BIOLOGICAL BULLETIN

KDITKI) BY

THE DIRECTOR
AND MEMBERS OF THE STAFF

'

OF

flDarine Biological laboratory

WOODS HOLL, MASS.

VOLUME II

BOSTON, U.S.A.

GINN & COMPANY, PUBLISHERS

CIjc SUbcnacttm |]rrt5G

1901

CONTENTS OF VOL II

No. i. — October, igoo.

PAGES

I. WM. MORTON WHEELER.

A Study of Sonic Texan Poncrinac \— 31

II. CARL H. EIGENMANN AND WINFIELD A.
DENNY.

The Eyes of the Blind Vertebrates of North
America. Ill 3 3-41

No. 2. -- November, igoo.

I. WM. MORTON WHEELER.

TJie Habits of Poncra and Stigmatomma . 43—69

II. EDWARD L. RICE.

Fusion of Filaments in tlic Lamellibranch

Gill 71-80

III. ADELE M. FIELDE.

Portable Ant Nests 81-85

No. 3. -- December, igoo.

I. R. R. BENSLEY.

The Oesophageal Glands of U rode la . . . 87-104

II. JOHANNA KROEBER.

An Experimental Demonstration of the
Regeneration of the Pharynx of Allolo-
bopliora from Endoderm 105-110

Ml

y

III. T. H. MORGAN.

Further Experiments on the Regeneration of

Tissue composed of Parts of Two Species . 111-119

IV. C. B. DAVENPORT.

Reviciv of Von Guaita 's Experiments in

Breedin Mice ......... 121-128

V. MARY HEFFERAN.

Variation in the Teeth of Nereis . . . 129-143

No. 4.- January, igoi.

I. MARTIN SMALLWOOD.

The Centrosome in the Maturation and

Fertilization of Bui la Solitaria . . 145-154

II. J. PLAYFAIR McMuRRicn.

Contributions on the Morphology of the
Actinozoa. VI. Halcurias Pilatus and
Endocoelactis ..... i55-l63

III. SAMUEL J. HOLMES.

Observations on the Habits and Natural

History of AmpJiithoe Longimana Smith 165-193

No. 5. -- February, igoi.

I. LEON J. COLE.

Notes on the Habits of Pycnogonids . . . 195-207

II. AIII.GAIL C. DIMON.

Experiments on cutting off Parts of the

Cotyledons of Pea and Nasturtium Seeds 209-219

III. CHAS. W. HARGITT.

Variation among Hydromedusae . . . . 221-255

No. 6. - June, igoi.

PAGES

I. EDWIN G. CONKLIN.

Tlic Individuality of tlie Germ Nuclei during

the Cleavage of the Egg of Crepidula . . 257-265

II. CHARLES ZELKNY.

The Early Development of the Hypophysis

in C lie Ionia 267—281

III. HARRY BEAL TORREY.

On Phoronis Pacifica (Sp. Arov.} .... 283-288

IV. ELIZABETH W. TOWLE.

On Muscle Regeneration in the Limbs of

Plethedon "... 289-299

V. T. H. MORGAN.

The Factors that determine Regeneration in

Antennularia 301-305

VI. C. B. DAVENPORT.

Mendel's Lazv of Dichotomy in Hybrids . . 307-310

VII. T. H. MORGAN.

Regeneration of Proportionate Structures in

Stentor 311—328

ALSO:

Abstracts of Papers presented at the Meetings of the
American Morphological Society at Baltimore,
December 27 and 28, I<JOO. Edited by the
Secretary of the Society 329-365

Volume //.] October, lyoo. [No. /.

BIOLOGICAL BULLETIN.

A STUDY OF SOME TEXAN PONERINAE.1

WILLIAM MORTON WHEELER.

THE observations hitherto published on the habits of the
Ponerine subfamily of ants are extremely meager. This gap
in our knowledge of ant life must be the more keenly felt by all
myrmecologists because the Ponerinae (including the Cera-
pachyi, a group allied to the Dorylinae, or driver ants) are,
to all appearances at least, the most generalized of existing
Formicidae. That the Ponerinae have a very primitive mor-
phological and social organization seems to be indicated by the
facts, first, that their colonies are composed of a relatively small
number of individuals - - like the incipient colonies of the more
specialized Myrmicine and Formicine ants, and, second, that the
polymorphism of the female is still, apparently, in an unstable
condition, i.e., the workers often differ little if any from the
queens, or females, in form and size, and the two phases are fre-
quently connected by individuals of an intermediate character
(ergatoid, or ergatomorphic females). It has also been sus-
pected by Professor Forel and Professor Emery that the breed-
ing habits of the Ponerinae would be found to be of a primitive
nature.

The common occurrence of three beautiful species of
Ponerinae in the environs of Austin, Texas, has induced me to
undertake the following study of their habits. The ants were
observed in a state of nature or kept in jars of earth or in

1 Contributions from the Zoological Laboratory of the University of Texas.

Director, W. M. Wheeler. No. 6.

i

WHEELER.

[VOL. II.

artificial nests of the Lubbock and of the Janet pattern. In col-
lecting material for these nests, I have been considerably aided
by two of my students, Mr. A. L. Melander and Mr. C. T.
Brues. Mr. Brues also aided me in drawing some of the figures

that accompany this
paper.

The three species
of Ponerinae which
I have studied are :
Odontomachus haema-
todes L. ; Pachycon-
dyla Jiarpax Fabr. (=
P oner a amp lino da
Buckley) ; and Lcpto-
genys elongata Buck-
ley.1 To European
myrmecologists both
O. haematodes and P.
harpax-dcco. well-known
species. They have
a wide geographical
distribution, occur-
ring from Texas
through Mexico, Cen-
tral America and Bra-
zil to Bolivia and Para-
guay. O. haematodes
is also recorded from Georgia, Florida, and the Antilles.2 In the
vicinity of Austin, P. harp ax and L. elongata are found under

1 Specimens of the last species were kindly identified for me by Mr. Pergande
as Lobopelta elongata Buck. ? The query seems to be due to Buckley's wretched
description of his Ponera elongata. Since Lobopelta is treated as a subgenus of
Leptogenys by Forel, and since the species is very common in Texas, and there-
fore, in all probability, the one described by Buckley, I have changed the generic
name and omitted the query. Should doubts arise, my figures (Fig. 4) will, I
trust, enable any future observer to recognize the species.

2 See Emery, " Beitrage zur Kenntniss der nordamerikanischen Ameisenfauna,"
Zool. Jahrb., Abth. f. Systematik., Bd. viii, pp. 268, 269 ; Patton, Amer. Nat., July,
1894, pp. 618, 619; Forel, Biologia Centrali-Americana, Hymenoptera, vol. iii,
April, 1899, p. 21.

FIG. i. — Odontomachus haematodes Linn. Worker.

No. i.] A STUDY OF SOME TEXAN PONERINAE. 3

stones or logs, on moderately dry hill-slopes, often in the shade
of trees or bushes, more rarely in the open fields. In these
same localities O. Jiaematodcs does not occur. The few nests
which I have seen belonging to this species were under stones
lying on a dry sandy loam in a different part of Travis County.
The exact distribution of these forms must be left for future
determination.

It is not an easy matter to ascertain the exact number of
individuals in any colony of the three species of Ponerinae.
When the stone or log that serves as a roof to the nest is
lifted, some of the ants always manage to escape, either into
the surrounding vegetation or into their burrows. This is
especially true of L. elongata and P. liarpax, both of which,
when excited, are very rapid in their movements. As a rough
estimate it may be said that the nests of L. elongata contain
from 10 to 50, those of P. Jiarpax from 15 to 100, of O. Jiaema-
todcs from 100 to 200 individuals.1

The nests of the three Ponerinae agree in being of a very
primitive structure. They consist of a few simple and irregular
burrows, or galleries, some of which run along the surface of the
soil immediately beneath the stone or log, while others extend
down into the soil obliquely or vertically to a depth of 8 or
10 inches. These burrows may anastomose, but they are not
widened at certain points to form chambers, as in the nests of
the more specialized ants (Atta, Pogonomyrmex, Camponotus,
etc.). Even in artificial nests of the Lubbock pattern the
Ponerinae dig only anastomosing galleries scarcely more than
a centimeter in diameter.

Owing to the interest which attaches to the sexual phases
of the Ponerinae, considerable effort was made to secure both
the winged and apterous forms of the Texan species. Of
O. haematodes only workers were seen, but as the full series
of phases of this species -- including the winged male, winged
female, and the apterous ergatoid female, as well as the worker

1 In tropical America (teste Forel, loc. cit., passim) P. harfax, race Montezumia,
Smith, O. haematodes, and the species of Leptogenys make their nests also in rotten
wood, and the first species is said to nest also in dry, hollow stems. The colonies
of O. haematodes are probably more numerous in individuals in the tropics than
at Austin.

//' 'HEELER.

[VOL. II.

-have been made known through the labors of European
myrmecologists, I fixed my attention on the other species. I
append a brief description of these species before recording
their habits.

P. Jiarpax (Figs. 2 and 3) is a large, rather robust, subopaque,
black ant, which carries its body close to the ground. It has a
peculiar habit of folding its antennae and of peeping out of
holes and crevices in the soil like a rat. In nearly all the nests

FIG. 2. — Pachycondyla harpax F'abr. Winged male and female.

that were opened only wingless individuals like the one repre-
sented in Fig. 3 were found. Except for a slight variation in
size they were all alike externally, and I at first regarded them
all as workers ; but on dissecting several of these ants during
the breeding season (from the ist of March to the end of May)
some of them were found to contain mature ova, while others
had abortive ovaries. In a nest consisting of twenty-nine indi-
viduals captured May 27, dissection revealed the presence of
mature eggs in thirteen, and one-third to one-half grown eggs
in four of the ants. The ants were also caring for several

o

No. i.] A STUDY OF SOME TEXAN PONERINAE. 5

packets of eggs which they had laid. Like the pigeons among
birds, P. karpax lays only two eggs at a time. I infer this from
the fact that only two eggs are found to be mature in the
ovaries. Since, in addition to these facts, wingless individuals
actually laid eggs in my artificial nests, I feel certain that forms

FIG. 3. — Pachycondyla harpax Fabr. Worker, dorsal and lateral aspect.

externally indistinguishable from the workers commonly func-
tion as females. These are, therefore, ergatoid or ergatomor-
phic females. There is not even a slight difference in size to
separate the two forms externally, since some of the ergatoid
females are smaller than some of the workers. There are,
however, in certain nests of P. harpax winged females of a less

6 WHEELER. [VOL. II.

worker-like aspect. These were seen only twice --once in a
large nest found under a log on March 12, and again in a nest
discovered in a very different locality by Mr. Brues, March 31.
In both nests there were many of these winged females
resembling Fig. 2, 9. They had small but distinct ocelli.
Several of them had lost some of their wings, while a slight
touch caused others to lose these appendages. Before finding
these winged females, I had found the winged males as early
as March 3 in a small nest containing ergatoid females and
workers. There were two of these males clinging to the

o o

under surface of the stone that covered the surface galleries
of the nest. They differ so much from the apterous females
and workers, that, if taken on the wing, they would scarcely be
regarded as ants (cf. Fig. 2, $, and Fig. 3). They are much
shorter and more slender than the workers, and their long,
slender, 13-jointed antennae are not geniculate, but straight,
with the basal and second joints short and the remaining
joints slender and subequal. The eyes and ocelli are very
large and prominent. These and other equally pronounced
structural peculiarities in the thorax and abdomen are shown in
Fig. 2, $. The small mouth-parts, the hypopygium, trochanters,
tibiae, and tarsi are honey-yellow ; the venter and sides of the
abdomen are piceous.1 On the 3ist of March I found four
nests containing winged males. Three of these nests each con-
tained three males, the remaining nest contained only two.

L. elongata is smaller and far more slender than P. Jiarpax,
with long delicate legs and antennae. The specific name is
very applicable not only to the adult ant but also to the egg,
larva, and pupa. The worker (Fig. 4, 9) is of a rich claret red
color, becoming yellow towards the tip of the abdomen. In
form, color, and movement this ant is certainly the embodiment
of elegance. The workers which were dissected --some thirty
specimens from different nests --contained only minute and
abortive ova, and for some time I failed to find the queen.

1 To my knowledge no description has hitherto been published of the male of
the typical P. harpax. I infer, however, from Forel's contribution (loc. cit., p. 12)
that Smith has described (Cat. ffym., vol. vi, p. 108, 1858) the winged male of his
P. monteziimia, a form which Forel regards as merely a Mexican race of P. harpax.

No. I.] A STUDY OF SOME TEXAN PONERINAE 7

Finally I noticed in each nest a single ant with a somewhat
larger abdomen than that of the workers, but, like them, with-
out any traces of wings. Dissection of three such individuals
disclosed mature eggs --only a single pair in each insect, as in
Pachycondyla. There could be no doubt that I had found the
hitherto unknown female of Leptogenys.1 They have no ocelli,

FIG. 4. — Leptogenys (Lobopelta) elongata Buck. Worker, winged male, and apterous female.

but the shape of the node and of the abdomen serves to distin-
guish them at once from the workers (see Fig. 4, j). They
are certainly ergatoid females, but as such they differ from the
ergatoicl females of P, harpax in several respects : First, they
resemble the workers less closely ; second, they occur singly in
the nests ; and, third, they have the status of true queens,

lrThus Emery's supposition (Emery, C., " Zur Biologie der Ameisen,"2?*<7/. Cen-
tralbl., Bd. xi, p. 174, 1891), that the females of this genus are apterous, is proved
to be true of at least one species.

8 WHEELER. [VOL. II.

because they are the only known egg-laying individuals of the
species. The winged, light-yellow male of L. clongata was de-
termined on circumstantial evidence, as it could not be found
in the natural nests, nor bred from the cocoons in the arti-
ficial nests. What seemed to be the males (Fig. 4, $) were
found during the autumn and spring flying about the lights in
the houses. They are somewhat smaller than the workers, have
very large eyes and ocelli, and the node is shaped like that
of the female. When placed in the artificial nests of L. clon-
gata they are not only not molested, but move about among
the workers as if they had always belonged among them. To
one who has witnessed the hostility of these ants to any ant of
a different species placed in their nest, this amicable reception
amounts to a demonstration of specific identity.

During the course of my observations I have had frequent
opportunity to observe the conduct of all three species of
Ponerinae towards ants of the same or of different species
introduced into their nests. They are always hostile to ants
of a different species, but the aggressive nature of this hostility
varies according to circumstances. If one or a few strange ants
be placed in a nest of Ponerinae, they will be at once attacked
and killed by concerted action, but if a single Ponerine be
placed in the nest of another species --say in a nest of the
agricultural ant (Pogonomyrmex barbatus) - - it will if attacked
defend itself valiantly with its mandibles and sting, but it will
prefer to seek refuge in some corner of the nest and remain
concealed. Single specimens of P. harpa.v, introduced at dif-
ferent times into the nest of P. barbatus, managed to evade the
pugnacious agricultural and to remain alive for days. When-
ever they left their concealment, however, they would be at-
tacked by a swarm of agricultural, but the Ponerinae were
so supple and fierce that they always managed to throw off
their assailants and to conceal themselves. Other investiga-
tors, like Forel and Wasmann, have called attention to this
difference in the pugnacity of ants according to whether they
find themselves alone or backed by numbers of their own species.

The species of Ponerinae differ in their reception of ants of
the same species from other nests. If two nests of workers

No. i.] A STUDY OF SOME TEXAN PONERINAE. 9

of P. Jiarpax (including ergatoid females) be thrown together
into the same jar, the ants of different nests at once begin to
struggle with one another. One ant will grasp a stranger by
the head with her mandibles and smite her with rapidly vibrating
antennae ; or the two foes will interlock mandibles and pull in
opposite directions till one gives up and runs away, only to
begin the same performance with some other stranger. This
peculiar tugging goes on between the ants of different nests
for hours, or even for three or four days. Then, without any
mutilations or deaths as a result of the struggle, the ants
abandon all this hostile play, and thenceforth form a single
peaceable community.

When two nests of workers of L. elongata are put together,
the struggle which ensues presents a slightly different aspect.
An ant from one nest will grasp a stranger by an antenna or
by one of her fore feet and, while tugging, smite her opponent
with very rapidly vibrating antennae. In this species there is
no interlocking of mandibles. After several days of tugging
and struggling the same peaceable conditions supervene as in
the case of P. harpax.

Workers of O. haematodes, belonging to different nests,
never display the slightest animosity towards each other, so far
as I have been able to observe : in a strange nest they are
treated exactly as if they had always belonged there. An
exception to this conduct of O. haematodes in the case of
friends deprived of their antennae will be mentioned below.

A few observations were made on the conduct of nests of
workers towards sexual individuals from other nests. One of
the two males of P. Jiarpax, taken March 3, was placed in a
strange nest of ten workers (including ergatoid females). The
male, after walking about on the earth for a short time, found
his way into one of the galleries in which six of the wingless
individuals were digging. On perceiving the male they at once
stopped their work and began licking the winged stranger with
signs of great agitation and affection. At the same time they
carefully massaged all his limbs and segments with their
mandibles. During this effusion the attitude of the male was
most peculiar. He lay on his back in what may be described

IO WHEELER. [VOL. II.

as a pupal attitude, with his legs drawn up against his body,
and his long filiform antennae folded against the mid-ventral
surface of his thorax and abdomen. This amusing scene
was still in progress when I had to quit my observations
half an hour later. The next morning I saw one of the
workers (or an ergatoid female?) carry the male in her man-
dibles a distance of four inches and disappear with him into
one of the galleries. During transport the male preserved the
motionless pupal attitude. This treatment of a strange male
contrasts strikingly with the treatment received by the male
which was allowed to remain with the workers and ergatoids
of his own colony. After eight days of captivity this male was
killed and devoured by one of the wingless individuals. When
I first observed this Amazon, she had already consumed the
small head of her victim and was hanging from the roof of a
gallery, twirling the torso about with her feet and rasping
away the thoracic muscles with her maxillae. Was this act
prompted by hunger? or had the male fulfilled his mission in
this nest already provided with several packets of eggs ? I
have already mentioned the fact that the male of L. elongata
was received by a strange colony of workers like an old ac-
quaintance. The absence of any display of affection in this
case may have been due to the fact that there were no females
in the nest. Two strange females introduced into this same
nest at different times were at once surrounded and attacked
by the workers, and although the colony was without a queen,
they were either maimed or killed during the course of the day.
Many of the habits of the Ponerinae closely resemble those
of other ants. All the habits relating to the cleanliness of the
individual ants and of the nests are the same as those of the
Myrmicinae and Formicinae. They clean themselves and one
another with great care. The males spend much of their time
drawing their delicate antennae through the strigils on the
fore feet. The nests of all the species are kept scrupulously
clean - - all refuse, inedible fragments of their insect prey,
empty cocoons, dead sister ants, etc., are carefully deposited
in one of the corners of the nest exposed to the light and as far
as possible from the chambers in which the young are reared.

No. i.] A STUDY OF SOME TEXAN PONERIXAE. II

In digging a gallery P. Jiarpax first uses its feet and power-
ful mandibles in removing the earth, after the manner of the
digger-wasps, but as soon as the gallery is sunk to a certain
depth, only the jaws are used in removing the lumps of earth.

Emery's observations1 show that the Ponerinae have the
power of stridulating. He has even been able to produce the
sound in preserved specimens of certain American species,
including species of Pachycondyla. Although the ants ob-
served by me occasionally moved their abdomens as if stridu-
lating, no sound could be detected.

The sting of the three species of Ponerinae, although of
formidable length, does not inflict severe pain, at least when
compared with the sting of the agricultural ant (Pogonomyrmex
barbatus Sm.). The pain may be acute for ten or fifteen
minutes, but then disappears, often without even reddening the
skin ; whereas the sting of Pogonomyrmex produces a throb-
bing pain which endures for hours and is sometimes accom-
panied by a sensation of sickness.

The Ponerinae do not seem to feed one another, like the
specialized ants. In captivity P. harpax would eat the yolk
of egg or even sugar, but it would not eat termites. L. elon-
gata devoured termites and small caterpillars with avidity,
but would not eat flies. O. haematodes is more omnivorous ;
besides caterpillars, house-flies, beetles, and small Hemiptera,
it will eat sugar, bread, cake, etc.

Pachycondyla and Leptogenys are like other ants in their
methods of killing their prey or in feeding upon it, but Odon-
tomachus is most exceptional in these particulars. Its whole
life, apart from the care of its young, appears to center in its
peculiarly constructed mouth. The long linear mandibles with
hook-shaped dentate tips are inserted close together. They
are usually carried wide open, as represented in Fig. I, while
the ant is moving about in search of food or while it is feed-
ing. The cutting edges of the mandibles are furnished with
some sensory hairs, two of which, nearly as long as the mandi-
bles, are inserted near the base and point directly forward

1 Emery, C., " Zirpende und springende Ameisen," Biol. Centralbl. Bd. xiii,
pp. 189-190. 1893.

12 WHEELER. [VOL. II.

when the mouth-parts are in the position represented in
Fig. i. The slender antennae, too, are carried in a peculiar
position, their tips being directed inwards. The touching of a
living insect or of any unfamiliar object with these incurved
tips calls forth a peculiar response, which seems to be largely
of a reflex nature. The ant darts forward and suddenly closes
its mandibles with a very audible click. The signal for closing
the mandibles seems to be given the moment the long sense-
hairs touch the object. The mandibles are brought together
with such force that if they strike a solid object the ant is
thrown backwards - - often to a distance of three or four inches
- occasionally even to a distance of ten or twelve inches.
The ant alights on its feet, like a cat, and again advances to
repeat the act. This remarkable clicking and leaping habit is
called into play on every occasion, and its study discloses some
interesting facts, as the following jottings from my notebook
will show :

May 12. Placed a living house-fly in the Odontomachus nest.
Its movements at once attracted several ants, which began snap-
ping at it like a pack of angry dogs. With each snap a leg or
wing was severed and often thrown to a distance of 2 or 3 inches.
In less than a minute all the limbs had been shorn from the
trunk. The fly was then seized and decapitated. Next the
following living ants were placed in the nest in succession:
Eciton sumickrasti, Pogonomyrmex barbatus, and Leptogenys
clongata. The first and last were reduced to torsos as rapidly
as the fly ; the harder legs and antennae of the Pogonomyrmex
offered greater resistance, but not for long. The stings and
strong mandibles of these various ants were no protection
against the singular method of attack adopted by the Odonto-
machus. A smooth green caterpillar was next introduced into
the nest. It was at once surrounded and attacked by a dozen
ants. At first the ants retreated from the writhing larva after
each snapping of the mandibles, but soon they grew bolder and,
retaining their hold, drove their stings through its velvety
skin. One ant with a dexterous movement closed its mandi-
bles in the caterpillar's brain. In a few moments the green
blood of the victim was oozing from a score of wounds.

No. i.] A STUDY OF SOME TEXAN PONERINAE. 13

May 25. O. Jiacniatodes reacts to lifeless objects just as it does
towards living insects. If the points of a pair of tweezers be
held in the nest and rapidly opened and closed, they at once
become the center of a swarm of snapping and leaping ants. A
small glass vial dropped into the nest at once begins to tinkle
under the blows of the alternately advancing and retreating
emmets. A lump of sugar is attacked in the same manner for
some moments, till the ants learn that it is edible, when they
settle down on it and lap its crystals with avidity. Even soft
viscid or liquid substances, like the yolk of egg, poured on
the floor of the nest, at once release the same peculiar response.
As soon as the ants perceive it with the contact-odor sense of
their antennae, they begin snapping at it as if it were a deadly
foe. In this case the tips of their mandibles are smeared with
the yolk. This causes them to feel some discomfort appar-
ently, and they wipe them on the hard floor of the nest, much
as a bird would wipe its beak in similar circumstances. It is
only after several repetitions of this performance that they
approach the yolk without closing the mandibles. Then they
begin to eat it, and some hours later they begin to cover it
up with little pellets of earth, for O. Jiaematodes, as well as
L. clongata and P. harp ax, share with the Myrmicine and For-
micine (and I may add, also, the Doryline ants!) the habit of
burying offensive substances.

Observers 1 have called attention to the fact that ants de-
prived of their antennae are in some cases attacked and killed
by ants of the same nest. A few experiments were tried for
the sake of testing these observations on Odontomachus. One
of these ants, deprived of both antennae, was replaced in the
nest from which it had just been taken. It was at once
fiercely attacked by several ants as if it had been a strange
insect. It stopped, as if dazed, unable to meet its sister work-
ers with the customary antennal greeting. The ants, however,
soon perceived their error, and a few hours later one of them
was licking and fondling the mutilated ant. By the next
day the mutilated ant was dead, probably as a result of the

1 Forel, Fourmis de la Suisse, p. 1 19, et al. • Wasmann, Die zusammengesetzten
Nester und gemischten Kolonien der Ameisen, pp. 151, 152, Miinster i. W., 1891.

I ^ WHEELER. [VOL. II.

extirpation of its antennae. Two other experiments termi-
nated in the same way.

Two inferences may, I believe, be drawn from these obser-
vations on the snapping and jumping habits of Odontomachus :
First, it seems evident that these are reflex actions ; second,
that they can be inhibited by the will of the insect. This
latter conclusion is supported by the experiment with the sugar
and the yolk. It also shows that these ants learn by experience,
and that they possess some form of memory.

The peculiar snapping and leaping habits of O. Jiaematodes
have been remarked by a few other observers. Emery has
published a brief note on this subject,1 and Forel 2 says of this
insect that " en Colombie on 1'appelle ' Fourmi tac ', a cause du
bruit qu'il fait en refermant brusquement ses mandibules. Par
le meme mouvement il ressaute en arriere lorsqu'il les referme
contre un objet, ce qui lui a fait attribuer a tort la propriete de
sauter." I cannot accept this last statement, which seems to
be a contradiction in terms. The ant makes use of its saltatory
powers for purposes of escape, as any one who tries to capture
a colony of these ants may readily observe. That it leaps
backward instead of forward, like other insects, is due to its
using a most unusual organ for leaping, for the mandibles,
which, as in other ants, are used for digging and transporting
the soil, carrying eggs and larvae, and for killing and cutting up
prey, have acquired an additional function as saltatory organs
in O. Jiaematodes.

The breeding habits and the characteristics of the eggs and
larvae of the Ponerinae exhibit striking deviations from those of
other ants. I have not seen the eggs of Odontomachus, but
throughout the month of May I have often happened on the
eggs of Pachycondyla and Leptogenys. These are white and
of a slender, oblong shape (Fig. 8, a], somewhat smaller in the
latter than in the former genus. They differ in shape from the
eggs of species of Eciton, Camponotus, Formica, Pogonomyr-
mex, Solenopsis, and Tapinoma ; for the ants of these genera,
representing several subfamilies, agree in having elliptical and

1 Biol. Centralbl. Bd. xiii, pp. 189-190. 1893.

2 Loc. cit., p. 20.

No. i.] A STUDY OF SOME TEXAN PONERINAE. 15

much less slender eggs than the Ponerinae. The Ponerinae
also keep their eggs in more regular packets, the long axes of
the different eggs being placed parallel with one another. I
have not been able to determine the length of embryonic
development with accuracy. It seems to be much protracted,
as in other ants. Some eggs of P. harpax in an artificial nest
had not hatched at the end of five weeks.

The larvae of the Ponerinae differ from those of all other
ants in several particulars, first made known by Emery. x Emery
describes and figures the larvae of Poncra stigma Fab. (New
Guinea), P. caffraria F. Sm. (Cape Colony), Diacamma rugo-
sum-gcometricum F. Sm. (Celebes), and Odontomachus liaema-
todes L. (Cayenne). All of these larvae agree in having the
mandibles powerfully developed for ant-larvae, the anterior
portion of the body long and slender and folded over the
abdominal portion, and in being covered with rows of peculiar
tubercles beset with more or less prominent bristles.

The larvae of Pachycondyla and Leptogenys are here figured
for the first time. I have seen fit to figure also the Odonto-
machus larvae because my material was evidently in fresher
condition than that depicted by Professor Emery. The larvae
of the three genera may be arranged in a series beginning with
Odontomachus and terminating with Pachycondyla.

The young larva of O. haematodes is represented in Fig. 5, a,
which shows the arrangement and character of the bristly
tubercules and the neck-like anterior portion, consisting of the
head, the three thoracic and the first two abdominal segments.
The remaining eight abdominal segments are much enlarged
and flattened ventrally. The larva is kept on its back, and the
neck-like anterior portion rests against the flattened ventral
surface. The shape of the tubercles, each of which is tipped
with a rigid bristle and encircled with bristles, is shown in Fig.
5, b. The larva was about to moult, so that the tubercle of the
succeeding cuticle is seen shining through the old one. The
adult larva is shown in Fig. 6, a. Compared with that of the
young larva, its head is very small in proportion to the body.

1 "Intorno alle Larve di Alcune Formiche," Mem. della Accad. delle Scienze
dell Istittito di Bologna, 7. Maggio, 1899, 2. Tavole.

i6

Jl" 'HEELER.

[VOL. II.

This seems to be the universal rule in ant larvae. The head
in dorsal view is represented enlarged in Fig. 6, b. The power-
ful dentate mandibles lie just below the outer edges of the
bilobed labrum ; still lower and projecting forward lies the
labium, bearing on its tip the opening of the spinning gland
(to be used in weaving the cocoon), and on either side two peg-
shaped tactile (?) organs. Similar but somewhat larger organs
are seen on the edges of the maxillae, which protrude on either
side below the mandibles. The ten tracheal stigmata, begin-

FIG. 5. — Odontomachus haematodes Linn, a, young larva ; />, tubercle of the same.

ning on the mesothoracic and terminating on the eighth
abdominal segment, are clearly shown in Fig. 6, a. The bristly
tubercles are essentially the same in structure as those of the
younger larva, but they are relatively shorter and smaller.
(Cf. Fig. 5, b, and Fig. 6, c.)

The larvae of Leptogenys (Fig. 7) are remarkably slender
and scarcely flattened on the ventral surface. In the young
larvae (Fig. 7, a) the tubercles are distinctly curved and pointed,
without apical bristle, and with only a few rather short bristles
encircling the base (Fig. 7, d). In the adult larvae (Fig. 7, b
and e) the tubercles are larger and shorter, with blunt or

No. i.] A STUDY OF SOME TEXAN PONERINAE. 17

acuminate apex and with relatively longer and more numerous
basal bristles. The head of the adult larva (Fig. 7, c) is remark-
able for its length and the narrowness of the labrum, which is
nearly as long as the slender mandibles and provided with a
median tooth at its tip.

The larvae of Pachycondyla (Fig. 8) are neither as slender as
those of Leptogenys nor as robust as those of Odontomachus.

FIG. 6. — Odontomachus haematodes Linn, a, adult larva ; l>, head of same
(dorsal aspect); c, tubercle.

The ventral surface of the abdomen is distinctly flattened.
The head (Fig. 8, e) resembles that of Odontomachus, especially
in the shape of the labrum and mouth-parts. There is a strik-
ing difference between the tubercles of the very young and the
adult larva. In the former (Fig. 8, b, c) the tubercles are nearly
or quite straight, and somewhat longer and more pointed than
those of Leptogenys. They lack the terminal bristle. The bris-
tles about the base are somewhat irregular in their insertion.

i8

WHEELER.

[VoL. II.

In the adult larva (Fig. 8, d) the tubercles are reduced to
large more or less flattened bosses, encircled with a regular
row of numerous, rather long bristles. In the stages between
those figured the gradual flattening of the juvenile spine-like
tubercles can be traced through the successive moults.

In this series of larval forms, Odontomachus seems to repre-
sent the most primitive condition. Here both young and old
larvae have pointed, bristle-tipped tubercles, and there is little

FIG. 7. — Leptogenys (Lobopelta) elongata Buck, a, young; b, adult larva : c, head of adult
larva (dorsal aspect) ; d, tubercle of young ; e, tubercle of adult larva.

difference between the tubercles of the young and adult. In
Leptogenys and Pachycondyla the apical bristle is absent, but
in both genera the young larvae have pointed tubercles. In the
adult larva of the former genus there is a perceptible blunting
of the tubercles, while in the adult larva of the latter the
tubercles have nearly subsided.1

1 Emery's observations on Ponera stigma, P. caffraria, and Diacamma geometri-
cttm, seem to indicate conditions the reverse of those which I have described. Of
the former species he says (loc. fit., p. 4) ; " Nello stado piu giovane, si vedono solo
deboli accenni del tubercoli cutanei ; ritengo che questo stado debba corrispondere
alle larve di prima schiusa e che lo stado seguente, di poco piu grande, sia quello

No. i.] A STUDY OF SOME TEX AX POXERINAE. 19

The bristly tubercles of the larvae of the Ponerinae are so
prominent as readily to suggest the question of their function.
Prof. L. Biro, who made some observations on the larvae of
P. stigma, which he sent to Professor Emery, believes that the
pointed tubercles are organs of defense. He saw these larvae
when disturbed by some termites move their long necks back
and forth with sufficient force to drive away the intruders.1

FIG. S. — Pachycondyla. harpax Fabr. a, eggs ; b, young larva ; c, tubercle of the same ;
. d, adult larva ; e, head of the same.

che segue la prima muta ; quest! si fan no successivamente piu numerosi e spor-
genti, a misura che la larva cresce." In the larvae of Diacamma a very different
condition is described : " Sopra ciascuno (segmento) di essi si trova un serie trans-
versale, irregolare di tubercoli conici, ineguali che, nelle larve piu sviluppate,
portano da uno a quattro peli. Xelle piccole larve, i tubercoli sono piccoli,
subcilindrici e senza peli; negli stadi intermedii passano per una forma acuminata
con pochi peli."

1 Professor Emery (Zoc. cit., p. 4) quotes from Professor Biro's letter: "Nelle
gallerie del nido scavato nel legno putrido, si trovavano le larve dal lungo collo,
coperte di spini singolari : abbandonate dai loro vigliacchi custodi, quelle larve
sapevano difendersi da se ; quando qualche termite (il nido di queste trovavasi
nello stesso legno) si avvicinava ad una di esse, questa batteva innanzi e indietro
col suo collo di cigno e tosto veniva lasciata in pace."

2O

WHEELER.

[VOL. II.

Biro's observations may be true of P. stigma without being
applicable to the three forms of Ponerinae which I have
observed. In artificial nests I have seen the neck movements
of the larvae, but they were often executed when the larva was
undisturbed, except perhaps by the pangs of hunger, and they
were not always made when termites or other insects were
running about and over them. Moreover, we should expect
to find the tubercles more highly developed on the neck than

FIG. 9. — Pogonomyrmex btrrbatits Smith, a, nearly adult larva ; /», young larva ;
c, serrated bristle of the same.

on the body, if they are really used as Biro suggests. I believe
that while they may be organs of passive defense, like the
somewhat similar tubercles and spines of certain caterpillars,
they also fulfill other functions ; they would seem to facilitate
the carrying of the larvae either singly --when full-grown — or
in batches --when young --by the worker ants. In the last
instance they would represent a peculiar form of the " poils
d'accrochage" carefully studied by Janet.1 Janet finds that
the young larvae of the more specialized ants are covered with

1 Les Fourmis. Conference faite le 28 Fevrier, 1896. Paris, 1896.

No. i.] A STUDY OF SOME TEXAN PONERINAE, 21

hooked bristles, which cause them to adhere together in packets
and thus facilitate their transportation by the workers. The
appearance of these peculiar hairs in the young and half-grown
larvae of one of our common Texas ants, Solcnopsis gcmi-
nata Fabr., is shown in Fig. 10. The very young larvae have
only simple bifurcated hairs, but when half-grown they have on
the dorsal surface of several of the segments, besides a much
greater number of these simple bifurcated hairs, several rows
of long and peculiarly contorted bristles, terminating in short

FIG. 10. — Solenopsis geminata. Fabr. a, very young larva; f>, furcate bristle of same ; r, half-grown

larva; d, contorted furcate bristle of same.

bifurcations. Still another modification of the " poils d'accro-
chages " is seen in Pogonomyrmex barbatns (Fig. 9), the young
larvae of which have the longer bristles serrate on the apical
half, so that they remind one of the hairs of certain mammals.
All of these modifications --the bristly tubercles of the Poner-
inae, the simple and contorted bifurcated bristles of Solenopsis,
the serrate bristles of Pogonomyrmex, and possibly also the
fascicles of uncinate hairs described and figured by Emery (loc.
cit.} for the larva of Sima natalensis F. Sm. --seem to subserve
the same purpose — a most interesting example of independent

22 WHEELER. [VOL. II.

lines of development terminating in organs of different struc-
ture but identical function.

The pupae of Odontomachus, Leptogenys, and Pachycon-
dyla are enclosed in elliptical brown cocoons, like the pupae
of many species of Formica and Camponotus. The pupa of
L. clongata is remarkable on account of its very slender shape,
a peculiarity not confined to the pupa, but, as we have seen,
extending also to the egg, larva, and imago.

We come now to a consideration of the breeding habits of
the Ponerinae. The little that has been made known concern-
ing their habits has led European myrmecologists to believe
that the philoprogenitive instincts of these ants must be less
highly developed than those of the Myrmicinae and Formi-
cinae. Thus, according to the above-quoted note of Professor
Biro, when the nest of Ponera stigma is disturbed the ants flee
and the larvae are "abbandonate dai loro vigliacchi custodi."
And Professor Forel has made what appears to be a somewhat
similar observation on our American Ponera coarctata Fabr. in
North Carolina : : "La Ponera coarctata americaine est tres com-
mune dans les troncs pourris et sous les pierres. J'ai fait chez
elle une observation qu'il est bien difficile de faire en Europe ;
mais ici elle est tout a fait constante. Lorsqu'on decouvre
un nid de Ponera dans un tronc pourri, on voit leurs cocons
jaunes assembles dans un coin, mais absolument abandonnes
des $ qui n'essaient pas de les sauver, ni de les recueiller.
Par contre, elles prennent le plus grand soin des larves qu'elles
emportent et cachent. Je soupqjonne que chez ces fourmis,
moins sociales que les autres, les nymphes sortent seules de
leurs cocons, sans avoir besoin de 1'aide des 9."

These observations relate to species of Ponera and are at
variance with the conclusions which I have reached from a
study of three other genera of Ponerinae. In many of the
nests which I have examined the total number of the eggs,
larvae and pupae, could scarcely be greater than one and one-
half to twice the number of the ants. This fact, together with
what has been said of the small number of eggs laid at one
time by a single female, shows very clearly that the Ponerinae

1 Ann. de la Soc. Entomol. de Belgique. Tome 43, p. 443. 1899.

No. i.] A STUDY OF SOME TEXAX POXERINAE. 23

are not nearly so prolific as the species of Camponotus,
Formica, Pogonomyrmex, Pheidole, Tapinoma, Eciton, etc.
Indeed, the small number of ants in the nests of the Ponerinae
is probably the direct result of this limited productivity. If
this is the case, it does not seem probable that these ants
would be more careless of their progeny than the very prolific
specialized ants. On the contrary, we should expect them to
extend even greater protection to their offspring. This my
observations show to be the case ; at any rate, P. harpax,
L. clongata, and O. Jiaematodes are in nowise inferior to the
Myrmicinae and Formicinae in this respect. The slightest
disturbance of the natural or artificial nests of these ants
causes them at once to seize their eggs, larvae, and cocoons,
and to make for their galleries. Occasionally some of the ants
escape without anything, but if they are watched for a few
moments, they will be seen returning, often in the very face
of danger, to carry off more of their young. They are, it is
true, most careful of their eggs, somewhat less careful of the
larvae, and least careful of their cocoons ; but these distinctions
are not always apparent and can only be affirmed as the result
of many observations. When the colony is agitated, it is
probably most easy for the ants to seize and remove the small
packets of eggs and the younger larvae, and least easy to carry
off the larger larvae and the awkward cocoons. Dead pupae
are often collected in one part of the nest and are there
allowed to lie unheeded. I am inclined to think that Professor
Forel may have seen such abandoned pupae in the nests of
P. c caret at a.

The strong development of the mandibles of the Ponerine
larva as compared with those of other ants led Emery remotely
to surmise the method which the Ponerinae employ in feeding
their young.1 But no myrmecologist could have predicted the

1 Loc, cit., pp. 8, 9. " Sembrami pertanto che lo sviluppo notevole della bocca
e particolarmente delle mandibole, nelle larve delle Ponerinae e dell' Acantho-
stichus inducano a qualche supposizione relativamente alia biologia di queste
formiche. Le larve delle specie europee che finora furono osservate vengono
alimentate col contenuto dell' ingluvie delle operaie che queste regurgitano sulla
bocca delle loro larve, e forse anche col secreto di ghiandole salivari. In queste
specie, 1 'airmen to delle larve consiste dunque esclusivamente di sostanze liquide o

24 WHEELER. [VOL. II.

remarkable and un-ant-like procedure which I have been able
to observe in the three Texan species.

My first observation on this singular method of feeding the
larvae was made on a large nest of Pachycondyla found under
a stone at the foot of Mt. Bonnell, near Austin, May 5.
Before the ants could carry them away, I had scooped up a
fine lot of larvae, together with the earth in which they were
lying. Among the larvae were several pieces, one or two seg-
ments long, of a recently killed myriopod (Scutigera). Into
these pieces the larvae, some of which were nearly full-grown,
had inserted their heads and were devouring the softer tissues !
This could be distinctly seen with the pocket lens through the
glass of the vial to which the larvae had been transferred. In
another nest of the same species, uncovered May 16, I ob-
served the larvae in the nest lying on their backs, devouring
the pieces of some insect which I could not identify.

The former of these observations made in the field led me
to observe the feeding of the larvae in my artificial nest of
Leptogenys. I had frequently wondered at the way in which
these ants decapitated termite nymphs or cut off their abdo-
mens and scattered these about among their larvae. It was
all quite clear to me now ; examination with the lens showed
that the larvae had inserted their long necks through the cut
surfaces into the soft parts of the termites and were feeding
exactly like the larvae of Pachycondyla.

During the month of May I had frequent opportunity to
see Odontomachus feeding its larvae in my artificial nests.
These larvae are placed by the ants on their broad backs, and
their heads and necks are folded over onto the concave ventral
surface, which serves as a table or trough on which the food is
placed by the workers. The following observations are tran-
scribed from my notebook :

semiliquide ; e tale e pure in massima 1'alimento delle stesse formiche allo stato
adulto, quando si cibano di sostanze zuccherine vegetali o degli escrementi liquidi
degli afidi. Pero, molte formiche vivono pure in parte di preda, e nulla prova che
si contentino di sorbire i succhi della loro vittima, e non digeriscano pure, medi-
ante la saliva, alcune parti solide." ..." Ora sarebbe pure possibile che For-
miche, le quale vivono principalmente di preda, diano in pasto alle loro larve pezzi
piii o meno triturati del corpo delle loro vittime come fanno le Vespe."

No. I.] A STUDY OF SOME TEXAN PONERINAK. 25

May 13. This evening several house-flies, placed in the
Janet nest of O. liacmatodes, were at once shorn of their legs,
then decapitated, and finally their thoraces and abdomens were
cut into smaller pieces and distributed among the larvae. One
was given a fly's head, which it kept twirling around in a comi-
cal manner, while it devoured the brain through the small
cervical orifice. Another was given a piece of a thorax with
one of the wings still attached, another a piece of an abdomen,
still another, a leg with a mass of muscle at its coxal end, etc.

May 16. This evening a small homopterous insect was
placed in the Odontomachus nest. One of the ants (A)
snapped at it, disabled it, and then left it. A few moments
later it was picked up by another ant (B) and carried into the
chamber containing the larvae and pupae. Thereupon a third
ant (C) took hold of it and began tugging at it with B till it
was torn open, but not into pieces. B then placed it on the
flat ventral surface of a medium-sized larva, which began feed-
ing at once, moving the homopteron around with its jaws.
After four minutes had elapsed, another ant (D) that had been
standing near by, apparently much interested in the feeding,
suddenly tore the morsel away and placed it on a small larva.
This larva was permitted to feed ten minutes, closely watched
during all this time by ant D and another (E) which had
come up in the mean time. Then ant D tried to tear the
morsel away from the small larva, but apparently unable to do
so, it took up the larva with the morsel and dumped them both
on the ventral surface of a large larva. This creature seized
the homopteron and forced the small larva to release its hold
and to drop to the ground. The large larva fed for fully
twenty minutes, closely watched by ant D and two others
(E and F). All of these ants tried at different times to
wrench the morsel away from the larva, but failed. Suddenly
a small ant (G) rushed up, tore it away, and ran off with it.
By this time very little was left of the homopteron and I lost
track of it.

May 23. A few crumbs of cake, moistened with water, were
placed in the Odontomachus nest at 11.7 P.M. A worker soon
carried one of the crumbs into the breeding chamber and gave

26 WHEELER. [VOL. II.

it to a large larva at 1 1.20. This larva fed but a few moments,
but the cake was not removed till 11.35, when it was carried
into another chamber, then at once brought back and placed
between three larvae, from one of which it had just been taken.
The smallest of these three larvae nibbled at it for a short
time, beginning at 11.40. But one minute later this larva was
carried away by a worker, and the cake was taken by another
worker and given to a small larva at 11.43. This larva, too,
was soon carried away (at 1 1.48), and the cake was taken to a
large larva, which would have none of it. It was not removed,
however, till 11.50. Then it was given by another worker to
a large larva, which did eat some of it. At 1 1.51 the piece of
cake, but little diminished in size after all its perambulations,
was taken to another large larva. The ant remained over the
larva holding the cake in place till 11.58, when another worker
came up and ran away with the larva. While the larvae were
feeding, the ants themselves could be plainly seen to partake
of the cake from time to time. During the whole period of
the above observations, and for some minutes later, i.e., for over
an hour, one little larva was permitted to feed without inter-
ruption on what seemed to be a piece of a house-fly.

These observations lead us to several interesting reflections.
First, it is certain that the feeding of the larva of the Ponerinae
is of a far more primitive character than in any other ants in
which this process has been studied. It is, in fact, even more
primitive than the corresponding habit of the social wasps,
which feed their larvae with masticated insect prey, for in the
Ponerinae the prey is cut into a few pieces only, for the pur-
pose of exposing the soft tissues and making them accessible
to the mandibles of the larvae. Myriopods or large insects are
disarticulated for this purpose, small insects are merely torn
open. Leaving the question of systematic affinities out of
consideration, the Ponerinae may be said to have habits of
feeding the young intermediate between the habits of the soli-
tary wasps, which provide their young with whole insects, and
the social wasps, which masticate the food for their larvae. In
this statement it may, perhaps, be more accurate to substitute
the Bembecidae for the solitary wasps, since the Bembecidae,

No. I.] A STUDY OF SOME TEXAN PONERINAE. 27

which feed their larvae from day to clay with entire Diptera in
a fresh condition, resemble the Ponerinae more closely than
do the solitary wasps, which merely enclose their eggs with
paralyzed larvae, spiders, grasshoppers, etc.1 From the con-
dition of the Ponerinae to that of the more specialized ants,
which feed their larvae with nothing but the liquid food regur-
gitated from their own crops or from their salivary glands,
the transition is very abrupt. But there are many ants whose
habits have not been studied, and some of these may yet be
found to bridge this chasm.

In the second place, the above-recorded observations seem to
show that the Ponerine method of feeding the larvae is of
a most capricious and irregular character. The quantity and
quality of the food given to a particular larva, and the time it
is permitted to feed, seem to be matters requiring no very
strict regulation. The ants that feed the young rarely act in
concert, but rather with a whimsical individualism that seems
at times to border on the ridiculous.

This irregular method of feeding suggests other consider-
ations of a wider bearing. It is generally admitted that the
polymorphism of the female sex in ants, i.e., the occurrence of
fertile females and of sterile females of one or more casts, is
in some manner correlated with the feeding of the larvae de-
veloped from fertilized eggs. In other words, the worker ants
can control the production of individuals like themselves and
of individuals like their queen. It is further maintained that
these differences are effected by the quantity and quality of
the food administered to the larvae at a certain period of their
development ; but here our knowledge ends. These data have
been accumulated from the study of the specialized Myrmicine
and Formicine ants of Europe and North America, and are

1 Fine descriptions of wasps (Polistes) and Bembecids feeding their young
are to be found in the charming works of Fabre (Souvenirs Entomologiques,
i° ser., 2m edit., Paris, 1894, pp. 126-128 and pp. 226 et seq.) and of Dr. and Mrs.
G. W. Peckham ("The Instincts and Habits of the Solitary Wasps," Bull. Wis-
consin Geol. N. H. Survey, No. 2, 1899, 245 pages, 14 plates). Janet has described
the corresponding habits of Vespa ("fitudes sur les fourmis, les guepes, et les
abeilles." 10. Note. Sur Vespa media, V. silvestris, et V. saxonica, Mem. de la
Soc. Acad. dc Z'Oise, tome xvi, 1895, P- 39)-

28

WHEELER.

[VOL. II.

supported by many valuable observations on the hive-bee.
Now, while we can, perhaps, understand how these more spe-
cialized ants may manage to control the quantity and quality of
liquid food regurgitated from their own crops and salivary
glands, it is not so easy to understand how ants can exercise
such control when they adopt a capricious method of feeding
like that of the Ponerinae. Such a method can hardly produce
clear-cut results ; i.e., either workers or fertile females. And
a comparative study of the better known species of Ponerinae
shows that in certain species at least there is no such sharp
distinction between the sterile and fertile female as we find in
the more specialized ants. Not only is the female sex in a
state of morphological and physiological instability, — i.e., di- or
even tri-morphic, — but the male sex also is sometimes dimorphic
- at least in the same genus, if not in the same species. For
the purpose of illustrating this singular instability of the sexes
I have compiled the following table from the literature to which
I have access.1 It includes twelve of the better known species

Species of Ponerinae.

Winged
Male.

Ergatoid
Male.

Winged

Female.

Ergatoid
Female.

Worker

Major.

Worker
Minor.

Odontomachus haematodes Linn

+

+

+

+

Pachycondyla harpax Fabr. . .

+

+

+

+

Cardiocondyla Emeryi Forel . .

+

+

+

+

Cardiocondyla Wroughtonii Forel

+

+

+

Cardiocondyla Stambulofii Forel

+

+

+

Leptogenys elongata Buck.

+

+

+

Ponera ergatandria Forel . . .

? +

+

+

+

Ponera ochracea Mayr

+

+

+

Ponera Eduardi Forel ....

+

+

+

+

Ponera coarctata Latr

+

+

+

+

Ponera punctatissima Rog. . .

+

+

+

Stigmatomma pallipes Hald. . .

+

+

+

1 Sharp, " Formicidae in Cambridge Natural History," Insects, vol. vi ; Emery,
" Sopra Alcune Formiche della Fauna Mediterranea," R. Accad. delle Scienze
deir htituto di Bologna, 21 Apr., 1895; Emery, " Beitrage zur Kenntniss der
nordamerikanischen Ameisenfauna," Schluss, Zool.Jahrb., Abth. f. Syst, Bd. viii.

No. i.] A STUDY OF SOME TEXAN PONERfNAE. 29

of Ponerinae. The presence of a particular sexual phase is
indicated by a cross.

Although it is by no means certain that the irregular poly-
morphism of the two sexes of the Ponerinae, as indicated in
this table, is due to an inability on the part of the ants to
regulate with precision the quality and quantity of the food
administered to the larvae, I nevertheless believe that there
is some causal connection between these two peculiar phenom-
ena. At any rate, we may assume this connection as a work-
ing hypothesis for future experimentation and observation. I
believe that continued study of the relatively undifferentiated
sexual conditions of the Ponerinae may lead us more rapidly to
a solution of the interesting problems of nutritional polymor-
phism than a study of the more specialized ants.

When the larvae of the Ponerinae are mature they are, like
the mature larvae of the Formicinae, buried in the soil till they
have spun their cocoons. They are then unearthed and the
small adherent particles of soil are carefully removed by the
workers. I have watched the burying of the larvae in Lepto-
genys and the unearthing and cleansing of the cocoon in Odon-
tomachus. The cocoons of the three species of Ponerinae are
usually kept together, but the ants are scarcely as careful in
this respect as the species of Formica and Camponotus which I
have observed (F. ncontfibarbis and C. castancus). Nor do
they keep their larvae assorted according to sizes, a peculiarity
which accentuates the irregularity of their feeding habits.

Forel, as we have seen, believes that Ponera coarctata
may escape from its cocoon without the assistance of the
workers. Unfortunately I had to leave my work at Austin
before the pupae of Odontomachus were ready to hatch, but I
am convinced that Leptogenys, at any rate, opens the cocoon
and draws out the pupa when ready to enter on its imaginal
life. I have not seen this operation under normal circum-
stances, as the two workers which appeared as callows in my
artificial nest left their cocoons when I was not present, but
for some reason the workers in this nest were continually
opening the cocoons near one end and pulling out the still
white pupae. Ten or a dozen workers would gather about

WHEELER. [VOL. II.

one of these extracted pupae and lick it for hours. Sometimes
one ant would take possession of the limp thing and hold it
astraddle for a long time. Ultimately these prematurely born
ants were either devoured by the workers or fed to the rav-
enous larvae. Nevertheless the deft manner with which the
cocoon was opened, the pupa extracted, and the empty cocoon
at once placed on the kitchen-midden, or rubbish heap, indicated
very clearly that this is also the method of procedure with
pupae that have reached their full growth.

A word in conclusion concerning myrmecophiles, for the
Ponerinae, like the other subfamilies of ants, are known to
harbor arthropod guests in their nests.1 No guests were
taken with Odontomachus and Leptogenys, but some six dif-
ferent species were observed in various nests of Pachycondyla
Jiarpax. Only one of these had not previously been found in
the nests of other species of ants near Austin. This was a
small yellow ant, a Solenopsis, allied to the European 5. fngax,
and found inhabiting some very minute galleries in the earth
between the huge burrows of the Ponerine. It is probably a
" Diebsameise," given to myrmecoclepsy like its European

1 The Ponerine guests enumerated by Wasmann in his Kritisches Verzeich-
niss der myrmekophilen und termitophilen Arthropoden, Berlin, 1894, are the
following : Typhloponemys hypogaea Rey (staphylinid beetle), with Typhlopone
oraniensis Luc., Palestine ; Apocellus (?) sphaericollis Say (staphylinid), with
Ponera coarctata Latr., North America ; Rlesotrochus paradoxus Wasm. (staphy-
linid), with Typhlomyrmex Rogenhoferi Mayr, Santa Catharina ; Euplectus Si-
korae Wasm. (pselaphid beetle), with Ponera Johannae Forel, Madagascar ;
Trichonyx sulcicollis Rchbch. (pselaphid), with Ponera coarctata Latr., Europe ;
Amauronyx Markeli Aube (pselaphid), with Ponera coarctata Latr., Switzerland ;
Araniops amblyoponica Brend. (pselaphid), with Stigmatomma pallipes Hald.,
Pennsylvania, North Carolina; Tmesiphorus formicinus McL. (pselaphid), with
Ectatomma sociale McL., Australia ; LeptotricJnis inquilinus Koelbel (isopod
crustacean), with Ponera senarensis Mayr, East Africa. More recently Was-
mann has described the following ponerinaphiles : Fauvelia permira Wasm.
(staphylinid), with Pachycondyla Fauveli Emery, Bolivia ("Die Ameisen- und Ter-
mitengaste von Brasilien," i. Theil, Verhandl. d. k. k. zool. hot. Gesell., Wien, Jahrg.
1895, PP- 4°>4I) > Lomechon Alfaroi Wasm. (silphid), with Pachycondyla aenescens
Mayr, Costa Rica (" Ein neues myrmekophiles Silphidengenus aus Costa Rica,"
Deutsch. Ent. Zeitschr., Heft, ii, 1897) ; Myrmedonia lobopeltina Wasm. (staphy-
linid) and Demera Fauveli Wasm. (staphylinid), with Leptogenys {Lobopelta}
mtida Sm., Natal (" Zwei neue Lobopelta-Gaste aus Siidafrika," Deutsch. Ent.
Zeitschr., Heft, ii, 1899).

No. i.] A STUDY OF SOME TEXAN PONERINAE. 31

congener. Two specimens of the very singular little ant,
Strumigenys louisianae Rog., were also taken from the earth of
this same nest. Their relations with the Pachycondyla were
probably of a more accidental nature. The other forms taken
are pleomyrmecophilous, i.e., they occur in the nests of several
other species of ants in the vicinity of Austin. These are,
first, a yellowish white species of Lepismina, quite common in
the nests of Pachycondyla, but even more abundant in the
nests of Camponotus castancns Latr., in the same localities.
This Thysanuran was also taken in the nests of Eciton coecum
Latr. Second, a white Collembolan, similar to, if not the same
as, Cyphodeira (Beckia) albinos Nicol. of Europe. This insect
is panmyrmecophilous, occurring in the nests of nearly all the
ants of Travis County. Third, Myrmecophila ncbrascensis
Bruner, rare in the nests of Pachycondyla, but very common
in the nests of Formica fusca, var. neorufibarbis Mayr. I have
no doubt that this singular little cricket had strayed from the
Formica to the Ponerine nests. Fourth, a small Trichoptery-
gid beetle was sometimes found in the nests of Pachycondyla.
As this same species was very common in the nests of Campo-
notus castaneus, in the same localities, I believe that it, too,
may have strayed from the nests of its typical host.

UNIVERSITY OF TEXAS MEDICAL SCHOOL,
GALVESTON, June 10, 1900.

THE EYES OF THE BLIND VERTEBRATES OF
NORTH AMERICA. III.

THE STRUCTURE AND ONTOGENIC DEGENERATION OF THE

EYES OF THE MISSOURI CAVE SALAMANDER, AN

ACCOUNT BASED ON MATERIAL COLLECTED

WITH A GRANT FROM THE ELIZABETH

THOMPSON SCIENCE FUND?-

CARL H. EIGENMANN AND WINFIELD AUGUSTUS DENNY.

A SINGLE specimen of a salamander was discovered in Rock
House Cave, Barrie County, Missouri, by Mr. F. A. Sampson
in July, 1891. The specimen was described by Stejneger in the
Proceedings of the U. S. National Museum, Vol. XV, p. 115,
as Typhlotriton spelaeus. His diagnosis reads as follows :
" Vertebrae opistocoelous ; parasphenoid teeth ; vomerine teeth ;
eyes concealed under the continuous skin of the head ; tongue
attached in front and along the median line, free laterally and
posteriorly ; maxillar and mandibular teeth small and numer-
ous ; vomerine teeth in two strongly curved series ; para-
sphenoid patches separate ; nostrils very small ; toes five ; six-
teen costal grooves, or eighteen if counting the axillary and
groin grooves ; tail slightly compressed, not finned ; toes
nearly half webbed ; vomerine teeth in two F-shaped series
with the curvatures directed forward ; gular fold strong, very
concave anteriorly ; color uniformly pale."

Stejneger fully appreciated the value and nature of his dis-
covery. He says : " Although many of our salamanders are
known to inhabit caves, this seems to be the only one, so far
discovered, which, like some of the other animals exclusively
living in caves, has become blind or nearly so." This was
written by him before he discovered the Typhlomolge in the
underground streams of Texas.

1 Contribution from the Zoological Laboratory of the Indiana University, No. 31.

33

34 EIGENMANN AND DENNY. [VOL. II.

A preliminary note by the present authors (Proc. Ind. Acad.
Set., 1898, p. 252, 1899) completes the list of papers dealing
with this species.

In the spring of 1897 Dr. Eigenmann visited Rock House
Cave and secured a number of larvae, which Dr. Stejneger pro-
nounced the larvae of Typhlotriton. Later Mr. E. A. Schultze
informed him that he had seen this salamander in the under-
ground passage to Bloncli's Throne Room in Marble Cave,
Stone County, Missouri. In September of 1898 he visited this
cave and secured four adults and three larvae of Typhlotriton.
A large number of the larvae were obtained from Rock House
Cave a few days later. Those from the latter cave were found
under loose stones and gravel in the rivulet at the mouth of
the cave. They had been exposed to the light. It is scarcely
supposable that those from Marble Cave had ever been affected
by the light. In the caves both larvae and adults are found
under the stones, the old ones in and out of the water.
Occasionally one is seen lying on the bottom of a pool.

In the aquarium the larvae creep into or under anything
available ; a glass tube serves as a " hiding " place. The
rubber tube admitting water to the aquarium is sometimes
occupied by several during a temporary cessation of the flow
of water. A wire screen sloping from the bottom of the
aquarium formed the most popular collecting place for the
larvae. They collected beneath this, although it was no pro-
tection from the light. From these observations it seems
probable that stereotropism rather than negative heliotropism
accounts for the presence of this species in the caves, and that
this reaction has been retained after the long stay of the species
in caves necessary to account for the changes in its eyes.

The eyes of the larvae when examined from the surface
appear perfectly normal, but they are little used in distin-
guishing objects. When hungry they will strike at a stick
held in the hand as they would at food. A stick lying at the
bottom of the aquarium undisturbed is not molested. They
strike at a worm when touched by it, or when it approaches
close enough for its motion to be perceived.

In the larvae up to 90 mm. long the skin passes over the

No. i.] BLIND VERTEBRATES OF NORTH AMERICA. 35

eye without forming a free orbital rim and the eye does not
protrude beyond the general contour of the head. In the
adult from 97 mm. on, the eye forms a bead-like projection.
There are in the adult distinct lids. These are closed over
the eye, covering it entirely, the slit being much too small for
the eye. The lower lid is free from pigment, but the upper
lid, which closes over the lower, is as thickly pigmented as any
other part of the body.

Stejneger says of the eyes that they are "small, only
slightly raised, and covered by the continuous skin of the
head, with only a shallow groove to indicate the opening
between the lids, the underlying eyes visible as two ill-
defined dusky spots."

In sections the lids are seen to overlap one another some
distance, forming an obscure, free orbital rim. Fig. i, a, is a
median section of the lids and corneal epithelium of an eye
.954 mm. in diameter, taken from an adult specimen 106 mm.
in length. In this section the upper lid overlaps the lower
lid .216 mm., or more than one-fifth the diameter of the eye.
Passing from the median section toward the corners of the
eye, the lower lid unites with the underlying tissue first.
When observed from the top the upper lid covers the entire
eye. The orbital slit is .17 mm. in length. The conjunctival
pocket extends some distance forward and backward beyond
the slit. The eye increases in size but little from the larval
to the adult stage, and its growth is not proportional to the
growth in length of the animal. (See comparative measure-
ments of the eyes at the close of the paper.)

The following is a series of measurements on the larvae of
Typhlotriton.

ROCK HOUSE CAVE. ROCK HOUSE CAVE. MARBLE CAVE.

Specimen ... 54 mm. long. 78 mm. long. 88 mm. long.

Size of pupil . . .432 mm. .640 mm.

Length of eye . . 1.30 mm. 1.50 mm. 1.60 mm.

From optic nerve

to front of lens . .80 mm. 1.20 mm.

Vertical diameter 1.248 mm. 1.28 mm.

Sections of the adult and larva from Marble Cave were made
in the usual manner. The six normal eye muscles were pres-

36 EIGENMANN AND DENNY. [VOL. II.

ent in Typhlotriton. The m. recti form a sheath about the
optic nerve in its distal part and spread out from it near the
eye. In the adult the sclera is a layer of uniform thickness
except in the region of the entrance of the optic nerve. It
is not usually separated from the adjoining parts of the eye,
but in places is retracted a short distance from the choroid coat
by the action of reagents. It is for the most part fibrous, with
few compressed nuclei, and varies from 1 8 to 40/0- in thickness.
In the larva a narrow cartilaginous band surrounds all but the
ventral wall of the eye. In a specimen 35 mm. long the width
of the band is about 3O//, its thickness i6/u. In three adult
specimens the sclera of only one had any traces of cartilage.
In the right eye of the adult specimen 103 mm. long a carti-
lage about 36/4 thick, 6o/u wide, and not more than 40/4 long
is found on the upper face of the eye. The absence of this
cartilage in the adult has probably no connection with the
degeneration of the eye. Its presence is probably a larval
characteristic which disappears as the gills disappear during
the metamorphosis.

The average thickness of the cornea is 40^. In the adult
it is covered by a layer of stratified epithelium, 25^ in thick-
ness, consisting of three rows of cells. The cells of the inner
row are columnar in shape, those of the middle row rounded,
and those of the outer row are very much flattened and
elongated (Fig. i, a).

In the adult the choroid coat is usually separated from the
pigment layer, but adheres closely to the sclera. In general it
is thicker at the back part of the eye, and quite decidedly so
at the entrance of the optic nerve. The lens is normal. Its
size is given in the table at the end of the paper.

The layers of the retina are well developed in the larva.
The retina of the larva differs from that of an Amblystoma
larva in the greater thickness of its ganglionic layer. This
layer is, in the young larva of Typhlotriton, composed of five
or six layers of cells. This thickness may in part be an arti-
fact, since the retinae examined are shrunken away from the
pigment epithelium, and the ganglionic layer is in contact with
the lens. In the larva 90 mm. long this layer has been reduced

No. i.] BLL\n VERTEBRATES OF NORTH AMERICA. 37

to not more than three series of cells. Aside from the differ-
ences noted above, the eye of the larvae of Typhlotriton is
apparently normal in all of its histological details. This rela-
tive thickness in the different sizes of the larvae may be gath-
ered from Figs. 2-5 and from the comparative table at the
end of the paper.

Figs. 2-5 are drawn with the same magnification and show
the relative thickness of the different layers in the retinae of
the larvae of different sizes and of the adult. The adult retina
is reduced in thickness by the absence of the rods and cones
and the (partial ?) atrophy of the outer reticular layer and by the
thinning of the ganglionic layer. The ganglionic layer in the
adult contains from two to five rows of cells. In this respect,
the adult approaches the condition found in Amblystoma more
than the young does. The inner reticular layer is compara-
tively thick, that of the young being thicker than that of the
adult.

In the adult the inner nuclear layer is continuous with the
outer nuclear layer. (See Fig. 5.)

The inner nuclear layer consists of about seven series of cells
in the smallest larva and of four to seven in the largest. The
cells in the preparations available cannot be separated into bipolar
and spongioblastic layers, nor are horizontal cell layers distin-
guishable. The outer reticular layer is well differentiated but
quite thin in the larvae, and is irregular in outline, adapting
itself to the overlying nuclei which encroach on its outlines.
In the adult this layer is indistinguishable by the same methods
that make it conspicuous in the larva. In places there appeared
an open space where the outer reticular layer should be (Fig. 9),
but none of its structure remains. It is fair to suppose that
the fibers forming this layer are resorbed during the meta-
morphosis. This layer seems to be the very first obliterated by
the processes of degeneration both ontogenic and phylogenic
in this as in other vertebrates with a degenerating eye.

The greatest change during and shortly after metamorphosis
takes place in the layer of the rods and cones. In the larva
35 mm. long, from the mouth of Rock House Cave, the rods
reach an extreme length of 50 /x. The relative sizes and

38 EIGENMANN AND DENNY. [VOL. II.

number of these as compared with the much smaller cones
may be gathered from Fig. 2, a.

In the larva 90 mm. long the outer segments of the rods are
much shorter and stain less conspicuously than in the younger.
The nuclei of the outer nuclear layer are distinctly in two
layers, whereas in the younger they are in three less regular
layers. The cones are correspondingly fainter than in the
young. It is surprising that whereas in the larva 90 mm.
long we find the rods and cones well developed they have
greatly degenerated or practically disappeared in the adult
only a few mm. longer. In an adult specimen 97 mm. long
the rods have retained their normal shape and position, but I
have not been able to detect any differentiation into inner and
outer segments. In longer ones most of the nuclei of the
outer series have become rounded at both ends. But one
cone was found in eyes of the adult over 100 mm. long. It
is shown in Fig. 6. In an adult specimen 103 mm. long
filmy rods are still evident. They appear as conical spaces
above the nuclei free from pigment rather than as possessing
any demonstrable structure. Just at the margin of the place
where the pigment has been torn from the retina one of these
is drawn out to a great length. The pigment in this individual
extends in places down between the cells of the cones. This
latter condition appears in a very exaggerated form in the
eye of Typhlomolge. In tangential section this condition
and the filmy rods give rise to the appearance represented in
Fig. 5, a.

Distinct signs of ontogenic degeneration are also seen in
other parts of the retina. For instance, many nuclei of the
inner series of the outer nuclear layer are shriveled. In some
eyes the ganglionic nuclei have for the greater part lost their
granular structure and show a homogeneous pasty condition,
only a few cells with granular nuclei being present (Fig. 5).
The same is true in large part of the inner nuclei of the inner
nuclear layer. This condition of the ganglionic nuclei is not
entirely confined to the adult but is also found in the larva.

Some of the modifications in the shapes of the outer nuclei
in the adult are shown in the figures. In Fig. 7 the upper

No. i.] BLIND VERTEBRATES OF NORTH AMERICA. 39

portion of the nucleus is very much elongated. This form is
of frequent occurrence. In Fig. 8 is shown the common
form where the nuclei are simple elliptical bodies, which give
no evidence whatever of any processes uniting them with the
other elements of the retina. The Miillerian fibers are pro-
fusely present and of very large size in both larva and adult.
In both adult and young the optic nerve enters as a single
strand and passes entirely through the layers. A heavy mass
of pigment is found following the optic nerve to within a short
distance of the brain.

AVERAGE MEASUREMENTS OF THE EYES OF TYPHLOTRITON.

LENGTH OF SPECIMEN-.

35
mm.

48
mm.

62
mm.

9o

mm.

97
mm.

103

mm.

1 06
mm.

Vertical diameter of eye ....

810

800

—

960

—

800

1170

From front of lens to back of eye

600

672

—

72O

720

72O

Hj4

Outer nuclear layer with the rods

76

42

I 12

36

28

28

—

Outer reticular layer

i

'J

Inner nuclear layer

76

72

So

5°

48

72

72

Inner reticular layer

16

2O

16

2A

8

8

I -3

Ganglionic layer

68

56

64

32

24

26

26

4

16

8

20

22

Optic nerve

2O

2C

2T.

2Q

Lens . ...

742

7OO

soo

4^2

4^O

"iO4

SUMMARY.

Typhlotriton is an incipient blind salamander living in the
caves of southwestern Missouri. It detects its food by the
sense of touch without the use of its eyes. It is stereotropic.
Its eyes show the early stages in the steps of degeneration
from those of salamanders living in the open to those of the
degenerate Typhlomolge from the caves of Texas. The lids
are in process of obliteration, the upper overlapping the lower
so that the eye is always covered in the adult. The sclera
possesses a cartilaginous band in the larval stages but not in
the adult. The disappearance of the cartilage is probably an
incident of the metamorphosis, not of the degeneration the eye
is undergoing. The lens is normal. The retina is normal in

4O EIGENMANN AND DENNY. [VOL. II.

the larva with a proportionally thicker ganglionic layer than in
the related epigaean forms. Marked ontogenic degenerations
take place during and shortly after the metamorphosis, a. The
outer reticular layer disappears, b. The rods and cones lose
their complexity of structure, such as differentiation into inner
and outer segments, and finally are lost altogether.

EXPLANATION OF FIGURES.

All drawings were made with the aid of the Abbe camera from sectioned
balsam preparations. The comparative measurements (p. 39) furnish the key
to the magnification :

ps. palpabra superior. pi. palpabra inferior,

i. pigment epithelium. 2. rods and cones.

3. outer nuclear layer. 4. outer reticular layer.

5. horizontal cell layer. 6. inner nuclear layer.

7. spongioblastic layer. 8. inner reticular layer.

9. ganglionic layer. 10. optic fibers.

FIG. i. Diagrammatic representation of the eye drawn to scale.

FIG. i, a. Vertical section through the cornea and lids of an adult.

FIG. 2. Section of the retina, exclusive of pigment cells, of a larva 35 mm.
long.

FIG. 2, a. Tangential section through the rods and cones about on a level
with the innermost extent of the pigment which is seen on the right, showing
the relative sizes and abundance of the rods and cones.

FIG. 3. Section of the retina of a larva 48 mm. long.

FIG. 4. Section of the retina of a larva 90 mm. long.

FIG. 4, a. Tangential section showing the rods and cones at about the inner
limit of the pigment which is seen on the left.

FIG. 5. Section of the retina of an adult 106 mm. long.

FIG. 5, a. Tangential section at about the inner limit of the pigment.

FIG. 6. The only cone found in the eyes of adults.

FIG. 7, 8. Difference in the shape of the outermost series of cells in the
outer nuclear layer.

FIG. 9. Section of the retina of an adult 97 mm. long.

No. i.] BLIND VERTEBRATES OF NORTH AMERICA. 41

Volume //.] November, 1900. \No. 2.

BIOLOGICAL BULLETIN.

THE HABITS OF PONERA AND STIGMATOMMA.1

WILLIAM MORTON WHEELER.

IN a recent number of the Biological Bulletin'2' I described
the habits of three Texan ants belonging to the subfamily
Ponerinae. During the past summer an excellent opportunity
presented itself to extend these observations to two other
forms widely distributed in the Eastern and Northern States,
vis., Ponera coarctata Latr. and Stigmatomma pallipes Hald.
These are of no little interest to the student of ant life, the
former as a member of the typical genus, the latter as the only
known North American representative of the most primitive
tribe (Amblyoponii) of the subfamily. European myrmecolo-
gists have long wished to gain some knowledge of the habits
of P. coarctata, but its rare and local occurrence on their con-
tinent has rendered this impossible up to the present time.
The European type of the genus Stigmatomma, S. dcnticn-
latum Roger, is also rarely seen, and for the same reason its
habits are all but unknown.

As both the ants to be considered in this paper are subter-
ranean and very timid, it is impossible to learn much about
them in their natural environment. It is therefore necessary
to keep them in artificial nests. This is, fortunately, an easy
matter, since the ants are very hardy. As the colonies are small,
it suffices to use for this purpose the Petri dishes employed by

1 Contributions from the Zoological Laboratory of the University of Texas, No. 10.

2 "A Study of some Texan Ponerinae," Biol. Bull. Vol. ii, No. i, pp. 1-31.
October, 1900.

43

44 WHEELER. [VOL. II.

bacteriologists for growing cultures of micro-organisms. The
ants are hastily scooped up, together with their larvae, pupae,
and much of the earth in which they have excavated their nest,
and the whole is transferred to a Petri dish. One or two glass
slides are then placed on the earth, which is spread out till it
forms a layer not more than about 5 mm. in thickness. The
Petri dish is kept covered to retain the moisture in the soil.
In the course of a day or two the ants excavate rough-walled
chambers under the slide and galleries in the adjacent soil, of
the same size and shape as those which they are in the habit
of forming in their natural nests. They also gather their eggs,
larvae, and pupae into these chambers, where they may be easily
seen. When the slides become smeared or covered with earth
they can at any time be hastily replaced by clean ones without
greatly disturbing the ants.

The Ponerinae may appear to lead very monotonous lives to
any one who has kept under observation the different species
of Myrmica, Pogonomyrmex, Lasins, Camponotus, and Formica.
But this very monotony is full of interest to the observer who
sees in the rudimental activities of these ants a certain picture,
however imperfect, of the simple stages through which the
higher ants have passed in attaining to their present remark-
ably differentiated social organizations. It can hardly be
doubted that there is a phylogeny of instincts, as there is a
phylogeny of structures, and there is certainly no single ani-
mal group which more clearly illustrates the truth of this
statement than the Formicidae.

PONERA COARCTATA LATREILLE.

Our American P. coarctata is considered by Emery 1 to differ
sufficiently from the European form to be ranked as a sub-
species, which he calls pennsylvanica Buckley. In the worker
the single node forming the pedicel of the abdomen is some-
what thicker and much broader behind and less narrowed
anteriorly than in the European forms. The punctation of

1 " I'.eitrage zur Kenntniss der nordamerikanischen Ameisenfauna " (Schluss),
Zool. Jahrb. Abth. f. System. Ed. viii, pp. 257-360, Taf. VIII. 1894.

No. 2.] HABITS OF PONERA AND STIGMA TO MM. 1. 45

the head is finer, that of the thorax and node much denser and
more distinct. Emery also mentions some differences in the
neuration of the wings of the male : " in den Fliigeln verbindet
sich aber die Costa rccurrcns etwas weiter von der Gabelung
mit clem hintern Ast der Costa cnbitalis, ungefahr wie bei der
europaischen P. piinctatissima."

Figs, i, 2, and 3, from camera drawings, represent the out-
lines of the male, female, and worker of the American coarctata.
The eyes of the worker are minute and vestigial, those of the
female considerably and those of the male very much larger.
The worker has no ocelli ; those of the female are small, while
in the male they are very prominent. The node in the male
and female is more slender than that of the worker, and of a
somewhat different shape. The antennae of the male are of
nearly uniform thickness throughout and 13-jointed, whereas
the geniculate antennae of the female and worker are 12-
jointed, with a long basal joint, or scape, and a club-shaped
funicular portion, with much shortened middle joints. The
worker and female are provided with a long sting ; while the
pygidium of the male ends in an acute point.

The coloration of the female and worker is highly variable.
Typical specimens have the head, thorax, node, and base of
the abdomen black, the mandibles, clypeus, frontal carinae,
antennae, legs, posterior third of the first abdominal segment,
and the tip of the abdomen from the base of the fourth seg-
ment, red or yellow. Very often the ventral portions of the
trunk are more or less suffused with red or yellow, especially
when the specimens are immersed in alcohol. Some specimens,
probably more or less immature, are red or yellow throughout.
The body is covered in all cases with short pale pubescence,
which on the head forms two lines, one on either side running
parallel with the straight lateral edges. These lines are appar-
ent only in dry specimens seen in a certain light. The male
is black, with the palpi, trochanters, knees, tips of tibiae, and
the tarsi light yellow. The genitalia and the incisures of the
segments of the slender abdomen are also more or less yellow
or piceous, as are also the stigma and veins of the colorless
wings, both in this sex and in the female.

46

IV HEELER.

[VOL. II.

P. coarctata is a small ant, the male and female measuring
scarcely more than 4 mm. in length, while the workers vary
from 3 to 3.75 mm.

According to Emery this ant occasionally presents ergatoid
females. He mentions 1 two of these wingless individuals from
Sicily, with eyes somewhat larger than those of the worker,
with ocelli and with the node somewhat higher and more slender
above. I have been unable to find any such specimens among

my American mate-
rial, although I care-
fully scrutinized no
less than two hundred
wingless individuals
from widely separated
localities and from at
least twenty different
nests.

P '. coarctata is the
most widely distrib-
uted of the Euro-
pean Ponerinae and
occurs even in north-
ern Africa (Algiers),
according to Emery.2
In this country, too,
its subspecies, pcnn-
sylvanica, is one of the most widely distributed forms in the
subfamily. Emery 3 has examined specimens from Pennsyl-
vania, New Jersey, Virginia, Maryland, Mississippi, Florida,
and Ohio. Forel has observed it in North Carolina,4 and I
can add to this list four other states, viz., Wisconsin, Illinois,
Massachusetts, and Connecticut. It may, I think, be safely
said to inhabit all the states east of the Mississippi, as well as
Canada.

FIG. i. —Ponera coarctata. Latr., subsp. pennsylvanica
Buckl. Male.

1 " Sopra Alcune Formiche della Fauna Mediterranea," Mem. letta alia R.
Accad. delle Science dclT Istituto di Bologna. Pp. 1-19, Tav. I. 2\ Aprile, 1895.

2 loc. fit., p. 6. 3 " Beitrage zur Kenntniss," etc., loc. cit., p. 268.

4 Aniia/cs de la Soc. Entomol. de Belgiqite. Tome xliii, pp. 438-447. 1899.

No. 2.] HABITS OF PONERA AND STIGMATOMMA. 47

It is undoubtedly far more common in this country than in
Europe. In July I found numerous nests at Rockford, 111.,
both under the bark of old logs and under stones along the
streets of the town.
It is not uncommon
in similar locations
at Woods H o 1 1,
Mass., and very
abundant under
stones on the slopes
of Mt. Pisgeh (alti-
tude 1450 feet), at
Colebrook, Litchfield
County, Connecticut.1

1 This last locality, to-
gether with the slope of a
small neighboring hill, is a
rich collecting ground for
ants. I give here the com-
plete list of the forms
taken there during August,
as it probably embraces
nearly all the species of
Formicidae that occur in
New England : Brachy-
myrm ex He eri F o r e 1 .
subsp. depilis
Emery ; Lasi.us
niger L. ; L. fla-
1'iis L. ; L. umbra-
tits Nyl., subsp.
mixtits, \zx.aphi-
dicola Walsh;
L. latipfs Walsh ;
Formica sangid-
nca Latr., subsp.
rubicunda Em. ;
F. cxsectoides
Forel, var. opaci-
ventris Em. ; F.

pallide-fulva Latr., subsp. Schaufussi Mayr ; F. pallide-fulva, subsp. nitidiventris
Em.; F. ftisca L., var. subsericea Say; F. fusca, var. subaenescens Em.; F. fusca,
subsp. subpolita Mayr, var. neogagates Em. ; Camponotus kerculeanus, subsp. ligni-
perdns Latr., var. novaeboracensis Fitch (= pictus Forel); C. Jierculeanus, subsp.
pennsylvanicus de Geer; Stigmatomma pallipes Hald. ; Ponera coarctata Latr.,

FIG. 2. — Ponera coarctata Latr., subsp . fennsylvanica
Buckl. Virgin female.

FIG. 3. — Ponera coarctata Buckl., subsp. fennsylvanica Buckl. Worker.

48 WHEELER. [VOL. II.

P. coarctata is not found in deep woods or in damp places,
but prefers rather dry localities more or less open to the sun-
light. The margins of woods and along stone walls are favor-
ite haunts, under stones rather deeply imbedded in a rich soil,
especially leaf mold. Here it excavates a small, irregular cham-
ber, from which a few straggling burrows run off into the
neighboring soil. In some cases the chamber and burrows
are found under the lower surface of the stone, but I have
gained the impression, from the examination of many nests,
that the ants often prefer the vegetable mold nearer the sur-
face, where it overlaps the sides of the stones. Chambers and
galleries of the same irregular pattern are excavated in the
rotten wood when the ants nest under the bark of old logs.
The larvae and pupae are reared in the chambers. In late
June and early July the nests contain eggs and larvae but no
pupae; during the latter half of July and the month of August
only cocoons are found, usually crowding the chamber so that
the ants have little space in which to move over and among
them. The imagines begin to hatch during the last two weeks
of August and the first week of September. Even by the
latter date I have seen no eggs nor larvae to represent a
second brood.

The number of individuals composing a colony varies in dif-
ferent nests and with the advance of the summer. As the
ants are very timid and at once seek refuge in their galleries
as soon as the stone that covers the nest is moved, it is not
easy to determine their precise numbers. None of the nests
opened at Rockford, 111., July i, contained more than eight or
ten ants, including a single female. As soon as the cocoons
begin to hatch, the colony increases rapidly. One rather
typical nest, opened at Colebrook, Conn., August 24, con-
tained six males, one female (with wing stumps and evidently
the mother of the colony), one callow virgin female, twelve

subsp. pennsylvanica Buckl. ; Myrmicuia Latreillei Curt., subsp. americana Em. ;
Formicoxenus nitidulus Nyl.; Solenopsis molesta Say; Crematogaster lineolata Say,
var. ; Stcimmma (Aphaenogastcr'} fnlvum Rog., subsp. aquia Buckl., var. piccnnt
Km.; Myrmica rubra L., subsp. scabrinodis Nyl., var. Schencki Em.; Tapinoma
sessile Say.

No. 2.] HABITS OF P ONER A AND STIGMATOMMA. 49

workers, and forty-four cocoons. A few nests examined some-
what later in the month contained a greater number of individ-
uals, so that fifty to sixty is perhaps not too great an estimate
for a large colony by the first week in September.

The winged males undoubtedly leave the nests like the males
of other ants, as I have taken them in the sweep-net in the
grass while collecting small Diptera. I have also seen the
males copulating with the newly hatched females in the same
nest. The small size of the nests in the early summer would
seem to indicate that the large number of workers in the late
summer and early autumn must split up into several detach-
ments, each with a young queen, and migrate to different
localities. My reasons for making this statement, apart from
the above-mentioned mating of the young queens within the
parental nest, are largely of a negative character, but they may
be given for what they are worth. First, I have observed no
tendency in the young queens, while they possess wings, to
leave the nests like the males ; second, the wings are often
lost very soon after hatching, sometimes before the queen has
acquired her deep adult coloration ; and, third, I have never
found a solitary queen in the act of founding a nest, either
of this or of any other of the five species of Ponerinae I have
studied, although I have frequently seen the young fertilized
queens of Camponotus, Formica, Lasius, Tapinoma, Cremato-
gaster, Stenamma, Myrmica, and Pogonomyrmcx starting their
colonies. The fact that the colonies seem to be annual instead
of perennial growths, as among other ants like those above
mentioned, is of considerable interest. It points to very primi-
tive conditions in the Ponerinae, especially as the same is also
true of tropical forms like PacJiycondyla and Leptogenys, which
can hardly be destroyed by severe winters. Thus what was
at one time erroneously supposed to be true of the more spe-
cialized ants, viz., the founding of a colony by a young female
leaving the parental nest like the young queen of the hive bee,
accompanied by a number of workers, may prove to be the
normal method of nest formation with the Ponerinae. If this
supposition is correct, there must be considerable inbreeding
in the colonies of these ants, as the females would be regularly

WHEELER. [VOL. II.

fertilized by males from the same nest. There may be some
connection between this condition and the limited productivity
of these ants, and the strong tendency to parthenogenesis seen
in some of the species (e.g., in the ergatoid females of Pachy-
condyla liarpax}.

The behavior of P. coarctata towards individuals of the same
species from different nests is very similar to that observed in
Pachycondyla. If two nests be thrown together into the same
dish, there may be no immediate signs of hostility ; but after a
few hours have elapsed, the ants are found struggling together
in pairs. They interlock mandibles or tug at each other's legs
and antennae, or even wrestle fiercely, intertwining their long
bodies and trying to use their slender stings. These contests
may be renewed from time to time for many days, whenever
two individuals from different nests happen to meet, but deaths
are rare, and ultimately the colonies fraternize completely.
Long before the ants have settled their various difficulties,
however, the cocoons and larvae of both nests are brought
together as common property. A dozen different nests can
be compounded quite as easily as two, and a few ants from one
nest can be induced to adopt a large number of cocoons and
larvae taken from half a dozen different nests.

The eyes of the workers of P. coarctata are so very small
that they can hardly be of much service as visual organs. The
actions of the ants indicate that they are guided very largely
by their extremely sensitive antennae. They are, undoubtedly,
able to detect the difference between light and darkness, but
the fact that they do not seem to mind exposure to the light,
provided they are covered with a glass plate, leads me to infer
that they are rather positively stereotropic than negatively
heliotropic. Of course their preference for an atmosphere
charged with a certain amount of moisture --their positive
hygrotropism-- leads them to seek refuge in dark places, under
stones or the bark of old logs.

I have not been able to ascertain the food of these ants in a
state of nature. They probably kill and imbibe the juices of
very small subterranean insects. In captivity they are omniv-
orous, feeding readily on raw or boiled meat, yolk of eggs,

No. 2.] HABITS OF PONERA AND STIGMATOMMA. 51

corn bread, or even on " Boston brown bread." They do not
appear to share the fondness of ants in general for sugar dis-
solved in water. When kept for a time without food they eat
their dead companions or their own eggs, larvae, and pupae.

The workers of Poncra are never seen feeding one another
with regurgitated food, like the different species of Formica,
Lasius, and Rlyrmica. Even the queen is obliged to feed her-
self. The workers bestow on her no special attentions, nor
does she enjoy any of the privileges of the queens of the
above-mentioned specialized genera, after they have • once
established their colonies. Like any one of the workers, she
takes part in digging the galleries, wanders out in search of
food, assists in transporting and cleansing the eggs, larvae, and
cocoons, and in feeding the larvae. Although not expressly
stated in my former paper, this is also true of the ergatoid
females of Leptogenys and Pachycondyla. This would seem to
indicate a decidedly primitive condition, since the activities of the
females of the Ponerinae never pass beyond the stage exhibited
by the females of the more specialized ants only while they
are raising their first batch of workers.1

In the scrupulous care of their nests, colonies of P. coarctata
closely resemble the more specialized ants. They bury their
food or any liquid or strong-smelling substance in their environ-
ment, and all refuse -- dead ants, dead pupae, empty cocoons,
etc. — is deposited in one corner of the nest.

The eggs of P. coarctata are oblong, like those of the other
Ponerinae I have described (Pachycondyla, Leptogenys}, and of
very large size — fully .6 mm. long, or nearly as large as the
thorax of the insect that lays them. The number produced
at one time is, however, relatively small. Only three were
deposited by one female in my nests July 20. As the
larvae found in nests in early July are of very different sizes,
we must assume that the queen lays a few eggs at a time,

1 In this connection it is interesting to note that, as Janet has shown (•' Nids
artificiels en platre. Fondation d'une colonie par une femelle isolee," Bull. Soc.
Zoi'/., tome xviii, p. 168, France, 1893), the female of the more specialized ants,
when separated from all her workers, may return to and repeat all the activities
which she displayed while founding her first colony.

WHEELER.

[VOL. II.

a

probably at intervals of a few days or a week. It is quite
possible that some of the workers, acting as ergatoid females,
may contribute unfertilized eggs which give rise to the males
that are found in nearly every colony late in August.

The larva (Fig. 4, a) is clearly of the Ponerine type, though
differing in a few important particulars from any of the larvae

of the five genera
(Leptogenys, Pachy-
condyla, P o n e ra,
OdontomacJms, and
Diet cam in a] de-
scribed by Emery 1
and myself.2 It is
rather robust, with
a large head suc-
ceeded by five dis-
tinct segments. The
remaining segments,
forming the swollen
abdomen, are not
distinctly marked
off from one another.
The body is fur-
nished with out-

FIG. 4.— a, larva of Poneracoarctata Latr.,subsp. pennsylvanica ,

Buckl. Nearly ready to pupate. 6, bristle-capped tubercle of gTOWthS OI . il TCC
same; c, head of same (dorsal aspect). different tVDCS The

first of these is represented by a number of pointed bristles
confined to the ventral surface of each segment. The
second type is represented by several longitudinal rows
of pointed tubercles, each of which, under a high magni-
fication (Fig. 4, b) is seen to consist of a short distal spine and
a long, tapering proximal base, directly continuous with the
integument of the larva, and covered with transverse rows of
serrated points. The distal spine is movably articulated with
the proximal portion, and is so easily detached that it may be

1 " Intorno alle I.arve di Alcune Formiche," Mem. letta alia A'. Accad. delle
Scienze dell' Istititto di Bologna. Pp. i-io, 2 Tav. 7 Maggio, 1899.

2 Loc. cit., pp. i 5-22.

No. 2.] HABITS OF PONERA AND STIGMA TOMMA. 53

overlooked. The third type of projection is found only on the
dorsal surface of the third to sixth abdominal segment as four
pairs of club-shaped structures which are glutinous to the
touch. That these are peculiar modifications of the tapering
tubercles seems to be indicated by the fact that they replace
on either side in each of the four above-mentioned segments
the more posterior of the two pointed projections seen in the
thoracic, first and second, and seventh and eighth abdominal
segments. The larva is usually kept on its back, so that the
four pairs of glutinous tubercles act as suckers and fix it to
the sides of the earthen chamber or to the glass of the artificial
nest. The ants have to exert a slight effort in pulling the
larva away from its attachment. The head of the larva in
dorsal view is shown in Fig. 4, c. It is broad, evenly rounded
behind, and beset with short stiff bristles. The labrum is
bilobed and does not extend beyond the tips of the powerful
tridentate mandibles. The fleshy maxillae and labrum pro-
ject somewhat beyond the mandibles, the former being pro-
vided with robust tactile cones, the latter with a prominent
median tubercle on which opens the duct of the spinning gland.
Comparison of the figures in this and my previous paper shows
that the larva of P. coarctata is peculiar in lacking the circlets
of bristles on the pointed projections and in possessing clavate
adhesive tubercles on the dorsal surface of the abdomen.

The larvae are fed in the very same manner as the larvae of
the large Texan Ponerinae, i.e., with pieces of food and not
with liquid regurgitated by the ants. In confinement I did
not succeed in inducing the ants to feed their larvae with
fragments of insects, but they carried crumbs of moistened
corn bread to them, and the larvae could be seen lying on their
backs, attached by their glutinous dorsal tubercles, slowly con-
suming the morsels which had been placed on their flattened
ventral surfaces. The fixation of the larva to the walls of the
nest seems to be an adaptation for giving freer play to the
head and slender neck during feeding.

The oblong elliptical cocoons of coarctata are of a light buff
or cream color, and vary from 2 to 3.5 mm. in length. They
closely resemble the worker cocoons of Lasius umbmttis mixtus

54 WHEELER. [VOL. II.

Nyl., var. apkidicola, an ant which in Massachusetts and Con-
necticut is often found under the same stones with the Ponera.
The larger cocoons belong to the males and females, the
smaller ones to the workers.

The eggs and larvae are looked after with great care by the
ants, as Forel has observed.1 On a former occasion, however,
I expressed doubt concerning the validity of Forel's further
statement : " Lorsqu'on decouvre un nid de Ponera dans un
tronc pourri, on voit leur cocons jaunes assembles dans un
coin, mais absolument abandonnes des 9 qui n'essaient pas
de les sauver, ni de les recueiller." I have since had frequent
opportunity to observe these ants, and I am convinced that the
master myrmecologist is in error. It is true that the slightest
disturbance of the nest causes the ants to retreat into their
galleries and to forsake their cocoons, but when one stops to
watch the nest a few moments, one is sure to see the ants
returning one by one and stealthily removing their charges.
This they do rather awkwardly, walking backwards and drag-
ging the cocoons away without lifting them from the ground,
in marked contrast with Lcptogenys clongata, which straddles
the cocoon with its long legs and carries it away with surpris-
ing dexterity. Simple experiment with the artificial nests
shows that the cocoons of Ponera, when removed to a distance
of three or four inches from the chamber in which the ants
have stored them, are taken back in the space of ten to thirty
minutes. Nevertheless, Forel certainly deserves credit for direct-
ing attention to this matter of the care of the cocoons, for if one
has observed the way in which a large and highly specialized ant,
like our northern Formica pallidefulva ScJianfnssi, e.g., when its
nest is uncovered, rushes out in the very face of danger to rescue
its cocoons, the slow and awkward methods of P. coarctata cer-
tainly indicate a more primitive or possibly degenerate condition
quite in harmony with the other habits of this feeble little ant.
Further evidence that these ants care for their cocoons is seen
in their habit of continually creeping in and out among them, and
in the time which they devote to licking and cleansing them when
there are no longer any larvae to require these attentions.

1 Loc. cit., p. 443.

No. 2.] HABITS OF POA'ERA AND STIGMATOMMA. 55

Forel goes on to say of P. coarctata l : " Je soupc_onne que
chez ces fourmis, moins sociales que les autres, les nymphes
sortent seules cle leurs cocons, sans avoir besom de 1'aide des
9." For the purpose of testing this supposition, I tried to
surprise the ants in the act of leaving their cocoons, but I was
not successful, notwithstanding numerous workers, males and
females, were hatching in my artificial nests. A large number
of cocoons from several different nests, and apparently in a
healthy condition, were isolated in small dishes, but they failed
to hatch. On the other hand, the workers opened many
cocoons, extracted the dead or moldy pupae, cut them up
into pieces, ate portions of them, and deposited the remainder
on the refuse heap. I shall show when I come to consider
Stigmatomma pallipcs that these acts, which resemble what I
formerly described for Leptogenys and regarded as indirect
proof that the living callows are assisted in their escape from
the cocoons, are of no value as evidence in this matter.

On hatching, the workers of Ponera are light dirty yellow,
and very gradually, in the course of several days, acquire their
dark color. The abdomen remains pale longer than the head
and thorax. The females are more mature on hatching, hav-
ing the head and thorax brown. The males are quite black
and fully mature soon after leaving the cocoon.

In concluding this account of P. coarctata a few myrme-
cophiles that dwell with this ant may be mentioned. In two
nests, one from Rockford, 111., the other from Colebrook, Conn.,
I found a small brown Pselaphid beetle. In a single nest in
the latter locality a minute Staphylinid was taken. In this
locality also were found some peculiar mites, often attached in
pairs on either side of the node and the first abdominal seg-
ment. Their symmetrical position resembles that of the
mites AntcnnopJiorus and Discopoma infesting the Lasius um-
bratus mixtus Nyl. of Europe.2 Wasmann3 enumerates as

1 Loc. cit., p. 443.

2 See Janet, fitudes sur les fourmis, les guepes, et les abeilles. Note 13.
Sur le Lasius mixtus 1'Antennophorus Uhlmanni, etc. Pp. 1-62, 1 6 figs. Limoges,
1897.

3 Kritisches Verzeichnis der myrmekophilen und termitophilen Arthropoden.
Berlin, 1894.

56 U'HEELER. [VOL. II.

myrmecophiles of P. coarctata in North America the Staphy-
linid beetle Apocellns (?) sphaericollis Say, and in Europe
the two Pselaphid beetles : Trichonyx sulcicollis Rchbch and
Amanrojiy.v Mdrkcli Aube.

STIGMATOMMA (AMBLYOPONE) PALLIPES HALDEMANN.

Although two species of Stigmatomma are known from south-
ern Europe (S. dcnticnlatiim Roger and 5. impressifrons Emery),
both are of rare and local occurrence. On the former species
Emery1 has published the following note: "Dr. Alessandro
Fosi had the good fortune to observe the nest of S. denticu-
latum while excavating antiquities at Verucchio near Rimini.
These nests were found on several occasions in the Umbrian
cinerary urns, and always at the surface of the layer of ashes
with bone fragments found beneath the earth which had per-
colated in between the lid and the original contents of the
urn. It was, however, impossible to obtain the winged indi-
viduals, nor could anything further be ascertained concerning
the habits of these singular subterranean ants. The popula-
tion of a nest which I had occasion to examine comprised
about forty individuals, three of which were females."

Our American species, too, is considered to be an uncom-
mon insect, although it is widely distributed through eastern
North America, from Canada to North Carolina.2 I have
found it both on the Island of Naushon, near Woods Holl,
Mass., and at Colebrook, Conn. In the former locality it was
very common in some rather open oak woods at the north end
of the island, under large stones, imbedded in rich vege-
table mold. Here, on August 6, I uncovered some thirty
nests in the course of three hours. At Colebrook, after
the most careful search, I succeeded in finding only three
nests (August 16 to 31), and these were in widely separated
localities. All the nests on Naushon Island contained great
numbers of cocoons, but in only three were there eggs or
larvae. One of these contained a few mature larvae ready to

1 " Sopra Alcune Formiche," etc., loc cit., p. 3.

2 Emery, " Beitrage," etc., loc. cit. ; Forel, loc. cit.

No. 2.] HABITS OF POXERA AND STIGMATOMMA. 57

pupate ; the two others contained several packets of eggs and
many young and half-grown larvae. One of the nests found
in Colebrook, August 29, contained numerous callows and
a few young larvae. These observations, together with the
fact that the cocoons collected on Naushon Island nearly all
hatched before August 20, show that Stigmatoinnia normally
produces two broods during the summer. In this respect it
may differ from P. coantata ; for, as I have said, no eggs or
larvae of this species were found in several nests examined the
last of August and the beginning of September.

>S. pallipcs seems to be so completely subterranean in its
habits that it does not come to the surface even at night.
Its nests are like those of P. coarctata, and it also often digs its
galleries in the vegetable mold overlapping the edges instead
of beneath the center of a stone. It is a much larger ant than
P. coarctata, measuring 5.5 to 7.5 mm. One of the colonies
taken on Naushon Island consisted throughout of very small
individuals (5.5 to 6 mm.). The Colebrook individuals were
all of smaller size than the majority of those from Massa-
chusetts.

The females (Fig. 5), of which each colony contains from
one to four before the hatching of the cocoons, are of the same
size or slightly larger than the workers. They differ from the
workers (Fig. 7) in having much larger lateral eyes, in having
ocelli and wings, and in the structure of the thorax. Both
females and workers are of a rich reddish-brown color ; in
older specimens the head, thorax, and node are almost or quite
black, while the abdomen and legs are much paler. The male
(Fig. 6) is black, with the two basal joints of the antennae, the
trochanters, tibiae, and tarsi yellow ; the remainder of the
antennae reddish. The head, thorax, and anterior portion of
the node are opaque and coarsely punctate or wrinkled, whereas
the pleurae, scutellum, posterior edge of the node, and the
abdomen are glabrous. The black stigma of the colorless
wings in both sexes is large and conspicuous. Interesting
morphological characters, such as the structure of the antennae
in the two sexes, the remarkable dentate mandibles and clypeus
in the female and worker, the venation, etc., are represented in

WHEELER.

[VOL. II.

the figures and need no further description. There were no
ergatoid females among the many specimens collected and
reared, but it is quite possible that some of the workers may

lay eggs, and, though lacking
ocelli, such individuals could
be called ergatoid females in

FIG. 5.

FIG. 6.

FIG. 7.
Stigmatotnma pallipes Haldem. Fig. 5, Virgin female ; Fig. h, male ; Fig. 7, worker.

the sense in which I have used that term (perhaps somewhat
inaccurately) in my description of Pachycondyld hat-pax.1

1 Li>f. <•//., p. 5.

No. 2.] HABITS OF PONERA AND STIGMATOMMA. 59

Much of what I have said concerning the size and growth of
the colonies of Poncni may be repeated for Stigniatouiuia.
One colony taken at Colebrook contained only two workers,
and another seven workers and a female ; but nearly all that
were collected on Naushon Island were larger, varying from
ten to twenty individuals. As some of these colonies had
from four to six times as many cocoons as ants, the colonies in
the artificial nests by the end of August contained from forty
to sixty individuals, including in some instances ten or a dozen
females and as many males.

Different nests of .5". pallipcs fraternize after a struggle in
much the same manner as other species of Ponerinae. When
my supply of vials gave out, while collecting the numerous
nests of this species on Naushon Island, Iwas obliged to put
some fifteen nests into a single glass jar. There was consid-
erable struggling among the ants of the different nests for a
few days, but eventually they settled down peacefully and
attended to their cocoons in common. Very few of the ants
were killed in the struggle, and these were usually small indi-
viduals. A portion of this compounded colony was taken in a
small dish on a day's railway journey, August 9. On arriving
at my destination I found most of the ants killed, and a few
that were still fighting died a few days later. The unusual
severity of the struggle in this case was probably due to the
close confinement of the ants in a small receptacle and the
jarring to which they had been subjected for several hours.
Four ants and a number of larvae taken at Colebrook, Conn.,
August 28, were placed in a Naushon nest which had just
hatched its last cocoons. The larvae were at once appropriated
by the Massachusetts ants, and later in the evening a sharp
struggle ensued between the members of the two colonies.
On the following morning one Connecticut ant was found
dead, the three others had gone over to the enemy, and the
whole colony was busy cleansing the larvae.

The eyes of the workers of Stigmatomma are even more
rudimental than those of Ponera. The reactions to light and
darkness, to contact and to moisture, closely resemble those of
the above-described species. The females, notwithstanding

6o IV HEELER. [VOL. II.

their much larger eyes and their ocelli, have the same timid,
groping habits as the workers. When fully mature the males,
like the males of Ponera, are positively heliotropic and nega-
tively stereotropic.

Stigmatoinuia appears to be very sensitive to low tempera-
tures. It passes the cold nights and mornings even during
August and September curled up, with its broad head covering
the tip of its abdomen. When dug out of the soil some
moments elapse before it straightens and begins to run about.
It probably passes the winter in the convoluted conditions.

For fully four weeks after my colonies were placed in arti-
ficial nests the ants refused to eat, although I tried a great
variety of foods. When small living insects were placed in
the nest they were cautiously attacked, the ant advancing,
snapping at them with its long mandibles, and then retreating.
This whole action was a very feeble imitation of the snapping
I have described for OdontomacJius.1 The insects thus killed
were not eaten, but covered with particles of earth. Finally
the ants consented to eat some of the larvae and pupae of
Formica pallide-fulva, and a little later they became very
fond of raw meat. While feeding, the huge mandibles are kept
closed with overlapping tips. The insects are obliged to slide
them over the food in order to reach it with the tongue and
maxillae. Like other ants, they are unable to swallow solid
food, but after rasping off a small mass with the tongue, they
press it in the subpharyngeal pocket, thereby extracting its
juices, and then spit out the small oblong ball of residue. The
whole process of forming and disposing of these " boulettes de

4

nettoyage " is exactly like that described by Janet for Formica
rtifn, Lasius jui.vtns, and Vespa crabro? The workers of Stig-
matomma were never seen feeding one another or their queens ;

1 Loc. cit., pp. 12-14.

Snr 1'organe de nettoyage tibio-tarsien de Myrmica rubra L.," Ann. Sof.
/:nt </<• /-'ranee, tome Ixiii, p. 697, 1895; " Sur Vespa crabro, histoire d'un nid
depuis son origine," Mem. Soc. Zool. de France, tome viii, p. 76, 1895; Sur le
Lasius mixtus 1'Antennophorus Uhlmanni, etc., pp. 16, 17. Limoges, 1897. — The
agricultural ant, Pcgonomyrmex I'nrbatiis Sm., when fed on starchy substances,
like rolled oats, literally sprinkles the floor and walls of its nest with snow-white
" boulettes."

No. 2.] HABITS OF POA'ERA AND STluMATOMMA. 6 1

nor can this be readily done by ants with such huge, horizontally
projecting mandibles.

Stigmatomma has a singular habit of vigorously shaking its
body from time to time, as if suffering with the ague. The
motion is not unlike that of the Termites when they are sup-
posed to be stridulating. It is possible that the action of the
Ponerine may produce a sound through the rubbing of the
various segments and
joints on one another, but,
though perfectly familiar
with the sounds produced
by the large Myrmicines
Pogonomyrmex and Atta,
I have not been able to
detect them in Stignia-
tomma.

What I have said con-
cerning the habits of
cleanliness in P '. coarctata
applies also to the species
under consideration.

The eggs of Stigma-
tomma, though deposited
by a larger insect, are
smaller than those of P.
coarctata, being only .5
mm. long. They are, how-
ever, more numerous. A packet deposited by a female in one
of my artificial nests contained thirteen eggs. In shape they
are oblong elliptical, like the eggs of other Ponerinae. They
are not arranged with their long axes parallel with one another,
as in Pachycondyla and Leptogenys, but in an irregular mass,
like the eggs of all the other subfamilies of ants, including the
Dorylinae (Ecitoii).

The larva has the appearance of Fig. 8, a. The body is
rather slender in alcoholic specimens, and the segments are
all quite distinct and clothed rather uniformly and densely
with yellowish hairs, which under a high power (Fig. 8, b} are

FIG. 8. — a, mature larva of Stigmatomma pallifes
Haldem ; b, bristle of same ; c, head of same (dor-
sal aspect).

62 ll'HEELER. [VOL. II.

seen to taper into very slender flexuous points. The head is
somewhat longer than broad and without hairs on its dorsal
surface, the labrum is bilobed, the maxillae provided with the
usual tactile cones. The outer one of these on either side
appears to be bifurcate. The young differs from the mature
larva only in having a relatively larger head and a sparser
covering of bristles. Comparison of the larva of Stigmatomma
with that of Poncra, Pachycondyla, Lcptogenys, and Odonto-
macJnts shows that it does not conform to the Ponerine type
but closely resembles, instead, the larvae of certain Myrmi-
cinae, which are also covered with hairs instead of bristly
tubercles.1

The cocoons of Stigmatomma are of a slightly darker, more
brownish color, and somewhat more oval than the cocoons of
Ponera. They measure from 4.5 to 6 mm. in length.

What I have said of the care of the eggs, larvae, and pupae
of P. coarctata is equally true of Stigmatomma.

The larvae are fed in the very same manner as other Ponerine
larvae. In one of the nests on Naushon Island a large larva
was seen with its head and neck inserted in the two last seg-
ments of a beetle larva (Tenebrionid}, which must have been
captured and disarticulated by the ants. In my artificial nests
the ants carried the larvae of Formica pallide-fulva to their
own young. The latter could be distinctly seen sucking the
juices of the Formica larvae till they were reduced to shriveled
skins. These were then carried away and placed on the refuse
heap.

I have succeeded in surprising the callows in the act of
escaping from their cocoons. This they do, as Forel believed
to be the case with P. coarctata, without any assistance from
the workers. Several cocoons were isolated in a watch-glass,
and I had an opportunity of seeing a female, two males, and
several workers emerge entirely by their own efforts. The
ant gnaws through the wall of the cocoon at a spot a short
distance behind the anterior pole. The shape of the incision
at once indicates whether a male or a female (or worker) is

1 Compare, e.g., my figures of the larvae of Pogonomyrmcx barbatits, loc. ri/.,
p. 20, Fig. 9.

No. 2.] HABITS OF PONERA AND STIC, MA I'OMMA. 63

about to emerge. In the latter case the opening, which is
produced by the huge mandibles, has the form of a transverse
slit, extending halfway round the cocoon. The small, sharp
mandibles of the male, however, gnaw a hole with irregular
edges and of much smaller size. The insect, after periods of
struggling, alternating with periods of rest, succeeds in getting
first one antenna, then the other, and then the fore legs through
the orifice, and finally, with considerable effort, creeps out.
After making this observation on isolated cocoons I had an

o

opportunity of making it in the artificial nests. In these the
hatching cocoons were often carried about and placed on or
under the stack of other cocoons, while the callows, struggling
to emerge, seemed to hold out their antennae and fore legs in
a supplicating attitude to the completely indifferent workers.
In a few instances the callows died while halfway out of the
cocoons and were carried to the refuse heap in this condition.
Occasionally, when the young callows had emerged with their
hind legs still enswathed and encumbered by the white pupal
skin, the workers would pull this away. They also occasion-
ally licked and fondled the newcomers, as if their bodies were
covered with some pleasant secretion, but beyond these acts
their helpfulness did not extend. These same workers, how-
ever, frequently opened cocoons and extracted dead immature
pupae, cut them up, and then placed them on the refuse heap.
This act shows that the statements concerning Leptogenys in
my former paper1 may require emendation.

The newly hatched Stigniatomma, as we should naturally
expect from the above observations, is not as feeble as the
callows of the more specialized ants. The males and females
issue with their wings fully expanded; the former have their
bodies completely pigmented and are able to run about briskly;
the latter, as well as the worker callows, although of a rich
yellowish-red, a color which they retain for several days, are
nevertheless soon able to run about and to join in the labors
of the colony. The queens show no tendency to leave the
nest and usually lose their wings (after copulation ?) while still
in the red callow condition.

1 Loc. cit., pp. 29, 30.

64 WHEELER. [VOL. II.

Careful search in and about the natural nests of Stigma-
tomma failed to reveal the presence of any myrmecophiles.
Wasmann, however, mentions 1 a Pselaphid beetle, Araniops
amblyoponica Brend., as occurring with these ants in Pennsyl-
vania and North Carolina.

GENERAL CONSIDERATIONS.

The foregoing observations on Ponera and Stigmatomma,
together with those contained in my former paper, suggest
some considerations of a more general nature.

First, the appearance of the larva of Stigmatomma, differing
so widely from the known larvae of other Ponerinae, is calcu-
lated to raise some doubts on the subject of taxonomy. Forel 2
has been of the opinion that the group of genera including
and allied to Amblyopone, viz., Stigmatomma, Mystrium, Priono-
pelta, and Myopopone, should be separated from the Ponerinae
as an independent subfamily, the Amblyoponinae. The main
characters used as an argument for this change are the presence
of two spurs on the hind tibia of these ants, and the very
broad and peculiar articulation of the node with the succeed-
ing abdominal segment. Emery, on the other hand,3 has pro-
tested against raising this group to the rank of a subfamily
and its separation from the Ponerinae.4 He argues that double
spurs occur also on the hind tibiae of nearly all Ponerinae
and of several Myrmicinae as well, that the articulation of the
node is highly variable among the Ponerinae, and that the
Cerapachyi (which Emery insists on placing with the Dory-
linae) present the same conditions of the node as in Amblyo-
pone. He also calls attention 5 to the important fact that the

1 I.oc. cit., p. 94.

Sur la classification de la famille des Formicides," Annales de la. Soc.

Tome xxxvii, pp. 161-165. 1893.

" Die Gattung Dorylus Fab. und die systematische Eintheilung der Formi-
Abth. f. System. Bd. viii, pp. 685-778, Taf. XIV-XVII,
41 text-figs. 1895.

" Die Abtheilung der Amblyoponinae darf nach meiner Ansicht nur als Tribus
in der Subfamilie der Ponerinen beibehalten werden."
5 Loc. at., p. 694, footnote.

No. 2.] HABITS OF PONERA AND STIGMA TOM M.I. 65

male genitalia of Stigmatomma sermtum Hald. are constructed
" soweit es ohne Zergliederung zu sehen ist, ganz nach dem
gewohnlichen Ponerinentypus."

It would seem, therefore, that there are no very cogent
reasons for adopting the subfamily Amblyoponinae, so far as
characters drawn from the adult structure are concerned. The
habits of Stigmatomma, as I have shown, are essentially the
same as those of the Ponerinae, so that there exist no eco-
logical grounds for accepting Forel's suggestion. The larva,
however, seems to me to show very clearly that there is a
greater gap between the Amblyoponii as a tribe of Ponerinae,
and the tribes Ponerii and Odontomachii, than between the
two last-mentioned groups. It must be remembered, however,
that the larvae of two tribes of Ponerinae, the Australian
Myrmecii and the cosmopolitan Ectatommii, have not been
described, and that when these are known the striking differ-
ences between the Amblyoponii and the Ponerii may be recon-
ciled. If, as Emery suggests,1 the Myrmicinae are descended
from the Ponerinae, it is obvious from a study of the larvae
that the former subfamily must have come from forms with
larvae like the Amblyoponii, a group which in other respects
also is generally regarded as very primitive.2 For the present
I can see no reasons for altering the excellent classification
outlined by Emery in his Dorylus paper.

There still exists an ecological (or biological, in the German
sense) connection between the Ponerinae and the Myrmicinae,
as I have lately ascertained. Since describing the peculiar
method employed by the Texan Ponerinae in feeding their
larvae, I have found that one of our New England Myrmicine
ants, Stenamma (Aphaenogaster) fuhnim Rog., subsp. aquia
Buck!., var. piccnm Emery --an ant very common under stones
and in rotten logs along the edges of woods --has essentially
the same method of feeding its young ! My attention was first

1 Loc. at., p. 773. " Dagegen liefern die Ponerinen offenbar die Wurzel, aus
welcher die iibrigen Unterfamilien der Ameisen (i.e., after excluding the very
primitive Dorylinae !) d. h. die Myrmecinen, Dolichoderinen und Camponotinen
entsprossen sind."

2 Emery, loc. cit., p. 774, suggests that the Myrmicinae may be related to the
Kctatommii, while the Dolichoderinae are more closely allied to the Ponerii.

66 WHEELER. [VOL. II.

called to this fact in an artificial nest belonging to Miss Adele
M. Fielde, at Woods Holl, Mass. One afternoon Miss Fielde
left a lot of queen pupae and larvae of Crematogaster lincolata
within reach of the Stcnainma colony. By the following morn-
ing the Stenammas had carried these into their nest, cut off
their heads and abdomens, and had distributed the pieces freely
among the larvae, which could be seen singly and in groups
of from two to five eagerly feeding on the juices in the same
manner as Ponerine larvae. Thinking that this might be a
very exceptional action, due to the confinement of the colony,
I opened numerous nests in the woods during the month of
August, while the, ants were rearing their second brood. In
nearly every one of these nests I found one or more larvae
feeding on substances left among them by the workers. In
one nest three larvae were feeding on a small Geometrid cater-
pillar ; in another several had their heads and necks inserted
into the thoraces of some small Carabid beetles that had been
decapitated by the ants ; in still another nest several larvae
were devouring the pulp of a blackberry, etc.1

Since making the above observation I find that Janet has
recorded some very similar facts.2 He saw several large larvae
of Lasius uii.vtns and L. flavits sucking the juices from the
cadavers of small larvae of their own species. Another more
detailed observation I quote in cxtcnso : " C'est clans un nid
artificiel de Tetramorium cacspituin que j'ai pu faire, a ce
sujet, 1'observation la plus precise. Au moment ou les Four-
mis recoltees venaient de terminer leur emmenagement, j'ai
vu, de la facjon la plus nette, une larve d'ouvriere sucer une
petite larve jaune de Coleoptere. La larve de Tetramorinm
n'etait pas tres eloignee d'avoir atteint sa taille definitive.
Elle etait suspendue par ses poils d'accrochage centre la paroi
du nid, immediatement sous le plafond en verre. Elle etait
place"e horizontalement, le dos en haut, mais un pen de cote.

1 Stenamnict fill-sum also brings into its chambers numerous seeds and the
corollas of small white flowers. I mention this fact because it shows that this
ant, which is normally carnivorous, nevertheless has proclivities that ally it by
instinct as well as by structure to the harvesting ants of the subgenus Messor.

2 l^e Lasius nii.\lits, etc., Inc. fit., pp. 10-12.

No. 2.] HABITS OF PONERA AND ST1GMATOMMA. 6j

Au-dessous d'elle, placee tete-beche, parallelement a son corps
et soutenue en partie par les poils d'accrochagc dc 1'abdomen
du Tetramorinm, se trouvait la petite larve jaune vermiforme,
ayant yz millimetre de diametre et 2 millimetres ^ de longeur.
Cette larve jaune avait certainement etc placee la par une
ouvriere, car pendant 1'emmenagement, j'en avals vu une qui
introduisait une larve semblable dans le nicl. La larve de
Tetramorinm avait sa tete inflechie et appliquee centre la larve
jaune. Elle laissait voir, tres nettement, sa bouche et ses
pieces buccales. Graces a ces circonstances exceptionnelle-
ment favorables j'ai pu examiner, avec une forte loupe, ce qui
s'est passe, et cela pendant plus d'un quart d'heure. J'ai
d'abord constate le mouvement incessant de la bouche et vu
nettement 1'absorption du liquide transparent qui sortait de la
plaie. Libre clans sa partie moyenne, la petite larve jaune
etait soutenue dans sa region cephalique par les poils d'accro-
chage de 1'abdomen du Tetramorinm. Ce dernier maintenait,
au moyen de ses mandibules crochues, 1'extremite anale de sa
proie, et cette extremite etait animee d'un mouvement rhythme
de balancement resultant des mouvements de succion. Pen-
dant ce repas, et sans que la larve du Tetramorinm parut en
etre derangee, une ouvriere est venue la lecher. Cette ouvriere
est allee, ensuite, degorger de la nourriture contre la bouche
d'une larve voisine. Au bout d'un quart d'heure j'ai du inter-
rompre 1'observation parce qu'une ouvriere est venue, malen-
contreusement, intercaler une nymphe entre la larve et le
verre. J'ai alors pris la larve avec un pinceau et une petite
cuiller et j'ai constate qu'elle avait ramene sa bouche contre
son corps, et que le repas etait interrompu. Quant a la petite
larve jaune dont j'avais vu le corps bien gonfle au commence-
ment de 1'observation, elle etait, maintenant, surtout dans la
region sucee, flasque et en partie viclee." Janet concludes his
observations on this subject with the words: "Cette observa-
tion, venant confirmer celles, un pen moins precises, que j'ai
faites chez d'autres especes, ne me laisse plus aucun doute sur
la faculte que les larves de Fourmis possedent de sucer directe-
ment les liquides contenus dans le corps de larves, probable-
ment blessees au prealable, qui les ouvrieres deposent aupres

68 WHEELER. [Vou II.

d'elles. Toutefois, cette maniere de prendre la nourriture doit
etre considered comme tout a fait exceptionelle."

The fact that this peculiar method of feeding the larvae is
the only method adopted by the Ponerinae, and, I believe, also
by the Myrmicine Stenamma fnlvum, and that it occurs as an
exception in such highly specialized ants as Lasius and Tctra-
moriiim, is of considerable interest from the standpoint of the
phylogeny of instincts. It not only spans a gap between the
generalized Ponerinae and the more specialized Formicinae, but
it would seem to indicate that the method of feeding the larvae
by regurgitation was grafted on to this original method in the
more recent ants, possibly in connection with their habit of feed-
ing one another by regurgitation in their adult conditions.1

In conclusion, the various peculiarities which indicate that
the Ponerinae are a very primitive and generalized subfamily
of ants may be enumerated 2 :

1. The colonies of the Ponerinae consist of a comparatively
small number of ants, like the incipient colonies of the Myrmi-
cinae, Dolichoderinae, and Formicinae.

2. These small colonies appear to be annual growths, formed
by swarming, as in the bees, and not by single fertilized female
ants unaccompanied by workers, as in the above-mentioned
subfamilies.

3. Two or more colonies of Ponerinae of the same species
can be fused to form a larger colony without much difficulty.
This is not so easily accomplished with many species of the
more specialized ants.

1 Since completing my manuscript I find that one of our Texan Myrmicine ants
(Pheidole sp. near P. fabricator Smith) resembles Stenamma fulvum in its manner
of feeding the larvae. September 27 I opened a small nest of the Pheidole near
Austin and found dozens of larvae feeding on fragments of different insects col-
lected and comminuted by the workers. This is of interest because the Pheidole,
like its congeners, is a harvesting ant, storing the large flat chambers of its nest
with many seeds.

2 While I have assumed in this and in my former paper that the Ponerinae
may be regarded as the group from which the Myrmicinae, Dolichoderinae, and
Formicinae (Camponotinae) have developed, I am quite of Emery's opinion that
the existing Ponerinae are not ancestral forms. Emery calls attention to the fact
that the palpi are aborted in all the tribes of Ponerinae except the Myrmecii,
whereas many of the genera of higher ants have retained the undiminished num-
ber of joints in these organs.

No. 2.] HABITS OF r OX ERA AND STIG MA TOM. MA. 69

4. The architecture of the Ponerinae is of a primitive char-
acter, consisting of a few irregular and unfinished galleries and
chambers. The latter are not even formed by all the species.

5. The queen and worker differ but little in size and
structure.

6. Ergatoid females, or forms intermediate between the
queens and workers, are of normal and comparatively frequent
occurrence in some species.

7. The habits of the queen and worker are very similar.
The female is not an individual on whom special attention is
bestowed by the workers.

8. The workers show no tendency to differentiate into major
and minor castes.

9. The Ponerinae are carnivorous and live by hunting (in
contrast with the various harvesting, fungus-growing, honey-
collecting, and aphidicolous members of the more specialized
subfamilies).

10. They do not feed one another by regurgitation.

11. The larvae are not fed by regurgitation, but are given
pieces of insects from which they suck the juices.

12. The cocoon is retained as a pupal envelope throughout
the group. (The Ecitonii among the Dorylinae have lost this
envelope, although the Dorylii still retain it ; it is lost in the
Myrmicinae, and is apparently in the process of disappearing
among the Formicinae.)

13. In at least one genus (Stigmatotnma) the callows are
able to escape from their cocoons without the assistance of the
workers.

14. The callows of the Ponerinae are more mature on
leaving the cocoon than the newly hatched Formicinae.

COLEBROOK, CONN., Sept. 5, 1900.

FUSION OF FILAMENTS IN THE
LAMELLIBRANCH GILL.

EDWARD L. RICE.

IT is far from the purpose of this paper to enter fully into the
discussion of the morphology of the lamellibranch gill ; but a
brief preliminary statement of a general nature may serve to
define the terminology employed, and to simplify the following
more detailed description of a single point in the gill structure.

When the animal is oriented in
the usual manner, with the hinge line
upward, the gills hang down on either
side, between the body and mantle,
as shown in Fig. i. Strictly con-
sidered, there is but one gill, or one
ctenidium, on each side of the body.
This ctenidium consists, fundamen-
tally, of a slight longitudinal ridge
along the side of the body, the gill
axis (Fig. i, a], and a double row of
ciliated filaments, which extend down-
ward into the mantle cavity of the
animal and are then reflected upward.
In common language each of these
rows of filaments, or the structure
arising therefrom, is termed a gill ;
and I shall retain this convenient,
though morphologically indefensible term, and designate that
half of the ctenidium next the body as the inner gill (Fig. i, b),
and that half next the mantle as the outer gill (Fig. i, c}.

In some few forms the adjacent gill filaments remain entirely
free from one another, e.g., Anomia ; in other forms, as Mytilus
and Pecten, there is a union by means of tufts of very long inter-
locking cilia — ciliated discs. In contrast to these two types,

71

FIG. i . — Diagrammatic frontal section
of lamellibranch. Shell omitted;
gills shaded, a, gill axis; b, inner
gill; c, outer gill ; d and d' , descend-
ing lamellae of inner and outer gills ;
e and e ', ascending lamellae of inner
and outer gills; f and _/", interla-
mellar connections of gill and dorsal
appendage ; g, dorsal appendage of
outer gill.

72 RICE. [VOL. II.

which may be grouped together as filamentous gills, we find in
the great majority of lamellibranchs a more or less strong
development of vascular connections between the filaments,
binding them together to form the lamellae from which the
class takes its name. Such gills may be termed lamellar.

As each filament consists of two limbs, a descending and an
ascending, so each gill is composed of two lamellae, correspond-
ingly designated as descending (Fig. i, ^and d') and ascending
(Fig. i, c and c'}. These lamellae are usually connected with
one another by more or less complicated interlamellar connec-
tions (Fig. i,/).

The ascending lamella of the inner gill may remain free, or
its upper margin may fuse with the body and, behind the body,
with the corresponding part of the gill of the other side. The
ascending lamella of the outer gill may lie free in the mantle
cavity or may be attached to the mantle on a level with the gill
axis ; or it may be continued dorsally above this line, forming
a dorsal appendage (Fig. i, g), which is finally attached to the
body, or rather to the fusion line of mantle and body. Thus
the dorsal appendage consists of the ascending lamella alone,
and may show a structure decidedly different from that of the
gill proper. Connections similar to the interlamellar connec-
tions of the gill attach the dorsal appendage to the body wall

(Fig. I,/)-

The lower free margin of some gills is smoothly rounded off ;

in other cases the border is deeply notched by a groove running
from end to end of the gill. This may be called the marginal
groove. In the diagram the groove is seen in cross-section in
the inner gill (Fig. i, //), while the outer gill presents a smooth
margin with no sign of a furrow. Exactly these conditions are
found in a large number of lamellibranchs, e.g., Astarte, Dreis-
sensia, Cardium, Psammobia. It may be suggested in passing
that the somewhat unintelligible distinction of "double gills"
and "single gills," in the classical paper by Williams,1 may
perhaps be based on the presence or absence of this marginal
groove.

1 Williams, T., "Respiratory Organs of Invertebrates," Annals and Magazine
of Natural History. Vol. xiv. 1854.

No. 2.] FILAMENTS IN THE LAMELLIBRAXCH GILL. 73

The surface of the gill lamellae may show either of two types.
It may be smooth, all the filaments lying in one plane ; or, with
the requirement of greater respiratory surface, the lamellae
may be folded parallel to the filaments. The two lamellae of
one gill are always folded symmetrically, so that the section
of the gill perpendicular to the filaments assumes an outline
reminding one, in its extreme form, of a string of wooden
button molds. Both filamentous and lamellar gills are subject
to this folding.

The filaments occupying the bottom of the reentrant folds,
the limiting filaments, are usually somewhat larger than the inter-
mediate filaments, and are often very much modified in form
as well as size. These filaments may be easily traced through
the whole height of the gill and afford a series of fixed points,
aiding materially in the study of the folded type of gill by means
of sections.

From the above resume it is evident that fusion or concres-
cence of parts has long been recognized in the lamellibranch
gill. Even in simple filamentous gills (Mytilus, etc.) the tips of
the filaments fuse to form a continuous band along the margin

o o

of the ascending lamella. In more complex forms it is a process
of fusion which transforms the rows of originally distinct fila-
ments into the characteristic lamellae. The fusion of the ascend-
ing lamellae with the mantle and body is also perfectly familiar.
But the particular type of fusion described below appears to have
escaped mention in the somewhat extensive literature upon the
lamellibranch gill.

Even a hasty study of serial sections of the strongly folded
gill of Cardium cditlc or Batissa tenebrosa shows the somewhat
surprising fact that each fold contains a far larger number of
filaments in the upper portion of the gill than in the neighbor-
hood of the free margin, this being equally true of ascending and
descending lamellae. (The extreme case was noted in Cardium,
where the number of filaments in one fold increased from eight
to thirty.) Yet the limiting filaments are continuous through-
out, showing that there is no reduction in the number of folds
correlated with the increase in the number of filaments in each

t

fold. These observations appear at first sight irreconcilable with

74

RICE.

[VOL. II.

the theory of gill development advanced by Lacaze-Duthiers1
and universally accepted, according to which the ascending limbs
of the filaments are essentially the reflected tips of the origi-
nally simple straight filaments. The larger number of filaments

in the upper parts of the descending
lamellae could be explained by the
assumption of the presence of a large
number of abortive filaments. But this
explanation, unsatisfactory even for the
descending lamellae, cannot apply to
the ascending lamellae.

A more detailed study of serial sec-
tions, and, better, of microscopic dis-
sections in which the lamellae are
separated and the folds spread out as
smooth as possible, shows clearly that
the filaments of the upper portion
gradually meet and fuse as the free
margin of the gill is approached. This
phenomenon is illustrated by Figs. 2
and 3. In the former we have a surface
view of a small portion of the inner gill
of Batissa tcnebrosa, extending upward
from the free edge. While the two
filaments at the left (Fig. 2, a and b]
are simple throughout, the next filament
(Fig. 2, c), indistinguishable from the
others at its lower end, is really formed
by the fusion of nine filaments.

In this species such groups of fusing
filaments are found principally in the
projecting folds of the gill, very seldom
in the reentrant folds. In Cardium
ednlc, on the other hand, the fusion
is more marked in the reentrant than in the projecting folds.

a.

FIG. 2. — Surface view of group of
filaments of inner gill of Batissa
tenebrosa. x 60. Lower edge
of figure represents the free mar-
gin of gill. Cilia and complex
interfilamentary c onnections
omiUr.l for ill,- sake of clearness.
n and /', simple filaments; c,
compound filament, formed by
I ..... n of nine simple filaments.

1 I .;n a/u-Duthiers, II.de, " Memoir sur le developpement des branchies des Mol-
lusques Acephales Lamellibranches," Annales des Sciences Naturelles. Zoologie
Scr. iv, Tome v. 1856.

No. 2.] FILAMENTS IN THE LAMELLIBRANCH GILL. 75

Fig. 3 represents diagrammatically the surface of a single fold
of this species, each filament being represented by a simple line.
At the gill margin (lower side of diagram) there are eight fila-
ments, including the limiting filaments (Fig. 3, a and //) ; in the
upper part of the gill (and diagram) there are thirty. Of these
thirty filaments, twenty-one unite to form the two limiting fila-
ments (Fig. 3, a
and Ji) and the
two adjacent in-
termediate fila-
ments (Fig. 3, b
and g), which

occupy the re- \/ V

entrant folds.

The remaining
four intermedi-
ate filaments
(Fig. 3, c, d, e,
and/"), those of
the projecting
fold, are formed
by the union of
only nine fila-
ments. The ir-
regularity of the
fusion is also
well shown by a.

V

V

\l

\

\

\

<*

f g-

k

tlllS dia°~ram. FIG. 3. — Diagrammatic view of surface of fold of inner gill of Cardinal

. editle, illustrating fusion of filaments. Filaments represented by

iNOte especially lines. Lower edge of diagram represents free margin of gill, a and

Hilt" ODP limit A, limiting filaments ; b-g, intermediate filaments.

ing filament is entirely simple, while the other is the resultant
of the fusion of seven simple filaments.

The fusion is usually almost exclusively limited to a some-
what narrow zone in the near vicinity of the free margin of the
gill. In this zone may also be noted a gradual reduction in the
folding of the gill as the margin is neared. In those lamelli-
branchs whose outer gill is provided with a dorsal appendage,
there is a second zone of fusion along the transition line from

76 RICE. [VOL. II.

the gill proper to the appendage. This line again marks a partial
or total obliteration of the folding of the gill - - a point of impor-
tance in the consideration of the meaning and cause of the
fusion of the filaments.

The term "fusion of filaments" has been employed repeat-
edly ; but the question arises whether the phenomenon under
discussion is really a fusion of a large number of once distinct
filaments or a branching of a relatively small number of original
filaments ; whether, in other words, the primary filaments are
represented by the maximum or minimum number. The answer
is not far to seek. At the free margin of the gill the continuity
of the filaments may be readily traced from one lamella to the
other, and the number of filaments in the two lamellae within
any given fold must necessarily be equal. A little higher, in the
zone of fusion, these numbers may become very unequal. For
example, in a section of the gill of Batissa seven filaments of
one lamella were observed to correspond to twelve in the other.
Neither of these facts, however, offers conclusive evidence ; both
may be explained as well on the supposition of branching as on
that of fusion. But as the serial sections are followed a little
higher in the gill, the inequality in the number of filaments, which
has increased from zero at the margin to a maximum in the zone
of fusion, begins to fall off again more or less rapidly, and finally
reduces itself once more to zero. This equality in the number
of filaments in the upper parts of the gill can be explained only
on the supposition that the phenomenon before us is a fusion,
not a branching, of the gill filaments. It is altogether too im-
probable that the independent branching of the filaments in the
two lamellae would lead to the same number in the two cases.

Granted that the phenomenon is a fusion, what is its mean-
ing ? Is it of systematic importance ? My observations are
not complete enough to permit a categoric answer ; but I am
strongly inclined to the belief that the fusion carries no more
of weight from the systematic standpoint than does the folding
of the gill.

Among smooth gills no fusion was observed, although a con-
siderable number of forms were studied. Especial attention
was devoted to Tellina and Scrobicularia, in which, for reasons

No. 2.] FILAMENTS IN THE LAM ELLIBRANL 11 ('.ILL. 77

detailed in an earlier paper,1 1 consider the simplicity and smooth
surface of the gill to be secondary characters --a retrograde
development from the folded gill type of the Veneridae. Fusion
was also never 6bserved in the filamentous type of gill.

The results of my own observations upon the folded lamellar
gills may be tabulated as follows :

1. Fusion strongly developed: Cardium edulc Lin., Chania
pcllncida Brod., Batissa tenebrosa Hinds, Psammobia vespertina
Lin., Donax serra Chemn.

2. Fusion moderately developed : Venus verrucosa Lin.,
Cyprina islandica Lin., in latter strongly developed on transi-
tion line from outer gill to appendage.

3. Fusion very slightly developed : E?isatc!laamcricanaVQrr\\\,
Mya arcnaria Lin. , Donax politus Poli, in latter perhaps accidental.

4. No fusion observed : Cythcrea cJiionc Lin., Donax trun-
culus Lin., Ostrca virginiana Lister, TJiracia papyracea Poli.

Lima and Ostrea should probably be added to the list of forms
in which fusion occurs, though on somewhat doubtful evidence,
which will be mentioned later.

Thus we find this fusion of filaments similarly developed in
widely separated forms, belonging to very diverse groups ; on
the other hand, within the single genus Donax we find all
grades of fusion, as we also find all grades of folding.

This parallelism of folding and fusion in the genus Donax
appears to me to furnish, in a certain sense, an epitome of the
whole matter, for I consider the fusion of the filaments to be
a mechanical correlative of the folding of the lamellae. The
process may be pictured in something this way. The folding of
the lamellae is gradually developed in the young lamellibranch
as the increasing number of filaments becomes too great to lie
in one plane. But at the free margin the filaments are bound
somewhat firmly together and the folding is somewhat reduced.
This leads to a crowding together of the filaments at this point
and eventually to an organic fusion of the same. How great
the crowding must really be in the strongly folded forms may
be inferred from the fact that the upper part of the inner gill of

1(1 Die systematische Verwertbarkeit der Kiemen bei den Lamellibranchiaten,"
Jenaische Zeitschrift fiir Naturivissenschaft. Bd. xxxi. 1897.

78 RICE. [VOL. II.

Cardium cdnle would measure, if flattened out, fully seven times
the length of the free margin of the same gill. The absence of
fusion in filamentous gills, even where the folding is extreme, as
in Pecten, may be easily explained on the ground of the looser
structure of the gill and the possibility of a displacement of the
filaments, with consequent relief of pressure.

It is an interesting point in this connection that no fusion
was observed in the outer gill of either Psammobia or Cardium,
although it is conspicuous in the inner gill. In both cases the
inner gill is provided with a deep marginal groove, while the
outer gill shows no sign of this structure. The presence of this
marginal groove is the mechanical equivalent of a shortening
(slight to be sure) of the margin of the gill, the bottom of this
groove being almost perfectly straight, and therefore slightly
shorter than the somewhat scalloped margin of the folded gills
in which no groove is present. Hence we find in this charac-
teristic a cause of increased crowding and increased fusion.

For a very short distance above each point of fusion one
observes a slight modification of the epithelium of those sides
of the fusing filaments which are turned toward each other.
The nuclei are somewhat larger and more closely crowded, as
shown diagrammatically in Fig. 5. Aside from this, no histo-
logical distinction can be drawn between the simple original
filaments of the upper portion of the gill and the compound
filaments of the lower margin. Even in the matter of size there
is no noticeable difference except in the immediate vicinity
of the fusion, where the compound filament is considerably
enlarged. This statement is made after comparison of a large
number of preparations, and seems to be the rule, although
there are considerable individual variations in different speci-
mens. The apparent larger size of the compound filaments in
Fig. 2 is to be explained in part on the ground of such indi-
vidual variations. It is due in larger degree, however, to slight
differences in the position of the filaments, which are elliptical
in cross-section, and appear of different size according as one
or another side is presented to view. It should also be noted that
the filaments are somewhat enlarged at the extreme margin of
the gill, and that these enlarged tips are shown in the figure.

No. 2.J FILAMENTS IN THE LAM ELLIBRAXCH GILL. 79

As regards the size and finer structure of the filaments, the
evidence of sections is more reliable than that of surface prep-
arations ; and in Figs. 4-8 are represented a series of sections
through a group of filaments of the inner gill of Cardinm ednlc.
The same filaments are shown in all the figures, although the
very complex interfilamentary and interlamellar connections,
differently cut in the different sections, cause a considerable
variety of aspect. These connections may be disregarded in
the present discussion. In Fig. 4, which represents the upper-
most section, six filaments are shown, all practically alike. In

FIG. 7. FIG. 8.

FIGS. 4-8. — Series of sections through group of filaments of inner gill of Cardium edule. x 95.
Cilia and nuclei somewhat diagrammatic. Fig. 4 is uppermost section. Asterisk marks
two simple filaments fusing to one compound.

Fig. 5 two of these, marked with an asterisk in all the figures,
show a slight modification of the epithelium, as described above.
This section is immediately above the point of fusion of the
two filaments. Fig. 6 shows these filaments so closely approxi-
mated that they may almost be described as forming a single
deeply grooved filament, a condition better shown by other
preparations. In Fig. 7 the fusion is complete, but the com-
pound filament is still considerably enlarged; while in Fig. 8
the compound filament has regained its normal size and form,
and the transition is complete.

Strangely, this very conspicuous fusion of the filaments
appears to have received little or no notice. I have been able
to find no reference whatever in the text of any articles at my

80 RICE.

disposal. Only in the figures is a possible hint of its observation.
Thus in the exquisite work by Deshayes 1 on the Mollusca of
Algiers there is a small figure of the gill of Pholas, in which two
incomplete filaments are interpolated among those which extend
through the whole height of the gill. In the one the free end
is turned upward ; in the other, downward. Do these represent
two filaments fusing with their neighbors ? If so, the difference
in direction in the two cases clearly distinguishes the phe-
nomenon from that described above. Moreover, the author's
description of the gill as entirely smooth makes a fusion
a priori improbable. An inaccuracy in the drawing, which is
very small, is the simpler explanation.

In a considerable number of more recent figures a single fold
of the gill is represented in section as containing an unequal
number of filaments in the two lamellae, thus in Cardium
(van Haren),2 Lima (Pelseneer),3 Dona.v trnncnhis (Sluiter),4 and
Ostrea (Kellogg).5 It will be noted that in the last two cases
the figures cited do not accord with my own observations. But
my purely negative verdict of " not observed " contains no direct
contradiction of the affirmative statements of these authors.
While I am convinced that my own preparations show no evi-
dence of a fusion of filaments in Ostrea and Donax trunculus, I
consider it probable that the figures of Kellogg and Sluiter, as
well as those of van Haren and Pelseneer, point to a greater or
less development of the phenomenon described. Unfortunately
no certain conclusions can be reached in the absence of corrob-
oration in the accompanying text.

1 Deshayes, G. P., " Exploration scientifique cle 1'Algerie," Zoohgie. Tome i,
No. 4. Paris, 1849.

- Haren-Noman, D. van, "Die Lamellibranchiaten, gesammelt wahrend der
Fahrten des Willem Barents, 1878 and 1879," ATiederlandisches Archiv fur Zoologie.
Supplementband I, 1881-82.

3 Pelseneer, P., "Contribution a 1' etude des Lamellibranches," Archives de
Biologic. Tome xi. 1891.

4 Sluiter, C. P., " Beitrage zur Kenntniss des Baues der Keimen bei den Lamelli-
branchiaten," Niederlandisches Archiv filr Zoologie. Bd. iv. 1878.

5 Kellogg, J. L., " A Contribution to our Knowledge of the Morphology of the
Lamellibranchiate Mollusks," Bulletin of the U. S. Fish Commission. Vol. x. 1890.

OHIO WF.SLEYAN UNIVERSITY, DELAWARE, OHIO.

PORTABLE ANT NESTS.

ADELE M. FIELDE.

IN order to keep ants under continued observation, and at
the same time to change occasionally the domicile of the
observer, it is necessary to have portable nests. The trans-
portation of either the Lubbock or the Janet nest is made
inconvenient by the water used for the isolation of the ants in
the former, and by the considerable weight of the latter.

Six colonies can be successfully carried on journeys of
several hours or days by the use of a case made of half-inch
pine boards and dovetailed at the corners, with a door hinged
upon its lower side and held shut by two buttons at its upper
edge. The case measures on the inside 1624 inches in length,
63^ inches in horizontal depth, and 4% inches in height. Three
shelves, each one-fourth of an inch thick, are mortised into the
ends of the case, making four compartments, each one inch
high. Several holes are bored in the door of the case and in
the side opposite the door, to admit fresh air. The case is car-
ried by a leather handle fastened lengthwise to its top, and it is
just filled by the six nests hereinafter described.

Of the six nests, there are two of each pattern, A, B,
and C. An A and a B nest together fill a shelf, while a C
nest fills a shelf alone.

The nests are all built of clear glass, and their parts are
readily cut by any glazier. The joinings are made with
Le Page's liquid glue, which must thoroughly dry before the
nests are used. The cost of material and the amount of labor
required in making these nests are much less than for either
the Lubbock or the Janet pattern.

The floors are panes of "double thick" glass, under which
is laid a thick sheet of white blotting paper of the same size as
the glass, to give opacity, elasticity, and a background against
which the ants are easily seen. The blotting paper should not

Si

S2 FIELDE. [VOL. II.

be fastened to the glass, as the latter must sometimes be raised
in order to get a view of the ants from below.

o

The walls stand half an inch from the edges of the base-pane,
for the greater security of the superstructure when the nest is
being lifted. The walls are built of strips of "double thick"
glass, half an inch wide, two horizontal layers in every wall
giving a perfectly level top. Apertures three-eighths of an
inch wide are left in the walls at points marked a on the plans.
These apertures admit the end of a three-eighths-inch glass
tube, which may be used as a safe bridge between nests when
ants are to be made to pass from one nest into another. When
not in use, the apertures are closed with plugs of cotton. The
outside of the walls, after thorough drying, are painted black,
to secure within the nest the darkness in which ants like to
keep their young.

The main partitions in the ant-dwelling take part with the
walls in the support of separate roof-panes, and they are there-
fore twice as wide as are the walls. Passageways (marked m
on the plans) permit the ants to go from one room to another,
as is necessary when a room is to be cleaned. The admission
of strong light will insure the removal of the ant family to the
darkened chamber adjoining. The passageways may be from
one-fourth to one-half inch in width, and they should be cov-
ered with very thin glass, well glued on.

The tops of the walls and main partitions are exactly cov-
ered with coarse Turkish toweling, cut in strips twice as wide
as the base on which it lies, and doubled so that its raw edges
meet in the middle of the glass it doubly covers. The under
layer of the toweling is glued to the glass, and it performs the
double office of keeping the ants within the nest and of admit-
ting sufficient fresh air for their breathing. Just over the pas-
sageways the toweling is left free, so that it may there be lifted
for observation of what is within. After the toweling is laid
on, the rooms have a uniform depth of less than half an inch,
and a hand lens can be focused upon any of their inmates.
The toweling being elastic and level, the roof-panes, of thin,
clear glass, lie closely upon it. They are not fastened down.

The roof-panes reach the center of the main partitions and

No. 2.]

PORTABLE ANT NESTS.

a

3
/>

A. 6% X 6 in. (i6X x 15^ ^-)

3

,5

m.

B. 10 x 6 in. (2^/2 X 15^ C.)

a

/

' 1

m

S?i

[

;~-i

'.

6 3 6

•

6 2
6

1

: ~,,:rr;rr"i

, ,

I ': " .,

a.

?n

?tl

•

,

I ''

i. Food-room.
a. Entrance,

C. i

6,

fc

X 6 in. (42 x 15^

2. Nursery.
b. Screen.

C.}

3. Sponge-room.
»t . Passage.

84 FIELDE. [VOL. II.

project one-fourth of an inch beyond the walls. All the rooms
except the food-room have an outer roofing of thick dark
blotting paper, which should be lifted only when actual study
of the ants is proceeding.

In the rooms numbered i the ants have, as in the Janet
nests, a chance to range in the light and to seek food; of which
it is well to put in the smallest sufficient quantity and several
kinds. If the ants are made to move into the darkened food-
room, leaving the other rooms free for cleaning, the passage-
ways (m) may, during the cleaning, be plugged with cotton.

In room 3 a soft, fine sponge, clean and wet, and less than one-
fourth of an inch thick, should nearly cover the floor, leaving a
passage all around it next the walls. This furnishes drink to
the ants and moisture to the air of their dwelling. If the ant
young are in the egg or the larval stage, or if the temperature
is high, the floor of room 2 should likewise be covered with
wet sponge ; but if the young are in cocoons, or if the temper-
ature is very low, then room 2 should have a layer of wadding
instead of sponge. The ants generally choose damp places for
the eggs and larvae, and dry places for the cocoons or pupae.

The screens (marked />) are substitutes for the ant-runs used
in the ground, and they gratify the disposition of the ant to
keep close to cover in going about in the nest. They are
made in the same way as are the walls, but are only one-fourth
of an inch thick, and are not topped with toweling.

The A nest, with base 6*/> X 6 inches, is designed for a
colony of very small ants, or for a few large ants. The B
nest, with base 10 X 6 inches, affords a home for a somewhat
larger family. The C nest, 16^ X 6 inches, can be used for
a multiplying and dividing colony, or for observing the activi-
ties of restless species. The ants should never be greatly
crowded in their habitation.

The ants in my nests appear sleek and healthy. I have
found these nests easier than others to keep free from the
molds that grow from particles of food that the ants convey
from the food-room to every other part of their nest. These
nests also lend themselves readily to experimental uses in
studying the instincts of their occupants.

No. 2.] PORTABLE ANT NESTS. 85

When the nests are to be carried on a journey the roofs
are securely fastened down by sewing narrow strips of cheese
cloth around the nest in such a way as to prevent the slipping
of the roof-pane. The fastenings must not exclude fresh air.
Having fastened the roof-panes each in place, the nests are
put upon the appropriate shelves in the case, where they may
be further secured by bits of wadding above the roof-panes and
at the ends of the shelves.

The weight of my case, with its six enclosed nests packed
for travel, is less than fifteen pounds. The strong local attach-
ments of the ants are undisturbed by their so journeying, and
at the end of the journey the study of their life processes may
be speedily resumed.

0

MARINE BIOLOGICAL LABORATORY,

WOODS HOLL, September, 1900.

Volume //.] December, lyoo. \_No. J.

BIOLOGICAL BULLETIN.

THE OESOPHAGEAL GLANDS OF URODELA.

R. R. 1-iKNSLEY.

FOR a long time the only known instance of glands occurring
in the oesophagus of an Amphibian was the familiar pepsin-
producing glands of the frog's oesophagus, discovered as early
as 1838 by Bischoff (i). In 1853 Leydig (8) described the
occurrence of saccular glands in the oesophagus of Proteus
anguineus, and, more recently, similar glands have been dis-
covered by Kingsbury (5) in the oesophagus of Necturus
maculatus.

In no other Batrachian has investigation revealed the exist-
ence of glands in the oesophagus, unless,, indeed, as Klein (6)
suggests, the highly branched glands found at the junction of
oesophagus and stomach in Triton, and termed by Langley (7)
the anterior oxyntic glands, are such.

Naturally, considerable interest has been evinced in the
question of the homology of these glands one with another,
and with those of the higher vertebrate classes.

In order that the problem to be solved may be clearly under-
stood, it may be as well to recapitulate briefly the facts as they
appear in the forms so far investigated.

In the frog, leaving out of consideration the pyloric glands,
there are two kinds of glands occurring in the foregut. The
oesophageal glands occur under a ciliated epithelium, and are
large compound glands, consisting each of a number of short
tubular acimi lined by pepsin-secreting cells, opening into a
common duct lined by transparent mucous cells. As we pass

87

88 BENSLEY. [VOL. II.

down the oesophagus we find that, at the point where the cili-
ated epithelium is succeeded by the ordinary mucigenous epi-
thelium of the stomach, there is a gradual transition from the
compound oesophageal glands to the more simple tubular
glands of the stomach, the second type. The secreting cells
of the two kinds of gland differ markedly from one another.
Those of the gastric glands contain few zymogen granules of
small size, while those of the oesophageal glands are more or
less filled with very large granules, and the cells themselves
are larger. Further, the oesophageal glands yield an alkaline
secretion, the gastric glands an acid secretion.

In Triton, again, there are two types of gland. At the
junction of oesophagus and stomach occur the anterior oxyntic
glands of Langley. The difference between these glands and
the other gastric glands (posterior oxyntic glands of Langley)
is not so marked. The former are more highly branched and
are separated from one another by a larger amount of connec-
tive tissue, but the differences in the size of the granules and
in the nature of the secretion, so conspicuous in the case of
the frog, are absent.

In Proteus a new structure makes its appearance in the
shape of isolated sac-like glands occurring in the oesophagus.
These have been fully investigated by Oppel (u), who de-
scribes them as follows : " Die Driisen des Oesophagus haben
eine rundliche Form. Sie bestehen aus einem grossen Acinus.
Die Drusen sind zusammengesetzt aus einem Ausfuhrungs-
gang und dem secernierenden Theil. Ich spreche von einem
Ausfiihrungsgang, da sich die Zellen desselben von denen der
Schleimhautoberflache unterscheiden. Der Ausfiihrungsgang
besteht aus Zellen von annahernd cylindrischer Form, und zwar
ist die Grenze zwischen conischem und cylindrischem Epithel
stets eine scharfe. Eine besondere Eigenthiimlichkeit liegt in
der Uebergangsstelle von cliesen cylindrischen Zellen des Aus-
fiihrungsgangs zu den secernierenden Zellen. Dieselbe liegt
namlich nicht an der Stelle, an welcher die Erweiterung des
engen Ganges zum Acinus stattfindet, sondern die Cylinder-
zellen gehen noch ein Stuck weit in den Acinus hinein, um
clann rasch zu den niedrigeren secernierenden Zellen abzufallen.

No. 3.] OESOPHAGEAL GLANDS OF URODELA. 89

Diese Zellen zeigen in ihrem 1'rotoplasma cinen kornigen Ban,
Korner, welchc sich mit verschiedenen Farben, z. B. Eosin,
S.-Fuchsin tingieren, mit Osmiumsaure braunen . . . ." No
glands resembling the anterior oxyntic glands of Triton are
present in the adult, but he found in the young animal, at the
junction of oesophagus and stomach, glands which are inter-
mediate in nature between the oesophageal and gastric glands.

In Necturus there are, according to Kingsbury, three kinds
of glands present. In the oesophagus are large saccular glands
in most respects like those of Proteus, except that Kingsbury
was unable, even after repeated trials, to demonstrate the pres-
ence of any granules capable of reducing osmic acid. At the
junction of oesophagus and stomach are richly branched glands
like the anterior oxyntic glands of Triton, and finally there are
the ordinary gastric glands.

There are thus three types of gland occurring in the oesopha-
gus of Batrachia, the relationship of which to one another, to
the gastric glands, and to the oesophageal glands of higher
vertebrates, is obscure. These are the compound pepsin-
forming glands of the frog's oesophagus, the saccular glands
of the oesophagus of Proteus and Necturus, and the anterior
oxyntic glands of Triton and Necturus. It might be claimed
for a priori reasons that no possible relationship could exist
between the oesophageal glands and the gastric glands, but that
position would necessitate a critical examination of the data
on which this anatomical division of the foregut in the forms
mentioned has been decided.

The writer found that Amblystoma combined, in a sense,
the conditions found in Proteus and Triton, inasmuch as the
glands in the larva resemble those of Proteus, the glands of the
adult those of Triton. The present memoir is a brief account
of the histogenesis of the glands in question.

Before passing on to a consideration of the histogenetic
phenomena it is necessary to describe briefly the structure of
the mucous membrane of the foregut in the adult animal.

The oesophagus is non-glandular, and is lined throughout
by a ciliated epithelium, in which many goblet cells may be
recognized. The ciliated epithelium is succeeded by the

9o

BENSLEY.

[VOL. II.

Duct{

Neck

ordinary cylindrical cells of the stomach at the point where
the first glands appear, and the oesophagus expands suddenly
into the stomach. The ciliated epithelium does not extend into
the stomach.

The gastric zymogenic glands are of two kinds. The ante-
rior oxyntic glands occupy the proximal portion of the mucous
membrane and form a zone about 2 mm. in width around the

oesophageal orifice of the stom-
ach. They are much shorter
than the other gastric glands
and, like the corresponding
glands in Triton, are more
highly branched, and contain
more mucous neck cells. The
rest of the stomach, with the
exception of the posterior third
in which the pyloric glands are
found, is occupied by the usual
tubular glands, consisting of
a body composed of granular
pepsin-forming cells, a neck
composed of transparent mu-
cous cells, and a duct composed
of cells resembling the surface
epithelium. These glands cor-
respond in all respects to the
excellent description given of
the corresponding structures
in Triton by Carlier (2). To
make clear the terminology
employed, a bi-tubular gland
is represented in Fig. i .
It was found necessary to resort to a new method of staining
the zymogen granules, as the conventional method, by the
employment of osmic acid, was not satisfactory when the cells
contained brown pigment, or a great deal of prozymogen, which,
as Langley (7) and Griitzner (3) point out, also reduces the
osmic acid, obscuring the granules if they be few in number or

Body I

FIG. i. — Gastric gland of Amblystoma Jeffer-
sonianum. Zeiss apoch. 2 mm., ocular 2.

No. 3-J OESOPHAGEAL GLANDS OF URODELA. 91

of small size. For this purpose the writer employed Reinke's
neutral gentian as follows : To a saturated solution of gen-
tian violet in water a solution of orange G is added in excess.
A brownish precipitate is formed which is very slightly soluble
in water. This may be collected on a filter and washed until
the wash water is only slightly tinged. The precipitate is
then dissolved in alcohol. For use a sufficient quantity is
added to twenty per cent alcohol to make a fluid of about the
same color as a good solution of haemalum. Sections fastened
to the slide are stained in this for twenty-four hours, all adher-
ent stain is then removed by pressing clown upon the sections
several folds of filter paper, absolute alcohol added and quickly
removed with the blotter, and finally oil of cloves added in
which the differentiation of the stain takes place. As soon
as the protoplasm of the epithelial cells appears orange, the
extraction of the stain may be checked by washing in ben-
zole, and the sections may then be mounted in the usual way
in balsam. The zymogen granules are stained of an intense
blue color, the nuclei blue, other portions of the cells orange.
The stain is most successful after fixation in aqueous
sublimate.

The earliest stages in the formation of the gastric glands
are difficult to discern, owing to the great number of yolk spher-
ules present which obscure the outlines of the cells. In a
larva 1 1 mm. long the glands are already visible as tubular
down-growths of the endoderm of the foregut. In this early
larva two kinds of glands are already to be recognized, those
occupying the anterior end just behind the tracheal groove,
and those at the posterior end, where the -stomach is' as yet
not clearly marked off from the general endoderm. The ante-
rior glands are of a flask-like shape, and have a distinct lumen
surrounded by a single layer of cells. Zymogen granules are
not yet to be recognized, and the yolk spherules are so abun-
dant that the outlines of the cells are not visible. In the lumen
there may often be seen one or two cells, which have been, so
to speak, squeezed out of the row of endoderm cells forming
the gland. These cells do not take any part in the formation
of the permanent histological elements, but may often be

BEA'SLEY.

[VOL. II.

recognized, even to a late stage of development, as disinte-
grating remains in these and the other gastric glands.

The posterior glands are simple tubes composed of a single
layer of large yolk-filled cells surrounding a cleft-like lumen.
Indeed, often it appears as if the endoderm of theforegut were
in several layers without any differentiation into glands 'and
epithelium. On careful inspection, however, it may be seen
that the nuclei are arranged in an orderly fashion, as if sur-
rounding the lumina of glands. Such an appearance is illus-
trated in Fig. 2. In this gland, in addition to the nuclei which
are clearly arranged in a row around the lumen, two others

may be seen which are nearer the center
of the gland ; these are the nuclei of cells
which will later be found as disintegrat-
ing remains in the lumen.

The flask-shaped glands do not, as one
proceeds caudad, abruptly give place to
the simple tubular glands, but there is a
gradual transition.

In a larva 12 mm. in length, although
the caudal portion of the stomach is still
undifferentiated and the cells crowded
with yolk, the yolk has sufficiently dis-
appeared from the anterior portion to
enable the shape of the glands and the cells composing
them to be clearly determined. The anterior glands arc
now distinctly saccular, with a large lumen surrounded by a
single layer of cells. The yolk spherules disappear from the
cells of the glands somewhat more rapidly than from the sur-
face epithelium, which as yet contains a considerable number.
Notwithstanding the presence of the yolk, one can clearly dis-
tinguish, at this early stage, several kinds of cells, which can
be readily referred to their analogues in the glands of the
adult. The flask-shaped body of the gland (Fig. 3, a) is formed
of a single layer of small cells, which vary from cubical to
fusiform in shape and are usually convex towards the lumen.
The protoplasm of these cells is granular and contains one or
more yolk spherules. The nucleus is round or oval and rich

gastric
gland, zeissapoch.

2 mm., comp. ocular 2.

No. 3.] OESOPHAGEAL GLANDS OF URODELA. 93

in chromatin. As the gland narrows into the duct (Fig. 3, b),
these are replaced by two or three slightly larger cells, in each
of which two zones may be recognized, an outer, wider, deeply
staining zone containing the oval nucleus, and an inner one
which stains but feebly. This inner zone exhibits a reticular
structure due to the presence of a secreted substance, probably
mucigen. In the more posterior tubular glands, likewise, two
kinds of cells may be found, similar in all respects to those of
the flask-shaped glands, granular cells occupying the body of
the gland, mucigenous cells the neck. In sections stained in
Reinke's neutral gentian it is found that already numerous
zymogen granules are present in the deeply
staining cells forming the body of the gland.
In the pancreas, also, zymogen granules may
be recognized long before the yolk granules
have entirely disappeared from the cells.
The epithelium of the foregut in the region

occupied by the flask-shaped glands is com- ^ ^ a

posed of two kinds of cells (Fig. 3, c), ciliated

cells and cells the outer ends of which stain

diffusely and intensely. These obviously FlG. 3.J^blystoma

represent the two characteristic elements of larvai 2mm- in length;

, c ..... .... oesophageal gland.

the future oesophageal epithelium, the cm- Apoch. 2 mm.,comP.
ated and the goblet cells. Over the tubular ocular2-
glands farther back there is only one kind of cell in the
epithelium, and this is without cilia.

Mitoses may be observed with equal frequency in all the
various kinds of cell composing the epithelium and glands, and
all are apparently equally capable of reproduction.

The important points to be learned from this stage are
that the characteristic elements of the glands are differen-
tiated very early, that no special groups of cells have, as yet,
assumed the mitotic function to the exclusion of the others,
and. that a portion of the glandular foregut bears a ciliated
epithelium.

In a larva 14 mm. in length the foregut has advanced to a
considerable degree beyond the stage last described. It is
now shaped like a letter U, with a long proximal and short

94 BENSLEY. [VOL. II.

distal limb, the latter curving cephalad above the ventral pos-
terior margin of the liver, before passing into the midgut.

Four regions may now be distinguished ; a very short ante-
rior region without glands, provided with a ciliated epithelium,
a region with flask-shaped glands and ciliated epithelium,
a third region with tubular or saccular glands and a mucige-
nous epithelium, and finally, at the posterior end, a region in
which no glands at all are to be discerned. The second and
third regions gradually merge into one another, but the pos-
terior non-glandular portion is sharply marked off and forms,
in part at least, the future pyloric gland region.

At this stage the two pulmonary diverticula open into a
capacious pouch lying below the foregut, into the floor of which
it opens. In longitudinal sections the first gland appears
immediately behind this sac. Farther back more glands make
their appearance, and at the point where the foregut begins to
enlarge into the stomach, it is completely encircled by six or
eight of these large flask-shaped glands. Farther back again
the glands become less and less flask-shaped and take on a
tubular or saccular character.

One of these anterior glands is represented in Fig. 4 as
seen after staining in haemalum, followed by neutral gentian.

The shape of the cells in the body of
the gland varies with the degree of dis-
tention. There seems to be in these
glands an accumulation of the secretion
in the lumen distending it, for it is only
by the application of a distending force
from within that the extreme stretching
of the cells, which may be commonly
observed, could be produced. In many
glands where this distention is great

FIG. -c-Ambiystoma larva ,4 the cells are quite flattened and spread

mm . in length ; oesophageal

gland. Apoch. 2 mm., comp. out over a great surface, reminding one

strongly of the appearance in the mam-

malian blastodermic vesicle at the time of its rapid expansion.
The explanation there of the flattening of the cells is clearly
the stretching caused by the rapid transudation of fluid into

No. 3.] OESOPHAGEAL GLANDS OF URODELA. 95

the vesicle, but in these glands it is difficult to explain why the
fluid is not discharged into the cavity of the foregut before
the pressure gets sufficiently high to cause a stretching of the
cells. A possible explanation is the viscidity of the secretion
owing to the large number of mucous cells in these glands.

Fig. 4 shows a gland only moderately distended, and here it
is seen that the cells at the bottom, where the gland is unsup-
ported by neighboring glands, are drawn out flat, while those
at the side still retain their approximately columnar shape.

The two kinds of cells noticed in the earlier larva may still
be recognized, the clear mucous cells occupying the top and
neck of the flask, the granular cells the sides and base of the
flask. The protoplasm of the latter now stains strongly in
haematoxylin, and exhibits a faintly striated or finely vacuolated
structure. This is due to the presence of prozymogen, which
may also be demonstrated by the use of acid alcohol, followed
by aqueous haematoxylin after the method of Macallum.1 The
inner end of the cell between the nucleus and the lumen stains
but slightly in haemalum, but in sections treated with neutral
gentian it is seen to be filled with perfectly round, deeply stained
granules of zymogen. The neck of the flask-shaped gland is
occupied by long mucous cells of a columnar shape, which also
extend into the gland and form the top of the flask. In these
cells two zones may be distinguished, an outer protoplasmic,
which stains strongly and which contains a quantity of masked
iron, and an inner transparent and reticular. The meshes of
the latter are filled with a substance which stains faintly in
indulin, more readily in Mayer's mucicarmine. In sections
stained with neutral gentian many deeply stained granules may
be seen in the mucigenous portions of these cells. These are
somewhat elongated and not perfectly round, as are the zymo-
gen granules of the other kind of cell. Their significance is
not clear ; it is possible that they may indicate an imperfect
differentiation of the zymogenic and mucigenic functions at
this stage of development.

The surface epithelium in this region of the foregut consists
of alternate ciliated cells and goblet cells. Tracing the foregut

1 Journ. of Phys. Vol. xxii. 1897.

96 BENSLEY. [VoL. II.

backward, the glands become gradually tubular or saccular,
without any appearance of distention, and the ciliated cells dis-
appear, so that the rest of the glandular portion, as well as all
the posterior non-glandular portion of the stomach, is provided
with a mucigenous epithelium.

Attention should be called at this stage to the remarkable
resemblance between the mucigenous border of the gastric
epithelium and the cuticula of the cells in the buccal cavity.
Both have a characteristic striated appearance, and one is
tempted to think that they cannot be very different chem-
ically.

The cells of the tubular glands do not differ in any respect
from those of the flask-shaped glands. The mucous cells are
less numerous, and a few glands may be entirely without them.
The cells of all the glands, even the very last, contain both
zymogen granules and prozymogen.

There are as yet no pyloric glands formed ; the epithelium
of the posterior portion of the stomach is perfectly smooth
and without glandular outgrowths.

Even at this stage there is a remarkable resemblance be-

o

tween the anterior flask-shaped glands and the oesophageal
glands of Proteus and Necturus, and as development proceeds
this resemblance becomes more and more striking.

The mouth and pharynx are lined in the aquatic Amblystoma
larva by a stratified non-ciliated epithelium, with cuticular cells
and goblet cells. In a transverse series it may be seen that
immediately behind the last gill slit this changes to a ciliated
epithelium. One may thus consider the first ciliated cell in a
longitudinal section as indicating where the oesophagus begins.
Measured from this point the foregut in a larva 16 mm. long
is about 3 mm. in length. Of this .49 mm. at the anterior
end is non-glandular. Behind this we have a portion .45 mm.
long extending from the anterior border of the first gland to a
point where the foregut begins to expand to form the stomach.
This would doubtless, but for the presence of glands, be regarded
as a portion of the oesophagus. Beyond this again, the ciliated
epithelium extends into the stomach for a distance of .35 mm.
The rest of the stomach is lined by a mucigenous epithelium,

No. 3.] OESOPHAGEAL GLANDS OF URODELA. 97

like that of the adult except in its great capacity for division. The
posterior portion 1.3 mm. in length is still quite devoid of glands.

Staining with haemalum and neutral gentian shows that at
this stage also the cells forming the body of every gland in
the foregut contain both abundant zymogen granules and pro-
zymogen, and it is impossible to discern any difference what-
ever in the cells composing the large anterior flask-shaped
glands and the smaller posterior tubular glands respectively,
except that in the latter there is no evidence of distention and
consequent flattening.

From this stage onward the changes proceed somewhat more
slowly and may be summed up briefly.

In a larva 25 mm. in length the foregut measured 5.6 mm.
in length. Of this the anterior 1.4 mm. was non-glandular,
showing a relatively more rapid growth in length in this
portion of the foregut. The ciliated epithelium extended a
further distance of .84 mm. into the stomach, the posterior
portion of which, 3.36 mm. in length, was lined by the usual
mucigenous epithelium.

Fig. 5 is from a larva 32 mm. in length. It is at this stage
that the resemblance to the oesophageal glands of Proteus and
Necturus is most marked. The duct of the gland and the por-
tion of the wall nearest
to the surface epithe-
lium are composed of
elongated cylindrical
cells forming a single
row. Four of these

Cells are rCOresented in ^IG' 5-~Larva°f Amblystoma 32 mm. in length; oesophag-
eal gland. Ross obj., -fa in., Leitz ocular No. i.

Fig. 6, A, as seen under

a high magnification. Each presents an outer granular proto-
plasmic zone in which the oval nucleus is imbedded, and an
inner more extensive zone which is coarsely reticular. They
obviously represent the large, clear mucous cells of the ordi-
nary gastric glands, and, as we shall see, are actually transformed
into these in the adult. The difference in shape is dependent
on external conditions, such as the grouping of the cells, and
is not inherent in the cells themselves.

98

BENS LEY.

[VOL. II.

At the end of the gland they are succeeded suddenly by the
zymogenic cells. In these the minute structure is obscured
by the large amount of prozymogen present. The cells are less
flattened than at the earlier stage of development, probably
because the initial distention has been compensated by the rapid
division and growth of the cells. A number of these cells is
represented in Fig. 6, B. They are now somewhat columnar in

shape, with convex ends projecting into
the lumen. The nucleus is round or
oval and placed in the center of the cell,
though in the more columnar cell it is
often nearer the lumen than the base of
the cell. The free end of the cell may
be seen in sections stained with neutral
gentian, to be filled with granules of
zymogen. Other granules may be seen
at the sides of the nucleus, and a few
are occasionally found in the base of
the cell. The rest of the cell is oc-
cupied by a deeply staining protoplasm,
which owes its ability to absorb nuclear
stains to the large amount of prozymo-
gen present, as may be shown by the employment of Macallum's
methods of detecting masked iron. The distribution of the
prozymogen determines the appearance of the cell, and three
main types are to be recognized ; in the first the stain is diffused
through the whole of the protoplasm, but more pronounced
at the base and sides of the cell, and on close examination
a very finely vacuolated structure may be made out ; in the
second the whole or part of the cell exhibits long deeply stain-
ing fibrillae ; and in the third type the prozymogen is distrib-
uted as small irregularly staining particles throughout the
protoplasm.

All the three main types of cells composing the glands and
surface epithelium are still capable of division, and numerous
mitoses may be seen in all.

Oppel's description of the structure of the oesophageal
glands of Proteus would apply word for word to these glands

FIG. 6. — Oesophageal gland of
Amblystoma larva. A, mu-
cous cells; B, zymogenic cells.
Zeiss apoch. 2 mm., comp.
ocular 8.

No. 3.] OESOPHAGEAL GLANDS OF URODELA.

99

of the larval Amblystoma, and in the case of Necturus I have
satisfied myself, by comparison of the actual objects, that the
structures are identical. «

The latest larva examined was 65 mm. in length. This
animal was apparently about to undergo metamorphosis, as the
stratified epithelium of the mouth had been replaced by ciliated.
The pyloric glands were fully developed, and the ordinary gas-
tric glands had assumed the appearance they present in the
adult. The anterior portion of the stomach, about a millimeter
in extent, was still ciliated, but the saccular glands of this
region had undergone considerable modification. One of these
is shown in Fig. 7. It
will be seen that the
base of the gland has
grown out into a num-
ber of short secondary
tubules, formed for the
most part of zymo-
genic cells, and the
gland now consists of a
large number of such
tubules, each similar in structure to an ordinary gastric gland,
opening into a large common cavity lined by transparent mucous
cells corresponding to the neck cells of an ordinary gland. In
short, the saccular gland of the embryo is being transformed
into an anterior oxyntic gland of the adult.

Two of the most anterior glands in this larva were included
in the oesophagus. All were in full physiological activity and
were filled with zy.mogen granules.

I have been unable to secure a specimen of Amblystoma
undergoing metamorphosis, or one that has just completed it,
and am therefore unable to state positively whether all the sac-
cular glands become transformed by subsequent branching into
anterior oxyntic glands, or some of them degenerate and dis-
appear. There is in the latest larva that I have examined no
evidence of changes of a degenerate nature, and I am there-
fore inclined to believe that the most anterior glands, as well as
the rest, are taken up into the stomach, and that the oesophagus

FIG. 7. — Oesophageal gland of 65 mm. larva. Ross obj.
IB in., Leitz ocular No. i.

I00 BENSLEY. [VOL. II.

of the adult is entirely formed by rapid growth from the short
non-glandular region of the early larva.

One cannot doubt that the large saccular glands of the larval
Amblystoma are the homologues of the so-called oesophageal
glands of Proteus and Necturus. The failure of Kingsbury,
however, to detect the presence of zymogen granules in the
glands of Necturus led me to reinvestigate these structures
with a view of determining whether or not this was a real
point of difference. At first I employed a number of speci-
mens of Necturus which had been kept in the laboratory tank
for several months without food. In these cases the results
were negative, no zymogen granules were present. I after-
wards obtained two specimens captured in the vicinity of
Toronto, and in a perfect state of nutrition. In these no
difficulty was experienced in demonstrating the presence of
zymogen granules in the cells of the oesophageal glands.
Kingsbury's failure is, in all 'probability, to be ascribed to the
inadequate method he employed to demonstrate the granules.
For this purpose he employed treatment of the fresh glands
with osmic acid. Now it has been noted by Langley (7) and
Griitzner (3) that the protoplasm of ferment-secreting cells
which contain a great deal of reserve material (prozymogen)
stains strongly in osmic acid. For this reason
its use, in cases where the granules are few in
number and small, and where there is a great
deal of prozymogen present, is of little value.
This is precisely the condition in Necturus.
By the neutral gentian method, however, the
granules are stained much more strongly than
the prozymogen, and no difficulty is experi-
enced in demonstrating them when present,
s.— Zymogenic Fig. 8 shows a number of cells from such a

cells from oesophageal

gland of Necturus, preparation.

It should be added, however, that the oeso-

ules in free border of

cells, zeiss apoch. 2 phageal glands of Necturus do not present the

mm., ocular 8. • -, f r , . .

evidences of strong functional activity seen in
those of the larval Amblystoma. The granules are much smaller
and may be quite absent from many of the cells of the gland,

No. 3-] OESOPHAGEAL GLANDS OF URODELA. IOI

even in a well-nourished animal. This is particularly the case
in those cells the inner ends of which exhibit signs of degenera-
tion in the shape of the structures described by Kingsbury as
mucous globules. It is probable that there is a tendency for
these glands in Necturus to degenerate rather than remain of
physiological importance.

The so-called oesophageal glands of Proteus and Necturus
are really gastric glands the development of which has been
arrested. There is also in these animals an arrested develop-
ment of the foregut, compensation for which has been, in a
measure, attained bv the conversion of the anterior portion
of the stomach into a functional oesophagus. Only a short
anterior non-glandular portion actually corresponds to the
oesophagus of other Urodela.

Two questions remain to be considered, the relation of these
glands to the oesophageal glands of higher vertebrates and to
the oesophageal glands of the frog/

The first question is a comparatively simple one. The oeso-
phageal glands of higher vertebrates have no features in com-
mon with those of Batrachia and are probably of secondary
origin. In Reptilia oesophageal glands are rare, and where
they do occur, as, for example, in Testudo graeca, are simple
crypts lined by cells similar to those of the surface epithelium,
namely, ciliated cells and goblet cells, the latter predominating.
In birds and mammals, where the epithelium is usually of the
stratified squamous variety, they are more or less complex
mucous glands. In no case, as far as I am aware, has investi-
gation revealed in the oesophageal glands of Sauropsicla or
Mammalia the occurrence of ferment-secreting cells. It is
probable that the oesophageal glands of higher vertebrates
have arisen in response to a demand, in a very long and rela-
tively narrow oesophageal tube, for a more efficient lubricating
mechanism, and an epithelium that will withstand friction.
The first step in this process is the formation of deep crypts
lined by ciliated cells and many goblet cells ; the second, the
disappearance of the ciliated cells from the crypts so as to
form a pure mucous gland, and their replacement on the sur-
face by a stratified squamous epithelium.

102 BENSLEY. [VOL. II.

The second problem is less simple. Because of the very
exceptional conditions introduced in the case of the frog by
the herbivorous diet of the tadpole, and of the very extensive
histolytic changes which take place in the whole intestine dur-
ing metamorphosis, it becomes difficult to discuss this question
from the standpoint of histogenesis. The question is, whether
the oesophageal glands of the frog, like those of Proteus and
Necturus, are to be regarded as somewhat modified anterior
gastric glands. Let us examine, in the first place, the ana-
tomical characters on which the subdivision of the foregut has
been determined in this form. According to Wiedersheim (12)
the stomach begins at a point where the foregut experiences
an abrupt turn to the left. This is found on examination to
correspond to the point where the ciliated epithelium is suc-
ceeded by the cylindrical epithelium of the stomach. There
is also a slight constriction at this point and a change in the
color of the mucous membrane. Of these the only character
of importance is the change of epithelium. This is not, in my
opinion, a valid criterion for the following reasons : In Amblys-
toma ciliated epithelium is found in the anterior portion of
the stomach up to a late stage of development. In the tad-
pole, according to Gage, the whole foregut is ciliated, and
several observers record patches of ciliated cells in the stomach
of the adult frog. In several of our American "ganoids,"
Hopkins (4) and Macallum (9) describe the ciliated epithelium
as being continued over a considerable portion of the stomach.

It is true that the differences in the cells of the oesophageal
and gastric glands of the frog are very striking ; but if we
compare the oesophageal glands of the frog with the gastric
glands of any Urodele or of Bufo, these differences are not
apparent. The same cellular elements are present, with almost
the same arrangement and structure.

The gastric glands of the frog are, in fact, unique among
the Batrachia, in the small amount of zymogen which they
contain. May this not be but another instance in which this
animal, as compared with other Batrachia, exhibits an unusual
degree of specialization, the anterior gastric glands (so-called
oesophageal glands) having retained and developed the zymogenic

No. 3-] OESOPHAGEAL GLANDS OF URODELA. 103

function at the expense of the oxyntic function, and the
posterior the oxyntic at the expense of the zymogenic function,
thus foreshadowing in a parallel way the histological differ-
entiation which is seen in the chief and parietal cells of the
gastric glands of mammals ?

The conditions obtaining in the foregut of Proteus, Necturus,
and the larval Amblystoma are of interest apart from their
purely histological bearing. For it is obvious that, if the con-
dition in these animals is primitive, the gastric glands of the
ancestral types must have occupied a much more extensive
portion of the foregut than is the case in existing forms.

Among fishes the subdivision of the foregut into oesophagus
and stomach is well marked, not only among the more highly
specialized Teleostomes, but also in the sharks and rays. No
glands are present in the oesophagus, and the epithelium is
different from that of the stomach. In Amia, Lepidosteus, and
Acipenser, according to Macallum (9), it is not only extremely
difficult to decide on superficial examination where the oesoph-
agus ends and the stomach begins, but on microscopic exami-
nation the former is found to have a similar epithelium to the
stomach and to contain glands. The nature of these glands is
at present in doubt. No doubt the investigation of the struc-
ture and histogenesis of the elements of the foregut in these
forms, and more particularly in Polypterus, will yield highly
interesting and instructive results.

BEXSLEY.

LITERATURE.

1. BISCHOFF. "Ueber den Bau der Magenschleimhaut." Mailer's

Archiv, 1838.

2. CARLIER. "The Newt's Stomach during Digestion." La Cellule.

Tome xvi.

3. GRUTZNER. " Ueber Bildung und Ausscheidung von Fermenten."

PJliigers Archiv. Bd. xx.

4. HOPKINS. " On the Enteron of American Ganoids." Jotirnal of

Morphology. Vol. xi.

5. KINGSBURY. " The Historical Structure of the Enteron of Necturus

maculatus." Proc. Amer. Micr. Soc. Vol. xvi. Pt. i.

6. KLEIN. "Oesophagus." In Strieker's Handbuch der Lehre von deti

G ewe ben.

7. LANGLEY. " On the Histology and Physiology of Pepsin-forming

Glands." Phil. Trans. Roy. Soc. Vol. clxxii.

8. LEYDIG. " Anat. hist. Untersuch. liber Fische und Reptilien."

Berlin, 1853.

9. MACALLUM. u The Alimentary Canal and Pancreas of Acipenser,

Amia, etc." Journ. of Anat. and Phys. Vol. xx.

10. MACALLUM. "On the Distribution of the Assimilated Iron Com-
pounds, etc." Quart, foitrn. Micr. Sci. Vol. xxxviii. N. s.

n. OPPEL. " Beitrage zur Anat. des Proteus anguineus." Arch. f.
Mik. Anat. Bd. xxxiv.

12. WIEDERSHEIM. Ecker's Anatomic des Frosches.

AN EXPERIMENTAL DEMONSTRATION OF THE

REGENERATION OF THE PHARYNX OF

ALLOLOBOPHORA FROM ENDODERM.

JOHANNA KROEBER.

SEVERAL recent investigators have shown with more or less
probability that the lining of the new pharynx which develops
during the regeneration of the head in certain earthworms

o o

comes from the endoderm, while the pharynx of the embryo is
lined by ectoderm.1 It seemed that by means of experimental
methods this relation might be definitely determined. In the
following pages I shall describe some experiments on Allolo-
bophora foetida that demonstrate, I think, that the lining of
the new pharynx is in fact derived from the endoderm.

Hescheler showed, as the result of observations made prin-
cipally on Allolobophora terrestris, that when the five anterior
segments are cut off the pharynx is regenerated by a growing
forward of the old digestive tract up to the third segment, and
that the new buccal cavity occupying the first three segments
is formed by an ectodermal invagination. The old pharynx
was not completely removed in these operations, since in the
normal worm its cavity frequently extends beyond the fifth
segment and its thickened muscular dorsal wall always goes
back into the sixth, so that Hescheler's results are open to
the objection that in his experiments a part, at least, of the old
pharyngeal zvall ahvays remained behind as a possible source
for the regeneration of the new pharynx.

Rievel in experimenting on certain Lumbricidae (Allolobo-
phora foetida, Allolobophora terrestris, Lumbricus rubellus)
cut off anterior ends consisting of between one-third to two-
thirds of the entire body. He arrives at the conclusion that
the pharynx is regenerated from the walls of the digestive tract

1 See Hoffman, Zeit.f. wiss. Zool. Bd. Ixvi. 1899.

105

106 KROEBER. [VOL. II.

at the point where this was cut, and that no ectodermal invagi-
nation whatever occurs, the endodermal diverticulum joining
the body wall to form the mouth at the very anterior end of
the worm.1

Haase showed that in Tubifex, when four to six anterior
segments have been removed, the pharynx grows forward out
of the walls of the digestive tract and meets an ectodermal
invagination of somewhat varying size. This ectodermal pouch,
which forms the buccal cavity, is small in all cases, never
extending quite as far back even as the region of the cerebral
ganglion.

Von Wagner's observations on Lumbriculus show that the
point of union of ectoderm and endoderm, originally at the
anterior end of the animal, subsequently comes to lie more
posteriorly, on account of the forward growth of the " Kop-
flappen " and accompanying turning in of the ectoderm.

The differences in the accounts cited above show clearly
that it is almost impossible to determine with certainty, merely
by observation, just how much of the regenerated pharynx
ultimately arises from the ectoderm and how much from the
endoderm. It is very easy to see where ectoderm and endo-
derm meet, but the point of fusion is lost soon afterwards, and
since the regenerated head continues to increase in size, it is
presumably possible that the point of union may come to lie at
some distance from its original position. At the time when
the pharynx opens to the exterior its walls are not sufficiently
developed for one to be able to determine whether the muscles
will grow around the endodermal part of the tube ; but if in
some manner the fusion of the ectoderm with the endoderm
could be delayed long enough for the pharyngeal muscles to
form around the latter, then the origin of the pharynx might
be determined. Hescheler affirms that it is possible to make
out the exact limits of the ectoderm by using stains which
bring out the cuticle covering this layer. I used the stain
which Hescheler mentions as giving the best results, but found
that, while my preparations showed in general an agreement

1 For criticism of Rievel's results, see papers by Morgan (Roux's Archiv,
Bd. v, 1897) and Hescheler (Jenaische Zeitsckrift, 1898).

No. 3-] THE PHARYNX OF ALLOLOBOPHORA. 107

with the figures of Hcscheler in regard to the extent of the
cuticle on the dorsal wall of the pharynx, they contained also
some alternating patches of what seemed to be cuticle and of
ciliated areas on the ventral wall and at points in the
digestive tract even further back than the regenerated
pharynx itself.

For this reason I have attempted to get more certain results
by the use of the following experimental methods. Worms,
from which the seven anterior segments had been removed, so
that no part whatever of the old pharynx was left behind, were
allowed to regenerate for a period of between twelve and
eighteen days. As a rule the fusion of the ectodermal invagi-
nation with the pharynx occurs about fifteen days after the
removal of the anterior end of the worm, --although there is
considerable individual variation in regard to this point, and
also some difference due probably to the temperature, etc. At
the end of this time the anterior tip of the new part of the
worm was removed in one of two ways : either it was burned
off by touching it with a hot needle, or it was cut off with fine
scissors. The latter method, though more difficult to carry
out successfully, proved to be the better one because the piece
cut off could be preserved to show whether the pharynx had
joined the ectoderm at the time of the second operation. The
worms were once more allowed to regenerate and were finally
killed between ten and fifteen days after the second operation.
In all cases the worms survived both operations and showed a
perfectly normal regeneration, — the only point of difference from
worms that had undergone only the first operation being that
the new pharynx had time to regenerate before the second
ectodermal invagination had fused with its anterior end. The
object of the experiment was to determine whether a normal
pharynx would develop from the endoderm if the fusion of the
ectoderm with it was prevented for a sufficient length of time
to allow this development to take place.

It is difficult to determine on the living object whether or
not the ectodermal invagination has met the endoderm, and
since for my purposes it was best to wait as long as possible
before the second operation, it happened in two or three cases,

loS

KROEBER.

[VOL. II.

as sections of the small pieces removed showed, that ectoderm
and endoderm had met. In the majority of instances, however,
I was fortunate enough to remove the invaginating ectoderm
just in time. In cases where this was done with a hot needle
there is, of course, nothing to prove that the fusion had not
taken place. There is ground for such a belief, however, in
the fact that, of a number of worms whose small anterior ends
were cut off at the same time after the first operation as when
the burning was done, and which were kept under exactly the
same conditions, there was not a single one in which the fusion

had taken place.

The same difficulty
presents itself again in
determining the time at
which the worm is to
be killed. I succeeded,
however, in getting a
number of cases where,
though the pharynx and
the ectoderm were just
on the point of joining,
they had not quite done
so when the worm was
killed. The accompany-
ing figures show two
worms in this condition.
Fig. i shows a verti-
cal longitudinal section
of a worm from which seven segments were removed on
January 15. On February 2, that is to say eighteen days
later, the tip of the newly regenerated part was cut off. This
piece was preserved and sectioned and was found to include the
whole of the ectodermal invagination besides the anterior end
of the pharyngeal diverticulum which had not yet broken
through to the exterior. Fourteen clays after this operation,
on February 16, the worm was killed.

Fig. 2 represents a vertical longitudinal section of a worm
from which the first seven segments were cut off on January

FIG. i.

No. 3.] THE PHARYNX OF ALLOLOBOPHORA.

15. The tip of the regenerated part was destroyed with
a hot needle on February 2, and the worm was killed on
February 17.

Both figures show the diverticulum which has grown out
from the walls of the oesophagus about to open to the exterior
by fusion with the ectodermal pit ; and a comparison with the
sections in the same neighborhood shows that these two repre-
sent the nearest approach of ectoderm and endoderm to be
found in the two specimens. The walls of the pharynx and
its musculature, especially
on the dorsal side, are
well developed. In both
worms a nerve cord and
a cerebral ganglion have
been formed, the latter
for the second time.
Owing: to the slight ob-

o o

liqueness of the section, as
shown in Fig. 2, the nerve
cord is cut for only a part
of its length. The muscles
of the body wall have be-
gun to differentiate and
there are clear indications
of metamerism. All of the
worms used in this set of experiments, as well as all those in a
later set made to test these results, present a similar condition
of things.

From these results we must conclude that the lining of the
pharynx is regenerated from the endoderm, while the new ecto-
derm turns in for a very short distance to meet the pharynx
and form the mouth.

The objection may be raised that the possibility of a later
pushing in of the ectoderm to form the ultimate lining of the
pharynx is in no way removed. But there is no evidence for
such an occurrence and, even if it did take place, the fact
remains that the musculature of the pharynx develops around
an endodermal tube, as my experiments have shown, while in

FIG. 2.

HO KROEBER.

the embryo the lining of this same region is derived from the
ectoderm.

The preceding work was done under the direction of Prof.
T. H. Morgan, to whom I wish to express my indebtedness.

BRYN MAWR, May 26, 1900.

FURTHER EXPERIMENTS ON THE REGENERA-
TION OF TISSUE COMPOSED OF PARTS
OF TWO SPECIES.

T. H. MORGAN.

THE experiments that I made a year ago were undertaken
in order to find out if regenerated tissue, made up of cells
derived from two species, showed any mixing of the specific
characters of the two species. For this purpose I grafted the
tail of a tadpole of one species of frog upon the posterior end
of a tadpole of another species. Later the tail was cut off in
such a way (as indicated by the line b-b in Fig. 5) that the
ectoderm left at the exposed edge belonged in part to one
species, in part to the other. When the new tail regenerated
there was found to be no mixing of the characters of the ecto-
dermal cells along their line of contact in the new part. The
results were unsatisfactory
from one point of view,
inasmuch as the small
piece of ectoderm left after
the operation is carried out
to the tip of the new tail
and increases proportion-
ally less in area than the
rest of the new part, so

FIG. i.

that although it is highly

probable that near the tip of the tail new ectodermal cells
are being formed by both kinds of ectoderm, still I did not
demonstrate that this is actually the case. Moreover, I found
that in the later stages the difference in color between the
two kinds of ectoderm was less marked than at first, so that
the experiment would have been more convincing had the tail
been cut off at an earlier stage. This I have done during
the present spring, and the results in regard to the ectoderm

1 1 1

112 MORGAN. [VOL. II.

confirm in every way those given in my former paper. In the
experiments made this spring my main object has been, how-
ever, to carry out the experiment in such a way that there
would be left at the exposed edge, when the grafted tail was
cut off, the internal tissues of two species. In this way I
hoped to be able to determine more definitely if, in the
newly regenerated part, the tissues mutually influence each
other.

The day after the grafting had been performed (i.e., after eight-
een to twenty-four hours) the tail was cut off at the region of
union of the two components, as shown by the line a-a in Fig. i.
In this way there is left at the exposed edge not only the ecto-
derm of the two species, but the inner tissues also. The regen-
eration that takes place from the exposed edge will include
material derived from both components. Two possibilities
presented themselves. First, would the new part be formed of

cells intermediate in
character between the
two species as the
result of an interaction
of the cells on each
other ; or, second,
would the new material
preserve the character-
istics of the region

FIG. 2. r . ,

from which it arises,

or, in other words, one-half of the tail show the characters
of one species and the other half of the other species ? It is
further possible that the new cells might intermingle, and if so
the tail might appear to be of a hybrid character.

Other experiments of minor interest have also been studied.
For instance, in several cases the grafted tail was cut off after
twenty-four hours very near its line of union to the major
component, as shown in Fig. 3, A, B. The experiment was
made in order to see if the major component might not have
some influence on the regenerated part from which it is sepa-
rated by only a narrow band of tissue of the minor component,
but no such influence was observed.

No. 3.]

THE REGENERATION OF TISSUE.

I have found a much safer criterion than before for distin-
guishing the inner tissues of the two species of tadpoles used
in these experiments. In my former experiments I used the
differences in color of the pigment cells. I find that this can-
not be relied upon under all circumstances. But the muscle
tissue of the tail of Rana palustris is, especially in the early
stages, golden-yellow, while in Rana sylvatica the same cells

A (April 24).

B (April 25).

are slaty-white. The two kinds of cells can be easily distin-
guished by means of this color difference. Other experiments
have shown also that this difference in color is transmitted to
the regenerating tissues in the new tail, so that it can be relied
upon in the grafting experiments.

In the first series of experiments the tails were grafted as
described in my former paper. After eighteen to twenty-four
hours, as a rule, the tail was cut off, as shown by the line a-a in
Fig. i. Before grafting it was found more convenient to cut

H^ MORGAN. [VOL. II.

off the tails obliquely, as shown in the figure — the more
anterior end on the dorsal side. Consequently, in order to
carry out the second operation of cutting off the tail through
the line of union, the cut was made also obliquely, but with
the ventral side forward. In a few cases the tails were first
cut off with the ventral side further forward (Fig. 2), and the
subsequent cutting off was made with the dorsal side forward, as
shown by a-a in Fig. 2, but the results were practically the same.

It was found easier to graft the tail of Rana palustris on the
posterior end of Rana sylvatica than the reverse. On an aver-
age five operations of the former succeeded to one of the latter.
The reason for this cannot be given, but it may be due to some
difference in the relative sizes of the two components that is
more favorable for union in one than in the other way. The
result recalls the experiments in cross-fertilization of the eggs
in different species, where the crossing can be more easily
carried out in one direction than in the other. In this case
also the results may be due in some cases to a gross, physical
difference, as Pfluger has tried to show for the frog's egg.

In the large majority of cases in which the experiment was
carried out as shown in Fig. I, the core of the new tail seemed
to be formed by the minor component, --i.e., if a yellow tail
(R. palustris) had been grafted upon a black tadpole (R. syl-
vatica) and then after twenty-four hours the tail had been cut
off obliquely (Fig. 4, A], the central part of the new tail would
be composed entirely of the yellow tissue derived from the minor
component (Fig. 4, B, C}. The small piece of yellow ecto-
derm is carried out on the new tail and remains near the tip.
It covers a larger area than at first, but it increases not nearly
so fast as the rest of the new, yellow tissue of the new tail.
The distinctive differences in color can only be seen in the
core of the tail, i.e., in the cells that form the muscles. On
each side of this axial core a broad fin is present containing
inside a gelatinous-like substance with scattered cells, but this
fin does not show any difference in color in the two species.
It is, therefore, probable that in many cases in which the core
of the new tail appears to be composed only of tissue from the
minor component that the ventral (or dorsal) connective tissue

No. 3.]

THE REGENERATION OF TISSUE.

of the fin is derived from the major component. The differ-
ences in the mesodermal pigment cells are at times very strik-
ing, and in all such cases the pigment cells are like those in the
tissues from which they immediately arise ; but while in many

A (.April 25).

B (May 2).

cases they furnish a safe criterion, in others the difference
cannot be made out with certainty. On the other hand, the
differences in the muscle tissue of the core can always be seen.
The explanation of the result, viz., that the new tail is in
most cases (in forty-seven cases out of sixty) like the minor
component, is that in nearly all of these operations too large
a piece of the grafted tail has been left. It has contained the

H6 MORGAN. [VOL. II.

notochord and nerve cord and the tissue immediately around
them, and from these the new tail has grown out. I did not
discover this until most of my material had been used. After
this I cut off the grafted tails nearer the line of union, and
although regeneration did not take place so well in several
cases, still those that regenerated showed more often both
parts contributing to the new tail. The same result followed
in a number of cases in the previous operations, and we may
now examine how regeneration takes place in such cases.

In thirteen cases (out of sixty) there was found evidence of
a dual or compound character to the new tail. In all cases
observed there was no evidence to show that the duality was
the result of the tissues being mixed in character either by
commingling of the cells (each cell retaining its specific charac-
ters) or by a hybridizing of the cells (due to mutual influence).
The duality consisted in each part, regenerating cells like itself,
so that definite regions of the new tail were made up of one or
of the other kind of tissue. For instance, the new tail might
be made up above of the slaty-colored tissue of R. sylvatica
and below of the yellow tissue of R. palustris. There is no
evidence of a shading of one kind of tissue into the other along
the line of meeting, but this point would be very difficult to
determine positively. There is further no evidence that the
two kinds of tissue are any more commingled at the distal end
of the tail than at the base.

In regard to the notochord and nerve cord it is extremely
unlikely that the cut would ever pass obliquely through the line
of union of the one or of the other, as these structures are very
small in cross-section. It is, therefore, probable that in nearly
every case the new notochord and the new nerve cord are made
up of cells belonging entirely to one component. Further-
more, these two structures lie so near together that it is not
probable that the cut would pass between them in such a way
that the nerve cord at the exposed edge would belong to one
component and the notochord to the other component.

The details of the successful experiments are as follows :
On April 14 and 15 nine grafts were made, as shown in Figs. I
and 2. On April 16 these were cut off, as indicated in Fig. i, a-a,

No. 3-] THE REGENERATION OF TISSUE. 117

and Fig. 2,a-a, but unfortunately the two lots were not kept sepa-
rately. On April 29 when again examined new tails had begun
to regenerate, and two individuals out of the nine showed that
the core of the new tail was compound in character. In both
the major component was black and the minor yellow. In one
of these the new tail was yellow on the dorsal side and black
on the ventral, and in the other the new tail was black on the
dorsal side and yellow on the ventral.

In another series the experiment was somewhat different.
The grafting took place on April 17. Two days later the ecto-
derm of the minor component had been carried out further on
the tail (Fig. 5), so that at the base of the tail the inner tis-
sues of the minor component were covered by the ectoderm of

FIG. 5.

the major component. At this time (April 19) the tail was
cut off obliquely, as indicated by the line a-a in Fig. 4, leaving
the inner tissues of both components exposed at the cut sur-
face. On May 19 all three of the tadpoles that had been
operated upon showed a compound tail. One of these tadpoles
was R. palustris and the other two were R. sylvatica, with
grafted tails of the other species, respectively.

In a third series of five individuals, grafted April 19 and cut
off April 20, as in Fig. i, two showed later a compound tail;
and in a third I was in doubt whether or not a few of the
yellow cells of the major component entered the new tail.

In three later experiments in which the tail had been cut
off, so that a smaller piece of the minor component was left
attached, a larger number regenerated compound tails.

In one of these experiments the grafting took place on
April 27, and the tail was cut off on the following day. One
of the three produced a tail composed of both kinds of tissue.

MORGAN.

[VOL. II.

In another experiment grafted April 28 (2.30 P.M.), as shown
in Fig. 2, and cut off April 29 (10 A.M.), two individuals formed
abnormal tails and a third a compound tail. The tail of this
individual is represented in Fig. 6. On the ventral side of the
new tail are found the slate-colored cells of the major compo-
nent, and on the dorsal side the yellow cells of the grafted
piece. (It is not possible to show this difference satisfactorily
in a simple uncolorecl drawing, since the principal difference is
one of color.) In addition to this difference one can see in the
region at which the grafting took place and where the new
tissue arises from the old that each component contributes its
half to the new tail. Moreover, in all these cases the tadpoles

had been carefully ob-
served from day to day
(and not only at the
intervals recorded in the
text) and the gradual
formation of the com-
pound tail observed.

In another experiment
on April 28 the tail was
cutoff on April 29. One
of the tadpoles did not
regenerate a new tail,
another (R. sylvatica) had a compound tail, and one had a bifid
tail, one branch being compound. Finally in another series in
which nine grafts were made, one produced a compound tail,
another may have contained a small amount of the major
component in the new tail, six regenerated entirely like the
minor component, and one was abnormal.

In addition to these cases there were three others (in the
total of sixty cases) in which there was an overlapping of the
two components in the tail, as in Fig. 6. In two of these
the core of the new tail came from the minor component, but
it is highly probable that a small addition came from the major
component also. In the third case the new tail contained at its
more distal end elements from both components. Unfortunately
this lot was killed accidentally before they regenerated further.

FIG. 6.

No. 3-]

THE REGENERATION OF TISSUE.

119

In several cases double tails grew out enclosed in the same
common fin, and lay usually in the same plane. In some cases
the core of one of the new tails was derived from one of the
components and the other from the other component. In
several cases one or the other new tail received material from
both components. In one of these cases it could be seen with the
greatest clearness that the compound tail received material from
both sources (Fig. 7).
Cases of this kind are
particularly convincing,
since they furnish all
the data for comparison
between the two kinds
of regenerating tissue
of the two components.
The dorsal tail was
yellow and the upper
part of the ventral tail
was also yellow, and its
tissue precisely like that of the dorsal tail. The pigment cells
also of the yellow component extended out on to both tails.

These cases of double tails are probably due to imperfect
grafting, - -the notochorcl and nerve cord of the two compo-
nents not being in the same position, so that when the tail is
cut off both sets of structures are exposed and a tail develops
from each.

Conclusions.- -The experiments demonstrate that a single
tail may be formed by the regeneration of tissue derived from
two species, and that in such cases there is no specific change
produced in the one kind of new tissue by the other. Each
kind of tissue regenerates its like, and the two kinds combine
to form a single morphological organ, --the tail.

FIG. 7.

May, 1900.

REVIEW OF VON GUAITA'S EXPERIMENTS IN

BREEDING MICE.

C. B. DAVENPORT.

IN the two latest volumes of the BericJite der Naturfor-
scJicnden Gescllschaft zu Freiburg, G. von Guaita gives the
results of his breeding experiments. He started in 1896 with
fifty-five Japanese walzing mice and with numerous wJiite mice
belonging to a race bred by Weismann since 1888, and made
crossings through seven generations to 1900. His main data
concern the inheritance of color ; incidentally, data were got
on the diminution of fertility with in-and-in breeding.

Diminution of Fertility.

These results were gained chiefly from Weismann's breed-
ings. The total number of young, the number of litters, and
the number of young per litter are given below for each decade
of generations.

ist to roth generation : 1345 young ; 219 litters ; avg. young per litter, 6.1
Iithto2oth " 352 " 62 " " " " " 5.6

2ist to 29th " 124 " 29 " " " " " 4.2

In von Guaita's breedings :

ist and 2d generations, . . . " " " " 3.5

3rd and 4th " ...."•'" " 3.7

5th and 6th " ....""" " 2.9

Thus in the successive generations there is a reduction in fer-
tility of about thirty per cent ; and this is probably due to too
close breeding.

121

122

DAVENPORT.

[VOL. II.

Color of Mixtures.

Japanese walzing mice vary in color, but are chiefly piebald
-black and white. White mice are without pigment (true
albinos) and breed very true.

Crosses of walzing 9 : white $ and white 9 x walzing $
gave twenty-eight young. All were of a gray color and indis-
tinguishable either in respect to color or size from the common
house mouse. Also in temperament they were like the house
mouse, for they were very wild and lively (unlike either parent)
and the walzing action was entirely absent from all the mice of
this second generation. Similar results were got by Haacke
('95) after crossing the same races. Haacke says : " When
you pair a blue and white spotted walzing mouse with a com-
mon white mouse you get either (and usually) uniformly gray
mice, which cannot be distinguished from the wild house
mouse, or else (more rarely) uniformly black mice." These
results, then, lead to the conclusion that when very unlike
races of mice are crossed the result is often or usually a
reversion.

A third generation was next produced by von Guaita by
mating two of the gray mice or reversions. Four pairs were
thus mated and forty-four young were reared --all having both
parents gray, and half their grandparents walzing and half
white. These fourty-four mice are placed in nine color classes,
as follows :

'•House"
or Gray

type.
Albino.

Walzing
type.

entirely gray,

gray with isolated markings,

black [essentially house type],

pure white, red eyes,

white walzers,

gray-white spotted walzers, .

gray walzers,

black-white spotted walzers, .

black walzers,

r

Per cent

'= 1

3

57

1 1

25

3 1

i

2 }> 8

I
I

44

18

100

The most striking phenomenon of this third generation is
the sudden occurrence of great variation. In the language of

No. 3.]

/:-.\7v-;AVj//-:.\v.s- /.\-

MICE.

123

plant breeders "the type is broken." Plant breeders (e.g.,
Focke and de Vries) have long ago observed that the progeny
of hybrids is extraordinarily variable.

Fourth Generation. - - Several pairs of the foregoing descend-
ants of the reverted gray mice were mated, and thirty-one
young sorted into eight classes were obtained, as follows :

Gray and

Third Generation Igp^

Gray 9
White-

Walzer $

G ray 9
White $

White 9
White $

Spotted 9
< iray and
Spotted $

Total

Sum

Per
Cent

Uniformly gray

2

12

'4

Gray with markings .

2

2

4

Black. . . .

o

2

2O

65

Albinos

i

4

5

16

Uniformly gray walzers

i

I

2

Gray walzers with spots

I

I

Black-white walzers . .

i

I

Black walzers ....

2

2

6

*9

9

16

4

2

31

IOO

As in the third generation, there is here great variation.
The results may be generalized as follows :

(1) All descendants of albinic parents are albinos.

(2) When both parents are gray and spotted all descendants
(2) are gray and spotted.

(3) Gray 9 :: white $ gives 88 per cent gray and 12 per
cent walzers ; the white is shut out.

(4) Gray 9 white walzer $ gives 44 per cent gray, 44
per cent walzers, and i 2 per cent white.

Fifth Generation.- -To save room we will henceforth make
use of abbreviations for the names of colors, as follows :

124

DAVENPORT.

[VOL. II.

S2

0)
N

.— i

OS
^

C/5 rt tfl

^ «« & h

qj ^ ^ D

N W 0) J-J

"rt ^J -ti os

& 5 -^ ^

2^1-3

'P nj rt rt

^ bJO fcJO -^

bJC

o
oS

en

bfi

rt

tr.

fc/j
C

S
03

o

o

c

be bfi 3 -° rt
II II II II II

O O M « <

^

"rt

c/3

(U

CJ

<u

oj

rt

O,

OJ

a

OJ

M-l

<u
Jl

H

1- ^

mo -

o

£8

^

o

1

N OO fO

CO - "

ro

"5

«fOOO\ « N N fO 10

o

ur> H-I

H

•£

S ^

VC

VO

10 O x

1-1

C

O

x ^

•i O

i^ CT\

-f

10 — - X

ro

-f

x is

< o

x

0

Is

•* i. x

M « r^

"

O C

x

x

X "*

M

ro

m -^ x

•£ K

X

O

«o fo

> ^

o # X

00 0 t^ ri LO

ri

x 0

0 ^

O PQ

X

~f K

^ x

ei i- -i >-

LO

x CH

^ |

U >— i ^

o.^m^ < l^l3^^

C rt

O r' ^ r' -^ ^

No. 3.] EXPERIMENTS IN BREEDING MICE.

Sixth Generation.

Gen.

Colors

Total

Sum

1'er

Cent

II

All gray or reversions

III

A, 2; Gw,4; G, 2

I A, i ; Gw, 2; )
JG,3;Ww,2J

Gw, 2 ; G, 3 ; Ww, 3

G,4; \Vw,4

IV
V

Gw, 4
Gw x Gw

!Gw,i;Bw,i; )
Wg,i;Wb,i )
Wgw x G

( Gw, i ;

1 B, i;
B x A

Bw, i;|
Wb, I \

Wb x B

Bw,i;B,2;W,i

A x Ww

G

I

I

2

Gw

20

I

21

B

Hw

A

10

7
16

18

I
I

4

8

I?

48
32

58
33

Ww

i

1

\Vb

I

I

3

4

31

3

43

2

4

83

IOO

Seventh Generation.- -The colors of only eight members of

this generation were determined

too few to be significant.

General Res ~n Its.

In the successive generations the percentage of walzing
individuals undergoes a steady decline from eighteen per cent
and nineteen per cent in the third and fourth generations to
eleven per cent in the fifth and four per cent in the sixth gen-
eration. Is this decline due to the elimination of an unstable
condition or to the circumstance that too little of the walzing
blood has been employed in the later crosses to keep up the
original proportion ? The question whether the normal law of
inheritance is followed here may, indeed, be asked of all the
colors. The normal law of inheritance, as defined by Galton,
is that one-half the heritage of any generation is derived from
the parents, one-fourth from the grandparents, one-eighth from
the great-grandparents, and so on, according to the formula :

Inheritance --. ' k1 + ' k2 + i k3 + TL k4 + etc.

126 DAVENPORT. [VOL. II.

To apply the normal law of inheritance it is convenient to
assume it and to compare the theoretical heritage with the
empirical. If the two agree, the validity of the law is estab-
lished in this case ; conversely, if the two do not agree, the law
does not hold. This method of testing the law is the same as
that employed by Galton ('98) in the case of Bassett hounds.
It may be illustrated by the calculation of the theoretical num-
ber of albinos in the sixth generation. Let us take the first
column. If one of the two parents were an albino, we should
expect at least \ x 50^0 of the progeny to. be such. If both
parents were A, | x 50/0 of progeny at least should be A.
If, in addition, all of the grandparents (Gen. IV) were A, we
should expect at least | x 50^0 + | x 25/0 of the VI Gen. to
be A. In general, if ;/v, wiv, 7/iU, etc., represent the number of
times an albino appears as ancestor in the different genera-
tions, then the proportion of albinos in the sixth generation
should be :

=-^ x 50/0 + Wiv x 2$fi + -i" x I2.5'/, +--1 x 6.25/0 .
24 1 6

+ x 3. 1 25/0 + i x 3-125/0.

*J "^

The last term is got by observing that the ancestors of half
of the first generation were exclusively albinos for many gen-
erations, while the ancestors of the other half were exclusively
walzers. The value of A is similarly calculated for each column,
and the theoretical number of individuals for each column
is found. Their sum should be equal to the observed number,
or, when reduced to percentages, to the observed percentage
of total. The closeness of theory to observation is some-
times striking. Thus if we compare column by column the
observed and theoretical frequencies of walzers in the fifth
generation we get :

«'"|.UMN i 2 3 4 5 6 TOTAL PERCENT

Observed 2 7 i 3 o o 13 n

Calculated 2.19 6.00 1.13 4.88 o o 14.19 12.5

In the following table are given for each generation the
observed and (in parenthesis) the corresponding calculated
percentages for each color :

No. 3-] EXPERIMENTS IN BREEDING MKi:. 127

GENERATION

i

II

III

IV

V

VI

Black alone

(B) . . .

\ o

0

(o)

7
(o)

7
(o)

25

08)

30
(40)

Gray less black (G)

1 °

IOO

50
(So)

58

(48)

48

(60)

28
(28)

Total gray

and black

C

IOO

(o)

57
(So)

65

73

(78)

58
(69)

Albinos

550

o

(50)

25
(25)

16

(32)

1 6
(9)

38
08)

Walzers ,

5 5°

0

(50)

1 8
(25)

'9

(20)

1 1
03)

4
03)

Several remarkable things come out of this table. In the
first place the most marked departure from Gallon's Law of
Ancestral Inheritance is seen in the second generation, where
the gray, non-walzing reversions suddenly made their appear-
ance. We know as yet little concerning the laws of the phe-
nomenon called "reversion"; but whether it be considered a
remote atavism or only an apparent "inheritance," it seems
equally to form an exception to Galton's Law.

Secondly, the case of the walzers does indeed look like an
exception to Galton's Law. It looks as though the walzing con-
dition were an unstable condition being rapidly eliminated. In so
far the result opposes the usual expectation of sport prepotency.

Thirdly, the albinos, likewise sports, apparently are pre-
potent, since there is twice the proportion there should be in
the sixth generation. The numbers are so large that one can
hardly object that these figures are not altogether significant.

Fourthly, the grays run close to theory, excepting always
generation II. They are nearest to the original type of Mus
muscnlns and seem to inherit in the most nearly normal
fashion.

In conclusion, then, we may say that the data afforded by
these breeding experiments indicate, so far as they go, that
Galton's Law of Inheritance holds only with form units which
are not very divergent from the type, and that among sports
we may have some that show a great stability and prepotency,
while we may occasionally have others which are physiolog-
ically so unfit that they are unstable and have less than normal
potency.

128 DAVENPORT.

LITERATURE CITED.

FOCKI . \V. O. ('81). Die Pflanzen-Mischlinge. Berlin, Borntraeger,

1 88 1. 569 pp.
GALTOX, F. ('97). " The Average Contribution of Each Several Ancestor

to the Total Heritage of the Offspring." Proc. Roy. Soc. London.

Vol. Ixi, pp. 401-413.
GI'.UTA, (i. vox ('98). •• \"ersuche mit Kreuzungen von verschiedenen

Rassen der Hausmaus." Ber. Naturf. Ges. su Freiburg. Bd. x,

3 Heft. pp. 317-332. April, 1898.
Gi'AiTA, G. vox ('00). " Zweite Mittheilung iiber Versuche mit Kreuzungen

von verschiedenen Hausmausrassen." Ber. A'aturf. Ges. zu Freiburg.

Bd. xi, 2 Heft, pp. 131-138. August, 1900.
HAACKE, W. ('95). " Ueber Wesen, Ursachen und Vererbung von Albin-

ismus und Scheckung und iiber deren Bedeutung fiir vererbungstheo-

retische und entwicklungsmechanische Fragen." Biol. Centralbl. Bd.

xv, pp. 44-78. 1895.

VARIATION IN THE TEETH OF NEREIS.

MARY HEFFERAN.

THE purpose of this quantitative study in variation is to
determine the modal condition of a character in a species of
Nereis commonly found on the west coast of the Atlantic.
The material was very generously placed at my disposal by
Professor Charles B. Davenport, who collected it during the
summer of 1899 at Cold Spring Harbor, Long Island. The
animals were found in the sand of the beach and were taken at
random, excepting that small ones were rejected.

These annelids went by the familiar name of Nereis virens,
but upon comparing them with Ehlers's ('68, p. 559) descriptions
and drawings of that species, I found that although they agreed
in most characters, an important difference occurred in the
length of the first or postoccipital segment ; that of N. virens
being twice as long as the second segment, while that of the
Cold Spring Harbor form is about equal to or even slightly
less than the second in length. In this character, as also in
that of certain parapodal bristles, the " Sichelanhange," which
are rather short and broad instead of slender and long as in
Ar. virens, the Cold Spring Harbor species agreed well with
Ehlers's description of A7", limbata, the distribution of which
also includes the west Atlantic coast. From these two points,
and from the fact that Cold Spring Harbor is slightly south
of the southern limit described for N. virens, and within the
range of N. limbata, it seems probable that we are here dealing
with the latter of Ehlers's two species. It may be possible
that the two species overlap in this region at the southern
limit of N. virens, and that my collection contained both.
However, nothing in the numerical results of my investigation
seemed to suggest two distinct or even transitional forms.

129

130 HEFFERAN. [ VOL. II.

I. Method.

The specific character selected for investigation was the
number of teeth on the jaw. This number is commonly stated
by authors in descriptions of species.

The jaws are two in number, from I to 3 mm. in length,
brown, horny, curved, and serrated along the inner or falcate
margin. They are at the extremity of a large exsertile pro-
boscis which is usually retracted in alcoholic specimens, so
that in order to free the jaws it is necessary to cut down the
median line of the head, ventrally, through to the inside of the
muscular proboscis. By turning the head backwards the jaws
can be made to extrude, and the teeth counted by means of a
hand lens. In the specimens killed later in the season by a
slowly killing fluid, the jaws remained extruded.

In counting, some difficulty was experienced in fixing a limit
in either direction, at the curved, distal end in those cases in
which very fine teeth ran to the tip, and particularly at the
proximal extremity where the outlines of the teeth were indefi-
nite. In order to count these it was necessary to pull out the
jaws gently with a forceps, and to free the bases from connect-
ive tissue, carefully, without entirely separating the jaws from
the head. Since the line of division between the free, promi-
nent teeth and the undeveloped ones, buried in a chitinous
band, was always distinct, the method was adopted of counting
them separately, and of including in each set all that showed
well-formed outlines when held up against a strong light and
viewed through a lens from the dorsal side. Those connected
by the chitinous band were called indefinite teeth, the rest
the definite teeth. Totals were found by adding.

2. Results.

I. TABLE OF DISTRIBUTION OF FREQUENCIES.

9 10 ii 12 13 14

95 116 100 39 7

27 IO2 114 89 46 10

Classes i

2

3

4

5

6

7

8

L. Def. . .

5

3°

88

128

IO2

41

6

L. Indef. .

12

3°

93

146

93

21

4

L. Total .

3

12

28

R. Def. . . i

4

37-

94

126

86

41

1 1

R. Indef. . 2

7

30

80

149

97

28

7

R. Total .

i

2

8

27

No. 3.] VARIATION IN THE TEETH OF NEREIS.

From the table of frequencies it will be seen that the num-
ber of definite teeth varies from I to 8, the indefinite from
i to 9, the total numbers from 5 to 14.

The following constants were obtained.

II. TABLE OF CONSTANTS.

LEFT DBF.

LEFT INDEF.

LEFT TOTAL.

RIGHT 1 h i .

KK.HT INDEF.

RlGHTToTAL.

«

400

400

400

400

400

400

M

5

5

10

5

5

IO

A

5 098 ± 0.040

4.905 ± 0.039

10.00 ± 0.044

5.043 ±0.043

5.013 ± 0.040

10.055 — °-°45

a

1.193 ± 0.028

1.179 ± 0.0279

1.306 ±0.031

1.267 — 0.030

i .191 ± 0.028

1.339 =*= 0.032

F

+ 0.7025

— 0.66 1 4-S

+ 0.0599

-i- 0.4339

— 0.8617

-0-53I4

Type

I

IV

I

I

IV

IV

Skewness

— 0.0369

— 0.0439

— 0.1341

+ 0.0153

— 0.0868

— 0.0509

Comparison of these numerical results suggests the follow-
ing conclusions :

The mode, 5, is the same for the definite and indefinite on
both sides, and for the total on each side it is 10. The aver-
ages also show little difference between the right and left jaws.
From the variability, however, as indicated by the standard
deviation, it appears that the number of teeth of the right jaw
is slightly more variable than that of the left, cr being 0.074
greater on the right side for the definite teeth, o.on (which
is less than the probable error) for the indefinite teeth, and
0.033 greater for the right total. The highest degree of varia-
tion is shown by the total number of teeth on the right side,
and the least variation is shown by the number of indefinite
teeth.

The form of the distribution curve of definite teeth on both
jaws falls under Pearson's Type I, with the peculiar result of
a slight negative skewness for the left side, and a positive
skewness, although a very slight one, +0.0153, for the right.
This indicates a tendency to the production of fewer definite
teeth than normal on the left side, and a faint tendency
towards a greater number on the right side. The distribution
curve of indefinite teeth on both sides is of Type IV, with a
negative skewness, —0.043, on tne l6^ side, anc^ twice this,
— 0.086, on the right ; i.e., there is on the right side a greater

132 HEFFERAN. [VOL. II.

tendency than on the left towards the production of few indefi-
nite teeth. On the right side, then, it is clear that the prob-
ability of a large number of definite teeth is associated with
that of a small number of indefinite teeth. The same thing is
shown for the left side, for although definite and indefinite
teeth both show negative skewness, the negativeness is much
greater in the indefinite than in the definite teeth. Therefore,
relatively, the skewness of the indefinite and the definite teeth
may be said to be, here also, of opposite sense. This agrees
with results shown in the correlation table, to be noted later.

A peculiar result is obtained in regard to the distribution
curve of the total number of teeth. The left total falls into a
curve of Type I, while the right total is of Type IV. The
negative skewness of the latter is — 0.050, while that of the
former is about two and one-half times as much. The table
of frequencies shows that the right total includes two classes
more, one at each end of the series, than the left total. There
is one individual in each of these two classes. It seemed
probable, by inspection of the calculation, that the critical
function, F, which was negative 0.5314, might be made posi-
tive by dropping these two extreme individuals, thus giving a
curve of Type I. I found this to be the case, and obtained
for F the value +0.210; but I found further that Type I
might be obtained by dropping only the individual of Class 5,
making F +0.0389. The skewness in this case was very
slight, only -0.00706.

In order to determine which was the closer fit of the ob-
served curve to the theoretical curve in the two types, I cal-
culated the theoretical curves from the observed data with the
following result.

TYPE IV.1

n = 400 d = 0.06822 M = 10.055

s = 25.66 m= 13.830 y0 = 96.34

a = 6.5798 ff -- 1.3386 zero ordinate = 9.1114 (M-md)

v = 3.6796 0 = 8° 9' 7" tan e = x/a

y = y0 (cos. 0)2m e-i-0

1 For the methods of calculating the results given in the following tables, see
Davenport, '99, pp. 20, 23, 24.

Xo. 3.] VARIATION IN THE TEKI'll OF NEREIS.

Polygon of observed frequency.

Polygon of theoretical frequency, Type IV.

Polygon of theoretical frequency, Type I.

HEFFERA1V. [VOL. II.

A. Calculated by Duncker's method, = 3.68%.

V-M f y 5i 5z

- 4.1114 i o-1 + °-87
-3.1114 2 1.2 +0.8
-2. 1 1 14 8 7-9 +0-1

- 1. 1 1 14 27 35-3 -8-3 - 0-09

- 0.1114 102 90.5 + 1.5 - 4-82
+ 0.8886 114 122.9 - 8.9 - 5.01
+ 1.8886 89 89.9 -0.98

+ 2.8886 46 35-5 + 7-43 - 0.85

+ 3.8886 10 10.8 -0.89 -0.79

+ 4.8886 i 2.3 - 1.36

400 399-7 4I-°5 11.56

TYPE I.

n = 399 = SOS-S6 b == 46.165

at = 21.663 1111=142.426 ff - 1.316

32 = 24.502 m2 = 161.134 d= 0.00929

y0 = 120.68 (calculated from approximate formula). M - d = 10.066

y = y ('

X

a\

)m2
(12

A, Calculated by

Duncker's method, = 3.58%.

V-M

/

y

8,

52

- 4.066

2

0.9

+ 1.07

- 3.066

8

7.8

+ 0.2

- 2.066

27

35-2

-8.24

- 0.19

- i. 066

102

87.2

+ 14-84

-5.29

- 0.066

114

121.5

- 7-46

-4.96

+ 0.934

89

93-i

- 4.08

+ t-934

46

41.1

+ 5.06

- 2.25

+ 2.934

10

10.3

- 0.3

- 0.29

+ 3-934

I

1.4

- 0.42

399 398-5 41.60 12.98

Since it is considered a sufficient agreement between obser-
vation and calculation when A < = %, which in this case is

v«

5 '/ , it is evident that these values show a moderate degree of
closeness of fit of the two curves. The difference between the
two values of A, o.io, is so small that we may conclude that it is
practically immaterial under which type this curve should fall.

No. 3] VARIATION IN THE TEETH OF NEREIS. 135

The fact that by dropping one individual from a series a
curve may be thrown from Type IV to Type I and may be
made to fit equally well in either case, raises a serious ques-
tion as to the biological importance of the distinction between
Pearson's Type I and Type IV. Pearson ('95) himself says :
" It seems very possible that discreteness rather than continu-
ity is characteristic of the ultimate elements of variation ; in
other words, if we replaced the curve by a discrete series of
points, we should find a limited range. It is the analytical
transition from this series to a closely fitting curve which
replaces the limited by an unlimited range. Exactly the same
transition occurs when we pass from symmetrical point bino-
mial to normal curve. Thus while Type I marks an absolutely
limited range, Type IV does not necessarily mean that the
range is actually unlimited."

It appears from the results obtained in the two calculations
given above that even less value can be placed upon any dis-
tinction between Type I and Type IV than is suggested by
Pearson. The difference of one individual actually causes,
here, the transition from one type to the other, the individual
being at the extreme of the series.

3. Correlation.

In the table on the following page every possible combina-
tion of teeth for the two sides is given, together with the actual
number of specimens for each combination of definite, indefinite,
and total teeth.

From this series of combinations the following results were
obtained for the coefficient of correlation. The calculations
were made by Pearson's method and checked by the briefer
method of Duncker.

Correlation between Right and Left Definite Teeth, r = + o.688±o.oi36

" " Indefinite " r= + 0.725 ±0.01 21

" " '• " " Total " r= + o.82o±o.oo8i

" Right Definite and Left Indefinite, r - - o.424±o.o23i

Bearing in mind that an index of i signifies a perfect corre-
lation, and that o indicates an entire lack of it, we see that the

36

HEFFERAN.

[VOL. II.

III. CORRELATION TABLE.

Teeth.

Speci-
mens
with

defi-
nite
teeth.

Speci-
mens
with
indef-
inite
teeth.

Teeth.

Speci-
mens
with
defi-
nite
teeth.

Speci-
mens
with
indef-
inite
teeth.

Teeth.

Speci-
mens
with
total
teeth.

Teeth.

Speci-
mens
with
total
teeth.

R.

L.

R.

L.

R.

L.

R.

L.

I

2

2

5

5

67

79

5

6

I

9

I

1

4

I

5

6

26

20

6

6

2

o

20

2

->

-->

5

5

7

6

2

7

7

7

i

60

2 3

I

2

5

8

I

7

8

i

*>

7

2

4

I

6

3

2

8

7

4

o

i

3

2

2

4

6

4

8

5

8

8

13

2

9

2

3 3

14

12

6

5

17

32

8

9

8

2

o

5

3

4

14

I I

6

6

49

54

8

10

2

2

i

14

3

5

6

2

6

7

9

7

9

8

I I

->

2

23

3

6

I

7

4

i

9

9

61

-7

13

2

3

7

I

7

5

4

5

9

10

25

3

I 2

6

4

2

I

I

7

6

13

15

9

1 1

5

13

13

4

4

3

7

9

7

7

21

8

10

7

i

14

12

i

4

4

4i

39

7

8

2

10

8

3

4

5

36

27

7

9

i

10

9

23

4

6

9

4

8

6

3

10

10

64

5

3

6

8

8

7

5

3

10

1 1

21

5

4

22

38

8

8

3

3

10

12

2

400

400

400

degree of correlation between the right and left sides is, on
the whole, rather high. It seemed at first a somewhat unex-
pected result that the correlation in the variability of what I
have called the indefinite teeth should be higher than in that
of the definite teeth. If the correlation had been perfect it
would have meant that those causes which produced a devia-
tion from the mean in the left sets acted in the same degree
on the right sets also. Is it possible then that different causes
may have acted or that the same cause may have been effect-
ive in different degrees in producing the variability in the
definite and the indefinite teeth ?

This question drew my attention more closely to a fact
noticed only incidentally in counting the teeth, i.e., in case of
animals having dark, heavy jaws, evidently older animals, the
definite teeth were fewer in number than in case of small,
young individuals. In the older jaws the teeth began farther
from the tip, leaving a smooth point, while the younger, more

No. 3-J VARIATION IN THE TEETH OF NEREIS. 137

delicate jaws were often finely denticulated to the extremity.
The serratures of the older jaws appeared to be worn off by
use. In order to determine whether or no the correlation
actually existed between the size or age of an animal and the
number of definite teeth, I made the comparison for one hun-
dred individuals, taking as an indication of size, and hence
roughly of age, the head length in millimeters. This was meas-
ured from the anterior edge of the first ring to the extremities
of the two apical feelers. The result was a negative correlation,
although a rather small one, —0.128. It is probable, then, that
age does come in as a factor in the production of a small number
of teeth, and that this decrease is due to wear. It is possible
also that the wear may be heavier upon one jaw than upon the
other, thus accounting for the comparatively slightly lower
degree of correlation between the definite teeth than between
the indefinite teeth. For wearing does not act at all directly
on the indefinite teeth, since they do not emerge from the
chitinous covering, and are often sunk in the tissue of the pro-
boscis. It would be interesting to know in what manner the
jaws are carried and work upon each other during the life of
the animal, for a certain habit of crossing them might also
account for the peculiar differences in skewness of the curves
of the right and left teeth noted in the discussion of con-
stants. The smallness of the negative index of correlation
between the age of an animal and the number of definite teeth
shows that age does not play a very important part in causing
variation.

An attempt to correlate the number of definite and the num-
ber of indefinite teeth on the right jaw resulted in a negative
index of correlation, —0.424. This fact indicates an inverse
relation between the definite and indefinite teeth on the same
jaw; that is, a jaw with a small number of definite teeth will
probably have a comparatively large number of indefinite teeth,
and inversely. It may be that indefinite teeth continue to be
laid down at the base of the jaw during the life of the animal,
in which case the number would tend to be greater with age,
while, as we have seen, the number of definite teeth is
smaller.

j^g HEFFERAX. [VOL. II.

o

4. Relation of Individual and Specific Variation.

Out of fifty different species of Nereis which I found de-
scribed by various authors, the number of teeth was stated for
forty-seven. The numbers ranged from o in two cases, in
which the edentalous condition of the jaws was an important
specific character, to 20, given by Audouin and Milne-Edwards
('29) for N. fucata. The number of teeth of TV7", fucata is
given by Ehlers ('68), however, as 7, by Johnston ('65) as 5 to
10, and in the Challenger ('85) reports as 14 to 16. It would
be difficult here, as in the case of a few others, to decide
which observer came nearest to the modal condition of the
species. It is also impossible to tell whether they counted
the total number of teeth including those covered by a chiti-
nous band, or whether they referred only to the prominent
definite teeth. Ehlers makes the distinction only in N. vircns,
where he gives the definite teeth as 5 to 6, total as 10. St.
Joseph ('88, '98) notes in description of N. divcrsicolor that of
8 teeth 2 are indefinite, and in N. floridana that of 9 teeth
the lower 4 are buried in a clear translucent covering. For
these two species, respectively, Ehlers has given 8 and 9 teeth,
evidently counting both definite and indefinite.

After attempting various methods of striking averages of
the statements made by different authors I finally decided to
use Ehlers' s numbers alone as most reliable, adding a few of
those given by St. Joseph in which there was less doubt that
the total number of teeth had been counted. Sedation of
twenty-two species gave the following results, total teeth.

Classes ...5 6 7 8 9 10 n 12
Frequencies 115263 2 2

CONSTANTS.

A == 8.727 F : + 1.354

a- - 1.838 curve - Type I

/3i == 0.000090 s = 5.862

/32 =- 2.323 skewness = + 0.00966

Any conclusions which can be drawn from these results are
necessarily of doubtful value. It will be seen that the mean

No. 3-] VARIATION IX THE TEETH OF . \EKE1S. 139

for the number of teeth in twenty-two species is lower than
the mean of total teeth in the one species which I have
described. The skewness of the curve instead of being nega-
tive is positive, although it is exceeding small. Had it been
negative, as I had thought it might be, it would have indicated
that in the species of the genus, as well as in the individuals
of the species, there had been a movement in the direction of a
smaller number of teeth, either from an excessive production
of individuals and species having few teeth, or from selective
annihilation of those having many teeth. The opposed positive
skewness is so small that it may mean little in regard to the
species, and particularly since the numbers are small and the
method of counting so doubtful no stress can be laid upon it.

5. Abnormalities.

Differences between the two jaws of the same animal in the
definite, indefinite, or total number of teeth were of common
occurrence. The accompanying drawings are intended to
show some irregularities of this kind. In Fig. i the right jaw
has four large definite teeth and five below which do not
emerge from the surrounding chitinous layer ; the left jaw has
only slight crenulations corresponding to five definite teeth,
although it has the same number of indefinite teeth as the
right side. Fig. 2 shows on the right jaw three definite
teeth, the edge above and distal to them having three very
slight elevations ; the opposite jaw ends in a long point with a
perfectly smooth edge and has only two large definite teeth
below. There is the same number of indefinite teeth on both
sides. Figs. 3 and 4 show variations of the same kind, the
numbers of both definite and indefinite teeth being different
for the two jaws.

So far the drawings have been made from old animals in
which the jaws are hard, strong, and very dark in color. It is
probable from the appearance of the jaws that the difference
in the number of definite teeth is due largely to the wearing,
on one side or the other, of the distal teeth. Fig. 5 shows a
common irregularity of equal totals with slight differences in

140

HEFFERAN,

[VOL. II.

the combinations of definite and indefinite teeth, the left jaw
having 8 to 4, the right 6 to 6. Figs. 5 and 6 are from small,
young animals, and the jaws are seen to be more slender with
numerous fine teeth, 13 on the right and 12 on the left jaw.

The specimen drawn in Fig. 7 was interesting in regard to
the indefinite teeth. The left jaw presented the usual appear-

FlG. I.

FIG. 2.

FIG. 3.

FIG. 4.

FIG. 5.

FIG. 6.

FIG. 7. FIG. 8. FIG. g.

FIGS. i-8. — Variation and abnormalities in teeth on opposite jaws.
FIG. 9. — Abnormal segment.

ance of six indefinite teeth placed fairly regularly and six
definite teeth. On the right jaw the point was worn smooth,
leaving only four definite teeth, while below three normal
indefinite teeth was a series of five small ones placed very
close together instead of three as on the opposite jaw. This
may indicate a tendency towards regulation by the production

No. 3] VARIATION I. .V THE TEETH OF NEREIS. 141

of an excess of teeth at the base of a jaw on which some of
the extreme teeth had been lost, but I found no other indica-
tion of such regulation. Another individual presented a par-
tial right jaw, Fig. 8, which was a stump of about half the
length of the left jaw. The normal jaw was dark brown,-
almost black, while the stump was light straw color character-
istic of a young jaw or of the very base or imbedded part of
an old one. The color indicated new growth or regeneration,
in which case one would expect to find a production of small
indefinite teeth crowded at the base, as in the specimen of
Fig. 7, if that could be interpreted as a regenerative process.
On the contrary, the stump had exactly the same number of
teeth similarly disposed as the part of the opposite jaw which
corresponded to it. It may have been, then, only the rounded
stump of a broken jaw, although this explanation does not
account for the peculiar color.

Abnormalities in other parts of the animals were looked for
only incidentally. I found no cases of double pairs of caudal
cirri, but all of the worms were not examined for this not
unusual abnormality, since the posterior parts of many of
them were not preserved.

Fig. 9 shows a case of an abnormal segment. The fifteenth
segment extended only a little more than halfway across
towards the left side of the body, leaving one broad segment
on the left side in place of two, and a partially double para-
pod. The axis of the body was bent at that point, compen-
sation being made gradually by the greater width on the left
side of the segments immediately preceding and following.

6. Summary.

The results of this study may now be summed up as follows :

(1) The typical condition for the total number of teeth of
N. limbata of Cold Spring Harbor, 1899, ig a curve of either
Type I or Type IV, with a slight skewness in a negative direc-
tion from the mode, 10.

(2) In case of the calculation of the right total teeth, a
transition from a curve of Type IV to an equally serviceable

142 HEFFERAN. [VOL. II.

one of Type I could be made by discarding one extreme indi-
vidual out of four hundred.

(3) The number of teeth on the right jaw appears to be
slightly more variable than that on the left.

(4) The degree of correlation between the two jaws is, on
the whole, rather high, 0.820. Correlation is closer between
the indefinite than between the definite teeth. An inverse
relation exists between the number of definite and the number
of indefinite teeth on the same jaw, and also one between the
number of definite teeth and the age of an animal.

(5) The class range of teeth as given by the different species
of the genus Nereis has a close agreement with the class range
of N. limbata. Hence this one species offers the material for
the modal condition of all species of the genus.

(6) The results of observations of many specimens showing
irregularities in teeth point to the conclusion that a process of
wearing away of the definite teeth takes place in mature ani-
mals, and therefore that age comes in to help produce small
number of teeth, but is not a large factor in causing variation.

Only one author, St. Joseph, makes note of a difference
between young and old specimens, the young having the
greater number of teeth. Thus the statements made in
regard to the number in many species in which only one ani-
mal or at most very few specimens were seen and described
by their discoverers, are of little value as criterions of the spe-
cific condition.

In conclusion, I wish to express my thanks to Professor
Charles B. Davenport, who not only generously furnished the
material for this investigation, but by his oversight and advice
greatly facilitated the progress of the work.

No. 3-] VARIATION IN THE TEETH OF NEREIS. 143

BIBLIOGRAPHY.

1. AUDOUIN ET MILNE-EDWARDS ('29). " Classification des Annelides, et

Description de celles qui habitant les cotes de la France." Ann. des
Sci. Nat. ie serie, tome xxix, pp. 195-221.

2. DAVENPORT, C. B. ('99). Statistical Methods, with Special Reference

to Biological Variation. New York, 1899. 148 pp.

3. EHLERS, ERNST ('68). Die Borstenwiirmer. Zweite Abtheil. Leipzig,

1868.

4. JOHNSTON, GEORGE ('65). A Catalogue of the British Non-Parasitical

Worms in the Collection of the British Museum. London.

5. M'INTOSH, WILLIAM C. ('85). " Report on the Annelida Polychaeta

Collected by H. M. S. Challenger." Challenger Reports. Vol. xii,
pp. 210-230.

6. PEARSON, K. ('95). " Skew Variation on Homogeneous Material."

Phil. Trans. Roy. Soc. London. Vol. 186 A, p. 389.

7. ST. JOSEPH, M. LE BARON ('88, '98). " Anne'lides Polychetes des cotes

de Dinard." Ann. des Sci. ATat., Zool. et Paleon. 7e sdrie, tome v,
pp. 266-269. Se serie, tome v, pp. 288-304.

Volume //.] January, 1901. [A7"^. ./

BIOLOGICAL BULLETIN.

THE CENTROSOME IN THE MATURATION AND
FERTILIZATION OF BULLA SOLITARIA.

MARTIN SMALLWOOD.

THE material upon which the following observations were
made was collected at Woods Holl during the seasons of
1898-1900. The greater part of the work was done at the
Marine Biological Laboratory under the direction of Dr. E. G.
Conklin, and I take this opportunity of thanking him for his
many valuable suggestions. I also wish to acknowledge my
indebtedness to Dr. C. O. Whitman and Dr. C. W. Hargitt.

A fuller account of my observations both upon the subject
of this paper and the cell lineage of Bulla, with a discussion of
the pertinent literature, will be published later.

The sketches illustrating the mitotic changes in maturation
were drawn from sections stained with Heidenhain's iron-
alum followed by an aqueous solution of Bordeaux. In order
to differentiate the sperm, it was necessary to use Conklin' s
mixture of haematoxylin and picric acid. The sperm, there-
fore, has been drawn from corresponding stages and inserted
into these figures.

In the interpretation and nomenclature of the centrosome
and sphere I have followed in the main Van Beneden. The
term " centrosome " will be applied to the body which occurs at
the pole of the spindle, etc., when that body has become differ-
entiated into a central corpuscle, surrounded by a clear area,
the medullary zone bounded by a definite wall. The body
occurring at the center of the aster is the central corpuscle.

MS

146 S. \rALLWOOD. [VOL. II.

The central corpuscle becomes the centrosome. Sections of
the ovotestis before copulation show the unfertilized egg lying
free in the follicles of the hermaphroditic gland. The large
germinal vesicle lies in the center of the egg ; it contains a
large vacuolated nucleolus, also basichromatin and oxychroma-
tin granules. The deutoplasmic spheres are equally distributed
in the cytoplasm and conceal its structure. I have not been able
to discover any evidence of a central corpuscle or centrosome
in the egg before mitosis begins. By the time the eggs are
laid the first polar spindle is in the end of the prophase. In
order to secure the earlier stages, a large number of animals
were collected and killed as soon as they began to lay. The
first polar spindle begins to form as the animals begin to lay.
Sections 'of the ovotestis taken from animals killed while they
were laying revealed the fact that every mature egg had
already passed through the early prophase of the first spindle ;
even those eggs in the most distant follicles, where it is prob-

able that the sperms from the receptaculum
seminalis had not penetrated ; I have found
the sperms in the anterior part of the herma-
phroditic duct, but not extending back to
any considerable distance. The earliest
stage thus far discovered had two well-
FI<;. i. -Taken from the formed central corpuscles and a definite cen-

tral sPittdle connecting each, which passed

rounded by the cortical through the germinal vesicle, the walls of

zone of the sphere. The , . , 1-1 • 'i_1 T-t

central spindle is weii which are plainly visible. The ring-shaped
formed and the chromo- chromosomes have begun to form from the

somes are forming into

the first equatorial plate, meshwork of linin and chromatin. The re-

Walls of the germinal -\ . • c . i i i i

vesicle still present, auction of the chromosomes has not been
sperm entering at vege- worked out in detail. These ring-shaped

tal pole.

chromosomes gradually take a deeper stain
and come to lie in the equatorial plate of the first polar
spindle (Fig. i). In a cross-section of the equatorial plate
I was able to count sixteen distinct chromosomes. There
is a distinct cortical zone surrounding the central corpuscle.
The astral rays pass through this clear area and extend
to the central corpuscle. At this stage I have been able to

i\o. 4.] BULLA SOLITARIA. 147

trace them nearly to the egg membrane. From now on,
there does not appear to be any appreciable change in the
mitotic figure, until after the egg has been laid, when it begins
to migrate to the periphery of the egg, where it assumes a
radial position. There is no difference in the character of the
two poles of the spindle. During this movement of the spin-
dle the chromosomes pass into the metaphase, and the cen-
trosome becomes differentiated into a central corpuscle and
a medullary zone which is limited by the walls of the old
central corpuscle. This is the first time that we have a struc-
ture to which we can apply the term "centrosome" in the sense
that I purpose to use the term. The cortical zone has enlarged
and become much fainter. The chromosomes do not divide at
once ; the activity is centered in the centrosomes. While the
chromosomes are still in the equatorial plate, the central cor-
puscle in each centrosome divides, having the dumbbell form.
The centrosome rapidly increases in size, the periphery is
limited by a definite line, which gradually becomes thinner.
The medullary zone does not take a plasma stain, as it did in
the previous stage (Fig. 9). The centrosome now begins to
assume an elliptical form and at the same time to rotate.
This rotation continues until the elongated centrosome, which
encloses the second polar spindle, lies radially and in the same
position that the first polar spindle did. The central cor-
puscles, connected by a central spindle, are so influenced by
this elongation of the walls of the centrosome that they come
to lie near the ends --at the foci of the ellipse. As the outer
pole of the second polar spindle nears the periphery of the
egg, the rays extend to the chromosomes, and they are pulled
into the spindle to form the equatorial plate of the second
polar spindle. It will be seen that in the main my results
corroborate those of MacFarland,1 Lillie,2 and Conklin.3

The changes which take place in the centrosome during this
stage are very interesting. The centrosome is so large and
comes out with such perfect clearness that I have been able to

1 " Cellulare Studien an Mollusken-Eiern," /.ool.Jahrb. 1897.

- " Centrosome and Sphere in the Egg of Unio," Zool. Bull. Vol. i, No. 6. 1898.

3 Science, March, 1898.

148 SMALLIVOOD. [VOL. II.

follow the details very carefully. The spindle is developed
when the central corpuscles separate. At this time a very
faint line can be seen connecting the new corpuscles; as
the distance between them increases, the line becomes more
distinct, until a central spindle can be clearly distinguished.
In the mean time the line limiting the centrosome has become
broken into pieces, which gradually become smaller and smaller
until they cannot be distinguished from the granules of the
cytoplasm. While these changes have been taking place, this
broken line has served to mark the outer limit of the medullary
zone. The old medullary zone has disappeared, and between

the central corpuscle and the broken wall of
the centrosome we have a new medullary
zone, which is the cortical zone of the
second polar spindle. The process is as
follows : as the walls of the centrosome begin
to break down, an area next to the central
corpuscles and at each end of the centro-
FIG. - Meuphase of some appears ; this begins to take a plasma

first maturation spindle. . /I—- \ -T-I

TWO central corpuscles stain (Fig. 3). The area gradually surrounds
at each pole. Cortical ^e central corpuscle and all of the space at

zone limited by a dotted

outline. The medullary thecudof the spindle between the central

zone does not take a i i ., i nr^i

plasma stain at this corpuscle and the wall of the centrosome.
stage, sperm head solid jn the mean time the central corpuscle has

and elliptical.

increased in size and is to become the cen-
trosome of the second polar spindle ; it ultimately becomes
differentiated into a central corpuscle and a medullary zone.

Fig. 2 shows two central corpuscles in the centrosome at
each pole of the spindle. Linville 1 shows a similar but not
identical stage. The centrosome of the outer pole of the
spindle does not enlarge more than is shown in the figure ; as
it reaches the surface of the egg, it breaks down and the cen-
tral corpuscles form the division centers in the first polar
body.

There is no telaphase in the first maturation spindle. The
nearest approach to such a stage is shown in Fig. 3, where the

1 " Maturation and Fertilization in Pulmonate Gasteropods," Bull. Mies. Comp
Zoo!., I/arrant. Vol. xxxv, No. 8.

No. 4.]

BL'LLA SOLITARIA.

149

chromosomes have become partly hollow vesicles. A few of
the interzonal fibers show at this stage, but they are very faint.
In the metamorphoses of the centrosome its attachment to
the astral rays is plainly evident ; the old
rays can sometimes be seen in a stage
younger than the one shown in Fig. 11,
when the new rays have already begun to
form and are attached to the central corpus-
cle. I believe that the rays of the first polar
spindle disappear and that the rays of the

FIG. •?. — Prophase second , n1 /

second spindle rise de now.

The metakinesis of the second polar spin-

The chromo-

m.uuration spindle. Wall
of centrosome broken in
pieces. Central corpuscles .

connected by a spindle, die takes place very rapidly.

New cortical zone form- SQmQS elon~ate divide transversely, and as

ing. Chromosomes are J

partly hoiiow. A few in- they move toward the poles, they assume a

terzonal fibers are present. . . ,

roundish form, and change into vesicular
bodies which fuse to form the female pronucleus. During the
time when they are fusing, the rays can be traced directly
into the areas immediately surrounding them. In the stage of
anaphase as represented in Fig. 5, the centrosome is evident,
although it does not stain as deeply as in Fig. 4. Immediately
after this stage the centrosome disappears and the cortical
zone enlarges and completely surrounds the
female pronucleus ; later both male and
female pronuclei come to lie in this clear
area. A single centrosome passes off with
the second polar body, which is much smaller
than the one given off in the first polar body
(Fig. 4).

„, . . . . ., FIG. 4. — Anaphase second

The eggs of this species are not especially maturation spindie. Cen.
favorable for a study of the problem of f er- trosome at each p°le with

J a single central corpuscle.

tilization. During all of the earlier stages, The medullary zone takes

a plasma stain.

During all of the earlier stages,
the sperm head lies completely surrounded by
deutoplasmic spheres. I have not been able to make out any
continuous clear area about the sperm head during its prog-
ress through the egg. In one instance there was a definite
clear area about the sperm nucleus after it had nearly ap-
proached the female pronucleus, otherwise it was unattended

SMALLWOOD. [VOL. II.

by anything that corresponds to the " Hellerhof " of MacFar-
land and others. The sperm enters at the vegetal pole, though
not in any definite place. The tail is lost before the sperm
enters the egg membrane (Fig. i). The head is a solid body
having a distinct angle in the middle. If there is a middle-
piece, it is practically indistinguishable. The only indication
that I have found of such a body is that on one end of the
sperm head sometimes one finds a narrow band that stains a
little denser than the rest of the head.

The sperm head becomes top-shaped as it begins to migrate
toward the animal pole with the point' leading. The head

becomes elliptical (Fig. 3) and stains very
black. It remains in this solid form for
some time, while the first polar spindle is
passing from the metaphase until the ana-
phase of the second polar spindle. During
the anaphase of the second maturation, the
solid sperm head becomes vesicular, and for
FU;. 5. - Late anaphase a very short time is accompanied by astral

of second maturation

spindle, centrosome rays. I have not been able to discover a

still present. Cortical j rpusde in connection with the

zone enlarging. The

sperm is composed of aster, nor have I ever seen an amphiaster.

three vesicles and ac- n

bv astral At this same time secondary asters usually

ravs.

appear in the egg, which are smaller than the
sperm aster. As the chromosomes of the second polar spindle
begin to assume the vesicular form, the sperm aster disappears,
and the sperm, consisting of one or more vesicles, rapidly ap-
proaches the inner pole of the second polar spindle. When
the sperm consists of more than one vesicle, these fuse into
one when the aster disappears. While the vesicular sperm is
shifting its position it does not increase in size to any notice-
able extent, but as soon as it comes near the female pronucleus,
which now consists of but three or four vesicles, it rapidly
increases in size until it is about twice as large as it was when
migrating toward the animal pole. From the time that the
sperm head enters the egg until it comes to lie in contact
with the female pronucleus (Fig. 6), it is not attended, so far
as I have observed, by any body which might be taken for a
central corpuscle or a centrosome.

No. 4.]

BULL A SOLITARIA.

FIG. 6. — The female pro-

inner pole of the second
maturation spindle. A
few astral rays are pres-
ent. Zwischenkorper
of second polar body
present.

The structure of the two pronuclei when they have come
together (Fig. 6) is the same. The male pronucleus is usu-
ally regular in outline and slightly smaller. The irregularities

in the outline of the female pronucleus often
persist until the central corpuscles of the
first cleavage appear. The chromatin stains
very slightly and is connected by delicate
linin threads. The changes through which
the chromatin passes before the equatorial
plate is formed may be described under
three stages. First, the chromatin rapidly
nucleus is irregular in increases in staining power, forming a dense

outline, surrounded by ....

the cortical zone of the reticulum, often containing stellate masses
of solid chromatin. Second, the chromatin
begins to assume a definite form. The most
conspicuous shape is the stage where the
masses of chromatin have begun to break up
into rings but are still attached to one another. The chromo-

O

somes have not yet become hollow. They stain uniformly.
Third, the chromosomes have broken apart from each other,
and have become hollow, round bodies. At first there is a
delicate meshwork connecting them (Fig. 7), but this is soon
lost and each pronucleus is filled with ring-
shaped chromosomes. While the chromatin
is undergoing the last two changes, the
central corpuscles (the so-called cleavage
centrosomes) of the first cleavage spindle
make their first appearance. I have found
them in a much earlier stage than the one

r , . . , -,, FIG. 7. — Origin of cleav-

figured, but in each case there was no con- age centrosomes. One

nection between them ; but these corpuscles

with their rays have a definite relation with

the pronuclei, that is to say, each pronucleus

has an aster and central corpuscle with

a faint cortical zone. My observations upon the cleavage

centrosomes would tend toward the position, first, that they

arise de novo ; and, second, that one arises in connection with

each pronucleus.

in connection with each
pronucleus. The cen-
tral corpuscles are
surrounded by a faint
cortical zone.

152 SMALLWOOD. [VOL. II.

Metamorphosis of tJie Centrosome in Maturation.

The definiteness and clearness with which the several
changes in the centrosome appear in Bulla make these changes
the most important of the various stages in maturation and
fertilization. In describing the changes of the centrosome,
tinder various stages, I have no theoretical points in considera-
tion. While the stages figured are clearly differentiated, still
there are intermediate stages which graduate imperceptibly
into one another.

In the earliest prophase that I found the central corpuscle
was a large solid mass (Fig. 8). Surrounding the central cor-
puscle there was a conspicuous area, the cortical zone, which was
sharply differentiated from the cytoplasm. The rays are not
lost in the cortical zone, as MacFarland has shown for Diaulula,
but extend to the central corpuscles, as Lillie has shown for
Unio, and Linville for Limnaea. However, I do not find a row
of microsomes, as in Unio, limiting the sphere, nor is the bound-
ary formed by the fusing of the astral rays, as in Limnaea.

Second stage (Fig. 9). The central corpuscle has become
clearly differentiated into a centrosome. It reacts to stain in
a very different manner from what it did in a previous stage.
There is now a medullary zone which takes on a plasma stain
and is limited by a distinct line. The small dark body in the
center is the new central corpuscle. The cortical zone has
increased in size and is less easily distinguished from the sur-
rounding cytoplasm. The line marking the periphery of the
centrosome is the limiting wall of the enlarged central corpus-
cle of the previous stage.

Third stage (Fig. 10). The centrosome has increased in size.
The line at the periphery is definite and whole. The central
corpuscle of the previous stage has divided into two central
corpuscles, which are connected from the first by faint lines.
The medullary zone does not take a plasma stain. The corti-
cal zone has become very faint and soon disappears as a
distinguishable area in the cytoplasm.

Fourth stage (Figs. 3 and 1 1). The periphery of the centro-
some loses its continuity, and openings occur in the wall ; while

No. 4-] BULLA SOLITARIA. 153

these changes in the wall are taking place, the centrosome
becomes much enlarged and assumes an elliptical shape. Im-
mediately after these breaks appear, there is a small part of
the medullary area which takes a plasma stain. This area is
somewhat triangular in shape and occurs at the end of the
centrosome between the central corpuscle and the broken
periphery of the centrosome. New astral rays are formed
which extend to the central corpuscle. The old astral rays can
be seen disappearing at this stage in the cytoplasm.

Fifth stage (Fig. 12). The central corpuscle of the second
polar spindle has enlarged and is still ^differentiated. The
pieces of the periphery of the old centrosome have become

FIG. 8. FIG. 9. FIG. 10. FIG. n. FIG. 12. FIG. 13.

FIGS. 8-13. — The changes through which the central corpuscle and centrosome pass from the
prophase of the first maturation spindle to the metaphase of the second maturation spindle.
In each case the solid dark body is the central corpuscle. The granular area surrounding the
central corpuscle is the cortical zone. Figs, q, 10, and 13 show a complete centrosome,
having a central corpuscle and medullary zone.

smaller. There is now a distinct cortical zone around the cen-
tral corpuscle, which has been derived from the medullary
zone of the centrosome of the first polar spindle. This stage
is identical with the first one described, except that the rim of
the old centrosome is still present and the central corpuscle is
only about one-half as large.

The sixth stage (Fig. 13) shows the rim of the old centro-
some still present, but in smaller pieces which do not stain
as deeply as in the previous stage. The cortical zone has
enlarged and become fainter. The centrosome is composed of
a medullary zone and a central corpuscle.

Summary.

The central corpuscle of the first polar spindle becomes the
centrosome of the second polar spindle with a medullary zone
and a central corpuscle. The medullary zone of the centro-

154 SMALLU'OOD.

some of the first polar spindle gives rise to the cortical zone of
the second polar spindle. The central corpuscle of the cen-
trosome of the first polar spindle gives rise to the centrosome
of the second polar spindle. Thus we may say that the cen-
trosome of the first polar spindle in Bulla solitaria not only
gives rise to the centrosome and mitotic figure of the second
polar spindle, but to the cortical zone or outer sphere substance
surrounding each centrosome.

ALLEGHENY COLLEGE,
October, 1900.

CONTRIBUTIONS ON THE MORPHOLOGY OF
THE ACTINOZOA.

J. PLAYFAIR McMURRICM.
VI. HALCURIAS PILATUS AND ENDOCOPILACTIS.

IN 1892 Carlgren showed that certain Edwardsiae, whose
tentacles were more numerous than the mesenteries, had these
tentacles arranged on the hexactinian plan, their arrangement
in this presumably primitive group of the Actiniaria seeming
to foreshadow what is characteristic of the phylogenetically
later group. In other multitentaculate Edwardsiae he found
what seemed to be an octamerous arrangement combined to a
certain extent with hexamerism, but later studies ('99) con-
vinced him that the octamerism did not occur, and that in all
cases the hexamerous arrangement obtained.

In the mean time an important discovery had been made by
Faurot ('95) in studying Edwardsia beautempsi and E. adc-
ncnsis, the former of which possesses fourteen to sixteen ten-
tacles, while for the latter the number is stated to be fifteen to
sixteen. Sections through the column showed the eight mesen-
teries, which have long been supposed to be the only mesen-
teries developed in the Edwardsiae ; but in the uppermost
portions a number of additional very short and narrow mesen-
teries were found which in E. beautempsi were placed in such
a way as to make with the perfect mesenteries an arrangement
recalling what occurs in Go nactinia prolifera. Thus there were
eight pairs of mesenteries present in the upper part of the
column, two of which, the directives, were formed of two per-
fect mesenteries, four of one perfect and one imperfect mesen-
tery, and one of two imperfect mesenteries. In E. adenensis
the additional short mesenteries were arranged in pairs in each
interval between adjacent perfect mesenteries, except in the
endocoel of the directives, so that in this form the arrangement
differed somewhat from that typical for the hexactinians.

'35

l$6 McMURRICH. [VOL. II.

These observations show reason for believing that in the
Kdwardsiae there is an intimate relation between the number
of tentacles and that of the mesenteries, and that when there
are more than eight tentacles there is a strong probability that
a number of short mesenteries are also present in the upper
part of the column. It is a general rule in the Actininae that
the number of tentacles in the fully developed condition is
double that of the pairs of mesenteries or, in other words, that
there is a tentacle corresponding to each endocoel and each
exocoel, the number of tentacles being equal to the total num-
ber of mesenteries. Exceptions, due to a lack of development
of the full complement of tentacles, are of common occurrence,
in many cases probably owing to the specimens examined not
having reached their full development, though even in some
adults, apparently, the number of tentacles never reaches that
of the mesenteries, as is the case, for instance, in Pcachia
Jiastata, which, with twenty mesenteries, never has more than
twelve tentacles.

The rule may be better expressed by saying that tlic number
of the tentacles never exceeds that of the mesenteries, and when
an apparent exception to this occurs the presumption is that
closer examination will reveal the existence of small mesen-
teries limited to the upper part of the column and in sufficient
numbers to fulfill the requirements of the rule.

Acting on this supposition, I have made a further study of
the upper portion of the column of Halcurias filatns, a form
which I have already described as possessing twenty mesen-
teries and a number of tentacles considerably in excess of that
of the mesenteries, having been estimated in one specimen
('93) to be about seventy, and in another ('98) to be about
sixty. Sections showed, as I had expected them to do, the
presence of a number of short and narrow mesenteries in the
upper part of the column, the number of these pins the twenty
perfect mesenteries being equal to the total number of the
tentacles, which proved to be sixty-eight.

The sections also revealed, however, a peculiarity which I
had not expected, and which, as may be seen from Fig. i, con-
sisted in the short, narrow mesenteries being developed in the

No. 4.] MORPHOLOGY OF Till-: ACTI. \OZOA. 157

endocoelic spaces bounded by the perfect mesenteries. The
sections did not, unfortunately, cut the column perfectly trans-
versely, but the arrangement which obtained may be perceived
from the representation of the half of one section shown in Fig. i ,
and from the diagram (Fig. 2) which represents a reconstruction
from a perfect series of sections. On each side of the median
line of Fig. i is one of a pair of directives, that to the right
being cut at the level of the oral stoma, as is also another
mesentery in the right half of the figure. On each side of the

T

FIG. i. — Transverse section through the upper part of the column of Halcurias pilatjis.
T = tentacle . D — directive mesenteries .

directives is a perfect mesentery with its muscle pennon on
the same side as that of the adjacent directive, and on the left
side this mesentery is succeeded by one which evidently forms
with it a typical pair, though it may be noticed that the endo-
coel enclosed by this pair is broader than the adjacent exocoels.
On the right side, where, owing to the obliquity of the sec-
tions, the column is cut higher up, the bases of some of the
tentacles (T) being cut, the mesenteries of the first lateral pair
are widely separated, and between them three imperfect pairs
occur, which evidently represent two cycles. The succeeding

McMURRICH.

[VOL. II.

FIG. 2. — Diagram showing the arrangement of the mesenteries
and tentacles in Halcurias pilatus.

exocoel is much narrower than either of the adjacent endocoels
and contains no imperfect mesenteries, while in the next endo-
coel two pairs of mesenteries are seen on the right side and
one on the left. By following through the series of sections
it is readily seen that the arrangement found in the first lateral
endpcoel of the right side is repeated in all the others, except
in the cases of the endocoels enclosed by the directives, and

the condition repre-
sented diagrammat-
ically in Fig. 2 is
that which obtains.
Owing to the rel-
ative widths of the
endocoels and exo-
coels, and the pres-
ence of imperfect
mesenteries in the
former, the first impression one receives is that of a form with
a large number of directive mesenteries. That such an inter-
pretation of the conditions is erroneous is clearly shown, how-
ever, by reference to the mesenteries on either side of the true
directives. It is interesting to note that the development of
the imperfect mesenteries, which are plainly arranged in two
cycles, follows the hexactinian rule, the smaller pairs being
developed in the intervals between the larger pairs and the
adjacent perfect mesenteries. It may be added that my sec-
tions show the existence of a marginal stoma in each perfect
mesentery in addition to the oral stoma already mentioned.

From the description given above, it will be perceived that
the arrangement of the mesenteries in Halcurias pilatus is
identical with that described by Carlgren ('97) for a form
from the Chinese seas which he refers to the genus Endo-
coelactis and to a new family, the Endocoelactidae. The simi-
larity to Halcurias is by no means confined, however, to the
arrangement of the mesenteries, and there can be no question
but that the two forms must be referred to the same genus,
to which, notwithstanding the greater appropriateness of Carl-
gren's name, the prior term, Halcurias, must be applied. The

No. 4] MORPHOLOGY OF Till'. ACT1. \OZOA. 159

specific identity of Carlgren's form with //. pilatns seems
improbable; for, apart from the difference in the localities for
which the two have been obtained, the tentacles of the Chinese
form are longer apparently than those of H. pilatns, and to
judge from Carlgren's figures, the longitudinal musculature of
the tentacles is weaker and its mesogloeal processes coarser.
It seems preferable at present to regard them as distinct, and
since Carlgren, in his brief notice, has bestowed no specific
name on his Endocoelactis, I would suggest that it be named
Halcnrias Carlgrcni, as a slight recognition of the admirable
work which that author has accomplished on the morphology
of the Actiniaria.

An examination of the arrangement of the tentacles of
H. pilatns with reference to the mesenteries was made in the
series of transverse sections and also by an examination of the
disk, and the results obtained were essentially the same as
Carlgren's. I was not able, however, to distinguish any dif-
ference in the position of the tentacles over the endocoels
bounded by the imperfect mesenteries, though on theoretical
grounds it is probable that some difference does exist, and,
furthermore, the study of sections seemed to indicate that the
tentacles over the directive endocoels were situated a little
nearer the mouth than were the others represented as being in
the same cycle in Fig. 2 ; an examination of the disk failed,
however, to confirm this appearance.

As regards the systematic position of Halcnrias, a few re-
marks are in order. I at first ('93) assigned it to the family
Halcampidae, but later ('98) deemed it advisable to separate
it from that family and refer it to Hertwig's Antheomorphidae.
Carlgren in the mean time had established for his Endocoe-
lactis the family Endocoelactidae. There are apparently three
courses open for the disposal of the genus. It may be referred
to a family already existent, the definition of the family being
changed, if necessary, to accommodate it ; or it may be taken
as the type of a distinct family, as Carlgren has done ; or,
finally, it may be separated altogether from the Hexactiniae
and regarded as the type of a separate tribe.

It seems to me that this last procedure is quite unnecessary,

McMURRICH. [VOL. II.

and would probably be entirely out of harmony with the phylo-
genetic relationships of the genus. We have learned within
recent years how extensively nearly allied forms may differ,
and how great are the modifications which the hexactinian
type may undergo. The entire facies of Halcurias is that of
an hexactinian, and it may furthermore be pointed out that
instances of the occasional endocoelous development of mesen-
teries have been already recorded by G. Y. and A. F. Dixon
('89) in Bunodcs thai Ha and by H addon ('98) in Actinioides
dixoniana and A. papuensis.

If, then, the third possibility be excluded, Halcurias must
either be assigned to an existent family, the endocoelous
development of mesenteries being regarded as of minor im-
portance, or this feature may be considered of sufficient
importance to warrant the establishment of a separate family.
I have already indicated my belief that the peculiar mode of
development of the secondary and tertiary mesenteries is of
minor importance and see no more reason for separating Hal-
curias as the type of a new family than I do for separating an
octamerous sagartian, or one with a multiplicity of mouths and
many siphonoglyphs, from the rest of the members of that
family.

It remains then to consider what the forms may be with
which Halcurias may be associated. As Carlgren has remarked,
and as I have indicated by the position to which I have referred
it in previous papers, Halcurias occupies a position near the
base of the hexactinian stem. The small number of perfect
mesenteries, the occurrence of reproductive organs on all of
them, the absence of a distinct sphincter muscle, the simplicity
of the margin, are features which, when combined in one indi-
vidual, indicate for it a somewhat low position. Are there
other forms which present a similar combination of peculiari-
ties associated with the development of an adherent base?

Two forms suggest themselves in this connection, namely,
Go nactinia prolifera and ProtantJica simplex ; but both of these
present peculiarities which render their association with Hal-
curias inadvisable. They both have but eight perfect mesen-
teries, the remaining ones, eight in Gonactinia and about

No. 4.] MORPHOLOGY OF THE ACTINOZUA. l6l

eighty-eight in Protauthca,1 being imperfect, and the ciliated
lobes are lacking in their mesenterial filaments. On account
of these peculiarities it seems to me that these two forms
must be grouped together in a family, Gonactiniidae, as Carl-
gren ('93) has proposed, and Halcurias cannot be placed with
them. The family Gonactiniidae must, I believe, be placed
among the Hexactiniae, as indeed must all the forms which I
have included in the past in the order Protactiniae, as well as
those which Carlgren has referred to the Protantheae. The
discovery of hexactinian mesenteries in certain Edwardsiae,
already referred to, necessitates either the abolition of both this
order and that of the Protactiniae, or else an extension of the
latter to include both the Edwardsiae and many of the Hal-
campidae, and it seems to me that the former step is the most
practical and the most in accord with a correct phylogenetic
scheme. Not that I mean by this that the stages of develop-
ment shown by the members of the group do not represent
phylogenetic stages in the evolution of the Hexactiniae. Cer-
tainly no one will imagine that what has so long been regarded
as the Edwardsian type of structure is not in reality a primary
phylogenetic condition, even though we are now obliged to
regard the existing Edwardsiae as true hexactinians which
secondarily in some cases may represent the more primitive
condition.2 The facts of embryology speak too strongly re-
garding the Edwardsian stage to allow of question as to its
past occurrence, and I believe that there can be as little ques-
tion regarding the stages which I have supposed to intervene
between the Edwardsiae and the typical Hexactiniae, even
though the forms which to-day represent these stages do so
possibly only on account of secondary modifications.

1 Protanthea has four imperfect mesenteries which make pairs with the four
lateral perfect mesenteries, and twelve others arranged in pairs in the primary
exocoels, all being fertile and provided with mesenterial filaments. In addition to
these there are, however, as in Halairias, a number of short, narrow mesenteries
confined to the upper part of the column and standing in relation to the tentacles,
of which there are about ninety-six.

- Compare Van Beneden, Les Anthozoaires in Ergebnisse der in dem Atlan-
tischen Ocean, etc., ausgefuhrten Plankton-Expedition der Humboldt-Stiftung.
II. 1898.

1 62 McMURRICH. [VOL. II.

It seems inadvisable then to associate Halcurias with Gonac-
tinia and ProtantJiea, but there still remains a possible asso-
ciation with Hertwig's Antheomorphidae. Unfortunately the
forms upon which this family was founded are insufficiently
known, but it seems to me that there are reasons for main-
taining the position I have already ('98) advocated, that the
nearest allies of Halcurias at present known are to be found
in the family Antheomorphidae. I find myself obliged, how-
ever, to recede from the position I held in 1898 as to the dis-
tinctness of this family and to return to my earlier opinion,
which has received the approval of so critical a taxonomist as
Haddon ('98), that the Antheomorphidae should be included
in the family Actiniidae, and if this view be accepted it is
necessary to refer the genus Halcurias to that family also.

This will necessitate no important modification of the defini-
tion of the Actiniidae given by Haddon ('98), but as I shall have
occasion in the immediate future to consider the family in
some detail, I shall postpone a discussion of its delimitation
for the present.

UNIVERSITY OF MICHIGAN,
November 10, 1900.

No. 4-] MORPHOLOGY OF THE ACT1.\( >SOA. 163

REFERENCES.

'92 CARLGREX, O. Beitriige /.ur Kenntniss cler Edwardsien. Ofvers.

AV/. Vet. Akad. Forliandl. Stockholm. 1892.
'93 CARLGREX, O. Studien iiber nordische Aktinien. Kgl. Svenska Vet.

Akad. Hantfl. XXV. 1893.
97 CARLGREN, O. Zur Mesenterienentwicklung der Aktinien. Ofi>ers.

Kgl. Vet. Akad. Forliandl. Stockholm. 1897.
'99 CARLGREX, O. Zoantharien. Hamburger Magalhaenische Saininel-

reise. 1 899.
'89 Dixox, G. Y. and A. F. Notes on Bunodes thallia, Bunodes verru-

cosus, and Tealia crassicornis. Sci. Proc. Roy. Dublin Soc. x.s.

VI. 1889.

y

'95 FAUROT, L. Etudes sur 1'anatomie. 1'histologie, et le de'veloppement

des Actinies. Paris. 1895.
'98 HADDOX, A. C. The Actiniaria of Torres Straits. Sci. Trans. Roy.

Dublin Soc. Series 1 1. VI. 1898.
'93 McMuRRiCH, J. P. Report on the Actiniae collected by the U. S.

Fish Commission Steamer A Ibatross during the winter of 1887-

1888. Proc. U. S. Nat. Mi/s. XVI. 1893.
'98 McMuRRiCH, J. P. Report on the Actiniaria collected by the Bahama

Expedition of the State University of Iowa, 1893. Bull. Lab. Nat.

Hist. State Univ. of Iowa. IV. 1898.

OBSERVATIONS

ON THE HABITS AND NATURAL HISTORY OF
AMPHITHOE LONGIMANA SMITH.

SAMUEL J. HOLMES.

IN the present paper I have given the results of my observa-
tions made at Woods Holl, Mass., during the past summer on
a species of amphipod, AmpJiitJioc longimana Smith. Com-
paratively little is known of the habits of amphipods, and most
of what is known has been collected from scattered and casual
observations. There is a value in getting together all the facts
that can be obtained concerning any one species of animal, so
that they may be viewed in their ensemble and thus give us
some idea of the general life of the creature. For this reason
it was deemed best to devote the short time that could be
given to the study of amphipod behavior mainly to the observa-
tion of a single species.

Throughout the paper I have used many terms which imply
the existence in the animal of certain psychical states, such as
hunger, fear, and courage, without intending to affirm that such
psychical states really exist in the animal's consciousness, or
even that the animal possesses consciousness at all. It is diffi-
cult to describe the behavior of an animal without the use of
terms which have certain psychological connotations. Such
terms are here used simply as a matter of convenience in
describing actions simply as actions. The Crustacea may or
may not be " Reflexmachinen," and Bethe and others may or
may not be right in denying that they possess consciousness ; but,
however this may be, descriptions of actions in psychological
terms stand for certain peculiarities of conduct that could not
otherwise be easily described, and if the sense in which such
terms are used is understood, no confusion need result.

Amp hit hoe longimana may be obtained in large numbers
from the eel pond near the laboratory by simply drawing a

.65

1 66 HOLMES. [VOL. II.

net over the eel-grass. It is abundant during the summer
months, a period when most of the other species of amphipods
suffer a marked diminution in numbers. It is quite hardy, and
may be kept alive for months in small glass dishes, if they are
kept covered and the sea water kept fresh by a small piece of
Ulva. Observations on this species were carried on for nearly
three months. Specimens were kept isolated in small dishes
and daily observations made and recorded. I was thus able to
follow the histories of quite a number of individuals for a
considerable period.

Specific Description.

Body slender. Eyes round. Lateral lobes of the head
truncated in front. Antennules slender, about as long as the
body, the second segment a little longer, but much more
slender than the first ; third segment from one-third to one-
half the length of the second ; flagellum much longer than the
peduncle. Second antennae shorter than the first, but usually
with a longer peduncle ; last segment of the peduncle a little
longer than the preceding one; flagellum shorter than the two
preceding basal joints.

Second, third, and fourth epimera much longer vertically
than wide ; fifth epimeron about as long as the fourth, but
broader and excavated at the upper posterior angle ; lower
margins of the epimera furnished with very short setae.
Postero-lateral angles of the abdominal segments not acute.

First gnathopods in the male elongated, the first joint pro-
duced into a rounded lobe at the antero-distal angle ; carpus
narrow, nearly as long as the hand, and thickly setose on the
posterior margin ; hand very long and narrow, slightly incurved,
and of nearly the same width throughout, although slightly
widened near the base ; lower margin setose ; palm very short,
transverse, and rounded at the outer angle ; dactyl very large,
dentate, and projecting far beyond the palm when closed.
Second gnathopods much stouter than the first ; first joint with
a rounded lobe at the antero-distal angle ; hand much broader
and stouter than in the first pair ; palm oblique, with a deep

No. 4.] AMPHITHOE LONGIMANA SMITH.

I67

sinus near the strongly produced outer angle ; dactyl scarcely
projecting beyond the palm.

In the female the gnathopods are much shorter and weaker
than in the male; the hand in the first pair is less elongated,
and the palm is more oblique and more broadly rounded at the
outer angle. In the second pair the sinus in the palm is not
so deep, and the outer angle not so prominent as in the male.

Peduncle of the first pair of pereopods rather slender, much
longer than the rami, and reaching nearly to the tip of the
peduncle of
the second
pair; inner
ramusof the
second pair
of uropods
about as
long as the

peduncle. Posterior pair of uropods with the rami
scarcely half as long as the peduncle ; rami subequal
in length, the broader, more 'or less oval inner one
with a short spine at the inner posterior angle and
several setae on the transverse distal margin ; outer
ramus with the usual stout hooks. Length 6-9 mm.

In the older specimens the antennae are relatively more
elongated, and the hands of the male relatively longer and
narrower. The eyes in the living specimens are red, but
become black in specimens preserved in alcohol.

Habitat.

The range of this species as reported by Professor Smith is
from Vineyard Sound to New Jersey, and it has been reported
from Provincetown, Mass., by Richard Rathbun. It is not
uncommon among the seaweed near the shore, and it has been
taken at the surface in the vicinity of Woods Holl in the tow
net. Its occurrence at the surface is probably due to its
having been carried away from the shore by tide currents, as
it has a strong tendency to keep among objects of shelter.

!68 HOLMES. [VOL. II.

Its most favored habitat seems to be the eel-grass, where it
finds a convenient substratum upon which to construct its
nests. This species is much more common in the eel pond
at Woods Roll than outside ; the abundance of eel-grass and
various algae and the quiet water being conditions which doubt-
less favor its perpetuation. It is not found on the muddy
bottom and does not occur abundantly in the seaweeds near
the bottom, but it maybe obtained in quantity from the masses
of eel-grass at the surface.

Enemies.

In common with most amphipods, Amphithoe is doubtless
preyed upon by fishes, and it certainly affords one of the prin-
cipal articles of food of the small but voracious jelly-fish Goni-
onemus. The latter form, however, owing to its unfortunate
attractiveness to the zoologists frequenting Woods Roll, is in
danger of not continuing to be a very destructive enemy. It
is very common to find Gonionemus with Amphithoe in its
stomach. This crustacean falls an easy victim to its enemy, as
it often makes surprisingly little effort to escape, owing possi-
bly to a narcotizing effect of the poison of the nettling organs
of its captor. I have seen an Amphithoe while swimming
vigorously strike against the tentacles of the jelly-fish, suddenly
stop, and remain almost perfectly quiet while it was being

engulfed.

Food.

Amphithoe lives chiefly upon seaweed. The alimentary
canal may usually be seen to contain numerous fragments of
red or green algae. Pieces of Ulva that are kept in dishes
containing these amphipods soon exhibit gnawed margins and,
after some time, a marked diminution in size. I have often
observed the process of feeding. The Ulva is gnawed directly
by the mouth parts, without being previously torn away by the
gnathopods. The quantity of algae eaten, judging by the
amount of excrement voided, is very considerable. In order
to ascertain how rapidly the excrement accumulated, a speci-
men with an abundance of Ulva was placed in a clean dish, and

No. 4-] AMPHITHOE LONGIMANA SMITH. 169

it was found that one hundred and forty-six masses accumulated
in twenty-four hours. These masses consisted almost entirely
of broken-up cells of Ulva, the contents of which had been
digested out. By making a very rough estimate based on the
size of these masses, it was calculated that the amount of food
eaten by the animal in twenty-four hours was about equal to
one-tenth of its bulk.

Amphithoe is by no means a strict vegetarian, for it will
devour animal food with great eagerness when it can be
obtained. It is very fond of bits of flesh of almost any ani-
mal, not excluding that of its own species. When aware of the
presence of food sufficiently near its nest to be seized without
letting go its hold, it will dart out.with a quick movement, grab
the food with its gnathopods, and suddenly retract itself inside
its domicile. When the food is brought in, it is held by the
gnathopods while being devoured.

Movements.

Of the movements performed by Amphithoe, the beating of
the pleopods is the most constant and uniform. Whether the
animal is swimming, crawling, or lying quiet, the pleopods are
continually engaged in their regular to and fro movement.
The motion of these appendages while the animal is at rest
serves to create a current of water past the gills in front, and
thus aids in respiration. The abdomen, except during swim-
ming, is held strongly flexed, forming a sinus, at the posterior
end of which the bases of the pleopods are attached, the tips
pointing forward. Small particles suspended in the water may
be seen to be drawn in at the sides of this sinus and thrown
out at the anterior end, thus .indicating the course of the
current.

The rhythm of the motion of the pleopods, like the respira-
tory movements of the higher vertebrates, goes on in a regular
way as a rule, but may be checked by impulses from the higher
nervous centers. When the animal changes its position, or
executes any other decided movement, the pleopods may cease
their action for a moment, but soon resume their regular beat

I*JQ HOLMES. [\'OL. II.

as before. Commonly during swimming the pleopods beat
more rapidly, but this is not always the case. When the
swimming ceases they drop back into their usual rhythm,
whether faster or slower than before. In their motion the
three pairs of pleopods act as a unit, keeping perfect time
like well-trained oarsmen. If the abdomen be removed from
the rest of the body, the pleopods, after a few spasmodic move-
ments clue to the shock of the operation, continue to beat
rhythmically for several minutes ; the three pairs all move with
the same rhythm, though somewhat more slowly than before the
operation. Moreover, if a single segment with' its pair of
appendages be isolated, the rhythmic motion of the appendages
still goes on for some minutes, but gradually becomes slower
and more irregular, until nothing but small twitches indicate
the existence of irritability.

When the animal is in a vigorous condition the beat of the
pleopods is rapid, but when the creature becomes weakened the
beat becomes slower, yet as long as life lasts the pleopods con-
tinue their movements. The beat of the pleopods may still
persist after the rest of the animal refuses to respond to any
sort of stimulation.

The swimming of Amphithoe is mainly effected by the pleo-
pods. The first impulse, however, is gained by the sudden
extension of the abdomen, which gives the body a rapid forward
movement. The abdomen is then held in an extended position
and the pleopods, which then hang nearly at right angles to the
body, serve to continue the forward motion. When swimming
against the force of gravity the motion of the pleopods alone
is not sufficient to keep the body going, and when the original
impetus becomes exhausted the abdomen is bent forward and
again suddenly extended, giving the animal a fresh start. The
flexure of the abdomen before every stroke tends to draw the
body backward. This, combined with the weight of the animal,
causes ground to be lost between every stroke. Swimming
towards the surface is therefore accomplished by a series of
springs, between each of which the animal falls back more or
less. While swimming horizontally the beating of the pleo-
pods is all that is required to keep up the motion ; specimens

No. 4.] AMT1I1 rtlOE LOXG1MAXA SMITH. 171

may be seen swimming about for a considerable time without
employing the abdomen.

Amphithoe has a decided disinclination for continuous swim-
ming. Ordinarily it makes only short excursions from one
place of concealment to another and generally stops upon
meeting with the first solid object that comes in its way,
although when situated where there is nothing to which it can
lay hold it may swim for some time in a uniform manner. It
may swim in various ways, on its side, or with either the dorsal
or the ventral surface uppermost, and apparently gets along
with about equal facility in any of these positions.

The beat of the pleopods tends to propel the body not in a
straight line forward but in a circular course. The pleopods
being on the ventral side tend to cause the body to veer around
towards the dorsal side. When the body is held somewhat
concave on the ventral side, as it often is, this tendency may
be balanced or overcome by the tendency to move in circles in
the opposite direction. Such a condition is analogous to a
person rowing on one side of a boat with the rudder turned
toward the side of the oar. By having the body extended to
the right degree a straight course may be maintained. The
direction of movement is often changed by the animal turning
now on one side and now on the other. Circular movements
in one direction are thus compensated for by circular move-
ments in another as the animal turns over and a certain general
direction of motion is maintained. When swimming on the
back a nearly straight course is kept by rolling the body
slightly to the one or the other side. Rolling is probably
effected by the movements of the hinder pairs of thoracic legs.
When the animal is swimming these legs project outward and
backward. A downward stroke of these appendages on one
side would push the same side of the body upward and roll it
over toward the opposite side. In a larger species of amphipod,
whose movements are not so exceedingly rapid as those of
Amphithoe, I was able to see that the rolling of the body was
effected in just this way. It is highly improbable that in
Amphithoe a different method would be employed to produce
the same result. However this may be, it is certain that

j-2 HOLMES. [\'OL. II.

Amphithoe steers itself while swimming by altering the exten-
sion of the abdomen and by rolling from side to side. Lateral
bendings of the body could not be seen to play a part in
directing the swimming motions, although I have observed
this method of steering employed by other amphipods.

Amphithoe longimana, like many other amphipods, is unable
to walk over a plane surface. When out of water it is able to
make indifferent progress by the characteristically amphipodan
gliding movements produced by alternately flexing and extend-
ing the abdomen. It is utterly incapable of leaping like the
sand fleas and some of their aquatic relatives. Owing to its
compressed form, it is unable to maintain itself upright while out
of water, or even in water, unless it has some object to which it
can lay hold. In its characteristic habitat among the seaweed,
Amphithoe crawls with considerable agility. The principal
organs for crawling are the second antennae, the two pairs of
gnathopods, the third and fourth pairs of pereopods, and to a cer-
tain extent the abdomen. The antennae are thrown over objects
and flexed, thus tending to pull the body upward and forward.
The gnathopods are used to seize objects in order to pull
the body along. The two following pairs of appendages are
employed much as the walking legs of insects, although they
move in a nearly vertical plane. The abdomen assists in loco-
motion by being thrust forward beneath the body until the tip
is hooked on to some irregularity of the surface over which the
animal is moving when it is extended, thus giving the body a
forward impulse. The movement recalls the leaping motion
effected by the abdomen in the sand fleas. In fact, very
similar motions are performed in both cases, but in Amphithoe
the motions are much less rapid and energetic. The ambula-
tory movements of this species are never rapid, owing doubtless
to the necessity for keeping the body from falling over on its
side. The last three pairs of thoracic legs, although not used
directly for locomotion, are indirectly of service in holding the
body upright. These appendages, which are bent over the back
and have their claws pointing forward, are used to hook on to
objects and thus act as organs of support while progression is
effected by the other appendages of the body.

No. 4-] AMPHITHOE LONG I MAN A SMITH. 173

Ordinarily Amphithoc lies in its nest, with the antennae pro-
truding from the opening at the end. The lower pair of
antennae are generally held motionless. The upper pair, how-
ever, are usually seen to be moving to and fro, sweeping about
in almost every direction. Occasionally their motion is sud-
denly checked and they are held motionless for a time, but
soon their moveme/it is resumed. The significance of these
movements will be discussed in a later section. The two pairs
of gnathopods are used for a variety of purposes. Occasion-
ally the antennae are bent downward and seized by the gnath-
opods and the flagella drawn through the space between the
dactyl and the palm. The function of this act is probably to
strip off any foreign bodies that may become attached to the
antennae. The gnathopods are frequently employed to grab
passing objects and to reach out and pull in bits of seaweed for
the construction of the nest. They are used also for holding
the food that the animal is eating and for carrying bits of food
to the mouth, where they are taken by the maxillipeds. While
not exercising any of their normal functions they may often be
seen going through the motions of grasping, with nothing to
grasp. This restless activity of the gnathopods seems to be
nothing but the exercise of the grasping reflex called forth by
some unknown stimulus and having no useful result. The act is
performed in all degrees of completeness, from a definite grasp-
ing motion to a mere nervous twitch. The mouth parts per-
form many motions when the animal is not masticating food.
These movements, which resemble the normal motions of
mastication so far as could be observed, apparently have no
functional significance. They take place in specimens kept
for a considerable time in dishes in which there is nothing
that could serve for food. Like the grasping actions, they are
movements which are called forth without the normal exciting
cause.

One of the most curious actions which Amphithoe performs
is its reversal of position in the nest. If the antennae be
somewhat roughly struck with a needle, or even if a threatening
object suddenly appears close in front, the head and antennae
will appear at the other end of the nest. As the nest is a tube

l-^ HOLMES. [VOL. II.

but little wider than the body of its occupant, no one who
watches the operation can fail to have a feeling of admiration
for the neatness and extreme quickness of this acrobatic feat.
The animal executes this " about face" with such rapidity that
it is only after watching the operation repeatedly that one can
determine how it is effected. As the animal lies in its nest
the abdomen is bent forward and the posterior pairs of thoracic
legs are directed backward, their claws being usually hooked
into the walls of the nest. When about to turn around, the
abdomen is thrust forward, its terminal hooks caught in the nest ;
the tip of the abdomen, therefore, forms a fixed point, and the
insertion of the thoracic legs forms another. The contraction of
the legs would therefore pull the middle and anterior parts of the
body backward. When the head is pulled back some distance,
extension of the body occurs, forcing the anterior part of the
body through to the other end of the nest. The head end
being reversed, the abdomen is loosened and quickly flexed
again under the body. The whole operation is completed in
less than a second, and the animal may be made to repeat the
performance several times in rapid succession.

Nests and Nest-Building.

The nests of Amphithoe are tubular structures which gener-
ally exceed somewhat the length of the animal. They are
made of a web-like material which, under the microscope, may
be seen to be a network of exceedingly fine threads. The
nests are usually constructed among the branches of the red
seaweeds or upon the leaves of eel-grass or the fronds of Ulva.
When built upon Ulva the nest is generally located in a wrinkle
or fold of the surface which affords a partial shelter. The nest
is open at both ends and is of about the same diameter through-
out. Foreign materials, such as bits of seaweed, are usually
woven into the nest and greatly add to its efficiency as a means
of concealment. Amphithoe frequently leaves its nest, but I
could find no evidence that it would return to its own nest
more readily than to any other ; it will simply enter the first
unoccupied nest that comes in its way. When established in

No. 4-] AMPHITHOE LONGIMAXA SMITH. 175

a nest Amphithoe is driven out only with difficulty. A mem-
ber of its own species that approaches is grabbed at and usually
driven off, and the creature appears to be on the alert to keep
out all intruders. The approach of a more formidable-looking
object causes the animal to retreat farther back into its nest.
If the antennae are stroked with a needle, a sudden somersault
will be executed and the head will appear at the other end of
the nest. Then it usually requires quite a series of pokes to
make the creature quit the nest entirely. The instinct to
remain in the nest when danger threatens presents a marked
contrast to the quickness with which flight is made when the
animal is roaming free.

A new nest is constructed in a remarkably short time, often
in less than a half hour. If a few specimens be placed in a
dish of sea water containing a little seaweed, nests will be
woven on the seaweed and on the lower surface of the dish,
and in a short time the number of nests may greatly exceed
the number of specimens. Those localities are chosen which
give the animal a maximum of contact with solid objects. In
dishes in which specimens were kept I have nearly always
found several nests along the angle between the bottom and
sides, although the seaweed kept in the dishes afforded locali-
ties better adapted for concealment. The choice of a spot for
a nest is apparently largely a matter of thigmotaxis. When
the animal remains in a spot for some time, the nest-building
activities begin, and where contact with different sides of the
body is secured, as between the branches of seaweed, in the
wrinkle of an Ulva frond, or in the angles of a glass dish, it
remains quiet. If Amphithoe is observed while constructing
its nest, the first and second pairs of pereopods will be seen to
be busily engaged in moving back and forth from point to
point along the surface on which the web is being laid down.
The first and second pereopods contain large glands which are
connected with a duct which opens at the tip of the claw. The
material for the web is secreted by these glands and probably
hardens soon after its emergence, like the web of a spider. A
very fine thread of web may frequently be seen passing out
from the small opening at the tip of the claw. As the tip of

1 76 HOLMES. [VOL. ii.

the claw touches one point after another, the web, as it is drawn
out, is fastened to different places. By moving back and forth
and rolling around during the weaving process, the animal
constructs its tubular dwelling. Several specimens from which
I clipped the claws from the first two pairs of pereopods were
kept for several days and did not construct a single nest.

During the construction of the nest, Amphithoe will reach
out and draw in bits of algae and other objects that lie near
and incorporate them into its dwelling. In a few cases I have
seen long pieces of algae bitten in two and used for this pur-
pose. As Amphithoe lives largely on algae, this biting may
not have had any special reference to nest-building, but may
have been a manifestation of the ordinary reaction to food.
In Microdeutopus, Smith has observed that the excrement of
the animal is worked into the web ; but in Amphithoe, whose
nest-building habits seem to be very similar, no such process
could be observed. The excrement is passed out of the nest,
accumulations of it usually being observable near the two ends.

Moulting.

Amphithoe was found to shed its skin more often than was
anticipated. Most of the specimens I kept isolated as long as
a week moulted once, and out of four specimens in which I
have records of the dates of two successive moults of the same
individual the interval between moults in three cases was seven
days, and in the other case eight days. These specimens were
of the usual size. How rapidly moults occur in different periods
of the life history of this species I cannot say.

The process of moulting in Amphithoe occurs in the same
manner as has been described in other species of amphipods.
The skin splits transversely along the line joining the head and
thorax, and on either side of the thorax is a longitudinal split
which occurs between the upper margins of the epimera and
the lower margins of the thoracic rings. This split may extend
along all the thoracic segments. The head and antennae are
pulled backward out of their investment and the posterior part
of the body is pulled out forwards, the old skin, after being

No. 4-] AMPHITHOE LONGLMANA SMITH. 177

shed, remaining intact except at the lines just mentioned. The
moulting process takes several minutes at least and is accom-
panied by considerable muscular effort to get out of the old skin.
In the several cases in which I observed the process, Amphi-
thoe leaves its nest to divest itself of its skin, and I have never
observed a moult in a nest but always some distance away.
After moulting the animal is rather quiet and cannot easily be
enticed from its nest by food. I have observed several cases
in which death occurred during the moulting process. In one
case moulting was not completed for several days. The
specimen was observed August 17 with the head and tail ends
drawn partly out of the old case. The next day the head and
antennae were still not completely drawn out, but the rest of
the skin was kicked off. On August 21 it was still in the same
condition, the feeble beating of the pleopods giving evidence
of failing strength. On the next day it died, the head and
antennae still only partly extricated from their old covering.
In several cases the antennae were observed to become broken
off in the process of moulting, but I have seen no cases in
which other appendages became lost in this way. The anten-
nae are the appendages most liable to more or less complete
loss from other causes, but owing to the rapidity with which
these organs can regenerate this loss can produce only a
temporary inconvenience. The cast-off skins are found some-
times on the bottom, and often floating on the surface of the
water, and in a short time after they are shed become filled
with swarms of protozoa.

TJie Seat of Smell.

Much has been written concerning the seat of the olfactory
sense in the Crustacea, but most opinions on the subject have
been based on morphological instead of experimental evidence.
The work of May and Bethe affords good evidence that in the
decapod Crustacea the seat of the olfactory sense, or, as Bethe
prefers to call it, of chemoreception, is in the first antennae, as
analogy with the insects would lead one to suspect. The first
antennae are not, however, according to Bethe, the only seat

jy3 HOLMES [VOL. II.

of chemoreception. In Carcinus the removal of the first
antennae as far as the first basal segment is followed by a
marked diminution of the power of reaction to chemical sub-
stances in the water. A Carcinus when the eyes are black-
ened over will find pieces of food when placed at some distance,
by the sense of smell. When the first antennae are removed
at the first basal segment, Bethe found that food maybe placed
as near as 10 cm. to the animal without calling forth any
efforts to obtain it. When the food is brought close to the
mouth or close behind the animal without contact with the body,
it is seized and eaten. The first antennae, therefore, while
they may be the main, are not the sole source of the reception
of olfactory stimuli.

My own observations on Amphithoe led me, before becoming
acquainted with Bethe's results, to infer a double seat of the
sense of smell. In Amphithoe, as in Carcinus, the first anten-
nae seem to be the most important olfactory organs. While
the animal is at rest in its nest the antennules are kept swaying
to and fro in different directions, as if they were being employed
to explore the surroundings. If a small bit of flesh is held on
a needle or in a fine pair of pinchers and carefully brought
near the animal, the antennae check their random movements
and make one or more strokes in the direction of the bit of
flesh ; often the antennae are held for some time in the direc-
tion of the object. On bringing the flesh nearer, the animal
may be seen to adjust itself in the nest for a sudden spring,
and if the flesh is sufficiently near to be touched by the
antennules the amphipod makes a sudden dart from the nest,
seizes the object, and draws quickly back again, never letting
go its hold, however, of the nest. The animal as a rule readily
distinguishes between the contact of flesh and that of a body
not serviceable for food. Only rarely does touching the anten-
nule with a needle call the animal forth from the nest. It
may be deceived more often if, when excited by the presence
of meat near by, one of its antennules be touched with a
needle; then it may dart out towards the needle and even
seize it. But the animal responds much more surely, as I have
found by repeated experiments, when the antennules come in

No. 4.] AMPHITHOE LONG I MAN A SMITH. \ 79

contact with the food itself, even when the animal is excited by
the presence of food in its vicinity. The darting forth, there-
fore, is apparently caused, not merely by a tactile stimulation,
but by a chemical stimulus from the food. The antennae are
delicate tactile organs, and tactile stimuli may assist in calling
forth the actions which result in the seizing of food, but tactile
stimulation alone generally fails to accomplish this result.

After Amphithoe has made a meal of fleshy diet it becomes
quite indifferent to the presence of that kind of food in its
vicinity and no longer darts forth to grab bits of flesh brought
in contact with its antennules. Different individuals present
very different degrees of eagerness for animal food, owing
doubtless to varying intervals of time since their last repast.

Sight has probably little to do with the food reactions of the
animal. When the head is completely withdrawn in the nest
the animals often give signs of perceiving food and dart after
it when brought in contact with the antennules. In many
cases the nest is so opaque that the animal cannot see through
it with any distinctness, and under these circumstances, when
the head was entirely withdrawn into the nest, I have often
brought bits of meat so they would be touched by the anten-
nules only when they were strongly bent backwards. Although
the meat was out of sight, the amphipod would dart out, bend
backwards, and seize the morsel. If the desired object is out
of reach of the antennules, the amphipod will not spring for it,
although it may be seen to make ready to do so. It will not
go to the length of leaving the nest to seize food, even if its
conduct betrays evidence of keen hunger. An object near
enough to be struck by the swaying of the antennules is suffi-
ciently near to be seized by the animal without letting go its
hold of the nest. It is a noticeable feature of the species of
Amphithoe and related genera that the antennules are, roughly
speaking, of about the length of the body. This feature is not
improbably correlated with the similar tube-dwelling habits of
these forms, the length of the antennules gauging the length
of a safe and successful spring.

The effect of removing the antennules of Amphithoe is
greatly to lessen responsiveness to olfactory stimuli. The

I So HOLMES. [VOL. II.

shock of the operation has a very temporary effect, for in a
few hours the animals behave with their usual activity. Meat
brought in contact with the second antennae is generally not
seized. This, however, is not always the case, for in several
instances I have found that contact with the second antennae
causes the grasping reflex. I was inclined at first to attribute
a certain olfactory sensibility to the second antennae, but I
found later that the animal reacts about as well to olfactory
stimuli when both antennae are removed as when the second
alone remain. In specimens with both antennae removed near
the base, leaving only the first joint of the peduncle of each
pair, there was no reaction to food placed in what would have
been within easy reach of the antennae before their removal.
If a piece of meat is placed about 2 mm. from the mouth of the
amphipod, it is generally allowed to remain untouched for
several seconds and then suddenly seized and eaten. The
morsel is seized by the gnathopods and at the same time
bitten at with the mouth parts. The reaction is not an imme-
diate one, such as is brought about by contact of the anten-
nules with food. It appears to be necessary for the food to
remain awhile in close proximity to the animal before its edible
nature is perceived ; when this occurs the seizing takes place
quickly enough. I have tried to induce the animals to take
bits of substance of the general appearance of fragments of
meat and brought very close to the mouth parts, but they are
apparently able to distinguish, before any contact with the
object occurs, whether or not it is of an edible nature. It is
true that Amphithoe often grasps objects that lie near by,
pulls them back, and incorporates them into the structure of
its nest, and it might be inferred that the seizing of meat lying
close to its mouth by a specimen with both antennae removed
is an expression of the nest-building instinct to seize any small
object within reach for building material. The reactions in
the two cases, which I have observed many times, differ. An
object used for the construction of the nest is reached for and
pulled back to the nest and not as a rule brought in contact
with the mouth. A bit of meat is grabbed at and bitten at in
the same act. This difference in reaction and the fact that

No. 4-] AMPHITHOE LONGIMANA SMITH. 181

the animals seize bits of meat when they cannot be induced
to pay attention to other objects of similar appearance con-
vinced me that the reaction to food was caused by chemical
stimulation from diffusing substances in the water.

Removal of the second pair of antennae, the first being left
intact, was not found to exert any marked influence upon
reactions to chemical stimuli. The second antennae may
transmit olfactory stimuli ; it would be difficult to prove they
do not in a certain degree, but the evidence obtained does not
justify us in attributing to them this function. When, after
removal of the first antennae, Amphithoe responds when food
is brought in contact with the second antennae, the reaction
may be due to the animal becoming aware of the presence of
food through some other organ, contact with the antennae
indicating that the food is sufficiently near to be seized. In
other words, the reaction may be clue to purely tactile stimula-
tion, the animal being keyed to this reaction by the excitement
of olfactory stimuli from some other organ. What other organ,
or organs, may serve to transmit olfactory stimuli is uncertain.
This has not been determined in the decapod Crustacea, which
afford the only other instance in which a double seat of the
olfactory sense has been suspected, or in fact in which, so far as
I am aware, any experimental evidence has been adduced as to
any location of this function at all. It seems probable that
some of the mouth parts have some olfactory function, as they
afford the most obviously appropriate location for such a sens'e.
Owing to its small size, Amphithoe is not a favorable form in
which to decide this question, and the attempt to do so was not
made.

Color and Color Changes.

One cannot but be struck, when examining a number of
specimens of this species, with the marked differences in color
presented by different individuals. Some are bright green, like
the bright green seaweeds ; others may be nearly colorless ; a
few are of a light blue green tint, and many range from a light
to a dark reddish-brown. The same individual may take on, at
different times, all these varieties of coloration. The color

HOLMES. [VoL. II.

differences are produced by the variation of five elements :
(i), the color of the chitinous integument; (2), the color of
the blood and tissues ; (3), the contents of the alimentary canal ;
(4), the color of the sex glands; (5), the pigment cells. The
first of these factors is, perhaps, the least important and is not
subject to great variation. The exoskeleton over most of the
surface of the body is colorless ; on the antennae it is marked
with transverse reddish-brown bands which give the light red-
dish-brown annulations of these organs. This color is seen as
distinctly in the shed skin as in the living animal.

The color of the tissues and blood is subject to great varia-
tion. The green color of Amphithoe, or the blue green tint
when it occurs, is due to some coloring matter that is uniformly
diffused throughout the animal. In some specimens there is a
sufficient amount present to give the animal a brilliant emerald
green, but many may be found in which not the slightest trace
of green coloration could be detected. This green color may
be seen to undergo marked changes in intensity if individuals
be watched for several days. The blue color is much rarer.
One specimen in which this blue coloration was strongly
marked was kept under observation for several days. After
five days most of the blue color had disappeared, the green
becoming more nearly like the typical green of other forms.
After six days the green was not to be distinguished from
the ordinary type ; the green color then gradually became
fainter, and on the ninth day the tissues were whitish, scarcely
a trace of green being visible. During all this time the speci-
men ate abundantly of green algae, judging from the amount of
excrement consisting of Ulva cells that accumulated in the dish.

The contents of the alimentary canal influence to a consider-
able extent the general color effect produced by the animal
when seen by the naked eye. If they consist largely of green
Ulva, they tend to give the animal a greenish appearance. If
the Ulva has been subjected some time to the action of the
digestive juices and become a yellowish color, it tends to give
the animal a corresponding yellowish aspect. Light becoming
colored by passing through the alimentary canal is reflected
and re-reflected in the tissues and tends to make them appear

No. 4.] AMPHITHOE LONGIMANA SMITH. 183

a corresponding color. I have several times cut off parts of
the body to see if their color might not have been due to this
cause ; but whatever effect this factor may have it is certain
that the green color of the blood and tissues is not entirely
caused in this way as it may easily be observed in the isolated
appendages. The part played by the sexual glands in the
coloration of this species varies greatly owing to the variation
in the size of these organs.

It is to the pigment cells that the most marked changes of
color are due. These cells are of two kinds, --reddish-brown
pigment cells, and cells with a pale green pigment. The latter
play an insignificant part in the coloration of the animal, as
they are pale in color and few in number, there being often
not more than a dozen on the entire surface of the body. The
pale green color appears most clearly in transmitted light ; in
reflected light they are of a silvery hue. Their size is about
the same as the largest cells with red pigment. Like the latter,
they are very richly branched, but were not seen to undergo
much variation in the distribution of their pigment. They are
mainly confined to the epirnera, being usually situated near the
lower margin.

The most important elements in determining the color
changes are the reddish-brown pigment spots. These spots
are scattered all over the body and are found also on most of
the appendages, especially towards the proximal end. When
extended the pigment spots are large and very richly branched,
forming most beautiful objects when seen under the microscope.
When fully contracted these spots assume the form of round
dots, and all stages of expansion may be seen in different
specimens, or even in the same specimen, between the most
contracted and the most expanded state. There are a few
large spots near the lower edges of the epimera that are
generally found in an expanded condition. Even when most
of the spots over the surface of the body are contracted, these
few spots, which may be not more than a dozen on each side,
are usually conspicuously large. This circumstance affords a
convenient means of distinguishing Amphitlioc longimana at a
glance from other species of the same genus.

1 84 HOLMES. [VOL. II.

The pigment spots of Amphithoe apparently contain but
one kind of pigment. The whole system of chromatophores is
much less complicated than that which Gamble and Keeble :
found in Hippolyte varians, and the power of sympathetic color
changes in relation to surrounding objects much less perfectly
developed. Amphithoe may be said to adapt its color to its
environment, but in only a rather rough way compared with the
remarkable protective color changes of Hippolyte. Specimens
taken from the eel-grass are very apt to be of a greenish color,
and specimens taken from among masses of red seaweed are
usually colored somewhat like their environment. While on
the eel-grass, they are usually exposed to the light and the pig-
ment spots are in a contracted condition. This allows the green
color of the tissues to be seen, and the animal has, consequently,
a greenish aspect. In the masses of red seaweed the animal
is usually more shaded and the pigment spots become expanded,
giving the animal a reddish tint which helps to conceal it in its
environment. Moreover the alimentary canal in specimens
found in the red seaweed often contains a greater or less
quantity of this alga, and this also helps to color the animal in
a protective manner. So far as the green color of the tissues
is concerned there appears to be no difference between the
animals from different habitats. The change from green to
reddish, or the reverse, completes the range of adaptive color
variations in this species so far as they are induced by changes
in the environment. And this kind of color change is the
one best adapted to afford protection to the species in its usual
habitat.

The pigment spots of Amphithoe change very slowly ; it
generally requires some hours to effect a change from the
expanded to the contracted condition. It was found by several
experiments that exposure to bright light causes the pigment
spots to contract, while specimens that have been kept in the
dark for several hours generally have the pigment spots much
expanded.

1 Quart. Journ. ilficr. -SV/., 1900.

No. 4-] AMPHITHOE LONGIMANA SMITH. 185

Sexual Habits.

Concerning the sexual habits of Amphithoe I have little to
add beyond what is known among other amphipods. The male
carries the female about for a considerable period and main-
tains his hold against efforts to dislodge him with great perti-
nacity. The instinct to retain hold of the female is sufficient to
overcome all fear, and it is difficult to separate the male without
injury. The posterior part of the body may be cut off, and yet
the anterior portion retains its hold of the female as long
as sufficient vitality remains. Ordinarily the male retains his
hold of the female by hooking the claws of his pereopods
beneath the edges of the epimera. The gnathopods are not
generally employed for this purpose, but are called into use
when a sudden disturbance renders the hold of the male inse-
cure. The female remains remarkably passive when carried
about by the male. Her body is usually held quite strongly
flexed and the male does the swimming for both, the female
being transported as so much dead weight. While carried by
the male the female seems much less responsive to stimuli
than when free. When poked by a needle she often makes
little motion, but the more alert male is generally aware of the
disturbance and carries her away from the seat of annoyance.

TJie Disposal of Excrement.

In its natural position Amphithoe lies so that the excrement
that is voided would be deposited in the nest. Yet the excre-
ment is never found in the nest but at some distance from
either end. At first I supposed that it was carried out by the
current of water produced by the movement of the pleopods,
but, after watching the animal, I noticed that when excrement
was extruded the abdomen was bent forward and the gnatho-
pods reached back and seized the mass as it was ejected from
the intestine, and passed it out of the front end of the nest.
This act was observed three or four times, but whether it is
always performed when excrement is extruded I cannot state.
In one individual lying outside a nest on the bottom of a glass

HOLMES. [VOL. II.

dish I noticed that when excrement was passed the abdomen
was bent forward and the mass seized by the gnathopods and
passed forward, just as it is when the animal is in the nest.
There was of course no use in seizing the excrement under the
circumstances, but the act was performed in the usual instinctive
way nevertheless.

Timidity and Pugnacity.

Specimens of Amphithoe are very ready to attack other
amphipods that come near and drive them away. The ani-
mals appear to be on the alert to prevent any other individual
from gaining access to the nest, --so much so that they very
frequently spring out at a passing amphipod and bite at it in
what appears to be a particularly vicious and hateful manner.
The individual attacked does not, so far as I have observed,
attempt any defense but precipitately flees from the spot. Any-
thing that properly could be called a fight is never engaged in ;
a passing nip with the gnathopods, or bite with the jaws, is all
that seems to occur in the nature of hostilities.

While ready to dispute the entrance of another amphipod
from in front, Amphithoe generally quickly flees from its nest
when an intruder enters from behind. One often sees the
occupants of nests routed out by others entering in this way,
and I have seen one individual that was expelled by another
entering behind it swim around, enter the other end of the
nest, and drive out the intruder. Courage in Amphithoe
depends in great measure on whether the attack is made from
in front or behind.

Outside the nest Amphithoe is very timid. It does not
attack its fellows except by giving an occasional nip when
accidentally colliding with them, and it flees quickly when
disturbed. It can very rarely be induced to seize meat, how-
ever hungry it may be, and however carefully the food be
presented. Even when in the nest, care has to be taken in
offering the animal food lest it be alarmed by one's move-
ments. This alarm is manifested by withdrawing a short dis-
tance in the nest. When an animal to which meat is presented
withdraws in this way, I find that it is useless to attempt to

No. 4-] AMPHITHOE LOXGIMANA SMITH. 187

induce it to take food for several minutes, until its fright wears
away. Experience in feeding these animals soon enables one
to tell whether or not they have become frightened by your
actions.

So far as my observations go they indicate that Amphithoe
has very little true pugnacity. It does not engage in a conflict
in order to overcome an adversary, as many decapod Crustaceans
do ; it fights only in self-defense. The attacks on other
amphipods passing by the nest are simply measures to keep
out unwelcome visitors. Had not these forms the instinct to
keep the nest to themselves, several individuals would often
crowd into the same nest much to their mutual inconvenience.

Phototaxis.

Amphithoe, like most of the aquatic Gammaridea, is nega-
tively phototactic. The specimens experimented with were
placed in an elongated, rectangular dish contained in a box
open at one end and above and blackened on the inside. When
placed near a window, either in direct sunlight or so that rays
of diffuse daylight fell obliquely into the dish, the animals
would swim towards the end of the dish farthest from the
source of light. When the dish was turned about, they swam
back again to the other end. In lamplight they may be driven
alternately from one end to the other by moving the lamp back
and forth to opposite ends of the dish. The endeavor was
made to make Amphithoe positively phototactic by altering the
temperature and concentration of the sea water, but only nega-
tive results were obtained. The animals remain negative until
the water is heated to about 90° Fahr., when little responsive-
ness to light remains. Further heating causes heat rigor to
supervene. Increase of temperature may be carried to the
point of producing death without changing the direction of the
phototactic movements. It is not probable that cooling the water
below the normal temperature would alter the phototactic
response. Decrease of temperature, other things equal, tends
to give rise to negative phototaxis in Orchestia, and increase of
temperature has the effect of making this form more positive.

1 88 HOLMES. [VOL. II.

Raising the temperature has the effect of making Gammarns
mncronatus positively phototactic, but an increase of only a few
degrees beyond the point where its phototaxis changes pro-
duces heat rigor. Possibly Amphithoe might be rendered
positively phototactic could it endure a somewhat higher
temperature.

Neither increasing nor decreasing the concentration of the
sea water was found to effect a change in the direction of
phototaxis. A more extended account of phototaxis in the
Amphipoda will appear in another paper.

TJiigmotaxis.

The tendency of Amphithoe to keep in contact with solid
objects is one of the most conspicuous features of its behavior.
This tendency is apparently one of the fundamental instincts
of the group, as it is exhibited very strongly by most of the
Amphipoda. When placed among branched seaweeds, Amphi-
thoe stops only after it works its way among the branches
where there is contact on several sides of the body. In Ulva
it comes to rest in a fold of the frond. When placed in a glass
dish containing nothing but sea water, it swims about restlessly
and eventually comes to lie quiet in the angle between the
bottom and side. The instinct to crawl into an empty nest
is an expression of the same tendency. Were it not for its
thigmotaxis, the whole conduct of the animal would be very
different from what it is. Many other instincts, like the nest-
building instinct and the instincts associated with it, are built
upon this fundamental reaction as a foundation. It does not
seem improbable that the instinct of the female to remain
perfectly quiet while carried about by the male, and even the
strong propensity of the male to seize and retain hold of the
female, may be but modified and specialized forms of thigmo-
taxis. Given variations of responsiveness to contact in differ-
ent parts of the body, and variations in the manner in which
the responsiveness is exhibited, we would have the means by
which thigmotaxis might be modified by natural selection into
more specialized forms of behavior. If the origin of the various

No. 4-] AMPHITHOE LONG/MAX. I SMITH. \ Sy

forms of amphipod behavior could be traced, it would be found,
I believe, that thigmotaxis is the mother of many instincts.

The Instincts of tlic Young.

When the young of Amphithoe quit the maternal brood
pouch, they have a quite different appearance from the adult.
They are whitish in color, with only a very few pigment spots,
located for the most part on the epimera, each epimeron con-
taining often but one spot. These pigment cells are of a
greenish-gray color and apparently have none of the reddish-
brown pigment which is found in those of the adult. By
reflected light they have a light greenish-silvery appearance
much like the large light green cells of the adult, but they
are very much smaller and much less branched. The head is
relatively large, but the eyes are small and red and composed
of only six ocelli, one of which is in the center surrounded by
five others. The number of ocelli, therefore, increases very
greatly as the animal grows older. Both pairs of antennae are
short, the flagella of the first pair consisting of four joints and
those of the second pair of but three.

The young when first hatched are in a feeble condition and
are carried about for a few days in the brood chamber of the
mother. Before they are hatched one can easily see the beating
of the heart and the peristaltic movements of the intestine.
Shortly before their emergence the young may be observed
flexing and extending the body in the effort to break the shell.
In one case I removed a lot of eggs from the brood pouch of
the mother while they were hatching. Those that had not yet
emerged were very vigorous in their movements within the
shell. When hatched they were unable to swim, and their
movements were irregular and little coordinated. The next
clay a few had died ; the others could swim feebly, but none
well, and if they became caught in the film at the surface of
the water they were unable to overcome the surface tension
and get free. In another case I removed from the brood
pouch of a female several young that had been hatched only
for a short time, as they were feeble and scarcely able to swim.

190 HOLMES. [VOL. II.

The swimmerets, however, beat rapidly, and the maxillipeds
and gnathopods were in constant motion. The antennules
were moved more than the antennae, but their movements were
more jerky and irregular than in the adult. The next day
several others came out of the brood pouch of the mother of
their own accord. They were a little more active than those
that had been removed the day before and exhibited apparently
a faint negative phototaxis. They had been out only a short
time when they began constructing nests. These nests were
the same in shape as those formed by the adult, and the
behavior of the young in relation to the nest was almost exactly
like that of the older individuals. One of these young impaled
on a needle was presented to one of the same brood lying in
its nest. At first the animal gave signs of timidity and with-
drew further into its nest. After some waiting the animal
emerged a little and began waving its antennules in the usual
manner. When they touched the food the creature darted out
quickly, seized it, dragged it back, and proceeded to devour it
at leisure. Apparently, it is only one or two days after hatching
that the young get effective control of their movements, and
they probably remain at least that long in the maternal brood
pouch. When they are sufficiently active to make their exit they
are equipped for the business of life. It was a matter of some
surprise to observe how perfectly endowed the young are with
the instincts of the parent forms. Their behavior in almost
every respect seems exactly like that of the adults. The nest-
building, movements within the nest, such as waving the anten-
nules, retraction, reversal of position, springing out after food,
jumping after passers-by, signs of timidity, as well as the
general behavior outside the nest, are all carried on just as in
individuals many times their size.

Regeneration .

No attempt was made to study the power of regeneration in
Amphithoe, but it may be worth while to record a few observa-
tions that were made incidentally on the regeneration of the
antennae. The antennae were removed in several specimens

No. 4-] AMPHITHOK LONG1MAXA SMITH. 191

while studying the reactions of the animal to olfactory stimuli,
and it was noticed that the regeneration of these appendages
took place with considerable rapidity. Several observations
were also made on specimens in which the antennae were
found to have been accidentally lost. In a specimen (Fig. 2)
observed August 13 the left antennule had been removed at
the end of the second joint, and the right one at the end of the
first. Of the second antennae the left member was off at the
end of the, fourth joint, which had a small, apparently regener-
ated knob at the end, and the right was gone at the end of the
third and had a black tip. On August 16 the specimen moulted
and the antennae appeared as follows : The right antennule
had regenerated the last two joints of the peduncle and a
flagellum of twelve joints ; the left antennule also had com-
pleted its peduncle and regen-
erated a flagellum of seven
joints, two of which were par-
tially constricted in the mid-
dle and may have later become
separated into two joints each. The second
antennae had both completed their peduncles,
the fifth joint in the right one being somewhat
longer than in the left. The left antenna as a
whole, however, was longer than the right, and its flagellum
consisted of nine joints, while that of the right was composed
of seven. How long the antennae had been injured when
the observation was first made I cannot state, but it was cer-
tainly since the previous moult, so these appendages were
probably regenerated in not much over a week and possibly
in a less time. In another specimen first observed August
13 the left antennule was gone at the end of the second
joint, and the right one at the middle of the second joint.
The left second antenna was off near the end of the fourth
joint, and the right one at the end of the third. After moult-
ing, which occurred on August 14 or 15, the right antennule
had a completed peduncle and a flagellum of six joints ; the
left antennule did not regenerate. Each of the second anten-
nae had a complete peduncle, and the left one had a flagellum

I92 HOLMES. [\'OL. II.

of three joints and a minute terminal fourth joint ; the right
one had a flagellum of four joints.

The Effect of Cutting- the Animal in Two.

It has been found in many of the lower animals that, after
removal of the posterior part of the body, the anterior portion
manifests little signs of pain and acts as if no injury had been
received. Many interesting cases of this kind have been col-
lected by Dr. Norman in his paper entitled " Do the Reactions
of the Lower Animals against Injury indicate Pain Sensa-
tions?" 1

In order to see how the removal of the posterior part of the
body affects the behavior of Amphithoe, I cut the animal in
two just behind the third thoracic segment. In the posterior
piece the thoracic legs moved but little, but the pleopods kept
up their rhythmical beating for fully a half hour. The anterior
part of the animal apparently suffered little discomfort, since,
after the operation, it behaved much as if forming a part of the
whole organism. The animal lay moving its antennules to and
fro and making the usual movements of its gnathopods, just as
an uninjured specimen would do. A piece of meat was brought
near so that it was struck by the movements of the antennules ;
it was first swept a little nearer by the second antennae, and
then quickly grabbed by the gnathopods, drawn in, and eaten,
although the food could pass through only the small part of
the alimentary canal that remained. Since this reaction in
the normal animal is readily prevented by fear, and only occurs
when the animal is hungry, it is somewhat surprising to find it
performed after such a serious injury as the loss of the poste-
rior half or more of the body. The animal behaves in such an
apparently normal manner after this operation that it would
seem as if the loss was scarcely felt. Owing to the loss of the
posterior part of the body, many of the actions of the animal
are naturally impeded or prevented. It cannot, for instance,
get meat except when it is placed quite near, as it is unable to
make its accustomed spring out of the nest. But in general,

1 Am.Jonrn. Phys. Vol. iii, p. 271. 1900.

No. 4-] AMPHITHOE LOXGIMANA SMITH. 193

so far as the actions of the animal are modified, they are
affected rather by the loss of certain organs than by any shock
to the central nervous system. The loss of blood consequent
upon the operation soon leads to weakness and finally death.
Were it not for the profuse bleeding that occurs, the separated
halves of the body could probably be kept alive for a long
period.

UNIVERSITY OF MICHIGAN,
ANN ARBOR, MICH.

Volume //.] February, ryoi. \.\ ',>.

BIOLOGICAL BULLETIN.

NOTES ON THE HABITS OE PYCNOGONIDS.1

J. COI.K.

THE Pycnogonids constitute a small and well-defined group,
especially interesting on account of their peculiar and unique
structure. In general they have been largely neglected by
naturalists, especially as compared with some other groups.
but they have received considerable attention from specialists,
and several excellent monographs have appeared dealing with
their structure and classification ; in fact, all the literature of
the group is very largely systematic or strictly morphological.
The embryology has been worked out for some forms by Dohrn
and Hoek, and in this country by Morgan,2 who has also given
an extended account of the metamorphosis of Tanystylum, and
considerations on the phylogenetic position of the Pycnogonida.3
The principal systematic work in this country has been done
by Wilson,4 who described some fifteen species found on the
New England coast. Practically nothing has been written on
their habits, which ought, it would seem from the isolated
position of these animals, to offer some very interesting com-
parisons with those of other arthropods. An excellent oppor-
tunity for this work was offered during the past summer at the

1 From the Zoological Laboratory. University of Michigan.
- Morgan. T. H.. •• A Contribution to the Embryology and Phylogeny of the
.ogonids Biol. Lab. J. Hopkins Unh-. V.. No. i. pp. 1-76. 1891.

3 Morgan. T. H.. he. cit., and "The Relationships of the Sea-Spiders," Biol.
Lfct. .^farinf Biol. Lab. for 1890. Seventh Lecture, pp. 142-167. 1891.

4 Wilson. E. B. (a) •• A Synopsis of the Pycnogonida of Xew England,"
Trans. Ccnn. Acad. V.. pp. 1-26. i $78. (p\ "The Pycnogonida of Xew Eng-
land and Adjacent Waters," C. S. Fish Com. Reft, for 1878, pp. 463-506. 1880.

'

COLE. [VOL. II.

Marine Biological Laboratory at Woods Roll, and at the sug-
gestion of Dr. S. J. Holmes, to whom I am also indebted for
much help and advice, I undertook to ascertain something of
the habits and reactions of the forms found there. The present
paper embodies some of these observations, which though far
from complete seem to be of interest. I hope to be able later
to supplement them by a more comprehensive account of the
biology of the group. Much of this work is rendered difficult
by the fact that the animals are not easy to observe under
natural conditions.

As has been noted by Morgan and other authors, there are
three species of Pycnogonids to be found at Woods Roll, repre-
senting as many genera. These peculiar animals may be found
in nearly every collection of hydroids made from the piles or
dredged up from the bottom, their long legs appearing to be
hopelessly tangled among the stems of the hydroid as they kick
slowly about in an aimless but persistent fashion. Perhaps the
commonest of these, and by considerable the largest (extent
40 mm. to 50 mm.), is the dark purple colored Anoplodactylns
Icntns Wilson (= Phoxichilidium inaxillare of Morgan and
others), which is especially abundant in colonies of Euden-
drium taken from the piles. A smaller species of a yellowish
color, Tanystylum orbiculare Wilson, measuring about 7 mm.
in extent, was found fairly abundant in a yellowish hydroid
almost the exact color of the Pycnogonid. Anoplodactylus was
on no occasion found among the light-colored hydroid, nor did
I ever find a specimen of Tanystylum among the dark colonies
of Eudendrium, where Anoplodactylus is fairly inconspicuous,
though I have found the latter among the much lighter colored
Bugula growing near the Eudendrium. I am not prepared to
say that this is a case of color adaptation, as my observations
were too limited to confirm this view, but merely throw out
the suggestion for what it is worth. And it is worth remarking
in this connection that the third representative occurring in
this locality, Pallenc brevirostris Johnston (=P. empusa Wil-
son), which is a slender whitish or more or less transparent
form and is very hard to see for this reason, was found to be
much more generally distributed than either of the other

No. 5.] THE HABITS Ol-' PYCNOGONIDS. 197

species, being found among both light and dark colored hydroids
and algae. Pallene impresses one as the smallest form of the
three owing to its extreme slenderness, though it is really
almost twice the extent of Tanystylum, measuring some 12 mm.
to i 3 mm. across.

Swimming and Crawling Movements.

/v

The activities of the three species of Pycnogonids under
consideration are in a way directly correlated with their struc-
tures. Tanystylum, a short-legged and compact form, is very
sluggish and inert ; if placed at the surface of a dish of water,
it kicks hardly at all, but sinks immediately to the bottom,1
where it does not attempt to crawl but usually draws its legs
together over its back and remains quiet. Pallene, on the
other hand, under the same conditions does not sink to the
bottom, but by vigorous kicking movements of its long slender
legs remains suspended in the water, for a considerable time at
least, its further movements being determined by the condi-
tions, one of the most important of which, as will be shown
later, is light. In the actions of Anoplodactylus there is great
individual variation, but in general it may be said that they
are intermediate between those of Tanystylum and Pallene.
Some specimens sink almost at once to the bottom, where they
rest in whatever position they may strike ; others may crawl
along upon the sand, or partly swim, touching the sand with
only the tips of certain of the legs ; or still others may swim
entirely free from the bottom. As with Pallene, just what the
animal does appears to depend largely upon the conditions.
Most of my observations were made upon Anoplodactylus, for
the reason that it was easiest to obtain and of convenient size
for observing the movements in detail.

Before going further it may be well to give a brief explana-
tion of the terminology which I shall use. Various authors
have used different names for the seven pairs of appendages

1 In these experiments the bottom of the dish was covered by a layer of fine
sand. The depth of water was usually about 5"cm. to 7 cm., though deeper water
was tried with no difference in the results.

198

COLE.

[VOL. II.

of the Pycnogonids, largely as they regarded them homologous
to the appendages of the Crustacea or to those of the Arach-
nids. Dohrn obviates this difficulty by simply numbering them
in their natural order, I-VII (Fig. i).1 For convenience I
shall speak of the third pair as the ovigerous legs (these are
absent in the female of Anoplodactylus), and of pairs IV-VII
simply as the first, second, third, and fourth pairs of legs

IV

VII

I.-IG ,_ _ Male Anoplodactylus lentns, dorsal aspect : at>., abdomen ; oc., oculiferous tubercle ;
pr., proboscis ; I, chelifori ; III, ovigerous legs ; IV-VII, walking legs, x 3.

respectively, or, to distinguish them from the ovigerous legs,
as the walking legs. Each of the walking legs is composed of
nine joints (including the terminal claw), and all four pairs are
essentially alike. As may be seen by reference to Fig. 2, the
first three joints in Anoplodactylus are short and capable of
comparatively little motion, while the fourth, fifth, and sixth
joints are long, most of the movement of the leg taking place

1 The second pair of appendages, the "palpi," are absent in Anoplodactylus
and Pallene.

No. 5-]

THE HABITS OF P

199

in these. The articulations are so arranged that there is little
chance for motion outside the vertical plane, the leg thus
moving up and down in the same plane in which it extends
from the body. There is, however, a possibility of movement
of the leg backward and forward in the horizontal plane to a
certain extent, this motion occurring chiefly in the articulation
between the first and second joints. The principal movement
of the leg, that is to say the movement in the vertical plane,
when the animal is swimming free, is shown approximately in
the diagram. Starting with the leg in the position shown at A,
and considering only that portion distal to the third joint, the
next movement is essentially a straightening out and raising

,B

J

FIG. 2. — Diagram showing movement of leg in Anoplodactylus.

dorsally of that part, bringing it into a position as shown at B.
The leg is now extended still farther and brought downward to
C, then inward, and at the same time flexed, to the original
position, A. This serves to indicate the movements in a rough
way, and it can be readily seen that so long as the specimen
is free in the water and all the eight legs are working with this
same treading motion, the tendency is to propel the animal
dorsalward, that is, in a direction perpendicular to its dorso-
ventral plane ; the fact that the legs extend radially from the
body (Fig. i) would help to keep the animal going straight in
this direction, provided tliey all beat ivith equal force. If the
claw of .any leg should grasp a solid object as it comes down
from B to C, the movement C—A would pull the animal in the
direction of that object.

200 COLE. [VOL. II.

We are now in a position to examine more carefully the
variations in the actions of the different individuals when
placed in the water. As mentioned before, if the animal treads
vigorously enough to overcome the force of gravity, it will swim ;
and so long as the body remains exactly in a horizontal posi-
tion it will only move directly up or clown according to the
vigor of the strokes ; but as soon as the body gets out of this
plane the animal will progress through the water in the direc-
tion in which its dorsal surface is turned. For it to remain in
a horizontal plane it is necessary that the legs should all beat
with equal force ; but as a matter of fact the anterior legs beat
of teucr and with more vigor than the posterior legs, thus raising
the anterior end and tilting the animal backward. I did not
make out any regular order of movement of the legs further
than this, that the posterior legs seem to lack the vigor and
strength and to be less under the control of the animal than

o

the anterior pairs. A specimen which does not tread fast
enough to raise itself from the bottom, or possibly one whose
specific gravity is greater, crawls or walks straight ahead upon
the sand, apparently, at first sight, much as an insect walks ;
but upon closer examination it may be seen that most of the
movement is accomplished by the first pair of legs, assisted to
some extent by the second, while the third and fourth pairs
seem to be a hindrance rather than a help. By reference to
Fig. 2 it can be seen how the anterior legs, by hooking into
the sand, can pull the animal forward ; while for the fourth pair
to help in the forward movement it would be necessary for
them to push, which would require a motion exactly the reverse
of that which has been described for them when free from the
bottom. Instead of this they drag along in a sort of helpless
fashion, seeming to attempt the same movement as before, but
hindered by striking the sand and by the forward movement of
the animal as a whole due to the stronger anterior legs. A
considerable backward and forward movement is now to be
observed in the second and third legs, but this is probably also
due to the pulling forward of the body after the legs are put
down onto the sand, and not to a direct action of the legs
themselves, the motion being allowed for, as before stated, by

No. 5-] TH1-: HABITS OF PYCNOGONIDS. 2OI

the specially arranged articulation between the first and second
joints. In order to prevent the specimens from swimming
while making these observations, and to force them to crawl,
a small collar of tinfoil was clasped around the body between
the second and third pairs of legs, care being taken that it was
small enough not to interfere with the movements.

Reaction to Light.

As has been stated, one of the most important factors con-
cerning the movements of the Pycnogonids when placed in the
water is the direction of the source of light. If the dish
containing them is placed near a window, the animals either
swim or crawl quickly to the light side of the dish.1 This fact
has been noted by Loeb,2 who says of Anoplodactylus : " Es ist
wie die meisten frei beweglichen Bewohner der Oberflache des
Meeres positiv heliotropisch," and he also states that when the
body was severed between the second and third pairs of legs,
the anterior portion still reacted in the same way, while the
posterior portion, which was comparatively inactive, moved
independently of the light. These results were easily verified,
and it was further ascertained that the oculiferous tubercle
(Figs. I and 4, oc.} is the photo-recipient organ, for when this
was cut off the animals failed entirely to show any response to
the light. Although there are, of course, individual variations,
it is surprising how quick this response usually is, especially
if the dish is covered above and on the sides away from the
window, so as to exclude all light from other directions. In
the diffuse light three or four feet from a northwest window,
lively specimens usually traveled an average of 12 cm. to 15 cm.
in thirty to forty seconds. The response seems to be more
pronounced when the light enters horizontally.

In moving towards the light the animals may adopt any one
of the modes of locomotion previously described : (i) they may

1 This is not the case if there are hydroid stems or similar objects which the
I'ycnogonid can grasp. The tendency to cling to anything of this character
seems to be stronger than the reaction to light.

- Loeb, J., " Bemerkungen iiber Regeneration," Arch. Entwick-Mech. Bd. ii,
pp. 250-256. 1896-97.

202 COLE. [VOL. II.

swim entirely free ; (2) they may partially swim, kicking along
on bottom with those legs that are down; or (3) they may
crawl with all the legs on the bottom. The second method
is the most common and the one in which the greatest speed
is made. But the striking thing to be noticed is that in the
first and second methods, those in which they swim or par-
tially swim, the movement is backwards, or nearly back-
wards, while in the third method of locomotion, when they
crawl on the bottom, they invariably go straight ahead, that is
to say, with the anterior end directed towards the source of
light. It thus appears that in moving towards the light they
orient themselves differently, according to whether they swim
or crawl ; and, as I have shown, whether they swim or crawl
depends directly upon the vigor or rate of the treacling move-
ment and not upon any difference in the direction of the stroke.
The question naturally presents itself, Why should there be
this difference in orientation in the two cases ? In order to
determine this, let us first consider specimens which are forced
to crawl by being weighted with tinfoil. If an animal so
weighted is placed in the water with its anterior end towards
the light, it crawls directly ahead without turning ; if its head
is pointed in any other direction, it gains this same orientation
by making a short circle, turning in the shortest direction
towards the light. Now for any animal to walk in a circular
path it is necessary for those legs on the outside of the circle
to act with a greater force than those on the inside, and thus
shove the body around ; in the case of an animal orienting
itself so as to head towards the source of light, this means that
those legs away from the light act stronger than those toivards
the light, they being the legs on the outside of the circle which
the animal describes in coming around. If this rule holds true
when the Pycnogonid swims, and I see no reason why it should
not, we have a simple explanation of its orientation with refer-
ence to light at all times. This can perhaps best be made
clear by taking a particular case and following it through.
An animal is placed on the bottom with the long axis of the
body at right angles to the rays of light. This is represented in
the diagram A (Fig. 3), in which we are supposed to be looking

Xo. 5.] THE HABITS OT P \ 'CXOC.OXI 1*S. 203

at one end of the Pycnogonid as it rests upon the sand. The
arrow indicates the direction of the light. The animal at once
begins to kick and the body is raised from the bottom, but
since those legs on the side from the light (a} beat stronger
than those towards the light (b), that side is raised more and
the body is tilted so that the rays of light strike approximately
perpendicular to the dorsal surface, as shown in B. Since the
regular movement of the legs tends to propel the animal dorsal-
ward, it moves toward the light. And now the fact that the
anterior legs beat more effi- /

ciently than the posterior must
be taken into account ; this
action tends to bring these legs A
uppermost, that is, around to
the place of the legs (a] in
diagram B. This is shown in
C, where we see the animal
from the side instead of end- B
wise, as in A and B\ in this
position the posterior legs kick
along on bottom. As a matter
of fact the anterior legs seldom
come entirely around so as to C

be directly Uppermost, but Only FlG. 3. _ Diagram representing the reaction

approximate that position, the of Anopiodactyius to light,

third and fourth legs of one side or the other being the ones
to touch bottom rather than the posterior legs ; so that the
animal does not move directly backwards but rather "corner-
wise." If the movement is more vigorous, or if the light
comes more from above, the animal may raise itself entirely
free from the bottom, keeping, however, the same relative
position.

If the orientation is different from the case given, it is easy
to see how the same result is brought about. In a case where
the animal is headed directly from the light, the anterior end
has but to raise from the bottom to bring it into this position ;
and in the other possible case, when the head is directed
towards the light, although the movements may be indefinite

204

COLE.

[VOL. II.

at first, it soon gets out of direct orientation and then turns
around the shortest way, as in the first case.

Pallene shows even a more marked positive phototaxis than
Anoplodactylus. If the light comes in nearly horizontally, it
usually tilts over to an angle of about ninety degrees, or until
the body is nearly perpendicular to the bottom, and moves
quickly towards the light by a rapid movement of the legs.
Pallene is much the better swimmer of the two and seldom
moves along the bottom in the manner described for Anoplo-
dactylus. So far as I could make out, the leg movement is
exactly similar in the two species.

«ju_£~

Transfer of the Eggs.

During the latter half of August the males of Anoplodac-
tylus may often be found bearing the egg-masses upon their
ovigerous legs. As a general rule, among Pycnogonids the
eggs are gathered into little spherical or spheroidal balls strung
along on the ovigerous legs, but in Anoplodactylus they are in
more or less irregular masses through which both of the ovig-

erous legs pass (Fig. 4) ;
their white color in this
form, clearly offset by the
dark body of the animal,
gives them much the ap-
pearance of little bunches
of wet cotton. Sars 1 says
of the genus, evidently
basing the statement on
A. pctiolatus, that there
are " several globular egg-
masses attached to the

**~«j

IM.,. J--Mul, A lent,,* from right side : walking legs I false legS in tllC male," and

'cd. Reference letters as in Fig, i.

1

jn ^Jg fjgure of this Species

there are five such masses shown on the right ovigerous leg.
It is possible that in A. Icntits the irregular masses may later

1 Sars, ('.. (>., " Pycnogonidea," The Norwegian North-Atlantic Expedition,
1876-78. Zoology, vol. vi, p. 25, and I'l. II, Fig. 2 />. 1891.

No. 5.] THE HABITS OF PYCNOGONIDS. 205

roll up into separate balls, for I have had opportunity to observe
them for only a few days after they were laid ; but this does not
seem to me probable.

Cases of the males carrying the eggs are rare among animals
and occur in widely separated groups; the male of the obstetric
toad (Alytcs obstctricans] winds the egg-strings about his body
and carries them till the tadpoles hatch ; the male of a South
American frog (Rhinoderma darwinii] takes the eggs into his
vocal sacs to develop ; and the males of some of the Lopho-
branch fishes have brood pouches for the reception of the ova.
In looking over the literature I have been unable to find any
reference to this habit among the invertebrates, aside from the
whole group of Pycnogonids. It seemed a matter of consider-
able interest to know just how such a seemingly intelligent act
as the transfer of the eggs to the male takes place in animals
whose movements in general seem to exhibit so low an order
of psychic development, and I kept close watch of them with
this point in view. So far as I am able to ascertain, the
process has never been described, though Hoek1 gives an account
of the copulation in a European species as follows : " In
regard to the way in which the eggs are laid, I had the
good fortune to observe the copulation of a male and female
PJioxichilns laevis Grube, when I was, last summer, in the
zoological station of Professor H. de Lacaze-Duthiers at Ros-
coff. The eggs are fecundated the moment they are laid, and
the copulation, therefore, is quite external, brought about by
the genital openings of the two sexes being placed against
each other. Half an hour after the beginning of copulation, the
male had a large white egg-mass on one of his ovigerous legs,
and about one hour later both masses were present." Only
once, on August 16, at 6.15 A.M., was I fortunate enough to
observe the pairing of Anoplodactylus. When first noticed
both animals were among the hydroids ; the male was clinging
to the dorsal surface of the female and headed in the same
direction. Both animals were kicking slowly in an indefinite
sort of way, but gradually the male drew forward and,

1 Iloek, P. P. C., "Report on the Pycnogonida, dredged by H. M. S. Chal-
lenger during the years 1873-76," Challenger Reports. Zoology, vol. iii, p. 131.

2O6

COLE.

[VOL. II.

passing down over the anterior end of the female, came to lie
beneath her, the animals being now headed in opposite direc-
tions and with their ventral surfaces opposed. The basal
joints of the legs of the female were approximated below, with
the mass of eggs between them. As the male came around
below the female, the ovigerous legs, which are curved at the
ends, forming a sort of hook (Fig. 5), fastened into the egg-

IV

oc

FIG. 5. - - Male A. lentus from below, showing egg-masses
on the ovigerous legs. ^i'«.¥.

masses, and as the
animals separated
pulled the eggs away
with them. The
masses did not pull
away clean, but
strung out more or
less, leaving a very
few eggs still on the
female. For some
time after they sep-
arated the male was observed to work the ovigerous legs slowly,
the effect seeming to be to get the eggs more firmly upon them
and into a more compact shape. The time from when the
animals were first observed until they had separated was only
about five minutes.

Some of the males have but one egg-mass on the ovigerous legs,
but more often there are two, as shown in Fig. 4. I am unable
to say with certainty whether this means that the male takes
the eggs from two females or that he gets them in two masses
from one; but from the fact that in those cases in which they
were examined the eggs in the two masses appeared to be in
the same stages of development, I am inclined to the latter
view. The genital openings are situated on the ventral side of
the second joint of all four pairs of legs, and it is easy to see
how the eggs of one female might gather into more than one
mass.

No. 5.] THE HABITS OF PYCNOGONIDS. 207

SUMMARY.

1. The three forms treated are more or less adapted in color
to their several habitats.

2. Swimming is accomplished by a treading movement of
the legs, which tends to propel the animal dorsalward.

3. The stroke of each of the legs is the same in character,
but is stronger in the anterior legs than in the posterior.

4. Crawling is accomplished by the same action of the legs
as swimming, when the action is not strong enough to raise
the animal from bottom. The anterior legs are most effective,
pulling the animal forward ; the action of the posterior legs is
a hindrance.

5. Both Anoplodactylus and Pallene are strongly positively
phototatic.

6. In crawling towards the light the animal proceeds with
the anterior end in advance. If not oriented in this direction
at first, it becomes so oriented by making a short circle, in
every case towards the light. This means that those legs away
from the light beat stronger than those towards the light.

7. In swimming towards the light the animal moves approxi-
mately backwards, with the anterior end somewhat raised.
The amount it raises depends upon the activity of the indi-
vidual and the slant of the rays of light.

8. This orientation is accomplished by the same actions that
produce orientation when crawling, except that they are more
vigorous, raising the animal from the bottom.

9. The transfer of the eggs from the female to the male is
a comparatively simple process.

ANN ARBOR, MICH., Nov. ^3, 1900.

EXPERIMENTS ON CUTTING OFF PARTS OF

THE COTYLEDONS OF PEA AND

NASTURTIUM SEEDS.

AI'.F.CAII, C. IUMON.

THE experiments to be described were undertaken as bear-
ing upon the general problem of the relation of food supply
to growth. They were carried on under the guidance of
Professor T. H. Morgan, to whom I am much indebted for
aid and suggestions. The variations in food supply were
produced in the pea and nasturtium, both dicotyledonous
plants, by cutting off part of the cotyledons, thereby reducing
the amount of food stored up by the parent plant for the use
of the seedling. The questions that arise relate to the effect
upon the size of the seedling, upon the differentiation of its
organs, and upon the number and size of the component cells,
caused by thus reducing the food supply. These questions may
be answered from the results of the experiments.

As already stated, the pea and the nasturtium were the
plants selected. Before deciding on them, however, the
morning-glory, sweet-pea, radish, common bean, buckwheat,
mustard, cucumber, and pumpkin were tested as to their
suitability, by planting specimens of each with portions of
their cotyledons cut off. It was found that the peas and
nasturtiums, possessing large cotyledons, were more easily
manipulated, and that their seedlings were hardier than those
of the other plants. Their seedlings, moreover, grew rapidly,
so that differences in the relative size of the plants were early
noticeable and were well marked. Under favorable conditions,
however, good results might be obtained from some of the other
species, and it would be interesting to see to what extent they
corroborate those from the two plants here discussed.

The seeds tested for availability, and subsequently all the
pea and nasturtium seeds, were treated as follows : After they

209

2IO

DIMOX.

[VOL. II.

had been soaked in water for from twelve to twenty-four hours
to soften them, their seed-coats were removed and, from some
of the seeds, parts of the cotyledons were cut off with a sharp
scalpel, while the others were left in their normal condition
save for the removal of their seed-coats. The normal seeds
underwent the soaking in water and removal of the seed-coats
to make their condition like that of the others except in the
one point of food supply. When thus prepared, all these seeds
were planted on sawdust, which was kept wet during their
growth. In the case of the pea and nasturtium, as soon as the
plants from these seeds were well started a second lot of nor-
mal seeds, treated in the same way as the normal seeds
described above, was planted. In a short time, usually about
two weeks, the plants of the second lot were found to be
about the same size as those of the first lot that had come
from the reduced seeds. Sections of the stems were then cut
freehand, and camera drawings made.

The amount of cotyledon cut off from the seeds varied very
considerably, and was not in all cases quantitatively deter-
mined. The variation may be seen from Table I, where are

TABLE I.

PART BY WEIGHT
LEFT, EXPRESSED
IN PERCENTAGES.

No. OF

PEAS.

No. OF NAS-
TURTIUMS.

j PART BY WEIGHT
LF.FT, EXPRESSED
IN PERCENTAGES.

No. OF
PEAS.

No. OF NAS-
TURTIUMS.

IO-I 5

—

2

3'~35

14

5

16-20

10

9

36-40

5

i

2I-2S

8

9

41-45

3

i

26-30

16

8

46-50

i

—

represented the percentages by weight of the part left in
the case of 57 peas and 35 nasturtiums. The seeds were
weighed after the removal of the seed-coats and again after
the removal of part of the cotyledons, and the percentages
express the ratio between the two weights. The percentages
in the case of the peas vary from 16 to 48, with more than
half the individuals between 26fo and 35^ ; those in the case

No. 5.] PEA AND NASTURTli'M .S7-V-./AS'. 211

of the nasturtiums vary from 10 to 41, with more than half
the individuals between i6'/i. and 25'^, and nearly three-
quarters between i6'/(, and 30'^. The chances, then, that
from (>5'/ to 74'^ of the bulk of any pea seed has been
removed, are even ; and the chances are three to one that
from /o'/r to 84/1- of any nasturtium seed has been removed.

The variation in the amount of cotyledon removed appeared
to influence the rate of growth of the seedling, but the num-
ber of plants of which a quantitative record of the develop-
ment was kept was too small to justify an attempt to lay down
a rule concerning the extent of this influence. The seedlings
will therefore be regarded as belonging to only two classes,
the normal and the dwarf, the latter composed of plants
growing from seeds that have been reduced by removing part
of their cotyledons. Plants of the two classes sprouted at
about the same time, and for a short time the differences
between them were not striking. As soon, however, as
leaves began to develop, the normal seedlings shot ahead,
surpassing the dwarf seedlings not only in size, but also in the
number and size of their leaves. A comparison of two groups
of pea plants, as given in Table II, from readings taken at
three different times, shows the relative rate of develop-
ment. The figures represent the height of the plants in
millimeters and their number of leaves, while at the foot of
each column is placed the average of the readings in that
column. The readings from the dwarf seeds are arranged in
the order of the fraction of the seed that is used for planting,
and the readings from the normal seeds, in the order of the
weight of the seeds, the smallest fraction and the smallest
weight being at the top of their respective columns. The first
period, two weeks after the seeds were planted, corresponds to
an early stage of development ; the second, five weeks old, is
the stage just before the production of flowers by the normal
plant ; and the third period, nine weeks old, is the period of
maturity, when the plants are bearing flowers and seeds. The
letters fl. and the word pod in the column marked "leaves,"
mean that the plant against which they are placed has a flower
or a seed-pod.

212

[VOL. II.

O

c.

T3
O

0

CA

oc

PI
O

CO
O

H
X

TJ
O
0,

Tf

-r

VO

rt
X

Pl
ro

CO

ON

0

0

s

CO
PI

i-O
M

00

0\

CO

00

w
_)

09

<!

H

CN

N
N

c\

\O

PI

O
P)

O\

No. 5-] 1'EA A XL) XASI'L'Rrii'M SEEDS. 213

From Table II we may collect several facts as to the rela-
tions in size and development between dwarf and normal
plants. In the first place, the ratio of height of dwarf to
height of normal, expressed in percentages, varies as fol-
lows : At two weeks, 6$'fo ; at five weeks, 36^ ; at nine
weeks, S2'/"- The ratios between the number of leaves of the
dwarf and of the normal are 58'^; and 56^ for the two and five
weeks old plants, respectively. No readings were made of the
number of leaves at the third stage, because some of the plants
had lost leaves, and the oldest leaves of all the plants were
beginning to wither, so that the number had not as much
significance as in the earlier stages. Similar measurements
of height and number of leaves made on another lot of peas
showed similar variations in ratios, so we may consider the
readings here given as typical. The advantage in size, then,
which the normal seedlings show in the first stage becomes
more marked as the plants grow older, while the differentia-
tion, so far as expressed by the number of leaves, though
markedly greater in the normal than in the dwarf plant, does
not increase much faster in proportion in the former than in
the latter. The time of the first appearance of flowers is, how-
ever, more significant of the differentiation of the plant than is
the number of leaves, for the latter may depend on the size of
the plant, as well as on its degree of development. The time
of flowering was unmistakably earlier in the normal plants, for
the first flower appeared upon them at five weeks, whereas the
first flowers appeared on the dwarf plants at eight weeks. At
this time three dwarf plants blossomed, while during the
same period five of the normal plants had borne flowers, of
which three had developed into pods. These facts suggest
the conclusion that the normal plants not only are larger
than are the dwarf plants but develop more quickly than do
the latter.

It remains, therefore, to be seen whether the foregoing con-
clusions to which the macroscopic examination has led are
supported by the examinations of sections under the micro-
scope. An examination of this sort may be looked to to
answer the following four questions:

214 DIMON. [VOL. II.

A. What is the relative size of the cross-section of a normal
and of a dwarf plant ?

/>'. What is the relation between the number of cells in a
normal cross-section and a dwarf one ?

C. What is the relation between a normal and a dwarf
plant, as regards the size of the cells ?

D. How does the degree of differentiation of the normal
compare with that of the dwarf plant ?

Attempts were made to compare sections of the dwarf
seedling under the microscope with sections of a normal seed-
ling of the same size, as a check, as well as with the nor-
mal seedling of the same age, but the only specimens of check
seedling available for comparison were so much larger than
the dwarf that allowance must in every case be made for a
discrepancy. The cross-sections studied were all cut free-
hand from a level less than an inch from the ground in the
growing plant.

The first question in the preceding paragraph relates to the
relative size of the cross-section of the stem of the normal
and dwarf plants. In the cross-sections represented by the
figures the diameters of the peas have the relative values of
36 and 53, or the diameter of the dwarf pea is .68 as great as
the diameter of the normal ; the diameters of the nasturtiums
have the relative values of 30 and 38, or the diameter of the
dwarf is .79 as great as that of the normal. Cross-sections of
the stem of the check plants have the values 46 and 33 for
the diameters of the pea and nasturtium, respectively. The
dwarf and normal plants were five weeks old and the check
plant two and one-half weeks old. The plants selected for
examination were typical ones, and the fact that the ratio
between the diameters does not correspond to the average
ratios between the heights given in Table II may be explained
in two ways :

i. The ratio between the heights, as was seen, decreases
with the increasing development of the plants, and the
degree of development of the plants from which the cross-
sections were taken was probably not the same as that of the
plants measured in Table II.

No. 5.] PEA AND NASTURT1LM SEEDS. 215

2. The stem of small plants is always thicker in proportion
to the size of the plant than that of large plants, so less
difference in cross-section than in height is to be expected.
Interpret this discrepancy as we may, the fact remains that
the stem of a normal plant has a greater diameter than the
stem of a plant sprung from a seed part of whose cotyledons
has been cut off.

The ratio between the number of cells in a normal cross-
section and a dwarf cross-section can be determined by count-
ing the number in a definite sector of each. The results of
counting the number of cells in a sector of 30° are : in the
normal pea (Fig. i), 410 cells ; in the dwarf pea (Fig. 2), 311
cells ; in the normal nasturtium (Fig. 3), 223 cells ; in the
dwarf nasturtium (Fig. 4), 208 cells ; in the check pea, 404
cells; in the check nasturtium, 219 cells. The dwarf plant
has, therefore, decidedly fewer cells than the normal ; in the
case of the pea .76 as many, and in the case of the nasturtium
.93 as many. If the ratio between the number of cells was
the same as the ratio between the diameters of the cross-sec-
tions, it would mean that the cells must be of the same size ;
since the former ratio is larger for both pea and nasturtium, it
means that the cells of the normal are larger than those of the
dwarf plant.

The conclusion as to the size of cells in dwarf and normal
plants may be confirmed directly by counting the number of
cells in a definite area. Proceeding to do this for Parenchyma
cells, the following results were obtained: In the normal pea
(Fig. i), 22 cells ; in the dwarf pea (Fig. 2), 43 cells ; in the
normal nasturtium (Fig. 3), 21 cells ; in the dwarf nasturtium
(Fig. 4), 32 cells ; in the check pea, 50 cells ; in the check
nasturtium, 29 cells. The regions counted were all in the
same part of the stem, and other counts were made that cor-
roborate the figures here given. These figures confirm the
conclusion reached in the preceding paragraph as to the size
of the cells in normal and dwarf plants ; but in the case of the
check pea plant it is found that the cells are smaller than those
of the dwarf plant.

Though the statistics from microscopic examination that

2l6

DIMOX.

[Voi.. II.

FIG. 2.

FIG. i.

No. 5.]

PEA AND XASTURl'U'.\r SEEDS.

2I7

¥•"'

FIG. 3.

DESCRIPTION OF FIGURES.

/../". = bast fibers. tr. = tracheid.

ph. = phloem. jrj'. = xylem.

(- = center of stem.

FIG. i. — Part of a cross-section of the stem of a normal

pea seedling.
FIG. 2. — Part of a cross-section of the stem of a dwarf

pea seedling.
FIG. 3. — Cross-section of a sector of the stem of a normal

nasturtium seedling.
FIG. 4. — Part of a cross-section of the stem of a dwarf

nasturtium seedling.

I-' it,. 4.

have been given and discussed were all drawn from three
plants of each species, nevertheless that the general results
are trustworthy is shown by another set of observations made
on different pea plants. These gave the following results :
Ratio of diameters of dwarf and normal plants, 111:62 (.56);
cells in a given sector (7°), 165 in normal and 106 in dwarf
(.64); cells in strips from center to circumference proportional
in width to the size of the cross-section, 219 and 144 respec-
tively (.66). The number of cells in a definite area was 29 in
the normal and 44 in the dwarf. These measurements con-
firm those already discussed, for they show that the normal
stem is larger in cross-section, is composed of a greater
number of cells, and of larger cells than the dwarf.

The next question to be examined is the degree of differen-
tiation of the various plants. This differentiation may be
studied in the fibre-vascular bundles, where we may note the

218

[VOL. 11.

appearance of the bundle as a whole, and the development of
its elements. The appearance as a whole is indicated diagram-
matically by the text-figures, which show especially marked
differences in the case of the nasturtium. In the dwarf nas-
turtium the fibro- vascular elements are arranged in separate
bundles around a central pith, while in the normal plant the
phloem of the different bundles has run together, making a
ring around the pith, the xylem being still discontinuous.
Bast fibers seem highly developed with thick walls, and the
tracheids are large, numerous, and clearly differentiated in the
normal nasturtium stem ; while in the dwarf stem the soft
bast has just begun to show signs of thickening into fibers,
and the tracheids are small, and comparatively few and poorly
differentiated. In both specimens of pea the fibro-vascular

Nasturtium

Pea

FIG. 5.

elements are arranged in an aggregate in the center of the
stem and in four small groups peripheral to the large group.
These bundles are more distinct and woody-looking to the
naked eye in the normal than in the dwarf plants, and both
bast fibers and tracheids are more numerous and highly differ-
entiated. The normal plant, therefore, is more highly dif-
ferentiated than the dwarf, as well as larger.

The conclusions reached from macroscopic and from micro-
scopic examinations are then in accord with one another, and
may be summarized by the statement that the removing of
part of the cotyledons of a seed retards not merely the growth
in size of the plant produced from that seed, but also its
development. The plant, however, is not the counterpart of
a younger normal plant, for it was found from comparing
dwarf plants with check plants that the dwarf plant of a
certain height was further developed than the check of the
same height. The same point is illustrated by the fact that

. 5.] PEA AND NASTURTIUM SEEDS. 219

a full-grown dwarf plant is smaller than a full-grown normal
plant, as is shown by the nine-weeks stage of Table II.

The effect, it would seem, of removing a part of its food
supply from the seed is not merely a transient one, but is one
that can be traced through the whole life of the plant, and
even increases as the plant grows older. The amount of food
supply in the cotyledons influences, perhaps, the early stages
of growth, while as the plant increases in size it becomes more
and more vigorous and tends to grow more and more rapidly;
so that a plant that is given an advantage over its fellow
at the start will increase this advantage during subsequent
development.

VARIATION AMONG HYDROMEDUSAE.

CHAS. W. HARGITT.

THE announcement of Bateson ('94), that " in the whole range
of natural history there is no more striking case of the dis-
continuity and perfection of meristic variation than in the
genus Sarsia, and the further proposition whether it is a mere
coincidence that the specimens presenting this variation, so
rare among the free-swimming Hydromedusae, should have
been members of the same genus," directed my attention to
this particular problem in conjunction with work upon this
group of coelenterates which had engaged my attention for
several years.

During the following years, therefore, collections of free
medusae of several genera were made with a view to testing
the problem raised by this observer. While as yet these col-
lections are not extensive, except in a few genera, certain facts
have been secured which may not be without value in their
general bearing upon this as well as still broader problems of
variation in general.

My collections have been restricted chiefly to the genera
Eucope, Obelia, Margelis, Pennafia, Gonionemus, Coryne
(Sarsia), and Hybocodon ; the specimens of several others
have been casually examined. Of the genera named, Obelia
has not as yet been examined in sufficient numbers and detail
to warrant any specific mention in this connection. And since
these observations have been under way a paper by Agassiz
and Woodworth ('96) on " Some Variations in the Genus
Eucope" has appeared which so fully covers the facts involved
in members of that genus, and are so coincident with my own,
that no special details will be offered in connection with it,
though the materials at hand are more abundant than upon
any of the others.

221

222 HARGITT. [ VOL. II.

What I shall have to offer in this paper, therefore, will be
upon the other genera named, namely, Coryne, Gonionemus,
Hybocodon, Pennaria, Nemopsis, and Margelis.

Coryne.

Of specimens of Coryne a comparatively few were avail-
able, though they were examined with unusual interest and
care as belonging to the genus to which, apparently among
the earliest, references to variation among Hydromedusae were
made, and which called out the rather remarkable proposition
of Bateson quoted in the opening paragraph of this paper.
While the specimens were too few to warrant any definite con-
clusions, they nevertheless showed a most remarkable con-
stancy in every morphological feature, not a single specimen
exhibiting the slightest variation in any of the more conspicu-
ous features, as tentacles, radial canals, manubrium, etc. If
this constancy is as marked in different regions of distribution
and for the large numbers cited by Bateson, it is not strange
that he should refer to the matter in the terms quoted, as it
would seem to be among the least variable of the free-swimming
medusae of this group. It will at the same time show how
very unsafe must be any such conclusion taken from so limited
a range of observation.

Hybocodon.

Of the genus Hybocodon Ag. the number of specimens at
my command has likewise been somewhat limited, slightly
less than two hundred, still they have been sufficient to show
some variation in certain features. This genus was insti-
tuted by L. Agassiz ('62), under which he included a Hydro-
medusa of very unique characters (cf. Contribution to the
Natural History of tJic United States, Vol. IV, p. 243), one of
which is the proliferous budding of medusae from the hydranth,
which in turn give rise to secondary and many later speci-
mens by a similar process of budding. (Cf. op. cit., PI. XXV,
Fig. 13.)

The specimens which came into my hands were all preserved
in formalin and had in consequence suffered considerable

No. 5.] VARIATION AMONG HYDROMEDUSAE. 223

distortion in the process, whereby minute variations of organs
often became difficult of detection, yet I was able to demon-
strate a fair degree of constancy in the general form of the
medusa, its radial canals, tentacles, etc. I desire to direct
attention to the number of tentacles. As stated by Agassi/,
there seems to be a single long tentacle arising from the mar-
gin of the bell at the terminus of one of the radial canals, from
the base of which arose later the proliferous medusae-buds, as
shown in the figure already cited. From these secondary medu-
sae other tentacles arose, giving to them the exact morpho-
logical equivalent of the primary or mother medusa. Hence, as
several of these proliferous specimens budded off from the
base of the primary tentacle, several tentacles would come to
be clustered near the same point, giving the impression of a
bunch of tentacles of the same nature. In several specimens
in which the medusae-buds had not yet appeared, or could be
detected as mere papilla-like bodies, these secondary tentacles
were nevertheless well developed, and of a length frequently
equal to that of the primary, one. Now whether this be a
variation, or whether it may not be rather fundamental, aris-
ing as a source from which the medusae are to spring, may
perhaps be an open question, to be settled by a more critical
examination of their development. Proceeding on the assump-
tion before stated, I venture to cite it as a case of varia-
tion, though it may later be found to be rather the normal
process.

The rather unusual character of this medusa, both in its
origin and proliferous progeny, led me to suspect that it might
exhibit more than the usual phases of variation ; but in this I
have been disappointed, except in the point just cited, - - its con-
stancy in almost every morphological detail being quite marked.
As stated, however, in connection with observations upon
Coryne, the limited number of specimens examined, and further-
more their distortion due to preservation, are barriers which
should suggest reasonable caution in the formulation of any
conclusion.

Pennaria. - - Of this medusa I have had an almost unlimited
number of specimens, having collected them during three

224 HARGITT. [VOL. II.

years devoted to the embryology of the species common in the
waters of the Massachusetts coast,-- P. Tiarella McCr. A criti-
cal study of these medusae is, however, rendered difficult and
tedious owing to their minuteness and form. In size they
are only about .8 mm. in diameter by about 1.5 mm. in length.
The highly oval form renders difficult a study of the aboral
surface and the junction of the radial, or chymiferous canals,
-a point of considerable variability in many cases in other
genera, notably Gonionetnus, to be noted later. In a study of
their morphology Smallwood ('99) has pointed out the variability
in the structure and development of these canals. He has
shown that in a considerable proportion of specimens there is
a tendency to atrophy both in the radial and circumferential
canals, especially the latter. These changes are not evident
in a surface study of specimens, the pigmentation which marks
their course being fairly constant. The principal variation to
which I desire to direct attention in this connection is a physi-
ological one, viz., a rather marked variation in habit and activity.
I have discussed elsewhere this feature ('00) and need only refer
here in brief to those observations. As there pointed out, there
seem to be two rather distinct features of habit ; namely, a
rather deep-water habit upon rocks, seaweed, piles, etc., and a
surface habit upon eel-grass or similar support, which serves to
bring the colonies to the surface, thus often in a low tide
exposing them directly to the action of the midsummer sun
and temperature.

Associated with these differences are correlated variations
in the form and color of the colonies, or, as Bateson would
designate them, " substantive variation." The surface or eel-
grass varieties exhibit more distinctly the pinnatifid character
which marks its specific peculiarity, due doubtless, in part at
least, to the prone or floating disposition of the colonies.
Associated also with this is the much higher coloration so con-
spicuous in these specimens, a variation extending not only to
the perisarc of the colonies but also to the medusae and the
eggs, which are rather bright orange, while those taken from
the deeper waters are a pale, creamy white, with the slightest
trace of pink in many cases.

No. 5.] VARIATION AMONG HYDROM EDTS. //•;. 225

Of the further physiological differences one of the most
marked is that of the relative activities of the medusae of the
afore-mentioned varieties ; those of the surface habit exhibit-
ing a much greater degree of activity and other vital phe-
nomena. These, as previously pointed out, are extremely
active, being liberated from the hydranths promptly upon
maturity, swimming with great ease and freedom, and discharg-
ing the sexual products with great promptness. On the other
hand those of the deep-water habit are passive, or even sluggish,
- in many cases the medusae never becoming free from the
hydranth, - - discharging the sexual products with much less
regularity and ease, and dying very soon after. These medusae
are short lived at best and never increase in size after libera-
tion from the hydroid. I would suggest the probable correla-
tion of some of these features of variation with the degenerative
tendency shown in both structural and physiological variations
already noted, especially in the atrophy of the chymiferous
canals.

A histological study of the tentaculocysts likewise shows
degenerative tendencies, as does also the very rudimentary
condition of the tentacles, which are barely distinguished as
bud-like protuberances upon the margins of the bell.

In connection with previous work upon the development of
Pennaria, attention was directed to variation in the rate of
cleavage and subsequent development. This would seem to
be a matter of considerable interest in connection with the
fundamental problems of physiological variation. It is well
known, of course, that cleavage is a phenomenon subject to con-
siderable variation as to rate, due to variable conditions, and
to some extent independent of sensible differences of environ-
ment. It seems to me, however, that in the case of Pennaria
there are presented such marked extremes in this respect that
it may well be considered as in some measure correlated with
other features of physiological variation.

The variable rate in the later phases of larval development
is also worthy of note. From data obtained during three sum-
mers of observation the range of time involved in the larval
history varies from about two days up to about two weeks.

226

HARGITT.

[VOL. II.

While in most cases these observations were made upon speci-
mens under artificial conditions, namely, aquaria of variable
sizes, etc., still the variations occurring were exhibited by
larvae under identical conditions, such as they were.

Under the head of abnormalities in immediate connection
with these observations, attention was also directed to certain
variations in the morphology of the larvae and early polyps.
Among these may be mentioned

1. Twin-planulae, - - planulae with bifurcated ends, irregular
bud-like outgrowths, etc. (Cf. op. cit., Figs. 4-6, and 8, PL i.)
It was suggested that they were probably due " to the intrin-
sic prepotency of hydroids to bud and branch." While this
is probably an explanation of the facts, that they exhibit
interesting variations from the ordinary is not discredited on
that account.

2. Attention was also directed to an interesting polyp form
(op. cit., Fig. i of text), which presented so marked a varia-
tion as to give rise to some doubt concerning its Pennarian
affinities. In view of the rather large range of variability
exhibited by the medusoid and larval persons already consid-
ered, I am still convinced that this is only a further illustration
of the same principle. Indeed, I have during the present sum-
mer observed in other polyps reared under similar conditions
the same variation from total annulations to less and less
degrees. A few additional annulations of the hydroid perisarc
is matter of no special surprise. A complete annulation of
the early and plastic colony, while quite unusual, need not be

regarded as improbable or especially strange.
Another feature which may perhaps rather
be designated as a monstrosity, or incidental
excrescence, may be noted in this connection ;
namely, certain wartlike or pustular vesicles
which often appear at various points on the
exumbrella of the medusae. These are fairly
represented in Fig. i. The figure indicates
relative positions where they are most likely to occur, though
in no case have I noted more than one upon a given specimen.
... A similar structure is referred to by Agassiz ('65) and

IMC;. T.

No. 5-] VARIATION AMONG HYDROMEDUSAE. 227

explained as due to distortion caused by the pressure of the
ova within the bell. This I am convinced is a mistaken virw,
for I have noted it upon specimens both living and on those
killed and preserved in formalin, in specimens with and with-
out eggs. It seems, moreover, to be wholly restricted to the
outer ectoderm only, in no case involving the inner ectoderm of
the subumbrella. There is nothing indicative of the cause or
character of these excrescences. Whether they are permanent
or merely transient features I am not able to say, the short-
lived condition rendering any determination difficult if not
impossible.

Ncmopsis.

Through the courtesy of Mr. Strong I had the privilege
of examining a small collection of about one hundred speci-
mens of Nemopsis Bachei taken in the tow off Tarpaulin
Cove. The variations here seemed quite as evident as in
Eucope and Gonionemus. Here again the variable features
included radial canals, manubrium, gonads, and tentacles.
Fully five per cent showed some feature of variation. About
two per cent had but three radials and three gonads. One of
these showed a definite correlation, including all the organs
named above. One, however, of the trimerous forms had a
fourth sensory bulb and tentacles, though these were less
prominent than were the other three sets. The oral tenta-
cles likewise shared in the correlation.

One specimen was a symmetrically pentamerous form with
a perfect correlation of all the organs under consideration.
Another specimen was quite as symmetrically hexamerous.

Several other specimens exhibited apparent symmetry in the
number of gonads. Frequently one of the series showed the
gonad of one canal very unequally developed as compared
with the others. But while approaching sexual maturity, and
in many cases fully so, it is of course impossible to say with
certainty that the short gonad might not have shown further
development with age. In any case it certainly showed varia-
tion as to development.

228 HARGITT. [VOL. II.

Concerning variation in the number and order of tentacles
it is difficult to determine definitely, since in Nemopsis they
constantly increase in number as the medusa grows, much as
in Margelis. So while there appears to be considerable varia-
tion in the number and arrangement it may be rather due to
variable development than to any actual meristic variation.
The same may be true as to the order of appearance. The
paired, capitate tentacles at the apex of the bulb appear uni-
formly first and seem to be fairly constant. The latter fila-
mentous tentacles appear to arise in pairs successively toward
the margins of the bulb. Since I was not able to follow this
development in the stages of growth of the medusa it is impos-
sible to determine definitely this point. So while there is upon
the preserved specimens considerable want of symmetry in
this respect, yet it may be due in part to slightly variable rates
of development. No constancy was apparent in the matter,
and it would seem therefore to be physiological rather than
morphological.

Similarly there was apparently some variation as to the
number and distinctness of the otocysts upon the sensory
bulbs. Normally there is a single eye-spot at the base of each
tentacle. But in many cases they were apparently absent.
And while it is not impossible that they had been rendered
indistinct by the formalin in which they were preserved, still it
remains quite certain that marked differences were distinguish-
able among various specimens of similar size and preservation,
and perhaps only critical histological examination will be ade-
quate to finally determine this point, and this I have not been
able to make.

Margelis.

Of these medusae more than five hundred specimens have
been examined, most of which were quite young, having only
the four pairs of marginal tentacles and four unbranched oral
tentacles and measuring only about 0.5 mm. in diameter. Typi-
cally this medusa may be characterized as having a high hemi-
spherical bell, four radial canals, at the distal or marginal ends
of which four clusters of filiform tentacles arise. The bell is

No. 5.] VARIATION AMOXC, 11 \ Y'A'( >M EDUSA /:'.

229

I-'u,. 2.

the

thick, velum not specially prominent. Manubrium subconical,
bearing four oral tentacles which divide dichotomously into

small clusters of tentacles. Figs.
2-4 give good general impressions
of the animal. The very minute
size made necessary the constant
use of the compound microscope
in all the examinations.

In general this medusa seems
to be fairly constant in form,
color, and organic symmetry.
Only the radial canals, and tenta-
cles - - marginal and oral - - were
noted, no gonads being present as
yet. Of the tentacles note was
taken of the number of groups,
number of individual tenta-
cles increasing in these with
age. The total of all sorts
of variations noted hardly

\exceecled two per cent.

KK;. 3.

The variation in radial
canals was usually corre-
lated with that of the ten-
tacles. In Fig. 2 is shown
a trimerous specimen, in
which there was a perfect
correlation of all the or-
gans, including oral tenta-
cles (several specimens
noted). In Figs. 3 and 4
is shown a pentamerous
form, in which there ap-
peared little meristic cor-
relation. For example, it will be seen that while five sets of
tentacles correspond with the five radial canals, two had but a

FIG.

230

HARGITT.

[VOL. II.

FIG. 5.

single tentacle each, while the other three sets contained the
normal (at this stage) number, two. A point not shown
definitely in the figure is the fact of only four oral tentacles.

In Fig. 5 is shown a specimen
with monstrously developed
manubrium, protruding beyond
the velum, and provided with but
two tentacles. In several speci-
mens similar enlargements were
noted, though not so pronounced
as in this, except in certain cases,
apparently pathological, in which
the entire bell was evaginated and
greatly shrunken, with the manu-
brium greatly enlarged. No account was taken of such
specimens, for they -were evidently due to conditions other
than those of health. It may be noted in this connection that
several specimens were found exhibiting similar vesicular or
pustular enlargement to those observed in connection with
accounts of Pennaria. Here is further evidence, if such were
necessary, that these enlargements, resembling distortions,
could not have been produced by the enlarged gonads, for in
the species under consideration the
sexual organs were as yet unde-
veloped.

In Fig. 6 is shown in diagram an
aboral view of a condition found in
several specimens, in which there
seem to be secondary, peripherally
directed radial canals, extending
nearly half-way over the bell. As
this medusa has normally but four
such canals this is a well-marked
case of variation in the direction of a condition quite common
in Rhegmatodes and many trachomedusae. I regret that
none of the specimens to which I have had access are of
approximate maturity so that such incipient variations might
be traced forward to their perfection, in order to ascertain

FIG. 6.

No. 5.] VARIATION AMONu HYDROMEDUSAE.

231

whether additional sets of tentacles, etc., would be found
correlated therewith.

The origin and symmetrical interpolation of such secondary
canals is very different, it seems to me, from that arising from
the bifurcation seen in Eucope and Gonionemus, though their
function in the economy of the organism may be the same.

Only one phenomenon more will be discussed in this con-
nection, namely, that of double or twin medusae, which is
shown in Fig. 7. Only one specimen of this character was
found in the entire lot. In every respect --size, general form,
organic relations, etc., the
double feature alone ex-
cepted --the specimen
seemed fairly normal, hav-
ing this one further feature,
that one was a tri-
merous specimen
and that one set of
tentacles contained
three as against the
two each of the other sets. The furrow:like depression along
the line of union presents the aspects of late coalescence
similar to that involved in artificial grafting (cf. Biological
Bulletin, Vol. I, p. 41).

But a more critical examination shows that the union is
much more profound, involving the gastric cavity and there-
fore the whole chymiferous system.

The specimen was not seen alive, and hence nothing can
be said as to coordination of physiological activities, mode of
progression, etc. But from what has been proved from ex-
perimentation on these points (cf. op. cit.} it may safely be
inferred that similar coordination at least prevailed in such a
case as that under consideration.

FIG. 7.

Podocoryne.

Of medusae of this genus I have had at my command only
a few more than one hundred specimens, a number too small

232 HARGITT, [VOL. II.

to warrant any formal deductions, but taken in connection with
others it will not be amiss as showing the facts of variation
pertaining to this genus.

Like Margelis this medusa is quite small when liberated,
indeed never attains a size of more than i mm. in diameter, so
far as I am aware. In general, it is similar in organization to
Margelis, has four radial canals and four primary tentacles,
between which a second series of the same number soon arises.
Like Margelis there is apparently a fair degree of constancy in
meristic features. Three trimerous specimens occurred in the
lot, which comprised the extent of variations along this line.
There were, however, considerable variations of size, form, etc.,
which increased the total to at least five per cent. It must
not be forgotten, however, that the extreme minuteness of
specimens, necessitating the constant use of the microscope,
might easily involve an oversight of specimens in sufficient
numbers to materially raise this ratio of variation.

Gonionemus.

It is among the members of this genus that I have been able
to work out the most numerous and detailed series of varia-
tions. The total number of specimens of Gonionemus exam-
ined during the progress of the work was more than two
thousand. (Of the first series studied no record was kept.)
While smaller than the number of Eucope, to which reference
has been made, the number is yet sufficiently large to insure
against an unusual per cent which might be clue to local or
other incidental influences. Moreover, the collections were
made during three summers, and by several collectors, so that
the results obtained may be considered as closely approximat-
ing the actual state of variation now under way. It is a pleas-
ure to acknowledge in this connection courtesies from Messrs.
Coe, Parker, Perkins, Gray, and others for permission to exam-
ine collections of these, medusae made by them, which has
facilitated my work.

In these studies attention has been directed chiefly to the
following structures : i, radial canals; 2, gonads ; 3, manubrium ;

NO. 5-] r.iAv.rr/o.v .i.uo. \\; HYDROMEDUSAE. 233

4, tentacles; 5, otocysts. Other features of subsidiary nature
will have mention in their appropriate places.

It ought to be stated in passing to details and tabulation of
results that only in the matter of radial canals and gonads
have the entire two thousand specimens been examined, and
not all of those with equal detail in each case. In number
and variation of tentacles, spurs, anastomosing of canals, etc.,
the tabulations were limited to one thousand ; and in the case
of otocysts to less than one hundred, a reason for which
will be given in the appropriate place. I may also state that
owing to the insignificant sexual differences, usually requiring
microscopic examination to certainly decide, no effort has been
made to determine the relation of variation to sex.

Gonionemus is a rather well-defined trachomedusa, first defi-
nitely described from the Atlantic coast by Murbach ('95). It
is characterized by the typical four radial canals, which have
the folded gonads suspended along their under surfaces, cruci-
form manubrium, as seen in transverse section, normally with
four oral lobes and sinuously folded lips. Tentacles are
numerous and similar, each characterized by a suctorial bulb
near the tip, beyond which it often makes a sharp bend.

These medusae abound from June to October in a small pond
near the Marine Laboratory, known as the "eel pond," which
communicates with the open harbor by a narrow inlet. Lately
they have been taken in the outer harbor, though in small
numbers. I may mention this fact of the localized habitat,
since it may well be a question whether its peculiarity may not
be an important factor in the physiological aspects of the varia-
tions to be considered.

Taking up now the consideration of the several points in the
order given, attention will first be directed to the

Radial Canals.

On this point an estimation of the ratio of numerical varia-
tion based upon fifteen hundred specimens gave nearly five
per cent (4.82). On the matter of their form or disposition,
i.e., whether in their course to the marginal canal the bells

234

H ARC ITT.

[VOL. II.

were divided into symmetrical segments, the results showed
no less than thirty per cent of variation. On the variations
shown in their aboral confluence, or union with the gastric
pouch, the calculations gave 14.4 per cent.

Of the entire number examined only a single specimen was
found having but two radial canals and two gonacls. These
were at an angle of 180 degrees, i.e., dividing the bell into two
symmetrical halves, as shown in Fig. 8, excepting alone the rela-

FIG. 8.

tive number of tentacles upon each half. One other specimen,
however, was found having only two canals, similarly disposed
and with a similar number and disposition of gonads ; but in it
there was a rather evident rudiment of a third springing from
the peripheral end of one of the canals, thus destroying the
bilateralism characteristic of the first.

The largest number of canals found was six. This, while
much more common than those with two, was much less so
than specimens with three and five. Seven specimens in all
were found having this number, and in one of these the sixth
was clue to the evident forking of one of the five apparently

No. 5.] VARIATION AMONG HYDROMEDUSAE. 235

primary canals, which divided the bell into approximately pen-
tamerous segments, as shown in PI. Ill, Fig. 12. In another
there was a very evident forking of two of the four primary
canals, as shown in PI. Ill, Fig. 9 ; for while the bell was
divided into hexamerous segments, the manubrium was sym-
metrically tetramerous. In every one of the other five hexam-
erous specimens the canals converged at the aboral pole in a
perfectly symmetrical way, though the hexamerism extended
to the manubrium in only two specimens, and in these only

FK,. 9. — • Hexamerous specimen showing pentamerous stomach and varying size and

distribution of tentacles.

the basal portion or gastric pouch was strictly hexamerous. In
at least two other of these hexamerous specimens the oral
lobes of the manubrium were four (cf. Fig. 9).

Of specimens having three and five canals there were by far
the larger number, with the preponderance slightly in favor of
the trimerous variety, but not sufficiently so to warrant any
conclusions as to the question whether the course of variation
was toward a trimerous rather than a pentamerous condition.
Of those with three canals there were twenty-one specimens,
while of those with five there were eighteen specimens. Of
those making up the total of seventy-two specimens there were

236

H ARC ITT.

[VOL. II.

twenty-five specimens from distinctly tetramerous forms, hav-
ing short forks less than one-third the length of the entire
canal, or of such other variable aspects as to warrant their
inclusion under this head. PI. II, Figs. 6-9, and PL III, Figs.
1—5, will best illustrate this point.

It now remains to consider somewhat more in detail the
individual variations exhibited by these several types. Direct-
ing attention first to those illustrated by the figures just cited,
it may be seen that the variation here seems to be in several

FIG. 10. — Symmetrically pentamerous specimen, but with the several series of
tentacles appearing at irregular intervals.

directions, (i) Atrophy, as shown in PI. II, Figs. 4, 6, 8, and 9.
In all these the evidences of degeneration are quite clear.
First, in Figs. 6 and 9 there is the atrophy of the connection
of one of the canals with the gastric pouch and the correlated
reduction of the fourth gastric pouch and the further failure
of the obsolescent canal to develop its visual gonad, the merest
rudiments of which are apparent. Furthermore, in the same
figure there is shown a still further atrophy of a second canal,
which extends only about halfway to the margin, and corre-
lated with that fact is the associated imperfect development of
its gonad. A still further illustration of this degenerative

No. 5-] VARIATION AMOXG IIYDKOMEDUSA /•/.

237

tendency is shown in PI. Ill, Fig-. 6, where only the vestige of
the fourth canal is shown, the reduction in extent of two
others, with the further correlation of the evidently bilobed
condition of the gastric pouch.

(2) Asymmetry. This is more or less consequent upon the
atrophy already noted, as will be seen from a comparison of the
figures just cited, and involves to a certain degree the entire
organism, gastric and oral symmetry, no less than that of the

FIG. ii. — Tetramerous specimen of very unsyfnmetrical type.

bell and tentacles. Some further reflections on these lines will
more naturally come up in connection with later discussions.

Another type of variation is shown in PI. Ill, Figs. 1-5. In
these specimens, while the tetramerous type is more or less
evident from the number of canals, gonads, or gastric pouches,
still there is a rather definite tendency toward a trimerous
aspect of the medusa as a whole, so far as the segmentation of
the body is concerned. In Fig. I, while there is a clearly
tetramerous condition exhibited which extends to the several
organs involved, there is yet such an approximation of those

238

HARGITT.

[VOL. II.

marked a and b as to leave the bell and number of tentacles in
a closely trimerous symmetry. In Figs. 3-5 this evolution
of trimerism is so evident that it would seem to point toward
a preponderance of variation in this direction. As, however,
will be seen later, facts of a very different kind seem to point
as clearly in the opposite direction. It may as well be pointed
out in this connection that the loopings of canals shown in
the figures under consideration are variously simulated by

FIG. 12. — Specimen of very unsymmetrical character.

structures shown in PI. II, Figs. 1-4, and 10. A critical
comparison however, while showing many unusual features
in these latter structures, will probably demonstrate their
fundamental likeness ; but this will be considered later.

We may next consider a type of variation fairly illustrated
by PI. Ill, Figs. 9 and 11. As will be seen from a glance,
there is here exhibited a clearly defined tendency toward an
increase in the number of canals, hardly less marked than that
of decrease just considered. Indeed, specimens with bifurcated
canals of this character were rather more common than those

No. 5-] VARIATION AMONG HYDROMEDUSAE. 239

of the last type, a fact, when taken in connection with the
closely similar number of distinctly pentamerous forms, of
great importance as showing that in neither case are we war-
ranted in concluding that the course of variation is in one
direction rather than in another. In order to reach anything
approximating conclusiveness on this point a larger number of
specimens studied through a successive series of years would
be necessary. As I have already intimated, the collections
forming the basis of the present discussion of this genus were
made during at least four years, and while I have not made
this a matter of critical comparison, it has not been at all
apparent that during this period there has been any appreci-
able ratio of difference.

Passing now to the consideration of other aspects of varia-
tion evident in the canals, attention is next directed to their
morphology. As is well known, the chymiferous canals in
medusae are tubular structures of fairly constant size in
members of the same species and of similar sizes, and their
courses are usually direct from the center to the margin in
most of the Hydromedusae. As Agassiz and Woodworth
('96) have shown in the case of Eucope, however, there are
not a few departures from this rule. The same is true of
Gonionemus, as a glance at Pis. I and II will demonstrate. Not
only does the diameter of the canal vary greatly in many speci-
mens, which is of only incidental concern, but in many cases,
as in PI. I, Figs. 9, 1 1, and 12, various loops and diverticula in
the form of spurs are formed at various points and at various
angles along their course. These are of varying sizes, lengths,
etc., and were found on between one and two per cent of all
the specimens examined. In the paper just cited the authors
suggest in these facts a possible simulation of a condition
"characteristic of the Discophores " (p. 122). Whether there
is in these structures anything more than simulation or paral-
lelism as compared with the Scyphomedusae, I shall not at
present discuss.1 As compared with the typical canal, however,

1 It may not be amiss, however, to state in this connection that in Rhegma-
todes there is a much more evident correspondence or resemblance in this matter
than in either Gonionemus or Eucope. While possessing a large number of

240 HARGITT. [VOL. II.

there seems to me to be little doubt that they are funda-
mentally of similar origin and function. While in many cases
there are extremely small serrations of the canal walls, in
other cases (PI. I, Figs. I and 2 ; PI. Ill, Figs. 4 and 5) they are
more prominent, even occasionally forming anastomosing con-
nections between adjacent canals. Similarly the loops already
referred to are probably in most cases anastomosed spurs.

An examination of the several figures of Pis. I-III will
bring to the attention some interesting and rather anoma-
lous illustrations of another phase of the structures under con-
sideration. As will be seen, there is here almost every degree
of intergradation between the perfectly symmetrical cruciform
aboral junction of the chymiferous canals and the perfectly
circular canal about the base of the gastric pouch into which
the radial s connect before their connection with the gastric
cavity. By careful injections through the radial canals I have
clearly demonstrated a direct continuity of the chymiferous
system throughout these several channels. Little doubt can
therefore remain concerning the fundamentally similar charac-
ter of these various structures. Nor is it more doubtful that
in function they are fundamentally similar ; and while con-
cerning the question of their significance in relation to the
affinities of the Hydro- and Scyphomedusae there may be room
for wide difference of view, that they serve similar functions
in both is highly probable, if not quite certain.

In passing to the consideration of a specimen of unusual
form, it should be noted that in the origin of spurs, extra
canals, etc., they were with very slight exceptions, which seem
to me easily explained, centrifugal, i.e., from the central toward
the peripheral portions of the body. The apparent exceptions
are shown in PI. II, Figs. 4 and 7, where portions of canals
extend from the margin toward the center. As will be noted,
however, there are in both cases spurs from the central region
in the line of the peripheral branches which would strongly

radial canals, many of them show bifurcations toward the margin, and in not a
few cases are there found centripetally developing canals similar to those of Car-
marina. This medusa likewise shows many other phases of variation, spurs, anas-
tonmsi •**. etc., of canals, but no details will he undertaken in this connection.

No. 5-] VARIATION A. \IOi\G HYDROMEDUSAE.

24I

suggest that there had been complete unions of these partial
canals at an earlier stage and that the present condition was
the result of atrophy such as is shown in Figs. 6 and 9. It
would seem therefore quite just to conclude that these several
structures, spurs, partial canals, loops, etc., have had their
development usually from the central pouches or canals and
not from the peripheral or marginal canal.

The unusual specimen, to which reference is made in this
connection, is shown in Fig. 13 of the text. It would appear
to partake somewhat of the nature
of a monstrosity and in some re-
spects of the nature of a marginal
bud, suggestive of a secondary
medusa. Aside from the general
form there is little to confirm this
possibility ; there is no sign of
manubrium ; and the canals and
tentacles are quite continuous with
those of the primary medusa. As
will be noted, there are vestiges of
gonads upon the peripheral termination of the median canal,
while the branches are wholly devoid of any signs of such
structures. Only a single specimen of this character was
found and it exhibits another aspect of erratic variation.

TIG. 13.

Gonads.

In the comparisons of gonads only specimens apparently
sexually mature were taken (as noted before, no distinction
was made between sexes). In the cases wholly devoid of
gonads the size and other organic conditions were considered
as sufficient to warrant the conclusion that they were probably
of such age and general development as are usually correlated
with perfect sexual maturity. In the whole number of speci-
mens examined 3.6 per cent showed numerical variation of the
gonads ; of specimens with less than the normal number, two
per cent; of those with more than the normal, 1.2 per cent;
of specimens without trace of gonads, .4 per cent. As will be

242 HARGITT. [¥OL. II.

seen in comparing these ratios with those concerning the radial
canals, there is here again a slight tendency toward the smaller
or trimerous forms, though not specially marked, especially
when account is taken of the fact that only on specimens with
more than the normal canals would additional gonads be
found, while it was not rare to find in tetramerous forms speci-
mens with only three gonads, or even less, one tetramerous
specimen having a single gonad. Concerning variations of the
several gonads of individual specimens no account was taken,
owing to the difficulty of determining relative differences in
organs loosely suspended in sinuous folds, as are these in Goni-
onemus, and by the further fact of the continued growth and
successive discharge of the sexual products, as seems to be the
case here.

Manubriitni.

As in most medusae, the manubrium is a rather prominent
and important organ. In correlation with the tetramerous
organization of the medusa, the manubrium, including in this
general term the basal gastric pouch and oral opening and lobes,
is of similar form and adjustment. As will be noted, how-
ever, by a comparison of the several tables, there are many
exceptions, or, in other words, considerable variation. In most
cases, however, as comparison will show, there is in the varia-
tion an obvious correlation with other variations, notably with
that involving the radial canals. But here again the exceptions
are sufficiently numerous to warrant the conclusion that there
is in this organ itself individual variation, apparently devoid of
any adaptive end or relation.

Aside from the facts of meristic nature above noted, there
are features of variation which would seem to be of a purely
substantive character. For example, in several specimens the
manubrium was greatly extended lengthwise, reaching in some
cases quite beyond the velum, occasionally as much as one-
fourth its total length. While of course this organ is very
extensile, yet in many hundreds of specimens examined alive,
in many cases while the animal was engaged in engulfing food,
I have never seen the manubrium extended beyond the velum.

No. 5-] VARIATION AMONG HYDROMEDUSAE. 243

While no emphasis is placed on this feature of variation, it is
yet worthy of note in comparison with such medusae as
Coryne, Dipurena, etc., in which the greatly elongated manu-
brium feature is rather distinctive.

In Fig. 14 is shown an interesting and anomalous feature
which is more or less monstrous, namely, a spike-like growth
from one side of the basal portion of the
manubrium. While in some respects it
might be comparable with an oral tentacle
of Margelis or Nemopsis, still only in a
somewhat remote way. It was rather
rigid, yet devoid of any chitinous or other Fu- M. --Semidi.ls,ammatk

sketch of manubrium of

rigid support. As will be seen, it has the medusa showing anomalous

r r i projecting spur, S.

form or an elongate, attenuate process,

about twice the length of the manubrium. It would seem, as
suggested above, to be a wholly unique if not anomalous struc-
ture, without evident correlation with, or adaptation to, any
other organ or function.

Tentacles.

As compared with Eucope, Obelia, Podocoryne, and many
other genera, there seems to be a very different order and
relationship among the tentacles of Gonionemus. In the small-
est specimen measured the diameter was but 2 mm., and the
number of tentacles twenty-nine. The largest specimen found
measured 19 mm., and the number of tentacles was sixty-eight.
That mere size is not, however, determinant of numbers will
be seen when it is stated that a specimen measuring 4 mm.
had thirty-nine tentacles, while one measuring 6 mm. had
but thirty. While, as stated above, the largest measured speci-
men had sixty-eight tentacles, two others measuring 15 and 16
mm., respectively, had each seventy-two tentacles, and a speci-
men measuring 14 mm. had seventy-one tentacles. While it
should be stated that these observations were made upon speci-
mens preserved in formaldehyde, which may have thereby suf-
fered some shrinkage, still since the preservative was in all
cases the same medium and of the same, or very nearly the
same, per cent, they would presumably be similarly affected.

244 HARGITT. [VOL. II.

However, the matter is not in any wise dependent upon data
of this character. Even a glance at Figs. 9-12 will show,
though diagrammatically, the relative number and distribution
of the tentacles about the margin, while an inspection of the
tables will show how very variable is this matter.

Bifurcation of tentacles, tentacular spurs, etc. - - In all some
fifteen specimens were found having variations involving one
or more of the features indicated under this head. As noted
by Agassiz and Woodworth ('96) in Eucope, the origin of
spurs is usually from the base, as is also the doubling of the
tentacles, as shown in PI. IV, Figs. 2, 7, and 10. In several
specimens there was an evident bifurcation of the terminal
portion, as shown in PI. IV, Figs. 5, 6, 8, and 9. In the speci-
mens shown in PL IV, Fig. 6, this had occurred in close
conjunction with the peculiar suctorial bulbs or pads so char-
acteristic of this genus, while in Fig. 8 it is shown as having
occurred somewhat proximal to these structures. A single
specimen was found having three of these organs on a given
tentacle at considerable intervals. In several specimens the
tentacular pads or bulbs associated with the bases of the tenta-
cles exhibited peculiar cordate lobings, sometimes on the outer
border, more frequently on the inner edge, or from both, as if
about to divide, though in no case was division found to be
complete in a given bulb. However, this feature will acquire
some significance in relation to the double and triple tentacles
shown in Fig. 2, where the pads are correspondingly double
and triple.

As compared with the specimens of Figs. 7 and 10, however,
there will be seen no such correlation, — a fact which would sug-
gest a measure of caution concerning the possible relation of
the apparent division of the basal pads and the doubling of
tentacles. This caution is further emphasized by the fact that
in their origin new tentacles appear wholly apart from these
pads, which only after some time are gradually developed on
their ventral bases.

I am unable to agree with Agassiz and Woodworth (op cit.,
p. 139) that these double and triple tentacles are due to coa-
lescence of the bases. Whatever may be the case with Eucope,

No. 5.] VARIATION AMONG HYDROMEDUSAE.

245

TAHI.K 1. TKTRAMKROUS $I>KCIMK.\S.

RADIAL
CANALS.

( HiXADS.

( IASTKK
1 ."HES.

1 ll'A [. KoHKs.

TENTACLES.

4

4

4

4

12, 12, 12, 12.

4

4

4

4

12, 12, 15, 16.

4

4

4

4

17, 15, 14, 5.

4

4

4

4

15, 14, 14, 14.

4

4

4

4

12, 11, 13, 13.

4

3

o
o

-»

.)

20, 18, 10, 13.

4

4

4

4

10, 10, 11, S.

4

4

4

4

16, 15, 17, 17.

4

4

4

4

14, 12, 12, 11.

4

4

4

4

16, 16, 16, 16.

4

4

4

4

14, 14, 14, 14.

4

4

4

4

15, 15, 15, 15.

4

4

4

4

14, 14, 14, 14.

4

4

4

4

12, 12, 12, 8.

4

4

4

4

17, 17, 21, 14.

4

4

4

4

17, 17, 19, 17.

4

4

4

4

16, 16, 16, 16.

4

4

4

4

11, 11, 11, 11.

4

4

4

i
6

15, 15, 16, 16.

4

4

4

4

12, 11, 12, 10.

4

4

4

4

13, 12, 12, 11.

4

4

4

4

16, 13, 15, 14.

4

4

4

4

15, 15, 15, 15.

4

3

4

4

13, 13, 10, 9.

4

4

4

4

13, 11, 12, 12.

4

4

4

4

11, 11, 11, 11.

4

4

4

4

12, 10, 11, 11.

4

4

4

4

13, 14, 14, 13.

4

4

4

4

17, 17, 17, 15.

4

4

4

4

14, 14, 14, 14.

4

4

4

4

15, 16, 14, 15.

4

4

4

4

16, 15, 16, 14.

4

3

3

3

12, 11, 8, 4.

4

4

3

3

11, 12, 11, 11.

4

4

4

4

12, 12, 13, 13.

4

4

5

5

21, 13, 13, 14.

4

4

4

4

11, 14, 11, 7.

4

4

4

4

14, 14, 13, 17.

4

3

4

4

14, 15, 14, 9.

246

HARGITT.

[VOL. II.

TABLE II.

RADIAL
CANALS.

GONADS.

GASTRIC
LOBES.

ORAL
LOBES.

TENTACLES.

3

3

3

3

14, 20, 22.

3

3

3

3

14, 19, 19.

3

i
j

3

3

20, 21, 21.

3

3

3

3

16, 20, 22.

3

•>
j>

i

j

3

14, IS, 24.

3

3

3

3

15, 15, 17.

3

3

2

2

13, 17, 24.

3

3

->
3

3

12, 15, 18.

3

3

o

o

3

13, 15, 18.

3

3

3

3

12, 12, 21.

3

3

4

4

13, 14, 18.

3

3

-»

j

3

15, 15, 16.

3

3

o

J

5

15, 17, 23.

*4

4

4

4

9, 13, 16.

*4

4

4

4

15, 15, 17.

5

5

4

4

11, 12, 14, 17, 18.

5

4

4

4

8, 12, 14, 15, 16.

5

5

5

5

7, 8, 9, 10, 11.

5

5

4

4

11, 9, 9, 11, 8.

5

5

3

3

4, 9, 11, 9, 15.

s

5

4

4

12, 10, 7, 8, 13.

5

5

5

5

12, 11, 11, 7, 11.

5

s

4

4

12, 10, 10. 11, 11.

5

5

4

4

11, 7, 9, 9, 11.

5

5

5

5

S, 9, 7, 9, 8.

5

S

4

4

6, 10, 14, 12, 10.

5

5

S

5

11, 13, 9, 10. 11.

5

5

S

5

S, 9, 15, 17, 16.

5

5

4

4

11, 10, 13, 13, 13.

6

6

5

5

6, 5, 6, 6, 6, 6.

6

6

4

4

7, 9, S, 12, 18, 9.

6

6

6

5

7, 9, S, 6, 4, 8.

6

6

6

5

7, 11, 7, 8, 9, 5.

6

6

S

5

11, 4, 9, 7, 7, 12.

* While in these specimens four canals were present, two in each were so
closely approximated as to divide the bell into trimerous sectors.

No. 5.] VARIATION AMONG HYDROMEDUSAE. 247

with Gonionemus all the facts would seem to point to their
independent origin, in many cases the tentacles being of con-
spicuously different sizes, and in others exhibiting all phases
of intermediate conditions between the simple bifurcation of
the terminal portion through the budding of a smaller tentacle
from the side of the larger, on to the symmetrical double ten-
tacles as shown in Figs. 3, 4, and 10.

In only one case was a trifid tentacle found. This is
shown in PI. IV, Fig. 9. However, while a trifid structure
there seems to be a degeneration of the median lobe, which
was in all probability the terminal portion of what was origi-
nally a normal tentacle, from which later were budded the two
lateral shoots, each in turn becoming more prominent than the
median tip and developing in the appropriate places the charac-
teristic suctorial pads. The degenerating middle tip would
very naturally suggest the probability that an injury might
have been the predisposing cause of the secondary tips ; on the
other hand, it must not be overlooked that in each of the other
specimens with double tips no such cause seems at all evident.
I am inclined to consider the cases as simply the expression of
the intrinsic capabilities of variation, more or less evident in
the several classes of organs already considered.

As yet no reference has been made to the order in which
second, third, and subsequent series of tentacles arise. Goni-
onemus, unlike Eucope, seems to have no such association of
tentacle and sensory bulb as serves to locate in part the pri-
mary set of tentacles and the order of appearance of the
subsequent series. While usually there is a single primary
tentacle at the terminus of each radial canal, this is not invari-
ably the case. An inspection of Figs. 9-12 will show that
there may be a very wide range of variation in this respect.
The following data, taken at random from many observations;
will further illustrate the same general fact.

248

HARGITT.

[VOL. II.

TABLE III.

Showing the number and order of succession of tentacles of three series,
counted from a primary set at some arbitrary point, as a radial canal. The large
figure gives the number of any given series occurring together, while the small
exponent figure indicates the series concerned, rv'c., primary, secondary, or tertiary.

A. 21, I2, P, 1', I1, P. 31. I3, 4', 2\ 2'. P, P, P, 4', P, 61, P, P, P, 21,

P, 4l, P, 4'.

B. 2\ P, P, 41. P. 41, P, S1, P, P, P, 2', P, P, 41, P, 2'.

C. P, P, P, 21, P, P, P, P, 2-, P, P, 23, P, P, 21, P, P, P, P, P, P,

P, P, 2', P, 31, P. P.

D. 31, P, P, P, 31, P, P, P, 51, P, 31, P, P, P, 51, P, 31, P, P, P.

E. 4!, P, P, P, P, P, P, 3', P, .V, P, 3', P, 21, P, P, P, 21, P, 31, P,

21, P, P, P, P, I3, P, P, P, P.

F. P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, 21, P, P,

P, 51, P, P, P, P, P, 21, P, P, I2, P, P, P, P, I1, P.

C. I1, P, P, P, P, P, P, P, P, P, P, P, P, 21, P, P, P, 21, P, P, P, P,
P, 21, P, P, P, 21, P, P, P, P, P, 21, P, 31, P, 21, P, 21, P.

// P P ?' P ?' P P P ?! P P P P P P P ?! P I1 P ?' P

Y7. 1, 1, £ , 1, £, , 1, 1, -L, L, , 1 , -L, 1,1, X, 1, 1, L, j 1, 1, 1, £ , 1,

P, P, P, P, P. P, P, P, P, P, 31, P, P, P, P, P, P, P, P, P,
21, P, 21, P, P. P, 2', P, P, P.

/ V I2 ?' P I' 13 71 p II p ?I P P P ?I p P p II p II 13

-*• *-* J •"•>**( * ' ' — « -i- j J- j J. ) — • i i-t -L ) -L) — ' ) •*- J J- ) -*- ' -L ) L) J. j J. •

P P 41 P 31 P P P P P ?' 1 3 (S1 P ^' P P P P P

-Li J-,1? -1-,^, -L. -1' -L, -1. A, ^ , J- j U j J. , Z» , i-, 1, 1, 1,1.

SUMMARY.

PRIMARY.

SECONDARY.

TERTIARY.

TOTAL.

^

35

6

7

48

B

26

4

6

36

C

IS

11

6

3^

D

26

6

4

16

E

2Q

8

s

4^

/'. . . . .

25

9

11

4^

G

29

9

12

^0

H.

36

13

13

62

/

39

7

14

60

As will be seen from these series of tabulated relations in
the appearance of tentacles, there is apparently no order what-
soever. Compared with the Figs., op cit., where is shown simi-
lar relation (not to say absence of relation), there can hardly
be discerned any such thing as definite series or sets of tenta-
cles arising in definite succession. On the contrary, new ten-
tacles seem to arise wherever and whenever in the growth of

No. 5.] VARIATION AMONG HYDROMEDUSAE. 249

the bell the adequate marginal space becomes available. And
the fact that this seems so variable would appear to warrant the
inference that growth occurs at irregular intervals and areas
over the body of the medusa. Might not this fact throw soim-
light upon the marked unsymmetry of such forms as those
shown in Figs. 10 and 11? This suggestion would seem to
find some further support in the fact that very young speci-
mens appear to be more constant in their symmetry than those
more mature.

Otocysts.

In formalin specimens there is a degree of opacity induced,
especially about the marginal area of the bell, which often
renders difficult any satisfactory examination of these sensory
bodies. Hence only a limited number of critical determina-
tions on this point were made, but these were sufficient to show
a degree of variation both in their number and arrangement
quite as marked as in that of other organs.

Normally they should occur in somewhat symmetrical order
between the bases of the tentacles. This, however, is rarely
the case. There seems about the same variation in their
occurrence and relations as in the case of the tentacles, though
I was not able to discover that the latter had any determining-
influence upon them. In only a few cases have I been able to
demonstrate the presence of more than a single otolith in a
given cyst, and in no case more than two. On this point,
however, the opacity above referred to, and the abundant pig-
ment about the bases of the tentacular bulbs were material
obstructions to such determinations, and suggest tentative con-
clusions. In matter of shape and size these organs present
likewise considerable variation. PI. IV, Fig. I, presents the
average aspect of form at a, while at b are shown forms not
unusual but variant.

Summary atid Rei'icic.

1. Variation among Hydromedusae is of wider extent than
had been supposed.

2. Variation is much greater in some genera than in others.

250 HARGITT. [Vol.. II.

3. Variation among Hydromedusae appears to be much less
symmetrical and less definitely correlated than among Scypho-
medusae.

4. Many phases of variation seem to be wholly devoid of
any adaptive features or tendencies.

5. The ratio of variation is higher among the tentacles than
among other organs, and in many species higher than in all
other organs combined, - - a feature which is perhaps the most
conspicuous case of adaptation apparent in the entire series.

Among the earliest references to variation in Coelenterates
is that of Ehrenberg ('37) relative to variation in Aurelia.
Later, Romanes ('74-'76) took up the subject with much more
detail, giving an extended account of the nature and extent of
variation, particularly in Aurelia, in which he figures and
describes many " monstrous forms of medusae " and points
out interesting correlations of radial canals, gonads, tenta-
cles, etc.

Within recent times these observations have been much
extended, notably by Brown ('94), who distinguished more than
two per cent.

Sorby ('94), Herdman ('94), and Unthank ('94) have each
recorded many interesting facts of variation in this medusa.

In 1895 Brown still further extended his observations upon
Aurelia, and in connection therewith undertook a comparison
of a large number of the Ephyrae. He was able to distinguish
no less than 22.6 per cent of numerical variation in tentaculo-
cysts, a ratio very close to that earlier determined for adult
Aurelia. The observations seemed to show upon the whole
a tendency toward an increase in meristic characters.

Ballowitz ('98) records extended observations upon Aurelia,
specially with reference to the gonads. While in general there
was more or less correlation in the numerical variation of these
organs with the actinal lobes, it was apparently less constant
than had been claimed by earlier observers. The highest
number noted was seven, while three was the minimal number.
One specimen in particular, which he names Ephyra abnor-
mitat, seems to be an unusual monstrosity, having a very large
balloon-shaped body with a correspondingly large manubrium.

No. 5-] VARIATION AMONG HYDROMEDUSAE. 251

He explains it as probably clue to an enormous expansion of the
top of the Ephyra, thus forming the balloon-like body.

It may not be amiss in this connection to record observa-
tions of a similar character as to numerical variations upon
the Aurelias of Woods I loll which quite confirm those cited.

Observations upon the Hydromedusae seem to have been
heretofore quite limited. Those of Forbes ('48), Agassiz ('49),
Hincks ('68), Romanes ('74-76), Agassiz and Wood worth ('96),
include all the more important observations which have come
to my knowledge. The latter would seem to be about the
only series made upon a large number of specimens with the
purpose of ascertaining the extent and character of variation
in a single genus.

SYRACUSE UNIVERSITY, September i, 1900.

REFERENCES.

AGASSIZ, A., and WOODWORTH. Bull. J///.c. Comp. Zool. Vol. xxx.

No. 2. 1896.

BALLOWITZ. Variation in Aurelia. Arch.f.Entunck.d. Organism en. 1898.
BATESON, W. Materials for Study of Variation, p. 424 et seq. 1894.
BROWX, E. T. Variation of Tentaculocysts of Aurelia aurita. Qnar.

Joiirn. Micr. Sci. Vol. xxxviii.
BROWX, E. T. Variation of Haliclystus octoradiatus. Ouar. Journ. Micr.

Sci. Vol. xxxviii.

BROWN, E. T. Variation of Aurelia. Xatnre. Vol. 1. 1894.
FORBES, EDW. British Naked Eye Medusae. London, 184*.
HARGITT, C. W. Nat. Hist, of Pennaria. Amer. Xat. May. 1900.
HERDMAX, W. H. Pentamerous Aurelia. Xatnre. Vol. 1. 1894.
HIXCKS, T. Clavatella. Hydroid Zoophytes, p. 71. 1868.
HORXELL, J. Abnormalities in Haliclystus octoradiatus. Xat. Sci. 1893.
HORXELL, J. Lucernaria as Degenerate Scyphomedusae. Nat. Sci.

1893.

MURBACH, L. Journ. of Morpli. 1895.
ROMAXES, J. G. Varieties and Monstrous Forms of Medusae. Journ.

Linn. Soc. 1874-76.

SMALL WOOD, M. Morphology of Pennaria. Amer. Xat. 189^.
SORBV, H. C. Symmetry of Aurelia aurita. Xatnre. Vol.1. 1894.
UNTHAXK, H. W. Pentamerous Aurelia. Xatnre. Vol. 1. 1894.

252

HARG1TT.

[VOL. II.

PLATE I. — Diagrammatic figures, illustrating variation in form, number, and arrangement

of the radial canals.

No. 5.] VARIATION AMONG U YDROMEDL .s. //, .

I'LATK II. — Showing various phases of atrophy, spur-like branches, etc. of radial

254

HARGITT.

[VOL. II.

PLATE III. — Figs. 1-6 show varying phases in the evolution of trimerism. Figs, n and 12
show at .V the development of spurs.

N o. 5 •] / - -1 K1A TIOX AMONG Jf YD ROM ED L X / /-;.

255

PLATE IV. — Fig. i, n, a, normal : /•, /', variable forms of otocysts. Fig. 2 showing variation in
tentacles. Figs. 3-10 show various phases in forking or budding of tentacles.

Volume //.] June, lyoi. \_No. 6.

BIOLOGICAL BULLETIN.

[From the Zoological Laboratory of the University of Pennsylvania.]

THE INDIVIDUALITY OF THE GERM NUCLEI

ERRATA.

In No. 5, p. 225, igth line, read sensory bulbs in place of
tentaculocysts.

On p. 228, 2oth line, read ocelli in place of otocysts.

.ioi. null.

ot other animals.1

•

1 Riickert calls attention to the fact that partially cleft nuclei are found in
the figures of various authors, particularly those of Fol ('79) on Toxopneustes,
of Bellonci ('84), and of Kolliker ('89) on Siredon. Of course no one of these
observers has interpreted these figures as showing the independence of the germ
nuclei, and some of the figures referred to by Riickert probably do not show this
phenomenon. For example, only one of Fol's figures (PI. VII, Fig. 7) shows a
dual nucleus, while the figure in Kolliker's text-book (Fig. 36) is probably a case
of the indentation of the nuclear membrane opposite the centrosomes in the early
prophase, a thing which frequently happens. Bellonci's Figs, i and 20 show an
indentation on one side of the nucleus which may correspond to a division between
the germ halves, though this must be regarded as more or less doubtful.

257

Volume //.] June, 1901. \_No. 6.

BIOLOGICAL BULLETIN.

[From the Zoological Laboratory of the University of Pennsylvania.]

THE INDIVIDUALITY OF THE GERM NUCLEI

DURING THE CLEAVAGE OF THE

EGG OF CREPIDULA.

EDWIN G. CONKLIN.

HAECKER ('92) and Riickert ('95) have made known the
interesting fact that the germ nuclei of Cyclops do not fuse
but preserve their individuality throughout a considerable por-
tion of the cleavage of the egg. Herla ('93) and Zoja ('95)
have shown that the paternal and maternal chromosomes of
Ascaris remain distinct at least as far as the 12-ceil stage.
These observations are of the greatest significance and, so far
as they go, establish Boveri's hypothesis ('91), "that in all cells
derived in the regular course of division from the fertilized
egg, one-half of the chromosomes are of strictly paternal
origin, the other half of maternal." So far as I am aware,
similar observations have not hitherto been made in the case
of other animals.1

1 Riickert calls attention to the fact that partially cleft nuclei are found in
the figures of various authors, particularly those of Fol ('79) on Toxopneustes,
of Bellonci ('84), and of Kolliker ('89) on Siredon. Of course no one of these
observers has interpreted these figures as showing the independence of the germ
nuclei, and some of the figures referred to by Riickert probably do not show this
phenomenon. For example, only one of Fol's figures (PI. VII, Fig. 7) shows a
dual nucleus, while the figure in Kolliker's text-book (Fig. 36) is probably a case
of the indentation of the nuclear membrane opposite the centrosomes in the early
prophase, a thing which frequently happens. Bellonci's Figs, i and 20 show an
indentation on one side of the nucleus which may correspond to a division between
the germ halves, though this must be regarded as more or less doubtful.

257

258

EggNvxdeus
Centrosewe,

CONKLIN.

FIGS, i and 2. — Prophase and Metaphase of First Cleavage. S.N. = Sperm Nucleus.
Ch. = Chromosomes. S. = Sphere. C. = Centrosome.

9 Nude

Lobq

FIGS. 3 and 4. — Anaphase and Telophase of First Cleavage, showing dual nuclei, centrosomes
(C)and spheres (S), " Zwischenkorper" (/.), bending of spindle axis, and progressive absorp-
tion of yolk lobe.

FIGS. 5 and 6. — Prophase of Second Cleavage, showing dual nuclei with central spindle tying in
groove between the halves ; Fig. 5 viewed from one side, Fig. 6 from animal pole. The
sphere's (S) lie over the nuclei and immediately under the cell wall ; the spindle axis is bent
on itself, and the "Zwischenkorper" (Z) is carried nearly to the vegetal pole; the nuclei
shows processes projecting toward the " Zwischenkorper," and the yolk lobe (L) is almost
completely absorbed.

No. 6.] INDIVIDUALITY OF THE GERM NUCLEI. 259

In Crepidula plana I have observed a separateness of the
germ nuclei in certain stages of the nuclear cycle which is of
such a character that it may lead to the discovery of similar
phenomena in other animals. This separateness is most easily

ON

FIGS. 7 and 8. — Anapliase and Telophase of Second Cleavage. Fig. 7, an abnormal egg in
which the left half has divided normally and the chromosomal vesicles of the daughter-nuclei
are fusing; in the right half the division figure is double with four spheres and two groups of
chromosomal vesicles which have not fused; there are thirty chromosomal vesicles in each
group. Fig. S, a normal egg showing the egg and sperm constituents in each of the
daughter-nuclei.

FIG. g. — Telophase of Second Cleavage ; centrosomes, spheres, and nuclei rotating in direction

of arrows.
FIG. 10. — Anaphase of Third Cleavage. Egg and sperm constituents of nuclei indicated.

observed in the telophase of each division, though in some
cleavage cells it may be seen in the prophase also, or even
throughout the resting period. At the time when the daughter-
nuclei are being formed the chromosomal vesicles fuse into two

260 CONK LIN. [VOL. II.

groups which are closely pressed together but are still separated
by a partition wall (Figs. 4, 8, et scg.}, as Ruckert has shown to
be the case in Cyclops. Gradually this partition wall dis-
appears, being preserved longest on that side of the nucleus
nearest the centrosome (Fig. 5). Here a groove is formed
on one side of the nucleus which marks the line of contact
between the two halves. In some cleavage cells this groove
is visible throughout most of the resting period (Figs. 4, 9) ;
in others it disappears during the greater part of the resting
period, though it may reappear in the following prophase
(Figs. 5, 6) ; in all cases, however, the partition wall and groove
reappear in the next succeeding telophase, when it is formed
again in the manner described above. I have observed the
double character of the nucleus in the telophase of every
cleavage up to the 29-cell stage (Figs. 1-16), and in several of
the later cleavages up to the 6o-cell stage, and I have no doubt
that it is found in all the later cleavages, though it becomes
more difficult to see as the nuclei grow smaller. While the
halves of these double nuclei occupy similar positions relative to
each other at corresponding stages in any cell generation, they
occupy different positions at different stages and in different
generations ; consequently the position of the groove or par-
tition wall which separates the halves of the double nuclei can
be satisfactorily studied only in preparations of entire eggs,
which may be observed from all sides. All the figures which
illustrate this paper are therefore of entire eggs, though many
isolated cases of double nuclei have been observed and studied
in actual sections.

On each side of the partition wall which divides these double
nuclei there is usually a single small nucleolus ; these two
nucleoli persist long after the disappearance of the partition
and frequently throughout the whole of the resting period. In
most if not all of the early cleavages there are two, and only
two, nucleoli present in the telophase (Figs. 3, 4) ; but if this
is succeeded by a very long resting period the number may
increase to more than two, or all may fuse into a single enor-
mously large one.

It still remains to show that these double nuclei really

No. 6.] INDIVIDUALITY OF THE GERM NUCLEI. 261

FIG. ii. — Telophase of Third Cleavage. Egg and sperm constituents of nuclei indicated; also
bending of spindle axis and rotation of centrosomes and nuclei.

FIG. 12. — Telophase of Fourth Cleavage and Prophase of division of First Quartette cells.
The nuclei in the telophase are dual, though from this stage on the egg and sperm con-
stituents cannot be identified with certainty.

FIG. 13. — Telophase of Division of First Quartette. Dual nuclei in each daughter-cell.

FIG. 14. — Subdivision of Second Quartette and formation of Third. Dual character of nuclei

shown in apical cells.

la.'

/a"1

FIG. 15. — Telophase of Division of second Quartette and of formation of Third; dual nuclei

shown in almost all of these cells.
Fig. 16. — Telophase of formation of the Mesentoblast Cell (40) and of the second division of

the First Quartette I'irt1, ia''2, etc.); dual nuclei shown in both cases.

262 CONK LIN. [VOL. II.

represent the egg and sperm nuclei which have not yet lost their
individuality. This cannot be demonstrated in Crepidula, for
the reason that this double character is not apparent at every
stage in the nuclear cycle, but it is extremely probable, as the
following observations will show :

1. In the first cleavage the germ nuclei do not fuse but
remain distinct throughout the prophase, and even in the meta-
phase they are represented by separate groups of chromosomes
(Figs. 1,2); in the early anaphase these groups of chromosomes
can no longer be distinguished, though I think they must still
remain separate, for the nuclei are clearly double in the imme-
diately following late anaphase and telophase (Figs. 3, 4).
The position of the partition wall in these double nuclei corre-
sponds to the plane of contact between the germ nuclei; the
egg nucleus always lies more or less above the sperm nucleus,
and in the telophase of the first cleavage one-half of each
double nucleus overlaps the other half to a greater or less
extent (Figs. 1-4). It is probable that the upper half repre-
sents the egg nucleus, and the lower half the sperm nucleus,
and in all the later cleavages it is probable that the half of the
nucleus which lies nearest the animal pole is from the egg, and
the other half from the sperm.

2. The groove which is found on one side of the nucleus in
the telophase of the first cleavage (Fig. 4) persists well into the
resting stage, and a corresponding groove is found in the same
position in the prophase of the second cleavage (Figs. 5, 6).
The central spindle for the second cleavage lies in this groove
(Fig. 6), and thus the amphiaster actually lies in the only plane
in which it would be possible to halve the two parts of the
double nuclei. This very fact shows that each half of a double
nucleus is represented in the daughter-nuclei, and it strongly
suggests that the two parts of the daughter-nuclei are derived
directly from the corresponding parts of the mother-nucleus
(cf. Figs. 6, 8). The fact that the central spindle lies in the
groove separating the halves of the nucleus has been observed
in the first, second, third, and fourth cleavages, and is undoubt-
edly a general phenomenon. There is no reasonable ground
for doubting that the two parts of every double nucleus are

No. 6.] INDIVIDUALITY OF THE GERM NUCLEI. 263

derived from corresponding parts of a mother-nucleus, and
so on back to the egg and sperm nuclei in the first
cleavage.

Since the descendants of the germ nuclei are halved at every
division, it follows that successive divisions of the double nuclei
cannot be at right angles to one another, since this would lead
to an unequal division of the halves, or even to a division along
the plane of contact between the halves. Such an unequal
division might be prevented in cleavages which successively
alternate in direction by the rotation of the nucleus during the
resting period, or by the rotation of the spindle in the early
stages of mitosis.1. As a matter of fact both of these methods
occur in Crepidula. The nucleus usually rotates during the
rest through 90°, so that although successive nuclear spindles
are at right angles to one another the axis of every spindle lies
in the same nuclear axis (cf. Figs. 3, 4, 8, 9) ; but in some
cases the nuclear spindle does not lie in its definitive position
when first formed but undergoes extensive rotation after its
formation. While it is not susceptible of absolute proof, since
the partition wall is absent during the later stages of the rest,
it is highly probable that the plane in which all nuclear spindles
lie is the plane of contact between the two halves of every
nucleus.

3. In certain abnormal cleavages the double nuclei are really
two entirely separate nuclei lying side by side within a single
cell. Such binucleated cells may occasionally be found with
the nuclei in the height of the rest, though they are more usual
in the telophase or early resting period. There is usually but
a single sphere and centrosome in such cells, though in one
case of pathological mitosis which I have seen (Fig. 7) there
are two mitotic figures side by side ; the chromosomes which
have reached the stage of the chromosomal vesicles have not
aggregated at the poles of these spindles, but are scattered
along their whole length. There are thirty of these

1 Riickert finds that the nuclei rotate in Cyclops even after the spheres have
reached their definitive positions at the poles of the spindle ; I have never observed
in Crepidula a rotation of the nuclei, independent of the spindles, at so late a
stage in the cell cycle.

264 CONK LIN. [VOL. II.

chromosomes in each spindle, the same number that is found
in each germ nucleus.

4. In each of the germ nuclei, before they come into contact,
there is a single nucleolus ; these nucleoli disappear in the pro-
phase of the first cleavage, but in the succeeding telophase a
single nucleolus generally appears in each half of each daughter-
nucleus. The same is true of the succeeding cleavages, so that
each nucleus throughout the cleavage usually has two nucleoli
in the telophase or early resting stage, though the number may
vary in the later resting period, as pointed out above. The
fact that there is a single nucleolus in each germ nucleus, and
that there is usually a single nucleolus in- each half of the
double nuclei of the cleavage, may possibly indicate that these
halves are each derived from one of the germ nuclei. Since
the nucleoli as such do not persist throughout the mitosis, may
it not be possible that there is some achromatic structure in
connection with them which does persist and form the basis
for the new nucleoli which appear in the daughter-nuclei ?

These facts make it very probable that the germ nuclei of
Crepidula preserve their individuality throughout the cleavage,
though their separateness may be apparent only or chiefly in a
single stage of the nuclear cycle, viz., the telophase. Further, it
is possible, even in an advanced stage of the cleavage, to deter-
mine with considerable probability which part of a double
nucleus is derived from the egg and which from the sperm,
the egg half always lying nearer the animal pole than the
sperm half. Finally the initial position of the mitotic spindle
seems to be determined by the relative positions of the halves
of the double nuclei, since the spindles when they first appear
lie in the plane of contact between the two halves ; the final
position of the spindle and the direction of division are deter-
mined by the movements of the cytoplasm.

No. 6.] INDIVIDUALITY OF THE GERM NUCLEI. 265

REFERENCES.

'84 BELLONCI, G. Interne alia cariocinesi nella segmentazione dell' ovo

de Axolotl. Atti delta Accad. dei Lincei, Memoire xix.
'79 FOL, H. Recherches sur la fecondation et le commencement de

Thenogenie. Geneve.
'92 HACKEK, V. Die Eibildung bei Cyclops und Canthocamotus. Zool.

Jahrb. Bd. v.
'93 HERLA, V. Etude des variations de la mitose chez 1'ascaride megalo-

cephale. Arch. Biol. Vol. xiii.

'89 KOLLIKER, A. Handbuch der Gewebelehre. 6 Aufl. Leipzig.
'95 RUCKERT, J. Ueber das Selbstandigbleiben der vaterlichen und

miitterlichen Kernsubstanz warend der ersten Entwicklung des

befruchteten Cyclops-Eies. Arch.f. inikr. Anat. Bd. xlv.
'95 ZOJA, R. Sulla independenza della cromatina paterna e materna nel

nucleo delle cellule embryonale. Anat. Ans. Vol. xi.

THE EARLY DEVELOPMENT OK THE
HYPOPHYSIS IN CHELONIA.

s ZKI.KNY.

THE following observations on the early development of the
hypophysis in Chelonia are offered at this time because they
throw positive light upon the derivation of the organ concern-
ing which we at present have several conflicting views. The
paper is based upon sections of Aspidonectes spinifer Ag.,
Chelydra serpentina L., and Chrysemys marginata Ag., which
show that in these forms the hypophysis is undoubtedly of
epiblastic origin. Incidentally some points regarding the rela-
tion of the preoral gut to the notochord and the head cavities
will be noted, but this subject will receive fuller treatment in a
subsequent paper. Kor the sake of convenience the subject
matter will be considered under the following heads :

1. A general outline of the literature dealing with the early
development of the hypophysis among vertebrates.

2. Material and methods of preparation.

3. Description of stages.

4. Summary and conclusion.

Literature.

It would be out of place in a paper such as the present
to give any detailed account of the views which have been
held regarding the subject under discussion. A bare mention
of a few of the upholders of each of the principal views, giving
the group upon which work was done, must suffice.

i. Those who have claimed a hypoblastic origin for the
hypophysis are Luschka ('69), Mammalia ; W. Miiller ('71),
Vertebrata in general ; Dohrn ('82), Teleostii, later extended to
Vertebrata in general ; Hoffmann ('§8), Lacerta ; Prather ('99),

Amia.

267

268 ZELENY. [VOL. II.

2. Those who have claimed an epiblastic origin for the
hypophysis are Mihalkovics ('77), Aves and Mammalia ; Balfour
('78), Elasmobranchii ; Orr ('87), Lacerta ; Lundsborg ('94),
Salmonidae ; Dean ('96), Amia ; Haller ('96), Vertebrata in
general ; Hoffmann ('96), Elasmobranchii ; Melchers ('99),
Lacertilia, Of these Orr ('87), although he describes the
hypophysis as of epiblastic origin and so figures it in his
sections, nevertheless considers it probable that hypoblast
cells may take some part in its development.

3. Those who have claimed that the hypophysis is partly of
epiblastic and partly of hypoblastic origin are Kupffer ('93),
Acipenser and Ammocoetes ; Valenti ('95), Amphibia (Bufo) ;
Nussbaum ('96), Mammalia ; and considered probable by Orr
('87), Lacerta.

It is of special interest to note that even this partial list
gives each of the three views a large number of the groups
of vertebrates upon which to base its general character.

Material and Methods.

The material upon which the following observations are
based was obtained in Minnesota during the summers of 1898
and 1899. The embryos of Aspidonectes are from Grey
Cloud Island in the Mississippi River below St. Paul, and
those of Chrysemys and Chelydra are from the neighborhood
of Hutchinson. The fixing fluid used was Gilson's mercuro-
nitric mixture. The embryos were stained in toto in haema-
calcium or in paracarmine. The series of sagittal sections
were found to be the most helpful in determining the cell-
layer from which the hypophysis is derived, and the following
descriptions are taken entirely from such sections.

Description of Stages.

The following stages will be figured and described :
Stage A. The hypophysial evagination has not yet

appeared.

Stage B. The hypophysial evagination is very evident and

the pharyngeal membrane has not yet been broken.

No. 6.] THE HYPOPHYSIS IX CHELOXIA. 269

Stage C. Slightly older than Stage B. The pharyngeal
membrane has been broken.

Stage D. The downward growth and enlargement of the
fore-brain have pushed the hypophysis back from its primary
relation to the broken ends of the pharyngeal membrane.

Stage E. The hypophysis has been shifted backward so
as to assume a position relatively far back in the pharyn-
geal cavity.

Stage F. The hypophysis has become differentiated into
a terminal, broad, sac-like part and a narrower connecting
stalk.

Stage A.

In an embryo with five or six mesoblastic somites and in
which the medullary folds have not yet united above to enclose
a medullary canal the hypophysial evagination has not begun to
form. A median sagittal section of such an embryo, however,
shows the relation of the parts surrounding the point at which
the hypophysis will appear at a later stage. Fig. i, PI. I, rep-
resents a diagrammatic median sagittal section of an embryo
of Chrysemys marginata 2.5 mm. in length and in which there
are five distinct mesoblastic somites with indistinct traces of a
sixth. The figure is a combination of the six sections nearest
to the median line. Although the medullary groove is still
open above, the medullary folds in the head region have grown
to a considerable height, as represented by the dotted line
D. At the same time the whole anterior region has been
folded and bent downward. The fold of the blastoderm which
comes up over the head as a result is the proamnion (Pa.), and
consists of epiblast and hypoblast. At the bottom of the head
fold the hypoblast has traveled back much farther than the
epiblast, leaving a space in which the cardiac mesoblast (Cm.)
develops. As we trace this hypoblast (End.\) forward along
the floor of the fore-gut (F.G.) we find that it is bent ven-
trally so as to come into contact with the epiblast. This point
of contact, represented by the double-headed arrow ( \ ), is the
region at which the mouth will be formed. In front of this
we may recognize a short, wide, preoral gut (Pr.G.). This is a

270 ZELENY. [VOL. II.

true preoral gut and not an apparent one caused by the down-
ward bending of the head region ; for we must consider the
sharp angle in the hypoblast at Pr.G. in the figure, and not
the part back of it where the mouth will appear, to have been
the original extreme anterior end. Thus we may consider all
the hypoblast below and posterior to this angle to belong to
the ventral wall of the alimentary canal, and all that above and
posterior to belong to the dorsal wall. Following the dorsal
wall we find that it immediately divides into two parts, one of
which is the notochord (Nc.) and the other the dorsal wall
of the gut (End. 2). For this reason the hypoblast forming the
ventral wall of the alimentary canal, and continued out at
the ends of the embryo into the flat outlying blastodermic
region, may be called the " primary hypoblast," and the hypo-
blast of the dorsal wall the " secondary hypoblast." Rex
('97), in his work on the duck, and Davidoff ('99), in the
embryo of Platydactylus, have found similar relations of the
notochord to the hypoblast. The terms " primary hypoblast '
and " secondary hypoblast," which are used above, are taken
from the paper of the latter author. It is important to note in
this connection that my sections show a distinct line of demar-
cation between the epiblast and the hypoblast in this region, so
that the mass of cells surrounding the preoral gut is distinctly
hypoblastic and not a mass of undifferentiated cells. The
arrow with a feathered shaft (Y) shown in Fig. i in the
epiblast directly under this point marks the position and direc-
tion of the future hypophysial pocket. The very plain line of
division between the epiblast and hypoblast excludes the possi-
bility that any of the hypoblast cells may take part in this
ingrowth of the epiblast.

Stage B.

In an embryo 5.2 mm. in length and with twenty-one meso-
blastic somites, such as is shown in median sagittal section in
PL I, Fig. 2, and PI. II, Fig. 3, the brain has developed with
great rapidity. The cephalic flexure having proceeded at the
same time, the region around the preoral gut is greatly com-
pressed. The cavity of the preoral gut itself has become very

No. 6.] THE HYPOPHYSIS IN CHELONIA. 271

small and its dorsal wall is doubled back on itself. The noto-
chord also at its anterior end near the point where it joins the
hypoblast has become very much twisted and curved, giving it
a knotted appearance in sagittal section. However, the parts
are easily recognized as having the same relation as those in
Stage A (Fig. i). In Stage B (Fig. 2), as before, the dorsal
wall of the fore-gut is the "secondary hypoblast" and the
ventral wall the "primary hypoblast," the division line between
the two being the point where the notochord joins the hypo-
blast. The pharyngeal membrane is still intact but breaks
up very soon after this stage. Directly in front of the place
where the mouth opening will appear the epiblast bends sharply
back on itself to follow the brain, forming a pocket in which
the cells are taller than in the neighboring parts of the epi-
blast. This is the beginning of the outpocketing which will
eventually form the hypophysis. The evaginating layer of
cells is clearly distinct from the hypoblast cells of the pre-
oral gut, as is well shown in PI. II, Fig. 3, which gives an
enlarged view of the hypophysial region of Fig. 2. Starting
from the condition of the stomodeal epiblast in the earlier
stage as shown in Fig. i, we see that an epiblastic groove
was originally formed at the point marked by the feathered
arrow (^) just in front of the future mouth by the forward
and downward bending of the fore-brain. It is at the bottom
and in the middle of this groove that the thickening and out-
pocketing of the cells start, and later form the epiblastic
pouch which becomes the oral portion of the hypophysis.

Stage C.

In an embryo but slightly older than the one last described
the pharyngeal membrane has already been broken. Fig. 4,
PI. Ill, represents a diagram of a median sagittal section
of such an embryo. On account of a lateral bend in the neck
region it was not possible to obtain a section which would show
the connection of the notochord and the preoral gut at the
same time with the mouth opening and the hypophysial evagi-
nation. The canal (H.C.) between the two premandibular

272 ZELENY. [VOL. II.

head cavities is, however, shown, and the small mass of cells
connected with it and directed toward the hypoblast of the
fore-gut is the strand which connects the head cavities with the
preoral gut. The preoral gut itself is not shown in this figure,
the cavity (F.G.') being a part of the fore-gut which has assumed
a position anterior to the mouth because of the bending of the
alimentary canal, which takes place at the same time with the
cephalic flexure. Except for the break in the pharyngeal mem-
brane the relation of the hypophysis to the epiblast is the same
as in Stage B. The hypophysial outpocketing is here, as before,
on the epiblastic side of the mouth opening and is undoubtedly
made up entirely of epiblast cells.

Stage D.

In Fig. 5, PL III, we have a median sagittal section of a some-
what later stage. The points/ and/' show the position of the
ends of the broken pharyngeal membrane, and the dotted line
(P.M.} between them represents the former position of the now
ruptured membrane. The hypophysial pouch (Hyp.} is shown
very distinctly on the epiblastic side of the membrane. It has,
however, been pushed back from its original position with
relation to the point /' by the rapid growth of the fore-brain.
The anterior end of the notochord (N.} is still more curved
and wrinkled than in the last stage, but it retains its connec-
tion with the anterior wall of the preoral gut (P.G.) by means
of a string of cells. In this string of cells we see the section
of the canal (H.C.) which connects the two anterior or pre-
mandibular head cavities. The hypophysial evagination from
the very beginning is in. close contact with the wall of the
infundibulum, but the two layers of cells always remain clearly
distinct. The epiblast cells of the hypophysis also remain
clearly distinct from the hypoblast cells of the preoral gut.

Stage E.

Fig. 6, PL IV, is a diagram of a section of Chelydra ser-
pentina. It shows a continuation of the same process of
enlargement of the fore-brain and the consequent pushing

No. 6.] THE HYPOPHYSIS IN CHELONIA. 273

back of the hypophysial evagination, so that the latter finally
appears to lie far back in the pharyngeal cavity. However,
here, as before, the dotted line //' shows the original position
of the now broken membrane. At this stage the notochord
has severed its connection with the fore-gut, but it is still
joined to the mass of cells which connects the two head cavi-
ties. These cells still surround a canal, so that there is a
passageway from the premandibular cavity on one side of the
head to the corresponding cavity on the other side.

Stage F.

Fig. 7, PI. IV, represents a median sagittal section of
Aspidonectes spinifer, and Figs. 8 and 9, PI. V, represent
sections of the same series, respectively, three and six sections
to the side of the median line. In these it is seen that there
is no evidence of the canal which connected the two head
cavities, and the notochord shows no sign of the former union
with the hypoblast. The hypophysis has begun to constrict in
the basal region and to enlarge in the terminal region so as to
show a division into a narrower basal stalk and a wider terminal
sac-like portion.

At a considerably later stage than the above the infun-
dibulum sends out a pouch-like evagination, which from the
beginning is in close contact with the wall of the oral sac
and forms the infundibular part of the hypophysis.

Summary and Conclusion.

The foregoing series of sections furnish a clear chain of
evidence in favor of the epiblastic origin of the hypophysis
in Chelonia. Stage B (Figs. 2 and 3) itself is a conclusive
proof of such an origin. Here with the pharyngeal membrane
yet unbroken we find that the evagination is on the epiblastic
side of the membrane. The distinct limiting line which marks
the inner border of the hypophysial pouch excludes the suppo-
sition that the hypoblast cells may take a part in its forma-
tion. Even at the early stage represented in Fig. i there is

274 ZELENY. [VOL. II.

a distinct limiting line between the epiblast and hypoblast at
the point where the hypophysis will later appear.

After the mouth opening has appeared (Stage C, Fig. 4) we
find the hypophysis at first in the same relative position as in
Stage B. Then the great increase in size of the fore-brain
forces the epiblastic pocket to a position far back in the
pharyngeal cavity, so that Stages D and E when considered
alone would lead one to believe that the hypophysis is of
hypoblastic origin in this group. That such is not the case is
made evident by following the whole series of changes through
all the different stages. There can be no doubt that in the
Chelonia at least the oral evagination which goes to form the
hypophysis is of epiblastic origin. As regards the infundibular
portion there is no essential difference of opinion and its devel-
opment need not be touched on here.

The bearing of the above conclusions on the paleostome
theory of Kupffer and the neostome theory of Dohrn is of
some interest. According to Kupffer, the hypophysis was
originally a canal connecting the fore-gut with the epiblast,
and represents an ancestral oesophagus which came up in
front of the fore-brain and was replaced by the modern oesoph-
agus at the time when the mouth was forced to a more ven-
tral position by the enlargement of the brain. Dohrn, on
the other hand, has picked out the epiphysis and hypophysis
as remnants of the old annelid oesophagus which went up
through the brain. Both of the above views presuppose some
connection of the hypophysis with the hypoblast. The sections
of turtle embryos which are described in the present paper
give no evidence of such a connection at any stage.

In conclusion I wish to express my sincere thanks to
Prof. H. F. Nachtrieb, who suggested to me the investi-
gation of chelonian development and has aided me in many
ways during the progress of the work.

DEPARTMENT OF ANIMAL BIOLOGY,

UNIVERSITY OF MINNESOTA, December, 1900.

No. 6.] THE HYPOPHYSIS IN CHELOX1A. 275

LITERATURE REFERRED TO IN THE TEXT.

BALFOUR, F. AI. A Monograph on the Development of Elasmobranch

Fishes. London. 1878.
DAVIDOFF, M. VON. Ueber praoralen Darm und die Entwicklung der

Pramandibularhohle bei den Reptilien (Platydactylus mauritanicus

L. und Lacerta agilis Alerr. ). Knpffcr Festschrift, pp. 431-454.

1899.
DEAX, BASHFORD. On the Larval Development of Amia calva. Zoo/.

Jahrb. 1 896.
DOHRX, A. Studien zur Urgeschichte cles Wirbelthierkorpers. Mittheil.

der Zool. Stat. zu Neapel. Bd. iii. 1882.
HALLER, B. Untersuchungen uber die Hypophyse und die Infundibular-

organe. Morph. Jahrb. 1897.
HOFFMANN, C. K. Reptilien, in Bronn's Klassen und Ordnungen des

Tierreichs. Bd. vi, Abth. iii. 1888.
HOFFMANN, C. K. Beitrage zur Entwicklungsgeschichte der Selachii.

Morph. Jahrb. Bd. xxiv. 1896.
KUPFFER, C. Studien zur vergleichenden Entwickelungsgeschichte des

Kopfes bei Kranioten. Hefte 1-3. Miinchen und Leipzig, Leh-

mann. 1893-95.
LUNDSBORG, H. Die Entwickelung der Hypophysis und des Saccus vas-

culosus bei Knochenfischen und Amphibien. Zool. Jahrb., Abth. f.

Anat. u. Ontog. Bd. vii. 1894.

LUSCHKA. Der Hirnanhang und die Stiezdriise. Berlin. 1860.
MKLCHERS, FRITZ. Ueber rudimentare Hirnanhangsgebilde beim Gecko.

Zeitschrift f. iviss. Zool. 1899.
AIiHALKOVics, V. v. Entwicklungsgeschichte cles Gehirns. Leipzig.

1877.
MCLLER, W. Ueber die Entwickelung und den Bau der Hypophysis und

des Processus infundibuli. Jenaische Zeitschr. Bd. vi. 1871.
NussBAi'M, J. Einige neue Thatsachen zur Entwicklungsgeschichte der

Hypophysis cerebri bei Saugethieren. Anat. Anzei^cr. Bd. xii.

1 896.
ORR, H. Contribution to the Embryology of the Lizard. Journ. of Morph.

Vol. i. 1887.
PRATHER, J. AI. The Early Stages in the Development of the Hypophysis

of Amia calva. Biol. Bull. Vol. i. 1900.
REX, HUGO. Ueber das Mesoderm des Vorderkopfes der Ente. . \i\liii'

f. Mikr. Anat. Bd. 1. 1897.
VALEXTI, G. Studio sull1 origine e sul significatio dell1 ipofisi. Atti d.

rAccad. med.-chir. d. Pai'ia. \'ol. viii. 1895.

276

ZEL ENY.

[VOL. II.

ABBREVIATIONS USED IN CONNECTION WITH THE FIGURES.

Ao. Aortic arch.

Cm. Cardiac mesoderm.

Ect. Epiblast.

End. i " Primary hypoblast."

End.z " Secondary hypoblast."

Ep. Epiphysis.

F. B. Fore-brain.

F. G. Fore-gut.

H. B. Hind-brain.

H. C. Canal connecting the two pre-

mandibular head cavities.

Ht. Heart.

Hyp. Hypophysis.

Hyp. S. Stalk of hypophysis.

/. Infundibulum.

M. Mouth opening.

M. B.
Md.

Nc. N.
Pa.
P. G.
P. M.

Mid brain.

Floor of medullary groove.

Notochord.

Proamnion.

Preoral gut.

Pharyngeal membrane.

Dotted line above Md. in Fig.

i represents the height to

which the medullary folds

have risen.
Position and direction of the

hypophysial evagination.
Pharyngeal membrane and

position of future mouth

opening.

No. 6.]

THE HYPOPHYSIS IN CHELONIA. •

277

PLATE I.
Ect.Endj Pa.

/ /'

FIG. i (PI. I), Stage A. — Combination diagram formed from the six sections
nearest to the median sagittal section. Embryo of Chrysemys marginata.
Mesoblastic somites = 5^. Lg. of embryo = 2.5 mm. x 62.

FB.'

FIG. 2 (PI. I), Stage B. — Diagram of the median sagittal section of an embryo
of Aspidonectes spinifer. Lg. = 5.2 mm. Mesoblastic somites = 21. x 50.

278

ZELENY.

[VOL. II.

PLATE II.

FB

End.

FIG. 3 (PI. II), Stage B. — Median sagittal section of Aspidonectes spinifer.
Part of Fig. 2, but with Magn. X 200.

No. 6.]

THE HYPOPHYSIS IX CHELO.X1A.

279

PLATE III.

FIG. 4 (PL III), Stage C. — Median sagittal section of A. spinifer.
Lg. = 5.7 mm. Mesoblastic somites =21. < 50.

PM.

FIG. 5 (PI. Ill), Stage D. — Median sagittal section of A. spinifer.
Lg. — 6.0 mm. x 50.

280

ZELENY.

[VOL. II.

PLATE IV.

MB.—

HB.

/

HC.

ffi/ft

i
/
;
_ _/

nyp. -

JTD —

Ht.

FIG. 6 (PI. IV), Stage E. — Median sagittal section of Chelydra serpentina.

Lg. = 7 mm. X 50.

MB.-—

Hyp.

FB

—KB.

Frc. 7 (PI. IV), Stage F. — Median sagittal section of A. spinifer.
Lg. = 7.5 mm. x 50.

No. 6.]

THE HYPOPHYSIS IX CHELONIA.

28l

PLATE V.

H.B.

FIG. 8 (PI. V), Stage F. — Sagittal section from the sajne series as Fig. 7, but
three sections (40 /u) to the side of the median line, x 50.

HypS.

-H.B.

•EG.

FIG. 9 (PI. V), Stage F. — Sagittal section from the same series as Fig. 7, but
six sections (So /JL) to the side of the median line. X 50.

ON PHORONIS PACIFICA, S/\ NOV.

HARRY HEAL TORREY.

DURING the past summer eight specimens of Phoronis came
into my hands. Five were collected in June, 1894, in Humboldt
Bay, California, by an expedition from the University of Cali-
fornia. Three were brought back from Puget Sound by the
Columbia University expedition of 1897. As the occurrence of
Phoronis on the Pacific coast has never been recorded, and it is
eminently desirable that all localities in which this interesting
form may be obtained should be made known to naturalists,
I have undertaken to describe this material, which represents
a single species hitherto unknown.

The following table will indicate the distribution and date of
first description of all the known species of Phoronis. For the
species presently to be described I propose the name P . pacifica.

P. hippocrepia Str. Wright.1 1856 Great Britain.

P. ovalis " 1856.

P. (Crepind) gracilis Van Ben.'2 1858.

P. Buskii Mclntosh.8 iSSi Philippines.

P. australis Haswell.4 1882 Port Jackson, N.S.W.

P. Kowalevskii (Caldwell) 5 Benham.8 1883 Naples.

P. psammophila Cori.r 1889 Messina.

Port Jackson ?

P. Sabatieri Roule.* 1889 Gulf of Lyons.

P. architecta Andrews. ;i 1890 North Carolina.

P. ijimai Oka.1" 1897 Japan.

P. pacifica. 1901 Humboldt Bay, California;

Puget Sound, Washington.

1 Proc. Roy. Phys. Soc. Edin., vol. i (1856), p. 165; Edin. Neiv Phil. Journ.,
vol. iv (1856), p. 313. '2 Ann. Sci. Nat., 4th ser., vol. x (1858), pp. 11-23.

3 Proc. Roy. Soc. Edin., vol. xi (iSSi), p. 21 1; Chall. Rep. Zool., vol. xxvii
(1888), 27 pp. 4 Proc. Linn. Soc. N.S. W., vol. vii (1882), pp. 606, 607.

5 Proc. Roy. Soc. Loud., vol. xxxiv (1883), pp. 371-383.

6 Quart. Journ. Jllicr. Sci., vol. xxx (July, 1889), pp. 125-158.

7 Dissert. Prag (Juli, 1889) ! Zeitschr.f. wiss. Zool., vol. li (1890), pp. 480-568.

8 Compt. Rend. Acad. Sci. Paris, vol. cix (1889), pp. 195, 196.

9 Ann. Mag. Nat. Hist., 6th ser., vol. v (1890), pp. 445-449.
10 Annot. Zool. Japan., vol. i (1897), pp. 147, 148.

283

284 TORRE Y. [VOL. II.

Cori has discussed in an interesting fashion all save the last
four species. Roule's P. Sabatieri and Andrews' P. arcJiitecta
were apparently unknown to him, though descriptions of both
were published before the date of his manuscript. P. Sabatieri
is known to me only through a meager description. It differs
from the other European forms in size and habit, and approaches
P. architecta in these respects. The latter possesses the simple
lophophore and comparatively small number of tentacles (60)
of the European species. It may be distinguished from them
(with the exception of P. Sabatieri, with which it may prove
identical) by its larger size, its straight tubes and solitary habit,
its strong longitudinal muscles (excepting P. psammophila}, the
presence of a ciliated groove in the digestive tract, and possibly
by a separation of the sexes. While it agrees fairly well with
P. Buskii in size, it differs from that species m the other char-
acters enumerated, as well as in the complexity of the lopho-
phore and the number of tentacles. It is thus more closely
allied to the European than to the Australian and Philippine
forms.

The differences between P. australis and P. Buskii are merely
of habit and size, which has caused Benham to suggest their
identity on the supposition that these differences are due to
dissimilar environmental conditions.

The description of the Japanese species has been inaccessible,
so that I can state nothing with regard to it save its existence.

It is an interesting fact that no one has cared to segregate
the species of Phoronis under more than one generic name, and
indicates the trifling character of the differences which serve to
distinguish them. We may separate them, however, into two
groups widely separated geographically. In the one belong the
European forms, including, perhaps, the American P. architecta.
In the other belong P. australis and P. Buskii. We may dis-
regard the Japanese species on account of dearth of information,
and the P. psammophila which Haswell has found at Port Jack-
son and which may have been brought from the Mediterranean
on a ship's bottom.

P. pacifica occupies a place intermediate between these two
groups both geographically and anatomically, but is somewhat

No. 6.]

PHORONIS PACIFIC A.

more closely related to the American than to the Australian
species. In size it resembles P. Bnskii, as well as in the com-
plexity of the lophophore ; though instead of three coils in the

FIG. j. — Diagrammatic cross-sections of the lopliophores of P. australis or P. Buskii (A),

P. pacifica (5).

c

FIG. 2. — Section through the upper third of P. pacifica; but one quadrant detailed ; semi-diagram-
matic, b, basement membrane ; c, circular muscle ; ec, ectoderm ; /, longitudinal muscle s ;
in, in', mesenteries; «7', median vessel ; «, nerve ; o, oesophagus; r, rectum.

286

TORRE Y.

[VOL. II.

spirals of the lophophore there are one and one-half to two,
and a correspondingly smaller number of tentacles (170-200),
Fig. r, A, B. In the strength of its longitudi-
nal muscles (Figs. 2, 3) it departs ivomP.Buskii
and even surpasses P. architecta, resembling
the latter species in the possession of a
specialized ridge in the digestive tract (though
this does not pass into the groove that Andrews
describes), in the structure of the nervous sys-
tem, lophophore organ and tube, in habit, and
in the possible separation of the sexes.

It is not my intention to enter into a full
and detailed anatomical description of Phoro-
nis, which the labors of Benham and Cori have
rendered largely unnecessary. But a few
words on some points may not be out of place.

In one of the Puget
Sound specimens sper-
matogonia and sperma-
tocytes were found
packed around the blood
vessels in the aboral regions of the body,
but no spermatozoa nor ova. The aboral
ends of all the Humboldt specimens were
wanting, so it was impossible to deter-
mine whether they were monoecious or
dioecious. In one nephridium three ova
were found unaccompanied by sperma-
tozoa ; the first polar body was forming
in one. As there was no sign of sperma-
tozoa in these eggs, it is probable that in
this species fertilization takes place either
externally or within the nephridium. It
is quite possible that the sexes may
be separate, or ova and sperm may be produced by the same
individual but not simultaneously.

The blood corpuscles have a conspicuous yellow color and
measure 10-15^ in diameter.

'•«N

FIG. 3. —Cross-section
of one longitudinal

FIG. 4. — Cross-section of the
oesophagus of P. j>aci_fica,
showing the ridge, its base in
contact with the median lon-
gitudinal blood vessel.

No. 6.]

PHOKOXIS PA CIF1 CA.

287

The ciliated ridge was present for a considerable distance in
the oesophagus, but could not be seen in the stomach either as
a ridge or a groove. In Figs. 2, 4 it is indicated in section,
where it appears to be a shallow groove, an appearance probably
due to the folding of the wall of the oesophagus. Its position
relative to the longitudinal blood vessels is identical with that
described for /'. arcJiitecta. The nuclei stain more intensely
with haematoxylin than the other
nuclei of the oesophagus, and are
crowded together usually in several
layers. These facts make the area
quite conspicuous in section.

The muscles reach their greatest
development in the oral third of
the body, where they form more
than eighty high narrow ridges. In
the aboral third these are reduced

. FIG. 5. — Semi-diagrammatic cross-section

to a very inconspicuous layer, though uf ioph0phore organ within the

still retaining their identity, being ity of the lophophore.
separated throughout their length by characteristic folds of
peritoneal epithelium.

There is a delicate peritoneum covering the muscle ridges,
the nuclei only (Fig. 3, ;/) being seen with ordinary powers of
magnification. Occasionally a similar nucleus is found within
the fold of muscle (;/').

The nervous system is constructed as in P. arcJiitecta, with
one interesting exception. The two longitudinal cords, which
are of exceedingly unequal length, instead of ending in the
nerve ring of the lophophore, are continuous across the median
line at the level of the median mass of ganglion cells. The
loop thus formed is closely applied posteriorly to the latter and
.touches the lophophore nerve on each side of the rectum, appar-
ently without fusing at either point. Just how intimate this
contact is cannot be determined from my poorly fixed material.
The brevity of the descriptions of this portion of the nervous
system in P. architecta and other species leads me to suspect
that the seemingly exceptional condition in P. pacifica may
prove to be of more general occurrence.

288 TORREY.

The lophophore organ is extremely variable and may be
present or absent, as in P. psammophila. It may resemble
that of P. australis, though differing somewhat in shape
(Fig. 5). In this case it is simple, with a thickened glandular
epithelium lying for the most part against the inner circle
of tentacles, and an outer free non-glandular edge of much
lower cells. In another case, however, it had the form of the
same organ in P. arcJiitecta and P. psammophila, as described
and figured by Andrews and Cori, being composed of a basal
lobe and a distal "carpel-like organ." This condition seems
to have been attained by the addition of the "carpel-like organ"
to the structure (basal lobe) which corresponds to the entire
organ in P. australis.

The following is a diagnosis of the species, from material
preserved in alcohol and formalin :

Total length may be 9 cm., of which the tentacles represent from 2^ to

4 mm.

Diameter, i \ to 2 mm.
Lophophore spirally coiled, each spiral possessing from ii to 2 complete

turns.

Tentacles 170 to 200.

Lophophore organ present or absent ; extremely variable in form.
Each animal occurs singly and completely fills tube.
Tube straight, cylindrical, composed of delicate chitin, encrusted with fine

sand grains.
Ridge of thickened epithelium in the descending limb of the digestive canal,

just beneath the median longitudinal blood vessel.
Longitudinal nerve trunks unite across median line between mouth and

anus.
Longitudinal muscles in numerous very high and narrow folds which reach

their maximum in the distal third of the body.
Sexes possibly separate.

Localities: Puget Sound, Washington; Humboldt Bay, California, on sand
and mud fiats that may be uncovered by the tide.

COLUMBIA UNIVERSITY, January, 1901.

ON MUSCLE REGENERATION IN THE
LIMBS OF PLETHEDON.

ELIZABETH W. TOWLE.

SPALLANZANI (1768) and Bonnet (1777) showed that a
salamander whose limbs have been cut off has the power to
regenerate new ones. This discovery has been confirmed by
later writers, and although some histological work has been
done, yet the method of regeneration of the muscle bundles
has not been worked out. There are several possibilities :
first, the old fibers might break clown at the cut ends and the
new ones develop from the indifferent tissue so formed, each
old muscle thus completing itself independently. Or, the
cut muscles might degenerate along their entire length, and
new ones take their place ; or some of the old muscles might
degenerate, new ones being formed from this tissue, while some
fibers might break up into smaller new fibers. An attempt
has been made in this work, not so much to follow the origin
of the cells in detail as to discover the general processes
taking place in the leg that lead to the formation of the new
muscles. The regenerating limbs of Plethedon cinercns were
used. They were studied by means of serial sections.

In addition to this histological study, I have also experimented
on a number of American urocleles in order to see in which ones
regeneration of the limbs takes place. For this purpose a
number of the commoner forms have been studied, and in
connection with these results a statement is given of the
previous observations on European forms.

I.

MetJiod.--Q\-\o. of the anterior limbs of Plethedon cinercns
was removed halfway between elbow and hand. The regen-
erating limbs were put up at intervals varying from four days

289

290 TOIVLE. [VOL. II.

to two weeks. They were fixed in corrosive acetic, hardened
for two or three days in 95^0 alcohol, and then decalcified for
from six to eight days in a nitric acid solution (HNO^ sp. gr.
1.42, 2 vols. + H2O, 98 vols.) which was changed daily. They
were finally hardened again for three days in 95^0 alcohol,
embedded and cut. Some limbs were stained in toto with borax
carmine, but the best results were obtained by the method
used by Byrnes ('82), viz., staining on the slide in Delafield's
haematoxylin, followed by a wash of picric acid in absolute
alcohol. This latter method differentiates the muscle substance
very clearly.

Eleven stages were preserved at the following intervals :

Time of Operation. Time of Killing. Age of Stump.

1. May 4, 1900.' May 14, 1900. 7 days.

2. " " 18, " it "

3. >l " " 29, " 22 "

4. Oct. 22, 1899. NOV. 20, 1899. 29 "

5. •• " Dec. 5, " 44 "

6. '' " 15, " 54 "

15, « 54 «

8. " " Jan. 5, " 75 «

9. - " " « 5, « 75 «

10. " " 20, " 90 "

11. March 22, '• 151 "

Transverse sections of this series were cut. Nos. i, 2, and
3 were stained with haematoxylin and picric acid ; 4, 5, and 6
with borax carmine ; 7 with borax carmine and picric acid ;
8 and 9 with Biondi-Ehrlich solution ;2 10 and n with haema-
toxylin and picric acic. In addition, normal limbs were cut
and stained in a similar manner and used for comparison.

Results. - - For convenience in description I shall consider the
sectioned limb as made up of three Regions : I, that between
the cut and the elbow; II, the Region just above the cut ; and
III, the growing end. In the earliest stages Region III does
not, of course, exist.

1 It will be noticed that stages i, 2, and 3 were preserved later in the year, but
observations will be described in the above order.

2 This stain was not successful, and the stages were replaced by one stained
with haematoxylin and picric acid.

No. 6.]

THE LIMBS OF PLETHEDOX.

291

In the first of the series of transverse sections changes in
the cut muscles are already noticeable, the most striking being
the increase in the number of nuclei, especially in the outer
fibers of the limb. This increase can be seen in Region I as
far up as the origin of the muscles at the elbow. In the outer
fibers the muscle tissue is becoming thinner and disappearing,
and while the outlines of fibers and bundles are not lost, they
are much less clear than in the normal limb. The inner
bundles, however, are but little affected, and extend unbroken
to the cut end, which is at this time not yet entirely covered
by ectoderm. No mitosis is seen in this section.

In stage 2 the changes are more marked. There has been
a continued increase in the number of nuclei in Regions I and
II, the outlines of the outer fibers and bundles are lost, while
the muscle substance has disappeared except for disintegrating
fragments here and there, contrasting sharply with the thin
cytoplasm of the neighboring cells. The inner fibers still extend
to the end of the limb, which is now covered entirely by several
layers of ectoderm. In the neighborhood of the cut three or
four mitotic figures are to be found. In the third stage Region
III begins to appear as a small knob of undifferentiated tissue
behind the cap of ectoderm. In this knob
and for a short distance above it among the
outer cells karyokinesis is not uncommon.

If we compare the following stage (4) with
the normal limb, the principal changes that
have taken place will be very clearly brought
out. In Region I the increase in number of
nuclei is very great. Even as far as the elbow
two to four nuclei may be found in a section
of a single fiber, often crowded together so as
almost to fill it (Fig. i). Many nuclei are also
scattered between the fibers. Below the elbow
the number of nuclei increases, the outlines
of the outer fibers are completely lost, and
the outer half of the limb, which is normally
solid muscle, is seen to be made up of a dense mass of nuclei
surrounded by loose protoplasmic substance, with here and there

FIG. i.

TOWLE.

[VOL. II.

clumps of disintegrating muscle tissue. The outlines of the
inner fibers are somewhat less distinct than in the preceding
stage, and some of the bundles seem to be splitting up. This
condition is represented in Fig. 2. Most significant is the fact
that at no stage is any karyokinesis found among the muscle
fibers, although the increase in the number of nuclei is enormous.
Passing through Region II of this stage, the number of old
fibers decreases and the scattered nuclei increase, until at the

FIG. 2.

plane where the cut was made all old fibers disappear and we
reach Region III, which is made up entirely of closely crowded
nuclei, each surrounded by a small amount of protoplasm.
Karyokinesis is first seen in II, a short distance above the cut,
among the outer cells, never in t/ie muscle fibers, and the num-
ber of dividing cells increases toward the growing tip until it
becomes quite large.

In the fifth stage a further difference is to be noted. We
find in Region I, in the inner part of the limb, the old muscle

No. 6.]

THE LIMBS OF P LETHE DON.

293

fibers, often with several nuclei, and broken frequently into
quite small fragments. Outside these are numerous nuclei, as
before, but now surrounded by very distinct muscle tissue
(Fig. 3). This tissue has formed often about several nuclei
in a group, and has no distinct walls ; there seem to be as yet
no distinct muscle fibers. But the line of separation between
the old muscle fibers and the new tissue is distinct, and a few
old fibers can be traced to the region of the cut, although at
that level the greatest
part of the tissue is
new. New muscle sub-
stance has appeared
about the nuclei for a
short distance below the
cut ; but it decreases as
we pass down, until we
find only crowded nuclei
and thin protoplasm. In
this region, as before,
numerous karyokineses
are seen.

In Region I of the
sixth stage all the mus-
cle fibers are small and
the definite line between
the old and new is lost.
The majority of the
fibers contain in cross-section but one nucleus, though some
may contain two or three, and in general the smallest fibers
and most nuclei are on the outer side of the limb. This is espe-
cially noticeable in Region II, where the outer (new) fibers are
exceedingly small. The muscle tissue decreases in amount as
we pass to Region III, until it is all lost. Cells dividing by
karyokinesis appear at this level.

Stage 7, though of the same age as the preceding, is
somewhat further differentiated, and in this the new fibers
are more rounded and have assumed a more characteristic form
(Fig. 4). A comparison with the normal shows smaller fibers

FIG. 3.

294

TOWLE.

[VOL. II

and great excess of nuclei, there being often two or three to
one fiber and many outside the fibers.

The later stages need not be described in detail. As the
limb grows longer the formation of new muscle tissue pro-
gresses farther down toward the tip, the new fibers being
always small and containing several nuclei. The number of
nuclei outside the fibers decreases, until in stage 1 1 the
muscles look quite normal, and the number of nuclei is exces-
sive only in the region of the foot, which is at this time

FIG. 4.

clearly differentiated. Karyokinesis is found, I believe, without
exception, near the growing end, never in the upper regions.

The first appearance of any definite grouping of cells appears
in stage 5, where the arrangement into bundles is foreshadowed.
As the fibers form, the division into bundles becomes more
distinct, until in stage 1 1 they are all differentiated as far
down as the foot, and here we can see by the arrangement of
nuclei where the bundles are to be.

In the process of regeneration described above there are
certain things to which I wish to direct especial attention. In

No. 6.]

THE LIMBS OF PLETHEDON.

295

the first place there is a great increase in the number of nuclei
zvit/iin the old fibers, but in no case is any karyokinesis found
there. This degenerative process in the old fibers must there-
fore take place by direct division of the nuclei. Instances of
this division are shown in Figs. 5 and 6. To this division and
to the disintegration of some of the old fibers is due the enor-
mous accumulation of nuclei in the outer part of the limb
(Fig. 2). The cells so formed then begin to divide by karyo-
kinesis in the region of the cut, and thus a further increase in
their number takes place. In these outer cells new muscle
tissue forms and the new fibers are built up. A certain

FIG. 5.

FIG. 6.

number of the old fibers remain in the middle of the limb,
and in these the muscle tissue never disintegrates, though
it splits longitudinally.

Again, as at any one level the number of nuclei far exceeds
the number of fibers in a normal muscle, a great number of
them must, between the early stages and the fully formed limb,
either degenerate or be transported (cf. Figs. 3 and 4). That
this is so is easily seen, for two reasons: (i) when the new
fibers form, at a given level several nuclei are often included
in one fiber ; when the limb is full-grown there is only one ;
(2) among the newly formed fibers, but between them, are many
scattered nuclei ; the majority of these disappear in later stages.

296

TOWLE.

[VOL. II.

One further point should be mentioned. In stage 5 the
line of distinction between new and old fibers is clear, owing
to their difference in size. In stage 7 this distinction has
disappeared. This is due not only to an increase in the size
of the new fibers, but to a decrease in the old. This decrease
is, I believe, due to the longitudinal splitting of such of the old
fibers as are left (Fig. /. a and b}.

Beside the stages described above, four others were preserved
and cut longitudinally. This was a somewhat difficult opera-
tion, for the new part forms at
an angle with the upper arm,
and it is hard to orient the
piece in such a way as to insure

FIG. 7 a.

FIG. i b.

true longitudinal cutting of the muscle fibers. The material
was preserved at the following intervals :

Time of Operation.

i. Jan. 29, 1900.

-. .. 4 i ..

-) t.

4.

Time of Killing.

Feb. 2, 1900.

Age of Stump.

4 days.
14 «

21 "
30 "

The first of these shows no transformation at the cut ends
of the muscles, but the ectoderm has closed in over the wound.
In the second there is a considerable increase in the thickness
of the ectoderm, and under it a collection of scattered nuclei,
exactly similar to the tissue in the growing end of later stages.
As the limb grows in length this tissue increases in amount
and a number of mitoses are seen in it. Degeneration of the
old fibers is distinctly noticeable in stage 3, and fibers are
found, as before, filled with nuclei, but there is no karyokinesis
in any muscle fiber.

No. 6.] THE LIMBS Ol- PLETHEDOX. 297

SUMMARY.

The main changes that take place in the muscles of Flcthcdon
cincrcus during the process of regeneration are as follows :

i (a] In the cut muscles the nuclei divide directly and in the
outer bundles the fibers disintegrate, leaving masses of nuclei
with a small amount of cytoplasm, (b) Some of the cells so
formed later divide mitotically, and by them new muscle sub-
stance is laid down, (r.) As the number of nuclei is, however,
far in excess of the normal number of muscle fibers, many nuclei
must degenerate or be transported.

2. The fibers of the inner bundles do not disintegrate, but
split longitudinally, giving rise to smaller fibers, which are soon
indistinguishable from those formed as described in i.

3. The arrangement into muscle bundles first becomes
clear at the end of about six weeks.

4. There is no change in the muscles of the upper arm.

II.

The following is a brief summary of the main observations
that have been made on the power of regeneration of the limbs
in European species of salamanders.

Spallanzani (8), in 1768, published a number of observations
on aquatic salamanders, presumably species of Triton. In these
he found that any or all of the limbs will regenerate, no matter
what the species, size, or age of the animal, the larger ones
regenerating more slowly than the smaller forms. Regenera-
tion will take place, he maintained, even when the limbs are
disarticulated.

Bonnet, 1777 (2), confirmed Spallanzani's observations, finding
that in Triton cristatus the hands and fingers will regenerate.

Von Siebold (7), in 1828, recorded abnormal regeneration of
the fingers of Triton cristatus.

Higginbottom, in 1847 (4- P- 29)> observed that in Triton the
limbs will regenerate at a temperature of from 48° to 57° F.

Philippeaux (6), in 1866, was the first to prove conclusively
that limbs when disarticulated will not regenerate. His experi-
ments were made on Triton cristatus and Axolotl.

298 TOWLE. [VOL. II.

Dumeril (3), in his paper of 1867, in the course of other
observations, notes the fact that in an Axolotl the two anterior
limbs regenerated after injury.

Wiedersheim, 1875 (r> P- 95 )> found that the toes of Triton
cristatus will regenerate, while there is no regeneration of the
limbs in Proteus and Siren lacertina.

Erber, 1876 (4, p. 34), notes regeneration of the feet of
Siren lacertina.

Goette, 1879 (5), records the regeneration of a leg of Proteus
after a year and a half. Regeneration also occurs in Amphiuma
and Siren, in Triton cristatus, T. tacniatns, and their larvae.

Weismann (9), in The Germ Plasm, 1893, says that the limbs
of Salamandra regenerate, while in Triton marmoratus regen-
eration is slight or absent.

Barfurth (i), in 1894, reports regeneration of the feet and
digits of Triton tacniatus and Siredon pisciformis . This regener-
ation is normal or abnormal according to the plane and method
of the injury.

I have experimented on the following forms, removing a fore
foot and a hind foot from different individuals of each species :
PletJicdon cincreus, Spelerpes ruber, S. giittolincatus, Desrnog-
natJtus ocJiropJiaea, Manculns quadridigitatns, Amblystonia
opacum, Diemyctylns viridesccns, AnipJiiuma means, and Nec-
tnrns inaculatus. Of these, all have regenerated.1 The regen-
eration in Spelerpes, Desmognathus, Manculus, and Ambly-
stoma was comparatively rapid, and new limbs were well
formed in four months, though they were somewhat smaller
than the old limbs. Diemyctylus was slower in reaction,
while the first Necturus to show a distinct regenerated
stump did so only after eight months. Other individuals of
the same species showed no regeneration even at that time.

I desire to thank Professor T. H. Morgan, under whose
direction this work was undertaken, for his kindly assistance
during its progress.

BIOLOGICAL LABORATORY OF BRYN MAWR COLLEGE.

1 Amphiuma means was observed for only eleven weeks. At that time regen-
eration was slight. The regenerated stump was not sectioned.

No. 6.] THE LIMBS OF PLETHEDON. 299

LITERATURE.

1. BARFURTH. Arch. f. Entwickelungsmechanik der Organismen. 1895.

p. 91.

2. BONNET. Mcmoire sur la reproduction des membres de la salamandre

aquatique. CEuvres completes. Tome xi. I. Memoire. 1777.

3. DUMERIL. Nouvelles archives du museum d'histoire naturelle de Paris.

Tome iii. 1867. p. 1 19.

4. FRAISSE. Die Regeneration von Geweben und Organen bei den

Wirbelthieren. Cassel und Berlin. 1885.

5. GOETTE. Ueber Entvvickelung und Regeneration des Gliedmassen-

skeletts der Molche. Leipzig. 1879.

6. PHILIPPEAUX. Comp. Rend. 1866. p. 576.

7. SIEBOLD, VON. Observationes quaedam de salamandris et tritonibus.

Dissert. Berolini. 1828.

8. SPALLANZANI. Precis sur les reproductions animales. 1768. p. 79.

9. WEISMANN. The Germ Plasm. New York, Charles Scribner. 1893.

THE FACTORS THAT DETERMINE REGEN-
ERATION IN ANTENNULARIA.

T. H. MORGAN.

THE following experiments were carried out at the Naples
Zoological Station during June and July, 1900. As there may
be no opportunity in the immediate future of completing the
observations, I have determined to publish them as they stand,
in the hope that the results may stimulate some one, so situated
as to obtain the necessary material, to take up the questions
here raised and to bring them to a more satisfactory conclusion.

Loeb's experiments on Antennularia, made in 1892, show
that pieces of the stem suspended in sea water always regen-
erate roots at the lower end and a new stem at the upper end.
The result was the same whether the apical or the basal end
of the piece was uppermost, i.e., whether the piece had a normal
or a reversed orientation. Similar results were obtained when
pieces were suspended obliquely, the high end producing
always the new stem and the low the new roots, etc. These
results are similar to certain results that have been obtained
in plants, although Vochting has shown conclusively in many
forms that the polarity of the piece is a much stronger factor
in determining the regeneration than is gravity. Loeb drew
the natural inference from his results, 77.7., that gravity deter-
mines the kind of regeneration that takes place at the ends of
the piece. Driesch,1 who examined later the regeneration of
Antennularia, found that when a piece of the stem is so placed
"that its basal end is freely surrounded by water," a large
number of roots are formed from that end. If the end with
its roots is cut off, there is generally formed from the cut end
a few new roots, but also always a more or less delicate stem
composed of a few tubes. This stem is negatively geotropic.

1 Driesch, H. Studien iiber das Regulationsvermogen der Organismen, I.
Rotix's Archiv. Bd. v, p. 383.

301

302 MORGAN. [VOL. II.

If the same end is again cut off, there develops rarely one or a
few roots, but generally two or three vigorous stems. If the
operation is repeated a fourth time, one or two stems are with-
out exception produced.

There is no statement made by Driesch as to how these
pieces were orientated with regard to gravity, but the results
show that another factor than gravity has an influence on the
regeneration. Unfortunately nothing is said with regard to
what has taken place at the other end of the piece. I shall try
to show that it is not improbable that this may be also a factor
in the result, and if so it is possible that Driesch's results are
due to this rather than to the action of the water on the free
basal end, or at least both factors may be present.

My experiments were primarily undertaken in order to see
how pieces would behave when fixed to a revolving wheel, but
on account of the apparent disagreement between Loeb's and
Driesch's results, it was first necessary to repeat the experi-
ment of suspending pieces with two cut ends in order to see
how far gravity acted upon them. In one series of experiments
pieces were suspended in an aquarium by means of a silk thread.
Some of these had the apical end upwards, others the basal
end upwards, and still others were suspended horizontally. In
nearly all cases roots developed in the course of a few days
from both ends. If the ends were cut off, new roots developed
again on both ends ; although in one or two cases in which the
apical end was uppermost a stem developed at that end. The
pieces were from 3 to 5 cm. long.

By means of another device the experiment can be much more
satisfactorily carried out. A small square piece was cut from a
sheet of cork and a hole bored in its middle. The end of a glass
rod, about 20 cm. long or longer, was pushed through the hole
in the cork. If the piece of cork is neither too large nor too
small, the glass rod, when put into an aquarium, will sink to the
bottom until one end touches, but the other end will be held up
in a vertical position owing to the buoyancy of the cork.

Pieces of the stem of Antennularia were fastened to the
sides of the cork by means of two dried cactus spines, that were
crossed over the stem and stuck into the cork.

No. 6.] REGENERATION IN ANTENNi '/.. I AY. /. 303

In these experiments, made with pieces of different lengths,
and from different parts of the old stem, the results were the
same as before. In nearly every case roots developed from
both ends,, and even after these ends had been once removed.
The experiments extended over two or three weeks. Whether,
if continued longer, a stem would develop at the upper end
among the roots there present, I do not know, but the results
suffice to show that the most characteristic thing that occurs
is the production of roots from both ends.

I was, therefore, not a little surprised to find in another
series of experiments that a different result occurred. I
placed in an aquarium a number of pieces of Antennularia
that remained attached to the stones on which they had been
found growing. Most of the pieces stood up vertically from
the floor of the aquarium with the apical end upwards ; a few
pieces I suspended in an inverted position, i.e., with the apical
end downwards and the attached basal end upwards. In the
former cases the apical ends did not produce roots at all, but
a new stem. In the latter cases, in which the pieces were
inverted, the apical end produced neither roots nor stem.
Although these pieces were observed for only ten days, the
time is ample to show that the pieces behave differently from
pieces with two cut ends. I regret that I could not carry
these experiments further.

One other result should be described, since it seems to have
a direct bearing on the last experiment. In one case a very
small piece had sunk to the bottom of a dish of water, where
it stood with its basal end in contact with the glass. It lay
there undisturbed, and attached itself at its basal end by means
of new roots. TJie apical end produced a shoot. This result,
taken in connection with the preceding experiment, seems to
indicate that the development, or the presence of roots on
the basal end, prevents the development of roots on the
apical end. This result, if it prove constant, opens the way
for several interesting experiments that so obviously suggest
themselves as to require no further mention.

A few experiments were made with a rotating wheel con-
structed for the purpose. The wheel consisted of two parallel

304 MORGAN. [VOL. II.

rings of wire between which, at equidistant points, were
sixteen paddles (5 cm. x 8 cm.) made of oblong pieces of
sheet-copper; spokes (13 cm. long) attached the rings to an
axis that rotated in two sockets. When the wheel was immersed
in the water of an aquarium, and a stream of water from the
tap was made to play (beneath the water) on the plates, the
wheel slowly revolved, making about five and one-half revo-
lutions in a minute.1

Pieces of Antennularia were attached to the wheel in the
following way. Sheets of cork of the same size as the copper
plates were attached to the underside of the latter by wire or
string. Pieces of the hydroid were fixed to the cork in different
positions by means of crossed cactus spines. The pieces were,
on an average, about 15 cm. from the axis of rotation. The
results were entirely negative. None of the pieces produced
either roots or stems ; and the pieces died sooner than did those
in other experiments. As this experiment was carried out in
a different aquarium, I cannot be certain that the death of the
pieces was not due to other causes than to the rotation.
Furthermore, it is not evident from the experiment whether
the rubbing of the moving ends against the water suppressed
the regeneration, or whether the result is due to the continuous
change of position in regard to gravity. The rotation was too
slow for the action of the centrifugal force to have played any
important part. Since other experiments have shown that roots
may develop at both ends of a piece suspended vertically, it
is improbable that in the rotating pieces the changing posi-
tion in regard to the action of gravity can account for the
result, and it is much more probable that the motion of the
piece through the water interfered with the regeneration at
the ends. The experiment needs to be repeated more often, and
other check experiments carried out in the same tank.

The work that I have done on the regeneration of Antennu-
laria, while incomplete in many ways, shows at least that other
factors than gravity enter into the result. I do not question the
main part of Loeb's results, for they seem to show that gravity

1 This wheel was left at the Naples Station in the hope that it might be used
by some one to continue the experiments.

No. 6.] REGENERATION IN ANTENNULARIA. 305

is a factor in the regeneration of this form ; but the development
of roots at both ends that first takes place, as I have found,
but which Loeb did not observe, and the behavior of pieces
attached at one end, as described in the preceding pages, show
that the factors determining regeneration are more involved
than previous results seem to indicate.

BKYN MANVR CUI.I.KCK, Feb. 4, 1901.

MENDEL'S LAW OF DICHOTOMY IN HYBRIDS.

C. B. DAVENPORT.

IN the study of hybrids we must, as De Vries (I900b) truly
says, no longer pay primary attention to the degree of differ-
ence between the forms united - - to whether they are species,
subspecies, or varieties-- but to the behavior of the peculiar
characters by which the crossed individuals and their ances-
tors are distinguishable. For each of these somatic characters
corresponds to some peculiarity of the germplasm. The be-
havior of the differing characters when united in the hybrid is
diverse ; three categories have long been recognized (Galton,
1888, p. 12). These are: (i) blending heritage, illustrated by
skin color in man ; (2) alternative heritage, illustrated by
human eye color ; and (3) mixed heritage, illustrated by the
piebald condition of the progeny of mice of different colors.
The law of dichotomy in hybrids applies only to the second
class, --alternative heritage, -- although it has recently been
brought forward by De Vries (1900) as the almost universal
law of inheritance in hybrids. The law itself was first enun-
ciated very clearly and completely by Mendel (1865) and
deserves to bear his name. The law was, however, forgotten.
It has been rediscovered independently by De Vries and by
Correns (1900), both of whom are able to add new evidence of
its validity (for alternative heritage !).

In his illustration of Mendel's law, De Vries first classifies
hybrids into monohybrids, dihybrids, and polyhybrids, accord-
ing as their parents differed in one character only, or in two
characters, or in many characters. The case of inheritance in
monohybrids is the simplest, and will be first considered.
Mendel's and De Vries's investigations have established the
following principles :

i. Of the two antagonistic peculiarities the hybrid exhibits
only one ; and it exhibits it completely, so as not to be

3°7

308 DA VENP ORT. [VOL. I L

distinguishable in this regard from one of the parents. Inter-
mediate conditions do not occur [in alternative heritage].

2. In the formation of the pollen and the egg cell the two
antagonistic peculiarities are segregated ; so that each ripe
germ cell carries either one of these peculiarities.

Of the two antagonisti-c peculiarities united in the hybrid,
that which becomes visible in the soma is called by Mendel the
dominating; that which lies latent is called the recessive char-
acter. What determines which character shall be dominating
is still unknown, and the determination of this point offers an
enticing field of inquiry. In some cases the dominating form
is the systematically higher; in others it is the older or ances-
tral form.

The law of dichotomy may now be developed. When a
hybrid (monohybrid) fertilization takes place the zygote con-
tains both the dominant quality (abbreviated d) and the reces-
sive quality (/'). In the early cleavages d and r are both passed
over into both the daughter-cells ; but apparently, at the time
of segregation of the germ cells, the somatic cells are provided
with d only, while the germ cells retain both qualities. In the
ripening of these germ cells, probably in the reduction division,
d and r come to reside in distinct cells, so that we have

of the female cells 50% d + 50% r, and
of the male cells 50% d + 50% r.

If now hybrids are crossed haphazard, a male d cell may
unite with either a female d cell or with a female r cell ; like-
wise a male r cell may unite with a female d or a female r cell.
Consequently in the long run we shall have of all the zygotes

25% d,d + 50% d, r + 25% r, ;-,

or 50^ of the zygotes hybrid and 5o'/< of pure blood, and of the
latter half exclusively maternal and half paternal. But since
the soma developed from the hybrid germ cell has the domi-
nant character, we shall have

75% of the cases with the dominant character
25% " '• " " recessive "

and this agrees with various empirical results, of which the
following from Correns is instructive. A cross was obtained

No. 6.]

LAU' OF DICHOTOMY AV HYBRIDS.

309

between a species of pea with a green (g) germ and one
having a yellow (y} germ. Yellow is dominating.

Gen. i.

Gen. 2.

Gen. 3.

(hybrid) peas; produced 12 plants;
these bore :

775,1' (hybrid + y) peas (= 75-8%)-
21 plants were produced.

7 (33 Pure-
blooded r,

because tl ey
bore:

292 y peas.

4 (66%) hybrids,
because they
bore :

247..V (pure-blooded)
peas ( = 24.2%).

20 plants bore:

A

462 y 149 g 670 green peas,

(hybrid + y) (pure-blooded)
peas (= 76.4%). peas (= 23.6%).

It is clear that if this process of crossing of the hybrids
continues, the propo rtion of hybrids to the whole population will
diminish ; for the share of pure-blooded forms breeds true ;
while the originally equal share of hybrids is repeatedly halved.

If the hybrid is crossed with one of the parents instead of
with another hybrid, we will

get (d + r) d = d, d + d, r, and
(d + ;-) r -- ef, r + ;-, r.

In the first case all of the progeny will appear of the dominant
type. In the second case one-half will appear of that type.
This again agrees with experiment.

In the case of dihybrids the law of alternative heritage is
somewhat more complicated. Imagine a lot of ripe germ cells
with the antagonistic qualities of any pair separated according
to the second principle stated at the outset. A indicates the
one pair of qualities and B the other ; then we shall have nine
classes of zygotes, the proportion of each of which is as follows :

A.

B.

, d

50% d, r

6.25% d',d' ; I 2.5% d, r; 6.25% r, r.

A. 25% r, r

12.5% d, d; 25% d, r ; 1 2.5';; /-, r.

B. 6.25% d, d; i 2.5% d, r; 6.25% r, r.

310 DAVENPORT.

Thus the first class has 6.25^0 purely dominant in both charac-
ters; the second class, 12.5^ purely dominant in one character
and hybrid in the other, and so on. Recalling that hybrid
zygotes produce somas with the dominant character, it follows
that the progeny appear as follows :

A. dom. + B. rec 18.75%

A. rec. + B. dom 18.75%

A. dom. + B. dom 56.25%

A. rec. + B. rec 6.25%

A result which again agrees with experiment. The resulting
mixtures of characters in tri to polyhybrids may be likewise
predicted, by extending the principles already laid down.

BIBLIOGRAPHY.

1900 CORRENS, C. G. Mendel's Regel iiber das Verhalten der Nachkom-

menschaft der Rassenbastarde. Berichte der dentscJien Botanischai

Gesellschaft. XVIII. Jahrgang. Heft 4, pp. 158-168. May 23,

1900.
1900 DE VRIES, H. Sur la loi de disjunction des hybrides. Comptes

Rendus de VAcad. des Sciences. Paris. March 26, 1900.
1900b DE VRIES, H. Das Spaltungsgestz der Bastarde. Berichte der

deutschen Botanischen Gesellschaft. XVIII. Jahrgang. Heft 3.

pp. 83-90. April 25, 1900.
1900c DE VRIES, H. Sur les unites des caracteres specifiques et leur

application a 1'etude des hybrides. Reinie generale de Botanique.

XII. pp. 257-271. July 15, 1900.
1889 GALTON, F. Natural Inheritance. New York. Macmillan & Co.

pp. 259.
1865 MENDEL, G. Versuche iiber Pflanzenhybriden. Verh. des A'atttr-

forscher- Vereins in Briinn. Bd. iv, p. i .

REGENERATION OF PROPORTIONATE
STRUCTURES IN STENTOR.

T. II. MORGAN.

THE important results of Gruber and of Balbiani on the
power of regeneration of pieces of Stcntor cocruleus opened
the way for further experiments; and the works of Johnson and
of Lillie on the same form have added some further results of
interest. There remained, however, one problem that had not
been touched upon by these investigators, an answer to which
is needed to make more complete our knowledge of the regen-
eration of unicellular forms. I refer to the question of the
proportionate development of the new organs in pieces of
different sizes, and from different parts of the body; and also
the no less important question of the change in size of old
organs that may be present on the piece at the time of its
removal. It is the purpose of the present communication to
describe certain experiments that bear on these questions.

Although it is evident, in a general way, from the figures
given by Gruber and by Balbiani that a small piece produces a
smaller peristome than does a large piece, yet their figures do
not show definitely that such is the case, and, in fact, it would
be difficult to determine that such is the case from observa-
tions made on the swimming animals. The figures that have
so far been published represent the new stentor as it appears
while contracted or when swimming. To obtain sufficiently
accurate data for the problems that I wished to examine, it was
necessary to make the measurements and drawings from the
stentor at rest when in a fully expanded condition. The
object of my work was to find an answer to the following
questions : i . Do small pieces produce a new organism having
the typical proportions of the normal; and does it make any
difference in this respect as to the part of the stentor from
which the piece is taken ? 2. If a piece containing the old

3"

312 MORGAX. [VOL. II.

peristome is cut off, will it retain the old peristome, or absorb
it and produce a new one of proportionate size ? If the old
peristome persists, will it decrease in size until it has assumed
the typical proportions? 3. If a part only of the old peri-
stome is left on a piece, will the missing parts be regenerated
from it, or will a new peristome develop ?

There appeared during January and February in one of the
aquaria in the laboratory a large number of stentors, whose
presence seemed to be connected with the appearance of vast
numbers of vorticellas, on which they fed. The operation of
cutting the stentor in two or more parts was carried out either
by means of small scissors, or, in most cases, by a sharp scalpel.
The latter operation is greatly facilitated by placing the ani-
mals in a dish of water, the bottom of which is covered by a
layer of paraffin.

The following measurements give the length of the normal
blue stentor and the greatest width of the peristome.

Length. Width of Peristome.

2.8 mm. .52 mm.

1.6 .46

.4 .40

•7 -50

•7 48

.6 .44

•9 48

If a stentor is cut in two by a cross-cut, as indicated in
Fig. i, A, a-a (the anterior piece, B, being smaller than the
posterior, C), the cut surfaces of each piece are closed almost
instantly by the outer layer bending over the exposed part.
Only a faint, clear line on the surface indicates where the cut
has been made. The history of the anterior piece is as fol-
lows: In the course of an hour or two the piece becomes some-
what more pointed at the posterior end, and then fixes itself
by a foot that appears at that end (Fig. i, B}. The posterior
end now begins to draw out into a stalk, and after thirteen
hours the piece has assumed the form shown in Fig. i, B1.
The piece is still proportionately too broad for its length, for
although the peristome has become reduced in size it is not as

No. 6.] PROPORTIONATE STRUCTURES IN STENTOR 313

B'

B;

B

C3

C1

FIG. i. — A, contracted stentor to show where the cut was made, a-a ; B, anterior end live
hours after the operation; £l, an anterior piece thirteen hours after operation ; />'2, B3,
anterior pieces twenty-four hours after operation ; B*, anterior piece forty-eight hours
after the operation.

C, posterior or foot end of A ; C1, posterior end five hours after operation ; C2, C3, pos-
terior ends twenty-four hours after operation.

314 MORGAN. [VOL. II.

yet reduced sufficiently to give the piece the typical propor-
tions. Two other head-pieces of this same series are shown in
Fig. i, Z?2, />3, that were drawn twenty-four hours after the
operation. It is even more evident in these (compare with
Fig. 2, A, for normal) than in the last that the peristome is too
broad for the length of the stentor. Even after another
twenty-four hours one of the pieces had still retained the same
form as shown in Fig. i, /?4.

The posterior piece, C, fixes itself at once by the old foot,
and may soon elongate to its full length. In the course of
two or three hours a clear band appears extending somewhat
obliquely over the rounded end of the piece (Fig. i, C). Cilia
appear along the band. In a few more hours, the rate depend-
ing on the temperature, the ciliated band moves forward around
the anterior end of the piece, and in doing so bends around on
itself into the characteristic peristome. A new peristome-
field, or disk marked by delicate parallel lines, appears on the
inner side of the band even before it moves forward, and as
the band bends around to make the terminal peristome, the
new disk comes to lie in its central part. A depression, that
appears at the basal end of the band, forms the pharyngeal
funnel. The new peristome is smaller than that of the original
animal, and, as the figures show (Fig. i, C1, C'2, Cs), it is, in
some cases, even smaller than the reduced peristome on the
anterior piece. The foot-piece is also at first very long as com-
pared with the size of the new peristome ; and this condition
may remain for several days.

These results demonstrate that, for some time after the new
organs have developed, the new stentors retain some of the
peculiarities of the part of body from which they have come.
When the pieces are contracted, or are swimming, these rela-
tions are scarcely evident and might easily escape detection.

The transformation of an anterior piece into a new stentor
is much more strikingly seen when only a small part of the
anterior end is cut off. One set of observations on the same
individual is represented in Fig. 2. The stentor fully extended
is represented in Fig. 2, A. The anterior end had been cut
off, as shown in Fig. 2, A\ while the animal was contracted.

No. 6.] PROPORTIONATE STRUCTURES IN STENTOR. 315

After its removal the anterior end, B, contracted still more, so
that its posterior cut-surface was quickly covered over by the
bending in of the sides ; the disk bulged forward. After a
few hours the piece became somewhat pointed at its posterior
end, and then fixed itself by a foot that appeared at the end.

FIG. 2. — A, 1.4 x .4 mm. Stentor fully extended; A1, same contracted, cut in two at a-a;
B, C, immediately after operation; Bl, .56 x .26, anterior end after twenty-four hours;
B2, .7 x .25, same after another forty-eight hours ; B*, .64 x .28, same after another four
days, i.e., seven days after operation ; C, 1.2 x .27, posterior end after twenty-four hours;
C2, 1.2 x .20, same after another forty-eight hours.

After twenty-four hours (it had been kept in the cold over-
night, i.e., for ten hours) it appeared as shown in Fig. 2, B1.
The old stentor measured 1.4 mm. by .4 mm. This new
stentor measures .56 mm. by .26. The old peristome has,
therefore, decreased nearly to half its original width. Two

i6

[VOL. II.

days later (Fig. 2, £P) the stalk was somewhat longer, and after
another four days (Fig. 2, />3) the form had not materially
changed. The development of the posterior piece of this same
individual is shown in Fig. 2, C1 and C2. The piece is about
twice as long as the anterior piece, but its peristome is about
the same size.

A similar operation was carried out on another individual ;
the results are shown in Fig. 3, A. A very small part of the

B

C1

FIG. 3. — A, 1.7x4. Stentor fully extended ; A1, same contracted, cut in two at a-a ; B,C,
immediately after operation; Bl, .5 x .21, five hours after operation ; B2, .5 x .24, three
days after operation; C1, posterior piece, 1.2 x .27, five hours after operation ; C2, same,
1.6 x .32, three days after operation. In Figs. 2 and 3 one individual used in each. In
Fig. i several individuals were used.

anterior end was cut off (Fig. 3, A1}. It contained, however,
the entire peristome and disk (Fig. 3, B}. About thirty hours
after the operation the anterior piece appeared as shown in

No. 6.] PROPORTIONATE STRUCTURES /A" STl-:.\ /'< >A\ 317

Fig. 3, f>1. After three days more the new stentor had about
the same form. The peristome is, as compared to that of the
original stentor, too wide for the length of the new individual,
although it is not much more than half the width of the old
peristome.

The development of the posterior piece is shown in Fig. 3,
Cl and £72. In this piece, particularly in the earlier stage, Cl,
the peristome is smaller than was the original peristome, ami
also relatively smaller as compared with the entire length of
the animal.

In order to be certain that the anterior pieces did not pro-
duce new peristomes during the night, they were kept in a
cold place when not under observation ; for I had found that
under these circumstances the formation of a new peristome is
greatly delayed, even although it may have begun to develop
before the piece is subjected to the cold. In this way I could
retard the development of the peristome for twelve hours, so
that I felt certain that a new peristome had not developed on
these anterior pieces in my absence.

In other experiments pieces of different sizes were cut from
the foot-end in order to see if the size of the new peristome
that is formed is in proportion to the size of the piece. It was
found that a smaller peristome develops on a smaller piece, and
a larger one on a larger piece; and this same relation holds
also for pieces of different sizes for other parts of the body.
It has been shown that cross-pieces from the anterior or
posterior ends retain some of their original peculiarities even
after the formation of a new individual, and that for several
days the stentors from anterior pieces are too broad for their
length, and individuals from the posterior end are too long for
their breadth. Some of these newly regenerated stentors from
the anterior pieces were kept for a longer period and supplied
with food. Their measurements for from one to seventeen
days after the operation are given in the following table.
The measurements of three normal individuals of this lot
were 1.6 x .5 ; 1.4 x .4 ; 1.1 x .4. The experiment began
February 3.

MORGAX.

[VOL. II.

f .6 X .22

Feb. 4. -I .85 x .28
[ .6 x .23

ANTERIOR PIECES OF HALF SIZE OR LESS.
Feb. 10.

.85 x .32
.76 x .28

r

Feb. 6.

f .6
.6

•
x

•23
•3

Feb. 13.

.68 x .2

•7 < -34

.8 x .35 x .25

.9 X .28 X .2

.64 x .28

1. 12 X .39
[ 1.2 X .28

Feb. 20.

.96 x .36
.00 x .39
.04 x .32
.00 x .40
.00 x .40
.00 x .39
•9 x -36

It will be seen from this table that, after feeding, the
stentors from anterior pieces grow larger. The increase takes
place both in the peristome and in the length of the piece, so
that the proportionate size of the disk to the rest of the piece
remains about the same. The new stentors had begun to
divide on February 13, and by February 20 there were about
twice as many present as at first. They have, however, about
the same proportionate size as at first. The question arises
whether in the normal stentor the ratio of the breadth of the
disk to the length of the pieces may not be less than in very
large individuals. I measured some of the smaller individuals
found in the aquaria with the larger ones and obtained the
following results: 1.04 x .36 ; .72 x .26; .7 x .28 ; .72 x .3 ;
.9 x .38 x .2 ; i.i x .28 ; .9 x .36 x .2 ; .98 x .28. There is
seen to be some variation in the relative size of the length to
the breadth ; that is due in part to the individuals not always
expanding to the same extent, and also in part to some of the
measurements of the peristome not having been made in
the widest part, but there are also actual differences, as some
very careful measurements have shown. It will be seen that
while in large stentors the greatest breadth of the peristome
is about one-fourth, or nearly so, of the total length, in the
small individuals the breadth is more nearly one-third of the
length ; therefore the peristome is proportionally somewhat
larger for smaller pieces. Comparing these measurements
with those of the sizes of individuals derived from pieces of
the anterior end, we see that they have reached in several cases
the characteristic form for a small individual. Since there is

No. 6.] PROPORTIONATE STRUCTURES IN STENTOR. 319

a good deal of variation in the proportion between the width of
the peristome and the length of the animal both for small
normal individuals and for those that have come from anterior
pieces, it may be stated that pieces from the anterior end may
produce new stentors whose proportions come within the
range of variation of size of normal small stentors of about
the same length. The measurements of posterior pieces, that
are at first too long for the size of the peristome, show that
they, too, assume more typical proportions. Thus one of the
posterior ends of the last series measured, on February 6,
i .o x .25. On February 10, two other individuals in the same
list measured 1.4 x .38 ; 1.2 x .4. The peristomial region had,
therefore, reached the full size.

A somewhat crude comparison may bring the results home.
If a man were cut in two at the waist and the pieces behaved
in the same way as those of stentor, two new individuals
would develop. The anterior half would produce a small man
with a head too large for his height, i.e., his legs would be too
short for a man with that sized head. Although the old head had
grown smaller, it would be still too large for the rest of the new
man. In fact, his proportions would be more like those of a
baby whose head is relatively too large for his length as com-
pared with that of a man, and his legs too short. It is just
this result that we have found for anterior pieces of stentor.
If the new man were supplied with food, all parts of the body
would grow larger ; but as he got larger his legs would grow
faster than his head.

The posterior end of our imaginary man would have at first
legs too long for his total length, and his new head would be
relatively too small ; but if he were fed his head, shoulders, and
arms would grow faster than any other part and continue to
grow until the proportionate size had been reached. If he were
not fed, it is possible that his head and upper part might
increase more slowly in size at the expense of the material in
his legs, and the latter would get smaller until a balance
was reached. The result would be that a boy rather than a
baby was produced.

In other experiments pieces were removed that contained

320

MORGA\.

[VOL. II.

only a part of the peristome. In one series these pieces were
cut off, as shown by the line a-a in Fig. 4, A ; so that there
was a smaller and a larger piece, the former, /?, containing a
part of the old peristome, but not any part of the pharyngeal
funnel, and the latter, the larger piece, C, containing also a

C1

FIG. 4. — A , stentor partially contracted, cut in two at a-a into small piece, B, and larger, C ' :
A1, piece B after seven hours; B2, B2, Bz , piece like B after twenty-four hours;
B$, />'3, B$, same after twenty-nine hours ; B*, B*, B*, same in part after forty-eight hours
(after operation). C1, C1, piece C after seven hours.

part of the peristome as well as the funnel. The cut surface
of each piece is quickly closed by the bending in of the sides,
and the cut ends of the peristome are generally brought together
to make a closed ring. A foot develops on the basal end of the
anterior piece, B, B 1, and a stalk is soon produced in that
region. This piece may remain for twenty-four hours in this

No. 6.] PROPORTIONATE STRUCTURES IN STENTOR. 321

condition. Sooner or later a new ciliated band appears on
the old wall behind the part of the old peristome, as shown in
Fig. 4, £>2. The band moves forward, fusing with the old ring
at one point and, replacing the latter, produces a new peristome.
Whether the piece of the old ring is entirely obliterated, or
whether a part of it remains to contribute to the new peristome,
I did not determine. The regeneration of a new peristome on
these pieces may be delayed for several days (Fig. 4, B^ ), and,
in general, does not appear as soon as on pieces that do not
contain any part of the old peristome. Five or six series of
experiments of this sort, each series of a number of pieces,
were made, and the smaller pieces followed with great care.
Only those smaller pieces were isolated that contained no part
of the old funnel. Nearly all the pieces behaved in the way
just described, but in one or two a small funnel developed
where the cut edges came together. This may have been due
to a very small piece of the original funnel having been cut off,
or to the piece having come from very near to the old funnel,
or, as seems more probable, to the development of a new funnel
from the old ciliated band. If the last interpretation is correct,
it shows that in exceptional cases the peristome may complete
itself. In the large majority of cases this does not occur and
a new peristome and funnel develop at the side and move
forward. In nearly all cases the cut ends of the old peristome
come together, meeting in a slight notch. In one or two
instances one band lay slightly below the other at the meeting
point, producing a peristome exactly like the normal in shape,
only the funnel was absent. If a small piece of the old funnel
is left, it assumes the characteristic position of the funnel, and,
in fact, becomes such to all appearance, although this peristome
is generally replaced later by a new one.

Gruber studied the regeneration of pieces somewhat similar
to these without a funnel, and states that the remaining part
of the old peristome gives rise to a new one, but I have not
found this to be the case. If the piece is kept under close
observation, the development of a new peristome is found to
take place in the way just described. The change is sometimes
so rapid that a few hours may suffice to bring it about.

322 MORGAN. [VOL. II.

The history of the complementary piece (Fig. 4, C) is as
follows : After the cut surface has been closed over and the
edges of the peristome brought together, the piece may imme-
diately fix itself by the old foot. The piece elongates to its
full length, which is the same as that of the former animal
(Fig. 4, C1}. In some of these pieces I have observed the
development of a new peristome in the course of a few hours
after the operation (Fig. 4, C1}. It seems that this takes
place sooner when only a small part of the old peristome and
funnel is left than when a larger part remains. In cases in
which a large part of the old peristome remains a new peri-
stome may not develop for several days ; and in some cases I
have not found it to appear at all, but I cannot state positively
that it does not ultimately appear. Since even normal indi-
viduals may produce a new peristome, the appearance of a peri-
stome on these new stentors after several days may be only the
regular process of renewal of that organ. In two cases, in
which a new peristome appeared after two days, the old one
had begun to break down while the new band was developing.
In all other cases the old part was still active and normal in
appearance up to the time of its replacement by the new cili-
ated band. These results show that even in a large piece the
new peristome is not regenerated from the old one. The
presence of the old pharyngeal funnel in these pieces does not
make any important difference in the end result, although it
may be that pieces of this sort regenerate less quickly than
when the piece does not contain the funnel portion of the old
peristome.

Another experiment that supplements the preceding one in
several respects consists in cutting the stentor in two, as indi-
cated by the line a-a in Fig. 5, A. In this case the smaller
piece, B, contains the funnel part of the peristome, while the
larger piece, C, contains the remaining part of the peristome.

The smaller piece, B, closes in, develops a foot, becomes
attached and produces a stalk. The edges of the peristomial
ring unite more or less, as shown in Fig. 5, B1. In one case
a new peristomial band appeared six hours after the operation,
moved forward, and produced a new peristome (Fig. 5, B1}. In

No. 6.] PROPORTIONATE STRUCTURES IN STENTOK. 323

other cases, in which the piece was small in comparison to the
size of the remaining part of the peristome, a new peristome
did not appear in one case until after thirty hours (Fig. 5, B 4) ;
in other cases a new peristome had not appeared at this time
(Fig. 5, £3).

The complementary piece (C, Fig. 5) closed in, fixed itself,
and extended to its full length. In pieces in which the

FIG. 5. — A, stentor partially contracted, cut in two at a-a into a small piece, B, and a large
piece, C; B1, piece like B after seven hours ; B2, piece like B after twenty-seven hours;
£3, B*, piece like B after twenty-nine hours ; C1, C1, piece like C after seven hours ; C2,
piece like C after twenty-nine hours ; O, C*, piece like C after fifty-one hours.

remaining part of the peristome, that had united to make a
ring, was quite small a new ciliated band appeared in four
hours; in others, in six hours (Fig. 5, C1 C1) ; and in pieces with
a larger peristomial region, after twenty-four and even after
fifty-one hours (Fig. 5, Cz). It is interesting to note that in
these pieces the region from which in the normal individual
the peristomial band is formed has been more or. less com-
pletely removed, yet a new peristomial band may very quickly

324

MORGAN.

[VOL. II.

appear. I did not attempt to determine the position of the
new band in its relation to the region of closure of the piece.
In several cases in the last two experiments, and in some
other experiments like those shown in Fig. 6, A, Bl, Cl, small
pieces were sometimes cut off that contained a part of the old
peristome, but which did not fix themselves, or assume the
characteristic form. As these pieces were generally small,
although not below the minimal size, there can be little doubt
that most of them did not contain any part of the nucleus, and in
several cases I proved this to be the case by staining the pieces
in picro-carmine. The result shows that in the absence of the
nucleus a piece containing a part of the old peristome cannot
complete the peristome from the remaining part. This is the

B1

Dl

FIG. 6. — A, anterior end of stentor cut off and then divided into three pieces. Two of these,
Bl, C1, were apparently without nuclei, and did not produce a new peristome or assume
the typical form.

less to be expected since it has been shown that even in nucle-
ated pieces the new peristome is produced not by the old one,
but by the development of a new peristomial band. The result
is interesting in connection with a result obtained by Gruber,
viz., that if a non-nucleated piece containing a part of the newly
forming ciliated band is obtained it produces from the band
a new peristome. My results show that a piece of the old
band cannot act in this way. That the presence of the
nucleus is connected with the formation of a new peristomial
band seems highly probable, but I can easily imagine that
could a non-nucleated piece be supplied with certain unformed
elements it might be capable of producing a new peristome.
The results do not seem to me to show more than that the
nucleus supplies certain products of metabolism that must be
present before the protoplasm can successfully carry out its

No. 6.J PROPORTIONATE STRUCTURES IN STEXTOR. 325

innate tendency to complete the typical form. We are not
justified, I believe, in drawing the conclusion, as Gruber has
done, that preformed elements of the peristome exist in the
nucleus and must be set free in order to initiate the develop-
ment of a new peristome.

Lillie has found that the smallest piece of Stcntor polymor-
plins that becomes a perfect form is equal to a sphere of about
80 /A in diameter. The average size of the stentors was equal
to 230 /J-. This makes the volume of the smallest stentor
about Jy of the normal. For Stcntor cocrnlcns the smallest
stentors measured 90 /u, (==T1T mm.), the average normal
stentor 280 //, (=^ mm.). Therefore the former is about
JT of the latter.

Although I have not worked specially on this problem, yet
I have obtained some small stentors that were proportionately
smaller than those obtained by Lillie. Thus one individual
measured when extended .25 x — .08 mm., and when contracted
into an oval or nearly into a sphere .08 x .08 ( = = j\ mm.).
The larger normal stentors measured about .4 x .32 x .32 when
contracted. Although it is only possible to give a general
estimate of the relative size of these two individuals, the smaller
cannot be over J-4- of the former. It would be a mistake to
infer from this, as well as from Lillie's calculations, that the
latter came from a piece ^ or even ^V of tne original stentor.
The protoplasm of stentor is so vacuolated that a piece losing
the fluid in the protoplasm might become much smaller than
when first removed.

Lillie states that he believes that it would be possible to
obtain a smaller individual of 5. cocrnlcns than TTT mm. The
one that I obtained was in fact somewhat smaller, vis., -^.2 mm.
The difference in our results depends, therefore, rather on the
size of the normal average stentor with which the comparison
is made than on the smallest individual obtained. Lillie says
that he does not think there can be much difference in the
absolute size of the smallest stentors, whether one uses the
largest or the smallest normal specimens. It seems to me that
this may or may not be true, according to what factors may
enter into the result.

326 MORGAN. [VOL. II.

The conclusion that a piece ^ or even JT of the entire
animal can produce a new individual can give only a most gen-
eral idea of the relative size of the smallest piece, since more
depends on the size of the normal individual than on that of
the smallest pieces, and there is for stentor a very wide range
of size that may be called normal. A large normal individual
may contain eight times the volume of a small normal indi-
vidual. More significant, therefore, is the absolute size of the
smallest piece capable of regeneration, and in this respect my
results are practically in accord with those of Lillie.

Several experiments were made in which pieces were cut in
two longitudinally. In a longer or shorter time most of the
halves produced a new peristomial band that became a new
peristome. As this experiment did not give promise of much
that is new, it was tried in only a few cases.

A few casual observations made during the course of the
work may be briefly mentioned. The stentors were observed
dividing on several occasions, but Johnson's excellent figures
and account of these stages leave nothing new for me to add.
I have often noticed that after division the two products are
found attached side by side, and if they are not disturbed a
little colony may arise in the same spot. Several times I have
observed that two individuals that have been formed by division
of one of the regenerated posterior pieces were unequal in size,
although I do not know whether the smaller individual was
the distal or the proximal one.

As Gruber has pointed out, the first steps in the process of
division and of regeneration are the same, and this holds also
for the physiological replacement of the old peristome. In all
cases a peristomial ciliated band appears on tlic side and moves
forward around the anterior end to become the new peristome.
We have here another illustration that shows that during the
process of regeneration the factors that appear in the normal
growth may take part in the regeneration, and this relation
appears to hold for unicellular as well as for multicellular
forms.

In many cases, especially where a somewhat oblique cut has
been made, the superficial blue stripes come together over the

No. 6.] PROPORTIONATE STRUCTURES IN STENTOR. 727

\J I

cut surface in a most irregular way, yet this does not appear to
interfere with the subsequent regeneration ; and after a time
the stripes appear to be more regularly arranged. That a
certain amount of absorption takes place, and possibly also
development of new stripes, seems probable, but I have not
studied these changes in any detail. It would be interesting
to find out if in cross-pieces of the body the number of the
stripes remains the same and their size becomes smaller, or
whether the number of the stripes is proportionately reduced.

On several occasions I have tried to graft together pieces of
different stentors, but the exposed surfaces close so quickly
that I have not been able to get the pieces to unite. It does
not seem altogether improbable that the result could be
brought about by cutting two stentors at the same time, one
about the other. A lucky cut might bring two exposed inner
surfaces together, and they might stick to each other, but
so far I have not been able to carry out successfully this
experiment.

In a few cases the stentor was split partially in two pieces,
but generally the halves soon fuse together. It is of some
interest to find that, although the peristome was cut in two
and had reunited, a new peristome was not produced, showing
that the operation alone does not initiate the changes that lead
to the development of a new peristome.

The development of a new peristome in a piece that contains
a part of the old one appears to be due to the lack of propor-
tion between the old part (even when it contains all the essen-
tial parts of the peristome) and the rest of the piece. This
result is unique, since in all other forms in which a part of an
old organ remains the new organ regenerates from that part.
In stentor this does not occur, but a new organ is produced.
It is important to observe, however, that this is the character-
istic way in which stentor produces a peristome, so that the
organisms make use of a process that already exists. The
reduction in size of the old peristome in pieces from the
anterior end is the result that has most interested me. It
seems to be due to the withdrawal of material from the ante-
rior region to form the body and stalk of the new stentor.

328 MORGAN.

The change in shape of the piece, i.e., the production of the
typical form, is primarily the result of a shifting of the material
that carries with it a loss of material in the old part. Other
so-called formative factors may have some share in the reduc-
tion in size of the old peristome to one proportionate to the
rest of the piece, but the simple loss of material will, I think,
account for the greater part of the change.

What primarily brings about this change in the material so
that the typical form is produced is a question to which at
present there is no answer.

BRYN MAWR, March 12, 1901.

ABSTRACTS OF PAPERS

PRESENTED AT THE MEETINGS UK THE

AMERICAN MORPHOLOGICAL SOCIETY

AT BALTIMORE, DECEMBER 27 AND 28, igoo.

I. FISSION AND REGULATION IN STENO-
STOMUM LEUCOPS.

C. M. CHILD.

THE single individual of Stenostomum differs considerably
in size, according to conditions. Well-nourished specimens
may reach a length of nearly one and one-half millimeters,
while specimens measuring only one-half millimeter are often
found when food is scarce. The length is about eight or
ten times the transverse diameter.

The animals usually occur in chains, the number of zooids
varying from two to nine. Ordinarily chains do not consist
of more than five zooids, the uneven number being due to the
fact that the anterior zooid precedes the others in division.
The short chains of two or three zooids occur when food is
scarce. In well-nourished specimens the fissions succeed each
other more rapidly, and longer chains are the result. Each
particular septum occurs, with little variation, in a definite,
characteristic position, this apparently being determined largely
by the relative degree of development of the two ends of the
zooid which it divides.

Entodermal tissue is necessary for regeneration. Portions
containing all the other tissues of the body except entoderm

329

330 ABSTRACTS OF PAPERS. [VOL. II.

do not regenerate, but if a small portion of the entodermal
digestive sac be present regeneration is complete, provided
the piece is above a certain size.

If a chain be artificially separated into its zooids before they
have attained their full development, each zooid undergoes a
form-regulation, assuming within a few hours what may be
called the normal proportions, i.e., the length becomes eight
or ten times the transverse diameter. This regulation does
not occur while the individual is a zooid in a chain, because
the whole chain is not simply a series of individuals, but in
some degree a single individual, and therefore possesses cer-
tain proportions differing from those which each zooid would
possess if single.

When the chain is cut at various points between the zones
of fission, the results differ according to the degree of develop-
ment of the particular zone of fission concerned and the parts
adjoining it. If a piece containing a very young septum be
cut out from a chain, the septum disappears and a single
perfect individual is formed from the parts, which originally
belonged to two different zooids. A head is regenerated at
the anterior cut surface, a tail at the posterior end. If the
included septum be more fully developed, it remains, and
the part anterior to it is completely absorbed by the part
posterior to it, the head of the new individual resulting being
the head which was forming just posterior to the septum.
The relative size of the parts anterior and posterior to the
septum does not affect the result, unless the posterior piece
be very small. It is always the anterior part which is
absorbed, never the posterior. If the septum be still more
advanced in development, the portion anterior to it is only
partly absorbed. It regenerates a new head and becomes a
perfect zooid, though at first it decreases in size, owing to the
partial absorption.

In general each zooid tends to absorb material from the
zooid anterior to it. Each zooid, however, offers a certain
resistance to this absorption, the resistance increasing as it
approaches the condition of independent individuality. When
the individuality of a zooid is destroyed or reduced to a lower

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 331

degree, e.g., by cutting it in half, the posterior half may be
completely absorbed by the zooids posterior to it.

The sexual individual arises as a zooid in a chain, but when
the sexual organs appear, asexual reproduction ceases. The
single sexual individual may, however, attain a length equal to
that of the longest chains. The power of regeneration is much
less in the sexual than in the asexual condition. Apparently in
the former the energy is chiefly directed toward the elabora-
tion of the sexual products.

II. THE OCCURRENCE OF GUNDA SEGMENTATA

IN AMERICA.

WTNTERTON C. CURTIS.

A SPECIES of Gnnda, which in its external features seemed
identical with the G. segmentata of Lang, was found in large
numbers at Sandwich on Cape Cod. The internal arrangement
is not, however, as regular as Lang describes for G. segmcntata.
From a comparison with Verrill's figure of Proccrodcs nlrac
collected in the same region (Trans. Conn. Acad., Vol. VIII,
January, 1893), it is probable that the two forms are identical
and that Verrill has figured the head incorrectly.

III. SOME DISPUTED POINTS IN THE ANATOMY

OF THE LIMPETS.

M. A. WILLCOX.

THE following points were made:

i . The space previously described by Willcox as the nephrid-
ium is lined throughout by more or less columnar cells pro-
vided with long, delicate cilia and loaded in the fresh condition
with dark green granules. The histology lends no counte-
nance to Haller's contention that the posterior part of this
sac represents coelom, the anterior nephridium.

332 ABSTRACTS OF PAPERS. [VOL. II.

2. The space which in most species of Acmaea underlies
the viscera on the left side in A. tcstudinalis stretches almost
across the body, and lacks entirely the ciliated cells character-
istic of the nephridium. This negatives the opinion that the
space in question is a paired structure whose fellow of the right
side is represented by the posterior part of the nephridium.

3. A sub'raclular organ, whose presence in the Docoglossa
has been denied, exists in both A. tcstudijialis and A. fragilis.
It is situated on the underside of the odontophore, just behind
the tip of the radula, and is a triangular, somewhat cushion-
shaped organ, divided by a V-shaped groove into an anterior
and a posterior part. The posterior part is marked by trans-
verse grooves and is covered by tall, columnar epithelial cells,
some of which seem to be ciliated, while others are somewhat
fusiform and have much the appearance of sense cells. The
innervation has not been traced, but no ganglia are to be found
in the organ. The subepidermal portion consists of connective
tissue with scattered and inter-crossing muscular fibers.

IV. THE HABITS AND LIFE HISTORY OF ARGU-
LUS WITH REFERENCE TO ITS ECONOMIC

IMPORTANCE.

CHARLES B. WILSON.

IN the town of Warren, Mass., is a small pond which was
stocked with carp and bass several years ago. The fish
seemed to thrive well until the fall of 1899, when they began
to die off in considerable numbers, with no apparent signs of
disease or injury. No clue to the cause of the devastation
could be obtained till the spring of 1900, when several suckers
were speared in the outlet of the pond whose gill chambers
were full of the parasitic copepod Argnlus, probably A. cato-
stovii. The gentleman who owned the pond stated that these
copepods were common on most of the fish caught there, and
his statement was afterward verified. On being put in an
aquarium with dace, roach, and bream, they attacked these fish

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY.

J

viciously and the next morning they were found dead. Argn-
lus deposits its eggs instead of carrying them around like other
crustaceans, arranging them in rows on sticks, stones, etc.,
with their long diameters parallel.

When laid they are covered with a jelly envelope consist-
ing of beads of jelly arranged in rows parallel with the long
diameter of the egg. These harden into a brittle shell. The
eggs are fertilized outside the body of the female and there
is no copulation. The egg hatches into a typical nauplius,
which after one or two moults changes into a metanauplius
having highly developed clasping organs in the shape of barbed
claws terminating the second maxillae. On putting two small
dace into the aquarium with about two hundred of these larvae,
the latter made no attempt to use their claspers, but the fish,
on recovering from their fright, ate up every one of the larvae.
On inquiry it was found that the pond in question had been
seined for three years, and that the dace and roach had been
sold for pickerel bait. First conclusion, the subject of para-
sitism is not so one-sided as would appear at first sight.
Second conclusion, the protection of small fish like dace and
roach in our fish ponds may be one of the best preventives
against such parasites as these.

V. THE ANATOMY AND DEVELOPMENT OF THE
POSTERIOR VENA CAVA IN DIDELPHYS
VIRGINIANA (KERR, LINN.).

C. F. W. McCLURE.

AN examination, thus far, of forty-eight opossums has brought
to light many interesting variations concerning the mode of
formation of their posterior vena cava.

These variations are so pronounced and so closely accord
with certain embryonic conditions described by Hochstetter
for Echidna aculeata, it seems to the writer as not improbable
that the development of the posterior vena cava may take place
in Didelphys and Echidna in substantially the same manner.

334 ABSTRACTS OF PAPERS. [VOL. II.

In all marsupials hitherto examined (Petaurus taguanoides
excepted) the posterior vena cava has been found to lie ventral
to the aorta between the renal and common iliac veins, and to
be formed through a union of the common iliac veins, which
takes place ventral to the arteries.

In DidelpJiys virginiana the posterior vena cava is not formed
in this manner.

In fact, the mode of formation of the posterior vena cava
was found to be so variable in DidelpJiys virginiana that it is
quite impossible to assign any one mode of origin for this
vessel which may be regarded as typical of the species.

For descriptive purposes the various modes of origin of the
posterior vena cava in DidelpJiys have been classified by the
writer under three main types as follows :

Type I includes those cases in which the internal iliac veins
unite with the external iliac veins to form the posterior vena
cava, ventral to the common iliac arteries or ventral to the
aorta.

Type II includes those cases in which the internal iliac
veins unite with the external iliac veins to form the posterior
vena cava, dorsal to the common iliac arteries or dorsal to the
aorta.

Type III includes those cases in which the internal iliac
veins unite with the external iliac veins to form the posterior
vena cava, both dorsal and ventral to the common iliac arteries
or both dorsal and ventral to the aorta.

So many variations of this type were met with that a
further subdivision of Type III was found necessary, as
follows :

Type III, A, includes those cases in which the principal
union between the internal and external iliac veins lies ventral
to the arteries in question.

Type III, B, includes those cases in which the principal
union between the internal and external iliac veins lies dorsal
to the arteries in question.

Type III, C, includes those cases in which the above-
mentioned dorsal and ventral unions are sub-equally de-
veloped.

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY.

335

The following table shows how the above-mentioned types
were distributed among the forty-eight individual opossums
examined (twenty-four males and twenty-four females).

TYPE.

$

<?

TOTAL.

Tvne I

5

9

' 1

'4

A /t'c

Tvoe II

6

7

'3

'3

j. j jj<~

Type III

A . . .

3

*>

5

21

B

8

5

'3

C

2

i

•-i

TOT A i

24

24

48

A comparison with the development stages in Echidna
aculeata shows, I think beyond the question of a doubt, that
the variations in the method of formation of the posterior vena
cava in Didclphys, so far as its posterior tributaries are con-
cerned, are modifications of a ground plan arrangement similar
to that described by Hochstetter for his Echidna embryo
No. 45.

The writer's investigations upon the development of the
posterior vena cava are as yet incomplete. So far as they
have gone, however (an examination of five 1 5-millimeter
embryos), they decidedly favor the above conclusions.

VI. THE CROSSING OF THE OPTIC NERVES

IN TELEOSTS.

G. H. PARKER.

IN ten species of symmetrical teleosts, in each of which one
hundred specimens were examined, the right optic nerve was
dorsal at the crossing about as frequently as the left. The two
types of crossing (right nerve dorsal and left nerve dorsal)
were not correlated with sex and were about equally frequent

336 ABSTRACTS OF PAPERS. [VOL. II.

in specimens taken from one school of fish. In one hundred
specimens of the winter flounder (Pseudopleuronectes ameri-
canns], whose eyes are on the right side, all had the left nerves
dorsal. In seventeen specimens of the summer flounder
(ParalicJithys dentatus), whose eyes are on the left side, all had
the right nerves dorsal. In one hundred specimens of the
stellate flounder (PlatiditJiys stcllatus} all had the left nerves
dorsal, notwithstanding the fact that fifty of these fish had their
eyes on the right side and fifty on the left. Although each
species of symmetrical teleosts examined showed about equal
numbers of the two possible types of optic nerve crossing,
the flounders showed only one type for each species.

VII. A NEW TYPE OF BUDDING IN ANNELIDS.

H. P. JOHNSON.

Two gigantic undescribed species of Pacific coast Syllidac
produce reproductive zooids by collateral budding from a defi-
nite proliferating region near the posterior extremity.

The single specimen of the larger species (Trypanosyllis
ingens, sp. nov.) at my command was too poorly preserved for
thorough study, but what I have learned about its budding
agrees essentially with the fuller knowledge acquired from the
other species. The buds numbered about thirty, all turgid
with nearly ripe ova. No very young buds were detected, as
in the succeeding species.

Of the other species (Trypanosyllis gemmipara, sp. nov.) I
have several specimens, but only one with buds. They develop
from a proliferating region twenty somites anterior to the
pygidium. The advanced buds are broadly elliptical, and much
flattened dorso-ventrally. Each is attached at its head-end by
a short pedicle. The somites number 20-28, with parapodia
which are miniatures of those of the parent. The prostomium
has large eyes, a pair of antennae, and brain. There are a pair
of ventral nerve cords, a muscular system, septa, and large
paired masses of sperm cells in every coelomic chamber from

No. 6.] AMERICA* .IfOA'PtfOLOGSCAL SOCIETY. 337

prostomium to pygiclium. Purely vegetative organs (e.g., mouth,
alimentary canal, anus, and nephridia) are absent, although a
rudiment of the alimentary canal may exist as a median strand
of tissue extending the length of the bud.

The youngest buds form a cluster of about twenty-five
attached to the right side of the zone of proliferation, on its
ventral aspect. The earliest-formed organs are the anal cirri,
at first two distal protuberances which elongate and become
moniliform before the bud segments. Apparently the bud
contains only ectoderm and mesoderm, which are continuous
with the same germ layers of the proliferating region.

In neither species are there any reproductive cells in the
body of the parent anterior to the proliferating region, but
sperm cells are present in T. gemmipara in the twenty parental
somites back of this point.

VIII. AMPHIBIAN STUDIES.

J. S. KINGSLEV.

THE following are the chief points made in the paper :

1 . The Salamandrina form the central Urodele stem, and the
Perennibranchs and Derotremes have been derived from this
stem by degeneration and the retention of larval characters.

2 . The Urodeles cannot have been the ancestors of the Anura ;
the anuran tadpole resembles the Urodele only in superficial char-
acters ; the Anura have descended directly from the StegoccpJiala.

3. Amphiuma has no tentacular apparatus at any stage;
what was described as such by Davison was a trematode
parasitic in the suborbital blood vessel.

4. The Caecilians differ from all Urodeles in the fact that
the palatine nerve receives a branch from the ophthalmicus
profundus instead of from the maxillaris superior nerve.

5. The Caecilians have not descended from the Urodeles, nor
is Amphiuma a neotaenic Caecilian. The Caecilians are degen-
erate in loss of limbs and tail ; in all other respects they are the
most primitive of living AmpJiibia.

338 ABSTRACTS OF PAPERS. [VOL. II.

6. No Stegocephalan as yet known can have served as ances-
tor of Urodeles, Annra, or Caecilians. The parent form must
have possessed characters intermediate between the known
Stegoccphali and the Crossopterygian ganoids.

7. The ancestry of the Amphibia must be sought in the
Crossopterygii and not in the Dipnoi.

8. The balancer of the Urodele larva is a modified external
gill belonging to the hyoid arch.

IX. PHAGOCYTOSIS IN A MAMMALIAN OVARY.

MAYNARD M. METCALF.

IT has long been known that in the ovaries of certain Mam-
mals and Fishes syncytia of young ova are found, and that of
the several nuclei in such a syncytium one or but few persist
as nuclei of definitive ova. Apparently, in these cases, the cells
which disappear are used as food for the persistent ova.

In the ovary of the common Cat somewhat similar conditions
have been observed by the author. Many of the young ova, in
the stage when they are surrounded by a follicle consisting of
but two layers of cells, are seen to have ingested many of the
follicle cells, and the nuclei of these ingested cells can clearly
be seen, some quite perfect (newly ingested), others apparently
in different stages of digestion. The nuclei of these ingested
cells, when almost completely digested, appear as groups of
granules, these granules being apparently the remnants of the
nodal thickenings of the chromatin network of the ingested
cells. Such an ovum with its ingested nuclei very closely
resembles the young blastomeres of a Salpa embryo, which
have the same habit of devouring follicle cells.

Many young ova with ingested follicle cells were found in
one Rat ovary. In another ovary of a White Rat no ingestion
of follicle cells was found, nor was ingestion found in the ovary
of a Gray Squirrel examined. Pressure of other duties has
prevented the author from determining if such ingestion of
follicle cells be normal in the ova of Rats and, if so, what

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 339

relation, if any, it may have to the age of the individual or
to its condition as regards reproduction. The observations are
reported in the hope that other observers may have them in
mind. If the phenomena are at all general among Mammalia,
they should be seen in many laboratories in the usual histo-
logical demonstrations.

Similar phenomena are, of course, well known in several
groups of Invertebrata.

X. THE MAMMALIAN LOWER JAW.

W. H. RUDDICK AND J. S. KINGSLEV.

IN no adult mammal, recent or fossil, is the lower jaw known
to consist of more than a single bone, and no author, save W. K.
Parker, whose observations appear to have been overlooked, has
shown the existence of more elements in its development. We
are able to confirm Parker in his account and to identify in the
mammals the following bones of the non-mammalian groups :
(i) articulare, (2) angulare, (3) splenial, (4) dentary. Of these,
articulare and angulare unite to form the malleus, while the
definitive lower jaw is composed of dentary and splenial. Two
cartilages participate in the formation of the lower jaw, --the
Meckelian and a second larger cartilage lying external to this,
which, like Parker, we homologize with one of the lower labials
of the Ichthyopsida. In its ossification this cartilage is strik-
ingly similar to amphibian cartilages, and the resulting bone-
a part of the dentary - - gives rise to the posterior part of the
lower jaw, including the coronoid and articular processes. In
the existence of this lower labial is to be found the explanation
of the shifting of the articulation of the lower jaw. It is note-
worthy that a lower labial occurs in about the same position in
the ganoid Polypterus.

340 ABSTRACTS OF PAPERS. [VOL. II.

XI. AN APPARATUS IN THE CENTRAL NERVOUS
SYSTEM OF VERTEBRATES FOR THE TRANS-
MISSION OF MOTOR REFLEXES ARISING
FROM OPTICAL STIMULI.

PORTER EDWARD SARGENT.

IN Antia, at about the time of hatching, there arises in the
anterior portion of the roof of the optic ventricle a group of
cells, eighty to one hundred in number, formed by the differen-
tiation of indifferent neuroblasts. During the first and second
days of larval life the axons develop from these cells as exceed-
ingly fine processes, growing directly toward and into the optic
ventricle. Early in the third clay the adjacent axons come
together in groups and coalesce at their tips, in their further
growth through the cerebro-spinal fluid appearing as a single
fibril. Later these fibrils coalesce with others similarly formed,
and in their growth posteriorly through the ventricles and
canalis centralis form what has been known as Reissner's fibre,
which is then a fibre tract made up of many axons closely
united and surrounded by a single medullary sheath. Through
the posterior portion of its course there come off from it fine
fibrils which pass through the canal obliquely backwards and
enter the tissue of the cord.

In the first clay after hatching there may be found in the
extreme posterior end of the canalis centralis a number of
small cells, three to four micra in diameter, lying in the lumen
of the canal and ventriculus terminalis. Some eight to ten
of these cells persist and continue to develop. Increasing
rapidly in size, they become spindle-shaped and send their
axons cephalad through the canal. The axons are at first
separate, but later coalesce as they grow forward, and,
eventually meeting the system of axons from the cells of
the tectum growing posteriorly, the two interweave in a way
not yet clearly made out.

The development of this apparatus in Amia is typical of
its development in all vertebrates, though in some groups

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 341

•

there are considerable variations. In the Skate the cells are of
conspicuous size, and three to four hundred in number. They
are multipolar, sending processes to the ectal portion of the
tectum, where they come in direct contact with the endings of
the optic nerve fibres and give rise to two fibre tracts, one of
which passes posteriorly to the cerebellum, the other anteriorly
and out into the optic ventricle to form the Reissner's fibre.
In reptiles and birds the apparatus appears at a late stage and
is not fully established until just before hatching.

In Pctromyzon this apparatus is not fully established until
the second month of larval life. The cells, about twelve in
number, form a well-marked nucleus. Reissner's fibre passes
through the optic ventricle, reenters the brain tissue, and again
emerges into the fourth ventricle. Thus Petrotnyzon furnishes
the connecting link between the condition in the Gnathostomes
and AmpJiioxus. In the latter the largest and most anterior of
the colossal dorsal cells lies across the central canal, is in direct
connection with the pigment spot, and sends its axon caudad
through the cord in the median plane just ventral to the canal.
This axon probably represents Reissner's fibre.

There are many lines of evidence which lead me to assign
the function I have to this apparatus.

1. The cells are in direct connection with the endings of

o

the optic nerve, and with the cerebellum. The axons pass by
the shortest path posteriorly through the central canal, and
probably out through the ventral roots to the musculature.

2. When the fibre is cut in Sharks or Dogfish they evidence
an inability to respond quickly to optical stimuli.

3. In the vertebrates of the cave fauna the apparatus degen-
erates as the eye degenerates.

4. In no animal does the apparatus reach complete develop-
ment until just before the animal attains free life.

5. In those animals which are sluggish at hatching (Petro-
inyzon, Amia), the apparatus is not fully developed until a con-
siderably later period.

6. In those mammals which are born blind (Mouse, Kitten),
the apparatus is not fully established until about the time the
eyes become functional.

342 ABSTRACTS OF PAPERS. [VOL. II

7. In any one group, as the Teleosts, the apparatus has its
highest development in those species which are most active.

8. The corpora quadrigemina of higher vertebrates are con-
cerned only with reflex functions ; therefore this apparatus
must have a reflex function.

Such a short circuit avoiding the loss of time in passing
through a chain of neurones must be of great importance in
saving time, amounting perhaps to a considerable fraction of
a second. An animal suddenly presented with some optical
evidence of danger from which it recoils in fear, does so reflexly,
calling into use this apparatus. When we consider that in the
struggle for existence the saving of a fraction of a second is
often a matter of life or death, it becomes evident that this
apparatus has played an important part in the survival of the
fittest, and in the whole evolutionary process throughout the
vertebrate series.

XII. THE SIGNIFICANCE OF THE SYNAPSIS
STAGE OF THE GERM CELLS.

THOS. H. MONTGOMERY, JR.

IN the germinal cycle of the Mctazoa may be distinguished
in succession the following main stages : the conjugation of the
maternal and paternal cells (fertilization), a number of genera-
tions of ovogonia (or spermatogonia), then the growth period,
and finally the stage of the two maturation divisions. The
reduction in the number of the chromosomes, i.e., the formation
of bivalent chromosomes, is not effected by either of the mat-
uration mitoses, but during that portion of the growth period
known as the synapsis stage. The bivalent chromosomes are
formed by a union, end to end, of every two univalent chro-
mosomes, as I have shown in a paper on the spermatogenesis
of Pcripatus, just published, and in another on the spermato-
genesis of the Hcmiptcra, now in press.

Heretofore no one has shown exactly how the bivalent chro-
mosomes are produced, and no one has given any adequate

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 343

explanation for the reason of their formation. My comparative-
studies on the spermatogenesis of a considerable number of
species of Hemiptcra, which have brought to light certain facts
of importance for determining these questions, render it probable
that in the process of formation of the bivalent chromosomes we
have a conjugation of paternal with maternal chromosomes. This
would then be the final stage in the fertilization of the germ cells ;
it would be a conjugation of the chromosomes of different parent-
age producing a rejuvenation of them as metabolic centers of
the cell ; and this rejuvenation finds its expression in the great
changes of the growth period. Then, probably, the reduction
division takes place, in order to separate again the conjugating
chromosomes, as two conjugating Infusoria unite and then sep-
arate after the accomplishment of the rejuvenescence.

In the space of a short abstract it is not possible to give the
evidence for these conclusions.

XIII. A STUDY OF THE PHENOMENA OF

FERTILIZATION AND CLEAVAGE

IN ETHERIZED EGGS.

EDMUND K. WILSON.

A.

IF fertilized eggs of Toxopnenstcs, after the formation of
the cleavage-figure, be placed in a 2.^/0 solution of ether in
SGSL water, the astral rays quickly fade out, as was long since
observed by O. and R. Hertwig in sea-urchin eggs treated by
solutions of chloral hydrate or sulphate of quinine. The clear
hyaloplasm masses forming the astral centers are thus left as
well-defined, slightly irregular, non-radiate areas, connected by
the spindle-area. If the eggs are replaced in sea water, the
rays are rapidly redeveloped and cleavage may proceed nearly
or quite normally. Even if left in the ether solution, how-
ever, the nuclear division may be completed, the daughter-
nuclei being re-formed and growing to their normal size, but
no cytoplasmic division oc curs, - - a result similar to the earlier

344 ABSTRACTS OF PAPERS. [VOL. II.

ones of Demoor on the division of plant cells in vacua, or in
an atmosphere of CO2, and to those of Loeb and Norman on
Arbacia eggs in sea water concentrated by the addition of a
small percentage of chloride of sodium or magnesium. If the
strength of the ether solution be now somewhat reduced, by
evaporation or by the addition of sea water, the asters reap-
pear, though not attaining full development, and progressive
nuclear division may occur without cytoplasmic cleavage. In
this case the egg may give rise to a syncytium, containing
from four to sixty-four or more nuclei, which migrate towards
the periphery so as to take up nearly the same position as
they would have had in a segmented blastula. This phenomenon
strikingly recalls that which normally occurs in the cleavage of
many arthropod and some coelenterate eggs. At each nuclear
division an attempt is made at a corresponding cytoplasmic
division, but this is usually unsuccessful ; or, in case division
occurs, the cells subsequently fuse together to repeat the
attempt at the next nuclear division. This, again, is closely
similar to the ineffective early attempts at cleavage in such
eggs as those of Renilla. If the eggs be replaced in sea water
when the process is not too far advanced (4-32 nuclei), cleavage
may occur of a multiple type almost exactly like that occurring
in Renilla, and a normal blastula may arise ; but the cleavage
is often irregular or incomplete.

These observations support the conclusion indicated in the
preceding paper, that the astral rays are not fixed and per-
manent structures, but an expression of a form of cytoplasmic
activity, partly in the nature of protoplasmic currents, that
may be inhibited by temporary paralysis of the cytoplasm.
They indicate also that the astral rays are connected with
cytoplasmic rather than with nuclear division, and support the
interpretation, offered by the author many years ago, of the
variations of cleavage observed in Renilla.

B.

If Toxopneustcs eggs be placed in 2.5/0 ether solution one
minute after fertilization, formation of the sperm-aster is com-
pletely suppressed. The sperm nucleus, however, slowly moves

No. 6.] AMERICA* MORPHOLOGICAL SOCIETY. 345

inwards and gradually enlarges, becoming finally (1-2 hours)
as large as the egg nucleus and indistinguishable from it. In
some cases, though this is not very common, the two nuclei
approach and finally completely fuse to form a typical cleavage
nucleus. If in the earlier stages of the process (before union
of the germ nuclei and while the sperm nucleus is still not
more than two-thirds the diameter of the egg nucleus) the
eggs be replaced in pure sea water, the sperm-aster is rapidly
developed, centering in a point at one side of the sperm
nucleus, and development may proceed normally ; but this
result was never obtained after the germ nuclei had united,
probably because the action of the ether had been too pro-
longed. In a few cases, after replacing the eggs in sea water,
the sperm-aster was observed to divide and form an amphiaster
before union of the germ nuclei. In this case the sperm
nucleus at the time of union had assumed the vesicular form,
though still somewhat smaller than the egg nucleus. One
such case was followed out and found to give rise to a normal
larva. In such cases the effect of the ether has been to trans-
form the type of fertilization from that characteristic of the
sea urchins into that observed in starfishes, or in many worms
and mollusks, where an amphiaster is formed before union of
the germ nuclei and the latter are approximately equal at the
time of union.

The foregoing facts show, in general accordance with the
early work of O. and R. Hertwig, that growth of the sperm
nucleus and approach and fusion of the germ nuclei may take
place quite independently of the sperm-aster ; further, that
approach of the nuclei is probably not a simple chemotactic
phenomenon, since it is very greatly delayed by etherization of
the egg.

C.

In some of the etherized eggs, after replacement in sea
water, the nucleus failed to divide at the first cleavage, the
whole of the chromatin passing to one pole and re-forming
as a single nucleus. Such eggs divide into a nucleated and
a non-nucleated half, the latter containing only an aster, as in
the case of some of the non-nucleated egg fragments fertilized

346 ABSTRACTS OF PAPERS. [VOL. II.

by a single spermatozoon observed by Boveri. In such cases
the asters in both halves multiply progressively at the same
rate, but complete division occurs in only the nucleated half.
In the non-nucleated half, however, each aster becomes sur-
rounded by a deep constriction which afterwards fades out.
This result stands intermediate between those of Boveri and
Ziegler. As in the case of the magnesium eggs, an aster un-
accompanied by nuclear material forms a division center of the
surrounding cytoplasmic area, but is here apparently unable to
effect complete division in the absence of chromosomes.

XIV. EXPERIMENTS UPON THE INFLUENCE OF

THE SEXUAL CELLS UPON THE

SOMATIC CELLS.

GEORGE WILTON FIELD.

1. Of iJic sexual cells of one individual upon the somatic
cells of another individual.

This is generally held to be a phase of Telegony. Several
instances have been cited by Gadow, Nathusius, and Bulman
to show that the spermatozoa affect the somatic cells which
secrete the eggshell. Thus, the shell of eggs of fowls which
normally lay brown eggs are said to become lighter in color
when the birds are mated with a male of a breed which lays
white eggs ; and conversely, white eggs become brownish, if
the birds are mated with a male of a breed which lays
brown eggs.

Our experiments were carried on at the Rhode Island Agri-
cultural Experiment Station upon Leghorn pullets (white eggs)
mated to a Light Brahma cock, and upon Light Brahma pullets
(brown eggs) mated to a Leghorn cock. By means of trap
nests the series of eggs laid by each individual was- followed ;
no departure from the ordinary normal color of the egg-
shell could be observed.

2. Of the sexual cells upon the somatic cells of the same
individual.

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 347

In addition to observations upon the changes in the sec-
ondary sexual characteristics induced by castration, the effect
of absorption of testicular material through the peritoneum
was the end in view.

Upon castration the testes of the cockerel were returned
free into the abdominal cavity. After six weeks a living ripe
testis, one inch long, was introduced into the abdominal
cavity through an incision anterior to the last pair of ribs.
The experiment was conducted on a scale too small to abso-
lutely determine how far the results were due to direct regen-
eration of the testes, and how far to absorption of the
introduced testicular material. Two points, however, were
demonstrated: (i) that the testes do regenerate; (2) that the
introduced testicular material was absorbed. In one case it
seemed clear that the secondary sexual characteristics devel-
oped solely as the effect of the absorption of the introduced
testicular material, without regeneration of the testes. In
another individual similar conditions were strongly indicated,
but were obscured by a pathological growth.

XV. THE MORPHOLOGICAL PHENOMENA IN-
VOLVED IN THE CHEMICAL PRODUCTION
OF PARTHENOGENESIS IN SEA URCHINS.

EDMUND B. WILSON.

IN accordance with Loeb's important discoveries on Arbacia,
unfertilized eggs of Toxopnettstes, when treated by the mag-
nesium chloride method, may segment and give rise to free-
swimming blastulas, gastrulas, and Plutei. There is always a
considerable proportion of abortive and monstrous forms, and
none of the stages are exactly like those arising from fertilized
eggs, though often closely similar to them. The Plutei pos-
sess, however, the characteristic arms, pigment, skeleton, and
divisions of the gut. That these eggs have not been acci-
dentally fertilized is proved by the fact, demonstrated to the
society by an exhibition of sections, that during cleavage they

348 ABSTRACTS OF PAPERS. [VOL. II.

show but half the usual number of chromosomes, namely,
eighteen instead of thirty-six. The same conclusion is reached
by the study of the other internal phenomena, which differ in
a characteristic way from those occurring in normal fertiliza-
tion, though showing an interesting parallel to them.

The eggs, even of the same individual, show a very high
degree of variability in their response to the solution. Great
numbers of incomplete or abnormal forms of mitosis occur.
The most interesting of these are cases in which the nucleus
becomes the center of formation of a single aster (monaster),
which never resolves itself into an amphiaster but nevertheless
passes through periodic changes parallel to those occurring in
complete mitosis. Thus, such an aster may appear, nearly dis-
appear, and reappear as many as six times in succession, the
nucleus simultaneously disappearing and re-forming. In such
cases the chromosomes divide, probably at each disappearance
of the nucleus, and may thus become very numerous, without
division of the nucleus or of the cell body. In other forms of
incomplete mitosis the single aster may give rise to an amphi-
aster, but the nucleus fails to divide.

In all the eggs capable of development, the initial change is
an irregularity in the cytoplasmic meshwork, followed by the
appearance of a primary radiation centering in the nucleus,
the gradual formation of a perinuclear clear zone of hyalo-
plasm, and the growth of the nucleus. In many of the eggs a
number of separate asters (cytasters, equivalent to the " arti-
ficial astrospheres " of Morgan), having no direct relation to
the nucleus, are formed in the cytoplasm in addition to the
primary radiation. At the centers of these asters hyaloplasm
likewise accumulates. Growth of the nucleus is followed by
disappearance of the nuclear membrane, the rays of both the
primary radiation and of the cytasters meanwhile becoming
much reduced and in some cases nearly disappearing. After
a short pause this is followed by a redevelopment of the rays,
and in typical cases the nuclear area (the center of the former
primary radiation) has now formed two centers of radiation,
producing a typical amphiaster. When no cytasters are present
(a relatively rare case) cleavage may proceed nearly as in

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 349

normally fertilized eggs, but in many cases complete cyto-
plasmic division does not occur until after two or more nuclear
divisions. Thus arise some of the forms of multiple cleavage
observed by Morgan and Loeb. If cytasters be present, one
or more of them may participate in the nuclear division, thus
forming triasters, tetrasters, etc. ; but such eggs are probably
incapable of producing an embryo. When the cytasters do
not establish a connection with the chromosomes, they never-
theless form, in many cases, ineffective centers of cytoplasmic
division, i.e., cleavage furrows appear between them but after-
wards fade out. Apparently strong evidence was, however,
obtained that in some cases complete division may occur
around asters unconnected with nuclear material. In any case
the cytasters persist for some time and may progressively mul-
tiply by division. The first division actually observed takes
place nearly synchronously with that of the cleavage asters,
at a time when the daughter-nuclei have been formed and are
rapidly enlarging. Division of the asters is in both cases pre-
ceded by a great reduction of the astral rays, leaving the clear
hyaloplasmic central mass surrounded by only short irregular
rays, and at the same time the aster migrates out towards
the periphery of the egg. The mass then draws apart into
two, and recrudescence of the rays from the two centers ensues.
A discrepancy, not yet fully cleared up, lies in the fact that,
although the cytasters divide synchronously with the cleavage
asters, they have not been observed in the living eggs to divide
at the time the dicentric nuclear figure is first formed ; but the
study of sections indicates that this is probably owing to a gap
in the observations. The cytasters ultimately disappear.

Asters are formed also in enucleated fragments, obtained by
shaking the unfertilized eggs into pieces, and such asters may
also progressively divide, though no case was observed of
cleavage, or even an attempt at cleavage, in such fragments.
Sections show that all the asters, whether cytasters or nuclear
asters, or those formed in enucleated fragments, contain cen-
trosomes which have the typical staining capacity and granular
structure observed in normally fertilized eggs. In the cytas-
ters, however, they are usually smaller than in the nuclear

350 ABSTRACTS OF PAPERS. [VOL. II.

asters, and those in the enucleated fragments are smaller still
and often not demonstrable.

The observations indicate that the astral rays, whatever be
their other functions, are in part an expression of centripetal
currents of hyaloplasm (continuous or interalveolar substance),
which lead to the formation of the perinuclear hyaloplasmic
zone, and of the clear centers of the cytasters--a conclusion
essentially in agreement with the early views of Fol. They
show further that the asters (centrosomes) must be regarded
as centers of cytoplasmic division, though not ordinarily effect-
ive unless connected with nuclear material. They seem to
leave no doubt, finally, of the formation de novo of functional
asters and centrosomes, capable of division, and show that such
formation may be entirely independent of the nucleus.

XVI. METAMORPHOSIS IN THE HERMIT-CRAB.

M. T. THOMPSON.

IN Eupagurus longicarpus only the first six larval stages are
distinct : the four zoeas, the important glaucothoe, and the
first of the adolescent stages. In the zoeas and the early part
of the glaucothoe stage, the "livers" or midgut diverticula
are cephalic and thoracic. There are two pairs of these ; a
pair of Lesser Lobes opening dorsally into the stomach, and
a pair of somewhat four-lobed Greater Lobes opening laterally
into the stomach.

During the glaucothoe stage, however, three of the divi-
sions of the greater lobes become atrophied. The fourth or
posterior division, at about the time of the second or third day
in the shell, grows back into the abdomen. But the lobe of
the right side of the body crosses under the intestine to the
left, so that both lobes lie on the left of the intestine, which
is thrown to the right. At this time the bladders of the
Green Glands also migrate into the abdomen. Then the append-
ages which will be lost become atrophied, and the body mus-
culature alters to the adult type. So the glaucothoe, which was

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 351

at first Macruran, attains the Eupagurid structural plan before
the moult to the sixth stage occurs.

The duration of the glaucothoe stage is dependent on the
time of entering the shell. Specimens which take the shell
within the first twenty-four hours after the moult from the
fourth zoe'a, spend only four or more often Jive days in the
stage. A delay of four days before the shell is taken pro-
longs this period to six or, in a few cases, to seven days. In
fifty glaucothoe kept without any shells, some remained in the
stage the minimum four days, but the majority remained six
and seven days, and one remained eight days.

The sixth and following stages introduce no important
changes in structure, except the branching of the liver. In
this branching, however, the majority of the diverticula of the
right lobe go to the right under the intestine, so that this lobe,
in the adult, apparently lies on this, its own side, of the intes-
tine. The lesser lobes branch later and finally form a small
tuft of tubules on each side of the stomach.

XVII. ESSENTIAL FACTORS IN THE REGEN-
ERATION OF PLANARIA MACULATA.

CHARLES RUSSELL BARDEEX.

REGENERATION of a new whole individual from a small piece
of a parent individual depends in Planaria rnaculata upon
the presence in the piece of a part of the central nervous system
and a part of the intestinal system. A small piece containing
these parts will regenerate from them new typical intestinal and
nervous systems. At the same time the parenchyma in the
vicinity of the cut surface becomes increased in amount, and is
symmetrically differentiated in relation to the new intestinal
system. A head may thus be formed anterior to a new axial gut,
and lateral and tail areas may be restored. Polarity of the piece
is determined by the central nervous apparatus which it contains.

A new pharynx is formed just posterior to the point where
intestinal contents collect when the whole piece contracts.

352 ABSTRACTS OF PAPERS. [VOL. II.

The pharynx may be formed in a region of the piece at some
distance from the cut surface. Head, lateral, and tail areas are
differentiated only at a cut surface.

The reproductive organs are not regenerated. Instead, they
disappear from a small piece isolated from a sexually mature
worm. The tail cut from a planarian in which the reproductive
organs are developing will give rise to regenerative forces which
overpower the forces giving rise to the sexual organs. Regen-
eration is equally rapid in sexually mature and in sexually
immature worms.

In regeneration in this animal the tissues seem to be spe-
cific, except that the new musculature probably comes from
parenchyma cells.

A full account of the Physiology of Regeneration in these
animals is given in the American Journal cf Physiology,
Vol. V (1901), p. i.

XVIII. THE HISTOGENESIS OF THE PERIPHERAL
NERVOUS SYSTEM IN SALMO SALAR.1

ROSS GRANVILLE HARRISON.

CELLS provided with pseudopodia-like processes wander out
singly from the dorsal surface of the medullary cord, and
collect together between the myotomes and the cord into
small groups, the spinal ganglia. Here the cells remain for
some time undifferentiated, but are transformed later into
bipolar cells, of which the centripetal processes grow into
the side of the medullary cord to form the dorsal roots.

Neuroblasts may be distinguished at an early stage as round
or polyhedral cells, lying in the outer zone of the cord. At
this period the cord is made up chiefly of epithelial cells, the
forerunner of the ependyma. These cells are still undifferen-
tiated, no specialized " Ra ndschleier " being present. As the
axones grow out from the neuroblasts, they bore their way

1 A full account of this work is published in the Arc/tiv fiir mikroskopische
Anafomit-, January, 1901.

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 353

through the substance of the epithelial cells, which with the
continued growth of new fibres become more and more honey-
combed. Their outer zone is finally transformed into a fibrous
framework, the " Randschlcier" which accordingly owes its
structure to the activity of the growing nerve fibres, and is
not pre-formed.

The dorsal cells or giant cells of Rohon arise in the clorso-
lateral portion of the cord next the outer limiting membrane.
They elongate, and for the most part each cell gives rise to an
ascending and a descending nerve fibre forming the beginning
of the dorsal funiculi. Gradually the cells leave their origi-
nal position and wander to the dorsal mid-line of the cord.
Through this movement the cells become unipolar, remaining
connected with their fibres by a slender process, which divides
in T-fashion at the point where the longitudinal fibres begin.
A large number of the cells form peripheral nerves also, which
are segmentally arranged. The dorsal cells are homologous
with the " Hintersellen " found in Petromyzon, and with the
bipolar cells of medium size in the cord of Amphioxus. They
are to be regarded as a primitive type of sensory cell identical
in function with the spinal ganglion cells, with which they are
genetically related.

XIX. THE SPERMATIC AND MESENTERIC

ARTERIES OF DIDELPHYS VIRGINIANA

(KERR, LINN.).

C. F. W. McCLURE.

IN mammals other than -marsupials the anterior mesenteric
artery supplies the small intestine and the proximal end of the
large intestine. The posterior mesenteric artery is given off
from the posterior division of the abdominal aorta, and supplies
the large intestine. In a large number of mammals the internal
spermatic arteries are given off from the aorta about midway
between the renal and posterior mesenteric arteries.

In Didelphys and other marsupials, so far as known to the
writer, the anterior mesenteric artery supplies both the small and

354 ABSTRACTS OF PAPERS. [VOL. II.

large intestines. In Didelphys and other marsupials the pos-
terior mesenteric artery is not present. Also in DidelpJiys and
other marsupials the internal spermatic arteries are given off
from the posterior division of the aorta, and at a point which
coincides with the point of origin of the posterior mesenteric
artery in other mammals.

In an adult DidelpJiys killed during the breeding season the
writer found present two pairs of functional internal spermatic
arteries. The anterior pair was given off from the aorta about
midway between the renal and posterior pair of internal sper-
matic arteries. The posterior pair, the so-called internal sper-
matic arteries of marsupials, was given off from the aorta in
the usual manner, as mentioned above. More recently the
writer has found another adult female DidelpJiys, in which, in
addition to two pairs of internal spermatic arteries, a large
posterior mesenteric artery was present.

In this individual the posterior mesenteric artery arose
from the aorta as a single vessel, and at a point which coin-
cided with the origin of this vessel in other mammals. On
arising from the aorta the vessel passed ventrad through a
foramen in the vena cava, and was distributed to the large
intestine. The anterior mesenteric artery in this individual
supplied the small intestine and the proximal portion of the
large intestine.

The relations of the spermatic arteries were as follows :

The anterior pair of internal spermatic arteries arose from
the aorta, and was distributed to the ovaries as in the above-
mentioned case. These arteries appear to be the homologues
of those spermatic arteries which in many other mammals
arise from the aorta about midway between the renal and
posterior mesenteric arteries. The posterior pair of internal
spermatic arteries in this opossum were branches of the posterior
mesenteric artery, and were given off from this vessel near its
point of origin at the aorta.

It appears to the writer as though in the marsupials, as the
result of an arrested development of the original internal
spermatic and the posterior mesenteric arteries, a new collat-
eral circulation has been established to the genital organs

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 355

and large intestine. The collateral circulation to the large
intestine has apparently been established through the anterior
mesenteric artery; that to the ovaries, through vessels which
may have been formed as the result of a modification of the
posterior mesenteric artery.

XX. SOME FACTS CONCERNING REGENERATION
AND REGULATION IN RENILLA.

H. B. TORREY.

DURING the past summer experiments were carried on at
Beaufort, North Carolina, preliminary to a more complete
investigation of the processes of regeneration and regulation
in Rcnilla. It was hoped that Rcnilla, being a polymorphic
colonial form, --an aggregate of polyps and zooids, -- would
behave like a simple metazoan individual, and at the same
time offer surer landmarks, during regulative processes, than
a metazoan individual --an aggregate of cells; for changes
in polyps as a whole may be perceived more clearly than
changes in their component cells.

The results may be summarized as follows :

1. Renilla colonies may regenerate lost parts readily.

2. They exhibit a strong polarity. When a peduncle is
removed by a transverse cut an axial polyp is never regen-
erated in its place, and vice versa.

3. There is an anterior limit beyond which anterior pieces
do not regenerate posteriorly, and a posterior limit beyond
which posterior pieces do not regenerate anteriorly. These
correspond to the limits of the budding zone.

4. The colonies regulate themselves in a plastic fashion
when cut in certain ways, obliquely, for instance. It is thus
possible to obtain two new colonies, one of which retains the
original peduncle with a lateral polyp displaced into the posi-
tion formerly occupied by the axial polyp. Whether or not
the colony develops symmetrically around this new axis is
not known.

356 ABSTRACTS OF PAPERS. [VOL. II.

If the oblique cut makes with the colonial axis an angle
larger than forty-five degrees, there is no displacement of the
lateral polyp, the extirpated axial polyp regenerating as though
it alone had been removed by a transverse cut.

5. When a lateral group of polyps is removed by a longi-
tudinal cut, it regenerates a new peduncle approximately at a
right angle to the cut surface, and approximately in the axis
of the chief lateral polyp of the group. The future of such
pieces is unknown. This is a case of heteromorphosis.

XXI. SOME POINTS IN THE BRAIN OF
LOWER VERTEBRATES.

J. B. JOHNSTON.

THE central olfactory apparatus of Petromyzon presents, in
all important features, an extraordinary resemblance to that
of Acipenser. In Petromyzon, on account of the great buccal
apparatus, there has occurred a sort of telescoping of the olfac-
tory lobes and areas upon the striatum and thalamus as fixed
points. The so-called cortex, described by Friedrich Mayer,
is nothing else than the epistriatum.

The cells of the olfactory lobe present more primitive char-
acters in Petromyzon than in Acipenser. The mitral cells are
only slightly differentiated, while the stellate and other cells
are very numerous and send their neurites, along with those
of the mitral cells, to the olfactory nuclei of the fore-brain.
Similar categories of cells have been described in Amphibia
(P. R. Cajal) and reptiles (Edinger's " Lobuscortex "), although
differently interpreted. The numerous, slightly differentiated
cells in the olfactory lobe of Petromyzon and Acipenser
represent the material from which the highly differentiated
elements of the olfactory lobe of higher vertebrates have
been developed.

Several authors have pointed out the close connection
between the cerebellum and acusticum in fishes. The study
of the minute structure shows that the cerebellum is derived

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 357

directly from the front end of the acusticum. Evidence for
this from Acipcnscr :

a. Gross continuity of cerebellum and acusticum.

b. The root fibres of the fifth, eighth, and lateral line nerves
enter and end in both.

c. The several categories of types of cells in the cerebellum
-Purkinje cells, granules, and cells of the second type --are

strictly homologous with similar cells found in the acusticum.

d. The development of the Purkinje cells in the acusti-
cum from the typical large cells of that nucleus is in actual
progress and may be studied in all its stages.

Additional evidence in Petromyzon :

a. The Purkinje cells in the cerebellum are not well devel-
oped, and their neurites run to the base of the mid-brain,
possibly having the same destination as the internal arcuate
fibres from the acusticum.

b. The tracttis tecto-cerebellaris seems to be absent and the
tractus lobo-ccrebellaris is small.

c. The cerebellum is little more than a dorsal arch and
commissure from the front end of the acusticum.

XXII. ASEXUAL REPRODUCTION OF PLANARIA

MAGULATA.

WINTERTON C. CURTIS.

THE fission of Planaria maculata, while it does not differ
essentially from the type found in other planarians where fis-
sion occurs, is of just the right sort to complete a very interest-
ing series and connect the fission of land planarians, which
is hardly more than a fragmentation, with the fission of
Planaria fissipara, in which the organs are completely formed
before the new individuals separate. This series is as follows :
(i) Land planarians, in which pieces of varying lengths are
pinched off from the posterior end; (2) Planaria maculata,
which divides always at the same place behind the pharynx,
with no preformation of organs; (3) Planaria subtentacula

358 ABSTRACTS OF PAPERS. [\'OL. II.

(Zacharias, Zeit.f. Wiss.Zool., 1886, Bd. XLIII, pp. 271-275),
where there is some rearrangement of the gut and the pharynx
is partly developed before separation ; (4) Planaria fissipara
(Kennel, Zool.Jahrb., Abth. f. Anat. u. Ontog., 1888, Bd. Ill,
pp. 447-486), in which a complete worm is developed out of
the posterior third of a large specimen and both reach normal
proportions before separation. In the last three cases the
division occurs at a corresponding place.

The division in Planaria macnlata seems to be brought
about by a contraction of the circular muscles, which pinches
the individual in two a short distance behind the pharynx.
The cut ends of either piece are as though they had been
produced by a knife-cut, and examination of sections shows
that the parenchyma at the scar is actually naked. There is
nothing like a furrow on the outside previous to the division,
which, nevertheless, is a regular and normal reproductive
process and not induced by any ordinary irritation or mutila-
tion of the animals. The large number of pieces in various
stages of regeneration, found in collecting, is sufficient evidence
of the occurrence of the fission under natural conditions.

Worms will not divide in the laboratory to any considerable
extent unless well fed. If the water has become foul and is
replaced by fresh, a considerable number of specimens will
usually be found divided within the next twelve hours. The
division usually occurs at night, whether the dishes are shaded
or not. The morpholaxis of the head and tail pieces resulting
from a division is rapid. Tail pieces may reach almost the
normal proportions and re-divide in from five to six days if
well fed, heads in not less than ten days. There is no regular
interval between the divisions.

In certain localities this species does not possess reproduc-
tive organs at any time during the year, but has during the
summer months an active period of asexual multiplication.
In other localities the worms develop these organs in the fall
and lay eggs in the spring ; and although all the specimens are
without these organs at the end of the summer, asexual repro-
duction has never been observed. In another sexual locality
the worms, when they are without reproductive organs at the

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 359

end of the summer, do reproduce asexually to a considerable
extent. Later this ceases and the reproductive organs develop.
These statements are based upon observations extending over
from two to three years.

A possible explanation is that the asexual reproduction may
be substituted for the sexual in certain localities during con-
siderable periods, but further data are necessary to confirm
this.

XXIII. VARIATION AMONG HYDROMEDUSAE.

CHARLES W. HARGITT.

OBSERVATIONS upon variations among the Hydromedusae
seem to have been of comparatively limited extent. Refer-
ences to the subject are to be found in the writings of Ehren-
berg, Forbes, Agassiz, Hincks, Romanes, and later by Agassiz
and Woodworth, but except in the last-named paper they are
rather incidental and fragmentary.

Of my own observations only the barest abstract and sum-
mary can be undertaken in this connection.

Among the genera studied the principal have been as
follows : Pennaria, Encopc, Obelia, Margclis, Gonionemns,
NemopsiSy, Rkegmatodes.

The principal organs examined were: (i) The Chymiferous
Canals, (2) Tentacles, (3) Gonads, (4) Otocysts. Among these
the greatest range of variation was noted in the tentacles, as
might naturally be expected, in some cases reaching as high as
90 per cent. In the forking and doubling of tentacles there
was least, rarely exceeding 5 per cent, and indeed seldom
reaching that ratio ; in Gonionemus 3 per cent.

In the looping, branching, and anastomosing of chymiferous
canals there was great variation in different genera, in some
being almost nil, while in others (Encope and Gonionemus)
varying from 5 to 10 per cent.

Considerable variation was found in the gonads, though less
than in the other organs already noted, varying in different
genera from 2 to 5 per cent. While considerable variation

360 ABSTRACTS OF PAPERS. [VOL. II.

was evident in the number, arrangement and correlation of
the otocysts, no attempt has been made to ascertain the exact
ratio, owing to the difficulty attending this determination in
preserved specimens.

The following summary will express in a general way some
of the more evident conclusions reached :

1. Variation among Hydromedusae is of wider extent than
had been suspected.

2. It is much greater in some genera than in others.

3. It seems to be much less symmetrical and correlated
than among Scyphomedusae.

4. Many phases of variation appear to be wholly devoid
of correlative and adaptive aspects.

XXIV. EXPERIMENTS ON MODIFYING THE

NORMAL PROPORTION OF THE SEXES

IN THE DOMESTIC FOWL.

GEORGE WILTON FIELD.

THIS is a brief report on a series of experimental attempts
to ascertain the factors which determine sex.

The normal proportion given by Darwin from observation of
100 1 chicks during eight years was 94.7 males to every 100
females. From 2105 chicks during two years, we found the
proportion to be 80.6 males to every 100 females.

These figures lead us to query whether the normal propor-
tion may not have changed during the past forty years as a
result of the breeders' desire to produce a larger proportion
of females.

In the experiments attempts were made to isolate the factors
so that the effects of each could be observed:

i . Absolute age of parents :

9 young females mated to male of same age.

2. Relative age of parents :

9 old females mated to young male.
9 young " " " old "

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 361

The proportion practically coincided with the normal except
in the case of young females mated to old male, where a slight
increase in the number of males appears.

3. Malnutrition (a result of feeding to one-half usual amount) : a very
marked increase in number of males ; rate of 1 76.6 males to every
100 females.

4. Scarcity of males, i.e., polygamy : 20 females mated to i male, a
marked increase in number of males; 139.6 males to every 100 females.

5. Conditions connected with time of year : 650 chicks hatched between
March 15 and May 15 gave a smaller proportion of males, — 73.9 males to
every 100 females; while 471 chicks hatched between May 15 and July 15
gave a larger proportion of males, — 88.4 males to every 100 females.

It is to be understood that these figures are for a relatively
small number of cases ; it is hoped to extend them to at least
10,000 individuals.

XXV. NOTES ON VARIATION IN THE
SHELLS OF PURPURA LAPILLUS.

R. P. BIGELOW AND H. S. CONANT.

Purpura lapillits is a species that presents great variations
in its diagnostic characters. It was thought, therefore, that a
study of its variations by statistical methods might be of value
in defining more exactly the limits of the species, and might
also bring to light facts of general biological interest. Collec-
tions were made at Prince's Cove and on the mainland opposite
Clark's Ledge, Eastport, at Kennebunk Beach, at Bass Rocks,
Gloucester, and at Newport. The sexes were separated for
each locality, and the following characters were measured :

(i) Angle of the apex and nuclear whorls, or nuclear angle;
(2) angle of the apex and the last whorl, or adult angle ; (3) total
length ; and (4) length of spire from the apex to the poste-
rior margin of the opening of the shell. Record was made
also of the presence of (5) imbrications, (6) sutures, (7) ribs,
(8) teeth, (9) of the curvature of the columella, and (10) of the
weight. Perhaps the most obvious variations are in the color
and the thickness of the shell, but no satisfactory method was

362 ABSTRACTS OF PAPERS. [VOL. II.

found for measuring these quantities. A special instrument is
being constructed which, it is hoped, will overcome the difficulty
in regard to thickness.

A preliminary study of the shells from Eastport and Glouces-
ter shows that for characters (i) to (4) the variations may be
represented by curves that are approximately normal. The
curves for the two stations at Eastport fit together pretty
closely ; while they differ distinctly from the curves for the
Gloucester specimens, the difference of the means being
greater than the standard deviation for each locality. In
each case the female shells showed, on the average, a wider
angle and a shorter spire expressed in per cent of total length,
than the corresponding males, and the same is true for the
Gloucester shells as a whole compared with the Eastport
shells.

As a measure of variability the coefficient of variation

f cv. — 100— ] gives contradictory results and appears not to

\ mJ '

be applicable to measurements expressed in degrees of a circle.

Judging from the standard deviations, the shells from Gloucester
are somewhat more variable than those from Eastport. The
relative variability of the males and females differs for the
different characters, and for the same characters in different
localities. In general, the females appear to be slightly more
variable than the males.

XXVI. VARIATION AND ELIMINATION IN
PHILOSAMIA CYNTHIA.

HENRY E. CRAMPTON.

SOME of the results were presented of a statistical study in
the case of a large Saturnid moth, P. cynthia, of the variability
of eliminated and surviving pupae and imagines. From a lot
of nearly 1 100 cocoons, only 310 living pupae were obtained;
632 contained dead pupae; while the remainder were shriveled
or otherwise abnormal larvae or pupae. The living pupae were
compared with an equal number of dead pupae with reference

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY. 363

to certain body characters : length, length of bust (to fifth
abdominal segment), width and depth of bust, frontal stature of
bust (ratio of middle to length), and sagittal stature of bust (ratio
of depth to length). In addition the length, width, and stature
of a typical organ, the left antenna, were determined. From
a tabulation of certain bases of comparison --mean, standard
deviation, coefficient of variability - - it appeared : (i) That elimi-
nated male pupae were on the whole more variable than the
surviving- males, and that the surviving females were far less
variable than the dead ones. (2) Only 180 of the 310 living
pupae produced perfect moths. The perfect male moths were
from pupae, which were far less variable than the others. This
condition was reversed in the case of the females, yet the sur-
viving females, though more variable than the eliminated ones,
were not as variable as the eliminated female pupae. (3) The
males of all groups were more variable than the females.

XXVII. THE ORIGIN OF TENTACLES
IN GONIONEMUS.

H. F. PERKINS.

SOME interesting data have been secured from the study of
the origin of tentacles in Gonionemus, a common Woods Holl
hydromedusa. Specimens \ mm. in diameter having from 8
to 1 6 tentacles are found in early summer, and in examining
these it was seen that there existed a definite relative position
of the tentacles and sense organs.

Two pairs of tentacles, the radial ones, are of equal size.
The other tentacles and sense organs are regularly graded
from large to small, so that it is possible to determine their
order of origin.

Looking at the marginal ring from below, in a normal
medusa, each newly formed tentacle is seen to lie next to a
sense organ and to precede it, as the hands of a watch move.
Fig. i shows a typical 8-tentacled medusa. Tentacles I and
II are radial in position; III follows I in the direction of the

64

ABSTRACTS OF PAPERS.

[VOL. II.

hands of a watch ; IV follows II, and the two pairs of sense
organs, i and 2, lie as if the cells that were to form them
had been crowded along to the right by the newly formed
tentacles.

In an older specimen (Fig. 2) the successive pairs of tenta-
cles and sense organs have arisen in corresponding positions,
as is indicated by the numbers on the diagram. The origin

FIG.

of tentacles and sense organs would seem then to be governed
by an attempt at radial symmetry which is constantly inter-
fered with by this sequence of formation from left to right,
along the bell-margin.

In full-grown medusae there appears a striking conformity
to this rule, with fewer exceptions than would be expected
from the frequency of other variations in all parts of the
creature.

No. 6.] AMERICAN MORPHOLOGICAL SOCIETY.

365

The comparison of other and allied forms, with this rule in
mind, may bring to light some interesting facts bearing on the
correspondence of parts and on radial and bilateral symmetry.

FIG. 2

MBI. WHOI LIBRARY

UH 17JE