrnl and one sinistral, and in all the left optic
nerve was dorsal, thus confirming the statement of Owen ('66, p. 300)
for this species. Of the bastard halibut, Paralichthys californicus,
twenty-six were examined, eleven sinistral and fifteen dextral, and in
all the right nerve was dorsal. Of the starry flounder, Platichthys
stellatus, one hundred were examined, fifty sinistral and fifty dextral,
and in all the left nerve was dorsal. It therefore appears that each
of these three species has a monomorphic chiasma irrespective of the
fact that it may be composed in part of sinistral and in part of dextral
individuals, and, therefore, the conclusion is that, at least in these
species, the manner of the crossing of the optic nerves is independent of
the type of migration shown by the eye.

The three species mentioned seem at first sight to be exceptions to
what has been said of the Pleuronectidae in general, but such is not
wholly true. Each species, as in the other Pleuronectidae examined,
has a monomorphic chiasma, and the nerve that is dorsal in each instance
is the one that would reasonably be expected to be. Thus, in the halibut
the species is essentially dextral, for sinistral individuals are extremely
rare,^ and in conformity with this the left nerve is always dorsal. The
bastard flounder belongs to a genus all other American members of which
are sinistral ; it is therefore natural to find that in this species, though
it contains both dextral and sinistral individuals, the rule for a sinistral
form holds, the right nerve being always dorsal. The starry flounder
is a member of the Pleuronectinae, a subfamily in which this species is
almost the only American exception to complete dextrality, and as
usual the rule for dextral species prevails, all left nerves being dorsal.
These species, therefore, conform perfectly to the rule for other Pleu-
ronectidae that prescribes a monomorphic chiasma, and though in them
the dorsal nerve is not always connected with the migrating eye, it is
always connected with that eye which in the greater number or nearest
of kin is the one to migrate. Thus these species are not so exceptional
as they at first appear.

Of the two conditions presented by each -of the three species men-
tioned one may be said to be typical and tbe other atypical. The
typical condition is represented by the dextral halibuts and stai-ry floun-
ders and by the sinistral bastard halibuts ; tlie atypical condition by the

1 The sinistral halibut examined by me was the only individual obtained dur-
iiict the winter of 1900-01 by one of tlic larjicst halibut estabhshmcnts in Boston.
It was certainly a single individual u\ many thousands.

PAKKER: OPTIC CIIIASMA IN TELEOSTS. 233

sinistral halibuts and starry flounders and by the dextral bastard floun-
ders. These two conditions are distinguished not only by differences in
the external symmetry of the fishes, but still more so by the optic chias-
niata. Thus, in a sinistral species, like Paralichthys californicus, the
typical individuals, having their right nerves dorsal, will have their optic
chiasmata somewhat uncrossed (Fig. 5), as already explained in dealing
with the soles (p. 226), and the atypical individuals, having their right
nerves also dorsal, will have their optic crossings emphasized (Fig. 6).
Converse conditions occur, of course, in dextral species, such as Pla-
tichthys stellatus (Figs. 7 and 8).

It might at first sight seem that the relations here pointed out are
like those already noticed in the Soleidae, but such is not precisely the
case. Wlien it is kept in mind that there are two types of cliiasmata
and that these may be combined with eyes either on the right or on the
left side of the head, it is clear that there must be four possible com-
binations. Tlie conditions in any species of sole can be thought of as
a combination of one of two types of nerve crossing with eyes always
on the same side of the head. The conditions in the three species of
Pleuronectidae may be described as a combination of one type of nerve
crossing witli the eyes either on the right or the left side of the head.
It thus follows that the two combinations in any one species of sole
cannot duplicate those in any one species of the Pleuronectidae in which
both dextral and sinistral individuals occur.

IV. The Asymmetry of the Heterosomata.

The older natiiralists assumed generally that the asymmetry of the
flatfishes was simply a question of tire migration of the eye. It is now
being recognized that the problem is a much more complex one. Thus
Cole and Johnstone (: 01, p. 8) have pointed out that the lack of sym-
metry of the mouth is quite independent of that of the eyes, though
both are probably adaptations to side swimming. The different colora-
tions of the two sides of the body, as well as the unsyra metrical form
of the skull, seem to be independent of the migration of the eye. This
is proved in pai't by tlie observations of Bumpus ('98, p. 197), who
noticed that many specimens of Pseudopleuronectes americanus were
marked with dark splotches on their light sides, though otherwise normal,
and also by those of Holt ('94) on a solo in which the typical coloration
and form of skull were present, though the eye had not migrated. The

234 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

independence of the type of chicasma and the kind of migration of the
eye, in some species at least, has been pointed out in this paper. It
thus appears that the asymmetry of a flatfish is made up of numerous
more or less independent elements, which in the typical individual are
brought together by a combination of events, but which may from time
to time show evidence of their independence by appearing in unusual
ways. What the factors are that control these elements in the asym-
metry of the fish is unknown, but how they may be discovered has been
indicated by Agassiz ('79, p. 12), who initiated experiments on the
unmetamorphosed fishes to ascertain the influence of light from below,
experiments which when carried out still further by Cunningham and
MacMunn ('94, p. 791) showed that this factor is of importance in
determining pigmentation.

Although it must be admitted that in the halibut, bastard halibut,
and staiTy flounder the evidence of the independence of the factor or
factoi'S determining the crossing of the optic nerves and those controll-
ing the migrations of the eyes is as complete as it well can be under
the cii'cumstances, it does not follow that in other species these factors
are so unrelated, nor that they have always been independent in tlie
three species named. The fact that in every species of Pleuronectidae that
turns in only one direction (Table III.) the nerve of the migrating eye is
always dorsal shows that there has been at least in the past a very
intimate relation between the process of chiasma formation and that of
eye migration. It seems beyond a doubt that in the ancestral Pleuronec-
tidae the process of forming a chiasma was narrowed down to the produc-
tion of that type which was mechanically most advantageous for the
migrating eye, and thus a stock arose in which a particular type of chiasma
was associated with a particular type of asymmetry. From this stand-
point the occurrence of reversed specimens, as in the three species already
mentioned (Table IV.), cannot be regarded a primitive trait, as implied
by Thilo (:02, p. 30G),but must be looked upon as a new departure, for
all these species show in their optic chiasmata the stamp of an ances-
tral condition uniform for each one.

Although phylogenetic questions, like taxonomic, are seldom well
answered on the basis of single characters, single characters are often
very important in the investigation of these questions. From this
standpoint the crossing of the optic nerves has a significant bearing
on the general questions of the origin and the present classification
of the flatfishes. The flatfishes have undoubtedly descended from sym-
metrical fishes, and, as Johannes MUller ('46) long ago pointed out.

PARKEK: OPTIC CHIASMA IN TELEOSTS. 235

their nearest present relatives are probably the Gadidae. The Gadidae,
however, have a body very differently formed from that of any living
flatfish, and if they were ancestral to the present flatfishes, there must
have been intermediate members whose bodies were flattened sidewise
and were probably symmetrical. A fish of such proportions is seen in
the modern Zeus faber. Without going the length that Thilo (:02)
does and assuming that this fish really represents the forerunners of the
flatfishes, it seems certain that the ancestors of these fishes must have
had much the proportions of Zeus. From fishes of such form the
unsymmetrical flatfishes have doubtless been derived. Their symmet-
rical ancestors, like all other symmetrical teleosts, probably had dimor-
phic chiasmata. That tliis feature was handed on to the flatfishes is
evident from the fact that it still characterizes the whole family of soles.
I am aware that the soles are usually regarded as degraded Pleuronec-
tidae, and they certainly are in many respects degenerate ; but, from the
standpoint of their chiasmata, they certainly present the most primitive
conditions seen in any flatfish, and I believe, therefore, that they are
degenerate descendants of the original stocli of flatfishes that had not
yet passed beyond the stage of dimorphic chiasmata. From this stock
was differentiated the Pleuronectitlae by a process whereby, amongst
other things, a monomorphic chiasma was produced. This type of
chiasma was differentiated in two lines so as to meet the requirements,

(1) of a sinistral type of symmetry, as in the Psettinae, or turbots, and

(2) of a dextral type, as in the Pleuronectinae, or flounders proper. In
the tribes thus established species here and there varied in their sym-
metry as in the starry flounder, etc., but in such instances the char-
acter of the chiasma indicates at once whether the species belongs to
a stock originally sinistral or dextral. Such changes as these must be
looked upon as the most recent realized by the flatfishes.

It would be a matter of great satisfaction if the ancestry of the flat-
fishes could be traced through their fossil remains. Unfortunately the
scantiness of such material renders this impossible, though the occurrence
of a Rhombus in the upper eocene and of a Solea in the miocene points
to the antiquity of these fishes among teleosts.

Throughout the whole of the preceding discussion on the Pleuronec-
tidae, it has been assumed that the dorsal position of the nerve con-
nected with the migrating eye is a real advantage to the animals
possessing it. In fact, the explanation of the prevalence of the mono-
morphic condition in the Pleuronectidae rests upon this assumption. It
is by no means easy to show that this assumption is, as I believe it to be,

VOL. XL. — NO. 5 2

236 bulletin: museum of COMrARATIVE ZOOLOGY.

perfectly sound, for there are not a few species, like tlic starry flounder,
the bastard halibut, etc., in which the ventral position of the nerve of
the migrating eye occurs in many adults. The death rate of these indi-
viduals, as compared with that of individuals having the nerve of the mi-
grating eye dorsal, would, however, be significant. Duncker (: 00, p. 339)
has determined this for Pleuronectes flesus. In a large collection of material
from Plymouth, England, including the dextral and the sinistral indi-
viduals in natural proportion, it was found that among the smaller, and
presumably younger, individuals the sinistral specimens were relatively
more abundant than among the larger ones, the proportion being about
one hundred to eighty-five. As Duncker correctly concludes, the death
rate of the sinistral individuals must therefore be higher than that of the
dextral ones. As this is a dextral species, it follows that individuals in
which the nerve of the migrating eye is ventral are more open to early
death than those in which this nerve is dorsal, and that therefore there
is good reason to suppose that the dorsal position of the nerve of the
migrating eye is a real advantnge in the Pleuronectidae.

Numerous attempts have been made to explain the phylogenetic pro-
cess by which the asymmetry of the flatfish has been established.
Most of these deal with the migration of the eye, and Cuimingham
('90, p. 51 ; '92, p. 193) has set forth in a clear way the two chief lines
of argument. One of these is based upon Darwinian principles, and
the other, which is on the whole favored by Cunningham, involves La-
marckian methods. This second explanation is somewhat elaborated by
Cunningham, in that he has ascribed the migration of the eye chiefly to
the action of the oblique eye muscles. In any fish that was flattened
sidewise and had taken up with side swimming, the oblique muscles
of the eye that faces downward would be continually brought into play
to lift the eye to a position of greater service, and if the effect of this
action could be inherited, the migration of the eye might thus be
accounted for. It would be hazardous in the present state of our knowl-
edge to assert that such changes cannot be inherited, though this does
not prove that they are. Granting that they are handed on from genera-
tion to generation, it is, in my opinion, conceivable that operations such
as those described by Cunningham may have -brought about the migra-
tion of the eye. But with the monomorphic chiasma the question seems
to me wholly diff"erent. The Pleuronectidae have descended from a stock
with two types of optic chiasmata essentially like those of the ])resent
symmetrical teleosts, and of these two types, that one has been retained
whicli in each group is mechanically advantageous for the migration

PAEKER: OPTIC CHIASMA IN TELEOSTS. 237

of the eye. The selection and preservation of this type seems to me
entirely inexplicable from the standpoint of Lamarckian factors, for the
optic nerves are in no way open to muscle influence as the eye is; the
whole change is, in my opinion, at once suggestive of a process of elimi-
nation. Hence I regard the origin of the monomorphic chiasmata of the
Pleuronectidae as an operation in which the Lamarckian factors have
played no part, but which may be entirely explained through natural
selection. Although natural selection seems to be the only way of
accounting for the origin of the monomorphic chiasmata of the Pleu-
ronectidae, I do not wish to be understood to imply that the whole
asymmetry of the flatfishes has been thus produced. I can see no
reason why continued muscle action may not in the end modify the
position of an eye or why some direct influence of the environment, such
as light, may not have much to do with pigmentation; nor am I con-
vinced that such changes may not be inlierited.

It seems to me entirely possible from our present knowledge that the
asymmetry of a flatfish may be in part the result of tlie action of La-
marckian factors and in part the outcome of natural selection, for these
two operations are not at all incompatible and may perfectly well work
together. But what I wish particularly to point out in this connection
is that in the origin of the monomorphic chiasmata of the Pleuronectidae
natural selection seems to be the only available means.

From another standpoint the flatfishes are biologically interesting.
Their asymmetry is of a very pronounced type, and its particular phase
sometimes characterizes a whole tribe, as the dextral Pleuronectinae
and the sinistral Psettinae. Notwithstanding this evidence of general
stability, species may occur almost anywhere among modern forms in
which a complete reversal of symmetry of external characters at least
may exist. This is well shown in P-leuronectes flesus, Platichthys stel-
latus, etc., and indicates that this group of animals is open to discon^
tinuous variation of a profound and fundamental kind. Flatfishes are
not peculiar in this respect, for discontinuous variation, as Bateson ('94)
has pointed out, has long been recognized in other groups. Thus in
the gasteropods reversed (sinistral) shells of the common Buccinum
and of the European garden snail have long been known. Reversed
specimens of this kind may establish themselves as a special race, as in
the case of Fusus antiquus of Vigo Bay, Spain. Sometimes whole
species are characterized by reversal, as among the Pupas, or even whole
genera, as in Clausilia and Physa. Not only do the gasteropods show
these differences, but some lamellibranchs, like Chama, are also reversed.

238 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Among arthropods the presence of enlarged chelae on one or other side,
as already mentioned, may involve discontinuity. The same is true of
the sexual asymmetry of the Cyprinodonts as worked out by Garman
('95), and it is probable that the condition in the human being known
as situs transversus viscerum is of like nature. Thus many other ani-
mals show in the reversal of asymmetrical conditions evidence of dis-
continuous variation not unlike that of the flatfishes ; but the flatfishes
di(fer from many of these in the relatively high degree of stability that
their asymmetry possesses, — a condition in part explainable, in my
opinion, as the result of the association of a special form of asymmetry
with certain advantageous internal conditions, like a particular type
of optic nerve crossing.

V. Summary.

1 . In each of ten species of symmetrical tcleosts the optic chiasmata
were dimorphic, in that in some instances the right optic nerve was
dorsal, in others the left.

2. In a thousand cases the right uerve was dorsal 514 times, the
left 486 times.

3. The two types of chiasmata are not correlated with sex.

4. In the Soleidae the chiasmata are also dimorphic, as in symmet-
rical tcleosts,

5. In the Pleuronectidae the chiasmata are monomorphic for each
species ; in dextral species the left nerve is dorsal, in sinistral species
the right uerve is dorsal.

6. All species of Pleuronectidae that turn in only one direction have
their dorsal nerves connected with their migrating eyes. In all species
that have both dextral and sinistral individuals (Table IV.), the dor-
sal nerve is connected with that eye which in the greatest number or
in the nearest of kin migrates.

7. The unmetamorphosed young of the Pleuronectidae are not sym-
metrical in the same sense that symmetrical teleosts are, for they have
monomorphic chiasmata.

8. The Soleidae are not degraded Plcnronectidae, but degenerate
descendants of primitive flatfishes, from which- the Pleuronectidae have
probably been derived.

9. The monomorpliic condition of the optic chiasma of the Pleu-
ronectidae can be explained only on the assumption of natural selection.

10. The flatfishes afford striking examples of discontinuous variation.

PARKER: OPTIC CHIASMA IN TELEOSTS. 239

BIBLIOGRAPHY.

Agassiz, A.

'79. On the Young Stages of Bony Fishes. Proceed. Amer. Acad. Arts and
Sci., Vol. 14, pp. 1-25, pi. 1-9.

Bateson, W.

'94. Materials for the Study of Variation. Loudon and New York.

xvi + 598 pp.

Bumpus, H. C.

'98. A Recent Variety of the Flatfish, and its Bearing upon the Question of
Discontinuous Variation. Science, New Series, Vol. 7, pp. 197-198.

Cole, F. J., and J. Johnstone.

:01. Pleuronectes. L. M. B. C Memoirs on Typical British Marine Plants
and Animals, [No.] 8. London, viii + 252 pp., 11 pi.

Cunningham, J. T.

'90. A Treatise on the Common Sole (Solea vulgaris). Plymouth,
viii + 147 pp., 18 pi.

Cunningham, J. T.

'92. The Evolution of Flatfishes. Natural Science, Vol. 1, pp. 191-199.

Cunningham, J. T., and MacMunn, C. A.
'94. On the Coloration of the Skins of Fishes, especially of Pleuronectidae.
Philos. Trans. Roy. Soc, Loudon, Vol. 184, pp. 765-812, pi. 53-55.

Duncker, G.

'96. Variation und Verwandtschaft von Pleuronectes flesus L. und Pi.
platessaL. Wissenschaftliche Mceresuntersuchungen, Neue Folge, Bd. 1,
Heft 2, pp. 47-103, Taf. 1-4.

Duncker, G.

:00. Variation und Asymmetric bei Pleuronectes flesus L. Wissenschaft-
liche Meeresuntersuchuugen, Neue Folgc, Bd. 3, Abt. Helgoland, Heft
2, pp. 333-406, Taf. 11-14.

Garman, S.

'95. The Cyprinodonts. Mem. Mus. Comp. Zool. Harvard Coll., Vol. 19,
pp. 1-179, 12 pi.

240 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Gegenbaur, C.

'98. Vergleiclieade Anatomic der Wirbcltbiere. Bd. 1. Leipzig, xiv +
978 pp.

Greeff, R.

:00. Die mikroskopische Anatomic dcs Selincrvcn und der Netzhaut. In
Graefe-Saemiscli Ilandbuch der gesamtcn Augcnheilkundc Zweite Auf-
lage. Bd. 1, Kapitei 5. Leipzig.

Herrick, C J.

'99. The Cranial and First Spnial Nerves of Mcnidia ; a Contribution upon
the Nerve Components . of the Bony Fishes. Jour. Comp. Neurology,
Vol. 9, pp. 153-455, pi. U-20.

Herrick, F, H.

'96. The American Lobster. Bull. U. S. Fish Commission for 1895, pp.
1-252, pis. A-J, and 1-54.

Holt, E. W. L.

'94. On an adult Specimen of the Common Sole (Soica vulgaris, Quensel)
with symmetrical Eyes, with a Discussion of its Bearing on Ambicolora-
tion. Proceed. Zool. Soc. London, 1894, pp. 432-446.

Jordan, D. S., and Evermann, B. W.

'96-00. The Fishes of North and Middle America. Bull. U. S. Nat.
Museum, No. 47, Ix + xxx + xxiv + ci + 3312 pp., 392 pi.

Muller, J.

'46. Ueber den Ban und die Grenzen der Ganoiden und iiber das natiirliche
System der Fische. Berlin. 100 pp., 6 Taf.

Owen, R.

'66. On the Anatomy of Vertebrates. Vol. 1. London, xlii + 650 pp.

Parker, G. H.

:01. The Crossing of the Optic Nerves in Teleosts. Biol. Bull., Vol. 2,
pp. 335-336.

Stannius, H.

'49. Das peripherische Nervensystem der Fische. Rostock, iv + 156 pp.,
5 Taf.

Thilo, O.

:02. Die Umbildung am Knocheugcriiste dcF Schollen. Zool. Anzeiger,
Bd. 25, pp. 305-320.

Wilhams, S. R.

:01. The Changes in the Facial Cartilaginous Skeleton of the Flatfishes,
Pscudoplcuronectes amcricanus (a dextral fish) and Bothus maculatus
(sinistral). Science, New Series, Vol. 13, pp. 378, 379-

PARKER: OPTIC CIIIASMA IN TELEOSTS. 241

Williams, S. R.

:02. Cbanges accompanying the Migration of the Eye and Observations on
the Traetus Opticus and Tectum Opticum in Pseudopleuronectes ameri-
canus. Bull. Mus. Comp. Zool. Harvard Coll., Vol. 40, No. 1, pp. 1-
57, 4 pi.

Yerkes, R. M.

:01. A Study of Variation in the Fiddler Crab Gelasimus pugdator Latr.
Proceed. Amer. Acad. Arts and Sci., Vol. 36, pp. 417-442.

242 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY,

P4REEH. — Optic Chiasma.

EXPLANATION OF THE PLATE.

All figures represent dorsal views of brains of teleosts from wliich the cerebral
hemispheres have been removed, thus exposing the optic nerves, chiasmata, and
parts of the tracts. The optic lobes, cerebellum, and medulla are sliown in each
instance, as well as the outline of the eyeballs.

Fig. 1. Gadus morrhua Linn. Left optic nerve dorsal.

Fig. 2. Gadus morrhua Linn. Right optic nerve dorsal.

Fig. 3. Lopliopsetta niaculata (Mitchill). Sinistral species. Right optic nerve

dorsal.
Fig. 4. Pseudopleuronectes americanus ( Walbaum). Dextral species. Left optic

nerve dorsal For the best exposure of the chiasma the brain is viewed

from an antero dorsal position , hence the optic lobes are somewhat

foreshortened.
Fig. 5. Paralichthys californicus (Ayres) Sinistral species. Sinistral individual.

Right optic nerve dorsal.
Fig. 6. Paralichthys californicus (Ayres). Sinistral species. Dextral individual.

Right optic nerve dorsal.
Fig. 7. Platichthys stellatus (Pallas). Dextral species. Sinistral individual.

Left optic nerve dorsal.
Fig. 8. Platichthys stellatus (Pallas). Dextral species. Dextral individual.

Left optic nerve dorsal.

Parker. -Optic Chiasma.

P

G. H. P. DEL.

MEIIOTYPE CO., BOSTON.

Bulletin of the Museum of Comparative Zoology

AT HARVARD COLLEGE.

Vol. XL. No. 6.

POLYDACTYLTSM IN MAN AND THE DOMESTIC ANIMALS,

WITH ESPECIAL REFERENCE TO DIGITAL

VARIATIONS IN SWINE.

By C. W. Pkentiss.

With Twenty-two Plates.

CAMBRIDGE, MASS., U. S. A. :

PRINTED FOR THE MUSEUM.

April, 1003.

MUS. COMP. ZOOL.
LIBRARY

JUN30 1967

HARVARD
UNIVERSITY

No. 6. — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY
OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD
COLLEGE, UNDER THE DIRECTION OF E. L. MARK. No. 141.

Polydactylisiii in Man and the Domestic Animals, with especial
Reference to Digital Variations in Swine.

By C. W. Prentiss.

TABLE OF CONTENTS.

Introduction . .

I. Historical survey . . . .
II. Polydactylism in man . .

A. Literature

B. Observations

III. Polydactylism in carnivora .

A. Literature

B. Observations

IV. Polydactylism in the fowl .
V. Polydactylism in swine . .

A. Literature

B. Observations

1. Manus in which the su-

pernumerary digits are
independent of the nor-
mal digits

a. One supernumerary
digit

b. Two supernumerary
digits

2. Manus in which tlie su-

pernumerary parts may
be more or less closely

PAOB

245
246
251
251
252
255
255
257
259
261
2G1
263

270
270
277

connected with meta-
carpal II

a. One supernumerary
digit

b. Two supernumerary
digits

C. Significance of variations

observed

VI. Polydactylism in ruminants

A. Literature

B. Observations

VII. Polydactylism in the equidae

A. Literature

B. Observations

VIII. Theories of polydactylism .

1. External influences . . .

2. Internal influences . . .

a. Reversion

b. Germinal variation . .
IX. Summary

Bibliography

Explanation of plates

PAOE

284

284

285

288
292
292
293
296
296
298
299
299
300
300
305
307
309
314

Introduction.

The frequent occurrence of extra digits on the extremities of both
man and the domestic animals has attracted the attention of many
anatomists during the past century. Various theories have been ad-
vanced to account for the appearance of these digital abnormalities,
and the opinions expressed by different investigators have been
remarkably contradictory,

VOf.. XI,. — NO. 6 1

246 bulletin: museum of comparative zoology.

Tliroiigh the great kindness of Dr. W. McM. Wooilworth, Keeper of
the Museum of Comparative Zoology at Harvard College, a valuable
collection of polydactyle specimens was placed at my disposal. The
investigation represented by this paper was undertaken with the view
to obtaining, from a study of these abnormalities, some clue as to the
causes leading to their occurrence.

In order to understand the phenomena of polydactylism, and to make
it possible to draw some general conclusions, a comparative study of
such abnormal structures is necessary. It has, tlicrefore, been considered
worth while to collate from the literature brief descriptions of poly-
dactylism in those forms of which we were unable to obtain suitable
material. In reviewing the literature, however, a resume is given of
only those papers which draw important and general conclusions.
Works concerned chiefly with descriptions of polydactylism in individual
animals are treated of in the separate accounts of digital variations in
man and the different domestic animals here referred to.

My research Avas carried on at the Zoological Laboratory of Harvard
University, and to Prof. E. L. jMark are due my sincerest thanks for
both the laboratory privileges I enjoyed, and his own kind direction
and most valuable criticism. To Dr. W. E. Castle I am also indebted
for important criticisms and revision of proof.

I. Historical Survey.

Allusions to polydactylism are to be met with as far back as the time
of Pliny. The first investigator who attempted to collect scientific data
on the subject was Struthers ('63). lie tabulated digital abnormalities
in man, and proved that they were strongly inherited.

Darwin ('76) accounts for the fact that supernumerary digits are more
numerous on the hands tlian on the feet by suggesting that the haml is
more specialized than the foot, and therefore more likely to vary. For
the same reason polydactylism is less common in women, tlie male
showing always greater ditferentiation, and therefore a greater tendency
to variation. Darwin at first assumed polydactylism to be reversion to
a more primitive ancestral condition ; but this assumption was later
withdrawn.

Gegenbaur ('80) criticises the theory which regards polydactylism as
atavistic. His arguments are : (1) tliat other parts of the manus or pes
shew no correlated modifications; (2) tliat man normally possesses five
digits, the typical number for vertebrates, and that the supernumerary

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 247

digits are produced by duplication or intercalation. He regards all
cases of polydactylism in the pig as due to the splitting of one of the
functional digits, and holds therefore that they are monstrosities.
Polydactylism in the horse, he admits, may be atavistic, as (1) the
reversion is to a closely related ancestor ; (2) in Hipparion, a three-
toed fossil horse, the second digit is better developed than the fourtli,
and in polydactyle horses the second digit is the one which most usually
appears ; (3) the rudiments of the extra digits may be present in the
embryo. Atavism Gegenbaur divides into two types: (1) Palaeo-
genetic, or cases wliere the fundament of an organ is always present in
the embryo, and may develop, or may degenerate (centrale of man) ;
(2) Neogenetic, or cases where the organ is absent even in the embryo,
(plialanges of digits ii and v in the horse).

Bardeleben ('85, '85% '86) answers Gegenbaur's objections to re-
versionary polydactylism in man, by advocating the prae-poUex theory.
He maintains that the cartilaginous elements found on the radial side
of the hand and the tibial side of the foot are rudiments of a " prae-
pollex " and " prae-hallux," respectively, and not sesamoids, as had been
previously maintained. Also tliat the pisiform of the carpus and the
tuberositas calcanei of the tarsus represent the rudiments of '' post-
minimi." The manus and pes of primitive mammals were therefore in
his opinion heptadactyle, and polydactylism in man and other mammals
is simply reversion to tins ancestral seven-toed condition.

Boas ('85, '90) considers polydactylism in the horse and ox as due to
reversion. The extra digits formed do not represent simply the per-
sistence of an embryonic condition, for in the polydactyle ox phalanges
are formed in the extra digits, and these elements are normally absent
in the embryo.

Albrecht ('86) points out that in man the greater number of poly-
dactyle cases consist in the duplication of a single digit. This he as-
sumes to be reversion to the bifid fin-rays of the elasmol)ranchs. He
distinguishes this type of polydactylism (false hyperdactyly) from that
found in animals where the number of digits is less than five (true
hyperdactyly). Albrecht is supported in his view by Kollman ('88).

Gegenbaur ('88) states that the discovery of the so-called " prae-pollex "
is not new, but was originally made by Cuvier, and he opposes the "prae-
pollex " theory of Bardeleben on the following grounds : (1) these doubtful
rudiments never form true fingers, and their development is secondaiy
to that of the other digital bones ; (2) polydactylism in man cannot
be explained by it, for supernumerary digits occur on the ulnar as well

248 bulletin: museum of comparative zoology.

as on the radial side of the carpus, and they may also be interpolated
between the other digits ; (3) when the *' prae-pollex " is present, no
correlated changes have been observed in the carpus and other parts of
the manus; (4) its inheritability is no proof of reversion to a palin-
genetic digit, for all monstrosities are inherited. Bardeleben's theory is
therefore an " unbegriindete Behauptung," and polydactylism in man is
due to doubling of the normal digits.

Zander ('91) describes in some detail a case of hexadactylism in man,
concluding that the abnormality was produced by the splitting or dupli-
cation of the fundament of the normal thun^b. He discusses at some
length the different theories which have been advanced to account for
polydactylism. Reversion and the assumption of Bardeleben he rejects
on the following grounds : (1) the rudiments of the prae-pollex are of
secondary formation, and therefore are sesamoids, not digital vestiges ;

(2) Kiikenthal ('89-93) has shown that the sixth digit found in Delphi-
nus leucas is produced by the splitting of the fifth digit in the embryo ;

(3) the most primitive fossil reptiles, the Ichthyopterygia, possessed,
according to Baur ('87), only five digits, and therefore the hexadactylo
condition must have been brought about later, either by duplication
of the primary digits, or by neomorphic development on the ulnar side of
the extremity ; (4) no case has been observed where the " rudiments " of
Bardeleben have developed into supernumerary digits. On the contrary,
the extra fingers of man are usually attached distally, where no rudi-
ments exist. Polydactylism in man, therefore, cannot be atavistic, but
is due to duplication of normal digits. This duplication is caused vi
utero by the pressure of amniotic threads.

This explanation was first proposed by Ahlfeld ('85-86), wlio observed
at the birth- of an infant with a divided thumb that an amniotic thread
■was still present in the fissure of the duplicated digit. This theory
accounts most satisfactorily for the different stages of division to bo met
with in cases of polydactylism and polymelia ; for, the earlier the amnion
presses upon an extremity of tlie embryo, the more complete and far-
reaching will be the duplication produced.

Marsh ('92), in treating of polydactylism inthe horse, gives little weight
to the fact that the ungual phalanges of tlio supernumerary digits never
revert to the partially cleft condition peculiar to the fossil horse. But
he concludes (p. 351) tliat "All the examples of polydactylism in the
horse which the writer has had opportunity to examine critically are
best explained by atavism, and many of them admit of no other ex-
planation. Taken together with their great frequency they clearly indi-

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 249

cate the descent of tlie horse from comparatively recent polydactyle
ancestry."

Blanc ('93) recognizes three distinct classes of polydactylism : (1) Ata-
vistic, or cases where ancestral digits reappear ; (2) Teratological, or
cases in which either normal digits or atavistic supernumerary ones are
duplicated; (3) Heterogenic, or cases belonging to neither (1) nor (2).

(1) Atavistic polydactylism. Bardeleben's theory is accepted without
reservation. Atavism is regarded by Blanc not as the neo-generation of
an ancestral digit, but merely as the development of rudiments normally
present in the embryo. From an examination of digital abnormalities in
mono-, di-, tetra-, and penta-dactylous animals he deduces the follow-
ing general principles : (a) the more simple the extremity, the more
varied and the more divergent from the normal are the forms of Polydac-
tyly. (6) In all species the thoracic limb presents ancestral digits more
frequently than the pelvic does ; this leads to the conclusion that the
manus has become simplified later than the pes. (c) In man the post-
minimus appears more frequently than the prae-pollex or prae-hallux ;
the reverse is true for other animals.

(2) Teratological Polydactylism. The proximate cause of these abnor-
malities Blanc regards as obscure, but he favors Albrecht's ('86) view of
reversion to the pterygian fin rays of selachians ; the single digit of the
higher animals represents two of these rays fused.

(3) Heterogenic polydactylism. This consists usually of the intercala-
tion of extra digits, and the producing cause is unknown.

If Albrecht's view is accepted, Blanc proposes the following classifica-
tion of polydactylism :

1. Atavistic polydactylism.

a. Eeversion to the pentadactyle or mammalian type.

b. Reversion to the heptadactyle or reptilian type.

c. Reversion to forms possessing a double series of phalanges or to

the selachian type.

2, Heterogenic polydactylism.

The supernumerary digits are monstrosities.
Bateson ('94) studied polydactylism in the cat especially, but cites and
fijiures a lar^e number of digital variations in the other domestic animals
and in man. His conclusions are : (1) Polydactylism occurs much more
frequently in certain species than in others. (2) Particular forms of
digital variation are peculiar to particular animals. (3) The abnormal-
ity usually occurs symmetrically placed on both sides of the body, and
often on both fore and hind extremities. (4) There is a tendency for

250 bulletin: museum of comparative zoology.

the abnormal digits to form systems of minor symmetry. (5) Tolydac-
tylism is due to variation, and not to reversion.

Wilson ('96) gives an account of five cases In man where polydactyl-
ism was transmitted through several generations, and conchules that the
abnormalities are generally constant in position, but variable in degree.
In reviewing the different theories advanced to account for polydactyl-
ism ho rejects that of reversion and Bardeleben's prae-pollex theory on
grounds similar to those put forward by Gegenbaur ('80, '88) and Zander
('91), and holds that germinal variation is the proximate cause.

If we summarize the conclusions of the various investigators whose
work we have briefly reviewed, it appears that three explanations liave
been proposed to account for the occurrcince of digital variation : (1) Re-
version, or Atavism. (2) External stimuli (pressure of amnion in xitero).
(3) Internal stimuli (germinal variation)/ A discussion of these theo-
ries Avill be more in place after we have examined for ourselves the types
of polydactylism occurring in the diiferent domestic animals. In pro-
ceeding with this examination we must keep these three theories clearly
in mind. If we are warranted, in rejecting Bardeleben's prae-pollex
theory, the possession of six digits by any domestic animal must be ac-
counted for on grounds other than reversionary. And only in animals
normally possessing fewer than five digits may we look for atavism to
restore, either partially or completely, the typical number of digits;
even in these cases the supernumerary parts may be produced by the
duplication of one or more of the normal digits. Throughout the fol-
lowing pages, therefore, we shall endeavor to determine as definitely as
possible the respective parts which these supposed causes play in pro-
ducing polydactylous abnormalities.

The special point which we have to determine is whether the extra
digits which appear in polydactylism are of palingenetic or neogenetic
origin, — whether they are returns to old structures, or represent new
variations. The term reversion has been loosely used to designate the
general phenomenon of heredity. To avoid confusion I shall limit its
meaning to the abnormal inheritance of palingenetic characters, while
heredity will be used in the. broader sense.- Beginning with the typi-
cal pentadactyle extremity characteristic of man and the Carnivora, we
shall take up in turn those forms in which the number of functional digits
has been reduced (fowl, swine, Euraiuantia, and Equidae).

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 251

II. Polydactylism in Man.

A. Literature.

On account of its importance to the medical profession, polydactylism
has been more often observed in man than in other vertebrates, numerous
cases having been described. Unfortunately the majority of the descrip-
tions are confined to the external appearance of the abnormalities, and to
the structure of the skeletal parts ; the anatomy of the muscles, and
still more important, that of the nerves, has seldom been thoroughly
worked out. Besides the many instances cited by Batesou ('94), the
observations of Morand (1773), Forster ('6l), Struthers ('63*), Ahlfeld
('85-86), Fackenheim ('88),WindIe ('9l), Zander ('9l), and Wilson ('96)
are of especial importance. From the descriptions of the above investi-
gators, it appears that the supernumerary digits are more frequently
found on the manus than on the pes, and on both the right and left
extremities than on one side only. But in those cases where the abnor-
malities are symmetrically placed, the structural conditions of each
extremity may be different from those of tiie others.

The most of the cases observed fall readily into two classes :

(1) A supernumerary digit occurs on the radial side of the extrem-
ity (Fig. A) ; this digit may be of two or three phalanges, and in
the latter case the pollex (i*) is often composed of three elements instead
of two. In most cases where an extra digit is present on the radial side
of the manus, the abnormality is evidently due to a duplication of the
pollex, and it is not possible to say that either of the digits is the normal
thumb. These conditions hold good for the foot as well as the hand.

(2) A supernumerary digit occurs on the ulnar side of the extremity
(Plate 1, Fig. 3). This digit may be (a) complete, of three phalanges,
and having its metacarpal articulating with the unciform (in the manus),
or (b) incomplete, of two or three phalanges which articulate with the
idnar side or distal end of metacai'pal v (minimus) ; in some cases the
extra digit may be merely attached to the minimus loosely by a peduncle
of the skin. Here again the digital variation usually occurs simulta-
neously on both hands, or both feet, or even on hands and feet ; the
conditions on the right and left sides, howevei-, may be different. It is
often impossible to tell whether the fifth or sixth digit is the true mini-
mus. In the well known case originally described by Morand (1773)
the muscular attachments peculiar to the minimus were transferred to

252

BULLETIX: MUSEUM OF COMPAKATIVE ZOOLOGY.

the sixth, or siipeniunierary, digit in the rir/ht hand, leading us to sup-
pose this to be the true minimus. But in the left hand the sixth digit
was rudimentary, and the fifth must therefore be taken as the normal
minimus. These abnormalities, which occur on the ulnar side of the
extremity, may therefore be best explained as due to duplication of the
minimus ; either one of tlie two digits produced may develop into an

III. II.

Fig. a. — Bones of riglit hand of man, showing duplicated thumb, i", i*", pollices;
cun., cuneiform; lun., lunar; os mag., os magnum; trz., trapezium; trz'., accessoiy trape-
zium; <rzrf., trapezoid; scph,, scaphoid; scph'., scpk"., accessory scaphoids; «»., unciform.
(After Bateson.)

apparently normal fifth digit. To this class belong the greater number
of digital abnormalities in man.

There are a few cases of polydactylism in man where one extra digit
has been interpolated. Bateson regards these cases as of doubtful origin.

B. Observations.

Through the kindness of Prof. W. F. Whitney, Curator of the Warren
Museum at the Harvard Medical School, I was permitted to study the
skeletal parts of twelve polydactyle extremities in man, and to obtain

PKENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 253

skiagraphs of the more important abnormalities. In every case ex-
amined the extra digit appeai'ed on the ulnar side of the manus
or pes.

The polydactyle extremities were from late foetal stages ; the carpals and
tarsals, therefore, show little or no calcification, and only the diaphyses
of the digital elements are ossified. The specimens were on exhibition
in the cases of the museum, and so could not be dissected.

Number 912 (Plate 1, Figs. 3-6) is an interesting case. This foetus
shows an extra digit on each hand and foot. In the right manus (Fig.
4) there are only five metacarpals, but the fifth shows evidence of dupli-
cation. It is abnormally large at its distal extremity, and from the ulnar
side of this end projects a bony process. This process is directed some-
what proximad, and with it articulates the supernumerary digit (v*),
which is little more than half the length of v", and consists of but two
phalanges. The other digits of this manus are apparently normal in all
respects.

The structural conditions of the right foot (Fig. 6) are very similar
to those of the right manus. The fifth metatarsal is short, and nearly
as broad as long ; a small protuberance on its ulnar side marks the
point of articulation for the extra digit. The supernumerary digit
shows only two ossification centres, but the incompletely calcified
condition exhibited by the normal digits leads one to suppose that three
phalanges might have been developed eventually. The supernumerary
digit (v*) is somewhat smaller than v", which may be interpreted as
the normal fifth digit.

The left manus (Fig. 3) presents a different skeletal structure. The
first four digits are normal as before, but the supernumerary one (v") is
apparently located on the radial side of the normal fifth digit (v^).
The two are entirely independent of each other, and are of nearly the
same size. From the appearance of the phalanges it is difficult to say
which is the normal digit ; however, the metacarpal of v" is ossified at
its distal end only, thus indicating that it is the interpolated digit.

The digits of the left pes (Fig. 5) resemble in their structure those of
the corresponding manus. There are six distinct digits, and all of the
metatarsal bones are well developed. The four external (ulnar) digits
are similar in structure, each being composed of a metatarsal and two
phalanges ; the ossification centre of the middle phalanx has not yet ap-
peared. The phalanges of digit v^ are smaller, and its metatarsal
bone is shorter than the corresponding skeletal elements of the other
digits. We may therefore consider it as the extra digit, and from the

254 BULLETIN : MUSEUM OF COMPAKATIVE ZOOLOGY.

conditions found in the hands and the right foot, it seems reasonable to
assume that the fifth digit has been duplicated.

These four cases of pulydactylisui are probably all abnormalities pro-
duced by the splitting of tlie fundament of the fifth digit; each instance
differs slightly from the others, but the manus and pes of the right side
are of somewhat similar skeletal structure, and the same is true of the
left appendages. In the appendages of the right side the fifth digit is
incompletely duplicated. In those of the left side the division is com-
plete ; in the manus the metacarpus of the more internal of the two
digits (v") is amorphous, while in the pes digits v" and v** are both
distinct and perfectly developed.

We are not warranted in assuming that either v" or v^ is the extra digit.
In the right hand v" is better developed, in the left hand v*, while in
the feet it is difficult to distinguish any difference between the two.

Number 5809 is a foetus which, like 912, exhibits a hexadactyle con-
dition in all four appendages. Both feet are identical in skeletal struc-
ture with the pes shown in Figure 6 (Plate 1) ; the fifth metatarsal is a
massive bone, as broad as long, and witli it articulate two digits of nearly
equal size, each consisting of two plialanges.

Tlie right manus (Plate 2, Fig. 8) resembles the left manus of number
912 (Plate 1, Fig. 3) ; the digits v" and v* are distinct, but the meta-
carpal of V" is amorphous. The left manus (Fig. 7) exhibits a peculiar
condition. Metacarpal v is abnormally large, especially at its distal
end; with it articulate the two digits v" and v^ v" is apparently
normal in form, size, and tlie number of its phalanges, v*, however, is
small, and directed proximad. Its three phalanges are small and the
distal one is double.

There are, thus, three instances in which digit v is incompletely
duplicated, and a single case in which tlicre is complete splitting of this
digit. Here, too, we are unable to say with certainty that cither v" or
V* is the extra digit.

In a third foetus, number 913 of the Warren Museum, only the
left manus and right pes were preserved. The manus (Plate 2, Fig. 9)
has a small supernumerary digit (v'') on the ulnar side of meta-
carpal V, but not articulating with it. Tliis digit is composed of three
skeletal elements, of which tlie two distal from their form may be inter-
preted as representing the first and third phalanges. Tlie proximal
element is a small nodule of bone, and may be the rudiment of a
metacarpal. IMetacarpal v is apparently normal, as is the digit v".

The right pes of the same foetus (Plate 2, Fig. 10) has six distinct

PRENTISS : POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 255

digits. Digits v" and v* show ossification centres of only one phalanx,
while in ii, iii, and iv, two or three may be seen. This may indicate
that the development of digits v" and v* had been retarded, v* is slightly
smaller than v", but otherwise their skeletal structure is identical.

Figures 1 and 2 (Plate 1) show a pair of feet from a fourth foetus
(number 6730), in both of which six distinct digits are present. The
right pes (Fig. 1) is noteworthy because of the condition of metatarsals
v" and V* ; these are nearly connected at their proximal ends, which
project further proximad than any of the other metatarsals. This is
another ground for assuming that v* and v* originated from the same
fundament. In the left foot (Fig. 2) these digits are considerably
smaller than the others and the proximal ends of their metatarsals also
project further proximad, i. e., toward the tarsus ; in both appendages
the first phalanx of digits v" and v* is the only one showing a centre of
ossification.

To sura up our observations on these twelve cases of polydactylism,
we find : (1) the abnormalities in every instance affect the ulnar (fibular)
side of the extremity and probably only the fifth digit ; (2) in five cases
metacarpal (metatai'sal) v bears two digits ; these may be equally well
developed, or the one on the ulnar side may be more or less rudimentary ;
(3) in seven cases v" and v** are distinct from each other, although
showing evidence of a common origin; either one of these'digits may be
completely formed, or rudimentary, and it cannot be said that one of
them is the normal, and the other the abnormal, digit.

There i^ no evidence of reversive modifications in the polydactyle ex-
tremities an account of which has been given here. Even if we admit
that the primitive ancestor of the mammalia was hexadactyle, there are
stili obstacles in the way of accounting for these abnormalities by rever-
sion. A discussion of these points- will be taken up in the theoretical
portion of this paper.

III. Polydactylism in Carnivora.

A. Literature.

Hereditary digital variations in the extremities of the cat were ob-
served by Poulton ('83, '86); the anatomy of the skeletal parts has been
studied by Bateson ('94); and Howe (:02) has given a detailed account
of the general anatomy of a single case. Such abnormalities are com-
paratively rare in the dog, and of the few cases which have been
observed I know of none which have been carefully described. Blanc

256 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

('93) figures a single case in which the hallux was developed and
duplicated.

In Loth the cat and dog the normal manus is composed of five
digits, but the pollex is much reduced in size. In the pes only four
functional digits are present, the liallux being represented by merely a
rudiment of metacarpal i. These animals are therefore tetradactyle in
the pes, and it is there only that we may look for evidence of reversion,
unless we assume the existence of a hexadactyle ancestor.

Most of the digital abnormalities in Carnivora occur on the radial
side of the manus or pes ; digits ii-v remain practically normal in all
cases. This is an important fact when the polydactyle conditions in
other animals are considered, for it shows that the digits which vary are
in most cases those which have been either reduced or modified in the
course of phylogenetic development.

In the pes of the cat the digital abnormalities fall into three classes :

(1) Five digits, each possessing three phalanges (Fig. B).

(2) Six digits, five of them possessing three phalanges each, the sixth,
which resembles a normal pollex (Fig. CT), exhibiting only two.

(3) Six digits, each having three phalanges. This is the condition of
most frequent occurrence ; the digits in this case are usually so formed
that the pes is bilaterally symmetrical. Bateson lays considerable stress
upon this symmetrical condition, which is brought about in the following
manner. Tlie distal phalanges of the normal extremities are retractile,
and are always drawn back to the ulnar side of the second phalanx (that
is, in the right extremity to the right, and in the left to the left). For this
retraction the second phalanx of each digit is hollowed out on the ulnar
side. The supernumerary digits, however, do not conform to this plan,
but their ungual phalanges are drawn back to the other (radial) side of
the manus or pes ; consequently the second phalanx is hollowed out on
the radial side to con-espond. This change in the symmetry of the
phalanges may extend also to the second digit (11).

In the manus of the cat we find the same three types of poly-
dactylism and in addition a fourth type, in which there are seven digits
present. Digits ii-v are always normal; on the radial side of 11 are
three extra digits, the most radial of which is amorphous (Bateson,
'94, Fig. 86, p. 319). Torrey (:02) describes a similar case in which
seven digits appeared, but the most radial was resorbed soon after birth.
In the case described by Howe (:02) three complete extra digits were
developed, which he considers similar in structure to digits iii, iv, and v.
To this class belong the majority of pijlydactyle cats. When six meta-

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 257

carpals are present in the polydactyle manus, the trapezium is ahnost
invariahly duplicated, and the length of the scapholunar is correspond-
ingly increased ; and the same is true respectively of the cuneiform and
navicular in the abnormal pes. \

cub
ec' cun.

IV. III.

IV.

Fig. B. —Right pes of cat, showing hal- Fig. C. — Right pes of cat, showing du-

liix abnormally developed, i, hallux; as(j.:; plicated hallux, i", i^*, duplications of hal-

astragalus; cac, calcaneum; cm6., cuboid; lux; rti^r., astragalus; coc, calcaneum ; CMi.,

ec'c«iM. , ecto-cuneilbrni ; enV<ni., ento-cunei- cuboid; ec'cun., ecto-cuneiform ; en^cun.,

form; jns'cwre., meso-cuneiform ; «a«., navic- ento-cuneiforni ; ms'cun., nieso-cuneiforni;

ular. (After Bateson.) nav., navicular. (After Batesou.)

B. Observations.

Although a number of cases of polydactylism in the cat have come
under my observation, it was not thought necessary to devote especial
study to them, the careful work done by Bateson making that unneces-
sary, Polydactylism in the dog, however, has never been adequately
described. On account of the difficulty of obtaining suitable material,
my own work on these abnormalities is far from being complete.

258 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

Digital variations arc extromely rare in the manus of the dog. Tho
pes, however, is quite often affected, and in the larger breeds (St. Bernard,
mastiff, and collie) the hallux is frequently present. All of the digital
variations which have come under ray observation were of the pes. As
we have seen, this consists of four digits, the hallux being normally
represented by only the proximal end of its metatarsal bone. The four
functional digits remain unmodified in all cases of polydactylism, and
the supernumerary digits occur on the radial side of digit ii, as varia-
tions of the hallux. We may distinguish three classes of these ab-
normalities : (1) Hallux, or "dew-claw," j)resent and formed of two
])halanges articulating with the distal end of a rudimentary metatarsal.
This digit does not articulate with the proximal rudiment of meta-
tarsal I, but is merely held in place by the skin. Six cases were
observed in the shepherd dog, and five cases in the St. Bernard.

(2) Hallux (Fig. D) presenting two well developed phalanges, of
which the proximal articulates with the rudimentary metatarsal bone;
this element is much longer than the normal phalanx. Three cases were
observed in the mastiff, and one case in the Scotch collie.

(3) Hallux present as in (1), and more or less completely duplicated, ex-
hibiting two phalanges and the distal rudiment of a metatarsal. This is
the common condition in the pes of the St. Bernard dog. The duplica-
tion of the hallux may give rise to the rudiment of only a single ungual
phalanx, or there may be complete duplication, with the formation of
two similar digits (Fig. E. i", i''). In some cases tlio two ungual pha-
langes of i" and 1^ bear but a single large claw, which, however, usually
shows evidence of duplication.

The cases of polydactylism which we have observed in Carnivora may
all be accounted for as modifications of the pollex and hallux. Except
for the change in symmetry of the phalanges of the extremities of the
cat, the rest of the manus or pes is unmodified. The conditions fuund
in the vianus of Carnivora are thus similar to the digital variations
whicli occur in the hand of man. In each case a functional, but reduced,
digit is affected. In man, however, it is the miniiuus which is normally
reduced, whereas in Carnivora it is the pollex.

In the pes of Carnivora the conditions are spmewhat different. Only
a vestige of the hallux is normally present ; in cases of polydactylism,
this is developed and duplicated to a greater or less degree. It would
seem, however, that the same underlying cause which produces poly-
dactylism iu the manus (variation of a reduced but functional digit),
brings about also the digital abnormalities in tlie pes (variation of a

PEENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 259

vestigial digit). Whether this underlying cause is reversion, will he
discussed Liter.

cun.

Fig. I). — Left pes of dog, showing hallux
fully developed. I, hallux; asg., astragalus; cac,
calcaneum; cub., cuhoid; ec'c!*?*., ecto-cuneiform;
en' cun., ento-cuneiform; ms'cun., meso-cunei-
fonn; tiav., navicular.

en

VIS

cun.

II.

Fig. E. — Left pes of dog, showing
duplicated hallux, i, rudimentary meta-
tarsal of hallux; i«, i*, accessory digits;
asg., astragalus; cac, calcaneum; cub.,
cuboid; ec't'wn., ecto-cuneiform; en' cun.,
ento-cuneiform; ms'cun., nieso-cunei-
form; nav., navicular.

IV. Polydactylism in the Fowl.

Although the domestic hen is tetradactyle, the fifth digit was lost so
early in phylogeny that it never appears in polydactyle ahnormalities.
As the hallux of the pes is reduced, however, polydactylism is entirely
limited to this digit ; the condition is thus directly comparahle to that

VOL. XL. — NO. 6 2

260 bulletin: museum of comparative zoology.

fouud in the pes of the dog and cat. The skeletal parts of the poly-
dactyle pes have been described l)y Cowpcr ('89), Howes ('92),
Bateson ('94), and Anthony (99). Tlie last-named writer also ex-
amined the pedal musculature of the Dorking.

Polydactylism, generally rare in birds, is quite common among the
Gallinaceae, especially tlie domestic fowl. It has become a fixed
chai'acteristic of the Dorking breed, and also occurs quite constantly
in the Houdan variety. In the normal fowl, as is well known, the hallux,
or first digit, is articulated at the side of the tarso-metatarsal, by a dis-
tinct rudimentary metatarsal element. Digits ii-iv have their meta-
tarsals fused together ; v is entirely wanting. In nearly all cases of
polydactylism in the fowl a supernumerary digit (sometimes two)
occurs on the tibial side of the hallux. The abnormalities may bo
grouped into three classes :

(1) Pes of five digits, metatarsal i bearing a normal hallux, and
tibial to this a digit of three phalanges (Cowper, '89, p. 249). This
is the most common condition.

(2) Pes of five digits ; the supernumerary digit is borne upon the
proximal phalanx of the hallux instead of articulating with its meta-
carpal. This condition is quite frequent.

(3) Pes of five digits; the hallux being completely divided into
two digits of two or three phalanges each (Howes, '92, Fig. 5).

Single cases have been described in which two extra digits occur.
Of these, one possesses three phalanges, is placed at the tibial side of
the hallux, and has an independent articulation with the tarso-meta-
tarsus ; the other exhibits only two phalanges and is formed by the
more or less complete duplication of the hallux.

Bateson and Saunders (:02) by crossing the polydactyloua Dorking
fowl with white and brown Leghorn varieties, found tliat in the resulting
offspring the polydactylous character is dominant, though not completely
so, over the normal pes of the Leghorn. In addition, the superjiumerary
digits of the crosshreds varied greatly from their structure iti the normal
Dorking. They are described as follows (p. 97) :

"When present the two hind toes may consist, as in the normal Dorking, of
a short toe, like the hallux of a 4-toe(l bird, with a long niany-joiiited digit
proximal to it pointing upwards. Tlie two, however, may often be both short,
pointing downwards, never both long. This condition ranges through many
stages of bigeniination down to mere bifidity of the nail. A form very rarely
seen is an elongation of the hallux without any extra toe being present.^

1 "[A chick lias lately occurrefl with n 'lonq' hallux bigeminus of tin's sort —
probably a hitherto unrecorded form.] March, 1902."

PEENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 261

" In such a hallux there is increase in the number of phalangeal joints.
This of course corresponds to the three-jointed pollex in man. ... In the
highest form of the reduplication the short toe is itself represented by two
digits, making six in all. Of this, also, there are many grades.

" Lastly, any of these conditions may be seen on one foot only, while the
other foot shows one of the other states or is normally four-toed. Generally
speaking, however, there is a fairly close symmetrical agreement between the
two feet."

Thus we see that a single cross between the Dorking and Leghorn
varieties produces all of the polydactylous abnormalities which investi-
gators have so far observed in the fowl.

The conditions presented are interesting and noteworthy from their
structural similarity to the digital variations found in man and the
Carnivora. For here, too, we find that the abnormalities are mainly
confined to a reduced or modified digit, which becomes partially or
completely doubled.

Howes ('92) and Anthony ('99) regard these abnormalities as due to
the splitting of the hallux, not as reversions to a five or six-toed ances-
tor. Bateson and Saunders (:02, p. 137) evidently agree with them, for
besides their allusions to " the reduplication " of the hallux, they class
the abnormalities as ''new characters " — "a palpable sport" (p. 137).

The significance of their experiments and the bearing of " Mendel's
law " upon polydactylism will be discussed later with other theoretical
considerations.

V. Polydactylism in Swine.

A. Literature.

Although polydactylism is quite common in the pig, and many cases
have been recorded, few careful descrii^tions have been given, and those
deal only with the skeletal parts. As a consequence, very conflicting
statements are made by different authors concerning the causes produc-
tive of the conditions, some maintaining that polydactylism in the pig is
atavistic, others that it is due to duplication of the whole foot, and still
others that it is to be accounted for only by haphazard variation.
Geoffrey St. Hilaire ('32-37), Gurlt ('77), Gegenbaur ('80), Bateson
('94), and Werner ('97) have observed instances of digital variation in
swine. Otto ('41), Ercolani ('8l), and Blanc ('93) have given good
descriptions of the skeletal parts of a few cases.

Ercolani obtained data as to the skeletal structure in twenty-five

262 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

cases. Of these, there was only one instance \vhere the supernumerary
digits occurred on the posterior extremity. In four cases the abuor-
mahty was found on botii fore feet ; and in all the specimens which he
himself examined, or which were described by other observers, the extra
digits occurred on the radial, or thumb, side of the manus. The ab-
normalities as figured by Ei'colani (Tav. 1, Fig. 1-G) consist in the
presence of from one to three supernumerary digits. He found also
that the trapezium of the carpus was well developed in most cases, and
occasionally duplicated. In two cases, however, it was entirely absent,
and Ercolani tlierefore concludes that its presence in connection with
the supernumerary digits is no proof that polydactylism is atavistic ;
for the trapezium is present also in most normal swine. Its absence is
a deformity by defect and may occur in the normal manus.

Blanc ('93) considers most of the cases of polydactylism in swine as
due to reversion. He figures four types : (1) Manus with an extra digit
of two phalanges, representing the developed pollcx (Fig. 7, p. 70).

(2) An extra digit of three phalanges, which he regards as the pollex
strongly developed ; digit ii is also abnormally large (Fig. 8). (3)
]\ranus resembling (2), but with a small digit of two phalanges and a
rudimentary metacarpal occurring on the radial side of digit i (Fig. 9).
(4) j\fanus of six completely formed digits, the two supernumerary
being large and of nearly equal size (Fig. 10). Blanc considers types

(3) and (4) as reversions to the hexadactyle ancestor of mammals.
Two other cases are figured to illustrate the duplication of digits
I and II.

Gegenbaur ('80) examined two cases of polydactylism in the manus
of the pig. In one specimen the carpals had been entirely removed, in
the other they Avere partly cut away. From this fragmentary material
he draws his conclusion, — that all cases of polydactylism in swine are
monstrosities and not due to atavism. The conclusions of Blanc and
Gegenbaur are thus completely contradictory.

If we reject the prae-pollex theory as untenable, the hexadactyle cases
regarded by Blanc as reversions must be accounted for in some other
way. On the other hand, (iegenbaur bases his arguments on the slender
evidence of two mutilated specimens; there is need therefore of further
investigation into the structural conditions peculiar to polydactyle swine,
before his refutation of reversion can be accepted. In proceeding with
our description of digital abnormalities in the pig we shall keep especially
in mind their bearing on this question.

PRENTISS: rOLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 263

B. Observations.

The thirty-six specimens of polydactylism in the pig which are to bo
described were collected at The Noi-th Pork Packing establishment,
Sonierville, near Boston, Mass., by Mr. Charles Ballard. In certain
cases the luanus was severed from the arm at the inter-carpal joint, and
conseqnently the upper row of carpals was lost. These bones, however,
are fortunately not so important for study as those of the lower row,
which were saved in all but one case.

In preparing the specimens for study they were first dissected merely
enough to allow a spreading of the digits, and were then skiagraphed.

I am indebted to the Director of the Jefferson Physical Laboratory of
Harvard University, and to Professor Sabine for kindly allowing me the
use of electrical apparatus for this purpose. After obtaining skiagraphs
of the more important abnormal types, the muscles and nerves were
dissected. Finally the bones of the carpus and metacarpus were studied
and separately compared, first with the corresponding parts of the nor-
mal manus, and next with those of the fossil swine figured by Kowa-
levsky ('73) and by Scott ('95). By the latter means it w^as possible to
ascertain whether or not the manus of the polydactyle pig reverts to
that of more primitive fossil forms in characters other than the presence
of extra digits.

Before passing to a description of the various abnormal specimens
which have been studied, it may be well to examine the normal manus
of the pig, and compare its skeletal elements with those of its fossil
ancestors.

The poUex, or digit i, is normally absent in all living artiodactyles,
and the remaining digits are arranged in two pairs (Plate 3, Fig. 11).
Of these, iii and iv are large, functional, and of equal length ; ii and
v arc only two thirds as long, and do not ordinarily reach the ground,

II being usually the smaller. Each digit consists of a metacarpal and three
phalanges. The metacarpals of digits in and iv are large and their
proximal extremities interlocked ; iv articulates with the ulnar side of

III and is partially over-lapped proximally by the large process of the
latter. In the same way a radial process from digit iii overlaps meta-
carpal II, and, as we shall see, is a distinguishing mark in the manus of
the modern pig. The phalangeal region of the manus is bilaterally sym-
metrical, the ungual phalanx and hoof being concave on the side facing
the median plane of the manus, and convex on the side turned away

264

bulletin: museum of comparative zoology.

from it. The hoofs of digits iii and iv are united posteriorly by means
of a horny pad.

The carpus (Fig. F) consists of two rows of four bones each ; in the
proximal row occur in succession, passing from the radial to the ulnar
side, the scaphoid, lunar, cuneiform, and pisiform. In tlie distal row,
which chiefly concerns us, the trapezium is most radial in position ; next

cun.

III.

FiQ. F. — Left normal manus of pig, showing carpals and metacarpals, ii-v, meta-
carpals; cun., cuneiform; lun., lunar; os mag., os magnum; jns., pisiform; scph., scaphoid;
trz. trapezium; <r2(/., trapezoid; m»., unciform, to natural size,

come in order the trapezoid, os magnum, and unciform. The trapezium
(Fig. F, trz.) is rudimentary ; it articulates with the postero-lateral sur-
face of the trapezoid and ends distally in a free, pointed process, which
projects distad of the proximal extremity of metacarpal ii. The trape-
zoid {trzd.) is functional but small. It articulates proximally with the
scaphoid, distally with metacarpals ii and in. Its distal extremity is

PKENTISS : POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 265

wedge-shaped and divided into two facets of nearly equal size, the radial
for articulation with metacarpal ii, the ulnar for the large process of
metacarpal iir. The os magnum articulates distally with the third meta-
carpal only ; the unciform has distally a small facet for the ulnar pro-
cess of metacarpal iii, a large one for metacarpal iv, and a small facet
laterally placed for metacarpal v.

lun.

mag.

III.

Fig. G. — Left manus of Ancodus brachyrhynchus, showing carpals and metacarpals.
i-v, first to fifth metacarpals; lun., hmar; os mng., os magiuim; scph., scaphoid; trz.,
trapezium; irztZ., trapezoid; ««., unciform. § natural size. (After Scott.)

If we compare the carpus and metacarpus of the pig with those of
fossil swine (Palaeochoerus and Hyopotamus or Ancodus) figured hy
Kowalevsky ('73) and Scott ('95), we find some remarkable differences.

In Hyopotamus (Ancodus of Kowalevsky) the trapezium (Fig G.) is
nearly as large as the trapezoid, and articulates superiorly with the
scaphoid, iuferiorly with the metacarpal of digit i. The trapezoid has

266 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

only a single facet on its distal end and articulates with metacarpal ii.
The pollex is present and is represented in the figure by metacarpal i.
Digits II and v are relatively large, especially at their proximal extrem-
ities ; II is better developed than v, and occupies the whole distal sur-
face of the trapezoid. It also articulates by a small facet with the
OS magnum.

The third metacarpal is longer than any of the others and proximally
there is no radial process for articulation with the trapezoid. In general
we may say that the digits of the fossil swine are confined chiefly to
their own carpal bones, while in the pig of the present day the third'
metacarpal has developed a radial process which articulates with the
trapezoid and has partially crowded out digit ii. In the same way
metacarpal iv has encroached upon the distal articular surface of the
unciform, and pushed the fiftli digit to oue side ; the third and fourth
digits thus come to occupy most of the carpo-metacarpal articulation in
the modern pig, a condition of evident advantage, as it strengthens the
joint between the carpus and the functional digits.

If complete reversion occurs in the skeletal parts of the pig's manus,
we should expect to find (1) an extra digit of two plialanges articulating
with the trapezium, and (2) metacarpals ii and in articulating with their
proper carpal bones (trapezoid and os magnum respectively) ; (3) meta-
carpal III should be longer than iv, and without a radial process, and
(4) digits II and v should be relatively larger than in the normal manus.

The normal musculature of the manus is quite complex. We need
mention here only those muscles which in the polydactyle manus pre-
sent variations from the normal. Anteriorly we have (1) the radial or
great extensor of the metacarpus (Fig. //, ext. mCcarp. mag.). This is
a large muscle and is inserted by a strong tendon into the proximal
end of metacarpal iii ; (2) the ulnar or oblique extensor of the meta-
carpus (Fig. II, ext. mCcarp. ob.), a small muscle, the tendon of which
crosses that of the magnum obliquely, and is inserted into the proximal
end of metacarpal ii ; (3) the extensor communis digitorum internus
(ext. com. dg.i.), a large muscle inserted by means of three tendons.
The main tendon bifurcates, the radial portion being inserted in the
third phalanx of digit ii ; the remaining portion of the tendon runs
some distance and again bifurcates, the two branches becoming attached
to the ungual phalanges of the third and fourth digits ; (4) the extensor
proprius internus (ext. prp. i.), a much smaller muscle than tlie preceding,
is inserted by two tendons, the larger going to the radial side of the tliird
digit, the smaller to the ungual phalanx of ii ; (5) extensor proprius

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 267

pollicis et indicis {ext.prp.) is a rudimentary muscle in the pig ; it arises
with the extensor metacarpi obhquus, and its threadlike tendon is lost in
that of the extensor communis digitorum internus.

Of the posterior muscles we may mention (1) the flexor perforatus, or
superficial flexor of the digits (Fig. /, fix. perf.) ; this is composed of two

ext. mt'carp.

ext. prp

Fig. n. — Left normal manus of pic^, showing extensor muscles, ext. com, dff. i., ex-
tensor communis digitorum internus; ext. mVcarp. mag., extensor metacarpi majiiuis; ext.
mt'carp. ob., extensor metacarpi obliquus; ext.prp., extensor proprius pollicis et indicis;
ext.prp. i., extensor proprius internus. 1 natural size.

distinct parts, the tendons of which are inserted into the second phalanges
of digits III and iv. These tendons form two sheaths for the large ten-
dons of the flexor perforans muscle {flx-perf.'), the deep flexor of the

268

bulletin: museum of comparative zoology.

digits. This divides into four tendons, two large and two small ; the
two large ones, after passing through the sheaths formed by the perfora-
tus, are inserted into the ungual phalanges of digits three and four ; the
two smaller tendons are attached similarly to the second and fifth digits.
As regards the innervation of the normal manus, we need concern our-
selves with the condition of the median nerve only, by which the digits

-fix. jicrf/

.rHx.perf.

Fig. I. — Left normal manus of pip, showing flexor muscles, fix. ptrf., tendons of
flexor perforatus; fix,, ptrf ., flexor perforans. i natural size.

are chiefly supplied. The trunk of the median nerve (Fig. J, n.m.)
passes between the two flexor muscles at the carpfil joint ; nearly at a level
with the proximal ends of metacarpals ii and v it gives off two lateral
branches (2, 5) to supply these digits. The main nerve, continuing dis-
tally, soon separates into two large branches (3, 4), ■which pass
together along the region between digits in and iv, to which they
are distributed. The lateral branches (2, 5) before pa-ssing to their

PKENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 269

respective digits divide, the larger of the resulting branches innervating
the lateral portions of the third and fourth digits.

In pentadactyle animals (Carnivora and Primates) the median nerve
gives off a fifth branch radial to 2 of the pig's manus, which divides
and supplies the thumb and index. No remains of such a nerve branch
could be detected in dissections of the normal manus of the pig.

n. m.

IV.

III.

Fig. J. — Posterior view of the left normal manus of pig, showing innervation, n. m.,
median nerve; 2-5, four branches of the median nerve supplying the corresponding digits.
§ natural size.

If the polydactyle manus of swine is due to reversion, we might ex-
pect to find reversive modifications in the muscles and nerves, as well as
in the skeletal parts.

The extensor of the thumb and index might be fully developed and
its tendon inserted into the phalanges of digits i and ii, as in penta-

270 bullf:tin: museum of compauative zoology.

dactylo animals ; the oblique extensor of the metacarpus might bo found
inserted into metacarpal i, and the flexor perforans muscle might send
a tendon to digit i. The pollex, if tlius supplied with muscles, should be
innervated by a branch from the radial side of the median nerve. In
examining tlie following cases of polydactylism in the manus of the pig,
we sliall see whetlier these theoretical conditions are ever fulliilcd.

Of the thirty-six instances of polydactylism which were studied, all
were of the manus ; in every case, also, the supernumerary digit oc-
curred on the radial side of the extremity. Digit ii is a])n(iniial in some
cases. The abnormalities might be divided into numerous types accord-
ing to the number and condition of the extra digits ; but as these types
grade into one another, we shall attempt to distinguish but two classes :
(1) cases in which the supernumerary parts are distinct from, and inde-
pendent of, the normal digits; (2) cases, where tliey are more or less
closely connected with digit ii. We shall see that even these are artifi-
cial groups, and that intermediate conditions link together the two. In
the following descriptions, we shall begin with the simplest forms, and
pass iu succession to the more complex types of polydactylism.

1. Manus in which the Super mimerary Digits are Independent of the

Normal Digits.

a. One Supernumerary Digit.

The simplest example of this condition is represented by a single case
(Plate 4, Fig. 12). Externally the extra digit (i) is inconspicuous, but
originally bore a small claw-like hoof. It is composed of two rudimen-
tary phalanges and a spheroidal element, which apparently represents the
distal end of a metacarpal. This does not articulate with the second
metacarpal, but is merely held in place by fibrous tissue and the skin.

In the carpus the trapezium is abnormally long ; it articulates with
the trapezoid laterally, and has a facet proxiraally for the scaplioid ; in
other respects the bones of the manus are normal. The muscles and
nerves are unmodified.

Figure 13 (Plate 5) shows a manus in which the pullox is fully de-
veloped. Of this type, four cases were examined. The pollex (i) is
smaller than digit ii and consists of the metacarpal and two phalanges.
The metacarpal bone articulates with tlie trapezium, which is abnor-
mally large and has three facets : a distal for metacarpal i, a latoial
for the trapezoid, and a proximal for the scaphoid. The relations of the
bones of this digit to those of the rest of the manus are thus identical with
the conditions found in fossil swine and in other pentadactyle animals.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 271

On examining the other skeletal elements of the manus, in order to
determine whether they show reversive modifications, one is at once
struck by the form of the trapezoid (Fig. A", trzd.). Although of normal
size, there is a remarkable change at its distal end ; instead of projecting
as a wedge between metacarpals ii and in (see normal manus, Fig. F,
trzd.., p. 264), and presenting two distal facets nearly equal in size,

cun.

OS mag. .._

II.

pis.

Ull.

III.

IV.

Fig. K. — Anterior view of left polydactyle manus of tlie pi,£^, showing carpals and
metacarpals, i-v, first to fifth metacarpals; cun., cuneifoiin; lun., lunar; os maf/., os
magnum; ^js., pisiform; scj)h., scaphoid; trz., trapezium; trzd., trapezoid; un., unciform.
3 natural size.

there is only one articular surface, which is slightly convex and occu-
pied entirely by metacarpal ii. The trapezoid barely touches metacarpal
III ; its form and relations to the other skeletal parts thus approach those
of the trapezoid of fossil swine (Fig. G, p. 265).

In correspondence with these carpal variations, the metacarpals show
some changes. The metacarpal of digit ii is slightly larger than nor-

272

bulletin: museum of comparative zoology.

mal, and its proximal end is relatively large. In digit iir the radial
process of the metacarpal bone, a special character of the nianns in recent
swine, is greatly reduced, and as a result scarcely touches the trapezoid,
Avhile metacarpal ii comes in contact posteriorly with the os magnum.
The trochlear ridges of the metacarpals are retained, and the phalanges
show no modifications in form.

n. m.

Fin. L. — Posterior view of left pnlydactvle nianiis of tlie pij?, sliowinc; iniiorvatioii.
n. m., median nerve; 1, branch of median nerve supplviiijj tlie supernumerary digit (i).
5 natural size.

The muscles are not much modified, for the extra digit is small and
functionless. In two instances, however, the tendon of the extensor
metacarpi ol)liqiuis muscle is inserted intii the proximal end of digit i.
This is an interesting condition, as in normal five-toed animals this
muscle is likewise always inserted into the metacarpal of the pollex.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 273

The innervation of the extra digit is also noteworthy. The median
nerve (Fig. L, n.in.) gives off on the radial side of its normal divisions
a small additional branch (1). This divides like the other branches,
sending one division to digit ii and the other to the pollex.

Closely resembling the cases just described, are two instances of poly-
dactylism in which the trapezium is fused to the supernumerary meta-
carpal. The extra digit is very small, and the metacarpal articulates
well up on the radial side of the trapezoid. This condition favors the

lun.

cun.

OS mag.

trzd
trz.

Fig. M. — Anterior view of left polydact3-le maiiits of the pic;, showing carpals and
metacarpals, i-v, first to fifth metacarpals; cun., cuneiform; lun., lunar; os ma;/., os
ma{j;num; pis., pisiform; sc/>/t., scaphoid; frz., trapezium ; iracZ., trapezoid; un., unciform.
I natural size.

theory that the trapezium of the manus of the pig may represent the
carpal element plus the rudiment of digit i.

Taking now a step further in our series, we come to a condition
in which the extra digit is still larger and consists of three phalanges
(Plate 6, Fig. 14). The four cases of this type studied showed practi-
cally the same anatomical conditions. Digit ii is relatively larger.
Digit I articulates with the trapezium, which is large and has facets for
the trapezoid, scaphoid, and metacarpal i (Fig. i/, trz.). The trapezoid

274

bulletin: museum of compaeative zoology.

has become enlarged to correspond with the increased size of its digit
(ii) ; it articulates chiefly with metacarpal ii, its facet for ni bein"-
small. The radial process of metacarpal iii is considerably reduced.
In another case (Plate 7, Fig. 15) the trapezium was fused to the
proximal end of metacarpal i.

In Figure 16 (Plate 8) is shown a raanus which exhibits an extremely
interesting structure. The extra digit is identical in its structure with
that of the manus figured in Plate 6, but the second digit is very
strongly developed, and is in fact more massive than either iii or iv.

trzfl. OS mag.

Fig. N. — Anterior view of left polydactj-le manus of the pig, showing lower row of
carpals and metacarpals, i-v, metacarpals; os mag., ps magnum; trz., trapezium; trzd,,
trapezoid ; un., unciform. | natural size.

Its hoof is large, convex on its radial, and flat on its ulnar surface ; it is
entirely independent of the hoof of digit in. The third plialanx of
digit II is al.so convex on its radial side; that of digit iii is indiff"erent,
and its hoof is flat on eitlier side. The other digits are apparently
normal. Of the carpals, the trapezium (Fig. N, trz.) is large and artic-
ulates with tiie scaphoid, trapezoid, and metacarpal i. The trapezoid
(trzd.) is nearly as large as the os magnum (os mag.), and its single
distal facet articulates with only metacarpal ii.

Of the metacarpals, i is small but v>ell formed ; ii is larger than in
at its distal end and shows evidence there of pathological hypertrophy.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 275

Metacarpal in has scarcely any radial enlargement at its proximal end
and does not articulate with the trapezoid.

Turning now to the musculature of these cases in which the super-
numerary digit is composed of three phalanges, we find that in every

ext. mfcarp.

I— — cxt. prp.

mt'carp. ob.

com. dg. i.

Fig. 0. — Anterior view of left polydactyle manus of the pig, showing extensor muscles.
ext. com. dg. t., extensor communis digitorum iuternus; ext. mt'carp. mag., extensor meta-
carpi magnus; ext. mt'carp. ob., extensor metacarpi obliquus; ext. prp., extensor proprius
poilicis et indicis; ext. pip. i., extensor proprius internus. iS natural size.

case the extensor metacarpi obliquus (Fig. 0, ext. vit'carp. ob.) has
shifted its insertion from the second to the first metacarpal ; the ex-
tensor proprius poilicis et indicis (Fig. 0, ext. prp.), which normally
is extremely rudimentary, is in two cases inserted into the distal
phalanges of digit i.

VOL. XL. — KO. 6 8

276

bulletin: museum of comparative zoology.

The flexors exhibit a very interesting condition ; in all cases the deep
flexor, or perforans (Fig. P, fix. perf.'), sends a small tendon lo the
extra digit ; this apparently is not formed by the division of the tendon
■which supplies digit ii, but is given off from the main tendon independ-
ently and more proximally. It may represent the radial portion of the
flexor perforans. In the three cases where the second digit is abnormally

Fig. p. — Posterior view of left polydactyle manus, showing flexor muscles, fix. pcrf.,
flexor perforatus; Jlx.perf., flexor perforans. I natural size.

large, the tendon of the perforans supplying this digit is much stronger
than usual. The superficial flexor, or perforatus, is normal in most cases,
but in one instance has three insertions, an extra tendon going to the
second digit.

The innervation of these cases is identical with that shown in Fig. L.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 277

A still greater development of digit i was exhibited in two of the
cases studied. Such a mauus is shown in Figure 17 (Plate 9). The three
phalanges and metacarpal of digit i are larger thau those of digit ii ;
the digit is borne on the trapezium, which is also large and articulates
"with the scaphoid and trapezoid. The other skeletal elements of the
manus are normal in structui'e. The musculature and innervation
of these two cases were identical with those shown in Figures 0, P,
and L.

The cases thus far described possess but one extra digit. Continuing
the examination of the polydactyle series, it is found that this digit may
be partially or completely doubled.

b. Two Supernumerary Digits.

Ten cases were studied. Fi'om the intermediate conditions found, it
seems probable that tliese forms of polydactylism are further modifica-
tions of those instances which have but a single extra digit. Figure 18
(Plate 10) shows the skeletal structure of one of the simplest of these
conditions. Tlie anatomy of the manus resembles in general that seen in
Figure 17 (Plate 9). Metacarpal i is large and articulates with the tra-
pezium, but instead of a single set of phalanges two series of bones are
present. One of these series (Plate 11, Fig. 19, i*) may be small, pollex-
like and composed of two phalanges, or both sets may be of nearly equal
size and each consist of three elements (Plate 10, Fig. 18, i", i''). Of
four cases examined, three showed the latter condition. The trapezium
and scaphoid are abnormally large in all cases. The musculature is like
that of the pentadactyle manus (Figs. 0, P), but the tendons which there
supply the single extra digit may here bifurcate, and be inserted into the
two digits. The nerve branch which supplies the first digit in Figure L
also divides (Fig. Q), so that in these cases there is undoubtedly a dupli-
cation of digit I. Eliminating this digit, the x'est of the manus, save for
the large size of the trapezium, would be entirely normal.

We now pass to a polydactyle condition in which digit i is completely
divided. The manus shown in Figure 20 (Plate 12) is interesting as
being a stage intermediate between the preceding cases and a complete
hexadactyle condition, and as additional evidence that the two exti-a
digits are produced by the duplication of digit i. For in this case, al-
though each is composed of a metacarpal and three phalanges, i" and
i** are alike in size and form ; still more noteworthy is the fact that the
two ungual phalanges are enveloped in a single hoof, and that the two
metacarpals articulate with the single trapezium. This carpal is large ;

278

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

the trapezoid, on the contrary, is small and laterally compressed, as is
also the proximal end of metacar^ml li.

The tendons of the muscles and the nerve of digit i bifurcate (Fig. Q, 1).

This intermediate stage leads up to conditions in which there are two
complete and entirely distinct digits. The duplication may extend even

n. m.

II.

i:.....i6.

Fig. Q. — Posterior view of left polydactj'le manus, showing innervation, i", i'',
supernumerary digits; 1, first branch of median nerve, which bifurcates twice, the branches
from the second bifurcation going to digits i" and i^ I \iatural size.

to the carpus, and the two digits thus formed may be nearly as large as
the functional digits (in and iv) of the manus. Six such cases were
examined. In the typical condition (Plate 13, Fig. 21) the supernu-
merary digits (i", 1*) are somewhat smaller than in and iv. Each

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 279

bears a large hoof, and the two hoofs are connected posteriorly by a cush-
ion of horny tissue, as are the functional digits. The trapezium, which
ai'ticulates with both extra digits, is very large, and shows evidence of
duplication ; the scaphoid also is abnormally large and broad. The

ext. mt'carp. mag.

ext. prp. i

•'ext. mfcarp. oh,
•' ext. com. dg. i.

Fig. R. — Anterior view of left polydact3'le manus of the pig, showing extensor
muscles, ext. com. dg. i., extensor communis digitorum internus; ext. mVcarp. mag.,
extensor mctacarpi magnus; ext. mV carp, ob., eyAensor metacarpi obliquus; ext. prp. i.,
extensor proprius internus; i", i*, supernumerarj' digits. 5 natural size.

trapezoid is narrow, being flattened by the large trapezium ; the proxi-
mal end of metacarpal ii also suffers in this respect.

When i" and i* are so large as to be functional, the muscles of the
manus show some important modifications. Extensor proprius internus
(Fig. B, ext. prp. i.) sends a tendon to i**; extensor metacarpi obliquus

280

BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

(ext. mVcarp. ob.) is large, and its tendon, instead of being inserted as
normally into the proximal end of metacarpal ii, continues down to the
distal phalanges of the supernumerary digits, into which it is inserted by
three slips. In two cases this muscle was strengthened by a strong slip
from the great extensor of the metacarpus. This is an interesting case

fix. per//

^flx.perf.

II.

- I"

Fro. S. — Posterior view of left polydactylc mantis of tlie pig, showing flexor
muscles. Jix. per/'., flexor perforatus teudons; Jlx. ptrf ., flexor perforans; i", i'', super-
numerary digits. \ natural size.

of adaptation, and shows what a strong influence the functional capacity
of the digits has on the development and structure of tiieir muscles.

Of the flexor muscles, the perforans (Fig. <S', fix. perf!) gives off a
large tendon to the extra digits : this divides, and a branch is inserted
into each ungual phalanx. The flexor perforatus {fix. per/.') also sends
a large tendon to the extra digits, which bifurcates in the region of the

PRENTISS: POLYDACTYLISxM IN MAN AND DOMESTIC ANLMALS. 281

second phalanges and forms a sheath for each division of the perforans
tendon. The innervation is shown in Figure Q.

With the increase in size of the extra digits of the polydactyle series,
goes a corresponding decrease in the size of digit ii. It is apparently
rednced, and partially, sometimes completely, atrophied on account of
the abnormal development of the supernumerary parts. In a case fig-
ured by Bateson ('94) tiie middle portion of metacarpal ii is gone. In
two front feet, from a single animal, I found that the left manus was
like that shown in Figure 20, the trapezoid and proximal end of meta-
carpal II being reduced ; in the right manus, however, metacarpal ii
was completely atrophied, but the three plialanges persisted and were
of nearly normal size. The trapezoid remained as a small flattened bone,
articulating chiefly with metacarpal in. The reduction is carried a
step further in another case, in which the three phalanges of digit ii
are present, but exceedingly small, and the hoof reduced to a claw-like
vestige (Plate 14, Fig. 22, ii).

The nerve branch which normally supplies the second digit innervates
this vestige (Fig. T, 2), making it reasonably certain that we have to do
with the rudiment of digit ii.

Figure 23 (Plate 15) represents the skeletal parts of a manus in
which the second digit has apparently atrophied completely. Three
specimens were examined which exhibited this condition. Such cases
have been described as dujjlications of digit ii, but a cai'eful study of
the manus shows that this is not the case. If we compare Figure 23
with Figure 22, the resemblance between the skeletal parts of the extra
digits is striking. In each case they both articulate with the trapezium,
and digit i* has taken nearly complete possession of the distal facet of
the trapezoid, which is normally occupied by digit ii. The trapezoid
itself is narrow and smaller than the'ti-apezium ; the scaphoid in Figure
23 is divided into two elements, a condition which is found 07ili/ when
two large functional digits are added to the normal number. Other im-
portant facts are that digits i" and i*" are of nearly equal size, symmet-
trical with reference to each other, and bear hoofs which are connected
posteriorly by a pad of horn.

The musculature and nerves also afford good evidence in favor of this
interpretation. The tendons which are normally inserted into the sec-
ond di(j;it ai'e wanting here. The second branch of the median nerve
(Fig. U^ 2), which normally supplies digit ii, still sends a large branch
to the radial side of digit in and may thus be identified. But dissec-
tions failed to disclose the small nerve which usually supplies the second

282

bulletin: museum of comparative zoology.

digit. We can only conclude, then, that digit ii, together with its ac-
cessories, has atrophied. This nianus is therefore only pseudo-penta-
dactyloiis, and belongs in reality to the hexadactyle abnormalities. This
conclusion is made possible only through the completeness of the poly-
dactyle series which I have studied, and emphasizes the futility of at-

n. m.

IV.

Fig. T. — Posterior view of left polydactyle manus of the pig, showing innervation.
I", I*, supernumerary digits; 1, first brancli of the median nerve, which bifurcates to the
extra digits; 2, second branch, a division of which innervates the rudimentary digit ii.
i natural size.

tempting to obtain general results from single cases of polydactylism.
Except for the intermediate stages at my disposal, the true significance
of the structural conditions shown in Figure 23 could only have been
guessed at.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 283

Conditions are rare where more than two supernumerary digits occur
in the polydactyle manus. Such a condition, however, is shown in
Figure 2-i (Plate 16). Digits i" and i'' are well formed and each con-
sists of three phalanges, but between them, and articulating with the

n. m.

Fig. U. — Posterior view of left polydactyle maniis of the pig. n. m., median nerve;
1, first branch of median nerve, supplying digits I", and i''; 2, second branch, innervating
digit III; its small radial division is wanting. § natural size.

proximal end of the second phalanx of i*, is an elongated bone, which,
from its position and form, may represent a first phalanx fused to a
portion of a metacarpal. In the carpus we find the trapezium repre-
sented by two elements (t7-z., tr-.'), and the scaphoid is also duplicated.

284 bulletin: museum of comparative zoology.

The other skeletal elements of this manus are normal. The muscula-
ture and innervation are identical with the conditions shown in Figures
Q, R, and S.

2. Manus in which the Supernumerary Parts may be more or less closely

connected with Metacarpal II.

^ a. One Supernumerary Digit.

This condition was observed in five cases. From a typical example
(Plate 17, Fig. 25) it might be inferred that all these cases were to be
interpreted as mere duplications of digit ir. The extra digit (i) pos-
sesses three phalanges and is of the same size as ii. Both are borne
on the same metacarpal, which is large and has two articular condyles
at its distal end. The digits, however, are nut symmetrical with each
other, as we should expect if they had resulted from duplication of
digit II ; in both, the hoofs and ungual phalanges are concave on the
ulnar, convex on the radial side. lu the carpus the trapezium is
larger than normal, and articulates above with the scaphoid, and below
with the radial portion of the proximal facet of metacarpal ir. This
condition is represented by only a single case. In four other specimens
the skeletal parts exhibited very interesting conditions which serve to
connect this class of abnormalities with the first part of the series we
are describing. In Figure 26 (Plate 18) it is seen that the extra
digit (i) is much larger than the second (ii), but, as in the preceding
case, both are borne on a single large metacarpal. They are not sym-
metrical with each other, and on examining carefully the metacarpal,
a dark irregular line will be seen, running nearly tlie whole length of
the bone and dividing it into two unequal portions. This line of separa-
tion, so clearly brought out in the skiagraph, is not, of course, a surface
marking but represents a complete bony septum. The two components
into which the metacarpal is thus divided, correspond in size with the
digits which they respectively bear.

The structure of the carpals furnishes important evidence as to
whether the extra digit is formed by the splitting of ii. If this were
the case, the trapezoid should show signs of duplication, while the tra-
pezium should remain normal. On the contrary the trapezium is large
and fused to the trapezoid. Comparing Figure 20 with Figure 17
(Plate 9), the similarity of the skeletal structures is striking, and we
can but conclude that the manus shown in Figure 26 differs from that
shown in Figure 17 only iu the fusion of its trapezium and traj^czoid,

PRENTISS : POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 285

and of its first and second metacarpals. This view is borne out by an-
other manus, in which the trapezium is fused to the proximal end of the
compound metacarpal, and also by a case figured by Ei-colani ('81,
Tav. I, Fig. 2). In this instance digit ii is of normal size, and its
metacarpal is fused with metacarpal i along its proximal half only.
This element (i) is large and bears three large phalanges. The com-
pound bone formed by the fusion of metacarpals i and ii articulates
above with the trapezoid, which is normal, and also with the trapezium,
which is abnormally large. If metacarpals i and ii of the manus shown
in Figure 17 were fused at their proximal ends, we should have a con-
dition identical with that figured by Ercolani.

The evidence of the skeletal parts is in the main confirmed by the
arrangement of the muscles and nerves. The condition of the muscles
is similar to that of cases where the extra digit is distinct (Figs. 0, P,
pp. 275, 276). In the five cases dissected, digit ii retained its own pe-
culiar muscles. In one case all the muscles were normal ; and in one in-
stance the most radial tendon of the flexor perforans (Fig. P, fix. per/.'),
which is normally inserted into digit ii, bifurcates and is attached to digit
I as well. In all cases the supernumerary digit was innervated by a special
branch given off independently from the radial side of the trunk of the
median nerve, as in pentadactyle animals (Fig. L, 1, p. 272). There is
little ground, therefore, for regarding these cases of polydactylism as due
to duplication of digit ii ; on the contrary, there is direct evidence against
this view. (1) Digit i varies in size, while digit ii always remains
normal; (2) they are not symmetrical with each other; (3) the
divisions of the metacarpal bone are unequal ; (4) the trapezoid is not
duplicated nor increased in size ; (5) there is no general duplication of
muscle tendons ; (6) the extra digit is innervated by an independent
branch of the median nerve.

In favor of the assumption that the extra digit represents the pollex
independently developed and later fused to metacarpal ir, is the fact
that the trapezium is of abnormal size, and always articulates with the
radial portion of the proximal facet of the compound metacarpal ; also
the striking resemblance of the skeletal, muscular, and nervous structures
to those of the cases in which the extra digit does arise independently.

h. Two SUPERNUMERART DiGITS.

Three cases were observed representing two types. Of the simplest
condition there was but one case. In this manus digit i» (Plate 19,
Fig. 27) consists of two small phalanges and the distal end of a meta-

286

bulletin: museum of comparative zoology.

carpal bone ; digits i^ and ii are of nearly equal size, each composed of
three phalanges and borne on a single large metacarpal, i" and i* are
enclosed in the same hoof, which shows evidence of duplication.

The phalanges of digit ii are of normal size and form ; the carpals
are practically normal, but the trapezium articulates with the proximal

n. m.

Fig. V. — Posterior view of left polydactyle inaiius, showinp; innervation, i", i*",
supernumerary dij^its; n. m., median nerve; 1, first branch of median nerve supplj'ing
digits I" and i". i natural size.

end of the compound metacarpal, and ends in a free distal process.
The musculature of digit ii is normal. The extensor proprius pollicis
et indicis divides and is inserted into the distal phalanges of both i" and
1*. The flexor perforans gives off an independent tendon to digit i^.
The innervation of the raanus (Fig. V) is identical with that of cases
in which the two extra digits are entirely distinct from ii (Fig. Q).

PRENTISS : POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 287

This abnormality may be accounted for in two ways : Either (1) digit
I" represents the developed pollex, and i* is formed by the duplication
of dio-it II, or (2) digits i" and i* are duplications of the pollex, and the
metacarpal of i'' is secondarily fused to that of digit ii. The first
hypothesis is supported by the similarity in structure of digits i*
and II, their symmetry with reference to each other, and the differences
existing between i" and i^ The second view, however, is supported
(1) by the fact that the extra digits are enclosed in the same hoof, and
therefore probably developed together, (2) by the fact that the trape-
zium articulates with the compound metacarpal, and (3) by the structure
of the muscles and nerves.

To another type belong two cases in which digit i" is completely
developed and articulates with the carpus (Plate 20, Fig. 28). Digits
I*" and II are borne on a single large metacarpal, but i** is much the
larger. The phalanges of ii are of normal size and unsymmetrical
with those of i^. The ungual phalanges of both i" and i* are enclosed
in separate hoofs, and are symmetrical with each other, although differ-
ing somewhat in size. The trapezium is large, and articulates with
metacarpal i" and with a portion of the compound metacarpal. The
musculature and innervation of this manus are similar to those of the
foregoing case.

Our view that these abnormalities are due to duplication of the
pollex and the subsequent fusion of the metacarpal of i^ to that of ii,
is favored by the structure of a manus figured by Otto ('41, Tab.
26, Fig. 12). In this case there are two extra digits of three pha-
langes ; i" is borne on a distinct metacarpal, which articulates with
the trapezium, and i* on a metacarpal which is almost completely fused
to metacarpal ii. Digit ii is of normal size. The phalanges of i" and
i^ form a single series of three bones, each of which is incompletely
divided into two ; the ungual phalanx evidently bore a single hoof.
The trapezium articulates with metacarpal i" and with two-thirds of
the pi'oximal surface of the compound metacarpal. The trapezoid is
smaller and articulates with the remaining third of the proximal facet
of the large metacarpal bone. In this manus, therefore, the digits
i" and i^ evidently developed together, and the fusion of metacarpal
i^ to that of II was of subsequent occurrence. This being the fact, it
is very probable that the foregoing cases which we have examined were
produced in a similar manner.

Having now briefly described the types of digital variation in the manus
of the pig, we shall next attempt to determine their significance.

288 BULLETIN : MUSEUxM OF COMrAliATlVE ZOOLOGY.

C. Significance of the Variations Observed.

The objections to explaining polydactylism in the pig by the theory
of reversion are based on anatomical, embryological, and palaeontological
evidence. They have been well summed up by Gegenbaur ('80):
(1) tlie accessory pollex is composed of three plialanges, whereas, if
due to reversion, it should consist of only two ; (2) the other parts of
the manus show no modifications toward ancestral conditions; (3) no
fundament of the pollex is present at any stage in the embryo pig, nor
is it present as a rudiment in any artiodactyle, living or extinct.
Gegenbaur, accordingly, concludes that the extra digit is not produced
by the development of a vestige, but can be formed only from the
duplication of one of the normal digits. Are these objections and
Gegenbaur's theory supported by the cases which we have examined ?

Pirst, as to the number of phalanges in digit i : in five of our cases
there was present a pollex of two phalanges. In the remaining twenty-
nine cases, however, there were three elements in each of the extra digits.
Gegenbaur is thus right in the main, but there are a few instances which
contradict his sweeping statement.

As regards the modification of the other parts of the polydactyle
manus, Gegenbaur is again correct in his general statement. But we
have seen that in a limited number of cases tliere are found the identical
conditions which he maintains never exist. The trapezium, trapezoid,
and third metacarpal of the polydactyle manus resemble in structure
the same elements in the manus of certain fossil swine (Ancodus, Palaeo-
choerus). But the troclilear ridge is found at the distal articular face
of the metacarpals in all polydactyle conditions, although it is partly
or completely wanting in fossil forms. Other peculiarities of the
phalanges of fossil forms are not reverted to.

The musculature also shows some interesting changes. Extensor
metacarpi obliquus is in many cases inserted into the metacarpal of the
extra digit (i) rather than into metacarpal ii. But wc know that in the
polydactyle manus of man tendons may shift from normal to abnormal
digits, although reversion plays no part in producing these abnormalities.
The development (1) of the extensor proprius ,pollicis et indicis (which
is rudimentary in the normal manus) and (2) of an independent tendon
from the radial side of the flexor perforans are the best evidences pre-
sented by the musculature that the extra digit is produced from a
vestige. But no great weight can be placed on the structure of the
muscles, as their modifications appear to be chiefly adaptive. They are

PKENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 289

most highly developed when the extra digits are functional, and often
to an abnormal degree.

Much greater stress can be laid on the innervation of the polydactyle
manus, for the structural conditions are singularly uniform throughout
this polydactyle series. In all cases the supernumerary parts are inner-
vated by an independent nerve arising from the radial side of the median
trunk, and at about the position where the nerve of the pollex is nor-
mally given off in pentadactyle animals. When two extra digits are
present in the manus, this branch bifurcates and supplies both. Thus
modifications exist in the skeletal, muscular, and nervous organs of the
polydactyle manus; they point towards the vestigial origin of the extra
digits, but there is little evidence of reversion in other parts of the manus.

Gegenbaur's third objection, that the pollex is absent in the embryo
and in all adult Artiodactyla, is well taken. For if these are facts, rever-
sion would have to produce a digit of which there is no fundament in the
embryo, and reproduce an organ characteristic of only extremely remote
ancestors. But Scott ('95) has shown in his work on the American
Anthracotheridae, that Ancodus brachyrhynchus has the pollex Avell de-
veloped. We do not, therefore, have to go back further than the Suinae
to find a pentadactyle form. As to the absence of the fundament of the
pollex in the pig embryo, I have confirmed Rosenberg's ('73) results by
examining the carpus of a large number of embryos in various stages of
development. For this material I am indebted to Prof. E. L. Mark.
There was absolutely no evidence of a pollex-fundament other than the
trapezium. This element is generally regarded as being simply the
carpal element of digit i, for it develops as a single cartilage. We
know, however, that the scaphoid and unciform bones develop in the
same way, yet that each represents two carpals fused. A careful study
of the trapezium in the embryo, in -the normal adult and in the poly-
dactyle pig, furnishes some evidence in support of the view that the so-
called trapezium represents a rudiment of the pollex as well as a carpal
element. (1) In the earliest stages of its development, the cartilage
which is to form the trapezium has the pointed distal end characteristic
of its adult condition, and jyrojects distad to the proximal limit of the
metacarims. (2) In the normal adult carpus the trapezium has always
the form of an elongated cone. Its distal end is free, and pointed,
instead of truncated, as we should expect if we had to do with only a
carpal element. Furthermore, its free end projects farther distad than
the other carpal bones and into the region of the nietacarpus. (3) In the
polydactyle manus one case was described in which only the distal

290 BULLETIN : MUSEUM OF COMPAKATIVE ZOOLOGY.

end of metacarpal i was developed ; yet tho so-called trapezium is ab-
normally long and projects well down by the side of metacarpal ii. In
three cases where the pollex is developed in a rudimentary condition tho
trapezium is fused to metacarpal I.

In other animals, such as the horse and ox, where there are well-
authenticated cases of vestigial polydactylism, the extra digits usually
represent the development of rudiments normally present in the embryo.
in the case of polydactyle swine, where the extra digits constantly make
their appearance in the region of digital reduction, it is but natural to
conclude that a rudiment of this digit, even though extremely vestigial,
is present in the embryonic manus.

In cases where two (rarely three) extra digits are found in the poly-
dactyle manus, there are no modifications in the other parts. Moreover,
it is out of the question to consider digit i" as representing a prae- pollex
and 1* a pollex. Granting that the prae-poUex existed, there arc still
hisurmountable difficulties in the way of this interpretation. Both extra
digits develop on a single carpal element, the trapezium. They are sup-
plied by bifurcations of the same muscle tendon, innervated by the
divisions of the same nerve-branch, and may even be enclosed distally in
the same hoof. In addition, they are usually of the same size and sym-
metrical with each other. Thus their structure, and the fact that con-
ditions exist intermediate between a single undivided digit and two
completely separate ones, make it almost certain that the two extra
digits arise from the duplication of the pollex.

Having found good evidence in favor of the vestigial origin of tho
extra digits, and that Gegenbaur's objections do not hold for all cases,
let us examine the evidence in favor of his theory that all cases of
polydactylism in the pig are due to duplication of the second digit.

On examining the structure of two digits which are known to be dupli-
cations of a single one, we find that they are of nearly the same size,
symmetrical with each other, often enclosed in the same hoof, and borne
always on a single duplicated carpal element. They are supplied also
by duplications of the same muscle tendons, and innervated by the
bifurcations of the same nerve-branch.

In the polydactyle cases which we have examined these are not tho
characteristic conditions. As we have seen, digits i and ii always differ
greatly in size, often in number of phalanges, and are not bilaterally
symmetrical. Digit i is never borne on the trapezoid, but on its own
proper carpal, the trapezium ; when the trapezium is apparently ab-
sent, it is really fused to metacarpal i, or to the trapezoid. The mus-

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 291

cular attachments and the innervation of the extra digit are entirely-
distinct from those supplying and innervating digit m. We can only
conclude, therefore, that in these cases the supernumerary digit is not a
duplication of digit ii. If it were such a duplication, why should not
the fifth digit be affected as often as the second 1 On the contrary, in
every polydactyle manus so far observed the siipernumei'ary digit is found
on the radial side of digit ii.

There is no doubt that abnormalities due to the duplication of a func-
tional digit may occur in the manus of the pig as in other mammals;
but in the majority of cases the origin of the extra digit must be vestigial.
By variation and duplication of this vestige in its development, two or
more supernumerary digits may be formed. Whether or not the develop-
ment of this digital vestige is due to reversion, we will discuss in the
theoretical portion of this paper.

Summing up the facts obtained as to polydactylism in the pig, it is
found that —

1. Polydactylism is confined almost entirely to the manus. (This fact
is interesting, as the condition restores that found in fossil swine. In the
pes of Ancodus the hallux is entirely gone, although in the manus the •
pollex is well developed. If we regard the extra digit as due to duplica-
tion of digit II we should expect this duplication to occur as often in
the pes as in the manus; but if the extra digit is vestigial in its origin,
the early and complete reduction of the hallux in fossil swine is good
reason for its never being developed in the pig of the present day.)

2. The supernumerary digits in every case occur on the radial side
of the second normal digit.

3. In nineteen of the thirty-six cases examined, a single super-
numeraiy digit is present ; in five instances this digit is composed of two
phalanges ; in nine cases, of three ; and in five instances its metacarpal
is fused to that of digit ii.

4. In the remaining seventeen specimens thirteen are hexadactyle,
although in three cases the metacarpal of one supernumerary digit (i^)
is fused to that of digit ii ; in three instances two supernumerary digits
are present, but digit ii is entirely wanting ; and in one specimen there
are evidences of tliree extra digits.

5. In more than a third of the cases examined, the skeletal, muscular,
and nervous organs of the manus give some evidence that the extra
digit is vestigial.

6. The trapezium (so-called) may represent this carpal element plus
the rudiment of digit I.

VOL. XL. NO. 6 4

292 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

7. Tlie extra digits articulate witli the trapezium in nearly every case ;
they therefore represent the development of a vestigial pollex, hut may
vary extremely from the normal pollex structure.

8. There may be cases where the extra digit is formed by the dupli-
cation of digit II, but there is strong evidence against this being the
general rule.

9. Two supernumerary digits may be formed by the duplication of
the vestigial pollex ; there are no grounds for considering one of them a
*'prae-pollex."

VI. Polydactylism in Ruminants.
A. Literature.

Observations have been made on polydactylism in ruminants and
descriptions given by Geoffrey St. Ililaire ('32-37), Goodman ('68),
Cliauveau et Arloing ('79), Boas ('90), Baumuller ('92), Blanc ('93),
and Bateson ('94). In the normal manus of ruminants, in and iv are
the functional digits, and in all forms save the water chevrotain their
metacarpals are fused to form a single " cannon bone." The pollex is
always wanting; digits il and v are reduced in varying degrees in the
different groups of ruminants. In the camel they are wanting ; in the
ox metacarpal v remains as a proximal rudiment ; the phalanges are
completely gone, but a " dew-claw " repi'esents each hoof. The sheep
has the two distal phalanges and hoofs of ii and v persistent, while in
the Cervidae these digits are represented by three well-developed pha-
langes and the distal ends of the metacarpal bones; the hoofs of digits
II and V are functional when the deer is running or travelling over soft
ground. In the water chevrotain there are four complete digits, each
formed of a distinct metacarpal and three phalanges.

I know of no instance of polydactylism in the camel, and there are
few descriptions of such abnormalities in sheep. Geoffroy St. Ililairo
('32-37) describes the manus of a lamb in which digits i, ii, and v were
developed ; digits i and ii were borne on the same metacarpal and
probably represent a duplicated condition of digit ii. The best de-
scription of polydactylism in the sheep is that of Chauveau et Arloing
('79). The manus of a lamb is figured, in which both the second and
fifth digits are developed, each being composed of a distinct metacarpal
element and three phalanges nearly as large as those of the functional
digits. This condition is certainly due to the development of vestiges,
and has been attributed to reversion.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 293

Bauniuller ('92) figures the manus of a roebuck (Cervus caprea)
which was composed of five digits. The abnormality was found on both
fore feet. Baumiiller regards the extra digit as a pollex, and attributes
its presence to reversion.

Bateson ('94) remarks with reference to polydactylism in the sheep
and ox, that the extra digits are in all cases formed by duplication or
variation. As to the development of digits ir and v he asserts that
" there is no such case."

In the ox, a number of cases of polydactylism have been observed
and described. They may be divided into two groups : (1) manus or
pes of three digits, all of nearly equal size, and borne on a single meta-
carpal bone (Bateson, '94, Figs. 114, 115, p. 375). In these cases the
presence of both accessory hoofs (rudiments of ii and v) in their
normal positions makes it certain that the vestiges of digits ii or v have
not developed, but that either in or iv has become duplicated. Four
cases are described by Bateson, and it is stated by Goodman ('68) that
the abnormality was common and frequently inherited in a herd of Eng-
lish cattle. (2) Manus of four digits, ii and v both being developed ;
the accessory hoofs are located at the distal extremities of the extra
digits ; each supernumerary digit is composed of a distinct metacarpal
element, and digit ii has in addition two small phalanges. Boas ('90)
describes two cases, and considers them good instances of reversionary
polydactylism.

B. Observations.

Two cases of polydactylism in the manus of the ox have come under
ray observation. Both specimens had been disarticulated at the
carpo-metacarpal joint, and the carpal bones were thus unfortunately
lost ; they were right and left fore feet and probably belonged to one
animal. Both are abnormally wide at the distal end of the cannon
bone ; in each the hoof of the radial side is very broad and incompletely
divided into two parts (Fig. W, p. 294, and Plate 21, Fig. 29). The
accessory hoof of the ulnar side of the manus is normal in position, but
that of the radial side is absent in both cases.

In the left manus (Fig. 29) the skeletal parts are well formed. The
metacarpus is of normal length, and is distinctly divided into three
elements, each of which bears an articular head for a corresponding digit.
These three elements represent three metacarpal bones, and we may
designate them as ii, in, and iv. in is larger than either of the
others; its distal articular surface is unsymmetrical, as the trochlear ridge

294

BULLETIN : MUSEUM OF COMPAKATIVE ZOOLOGY.

com, dg.

ext. prp. ex.
ext. prp. i.

has shifted towanl the iihiar side. The supernumerary metacarpal (ii)
is the smallest of the three ; it is fused to iii througliout its whole
length, and can be traced to the proximal extremity of the metacarpus,
where it takes part in forming the articular facet for the carpals. The

distal epiphysis extends beyond those of
the normal metacarpals, has a flattened
instead of a convex articular surface, and
no trochlear ridge. The fifth metacarpal
(Fig. IF, v) is, as normally, a rudimentary
stylet articulating at the ulnar side of the
proximal extremity of iv.

All three digits are composed of three
phalanges. Digit iv is apparently normal ;
digit III is more massive, and the sym-
metry of its phalanges and hoof is affected
by the presence of the abnormal digit.
Instead of being optical images of those
of digit IV, these bones are indifferent in
their conformation, curving neither to the
right nor to the left. The hoof in Avhich
the ungual phalanx is enclosed is coniraou
also to digit ii. The extra digit (ii) is
shorter and not so massive as the normal
ones ; its ungual phalanx is flattened lat-
erally, and more pointed than the normal
phalanges ; the sesamoitls are absent.

Dissection of the musculature of this
manus shows that the flexors are entirely
normal ; the extensors, however, exhibit an
important modification. The tendon of the
extensor proprius internus (Fig. W, ext.
prp.i.) divides, and the more radial of the
two slips thus formed is inserted into the
second and-ungual phalanges of the super-
numerary digit. Before its insertion this
tendon is joined by a division of the suspensory ligament. The anatom-
ical relations of this tendon thus resemble the normal condition in
four-toed animals. If the supernumerary digit is a duplication of digit
III, we sliould expect to find the extensor communis digitorum (ext.
coin, dg.) and the flexor tendons bifurcated ; but they are unmodified.

II. III. IV.

Fig. W. — Anterior view of the
left polydactylc niamis of a calf,
sliowing the extensor muscles, ii,
supernumerary digit ; v, nietacar-
pai of digit five; ext. com. dg.,
extensor communis digitorum ; ext.
p7-p. ex., extensor proprius exter-
nus; ext. prp. i., extensor pro-
prius internus. i natural size.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 295

71. m. —

The nerves of this manus also show important modifications. The
normal manus, like that of swine, is innervated by four branches of the
median nerve ; the most radial and most ulnar branches (compare Fig.
X, 2, 5) give off small twigs to the rudiments of digits ii and v.
Brancli 5 is joined by the ulnar nerve im-
mediately before it divides to form 5 a and
5 b. In the polydactyle manus (Fig. X, 2,
5) the modification is in connection with
the small fasciculus (2°), which normally
innervates the radial accessory hoof (rudi-
ment of digit ii). This is no longer a
mere filament ending at the distal end of
the metacarpus, but a moderate-sized
branch, which continues to the hoof and
ungual phalanx of the supernumerary
digit. The condition of this nerve branch,
together with the fact that the accessory
hoof of this side is absent, affords most con-
vincing proof that this abnormality is not
a monstrosity, or a duplication of digit in,
but is due to the development of digit ii.

The second case, a right manus, con-
firms by its structure the conclusion which
we have drawn from the first.

The line of demarkation between the
second and third metacarpals is even more
distinct (Plate 22, Fig. 30) ; the first
and second phalanges of digit ii are fused
together and. are abnormally short.

Rosenberg ('73) states that metacarpals
II and V are present in tlie embryo of
the sheep and ox, but later partially de-
generate and fuse to the cannon bone, a
small portion of v remaining distinct in
the ox. In the Cervidae the distal ends
of the metacarpals persist in the adult.
It is not surprising therefore that we find
these digital rudiments occasionally developed in the adult ruminant.

Polydactylism in ruminants is thus of two types: (1) vestigial, due
to the development of either digit ii or v (or both) ; (2) teratological,
produced by the duplication of one of the functional digits (iii or iv).

IV. III. II.

Fig. X. — Posterior view of
left polydactyle manus of the calf,
showing innervation. ii, extra
digit; V, metacarpal of fifth digit;
n. m., median nerve; n. u., ulnar
nerve; 2-5, four branches of me-
dian nerve; 2«, division of second
branch which supplies the extra
digit (ii); 5", division of fifth
brancli which innervates the ac-
cessory hoof (digit v). i natural
size.

296 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

VII. Polydactylism in the Equidae.

A. Literature.

The anatomy and diseases of tlie liorso have been studied almost as
thoroughly as those of man, and consequently we find that polydactylism
in the Equidae has received considerable attention. Aside from the classi-
cal allusion of Suetonius ('86) to the horse of Julius Caesar "which had
feet that were almost human, the hoofs being cleft like toes," the first
account of polydactylism is that of Winter (1703), who describes two cases.
Geoffroy St. Hilaire ('32-37) records a foetus whicli Avas polydactyle
in the fore feet, the left foot bearing three nearly equal digits, and the
right two. Numerous instances have since been noted, tlie more im-
portant descriptions being tlioso of Arloing ('67), Wood-j\Iason ('71),
Marsh ('79, '92), Ercolani ('81), Boas ('85), Putz ('89), and Ewart ('94).
Blanc ('93), and Bateson ('94) review the general subject.

The normal functional digit of the Equidae is iii of the typical
mammalian mauus ; it consists of a long metacarpal bone and three
phalanges. The ungual phalanx is completely enclosed in a massive
hoof. Two splints, representing the metacarpals of digits ii and iv,
articulate at each side of the cannon bone posteriorly and with the
carpus. The trapezium is a small pea-shaped rudiment lying posterior
to the trapezoid and often wanting. The os magnum is very large, and
■with it, chiefly, the cannon bone articulates.

The polydactyle cases cited by various investigators fall into two
groups, the first of which may be subdivided into three:

(1) Supernumerary digits representing the development of digital

vestiges.

a. Three metacarpals, the extra digits being home on ii and iv. The
condition of an extra digit borne on metacarpal ii may occur on all four
feet (Marsh, '92) or be limited to the raanus (Arloing, '67). The extra
digits are always smaller than in and do not function in locomotion ;
this condition is of quite frequent occurrence. A single case is cited by
Wood-Mason ('71), in which an extra digit of three phalanges occurs on
metacarpal iv ; the radial splint bone (ii) was also somewhat better
developed than in a normal nianus. Cases of three digits (both n
and IV being developed) are cited by Geoffroy St. Hilaire ('32-37) and
Marsh ('92), but no good anatomical descriptions are given.

PKENTISS: rOLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 297

trzA

h. Four metacarpals ; digit i is represented by a splint radial to
digit 11, wliicli is fully developed and composed of three phalanges
(Fig. Y).

In these cases there are four large bones present
in the distal row of carpals. Digit ii is large, and
its metacarpal is fused throughout most of its length
to that of digit iii. Four cases are cited by Marsh
('92), and one is carefully described by Bateson ('94).

A different interpretation from that here as-
sumed may be brought forward in explanation of
these cases. The digit designated as ii in Figure
Y may be regarded as a duplication of digit iii, and
the so-called trapezoid of the carpus may represent
a duplication of the os magnum. Then the bone
designated as trapezium must be the true trapezoid,
and its splint bone the second, not the first, meta-
carpal. Only by a careful examination of the
skeletal, muscular, and nervous structures can we
determine which interpretation is correct ; whether
digit II is of vestigial origin, or due to a duplication
of digit III. The fact that in phylogeny the pollex
disappeared long before the iifth digit is a strong
argument against the former interpretation. For
by that interpretation we should here have the pol-
lex reappearing, and the second digit almost as large
as the third, while the fourth digit is unmodified
and the fifth is entirely absent.

c. Five metacarpals ; one supernumerary digit,
home on metacarpal ii. One case i^ described by
Piitz ('89) in which the trapezoid bears digit ii ;
this consists of a well-developed metacarpal bone and three phalanges.
Radial to this is a large trapezium, articulating with the scaphoid and
trapezoid and bearing a splint six cm. long ; metacarpal iv is normal,
and on its ulnar side is another metacarpal element supposed to rep-
resent digit V. The supernumerary elements in this case can only be
exi)lained as of vestigial origin.

(2) Two digits borne on metacarpal iii.

These are clear cases of dui)lication, and have been described in the
manus only. The doubling may extend to the metacarpal bone, but is

Fig. F. — Anterior
view of left polydac-
t^'le manus of horse.
I, metacarpal of first
supernumerary digit
(pollex) ; II, second
supernumerary digit;

III, functional digit;

IV, metacarpal (splint)
of fourth digit; r,
radius; tvz., trape-
zium; ««., unciform.
(After Marsh.)

298

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

usually limited to the phalanges. Such conditions have beeu described
by Struthers ('63), Arloing ('67), and Boas ('85).

— pJvx. 1.

B. Observations.

Through the kindness of Dr. Frothingham, of the Harvard Veterinary
School, an abnormal manus of a polydactyle colt came under my observa-
tion. The specimen came from Texas. Externally the hoof was almost

completely divided into two : each
mt' carp. \ •' , • , ,

portion was several inches long, and

curved away from the other. On ex-
amining the skeletal parts (Fig. Z),
they were found to be normal down to
tlie distal end of the first plialanx,
which was bifurcated and bore two
articular surfaces. Each of these car-
ried two phalanges, which resembled
tlie median and ungual phalanges
of the artiodactyle digit. The two
series were mirrored images of each
other ; each os pedis was slightly con-
cave on the surface facing the median
plane of the digit, and convex on the

■phx. 2^ opposite side, so that the two fitted
together would give a phalanx of
nearly normal form. A navicular of
about half the length of the normal

'p\x. 3* l)one articulated with the posterior

face of eacli os pedis, thus resembling

the condition of ruminants.
Fio. Z. — Anterior view of left polv- _,, . . i i i i • i i

dactyle manus of the horse, showing dupli- This specimen had been dried be-

cation of digit in. mVcarp., distal end of fore it waS examined, and the inner-
third metacarpal bone; pAx. 1, first phalanx . 1 , .4- -u^ „4.„ i:„.i i,,,f r>x-
of third digit; phx. 2'S 2^ duplications of nation COuld not be studied, but ex-
second phalanx; phx. 3", 3^, duplications aminatiou of the chief luuscle tendoiis
of ungual phalanx, i natural size. showed that the extensor pedis and

flexor perforans were duplicated at their distal ends. This case is there-
fore simply an example of duplication of digit in.

It has long been known that the " splint bones " of the equine manus
represent rudimentary metacarpals, but until recently the presence cf
phalangeal vestiges in the manus of the embryo has been denied.

phx. 2° — •

phx. 3"—"

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 299

Rosenberg ('73) searched for sucli vestiges, but without success.
Ewart ('94), in tracing out the skeletal development of the limbs of the
horse, found cartiliigimnis nodules articulating iu an imperfect manner
with the distal epiphyses of metacarpals 11 and iv. The vestige
attached to digit 11 was the larger, and in some instances showed evidence
of division into two or three parts, which Ewart takes to be the funda-
ments of as many phalanges.

Tins is an interesting and important discovery, since, if digit 11 is better
developed than iv in the normal embryo, we have a good explanation
for the fact that in polydactyle horses it is the second digit which is
of most frequent occurrence. Dissection of the manus of a foetus 35 cm.
long enabled me to confirm Ewart's work. There is thus conclusive
evidence that in the horse extra digits are frequently of vestigial origin.
The digital abnormalities of the Equidae can therefore be divided into
two distinct classes :

(1) Vestigial cases, in which the extra digits are developed from
rudiments normally present in the manus of equine embryos and
extinct ancestors.

(2) Teratological cases, which are malformations usually due to the
partial or complete duplication of the functional digit (m).

VIII. Theories of Polydactylism.

The occurrence of polydactylism has been attributed to two proximate
causes : (1) External influences, (2) Internal influences.

1. External Influences.

The supporters of this theory (Ahlfeld, '85-86, and Zander, '9l) would
explain all cases of digital variation as due to the pressure of amniotic
threads in ntero. This view accounts satisfactorily for the variation in
degree of digital duplications, but utterly fails to explain their fixed
position with reference to certain digits, and cannot apply to the
development of digital vestiges. Pressure from an amniotic thread
would naturally affect any finger or toe, whereas we know that poly-
dactylism in mammals is practically limited to the first or fifth digit, is
often bilaterally symmetrical in its occurrence, and may a0"ect both
manus and pes in the same individual. The abnormalities are also
strongly inherited, and the amniotic theory, if correct, would necessitate
admitting the inheritance of acquired characters. Although the duplica-
tion of organs has been artificially produced by Dareste ('91) and others, it

300 BULLETIN : MUSEUM OF COMrARATIVK ZOOLOGY.

has yet to be proved that such modifications are inlierited. Certain cases
of digital duplication are undoubtedly caused by the pressure of amniotic
threads. Such abnormalities are true malformations, and usually alFect
a normal, unreduced digit. An assured case is that of a duplicated
thumb described by Ahlfeld, in which a fold of the amnion was found
at birth still adherent between the duplications of the poUex. It is
possible that certain cases where a single functional digit is duplicated
are produced in a similar manner. Such examples of polydactylism,
however, are the exceptions ratlier than the rule, for in both mammals
and birds we have seen that the typical, unmodified, functional digits
vary but rarely. Under this class might come the cases of partial or
complete duplication of digits ii-iv in birds and man; of digits ii-v in
carnivores ; of digits in and iv in artiodactyles, and of digit in in the
horse. Some cases of the duplication of digits i and v in man and of
digits II and v in swine may also be included in the above categor}^ ; but
it may be that all the symmetrically placed, hereditary digital abnormali-
ties are produced by some internal influence emanating from the germ
itself.

2. Internal Influences.

One of the most important facts brought out by the comparative stmly
of polydactylism is its limitation chiefly to the variation of digits which
normally are either modijied, rudimentary^ or vestigial. It is natural to
conclude that all such variations are due to one and the same cause.
But on comparing the diff"erent types we find that it is only in the liorse,
ruminants, swine, and the pes of carnivores that extra digits arise as
vestigial developments ; whereas, in man, the fowl, and the manus of
the cat they are formed as duplications of functional digits.

a. Reversio7i.

The theory of reversion, first proposed by Darwin to account for poly-
dactylism in man, has been supported, and extended to all mammahan
forms, by Bardeleben ('85), Albrecht ('86), Kollman ('88), Cowper ("89),
and Blanc ('93). Boas ('85, '90) limits reversionary polydactylism to
the horse and ox. Marsh ('92) asserts that the digital variations in the
Equidae can be accounted for in no other way. Gegenbaur ('80, '88),
while strongly opposed to the theory in general, admits that it may
bo applicable to polydactylism in the horse.

Reversion, as generally understood, is but heredity carried to an
extreme in point of time. It is the inheritance by an individual of

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANLMALS. 301

qualities peculiar to a distant ancestor, — qualities which were once
characteristic of the species, but have been lost in the evolution of
varieties. Consequently, the best-authenticated instances of reversion are
those in which individuals of a certain variety or breed return to the
characters of the original species. Well-known examples are the rever-
sion of domestic varieties to the character of the wild rock-pigeon ; the
recurrence of shoulder-stripes and a dun coloration in the horse and mule;
the appearance of longitudinal stripes on the backs of young domestic
swine when allowed to return to the feral state, — a coloration pecu-
liar to the sucklings of the wild ancestors of the hog, but normally want-
ing in the young of the domestic pig. In these cases, which we know
are reversionary, it may be observed (1) that the phenomenon is simply
the return of individuals of a variety to the original characteristics
of the species ; (2) that the variation in such reversions relates merely
to the degree of completeness with which the atavistic qualities are
transmitted ; monstrous conditions, or malformations, are never thus
produced.

In animals in which the typical number of functional digits is normally
reduced (pes of Carnivora, swine, ruminants, and Equidae), the super-
numerary digits in the majority of cases are developed independently of
the normal digits, but in connection with embryonic vestiges or rudi-
ments. Is not reversion, then, the factor which is operative here, caus-
ing the development of degenerate digits, and thus tending to restore
the original pentadactyle condition 1 The objection is raised, however,
that tliere is too great a disttince in point of time and relationship between
the polydactyle animal and the pentadactyle ancestor to which it is sup-
posed to revert. According to the old idea of heredity this might seem
true, but in the light of Mendel's law (recently fully confirmed) it is no
longer a serious objection. As pointed out by Bateson and Saunders (:02)
and Castle (:03), the important facts discovered by Mendel are that a
single parental character may be segregated in the germ-cells of the off-
spring, and that one of a pair of parental characters may regularly domi-
nate over the other ; further that each of the offspring, though exhibiting
the dominant character only, produces ripe germ-cells half of which bear
the dominant character of one parent, the other half, tlie recessive charac-
ter of the other parent. Thus, if the polydactylous Doi'king is crossed
with the normal Leghorn, nearly all of the hybrids will be polydac-
tylous — not quite all, however, for the extra toe in this case is not
complett'.ly dominant. But continued breeding shows that the sperm
and ova of the crossbreds will bear eitlier the dominant polydactylous

302 bulletin: museum of comparative zoology.

character, D, or tlie normal recessive character, i?, and that equal num-
bers of D's and i?'s will be produced. Offspring of the crossbreds will
therefore show these characters in the following ratios : — \ D -.2 DR : 1 R.
But the character D being dominant, not only the 1 Z)'« but the 2 DR'^
will be polydactylous and therefore oidy one-fourth of the chicks will
have normal toes. Bateson's experiments show that this is really the
case.

To us the significance of Mendel's law lies in the fact that a certain
character may be transmitted pure from generation to generation of
germ-cells in a latent condition ; that is, the character may not appear
in the structure of the animal, tiiough present in its germ-cells.

The occurrence in a latent condition of characters which when active
are dominant may thus explain the constant outcrojjping of these
characters, such, for example, as the continual a})pearance of "rogues,"
in apparently pure races of plants and in animals which have been
selectively bred for generations. The appearance of reversionary poly-
dactylism may be explained in this way.

Although we know that in the horse, ruminants, swine, and the pes
of carnivores the extra digits may be of vestigial origin, yet Gegenbaur
has objected that there is no other evidence of reversion, either in the
polydactyle extremity or in the general appearance of polydactyle animals.

We have shown that in polydactyle swine the abnormality is con-
fined to the manus, and that in most, if not all, cases the extra digits
represent the development of the normally vestigial pollex. In a third
of the cases a well-formed digit of two or three phalanges is found, and
when these conditions are compared Avith those of the manus of the
earliest fossil swine, it appears that the two are similar ; for a pollex is
found in the manus of the fossil pig, while in the pes the hallux is
entirely wanting. In addition to the development of the pollex, other
modifications were found in the structure of the polydactyle manus,
which seemed to reproduce a primitive, ancestral condition. We have
also seen that in most cases of polydactylism in the ox and horse the
extra digits represent the development of digital parts normally rudi-
mentary,— a development winch might bo regarded as duo to rever-
sion, for other parts of the polydactyle member show correlated variations,
and related fossil ancestors also have the same digits normally developed
and functional. Moreover, according to recent discoveries in hereditj'^,
single segregated characters may be inherited, without general modilica-
tion of the germ-plasm. This has been proved by Bateson and Saunders
(:02), Castle (;03, :03") and others in agreement with Mendel's law.

PKENTISS: rOLYDACTYLISiM IN MAN AND DOMESTIC ANIMALS. 303

The least answerable of the arguments against the general occurrence
of reversionary poly dactyl ism is the fact that more than five digits are
found in certain cases of polydactylism (man and cat), and that in other
cases the extra digits, though of vestigial origin, are exceedingly vari-
able, and often duplicated (swine and pes of Carnivora). Some factor
other than reversion must enter here, unless we assume with Albrecht
('86) that the tendency to digital duplication is reversion to the bifid
fin-rays of elasmobranch fishes, or with Bardeleben ('86) that the sixth
and seventh digits represent reversions to a hypothetical six-toed or seven-
toed ancestor. Albrecht's assumption seems absurd, for we know that
such duplications are of common occurrence in the development of other
structures to which his explanation of reversion cannot apply. Likewise,
it has been clearly shown by various investigators that Bardeleben's " prae-
pollex " theory is a mere assumption unsupported by the evidences of
anatomy, embryology, or palaeontology. For (1) the "prae-pollex " rudi-
ments never develop into digits and are not located in the region where
the supernumerary digits appear in man (Forster, '61; Gegenbaur, '88;
Zander, '9l). (2) They are not the vestigial remains of a degenerating
digit, but secondary developments, or neomorphs (Tornier, '89 ; Carlsson,
'90 ; Wiedersheim, :02). (3) The most primitive reptilian fossils (the
Ichthyopterygia) possess only five digits (Baur,'87). The "prae-pollex "
theory is thus rightly rejected by such eminent anatomists as Gegenbaur
and Wiedersheim. With it, as a consequence, must go the assumption
that polydactylism in pentadactyle extremities is a reversion to a hepta-
dactyle type.

In comparing the skeletal parts of the polydactylous manus shown in
Figure 13 (Plate 5)' and in Figure K with the normal and fossil condi-
tions (Figs. F and G), no one can doubt that reversion is the true cause
of such abnormalities. The same- conclusion holds true for a fully
formed hallux in the dog and for the cases of vestigial polydactylism in
the liorse and ruminants. It seems probable, however, from the varia-
tions which we have described in swine, that the character of digits pro-
duced by reversion is not firmly fixed in the germ, and that on crossing with
normal animals, the abnormal character, since it is dominant in Mendel's
sense of the word, is transmitted to the offspring, but in diflerent de-
grees of variation and duplication. Experimental breeding may settle
this question, but at present we can only argue from analogy with other
forms. Thus, Bateson found that the extra digits of the fowl varied
greatly on crossbreeding. But in the case of the fowl the extra digits are
sports, not palingenetic structures.

304 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

We have suggested the possibiUty that a factor in tlie production of
polydactylism in man, the cat, and the fowl may be reversion, not to a
Jiypothetical heptadactyle ancestor, but to the uuraodified minimus,
pollex or hallux of a not distantly related peutadactyle form. The re-
acquired structures might prove to be in their germinal characters, like
those of many neomorphs, so unstable as to lead to variations in tlie
next generation, such as polydactylous duplications.

We have evidence to show that in man, the cat, and tlie fowl it is not
a definite number of extra digits, but a tendency to digital variation and
duplication which is inherited. In man the minimus may l)o duplicated
on all extremities, but to a different degree iu each case, and the varia-
tions may increase in succeeding generations. Thus, Fackenheim ('88)
cites the case of normal parents whose daughter had a rudimentary sixth
finger on the idnar side of each hand. Of her two sons, one liad six
fully developed digits on each hand, tlie other six digits on all four
extremities! In another family the first parent observed had six toes on
each foot. Of eight children three were normal, three had six toes ('\n
one case correlated with hare-lip), and two had six fingers ; all the extra
digits were of symmetrical occurrence. In the three succeeding genera-
tions extra digits appeared now on the feet, now on the hands, and in two
cases on all four extremities. In two cases also, seven toes were present
on one or both feet.

In a family of cats observed by Poulton ('86) the abnormality ap-
peared in the third generation (number of extra digits not stated). Iu
the fourth generation six toes appeared on all four extremities. In the
fifth generation there were many individuals with seven toes on all paws,
and evidences of further duplication in the existence of doubled claws.
All gradations occurred between the extreme and normal form. This
condition prevailed up to the ninth generation, although in every case
the male parent was normal.

Torrey (:02) describes a similar case in which the offspring of a female
cat with six toes on the nianus and five on the pes showed all gradations
between the normal and a seven-toed condition. Often in these cats the
pollex was abnormally long and composed of tjiree phalanges instead of
two. In all cases digits ii-v were apparently normal in structure.

Bateson's breeding experiments show the same to be the case in the
polydactylous fowl. On crossing with normal birds all degrees of
variation are exhibited by the hallux, from simple elongation to
complete duplications and reduplications.

These observations bring out the important fact that often no extra

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 305

digit is produced, but simply a variation in the structure of the pollex,
hallux, and nmiimus. It would seem, therefore, that it is tins tenJeucy
of tlie modified digits to vary which is inherited.

We know that such digital variations occur also in the offspring of
normal individuals, and that they are inherited. Bateson cites the
occurrence of such a case in cattle and the formation of a three-toed
race thereby. The duplication of appendages is common in the lower
animals, and variation is of frequent occurrence in all neomorphic organs.
Well-known examples are tlie duplicated claws of arthropods and the
doubled horns of sheep. Polydactylism according to Fackenheim ('88)
is often correlated with abnormality by defect.

jSTone of these variations can be attributed to reversion. The law
of Mendel, as Bateson and Saunders (:02, p. 150) have pointed out,
" applies only to the manner of transmission of a character already
existing. It makes no suggestion as to the manner in which such a
character came into existence." Bateson regards the polydactyle fowl as
"a palpable sport;" tlie usual digital abnormalities of the fowl, the
cat, and of man undoubtedly belong to the same class of polydactylous
abnormalities. It is possible that reversion may be the primal cause in
producing certain of these digital variations, but the present evidence
does not warrant a positive statement to that effect.

h. Germinal Variation.

This has been regarded as the chief factor in polydactylism by Forster
('61), Darwin ('76), Gegenbaur ('80), Howes ('92), Weismann ('93),
Bateson ('94), Wilson ('96), and many others. Weismanu's view ('93,
p. 329) is, that excessive nutrition in the cells of the embryo may cause
the duplication of a group of determinants which are to form a particular
digit; the doubled condition of the determinants might then be in-
herited, and thus the inheritance of these digital abnormalities accounted
for. This, however, does not explain the changes in position which
digital variations in man may undergo in the course of hereditary trans-
mission (that is, from fingers to toes). Wilson ('96) attempts to clear
up this point by assuming tliat there may be variation in those determi-
nants which affect the nutrition of the digital fundament, and that it is
the tendency of these determinants to vary which is transmitted, rather
than the doubled condition of the digital determinants themselves.

There is some direct evidence that germinal variation is due to an
excess of nutrition. It has been observed by Ercolani ('81) and Boas
('85, '90} that certain polydactyle conditions in the ox and horse

306 BULLETIN : MUSEUM OF COMrAllATIVE ZOOLOGY.

occurred along with the atrophy, partial or coinplete, of the functional
digits, which apparently caused the subsequent development of the
normally rudimentary ones. In these instances it would seem that the
nutriment which is normally appropriated Ly the functional digits is
transferred to, and utilized by, the digital rudiments, tlius enabling
them to continue their development. We are familiar with the same
l)]ienomenon in plants, where, if the terminal bud is removed, lateral
buds, which would otherwise have remained dormant, are stimulated
to development by the extra supply of nutriment which they receive.
Again, polydactylism very often accompanies acephalic conditions, and
other abnormalities due to defect of some organ, as recorded by
Fackenheim and others. Here the same law is applicable ; on account
of the abnormal absence of certain organic fundaments, the remaining
ones receive more than their usual amount of nutrition ; as a result, an
increased development of normally reduced or otherwise modified digits
may be brought about. But these cases of polydactylism may also bo
explained as due to external influences acting in utero. Fackenlieim has
shown that in a certain family polydactylism did not appear as a correla-
tive of inherited abnormality by defect, until one of its members married
into another family in which digital' abnormalities were of frequent occur-
rence. Then only did offspring appear afflicted with both polydactylism
and defective teeth. From such cases the evidence that excess of nutri-
ment causes germinal variation loses much of its weight.

Any explanation of the phenomena of germinal variation must neces-
sarily be theoretical, as long as our practical knowledge of the germ-plasm
is so limited. We know, however, that all neomorphs are prone to varia-
tion. In polydactylism all the digital abnormalities produced by internal
causes vary greatly, and the tendency to variation is inherited. By
Mendel's law tlie inheritance of these variations is explained, and the
puzzling point whicli Wilson ('96) attempted to clear up by his theory of
nutritive variation, is made plain, — the fact that in man an individual
having a polydactyle man us may produce offspring with abnormal pes or
Avith all extremities abnormal. In this case we may assume that the
variation first appeared on all extremities as. a duplication of the mini-
mus, due to the doubling of the determinants of these digits. On
marrying with a normal individual the abnormal character would be
dominant, but not completely so (Bateson found this to be the case with
the polydactyle fowl). Of the DR offspring produced, some would be
abnormal like the D parent, but in others the usually dominant character
might be recessive ; their extremities might be entirely normal, or only

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 307

the hands polydactyle. In either case, however, they would be capable
of producing other DR offspring, if married to normal individuals, and
these offspring might themselves be normal or polydactyle ; should they
marry with recessive individuals like themselves, pure Z^'s wouhl be pro-
duced as well as RD's, and such individuals again would be polydactyle
on both hands and feet. "Wilson's theory of nutritive variation is thus
rendered unnecessary, as Mendel's law explains how all cases of polydac-
tylism, not due to external causes, may be the result of inheritance.

All such inherited types of polydactylism are thus ancestral. But
only those forms in which the extra digits develop directly from rudi-
ments and vestiges may be attributed to palingenetic reversion. In
those cases in which digital rudiments and vestiges are duplicated, rever-
sion and germinal variation may occur together ; but the duplications of
functional digits are probably caused by germinal variation alone. As
to tlie cause of these germinal variations, or sports, we know little or

nothing.

IX. Summary.

1. Polydactylism consists in an excess in the number of digits pos-
sessed by the individual over the number peculiar to the species.

2. The supernumerary digits generally occur symmetrically placed on
tlie right and left extremities, either in the manus, in the pes, or in both ;
they are found most frequently in the manus.

3. The extra digits are formed most frequently in connection with the
fifth and first digit in man ; with the first digit in the fowl, Carnivora,
and swine ; witli the second digit in ruminants and the Equidae. In
general, polydactylism may he said to aff'ect digits which are normally
much reduced or modijied.

4. Cases of polydactylism in which more than five digits occur cannot
be attributed to reversion alone (a heptadactyle ancestop is hypothetical,
the so-called prae-poUex and post-minimus are rudiments of secondary
development, and they have never been known to produce functional
digits).

5. PaHngenetic polydactylism is limited to those forms in which —
the number of functional digits being normally reduced to fewer than five
— the digital rudiments develop and reproduce, more or less completely,
the sti'ucture of homologous digits typical of some ancestral form. The
evidences of comparative anatomy, embryology, and palaeontology show
this to be the case in the horse, ruminants, and swine ; possibly in the
pes of Carnivora.

6. Tliis eventual dominance of a digital character, which has been

VOL. XL. — NO. G 5

308 BULLETIN : MUSEUM OF COMPAllATIVE ZOOLOGY.

transmitted in a recessive condition through many generations, is in strict
accordance with Mendel's law of heredity.

7. Neogenetic and palingenetic forms of polydactylism are, like other
new characters, extremely variable ; as they are hereditary, we may con-
clude that duplications of both functional and vestigial digits are due to
variations in the gametes.

8. The poly dactyl e abnormalities of man and the domestic animals
may be classified as follows :

I. Teralological polydactylism includes those cases of digital duplica-
tion and malformation which are produced by external influences ; it occurs
rarely in all animals, often in correlation with other monstrosities.

II. Neogenetic polydactylism includes those digital variations, or
sports, which are produced by some internal cause, presumably germinal
variation,

a. Duplication of unmodijied functional digits occurs occasionally in
all animals and is transmissible.

h. Variation of modijied but functional digits is the ordinary form
of polydactylism in man, the cat, and the fowl (pes), and it also is
transmissible.

III. Palingenetic polydactylism includes those cases in which digital
rudiments, or vestiges, develop into extra digits.

a. The extra digits reproduce more or less completely the structure of
the homologous functional digits of related fossil ancestors; this condi-
tion is found in the horse, ruminants, swine, and the pes of the dog.

h. The extra digits arise as variations or duplications of rudiments, or
vestiges ; they are neogenetic in so far as they do not reproduce ancestral
conditions. Examples are the hallux and pollex having three phalanges
and the various duplications of these digits found in the manus of swine
and the pes of Carnivora.

PRENTISS: FOLYDxVCTYLISM IN MAN AND DOMESTIC ANIMALS. 309

BIBLIOGRAPHY.

Ahlfeld, F.

'85-86. Die Verwaclisuugen des Amnion mit der Oberflaclie der Fmcht.
Berichte u. Arbeit, geburtshilfiicb-gyuakolog. Kliuik zu Marburg, Bd.
3, pp. 158-165.

Albrecht, [P.]

'86. Ueber den morpbologischen "Wertb iiberzabliger Finger und Zehen.
Centralbl. f. Chirurg., Jabrg. 1886, Beilage zu Nr. 2tt, pp. 105-107.

Anthony, R.

'99. litude sur la Polydactylie cbez les Galliuaces. (Poulet domestique.)
Journ. d. I'Auat. et Pbysiol., torn. 35, pp. 711-750, 25 fig.

Arloing, S.

'67. Contribution k I'etude de I'organisation du pied cbez le cbeval. Ann.
des Sci. Nat., ser. 5, torn. 8, pp. 55-81, 2 pi.

Bardeleben, K.

'85. Zur Morpliologie des Hand- und Fussskelets. Sitzungsber. d. Jeua.
Gesell. f. Med. u. Naturw. fiir das Jabr 1885, pp. 84-88.

Bardeleben, K.

'85". Ueber neue Bestandteile der Hand- und Fusswurzel der Saugetbiere,
so wie die normale Anlage von Rudimenten " iiberzabliger " Finger und
Zelien beini Menscben. Sitzungsber. d. Jena. Gesell. f. Med. u. Naturw.
fur das Jabr 1885, pp. 149-164.

Bardeleben, K.

'86. Hand und Fuss. Tagcbl. d. 59. Versamml. deutscber Naturf. u.
Aerzte zu Berlin, Jabrg. 1886, pp. 96-102.

Bateson, W.

'94. Materials for tbe Study of Variation ; treated with especial Regard to
Discontinuity in tbe Origin of Species, xv + 598 pp., 209 fig. London
and New York.

Bateson, W., and Saunders, E. R.

:02. Experimental Studies in tbe Physiology of Heredity. Reports to tbe
Evolution Committee of tbe Royal Society. Report I, 160 pp. London.

310 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Baumiiller, B.

'92. Polydactylie beiin Reh. Abh. naturhist. GescU. zu Niirnberg., Bd.
9, pp. 51-71, 1 Taf.

Baur, G.

'87. On the Morpliology and Origin of the Ichthyopterygia. Amer. Na-
turalist, Vol. 21, pp. S37-S40.

Blanc, L..

'93. Ltude sur la Polydactylie chez les Mammiferes. Ann. Soc. Linn. Lyon,
toni. 40, pp. 53-88, 29 fig.

Boas, J. E. V.

'85. Benierkungen iiber die Polydactylie des Pferdes. Morph. Jahrb.,
Bd. 10, pp. 182-184.

Boas, J. E. V.

'90. Ein Fall von voUstiindiger Ausbildung des 2. und 5. Metacarpale beira
Rind. Morph. Jahrb., Bd. 16, pp. 530-533, 2 fig.

* Carlsson, A.

'90. Von den weichen Theilen des sogenannten PrapoUex und Praliallux.
Eine vorlaufige Mittheilung. Verh. d. Biol. Vereins in Stockholm, Bd. 2,
No. 8, pp. 117-124.

Castle, W. E.

:03. Mendel's Law of Heredity. Proceed. Amer. Acad. Arts and Sci., Vol.
38, No. 18, pp. 533-548.

Castle, W. E.

:03*. The Heredity of Sex. Bull. Mus. Comp. Zool., Harvard Coll., Vol.
40, No. 4, pp. 187-218.

Chauveau, A., et Arloing, S.

'79. Traite d'auatoinie compare des aniraaux domestiques. 3'"'' ed., Paris.
1036 pp., 406 fig.

Cowper, J.

'89. On Hexadactylism. With especial Reference to the Signification of
its Occurrence in a Variety of Gallus doniesticus. Jouru. Anat. and
Physiol., Vol. 23, pp. 242-249, 2 fig.

Dareste, C.

'91. llecherches sur la production artificielle des monstruosites : ou, Essais
de teratogenic experimeutale, 2* ed. xvi + 590 pp., 16 pi.

Darwin, C.

'76. The Variation of Animals and Plants under Domestication. 2d ed.,
revised. ' New York. 2 vols, xiv + 473 ; x + 495 pp.

Ercolani, G.

'81. Delia polidaetilia e della poliinelia ncU' uomo e nei vertebrati. Mem.
Accad. Sci. Istit. Bologna, ser. 4, tom- 3, pp. 727-824, tav. 1-4.

PKENTISS : POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 311

Ewart, J. C.

'94. The Development of the Skeleton of the Limbs of the Horse, with
Observations on Polydactyly. Journ. Anat. and Physiol., Vol. 28, pp.
23G-256, 342-369, 25 fig. and pi. 12.

Fackenheim, J.

'88. Ueber einen Tall von hereditarcr Polydactylie mit gleichzeitig erblicher
Zahuanomalie. Jena. Zeitschr., Bd. 22, pp. 343-385, 7 Abbildungen.

Fbrster, A.

'61. Die Missbildungen des Menscheu. Jena, iv + 171 pp., 26 Taf.

Gegenbaur, C.

'80. Kritische Bemerkungeu iiber Polydactylie als Atavismus. Morph.
Jahrb., Bd. 6, pp. 584-596.

Gegenbaur, C.

'88. Ueber Polydactylie. Morph, Jahrb., Bd. 14, pp. 394-406.

Geoffroy-St. Hilaire, I.

'32-37. Histoire generate et particuliere des Anomalies de I'organisation
chez Phomme et les animaux, etc. 3 torn. Paris, xv + 746 ; 571 ;
618 + XX pp., avec Atlas, 20 pi.

Goodman, N.

'68. On a Three-toed Cow. Journ. Anat. and Physiol., Vol. 2, pp. 109-113.

Gurlt, E. F.

'77. Thierische Missgeburten. Ein Beitrag zur pathologischeu Anatomic
und Entwickelungsgeschichte. Berlin. 97 pp., 20 Taf.

Howe, F., Jr.

:02. A Case of Abnormality in Cats' Paws. Amer. Nat., Vol. 36, pp.
511-526, 18 fig.

Howes, G. B.

'92. On the Pedal Skeleton of the Dorking Fowl, with Remarks on Hex-
adactylism and Piialangeal Variatit)n in the Amniota. Journ. Anat. and
Physiol., Vol. 26, pp. 395-403, 5 fig.

Kollman, J.

'88. Haudskelett und Hyperdactylie. Verh. Anat. Gesell. 2te Versam.
in Wiirzburg, 1888, pp. 25-40, 1 Taf.

Kowalevsky, W.

'73. On the Osteology of the Hyopotamidae. Phil. Trans. London, Vol.
163, Pt. 1, pp. 19-94, pi. 35-40.

Kukenthal, W.

'89-93. Vergleichend-anatomische und entwickelungsgeschichtliche Un.
tersuchungen an Walthieren. Denkschr. med.-naturw. GeseU. zu Jena,
Bd. 3, 8 + 448 pp., 25 Taf. u. 124 Abbild.

312 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Marsh, O. C.

'79. Polydactyle Horses Recent and Extinct. Amer. Joum. Sci. and Arts,
scr. 3, Vol. 17, pp. 499-505, 4 fig., 1 pi.

Marsh, O. C.

'92. Recent Polydactyle Horses. Amer. Journ. Sei. and Arts, scr. 3, Vol.
43, pp. 339-355, 21 fig., 1 pi.

Mendel, G.

'66. Versuclie iibcr Pflauzeu-IIybrideii. Verb, naturf. Vercincs in Briinn,
Bd. 4, Abbandl., pp. 3-47.

Morand, [S. F.]

1773. Recliercbes sur quelques conformations monstrueuses des doigts dans
I'homme. Mem. Acad. Roy. Sci., Paris, pour I'auuee 1770, pp. 137-150,
pi. 4-12.

Otto, A. W.

'41. Moustrorum sexcentoruni descriptio anatomiea. xx + 335 pp., 30 tab.
Vratislaviense.

Poulton, E. B.

'83. Observations on Heredity in Cats witb an Abnormal Number of Toes,
Nature, Vol. 29, pp. 20-21, 8 fig.

Poulton, E. B.

'86. Observations on Heredity in Cats witb an Abnormal Number of Toes.
Nature, Vol. 35, pp. 38-41, 8 fig.

Piitz, H.

'89. Einc iiberzablige Zelie bei einem Pferde. Deutsch. Zeitschr. f. Tliier-
medicin, Bd. 15, pp. 224-232, Taf. 5.

Rosenberg, A.

'73. Ueber die Entwicklung des Extremitaten-Skeletes bei eiuigen durch
Reductionen ihrer Gliedmassen characterisirten Wirbelthicrcn. Zeitscbr.
f. wiss. Zool., Bd. 23, pp. llG-166, Taf. 5-7.

Scott, W. B.

'95. The Structure and Relationships of Aneodus. Journ. Acad. Nat. Sci.,
Philadelphia, ser. 2, Vol. 9, Pt. 4, pp. 461-497, pi. 23, 24.

Struthers, J.

'63. On the Solid-hoofed Pig ; and on a Case in whicb the Eore Foot of the
Horse presented Two Toes. Edinb. New PJiil. Journ., New Ser., Vol.
17, pp. 273-2S0, 2 fig.

Struthers, J.

'63^ On Variation in the Number of Fingers and Toes, and in tlic Number
of Phalanges, in Man. Edinb. New Phil. Journ., New Ser. Vol. 18, pp.
83-111, pi. 2.

PRENTISS: POLYDACTYLISM IN MAN AND DOMESTIC ANIMALS. 313

Suetonius (Caius Suetonius Tranquillius).

'86. De vita Caesaruiu. (C. von llotli) Lipsiae. civ + 357 pp.
[Original appeared about the year 120 a.d.]

Tornier, G.

'89. Giebt-es eiu Prahallux-rudimeut ? Sitzuiigsber. Gesell. naturf. Freuudc
zu Berlin, Jahrg. 18S9, pp. 175-182, 1 Fig.

Torrey, H. B.

■■02. Prepotency in Polydactylous Cats. Science, N. S., Vol. 16, pp.
554-555.

Verrier, E.

'85. Des anomalies symmetrique des doigts et du role que Ton pourrait
attribucr k I'atavisme dans ces anomalies. Compt. Rend. Acad. Sci.,
Paris, torn. 100, pp. 865-867.

Weismann, A.

'93. The Germ-Plasm. A Theory of Heredity. Translated by W. N.
Parker, xxii + 477 pp., 24 fig.. New York.

Werner, T.

'97. Polydactylie beim Schwein. Sitzungsber. Gesell. naturf. Freunde zu
Berlin, Jahrg. 1897, pp. 47-48.

Wiedersheim, R.

:02. Der Bau des Menschen als Zeugniss fiir seine Vergangenheit. 3te
Aufl., Freiburg, i. B. 243 pp., 132 fig.

Wilson, G.

'96. Hereditary Polydactylism. Journ. Anat. and Physiol., Yol. 30, pp.
437-449, 2 fig.

Windle, B. C. A.

'91. The Occurrence of an Additional Phalanx in the Human Pollex.
Journ. Anat. and Physiol., Vol. 26, pp. 100-116, pi. 2.

Winter, G. S.

1703. Tractatio nova et auctior de re equaria. Niirnberg, 14 + 223 pp.,
34 tab.

Wood-Mason, J.

'71. [A case of Polydactylism in a Horse.] Proceed. Asiat. Soc. Bengal,
1871, pp. 18, 19, pi. 1.

Zandev, R.

'91. 1st die Polydactylie als theromorphe Varietal oder als Missbildung
anzusehen ? Arch. f. path. Anat., Bd. 125, pp. 453-487.

314

bulletin: museum of comparative zoology.

EXPLANATION OF PLATES.

The figures are all reproduced from natural size skiagraplis of the polydactyie
specimens; in every plate the distal ends of tiie extremities are down, but right
and left are reversed. liiglit extremities therefore appear as left in tlie figures,
and vice versa.

asg

cac

cub

cun

ec'cun. . . .
en'cim. . . .
ext. com. dg. i. .

ext. mt'carp. mag.

ext. mt'carp. oh.

ext.prp. . . .

ext.prp.ex. . .

ext. prp. i. . .

fix. per/. . . .

ABBRE

Astragalus.

Calcaneum.

Cuboid.

Cuneiform.

Ectocuneiform.

Entocuneiform.

Extensor communis
digitorum internus.

Extensor metacarpi
magnus.

Extensor metacarpi
obliquus.

Ext. proprius poUicis
et indicis.

Extensor proprius ex-
tern us.

Extensor proprius in-
ternus.

Flexor perforatus.

VIATIONS.
Jlx. per/.'
lun. .
ms'cun.
mt'carp.
vit'tar.
nav. .
n. m. .
n. uln.
OS. mag.
phlx. ,
pis.
scph. .
trz. .
trzd. .
un.

i-v .
1-6 .

Flexor perforans.
Lunar.

Mesocuneiform.
Metacarpal.
Metatarsal.
Navicular.
Median nerve.
Ulnar nerve.
Os magnum.
Phalanx.
Pisiform.
Scaphoid.
Trapezium.
Trapezoid.
Unciform.
First to fifth digits.
First to fifth branches
of the median nerve.

Prentiss. — Polydactylism.

PLATE 1.

All figures are skiagraplis of human appendages.

Fig. 1. Kiglit foot of foetus, No. OTuO.

Fig. 2. Left foot of foetus, No. 6730.

Fig. 3. Left hand of foetus, No. 912.

Fig. 4. Riglit liand of foetus, No. 012.

Fig. 5. Left foot of foetus, No. 912.

Fig. 0. Right foot of foetus, No. 912.

Prentiss- PoLYDACTYLiSM.

Plate 1.

II Vb

III IV Va

II

Vb

Va IV 111

« ^

Vb

'^' Va '^ " n
III

TV

II

Va

111 IV

n

Vh

Va

Vb

Va

IV

m

IT

IV

III

Pbentiss. — Polydactylism.

Fig.

7.

Fig.

8.

Fig.

9.

Fig.

10.

PLATE 2.

All figures are from skiagraphs of human foetal appendages.

Left hand of foetus, No. 5809.

Right hand of foetus, No. 5809. Note. — Tlie metacarpal mentioned in
the text (p. 254) lias failed of reproduction in the printing of this plate.
Right hand of foetus. No. 913.
Left foot of foetus, No. 913.

Prentiss.-Polydactylism.

Plate 2.

VI-.

8

■m

0

V

i.

w

Va

IV

in n

II

III

Vb

\^^

tv

^Br

0

1 "^^

M,

I

II

Va

IV

in

\i,

Va

n

IV III

Prentiss. — Polydactylism.

PLATE 3.

Fig. U. NorniJil left manus of the pig, anterior view, showing skeletal structure
of the digits.

Prentiss.-Polydactylism.

Plates.

11

IV'

in

Prentiss. — Polydactylisin.

PLATE 4.

Fig. 12. Anterior view of left polydactyle manus of the pig, showing a small
supernumerary digit (i) and the lower row of carpals.

Prentiss.-Polydactylism.

Plate 4.

12

nn.

.-ON. IlKltJ.

trsd.

II

IV

III

Prentiss. — Polydactylism.

PLATE 5.

Fig. 18. Anterior view of the left polydactyle niiinus of the pig, showing a fully
developed pollex (i) and the bones of the carpus.

Prentiss- PoLYDACTYLisM.

Plate 5.

13.

II

IV

III

Pekntiss. — Polydactylism.

PLATE 6.

Fig. 14. Anterior view of Ivit polydactyle nianus of the pig willi one super-
numerary digit (i), and digit ii abnormally large.

Prentiss -PoLYDACTYLiSM.

Plate 6.

! -i

I

k

J

trzO.

^

trz.

V

IV

11

III

Pbbntiss. — Polydactylisni.

PLATE 7.

Fio. 15. Anterior view of the left polydactyle nianus of the pig, sliowing a
supernumerary digit (i), to the proximal end of which the trapezium
is fused.

r'DACTYLISM-

Plate 7.

IV

in

Prentiss. — Polydactylism.

PLATE 8.

Fig. 16. Anterior view of the left niantis of a polydactyle pip, showing the
lowor row of carpals, a supernumerary digit (i), and digit (ii) abnor-
mally developed.

Prentiss.-Polydactylism.

Plate 8.

-trzd.

In

V

111

IV

II

Prentiss. — Polydactylism.

PLATE 9.

Fig. 17. Anterior view of the left manus of a polydactyle pig, showing the
lower row of carpals and a large superniinurary digit (i).

Prentiss- Polydactylism.

Plate 9.

irs'l.

in

iv

Prentiss. — Polydactylism.

PLATE 10.

Fio. 18. Anterior view of the right manus of a polydactyle pig, sliowing the
lower row of carpals and two supernumerary digits borne on meta-
carpal I.

Prentiss -PoLYDACTYLiSM.

Plate 10.

trzd. ;

d

la

I(

rii

IV

Prentiss. — PolyJactylism.

PLATE 11.

Fig. 19. Anterior view of the left polydactyle manus of a polydactyle pig,
sliowing the lower row of carpals, and two extra digits borne on meta-
carpal I.

Prentiss.-Polydactylism.

Plate 11.

9.

II

IV

III

PuENTiss. - I'ulydactylism.

PLATE 12.

Fig. 20. Anterior view of the left manus of a polydactylc pig, showing two
complete supernumerary digits enclosed distally in a single hoof.

Prentiss.-Polydactylism.

Plate 12.

trzd.
-- trz.

la

IV

II

III

Prentiss. — Polydactylism.

PLATE 13.

Fig. 21. Anterior view of tlie right manus of a polydactyle pig, showing two com-
plete supernumerary digits.

Prentiss.-Polydactylism.

Plate 13.

21

Irz

trzd

\

V

II

la

lb

III

IV

PRBNT183. — rolydactyliBm.

PLATE 14.

Fig. 22. Anterior view of the left polydactyle manu8 of a polydactyle pig,
showing the lower row of carpal bones, two supernumerary digits, and
the rudimentary phalanges of digit ii.

Prentiss- PoLYDACTYi ism

Plate 14.

;,■,;■:'.

22.

trz.

n

*|k

lb

IV

Hi

Pbentiss. — Polydactyligm.

PLATE 15.

Fio. 23. Anterior view of the left manus of a polydactyle pig in which two
large supernumerary digits are present, but digit ii is absent.

Prentiss -PoLYDACTYLiSM.

Plate 15.

scph.

23.

scpK

. trzd.
Irz.

\

"Hx

la

IV

lb

III

PuBMTiss. — PolydactyliBui.

PLATE 16.

Fio. 24. Anterior view of the kft nianus of a polyJactylc pig, sliowiiig two
fully formed supernuuierary digits, and the rudimeuta of a third.

I

Prentiss- PoLYDACTYLiSM.

Plate 16.

..scph.

24

ftcpJi'.

.. trz.
...trz.

I.

\c

^

II Ih

la

IV

HI

Prentiss. — Polydactylisui.

PLATE 17.

Fig. 25. Anterior view of the right nianus of a polydactyle pig, showing an extra
digit borne on metacarpal ii, and the lower row of carpals.

Prentiss.-Polydactylism.

Plate 1 7.

IH

IV

Pkbhtisb. — Polydactylism.

PLATE 18.

Fio. 26. Anterior view of the left manus of a polydactyle pig, showing a large
supernumerary digit, the metacarpal of which is fused to that of
digit II.

Prentiss.-Polydactylism.

Plate 18.

26.

^

.trzd.

s

_.trz.

#

II

IV

III

PuENTiss. — Polydactylism.

PLATE 19.

Fig. 27. Anterior view of the left manus of a polydactyle pig, showing two
extra digits, one of which is borne on metacarpal ii.

Prentiss.-Polydactylism.

27

in

Plate 19.

trzd.

■ trz.

I

X

la

lb

11

IV

III

Pbentiss. — Polydactylism.

PLATE 20.

Fig. 28. Anterior view of the left maiius of a polydactyle pig, sliowing two
extra digits, one of wiiicli (i*) is borne on the same metacarpal with ii.

Prentiss- PoLYDACTYLiSM.

Plate 20.

28.

^W

i^

•^.

la

IV

lb

111

PitENTiss. — PolydactyliBiu.

PLATE 21.

Fig. 29. Anterior view of tlie left manus of a polydactyle calf, showing only tlie
distal extremity of the metacarpus, and a supernumerary digit (ii).

Prentiss.-Polydactylism.

Plate 21,

II

IV

III

Pbkntiss. — Polydactyliem.

PLATE 22.

Fig. 30. Anterior view of right raanus of same calf as Fig. 29, sliowing one
extra digit (ii).

Prentiss- PoLYDACTYLisM.

Plate 22.

II

III

S

IV

i

Bulletin of the Museum of Comparative Zoology

AT HARVARD COLLEGE.

Vol. XL. No. 7.

THE CHANGES WHICH OCCUR IN THE MUSCLES OF A

BEETLE, THYMALUS MARGINICOLLIS CHEVR.,

DURING METAMORPHOSIS.

Bt Robert S. Breed.

With Seven Plates.

CAMBRIDGE, MASS., U.S.A.:

PRINTED FOR THE MUSEUM.

October, 1903.

No. 7. — CONTRIBUTIONS FROM THE ZOOLOGICAL LABORATORY
OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD
COLLEGE, UNDER THE DIRECTION OF E. L. MARK, No. 145.

The Changes which occur in the Muscles of a Beetle^ Thymalus
marginicollis Chevr., during Metamorphosis.

By Egbert S. Breed.

TABLE OF CONTENTS.

Introduction

Part I. — Anatomy

A. Historical Survey ....

B. Observations

1. Material

2. Methods

3. Anatomical changes of the

muscles

a. Metathorax

( 1 ) . Dorsal antero-posterior

muscles

(2). Lateral dorso-ventral

muscles

(3). Ventral antero-poste-
rior muscles . . .
6. Mesothorax

c. Prothorax

d. Head

e. Abdomen

/. Appendages

4. Discussion of results . . .
Part II. — Histology

A. Historical survey ....

B. Observations

1. Methods

2. Histological changes of the

muscles

PAGE

317
318
318
319
319
321

321
322

323

324

336
337
338
338
338
339
339
340
340
347
347

348

PAQZ

a. Muscles that pass unal-
tered from the larva to
the imago ....
h. Metamorphosis of larval
muscles into
(1). Muscles of the wing

type

o. Larval period . . .

/3. Pupal period . . .

y. Imaginal period . .

(2). Muscles of the leg type

a. Larval period . . .

/8. Pupal period . . .

7. Imaginal period . .

(3). Metamorphosis of the

intestinal muscles . .

c. Histolysis of the larval

muscles 361

d. Histogenesis of the imag-

inal muscles . . . 363
3. Observations on other Cole-

optera 364

C. Discussion of results . . . 366

Summary 371

Bibliography 375

Explanation of plates 380

349

349
349
353
355
356
357
357
358

358

Introduction.

While there have been numerous researches on the changes which
occur during the metamorphosis of insects, many points remain not
clearly understood, and others are in dispute. The present investigation

VOL. XL. — NO. 7 1

318 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

has been undertaken with the purpose of aiding, if possible, in the ex-
planation of some of these alterations, and thus to untangle the confusion
in regard to them. A detailed study has been made on Coleopterous
material, since beetles were found to present a fairly simple metamor-
phosis of the muscular system.

These changes naturally fall into two groups ; the anatomical and the
histological. Previous papers on this subject have ignored almost com-
pletely the anatomical side of the question. This one-sided method has
been responsible for much of the confusion which has arisen.

In connection with this neglect of the study of the anatomy of the
muscles, most authors have assumed that all of the muscles of any one
insect undergo similar changes during pupal life. Yet, it is conceiv-
able that any one of, or any combination of, the following conditions may
be found in a single holometabolic insect :

a. The larval muscles may not be changed, but pass unaltered into the
imago.

b. The larval muscles may undergo a more or less complete metamor-
phosis into the imaginal muscles.

c. The larval muscles may degenerate entirely, and the imaginal
muscles form anew in the pupa.

As the results of this research show that a combination of these three
methods is found in Coleoptera, and as the remaining orders of metabolic
insects are probably fundamentally like Coleoptera, it is not strange that
contradictions have arisen. It is possible that two investigators, even
though working on the same species, have, in studying different muscles,
studied different conditions.

This investigation was undertaken at the suggestion of Dr. E. L.
Mark. During the three years that I have been engaged in the work,
he has constantly aided me by his advice and criticism. To him, my
heartiest thanks are due. I also wish to express my thanks to Mr.
Samuel Henshaw, of the Museum of Comparative Zoology, for his many
kindnesses.

Part I— Anatomy.

A. Historical Survey.

The dissections of the muscular system of insects are not very numer-
ous, and, as the homologies of the muscles are difficult to determine, the
comparative myology of insects is not in a very satisfactory condition.
Those investigations which have been published are, with few exceptions,

bkeed: metamorphosis of the muscles of a beetle. 319

based on dissections in which only imaginal forms liave been used. The
few exceptional cases in which larval forms have been used happen to be
dissections of larvae from orders of insects other than Coleoptera. The
best attempt that has been made as yet to establish the homologies of
the imaginal forms is that of Petri ('99), who has studied the muscular
systems of Trichoptera, Diptera, and Hymenoptera, On account of this
unsatisfactory state of the comparative myology, no attempt will be made
to homologize the muscles of Coleoptera with those of otlier orders.
Consequently, only those papers that deal with Coleoptera will be men-
tioned. A very good review of the whole ground is given by Petri ('99).
Of the three papers that deal with the imaginal muscular system of
Coleoptera, the monumental work of Straus-Diircklieira ('28), on Melo-
lontha vulgaris, is the first and most important. The nomenclature used
by him is, however, unsatisfactory, as it is not generally applicable. The
next paper in importance for us is that of Luks ('83), who gives good
figures and a short description of the thoracic musculature of Dytiscus
marginalis Linn. He finds the musculature much the same as in
Melolontha, with the exception of the coxal muscles of the metathorax.
Owing to the firm fusion of the coxae to the metasternum, the func-
tions of the coxal muscles have changed. These muscles serve either
as indirect wing muscles, or as flexors or extensors of the trochanter.
The Latin nomenclature used by him is founded principally on the func"
tions of the muscles. It is the best nomenclature available, and is there-
fore used as far as practicable in this paper. When the homologies shall
have been made clear, probably a modification of the nomenclature of
Amans ('85), founded on the attachments and positions of the muscles,
will be used for all orders of insects. In his paper, Amans gives a short
description of the wing muscles of beetles..

Observations.

1. Material.

The principal material used has been Thymalus marginicollis Chevr.,
one of the Trogositidae. Marginicollis (Chevr. 1842) is used as the
specific name of this species by the authority of Leveille ('88), who, in
his catalogue of the Temnochildes (=Trogositidae), substitutes this name
for fulgidus (Erich. 1844), the name in most common use. Inasmuch as
marginicollis is figured in the original description, and has priority, it
certainly ought to be used. This species lives in Polyporus betulinus,
the common shelf fungus growing on white birch (Betula populifolia

320 bulletin: museum of comparative zoology.

Ait. ; Dr. Eoland Thaxter tells me that it is also sometimes found on B.
papyifera Marshall). This species of Tliymalus is entirely North Ameri-
can, so far as recorded, being found within, and limited to, the re^jions
occupied by these species of white birch. The localities recorded are
Canada, Maine, New Hampshire, Vermont, Massachusetts, New York,
Pennsylvania, New Jersey, Michigan, Wisconsin, and Iowa.

The only account of its life history is that of Beutenmueller ('90),
who gives little more than an accurate description of the larva and pupa,
^ly specimens agi'ee Avith his in every particular, excepting in regard to
the size of the larva. He states that the larvae are 6 mm. by 3 mm.,
whereas my specimens of full grown larvae are not as broad, being only
2-2.5 mm. broad by 6-7 mm. long. Material has been obtained in the
spring from three localities aboxit Cambridge; viz., Middlesex Fells,
Arlington Heights, and Belmont. The eggs are deposited in the fall and
liatch in the spring. Young larvae, 2-5 mm. long, were found in the
fungi as early as the 17th of April, 1901, and the 4th of April, 1902.

The larvae grow rapidly, bore through the fungus in various directions,
and finally excavate a chamber at the end of the burrow, in which to
pupate. These chambers are usually made in the upper portion of the
fungus. A drawing of a resting larva, taken from one of the chambers
is shown in Figure 6 (Plate 2). Peculiar hooked hairs are found on the
under side of the abdomen, as shown in the drawing. These hairs are
found on all of the older larvae, but not on the younger ones (2-4 mm.
long), nor on the pupae. Inasmuch as the points of the hooks are
turned forward, it seems as if these hairs woidd seriously impede the
forward locomotion of the larvae. However, this would probably not
be a great hinderance to the larvae, since they move but a few inches
during the month or more of their existence. No use for these hairs
can be suggested until further knowledge of the habits of the larvae is
obtained.

The first pupa from the larvae obtained April 17, 1901, appeared May
9th. These larvae, kept in a laboratory where the temperature was from
15°-22° C, had all pupated by the 13th of May. A drawing of one of
the pupae is shown in Figure 8 (Plate 3). These pupae took from 8-10
days to mature, the first imago appearing'May 19th. There is consider-
able variation in the date of the appearance of the imagines of this
species, as larvae were obtained out of doors on May 29th. These did
not begin to pupate till June 4th. The first of the beetles appeared in the
imaginal state June 11th, while several did not appear until a few days
later. It is probable that the beetles appear normally about the first of

breed: metamokphosis of the muscles of a beetle. 321

June. As long as they were under observation, i.e., till the first part of
July, they showed no signs of leaving the protected places about the
fungus from which they hatched. Inasmuch as the Polyporus which
serves the larvae as a food plant is an annual, there is probably but one
brood during the year, the eggs not being deposited until fall.

Thymalus is a particularly good form for histological study, inasmuch
as material seems to be plentiful wherever there is a food supply. It is
of convenient size and has a relatively thin cuticula at every stage.

2. Methods.

Since Thymalus is a small beetle, it has been necessary in studying
the anatomy of the musculature to resort to reconstructions from sections
in place of dissections. Material killed in hot water, or by some method
which gave no distortion, was used, and serial sections cut 16| ^u. in
thickness. To obtain a plane for reconstructiou, a " definition appa-
ratus " made by Zimmermann has been used. By means of this apparatus,
the lateral faces of the paraffin block were cut exactly perpendicular to
each other and to the proposed plane of sectioning. Two adjacent lateral
surfaces were then painted with a mixture of soft paraffin and lampblack,
melting at about 51° C, after which each face was again trimmed in the
" definition apparatus " so that only a very thin layer of paint was left.

The sections were cut on a Minot microtome in a plane perpendicular
to that of the painted surfaces. In mounting the sections, much of the
lampblack washes away, but, with ordinary care in the staining and
other processes, enough adheres to the albumen affixative to give a very
definite line at the outer edge of the lampblack area. A magnification
of 120 diameters was used in all of the reconstructions, as this made the
thickness of each section equivalent to 2 mm. The drawings made from
the reconstructions have been reduced to -^g of their original size in the
process of reproduction, so that the ultimate magnification in the plates
is about 67.5 diameters.

Whole and partial preparations have been used in checking the results
of reconstruction.

3. Anatomical Changes of the Muscles.

Early in my study of the histological alterations of the muscles in
Coleoptera, it was found that all of the muscles do not undergo the same
changes. Some remain unchanged from larva to imago, many metamor-
phose, and a few degenerate. Whether or not there were any newly

322 bulletin: museum of comparative zoology.

formed in the pupa, it was impossible to say without a systematic
search. To settle this question, and also to find out precisely which
muscles remain unchanged, which metamorphose and which degenerate,
a detailed study of the musculature of the metathorax was made. This
is for Coleoptera, the most important somite as far as the muscular
system is concerned. After completing the study of the metathorax, it
was found to be unnecessary to investigate the anatomical changes of
the muscles of the other somites except in a general way.

In connection with this study of Thymalus, a dissection of Colymbetes
sculptilis Harr., one of the Dytiscidae, was made in order to permit a closer
comparison with the dissection of Dytiscus marginalis by Luks. The anat-
omy of the imaginal musculature of Synchroa punctata Newm. (Melan-
dryidae) and of Bruchus obtectus Say (Bruchidae) has also been studied.
The two latter species have been studied from serial sections, both being
too small to be dissected successfully. This gives five beetles, of as many
different families, for comparison, to which may be added the dissection
of Melolontha by Straus-Durckheim. Several points of difference in
various muscles were found among these beetles, which are noted at the
end of the description of the muscle in question. Where nothing is
stated to the contrary, it may be understood that the conditions in the
other forms agree essentially with those in Thymalus.

a. Metathorax.

The muscles of the larval metathorax, or of any larval somite, may be
naturally separated into three groups; the dorsal antero-posterior, the
ventral antero-posterior, and the lateral dorso-ventral.

The function of most of the muscles of the larval metathorax is to aid
in locomotion. Some of the lateral dorso-ventral muscles are attached
to the legs and serve as flexors or extensors. The antero-posterior
muscles of both groups serve to bend the body in one direction or an-
other. All of the muscles are employed in a not very successful creeping
movement, similar to the creeping movements of certain Annelids, such
as the earthworm. That is, the longitudinal muscles oppose the dorso-
ventral muscles through the medium of th6 body fluid.

In the imago the muscles may, or may not, retain their larval func-
tion. Most of the leg muscles retain their former function, but many of
the others, including all of those which form tlie imaginal wing muscles,
change their function during pupal life. From this, it is readily seen
that many of the names of these muscles, given from their function in

breed: metamorphosis of the muscles of a beetle. 323

the imago, are misnomers when applied to the muscle in its larval state.
Even though such misnomers may cause confusion, they are retained in
this paper because no better nomenclature is available at present.

In the detailed description of the muscles, the order followed is :
(1) dorsal antero-posterior, (2) lateral dorso-ventral, and (3) ventral
antero-posterior. By this arrangement, the wing muscles of the imago,
both direct and indirect, are spoken of first.

(1) Tlie dorsal antero-poderior group of muscles is shown in Figure 1
(Plate l), which is a view of the left side of the larval metathorax seen
from above (dorsal), anterior being up on the plate. Figure 2 is a
similar view of the pupal metathorax. In the upper portions of Figure 9
(Plate 4) and Figure 1 1 (Plate 5) is shown the same group of muscles
in the imago as they would appear when seen from the left side of the
thorax, after cutting away the lateral wall of the metathorax.

Musculus metanoti of Luks.
(Abaisseur de Vaile of Straus-Dlirckheim ; dorsal of Amans.)

The musculus metanoti is one of the most important of the indirect
wing muscles, since it functions as the principal depressor of the wing
in the imago. In the larva (Plate 1, Figure 1, mt'nt.) it exists as three
distinct muscles, extending from the anterior to the posterior boundary of
the metathorax. At this stage the three muscles do not even lie parallel
to one another. It is their subsequent history only which shows that they
constitute one imaginal muscle. Just before pupation, in a larva which
is no longer feeding, these three muscles show histological evidences of
metamorphosis, which will be described later. There is very little
change anatomically, till pupation, when there is a quite rapid shifting
of the attachments of the three muscles, caused by the unequal growth of
the hypodermis. In the pU2m (Figure 2, mfnt.) they still extend
throughout the entire length of the somite, but have changed their rela-
tive positions so that now they lie parallel to one another. In the older
pupa they grow in size until they touch each other, and in the young
imago (Plate 4, Figure 9; Plate 5, Figure 11, mt'nt.) they become so
united as to be almost indistinguishable. Each of the three original
muscles has divided lengthwise into from three to nine fibres, so that the
entire adult muscle is composed of about fifteen fibres.

During pupal life thei'e is- formed an ingrowth of the hypodermis
along the dorsal portion of the suture between the meso- and metathorax,
and from this is formed the mesophragma of the imago (Plate 4, Figure 9,

324 bulletin: museum of comparative zoology.

ms'phg.^. Since the infolding hypodermis of the pupa carries witli it
the attachments of the anterior end of this muscle, the musculus
metanoti is attached in the imago to the posterior face of the mcso-
phragma. The metaphragma {mfphg.) is formed by a similar infolding
at the posterior margin of the somite', and consequently the posterior
end of the muscle is attached to tlie anterior face of this ingrowth.

Musculus lateralis metanoti of Luks.
(Pretradeur de Vaile of Straus-Durckheim ; latero-dorsal of Amans.)

This muscle is present in the larva (Plate 1, Figure 1, Z. mf'nt.) a.s
two, or occasionally three, fibres. "When three fibres are present, the
two more lateral are always closely approximated, as in the case figured ;
this, then, is a simple doubling of the more usual single fibre. These
fi,bres do not stretch through the full length of the nietathorax, but
extend from a suture (Plate 1, Figure 2, suf. a.) — which probably
represents the posterior boundary of theprescutum — posteriorly and later-
ally to the posterior edge of the somite. In the JW^ (Figure 2, I. mfnt.,
drawn from an animal which had but two fibres in the larva) these two
or three fibres become approximated, and in the old pupa fuse to form a
single muscle. In the imago (Plate 4, Figure 9, I. mfnt.) the attach-
ments of this muscle are, anteriorly, to the anterior portion of the scutum,
and, posteriorly, to the postscutellum and metaphragma.

The muscles which degenerate (Plate 1, Figure 1, a, /3, y, 8, c, ^, rj) are,
in general, those of the deeper layer, and all of them except a extend
the full length of the somite. In the young jm^ja (Figure 2, a, /?, y, 8,
c, ^, 77) they are still present, showing, however, even anatomical evidences
of degeneration. They are very irregular in outline, and do not extend in
a straight course from origin to insertion, because they are greatly re-
laxed. No traces of them can be found in old pupae and imagines.

(2) The lateral doi'so-ventral group of muscles of the larva is by far
the most important of the three groups, since from it are developed
nearly all of the muscles of the metathorax of the imago. This group is
shown in lateral aspect for the larva in Figures 3 and 4 (Plate l) ; for
the pupa in Figure 5 (Plate 2) and Figure 7 (Plate 3), and for the imago
in Figure 9 (Plate 4) and Figure 11 (Plate 5). Figures 4, 5, and 9 show
the more superficial lateral layer of muscles in their respective stages.
The group embraces no less than twenty-seven muscles on each side of
the metathorax : viz. :

bkeed: metamokphosis of the muscles of a beetle. 325

Musculus lateralis metatharacis anterior of Luks.
(Elevateur de I'aile of Straus-Diirckheim ; stei-nali-dorsaux of Amans.)

In the larva (Plate 1, Figure 3, I. mfthx. a.) this muscle is composed
of two fibres, extending vertically downwards from the antero-dorso-
lateral portion of the metathorax. to their attachment near the anterior
edge of the metathoracic leg. It serves as an extensor of the leg. Even
in the young p^pa (Plate 3, Figure 7, I. mfthx. a.), these two fibres
become so fused that they cannot be distinguished from each other, ex-
cept in cross sections of the muscle. In common with the corresponding
attachments of all of the dorso-ventral muscles, the ventral attach-
ment of this muscle becomes shifted posteriorly by the very consider-
able posterior growth of the ventral portion of the metathorax. The
muscle, therefore, changes in its general direction, becoming directed
obliquely downward and backward. In the imago (Plate 5, Figure 11,
Z. mfthx. a.) this muscle forms the anterior portion of the musculus
lateralis metathoracis, which serves for the elevation of the wings. At
its dorsal end, it attaches to the anterior lateral part of the scutum.
Ventrally, it attaches near the median line of the metasternum ; but,
contrary to the condition found by Straus-Diirckheim in Melolontha and
by Luks in Dytiscus, no fibres attach to the lateral faces of the median
lamina of the metafurca (inffur. 4).

Musculus lateralis metathoracis posterior of Luks.
(Synonymy as with the anterior muscle.)

This muscle is found in the larva (Plate 1, Figure 3, I. mfthx. p.)
as a single fibre immediately posterior to musculus lateralis metathoracis
anterior, with which it is nearly parallel. This relation is continued in
all stages of the pupa (Plate 3, Figure 7, I. mfthx. p.) and in the
imago (Plate 5, Figure 11, I. mfthx. p.). The muscle attaches in
the imago, dorsally, to the lateral portion of the scutum and, ventrally,
near the median line of the metasternum. In the adult Thymalus, the
anterior and posterior muscles are separated farther from each other than
in tlie larva ; but in the other beetles examined, as well as in Dytiscus
(Luks), they may be so fused that they cannot be readily distinguished
from each other.

Flexor coxae vietathoracis secundus of Luks.
(Second flechisseur de la hanche of Straus-Diirckheim.)

While this muscle acts as a flexor of the posterior coxa, it also acts in
the imago as an elevator of the Aving. It is, therefore, described here

326 bulletin: museum of comparative zoology.

among tlie wing muscles. In the larva (Plate 1, Figure 3, fix. cox.
niftJix. 2') it is composed of three fibres, extending from the dorso-lateral
portion of the metathorax vertically downward, and attaching to the
posterior side of the leg. It serves in this stage exclusively as a flexor
of the coxa, since no wings are present. The three fibres become closely
approximated during impal life (Plate 3, Figure 7, fix. cox. mftlix. 2).
The dorsal attachment in the imago (Plate 5, Figure 11, fix. cox. mt'fhx. 2)
is to the posterior part of the scutum, from which it extends downward
and backward to attach to the ventral surface of the middle of the coxa.

Extensor alae magnus metathoracu of Luks.
{Extensor anterieur de Vaile of Straus-Dilrckheim ; preaxillaire of Anians.)

The great extensor of the wings is composed in the larva (Plate 1,
Figure 4, ext. al. mag. mt'thx.) of either three or four fibres, there being
individual variations. These fibres, which are very short, are found in
the lateral ventral portion of the metathorax, immediately above the base
of the larval leg, and extend nearly vertically. They probably have
some connection with the leg movements. These fibres elongate very
rapidly in the pupa (Plate 2, Figure 5, ext. al. mag. mt'thx.) and fuse
completely at their dorsal ends. During this growth, the dorsal end
shifts its position very noticeably, so that its attachment comes to lie in
the antero-lateral portion of the somite. By the time the imaginal state
(Plate 4, Figure 9, ext. al. mag. mftlix.) is attained, the muscle has in-
creased still more in size, and its fibres are so fused as to show but two
parts, which are separated at the ventral end only. It extends from
what is known as the large cupule — a tendon formed during pupal
life — backward and downward to the middle of the lateral expanse of
the metasternum. Tlie posterior portion of the muscle at its ventral end
attaches to a chitiuous ingrowth from the metasternum.

This muscle in Colymbetes is also very plainly divided into anterior
and posterior portions, the division being much plainer than Luks has
shown for Dytiscus. The division into two parts is not as apparent in
Synchroa and Bruchus as in Thymalus.

Extensor alae parvus metathoracis of Luks.
(Troisieme fiechisseur de la handle et extenseur posterieur de Vaile of
Straus-Diirckheim ; postaxillaire of Amans.)

Besides acting as an extensor of the wing in the imago, this muscle is
also the third fiexor of the metathoracic coxa. It is composed in the la^'va

BREED: METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 327

(Plate 1, Figure 3, exi. al.pa. mfthx.) of two fibres, which extend from
the posterior lateral surface of the raetathorax ventrally, and a little
toward the median plane to attach to the posterior edge of the
leg, very close to the attachment of the second flexor of the coxa. At
this stage its only function is that of flexor of the coxa. In i\iQ pupa
(Plate 2, Figure 5, ext. al. pa. mt'thx.) a fusion of the two fibres takes
place, and a very considerable shifting of position. The attachments of
this muscle in the imago (Plate 4, Figure 9, ext. al.pa. mt'thx.) are, dor-
sally, to the small cupule, which is placed immediately posterior to the
large cupule, and, ventrally, to the ventral surflice of the coxa just lateral
to the insertion of the second flexor of the coxa.

ReJaxator extensoris alae of Luks.
{Releveur de la grande cupule of Straus-Durckheim ; dorso-preaxillaire

of Amans.)

There is some doubt as to the larval condition of this muscle and the
few muscles next described ; this is due principally to their small size.
During pupal life, this muscle and the relaxator alae metathoracis are so
closely united as to be indistinguishable. In fact, there is little more
than a mass of tissue containing remains of larval muscle and having
about the position indicated in Figure 5 (Plate 2) by rlx. ext. al. and
rlx. al. mt'thx. Out of this mass are differentiated the two muscles men-
tioned above. In the imago the relaxator extensoris alae (Plate 4, Figure 9,
7'lx. ext. al.) is inserted on the edge of the large cupule to which the
extensor alae magnus metathoracis is attached. Its origin lies almost
directly dorsal to this point on the wing-bearing apophysis.

Relaxator alae metathoracis of Luks.
(Relaxateur de Vaile of Straus-Durckheim ; muscles du tampon of Amans.)

The attachments of this muscle in the imago (Plate 4, Figure 9, rlx.
al. mfthx.) are as follows. Its origin is on a small cupule placed near
the dorsal attachment of the musculus lateralis metathoracis anterior
(Plate 5, Figure 11, I. mt'thx. a.), from which it extends laterally, and
somewhat ventrally, to attach on the base of the wing.

As to the larval condition of the two muscles last described (rlx. exf.
aL, rl.c. al. mt'thx.), it seems probable that they are derived from three
fibres. It is possible, and even probable, that the two fibres so marked
(Plate 1, Figure 4, rlx. ext. al. ?) give rise to the relaxator extensoris
alae of the imago, and that the other fibre (Plate 1, Figure 4, rlx. al.

328 bulletin: museum of comparative zoology.

mt'thx. ?) gives rise to the relaxator alae metathoracis. If this "be so,
then the two muscles probably remain distinct throughout pupal life.
Certainly the positions of these larval fibres correspond very closely -with
the positions of the two muscles in the imago, and the identification
.seems the more probable when one takes into account the shifting in
positions of the extensor alae magnus metathoracis and other muscles
which attach near by. There is no doubt but that both of the muscles
under discussion are metamorphosed larval muscles, not muscles newly
formed in the pupa.

Flexor alae metathoracis primus et secundus.
(^Flechisseur de Vaile of Straus-Dlirckheim ; entopleuro-dorsal of Amans.)

Larva (Plate 1, Figure 4, fix. al. jnfthx. 1, 2). These flexors are
found in the larva as single fibres, running nearly parallel with each
other. They extend almost vertically from the dorso-lateral portion
of the somite to the ventro-lateral portion. The positions in the pw/>a
(Plate 2, Figure 5, fix. al. mfthx. 1, 2) are changed but slightly. In
the imago (Plate 4, Figure 9, fix. al. mfthx. l, 2), they extend from the
posterior portion of the base of the wing, ventrally and posteriorly, to
attach to the dorsal edge of the episternum.

Flexor alae metathoracis tertius.
(Synonymy as in primus and secundus.)

The facts concerning this muscle are much the same as those con-
cerning the relaxator extensoris alae and the relaxator alae metathoracis.
In the larva (Plate 1, Figure 3, fix. al. mVthx. 3 ?) there are usually
three fibres, sometimes two as shown in the figure. These fibres lie
parallel and close together, extending from the antero-lateral portion
of the metathorax to the antero-ventro-lateral portion, and show all the
evidences of metamorphosis in older larva. In the young pupa it is very
difficult to trace their development, but it is probable that they form
the mass of tissue shown in Figure 5, fix. al. mfthx. 3 (Plate 2). From
this mass of tissue is developed the third flexor of the wing in the
imago (Plate 4, Figure 9, fi.c. al. mfthx. 3). This muscle in its adult
condition is composed of three parts, which attach by a common tendon
on the anterior part of the base of the wing.

These flexors are so different from those described by Straus-Dilrck-
heim for Melolontha that their homologies are somewhat uncertain. The
third flexor in Thymalus is probably homologous with the three flexors

breed: metamorphosis of the muscles of a beetle. 329

of Melolontha, though possibly the three flexors of Thymalus are re-
spectively homologous with the three of Melolontha.

Luks states that he is uuable to find more than one flexor of the wing
in Dytiscus. As a matter of fact, the muscle which he has described as
the flexor of the wing is the fourth flexor of the posterior coxa. This
may be seen in his own figure (Tafel 23, Figur 12, fa.), where this
muscle is shown attaching to the lateral edge of the posterior coxa, and
occupying a position exactly similar to that of the fourth flexor of the coxa
as shown by Straus-Diirckheim and myself (Plate 4, Figure 9, fix. cox.
mt'thx. 4). This conclusion is corroborated by the dissection of Colym-
betes, where not only the fourth flexor of the coxa, but also the three
flexors of the wing are found occupying their usual positions. Inas-
much as the muscles of Colymbetes are almost exactly identical with
those of Dytiscus, it is certain that Luks overlooked the flexors entirely.

The conditions in Synchroa and Bruchus are much like those in Thy-
malus, except that in both of these beetles the second and third flexors
are fused into a single muscle. The third flexor is divided in both cases
into three parts, which attach on the base of the wing by a common
tendon.

The muscles described thus far are all muscles of flight, acting either
directly or indirectly on the wing. Those now following have very little,
if any, action on flight.

Musculus mesofurcae dorsalis.
(^Ahaisseur du diaphragme of Straus-Diirckheim ; musculus furcae dor-
salis of Luks.)

In the larva (Plate 1, Figure 3, ms'fur. d.), this is one of the muscles
which extend dorso-ventrally along the suture between the meso- and
metathorax. It attaches laterally, and extends to a ventro-lateral posi-
tion. The position of this muscle changes very little during pupal life
(Plate 3, Figure 7, vis'/ur. d.), but there are ingrowths of hypodermis at
both dorsal and ventral attachments. The dorsal ingrowth forms in the
imago the inferior process of the mesophragma {pre. if. ms'phg.), to the
tip of which this muscle (Plate 5, Figure 11, ms'fur. d.) attaches. The
ventral attachment is to the ventral ingrowth which forms the meso-
furca (jns'fur.) in the imago.

Musculus lateralis processus inferioris mesopTiragmatis.

In the larva, this muscle (Plate 1, Figures 3, l.prc.if.ms'phg.) is a
simple fibre, whose dorsal end attaches to the suture between the meso-

330 bulletin: museum of comparative zoology.

and metathorax in a dorso-lateral position, and whose ventral attachment
is on the autero-ventro-lateral surface of the metathorax. In the ^jw^va
this fibre (Plate 3, Figure 7, I. prc.if. ms^pluj.) shortens very consider-
ably, but no more than would be expected from the growth of the
extensor alae magnus metathoracis during the same period. The dorsal
attachment of the extensor is just ventral to the ventral end of this
muscle, so that dorsal growth of the former, necessarily means a shorten-
ing of the latter. The attachments of this muscle in the imago (Plate 5,
Figure 11, I. pre. if. ms'phfj.) are, medianly, to the inferior process of the
mesophragma, and, laterally, just posterior to the metathoracic stigma.

This muscle was not found by Straus-Dilrckheim in Melolontha, nor
by Luks in Dytiscus, nor was I able to find it in Colymbetes. It may
be present in some of these beetles, however, as it might easily be over-
looked in the dissections, on account of its small size. It is present
in both Synchroa and Bruchus, occupying the same position as in
Thymalus.

Musculus lateralis mesofurcae.

In the larva (Plate 1, Figure 4, l.ms^fur.') this muscle is found as
two nearly parallel fibres which extend from the antero-ventro-lateral
portion of the metathorax, anteriorly and ventrally, to the suture be-
tween the meso- and metathorax near the ventral attachment of the
musculus mesofurcae dorsalis. The two fibres fuse so as to be indis-
tinguishable in the pupa (Plate 3, Figure 7, I. nisfur.), maintaining, how-
ever, a closely similar position. The attachments in the imago (Plate 5,
Figure 11, I. msfur.) are, medianly, to the tip of the mesofurca (7m'/ur.),
and, laterally, just posterior and ventral to the metathoracic stigma
(sfg. mfthx.).

This muscle is not mentioned by either Straus-DUrckheira or Luks.
It also did not show in my dissection of Colymbetes, nor could it be
found in the sections of Bruchus. It is present in Synchroa, however,
extending from the mesofurca to the lateral wall of the metathorax as in
Thymalus.

Depressor tevgi.
(Abaisseur du iergum of Straus-DUrckheim.)

In the larva the depressor tergi (Plate 1, Figure 3, dep. trg.) is a sin-
gle fibre, extending dorso-ventrally along the suture between the meta-
thorax and the first abdominal somite. In the young pupa (Plate 3,
Figure 7, dep. trg.) there is a very evident bend both in this muscle and

breed: metamorphosis of the muscles of a beetle. 331

In flexor processus postero-lateralis metafurcae, the muscle next to be
described. This bend is caused by the presence of a large trachea, a
branch from the trunk arising at the first abdominal stigma. The tra-
chea lies in such a position that tlie muscles are bent around it when
their ventral attachments shift posteriorly. In older pupae the relations
of these parts become readjusted so that there is no bend in the muscles.
The metafurca commences to form very early in the pupa, and by its
ingrowth carries in the ventral attachments of this muscle, together with
that of several other muscles. On account of the ingrowth, this muscle
is shortened in later pupal life until, in the imago (Plate 5, Figure 11,
dep. trg.), it has about one third of its original length. The attach-
ments are, dorsally, to the suture between metathorax and abdomen, the
same as in the larva, and, ventrally, to the tip of the posterior lateral
horn of the metafurca (mffur. 2).

The depressor of the tergum is frequently fused with the muscle next
to be described, this being the case in Bruchus and Colymbetes. This
condition is probably found in Dytiscus, though Luks does not figure
either of the muscles.

Flexor processus postero-lateralis metafurcae.
(Flechisseur lateral de Vapophyse episternale posterieure of Straus-

DUrckheira.)

This muscle in the larva (Plate 1, Figure 3, Jlx. pre. p-l. mffur.) has
a position exactly parallel with that of the muscle last described, but is
shorter, lying more laterally. Dnvrng pupal life (Plate 3, Figure 7,Jfx.
pre. p-l. mffur.) there is an ingrowth of the hypodermis at both dorsal
and ventral attachments, so that in the imago (Plate 5, Figure l\, fix.
pre. p-l. mffur.) this muscle lies in a horizontal position instead of a
vertical one as formerly. This change in position is in such a direction
that the fo'" ner ventral end lies mediad. The process formed ventrally
is the metafurca, this muscle being attached to its posterior lateral horn
(mffur. 2). The lateral attachment is to the inferior process of the meta-
phragma (jp/-c. if. mfplig.).

The flexor of the posterior lateral horn of the metafurca was found by
Straus-Durckheim, but not by Luks. It is certain that it is present in
Dytiscus, however, since it is present in Colymbetes, extending from the
posterior lateral horn of the metafurca to the inferior part of the meta-
phragma, there being no inferior process. In Colymbetes, as also in
Bruchus, the depressor tergi and this muscle are fused, the development

VOL. XL. — NO. 7 2

332 bulletin: museum of comparative zoology.

of their attachments being such that they lie parallel and close together.
The conditions iu Synchroa and Melolontha agree with those in
Thymalus.

Musculus episternaliii.
{Muscle expirateur dans le metatliorax of Straus-Durckheira ; Expira-

tionsmuskel of Luks.)

This is a muscle of which no trace can be found in the larva or young
pupa. Therefore it is probably a muscle of new formation in the pupa.
In the imago (Plate 4, Figure 9, e'stn.') it is found just beneath the
episternum. Its origin is near the dorsal edge of the episternum, from
which it extends obliquely downward and mediad to attach to the ven-
tral edge of the episternum. It was described and figured by Straus-
Durckheim ('28), who ascribed to it the function of an expiratory muscle.
In his own words (p. 164), "It is only by conjecture that I regard this
muscle as acting in respiration, not being able to ascribe to it any other
function." Also (p. 165), "This muscle, being placed between two
pieces of the case which forms the thorax, does not appear to act either
in flight or in the movements of the legs, and, as it compresses the tho-
racic cavity, and so necessarily compresses the ti-achea, I believe it ought
to be regarded as an expiratory muscle." Luks adopts these views with-
out comment.

That this is not the function in Thymalus, is shown by a cross section
of the thorax in the region of this muscle (Plate 6, Figure 13). Here
the elytron (ely.) is shown hooked into a fold {21U.) on the episternum
by means of a ridge (loph.) on the inflexed edge of the elytron. The
elytron after being hooked into the fold is held firmly in place by the
interlocking of the teeth along the inner surfoce of the elytron with
those on the outer sui'face of the metathorax at the place indicated by a
star (i^) and by the teeth on the inner side of the fold {]'U.). This
fold extends antero-posteriorly along the episternum as far as the muscle
reaches. Tlie contraction of the muscle releases the elytra by bringing
the cuticula into the position shown by the dotted lines. This muscle
is aided in its action by a pull on the bases of the elytra by their exten-
sor muscles. The contraction of this muscle would be necessary in re-
placing the elytra, as it would depress the fold for the reception of the
ridge.

The episternal muscle is present in all of the beetles examined, as also
in Melolontha and Dytiscus. Yet the elytra of some of these species do
not lock into a fold when closed, so that in such cases the muscle is
probably functionless.

BREED : METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 333

The remaining muscles of the lateral dorso-ventral group are all leg
muscles, either flexors or extensors. The homologies with the muscles
of Dytiscus are not all entirely certain, because the leg muscles of Dy-
tiscus are so different from those of Melolontha and Thymalus, that
the homologies are not always evident.

Flexor coxae metafhoracis primus.
(Premier flecMsseur de la Tianche of Straus-Diirckheim ; extensor trochan-

teris metathoracis of Luks.)

This muscle is found in the larva (Plate 1, Figure 4, Jlx. cox. mftJix. l)
as one fibre, whose origin is on the ventral portion of the suture between
the metathorax and the abdomen, and whose insertion is on the outside
surface of the leg on a portion which later forms the coxa of the adult.
In the piqm (Plate 3, Figure 7, fix. cox. mftlix. 1) its position is changed
greatly by the formation of the metafurca, and the shifting of the leg
posteriorly. The origin of this muscle in the imago (Plate 5, Figure 11,
fix. cox. mt'thx. 1) is on the posterior part of the median lamina of the
metafurca {mffur. 4), and its insertion, on the anterior ventral edge of
the coxa about one third of the distance from the trochanter to the
lateral edge of the coxa.

For an account of Flexor coxae metathoracis secundus, see page 325, and
for an account of Flexor coxae metathoracis tertius, see page 326.

Flexor coxae metathoracis quattuor.
(Quatrieme fiechisseur de la hanche of Straus-Durckheim ; Jlexor alae

metathoracis of Luks.)
This is the second muscle of the imaginal metathorax which has not
been found in the larva. It is found in younger pupae than is the first
muscle (musculus episternalis), but it is probably a muscle of new forma-
tion in the pujm (Plate 2, Figure 5, fix. cox. mt'thx. 4). In the imago
(Plate 4, Figure ^, fix. cox. mfthx. 4) it takes its origin near the middle
of the dorsal side of the episternum, and, extending caudad and a little
ventrad, is inserted on the extreme anterior lateral edge of the coxa.
This is the muscle which Luks has incorrectly described for Dytiscus as
the flexor of the wing.

Flexor coxae metathoracis quintus.
{Cinquieme fiechisseur de la hanche of Straus-Diirckheim ; musculiis

furcae dorsalis of Luks.)
The fifth metathoracic flexor of the coxa is found in the larva (Plate 1,
Figure 4, fix. cox. mt'thx. 5) as a single fibre, extending from the latero-

334 bulletin: museum of comparative zoology.

ventral portion of the suture between the metathorax and abdomen to
the postero-lateral portion of the metathorax. In the jtupa (Plate 3,
Figure 7, Jix. cox. nd'thx. 5) this muscle has changed its position con-
siderably, extending more nearly laterad from the newly forming nieta-
furca. Its origin in the imago (Plato 5, Figure W, fix. cox. mt'thx. 5) is
on the anterior portion of the median lamina of the metafurca (mffur. 4).
From this it extends laterad and a little caudad, attaching by a long ten-
don to the suture between the metasternum and coxa, a little dorsal to
the insertion of the muscle last described.

Extensor coxae metathoracw primus.

{Premier extenseur de la hancke of Straus-Diirckheim ; extensor tro-

chanteris metathoracis of Luks.)

This extensor is composed of a single fibre in the larva (Plate 1,
Figure 4, ext. cox. mt'thx. 1), whose origin is on tlio ventral portion of
the suture between the metathorax and abdomen ; its insertion is on the
postero-lateral surface of the upper part of the larval leg. In the pupa
(Plate 3, Figure 7, ejii. cox. mt'thx. 1) its position has changed to some
extent, as a result of the changes in position of both its attachments. Its
origin in the imago (Plate 5, Figure 11, ext. cox. mt'thx. i) is on the
posterior face of the lateral wing of the metafurca (mt'fur. 3), from which
it extends ventrad and caudad to its insertion on the posterior median
surface of the coxa.

Extensor coxae metathoracis secundus.
(^Second extenseur de la hanche of Straus-Diirckheim ; extensor trochanteris

metathoracis of Luks.)

This muscle properly belongs to the first abdominal somite, but since
it acts as an extensor of the coxa in some beetles, it is spoken of here
among the muscles of the metathoracic leg. In the larva this muscle
forms part of the ventral antero-posterior group of muscles of the first
abdominal somite. During pupal life (Plate 3, Figure 7, ext. cox.
mt'thx. 2') there is a great change in this group of muscles. Some de-
generate, while the remainder metamorphose, to form this so-called
extensor of the coxa, which in the imago (Plate 5, Figure 11, ext. cox.
mVtlix. 2) is divided into two parts. The origin of these muscles is on
the posterior side of the posterior lateral horn of the metafurca (int'fur. Si)
and their insertion, on the boundary between the first and second
abdominal somites, very close to the median face of the metacoxa.

BREED : METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 335

At first sight it seems impossible that larval muscles, extending antero-
posteriad the full length of the first abdominal somite, should be trans-
formed into extensors of the coxa of the imago. In Thymalus, indeed,
these muscles have no such function in the imago, but in forms in which
the ventral plate of the first abdominal somite becomes completely
eliminated, it does not seem improbable that such a shifting of position
takes place. In Thymalus their function is that of ventral protractors
of the second abdominal somite.

Extensor coxae metathoracis tertius of Luks.
(Troisieme extenseur de la handle of Straus-Dtirckheim.)

The third extensor of the coxa is present in the larva (Plate 1, Figure 4,
ext. cox. mt'thx. 3) as two fibres extending dorso-ventrally from the dorso-
lateral part of the metathorax to the ventro-lateral part. In the jnipa
(Plate 2, Figure 5, ext. cox. mfthx. 3) the ventral attachment is shifted
posteriorly, so that the muscle extends obliquely from an antero-dorsal
to a postero-ventral position. The origin of this muscle in the imago
(Plate 4, Figure 9, ext. cox. mfthx. 3) is on the lateral edge of the
scutum and the insertion, on the dorso-median edge of the coxa.

Extensor trochanteris metathoracis of Luks.
{Extenseur du trochanter of Straus-Dtirckheim.)

The extensor of the trochanter in the imago is divided into two parts, —
the long and the short heads. In the reconstruction only the pupal and
imaginal conditions of the long head have been determined. In the pupa
a muscle (Plate 3, Figure 7, e.xt. trchn. 7nt'thx.) is found which shows
histologically that it is a metamorphosed larval fibre ; this forms the long
head of the extensor trochanteris in the imago (Plate 3, Figure 7, ext.
trchn. mfthx.). Its origin is on the posterior face of the lateral wing of
the metafurca (mffur. <?), very close to the origin of the first extensor of
the coxa. Its insertion is on an apodeme which projects from the median
side of the trochanter. The short head of this muscle attaches to the
same apodeme, and would show in the same figures as the long head, if it
had been reconstructed.

The flexor trochanteris metathoracis would likewise have been visible in
Figure 7 (Plate 3) and Figure 11 (Plate 5), if it had been reconstructed.

The remainder of the imaginal leg muscles are metamorphosed larval
muscles. The details of their changes have not been studied out.

This ends the description of the changes of the lateral dorso-ventral

836 bulletin: museum of comparative zoology.-

group of muscles, with the exception of three larval muscles which
degenerate during pupal life. Two of these muscles (Plate 1, Figure 3,
A, fj.) extend dorso-ventrally along the suture hetween the meso- and
metathorax. They do not disappear for some time, and are shown in
the figure of the pupa (Plate 3, Figure 7, X ; Plate 2, Figure 5, /a). The
third of these degenerating muscles (Plate 1, Figures 3, 4, v) extends the
full length of the metathorax. It lies in the lateral part of the somite
extending obliquely from antero-dorsal to postero-ventral. This muscle
is one of the first to disappear, and so is not shown in the figure of the
pupa.

(3) TJie ventral anfero-posterior group consists in the larva of eight
muscles, five of which fuse to form the single representative of this group
in the imago. This muscle is shown in the reconstruction drawings
only in the pupa (Plate 3, Figure 7, rfr. ms'thx. if.) and in the imago
(Plate 5, Figure 11, rtr.ms'thx.if.); in both the view is from the left
side of the insect. Cross sections of this group {rtr. ms'thx. if., 0, i, k)
are shown in Figure 10 (Plate 4) for the larva, and in Figure 12
(Plate 5) for the young pupa.

Retractor mesothoracis inferior of Luks.
(Pretradeur de Vapophyse episternali posterieure of Straus-Dilrckheim.)

The five larval muscles (Plate 4, Figure 10, rtr. ms'thx. if), all of
which extend the full length of the somite, become in the pupa (Plate 5,
Figure 7, rtr. ms'thx. if.) closely approximated to form a single muscle.
This, by the ingrowth of the meso- and metafurcae, comes to have in the
imago the position shown in Figure 11, rtr. ms'thx. if (Plate 5). Here
its origin is seen to be on the anterior lateral horn of the metafurca
(inffur. 1) and its insertion on the mesofurca (msfur.).

The three remaining larval muscles of this group (0, i, k), degenerate
during pupal life (Figure 10, larva; Figure 12, pupa). These muscles
extend the full length of the somite, form the deeper layer of this group,
and present in general the same characteristics as the degenerating
muscles of the dorsal group.

Summing up the changes which take place in the muscles of the meta-
thorax during pupal life, we find :

a. That not a single larval muscle persists unaltered from larva to
imago.

h. That the great majority of the larval muscles metamorphose into
adult muscles, and

breed: metamorphosis of the muscles of a beetle. 337

c. That thirteen of the larval muscles degenerate, these being in
general dorso-ventral intersegmental muscles and the inner layer of the
antero-posterior muscles. Two of the imaginal muscles (musculus
episternalis and flexor coxae metathoracis quattuor) are muscles of new
formation in the pupa.

6. Mksothorax.

In the mesothorax the muscles are arranged similarly to those ot the
metathorax. For tlie dorsal group of antero-posterior muscles, the figures
of the similar group of the metathorax (Plate 1, Figures 1, 2) would
serve with only minor changes. It is very interesting to find that the
serial homology is practically complete even to the changes which take
place during pupal life. The three muscles which in the metathorax
metamorphose into musculus metanoti have counterparts in this somite
which metamorphose into musculus mesonoti. The same relations hold
true between musculus lateralis metanoti and musculus lateralis mesonoti
(retracteur de Vaile of Straus-Diirckheim). The remaining mesothoracic
muscles of this group degenerate during pupal life, as do their counter-
parts of the metathorax.

The close similarity of the muscles of the lateral dorso-ventral groups
in the two somites is likewise remarkable. A careful comparison be-
tween these muscles in a series of frontal sections of a resting larva
showed only the following slight anatomical differences. The muscle in
the mesothorax corresponding to the third extensor coxae metathoracis
(Plate 1, Figure 4, ext. cox. mfthx. 3) was composed of three fibres in-
stead of two, and the muscle corresponding to the oblique muscle v
(Figure 4) was divided dorsally into two parts. The changes of the
mesothoracic muscles of this group do not correspond exactly to the
changes of their counterparts in the metathorax. A greater number of
muscles degenerate in the mesothorax than in the metathorax. The
additional muscles of this somite which have been noticed to degenerate
are the musculus lateralis mesothoracis and the second flexor of the coxa.
It is evident from the muscles which are present in the imago that a few
others degenerate also, but their identity has not been established.
These additional degenerating muscles are such as would function in the
imago as muscles of flight, if the elytra were used as organs of flight.

In the ventral antero-posterior group, only seven muscles are found
in the larva ; three of these degenerate, while the remaining four meta-
morphose to form the retractor prothoracis inferior. The only difference
between the metathorax and the mesothorax in this case is, that in the
latter there are only four metamorphosing muscles, whereas, in the

338 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

former, there are five. The outline of the retractor of the prothorax is
shown by the dotted lines in Figure 11, rtr.prothx.if. (Plate 5). This
shows the iraaginal position of the muscle, its origin being on the
mesofurca and its insertion on the antefurca.

c. Prothorax.

The serial homology between the muscles of this somite and those of
raeso- and metathorax is not so marked as between those just compared.
Yet, in general, muscles in similar positions undergo similar changes.
The great majority of the larval muscles of the prothorax metamorphose
into imaginal muscles, but a number degenerate. None of the larval
muscles pass unchanged into the adult.

d. Head.

The muscles of the head of the larva are probably all metamorphosed
into imaginal muscles, for there is no evidence that muscles degenerate,
nor do any of the muscles remain unchanged. One point in regard to
the adductor of the mandible may be of interest. In the larva this
muscle is composed of about fifty fibres, whereas in the imago the same
muscle has from two to three Inindred fibres of smaller calibre, which
have been formed by the longitudinal splitting of the larval fibres.

e. Abdomen.

The abdomen is the only region of the body where any muscle remains
unaltered from the larva to the imago. The abdominal muscles which
have this fate occupy in general positions homologous with those of the
muscles of the thoracic region which undergo degeneration. They are
the inner muscles of the dorso-ventral intersegmental muscles and the
inner layer of the autero-posterior muscles. Most of the remaining
larval muscles in the abdomen metamorphose into imaginal muscles ;
there are a few, however, which degenerate. The latter are found in the
somites in which the greatest changes in external form take place during
pupal life, i. e., the first and last abdominal somites. No muscles newly
formed in the pupa have been observed, though some may be present.
Such are quite probably to be found in connection with the sexual
organs, — ovipositors, etc.

Two of the metamorphosed muscles of the first abdominal somite are
shown at ah, in Figure 9 (Plate 4). The metamorphosis of extensor
coxae metathoracis secundus from muscles of the first abdominal somite
has already been described (page 334).

bkeed: metamorphosis of the muscles of a beetle. 339

f. Appendages.

The imaginal appendicular muscles of Thymalus are apparently all
metamorphosed larval muscles. No evidence of the degeneration of
larval muscles nor of the new formation of imaginal muscles in the pupa
has been observed. The changes of these muscles in some beetles are quite
different from those of Thymalus. This is especially true of the forms
with legless grubs. In these, the imaginal leg muscles are of new
formation in the pupa.

4. Discussion of Results.

Summing up the anatomical changes which the muscles of Thymalus
undergo during pupal life, we find that :

1. The only larval muscles which remain unchanged in both position
and histological structure are found in the abdominal region, this being
the region of least change in external form during pupal life. This
persistence of the larval muscles might have been inferred from the fact
that the pupa retains throughout life the power to roll itself about by
means of the movements of the abdominal somites on each other.

2. However, only about half of the larval muscles of the abdomen
remain unchanged, those of the more peripheral layers undergoing a
metamorphosis into imaginal muscles. Most of the muscles of the lar-
val thorax and all of the muscles of the head and appendages metamor-
phose into imaginal muscles.

3. The larval muscles which degenerate are found in the thorax and the
first and last abdominal somites. They occupy in nearly every case
positions similar to the positions of the muscles of the abdomen which
persist unaltered by the metamorphosis. , Exceptions to this statement
have been noted in the mesothorax, where there is a degeneration of
dorso-ventral muscles other than intersegmental ones.

4. Probably two new metathoracic muscles are formed during pupal
life, one being a iiexor of the metathoracic coxa and the other, the
muscle which operates the fold of the episternum into which the elytra
catch when closed.

The most radical changes in the musculature are found in the thoracic
region. This is to be expected as the imaginal thorax differs greatly
from the larval in both form and function. The least radical changes
are found in those somites of the abdomen whose larval condition most
resembles the imaginal. The serial homology between the degenerating
muscles of the thoracic region and the persistent larval muscles of the

340 bulletin: museum of comparative zoology.

abdominal region is a curious fact of which no explanation can be
offered.

The direct descent of most of the imaginal muscles from larval
muscles, which has here been shown, will help in solving some of the
difficult problems of the comparative myology of insects, — a subject
about which little is known. Hitherto the only basis of comparison be-
tween the muscles of metabolic and ametabolic insects, or between the
muscles of different metabolic insects, has been the origin and insertion
of the muscles in the imago. No attention has been paid to the larval
musculature, since this has been generally supposed to have no connec-
tion with the imaginal. But, as this paper shows, there is a close
connection between the larval and imaginal musculature in Coleoptera,
and a similar connection will probably be found to exist in most of
the metabolic insects. With this relation as a basis for comparisons,
the simpler conditions — the larval — may be used in establishing the
homologies instead of the more complex, — the imaginal. And this, not
only for comparison between different metabolic insects, but also between
metabolic and ametabolic insects.

A word ouglit, perhaps, to be added to meet the possible criticism,
that in some of the muscles there are such radical differences between
the conditions in the stages figured that the identity of the various
muscles in successive stages is doubtful. In answer to this, it may be
stated that not only the stages figured, but also several intermediate
stages, have been studied. The dorso-ventral metathoracic muscles have
been identified with the help of camera sketches in four individuals in
stages of development intermediate between the stages used in making
the reconstructions. Numerous other animals have been used in which a
part of these muscles have been identified. The antero-posterior mus-
cles are much simpler, and have been identified in as many as twenty
cases.

Part II. — Histology.

A. Historical Survey.

This review of researches on the histological changes of the muscles
during the metamorphoses of insects has been arranged in four parts
corresponding to the four principal groups of holometabolic insects. Such
an arrangement is used rather than a simple chronological one, because so
little comparative work has been done that the mutual relations of the
changes of the various groups are not entirely understood. The studies

BREED: METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 341

on Coleoptera will be spoken of first, and in greater detail than those on
the other groups, as they are of more interest in connection with this
paper. None of the researches on Coleoptera had, as a main object, the
study of the muscular changes, and most of the investigators speak of
them only incidentally.

Coleoptera. The first paper in chronologcial order is that of Rengel
(*96), who describes the changes which occur in the midiutestine of
Tenebrio during metamorphosis, including a description of the changes
of the intestinal muscles. The muscle layer of the larval intestine de-
generates into a structureless protoplasmic zone in the late larva and
early pupa. In this protoplasmic zone the individual muscle fibres can no
longer be distinguished, though the nuclei of the larval fibres remain
unaltered. No phagocytes (" Korachenkugelu " of Weismann, '64) are
present, this degeneration being entirely chemical. The intestinal mus-
cles of the imago develop in this protoplasmic zone, but the exact
method of their formation is somewhat in doubt. Apparently, part or
all of the nuclei of the larval muscles remain and form the new muscles
out of the material in which they are embedded.

De Bruyne ('97), speaking of phagocytosis in the development of in-
vertebrates, treats of the changes in the hypodermal muscles of Tenebrio
during metamorphosis. He finds a degeneration of the larval muscles,
which begins with a chemical alteration of the muscle substance. The
muscles soon break into fragments, which later are engulfed in leucocytes
acting as phagocytes, thereby forming " Kornchenkugeln." These mus-
cle fragments undergo fatty degeneration in the phagocytes, each becom-
ing surrounded by a vacuole. Tlie vacuoles with their contents fuse
with one another until each phagocyte contains a few large vacuoles
with correspondingly large fat globules .inside. These fat globules are
then dispersed to the growing tissues, leaving the large vacuoles in the
cytoplasm of the phagocyte. This is the beginning of degeneration for
many of the phagocytes.

Krtlger ('98), describing the development of the wings in beetles
(Tenebrio, Lema), states that he finds two larval muscles at the base of
the wing (the flexor alae metathoracis, judging from his figures) which
metamorphose into wing muscles of the imago. He concludes from this
that the wing muscles of the adult are metamorphosed larval muscles.
He also finds in the blood what he calls " Weismannsche Korncheu-
zellen."

In an article on the anatomy and metamorphosis of the intestinal
canal of Auobium, Karawaiew (*99) states that there is no phagocytosis

342 bulletin: museum of comparative zoology.

of the muscles of the larva. The changes of the muscles are similar to
those in Lasius, as described by himself ('98).

Deegener ( :00) describes the metamorphosis of the intestine in
Hydrophilus. His observations on the changes of the intestinal mus-
culature differ in many fundamental points from those of Rengel on
Tenebrio. He finds typical phagocytosis, sucli as Kowalevsky ('87) and
Van Rees ('88) found in Muscidae. The phagocytes make their appear-
ance in the old larvae, engulfing both sarcolytes (muscle fragments) and
muscle nuclei. They then do not become scattered through the body,
but degenerate — in larger part at least — in the lumen of the pupal
intestine. Spindle cells whose origin is uncertain, but which cannot have
been derived from the nuclei of the larval muscle, appear in the old
larvae. In the muscle layer of the pupa, the changes are difficult to
follow on account of the close intermingling of diverse elements. The
spindle cells give rise to the imaginal musculature, but he does not
describe the process clearly, nor give figures.

In tlie midintestinal region, there are so few phagocytes that they are
not sufficient to entirely account for the disintegration of the muscles, so
that, in this case, there must be chemical degeneration as well. The
source of the imaginal musculature in this region is doubtful, as no
spindle cells could be distinguished. Deegener thinks, however, that
spindle cells are present in the closely intermingled elements of the
muscle layer, and that the imaginal muscles are derived from them.
Berlese (:00, :01, :02^) speaks of the histolysis and histogenesis of the
hypodermal muscles in Aphodius and other Coleoptera. He states
that the larval muscles are dissolved, but that the nuclei resist dissolu-
tion. These nuclei emigrate from the degenerating larval muscles,
acquiring cytoplasm and a cell membrane, and thus become " sarcocytes."
By division, the "sarcocytes" form spindle-shaped "myocytes," which
give rise to the imaginal muscles by fusing in rows to form muscle fibres.
The " myocytes " at one stage closely resemble leucocytes, so that there is
a possibility of confusing them ; but Berlese, reasoning from his similar
studies on Muscidae, feels confident that their origin is, as has just been
stated, from the nuclei of the degenerating'larval fibres.

Needham (:00) states that in Mononychus vulpeculis the fat cells of
the abdominal region, after getting rid of their surplus food supply, be-
come associated with the new muscle rudiments, and that their nuclei
become nuclei of the developing muscle fibres.

Diptera. The most important of the investigations concerning the
postembryonic development of insects have been made on Diptera.

breed: METAMOKPHOSIS OF THE MUSCLES OF A BEETLE, 343

After the classical researches of Weismann ('62, '64, '66), the more
importaut of the earlier authors are Kiinckel d'Herculais ('72, '75),
Gauin ('76), and Viallaues ('81, '82). Later authors have shown that
the results of these papers on the histological clianges of the muscles
daring pupal life are not of great importance, so that they need not be
mentioned in detail here. The higher (cjclorraphic) and the lower
(orthorraphic) Diptera seem to present, together with other dilferences,
two distinct types of muscle degeneration, and so the papers on each
group are here reviewed separately.

a. Cydorrapha. Van Eees ('84, '88) and Kowalevsky ('85, '87) both
find in Calliphora that the larval muscles undergo phagocytosis. The
leucocytes penetrate the muscle fibres, which they break up into frag-
ments ; these, together with the muscle nuclei, are engulfed by the
leucocytes and digested. The leucocytes with their inclusions are the
" Kornchenkugeln " of Weismann ('64). Van Rees finds that three
pairs of muscles in the dorsal part of the mesothorax are exempt from
this fate, and that they metamorphose to form the indirect wing muscles
of the adult.

Lowne ('90-95) confirms the two preceding authors in regard to the
phagocytosis of the larval muscles, but denies the metamorphosis of the
three pairs of muscles of the mesothorax described by Van Rees. He
states that all of the imaginal muscles are newly formed in the pupa,
being produced from mesoderm cells which are derived from the imaginal
disks.

De Bruyne ('97) practically agrees Avith Van Rees and Kowalevsky,
except that he finds that the leucocytes are not the active agents in
breaking up the muscle substance into fragments, the muscle being
frequently broken up before the arrival of the leucocytes. He also finds
that some of the nuclei of the larval muscles are not immediately de-
stroyed. These, collecting a portion of the sarcoplasm of the fibre about
themselves, act as myoblastic phagocytes, engulfing and digesting the
muscle fragments. He calls this " autophagocytosis," to distinguish it
from ordinary or leucocytic phagocytosis.

The results of the studies of Noetzel ('98) accord with those of De
Bruyne in regard to the breaking up of the muscle before the arrival of
the leucocytes.

Berlese ('99, -.00, :00% :01, :02, :02*) diff'ers from the above authors in
many essential points. He states that there is no phagocytosis, the
ingestion of the sarcolytes and muscle nuclei by the leucocytes being for
the purpose of distributing those elements to all parts of the body. The

344 bulletin: museum of comparative zoology.

muscle nuclei are never digested by the leucocytes, but divide and form
cells — the " sarcocytes " — which give rise to " myocytes." The
" myocytes " then fuse with each other, either developing into imaginal
muscles or undergoing fatty degeneration to form the imaginal fat-body.

Vaney (:00), who studied Gastrophilus, describes the larval muscles
as undergoing, during pupal life, a phagocytosis accompanied by the
formation of " Kurnchenkugeln."

h. Orthorrapha. Hurst ('90) states that all of the imaginal muscles
are present in the young pupa of Culex.

Miall and Hammond ('92, :00) find in Chironomus cells which re-
semble "Kornchenkugeln," but these do not result from the phagocyto-
sis of the larval muscles. The larval muscles of the head and thorax
seem to waste away gradually and uniformly while undergoing for a long
time no external chaugo of form. Some of the larval muscles remain in
the adult.

Kellogg (:01) finds in Holorusia, with a generalized larval form, that
there is no phagocytosis. The larval muscles of the thorax undergo a
" selbstandige Degeneration " (Karawaiew, '98), while many new muscles
are added in the head and thorax during pupal life. In Blepharocera,
with a highly specialized larval form, he finds active pliagocytosis, but
apparently without the formation of " Kornchenkugeln."

Lepidoptera. In a paper on the changes of the muscles in Tinea,
Korotneff ('92) states that all of the imaginal muscles are to be regarded
as metamorphosed larval muscles. The resorption of the muscles takes
place as follows : the nuclei and sarcoplasm of each fibre accumulate on
one side, and finally become separated from the fibrillar substance by a
longitudinal splitting. The imaginal muscles originate from this de-
tached strand, which is composed of the undifferentiated sarcoplasm con-
taining the nuclei, whereas the strand which is composed of contractile
fibrillar substance undergoes a chemical degeneration in which the
leucocytes take no part.

De Bruyne ('97), in his study of Bombyx, finds that the initial cause
of the muscular destruction lies in the muscles themselves. There is
both autophagocytosis and leucocytic phagocytosis of the muscles, the
latter taking place only at a late stage in the destruction of the muscles.

Berlese (;00, :01, :02*) obtains in Lepidoptera results similar to those
which he found in beetles.

Perez (:00) states that he finds typical phagocytosis, and denies the
truth of Korotneff's observations. The results of these papers on Lepi-
doptera are apparently irreconcilable.

breed: metamorphosis of the muscles of a beetle. 345

Hymenojptera. The first, and one of the most important, of the re-
searches on Hymenoptera is that of Karawaiew ('97, '98,) on Lasius.
He finds that there are two kinds of nuclei in the muscle fibres of the
old larva, one larger than the other. During metamorphosis the larger
nuclei degenerate, while the small ones, which are imaginal myoblasts,
divide amitotically and after the fibrillar substance of the larval muscle
has been dissolved, form the imaginal muscles. The imaginal muscles
are, therefore, metamorphosed larval muscles, except in the case of the
appendicular muscles, which are of new formation in the pupa.

Terre ('99, :00, :00*) confirms most of Karawaiew's results. He adds,
among other new observations, that the two kinds of nuclei are present
in the muscles of larvae which had but just escaped from the egg.

Anglas ('99, '99*, :00, :01, :01% :02) and Perez ('99, :00) dispute
the observations of the two authors last cited, stating that there is an
invasion of the larval muscles by leucocytes. Perez speaks of this in-
vasion as the beginning of an active phagocytosis which destroys the
muscles. However, according to the statements of Anglas, the substance
of the muscles is digested by the secretions of the leucocytes without
any ingestion of solid particles. This is not true intracellular digestion
or phagocytosis, but, rather, an extracellular digestion, for which he pro-
poses the term " lyocytosis." There are no " Kornchenkugeln " formed,
a statement in which all of the authors concur. Anglas finds that this
lyocytosis totally destroys certain muscles (those of the pharynx, of the
anterior part of the thorax, of the posterior part of the abdomen, the rectal
spliincter, and the transverse muscles ) ; while in the thoracic and intes-
tinal muscles the nuclei of the larval muscles survive and give rise by
fragmentation to small nuclei. These in turn form the imaginal muscles
in the midst of the mass left from the destruction of tlie remainder of
the fibre. The abdominal muscles do not undergo so deep-seated a
metamorphosis, inasmuch as the leucocytes never invade their substance.
The imaginal muscles in this case likewise are derived from nuclei which
arise by the direct division of the larval nuclei. There are some muscles
of new formation in the pupa which are derived from indifferent mesoderm
cells.

The results of Berlese's ( :01, :02'') observations agree more with
those of Karawaiew and Terre than with those of Anglas and Perez.
According to Berlese, the imaginal myoblasts of Karawaiew are the same
as bis " sarcocytes," and are derived from the larval muscle nuclei by
direct division. These may remain in the place where they are formed
and give rise to " myocytes," which then develop into the imaginal muscles

346 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

(the metamorphosing muscles of Aiiglas), or they may emigrate and form
muscles elsewhere in the body (the degenerating muscles and the muscles
of new formatiou of Auglas).

No very important generalizations can be made from this review.
The subject has reached a stage where it is evident that the muscular
changes differ in the various groups of insects, and that not all of the
muscles of the same insect undergo the same changes. Yet the impor-
tance and significance of these differences are not known. Comparative
researches are therefore needed. Two of the investigators have already
attempted such researches, but both attempts are unfortunate. De
Bruyne's results, both his observations and his interpretations of the
phenomena observed, have already been shown by Berlese to bo untrust-
worthy. Berlese has given us an elaborate memoir full of interesting
observations, and as accurate as could be expected when the phenomena
observed are so complicated. His interpretations of these phenomena
are not so fortunate, however. Judging from my observations on Cole-
optera, as well as from personal observations on all of the groups of in-
sects which he has studied, and from the numerous authors whose
interpretations of phenomena lie has contradicted, his fundamental idea
of the formation of " sarcocytes " from the larval muscle nuclei, and the
development of imaginal "myocytes" from the "sarcocytes" is not true
in many cases, if at all. The reasons for this statement, as fiir as Cole-
optera are concerned, will be given in detail in discussing the results of
the present paper, while the results of my comparative studies on other
insects I hope to publish in the not far distant future. The fundamental
correctness of the interpretations of the present paper, as contrasted with
those of Berlese, is indicated by the fact that they are in complete accord
with the statements of three (liengel, Krliger, Karawaiew) of the seven
authors who have previously mentioned these changes, while the results
of Berlese are not in accord with those of any of the other investigators.

Some confusion has arisen from the careless use of the word " Korn-
chenkugeln," for which there is no really satisfactory English equivalent.
Some authors have used it to signify {jny leucocyte containing solid
bodies of whatever nature, or, worse yet, some have used it in cases
where it does not appear that the cells in question are even leucocytes.
The " Kijrnchenkugeln " which Weismanu found and so called are leu-
cocytes containing fragments of muscle, either pieces of the contractile
substance or occasionally muscle nuclei. As this is the generally accepted
use of the word, carelessness in its use ought not to be permitted. With

bkeed: metamorphosis of the muscles of a beetle. 347

such a meaning of the word, the presence of " Kornchenkugeln " in an
animal implies, as a necessary corollary, the breaking up of muscles into
fragments somewhere in the body, and the ingestion of these fragments
by the leucocytes. This corollary is probably not generally true in any
of the insects except the higher Diptera, and statements as to the
presence of typical "Kornchenkugeln" in other groups of insects must
be taken with reserve, unless some evidence is offered that they are
" Kornchenkugeln " and not leucocytes containing bodies derived from
some other source than degenerating muscles. According to this defini-
tion, " Kornchenkugeln " is not equivalent to " phagocyte," since it in-
cludes only a particular class of phagocytes, or, if Berlese's idea of the
function of the cells be correct, they ought not to be called phagocytes
at all.

Another cause of confusion is found in statements that muscles de-
generate when, from later observations, it is evident that metamorphose
or some equivalent word is intended. In the present paper, whenever it
is stated that a muscle degenerates, the meaning is that no part of its
substance retains its morphological integrity to function as part of a
muscle or as any other tissue. By metamorphosis of muscles is signified
that some part, or all, of the muscle substance persists, with more or
less change in structure, and functions in the adult either as muscular
tissue or — if Berlese's idea in regard to the development of the imaginal
fat body in Muscidae be correct — sometimes as fat tissue.

B. Observatioxs.

1. Methods.

Serial sections of either the entire insect, or of a large part of its body,
were used, in order that any particular muscle might be identified.
Nearly all of the usually recommended fixing fluids were tried. The
best results were obtained by killing in hot (70° C.) water and fixing
in a cold, saturated solution of corrosive sublimate in 35% alcohol,
or in cold picro-sulphuric acid. It is necessary to cut the animal
open, in order to allow the fixing fluids to penetrate. Objects were
left in the fixing fluids for several hours, even as long as twenty-
four hours in many cases. Hermann's platino-aceto-osmic and Flem-
ming's chromo-aceto-osmic mixtures are good for special purposes, but,
on account of their lack of penetrating power, they are not as good for
general results.

VOL. XL. — NO. 7 8

348 bulletin: museum of comparative zoology.

The serial sections were cut 6| or 10/a in thickness and stained on
the slide. Borax carmine, safranin, haemalura, and several haematoxylin
stains, including iron haematoxylin, were tried, but none gave as good
results as a saturated aqueous solution of thionin. This is very selective
and does not stain the cytoplasm of the growing tissues as deeply as
most of the other stains. My thionin preparations have not faded
much, though some of them are three years old. The preparations in
which the stain has a greenish tinge fade more quickly than those in
which it is of a deep blue. All of the preparations used in making
drawings were stained in thionin. Haemalum and safranin are also
very satisfactory stains.

2. Histological Clianges of the Muscles.

The hypodermal muscles of insects exhibit three varieties which,
though fundamentally alike, present quite dififerent appearances under
ordinary magnifications. Weismann ('62) has designated these types
as the larval, the leg, and the wing muscles, from their principal
distributions.

The muscles of the larval type include in Coleoptera not only all of
the muscles of the larva, but also some of those of the pupae and
imagines. Those found in the pupa and imago exist in the abdominal
region only, and are muscles of the larva which have persisted unaltered
during the metamorphosis. All of these muscles are composed of a few
relatively large fibi'es with a well-marked sarcolemma, and usually with
the nuclei at the periphery of the fibres.

The muscles of the second, or leg, type are formed during pupal life,
and are found not only in the legs but also in other parts of the body.
In the imaginal form of Thymalus all of the skeletal muscles are of this
type, except the few metathoracic muscles mentioned below, and the
persistent larval muscles of the abdominal region noted above. These
muscles are composed of numerous small fibres frequently arranged in a
penniform or bipenniform manner and attached by a common tendon.
The nuclei are found at the surface of the fibres in Thymalus, but in
many other insects, including many Coleopterous forms, they are
arranged in rows along the axis of the fibres.

The muscles of the third, or wing, type are frequently spoken of as
the fibrillar muscles, since they separate very readily into their primi-
tive fibrillae. They are composed of very large fibres with nuclei scat-
tered throughout their substance. Numerous tracheoles penetrate the
fibres of these muscles. The following muscles are of this type in the

bkeed: metamorphosis of the muscles of a beetle. 349

imagines of Coleoptera (compare Aubert, '53) : musculus metanoti,
musculus lateralis metanoti, musculus lateralis metathoracis, flexor coxae
metathoracis (secundus), extensor alae magnus metathoracis, and exten-
sor alae parvus metathoracis.

a. Muscles that pass unaltered from the Larva to the Imago.

The larval muscle fibres of Thymalus have the structure of this type
of cross-striated muscle. Cross and longitudinal sections are shown in
Figures 16, 22 (Plate 6) and Figure 33 (Plate 7). A granular sarco-
plasm containing the nuclei is found unevenly distributed just beneath a
well-marked sarcolemma. Occasionally the nuclei are embedded deep
in the fibres, but these exceptions are practically limited to a certain few
muscles ; as, for instance, the adductor mandibularis, where the fibres are
larger than usual and frequently have their nuclei embedded in the
contractile substance. The cross striations are well marked (Figure 33),
and may show all of the usual bands (Z, E, N, J, Q, H of Rollett, '85).
The muscle columns are flattened and of irregular shapes, so that the
Cohnheim's areas seen in cross sections (Figures 16, 22) make a peculiar
pattern.

The trachae supplying the larval muscles break up into fine intracel-
lular tracheoles at the surface of the fibres. Whether these tracheoles
penetrate the sarcolemma or not, is difficult to determine with the
methods used. From cross sections (Figures 16, 22, trl.) it appears as if
they penetrated the sarcolemma (sar'lem), but remained in the super-
ficial layers of the sarcoplasm (sar'pl.).

The muscle fibres of the abdomen, whose anatomical positions have
been described on page 338, preserve the structure just described in all
of the stages of the pupa and the imago.

6. Metamorphosis of Larval Muscles into

(1) Muscles of the Wing Tijpe.

a. Period of the resting Larva or Period of Destructive Changes. In
the feeding larva the muscles which metamorphose into imaginal
muscles of the wing type show the same structure as the larval mus-
cles described above. When the larva ceases feeding, and the wings
have been evaginated from their hypodermal pockets, these muscles
undergo several rapid changes. Perhaps the most striking of these
changes take place in the contractile substance. This, in the course
of a few days, divides lengthwise into from four to ten strands, the

350 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

division being completed at a stage when the wings have grown so
large that they begin to be crumpled and folded. Figure 14 (Plate 6)
and Figure 34 (Plate 7) show, respectively, cross and longitudinal sec-
tions in which this division has been partially accomplished. Figure 14
shows the cross section of six angular strands, the larger of which again
divide to form the usual eight or nine fibres of this muscle in the imago
(Figure 15), The rounding of these more or less angular strands into
the cylindrical form of a muscle fibre takes place in the very young
pupa.

At an early stage in the division of the fibre, the sarcolemma is broken
up and soon disappears.

The changes in the finer structure of the muscle substance during the
time in which the fibres undergo this division are very noticeable. These
changes are illustrated by a series of drawings magnified 1,600 diameters,
in which both cross and longitudinal sections are shown at three differ-
ent stages of the resting larva. Stage one (Figures 23, 26, Plate 6)
represents the condition before any change has taken place. Cohnheim's
areas (aa. Cohn.) are very plainly shown in the cross section, while the
longitudinal section shows both longitudinal fibrillation and cross
striations.

Stage two (Figures 24, 27) is from a resting larva several days before
pupation. The figures are drawn from muscles which correspond in
their stages of development with those shown in Figures 14 (Plate 6) and
34 (Plate 7). In the figures at the higher magnification (Figures 24, 27)
it is seen that the muscle columns have partially separated into their
primitive fibrillae, Cohnheim's areas appearing in only a few places.
The cross striation has disappeared entirely, whereas the longitudinal
fibrillation shows nearly as plainly as before. The sarcoplasm between
the fibrillae has meanwhile increased in amount and now begins to take
a stain with thionin, a characteristic of the cytoplasm of all actively
growing tissues. This is a strong reason for believing that the sarco-
plasm is itself in an active metabolic condition, and therefore the agent
which is causing the solution of the fibrillae.

Figures 25 and 28, which represent stage three, are drawn from a
series of sections of a larva which would have pupated in a few hours.
These figures show only a finely granular sarcoplasm, in which there is
no trace of the fibrillae of the previous stage, not even a suggestion of
longitudinal fibrillation remaining. The muscle as a whole appears still
more deeply stained than before, since none of the non-staining fibrillae
remain.

breed: metamorphosis of the muscles of a beetle. 351

The course of events in the destructive changes of the contractile
substance is quite evident from these three stages. The muscle columns
break up into their primitive fibrillae, and these then undergo dissolution.
The sarcoplasm increases in amount during this process, but not enough
to balance the loss in volume caused by the dissolution of the fibrillae, so
that each fibre shrinks in actual volume. This is shown by a determi-
nation of the volume of the largest fibi'e of musculus metanoti (Plate 1, Fig-
ure 1, mt'nt.) in each of the three stages described. Of course there is a
chance for error in this determination, in that the muscle fibres vary in size
in different individuals ; but the ratios of the volumes in the three stages
will at least give an indication of the amount of shrinkage. The ratios

-, ■ 1 1 stage i : ii : iii

of the volumes are m the case determined very nearly, — r^^ — - — ^r-

•^ volume 4 : o : 2

From this it seems probable that not all of the material derived from the

dissolution of the fibrillae is transformed immediately into sarcoplasm,

but that some of it remains for a time in solution. It is suggested above

that the agent which causes this dissolution is the sarcoplasm. There is

no evidence of the action of leucocytes, either phagocytic or lyocytic,

since they come into the neighborhood of the muscles only occasionally ;

nor is there reason for supposing action on the part of other outside

agents.

During the whole period of these destructive changes the muscle
nuclei undergo frequent amitotic divisions. The larval nuclei (Plate 7,
Figure 34, nl.) before division are comparatively large, with usually a
single definite nucleolus. Figure 34 shows a nucleus dividing amitoti-
cally (nl.^) and three pairs of smaller nuclei (nl.^), the resultants of
such divisions. At pupation very few of the nuclei presenting the
characteristics of nl. are found, whereas very much elongated nuclei
(Plate 6, Figure 25, ??/.,* shows one that is comparatively short) are found
associated with strings of nuclei which have arisen from the division of
such elongated ones. Many of these nuclei no longer lie at the periphery
of a fibre, nor even at the periphery of one of the strands which have
arisen from the division of a fibre, but are deeply embedded in the
muscle substance (Figures 14, 27, 28).

The sarcoplasm found at the surface of the larval fibres becomes lost
at an early stage, intermingling with the increasing amount of sarcoplasm
between the fibrillae.

The only tissues, other than the muscular, which need to be considered
in this connection are the tracheae and the embryonic tracheal cells.
The tracheal endings on the muscles before any change takes place have

352 bulletin: museum of comparative zoology.

been described. Immediately on the division of the muscle into strands,
the cells of these finer tracheoles begin very rapid mitotic division.
Cells in various stages of division (cl. init.) are to be found in nearly
every section of a muscle in a stage similar to Figure 14 (Plate 6) and
Figure 34 (Plate 7). Most of the new cells so formed become either
actually or apparently detached from the tracheoles, and penetrate into
the fissures between the muscle strands (cl. tr.). Some, however, re-
main connected with the tracheae and show tracheoles, running through
their cytoplasm (Figure 14, cl. tr.^). Especially in longitudinal sections
(Figure 34, cl. tr.) they show long processes, which frequently connect
with each other. These processes cause the cells to be of irregular forms,
the spindle form being, however, the most frequent. The cytoplasm
stains so deeply in thionin that the limits of the nuclei ai-e in many
cases difficult to determine.

J^

foO So °of >>°0-||o<, O OOO^

o ° o^ ■" " 6

Fig. a.

Other considerations than those mentioned above point to the origin
of these cells from the cells of the walls of the tracheae. Figure A is
a projection of the nuclei of the tracheal cells (represented by the small
oval outlines) on an optical longitudinal section of the largest of the fibres
of musculus nietanoti (Plate 1, Figure 1, mfnt.) to show the positions
and numbers of these cells. The particular fibre chosen for this recon-
struction was in an early stage of its metamorphosis, the reconstruction
being made from a series of cross sections. similar to Figure 14 (Plate 6).
From the textfigure it is seen that near the places where the tracheae
join the fibre, tracheal cells are much more numerous than elsewhere,
and that they are distributed in just such positions as would be expected
if they were being formed from the intracellular tracheoles which arise
from the tracheae. This uneven distribution of the tracheal cells can
scarcely be explained by assuming an origin of these cells from nuclei of
the muscle fibre or from leucocytes. Mitosis is found in the cells of

breed: metamokphosis of the muscles of a beetle. 353

the walls of the tracheae, the tracheal cells, and in the cells of the hypo-
dermis, the latter being, of course, the tissue from which the traclieae
were derived. Few of the other tissues show mitosis, amitotis being the
method of division in both leucocytes and muscle nuclei. Moreover',
there is little chance of confusing the tracheal cells with leucocytes, as
the latter are readily distinguishable by their more rounded form and
finely vacuolated cytoplasm, which does not stain as deeply as the cyto-
plasm of the tracheal cells. The sudden appearance of the tracheal cells
in all parts of the body at once, precludes any possibility of a local place
of origin, such as the base of the wing, etc. Finally their fate, i. e.,
development into tracheae, indicates their origin from tracheae.

The question might be raised, whether or not these cells are the active
agents in the splitting of the muscle into strands. This can scarcely be
so, because the earlier the stages in the changes of these muscles, the
fewer are these cells in the spaces between the strands. Moreover, ia
the earliest stages there are numerous fissures in which there are no
tracheal cells.

The relationships of these tracheal cells to the mesenchyme, mesoderm,
embryonic cells, myocytes, etc., which other investigators have found in
connection with the postembryonic development of insects, cannot be
entirely settled. The tracheal cells are doubtless the same as the
spindle cells of Deegener. It is also probable that they ai'e the same
as the so-called myocytes of Berlese ; at least, the same as those that he
has described for Coleoptera. That entirely different kinds of cells have
been described under these various terms, is almost certain. For my-
self, I am disposed to think that there are present during the metamor-
phoses of holometabolic insects, two distinct kinds of embryonic cells,
which resemble each other in form and" structure, but which have differ-
ent origins and fates. One kind might properly be called mesenchymal ;
these are cells which arise singly from the tracheae or hypodermis and
rise to tracheae, leucocytes, and other related tissues. Such cells are
to be expected in most cases. The other kind may be called mesodermal.
Their origin is not established as yet, but probably they are derived
from cells of the embryonic mesoderm which persist until pupal life.
They give rise to muscles and possibly other tissues in the pupa and are
found principally in those insects in which muscles are newly formed
during pupal life. There are many facts to support such a view, but it
cannot be definitely proved with the material at hand.

/?. Pupal or Reconstructive Period. The time of pupation agrees
closely with the change from destructive to reconstructive changes in

354 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

the wing muscles, destructive changes taking place for only a short time
after pupation. As we have seen, the so-called wing muscles are at the
time of pupation composed of a few cylindrical strands or fibres of undif-
ferentiated sarcoplasm which contain many nuclei undergoing rapid
amitotic division. For some time in the pupa no very evident changes
occur. Many of the elongated muscle nuclei and numerous chains of
nuclei (Plate 6, Figure 30) are present. The tracheal cells are still
increasing rapidly by mitosis, and in a two- to three-day pupa have be-
come numerous, occupying most of the space between the strands
(Figure 19, cl.tr.).

At a stage when pupal life is nearly half over, the fibrillae of the adult
muscles begin to show. Figures 29 and 30, represent the appearance of
the muscles at this period. The cross section (Figure 29) shows scattered
through it the cross sections of newly formed fibrillae of various
sizes. The longitudinal section (Figure 30), taken from another muscle
of the same series of sections, shows longitudinal fibrillation. Sections
of stages a little younger than this, e.g., the stage shown in Figure 19, re-
veal only the faintest hint of these structures under high magnifications.

During the last half of pupal life, a number of important changes take
place, the most noteworthy being growth in size. In some muscles the
area of cross section doubles or even quadruples during this period
(compare Figure 19 with Figure 21, the latter showing three fibres of
the former, the magnification being in each case 800 diameters). This
increase in area of cross section is accompanied by a lengthening of the
muscles, sometimes to even twice their former length, so that their
volume increases many fold. A rough estimate of the changes in
volume during metamorphosis of any metathoracic muscle can be made
from the series of anatomical drawings given on Plates 1-5, as these are
all drawn to the same scale.

The tracheal cells in a stage a few days before the emergence of the
imago (Figure 21, cl.tr.) arrive at a condition in which there are no
more cell divisions. In cross sections of the muscles at this stage the
tracheal cells are not as numerous as in the earlier stages (Figure 19).
This does not mean that they are fewer in jiumber in the whole muscle,
however, as the volume of the muscle has increased witliout a corres-
ponding increase in the number of tracheal cells. Nearly every tracheal
cell in Figure 21 shows its future plainly. Some {cl. ir.^) have formed
tracheoles through their cytoplasm and show connections with tracheae.
Most of tlie others are connected with tracheae, but their connections are
severed by the plane of the section (d. tr.^. There are a few, however,

breed: metamorphosis of the muscles of a beetle. 355

which {cl.tr.) do not show their tracheal nature in the least, these
forming a direct transition to the tracheal cells of the previous stages
{cl. tr.y Figures 14, 19, etc.). The processes of these cells are embedded
in the muscle substance, and even some of the cells {d. tr.^) may be
entirely embedded in the muscle. All through the substance of the
muscle are found the processes {pre.) of these cells detached from the
cell body by the plane of the section. Some of these processes are solid,
but most of them are already tubular tracheoles, which show prominently
in the sections because their walls stain deeply. They may be seen
better in the more enlarged representation (Figure 32, pre). This
penetration of the wing muscles by the tracheoles has long been known,
but their development has never before been described. A similar
development of the intracellular tracheoles in other parts of the body has
been noted in several cases.

It is probable that some of these tracheal cells become leucocytes at
about this period. Certainly the large vacuolated leucocytes which have
persisted from the larva, such as are shown in Figure 51, leucyt.
(Plate 7), disappear in old pupae, and their places are taken by smaller,
less vacuolated leucocytes which resemble the tracheal cells. These
new leucocytes grow in size, and soon are characteristically vacuolated
(Figure 36, leu'cyt.).

The finer structure of the muscle substance at a stage corresponding
to Figure 21 (Plate 6) is shown in Figure 32. The fibrillae are much
more numerous than before (Figure 29), and show more plainly in cross
section, while the amount of stainable sarcoplasm between them is
relatively less, so that the muscle as a whole stains fainter than before.
In longitudinal sections the fibrillation is plain, but no cross striation is
visible. In none of my sections of pupae does the cross striation show
in these muscles, but it appears in a series of sections of an imago a few
hours old (Figure 31), so that possibly this striation is formed during
the last stages of pupal life.

In the stage shown in the longitudinal section the muscle nuclei
(Plate 7, Figure 35, wZ.^) are still dividing amitotically, but in the
somewhat older stage, shown in cross section only (Figure 21, Plate 6),
amitosis is rare. The nuclei in this older stage are numerous and are
scattered throughout the substance of the muscle. They are short oval
in form, the elongated nuclei of the preceding stages having disappeared
entirely.

y. Imaginal Period. The structure of the wing muscles of insects has
been described so well by various authors that it need not be repeated

356 bulletin: museum of comparative zoology.

here (see Heidenhain, '98, for a bibliography of papers on cross-striated
muscle). Cross and longitudinal sections of these muscles in Tliymalus
are given in Figures 15 and 36, respectively. The changes since the old
pupa are few. Cross striation is readily distinguishable, showing the
J and Q bands. The fibrillao show clearly in both cross and longitudinal
sections, and are nearly all of one size. In Thymalus they are about 1 /a
in diameter, which is smaller than in many other insects. No sarcolemma
could be demonstrated, though it has been described for this type of
muscle (see Cajal, '88, p. 268).

The tracheoles {trl.) are fully developed and are often to be seen in
the muscle substance. It is, however, much more difficult to distinguish
them than it was earlier, since they have thinner walls and these do not
stain as deeply as in the earlier stage.

(2) Muscles of the Leg Type.

The figures already described as showing the structure of the larval
muscles (Plate 6, Figures 16, 22, and Plate 7, Figure 33) will serve as
a starting point for the description of this type also ; for, as already
stated, both the wing and the leg muscles are at first alike. In some of
the larval muscles which are destined to metamorphose into muscles
of the leg type, changes begin at the same time that they do in those of
the wing type, i.e., at about the time the larva ceases feeding; but in
others of the leg type metamorphosis does not begin until later. The
muscles which are to undergo the greatest changes in position at the
time of pupation begin to show alterations first. The others start their
changes during the resting larval period, though some of them are not
greatly changed even at the time of pupation. On account of this varia-
tion in the time of the beginning of the metamorphosis in different
muscles, it is of great importance to be able to identify these muscles at
every stage of development. The details of their metamorphosis are, how-
ever, apparently the same in all instances, there being in no case which
has been observed transitional conditions between these metamorphosing
muscles and the muscles which pass unaltered from the larva to the

imago.

These muscles may be somewhat artificially divided into three groups,
according to the period in which they begin their metamorphoses.
Those of Group I. begin their metamorphosis at the same time as the
muscles of the wing type. This group includes, among other muscles,
the adductor of the mandible, and the following metathoracic muscles :
the third flexor of the wing, the relaxator of the wing, and the relaxator

breed: metamorphosis of the muscles of a beetle. 357

of the extensor of the wing. Group IT. includes those muscles which be-
gin their metamorphosis soon after the muscles of Group I. have begun
theirs, but which retain their cross striation until the time of pupation.
Examples of metathoracic muscles of this group are : the first and second
flexors of the wing and the third extensor of the coxa. The remaining
group (III.) includes the muscles which show little evidence of metamor-
phosis even at the time of pupation. Among these may be mentioned
the dorsal muscle of the mesofurca, the lateral muscle of the inferior
process of the mesophragma, the lateral muscle of the mesofurca, the
depressor of the tergum, and the flexor of the postero-lateral process
of the metafurca. It will be noticed that the examples of Group III.
include all of the intersegmental muscles which lie between the meso- and
metathorax, and also all of those between the metathorax and the first
abdominal somite. Why these muscles should all belong to the group
which is the most retarded in beginning its metamorphosis, is not
evident.

a. Larval Period. In the muscles of this type the larval existence
does not include the entire period of destructive changes, these extend-
ing into the pupal stage. In the destructive alterations, the differences
between those larval muscles which metamorphose into muscles of the
wing type and those which assume the leg type are not great; these
differences alone need be mentioned. Figure 49 (Plate 7) shows a
cross section of the second flexor of the wing drawn from an older larva
than the one from which Figure 14 (Plate 6), of the wing-muscle series,
was drawn. These muscles are at nearly the same stage of development
and will serve to illustrate the differences in the metamorphoses of the
two types. These differences are chiefly, that the muscles of the leg
type divide into a greater number of smaller longitudinal strands (19-22
in the particular muscle figured), and that the fibrillae of most of the
leg-type muscles do not disappear as quickly as those of the wing type.

p. Pupal Period. Eventually the substance of these muscles reaches
a structureless condition, the same as is shown in Figures 25, 28 (Plate
6) for the wing muscles, though this stage in some cases is not attained
until the middle of pupal life. In fact, the structureless condition has
not been observed in all of the muscles of Group III. mentioned above.
It is even possible that in some cases the fibrillae of the larval muscles
of this group may persist as fibrillae in the imaginal muscles. If
so, these muscles would form a transition, so far as the contractile
elements are concerned, to those which remain entirely unchanged from
the larva to the imago. The structureless period is certainly of shorter

358 bulletin: museum of comparative zoology.

duration in some muscles than others, and is not found in all of the
muscles at the same instant.

During the period of these destructive changes in the contractile
muscle substance, the angular strands become more rounded and
separated, precisely as in the wing muscles during the same period.
However, the nuclei, with rare exceptions, remain at the periphery of
the strands. The tracheal cells are never formed as numerously as is
shown for the wing muscles in Figure 19, and, in fact, are fewer at all
stages than in the wing muscles at the corresponding stages.

The reconstructive changes begin in the pupa, at varying times for the
different muscles, the same as has been shown concerning the beginning
of the destructive changes. It is difficult to determine much about the
reconstruction of the fibrillae of these muscles, because the fibrillae are
so small. In fact, it is not certain that they have been recognized. In
cross sections of these muscles from old pupae there appear irregular
polygonal areas of small size (less than 1 /a in diameter), which, how-
ever, are presumably Cohnheim's areas, rather than the cross sections of
separate fibrillae. These become more evident in later stages, and show
plainly in the imaginal muscles (Figure 18). Longitudinal fibrillation
appears at the same time that the polygonal areas begin to show, whereas
cross striation is not seen until the day before the emergence of the
imago. A longitudinal section of a stage corresponding to that shown in
Figure 18 is given in Figure 17. This presents the usual appearance of
the cross-striated muscles of the legs of insects.

y. Imaginal Period. The same muscle that is shown in cross section in
its larval state in Figure 49 (Plate 7) is represented in its imaginal state
in Figure 50. A comparison between the two figures will reveal how
simple the changes between the two stages really are. In the imaginal
muscle, there is evident a superficial layer of sarcoplasm with the nuclei
embedded in it. A sarcolerama is present about each fibre, having been
formed during the late pupal stages. The tracheal cells have developed
into tracheae, which, however, do not penetrate the muscle substance as
in the case of the indirect wing muscles. ~ Most of the muscles of the leg
type increase somewhat in size during metamorphosis, but this increase
is small compared with the growth of the majority of the wing muscles.

(3) Metamorphosis of the Intestinal Muscles.

The intestinal muscles undergo changes precisely similar to those
described for the leg type of muscles. INIy observations are in almost
exact accord with those of Eengel ('96), so far as he has described the

breed: metamorphosis of the muscles of a beetle. 359

changes in the muscles of the intestine. I have studied especially the
region of the proventriculus, where the muscle layers are well developed.
No differences were discovered between the changes of the muscles of this
region and those of the remainder of the intestine. Two general figures
are given. Figure 51 (Plate 7) is a portion of the wall of the proventri-
culus in a larva about to pupate, and Figure 52 is a similar figure from
an old pupa. The muscle fibres are found in two layers : a circular layer
inside (mu. crc), and a longitudinal layer outside {mu. Ig.). Their
structure is similar to that of the other larval muscle fibres, except that
the nuclei are more frequently found at the centre of the fibres and that
Cohnheim's areas are arranged similarly to those shown in Figure 20
(Plate 6) ; this particular figure, however, is not from one of the larval
fibres. The principal difference between the destructive changes in these
muscles and in those of the leg type is, that they are still slower in
being completed than the latter. The larval fibres rarely, if ever, divide
lengthwise to form new fibres, those in the larva being apparently as
numerous as those in the imago. The tracheal cells are slower in mak-
ing their appearance, and only a few are found in this region at the time
of pupation (see Figure 51, which does not show any of them); whereas,
even before this time, they are numerous in the regions of the other
metamorphosing muscles. Compare Figure 14 (Plate 6) and Figure 49
(Plate 7), which are from younger pupae than Figure 51. The intestinal
muscles show cross striation much longer than any of the other metamor-
phosing muscles, as the striation does not disappear until the pupa has
undergone nearly half of its development. Longitudinal fibrillation dis-
appears almost as quickly, and thus a structureless stage, shown in
Figure 52 {mu. crc), is reached.

During all the time in which the destruction of the contractile ele-
ments is taking place, the muscle nuclei show no apparent changes.
No cases of amitosis have been seen, though they are common in the
other metamorphosing muscles ; nor is there any evidence of degenera-
tion and phagocytosis such as Deegener (:00) states that he finds. It
seems as if Deegener's statement, that there is phagocytosis of these
muscles, such as Kowalevsky ('87) and Van Rees ('88) found in Mus-
cidae, must be strongly questioned. For, in the first place, both Rengel and
I have failed to find evidence of it in Coleoptera. Secondly, it is evident
on reading Deegener's paper that this statement is based more on infer-
ence than actual observation. No satisfactory figure nor description is
given of the phenomena which take place when the leucocytes attack
the muscles. Apparently the only ground for the statement is that be

3 GO BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY.

has found what he calls " Kornchenkngelu." Judging from his figures
of them, they do not look much like the " Kornchenkugelu " of the
Muscidae, nor does their migration into the lumen of the intestine agree
with what has been found in Diptera. Moreover, he states that these
phagocytes are not numerous enough in the region of the midintestine
to account for the degeneration of the muscles of this region, and conse-
quently infers that there is chemical degenex'ation as well as phagocyto-
sis. Such different methods of degeneration in similar muscles of the
same animal is improbable. But the principal reason for believing that
there is no phagocytosis of these muscles in Thymalus and other Cole-
optera lies in the exact similarity of all their changes to those occurring
in the muscles of the leg type. In these muscles it can be stated Avith
certainty, not only that there is no phagocytosis, but also that the
larval muscles metamorphose into the imaginal muscles instead of
degenerating.

The typical " Kornchenkugeln " which Deegener finds, but which
Rengel could not find, are met with in Thymalus. That is to say, there
are to be found leucocytes containing bodies many of which would
answer the description given by Deegener, but these leucocytes are not
such " Kornchenkugeln " as Weismann found. This is evident from
some of the appearances reproduced in Figures 40-48 (Plate 7).
These all represent leucocytes found in old pupae magnified 1600
diameters. Figures 43 and 46 look like leucocytes containing de-
generating nuclei, and there is a possibility that such may be the true
explanation of some of them ; none of them, however, are nuclei from
the intestinal muscles. Figures 40, 42, and 47 show inclusions which
certainly are not degenerating nuclei, and since there are found transi-
tional stages (Figure 48) to the first mentioned conditions, it is probable
that all of the inclusions are of the same kind. The most probable
interpretation of them is that they are intracellular parasites. This
view is strengthened by the presence of apparently similar bodies in
the intestinal epithelium of resting larvae. Also, bodies similar to the
deeply stained portions of Figure 40 aro found very numerously in the
body cavity and lumen of the intestine of old pupae and young imagines.
The true nature and relationship of these bodies cannot be stated with
certainty as yet, but whatever they may be, very few, if any of them,
can be called " Kornchenkugeln."

Concerning the formation of the intestinal muscles of the imago, my
observations, again, are in harmony with those of Reugel, and disagree
with those of Deegener. The reconstruction of the intestinal muscles

bkeed: metamoephosis of the muscles of a beetle. 361

from the structureless muscle substance containing the larval nuclei is
the same as the reconstruction of the leg muscles. That is, longitudinal
fibrillation appears first, then cross striation, the latter appearing about
the time of the emergence of the imago. At the same time Cohnheim's
areas become plainly distinguishable, and have the pattern shown in
Figure 20 (Plate 6), which is drawn from the cross section of a single
fibre of the foreintestine of the imago. The muscle substance, when
structureless, stains deeply with thionin, but after the fibrillae are
formed, it stains scarcely at all. The nuclei remain as they were, while
a new sarcolemma is formed about each fibre in the old pupa. The
tracheal cells of this region give rise to the new tracheae and possibly,
as stated before, to imaginal leucocytes.

Deegener, who speaks of these tracheal cells as spindle cells (page 146, .
et seq.), derives the intestinal musculature of the imago from them. He
gives no conclusive proof of this derivation in any case, however. In
the region of the midintestine he was unable to distinguish these
spindle cells with certainty, so that his conclusion that the muscles of
this region are formed from these cells is pure assumption. He is
forced to make such an assumption by his conclusion, — which has
already been shown to be incorrect, — that there is a phagocytosis and
total destruction of the larval muscles. Thei'e is no reason for suppos-
ing that these cells form the intestinal muscles of the imago any more
than that they form the muscles of the remainder of the body, and this,
as has been shown, is not true.

c. Histolysis of the Larval Muscles.

The muscles which undergo histolysis in the pupa present great indi-
vidual variation as to the time when degeneration begins. There are
also variations in the details of the degeneration, which are of such a
nature that they form a partial transition to metamorphosing muscles.
However, no instance of a muscle which sometimes degenerates and
sometimes metamorphoses into a rudimentary imaginal muscle has been
found, though it does not seem improbable that such may be present
in some of the beetles.

The group of muscles of the metathorax designated in Figure 1 (Plate
1) by the Greek letters /3, y, 8, e, ^ rj belong to a class of degenerating
muscles which are very distinct from the metamorphosing muscles. This
group will servo as a type in describing the degeneration and the differ-
ences between these and the other degenerating muscles noted later.
The substance of these degenerating muscles never stains with thionin.

362 BULLETIN: MUSEUM OF COMPAEATIVE ZOOLOGY.

For this reason, they stand in sharp contrast with the nearhy metamor-
phosing muscles. No other evidence of degeneration manifests itself
until the pupal stage is reached. Then there begins a gradual atrophy
of the muscles, during which the substance of the muscle becomes some-
what broken, as is shown in Figure 39 (Plate 7). This figure, drawn
from a cross section, is of muscles ^, -q (Plate 1, Figure 2), and Figure
37 (Plate 7) is a longitudinal section of one of the similar group of
mesothoracic muscles, both taken from pupae a few days old. The size
of the area of cross section has diminished nearly one half at this stage ;
this, however, does not mean a proportional shrinkage in volume, because
the length of the fibres increases at pupation. Cross sections at this
stage show Cohnheim's areas, but only where viewed with a higher
magnification than that used in making Figure 39. Longitudinal sec-
tions (Figure 37) show fibrillation distinctly and cross striation faintly.
The nuclei are apparently unchanged, retaining the nucleoli found in the
nuclei of the larval muscles. In longitudinal sections they commonly
project from the surface of the fibres, as shown in the figure. Sarco-
lemma can usually be distinguished even at this stage. Tracheal cells
are sometimes found in the fissures of the muscle substance (Figure
39, d. tr.), though this is not common. There can be little question
of the identity of these cells with the tracheal cells of the remainder of
the body, or of the fact that they are not leucocytes. There is no
evidence of phagocytosis at any stage.

From this period of the young pupa, until the old pupa, there is a
gradual atrophy of the muscle substance of each fibre, until only a
slender strand is left. This strand has in connection with it all the
nuclei of the original fibre, these nuclei showing little evidence of de-
generation until practically all of the remainder of the fibre has entered
into solution. They then undergo a typical chromatolysis, as shown in
Figure 38, nl. Inside the nuclear membrane, the chromatin grains col-
lect into masses of various sizes which at first stain deeply. These
masses seem to persist for a short time after the dissolution of the
nuclear membrane, for there may be found such chromatin masses (chr.)
around which no nuclear membrane can be distinguished. No trace of
these muscles can be found in pupae shortly before the emergence of the
imago. The possibility that leucocytes may engulf some of these degen-
erating nuclei ought to be mentioned. Such an engalfment of loose
debris would agree with the well-known habits of leucocytes, and it might
be contended that such appearances as are represented in Figures 41,
44, and 45 (Plate 7) are due to this cause. No direct evidence can be

breed: metamorphosis of the muscles of a beetle. 363

given for or against this view, but it seems to me that more probable
explanations of the source of these leucocytes can be given.

Transitional conditions between degenerating and metamorphosing
muscles have been noticed, especially in the musculus lateralis meso-
thoracis and other mesothoracic muscles whose counterparts in the meta-
thorax metamorphose into imaginal muscles. Until a few days before
pupation, there are few differences between the changes of these meso-
thoracic muscles and those of their counterparts in the metathorax. That
is, the changes of the mesothoracic muscles differ from those of the type
of degenerating muscles just described in the following particulars : they
begin their changes in the early resting larva, instead of at the time of
pupation ; they split into a definite number of longitudinal strands ;
their nuclei divide amitotically, though not as abundantly as in most
of the metamorphosing muscles ; the muscle substance stains with
thionin ; and the tracheal cells are present in considerable numbers.
All these features so resemble those of the metamorphosing muscles that
for a long time I supposed that these muscles likewise metamorphosed.
It was only by tracing the history of each muscle individually that I was
able to establish their final and total disappearance. Their final disin-
tegration takes place in the old pupa at tlie same time, and in the same
manner, as that of the other degenerating muscles. The fate of the
tracheal cells connected with them is not certain, but eventually they
must become free in the blood plasma, where they presumably form
tracheae or leucocytes.

The probable explanation of the similarity of these degenerating muscles
to the metamorphosing muscles is, that in some ancestral form not far re-
moved, the former also metamorphose to become imaginal muscles. That
such a condition (i. e. a metamorpliosis^o^ the l.mfthx. and the other
degenerating mesothoracic muscles) will be found in some of the hemimet-
abolic insects, is very probable. A similar relation between the fibrillar
■wing muscles of certain beetles is almost certain. In Thymalus these
fibrillar muscles are metamorphosed larval muscles, but in the imagines
of certain wingless beetles they are not found (Aubert, '53). It is prob-
able, therefore, that investigation would show their presence in the larvae
of these forms and that they degenerate in the pupa.

d. Histogenesis of the Imaginal Muscles.

Nothing has been determined with certainty about the origin of the
two metathoracic muscles of Thymalus which were absent in the larva.
They probably are derived in the same manner as the muscles of new

VOL. XL. — NO. 7 4

364 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

formation in the pupa of other beetles; that is, from cells resembling the
tracheal cells, but probably having a diflerent origin.

3. Observations on other Coleoptera.

Bruchus obtectus Say, the common bean weevil, was chosen for com-
parison with Thymalus chiefly because of the different conditions which
might be expected in tlie leg muscles. Thymalus is a form with an un-
modified larva possessing six well-developed legs. Bruchus, on the other
hand, has a more highly specialized larva, which has legs when it hatches
from the egg, but at the first moult loses all except the merest rudi-
ments of them. During the remainder of larval life, these rudiments
are barely visible. The legs of the first larval form are scarcely larger
than the hairs which are found on other parts of the body. They do
not show all the joints of the adult leg, but only the femur and tibia,
the latter possessing an enlargement at the distal end which represents
the tarsus. In whole preparations, no muscles can be distinguished in
these legs, and it is probable that they are functionless as locomotor
organs. (For descriptions and figures of the larval stages of this insect,
see Chittenden, '99.)

Sections of half-grown larvae — the youngest used in sectioning — show
rudiments of legs, at the bases of which are found masses of cells. Tliese
masses are principally composed of the small spindle-shaped cells which
later give rise to the muscles of the imaginal legs. These cells have a
somewhat oval nucleus surrounded by a small amount of cytoplasm. A
few tracheae aerate this mass, while an occasional leucocyte is also found.
The origin of the spindle cells has not been traced, but they are pre-
sumably the embryonic mesoderm cells which would have formed the
muscles of the legs, had muscles been functionally developed in the legs
of the larva.

At the time of pupation, three kinds of cells are found in these masses.
There are (1) the leucocytes, which are readily distinguished. They are
several times larger than the other cells, have a more rounded form,
an abundant cytoplasm, and a spherical nucleus, in which the chro-
matin network lies chiefly at the periphery. The remaining cells are
spindle-shaped and apparently all alike ; but later stages of development
indicate that they are of two kinds, which probably have different origins.
These are (2) the mesoderm cells mentioned above and (3) mesenchy-
matous tracheal cells. The mesoderm cells probably have an embryonic
origin, and they develop into muscles. No direct proof of the origin of
the tracheal cells can be given, because in their young stages it has been

BREED: METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 365

impossible to distinguish them from the mesoderm cells. But from
analogy with the remainder of the body, it is very likely that they have
not persisted from embryonic life, but are developed during the period
of the resting larva from the tracheae which supply the masses of tissue
at the bases of the legs. They develop into the tracheae of the legs of
the imago.

In young pupae in which the legs have grown to some size, in the
places where new muscles are to be formed, there may be found groups
of cells already transforming into muscle fibres. Between these form-
ing fibres are to be seen free cells, many of which are dividing mitotically.
These may now be recognized as tracheal cells, which are precisely like
the cells found associated with the metamorphosing muscles of the
remainder of the body. The muscle nuclei in the earliest stages in which
they can be recognized as such are seen to be undergoing frequent
amitotic divisions. From this time on the amitotic is their only
method of division : a thing which is characteristic of the nuclei of all of
the muscles which have been studied. The muscle fibres increase rapidly
in size, and it very soon becomes impossible to distinguish them from the
metamorphosing muscles of the leg type, which meanwhile have com-
pleted their destructive changes, and are starting on their reconstruction.
The tracheal cells remain as free cells between these fibres until a late
stage of the pupa, when they form tracheae in a manner similar to that
already described for Thymalus.

The question whether each muscle fibre is developed from a single
cell or not, is almost impossible to settle in this case. There cannot be
much fusion, however, as the fibres of the completed muscles are almost,
if not quite, as numerous as the cells from which they are developed.

The metamorphosing, degenerating, and persistent larval muscles of
Bruchus obtectus show conditions exactly comparable with those of Thy-
malus. The fibrillae of the indirect wing muscles are larger in Bruchus,
and their development in the structureless sarcoplasm of these muscles in
the pupa is much more obvious than in Thymalus. No leucocytes with
inclusions have been found at any stage, though a careful search has been
made for them.

Sections of larvae and pupae of Synchroa punctata Newm., a Melan-
dryid oak-bark borer, and Cylleue pictus Drury, the common Cerambycid
hickory borer, have also been examined. The muscular changes of
these forms are essentially like those already described. A sharp look-
out has been maintained for " Kornchenkugeln," or similar bodies, hut
none have been seen in these forms.

366 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

C. Discussion of Results.

An attempt will now be made to harmonize the results of the various
investigators of the muscular changes of Coleoptera. The researches of
those who have studied the remaining groups of holometabolic insects,
though treated of first, will not be considered in detail, because tlie
relation of the changes in Coleoptera to those in the other groups are
not yet perfectly clear. It is sufficient to state that the results of this
paper are not fundamentally at variance with those obtained by many
of these investigators.

Concerning the state of affairs in Diptera, the following facts are
evident from the papers on the subject. In the orthorraphic Diptera
there is a persistence of many of the larval muscles. The degeneration
of those muscles which disappear during pupal life does not seem to be
diflferent from that found in Coleoptera. In the cyclorraphic forms no in-
vestigator has found a persistence of larval muscles. Degeneration seems
to be the common fate of the larval muscles, a degeneration which
takes place by a method different from that found either in Orthorrapha
or in other insects. Muscles newly formed in the pupa ai'e very common
in Diptera, especially in the higher forms. A true metamorphosis of
larval muscles into imaginal muscles has been noted by Van Kees ('88)
only. I can confirm from my own observations the metamorphosis of
the three pairs of muscles which Van Eees has noted. Contrary to his
statement, however, these do not form all of the indirect wing muscles,
but only musculus mesonoti, each of the three larval muscles dividing
into two fibres, and thus giving rise to the six fibres composing the
imaginal mesonotal muscles of each side of the body. A similar
development of musculus mesonoti from three pairs of larval muscle
fundaments is found in Culex sp. and Chironomus sp. The metamor-
phosis of the undoubtedly homologous three pairs of larval muscles in
both meso- and metathorax of Thymalus has already been noted
(pages 337 and 323, respectively).

The results of the investigators who . have studied Lepidopterous
material are so greatly at variance with one another that little can be
stated definitely. The probabilities seem to favor the authors who state
that there is a metamorphosis of many of the larval muscles. Perez
(:00) states, and probably correctly, that many of the larval abdominal
muscles pass into the adult with no changes except a proliferation of
their nuclei.

It is my belief that not one of the investigators of Hymenopterous

breed: metamorphosis of the muscles of a beetle. 367

forms has interpreted entirely correctly the phemomena which he has
seen. I affirm this the more confidently hecause in the controversy
which has arisen among these authors neither side has satisfactorily
explained the observations of the other. They all agree in describing
phenomena which are so like those of which I have here given an
account for Coleoptera, that it does not seem possible that there should
be any fundamental differences between the two groups. It is evident,
chiefly from the completed paper of Anglas (:01), that there is in
Hymenoptera a metamorphosis of most of the larval muscles, a degener-
ation of the remaining ones, and a new formation in the pupa of some
imaginal muscles. There are no persistent larval muscles such as exist
in Coleoptera, Lepidoptera, and orthorraphic Diptera, the abdominal
muscles undergoing a less complete metamorphosis than the metamor-
phosing muscles of the remainder of the body.

The settlement of the whole controversy between the five authors
(Karawaiew, Terre, Anglas, Perez, Berlese) depends on the interpreta-
tion of the nature of certain cells found in the regions of the metamor-
phosing and degenerating muscles, these cells being apparently exactly
comparable to the cells in Coleoptera which have been spoken of in the
present paper as tracheal cells. N'one of the five authors mentioned
above has considered the possibility of the tracheal nature of these cells.
Nevertheless, none of their observations preclude such an origin.
Karawaiew, Terre, and Berlese contend that these cells are not leuco-
cytes, hut are developed from the nuclei of the larval muscles ; whereas
Anglas and Perez contend that they are not developed from the nuclei of
the larval muscles, but are leucocytes. Is it not possible that both sides
are correct in their negative conclusions and incorrect in their positive
affirmations 1 May not these cells be developed from the tracheoles of
the larval muscles, instead of from either of the tissues mentioned 1
None of these investigators has described the origin of the tracheae of
the imaginal muscles. Yet these tracheae are so exceedingly abundant
in the region of the wing muscles, that their origin cannot be so incon-
spicuous as to have been overlooked entirely, nor ought it to have been
neglected, as it has been. It is to be hoped that some of these authors
will at least consider the possibility of the explanation which I have
suggested, since, if correct, it will straighten out what otherwise is an
apparently hopeless controversy.

We will now consider the researches on Coleoptera. A review of the
disagreements of Rengel ('96) and Deegener (:00) has already been
given in considering the changes of the intestinal musculature. It is

368 BULLETIN: MUSEUM OF COMPARA.TIVE ZOOLOGY.

rarely possible to confirm the results of another investigator's work more
completely than Rengel's results have been confirmed by my own
investigation.

The results of De Bruyne's ('97) investigation of Tenebrio may be
entirely disregarded, because there can be little doubt but that he has
mistaken tlie fundamental nature of the changes with which he was
dealing. Misled by the similarity in appearance of cross sections of
metamorphosing muscles (such as my Figure 15, Plate 6) to cross sec-
tions of the degenerating muscles of Muscidae (see figures given by
Kowalevsky, '87, Van Rees, '88, and others), he has concluded that the
muscles in Tenebrio likewise degenerate. As a matter of fact, there can
be no doubt but that he was dealing with metamorphosing muscles
which retained their individuality thoughout pupal life, as is indicated by
Kriiger's ('98) results on the same insect, as well as by the present study
of Coleopterous forms. The probability is that his leucocytes, which
he found engulfing fragments of muscle, are the same as the tracheal cells
of the present paper, and that his " Kornchenkugeln " are the same as
the detached fat cells described by Kriiger ('98, p. 16).

Kriiger ('98) was venturesome in generalizing from such meagre data,
but his conclusion is entirely confirmed by the present research. All of
the imaginal wing muscles are metarnorphosed larval muscles, though
some of the other metathoracic muscles nearby are not. However, it is
questionable if the cells which Kriiger ('98, p. 1 7) describes as " "Weis-
mannsche Kornchenzellen " are such in reality. He has given us no
evidence to support the view that the inclusions in these cells are
muscle fragments. Other, just as probable, explanations of the nature
of these cells might be given.

Karawaiew's statement ('99, p. 202), that he finds no phagocytosis
of the muscles of Anobium, agrees with what has been found in
Thymalus.

It was impossible to explain the disagreement of Berlese's results with
the results of the present research, until a copy of his last paper (:02'')
was received. His idea, that there is, in the metamorphosis of the
muscles of all the metabolic insects : first, an emigration of nuclei from
the larval muscles ; secondly, a formation of " sarcocytes " from these ;
thirdly, a transformation of these " sarcocytes " into " myocytes ; " and,
finally, a production of new muscles from these, meets a fatal objection,
as far as Coleoptera are concerned, when the anatomical changes of
these muscles are considered. The first half of my paper is taken up
with tracing individual larval muscles in their metamorphosis into

BREED : METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 369

imaginal muscles. At no stage do these metamorphosing muscles lose
their identity, so that a dissolution of these muscles and a survival of
their nuclei only, is impossible.

Berlese's mistake may be easily explained, however. He has neglected
entirely the study of the anatomical changes ; these would have immedi-
ately revealed the falsity of his view. Moreover, he is unfortunate in
his choice of the adductor of the mandible, as a muscle in which to study
these changes. This muscle is composed of numerous fibres (50 in the
larva, 250 in the imago of Thymalus), so that it is impossible to follow
any particular one of them in its development. When the destructive
changes in the metamorphosis of this muscle are completed, there re-
mains simply a confused mass of these fibres still retaining their nuclei,
with numerous spindle-shaped cells scattered between the fibres, pre-
cisely as Berlese describes and figures (:02% p. 65, Fig. 253). His
mistake arises from his imagining that spindle cells are derived from the
muscle nuclei, a mistake very easily made. In some of the beetles
which I have examined, the diS"erence between these cells and the
muscle nuclei is not obvious at first sight. In Thymalus, however, there
can be no doubt of a difference between them at all stages. As already
shown, the spindle cells develop from tracheae and into tracheae, while
the muscle nuclei persist as they are in the nndi£ferentiated sarcoplasm
and form the imaginal muscles. The conditions which Berlese shows in
his second figure (Fig. 254) are different from anything observed in
Thymalus. That all the cells pictured in this figure are of the same
nature, is open to question. It has also been shown that there is no
need of supposing a derivation of complete cells from nuclei alone, as
Berlese has done. This assumption itself is enough to shake one's
confidence in his views.

He also lays great stress on the simplicity of his idea, and the fact that
he has been able to make it apply in every case which he has studied.
But there may be a fault in too great simplicity, as well as in too great
complexity. The reasonableness of the ideas of the present paper, as
contrasted with those of Berlese, may best be shown by tracing what
may have been the phylogenetic development of these muscular changes.

It is fair to assume that in primitive insects the muscles were the
same in number, function, and position, when the larva escaped from the
egg, as they were when the imaginal form was attained, since there
doubtless was little difference between the two stages except in size.
Now, in the development of such primitive insects into hemimetabolic
forms, and the development of these into holometabolic forms, it has

370 bulletin: museum of comparative zoology.

come about that the imaginal form is exceedmgly diflFerent from the
larval. This has necessitated great changes in the muscular system.
It is easy to see that iu this evolution many muscles must have reached
a stage where, if they were to be useful in the imago, they must be
stronger, or their attachments must be shifted, or they must be changed
iu some other manner, which would necessitate a greater or less meta-
morphosis. In this metamorphosis nothing could be more probable
than that there should be, first, a proliferation of the nuclei, second, a
longitudinal splitting of the original fibre into as many new fibres as
were needed, and, if an extensive metamorphosis was required, a de-
struction of the original fibrillae and the formation of new fibrillae by
the undifferentiated sarcoplasm remaining. Such is the metamorphosis
which has been described in the present paper for Coleoptera, and I can
conceive of nothing simpler or more probable.

The presence of degenerating muscles is quite as easily explained.
In the development of holometabolic insects, it must have happened
many times that a muscle which was useful in the larva became function-
less in the imago. It is evident that the ultimate fate of such a muscle
would be degeneration at the end of larval life. The method of degen-
eration might be different iu different cases, but no one can deny suc-
cessfully that such muscles would exist, though Berlese has attempted to
do so. The converse of this might also be expected, that is, muscles
which are useful in the imago but functionless in the larva. Such
muscles would tend naturally to be retarded in their development until
they came to be muscles newly formed in the pupa ; but in their final
development they would arise from the cells which had previously
formed them. How it could come about that these muscles of new
formation in the pupa should be developed from cells furnished by the
degenerating muscles of other jmrts of the body, as Berlese states, is
something which I cannot understand.

From what has been said, it is evident that there is little doubt as to
the incorrectness of Berlese's main idea in other groups of insects, as
well as in Coleoptera.

Needham's ( :00) statement that the nuclei of fat cells become associ-
ated with the developing muscles, does not seem probable. The develop-
ment of such highly specialized cells into a tissue of such an entirely
different nature, is an exceedingly rare phenomenon. Nothing that
would indicate such a development has been seen in the present
study.

breed: metamorphosis of the muscles of a beetle. 371

Summary.

During the metamorphosis of the larvae of Coleoptera into the
imagines, some of the larval muscles remain unaltered during the meta-
morphosis, a few degenerate, while many metamorphose into imaginal
muscles. Imaginal muscles are formed in the pupa from cells of an
embryonic nature, but they are few in number.

I. Anatomical.

1. The muscles which remain unaltered by the metamorphosis are all
found in the abdominal region. They compose the inner layer of the
antero-posterior muscles, and the inner muscles of the dorso-ventral
intersegmental muscles. Exceptions to this statement are found in the
first and last abdominal somites, where muscles occupying these positions
are found to degenerate. This is explained by the greater changes of
external form which these somites undergo.

2. The typical degenerating muscles are found in the thorax and
the abdominal somites just mentioned. They occupy positions in these
somites serially homologous to the positions of the persistent larval
muscles of the abdomen. There are some cases of the degeneration of
dorso-veutral muscles other than intersegmental muscles. These were
noticed especially in mesothoracic muscles whose counterparts in the
metathorax metamorphose into imaginal muscles. Their histological
changes show transitional stages between metamorphosing and degenerat-
ing muscles. The muscles which show these conditions are such as
would be functional in the adult, if the elytra were used as organs of
flight, as presumably was the case in the ancestors of beetles.

3. Imaginal muscles of new formation in the pupa are not very com-
mon, only two somewhat questionable cases having been observed in
Thymalus. In Bruchus and other forms with legless larvae, the leg
muscles belong to this class.

4r. The metamorphosing larval muscles are by far the most numerous,
and include all of the remaining larval muscles. In general, these are
the muscles of the head, the peripheral layers of the hypodermal muscles,
and the intestinal muscles. There is a metamorphosis of larval muscles
into imaginal muscles of both the wing and the leg types.

II. Histological.

1. The fibres of the larval muscles which pass unaltered from the
larva to the imago, present the usual structure of this type of muscle

372 bulletin: museum of comparative zoology.

fibre. Each muscle is composed of a few fibres whose nuclei are
placed at the surface of the fibre in an abundant sarcoplasm. They show
a well-marked sarcolemma and evident cross and longitudinal striations.
The intracellular tracheoles which supply the muscles apparently pene-
trate the sarcolemma and ramify in the superficial layer of the sarcoplasm.

2. The larval muscles which metamorphose into muscles of the iving
type begin their metamorphosis at an early stage of the resting larva.
The metamorphosis consists of (1) a longitudinal division of the original
fibre into from four to ten fibres, (2) the destruction of the fibrillae of the
larval muscles, and the formation of the larger separate fibrillae of the
imaginal muscles in the remaining structureless sarcoplasm, and (3) a
great increase in the number of the nuclei, which become redistributed
throughout the substance of the muscle. All of the muscles of this type
increase in size during these changes. At an early stage in the meta-
morphosis, mesenchymatous cells derived from the intracellular tracheoles
make their appearance between the newly divided fibres. These cells
increase rapidly by mitotic division, and, in a late stage of the pupa, form
the abundant new tracheoles which supply these muscles in the imago.
Possibly some of these mesenchymatous cells become imaginal leucocytes.

3. The metamorphosis of the larval muscles into muscles of the leg
type does not differ essentially from that of muscles of the wing type.
The principal difference is that the muscles of the leg type divide into
smaller fibres, and a greater number of them, fifteen to twenty fibres
being frequently formed by this division. The nuclei divide frequently
by amitosis, and in the redistribution may take either of two positions in
the new fibres. They may come to lie at the periphery, as in Thymalus,
or in a row along the axis of each fibre, as in Bruchus. There is in
different muscles a great variation in the time of the beginning of this
metamorphosis. Some begin their changes as early as those which meta-
morphose into imaginal muscles of the wing type ; others begin their
changes at various periods during the resting larva; while a few show
scarcely any evidence of metamorphosis, even at the time of pupation.
It is barely possible that in the muscles last mentioned some of the
fibrillae of the larval muscles may persist as fibrillae of the imaginal
muscles. This cannot be commonly the case, however. In the region
of the leg muscles the mesenchymatous tracheal cells are not as nu-
merous as in the wing muscles, and the tracheae developed from them
do not penetrate the substance of the muscle fibres.

4. The metamorphosis of the intestinal muscles is later in starting
than tliat of any of the other muscles. Not until well along in pupal

BREED: METAMORPHOSIS OF THE MUSCLES OF A BEETLE. 373

life are the fibrillae of the larval muscles entirely dissolved. There
seems to be no increase in the number of muscle fibres by longitudinal
division, and the nuclei were not observed to divide amitotically, as in
the other metamorphosing muscles. The usual tracheal cells are found
accompanying these muscles.

5. The degeneration of the larval muscles is entirely chemical, there
being no evidence of phagocytosis. In the early pupa, there com-
mences a gradual atrophy of the muscle substance, during which the
muscle is partially divided into longitudinal strands. The nuclei show
no evidence of degeneration until practically all other parts of the
muscle have disappeared. They then undergo a typical chromatolysis.
This happens in the late pupa. Occasionally, tracheal cells are found in
the fissures formed by the breaking up of these muscles.

In those cases which presented transitional conditions between degen-
eration and metamorphosis, the muscles underwent changes exactly
similar to those of the metamorphosing muscles, until the stage was
reached where the reconstructive changes begin. Then the degenerating
muscles seemed to lack the stimulus to start this reconstruction, and,
therefore, continued to atrophy, and finally disappeared at the same time
and in the same manner as the more typically degenerating muscles.

6. The histological changes of the muscles of new formation in the
pupa were observed principally in the leg muscles of Bruchus. These
muscles are formed from spindle-shaped mesoderm cells found in the
larva at the bases of imaginal folds which represent the legs. These
cells probably are derived from the embryonic mesoderm. In the
young pupa these mesoderm cells form the muscle fibres, each cell possibly
giving rise to a single fibre. In the youngest stage in which the muscle
fibres can be distinguished with certainty, it is evident that there are
two kinds of cells in this mass : one, the mesoderm cells which form
the muscle fibres ; the other, tracheal cells which form the tracheae of the
leg. The latter are presumably derived from the same source as the
tracheal cells of the rest of the body, that is, from the intracellular
tracheoles of the resting larva. These cells may be distinguished as
mesenchyme.

III. Additional.

1. Incidentally some other points have been noted. The musculns
episternalis of the metathorax, whose function former authors had sug-
gested to be that of an expiratory muscle, was discovered not to have
this function. In the imaginal form of Thymalus, the pair of episternal

374 bulletin: museum of comparative zoology.

muscles lie in such positions that their contraction depresses the folds on
the metaepisterni into which ridges on the elytra catch when these are
closed. This depression of the folds releases the elytra, or, if these are
open, it allows them to be closed.

2. Phagocytosis of the muscles of Coleoptera does not exist. No
" Kornchenkugeln " have been found, though leucocytes containing
what are evidently foreign bodies have been found in Thymalus. These
inclusions are possibly to be explained as intracellular parasites.

breed: metamokphosis of the muscles of a beetle. 375

BIBLIOGRAPHY.

Amans, P. C.
'85. Comparaisons des organes du vol dans la serie auimale. Ann. Sci. Nat.,
ZooL, Ser. 6, Tome 19, pp. 9-222, pi. 1-8.

Anglas, J.
'99. Sur I'histolyse et I'histogenese des muscles des Hymenopteres pendant
la metamorphose. Comp. Rend, et Mem. Soc. Biol, Paris, Ser. 11, Tome
51, pp. 931-933.
Also in Bull. Soc. Entomol. Prance, Annee 1899, pp. 348-350.

Anglas, J.

'99*. Sur I'bistogenese des muscles imagiuaux des Hymenopteres. Comp.
Rend, et Mem. Soc. Biol., Paris, Ser. 11, Tome 51, pp. 947-949.

Anglas, J.

:00. Sur la signification des termes " phagocytose " et " lyocytose." Comp.
Rend, et Mem. Soc. Biol., Paris, Ser. 11, Tome 52, pp. 219-221.

Anglas, J.

: 01. Observations sur les metamorphoses internes de la Guepe et de rAbeille.
Bull. Sci. Prance et Belgique, Tome 34, pp. 363-473, pi. 19-23, 8 fig.

Anglas, J.

: 01". Quelques caracteres essentiels de I'histolyse pendant la metamorphose.
Bull. Soc. Entomol. Prance, Auuee 1901, pp. 301-304.

Anglas, J.

:02. Nouvelles observations sur les metamorphoses internes. Arch. d'Anat.
Micr., Tome 5, pp. 78-121, pi. 4, 1 fig.

Aubert, H.

'53 Ueber die eigenthiimliche Structur der Thoraxmuskeln der Insecten.
Zeit. f. wiss. Zool., Bd. 4, pp. 388-399, Taf. 15.

Berlese, A.

'99. Osservazioni su fenomeni cbe avvengono durante la ninfosi degli insetti
metabolici. Parte I. Tessuto adiposo. Rivista Patolog. vegetale. Anno
8, pp. 1-155, tav. 1-6, 42 fig.

376 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

Berlese, A.
:00. lutoruo alle modificazioni de alcuni tessuti durante la ninfosi della Calli-
phora erythrocepbala. Bull. Soc. Eatomol. Ital., Auno 32, pp. 253-288,
7 fig.

Berlese, A.
: 00*. Considerazioiii sulla Fagocitosi negli insetti metabolici. Zool. Auz.,
Bd. 23, pp. 441-M9.

Berlese, A.
:01. Vorgauge, welche ■\\'aiirend der Nymphosis der metabolischen lusecten
vorkommen. Zool. Aiiz., Bd. 24, pp. 515-521.

Berlese, A.
: 02. Osservazioni su feuomeui die avveiigouo duraute la uiiifosi degli iusetti
metabolici. Parte I. Tessuto adiposo, Memoria socouda. lUvista Patolog.
vegetale. Anno 9, pp. 177-344, tav. 7-10, 53 fig.

Berlese, A.

:02''. Osservazioni su fenomeui clie avvengono durante la ninfosi degli insetti
metabolici. Parte II. Tessuto musculare. Rivista Patolog. vegetale,
Auno 10, pp. 1-120, tav. 11-14, 4 fig.

Beutenmueller, W.
'90. Descriptiou of the larva of Thymalus fulgidus. Entomol. Amer., Vol.
6, p. 57.

Bruyne, C. de.

'97. Sur I'iutervention de la pbagocytose dans le developperaent des inverte-
bres. Arch, de Biol., Tome 15, pp. 181-300, pi. 7-11.

Also in Mem. cour. et Mem. savants etraugers Acad. roy. de Belgique,
Tome 56, pp. 1-114, pi. 1-5, 1898.

Cajal, R. y.
'88. Observations sur la texture des fibres musculaires des pattes et des ailes
des insectes. Internat. Mouat. f. Anat. u. Physiol., Bd. 5, pp. 205-232,
253-270, Taf. 19-22.

Chittenden, F. H.

'99. Insects injurious to beans and peas. Yearbook U. S. Dept. Agri. for
1898, pp. 233-260, 17 fig. in text.

De Bruyne, C. See Bruyne, C. de.

Deegener, P.

:00. Entwicklung der Mundwerkzeuge und des Darmkanals von Hydrophilus.
Zeit. f. wiss. Zool., Bd. 08, pp. 113-168, Taf. 8-10.

Ganin, M.

'76. (The postembryonic development of insects.) (Russian.) Arbeit, d.
V. Versammluug russ. Naturf. u. Artze iu Warschau.

Reviewed in Zeit. f. wiss. Zool., Bd. 28, 1877, pp. 386-389.

breed: metamorphosis of the muscles of a beetle. 377

Hammond, A. R.

See Miall, L. C, and Hammond, A. R.

Heidenhain, M.
'98. Struktur der kontraktilen Materie. 1. Struktur der quergestreiften
Muskelsubstanz. Auat. Hel'te, Ergeb. d. Anat. u. Eutwick., Bd. 8, pp.
3-111, 19 Textfig.

Hurst, C. H.
'90. The pupal stage of Culex. An inaugural dissertation for the degree of
Pb. D. in the Univ. of Leipzig. 26 pp., 1 pi, Manchester.

Karawaiew, 'W.

'97. Vorlaufige Mittheiluug iiber die innere Metamorphose bei Ameisen.
Zool. Anz., Bd. 20, pp. 415-422.

Karawaiew, W.
'98. Die naohembryonale Eutwicklung von Lasius flavus. Zeit. f. wiss.
Zool., Bd. 64, pp. 385-478, Taf. 9-12, 15 Textfig.

Karawaiew, W.

'99. Ueber Anatomie und Metamorphose des Darmkanals der Larve von
Anobium paniceum. Biol. Ceutralbl., Bd. 19, pp. 122-130, 161-171, 196-
202, 19 Textfig.

Kellogg, V. L.
-.01. Phagocytosis in the postembryonic development of the Diptera. Amer.
Nat., Voi. 35, pp. 363-368, 2 fig. in text.

Korotneff, A.

'92. Histolyse und Histogenese des Muskelgewebes bei der Metamorphose
der Insekten. Biol. Ceutralbl., Bd. 12, pp. 261-265, 5 Textfig.

Kowalevsky, A.
'85. Beitrage zur nachembryonalen Eutwicklung der Musciden. Zool. Anz.,
Jahrg. 8, pp. 98-103, 123-128,153-157.

Kowalevsky, A.
'87. Beitrage zur Kenntniss der nachembryonalen Eutwicklung der Musciden.
Zeit. f. wiss. Zool., Bd. 45, pp. 542-594, Taf. 26-30.

Kriiger, E.

'98. Ueber die Entwicklung der Pliigel der Insekten mit besonderer Beruck-
sichtigung der Deckfliigel der Kafer. Inaugural-Dissertation, Uuiv. Got-
tingen. 60 pp., 19 Textfig. Gottingen.

Kiinckel d'Herculais, J.

'72. Sur le developpement des fibres musculaires striees chez les insectes.
Coinp. Bend. Acad. Sci., Paris, Tome 75, pp. 359-362.

Kiinckel d'Herculais, J.
'75. Recherches sur Torgaiusation et le d6veloppement des Volucelles. vi+
208 pp., 26 pi., Paris.

378 bulletin: museum of comparative zoology.

L^veille, A.
'88. Catalogue de la famille des Temnocliildes. Aua. Soc Eutomol. Prauce,
Ser. 6, Tome 8, pp. 429-448.

Lowne, B. T.

'90-95. The Anatomy, Physiology, Morphology, and Development of the
Blow-fly (CaUiphora erythrocephala). Vol. 1, ix+350 pp., 21 ])1., 47
fig. in text; Vol. 2, viii+428 pp., 30 pi., 60 fig. in text. Loudon.

Luks, C.
'83. Ueber die Brustmuskulatur der lusecten. Jena. Zeit., Bd. 16, pp. 529-
552, Taf. 22-23.

Miall, L. C, and Hammond, A. R.

'92. The Development of the Head of the Imago of Chironomus. Trans.
Linn. Soc. London, Ser. 2, Vol. 5, pp. 2G5-279, pi. 28-31.

Miall, L. C, and Hammond, A. R.

:00. The Structure and Life-history of the Harlequin Fly (Chironomus).
vi-hl96 pp., 1 pi., 129 fig. in text. Oxford.

Needham, J. G.

:00. Some general Features of tlie Metamorphosis of the Flag Weevil, Mon.
onychus vulpeculus Fabr. Biol. Bull., Vol. 1, pp. 179-191, 10 fig. in text.

Noetzel, W.
'98. Zur Kenntniss der Histolyse. Arch. f. path. Anat. u. Physiol., Bd. 151,
pp. 7-22.

P6rez, C.
'99. Sur la metamorphose des insectes. Bull. Soc. Entomol. France, Anuee
1899, pp. 398-402.

P6rez, C.

:00. Sur I'histolyse musculaire chez les insectes. Comp. Rend, et M^m.
Soc. Biol. Paris, Ser. 11, Tome 52, pp. 7-S.

Petri, L.
'99. I muscoli delle ali nei ditteri e uegli imenotteri. Bull. Soc. Eutomol.
Ital., Anno 31, pp. 3-45, tav. 1-3, 9 fig.

Rengel, C.
'96. Uber die Veranderungen des Darmepithels bei Tenebrio molitor wahrend
der Metamorphose. Zeit. f. wiss. Zool., Bd. 62, pp. 1-60, Taf. 1,

Rollett, A.

'85. Untersuchungen iiber den Bau der quergestreiften Muskelfasern.
Denkschr. Kaiser. Akad. Wissen. Wien, Bd. 49, pp. 81-132, 4 Taf.

Straus-Diirckheim, H.
'28. Considerations gene rales sur ranatomie comparee des animaux articules.
xix -I- 434 -f 36 pp., 9 pi. Paris.

' breed: metamoephosis of the muscles of a beetle. 379

Terre, L.
'99. C oatribatiou a I'etude de I'histolyse et de Thistogeaese des tissus miiscu-
laire cliez I'Abeille. Comp. Reud. et Mem. Soc. Biol Paris, Ser. 11,
Tome 51, pp. 896-898.
Also ia Bull. Soc. Eatomol. Erance, Auuee 1899, pp. 351-352.

Terre, L.
-.00 S 111- I'histolyse muse ulai re des Hymeuopteres. Camp. Rcud. et Mem.
Soc. Biol. Paris, Ser. 11, Tome 53, pp. 91-93.
Also ia Ball. Soc. Eiitomol. Erauce, Amies 1900, pp. 23-25.

Terre, L.

:00^. Metamorphose et pingocytose. Comp. Reud. et Mem. Soc. Biol.
Paris, Ser. 11, Tome 52, pp. 158-159.

Van Rees, J.

'84 Ov^er iutra-cellulaire spijsverteering en over de beteekenis der witte
bloedlicliaampjes. Maaudblad voor Natuurwetenschappeu, Jaarg. 11,
28 pp.

Van Rees, J.

'83. Bjitraga zur Kenutuiss der iuuereu Metamorphose von Musea vomitoria.
Zool. Jahrb., Abth. f. Auat. u. Ontog., Bd. 3, pp. 1-131, Taf. 1-2, 15
Textfig.

Vaney, C.

:00. Contributions a I'etude des pheuomMies de metamorphose chez les
Dipteres. Comp. Reud. Acad. Sci., Paris, Tome 131, pp. 753-761.

Viallanes, H.
'81. Sur I'histolyse des muscles de la larve durant le developpement postem-
bryonnaire des Dipteres. Comp. Read. Acad. Sci., Paris, Tome 92, pp.
416-418.

Viallanes, H.
'82. Recherches sur 1' histologic des insectes et sur les phenomenes histolo-
giques qui accompagnent le developpement post-embryonuaire de ces
auimaux. Ann. Sci. Nat., Zool., Ser. 6, Tome U, pp. 1-348, pi. 1-18.

Weismann, A.
'62. Ueber die zwei Typen contractilen G^webes und ihre Vei-theilung inder
grossen Gruppen des Thierreiclis, sowie iiber die histologische Bedeutuug
ihrer Eormelemente. Zeit. f. ratiou. Med., Bd. 15, pp. 60-103, Taf. 4-8.

Weismann, A.
" '64, Die uachembryouale Eutwicklung der Musciden nacli Bsobaclituugen an
Musca vomitoria und Sarcophaga carnaria. Zeit. f. wiss. Zool, Bd. 14,
pp. 187-336, Taf. 21-27.

Weismann, A.
'65. Die Metamorphose der Corethra plumicoruis. Zeit. f. wiss. Zool., Bd.
16, pp. 45-132, Taf. 3-7.

380 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

EXPLANATION OF PLATES.

All figures were drawn with the aid of the camera lucida from preparations of
Thi/malus mar(jinicollis Chevr. The magnifications are given witii the descriptions
of the several figures.

In Plates 1-5, Figures 1-5, 7, 0, and II are drawn from reconstructions of serial
sections. They form two series of figures illustrating the anatomical clianges of
the dorsal antero-posterior (Figs. 1, 2) and lateral dorso-ventral (Figs. 3-5, 7, 9, II)
groups of metatlioracic muscles during metamorphosis. These figures are all
magnified 67.5 diameters.

ABBREVIATIONS.

aa. Colin .... Cohnheira's areas.

al Wing.

cd.n Nerve cord.

chr Chromatin masses left after tlie disintegration of nuclei.

cl.mit Tracheal cells in stages of mitotic division.

cl. tr Traclieal cell.

cl. tr.^ Tracheal cells showing connections with tracheae.

cl. tr.'^ Tracheal cells whose connections with the tracheae have

been severed, but which show tracheoles througli their

cytoplasm.

cl. tr.^ Tracheal cell entirely embedded in the muscle.

cp. adp Fat body.

cr Heart.

eta Cuticula.

dep. trg Depressor tergi.

elif Elytron.

e'stn Musculus episternalis.

e'th Epithehal lining of the foreintestine.

ext. al. mag. mt'thx. . Extensor alae magnus metathoracis.

ext.al.pa.mt'thx. . E.xtensor alae parvus metathoracis.

ext. cox. mt'thx. (^-3) Extensor coxae metathoracis (primus, secundus, tertius).

ext. trchn. int'tkr. . . Extensor trochanteris metathoracis.

Jlx. al, mt'thx. (1-^) Flexor alae metathoracis (primus, secundus, tertius).

Jlx, cox. mt'thx. (-^-5) Flexor coxae metathoracis (primus, secundus, tertius, quat-

tuor, quintus).

breed: metamorphosis of the muscles of a beetle. 381

fix. pre. p-J. mt'fur. . Flexor processus postero-lateralis metafurcae.

Jix. trchn. mCthx. . . Flexor troehanteris metathoracis.

htfdrm Hypodermis.

in Intestine.

lexi'cyt Leucocyte.

I. vis' far Musculus lateralis raesofurcae.

;. mVnt Musculus lateralis metanoti.

/. mVthx. a. ... Musculus lateralis metathoracis anterior.

/. mt'thx. p. ... Musculus lateralis raetatlioracis posterior.

loph Cross section of ridge on elytron.

l.prc.if.ms'phg. . . Musculus lateralis processus inferioris mesophragmatis.

ms'fur Mesofurca.

ms'fitr. d Musculus raesofurcae dorsalis.

tns'phg Mesophragma.

mffur Metafurca.

mt'nt Musculus metanoti.

mt'phg Metaphragma.

mu. crc Circular layer of intestinal muscles.

mu. Ig Longitudinal layer of intestinal muscles.

n Cross section of the main branch of the sympathetic nervous

system.

nl Nucleus of larval muscle fibre before division.

jj^.i Nucleus of muscle fibre undergoing amitotic division.

n/.2 Pairs of nuclei resulting from amitotic division.

nL* Elongated nucleus common in metamorphosing muscles.

nZ.* Nucleus of degenerating muscle undergoing chromatolysis.

n/.5 Nucleus of leucocyte.

pJi Cross section of fold on episternum.

pre Processes of tracheal cells detached from cell body by the

plane of the section.

pre. ms''phg. if. . . Processus mesophragmatis inferior.

pre. mt'phg. if. . . Processus metaphragmatis inferior.

rlx. al. mt'lhx. . . . Relaxator alae metathoracis.

rlx. ext. al Relaxator extensoris altfe.

rtr. ms'thx. if. . . . Retractor mesothoracis inferior.

rtr. prothx. if. . . . Retractor prothoracis inferior.

sar'lem Sarcolemma.

sar'pl Sarcoplasm.

sty. ah. 1 .... Stigma of the first abdominal somite.

stg. mt'thx Metathoracic stigma.

sut. a Suture of the larval metathorax, probably equivalent to the

suture between prescutum and scutum.

sut. p Suture probably equivalent to the suture between the scutum

and scutellum.

tr Trachea.

trl Intracellular tracheole.

a, ;S, y, 5, €, etc. . . Larval muscles which degenerate during pupal life.

1 Anterior lateral horn of the metafurca.

382

BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY.

2 Posterior lateral horn of the metafurca.

3 Lateral wing of the metafurca.

4 Median lamina of the metafurca.

The >|< is used in Figure 13 to indicate the place where teeth on the inner sur-
face of the elytron interlock with teeth on the outer surface of the thorax, thereby
holding the elytron in position.

The table given below shows in a comprehensive manner the relative develop-
ment of all of the animals used in making drawings. Where figures are bracketed
together, all of the figures embraced in the bracket were drawn from the same
animal. In all, twenty-three specimens were used in making the fifty-three
figures.

Feeding

Resting Larva

Pupa.

Imago.

Larva.

Young. Old.

Yuung.

Old.

Fig. 1

Fig. 6

Fig. 2

Fig. 8

frig. 21

- Fig. 32
(Fig. 40

(Fig. n

] Fig. 11

(Fig. 5
)Fig. 7

<Fig. 3
iFig. 4

Fig. 14

Fig. 35

\ Fig. 13
\ Fig. 15

Textfig. 1

Fig. 12

Fig. 38

Fig. 10

fFig. 25
J Fig. 28
iFig. 51

\ Fig. 43

Fig. 45

( Fig. 20
\ Fig. 36

(Fig. 16
1 Fig. 22

iFig. 48

( Fig. 19
\ Fig. 39

fFig. 20-30

' Fig. 26
iFig. 33

/■Fig. 24
Fig 27

Fig. 41-42

(Fig. 17
1 Fig. 18

-

Fig. 44

" Fig. 34

Fig. 46-47

- Fig. 31
[Fig. 50

Fig. 23

.Fig. 49

Fig. 37

[Fig. 52

Brked. — Muscle MetamorphoBis.

PLATE 1.

All of the figures magnified 67.5 diameters.

Fig. 1. Dorsal view of the dorsal antero-posterior muscles of the left side of the

metathorax of a, feeding larva. Anterior is up on the plate.
Fig. 2. Yoiuuj pupal stage of the muscles shown in Fig. 1. Similar view.
Fig. 3. Deeper layer of the lateral dorso-ventral muscles of tlie left side of

the metathora.x of a feeding larva seen in lateral aspect. Anterior at

the left.
Fig. 4. Superficial layer of the group of muscles whose deeper layer is shown in

Fig. 3.

Breed.— Muscle Metamorphosis.

Plate. I.

mt'nt.

/

l.mt'thx.p.
l.prc.if. i^fnVthX.a. ; sut.a.

ms'phg.

l.mVnt

/.A

n.e.a,r.m.t'thx.i Tf i.mrnt. g 7

xut.a. sut.p.

-^

ex.t.cox.mVthx.H

dep.trg.

rlx.ext.al.1

flx.prc.p-l.

™''^"^- rlx.al
■mVthx.?
ext.al.pa.
int'thx.

flx.al.mt'thx. 1^

\

flx.eox.
mt'thx-S

l.ms'fur.

ext.al.mad.
mVthx.

ext.cox.
' mVthx.l

flx.eox.
mt'tlix.l

RSi.del.

Breed. — Muscle Metamorphosis,

PLATE 2.

Fig. 5. Superficial layer of the metathoracic lateral dorso-ventral muscles of the
left side of a young pupa as they would appear with the lateral wall of
the metathorax removed. Anterior at the left. X G7.5.

Fig. 6. Side view of the resting larva of Thymalus. X 13.

Breed —Muscle l\feTAMORPHosis.

flx.al.mt'thx.l^ ext.cox.mVfhx.S

Plate 2.

Ix.ext.al. 'jSf-
. mVtIix.W

l.mt'thx.

mt'thx.
il.maij.mVthx.

ms'/wr.
c.cox.ml'thx.4

stg.ab.l

ext.al.pa. mVthx.

Bbeed. — Muscle Metamorphosis.

PLATE 3.

Fig. 7. Young pupa. Deeper layer of the group of muscles whose superficial layer

is shown in Fig. 5. X 67.5.
Fig. 8. Side view of the /"//w of Thymalus. X 13.

Breed — Muscle Metajv^of^hosis.

l.mt'thx. p. flr.rM.mfthr. J.

-:.-■"'-: 3

mt't)i.r. «■

.if. ms'i>luj

/Jcp'trp.

,fl.r.prc.p-I.
mt'fiir.

stg.a h. 1

' rt.e.rir.mt'th.r. •>.
'I.r.ct>.r.mt'th.r. 5

flx.cox.int'thx. i

I rt.trchn.

rt.eox. mi'lhx.i

^ ^

R 5; R ,^f>l

i

Breed. — Muscle Metamorphosis.

PLATE 4.

Both figures magnified G7.5 diameters.

Fig. 9. Superficial layer of the inetathoracic muscles of the left side of an imago

as they would appear with tlie lateral wall of the metathorax removed.

Anterior at the left.

Fig. 10. Portion of a cross section of tlie metathora.\ of a larva showing the

cross section of the ventral anteroposterior muscles. Dorsal up on

the plate.

Breed. — ^Muscle Metamorphosis.

Plate 4.

Bkbbo. — Muscle Metamorpliosis.

PLATE 5.

Both figures magnified 67.5 diameters.

Fig. 11. Imago. Deeper layer of the muscles whose superficial layer is shown in

Fig. 9.
Fig. 12. Portion of a cross section of the metathorax of a pupa showing the cross

section of the ventral antero-posterior muscles. Dorsal up on the

plate. Compare with Fig. 10.

Breed.— Muscle Metamorphosis.

Plate 5.

R.S.B. DEL.

Bbebd. — Muscle Metamorphosis.

PLATE 6.

Fig. 13. Posterior face of lateral (right) portion o- cross section of the meta-
thorax of an imago showing tiie parts affected by the contraction of
musculus ei)isternalis [esln.). X loO.

Fig. 14. Cross section of the largest fibre of musculus metanoti. Drawn from a
resting larva about midway in its development. X 800.

Fig. 15. Cross section of that portion of musculus metanoti which has been de-
rived from the largest fibre of this muscle in the larva. Drawn from
an imago. Compare Fig. 14. X 800.

Fig. 16. Cross section of a functional larval muscle fibre. Feeding larva. X 800.

Fig. 17. Longitudinal section of a fibre of retractor mesothoracis inferior. Drawn
from an imago. X 1600.

Fig. 18. Cross section of a fibre of fle.xor alae metathoracis secundus drawn from
tlie same series of sections. X 1600,

Fig. 19. Cross section of flexor coxae metathoracis secundus. Drawn from a
young jmpa. X 800.

Fig. 20. Cross section of a circular muscle fibre of the foreintestine of an imar/o.
X 1600.

Fig. 21. Cross section of three fibres of flexor coxae metathoracis secundus.
Taken from an old pupa. Compare Fig. 19. X 800.

Fig. 22. Cross section of a functional larval muscle fibre. Feeding larva. X 800.

Figs. 23-32. Of these figures, Figs. 23-25, 30 and 31 form a series of longitudinal
sections, and Figs. 26-29 and 32 a series of cross sections, of small por-
tions of muscle fibres of the wing type. These drawings illustrate the
changes in the finer structure of these muscles during tlieir metamor-
phosis. All of tlie figures are magnified 1600 diameters.

Fig. 23. Feeding larva. ■ Longitudinal section of part of a functional fibre.

Fig. 24. Resting larva. Longitudinal section of part of musculus metanoti.

Fig. 25. Resting larva a few hours before pupation. Longitudinal section of part
of musculus lateralis metathoracis anterior.

Fig. 26. Feeding larva. Cross section of part of a functional fibre.

Fig. 27. Resting larva. Cross section of part of flexor coxae metathoracis
secundus.

Fig. 23. Resting larva a few hours before pupation. Cross section of part of mus-
culus metanoti.

Fig. 29. Midway pupa. Cross section of part of musculus lateralis metathoracis
posterior.

Fig. 30. Midway pupa. Longitudinal section of part of musculus metanoti.

Fig. 31. Young imago. Longitudinal section of part of musculus metanoti.

Fig. 32. Old pupa. Cross section of part of extensor alae metathoracis.

Breed —Muscle Metamorphosis.

Plate (

Breed. — Muscle Metamorphosis,

PLATE 7.

Fig. 33. Longitiulinal section of a functional muscle fibre. Feeding lan'a. X 800.

Fig. 34. Longitudinal section of the largest of the fibres of musculus metanoti.
Taken from a resting larva. X 800.

Fig. 35. Longitudinal section of a portion of musculus metanoti. Taken from an
old pupa. X 800.

Fig. 86. Longitudinal section of a part of flexor coxae nietatlioracis secundus.
Drawn from an iinarp. X 800.

Fig. 37. Longitudinal section of one of the degenerating larval muscles of the dor-
sal antero-posterior group in the mesothorax. Drawn from a jjonng
pupa. X 800.

Fig. 38. Remains of the degenerating larval muscles €, tj (see Fig. 1). Drawn
from an old pupa. X 800.

Fig. 39. Cross section of the degenerating larval muscles €, tj. Drawn from a
i/oitng j>Hpa. X 800.

Figs. 40-48. Leucocytes containing foreign bodies, all of them being taken from
old pnpae. X 1600.

Fig. 49. Cross section of flexor alae nietatlioracis secundus. Drawn from a rest-
ing larva. X 800.

Fig. 50. Cross section of the same muscle in the imago. X 800.

Fig. 61. Cross section of a part of tlie wall of the proventriculus of a larva about
to pupate. X 1200.

Fig. 52. Dorsal part of a cross section of tlie proventriculus of an old pupa.
Ventral is uppermost on tiie plate. X 1200.

Brred — Mn.s'-LE Metamorpho^t?,

/***^,

Plate 7

5'^

,H,,1 fa, "^\'

«:■'•) p .''^^••■ «

i "f

f

I«u'ovt.

t. ^

i' I

1*1 Pi

».?. , , .., V'l I. ;."i -' ■ ' f,

*■

.^ %

S!,

^

*^ ^&'%

^^

m

^

# <r-

;5W

ji>.^

> *;

/</

©

% 0^'Q^ ^""''

"^'Kk^

300 S/.0/-

BOOKBINDING CO., INC.
NOV 2 9 1983

100 CAMBRIDGE STREET
CHARLE3T0WN. MASS.

Harvard MCZ Libra

3 2044 066 302 944

Date Due