ANIMAL AGGREGATIONS

THE UNIVERSITY OF CHICAGO PRESS
CHICAGO, ILLINOIS

THE BAKER & TAYLOR COMPANY

NEW YORK
THE CAMBRIDGE UNIVERSITY PRESS

LONDON

THE MARUZEN-KABUSHIKI-KAISHA
TOKYO, OSAKA, KYOTO, FUKUOKA, SENDAI

THE COMMERCIAL PRESS, LIMITED
SHANGHAI

ANIMAL

/

/'

AGGREGATIONS

A Study in General Sociology

By W. C. ALLEE

The University of Chicago

THE UNIVERSITY OF CHICAGO PRESS i

CHICAGO • ILLINOIS

COPYRIGHT I93I BY THE UNIVERSITY OF CHICAGO
ALL RIGHTS RESERVED. PUBLISHED MARCH I93I

COMPOSED AND PRINTED BY THE UNIVERSITY OF CHICAGO PRESS
CHICAGO, ILLINOIS, U.S.A.

IN MEMORY OF MY SON

WARDER ALLEE

1913-23

IN APPRECIATION OF HIS BOYISH ENTHUSIASM

OVER THE PARTS OF THIS WORK

WHICH HE UNDERSTOOD

PREFACE

The attempt to summarize knowledge concerning the relations
within and between different sorts of animal societies is not new.
Espinas in 1878 undertook such an effort, and shows in his introduc-
tion that Aristotle, Spinoza, Leibnitz, Montesquieu, Kant, Hegel,
A. Comte, Herbert Spencer, and others had preceded him in the
consideration of certain aspects of this problem. Since the time of
Espinas, knowledge concerning the social insects and the develop-
ment of insect societies has greatly increased. Wheeler, Forel,
Buttel-Reepen, and many others have contributed both personal ob-
servations and generahzing summaries of value ; and I have no desire
to enter into a field so ably covered. There does remain, however, a
field of social, or perhaps subsocial, life almost entirely unJ;ouched by
these students. They have been concerned with the fascinating prob-
lems and intricate relationships presented by fairly well-developed
societies. Here, I propose to investigate the relationships existing
among the more loosely integrated collections of animals, which may
rightly be designated as "animal aggregations," with regard to their
ecological and behavioristic physiology, as well as with regard to
their strictly social implications.

This book is built about a phenomenon or a series of phenomena,
rather than about a philosophy. In the present form it may even be
designated as notes on an unsolved problem; but since a presenta-
tion of a problem is necessary for its ultimate solution, and since an
inquiry into the universahty of a given problem is imperative before
undertaking laborious experimentation directed toward finding a
solution, no apology is offered for summarizing our growing knowl-
edge on the subject of animal aggregations at the present stage of
inquiry into the problems involved.

My own experimental work within the field covered by the pres-
ent book began in 191 1 and has continued intermittently to date.
The investigation of animal aggregations has been at the center of

viii PREFACE

my research program for the last twelve years, during which time
work has been actively carried on with the aid of a number of gradu-
ate students, with facilities furnished by the University of Chicago,
the Marine Biological Laboratory, and, more recently, with financial
aid from a grant from the Rockefeller Foundation to aid investiga-
tions in the biological sciences at this university. The preparation of
the manuscript of this book has been, in part, supported by aid from
this latter source.

In addition to the loyal co-operation of students and colleagues in
the accumulation of experimental data, of citations, and of criti-
cisms, aid with the scattered literature has come from friends and
acquaintances from five continents. The extended hterature hst is
incomplete, but the labor of gathering and selecting the references
used has been appreciably decreased by this cordial co-operation.

Certain specific acknowledgments I have made in the text. I am
also indebted to Drs. Marie A Hinrichs, A. M. Holmquist, Walburga
A. Petersen, J. M. Shaver, and O. Park; to Messrs. M. R. Garner,
J. R. Fowler, J. F. Schuett, W. A. Dreyer, Carl Welty, W. H. John-
son, E. O. Deere, Ralph Buchsbaum, D. A. D. Boyer, J. F. W. Pear-
son, and T. Park; to Mrs. Frances Church van Pelt and Mrs. Gret-
chen Shaw Rudnick; and to Miss Edith Bowen, for citations to per-
tinent hterature, for permission to give results before pubhcation, or
for criticism of parts of the manuscript, or for all three; to Professor
F. R. Lillie, who read critically chapters xvi and xvii; to Professor
A. E. Emerson for similar service with chapters xix and xx; to Mr. K.
Toda, who drew or copied the text figures; and to Marjorie Hill
Allee for editorial assistance with the manuscript. Acknowledgment
of courtesy in permitting reproduction of figures will be given else-
where.

W. C. Allee

Whitman Laboratory or Experimental Zoology
University of Chicago
August, 1930

CONTENTS
INTRODUCTION

CHAPTER PAGE

I. The General Background 3

II. Classification of Animal Aggregations 12

III. Formation of Animal Aggregations 38

IV. General Factors Conditioning Aggregations .... 65
V. Integration of Aggregations 81

HARMFUL EFFECTS OF AGGREGATIONS

VI. Harmful Effects of Crowding upon Growth . . . . 10 1

VII. Retarding Influence of Crowding on the Rate of Repro-
duction 120

VIII. Crowding and Increased Death-Rate 136

BENEFICIAL EFFECTS OF AGGREGATIONS

IX. Stimulation of Growth by Crowding 147

X. Stimulating Effects of Crowding on the Rate of Repro-
duction 161

XI. Effect of Crowding on Survival and Oxygen Consumption 181

XII. Protection from Toxic Reagents 201

XIII. Resistance to Hypotonic Sea-Water 222

XIV. Relation between Density of Population and Insect Sur-
vival 236

XV. Communal Activity of Bacteria 247

XVI. Mass Physiology of Spermatozoa 263

GENERAL EFFECTS OF AGGREGATIONS

XVII. Influence of Crowding upon Sex Determination . . . 289

.XVIII. Morphological Effects of Crowding 311

CONCLUSION

XIX. Animal Aggregations and Social Life 337

XX. The Principle of Co-operation 352

Bibliography 365

Index 413

.'^T6G3

INTRODUCTION

CHAPTER I
THE GENERAL BACKGROUND

INTRODUCTION

This study of animal aggregations is concerned with some of the
physiological effects of crowding upon the individuals composing the
crowd, and is offered as a contribution toward the development of
general sociology upon a physiological basis. A few years ago it
would have been possible to summarize the knowledge' then existing
on the subject with the statement that, except in hibernation or at
breeding time, the physiological effects of crowding are uniformly
harmful, whether attention is given to the effect upon rate of repro-
duction, rate of individual growth, or longevity. Data on these
harmful results will be presented later, but they are no longer ac-
cepted as a complete picture; in this study they are needed to ob-
tain a correct perspective for the recent discoveries of beneficial
effects of relatively unorganized crowds of animals.

Much attention has deservedly been given to the study of or-
ganized societies, particularly those of mammals, birds, and insects —
sometimes with relation to the light they may throw upon the social
relations of man, but frequently on account of their own inherent
interest. In the main, consideration of these highly organized social
groups falls outside the interests of the present discussion, which will
be limited, so far as possible, to the physiological effects of crowding
upon organisms whose interrelations have not reached the level of
development usually called "social."

The general physiologists contend with justice that one cannot
understand the physiology of man without a knowledge of the gen-
eral physiology of all animals and much of that of plants as well. The
comparative psychologists conclude similarly that one cannot under-
stand the working of the human nervous system without knowing
how other nervous systems function. Similarly, an increasing num-

3

4 ANIMAL AGGREGATIONS

ber of investigators are convinced that without a knowledge of gen-
eral sociology we are likely to regard the social traits exhibited by
man or by the ants as being peculiarly human or peculiarly formi-
cine, when many of them are merely human or ant variations of
social traits common to animals in general. Again, it is difficult to
evaluate properly the origin and function of many of these general
social traits without a proper understanding of their physiological
antecedents among animals not usually regarded as having reached
the social level. It is quite easy to consider certain bits of behavior
as definitely social in origin and inherent in the social type of organi-
zation which may be merely specialized developments of general be-
havior common to most animals when crowded.

One fallacy may be suggested at the beginning. All too frequently
one gains the impression that sex forms the main, if not the only,
physiological connecting link between the infrasocial and the social
animals. I beheve that a consideration of the facts to be presented
will allow us to place this important social factor more nearly in its
proper relation to other factors equally important.

The problems dealt with in the present study are of interest also to
the large group of students of animal ecology. It is generally known
that ecology deals with the relations between the organism and its
environment. This environment is roughly divided into two parts—
the non-living and the Hving — which are commonly referred to as
the "physical" and the "biotic" elements of the environment. Of
these, the former has received particular attention at the hands of
modern animal ecologists, since such factors as light, hydrogen-ion
concentration, humidity, wind velocity, and temperature are more
or less readily and definitely measured, and since others, such as soil
type or the chemical composition of the waters of a lake or river,
though less readily analyzed, are still capable of being studied on a
quantitative basis. Meantime the analysis of the biotic relations of
the environment has lagged, probably on account of the greater
difficulties involved in the quantitative treatment of this exceedingly
complex part of the environment. Even so, marked progress has
been made in this analysis by recent students of the ecological rela-
tions within animal and plant communities (Smith, 1928; Shackle-

THE GENERAL BACKGROUND 5

ford, 1929). In the present studies we shall find ourselves concerned
with animal communities which, from their concentrated nature,
necessarily make the biotic elements an important aspect in the en~
vironment of any particular individual, while the physical elements
of the environment act mainly through their iniluence on the entire
aggregation or crowd. Such a situation must frequently obtain in
assemblages of sea anemones, of Mytilus, of ascidians, or of crabs.
In working at the aggregation level here considered, we find the
ratio of importance of the physical and the biotic environment in a
transition stage between that present in definitely social groups and
that occurring in the more typical animal community, or biocoenose
of the ecologist.

We must emphasize the fact that all studies dealing with the
biotic elements of the environment are likely to be less definitely
quantitative than those dealing with the non-living environment.
This is no reason for their neglect, but it is a reason why we may not
expect their treatment to be precise and final. The present summary,
gained from pioneering in this relatively new field, must be regarded
as tentative in many respects. My own research program dealing
with various aspects of the subject is only well under way; the pres-
ent statements furnish a point of departure, rather than a gathering-
in of conclusions. With the accumulation of evidence now being
actively collected, the conclusions tentatively advanced here may
be further confirmed, or they may soon be modified or entirely aban-
doned. This must always be the case, even in the well-developed
fields of physics and chemistry; and does not prevent summaries of
knowledge to date having definite value, if they stimulate further
research or give point to researches already in progress.

TERMINOLOGY

The general terminology causes unexpected difficulty. One usual-
ly thinks that such words as "society," ''association," and "com-
munity" have a relatively stable meaning and that "biocoenosis,"
for example, might be expected to be a quite exact term; but this is
unfortunately not the case.

According to writers on human sociology (Park and Burgess,

6 ANIMAL AGGREGATIONS

1921), "the terms society, community and social group are now used
by students with a certain difference of emphasis, but with very
Httle difference in meaning. Society is the more abstract and inclu-
sive term, and society is made up of social groups, each possessing
its own specific type of organization but having at the same time all
the general characteristics of society in the abstract. Community is
the term applied to societies and the social groups where they are
considered from the point of view of the geographical chstribution of
the individuals and institutions of which they are composed. It fol-
lows that every community is a society, but not every society is a
community. An individual may belong to many social groups but
he will not ordinarily belong to more than one community, except in
so far as a smaller community of which he is a member is included in
a larger of which he is also a member. However, an individual is not,
at least from a sociological point of view, a member of a community
because he lives in it but rather because, and to the extent that, he
participates in the common life of the community."

The same authors evidently do not consider ''association" as be-
ing a sufficiently significant term to be given formal definition. In
general sociology the contrast between what is ordinarily called an
"association" and a "society" is important. Students differ concern-
ing the proper criteria to use in making this distinction. Espinas
(1878) recognized that there was a difference, and called "associa-
tions" accidental societies between animals of different species. Ac-
cording to this pioneer in the field of general sociology, the charac-
teristic trait of social fife is to be found in habitual reciprocity be-
tween activities which are more or less independent. He recognized
certain similarities between associations and societies but regarded
the former as less necessary for their constituent elements. Associa-
tions, according to Espinas, are groups of convenience, not of neces-
sity. Deegener sharpened this distinction (1918) on the basis of use-
fulness of the animal group to the individual members. He designat-
ed an "association" as a collection of similar or dissimilar animals,
which does not have value for the individuals composing the group ;
and a "society" as one in which the collection does have distinct
value for the individuals of which it is composed.

THE GENERAL BACKGROUND 7

Deegener's criteria for the social value of his categories were far
less sensitive than those which were shortly developed by other
workers in this field, and which will be summarized in the body of
the present discussion. Apphcation of such distinctions, even in their
present incomplete form, would necessitate a marked revision in
Deegener's scheme.

Later, Deegener recognized that certain groups of animals are
held together by a social force or instinct of which we know at present
relatively httle. The arrangement of such groups in his original sys-
tem is obviously difficult. One may think of the satisfaction of the
so-called "social force" or "instinct" as having definite value for the
animal so satisfied. According to this reasoning, the group collected
by social instinct would be a "society"; although, since there is no
other demonstrable advantage accruing to the members of the group,
Deegener at first was inchned to regard such an aggregation of in-
dividuals as an "association." Faced with this dilemma, he decided
(191 9) that associations whose occurrence depends upon a social in-
stinct may be designated as "instinctive associations." They are
opposed to aggregations of purely accidental character which are
formed not because of instinct but because of hmited space or local-
ized food. If the aggregation is formed from obvious mutual attrac-
tion but without any recognizable objective benefit to the members,
Deegener calls it an "instinctive association," as with young spiders,
young ticks, or groups of grasshoppers.

Alverdes (1927) understands by "associations" the chance gather-
ings produced solely by external factors, such as insects collected
around a lamp, while "societies" are genuine communities held to-
gether by the force of a social instinct. "In short," Alverdes says,
"no social instinct, no society!" According to this point of view, the
individuals are collected into an association because of their re-
sponses to environmental factors, but they collect into a society
primarily because of the presence of other similar animals and only
secondarily because of the action of environmental forces. Alverdes
v/ould consider the lack of a social instinct all the evidence necessary
for calling such a group an "association."

Wheeler (1928), commenting on these two classification schemes.

8 ANIMAL AGGREGATIONS

doubts the applicability of Deegener's basic principle of benefit or
no benefit, but commends Alverdes' position as being essentially
sound.

The ecological use of these and related terms must needs be con-
sidered. Modern ecological work has shown that each different kind
of a habitat contains a more or less characteristic set of animals
which are not mere accidental assemblages but are interrelated com-
munities. When these are geographic in extent, they are usually
spoken of as a "formation," within which may be recognized smaller
units or "associations," which are composed of groups of habitat
strata that are uniform over a considerable area but smaller than a
formation. These associations are frequently composed, at least in
part, of developmental stages, such that an orderly succession of
communities can be recognized. Such a series, forming a unit of
succession from initial to climax stages of an association, is some-
times called a "sere"; and the different developmental units of the
whole association are called "associes." Thus, we have the animal
communities or associes of the open sand, the foredune, the pines,
the oaks, and fmally the beech and maple forest, forming one de-
velopmental sere arranged in the order given, within the beach and
maple association near Chicago (Smith, 1928; Shackleford, 1929).

In ecology the term "animal society," according to the most re-
cent usage (Smith, 1928), has been divided into two parts, one of
which is called a "pre-society."^ This is a community of organisms
living among the plants of an association and subordinate to the
plant dominants. The "plant association" is named for one or more
of the dominant plants, while one or more of the predominant ani-
mals give the name for the "super-society." The super-society, like
its accompanying plant association, generally covers an extensive
area with an essentially uniform taxonomic composition. Within such
a super-society one finds animal societies which are communities of
lesser magnitude and which may be seasonal, or stratal, or confined
to a given locality. When these societies, or recognizable subdivi-
sions of them, are composed of animals closely bound together by
biotic relationships such as have been described in general terms as

' "Super-society" would appear to fit the meaning more exactly.

THE GENERAL BACKGROUND 9

composing a "web of life," they are frequently called "biocoenoses."
Food and shelter relationships, climatic and edaphic factors, are im-
portant determining conditions for a given biocoenosis.

With the growing complexity of ecological terminology and the
growing precision with which different terms are applied to various
recognizable groupings of animals, a need has developed for some
term which could be used in a general sense to cover any one of the
named units from the largest to the smallest. The word ''communi-
ty" has been reserved for this general purpose, and one may speak
with equal propriety of the animal communities of the Amazon
rain-forest or of a decaying tree within that forest.

In the border-hne field where general sociology meets and over-
laps general physiology and ecology, the field which is being con-
sidered in the present discussion, it seems desirable to have a term
which may be applied loosely, but not incorrectly, to any of the
recognized units lying below the groups accepted as definitely social,
just as the term "community" is applied by the animal ecologists
with equal propriety to strata, super-society, society, association,
and what not. It is in this general sense, for this level of social or
subsocial hfe, that I propose to use the term "aggregation." I am
not concerned with defining it closely in terms of the association or
society of Deegener or Alverdes. It may be used with equal pro-
priety in speaking of a group of frogs collected as a result of sexual
attraction during the breeding season ; or of a concentration of May
flies about a fight, where they have been collected by forced move-
ment as a result of their strongly positive phototropism. There is in
the term itself a strong suggestion that the groupings involved are
not closely integrated, which is in keeping with the facts in the field
to be covered.

INSTINCTS

In the course of this discussion we shall have reason to refer to
"instinct," a term deservedly in disrepute among careful thinkers be-
cause of the slipshod way in which it has been used. Early students
of human sociology and recent zoological commentators on socio-
logical phenomena have sadly overworked the word by referring any
unanalyzed social behavior to the working-out of a social instinct.

lo ANIMAL AGGREGATIONS

That social instinct may be acting in given cases is not to be denied,
but there has been an increasing and wholesome tendency to depre-
cate the use of this term to cover ignorance.

"Instinct" is hard to define. The most satisfactory definition
known to the writer is that of Wheeler, who says (1913a): "An in-
stinct is a more or less complicated activity of an organism which is
acting (i) as a whole rather than as a part; (2) as a representative of
a species rather than as an individual; (3) without previous experi-
ence;' and (4) with an end or purpose of which it has no knowledge."

It is obvious to one who has observed the reactions of animals
that there are two types of behavior: the learned and the unlearned.
Much of the latter is frequently called "instinctive," with propriety,
though in the case of many highly organized animals, including man,
there has been an unfortunate tendency to regard, as instinctive or
unlearned, behavior that is in reality based on very early training
which has been entirely forgotten or overlooked.

In man breathing, swallowing, gland secretion, and muscle con-
traction are all unlearned; and some of these, for example the secre-
tion of certain glands, cannot be effected by learning. These un-
learned reflex actions of parts of organisms seem to be the simplest of
a series of unlearned responses whose other categories are those re-
flexes of an entire organism commonly called "tropisms," and the
more complex behavior usually called "instinctive."

It is becoming increasingly difficult to draw hard and fast lines
between instincts and tropisms, or between either of these and the
general functioning of Hving cells. It is further impossible to dis-
sociate any of these three categories of behavior from the activities
concerned with growth and development. If one considers in this
connection the metamorphosis of a larva into an adult, which is
usually regarded as the function of growth and development, one
finds the processes concerned so inextricably bound with major and
minor activities of the animal that the instinctive behavior cannot
be clearly separated from the other processes going on at this time.
Is the production of the silk cocoon of the moth an instinctive action,
while the production of the thickened hypodermis to form the chrys-

' Or without modification caused by experience (W. C. A.).

THE GENERAL BACKGROUND . n

alis of the butterfly is only a growth process? What is the essential
difference between the two?

In so far as is possible, we shall avoid dwelling upon the aspects of
behavior usually called "instinctive," except in reference to the
literature. This is not due to a disbelief in the reality of instinctive
social behavior, but rather to a conviction that progress Hes in a
field where the elements of behavior can be more ^actly ascertained.

The drive which leads an animal to exhibit such behavior as is
usually classified as being due to the operation of social instinct I
prefer to regard, as does Wheeler (1928), as an expression of appe-
tite. Wheeler says in this connection: "It thus takes its place with
the other appetites like hunger and sex, though it is feebler and more
continuous, i.e., less spasmodic and, therefore, less obvious. It is
most strikingly displayed, however, in the restless behavior of the
higher social animal when isolated from the continuous, customary
stimuli of its kind." From this approach, the strength of the social
appetite can become a subject for objective investigation, such as
Warner (1928a) has recently made for the relative strength of the
drives furnished by food or sex hunger; but such an objective
investigation of the general social appetite has not yet been con-
ducted.

The scope of the discussion, some concepts, and a part of the ter-
minology having now been considered, we may plunge directly into
the mass of material awaiting analysis.

CHAPTER II
CLASSIFICATION OF ANIMAL AGGREGATIONS

It has long been known that animals not naturally bound together
in organic union may aggregate into groups or clusters more or less
closely associated, in which physical contact may or may not occur.
Actual physical contact is normally found as part of the aggregation
phenomenon among many Protozoa, as, for example, in Paramecium;
in fiatworms, such as the planarians; in earthworms; in echinoderms,
such as starfish; in mollusks; in arthropods; and among many
chordates, including ascidians, fish, frogs, reptiles, birds, and mam-
mals.

Among other animals similarly widely distributed through the
animal kingdom, collections occur in which physical contact is not
the rule. These may be illustrated by the jellyfish, ctenophores, or
copepods that may discolor the ocean for miles; by collections of
leeches, snails, or ostracods; by the swarms of gnats that dance to-
gether like particles in brownian movement; by ants, bees, schools
of fish, flocks of birds, herds of ungulates, and groups of various other
mammals, including man. The highest development of aggregations
not based on physical contact requires the possession of highly de-
veloped sense organs.

These two t3^es of animal aggregations are not mutually exclu-
sive, even when reactions associated with copulation are disregard-
ed; for animals may be involved in first one and then the other in
different phases of their life-cycle or seasonal history. With many
birds the loose flock of the daytime may be replaced by close physi-
cal contact during the night roost. At times this may be due to the
lack of adequate perching space, and show merely toleration of close
proximity; but in other instances, as, for example, the Indian tree
swift, there is a positive movement together even in the presence of
abundant roosting space. Bats may show the same phenomenon
during their daytime sleep.

CLASSIFICATION OF ANIMAL AGGREGATIONS 13

There are abundant examples of animals that lead wholly or par-
tially solitary lives during part of their seasonal- or life-cycle but at
another period come together into flocks or in actual physical con-
tact. This is true of the cowbirds, reared singly from eggs surrepti-
tiously laid singly in the nests of other species of birds. The young
cowbirds develop quite out of touch with other members of their own
kind and yet collect into definite flocks when adult. Another aspect
of the same kind of behavior is shown by the grackles, which nest
fairly separately but join in large flocks before the fall migration; by
deer, which summer separately or in partial family groups but winter
in herds; by frogs, which remain practically sohtary during the year
except for possible hibernation groups and then aggregate during
the breeding season; by solitary bees or wasps, which for the greater
part of the year are out of physical contact with their fellows and
yet during the summer may form overnight aggregations in closest
physical proximity; or, to give one more of many possible examples,
by land isopods, which congregate into dense bunches when their
habitat becomes dry.

The aggregations of the physical-contact type are, of necessity,
transitory in character in motile organisms; but in sessile animals,
such as the ascidians, or the marine mussel Mytilus, this may well be
the normal way of living. The physical-contact type of aggregation
finds its most complete expression among the sessile colonial organ-
isms that grow in dense stands of many individuals, which are physi-
cally connected with each other throughout Hfe. Obelia hydroids
represent this growth form.

Collections without physical contact, such as the flock or the
herd, may be constant and normal for some species; and the animals
in these are usually said to exhibit the social habit. This social habit
finds its best development in the insects, such as the ants and ter-
mites, among whom division of labor is carried out to its logical end,
in that polymorphic forms have evolved of which some do not com-
plete their sexual development while others specialize upon repro-
duction. These have been well described by Wheeler and Forel.

Animal aggregations may be classified on many other bases be-
side that of the degree of physical contact. Deegener (1918) has

14 ANIMAL AGGREGATIONS

made an exhaustive classification of the different forms of animal
groupings {V ergesellschaftung) in which he undertakes to arrange
logically all such associations, ranging from the relatively simple
colonies of the protozoans — Synura or Carchesium, where all the
individuals are similar and all arise from the same parent-cell and
are organically connected with each other — to colonies of ants with
their complicated social structure, which may include, in addition
to the ant castes themselves, their slaves, their commensals, their
tolerated guests, parasites, parasites of the parasites, or parasites
of other associated forms.

A summary of the classification of animal aggregations as worked
out by Deegener is given here at some length, not because I accept it
entirely with all its implications, but because it is the most complete
classification yet produced and because I am in hearty accord
with the principle underlying this scheme of organization: that no
hard and fast line can be drawn between well-integrated social or-
ganizations and loosely integrated aggregations which are usually
regarded as being definitely non-social. Further, experience with pre-
senting this material to seminar students has shown the desirability
of wading through a detailed outline, such as that of Deegener's, in
order to acquire a comprehensive view at one and the same time of
the ramifications of the subject matter and of its inherent unity.

It is the custom at present to ignore this work of Deegener or to
fail to appreciate its essential value (Wheeler, 1928) because of ob-
vious defects in its cumbersome terminology, in the criteria used to
distinguish between major groupings, and because the categories are
not clean cut and mutually exclusive. Many of these faults are in-
herent in a pioneering classification of subject matter in any field,
and others were caused by the lack of definite knowledge in 1918 of
the relationships involved. On this latter count we are in a position
to make improvements on Deegener's classification at the present
time, but we do not appear to be able to refine it sufficiently as yet
to repay the trouble involved.

The account given below is not a direct translation of Deegener's
1918 outline; but it follows that outline and gives his point of view,
criticisms of which have been suggested and will later be elaborated.

CLASSIFICATION OF ANIMAL AGGREGATIONS 15

deegener's classification of aggregations

Part I. Accidental unions or associations are groups of animals
without mutual benefit for individual members. "Accidental" is, to
Deegener's mind, a better term for these aggregations than "in-
different," because to him it plainly indicates the method of their
formation, and also because the members of accidental aggregations
are not always indifferent to each other. Accidental aggregations
will be seen to be of various kinds, formed in various ways. They
may consist of one or of a number of species. One cannot always be
sure concerning the proper classification of a given association, which
may as yet be merely a matter of opinion. Deegener recognizes that
even the major distinctions are not always clean cut and that one of
a pair of apparently closely similar groupings may be assigned to the
accidental associations while the other is called an "essential so-
ciety." In the minor categories the methods of formation determine
the classification to a considerable degree.

A. Eomotypical associations consist of members of the same spe-
cies which have arisen either sexually or asexually, which may have
remained together because they are the oft'spring of the same parent,
or which may have become accidentally associated together although
of different parentage. The former are called "primary," and the
latter "secondary," associations.

Alpha. Kormogene associations^ are confined to invertebrates and
do not occur in arthropods, echinoderms, and mollusks. They are
those colonial forms in which the different individuals remain mor-
phologically attached to each other. The advantages of the colony
are not always clear. In Protozoa, relationships of individuals in the
colony are not such as to guarantee nourishment for the entire
colony; thus there is no advantage in this respect with this phylum.
In the hydroid colonies, nourishment is better assured for the in-
dividual by the colonial form. The colony does not appear to be
formed necessarily because it is a more favorable adaptation to living
conditions but because of the failure of the different elements to
separate at fission. The tendency toward colony-building increases

' Budded colonial forms, as among the hydroids, cannot be regarded as "accidental"
in the usual usage of that word.

1 6 ANIMAL AGGREGATIONS

as habits become sedentary, and is also more marked in relatively sim-
ple animals having strongly developed skeletal parts, as the sponges,
hydroids, bryozoans, and tunicates.

I. Primary colonies arise as the result of division in which the
smaller pieces remain together, or as a result of budding in similar
fashion.

1. Homomorphic colonies result when the divisions are equal and
all members of the colony are similar, as in Synura, Carchesium, and
Salpa chains. Such colonies as Zodthamniiim may represent true
societies, since all individuals may contract if one is stimulated, and
so all may escape harm; while Carchesium does not, and so is placed
in the present category.

2. Heter amorphic colonies are formed when the divisions are un-
equal, as is the case with the strobila of the Scyphozoa, or during the
processes of asexual reproduction of certain worms, such as Autolytus.

II. Secondary colonies, or concrescence colonies, arise by the sec-
ondary union of individuals which are entirely separate for at least a
brief period.

1. Concrescence colonies having a genetic basis, in that the individ-
uals composing the colonies originated from the same mother, are
shown in Proteriodendron, Dinobryon, and secondary Salpa chains.
The fact that identical or related forms have survived and can live as
separate individuals indicates that these animals are able to live
without the small and perhaps accidental benefit arising from their
communal life.

2. Concrescence colonies without a genetic basis are those in which
the animals that later become attached together in one colony are
not descendants of the same mother. These commonly occur in ses-
sile animals, such as the ascidians, sea anemones, sponges, oysters,
and Mytilus. If no organic union takes place, causing a real fusion
between the different animals composing the colony, then the asso-
ciation remains accidental.

Beta. Associations of free individuals.

I. Primary associations arise through asexual or sexual reproduc-
tion when individuals descending from the same parent or parents
remain near the place of origin and form an aggregation which varies

CLASSIFICATION OF ANIMAL AGGREGATIONS 17

from a loose to a firm integration. The primary cause of their being
together hes in their common origin, but the cause of their remaining
together is not of a genetic nature but may depend on the favorable
character of the place or on the presence of food. In other cases one
must assume the operation of a social instinct which holds the ani-
mals together.

1. Syngenia are primary associations which arise by means of
asexual reproduction. This may be illustrated by Stcnior coeruleus,
which Hves on decayed water plants and occurs frequently in such
abundance as to give a blue color to the surface of the water. The
aggregation is located in space by favorable food conditions. So long
as there are only offspring from a single mother present, the aggre-
gation would be called a monosyngenium; but when second and third
generations appear from the same stem-mother, the group becomes a
polysyngenium. Other unrelated individuals may wander into this
favorable niche, forming a secondary association. Similar relations
hold with Vorticella, but with both these aggregations there may be
some social value accruing to the different individuals, since the
combined vortex action of the cilia brings more food to each animal.
This does not occur in hydroids, such as the common fresh-water
Hydra, which reproduces asexually and remains in a purely acciden-
tal aggregation in which there is no reciprocal relationship before
sexual reproduction begins. Similar relations hold with various other
simple coelenterates whose slight powers of locomotion tend to con-
fine them close to the place in which they are budded free, providing
it is a generally favorable location.

2. Primary associations arising from sexual reproduction may form
close unions which may rise to the widely extending reciprocity of
the highest types of society found among animals. In the inverte-
brates these are represented by the conditions obtaining in ant and
termite colonies; in the vertebrates, by human societies. This part
of Deegener's outHne undertakes to consider only the more primi-
tive, purely accidental forms of this family union, in which the par-
ents need not necessarily be concerned. Various combinations of sim-
ple families where the young all originate from the stem-mother may
be distinguished and divided as follows:

i8 ANIMAL AGGREGATIONS

a) Sympaedium, in which the offspring of the same mother form
the aggregation without the presence of either parent. This concU-
tion is seen in some spiders and insects, where the young of the same
mother remain together for a longer or shorter period. If the mother
remains with the offspring, the group belongs to another category.
Lophyrus caterpillars, which feed on pine needles, form an aggrega-
tion due in the first place to the eggs being laid together. No obvious
benefit accrues to the individuals. They are more conspicuous as a
result of the grouping and cannot defend themselves better than if
alone. The causal factors in such an aggregation are obscure. The
fact that the eggs are laid together is not sufficient in itself, since
other forms have their eggs laid similarly close together and yet
separate immediately on hatching. It may be that the sluggishness
of the animals and the lack of disrupting stimuli explain a large part
of the behavior; while, on the other hand, there may be a social
appetite which holds the groups together. The problem becomes
more difficult with those larvae which remain together during the
early larval hfe and separate when partly grown.

Many lepidopterous larvae that remain together during part
or all of their larval life spin a common nest. The formation of
such a nest may be due to the fact of living together rather than the
living together being due to the need or use of a common nest. The
abihty to spin a common nest does not guarantee the actual building
of one, for many spinning animals live alone. These larval colonies
are common among animals in which the adults are winged, and
hence are readily distributed during that phase of their life-history.
Such a sympaedium occurs in sohtary bees which lay eggs in cells.
The resulting larvae and pupae form an accidental association, living
together as offspring of a common mother. When adult, they fly
away separately.

h) A gynopaediiim is composed of a mother and her offspring that
remain together for a period. This grouping is not concerned with
the relationship between mother and offspring beyond the fact that
they remain together without obvious benefits accruing to the group
from the association. The aphid stem-mother in the spring gives

CLASSIFICATION OF ANIMAL AGGREGATIONS 19

birth to young parthenogenetically. This gynopaedium, consisting
of one female and her immediate offspring, may be designated a
monogyno paedium. The young also reproduce parthenogenetically,
and such a complex group may be called a polygyno paedium. These
colonies are homomorphic; but as winged forms appear, heteromor-
phic colonies are formed. In the autumn sexual generations appear
and produce a resistant over-wintering egg, which carries the colony
over the winter season. In this aggregation there are no benefits im-
mediately apparent. The brood is not cared for by the older mem-
bers or by each other. The individuals composing a crowd of aphids
are more easily cared for by ants of the myrmecocolous species when
together, but also are more easily preyed upon by their numerous
enemies. The massed aphids also tend to destroy the food plant on
which they cluster, to their own disadvantage. Deegener recognizes
no social advantage, and therefore regards the aggregation as ac-
cidental.

c) Patrogynopaedia occur when both parents remain with their
offspring in groups. Those with no social benefits for their members
belong here, but this type of aggregation often carries with it some
social advantage, and so usually belongs in a later category. Necroph-
orus beetles live with their young in decaying animal bodies. This
association may confer social benefits under certain conditions, but
they are not recognizable in all cases. In these scavenger beetles, the
presence of a dead body seems to release a digging reaction whether
the individual is solitary or in company with others. Each individual
digs without reference to the others. The results may have no sig-
nificance for the assisting beetles, but only for the pair leaving their
eggs with the dead body. Obviously the whole has racial significance,
although without significance for many of the participating in-
dividuals.

Combination family groups also occur in which the individuals
composing the aggregation come from more than one stem-mother.

d) Synchoropaedia are formed when eggs laid by different females
in a favorable place hatch out and the larvae remain together from
the very first, not as separate families, but freely mixed into a com-

20 ANIMAL AGGREGATIONS

mon aggregation. Mosquito {Culex) larvae in a rain barrel are an
example of a synchoropaedium. When larvae of different species are
present in the same rain barrel, we have a heterosynchoropaedium.

e) Similarly, symphagopaedia may result from several groups of
the same species laying eggs on the same food material except that
here the favorable food rather than the favorable place becomes the
integrating factor. This type of aggregation may be illustrated by
flesh flies and, according to Deegener, by Drosophila.

II. Secondary associations may be distinguished from primary as-
sociations because they are the result of a coming-together of free
individuals rather than their merely remaining together. The classi-
fication is based on the integrating factor judged to be most impor-
tant.

1. Sysyngenia arise from the secondary fusion of two or more
syngenia.

2. Sysympaedia consist of fused "children-famihes" and arise when
one sympaedium meets with another. Deegener observed such in
juvenile spiders of Epeira (1919&). The members of both groups
mixed peaceably and gave no sign in their conduct that they were
influenced by the foreign spiders; indeed, they did not seem to notice
that their membership had been doubled, and new and old alike ag-
gregated into one close mass. Another sympaedium was added to
these two with similar results, although it was not ascertained
whether or not the individuals of a given sympaedium remained for
the most part together.

Two sympaedia of caterpillars of Malacosoma castrense L. are not
mixable when the larvae of one sympaedium are in the molting
period; otherwise they mix without the caterpillars of the two broods
appearing to sense the change in their association. Schulz (1926),
in studying the reaction of caterpillars of Vanessa io L., V. urticae L.,
and Araschnia levana L., found, with the methods he used, no recog-
nizable value to rest in the aggregations other than the satisfaction
of a social instinct; and this value had lost much of its meaning,
since the caterpillars are able to live if isolated, under which condi-
ditions they spin small coverings in place of the usual communal
nests. They will again take up communal life after an experimental

CLASSIFICATION OF ANIMAL AGGREGATIONS 21

isolation of four days. Marked sympaedia, some of which differed
from each other in size of individuals, fused to a single sysympae-
dium. When this divided late, the resultant groupings usually con-
tained members derived from different original sympaedia.

3. Sympolyandria are accidental polyandric associations formed
on a sjTichoric basis, as that of Alcippe, a barnacle which dwells on
the deserted snail shells occupied by hermit crabs, forming an ac-
cidental hetero typical association; but the barnacles, considered
alone, form a sympolyandria. Polyandria form a type of essential
mating society to be discussed later in this outline.

4. Synchoria are locality aggregations formed primarily because of
a limited expanse of particularly favorable locations for living. Bar-
nacles gathered together on available rocks are a good example.

5. Syncheimadia are hibernating aggregations, such as those of
snakes or salamanders.

6. Synhesia are swarming aggregations under the influence of the
breeding season, as illustrated by palolo worms. Factors concerned
here include the simultaneous ripening of the sex cells, a limited
favorable area, and the correct external conditions,' which are fre-
quently associated with lunar rhythms. Similarly, the swarms of
May flies are due at least in part to simultaneous pupation rather
than to sex attraction.

7. Symphagia are aggregations about a favorable food supply, as
flies collect about carrion or sugar. Here there is no obvious benefit
from the association.

8. Symporia are migration aggregations joined either because they
originated in the same place or because they are going in the same
direction, and may be illustrated by the migrating masses of fiddler
crabs, of butterflies, or of salmon.

9. Symphotia^ occur when the aggregations collect about a source
of fight. Such a reaction is given by a great many insects, as well as
by other animals (Mast, 191 1).

' Considerations given later, particularly in chapters xvi and xvii, indicate that such
swarms have a rather obvious survival value and hence should not be placed among the
"accidental" groupings.

^ If this type of category be included, it is necessary to include similar headings for
tropistic collections due to the reaction to other environmental factors, such as heat,

22 ANIMAL AGGREGATIONS

lo. Synaporia are collections due to unfavorable conditions, as
when beetles are collected by the wind and deposited in beetle drifts
in the same way that snow is drifted.

Krizenecky (1923) recognizes two different types of synaporia,
the passive and the active. The first are formed when the animals
are passively carried together, as by wind or wave action. The latter
are formed when animals faced with unusual disturbing con-
ditions collect together. Such aggregations may be noted in the
worm Enchytraeis humicolor, which ordinarily hves singly in the
soil but which aggregates into symphagia about decaying food ma-
terial. If the worms are placed in a dish of water, they aggregate
into larger or smaller masses with the worms closely intertwined.
Such clumped masses do not remain together; but after the group
is closely formed, there comes a disintegrating movement which re-
sults in the animals finally coming to rest scattered singly over the
bottom of the dish. The animals remain thus scattered as long as the
water is undisturbed. When subjected to renewed stimulation by
adding chemicals or by mechanically disturbing the water, another
aggregation cycle is set up.

B. Heterotypical associations consist of collections of unlike species
which may occur for the reasons given above, and which may be
designated by adding the prefix hetero- to the proper term for the
homotypical aggregation, as: heterosymphagopaedium, heterosyncho-
rium, heterosyncheimadium, etc. Deegener recognizes also co-incuha-
tia, which are breeding aggregations of different species of birds, for
example, selecting a common, restricted nesting site. Finally he adds
symphoria, which are formed when one or more species of animals
settle upon another of different species, forming a heterotypical
aggregation without obvious mutualism or parasitism, and are weU
illustrated by the barnacles, hydroids, snails, bryozoans, and others
growing on the shell of an old horse-shoe crab {Limuliis poly pJiemiis) .

chemicals, touch, gravity, and the like. Rather, it seems preferable to replace this cate-
gory by some such term as syntropia, meaning those collections which are brought about
by tropistic reactions to some environmental factor. Such collections occur, due to a
combination of elements, including that of a limited space into which tropistic reactions
lead animals to assemble and the incidental presence of numerous individuals in the
region at one and the same time. In all these collections there is this time factor
working; otherwise we could not recognize them as aggregations.

CLASSIFICATION OF ANIMAL AGGREGATIONS 23

Some of the overlapping inherent in this type of subject-matter
classification appears when one considers a heterosynaporium. a col-
lection of different species due to the action of unfavorable condi-
tions, which Deegener illustrates by the growing concentration of
water animals in a drying pond. Obviously, such a collection would
be at the same time a heterosymphagium and a heterosynchorium.
Apparently, Deegener would classify the animal community of
modern ecology as a heterosynchorium, since it is composed of sev-
eral species occupying the same place, although the individuals of the
group are not of obvious advantage to each other. He does not ac-
tually say that an ecological community should be so classified; he
does use the term hiococnosis in connection with his discussion of a
coral reef heterosynchorium.

Part II. Essential aggregations or societies are communities of spe-
cies of similar or dissimilar animals which have a real value for the
individuals composing them, thereby differing from the "associa-
tions" treated in the previous sections.

A. Homotypical societies are composed of the same species.

Alpha. Kormogene societies have the different individuals compos-
ing them organically connected with each other.

I. Primary colonies have arisen from the same mother.

I. Reciprocal colonies are those in which all the individuals repre-
sented stand in reciprocal relationship to each other.

a) Homomorphic colonies have all the individuals morphologically
similar and may be found among sponges and at certain times among
hydroids and bryozoans.

(i) Colonies formed by division may be illustrated by the colonies
of Volvox so long as they remain free from specialized reproductive
cells.

(2) Colonies formed by budding occur commonly among the Hy-
drozoa, the Bryozoa, and in many colonial chordates.

b) Heteromorpliic and polymorphic colonies are formed when there
is a differentiation between the different members of the colony, as
occurs in the hydroid Hydractinia, in which feeding, reproductive,
and protective zooids may be recognized. Polymorphism is carried
much farther in the Portuguese man-of-war, Physalia, and its aUies.
Here again we may recognize (i) colonies formed by division, as in

24 ANIMAL AGGREGATIONS

Volvox, when reproductive cells appear, and (2) colonies formed by
budding, as in the hydroids.

2. Irreciprocal colonies must be recognized in which all members
do not contribute equally to the welfare of the whole. This is simply
illustrated by the case of a budding fresh-water Hydra, where the
new individual, the developing bud, has a parasitic relationship with
the mother.

II. Secondary colonies develop by concrescence, as when the young
fresh-water sponges developing from different gemmules coalesce,
due to their proximity, and form one sponge body originating from
several gemmules.

Beta. Societies of free individuals may be classified as follows:

I. Societies based on a sexual or genetic foundation.

I. Primary societies: families in which the young are descended
from a common father or a common mother or from common par-
ents, and which remain together from the very first.

a) Reciprocal families in which all members benefit from the social
connection.

(i) Sympaedia are composed of young of the same brood, but
without either of the parents present. Such societies may be homo-
morphic, as in the case of minnows or young birds, or heteromorphic,
as in bee colonies after the queen's swarm has departed.

(2) Gynopaedia are composed of the mother and her immediate
offspring, which may again be divided between homomorphic and
heteromorphic groups. The former is represented by the mole crick-
ets {Gryllotalpa), the earwigs {Forficula), and many birds and mam-
mals; the latter group, by colonies of bees or ants.

(3) Patrogynopaedia consist of a male and a female and their off-
spring, and may be divided into monomorphic, dinwrpliic, stud. poly-
morphic societies. Monogamous monomorphic societies of this sort
are common among birds where both parents remain with the young.

Polygamous monomorphic families are similarly common among
many large animals, although monogamous families occur there, too,
as among foxes. In dimorphic patrogynopaedia the oftspring living
with the parents are true larvae, as, for example, in the passalid
beetles. The best example of a polymorphic colony of this type is

CLASSIFICATION OF ANIMAL AGGREGATIONS 25

given by the termites, where sexually mature males and females of
one or more grades occur in the same nest with soldiers and
workers.

(4) In a patropacdium the male remains with his offspring for
some time. Schulz (1926), in his analysis of the situation in the
brooding stickleback fish {Gastcrosteus aculeatits and G. pungitius),
concludes that the value to the male is in the psychological realm, and
quotes Deegener with approval as saying that the nest and young are
of lively interest to the male stickleback, their loss is a misfortune,
and the nesting and brooding phenomena are a source of inner peace.
Obviously, such assertions are not susceptible of demonstration. To
the eggs there is the benefit of added certainty of fertilization, of
protection from other fishes, of aeration, with resulting protection
from fungus growth; while the young find a favorable place for
development, passive protection by the nest, and active protection
by the guarding male. The relation between eggs and young and the
brooding male is essential rather than accidental, and therefore
forms a true society. It is reciprocal, and the female is not concerned
after the eggs are laid; therefore a patropacdium, which had its
origin in a polygamous connubium existing merely as a mating rela-
tionship, but this connubium is an association rather than a society.
If the male dies, the society becomes a simple sympaedium, which
would be accidental in nature, since the association of the young has
no value for them. The relation of the young to the nest has a syn-
chorium factor. The existence of the patropacdium is necessary for
the well-being of the eggs but not of the young fishes. The relations
between the males of the large- and small-mouthed black bass, the
bullheads, and the fresh- water dogfish {Amia calva) and their nests
and young give an opportunity for similar analyses.

b) Irreciprocal families are those in which the social values rest
only with the young.

(i) Gynopaedia of this sort are to be found in the leeches (Glossi-
phonia), according to Deegener; but Schulz detected evidence which
led him to conclude that the female leech is somewhat interested in
her eggs, and on this account he places these leech gynopaedia among
the reciprocal societies. Similarly, careful observation might show

26 ANIMAL AGGREGATIONS

the same sort of value, if such it can correctly be considered, in the
other cases cited by Deegener, such as the amphibians, Hylodes
lineatus and Pipa pipa.

(2) Patropaedia of this sort are thought by Deegener to be illus-
trated by the relations in the obstetric toad Alytes, in which the
male carries the strings of eggs twisted about his legs, and in Rhino-
derma darwini, a small cricket-like frog of the moist beech forests of
Chile. The male of the latter species takes the fertilized eggs and
crams them into his singing pouch, which becomes greatly enlarged
during the breeding season. Here they develop and transform, hop-
ping forth from their father's mouth as fully developed small frogs
(Barbour, 1926).

2. Secondary societies are those in which the individuals are not
together from the very beginning, or at least those in which the
primary social group becomes modified by secondary additions.

a) Sexual societies of the Protozoa are such as are shown in ciliate
conjugation.

b) Connubium simplex of the Metazoa is a grouping in which mat-
ing occurs between animals of the same species but of different sexes,
or between hermaphroditic animals.

(i) Polygamy includes polygyny, or the mating of one male with
more than one female, as in polygynous birds, such as the domestic
fowl, and in many mammals; and polyandry, in which several males
mate with a single female without the female being free to all males.
Among Deegener's examples are the cases of double copulation in
insects. In the case of Alcippe, a barnacle, the females as a rule live
near each other, and from three to twelve dwarf males join each
female and remain with her during their lives. Alverdes (1927) states
that this sort of relationship is rare, but adds the case of Boncllia, a
worm of which more will be said in a later section, and with which as
many as eighteen males attach themselves to a given female and
remain so for extended periods. Polyandry has also been observed
among some spiders.

(2) Monogamy is fairly widespread, at least in the form of seasonal
pairings. It is found among beetles, as for example, the monogamous

CLASSIFICATION OF ANIMAL AGGREGATIONS 27

Passalidae, which remain with the larvae and the pupae. Alverdes
Hsts also cases of at least seasonal monogamous mating among
spiders, fishes, amphibians, reptiles, birds, and mammals.

(3) Communal connubium, or promiscuity, occurs among many
fishes at the spawning grounds, among certain lizards, and among
gregarious bats. It is also reported for the American bison, for the
American cowbird, and among various other birds (Alverdes).
Miller (1928) summarizes evidence that this is a common state
among anthropoid apes and certain monkeys; unlike most modern
sociologists, he believes this represents the original mating relation-
ship among Homo sapiens.

(4) A conconnubium is formed when monogamous animals collect
during the breeding period, forming small societies that continue
during copulation. Deegener gives as examples the viper (probably
Pelias) and birds, such as gulls, which move at mating time to a
restricted location and there form seasonal pairs.

c) Perversum simplex applies to those cases where males attempt
to mate with each other, as has been observed for drones of the honey
bee, after they are driven out of the nest in the autumn, and for
various other insects, including certain beetles and house flies.

d) Preconnubia occur when individuals of one sex collect at one
place before the mating season, or both sexes may be present, but
without mating. Such preconnubia occur among many frogs and
birds.

e) Synhesmia are swarming societies which collect under the in-
fluence of reproductive drives. Androsynhesmia, male swarms; gyno-
synhesmia, female swarms; and amphoterosynhesmia, or mixed
swarms, are known to occur.

II. Societies that are not immediately based on a sexual or genetic
basis are also known, as follows:

1. Sysympaedia are combinations of sympaedia, such as occur in
minnows.

2. Syngynopaedia consist of two gynopaedia which have united as
may happen with ants, or seals {Phoca gruenlandica) , or wild hog^
{Sus scrofa)

28 ANIMAL AGGREGATIONS

3. Sympatrogynopaedia are combinations of at least two patro-
gynopaedia, and are known in monkeys, marmots, elephants, ante-
lopes, and many other mammals.

4. Adoption societies are those in which a female takes offspring
from the same species. They are known for birds and mammals, for
example among the wild hogs (Stis scrofa).

5. 5}'waw(/r/a are groups of males which herd together. Thus, male
birds of several species are known to have this habit; and it is re-
ported to be common also among mammals, as in seals and ante-
lopes.

6. Syngynia are similar groups of females, such as are formed by
the stickleback fishes.

7. Symphagia, again, are feeding societies formed of several in-
dividuals, and illustrated by Necrophorus beetles during a portion of
their life.

8. SyncJwria are societies united around a common place which
has some peculiarly favorable quality or qualities. They are well
illustrated by the common bird roosts, as of crows and robins, and,
among insects, as wasps and Mellisodes bees. (See chap, iv.)

9. Syncheimadia are combined over-wintering societies, and may
be illustrated by solitary bees and coccinellid beetles.

10. Symporia, again, are migration societies, such as swarms of
bees or flocks of migrating birds or mammals.

11. Synepileia are marauding societies or hunting bands, such as
those of jackals and wolves.'

12. Sympaigma are groups of individuals brought together in or-
der that they may engage in common play. Deegener cites the whirl-
igig beetles (Gyrinus) as examples. Schulz (1926) has investigated
this aggregation somewhat and concludes that play is not the prin-
cipal integrating factor; he believes that the greater security fur-
nished is the more important cause. Therefore he places them in the
next category. Brown and Hatch (1929) think that the collection of
gyrinid beetles is an example of a reaction to a general environ-

' The American wolf pack apparently is usually a family affair, but may not always
be so (Seton, 1929).

CLASSIFICATION OF ANIMAL AGGREGATIONS 29

mental pattern which they regard as more important than the bio-
logical values involved.

13. Symphylacia are societies that furnish protection for the in-
dividuals composing them.

B. Heterotypical societies are composed of individuals of different
species.

Alpha. Reciprocal societies.

I. Integrated by sexual drives.

1. Connubium confusa are societies of both sexes, but of different
species, brought together Tor the breeding season. Thus, male frogs
will attempt to mate -v^/ith females of other species, or with toads,
or even with fish. Or another taxonomic level, coccinellid beetles of
different species ha/e been observed to attempt copulation.

2. Perversum confusa are formed when individuals of the same
sex congregate during the breeding season, although of different
species, as for example, male frogs and toads, Rhagonycha melanura
Oliv. with Luciola luistanica Charp.

II. N on-sexual combinations .

1. Phagophilia are heterotypical reciprocal societies wherein each
species benelits, although at least one of the two receives its food
through its association with the other. Thus a passive species is
freed of its parasites through the efforts of its active associates,
showing one variety of mutuahsm. This is illustrated by cow-
birds following cattle and feeding on the flies which infest the
latter.

2. Synsitia are also symbiotic societies in which one of the asso-
ciates lives on the shell or the outer covering of the other, without
being parasitic and without the type of relationship found in a
phagophilium. Deegener regards the relationship between a hydro-
zoan and a hermit crab, such as Hydractinia growing on the shell
occupied by Eupagurus, as a synsitium. The former clearly receives
transportation and fragments of food, while the latter may be
protected by the nematocysts of the dactylozoids, as Deegener
suggests.

3. Phylacobia occur when two species live together in the same
cavities, as Campanotus punclulatus termitarius Em., an ant, is said

30 ANIMAL AGGREGATIONS

to live (Wasmann, 1901-2) in the runways made by various ter-
mites, receiving shelter and giving increased protection. Wheeler
(1913a), Emerson, and other students of social insects are agreed that
cases of reputed association in compound nests are in need of further
careful investigation. Wasmann calls this relationship phylacobiosis.

4. Trophobobia exist when one species feeds upon the excretions
of or the waste of the other, and in turn provides protection for the
weaker species. This relationship is found between certain species of
ants and aphids.

5. Symphilia are formed when one spfcv':ies receives food, protec-
tion^ and shelter from another, and in turn supplies excretions which
are apparently narcotic in nature. This relationship exists between
many ants and their myrmecocoles and between termites and ter-
miticoles.

6. Dulobia are illustrated by the slave-making ants which raid
other colonies and carry off the young, which in time take over the
routine work of the colony into which they are carried, receiving in
return the advantages of being members of the given society.

7. Adoption societies are formed by mutual adoption freely entered
into by both species, and without recognizable advantages or notice-
able harm for either. The ants, Formica consocians and F. incerta,
are said to form such societies.

8. Heterosymphylacia , as in the homotypical symphylacia, furnish
increased protection for all individuals as a result of the social union.
Thus zebras and ostriches, or giraffes and elephants, are reported to
live together, thereby increasing the security of both constituent
species.

9. A heterosynepileium occurs when more than two species of ani-
mals join forces and gain greater hunting efficiency for the group.
Different species of storks over-wintering in East Africa have been
observed to form common hunting bands and to conduct more or
less organized drives for concentrating scattered grasshoppers.

10. Confoederata are recognized by Deegener as being societies of
unlike species united by mutual friendship or sympathy, and as
having no other basis. Crows and jackdaws, alone or with starhngs,
golden-crowned kinglets and titmice, common creepers and wood-

CLASSIFICATION OF ANIMAL AGGREGATIONS 31

peckers, are given as examples. Obviously, such a category is with-
out secure foundation, but is perhaps to be expected from a worker
who beheves that the future belongs, however the present resists, to
the psychic and not to the mechanistic (Deegener, 19206).

II. Heterosymporia are mixed migration societies, such as occur in
birds and mammals, and are especially well marked on the plains of
South Africa.

Beta. Irreciprocal societies occur when the benefits extend mainly
to one species, while the other may be decidedly harmed from the
association.

1. Synclopia, or thieving societies, are those in which one species
feeds upon the stored food supplies of another, as thieving ants are
known to prey upon stored termite food, or as thieving species of
termites take the food of other termites. Wheeler calls this clepto-
hiosis; Forel designates it as lestohiosis.

2. Syllestia are societies containing robber guests which prey upon
the eggs or the young of the species with which they are associated.
Thus staphylinid beetles may prey upon the brood of the ant colonies
whose nests they inhabit, as Wheeler's "synechthren." In somewhat
similar relations are the hawks that prey upon flocks of migrating
birds. The flocks of wandering grasshoppers, springbok, and the like
are each set upon by its own particular set of predators which ac-
companies the food flock on its migrations.

3. Paraphagia are societies composed of harmless companions of
their host feeding commensually on fragments neglected by the
host. Alcippe, a boring barnacle, inhabits the snail shells which have
been appropriated by hermit crabs, and feeds on fragments escaping
from the feeding of the latter. Dermestid beetles occupy nests of
other insects, feeding on waste material such as molted skins. The
so-called synoektes of ants form paraphagia with the ants with which
they live.

4. Synoeciiim is the term given by Deegener to the association be-
tween certain animals and the nests of other animals. This is known
to be a widespread relationship. The crab Pinnixa hves in holes oc-
cupied by marine mollusks. Birds' nests have m.any animals regular-
ly living in them; sparrows may build in storks' nests. Fishes build in

32 ANIMAL AGGREGATIONS

the nests of other fishes (Reighard, 1920). Many similar examples
could be given for other nests, such as those of ants and termites.

5. Paroecia, or neighborly groups, are formed in which the less
conspicuous animal species finds protection from the other without
occupying a part of its nest. Thus, small fishes are frequently as-
sociated with medusae or with the Portuguese man-of-war Physalia;
while many animals, such as fish, worms, snails, and starfish, have
similar relationships with coral colonies.

6. Metrokoinia occurs in ants when the fertilized female of one
species who has lost the ability to start a new colony joins herself
with the fertilized female of another species that has retained this
power, and is thus associated with a colony development which she
would be unable to secure alone, and to which she contributes Uttle
or nothing. This relation has been described for Strangylognathus
testaceus Sch., which has lost the power of colony formation, living
in mixed colonies with Tetramorium caespitum L.

7. Irreciprocal sym porta occur when one animal species attaches
itself to the surface of another without becoming parasitic and with-
out contributing aid to the animal on whose back it grows. This
relationship may exist between barnacles growing on whales, be-
tween hydroids and crabs, and between stalked protozoans, such as
peritrichs and suctorians, and the snails, crustaceans, or hydroids
supporting them.

8. Syncollesia are cemented societies in which one animal cements
into its own covering the case or shell of another species of animal
without killing off the original owner. Small mussels {Sphaeridae)
and snails may be worked into the cases of caddis-fly larvae.

9. Parachorium is the name given to the relationship that exists
when one animal lives within the body of another without being
parasitic upon it. Hydroids, sea anemones, polychaete worms, ophi-
urids, and crustaceans live in the canal systems of sponges; and Pin-
notheres, a crab, lives in the mantle cavity of Mytihis, the sea mussel.

10. Parasitism is not easily separated from several of the preced-
ing categories. A parasite, in the restricted sense used here, obtains
its nourishment, at least, from the host with whose continued exist-
ence the parasite is more or less closely bound. Frequently the nour-

CLASSIFICATION OF ANIMAL AGGREGATIONS 33

ishment of the parasite comes from the li\ang substance of the host.
Many categories of even such restricted parasitism are recognized,
and may be found Hsted in reference works on the subject (Hegner,
Root, and Augustine, 1929).

We have given here an outline of Deegener's classification of ani-
mal groupings in detail, but it is not our intention to fit the different
aggregations to be discussed later into their appropriate niches in
this classification. In fact, certain of its more detailed aspects will
not be referred to again. But it is upon the idea that there is an es-
sential unity within the phenomena to be discussed that the present
summarizing account has been prepared; this concept, although
foreshadowed by Espinas, was first fully expressed in Deegener's out-
line. We shall return to it in the concluding chapters.

CLASSIFICATION OF ALVERDES

When we turn to the analysis of social phenomena by Alverdes, we
find, as suggested in the introduction, that the relations composing
the first part of Deegener's outline are omitted as without social
significance, since in them Alverdes cannot recognize the expression
of a social instinct and since the entire discussion of these so-called
associations is limited to a definition and slightly more than two
pages of text. This omits consideration of much of the material to
be presented in the present discussion, and Hmits markedly the
field of general sociology. Even under these sharper limitations, the
criterion of social life suggested by Alverdes, that of the possession
of a social instinct, must necessarily be vague and easily capable of
misinterpretation.

The material which Alverdes believes to form the subject matter
of general sociology is organized in the main about sexual relations,
in which he recognizes such categories as monogamy, polygyny,
father-families, mother-families, and other similar divisions which
were also found in Deegener's more inclusive outline. In addition,
he recognizes that animal societies may be closed or open. In the
former, new members are admitted only under special conditions, if
at all; insect states are such. Within a closed community there is
frequently an established hierarchy, as has been shown for birds

34 ANIMAL AGGREGATIONS

which, to be sure, are only partially closed communities (Schjel-
derup-Ebbe, 1922, 1923). In the open societies membership is much
less exclusive, and chance alone determines whether or not its mem-
bers shall unite or separate. The open societies may be organized,
like those of the saiga antelopes, which have guards and leaders; or
unorganized, as in many groupings of what Alverdes regards as,
strictly speaking, non-social insects, as when grasshoppers, butter-
flies, caterpillars, and the like unite in migrating swarms.

CLASSIFICATION OF ESPINAS AND WHEELER

Wheeler, in his discussion of animal societies (1930), gives a sum-
marized scheme of classification of social and subsocial groupings,
based upon the work of Espinas, which is reproduced herewith in a
somewhat modified form. The principal modifications made have
been the placing of all distinctions between homotypic and hetero-
typic groupings in the third and least important column, the re-
arranging of the categories under associations, and the substitution
of "anthropoid" for "human" in the last category. Wheeler does not
believe that the societies arose from associations, although he says
that the ancient aggregative or associative proclivities may have
been retained by many species and may serve to reinforce their
specifically social behavior. This subject will receive more detailed
attention in the last two chapters.

CLASSIFICATION ON BASIS OF INTEGRATION

It is illuminating to attempt a classification of social grouping on
the basis of the type or the degree of integration of the social group.
Some of the available knowledge on this point will be set forth later.
From many points of view this seems a most desirable basis of
classification, but there is not at present sufficient exact knowledge
to justify an elaborate attempt in this direction. When made, such a
classification would follow the general outlines suggested by Deege-
ner, at least to the extent that such a scheme would present the
social organization of animals from the loosely organized, apparently
chance aggregations due to collections around favorable locations or
on account of physical limitations which prevent separation, through

CLASSIFICATION OF ANIMAL AGGREGATIONS

35

a series of small quantitatively, rather than qualitatively, different
degrees of integration, up to the closely organized societies of ants
and termites and the more extensive group societies of man.

Simplified Schematic Aerangement of Types of Associations

AND Societies

(Modified from Wheeler, 1930)

1. Passively collected aggregations or [ Homotypic
agglomerations, e.g., wind collected

2. Actively collected aggregations or '
agglomerations, e.g., tropistically
collected

3. Food chain associations

a) Predatory

b) Parasitic

4. Commensal associations

5. Mimetic associations

6. Symbiotic or mutualistic associa-
tions

7. Communities (biocoenoses)

A. Associations:

Loosely integrated,
relatively unstable,
and temporary sys-
tems primarily de-
pendent on the reac-
tions of individuals
to environmental
stimuli

_ Heterotypic

Heterotypic

B. Societies:

More closely inte
grated, more sta
ble, and permanent
systems primarily
dependent on reac-
tions of individuals
to each other

I. Persons (multicellular)

Organically interconnected colonial ^
organisms forming closed societies
chiefly nutritive in function, e.g.,
sponges, colonial hydroids

Mainly reproductive societies closed,
e.g., subsocial insects and social in-
sects such as bees, ants, and termites

Mainly protective societies, closed
and open, e.g., flocks, herds, and
schools

Homotypic

5. Anthropoid societies; group societies

Homotypic or hetero-
typic; i.e., may be
pure or mixed colo-
nies of dominant
animals; dominants
may be accompanied
by social parasites or
by various other
sorts of associates

The outlines of such a scheme of classification can be sketched.
In doing so, its limitations in the present state of knowledge become
the more evident.

Alpha. Individuals organically connected.

I." Individuals with true organic union, as in the hydroid Ohelia.

II. Individuals only superficially connected, as in the mollusks

36 ANIMAL AGGREGATIONS

Mytilus or Ostraea.

Beta. Individuals not organically connected.

I. Aggregations primarily due to reactions to environment. Ani-
mals live at this level of group integration in a common habitat but
without marked organization into groups. This category would in-
clude the habitat communities of the ecologists. To some extent
the plants share with the animals in the organization of this com-
munity, usually, in fact, being the conspicuous factors in land com-
munities; hence the modern emphasis by ecologists upon the biota.
Further classification would depend on the physical or biotic factors
in the environment which dominate the habitat.

II. Aggregations primarily due to reactions to other organisms.
These are generally recognized to be more closely integrated than are
habitat communities, being bound together by biological relation-
ships as well as by those of habitat. There is no sharply defined line
to be drawn between the two.

In addition to the subdivisions based upon the method of inte-
gration, three fairly definite subdivisions can be recognized, based
on degree of integration.

1 . Relatively slightly integrated groups in which the primary (in-
dividual) reactions predominate, and whose survival value is ap-
parent only after experimentation. The aggregations of isopods,
OpJdoderma, and Proccrodes, to be discussed later, are examples.
Further classification would depend on method of formation of the
group and on the type of integration, as well as on the different sorts
of animals of which it is composed. Many of Deegener's groupings
could be taken over here and in the next two categories.

2. Moderately well-integrated groups in which the secondary
(group) reactions predominate although primary reactions are still
strongly in evidence. The survival value of the group is more ob-
vious. Schools of fish, flocks of birds, and the like would frequently
come under this category.

3. Highly integrated groups in which the primary reactions are
decidedly in the minority and the social value is strongly in evidence.
Here would be classified the different insect societies, together with

CLASSIFICATION OF ANIMAL AGGREGATIONS 37

the societies of man and those of the other vertebrates which ap-
proach these standard societies in their social organization.

The difficulties inherent in the further elaboration of this scheme
reveal at once the lack of natural divisions between the different
levels of organization with which we are dealing. It is apparent that
we must recognize that the whole field of interrelationships of organ-
isms must be taken as the content of general sociology; we can only
arbitrarily single out some particular level of social appetite, group
reaction, community integration, social value, or exhibition of divi-
sion of labor, as forming the beginning of social life.

CHAPTER III
FORMATION OF ANIMAL AGGREGATIONS

The method of formation of animal aggregations differs with the
degree of integration and with the different types of integrating
factors. The discussion to be given here is not necessarily exhaustive,
but the examples included may serve to illustrate the common meth-
ods and some of the problems involved.

One whole group of aggregations of individuals that are ordinarily
solitary is caused by tropistic responses to environmental stimuli.
Deegener recognizes one phase of this type of aggregation in his
grouping called "symphotium" which occurs when individuals col-
lect about a source of light. Aggregations of this general type may
be called "syntropia," as suggested earlier. The method of forma-
tion of such aggregations attracted much attention in the three dec-
ades and a half of J. Loeb's work in this field, from about 1888 to
1923. Loeb and his immediate followers were concerned chiefly with
aggregations which result from environmentally forced orientations
and movements.

FORCED MOVEMENTS

When exposed to certain stimuli, some animals react as if they
were automatons forced by the interaction between their own organi-
zation and their environment to move in a certain direction and to
aggregate when available space is limited. The term "tropism" was
at one time reserved for such reactions. These are well illustrated by
the response of the larvae of the annelid worm Arenicola to light.

These worms burrow as adults in the sandy tidal flats of the
Atlantic Coast south of Cape Cod. The eggs are deposited in large
numbers in a jelly-like mass which is attached at the opening of the
burrow. The eggs develop into free-swimming ciliated larvae having
two eye-spots symmetrically placed near the anterior end. Immedi-
ately after hatching, the larvae are strongly positive to light and
negative to gravity. Accordingly, they travel to the surface of the

38

FORMATION OF ANIMAL AGGREGATIONS 39

water, where they may collect in great aggregations unless scattered
by waves or by tidal currents.

These larvae swim in a long spiral path orienting quite accurately
to light. The orientation, Mast says (191 1), is not entirely accurate
but is subject to frequent muscular turnings which result in re-
orientations. The general course is toward the light, as shown by
the diagram (Fig. i). The following account of the details of this
reaction is taken from Mast's description (1911), since he has been
consistently critical of interpreting any animal reaction as approach-
ing automatonism.

"If the direction of the rays of light is changed after the larvae are
oriented, they all appear to turn directly toward the source of light
in its new position without preliminary trial movements." Ordinari-
ly, these larvae swim so rapidly that the exact details of their path
are hard to follow. When caught under a sloping cover slip so that
they can no longer swim spirally, if the larvae are caught lying on
one side no definite movement is seen except a slight forward mo-
tion; in those lying on either dorsal or ventral surface, the anterior
end is seen to move constantly from side to side with a slight jerky
motion, a movement undoubtedly due to muscular contractions. If
light is thrown on such an organism at right angles, the lateral move-
ment toward the illuminated side is at once increased, and the larva
turns in that direction. "By using two sources of light so situated that
the rays cross at right angles in the region where the specimen is lo-
cated, and then alternately intercepting the light from each of the two
sources, it can be seen clearly that the larva, by muscular move-
ment, turns the anterior end toward the source of light directly.
There is no trial reaction in this process. It is an asymmetrical re-
sponse to an asymmetrical stimulation. The movement of these an-
nelid larvae appear little more voluntary than the precise movement
of algal swarm spores."

Galvanotropic reactions frequently produce aggregations in a
diagrammatic fashion. Thus Paramecium, a protozoan well known
to react usually by a reflex type of behavior which suggests Jennings'
designation of a "trial and error" reaction, exhibits a forced-move-
ment type of behavior under the influence of a continuous electric

40

ANIMAL AGGREGATIONS

current. Jennings (1906), in his discussion of this reaction, says,
''When a Paramecium is transverse or obUque to the direction of a

1 2

Fig. I. — (i) Arenicola larva in the free-swimming state proceeding on a spiral
course, m, n, Directions of light; a-/i, positions in the spiral. Larvae react to changes
in ray direction in positions b or d, but not in positions a and e. (2) Much enlarged
sketch of larval head. The eye-spots are composed of a dark-brownish part y and a
clear part x. Note the ciliary bands on (2) which are a part of the locomotor system.
(From Mast, Liglit and the Behavior of Organisms; courtesy of Wiley & Sons.)

current at the time when the circuit is closed (Figure 2) certain strik-
ing effects are produced. If a current of medium strength is em-

FORMATION OF ANIMAL AGGREGATIONS

41

ployed, such as causes reversal of about half the cilia, the following
results may be observed. On the anode side the cilia strike back-
ward as usual. On the cathode side the cilia strike forward. As a
result the animal, when in a transverse position, must turn directly
toward the cathode side, the cilia of both sides of the body tending

~^

4-

Fig. 2. — Effects of electric current on the cilia of Paramecia and the direction of
turning in different positions (large arrows). The small internal arrows show the direc-
tion in which the cilia of the corresponding quarter of the animal tend to turn the
animal. At/ the impulse to turn is equal in both directions and there is no result until
the revolution on the long axis brings the aboral side to the cathode. (From Jennings,
Behavior of Lower Organisms; courtesy of the Columbia Press.)

to produce this effect, as indicated by the arrows in Figure 2. This
happens even when the oral side is directed toward the cathode
(Figure 2e). The animal turns toward the oral side — a result never
produced by other stimuli, and due to the peculiar cathodic effect of
the current."

42 ANIMAL AGGREGATIONS

Once oriented so, the animals swim toward the cathode; if the
current is reversed, a reversal is caused in the orientation and loco-
motion of the animals. Many similar cases of forced orientation and
locomotion under the influence of the galvanic current are to be
found in the literature; in certain cases the animals move to the
anode rather than to the cathode. A general summary of galvano-
tropic reactions has been given by Loeb (1918).

Similar forced movements which lead to aggregations under favor-
able conditions are given in response to other stimuli, particularly
those of chemicals and of gravity. They are not given by all members
of the animal kingdom, and are more likely to be exhibited by those
animals which, like the insects, have a disproportionate development
of the sensory system in comparison with the central nervous sys-
tem, so that the animal becomes the creature of its sensation (Ken-
nedy, 1927).

RANDOM MOVEMENTS

On the other hand, animals may congregate as a result of a series
of reactions, which suggest the method described by Jennings (1906)
as "trial and error" or, as Holmes (1905) has put it, by "the selec-
tion of random movements." The classic case is that originally given
by Jennings, of Paramecia collecting in the more-acid portion of the
water they occupy. This reaction is in part, at least, a trap reaction,
in that the animals do not react upon entering the more-acid region,
but respond by the characteristic avoiding reflexes when they come
in contact with less-acid water, and hence are caught in the region
of higher acidity (Johnson, 1929). This reaction by Paramecia is so
well known as to have been diagramed in all the current textbooks of
zoology. It is worth emphasizing that such a method of formation
of an aggregation, while less spectacular, is not necessarily less mech-
anistic than is the type of reaction given by Arenicola larvae when
they collect under the influence of directive light stimulus. It is also
of interest to us that, as the Paramecia aggregate, the carbon dioxide
given off as a result of their normal metabolic activities tends to keep
the region more acid and thus the aggregation tends to perpetuate
itself.

When there is a limited space available, or a limited amount of

FORMATION OF ANIMAL AGGREGATIONS 43

optimum space, aggregations may form from either of these two
reaction methods, the method used depending in part on the nature
of the stimulus emanating from the favorable locality, but, in the
main, on the reaction system of the animals involved. If the condi-
tions are such that directive stimuli are absent, aggregations, if
formed, will result only from the methf: d of "trial." This apparently
happens many times in nature and in lx^c laboratory.

Land isopods (Allee, 1926) tend to collect in aggregations in the
hot, dry summer and in the cold, and often physiologically dry,
winter. These aggregations are frequently such as might result when
shelter is limited, provided there is a tolerance for the presence of
other similar animals; but at times these animals collect in much
closer units than can be entirely explained on this basis. That is to
say, the isopods do not occupy all the available and apparently
equally desirable space, but clump together in one part of this.

When the method of formation of the aggregations is studied in
the laboratory, the grouping is found to be brought about by the
"selection of random movement" type of behavior. Usually the iso-
pods wander over the surface of their container, preferably around
the margin, and come to rest in the position in which they are ap-
parently less stimulated. Downs (Allee, 1926) made a long series of
observations in an attempt to find the method of formation of ag-
gregations when conditions were as nearly uniform in all parts of the
container as they could be made. Under these uniform environ-
mental conditions the land isopods usually wandered about until
one came to rest for some reason or other. Sometimes inequalities
developed in an originally uniform environment; at other times the
isopod apparently stopped for internal reasons. After one became
quiet, there was a distinct tendency for others to come to rest near-
by. These might or might not be in physical contact with the first;
frequently they had crawled over it immediately before stopping. In
their incipient stages these bunches were frequently quite loose. The
isopods would then alternate periods of rest and of motion. During
the latter, many, or perhaps all, might start up again; but often a
nucleus remained, consisting of the original individual and one or
more others. Around such a nucleus the isopods would again gather,

44 ANIMAL AGGREGATIONS

and the bunch would at last become consolidated by slight move-
ments on the part of those on the periphery. Partially successful
attempts were made to control the place of bunch formation on a
uniform field by gluing a recently killed isopod to the substratum.

When a drop of water was introduced on a dry background, the
isopods tended to occupy all of that favorable location regardless of
whether or not they were in contact. The bunching in close physical
contact came later, and might take place as a thigmotropic reaction,
perhaps modified by chemical stimuli, or might have been condi-
tioned by the drying of the small moistened region.

Similarly, detailed studies have been made on the bunching be-
havior of the ophiurid starfish, Ophioderma brevispina Say (Allee,
1927), which lives in the eelgrass along the Atlantic Coast of North
America from Cape Cod southward. Individuals of this species have
not been found in physical contact in nature during the summer and
late autumn, but the collectors for the Marine Biological Laboratory
report that large numbers may aggregate in late November and
December. In the laboratory the tendency to collect in bunches dis-
appears as conditions approach those obtaining in nature. Thus,
bunches were absent or rare when eelgrass was present in approxi-
mately natural condition. These relations held even under the tem-
peratures of about 10° C. obtaining in laboratory aquaria in late
December.

When, however, the Ophioderma were placed in bare containers,
bunches formed within a short time. The speed of formation was
retarded by the slower movement accompanying low temperatures
and dim illumination. The efi"ect of changes in illumination are
shown by the following example : With a constant temperature near
20° C. one lot formed a compact aggregation in from i to 10 minutes
in different trials in direct sunlight; in from 14 to 25 minutes in
diffuse light, and in from 27 to 56 minutes in complete darkness.

Detailed observations of the method of formation of a large num-
ber of these aggregations made under a variety of conditions show
that the collections occur in the less illuminated part of the container
when there is a difference in light intensity. When conditions are
uniform, the starfish cluster about one of the least active individu-

FORMATION OF ANIMAL AGGREGATIONS 45

als of the lot. In both cases the aggregation forms after a large num-
ber of apparently random movements in which the individuals react
to the others present in much the same way that they do to pieces of
glass rods or to eelgrass. Once formed, these aggregations tend to
move together and so to form a more compact bunch. This may
smack of a social tendency, although similar behavior is shown to
occur when isolated individuals are adjusting themselves to the in-
equalities found in a tuft of eelgrass or a loose pile of glass rods.
These bunches of Opiiioderma are formed in the same general manner
already described for land isopods.

Such behavior as that of the land isopods or of these starfish is
obviously to a large extent conditioned by the reactions of the ani-
mals to their physical surroundings. In the absence of elements usu-
ally found in the normal physical environment, animals may so react
to each other as partially to substitute for the normal environment;
that is, other individuals may take the places usually occupied by
non-Hving environmental items. Two types of explanation have been
advanced for this kind of phenomenon, one of which implies some
innate social tendency. The other explains such aggregations in more
objective terms.

THE FORMATION OF CELL AGGREGATES

Roux (1894), a distinguished experimental embryo logist, observed
that when cells of the frog's egg are shaken apart during early stages
of cleavage and placed in water only a short distance apart, they
slowly approach each other until they come in contact. He termed
such cell behavior "cytotropism." In normal development this tend-
ency acts to help keep the cells close together in a compact mass.
Later Wilson (1910), Galtsoff (1925), and Child (1928), among
others, have observed the behavior of dissociated tissue cells of
sponges and hydroids. Some of these thoroughly dissociated cells
move about and collect in cell aggregations which under proper con-
ditions regulate into new organisms. Galtsoff for sponges and Child
for the hydroid Corynwrpha have concluded that these cells come
together as a result of chance movements on the part of certain cells
which incidentally collect other cells as they move and by chance

46 ANIMAL AGGREGATIONS

come together to form viable aggregations. Galtsoff 's statement con-
cerning sponge cells is: "The examination of the behavior of dis-
sociated cells shows that the formation of aggregates is chiefly due
to random movement of the archaeocytes which collect all the cells
lying in their route." Child is more certain of the absence of definite
cytotropism around dissociated Corymorpha cells and has observed
cells when near together to move apart without aggregating ap-
parently as often as he has found movement in the opposite direc-
tion.

The cytotropism observed by Roux can be interpreted as analo-
gous with a very simple social appetite, or at least showing that mu-
tual attraction between living units extends to dissociated embryo
cells. From forces similar to those causing such simple mutual at-
tractions of cells, we might expect social appetites to develop. Such
a reaction may be regarded as a forerunner of the social instincts of
many observers. On the other hand, in the formation of aggrega-
tions of dissociated sponge and hydroid cells there is no evidence of
such mutual attraction. The method is essentially the same as that
just outlined for the formation of aggregations of land isopods and
of starfishes. Yet under favorable conditions these cell aggregates
formed without evidence of mutual attraction may develop into well
integrated animals.

PROTOTAXIS AND INSTINCT

Wallin (1927) has postulated a factor or principle which he re-
gards as of fundamental importance for many interrelations between
cells or between whole organisms and which he has called the prin-
ciple of "prototaxis." This is defined as "the innate tendency of one
organism or cell to react in a definite manner to another organism or
cell." This reaction may be either positive or negative. The latter
results in a mutual repulsion of organisms or cells, for, since organ-
isms may be found separated for a number of reasons, Wallin recog-
nized that negative prototaxis can be demonstrated only if the actual
process is observed. On the other hand, positive prototaxis, which is
"the affinity of one organism or cell for another organism or cell,"
may result in such well-known phenomena as those of conjugation.

FORMATION OF ANIMAL AGGREGATIONS 47

symplasm, cell fusion, parasitism, and symbiosis. Obviously these
are real phenomena; but, as detailed information concerning the proc-
ess by which the cells or organisms come together is lacking, the fact of
their being together is no more evidence for the existence of a posi-
tive prototaxis than the separateness of other cells or organisms is
proof of the existence of a negative prototaxis.

If we waive this objection to accepting the principle of prototaxis
as an all-inclusive explanation of all aggregations whether of cells or
of organisms, and proceed to examine the nature of prototaxis, we
find that, instead of a simple tropism which may be best understood
as a reflex action of an entire organism, prototaxis is a compound or
complex tropism which Wallin says cannot be analyzed. Certainly
we can recognize different elements, such as chemotropism, thigmo-
tropism, stereotropism, as well as reactions due to surface tension,
temperature, light, moisture, and electrical potential. In fact, such
an analysis indicates that WalUn's conception of prototaxis is merely
another name for the type of reactions referred to by many writers
as being instinctive, except that no one would ordinarily regard the
reaction of tissue cells as belonging in this category. Logically there
is no real reason why they should not be so regarded, but the usage
has been otherwise.

Wallin's conception of the formation of aggregates, whether of
cells or of organisms, as being due to the expression of a fundamental
biological tendency or principle, has two merits. In the first place
it recognizes rightly that there is no logical line to be drawn between
the behavior of tissue cells forming an animal body and that of plants
or animals forming a close aggregation like those seen in symbiotic or
parasitic relations. This is in line with the conclusions of Espinas
and of Deegener, which I believe to be essentially correct, that
there is no hard and fast line that can be drawn between the
social and the infrasocial. Further, Wallin specifically recognizes
that the ideas that have developed about symbiosis and parasitism
have usually been based on the utility of the relationships and have
also involved the idea of purpose. When such phenomena are con-
sidered from the point of view of prototaxis, then parasitism and
symbiosis and presumably all their related phenomena are merely

48 ANIMAL AGGREGATIONS

different end-responses in the expression of one and the same bio-
logical principle, involving therefore only the vague type of utility
necessary for the cumbersome working of natural selection and with
no more suggestion of purpose than is inherent in scientific concep-
tions generally.

This analysis of prototaxis shows that it is in the main a renaming
of the type of activities usually called "instinctive," with the exten-
sion of this sort of action to include the behavior of tissue cells and
with a deprecation of the tendency to include a distinct teleological
element which is usually present in discussions of instincts. The
question immediately arises as to what social instincts or appetites
may mean, and .whether or not they are capable of analysis.

Szymanski (1913) undertook to investigate this problem by com-
paring the reactions of isolated caterpillars of Hyponomeuta and of
Arye with those given by groups when placed under the same general
conditions. Recognizing the fact that social reactions are not readily
analyzed, Szymanski undertook to separate them into two cate-
gories: (i) those peculiar to the individual, which, if fortunate, make
possible the living-together of individuals as a social group, and
which may be called "primary reactions"; and (2) responses which
arise as the result of the living-together of many individuals, and
which may be called "secondary reactions."

In order to distinguish primary and secondary responses in a social
group, Szymanski suggested and used the following procedure. The
reactions of the individual are first studied with a view to finding the
usual responses given to various stimuli; thereafter one studies the
behavior of individuals as members of a group. In the latter study it
is frequently possible to recognize elements of behavior which have
been observed in the isolated individuals. If all the reactions given
by the individuals of a colony can be recognized as primary re-
sponses, such as would be given were all the animals isolated, the
problem of group behavior is solved without the need for recognition
of secondary or essentially social behavior; but if there is a residue of
behavior which cannot be recognized as primary, then this is to be
regarded as the secondary or true social behavior.

Szymanski so analyzed the responses given by caterpillars of

FORMATION OF ANIMAL AGGREGATIONS 49

Hyponomeuta, which inhabit an irregularly rounded web usually
placed between several branches of the food plant. The individual
larvae exhibit no tropisms except a strong negative stereotropism.
When placed singly on the ground, the larvae make a looplike path.
They stop at almost every point of the loop and test out their en-
vironment with their heads, selecting thus a place to lay their silk
thread. Single caterpillars spinning their web behave similarly. The
individual reaches out as far as possible from the place of beginning
and lays down its thread. This action is repeated and results in a
spreading of the web. Such a response to space Szymanski regards
as a negative stereotropism.

In one experiment eight caterpillars were observed in nest-build-
ing. Six were placed together at one place, and one each at two
slightly distant points. All began spinning webs as described above.
The six spun a common web, which finally reached and fused with
the webs of the isolated caterpillars, so that a joint web resulted.
This probably happens in nature. So far, there are no reactions re-
maining over and above the individual responses, and Szymanski
concludes that in the formation of this common web there are no
secondary or purely social reactions.

Deegener (1922) disagrees with Szymanski 's observations and
with his interpretations, particularly the latter. He, too, found that
isolated Hyponomeuta larvae can spin webs, but concluded that they
do not begin spinning as soon as if grouped together, and that the
web spun by a group is smaller than that made by the union of webs
spun by the same number of isolated larvae. In both respects he
would recognize the working-out of a social instinct. Further, he
believes that the caterpillars actively seek out the company of
others, guided by sensing vibration waves, which may be merely
refined touch perception. Back of all this he believes there is a need
for association which leads the isolated larvae to seek their own kind.
If they do not find their fellows, they build their own individual
nests, which later they may abandon, wandering and seeking in
order to associate themselves with other larvae. If none are found,
they may remain solitary for days without losing their social in-
stinct.

50 ANIMAL AGGREGATIONS

Szymanski (19 13) further studied the formation of feeding aggre-
gations of Arye caterpillars. Groups of these young caterpillars
gather on their species of food plant and arrange themselves on the
leaves so that they cling with their thoracic legs on the upper surface
while the posterior end hangs down curled around the edge of the
leaf. They arrange themselves so, side by side along the margin of
the leaf. When the larva at the tip has eaten to the main vein, it
may do one of three things: (i) turn around and go to the base of the
leaf and begin feeding there; (2) leave the leaf entirely; or (3) cross
over to the opposite side and begin feeding there. The older cater-
pillars tend to lose this regular arrangement and behavior.

By the usual type of analytical experiments the Arye caterpillars
are shown to be positive to light, negative to gravity, and positive
to certain touch stimuli. The method of locomotion consists in the
extension of the anterior end and the drawing-up of the posterior.
The posterior end shows a definite motor reflex upon stimulation.
Thus, if touched at the posterior end, the posterior half of the body
is raised. A similar reaction is given if the substratum is gently
shaken. If one side is touched, the same response may occur, to-
gether with a bending-away of the touched part.

If one tests out the method of colony formation, one finds that
when the larvae are placed at the base of the food plant they will
crawl up on it, since they are positive to light and negative to grav-
ity. When the first leaf petiole is encountered, they will turn aside
onto that because it is narrower than the main stem, and for the
same reason they will move along the edge of the leaf. On the leaf
they move to the side most strongly illuminated, or to the side far-
ther from the ground, as the case may be.

The larvae crawl here and there over the leaf, passing over each
other; or they may touch the larvae ahead and cause them to move
forward. Finally one begins to eat, and gradually all settle to eating.
The positions taken may be accidental, for wide spaces may occur
between larvae, while others are closely crowded. The piece of leaf
between two larvae becomes eaten away, so that eventually the
head of the second larva touches the posterior end of the first. This
causes the latter to raise its posterior end, as in the test experiments

FORMATION OF ANIMAL AGGREGATIONS 51

described above. The reaction will be repeated whenever the posteri-
or end is stimulated, and only ceases when the abdomen curls over
the edge of the leaf. In this way, and as a result of these reactions,
the colony takes on its well-organized appearance, which depends on
the interaction of the following factors: (i) the crowding of many
individuals into a small space; (2) the tropic reactions of the larvae;
(3) the character of the anterior and posterior end reflexes; and (4)
the manner of locomotion and of feeding.

Here, as in Szymanski's analysis of the group formation in Hypo-
nomenta, primary reactions play the principal role in the colony
formation ; but there are some elements of the behavior of the colony
that Szymanski thinks may be due to secondary or social behavior.
Thus, when the leaf is shaken, the posterior end of each larva is
raised simultaneously. When we remember the great individual dif-
ferences usual in behavior, the synchrony of this response suggests
that there may be a social factor at work. However, it is possible
that this, too, is merely an expression of primary or individual reac-
tion, with the synchrony either more apparent than real or due to
the proximity of the responding larvae.

These investigations of Szymanski's lead to the same conclusion
as my own, formed independently, concerning the method of forma-
tion of aggregations of land isopods and of Ophiodcrma. In these
cases it is the primary, individual reactions that produce the group-
ings, not the expression of a community spirit or of a social appetite.
The only social trait necessarily present is that of toleration for the
presence of numerous other similar animals within the same region.
If this analysis be sound, as it appears to be, then one of the early
stages of mutual interdependence is the appearance of toleration for
the presence of other animals in a limited space, where they have
collected as a result of tropistic reactions to environmental stimuli.
Once formed, aggregations may persist for a considerable time, mere-
ly because of the lack of disruptive stimuli.

The conclusions of Szymanski are supported by Krizenecky (1923)
in his work on the transitory aggregations of the enchytraeids al-
ready mentioned in the chapter on classification. He thinks that
individual reactions are important in the formation of these aggre-

52 ANIMAL AGGREGATIONS

gations, and that thigmotropic reactions are largely concerned. The
observations of Essenberg and of Riley on water striders, of Clark on
Notonecta, of the Severins on Belostoma, as well as the tremendous
general literature on animal behavior (see Loeb, 1918, for a partial
bibliography), show that aggregations do form in many cases with-
out evidence of a positive social instinct or appetite, although this is
not to be taken as proof that in other instances aggregations may
not form as a result of social appetite.

AGGREGATIONS OF ASELLUS IN NATURE

The analysis of one other case is illuminating. For a number of
years I had been seeking a favorable opportunity to apply in the
field certain analytical methods worked out in studying aspects of
the laboratory ecology of animal aggregations, and accordingly
welcomed the information that a great aggregation of the common
fresh-water isopod, Asellus communis Say, had been found in mid-
winter in the Indiana dune country near-by.

At the point where this collection occurred, a low sand ridge had
been thrown up to serve as a roadway across an extensive cat-tail
swamp, here about a quarter of a mile wide. To the east, the swamp
stretched as far as could be seen from the low elevation of the road-
way. To the west, there was also a very extensive continuation of
the cat-tail swamp for at least a half-mile. The whole formed a major
part of the headwaters of a small stream.

The roadway was pierced at several places by culverts, introduced
to relieve the water pressure above. These had proved inadequate,
and at one place the water had washed away the ridge of sand and
flowed over the roadway through an opening about 5 meters wide,
with a current there sufficient to prevent complete freezing. When
first seen, the ice was about 6 cm. thick and the effective stream was
reduced to about 1.5 meters width.

Here on the under side of the ice were tens of thousands of isopods,
oriented to face upstream, and showing by their arrangement the
definite lines of force of the current below. Thousands of other iso-
pods were resting on the bottom in protected places, and many more
were being swept downstream by the rapidly moving current. Some-

FORMATION OF ANIMAL AGGREGATIONS 53

times these collected into small balls of from 6 to 20 isopods, which
rolled along the bottom until they found a lodging against some ob-
struction or settled into a deeper pool where the current was less
strong. There were many isopods on the sandy bottom of the stream,
mostly facing against the current, but making very little progress
against it. Chopping through the ice above or below the roadway
revealed no comparable collection of isopods, although there was
evidence of an increase in numbers as one neared the narrow chan-
nels of the washout either from above or below.

After the break-up of the winter ice, the majority of the isopods
disappeared, although traces of the aggregation could still be seen,
particularly in the sheltered places just below the opening of the
stream into the lower swamp. There the isopods were mainly travel-
ing downstream with the current, or were collected in sheltered
places in deeper water or about lodged debris. As before, few were
found in the open above the roadway.

After the ice was entirely gone, and with the usual rise in water
level, the aggregation re-formed. In early April a few were being
carried downstream through the washout. Several more were to be
seen along the margins, for the most part headed upstream, where
some were able to make their way for a considerable distance. At
the lower edge of the roadway, great masses of isopods had collected
about willow shrubs, old cat-tails, or in deeper pools, wherever they
might find a lodging. The largest of these masses was about 75 cm.
across the current, 30 cm. up and down stream, and over 10 cm.
deep, a solid writhing mass of isopods. This was loosely joined with
other similar units, each formed about some basis of support from
the force of the current, the whole making an isopod barrier all along
the lower margin of the washout, over 5 meters in length and about i
meter wide. The numbers concerned were unbelievable. They were
to be measured by liters rather than by individuals. The mass can
be imagined by thinking of the full swarms of some twenty or more
beehives settling near each other. Conditions remained much the
same for the next 3 weeks, with but a slight variation in the position
of the largest mass, depending apparently on the strength of the
current.

54 ANIMAL AGGREGATIONS

In late April the water level had again been raised by rain, and in
the main current stood about 45 cm. deep, in place of the more usual
15-18 cm. A new, smaller overflow had been formed near-by. The
isopods were all gone from their place of aggregation ; and although
they were still plentiful all around the edge of the fan of sand washed
down by the recent rains, they were not collected into the great
masses found heretofore. In the slacker current just preceding the
rains, isopods were no longer being carried downstream across the
roadway; and one could not have collected more than 50 such drift-
ers by watching all day. Now with the higher water level and the
swifter current they were again being swept downstream from the
upper swamp in numbers.

With the higher water level of early April, smaller aggregations
had appeared about the lower ends of the central iron culverts
piercing the roadway, but now there, too, were dissipated. With the
higher water of late April, the culverts situated at the edges of the
swamp showed a marked current for the first time — not nearly so
strong as that in the center culverts, but corresponding in strength
to the latter when aggregation of isopods occurred near them. Now,
for the first time, sizeable aggregations were present at the lower
ends of these side culverts. At the upper end of one of these there
was a log and much plant debris on and about which isopods borne
down from the upper swamp might have lodged; but none were
there, while they occurred in large numbers at the lower opening,
particularly in eddies out of the main current. The current ceased to
flow through the north marginal culvert within a few days, and the
aggregations there disintegrated. Those at the opposite margin per-
sisted for about two weeks. The aggregations below the main over-
flow did not re-form, although many individuals could be seen at any
time unsuccessfully attempting to make their way up over the shift-
ing sandy bottom.

The final breaking-up of the aggregations was not observed,
though at any time small groups or single individuals might be seen
becoming detached and borne away by the current. When the water
rose, the increased velocity probably carried the whole lot off in a
similar manner. At the end of the season some of the aggregated ani-

FORMATION OF ANIMAL AGGREGATIONS 55

mals died in situ, especially if located at one side where the current
became cut off.

An increased flow following heavy rains in late May produced
physical conditions similar to those of late April, but no aggregations
were formed, although a few large isopods were carried downstream
through the main spillway.

The favorable localities were well watched the next winter and
spring, but no large aggregations were found. In early April a small
aggregation occurred below the culvert at the extreme south side,
where the last one had formed the spring before. With the passing of
time, the main washout had deepened so that a stronger current was
running there than when the isopods had aggregated the preceding
year. In general, there appeared to be fewer Aselli in the swamp,
and one is led to suspect that there may have been an unusually
large production of isopods preceding the formation of the monster
aggregation observed in 1927.

SEX RATIO

In early spring one can usually determine the sex of Asellus by
considering the size, shape of thorax, and presence or absence of the
brood pouch. In the laboratory, sexes are easily and accurately de-
termined. Careful observations showed that during the time of the
great spring aggregations the ratio of the collected isopods ran as
high as 25 males to i female, and never ranged below 9:1.

November coflections from the scattered isopods, both above and
below the culverts, showed a i : i ratio. In early April of the next
year, random collections from the relatively small aggregation at
the lower end of a lateral culvert showed a sex ratio of 12 males to
each female. At the same time, similar collections both above and
below the aggregation showed a ratio of approximately 1:1.

Five suggestions readily occur to account for the high ratio of
males to females in the great bunches:

1. The aggregation may be due to a mating or other social im-
pulse acting more strongly in the males than in the females, which
impels them to gather in these large groups.

2. The males may tend to move about more and to come into

56 ANIMAL AGGREGATIONS

contact with the current and be swept off their feet, regaining a foot-
hold only when the current slackens or when they reach a solid foot-
ing.

3. The males may possess less clinging power than the females.

4. The females may be carried downstream as well as the males,
but may escape from the bunches to the lower swamp.

5. The aggregations may be formed from isopods that start up
from the lower swamp and are unable to make progress when the
swifter current is encountered. If this is a factor, it would imply that
the males are more strongly positive in their rheotropic reaction than
are the females.

The last four possibilities would account for the formation of the
aggregations through the operation of tropistic reactions of the iso-
pods as individuals, the so-called "primary reactions" of Szymanski
(1913); while the first would bring in a secondary or group reaction.
The different possibilities may be considered in reverse order.

The rheotropic reactions of both sexes were tested according to
methods developed earHer (Allee, 191 2). These tests indicated that,
in the breeding season at least, the males are somewhat more strong-
ly positive in their rheotropic reaction than are the females, and
that they respond positively to stronger currents. In so far as the
aggregations form as the trapping of positive isopods moving up-
stream from the lower swamp, this helps to account for the great
discrepancy in the sex ratio. However, this is not the whole story.

There is little evidence for the assumption that the females may
be carried down from the upper swamp in the same numbers as the
males but escape from the aggregation to the lower swamp. There
were very few females found among the many isopods collected
while being carried downstream.

The supposition that the males have less clinging power than the
females, at least in the breeding season, was subjected to direct ex-
perimentation, using the method described by myself in 1914. The
results indicated that there is little, if any, difference in the cHnging
ability of the males and females under the conditions of this test,
with whatever advantage that may exist favoring the males. Such
results are to be expected from a consideration of the mechanical

FORMATION OF ANIMAL AGGREGATIONS 57

difficulties of maintenance of position by females carrying a large
brood pouch between their anterior thoracic legs.

Of the tropistic non-social suggestions advanced as possible ex-
planations of the greater proportion of males than females in the
spring aggregations, one more remains for detailed consideration.
This is the suggestion that the males move about more and so come
into contact with the current more frequently than the more passive
females. Such ditTerential action would result in more males being
swept off their feet and carried down from above, and also in more
males coming in contact with a current strength which would call
forth a positive rheotropic response and so bring them up from the
lower swamp. This possibility is supported by the following kinds
of evidence. The direction of the current impinging on a large bunch
was artificially changed, and the current change resulted in a re-
organization of the bunch of isopods in a new position. At a time
when the main bunch showed a ratio of males to females of 25:0,
25:3, 25:2, 25:2, with a total of 100:7, the reorganized bunch
showed ratios of 50: i and 45 : 4, with a total of 95 : 5, which is nearly
twice the number of males per female as found in the bunch of
longer standing. Again, I pulled from near a large aggregation a tuft
of grass heavily covered with isopods. The sex ratio of those that ac-
tively crawled from the grass onto my hand proved to be 4 males to
each female. The sex ratio of all the isopods on a similar tuft was
found to be i male to 3 females. In both cases the males showed a
higher degree of activity. It is also true that the vast majority of
animals taken while being carried downstream by the current were
males, and that the sex ratio of the isopods on the water plants out-
side the main current, but above the roadway, showed a higher
number of females than males.

Regarding the possibility that the males may be responding to a
stronger internal sexual stimulant than the females, there is evi-
dence from earlier work that in the breeding season the males do tend
to cling to any passing isopod, and apparently have this tendency
more strongly developed than do the females. The tendency to col-
lect in bunches is so strong that spring isopods must frequently be
tested singly for rheotropism or they will fail to respond to the cur-

58 ANIMAL AGGREGATIONS

rent at all. I have seen males which were responding definitely to a
water current behave as if they perceived another isopod at a dis-
tance of some 2-4 cm., discontinue their rheotropic reaction, and
move directly to the nearby isopod and cling to it. I have no evi-
dence of such reactions at distances greater than 5 cm., so that their
effect would be operative in bunch formation only, after the isopods
had been brought close together through the operation of some other
factors.

I have no knowledge of such isopod aggregations except in winter
and spring, and unfortunately the sex ratios of the winter aggrega-
tions were not taken. In this connection it must be remembered that
the isopods do not start their breeding season in December in nature.
Yet large aggregations were found at that time. The observations
show clearly that the ratio of males to females is high in the spring
aggregations, and suggest that this is due to the tendency of males to
move about during the breeding season, which makes them more
likely to be caught in the current and swept down from the upper
swamp, and, on the other hand, more Ukely to come into contact
with a current sufficiently strong to cause them to react positively,
and so move upstream to the place of aggregation from the lower
swamp.

METHOD OP FORMATION OF THE AGGREGATIONS

This subject obviously overlaps consideration of the preponder-
ance of males, and the conclusions reached from much consideration
of the problem are the same as those indicated there. As was to be
expected, disturbances in the swamp just above the opening of one
of the outlets caused a marked increase in the numbers of isopods
carried down. These might lodge in slight depressions in the stream
bed where the current was less strong; some 50 were observed to
collect in a small depression less than 12 cm. in diameter within 5
minutes following a disturbance in the upper swamp.

Others were carried on by the current until they found physical
support against rushes or other debris, or against other isopods which
were in turn supported by the rushes. Thus, the bunch may be seen
to grow on its upstream side, the newcomers using the other isopods
as an extension of the support furnished by the lodged debris.

FORMATION OF ANIMAL AGGREGATIONS 59

But this is not the only method by which the aggregations are
formed. ISIention has been made already of the finding of a large
aggregation at the lower end of a culvert whose upper opening was
well protected by the presence of logs, grass, and other debris,
through which the water ran easily, but upon which few isopods
collected even at the sides where the current was certainly not of
sufficient strength to tear them loose from available support.

In laboratory experiments with artificial streams some isopods,
mostly males, traveled against the current and collected in the more
quiet water at the upper end of the trough. Similar behavior was
repeatedly seen in nature. After the ice left, isopods from the great
spring aggregations could be seen laboriously moving against the
current over the sandy bed of the stream ; while those located below
the opening of one of the streams into the lower swamp, if not pres-
ent in sufficient numbers to form an aggregation of three dimensions,
were frequently spread thickly over the bottom, with all individuals
headed upstream.

Of all the isopods moving upstream, those near the margin were
most successful. Usually, however, all were swept down sooner or
later to the main group below. When a board was placed with one
end resting in an aggregation so that it furnished a solid substratum
on which the isopods might crawl, they immediately started up-
stream as closely as they could stick on the board. On reaching the
upper end, many were immediately washed down by the current,
while others would continue over the precarious bottom for a short
distance before they, too, lost their footing. If dikes were built so
that the current impinging on an aggregation was slackened, the
isopods started upstream in numbers, only to be swept down again
when a stronger current was encountered.

There is also a fatigue factor which causes the failure of these
isopods to continue their journey upstream even in a fairly weak
current. The length of time before reversal is roughly correlated
with the physical condition of the isopods. In laboratory tests with
isopods from these aggregations, reversal in a straight current oc-
curred after an exposure of about an hour.

If the impinging current is cut off completely by the construction

6o ANIMAL AGGREGATIONS

of a dam, the aggregated individuals begin a rearrangement which
usually results in new aggregations being formed in depressions, or
about some quiet individual or a quiet group, just as such aggrega-
tions form in the quiet water of a laboratory tank. These groupings
are usually less dense than those exposed to the drive of the current.
The negative reaction to light is one of the factors conditioning this
reaction ; positive thigmotropism is another. If grass or other debris
is present in abundance, the isopods usually collect in contact with
the inanimate matter rather than piling up in great isopod aggrega-
tions. There may be some collections due to positive chemotropism,
for these aggregations cause measurable differences in their chemi-
cal environment.

I was much impressed, in all the observations made upon these
groups, by the fact that so large a part of the formation of the aggre-
gations could readily be explained on the basis of individual tropistic
reactions to environmental stimuli largely produced independently
of the massed isopods themselves. Relatively few of the causes of
aggregation were left to be explained by the reactions due to social
appetite. In this respect the situation is wholly similar to that found
with land isopods, with Ophioderma, and with Szymanski's cater-
pillars. Again the main social trait exhibited appears to be that of
tolerance for the presence of many other individuals in a limited
space where they have collected, or — one might almost say — where
they have been collected. The same idea can be expressed by saying
that almost the sole social trait exhibited is immunity to injurious
effects resulting from the presence of many others in a limited
amount of space. It is interesting to note that there were also leeches,
snails, and other animals collected in the same location and, to a
large extent, by the same combination of physical forces and tropis-
tic reactions that had brought the isopods together.

GYRINID BEETLES

The reactions concerned in the formation and maintenance of two
more complex, more closely integrated types of aggregations have
also been made. In the case of the gyrinid or whirhgig beetles, giant
aggregations may occur on the surface of streams or of still water,

FORMATION OF ANIMAL AGGREGATIONS 6i

where the animals may be resting quietly or where they may exhibit
what appears to be a perfect frenzy of erratic activity. As stated
above, Deegener regarded these as forming play societies, while
Schulz thought of them as having protective values.

From the analysis of Brown and Hatch (1929) it appears that the
aggregating behavior of these beetles is largely due to visual stimuli,
since the aggregations break up in the dark. Further, the position
they occupy in the laboratory tanks, though not necessarily in na-
ture, may be determined by the lighting. These authors believe that
the gyrinids are exhibiting a more complex type of behavior than
that which is usually called "tropistic," and refer it rather to some
sort of configurationist behavior, in which orientation behavior con-
sists of movements so co-ordinated that an invariant relationship is
maintained between movements and variations of the visual field.

They find evidence of two sorts of orientation : one in which the
body axis is maintained in a relatively fixed position with respect to
the base of orientation, and another in which the body is maintained
in a relatively fixed region but without body orientation. The former
is like the orientation called for by the tropistic theory. The latter
bears at least a superficial resemblance to those cases where organ-
isms move along a physical or chemical gradient in one direction
without reaction to it but execute negative "avoiding" reactions
when moving in the opposite direction, like the trapping of Parame-
cia in weak acids, as described by Jennings.

They believe that the location of an aggregation in nature is due
to habituation to certain visual patterns, possibly of light and shade,
to which the animals respond; these patterns are not significant in
themselves but are a sign of the location of general environmental
conditions which are of vital importance to the beetle and the spe-
cies. If the patterns are slowly changed, the beetles may remain in a
given position; and collections have been observed not to shift their
position as much as a meter during a whole day, although the pat-
tern of the field of vision changed radically in that time. If, however,
the patterns are rapidly changed by a sudden increase in the com-
plexity of the visual pattern, marked stimulation to activity results,
which may cause a breaking-up of the aggregation.

62 ANIMAL AGGREGATIONS

Brown and Hatch, in their report, do not discuss the importance of
the presence of other individual gyrinids in the immediate neighbor-
hood in connection with the pattern complex. Rather, they give the
impression that each beetle is reacting as an individual to a general
environmental pattern, which traps it somewhat as Jennings regards
individual Paramecia as trapped by a drop of acid, until a collection
is formed.

CATFISH AGGREGATIONS

The young of the silurid fishes, the catfishes and the bullheads,
exhibit a striking type of aggregation, which has been analyzed by
Bowen (1929). In the species Ameiurus melas used in this work the
young may be observed in the summer months swimming in close
bunches near the surface of ditches or small ponds, packed together
in a more or less spherical mass. A single fortunate dip has yielded
over 500 of these minnows.

If such a group is scattered, within a few minutes 2 or 3 individuals
appear singly and come together somewhere near the original loca-
tion of the entire group. Gradually they are joined by single fishes
or by small groups which come into the same locality, apparently
swimming at random. These show no reaction to the larger group
until they are within 2 or 3 feet of it, when they swim directly toward
the larger aggregation and join it. Within 30 minutes to i hour the
original aggregation will have re-formed.

Appropriate tests showed that the individual fish were not react-
ing to a gradient of chemical emanations from the group, for they
do not respond to fish-conditioned water (i.e., water in which fish
have stood until it exhibits various chemical and perhaps physical
changes), even when the conditioning is greater than could be the
case with an aggregation in nature. Cutting the olfactory nerves had
no effect either on normal or on blinded fish.

With these bullheads, as with the gyrinids, vision is the important
factor in the formation of the aggregation and in its subsequent inte-
gration. Neither blinded fish nor normal fish in the dark ever ag-
gregate, and normal fish will follow a moving fish model in a way
similar to that which results in aggregation when in the company of
other normal fish.

FORMATION OF ANIMAL AGGREGATIONS 63

Ameiurus melas can sense through the skin the presence of another
fish in motion, probably by detecting the vibrations set up by the
tail of the nearby fish. Shght positive responses to others of a
group are shown by blinded fish; this reaction is not afifected by the
destruction of the lateral-Hne organs but is almost ehminated when
the skin is anesthetized with magnesium sulphate.

When the bullheads come into actual contact with another object,
a positive thigmotropic response is given. The barblets are dragged
over the object; and by means of the sensations received, apparently
chemical in nature, the fish is able to discriminate between paraffin
models and live fishes, but it is apparently unable to distinguish be-
tween fish of the same or of different species.

Bowen sums up her observations in practically these words: "In
the evening, as soon as it begins to grow dark, the aggregated young
catfish separate and swim about, sweeping through the water or
along the bottom with the barblets, giving a feeding reaction similar
to that given by blinded fish at any time of the day. As soon as it
begins to grow light the young fish come together into aggregations
in which they remain for the entire day, re-forming in a short time if
scattered by a disturbance. Some feeding may occur while the fish
are aggregating, but it is doubtful if this occurs to any extent. Usu-
ally the fish are in a close bunch actively pushing against each other,
or resting at the surface in contact or close proximity. A thigmo-
tactic reaction seems to be at the base of this behavior. Unless dis-
turbed, older fish in the aquarium rest during the day in contact with
the substratum, or more often in contact with one another. By
means of aggregations the young fish can satisfy their positive thig-
motaxis even while in motion. The pushing in a group suggests the
importance of this. Catfish will also push against other species of
fish, which, however, do not reciprocate. This contact reaction is
largely one of pressure, but gustatory response apparently plays
some part, as shown by the different responses to paraffin models and
to fish. Whether this factor is instinctive or is influenced to any ex-
tent by conditioning is yet to be determined. The reaction is ap-
parently not species specific, since there is no evidence that young
catfish show different behavior toward members of their own species

64 ANIMAL AGGREGATIONS

than toward other forms." The aggregation arises from the tend-
ency of the other catfish to respond by appropriate positive reac-
tions, instead of making-off as fishes of other species do when ap-
proached.

With these fairly well-integrated aggregations of young bullheads
analysis shows that the social appetite is diffuse rather than specific,
and that in the normal fishes, aggregation involves a sight reflex, a
touch reflex, and possibly a low-frequency vibration reflex, all of
which may be given to other moving objects, non-living as well as
living; a chemical reflex, the sign of which is reversed with non-
living models; and, finally, reciprocal behavior on the part of the
different individual members of the aggregation. The matter of re-
ciprocal responses contributes the distinctly social element in this
behavior. The extent to which the combination of these reflexes into
a functioning whole depends upon the presence of an inherited social
appetite, or upon early conditioned behavior, remains to be investi-
gated.

CHAPTER IV
GENERAL FACTORS CONDITIONING AGGREGATIONS

In many animal species the formation of an aggregation depends
on the physiological state of the animal. This may be controlled by
internal developments, such as the maturing of the sex products, or
by external factors, as when land isopods are made to bunch by con-
trolling the moisture of the substratum; but more commonly the in-
ternal and external conditioning factors work together closely. Some
of the more outstanding of these are discussed here.

THE BREEDING SEASON

Water isopods. — My own attention was drawn to the general prob-
lem of animal aggregations in 191 1, while studying the factors con-
trolling the rheotropic reaction in the common water-isopod Asellus
communis. As spring came on, the stream isopods no longer gave
highly regular, positive responses to the water current; but, as stated
in the preceding chapter, one might strike across a strong current,
guided apparently by sight, and seize another isopod, male or fe-
male. From such a beginning one might soon have all the isopods
under observation gathered into a compact rounded cluster, rolling
over and over in the water.

During the height of the breeding season stream isopods disregard
the stimulus of a water current almost completely unless they are
relatively isolated. On the other hand, I have repeatedly tried to
induce half-grown Aselli to form such a cluster, even placing them
in a watch glass with rounded, smooth bottom, where they were con-
tinually brought in contact with each other, but no real aggregation
resulted. Bunching may be induced in adults out of the breeding
season, but many conditions that favor it in April during the height
of the breeding season have little or no effect in late May (Allee,
1923)-

65

66 ANIMAL AGGREGATIONS

Mosquitoes and midges. — Culicidae and Chironomidae form swarms
of males which maintain position as groups, although the individuals
within the swarm are continually darting from one part of the swarm
to another. Such swarms have been known for years, although their
significance has not been generally understood. Knab (1906) cites
his own observations on the swarming of mosquitoes and reviews the
literature to show that the swarms are composed of males which
hover over or near prominent objects such as trees, corn shocks,
house gables, or people. Enormous numbers of these dipterans may
be present in the collections, which occur generally in the early
evening of quiet and almost windless days. Straight-flying females
dart into these irregularly gyrating swarms of males and emerge in
copula with one male. One such newly mated pair was observed to
emerge from one swarm only to enter accidentally another near-by.
The copulating pair appeared to be greatly stimulated and flew into
the open as soon as possible. The swarming males which were as-
sociated for even so short a time with the mated pair also increased
their rate of flying and "danced up and down at a furious pace for
some time" before again quieting down to their normal rate of
gyration. With growing darkness the activity of the swarms in-
creased, but fewer successful matings took place; the entering female
would be set upon by two or three males, and all would fall together
to the ground, where they would separate. Later, females ceased
entering the swarms, and the males gradually dispersed. Counted
sex ratios of Culex were 897 males to 4 females (Knab), and of Chi-
ronomus 4,300 males to 22 females (Taylor, 1900). Mosier and Sny-
der (1919a) interpret the large morning swarms of tabanid flies which
they observed in the Florida Everglades as aggregations of males to
which females are attracted and into which they dart for the purpose
of mating.

Frogs. — With the approach of spring frogs desert their hibernation
quarters for breeding places in the shallow ponds (Cummins, 1920).
Many hibernate in the mud at the bottom of these same ponds; but
others winter elsewhere, perhaps in nearby bodies of water or on
land among masses of dead vegetation, or in localities similarly
favorable. Cummins suggests that such frogs may migrate to open

GENERAL FACTORS CONDITIONING AGGREGATIONS 67

water caused by the early melting of ice in a pond with proper ex-
posure. Banta (1914), Yerkes (1903, 1905), and Noble (1923) find
evidence that frogs may respond to frog calls and splashings, par-
ticularly during the spring breeding season. Studies on the breeding
migration of toads indicate that with them the voice serves as a sex
call (Courtis, 1907; Miller, 1909; Wellman, 191 7). Boulenger (191 2)
concluded that the voices of frogs and toads do not control migra-
tions toward breeding grounds or movements of individuals at the
grounds. Cummins later came to the same conclusion as a result of
his observations on a partially fenced pond, since he found that
heavy migrations followed periods in which there was no croaking in
or near the pond, and that, on the other hand, great vocal activity
was not accompanied by increased migration. Certainly, vocal ac-
tivity cannot account for the similar spring migration of the voice-
less Arnhystoma.

The immediate inception of the migratory impulse must be in-
trinsic and is probably associated with the conditions of the sexual
glands. In frogs it is secondarily conditioned by weather, since waves
of migration are coincident with high relative humidity and with a
temperature of from 41° to 52° F. The migration is independent of
daylight. All of Cummins' illuminating observations still give no
information as to why the frogs congregate in a given pond or how
they learn of its existence. He does record that migration routes are
not direct, so that we may assume that we are dealing, at least in
part, with random movements, probably controlled largely by tem-
perature. Blan chard (1930) concludes that the external control for
the breeding migration is to be found in rainfall rather than in tem-
perature relations.

During the breeding season a gregariousness appears among frogs
which does not exist under usual circumstances. This is not entirely
accounted for by the tendency which the animals exhibit to seek a
similar habitat for breeding, for if there are only a few pairs of frogs
in a given place, they force themselves together as closely as possible
and the eggs form a continuous mass.

At the height of the breeding season several males will struggle for
the possession of a single female (Banta, 1914) ; the struggles attract

68 ANIMAL AGGREGATIONS

other males, and one female may become the center of a struggling
mass. One such group which Banta caught had 6 males fastened
together about a single female and 5 others nearby but not yet at-
tached. The actual egg-laying and fertilization of the eggs is ac-
companied by the formation of a close aggregation (Fischer-Sigwart,
1897). In addition to the male that has been in copulo for some time,
these supernumerary males gather and, despite kicks from the first
male, still manage to form a close clump. In Rana fusca one may
find single pairs, but as a rule fertilization is a community matter.
Supernumerary males also crawl over and among the egg masses and
effect the fertilization of ova which may not have been reached by
spermatozoa at the time of their discharge.

At the close of the breeding season frogs scatter and resume a
solitary, non-social existence.

Fish. — Similar breeding clusters of fish have been described by
Reeves (1907) with many identical details. With the rainbow darter
supernumerary males crowd about the spawning pair and appear
also to shed spermatozoa. Reighard (1903) has seen such behavior;
but in the main his studies (1903, 191 5, 1920) emphasize the orderly
spacing of breeding holdings in lish, a phase of the aggregation
phenomenon with which the present summary is not greatly con-
cerned. The close contact between males and females of fresh-water
animals with external fertilization is made necessary by the extreme-
ly short life of the gametes shed into fresh water. Reighard has
stated that fish sperm can remain functional for less than a minute
under these conditions.

Snakes. — Snakes are reported to form bunches in the breeding
season similar to those described for frogs, except that they occur out
of the water (Ditmars, 1907; Ellicott, 1880; Ruthven, 1908). EUi-
cott records: ''I first saw such a bunch of snakes on the stony banks
of the Patapsco River, heaped together on a rock and between big
stones. It was a warm and sunny location where a human being
could scarcely disturb them. I reasoned that the warmth and the
quiet of that secluded space had brought them together. Some hun-
dreds could be counted, and all in a very lively state of humor,
hissing at me with threatening glances and with such persistency

GENERAL FACTORS CONDITIONING AGGREGATIONS 69

that stones thrown at them could not stop them nor alter the posi-
tion of a single animal. They would make the proper movements and
the stone would roll off; all the snakes in this lump were common
garter snakes {Eutaemia sirtalis L.).

"The second time I noticed a ball of black snakes rolling slowly
down a steep hillside on the bank of the same river. Some of the
snakes were of considerable length and thickness and as I noticed
clearly, kept together by procreative impulses."

Lunar periodicities. -Such breeding aggregations are much more
important in fresh-water and land forms, with whom the surround-
ings are more injurious to shed sperm or eggs, but they do occur
among marine animals. With marine organisms the most spectacular
expression of breeding aggregations is to be found in the case of the
large number of animals whose breeding rhythms coincide to some
extent with lunar periodicities. The Hterature on this subject is ex-
tensive (Woodworth, 1907; Fox, 1923; Legendre, 1925; B. H. Grave,
1922, 1927); but while the facts are plain enough, the fundamental
causal relations remain unknown. One illustration must suffice,
based on the account given by Lillie and Just (191 3) for the swarm-
ing of the sea worm Nereis limbata in waters around Woods Hole.

Nereis limbata has its swarming period only after twilight. Each
run begins near the time of the full moon, increases to a maximum
during successive nights, and sinks to a low point about the time of
the third quarter, again rising and falling to extinction shortly after
the new moon. They appear in four periods or cycles during the
summer, corresponding to the lunar cycles in the months of June,
July, August, and September.

Only fully mature animals swarm. The swarming begins shortly
after twilight and lasts for only an hour or so. The swarming animals
are attracted by the light of a lantern. Males appear first, darting
through the water in curved paths in and out of the circle of the
hght. Females are fewer in number and swim more slowly. The
males outnumber the females hundreds to dozens. In the next few
minutes the numbers increase, waning again after about three-
quarters of an hour.

New females appear each night, but some males may presumably

70 ANIMAL AGGREGATIONS

reappear on several successive nights. A swarming female is soon
surrounded by several males. These swim rapidly in narrow circles
about her. In a little while they begin to shed sperm^ probably in
reaction to some secretion from the female, rendering the water
milky. Soon the female begins to shed her eggs, shrinking in bulk as
she does so, until, a shadow of her former self, she sinks through the
water to die. Lillie and Just, following a lead from Hempelmann
(191 1), assume that the maturing of the animals is dependent on
some relation of the life-history to the phases of the moon, involving,
probably, through lunar tidal variations, rhythmical alterations of
conditions of nutrition.

HIBERNATION

Over- wintering aggregations of animals have long been known.
This phenomenon in social bees has been noted in scientific litera-
ture for almost two hundred years (Reaumur, 1734-42). Barkow
(1846), in his monograph on hibernation written over three-quarters
of a century ago, has a short chapter in which he calls attention to
the winter aggregations of lepidopterous larvae, adult ants, bees,
true bugs, beetles, including the frequently observed case of the
coccinellid beetles, carp and the eel-like Muraena anguilla, snakes,
frogs, and a few mammals, including marmots and bats. Barkow
advances no theory to account for the congregation of these animals
but does state that there is a suggestion current that the animals
come together as a result of response to their sense of smell.

This list of over-wintering aggregations has since been much ex-
tended, especially by Holmquist (1926), who has made extensive
studies on hibernating arthropods in the Chicago region. He reports
that of 329 identified species taken during the winter season, nearly
17 per cent were more or less closely aggregated. Omitting those
known to be of a somewhat social habit at other times of the year,
about 9 per cent of the species ordinarily solitary in the summer were
aggregated in winter.

In the social bees careful experiments have shown that tempera-
ture-control results from such clusters (PhilHps and Demuth, 1914;
Phillips, 191 7); and Holmquist (1928) has demonstrated that protec-

GENERAL FACTORS CONDITIONING AGGREGATIONS 71

tion from flooding, and other benefits, may accrue from the cluster
formation of hibernating ants.

In many cases these over-wintering groups are essentially shelter
aggregations, apparently due to the small amount of serviceable
shelter available. Often, however, all the apparently equally desir-
able space is not occupied, so that the aggregation cannot be entirely
explained on the basis of unavoidable crowding. In other cases
Holmquist has been unable to find any environmental differences to
account for the location of the hibernating aggregation. These
groupings are partially under temperature control; but, as with other
phenomena connected with hibernation, the temperature control is
incomplete, and the problem of the exact nature of the causal factors
remains open,

AESTIVATION

Aestivating aggregations have been less studied. Land isopods
will form aestivating groups which may be either homotypic or
heterotypic. Dr. C. H. Abbott has informed me personally that they
collect in large numbers in protected places, and so pass the long,
hot, dry summer of southern California.

AGGREGATIONS CONTROLLED BY MOISTURE

The chief controls of the aestivation reaction of these isopods are
temperature and moisture. Of the two, laboratory experiments show
the latter to be more important (Allee, 1926). When land isopods of
various species are placed on air-dry filter paper, they collect in
bunches within a few minutes, unless the substratum is too dry,
when they will run about actively until at the point of death. If the
substratum is moist, the same isopods will remain quietly scattered.

These relations are shown in Figure 3. In the upper picture there
are 25 isopods in a crystallization dish photographed 30 minutes
after being introduced into the cUsh, which had the bottom covered
with dry filter paper. In the meantime they were in a darkened
room and the exposure was by flashlight. The lower photograph
shows the effect of adding enough water to make the filter paper
thoroughly moist without being sloppily wet. The same animals are
shown as in the preceding photograph, but 15 minutes later, and 5

72

ANIMAL AGGREGATIONS

minutes after the background was moistened. The animals not
shown in this photograph have crawled up the sides of the dish.

A somewhat similar effect of drought in nature is reported for the
California quail (Evermann, 1901). In an unusually dry season these

quail do not breed but remain in
flocks during the entire summer.
The opposite type of moisture
control is also observable. Too
much moisture may produce
well-defined aggregations. Thus
Solenopsis geminata (von Ihering,
1894), a species of ant which
often nests in lowlands, will, if
the nest is flooded, aggregate in
a ball of some 15-20 cm. in di-
ameter, with the larvae and
pupae inside. By constant rota-
tion they avoid too long sub-
mergence, and at length may
come against some solid object
and so escape from the water.
Wheeler (1913a) cites this case
and mentions similar instances
in this and other species of ants.
The formation of the dancing
bunches of midges already men-
tioned, which one frequently
sees aggregated in the space of
a half -bushel basket, appear to
be in part conditioned by the at-
mospheric humidity, although
the absence of wind is another
obvious prerequisite.
In both these cases the environmental conditions are uniform; and
the animals, in grouping together, react to each other only. There
are also the place aggregations controlled by moisture, when animals

Fig. 3. — (i) Land isopods in darkened
room on dry background of filter paper. (2)
Same animals and conditions as in (i) except
that the filter paper has been moistened.

GENERAL FACTORS CONDITIONING AGGREGATIONS 73

will collect in a limited area because it provides an oasis of moisture
or of dryness in an otherwise overdry or overwet environment. Thus
land isopods can be made to collect at will in a given spot by making
it moist. Selous (1907) gives a striking picture of the congregating of
large ungulates about an African drinking-hole in the dry season.
The common fruit fly, Drosophila, struggling to escape too great
moisture, aggregates in shifting masses at the top of a projection;
these masses continually fall apart and re-form as the flies move up
again. Under optimal conditions all of these move out of contact
with their fellows.

LACK OF NORMAL ENVIRONMENT

The snake starfish, Ophioderma, lives in eelgrass in certain loca-
tions along our eastern coast. Repeated attempts have failed to
find this animal in contact with others of its own kind in nature
during the summer (Aflee, 1927). They are often found near together
but never aggregated.

Ten of these starfish were introduced into a laboratory aquarium
made to approach normal living conditions by the introduction of
eelgrass. Nineteen hours later 7 of the 10 animals were sighted after
a search lasting half an hour. One was found on the bottom at the
side away from the strongest light ; 6 animals were in the densest part
of the vegetation in the same region; and, although not in immediate
contact, all of them could probably have been inclosed in a 5-inch
cube. The exact location of the other 3 animals could not be ob-
served without disturbing them. These animals in the field may
also be close together without actually touching. Only such loose
collections were ever seen in this eelgrass aquarium. Extended ex-
perience with these animals in the laboratory leads me to conclude
that the tendency to bunch is greatly reduced in proportion as
favorable natural conditions are approximated, and that the animals
so congregated are usually found in regions to which they have been
directed by their tropistic reactions.

When, however, Ophioderma are placed as they are collected in a
glass or similar container, they form dense mats of bunched animals
with arms closely interwoven. The aggregations form in the shadiest

74 ANIMAL AGGREGATIONS

part of the dish and are to be explained in part by the fact that the
lower animals are shaded by the upper ones, and so, having satisfied
a negative phototropism and a positive thigmotropism, they remain
quiet.

The position of the arms shows the strong thigmotropic reaction
of these animals. In the recently formed bunches there are a larger
number of free arms than in older aggregations ; at first the arms tend
to extend out and up into the water. They may be entirely free, or
they may touch another arm only where the two cross. Even in the
early stages of the bunching some of the arms lie nearly parallel with
each other. In bunches of longer duration there are practically no
free arms. In one case I saw two starfish with four pairs of arms
paralleling each other, and only two free. Larger bunches become
ropelike masses composed of parallel arms or of arms intertwined
like basketry. In these older aggregations the arms of the animals, at
first extending freely, are turned back and interwoven with the
others so that the outer edge presents a relatively regular line. When
these starfish are isolated and left for a week or more in separate
dishes exposed to light, frequently the arms are moved into contact
until they present a sort of self-bunching.

Laboratory aggregations occur in a large number of animals.
May-fly nymphs, various isopods, earthworms, frogs, and others
may readily be observed to form such bunches. The behavior ap-
pears due to similar causes to that which results in the collection
of foreigners into communities of their own nationality in our large
cities; that is, a group of similar animals tend to minimize for each
other the disturbing effects of unusual surroundings.

"sleep" aggregations

The sleep aggregations of insects have been relatively little written
about, even in research journals; so it seems important to bring to-
gether a more extended summary concerning such slumber aggrega-
tions than is needed for the better-known overnight assemblies of
birds.

Fabre (191 5) found some hundreds of the wasp Ammophila (Sphex)
hirsuta assembled under the shelter of a stone on the mountain side,

GENERAL FACTORS CONDITIONING AGGREGATIONS 75

and speculated much concerning this gregarious condition of a soli-
tary wasp. The Raus (191 6) found three related species sleeping in
such assemblies, from which it would seem probable that Fabre was
observing a slumber aggregation. With Chalyhion caerulum both
males and females may be found aggregated at night in about equal
proportions. As many as a thousand have been found in one colony.
Marked individuals will return to the same sleeping place for at least
2 weeks. No one knows how the male of the species passes the day;
the female labors about the nest.

The solitary Sphcx wasps appear to choose their sleeping quarters
independently; but since they select the same sort of place, they
tend to form spaced aggregations. Prionyx sleeps sometimes singly;
sometimes gregariously crowded close together on the top of a weed,
with equal numbers of males and females present but without ob-
served copulation. The males and females of the horse fly Tabanus
sulcijrons are reported also to collect in favorable places to sleep
(Hine, 1906). Similar observations are on record for various other
insects.

There is no evident protection from enemies in such assemblies.
The sleep may be sound, and may extend so late that early birds
could pick off the sleeping insects in numbers, as beetles are reported
to kill off sleeping butterflies (Floerscheims, 1906).

Schrittky (1922) observed in Paraguay an aggregation of from 20
to 27 butterflies (genus Heliconus) that gathered nightly during
August and September. The butterflies could be hancUed in the early
mornings without waking them. The temperature then ranged
around 5° C. The butterflies were quite restless in the evening long
after dark, when the temperature was higher than in the morning.
He also observed males of the genus Tetrapedia in aggregations at
night; females are found in temporary aggregations until the time
of fertilization, after which they separate.

Banks (1902) gives observations on males of the solitary digger
bees of the genus Melissodes. He saw these bees at dusk in his back
yard clinging with mandibles and feet to grass blades. He records
three or four returning for several nights. He cites the record of
Schwarz (1896) to show that Melissodes pygmeus clasp twigs with

76 ANIMAL AGGREGATIONS

their mandibles. Bradley also records finding Melissodes agilis cling-
ing on dried blades of wild oats alongside a newly cut grain field. In
this same patch were large aggregations of a number of species of
wasps — no two species on the same blade. He accounts for the aggre-
gations of wasps as being caused by the cutting of nearby grain.
Boyer and Buchsbaum of this laboratory, from their unpublished
observations, think that the Melissodes which Bradley found ag-
gregated were present because of the cutting of the flowers on which
they ordinarily collect.

Von Frisch (1918) gives observations on 6 solitary male bees,
Halictus, which returned for 4 days to the same dry stem of a plant.
He records that the bees would return to this plant if the weather
grew bad or if the temperature became low, even during daylight.
For 4 evenings after his original observation he observed 5 bees pres-
ent on the same stem, although there were other similar stems near-
by. One bee had been taken for identification. He cannot be sure
that the same five returned each evening.

The Raus have several notes on bees. Concerning Melissodes ob-
ligua they s,a.y: ". . . . We found the twenty-eight bees clustered near
the tops of a small clump of stalks. Since it was now almost dark my
presence did not disturb them. They were huddled together in
groups of two to five, with only three insects occupying their sites
singly.

"The next evening twenty-nine bees, only one more, were asleep
on these five stems, all clustered on the apical three inches of the
dead plants. At the top of another plant ten feet away, two were at
rest. If they had chosen this site for protection alone they would
have rested singly on the plants, but since they huddled in groups
they must have sought sociability also. They were so close together
in some cases as to arouse suspicion about their mating, but a close
examination proved the idea false.

''The following night, July 21, twenty-four of these bees were here
to spend the night in the same way. On the 22nd, thirty were pres-
ent. On this evening I marked part of them with white paint

As fate would have it, the next evening a cow had broken down their
chosen stems, so none of the bees were there. However, fifteen were
found on similar weeds nearby; seven of these bore the white mark-

GENERAL FACTORS CONDITIONING AGGREGATIONS 77

ings. This gave evidence sufficient to prove that the same bees re-
turn to their chosen spot regularly .... all were males."

Here Boyer and Buchsbaum took up the problem, using the soli-
tary digger bee, Melissodes agilis or aurigensia, which Professor
Cockerell in a personal communication says are variants of the same
species. They found Melissodes active in the field only on warm
sunny days, with the temperature 16° C. or above, depending on the
light. When it is cloudy or cool, the bees remain inactive on the sun-
flowers Helianthus annuus and petiolaris, which they frequent. The
male bees usually arrange themselves so that two are in contact when
two or more become inactive on the same blossom. Several groups
of two's have been found on the same flower, isolated from each other.

The bees are invariably inactive between twilight and sunrise.
The beginning of activity depends on the amount of light and on the
temperature. Controlled laboratory experiments showed that in
bright sunlight activity started under stimulation at about 7° C,
while in dim light the first activity came at 9° C. Similarly, spon-
taneous activity began at 18° C. in the sunlight and at 21° in dim
hght.

Aggregations of males at night were recorded as follows for one
particular group of sunflowers:

Year

Singly

Pairs

Groups of 3 Groups of 4 Groups of 5

Groups of 6

1927.
1928.

100
39

31

5

Boyer and Buchsbaum marked some of these bees with paints of
different colors so that they could follow individual reactions. There
were some 500 flowers in this particular group. Each of these was
plotted and followed night after night for its bee population. In all,
201 bees were observed in 1927. Thirty-four of these were success-
fully painted. Of the 20 painted bees which were seen again, 10 bees
returned to the flower they occupied when painted. Eight others
returned to the same plant but to a different flower. Fourteen re-
turned at least twice to the same plant but to a different flower from
that on which they had been painted. In all, 37 returns were noted
to the same flower or to a flower within 10 feet of the original one.

78 ANIMAL AGGREGATIONS

while 24 returns were noted to some flower more than 10 feet away
from the original.

In 1928, 14 bees were successfully painted. Of these only 4 were
seen again; and of these, two returned to the same flower where they
were painted and the other two returned to a nearby flower. A study
of the details of the observations shows that males of Melissodes
frequently return to the same flower night after night or in cool or
cloudy weather. They are generally found in the same vicinity on
successive nights, even if not on the same flower. They must neces-
sarily return to a different flower if the one on which they have been
staying is destroyed or dries up. No bees were observed on withered
flowers. If they are blown to a distant part of the sunflower patch,
they tend to remain there in a narrowly circumscribed area for the
next several days.

It must be noted that these overnight aggregations in Melissodes
were composed of males only.' They cannot have sexual significance.
It seems entirely possible that we are concerned here with an in-
cipient social habit which does not extend to many solitary species
and is not found in all individuals of the species in which it occurs.

Swarming locusts of several different species are known to pass the
night in dense masses both as nymphs and later when they become
adults. Much of this literature is reviewed by Uvarov (1928). Re-
garding the overnight aggregations of these locusts, Uvarov says:
"The night is passed on plants in dense bands, which are extremely
conspicuous on the background of the vegetation owing to their
blackish general color ; during the night the hoppers are in a state of
torpor caused by the cold." If the day is cool, the slumber bands do
not break up as they do on warm sunshiny days. Even when con-
siderable numbers of the South African locust, Locustana pardalina,
have become adult, they collect at night near the main nymph
swarm, although they may range at considerable distance during the
day. The night clusters of the flying adults are not so dense as those
of the hoppers.

Faure (1923), in describing the night collecting of nymphs, says
they gather slowly together into fairly dense masses, forming clusters

' J. F. W. Pearson has taken 3 female Melissodes and 90 males from early morn-
ing collecting on Helianthus in this locality.

GENERAL FACTORS CONDITIONING AGGREGATIONS 79

that closely resemble close masses of bees. They swarm on the tall
grass, or, if this is lacking, they pack together in the low grass, on
stones or on the ground. At sunrise the swarm gradually breaks and
continues its migration. These night clusters are conspicuous objects
showing up as reddish-brown patches on the veldt. Man and pre-
sumably other animals take advantage of these aggregations to de-
stroy great numbers of the grasshoppers. The benefits accruing are
not known. Nikolsky (1925, vide Uvarov, 1928) thinks that they
conserve animal heat. Holmquist's observations on mass collection
of ants (1928a) suggest that they may at least slow down the rate of
change of temperature.

The congregation of birds for sleeping has been widely observed
(Brewster, 1890; Davis, 1894; Bates, 1895; Widman, 1898, 1922;
Allen, 1925), particularly for martins, robins, grackles, and crows.
Many other birds are reported to gather in the roosts dominated by
martins and robins. Extreme cases of close crowding in these roosts
are reported by Baker (vide Allen, 1925) for the crested tree swift of
India.

"On arriving at their proposed meeting place," Baker says, "they
fly round and round, gradually lowering their flight until one bird
makes a sweep and settles on some part of the tree near the top. This
is the signal for the rest to perch, and in a few minutes they are all
dotted about the higher branches. They then begin to close up with
the bird which first ahghted on the tree, finally collecting in a feath-
ery ball, one on top of the other. Sometimes this happens again and
again before they get settled, but at last the twittering stops and
they are asleep for the night. It is wonderful how compactly these
birds close up; a flock of eleven appeared not to take more than a
foot long by half that breadth."

The Indian swallow shrike is said on the same authority to have a
similar habit. Sharp records that the colonies of mouse birds of
Africa, small birds resembhng parrots, roost in small parties that
cling together.

It is well known that bats also gather into sleeping aggregations
(Goldman, 1920; Howell, 1920; Allen, 1921). They may congregate
in clusters comprising only a few individuals, or hundreds may hang
with bodies touching. The groups may be homotypic or heterotypic.

8o ANIMAL AGGREGATIONS

To the human senses these bird and bat roosts are easily detected
by their odor, and perhaps that is a factor in guiding the bats to the
common sleeping place.

Allen (192 1 ) has banded clusters of these bats. He records re-
covering three of a group of four from the same place where they
were banded, after an interval of three years.

These sleeping aggregations appear to be without mating signifi-
cance. The Raus did not see copulation among the insects they ob-
served; and, in fact, in many cases the sleeping groups were com-
posed of males only. The robin roosts may contain both sexes and
all ages of birds above the nestlings. With crows the common roost
ends with the beginning of the breeding season, except for the bache-
lors; and in general these roosts are not occupied by the breeding
birds. After the breeding season the birds may return in family
groups, a situation to be discussed later at some length. Among bats
the sexes are segregated (Howell, 1920) during the time of gestation
and of the care of the young, at a time when contact sleeping aggre-
gations were observed.

At the low level of integration of aggregations, with which we are
especially concerned, the appearance of social appetite is an inter-
mittent phenomenon. It may be awakened by gonadal activities
that precede the breeding season or by the conditions which induce
hibernation or aestivation. These varying exhibitions of a stronger
social appetite are ordinarily part of an annual rhythm, but in many
marine forms the rhythm may be a lunar one during the warmer
season of the year. In the slumber aggregations, the periodic
strengthening of the social appetite has a diurnal rhythm. Aggrega-
tions may be induced or controlled by conditions of moisture or by
the lack of normally favorable conditions ; this phenomenon may or
may not be rhythmical in nature. At this low level of social integration
the social appetite is not constant in appearance and in this regard
becomes more like the sex and hunger appetites, in which rhyth-
mical or spasmodic appearance is one of the usual characteristics.
In more highly integrated social groups the action of the social
appetite is steadier, and therefore less spectacular and less easily
recognized.

CHAPTER V
INTEGRATION OF AGGREGATIONS

THE COMMUNITY LEVEL OF INTEGRATION

It is instructive to regard an animal as a physiological system of
physicochemical processes in dynamic equilibrium. When this is
understood, one is prepared for the definition of an "animal society"
or an "ecological animal community" as a system of organisms which
is in the process of dynamic equilibration.

In the case of the animal considered as an organism, the different
parts are integrated more or less perfectly into a unit, which has
been receiving considerable attention in the last decade in studies on
the organism as a whole as contrasted with the study of different
parts of organisms. One can readily see that there are highly inte-
grated organisms under close control of the nervous system or of
hormones, the loss of any major part of which will strongly affect the
whole system and frequently will cause death; but, on the other
hand, there are the lower organisms much more loosely correlated,
where the loss of even a major part of the body causes only tempo-
rary inconvenience pending the regeneration of replacement tissues.
Many of these more loosely organized animals are so poorly inte-
grated that different parts may be in active opposition to each other.
Thus, when an ordinary starfish is placed on its back, part of the
arms may attempt to turn the animal in one direction, while others
work to turn it in the opposite way. With sponges, the pores ad-
mitting water to the canal system may be open and the flagella
engaged in pumping water into the canals, while the ostia remain
closed so that no water can be brought in (Parker, 191 9). On ac-
count of its loose integration, the sea anemone may move off and
leave a portion of its foot clinging tightly to a rock, so that the ani-
mal suffers serious rupture.

It is to such relatively slack systems that an ecological animal

81

82 ANIMAL AGGREGATIONS

community is to be compared, rather than to the highly integrated
ant or bird or man. In human society we are accustomed to the idea
of community integration. Thus a village is composed of a number
of families which are connected as a unit not only because they oc-
cupy a limited amount of contiguous space but also because they
are bound together by social organizations such as church and
school, by economic relationships of kinship or of marriage; all of
these knit the community into a working unit. The organization
is loose. Individuals may come and go. Whole famihes may depart
and others move into the village, and yet the village retains a defi-
nite unity, with a more or less marked individuality which may be
quite distinct from that of neighboring communities.

In such a community as the village, men are associated not only
with each other but with other animals. There are the horses that
supply part of the draft power; cattle that give meat and milk;
dogs and cats that provide companionship and amusement to man,
feed on his surplus food or on other associated animals, add to the
dirt of his household and scatter bacteria and parasites; flies that
feed on the refuse of man and breed in the excreta of his commen-
sals; mosquitoes that breed in water reservoirs and feed on man
and other animals; birds attracted by the nesting sites and food to
be found near man; rats and mice similarly attracted, and snakes
attracted by the birds and rats and mice; insects that prey on gar-
dens and orchards, and insects that prey upon these; as well as
other animals with little direct relationship to the community but
occupying the same general space.

If a progressive town board decides to instal a hydroelectric plant,
the river is dammed and a breeding place is furnished for thousands
of mosquitoes; if some of these are Anopheles, the malarial parasite
may become prevalent. The breeding range of fish and of pond in-
sects is extended at the same time that the human population is ad-
justing to the use of cheap water power ; the dam is a matter of con-
cern to the whole animal community.

The consequences of an unusually mild winter ramify also through
the entire community. One result is that many insects live over
which would ordinarily be winter-killed. These attack orchard, gar-

INTEGRATION OF AGGREGATIONS 83

den, and farm, affecting the food of grain- and fruit-eating birds and
mammals and of man himself. These are finally checked by the sub-
sequent increase in predaceous insects and birds that live on garden
and orchard pests, and the rough biotic balance characteristic of
animal and plant communities in nature thus tends to be restored.

These instances are enough to illustrate the interdependence and
general type of organization of an animal community of which man
is a dominant element. The removal or introduction of animals,
whether by accident or by purposive action by man, may upset the
whole equilibrium, as has happened with the introduction of rabbits
into Australia.

A similar organization has long been recognized to exist in animal
communities of which man is only a minor part, or perhaps no part
at all. This type of organization has been called by J. Arthur Thomp-
son the "web of life." Frequently the food relationships are the most
easily demonstrated in a group of this kind. A partial idea of the
complexity of such organization is given by a consideration of the
food relations of the black bass as summarized from Forbes' excel-
lent essay on "The Lake as a Microcosm."

INTEGRATION OF THE BLACK-BASS COMMUNITY

The organization of animal communities is more marked in the
case of inhabitants of small bodies of water than of equal bodies of
land, since conditions tend to isolate such aquatic animals and since,
through long evolution, they have become closely integrated and
highly independent of the newer societies of the land. The hfe in
such a body of water represents an islet of older, lower life in the
midst of the higher, more recent life of the surrounding region. It
forms a microcosm, a little world in itself. The play of Hfe is full, but
on a smaller scale and less confusing to observe.

In such a community one can see fully illustrated the degree of
sensitivity characteristic of an organic complex, which has just been
demonstrated for a man-dominated community. Whatever affects
one species must have its influence on the whole assemblage. It thus
becomes apparent that it is impossible to study any one animal
completely if it be out of its relations to other animals and to plants.

84 ANIMAL AGGREGATIONS

even though the animal selected for study belongs to what is usually
regarded as a non-social species.

It is relatively easy, though sufficiently exciting to be called sport,
given the right body of water and the proper season and bait, to lift a
large-mouthed black bass from the water; but if one should under-
take to trace out all the interrelations from which the black bass has
suddenly been removed, he will have seen the whole complicated
mechanism of the aquatic life of the locality, both plant and animal,
of which the black bass forms a part.

In the food of the black bass are to be found fishes of different
species at different ages of the individual, representing all the im-
portant orders of the fishes; insects in considerable number, especial-
ly the various water bugs and larvae of the May flies; fresh-water
crayfishes, shrimps, and a multitude of the small crustaceans called
Entomostraca, of many genera and species.

Looking at the food of the fishes upon which the black bass feeds,
one finds that one of these eats mud, algae and Entomostraca, and
another takes nearly every animal substance in the water, including
mollusks and decomposing organic matter. The crayfishes are nearly
omnivorous; of the other Crustacea, some eat Entomostraca and
some algae and Protozoa. The insects eaten by the bass eat each
other, other insects, and Entomostraca. At only the second step,
therefore, do we find the black bass directly related to every class
of animals, many plants, and the decaying vegetal matter of the
water.

Turning now to competitors, which are extremely numerous, we
find that all the young fishes, except the suckers, feed at first almost
wholly on Entomostraca, so that the young black bass finds himself
at the very beginning engaged in a scramble with almost all the
other fishes in the lake for food and, in fact, not only with the fishes
but with the insects and mollusks and larger crustaceans that also
live on these small entomostracans. The Mollusca are not in such
direct competition; but they do compete, since they feed upon the
microplankton which the Entomostraca themselves take as food.

But the competitors of the bass are not limited to those which
take the same food, for predaceous fishes, turtles, water snakes,

INTEGRATION OF AGGREGATIONS 85

wading and diving birds, and the large beetles, dragon-fly nymphs,
and giant water bugs feed on the young bass at every opportunity.

An illustration of remote and unsuspected rivalries is found in the
relation of the black bass to the bladderwort {U tricularia) , which fills
many acres of the northern Illinois lakes. Upon the leaves of this
plant are small bladders, several hundred to the plant, which are
tiny traps for the capture of entomostracans and other minute ani-
mals. The plant usually has no roots and lives largely on the animals
taken through these bladders. Ten of these sacs, taken at random,
upon examination gave 93 animals of 26 different species, of the
Entomostraca and insect larvae. Hence, the bladderwort competes
with the fishes for food and, by destroying large amounts, helps keep
down the number of black bass in an otherwise favorable lake; and
they have an especial advantage since, when the Entomostraca be-
come scarce, they may grow roots and live as other plants.

These simple instances sufhce to illustrate the intimate way in
which the living forms of a lake are united.

A different phase of the story is shown by the study of fluviatile
prairie lakes which are appendages of river systems and form in
oxbow cut-offs or bayous, or in other regions where the usual deposi-
tion of materials has been retarded. Normally they are connected
with each other during the rainy period and for a longer or shorter
time during the summer. The amount and variation of animal Ufe
in them is dependent chiefly upon the frequency, extent, and dura-
tion of the overflows. In them we may see illustrated the method by
which the flexible system of the animal community adjusts itself to
widely and rapidly fluctuating conditions.

Whenever the waters of a river remain for a long time outside its
banks, the breeding grounds of the fishes and other animals are cor-
respondingly extended. The slow and stagnant waters of such an
overflow, frequently enriched by sewage to a limited extent, form
the best possible place for the growth of myriads of algae and Pro-
tozoa. This development allows a similarly great development of
Entomostraca. These animals increase with tremendous rapidity due
to the pace at which their life-circle is run and to their high rate of
reproduction. The sudden development of food resources allows a

86 ANIMAL AGGREGATIONS

corresponding increase in the rapidly breeding, non-predaceous
fishes; and at last the game fishes which derive their principal food
from the non-predaceous fishes also increase in numbers. Evidently
the multiplication of each of these classes acts as a check on the one
preceding it. The development of Protozoa and algae is arrested and
sent below normal by the swarm of entomostracans; the latter are
met and checked by the vast swarm of minnows, which are in turn
checked by the increase in predaceous fishes. In this way a gradual
readjustment of the conditions will occur; but usually, long before
this new equilibrium is reached, a new disturbance of the water level
results in the recession of the water. As the lakes grow smaller and
the teeming life they inclose is daily restricted within narrower and
narrower bounds, a fearful slaughter ensues. The predaceous fishes
thrive for a time, since their food is more easily caught; but finally
they too are thinned out by the lack of food and of space.

Year after year in such lakes and in other animal communities
there is a fairly steady balance of organic life. The community re-
mains in dynamic equilibrium. The rate of reproduction about
equals the death-rate. Every species must fight its way from hatch-
ing to maturity. Adults are as rare as human centenarians; yet no
species is exterminated, and each is maintained at the average num-
ber, for which we have reason to think there is sufficient food year
after year. Two ideas explain the order that is evolved in such com-
munities. First, there is the background of common interests among
all elements of the community. New evidence concerning the nature
of some of these common interests will be presented shortly. Second,
there is the struggle for existence and the elimination not only of the
less fortunate but, at times, of the less fit animals.

Upon such a foundation as this, modern comparative sociology is
built in part, and must be built in entirety if it is to be sohdly ground-
ed. With this conception of the type of integration existing in eco-
logical animal communities, and with the realization that even such
loosely knit communities can be regarded as constituting a unit, we
are better prepared to search for integrations in animal aggregations
and to evaluate those found.

INTEGRATION OF AGGREGATIONS 87

AGGREGATION INTEGRATION

As has been said before, a decided advance toward social life is
made by the appearance of tolerance for other animals in a limited
space, where they have collected as a result of random movements or
of tropistic reactions to their environment. This may occur in con-
nection with some phase of breeding activity, but it may also be
exhibited without sexual significance. Some of the less complex of
these aggregations may exist because there is an absence of dis-
sociating factors among a group of animals that have been hatched
out in a restricted locality or that have been brought together by
any other process. Thus, some of the aggregations resulting from
tropistic responses may well owe whatever permanency they possess
to the absence of disruptive factors rather than to any inherent gre-
garious tendency or apparent advantage.

Another advance in social life is made when these groupings con-
fer especial survival values upon at least some of the individuals
composing them. Such an advantage is illustrated by the slower
rate of moisture change in an aggregation of land isopods out of
water equilibrium with the surroundings. Under conditions of
drought this results in a definite prolongation of life for the members
of a group. Other examples will be discussed later.

The land isopods and Ophioderma have gone little beyond such a
stage in their social development. There is some slight evidence of
mutual attraction, but the experiments to date do not indicate how
much of this would also be exhibited toward similar inanimate ob-
jects. There is also slight evidence of integrated group behavior, in
that the bunch shows occasional periods of activity apparently
originating in one individual and passed mechanically through the
group. Such activity may be the beginning of disintegration of the
group; but it frequently results in a closer aggregation, because the
animals may move closer together during their brief period of ac-
tivity.

The state of development of integration by means of which the
group, once it appears, acts as a unit is a very important criterion of
the degree of social development it has attained. When there is no

88 ANIMAL AGGREGATIONS

integrative action, one is dealing with a crowd, a mere collection of
individuals within a limited area. Apparently it was this aspect that
Szymanski (19 13) had in mind in distinguishing between primary
reactions, the reactions of the individual, and secondary reactions,
the reactions of the individuals as members of a group.

TACTILE INTEGRATION

The simplest form of group organization is found when animals in
physical contact respond as a group to touch stimuli passed from one
to another. Such organization may be sufHciently refined for the
whole group to show definitely synchronous behavior. Collections of
Liobumi7?i, the harvestman, have been observed by Newman (191 7),
and later by myself, to give such reaction. One group was found by
Newman resting on the under side of an overhanging shelf of rock on
a steep hillside. The harvestmen were closely packed together within
an area of about 5 feet in diameter. When first seen, they were hang-
ing from the rock roof perfectly motionless. When the observer came
nearer, they began a rhythmic stationary dance practically in uni-
son. This died down shortly but could be started again by appro-
priate mechanical stimulation.

When the colony was first seen, the long legs of neighboring in-
dividuals were interlocked, which would sufficiently account for the
transmission of stimuli through the group. It should be noted, since
we are interested in the state of integration of the aggregation, that
the rhythm was not perfectly synchronous at the beginning but be-
came practically so after a few seconds.

Such integration, due to tactile transmission, must be present to
some degree in all cases of aggregation in close physical contact. It
is highly developed in the sleeping groups of bats (Allen, 1921),
which may hang in compact clusters, as already mentioned. If one is
touched, the whole cluster may drop. Allen caught eighteen by hold-
ing an insect net under the group and touching only one of the outer
bats.

The effect of physical contact in establishing synchrony in two
reacting systems is illustrated by the observation of Fischer (1924),
who found that two pieces of embryonic heart planted out in tissue-

INTEGRATION OF AGGREGATIONS 89

culture media beat at different rhythms even when taken from the
same individual and kept as far as possible under identical cultural
conditions. When two such pieces succeeded, by outgrowths from
each, in estabHshing close organic union, the two beat in unison.
Such a modification of behavior may involve factors of transmission
distinct from those we usually regard as tactile.

CONTACT AND ODOR INTEGRATION

Sex recognition frequently causes animals to give characteristic
group reactions; often there are only two animals forming a diminu-
tive group. Sex recognition is frequently accomplished by contact
relations alone. Such behavior is recorded for crayfishes (Pearse,
1909; Andrews, 1910), spiders (Montgomery, 1910), frogs (Banta,
1 91 4), amphipods (Holmes, 1903), as well as others.

Among other methods of sex recognition, that due to chemical
sense deserves prominent mention. This is well illustrated by the
long distances certain male moths will fly to cluster about a female
ready to copulate (Kennedy, 1927). Animals may aggregate at other
times than the breeding season, due to the same sort of stimulus;
and this stimulus is also frequently eft'ective in maintaining the
aggregations once formed. In fact, it is common for the principal
stimulus causing animals to congregate to be the effective one in in-
tegrating their aggregation.

These two senses, odor and contact, are sufficient means of group
integration to form the basis of well-unified societies. Much of the
social organization of the ants and the termites appears to be based
on them. The ants apparently live in a world of contact-odor shapes,
as we live primarily in a world of color-shapes (Wheeler, 1913a).

VISUAL INTEGRATION

Sight plays an important role in the organization of many animal
groups. When one vulture, soaring aloft, sees another swoop miles
away, he moves over and also swoops; his action is seen by others,
and thus these scavenger groups congregate rapidly, although they
are practically lacking in a sense of smell.

Aggregations of male frogs in the breeding season will follow and

go ANIMAL AGGREGATIONS

frequently tightly clasp any moving object, whether salamander,
fish, or other males; and this reaction is based at least in part on
sight. Aggregations of young catfishes are primarily integrated by
sight and secondarily by water vibrations and chemical-touch sen-
sations (Bowen, 1930).

Other instances might be multiplied; but one spectacular one, that
of the synchronous flashing of fireflies, must sufiice. A considerable
controversy has been waged over this subject, but the observation
experiments of Hess (1920) seem to have established the fact of its
occurrence. He found a valley of fireflies flashing in unison, with the
flash apparently initiated on a hill at one side, from which it spread
almost instantaneously over the valley. The next night in the same
place the observer was able to obtain at least partial control of the
flash and to alter to some extent the intervals between flashes. With
a pocket flashlight he gave the initiating signal just before it would
normally have occurred, and the insects followed the artificial lead
until the interval was reduced to three-quarters its original duration,
and then one-half. At the second trial at one-half the original period
fewer insects followed the flashlight, and after that the flashing in
unison was broken.

Such synchronous flashings of fireflies are apparently more com-
mon in the orient. Morrison (1929) has published a recent note upon
their occurrence in Siam, based upon three years' experience there.
His account follows:

"During the months of July, August, September, and until the
heavy rains set in, on any dark night it is possible to see whole
stretches of the river or canal banks fit up by the flashing of myriads
of insects. These areas of synchronism may extend for several hun-
dred yards at a stretch or may be confined to single trees, glowing
and being extinguished with surprising regularity. Actual timing of
this intermittence showed that luminescence occurs at the rate of
approximately 120 times a minute. During the period between the
flashes the light of the fireflies reached almost complete extinction,
the intensity being so low that at a few feet from a tree of actively
luminescing insects it is quite invisible.

"Perhaps one of the first things which is called to the attention of

INTEGRATION OF AGGREGATIONS 91

the observer is the fact that this synchronism is confined to locaHties

bordering on streams, or to low, water-saturated ground

Around Bangkok it is commonly known that the synchronal flashing
of fireflies is confined to one particular tree, the 'ton lampoo' of the
Siamese — Sonneratia acida. In all of the observations which the writ-
er has made, no exceptions to this have been found, but whether this
particular tree is the gathering-place of the insects in cases of syn-
chronism reported from other parts of the East is a question.

''The fact that Sonneratia acida is the tree on which the insects
congregate around Bangkok leads one to question the statement that
has been frequently made to the effect that the synchronal flashing
of the fireflies is a mating adaptation. S. acida is found both in man-
grove associations, and also as a solitary tree growing along the
banks of streams. In these latter cases the roots of the tree are often
immersed in water, the tree at times standing several feet from the
bank. If the females of the species are wingless, as is the case with
the majority of the North American Lampyridae, there would be no
opportunity for them to approach the tree. Furthermore, at no time
have females been found on a tree of actively synchronizing insects,
or within its vicinity. Observations on this point have been repeated-
ly made and have been corroborated by local entomologists who
have become interested in the problem.

"Perhaps one of the most popular theories as to the cause of
synchronism is that of 'sympathy.' According to this idea there is
some particular insect which acts as a pace-maker for the rest, and
they follow him, regulating their flashes by his. However, due to the
fact that the insects are scattered quite generally over a tree and are
not within sight of any one particular animal, this appears to be
quite impossible. Furthermore, any follow-the-leader action on the
part of the insects would result in a wave of light passing over the
tree and originating from a definite point, a fact which is not the
case once the synchronism has begun.

"It is possible to inhibit the synchronism of a tree of insects by ex-
posing them to a bright light for about a minute. When the fight is
turned off, the synchronism returns having its origin, apparently, in
some individual or group generally located in the central part of the

92 ANIMAL AGGREGATIONS

tree. From this group, then, the synchronism extends over the entire
tree in an irregular wave until all of the insects are flashing in
unison.

"Synchronism usually begins shortly after darkness has set in, the
fireflies emerging from the nearby thickets and flying in an indirect
course to the Sonneratia trees. During this flight to the trees there
is no sign of a concerted flashing, the actions of the insects being
similar to those found in our local forms during flight."

It seems probable that with these fireflies we are dealing with a
phenomenon of two distinct aspects (Blair, 19 15). One is a recovery
response similar to recovery from fatigue. Such flashing would rarely
be synchronous or near-synchronous. On the other hand, there ap-
pears to be a releasing stimulus which, in the cases observed by Hess,
might come either from the pace-setting flash of a firefly or of an
electric torch. This brings up the problem of the leader in group inte-
gration, for which we have not space here. It is discussed at some
length by Child (1924).

INTEGRATION BY SOUNDS

Among many animals group organizations occur as the result of
sound production. To be sure of this, one must have evidence that
behavior is altered as a result of sounds. The fact that collections
of animals, such as frogs or insects, are producing sounds which are
loud to the human ear is not good evidence that they have group
significance (Lutz, 1924). There is evidence that among some ani-
mals sounds may be used in sex recognition. Perhaps they are more
often of sexual significance in general sex stimulation which, while of
advantage to the group, may yield no advantage to the producer of
the sound ; and may even result disastrously in the case of the young
deserted by a nesting bird which had been stimulated to renewed
sexual activity by an outburst of song. Such cases have been re-
ported by creditable ornithologists (Sherman, 1924).

Ohaus (1899-1900) and Wheeler (1923) report that the Passalus
beetles, which have the habit of boring in logs, are kept together by
auditory signals; and Professor Wheeler has more than once spoken
of his observations, indicating that aerial sounds may play a part in

INTEGRATION OF AGGREGATIONS 93

the organization of ant colonies. But on this point there are other
observations to the contrary (Fielde and Parker, 1905).

Beebe (1916) thinks that there is a close correlation between habi-
tat and habits of tropical birds and the development of their voices,
which are popularly supposed to be one of the most striking attri-
butes of tropical birds. He reports that sohtary birds, living in the
open country where the view is more or less uninterrupted, have a
tendency to possess negligible voices. Inhabitants of dense jungle,
if relatively solitary, have remarkable vocal powers, with loud
staccato calls or with insistent rhythm, by means of which they com-
municate with their unseen fellows. Such birds may be nocturnal in
habit. Birds living in pairs or in families have, for the most part,
vocal organs which they use to good effect; but they lack the super-
lative voice development of solitary birds. Birds living in flocks have
voices that are still less in evidence, though there are notable excep-
tions to this rule, as, for example, the parakeets.

In the matter of vocal performance, as with tactile and visual
integrations, group unisons have been reported. The group singing
of the western meadowlark is an example among birds. One of the
most interesting cases is that of the snowy tree cricket, which has
been much studied and which Fulton (1925) reported to effect
changes of chirping rate in order to chirp in unison.

ShuU (1907), a careful and critical observer, concluded early in
his studies that real synchrony does exist in the chirping of the tree
cricket; but later he somewhat modified this opinion, saying that
while he still believed that the singing insects do influence one an-
other, he believed that cases of exact synchronism were usually
accidental. Lutz (1924) was skeptical both concerning the fact of
synchronism and concerning its importance with tree crickets. Ful-
ton (1928, 1928a) in recent studies appears to have furnished con-
clusive evidence that the Oecanthus song is both rhythmical and
synchronous. After the usual listening tests, revealing almost per-
fect synchronism, a number of the singing insects were placed in
another cage at some distance, and the front tibiae containing the
auditory organs were removed. This effectively broke up the syn-
chrony except at those times when the individual rhythms appeared

94 ANIMAL AGGREGATIONS

to coincide for a brief period. Fulton records that "when three or
more mutilated males were singing at once an utter confusion of
notes resulted, so that the rhythmical quality of their songs was
entirely obscured." The removal of the tibiae did not seem to affect
the general health of the insects. The loss of one or more legs ap-
pears to be a matter of relatively small importance among these
insects; they lived as long as did those with the ordinary quota of
legs. Similar observations were made by Fulton on a katydid and on
a grasshopper known as the "Nebraska conehead."

Synchronic behavior may, of course, merely mean that the group,
while reacting as individuals, receive the stimulus at the same time
and so react simultaneously. This is illustrated by the responses
Minnich (1925) obtained when he exposed aggregations of caterpil-
lars to various sounds. Such synchronism has no bearing upon the
problem of group integration; but synchronism, such as described by
Fulton, of responses by members of the group to each other may well
have group significance.

Buxton (1923) records an observation made some years before
upon the production of rhythmical sounds by termites. 'T noticed,"
he says, "small numbers of winged termites emerging at one p.m.
from a subterranean nest under stones in a shady place by the road-
side. The ground round the mouth of the nest over a radius of three
feet was covered by thousands of small soldiers and a small number
of large soldiers. All of these were making a rhythmical sound which
resembled the noise made by sand falling on brown paper and which
was caused by tapping their heads on the dead leaves on which they
were standing. The sound was produced in perfect time at a rate of
about 48 beats per minute, and in the intervals between the beats
there was complete silence. This remarkable performance was not
disturbed by my collecting a considerable series of the performers,
but an hour later when I passed the spot, the emergence of
winged adults had ceased and not a soldier was to be seen above
ground."

The termites were determined by Silvestri to be Acanthotennes
militaris Hag. Buxton does not believe that the rhythmical nature
of the sound production could be explained by substratal vibrations.

INTEGRATION OF AGGREGATIONS 95

since the termites were standing on many different dead leaves scat-
tered over a considerable radius. Gounelle (1900) had previously
described the sound produced by termites by tapping their heads on
plants as being Hke the sound produced by a pinch of sand hitting
paper, but he did not record synchrony. Emerson (1928) found that,
despite the possession of the so-called "auditory organs" on the
tibiae, N asutitermes giiayanae did not respond to a wide range of
aerial sounds but did react to substratum vibrations.

Much emphasis has been placed on the role played by the human
voice in the integration of human society; some social psychologists
prefer to define man as a language animal. In this, man does not
appear to be unique except in the degree to which language has been
developed in his species. Craig (1908), in discussing voices of pi-
geons as a means of social control, finds that in animals with so
highly developed instincts as birds there is still much of the social
life that cannot be explained on an instinctive basis. The reaction of
the individual pigeon must be adjusted to meet the activities of other
birds, its parents, its mate, its young, its neighbors, and chance
strangers. The adjustment is very delicate and requires that each
individual must be susceptible to the influence of others, an adjust-
ment which is largely accomplished by vocal means.

Perhaps more time has been spent on the vocal-auditory method
of group integration than is justified by the conditions obtaining
at the aggregation level with which this study is immediately con-
cerned. Its interest by reason of its importance with the higher ani-
mals must be the excuse.

INTEGRATION BY LOW-FREQUENCY VIBRATIONS

Much experimentation shows that animals that give little or no
indication of perceiving sound vibrations coming through the atmos-
phere respond definitely to vibrations of similar or lower frequency
coming to them through water or through the substratum. With
catfish, Bowen (1930) finds that blinded animals give definite reac-
tions to the passing of another fish or of a model with a posterior
part vibrating somewhat as does the tail of a fish. Such reactions are
dependent on the presence of the sense organs of the skin. When

96 ANIMAL AGGREGATIONS

these are anesthetized, the Winded fish respond very httle, if at all,
to the passing of others.

Various insects and other animals give no responses to aerial
vibrations easily detected by the human ear, but readily respond to
the same sounds when their receptacle is placed upon the piano pro-
ducing the vibrations. Emerson (1929) has demonstrated that in
the social termites mechanisms exist for producing substratal vibra-
tions which can be detected at times by the unaided human ear, and
easily when a microphone is used. He suggests that this may be one
means of communication between these insects. Rabbits have long
been known to signal by ground thumpings. The extent to which
this kind of vibration is used in the aggregations with which we are
specifically concerned awaits investigation.

Buxton (1923) records an instance of co-ordinated movement
among arctiid moth larvae which illustrates some of the possibilities
of this type of integration. These caterpillars live in webs on herb-
age in groups numbering several scores. If the web is disturbed,
the larvae jerk the anterior ends of their bodies sidewise with a sharp
flicking movement. All jerk together and maintain the reaction at a
rate of about twice a second for as much as 20 to 30 seconds. Then
they cease this movement and resume feeding. If they wander even
an inch or so from the web, they do not take part in this movement.
If an elongated web is chosen for the experiment (for example, a web
4X12 inches), the movement of the larvae is not simultaneous, but
waves of movement may be seen to pass through the mass of larvae
from the point of disturbance so that the movement is organized
but not synchronous. Obviously, the stimulus is conducted along
the web. More mature larvae that have left the web do not generally
give this movement when disturbed, although they may so respond
when another crawls over them.

POSSIBILITY OF BIOPHYSICAL INTEGRATION

It is probably too early as yet to speculate with profit concerning
the possibility of other, more subtle methods of group integration,
such as the observations of Gurwitsch (1926), Borodin (1930),
and others suggest may result from exploration of the field of bio-

INTEGRATION OF AGGREGATIONS 97

physics. These workers believe they have demonstrated that rapidly
growing plant and animal tissues give off radiations which are able to
stimulate other tissues completely separated from the so-called
"senders" by being inclosed in quartz tubes so that these "detectors"
show a decided increase in mitoses on the radiated half as compared
with the non-radiated part of the same stem.

A favorite experiment consists in placing a moist onion root,
attached to part or all of the bulb from which it grew, into a small
tube made of quartz. One or more onion roots from different bulbs
are introduced into open-ended glass tubes and are also kept moist.
The former is to be the detector; the latter, the sender. The sender
is carefully centered so that its growing root tip points directly to-
ward and at right angles to the detector root, and is allowed to re-
main so for from i to 2| hours. The detector is then marked with
India ink on the side away from the sender and is killed in an ap-
propriate fixing-solution and sectioned. In fixed and stained sections
the number of mitoses on the exposed and non-exposed sides are
compared. Most of the work reported to date shows uniformly a
greater number of mitoses on the exposed side ; the work of Rossman
(1928) is an exception. These are believed to have been induced by
the action of mitogenetic rays. Definite but conflicting wave-lengths
have been announced for these rays.

This field needs further clarification before we can begin building,
with a sense of security, upon the suggestions opened by this work.
If the presence and importance of mitogenetic rays are finally es-
tablished, we shall then have to inquire carefully whether or not we
have similar subtle means of group integration in the field of bio-
physics which may help us resolve the problems in social and sub-
social behavior that are epitomized by Maeterlinck's phrase "the
spirit of the hive."

HARMFUL EFFECTS OF AGGREGATIONS

CHAPTER VI
HARMFUL EFFECTS OF CROWDING UPON GROWTH

Our knowledge concerning the methods of aggregating and the
factors conditioning the formation of aggregations has grown steadi-
ly and gradually, as has our information concerning their integra-
tion. On the other hand, marked advance has been made since 1920 in
the investigations of the physiological effects which such aggregations
produce upon the individuals of which they are composed. The type
and extent of such effects make one of the crucial tests of the impor-
tance of the phenomena. If these aggregations are merely forced
reactions resulting from limited space or from blind tropistic be-
havior, or if they result only as an expression of a social appetite or
instinct, their significance is more remote and the problem of their
origin is more difficult of solution than if they can be shown to have
group value even in their poorly integrated stages. Failure to ob-
serve such values for many aggregations led Deegener to conclude
that their formation must be due to some inexplicable instinct.

In the investigation of this problem we must first inquire whether
or not the aggregations with which we are deahng have positive or
negative survival value which can be recognized. Even if positive
survival value is found in a number of cases, the problem is by no
means solved; but the methods to be used in its solution will be more
clearly indicated than if we are forced to rely upon the postulate of a
former survival value, of which the only remaining evidence is a
weak social appetite persisting frequently in the face of present
negative survival values.

Even with the recently devised methods of analysis, the harmful
effects of such aggregations are frequently more easily apparent than
are the benefits. To the eye of the naturalist depending on field ob-
servation for his data, benefits do not become obvious until the ag-
gregation is sufficiently well integrated so that members may be
warned of the approach of danger by some group attribute, such as

I02 ANIMAL AGGREGATIONS

the multiplicity of eyes in the group, or can attack or defend them-
selves more effectively by the multiphcity of claws or of teeth. Most
of the experimental approaches to this subject have been similarly
limited.

The impressive array of facts, accumulated by observation and
experiment, which indicates that loosely integrated aggregations
have harmful effects will be summarized in the present chapter so far
as the rate of growth is concerned. The next following chapters will
give other facts concerning harmful effects upon the rate of reproduc-
tion and upon longevity.

Dermestes beetles feeding on a limited amount of carrion exhaust
their food sooner when more than one is present. This is also true of
leaf-eating caterpillars, sap-sucking aphids, or tissue-filling parasites.
It is only with well integrated groups of predators catching lively
creatures as food that the feeding aggregation becomes of value. A
school of young minnows is much more likely to catch a given Daph-
nia than is a single individual, and each member of the group is more
likely to feed upon the Daphnia stirred up than if he swam alone.
This type of group advantage increases with group organization, as
shown by the grasshopper drives of African storks.

The same number of relatively defenseless individuals are more
easily gobbled down by an enemy when aggregated than when
scattered. One of the insect sleeping-clubs described by the Raus
would provide a substantial breakfast for the proverbial early bird,
and a hungry centipede would have easy picking in a group of aesti-
vating land isopods. In locust control measures, men take advantage
of the tendency of locusts to collect in dense overnight aggregations.

There is a general ecological assumption that the accumulation of
the waste products of a given species in their habitat tends, with
most animals, to limit the time of their occupancy, at the same time
preparing the way for another species to come in. This is sometimes
considered one of the major biological factors causing ecological suc-
cession, a process well illustrated by the sequence of fauna in a pro-
tozoan infusion.

PLANT TOXINS

It has long been thought that one of the major causes of such suc-
cession among plants is the accumulation of more or less specifically

HARMFUL EFFECTS OF CROWDING UPON GROWTH 103

toxic root secretions. Almost a century ago De Candolle, the French
botanist, suggested that the reason for the decrease in yield following
the continued growth of the same crop on the same soil is due to the
accumulation in the soil of harmful material given off by the growing
plants. Liebig apparently adopted this view for a time but aban-
doned it later, thinking that the observed benefits of crop rotation
were due to the different nutrient requirements of the crops rather
than to the accumulation of poisons in the soil. Pickering (191 7)
gives conclusive evidence that root excretions may have a toxic effect
upon growing plants. In his work he used mustard plants growing in
earth, on the surface of which rested a tray with a porous bottom,
with a large central walled opening through which the plants grew.
This tray held 5 inches of earth. The presence of such a tray made
practically no difference in the growth of the plants in the pot below,
even when the tray itself contained a growth of mustard plants,
providing their roots were kept out of contact with the soil of the
lower pot and that water from around the roots of the upper plants
was not allowed to reach the lower soil. When washings from the
upper growth were allowed to drain into the lower pot, carrying
leachings from the plants grown in the upper tray, growth of the
experimental seedlings was reduced to o.oi of that given in control
pots. Pickering found such results common and widespread, and
especially well shown by the effect of grasses on the growth of apple
trees. In summarizing all the evidence on the subject, Russell (1927)
concludes that, while a toxin can be shown to be present, the toxin
concerned'is not stable and is non-specific.

CESSATION OF GROWTH IN BACTERIAL CULTURES

The long-recognized failure of cultures of micro-organisms, such
as bacteria and molds, to continue growing indefinitely has been
attributed to three main causes: the exhaustion of foodstuffs, the
accumulation of metabolic wastes or specific "autotoxins," or the
limitations imposed by actual physical crowding. Henrici (1928)
gives a good summary of the present state of knowledge in this
field.

The possible effect of physical crowding in limiting growth must
be excluded because much more dense growths can be obtained with

I04 ANIMAL AGGREGATIONS

organisms on filter paper or on agar than when they are grown free
in hquid broth; and when the organisms are repeatedly filtered off so
that the physical effects of crowding are periodically eliminated, the
growth period is not thereby prolonged. There can be no doubt but
that the exhaustion of food materials does play an important role
in the limitation of cultures, but the question as to how important
this is in comparison with the accumulation of waste products or
"autotoxins" has not been decided.

Henrici says, "The idea that growth is limited by the accumula-
tion of some toxic substance is the one that seems to be most general-
ly accepted, though the evidence for it is far from being convincing."
The evidence supporting the idea that the toxic substances are im-
portant is as follows: Eijkman (1904, 1906, 1907), working with a
number of species of bacteria, grew them in gelatine until the culture
was densely crowded. He found then that if he took a part of this,
heated it to boiling and, after cooling, reinoculated it, it would then
support growth; but that another part, heated only sKghtly and then
allowed to resolidify, would not produce growth in a new surface
inoculation. Since heating to boiling-point would add no new food
material, Eijkman concluded that he was dealing with some ther-
molabile product of metabolism or a more specific growth-inhibiting
substance.

Further experiments showed that the toxic material would not
pass through a porcelain filter, that heating which killed the living
organisms destroyed the toxicity of the medium, and that treatment
with such volatile agents as ether and ammonium sulphide not only
killed the bacteria but rendered the medium again capable of sup-
porting growth after the volatile material was driven off. They also
showed that if gelatine in which Bacillus coli had grown was resolidi-
fied into a plate and reinoculated with more of the same organisms
and covered with a layer of fresh gelatine, there would be no growth;
or, if fresh gelatine was inoculated with B. coli and then had one part
covered with fresh gelatine and another with the so-called "coH-
gelatine," growth would take place in the former only. Inoculation
of paper dipped in agar and then placed over the coli-gelatine did not
yield a growth unless the coli-gelatine had first been heated.

HARMFUL EFFECTS OF CROWDING UPON GROWTH 105

Rahn (1906), with B. fltiorescens liquefaciens and three other
species of bacteria, obtained essentially similar results except that in
his experiments treatment with ether killed the bacteria without
removing the unstable toxic material. The ether could be evaporated
off but no growth would occur on reinoculation. Appropriate con-
trols made by treating sterile broth with ether, which was then
evaporated, showed growth on inoculation. Heated cultures gave
growth when similar cultures treated with ether gave none, show-
ing apparently that the food value of the medium was not ex-
hausted.

Chesney (19 16), working with pneumococci, found that if the
organisms are removed by centrifuging a rapidly growing culture,
those remaining will continue to grow at the same rate; but that if
the culture be similarly treated after the period of maximum growth
is past, growth is delayed and some of the cells may die off. Filtrates
from 24-hour cultures inhibit growth of new inoculations of similar
organisms, but lose this property if the filtrates are allowed to stand
for a time in the incubator. Chesney concludes that the cells do
produce an unstable, toxic, growth-inhibiting material. Some inves-
tigators have believed this substance inhibiting growth in pneumo-
cocci is fairly specific in its action; but Henrici cites later work show-
ing that the limitation of growth in pneumococcus cultures is due to
three factors: the accumulation of acid, the production of peroxide,
and the exhaustion of the nutrients. The first two come under the
general heading of "toxic products of metabolism," which are here
shown to' limit growth of this organism. Hajos (1922) also demon-
strated growth-inhibition due to the accumulation of products of
metabolism among the colon-typhoid type of bacteria.

Curran (1925), again using B. coli, found a thermolabile growth-
arresting material readily adsorbed by bacterial filters. The first
50 cc. of filtrate from a 200 cc. solution which had supported bac-
terial growth for 3 days was found to support growth on reinocula-
tion much better than did the last 50 cc. of the same solution. It
would appear that the filter became loaded with the growth-inhibit-
ing material and so allowed material that was stopped at the begin-
ning to pass at the end of the filtering. Using a different sort of or-

io6 ANIMAL AGGREGATIONS

ganism, Kuester (1908) finds that molds grown in a nutrient solution
produce conditions which check the growth of further inoculations
before the nutrient supply is exhausted.

In the face of this evidence, Henrici is still unconvinced of the
general apphcabihty of the theory of the production of a toxic ma-
terial serving to limit growth, and holds, rather, to the idea that the
exhaustion of the nutrient material is the crucial point. He suggests
rightly that the Eijkman type of experiment may only show that
media may contain sufficient material to support a heavy population
without growth and still be able to support growth in a smaller
population; that heating the medium to kill the bacteria may cause
a release of nutrient material, making it available for the reinoculat-
ed organism; and finally cites the work of Graham-Smith (1920) in
which he was able to revive staphylococci by adding concentrated
meat extract and thus inducing new growth after the period of maxi-
mum growth had passed, and could postpone the death phase in-
definitely by small daily additions of meat extract. Henrici is prob-
ably correct in concluding that different factors may limit growth in
different cases and that there is no sound basis for believing in the
production of specific autotoxins.

EVIDENCE FROM TISSUE CULTURE

In tissue-culture work Carrel and Ebeling (1923) and Mottram
(1925) report a substance which inhibits growth of explants present
in extracts of aU adult tissues, in serum and even in extracts of em-
bryos. The latter have usually been found to favor growth in such
cultures. Heaton (1926) has found such a substance in yeast extracts
and in a number of adult-animal tissues, especially the liver. He
thinks that the failure of adult tissues to grow easily in vitro, and the
stoppage of growth of connective tissue in vivo, as contrasted with
the continued growth of epithelia, is to be attributed to this growth-
inhibiting substance. Heaton finds it to be thermostabile, though
destroyed by heating up to 125° C. It is soluble in water and alcohol
up to 75 per cent strength, but is insoluble in 97 per cent alcohol. It
seems to be destroyed by autolysis. Its action is greater on older
than on younger embryonic tissues.

HARMFUL EFFECTS OF CROWDING UPON GROWTH 107

GROWTH INHIBITION IN ANIMAL CULTURES

The work on animal cultures most closely connected with these
investigations on bacteria and on tissue culture is that dealing with
the growth in a protozoan infusion. In two studies (1911 and 1914)
Woodruff demonstrated that Paramecia excrete substances that are
toxic to themselves when present in their environment and that
probably play an appreciable role in determining the time of maxi-
mum number, rate of decline, and other characters. Similar conclu-
sions were reached as a result of work with the hypotrich as regards
their own excreta, but they are immune to the effects of Paramecium

Fig. 4. — ^A record of the rate of division of Paramecium aurelia in a series of four
experiments (A, B, C, D) to determine the effect of different volumes of culture medium,
changed every 24 hours, on the rate of reproduction. The ordinates represent the
average daily rate of division of the four lines of organisms in the respective volumes

of medium, averaged for 4-day periods. Rate of division in 2 drops, ; 5 drops,

; 20 drops, ; 40 drops, -• • (From Woodruff, 191 1.)

excreta. In a protozoan infusion the appearance of dominant Pro-
tozoa at the surface runs in this order: Monad, Colpoda, Hypo-
trichida, Paramecium, Vorticella, and Amoeba. This ecological se-
quence is due in part to accumulation of toxic material and in part
to the supply of available food.

This problem is closely related to the consideration of the effect
of the size of the effective environment, whether lake, pool, or labo-
ratory container, upon the contained organisms, which in turn is
closely related to the whole problem of crowding.

io8 ANIMAL AGGREGATIONS

In 1854 Jabez Hogg with some right apologized to the London
Microscopical Society for taking their time with observations on the
subject of the pond snail Limnaeus stagnalis, which had already been
well studied; but in the midst of his tedious record he states that a
snail kept in a "small narrow cell will grow only to such a size as will
enable it to move freely." This is the first recorded observation that
has come to my attention of the limiting effect of volume on growth.

Semper (1874, 1881) took up the problem twenty years after
Hogg's observations, using the fresh-water isopod Asellus and the
pond snail Lymnaea stagnalis. With the former he found that when
animals living in a balanced aquarium were sealed into glass dishes,
they might be left for nearly 2 years, with an adequate food supply
and, he beheved, an adequate oxygen supply for the three or four
generations that would be produced; but under these conditions the
last generation was abnormally small. With the snails Semper divid-
ed the same mass of eggs into different lots, which were placed in
variously sized containers ranging from 100 to 5,000 cc. Food was
kept at an optimum, but the snails placed in the containers of smaller
volume grew more slowly than did their fellows placed in the larger
vessels. Similar results were obtained regardless of whether the
snails were isolated into a given volume or were put in groups, so
long as the volume per snail was the same in both cases. Semper
found that optimum growth lay between 400 and 500 cc. per snail
and that increases beyond this point gave no further increase in
growth. The effect of increase in volume was much more marked in
the smaller volumes. Later workers are agreed that relatively large
volumes of water per snail are necessary for optimum growth.

Semper recognized the complex nature of the problem and at-
tempted, by chemical analyses made by a trained chemist, to find a
chemical cause. Failing in this attempt, he advanced the hypothesis
that some substance unknown to him was present in the water, prob-
ably in a very minute quantity, "which, by its relations to the water
which holds it in solution, and by its osmotic affinity to the skin of
the animal, can be absorbed only in a determined and extremely

small quantity Since, according to this hypothesis, the

amount of the substance absorbable in a given time depends on the

HARMFUL EFFECTS OF CROWDING UPON GROWTH 109

volume of the water .... the attainment of full size within a definite
period would only be possible if the volume of water were so great
that the Lymnaea could at all times absorb this unknown stimulant
from the water." This hypothesis, in some form or other, has been
proposed, apparently independently, by a number of workers since
Semper's time.

Semper seems to have been certain of the evolutionary significance
of the Hmitation of growth by volume. He found it impossible to
obtain full-sized individuals from snails stunted during the first year
of their lives; and if the causes checking growth were repeated regu-
larly through the succeeding generations, he felt that a dwarfed race
must arise. Whitefield (1882) came to the same conclusion, using
Lymnaea megasoma from Vermont. Whitefield continued the crowd-
ing for four successive generations, during which time the snails be-
came successively smaller and more slender, so that an experienced
conchologist did not recognize their relation to the shells of the
parent stock.

Yung (1878, 1885) concluded from his experience in raising tad-
poles in containers of various sizes and shapes that dwarfing is due to
a lack of aeration. De Varigny (1894) took up the problem with
Lymnaea again and in general obtained the same sort of results
reported by both Semper and Yung. A snail kept in a liter of water
with a surface of 257 sq. cm. for 5 months was nearly twice the
length of one kept in the same volume of water but with a surface
area of 3.14 sq. cm. In order to facilitate the analysis, De Varigny
suspended a glass tube 2-3 cm. in diameter in containers of various
sizes. The glass tubes were closed over the bottom with musHn, and
each contained a single snail. Each day these tubes were hfted from
the water and replaced two or three times in order to secure complete
mixing of water. Even so, the contained snails grew approximately
the same regardless of the volume of water with which they were in
contact through the muslin screen. In one instance the growth was
the same in such a muslin-bottomed tube as compared with that of a
snail in a corked tube which prevented all exchange between the
inner and the surrounding water. From these experiences he con-
cluded that Semper's explanation would not hold and that the size

no ANIMAL AGGREGATIONS

to which snails grow depends in some way on the actual volume to
which they are exposed and on the surface area of such water. His
explanation was that in the small tubes the snails needed to move
about less to obtain their food and that, with this decrease in exercis-
ing, there came a decreased rate of growth. According to De Va-
rigny, dwarfing from crowding is not so much due to the actual
numbers in the vessel as to the "psychological" influence of num-
bers, which inhibit exercise, just as a man is less likely to walk a
considerable distance on a crowded street than on a deserted one.
He also believed dwarfing to be affected by the accumulation of
faeces.

Willem (1896) bubbled air through his snail cultures and found
growth of the contained snails greatly increased. He concluded that
aeration is important because, even in lung-breathing pond snails,
he believed cutaneous respiration to be more important than lung-
breathing and alone sufficient for the animal.

Davenport (1899) reviewed much of the evidence on the relation
between crowding and rate of growth, and concluded with Hogg that
in respect to the size attained, as in other qualities, the snail has the
power of adapting itself to the necessities of its existence.

Vernon (1895, 1899, 1903), working with echinoderm larvae, con-
cluded that dwarfing is due to a concentration of the excretory prod-
ucts in the media. He found that if eggs of echinoderms were allowed
to develop in water which had previously contained other eggs for a
considerable period of time, the larvae of the second batch were dim-
inished in size as compared with the control. The growth of the
larvae appeared to be reduced by their own excretory products, or
especially by those of adult echinoderms, the more so if these belonged
to the same species. On the other hand, he found that the excretory
products of two species not closely related were favorable to
growth.

Warren (1900), working with the common entomostracan, Daph-
nia, found that continued breeding in small aquaria with the medium
unchanged caused dwarfing. This result he attributed to the action
of the excretory products, which he found to be somewhat specific,
since ostracods and copepods flourished in cultures of Daphnia in

HARMFUL EFFECTS OF CROWDING UPON GROWTH iii

which the latter were dying out. Such results are similar to those
reported by Woodruff and others for protozoan infusions.

Legendre (1907) returned to the problem of the effect of crowding
on the growth of snails, using Lymnaea stagnalis and Planorhis
corneus, raised in one series of experiments in stagnant water and, in
the other, in water changed periodically. As in the case of previous
workers, he found the smaller shells in the stagnant water, and at-
tributed the cause to the accumulation of excretions. In further work
reported the following year, using another species of Lymnaea,
Legendre changed the water every 2 hours in order to avoid the
accumulation of excreta, and varied the factors of volume of water,
surface area, and number of individuals. After 51 days he obtained
the same shell size in all such experiments. He recognized that a
number of factors might bring about retardation in crowded ani-
mals, but laid particular emphasis upon the retarding effect of the
excretions.

Colton (1908) continued work on the effect of crowding on growth
in Lymnaea. Food was recognized as an important element, but just
how important Colton's work does not reveal. He did find that snails
need a certain amount of sediment to aid in grinding their food, and
that certain salts, for example calcium sulphate, aid growth. Colton
found that washed and filtered snail faeces placed in aquaria has-
tened the growth of the snail, probably due to the increase in algae
caused thereby. His aeration experiments support the conclusions of
Willem that these pulmonate snails have a large proportion of cutic-
ular respiration. Concentrated excretory products caused dwarfing;
accordingly decreases in volume of water per individual present,
whether in isolations or in crowded cultures, caused a decrease in
growth rate. Popovici-Baznosanu (1921) minimized the effect of ex-
crement, thinking the amount of food more important.

Crabb (1929) has recently reinvestigated this entire problem with
the pond snail Lymnaea stagnalis appressa, taking care that his snails
were free from trematode parasites, and supplying them with food
known to be favorable for growth in laboratory conditions. He used
eggs from the same egg mass for experiments run simultaneously;
since this snail reproduces by self-fertilization, individuals obtained

112 ANIMAL AGGREGATIONS

from the same egg mass would be expected to have similar genetic
constitution. He concludes that food insufficiency and foul media
are the most common growth-inhibiting factors in snails reared in
otherwise favorable media. Extreme crowding markedly retards
growth, but the individuals rapidly reach normal size after transfer
to standard conditions, unless they are too old. The volume of me-
dium has little effect on the growth of isolated snails providing foul-
ness is not permitted. Aeration promoted growth through reducing
foulness rather than by increasing the respiration of the snails.
Daphnia introduced into the culture are beneficial to snail growth,
since they retard fouling of the medium. He found no evidence that
environmentally induced dwarfing is transmitted, though on this
the experiments were not continued through enough generations to
be conclusive.

Crabb, in his work, continued the general methods of study of this
problem which have been used since the time of Hogg, adding re-
finements which make his conclusions the more trustworthy. Un-
fortunately, he did not take advantage of the method originated by
De Varigny (1894) and used extensively by Goetsch (1924), which
allows a separation of the factor of available space from that of
available volume. In this procedure Goetsch placed animals in the
experimental aquaria in separate tubes thrust through corks to keep
them afloat and covered at the lower end with gauze, which allowed
diffusion connection with the entire aquarium while limiting the
amount of available space.

Goetsch was led to this method by the experience of Bilski (192 1),
who found that the relatively active tadpoles of Bufo and of Rana
esculenta grew less rapidly when subjected to frequent changes of
water than they did when metabolic wastes were allowed to accumu-
late. Bilski also found that an increase in numbers slowed down the
rate of growth more than would be expected by the change in vol-
ume relations involved, when the rate of growth was compared with
that given by an equal number of animals placed in different aquaria.

Goetsch experimented upon sessile Hydra, upon the relatively
slow-moving flatworms, and upon amphibian larvae which are capa-
ble of rapid locomotion. As might be expected, he finds different

HARMFUL EFFECTS OF CROWDING UPON GROWTH 113

factors important for different animals. Thus, with Hydra, volume
per animal is the controlling factor because of the restriction of food
which it conditions. There is no stimulation or depression caused by
the crowding of Hydra into a narrow space; and, within reasonable
limits, concentration of excretory products are not effective. With
Planaria food is again the most important factor, but growth is in-
hibited by the concentration of excretion products or of stale food.
With the active amphibian larvae, if food is controlled, the major
limiting factor is furnished by the more frequent collisions in a dense
population or in a restricted area, and the concentration of excretory
products plays a wholly secondary role.

Church (1927) extended these experiments to include the rate of
growth of the tropical fish Platypoecilus maculatus rubra in connec-
tion with other experiments upon the effect of crowding upon the
rate of growth of fishes. Eight liters of water were used in glass
aquaria, each of which contained 2, 8, or 16 fish. In each series of
experiments, one set of aquaria contained small fish 8-10 mm. long,
another set held fish 12-14 mm. long, and the third set was supplied
with fish 20-23 i^n^- Adult Platypoecilus range from 30 to 35 mm.
The amount of oxygen and the pH of the different aquaria did not dif-
fer significantly. The water was left unchanged during the entire ex-
periment, which ran in some cases as long as 70 days, except that
there were slight additions to replace the small amount lost by evap-
oration. The fish were fed the same number of Daphnia per fish
per day.

Under these conditions the large fish always grew less rapidly the
more fish there were present in a given container. With the small
and medium fish there was some indicatfon of more rapid growth
early in the experimental periods among the fish grouped 8 to the
aquaria; but as the experiment progressed, the rate of growth was
always greatest when the fewest fish were present. Shaw (1929) has
repeated these experiments, with similar results. The experience of
these two workers demonstrates that when there is sufficient con-
centration of waste products the rate of growth is retarded.

In following out the Goetsch type of experiment, Church placed
transparent celluloid containers in the center of each aquarium.

114 ANIMAL AGGREGATIONS

These were 4.5 cm. in diameter and were covered with coarse scrim
at the bottom. They were suspended by wires so that each extended
1.5 cm. below the surface of the water, thus giving to the contained
fish a volume of 24 cc. in which to move about, as contrasted with
the 4,000, 1,000, or 500 cc. volume per fish to which the 2, 8, or 16
fish were exposed in the surrounding aquaria. A single medium-sized
fish was transferred to each of these tubes, regardless of whether
those in the surrounding aquaria were large-, small-, or medium-
sized.

Under these conditions the fish within the small tubes grew less
than did those in the aquaria. At the end of the first 10 days the
average length of the 34 fish in the tubes showed 1.35 per cent in-
crease, while the medium-sized fish in the surrounding aquaria grew
6.51 per cent. At the end of 20 days the difference was still more
striking. The 25 fish in the tubes had grown in this period on the
average 2.78 per cent, while those of the same original size in the
larger volume of water had grown 12.83 per cent. So far as known,
the size of the container was the only variable. The meshes of the
scrim cloth were open throughout the experiment; but to guard
against the possibiUty of lack of adequate diffusion, the tubes were
raised once daily to insure a complete change of water. Such results
are similar to those Goetsch secured for the relatively swiftly moving
tadpoles, and are probably due to the effect of overstimulation
caused by frequent contact with the walls of the small tube.

As stated above, Bilski (192 1) points out that when limitation of
growth rate is caused primarily by stimulation from repeated con-
tacts, and when the number of individuals present is proportional to
the different sizes of the vessels, the rate of growth is not the same.
If we take two vessels of different sizes, a and h, and populate them
with a and h number of animals respectively, so that each animal has
the same amount of space available, in the simplest case the stimu-
lation will come from the contact, or near approach, of two animals.
The relation of the size of the two containers will be a:h, which in a
simple case might be 2:3. The stimulation possibilities from group
interference would he a{a—i):h{h—i). Substituting the values sug-
gested above, we get a stimulation possibility of 2:6. Under such

HARMFUL EFFECTS OF CROWDING UPON GROWTH 115

conditions one would expect to find growth retardation with increase
in numbers to be much greater than if volume relations alone were
the responsible factor.

Inspection of his experimental results in comparison with a simple
formula built on the assumption that the growth would be inversely
proportional to the group stimulation,

'^'{^

i)

in which y represents size, x stands for the number of animals in a
given space, and ^ is a constant, shows that the influence of the
stimulation is not on this order but is approximated by taking an
exponential value of x, namely x^^'. The equation then becomes

X3/2

x{x—i)

o.

y = k

Vx

Values calculated from this formula fit fairly well with Bilski's ob-
servations on the effect upon the growth of differing numbers of
tadpoles in jars of equal size; the observations of Semper on the
growth of snails in relatively small vessels; and, according to Bilski,
with the observations of HofTbauer on growth in carp. Another for-
mula derived by a continuation of the same reasoning better fits
Semper's results with snails in larger volumes.

Bilski recognizes the general significance of his results and believes
that such diverse phenomena as the reported dependence of size of
mammals upon available land, and other similar relationships, in-
cluding even a correlation between the size of children and available
space, nmy depend upon an application of this principle. Farr (1843,
1875) worked out an equation essentially similar to that of Bilski to
describe the relation between death-rate and the density of human
populations. Brownlee (1915, 1920) finds that Farr's law fits a wide

ii6 ANIMAL AGGREGATIONS

range of biological and biochemical relationships, including even the
relation worked out by Kennealy (1906) between the racing record
for a particular distance and the length of the race. Pearl (1925)
finds that essentially the same equation describes the effect of crowd-
ing upon the rate of reproduction in Drosophila.

The problem is obviously complicated by many factors, but it is
interesting and probably significant that the relationship can be ex-
pressed mathematically in a similar way for such a wide range of
phenomena. It is almost an anticlimax to have to record that physi-
cal disturbance due to numbers is not the only factor controlKng
growth in rapidly moving animals, such as fish, under crowded con-
ditions. The careful work of Church and of Shaw, already summa-
rized, demonstrates that the accumulation of waste products is also
effective with fish, just as a long line of evidence culminating in that
given by Goetsch proves that it is effective in the slower-moving
planarian worms.

More recently Wilier and Schnigenberg (1927) and Kawajiri
(1928) have independently tested the effect of crowding on the rate
of growth of young trout in running water. Both report essentially
similar results; the work of the former will be reviewed here, since it
is the more comprehensive. These workers used young of the brook
trout during their prehatching, yolk-sac, and early feeding stages.
In their experiment they tested a wide range of conditions. They
used the same number of eggs or of young in different volumes and
with different surface areas, and in other tests used different num-
bers in the same volumes. All experiments were carried on with
water running at a rate of from t^.t^ to 65 cc. per second.

Their results show that moderate crowding after hatching has no
adverse effect upon fish whose prehatching development has been in
equally crowded conditions. In fact, under these conditions, one set
of experiments show an apparently beneficial effect. On the other
hand, crowding the eggs produces definite retardation in length and
perhaps also in weight at hatching time. Such retardation is corre-
lated with the volume of water rather than with the area of the
screen on which the eggs rest.

Exposure of uncrowded eggs to water that has flowed over a mass

HARMFUL EFFECTS OF CROWDING UPON GROWTH 117

of developing eggs is found to produce about the same degree of re-
tardation as is furnished by crowding. Under these conditions the
dwarfing effect must be a result of toxic materials accumulated in
the water. The general importance of these results is enhanced be-
cause of the fact that they have been obtained from animals grown in
running rather than in stagnant water. There was an indication of a
condition of optimum crowding^ in the early experiments which was
not sustained by later work, although specific experiments designed
to test this point were not attempted.

Peebles (1929) has taken up the problem of effect of numbers
present upon the rate of cleavage of echinoderm eggs, and upon the
rate of growth of arms of plutei, in the light of developments in
tissue-culture work. She finds, as did Vernon (1895) and Springer
(1922), that embryo-water contains substances which check growth,
but adds the observation that some of the inhibiting effect is counter-
acted when living larvae are present. She produces evidence that the
growth-inhibiting substances are associated with the lipoids and
that, after their removal, growth-promoting substances can be dem-
onstrated to be present. These latter will be discussed in chapter ix.

The relation between the size of the effective environment and
that attained by the animals living therein has more than laboratory
interest. The belief is widespread that fish grow larger in large lakes
than in small ones. Pearse and Achtenberg (1920) report such a
correlation between size of lake and size of contained yellow perch.
This correlation is not uniform, for numerous exceptions could be
cited; for example, Jewell and Brown (1929) find no such relation-
ship holding between size of fish and the size of the small Michigan
lakes in which the fish live.

Hesse (1924) states that the same relation holds for mammals with
regard to the size of available range: those living on small islands
attain a smaller adult size than related forms on larger bodies of
land. In many cases the reduced amount of available food in the
smaller habitats has been recognized as being sufficient to explain
the observed phenomena. Semper (1879) critically discussed this

• Kawajiri reports that the survival-rate increases as the number of fry in a box
increases.

ii8 ANIMAL AGGREGATIONS

general situation long ago and left the impression that the suggest-
ed relationship was either not proved or only indirectly related
to the suggested space factor. The idea that there is a direct con-
nection between available space, and size attained in land animals,
still has, however, considerable vitality, as is shown by Bilski's
suggestion (1921), following his careful statistical analysis of the re-
lations between available space and growth in tadpoles, that the
smaller size of children reared in slums, as compared with that of the
children of more fortunate parents, is to be accounted for by the
smaller space available per child for the former and the resulting
greater degree of stimulation by repeated contacts, such as have been
shown to result in decreased growth in tadpoles, fish, and other
rapidly moving animals.

There can be no doubt that crowding decreases the rate of growth
in many instances, and any interpretation of the facts to be present-
ed later concerning beneficial effects of crowding up to an optimum
population must take this fact into consideration. When one at-
tempts to summarize the evidence concerning the factors causing the
retarded growth in crowded conditions, he finds a decided lack of
unanimity among the different investigators, indicating that in all
probability there are many factors which may produce the same
result.

It is instructive to review the retarding factors suggested to date.
They are of two kinds: the vague and the definite. In the former
category one must put the suggestion of Hogg, working with snails
in 1854, that they adapt themselves to the necessities of their exist-
ence, which Davenport, 45 years later, said still summarized the
state of knowledge on the subject at that time. There is also Sem-
per's postulated X-substance necessary for growth in snails and
water isopods (1874, 1881) ; the autotoxins of the bacteriologists; and
the growth-inhibiting substances of the tissue culturists (Heaton,
1926) and of Peebles (1929) for echinoderm larvae; as well as a
"space factor" seriously discussed by many observers (cf. Wilier and
Schnigenberg, 1927). As commonly used, this space factor is about
equivalent to Hogg's conception.

Regarding this group of suggested retarding factors, the best we

HARMFUL EFFECTS OF CROWDING UPON GROWTH 115

can say at present is that they are unproved. We shall find the sug-
gestion of an .Y-substance made in many different connections be-
fore we have finished this discussion. It is useful as a hypothesis but
is not to be confused with concrete fact. However, the recent de-
velopments concerning the importance of small traces of vitamins,
and the work upon "bios" and upon tissue-culture inhibitions, will
keep us from dismissing this hypothesis too hastily.

Of the definite factors suggested, we have lack of sufficient aera-
tion, in addition to undernutrition, reported as operating in crowded
tadpoles (Yung) and among snails (Willem, Colton, Crabb). There
can be little doubt but that insufficient aeration is an effective factor
under many conditions. The suggested harmful effects of lack of ex-
ercise in snails (De Varigny) now appear groundless. The accumu-
lation of excretory products reported as an effective agent by many
workers appears to have undoubted and marked influence, whether
in echinoderm larvae (Vernon, Peebles), in Daphnia (Warren), in
snails (Legendre, Colton, Crabb), in planarians (Goetsch), or in fish
(Church, Wilier and Schnigenberg, Shaw). Evidence in favor of this
conclusion will accumulate as we proceed.

The reduction of available food correlated with crowding, whether
caused by increase in numbers or decrease in volume, is another un-
doubted factor in the situation, as shown for snails by Colton and
Popovici-Baznasanu and for Hydra by Goetsch. With some animals,
such as Hydra, it may be that this is the only factor operating. With
rapidly moving animals, the effect of frequent contacts resulting in
overstimulation of some sort also contributes to the retardation of
growth in crowded animals, as in tadpoles (Bilski, Goetsch) and in
fish (Church).

CHAPTER VII

RETARDING INFLUENCE OF CROWDING ON
THE RATE OF REPRODUCTION

In the preceding chapter we have assembled evidence to demon-
strate that among many animals overcrowding tends to produce
dwarfed individuals, and have discussed the factors that have been
suggested as operating to produce this effect. As might be expected,
there is frequently a retardation of the rate of reproduction as well
as of the growth-rate of the individual. In many respects the two
phenomena overlap. The evidence for the slowing-down of repro-
ductive rate under crowded conditions will be examined in part in
the present chapter. At another place consideration will be given to
the data brought forth by Robertson and others which indicate that
under certain conditions the rate of reproduction is increased in early
stages of protozoan or other cultures when more than one animal
is present in a limited amount of medium.

REDUCED DIVISION RATE IN INFUSORIA

Balbiani (i860) reported from a single experiment on Parame-
cium that this infusorian must be in not less than 2-3 cc. of medium
for the greatest productivity to be realized. Kulagin (1899) sug-
gested that this is due to the accumulation within the medium of
excretions analogous to toxins, which gradually accumulate until the
nucleus is affected.

Woodruff took up this problem in 1911 in an effort to find the
effect of excretion products of Paramecium on its rate of reproduc-
tion. Since the experiments of Woodruff usually form the starting-
point for present-day citations on this subject, they deserve to be
given in some detail.

The reproduction of P. aurelia was followed for from 16 to 20 days
in four volumes of hay infusion: 2, 5, 20, and 40 drops, which were
changed at 24- and 48-hour intervals in different series of experi-

RETARDING INFLUENCE OF CROWDING 121

ments. The results are given graphically in Figure 4. For the ex-
periments in which the medium was changed every 24 hours the
Paramecia in 5, 20, and 40 drops are shown to have divided 2.4,
6.4, and 7.4 per cent more rapidly, respectively, than did those in 2
drops. When the medium was changed every 48 hours, the per-
centages for the same volumes were 5.3, 9.3, and 9.25. The results
are given throughout as averages for 4-day periods.

From these experiments Woodruff concludes, "The rate of repro-
duction of specimens from pure lines of Paramecia when bred under
identical conditions of temperature and culture medium is influenced
by the volume of the culture medium (within the limits tested in the
experiments) and the greater the volume, the more rapid is the rate
of division." The slight discrepancy with the 40-drop cultures
changed every 48 hours is unexplained, but the suggestion is offered
that the bacteria always found in such cultures, and which are used
as food by the Paramecia, develop so rapidly under these conditions
that they may exhaust their own food or produce sufficient excretion
products to be injurious to the associated Paramecia. Otherwise,
Woodruff believes that by his culture methods, which included cross-
inoculations between the different cultures, he has eliminated the
bacteria as agents causing the observed difference in rate of Para-
mecium division.

The conclusion that the recorded effects are due to the accumula-
tion of Paramecium waste products rests on three lines of evidence.
In the first place, as we have just seen, the rate of division is higher,
for the periods and amounts tested, the larger the amount of avail-
able medium. Second, the rate averaged 8 per cent greater in the 2-
drop cultures changed daily than in similar cultures changed every 48
hours. The other cultures similarly showed a 6 per cent increase if
changed daily. Finally, culture media in which Paramecia had flour-
ished for 10 days before removal were shown to have a depressing
effect upon the reproductive rate of Paramecia replaced in it, as
compared with the effect of an infusion which had contained no
Paramecia but which otherwise was as nearly comparable as the two
could be made.

Woodruff (19 13), as a result of further experience, concluded that

122 ANIMAL AGGREGATIONS

the substances which Paramecia excrete into their medium are es-
sentially species-specific, at least to the extent that they do not uni-
formly influence the rate of reproduction of the hypotrich Stylony-
chia. This hypotrich produces conditions within its own culture me-
dium which are definitely depressing for hypotrich reproduction and
without necessarily affecting the rate of reproduction of Paramecium.

The question of species specificity has not attracted the work it
deserves; but, stimulated by the researches of Robertson, to be re-
ported in a later chapter, several workers have retested the effect of
crowding upon the rate of reproduction of Paramecia and of other
protozoans. Without exception, all the workers reporting so far,
Robertson included, have confirmed the conclusions reached by
Woodruff for cultures running the length of time for which his
averages were taken. A detailed discussion of this later work is post-
poned for the present.

From general considerations it appears highly probable that the
relationships outlined above and in the preceding chapter, if properly
adjusted, could so affect an animal (for example, Paramecium) that,
while it might be able to continue to live, its powers of reproduction
would be lost. Crampton (191 2) tried this experiment. He found
that a single Paramecium confined in a capillary tube could be kept
from fission for as long as 32 days, while controls relatively un-
restricted as to space were dividing at a rate that would produce
4,300,000,000 animals in the same time. He recognized three factors
as working to produce this effect: lack of sufficient nutrition, ac-
cumulation of waste products, and stimulation from the narrow
limits. That lack of sufficient or proper food is not the sole cause is
shown by his experience that the confined animals could be released
to swim about in a culture of Bacterium iermo daily for as long as 1 2
hours out of the 24, without division, if the remainder of the day
were spent in the confinement of the tube; and that they could be
held so without division for a week, while controls were dividing on
an average of once a day. Such Paramecia remained plump and well
nourished in appearance ; those left in the tubes for long periods with-
out changing became transparent and emaciated. Stylonychia gave
similar results. It is significant that Crampton centrifuged his ani-

RETARDING INFLUENCE OF CROWDING 123

mals, which must have brought them into violent contact with the
walls of the capillary tubes.

Crampton's work was in many ways an extension of Conklin's
earher observations concerning the size attained by the gasteropod
Crepidula plana with relation to the amount of available space. The
dwarfing of these snails when crowded, Conkhn thought, should be
interpreted as due to space inhibition of cell division.

These facts were reported by Conklin in 1898. Crepidula plana
lives within the shells harboring herinit crabs. If the shells are small,
the contained Crepidula are few in number and are dwarfed; if
large, the Crepidula may be present in numbers and be large. Since
there may be but i small individual in the small shells, while there
may be 4 "giants" in a large one, Conklin believed that the difference
in size is not due to differences in available food; nor is it due to the
presence of accumulations of excreta, since both shells are equally
open to the surrounding ocean. Neither is the result due to the lack
of room to move about in, since both large and small Crepidula are
relatively firmly attached to their substratum. Rather, there is a
space retardation of cell division, since the cell sizes of the one are no
larger than the other. If the small Crepidula are transferred to a
larger space, they will increase in size. The stimulus acting to retard
cell division in these dwarfed Crepidula is more obscure than in the
case of the rapidly darting Paramecium confined in a capillary tube.

Kalmus (1929) has added two other factors to Crampton's three,
in reporting his own studies on the effect of inclosing Paramecium
caudatum, Stylonychia, and Spirosiomum in capillary tubes. He finds
that the age of the culture and the solubihty of glass in the culture
medium have decided effects.

Capillary tubes made of two kinds of glass were used: "Schot-
schem, nr. 20" and the Bohemian glass made by CavaHer. The tubes
measured 100-200 /it and the length of the contained column of hquid
was from 8-30 )U. Some of the observations are summarized in
Table I. These results and others show that the type of glass in
which small amounts of culture medium are held may affect the con-
dition of the contained animals.

Kalmus concludes that his observations show that the retardation

124

ANIMAL AGGREGATIONS

of division in small volumes is approximately proportional to the
ratio of total surface to the volume of medium. Animals from young
cultures are more sensitive to limited volumes than are those from
old cultures. A fully bacterized medium retards the poisonous effect
of small volumes by furnishing more food and by tending to keep
the Paramecia out of the most toxic region next the glass, and by
binding the toxins present, thereby rendering them relatively harm-
less. These toxins may be of two sorts: there are the poisons which
may leach out of the glass into the limited amount of medium in

TABLE

I

Schotschem Glass

Cavalier Glass

24 hours:
Division of 25 pairs

II animals
2 pairs

4 animals

17 animals
I pair

5 animals

19 animals

I animal

Conjugating of 25 pairs

4 pairs

Dead of 25 pairs

9 animals

48 hours:

Division of 25 pairs

9 animals

Conjugating of 25 pairs

3 pairs

Dead of 25 pairs

15 animals

72 hours:

16 animals

Dead of 25 pairs

14 animals

29 animals

sufficient quantity to have decided effects, and there are the meta-
bohc products of the Protozoa themselves. The question of the fixing
of toxins takes us somewhat afield from our present considerations
and will be left to be taken up later in detail.

Unhke Crampton, Kalmus found that divisions of protozoans are
possible even when they are contained in small capillary tubes. Ap-
parently he did not subject his animals to the action of the centri-
fuge, which may partly account for the difference in results. How-
ever, when there are so many different factors operating, such as
composition of glass, age of culture, and bacterial flora, one cannot
be sure of the precise factor or factors causing the differences in
observed results. From his observations, Kalmus challenges the en-
tire conception that a small amount of available space, acting direct-
ly, may limit the rate of cell division and thereby the size of meta-
zoans. In this he overlooks the important results obtained by

RETARDING INFLUENCE OF CROWDING 125

Goetsch' on the effect of stimulation by contact with the walls of a
small container in limiting the growth of active animals, even though
the contained liquid be effectively connected with that of a much
larger vessel.

Undoubtedly there may be a limiting toxic effect of materials
leached out of glass, particularly from soft glass. The dangers result-
ing from the use of such glassware have been known for years. In ad-
dition there may be a physical as well as a chemical effect from the
glass walls of an inclosing vessel. Such effects are shown in the recent
work of Drzewina and Bohn (1927). They base their experiments on
the report of Norrish (1924, vide Taylor), who found that bromine
combines with ethylene about twice as fast in contact with a surface
of stearic acid as with one of glass, and that, on the other hand, the
reaction within a paraffin-lined dish is about one-thirtieth of that
given when the exposed surface is one of stearic acid. Using the
marine flatworm Convoluta, Drzewina and Bohn found that these
small worms survive only about half an hour when placed in sea
water in a glass dish coated with stearic acid. In this instance the
worms are not affected by dissolved chemicals, since stearic acid is
insoluble in water. There is no change in the pH of the water, and
water which has stood in such dishes is non-toxic when removed. A
glass dish coated with paraffin becomes less toxic than is a plain glass
dish. If the Convoluta in a glass dish on a white background are ex-
posed to sunlight, they do not maintain their normal activity so
long as when they are in a paraffined dish. There is also greater pro-
tection in the latter against the toxic action of metallic silver. Para-
mecia behave similarly. They die more rapidly in a dish covered with
stearic acid, whether in light or in shade. Paraffin protects them
against the action of metallic silver and of neutral red, even though
they take up as much neutral red in a paraffined dish as in a plain
glass dish. Drzewina and Bohn conclude that stearic acid catalyzes
reactions of living animals but that paraffin inhibits them. They sug-
gest that the action is similar to the action of paraffined glass in
preventing the coagulation of blood, and advance the theory that

' It is not yet definitely proven by chemical tests that the tubes with one end covered
by cloth do allow free diffusion of excretory products.

126 ANIMAL AGGREGATIONS

both efifects may be attributed to the electrical charge carried by the
paraffin.

Warren (1900), in his work on the effect of crowding on Daphnia,
had previously found that media in which excretory products are
allowed to accumulate cause a decrease in the number of genera-
tions and the number of offspring in a brood, and that reproduction
ceases long before the animals die. Such water is injurious, though
not usually fatal to fresh Daphnia; and the reproductive power of
the newly introduced Daphnia is soon reduced. The injurious nature
of the water seems to pass off after a sufficiently long period.

Our experience in growing Daphnia in quantity for fish food in a
considerable volume of water, of perhaps 10-100 liters, accords with
the experimental results of Warren. Events run as follows: A month
or 6 weeks after having stocked such an aquarium with a few Daph-
nia, conditions being favorable, several hundreds of animals may
be living in good condition and reproducing. Then suddenly a change
begins. The greater number die, young and old alike. Perhaps from
I to 3 per liter survive, and these will live for months without pro-
ducing eggs. After a very considerable time eggs are formed and
Daphnia may become fairly plentiful again, but the second swarm
is never as numerous as the first. During the time when the Daphnia
have ceased to reproduce and have, for the most part, died off, the
water may be teeming with other entomostracans, ostracods or cope-
pods.- This indicates a certain specificity in the effect of the Daphnia
metabolic wastes. The duration of the period of depression of repro-
duction is greatly shortened by keeping the food value of the medium
at a high level.

EFFECT OF CROWDING ON RATE OF EGG-LAYING OF HENS

The effect of density of population upon rate of reproduction in a
different medium and with animals far removed in habits and in the
evolutionary scale from Protozoa or Entomostraca was reported by
Pearl and Surface (1909) from the experiments of Professor Go well
of the Maine Agriculture Station. These men report the result of
investigations concerning egg production extending over several
years. The chickens studied were kept in pens containing 50, 100,

RETARDING INFLUENCE OF CROWDING

127

and 150 hens each. The pens with the smaller flocks provided 4.8 sq.
ft. of floor space per hen. In the largest flock this was reduced to
3.2 sq. ft. per individual. The number of pens is shown in Table II.

In all there were 700 pullets placed in the 50-bird pens, 500 in the
IOC-bird pens, and 750 in the 1 50-bird pens. Conditions varied some-
what from year to year, so that Pearl and Surface warn that "wher-
ever comparisons between years are instituted, great caution must
be exercised in drawing conclusions."

Due care was taken to select the members of the different pens
with hereditary constitutions equally disposed to egg-laying, so far
as this factor could be regulated. All were from the same breed, and

TABLE II

Year

So-Bird Pens

loo-Bird Pens

ISO-Bird Pens

1904-5

6
4
4

I
2
2

I

1905-6

2

1906-7

2

individuals were distributed among the different pens so that the
percentage from ancestors of different productivity were the same
throughout. The experiment with which we are concerned ran three
seasons. Results are graphically given in Figure 5, which shows the
mean annual egg production per hen. An inspection of this figure
shows that each year there is a trend toward reduced egg production
in the pens with the greatest number of birds. During two of the
three years the decrease in rate of laying is practically the same
between the 50- and the loo-chicken pen as it is between the 100 and
J50. The results obtained the first season, 1904-5, are different and
affect the mean differences, as is seen from the fact that the pens
with 50 birds produced on the average 129.69 eggs per season;
those with 100 produced 123.21, while those with 150 gave in. 68;
The mean difference between the pens with 50 birds and those with
150 amounted to 18.01 eggs per year.

The difference between the loo-bird pens and the 1 50-bird pens,
where there were two factors acting — increase of numbers and de-
crease of floor space — is approximately twice as great during these

128

ANIMAL AGGREGATIONS

three seasons as is the difference between the 50- and the loo-bird
pens, where numbers only were varied. The experiments on the
whole indicate, as Pearl and Surface conclude, that the mean an-
nual egg production is much influenced by the differences in environ-
mental factors present in the experiment.

140

\

\

N

i

^ Ii;^j:££f-^j-

f::

<0 /30

N 0^

^

>

^"^\

^

^

^■^ \

^ 120

\^^

\ ^

1

^

\

e)

>

,

kj

<

'■■■■■?._;'

//o

""-■■■X

>..

"^

kj

^

ZOO

'x.

)

so

£//?£> P£N5

/OO
B/RD P£NS

/SO
B/RD P£A/S

Fig. 5. — A graphic summary of the relation between size of flock and mean annual
egg production in the domestic fowl. (From Pearl and Surface, 1909.)

In an attempt to get at the underlying factors they suggest that
there is another element involved besides the physical density of the
population, which they are inclined to place on the psychological
level, and which works even when the amount of floor space per in-
dividual is equal. The conditioning of the surrounding medium is of
a different type from that of crowded aquatic animals, where the

RETARDING INFLUENCE OF CROWDING 129

excreta and glandular secretions are dissolved. in the surrounding
liquid medium and come of necessity into intimate contact with each
of the contained animals. Presumably, with chickens we are free
from inequalities in food, although in the larger pens some may
have fared better and others more poorly, especially in view of the
flock organization which Schjelderup-Ebbe (1922) has described.
Availability of equal floor space does not insure equality of use, and
crowding was probably greater the greater the numbers present.
Even so, the significance of these observations is not lessened, and
the conclusion of Pearl and Surface may be justified that we are here
dealing with physiological effects on the reproductive system pro-
duced by physiological effects on the nervous system of the order
usually spoken of as "psychological."

It becomes important to follow the differences in egg production
during the course of the year with these pullets housed with dift'erent
degrees of crowding. The results of such analyses are published by
Pearl and Surface (191 1) and are summarized in Figure 6. The
months from November to July are based on the averages of records
for 4 years; from July through September on the records of 3 years.
October is not included because records for only 2 years were avail-
able.

The data summarized in these graphs show that there is no harm-
ful effect from keeping pullets in large and crowded flocks during
early winter egg production near the beginning of the laying period.
In fact there appears to be a significant advantage accruing from the
crowding in the first really cold winter month, December. On the
other hand, the 50-bird pens show a distinctly better production
than do the other lots in late winter and early spring, about the time
of heaviest egg production. This difference does not obtain between
the birds kept in lots of 100 and those in lots of 150. The harmful
effects of summer crowding on egg production shows plainly when
the most crowded pullets are compared with less crowded lots.
Overcrowding affects summer egg production in a distinctly ad-
verse manner.

There would thus seem to be three distinct aspects of the effects
of crowding on egg production in the domestic fowl. First, in early

I30

ANIMAL AGGREGATIONS

winter at a time of relatively low egg production, when the nights
are becoming increasingly cold, the large crowded flocks apparently

c,2

J5

-1

ll
ij
ll

ll

1

;

/
1
/

1

1

(
/
/
/
/
/

I' 1

1 1
1 1

1

<

1

1 I
1

1 .••■

<

1 /

X

_^

/

^

i '

/ /
' /

V

C

\

/

\

\ \
\ \
\ \

\ \
\ \
\ \
\ ^

/ /
/ /
( /
/ /
/ /

/ /
I /
/ /
/ /
/ /

■

>■'

'

1

/VOV. DEC.

JAN

r£B. MAR. AP/i. nAY JUNE JULY AUG. 6EPr

Fig. 6. — Diagram showing the average excess in mean egg production of different
sized flocks of barred Plymouth Rock pullets in the first 1 1 months of their first year
of laying.

, 50-bird pens compared with loo-bird pens.

J 50-bird pens compared with 1 50-bird pens.

, loo-bird pens compared with 150-bird pens.

The broken lines running parallel with the zero line approximate the mean probable
error. Points below the zero line indicate that the larger number per pen gave a higher
average egg production. Points above the zero line show that the pen with the smaller
numbers gave the higher average. (From Pearl and Surface, 191 1.)

conserve animal heat, so that greater egg-laying occurs in the more
crowded pens. Second, as the period of maximum egg production is

RETARDING INFLUENCE OF CROWDING 131

reached, crowding has the opposite effect, for reasons not yet clear.
The nights are still cold, frequently colder than in December, when
the opposite results were obtained. Probably the differential effect
of crowding is associated with acclimatization to the cold ; the gen-
eral physiological condition of the hens must be different at the
height of the laying season from that at its beginning, and this shift
in physiological state may account for the reversed effect of crowd-
ing; still, perhaps the psychological factor invoked by Pearl and Sur-
face to cover admitted ignorance may be the only feasible suggestion
as yet. Third, following the approach of warm weather and the com-
ing of the hot summer months, the birds of the crowded pens prob-
ably have difficulty in maintaining comfortable temperatures, par-
ticularly while roosting.

In concluding this discussion of the effect of crowding on the rate
of egg production in chickens, it is of interest to note that the varia-
bility in the rate of egg production increases with crowding when the
annual egg production is taken as the unit. When this is broken into
monthly periods, it is seen that the greatest effect of crowding is to
be found at the beginning and the end of the laying-year, at a time
of low production. From February to July, at the time of heaviest
laying, the environmental differences implied by flock size as used in
these experiments do not affect the relative variability of produc-
tion. Unfortunately there are no data concerning egg production of
chickens isolated in pens with the floor areas per individual used in
these experiments.

EFFECT OF CROWDING ON RATE OF REPRODUCTION IN DROSOPHILA

Pearl and Parker (1922) have contributed another bit of signifi-
cant evidence to our problem by their work upon the influence of the
density of population upon the rate of reproduction in Drosophila.
In this work mass matings were made from a given line. The off-
spring from this mass mating were used in making up the matings in
the experiments to be described. Half-pint milk bottles were used as
containers. The procedure was definitely standardized throughout.
Sets of four bottles were started, each containing i, 2, 3, .... 9,
mated pairs of flies. Sets of three bottles contained, respectively, 10,

132 ANIMAL AGGREGATIONS

12, 15, 20, and 30 mated pairs; two bottles held 50 mated pairs each;
and one bottle had 25 mated pairs. At the end of 8 days at 25° C. the
surviving parent flies were transferred to fresh bottles for a second
breeding period of 8 days. The only variable known to be significant
throughout this series was the density of the population. All the
offspring from the two breeding periods were counted and sexed.
The results tabulated as the rate of reproduction per female per day
during the first 16 days of life are shown in Figure 7.

In this figure the circles give the observations, and the curve is the
graph of the following equation fitted by the method of least squares:

>' = 34-S3 g~''-"l\;~°''58 J

which in logarithmic form becomes:

log y = i. S4 — 0. 008.V — 0.658 log X ,

when y signifies the flies per mated female per day and x is the num-
ber of mated flies per bottle, taken over the whole 16-day period the
experiment ran.

The observations include a total of 23,922 progeny flies, which is
a large enough number to cause the results to be treated with re-
spect. Further, it is apparent that the curve fits the observed facts
closely. In the preceding chapter, I have called attention to the
fact that this formula is related to that which Bilski developed to
describe the effect of crowding on the rate of growth in tadpoles, to
that of Kennealy for the relation between length of race and the
record established for that distance, and to that of Farr for the rela-
tion between density of human population and the death-rate. These
phenomena must be based on a common fundamental biological
relationship.

When these results are analyzed further, we find that the greatest
drop in rate of reproduction of adult flies per female per day comes as
the number of original mated pairs per bottle increases from i to 2,
and the next greatest drop comes between the bottles having an
initial population of 2 and 3 mated pairs. This result cannot be due
to larval crowding, since the bottles containing 9 mated pairs of flies

RETARDING INFLUENCE OF CROWDING 133

22

80 90

/^EA/V Fl/ES P£/? BOTTLE

Fig. 7.— Pearl and Parker's (1922) curve showing the decrease in rate of reproduc-
tion in Drosophila as cultures become more crowded.

134 ANIMAL AGGREGATIONS

produced 2,117 adult offspring. The 80 cc. of banana-agar food with
an exposed surface of 23.76 sq. cm. per bottle must therefore have
been capable of supporting this number of larvae in the time avail-
able. The bottles with i, 2, and 3 original mated pairs produced, re-
spectively, 1,348, 1,124, and 1,877 total imagoes in 16 days. The food
available would have allowed at least 2,117 larvae to pupate and
produce adults.

The exposed area, as well as the amount of food present, has been
shown to have a distinct effect upon the numbers of Drosophila pro-
duced. Harnly (1929) varied the area of standard food with the

Fig. 8. — Showing the relation between productivity and the area of food with
Drosophila. The vertical column of figures gives density; figures on the graph show the
area of food surface in square centimeters. The corresponding total volume capacities of
the containers are: , 1181, 2365, 473, and 250. (Data and figure from Harnly.)

depth kept constant at 25 mm. The live areas tried were those fur-
nished by culturing the fhes in vials, 4-ounce bottles, half-pint, and
pint milk bottles, and in 250 cc. Erlenmeyer flasks. These different
culture-containers gave food surface areas of 4.4, 11, 24, 40, and 52
sq. cm., respectively. Summarized results are given in Figure 8,
which shows graphically the effect of surface food area upon the total
yield from a single pair mating for a period of 10 days. The largest
yield under these conditions was given by a surface area of 40 sq.
cm.; 52 sq. cm. had about the same productivity as 24 sq. cm. The
viability was greatest in the flies reared with the largest amount of
space.

RETARDING INFLUENCE OF CROWDING 135

The explanation of Harnly's results is not necessarily obvious or
easy. It may be that there is actually a surface-population optimum
which stands below the largest surface and volume of food available.
A possible factor may be that with greater area and volume wild
yeasts or molds grow too rapidly for the Drosophila to control. Be-
fore coming to this conclusion, it is well to note the sizes of the dif-
ferent containers, which were: vials, size not stated, 118, 236.5, 473,
and 250 ml. The population curve may be a result of the available,
space rather than of the food surface acting alone. Such an interpre-
tation would be in line with the data of Pearl and Parker shown in
Figure 7. More work is needed, however, before one can draw as-
sured conclusions.

The tendency toward universality of the effect of crowding upon
the rate of reproduction is shown by the fact that Hill (1926) and
Sarles (1929), working with hookworms, have reported counts on
population density of these parasites in relation to egg production
which show that as the number of worms in a given host increases,
the egg output per worm decreases.

Pearl and Parker conclude the account of their work upon crowd-
ing and the rate of reproduction in Drosophila with the following
statement, which Pearl repeats in a later book: 'Tn general there
can be no question that this whole matter of influence of density of
population in all senses, upon biological phenomena, deserves a great
deal more attention than it has had. The indications all are that it
is the most important and significant element in the biological, as
distinguished from the physical, environment of organisms." With
this position I am in complete accord.

CHAPTER VIII
CROWDING AND INCREASED DEATH-RATE

Measurements of growth, reproduction and length of hfe, sum up
many of the physiological processes that may be affected by crowd-
ing; the first two of these have already been considered in some de-
tail. These three functions are closely connected, and it has been
impossible to keep their treatment entirely separate. Thus, in the
preceding chapter, the discussion of the effect upon the rate of re-
production of confining Paramecia within a small capillary tube was
extended to include a partial discussion of such treatment upon the
longevity of the animals in order to get suflficient control of the avail-
able evidence to be able properly to evaluate factors affecting the
decrease in rate of reproduction brought about by crowding.

Inspection of the material previously presented demonstrates that
the ability of adult organisms to live is not necessarily the same as
their ability to reproduce. Kuczynski (1928), in studying the bal-
ance of birth and deaths among the human population of Western
Europe, describes the differential effect of changing conditions upon
fertihty and upon the death-rate, and concludes that human fertility
has become a problem in itself largely divorced from the problem of
mortality.

The experience of Warren that Daphnia lose their reproductive
capacity long before they die, and that in such a condition they may
be able to live through adverse conditions produced by overcrowding
and again take up reproduction when conditions become more favor-
able, is a case in point. Kalmus adds observations upon Paramecia
along the same line. The usual interaction between fertility and mor-
tality is such that in a given amount of liquid medium the popula-
tion increases to a maximum whose size depends on the volume of the
medium and the concentration of the food material, and then gradu-
ally falls to complete or nearly complete extinction. This course of

136

CROWDING AND INCREASED DEATH-RATE

137

events is shown in Figure 9, which is taken from the work of Myers
with Paramecia.

If the initial volume is relatively large (16 drops, or about 0.8 ml.),
Myers finds that fission begins at about the same rate, so far as 1 2-
hour periods of observation show, regardless of whether the seeding
be with I, 2, or 4 individuals. The population of such cultures rises
rapidly to a peak which is essentially the same for all the seedings
just mentioned; at its peak it ranges from 123 to 126 individuals and

no

^.•/<}

Z:

'6

/

i/.V<^

1

\

/

100

^

/

'\

/

80

/

/

'■

,'

/

K

\

\

. 8/<5

''^

/

\

,;;

/

^N

\

1/

\

\

j/

/

>

'

*

A

^

''

/

\

\

/

y^/

/

\

\ \

\

y

,'

-''

^

\

■^

^ \

\

^ii

-s^

'^

N

'■■•

. >

t^-

^

It

IB

Fig. 9. — Showing the rise and decline of populations of Paramecia in 0.8 cc. of
culture fluid by, respectively, i, 2, 4, and 8 individuals. The horizontal axis shows
successive periods of 12 hours each; the vertical axis gives numbers of individuals in
the populations. (From Myers, 1927.)

then falls off at about the same rate. The cultures seeded with a
single individual take longer to reach the maximum than do those
seeded with 2 or 4, but otherwise the course of the population history
is similar.

When 8 individuals are introduced in place of i, 2, or 4, the maxi-
mum in Myers' cultures showed a population of only half that given
with the smaller seedings. The reason for this difference is not clear.
On the surface it appears that the larger initial seeding either ex-
hausts the food supply more speedily or poisons the culture medium

138 ANIMAL AGGREGATIONS

more rapidly than do the other seedings or acts in both ways at the
same time.

The point of especial interest to us in these observations is the
fact that with certain initial densities of populations, even in an un-
changed medium, the maximum reached is practically identical and
independent of the numbers originally introduced.

The effect of the exhaustion of food supply has been eliminated by
Chapman (1928) in his work with the confused flour beetle, Triboli-
um confusum. Chapman introduced varying numbers of these beetles
into a definite amount of whole-wheat flour and found for this insect,
as Robertson, Cutler and Crump, Myers, and others had previously
found with various Protozoa, that there is a definite limit to the
number of organisms that will develop in a unit volume of culture
medium.

Chapman's work does, however, introduce one new fact into the
situation. His choice of experimental material is particularly for-
tunate in that the beetles can be screened out of their floury environ-
ment and the eggs, larvae, and pupae, as well as imagoes, can be
counted and the flour renewed at each observation. Hence, in place
of the usual more or less symmetrical population curve found by
other workers dealing with a population composed of all age groups,
in which the population, after rising to the maximum determined by
the nature of the culture medium and the amount of space available,
falls away to approximate or total extinction on account of the ex-
haustion of food or the addition of excretory products. Chapman is
able, by periodically renewing the environment, to carry his beetle
population along for extended periods, perhaps indefinitely, with
approximately the same number of individuals present per gram of
flour. In his terminology, "a condition of equilibrium is attained in
which the biotic potential is equalled by the environmental resist-
ance and the population remains relatively constant."

Appropriate tests showed that the stationary character of the
population when in equilibrium was not due to absence of eggs or to
their lack of fertility. Rather, the lack of increase in population be-
yond a certain point was due to the eggs, pupae, and, to some extent,
the larvae, being eaten by the adult beetles. When eggs were placed

CROWDING AND INCREASED DEATH-RATE

139

in flour cultures containing only male beetles, the percentage of
eggs eaten varied directly with the population of adults per gram of
flour.

Chapman's experience with Triholium can be shown in a number
of ways; perhaps as significant as any is the result of carrying to
equilibrium a series of beetle environments of different sizes, of
4-128 grams of whole-wheat flour, seeded with i, 2, 4, 8, 16, and 32
pairs of beetles each, making one pair of introduced beetles per 4
grams of flour. This experiment may be followed in Table III, which

TABLE III

Beetles {TriboUitm confiisum) per Gram
(Data from Chapman

OF Whole-Wheat Flour
, 1928)

Days

4G.

8G.

16 G.

32 G.

64 G.

128 G.

I c

15
30

35
39
35
40
48
37
38
46

O-S
17
25
33
39
41
46

45
50
49
49

0-5
20
26
32

34
39
38
36
41
46
46

0-5
17

22

35
39
36
44
43
41
44
43

O-S
21

24
32
40
37
49
40
48
45
42

05
19

30

50

64

78

lOI

23
34
37
39
39

"4

I J4

40
45

156

171

47
40

gives the total number of the beetles present per gram of flour at
different times in the history of the cultures, regardl-ess of the de-
velopmental stage of the beetles.

In all the conditions tested by Chapman, the mean number of in-
dividuals per gram of flour after equilibrium was established was
43.97, with a standard deviation of 4.27, and a probable error of 2.88.
Chapman found, as will be shown in detail later (chap, x), that the
time taken to reach this equilibrium differed with the initial seeding
per gram of available flour, while the equilibrium population re-
mained constant per gram of flour; and that, with the same initial
seeding throughout, the equilibrium population per gram of flour re-
mained constant but the time taken to reach equilibrium varied
directly with the quantity of flour available. This equilibrium is
primarily a food relation, or a food and space relation, since the

I40 ANIMAL AGGREGATIONS

metabolic products are removed by the periodic changes of the flour.
The equilibrium is apparently based upon a competition between
adults and larvae, as is shown by the fact that in one 1 6-gram en-
vironment the number of adults was accidentally reduced on the
seventy-eighth day of the experiment and was never returned to its
place in the geometric series, while the number in its total popula-
tion — eggs, larvae, pupae, and adults — did so return.

Following current tendencies. Chapman interprets his findings in
terms of a mathematical formula, C = Bp{R), when C is the concen-
tration of insects, R is the environmental resistance, and Bp is the
biotic potential, which he defines as the mean maximum rate of re-
production in a given period under given conditions. Substituting
and solving, we find :

^^ (43-97X8.4). 25 ^^ ^
43-97

The concentration of insects per gram of flour is 43.97. The aver-
age number of eggs laid per day in these experiments is 8.4, and one-
fourth of the population are egg-producing females. The formula so
given represents the state of equilibrium only.

The work we have been discussing summarizes the effect of the
environmental factors associated with crowding upon the total popu-
lation. The work which deals most directly with the harmful effects
of crowding on length of life is that of Drzewina and Bohn. In con-
nection with their studies on the relation existing between mass of
toxic liquids and the contained mass of animals, Drzewina and Bohn
have found that many cases of protection are furnished by increasing
the numbers of animals present in the same solution. These wifl be
reported later. In some instances they record the opposite results
(i92i(^, 1922).

When KCl was used as a toxic agent with cultures of Convoluta, a
smaU marine planarian, other things being equal, those in the solu-
tion containing the larger number died first. Similar relations hold
when the same number of individuals are placed in differing amounts
of the same strength of KCl solution: those in the smaller amount of
liquid die more rapidly. The fresh-water planarian Polycelis nigra

CROWDING AND INCREASED DEATH-RATE 141

reacts similarly. These investigators believe that the planarians give
off a substance which causes autodestruction, and that, if this be
true, such destruction is hastened by increasing the mass of indi-
viduals in proportion to the amount of liquid.

Their interpretation is supported by the observation that if fresh
Planaria are introduced into a solution of KCl which has already
contained others, their death is hastened. If, after some time in such
a solution, a part are removed to a new solution of similar strength
of KCl, these die more slowly than do their fellows left in the original
solution, which contained not only KCl but also some substance or
substances given off by the worms themselves.

Later (1928) they observed that around cytolyzing flatworms the
H-ion concentration (acidity) of the solution is greatly increased, and
they considered this to be the factor which causes the increased
mortahty of the groups. The larger the number of cytolyzing in-
dividuals, the more rapid and the greater the increase of the H-ion
concentration, and consequently the more rapid and pronounced are
the lethal elTects involved. Such a process accelerates itself in the
presence of many individuals, or, on the other hand, is not effective
in the case of a few scattered animals. If the latter die, they do so
because of the lethal effect of the KCl alone.

Fowler (1927, 193 1) undertook to test the effect of a large number
of electrolytes upon the rate of survival in certain crustaceans, using
mainly a species of Daphnia. His results show that there is a distinct
correlation between the survival value of the group and the degree
of toxicity of the salt solutions employed. Tests upon the rate of
oxygen consumption have shown that crowded Daphnia, in the con-
centrations tested, use less oxygen per individual than do isolated
Daphnia under similar conditions. When the toxicity is sufficiently
great so that death occurs within a relatively short time, this group
depression tends to favor group survival. On the other hand, when
the concentration is low and the effect of the chemicals is de-
ferred, the isolated individuals live longer than the group. Such re-
sults are in accord with Child's (191 5) differential susceptibility find-
ings when planarians are subjected to relatively strong or relatively
weak concentrations of various toxic agents, particularly KCN.

142 ANIMAL AGGREGATIONS

Under conditions of high toxicity the animals, or, in the case of
Child's worms, certain regions, which have the lowest rate of general
metabolism are least affected by the toxic agent and survive longest.
With weaker solutions, on the other hand, the most vigorous indi-
viduals, or, in the case of the worms, the most vigorous regions, can
acclimate most readily and hence survive longer. Fowler's results
show that depression due to crowding may have definite survival
value under certain conditions but that with weaker concentrations
of even the same salts crowding decreases the chance of survival. In
the latter aspect his results support the facts reported by Drzewina
and Bohn in their experiments with KCl. Fowler's results fail to
support the hypothesis that a specific autodestructive material is
produced. They extend the later explanation of Drzewina and Bohn
by indicating that the unknown autodestructive substance is the
carbon dioxide produced by the animals, which does raise the H-ion
concentration as Drzewina and Bohn determined.

For further consideration of the relation between density of popu-
lation and the death-rate it seems best to take up the case with re-
spect to man, since with human populations this relationship has
attracted particular attention for a considerable period of time. We
have already referred to the generalization known as Farr's law; this
states that if the death-rate be represented by R and the density of
population per unit area by D, then R = cD"\ where c and m are con-
stants.

Brownlee (191 5) rehabilitated this law by showing that the statis-
tics used by Farr, which came from the decade 1861-70, compared
favorably, so far as the relation between population density and
death-rate was concerned, with those of the decade 1891-1900, as
given by Tatham. Brownlee's republication of these tables and his
calculations are given herewith; see Table IV.

Brownlee calls attention to the fact that the values of m corre-
spond roughly for each type of analysis in the two periods but that
in the case of the life-table death-rates they correspond to the third
decimal place, which is as much as could be statistically expected.
He concludes that Farr's law is thus shown to be a definite law oper-
ating independently of the changes due to sanitary progress. Re-

CROWDING AND INCREASED DEATH-RATE

143

gardless of improvements in sanitation and in medicine, the ex-
ponent does not vary, but only the multiplying constant. Therefore
m represents the law, while c represents rather the coefficient of

TABLE IV^

No. of
Districts

Persons per
Sq. Mi.

Corrected

Death-

Rate

ame Fitted

Crude

by Least

Death-

Squares

Rate

Same Fitted
by Farr

Life-Table

Deatii-

Rate

Same Fitted
by Least
Squares

A. Showing Figures Relating to Density and Death-Rate, 1861-70

53

345

137

47

9

1

I

*

t

X

166

1550

16.70

16.75

18.90

19.90

20

186

17 .02

17

00

19

16

19

lb

21

07

20

379

20.52

18

99

20

87

20

87

23

97

22

1,718

24-35

24

03

25

02

25

02

25

09

26

4,499

27.94

27

92

28

08

28

08

28

54

28

12,357

33 98

32

67

32

70

32

70

32

67

31

65,823

40.55

42

39

38

74

38

74

37

17

37

E% = 3.79§

E% = 2 . 70

K% =

A =

1. 17

A =

.90

A =

B. The Same for 1891-1900

27
112
121

92

53
56

31

40

31
21
18

13
6

5
5
4

**

It

136

11.63

13.06

14.20

14.16

17.38

161

12

54

13

43

15

05

14

SI

18

01

181

13

44

13

70

15

44

14

68

18

62

261

14

52

14

56

15

46

15

38

19

36

407

15

53

15

68

16

08

16

28

20

OS

457

16

53

15

99

16

67

16

52

20

24

734

17

58

17

32

17

64

17

56

21

45

1,303

18

S3

19

05

18

04

18

88

22

10

1.705

19

42

19

93

18

61

19

54

22

71

2,339

20

37

21

00

19

50

20

35

23

3t>

4,424

21

56

23

37

20

21

22

08

24

18

4,884

22

36

23

76

20

69

22

35

24

72

4,194

23

48

23

16

22

05

21

93

25

49

2,925

24

33

21

80

23

29

20

94

26

07

7,480

26

54

25

51

24

74

23

60

27

S8

55,563

34

82

32

67

23

67

30

49

33

25

E% = 4.3

E% = 3.8

A = I . OS

A = i.i4

XX

17.18
18.12

18.33

19.02

19.90
20.13

21 . 12
22.31
22.99
23.72
25-31
25 56
25.10
24. 21
26.68
32.58

E9o = 2.03
A= .63

H From Brownlee, Journal oj Hygiene, XV, 16.

§ E%=mean experimental error; A ='V of the mean of the squares of the errors.
**i? = i2.4oZ)-""s tt^ = i3-57£'-"'" tti? = io.83£>.">«'»

sanitary and general conditions of health. The coefficient c decreased
from 12.42 in the earlier period to 10.83 for the later. In other words,
in the same general area, conditions of living have so improved in

144 ANIMAL AGGREGATIONS

the thirty years' interval shown in Table IV that density of popula-
tion had only 0.875 the effect during the 1890's that it had in the
i86o's. Here we have evidence that relatively mild crowding affects
longevity in men.

This is to be expected when we consider the relatively greater
ease of transmission of contagious diseases in the more crowded
areas. Such dangers from the crowd are illustrated in a simplified
form by one of the Ophioderma experiments to be reported in full in
another connection. In these experiments the survival of 8 isolated
brittle starfish, each in a i-liter Erlenmeyer flask, was compared with
a group of 8 similar starfish in an 8-liter bottle. Usually the group
outlived the isolated individuals; but on one occasion one of the
members of the group died soon after the daily inspection and change
of water, and so polluted the whole 8 liters that all the remainder
were dead on the following morning. When an isolated animal died
similarly, the effects of its death could not extend beyond the limits
of its single flask.

When we pass in review the materials presented in the chapters of
which this is the third, we find much evidence supporting the gener-
ally accepted dictum that crowding is harmful for poorly integrated
groups of animals, breeding and hibernation seasons excepted. We
have seen that crowding may slow down the rate of growth and may
result in dwarfed individuals, that the rate of reproduction may be
decreased, and that the death-rate may be greater. These effects
have been reported for so many different animals from such a wide
range of the animal kingdom that there can be no doubt of their
general significance. But this is not the whole story. In many of the
experiments to be reported in our next section, we shall find that
crowding does not always produce harmful results; and that under
many conditions there are distinctly beneficial results, providing the
crowding be not too great. When considering these beneficial re-
sults, we must, however, always keep in mind the harmful effects of
overcrowding.

BENEFICIAL EFFECTS OF AGGREGATIONS

CHAPTER IX
STIMULATION OF GROWTH BY CROWDING

Having sketched in some detail the harmful effects resulting from
the crowding of many animals into a relatively small space, it is now
possible, with a better perspective, to look into the more recently ac-
cumulated evidence that harmful results do not necessarily follow
the formation of such aggregations and that they are often useful and
even necessary to the welfare of the individual.

The extent to which the phenomenon of aggregation affects the
rate of growth in a positive manner has been relatively little in-
vestigated. In the work of Colton (1908) upon Lyninaea it will be re-
called that he found crowding generally decreased the rate of growth
in snails. He found, however, that the snail faeces, if washed free of
easily soluble material and placed in weak solutions with snails,
tended to increase the rate of growth. With concentrated solutions
of faeces the results were reversed. Similarly, weak solutions of urea
favored snail growth, though stronger solutions retarded it.

Popovici-Baznosanu (192 1) also found that under certain condi-
tions snails grew more rapidly in stagnant water, conditioned by the
snails, than in fresh water. In short experiments (1914) 10 young
Lymnaea attained a length of 9.5 mm. in fresh water while those
living in stagnant water grew to 10 mm. Later he tested this effect
for a longer period. The young Lymnaea from three egg masses were
placed in three culture dishes of identical dimensions as regards vol-
ume and surface of water; after a long sojourn, when the water was
thoroughly snail-conditioned, Popovici-Baznosanu took half of the
individuals and placed them in better conditions of existence, in
culture jars with a large volume and a relatively large surface, and
containing fresh pure water. Elodea was used as food both for those
in the stale and those in the fresh water, and in the same quantity
for both. After 106 days the results were as given in Table V.

147

148

ANIMAL AGGREGATIONS

In only one of the three cases was there a clearly significant dif-
ference; yet Popovici-Baznosanu interpreted these results as mean-
ing that in the stagnant, snail-conditioned water, the higher plants
present, as well as the walls of the jar, are covered by growths of
microflora, which he regards as forming the chief food of the snails;
and that the snails therefore grew more rapidly in cultures contain-
ing a rich microflora than in those with only a scanty supply. Colton
had interpreted his results similarly.

The observations of Eigenbrodt (1925) that Drosophila grow larg-
er in small culture vials when present in numbers of from 8 to 16 than

TABLE V*

Brood

Surface of Water
(Sq. Cm.)

Volume of Water
(Cc.)

Condition of
Water

Length of Largest
Shell (Mm.)

I

/ "3

1,300
1 ,900

1,300
4,510

1,300
1,640

Conditioned
Raw

Conditioned
Raw

Conditioned
Raw

19

II

\ 133

/ 113

1 208

/ 113

\ 116

95
19

Ill

18

15
15

* Data from Popovici-Baznosanu.

at other population densities may be explained on the assumption
that too few Drosophila larvae per culture fail to control the growth
of harmful elements of the yeast or bacterial flora as well as optimal
numbers do, while overcrowding overcontrols the growth of the food
plant. This would result in a growth optimum occurring, as sug-
gested, at a relatively low population density but distinctly above
the minimum populations studied. These results should be compared
with the relation between numbers present and Drosophila survival
given in chapter xiv.

Bilski (1926) tested the effect of crowding upon the rate of re-
generation of the tails of Rana esculenta tadpoles. Five of these tad-
poles were kept isolated, and five similar ones were placed together
in the same sort of dish and with the same amount of water which
was given to each of the singles. Although in most cases there was a
decrease in length of body from tip of head to the root of the tail,
there was growth both of the tail stump and of regenerated material.

STIMULATION OF GROWTH BY CROWDING

149

The proportions of decrease and of growth or regeneration differ be-
tween isolated and grouped animals. The results for the 7 days the
animals were observed are given in Table VI. The results indicate, as
much as a single experiment is likely to, that there is a greater re-
generation with the decreased volume per animal, which is compen-
sated by the greater growth of the tail stump when the animals are
isolated. Bilski states that this experiment is supported by his gen-
eral experience in many similar experiments in other phases of the
work, but cites no direct support of these results.

TABLE VI

Showing the Effect of Crowding on the Rate of Regeneration of

Tails of Frog Tadpoles in 7 Days. (In Millimeters)

(Data from Bilski)

Difference

Percentage of Difference*

Conditions

Body Length
(B)

Tail Stump
(S)

Body Length
(S)

Tail Stump
(S)

Regenerated

(R)

Isolated

Grouped

-1.4

-0.7

2-5
2.0

-II. 8
- 6.2

151
12.6

151
1B.3

B, length from tip of head to root of tail.

5, length of tail stump after cutting. R, length of regenerated material.

* Percentage B is calculated on the basis of the original body length; percentages 5 and R are in terms
of the original tail length before operation.

CROWDING IN TISSUE CULTURES

Work with tissue cultures has yielded pertinent evidence concern-
ing the beneficial effects of crowding on growth. The literature in
this field is enormous; and no attempt will be made to cover the
different ramifications of the subject, with most of which we are not
immediately concerned. It has been known for some years (Carrel,
1924) that tissues to be grown in vitro must have a proper back-
ground on which to creep. One of the most used backgrounds is of
fibrin network. In many recent studies this is placed as blood-plasma
in a thin layer over the bottom of a special culture flask. The tissue
to be cultured is introduced aseptically into this sterile medium,
which is then covered with a sterile fluid that has Tyrode solution as
its main ingredient but which contains other materials such as serum
or a saline extract of embryonic tissues. The latter, or some fraction
thereof, appears to be necessary for real growth of such cells as

I50 ANIMAL AGGREGATIONS

fibroblasts or epithelial cells. Extracts of sarcomas are superficially
similar to extracts of embryos in their growth-producing qualities.
Leucocytes (macrophages) behave in reverse fashion, growing per-
manently in pure serum and being inhibited by the presence of em-
bryonic extracts.

Inorganic substances, oxygen excepted, apparently do not affect
growth-rate of cells in vitro when present in approximately the same
concentrations as in the blood of the animal furnishing the tissues
under cultivation. Any departure from such concentrations yields
adverse results. Only approximately isotonic media allow indefinite
survival. The exact nature of the growth-promoting substance found
in embryonic extracts is still unknown. It appears to be associated
with the protein fraction and is particularly associated with pro-
teoses which result from a brief digestion of the protein with peptone
(Carrel and Baker, 1926). Prolonged digestion destroys the effec-
tiveness of this material. Willmer (1928) has been unable to confirm
this work, but concludes from the evidence furnished by Carrel,
Baker, and others that tissues can get some energy from amino-
acids but that their nitrogen supply is chiefly obtained from pro-
teoses (embryo extract contains both elements). These are heat-
stable substances; but most workers find that there is present in
embryonic juice a thermolabile growth-promoting substance which
is easily destroyed by heat or is adsorbed when heated, which does
not pass through a Chamberlain filter, and which is destroyed by
prolonged shakings. Carrel (1924) has called such substances "tre-
phones"; Fischer (1925a) calls supposedly similar substances "des-
mones"; and Burrows and Johnson (1925) named them the
"archusia."

Tissue-culture workers appear to be agreed upon the necessity of
keeping the cells from normal tissues in numbers, for successful cul-
tivation in vitro. Harrison (1928) says in this connection: 'Tt is a
very interesting and at present inexplicable fact that single somatic
cells isolated in culture media do not proliferate. Experiments to
this end made in my own laboratory some years ago but not pub-
lished did not succeed and other workers have reported similar ex-
perience. As Fischer puts it, a colony of fibroblasts cannot arise

STIMULATION OF GROWTH BY CROWDING 151

from a single cell even when the nutrient conditions are most favor-
able. Likewise small groups of cells if isolated do not undergo divi-
sion and their growth remains at a standstill. On the other hand
certain tumor cells (Rous chicken sarcoma) are capable of multiply-
ing and producing colonies when isolated singly." Similarly, we
know that in nature single egg cells will grow. The germinal area of
the hen's egg is an excellent example of an isolated bit of protoplasm
which, under fav^orable conditions, will grow. It is of interest to us
to note that Wright (1926) has found by dialysis a growth-stimulant
in the incubated yolk of hen eggs which is not shown when such
yolk is added directly to tissue-culture medium without dialysis.

Haberlandt {vide Fischer), in his work with plant cells, could se-
cure increase in size from certain isolated cells but did not find cell
division in such cultures. He (Haberlandt, 1919-22) reports a direct
relation between the size of the piece of plant tissue transplanted, or
the number of cells within it, and the number of cell divisions. From
these studies this investigator has concluded that the inciting to cell
division comes from substance given off by injured cells, which he
terms "wound hormones" or "division hormones."

A dramatic instance of the effect of heterotypic crowding upon
growth of tissue cells in vitro is furnished by Carrel and Ebeling
(1923). Cultures of leucocytes and of fibroblasts were made together
in the same flask of plasma. As usual under these conditions, the
fibroblasts did not grow, while the leucocytes grew well. In time they
spread until they came in contact with the languishing fibroblasts,
when a marked revival and initiation of growth took place in the
latter cells. This agrees with the generally known fact that in the
tissues of early embryos, when growth is taking place most rapidly,
there is a mass of growing tissue tightly packed together which is
supplied with a relatively small amount of blood. In tissue cultures
growth takes place best when the cells are present in relatively large
numbers in a small amount of medium which is stagnant but proper-
ly supplied with oxygen. Both kinds of observations suggest that
the cells forming metazoan tissues are dependent greatly upon one
another for their growth.

Fischer (1925) has suggested that this dependence is due to the

152 ANIMAL AGGREGATIONS

slow diffusion of products of metabolism or secretions from one cell
to another. He thinks these travel by protoplasmic bridges and are
independent of Carrel's "trephones," since fibroblasts that cease to
grow in the presence of an abundance of these trephones may be
restored to rapid growth by the presence of active healthy cells.

Burrows and his co-workers (1925, 1926) have put forward an in-
teresting and ingenious suggestion which explains many aspects of
the interrelations between cells and the fact that they must be pres-
ent in numbers before growth will occur and at the same time ex-
plains other characteristic activities of cells in tissue cultures. These
workers suggest that in the presence of a sufficient amount of oxygen,
about one-third of an atmosphere, the cells secrete a hypothetical
substance or group of related substances which as stated above are
called "archusia," which are supposed to function somewhat like the
desmones of Fischer except that the function of archusia is profound-
ly modified by their concentration. If present in high concentration,
they display an enzyme-like action which causes self-digestion of the
tissues; if the concentration is somewhat lower, the presence of ar-
chusia allows the cells to digest fats and proteins and to grow, pro-
viding the medium is otherwise suitable. In more dilute solutions,
tissue growth ceases; but the cells display their characteristic mi-
grating ability, which is frequently shown in cultures, or parts of
cultures, in which no growth is going forward. In yet more dilute
concentrations, the cells lose their power of carrying on their ordi-
nary activities, round up, and become dormant.

Archusia are water soluble, are secreted by cells, and can diffuse
through cell membranes to the outside medium. They tend to collect
in quantity when many cells are together in a minimum volume
under stagnant conditions, which are known to favor growth of
tissue cultures. When too great a volume of medium is present in
proportion to the number of cells, or if such cells as fibroblasts are
isolated, archusia escape into the surrounding medium and growth
ceases. Cells isolated into sufficiently small volume should grow,
according to the implications of this hypothesis; but the needed
volume may be so small that other complicating factors arise. Such
a substance would also be carried away by repeated washings, which

STIMULATION OF GROWTH BY CROWDING 153

are known to be harmful to cells whether grown in vitro or in vivo.
Archusia have properties resembling bios and vitamin B and have
been thought to be identical with the latter.

The whole concept of archusia is in the hypothetical stage at pres-
ent, and more evidence is needed before coming to a definite con-
clusion concerning its validity.

Heaton (1926) has worked upon the effect of vitamin B upon the
growth of cells in vitro. He finds two elements present in extracts of
yeast and of liver — one which stimulates growth and another which
depresses it. The two can be separated by their different solubility in
alcohol. Burrows and Jorstad (1926) think that vitamin A is neces-
sary for the functioning of cells and is produced when cells are digest-
ing fats and growing under the stimulus of relatively high concentra-
tions of archusia (vitamin B?). They regard vitamins A and B as
antagonistic, and balanced in cells that are functioning. In fact,
most observers are agreed that fats and lipoids are associated with
substances which inhibit the growth of cells, while some portion of
the protein molecule is associated with the promotion of growth.

EFFECT OF CROWDING ON GROWTH OF SEA-URCHIN PLUTEI

Certain of these results of the tissue culturists have been applied
to the problem of the effect of crowding upon the rate of cleavage and
of the growth of the arms of sea-urchin plutei by Peebles (1929). By
treating extracts of sea-urchin eggs and larvae with alcohol or with
acetone, a growth-inhibiting substance was obtained which definitely
retarded the rate of growth of eggs or of plutei. When this fraction
containing lipoids was partially removed, growth acceleration was
observed, as shown in Figure 10. Further experiments showed that
there was a decided difference in growth, depending on whether the
alcohol-soluble or alcohol-insoluble fractions of extracts of echino-
derm plutei were used. These results are shown graphically in Fig. 1 1 .
Peebles was also able to remove growth-inhibiting substances from
such extracts by adsorption, as shown in Fig. 12, but has not been
able as yet to isolate either the growth-inhibiting or the growth-
promoting principle.

Peebles, in summarizing her work, says: "The eggs and larvae of

154

ANIMAL AGGREGATIONS

<o

50

60

80 90 100

Time I a Minutes

Fig. io. — Showing the rate of cleavage in eggs of the sea urchin treated with acetone
extract from which the fat has been partially removed. Acetone (extract) minus fats, O .
Control, A. (From Peebles, 1929.)

a

I

60

70 80 90 100

Time in Afinutes

Fig. II. — A comparison of the results obtained by Peebles (1929) by using the filtered
alcoholic extracts of plutei (AA) with that of the precipitate (x, v). Ninety-seven per
cent ethyl alcohol (filtrate), A; 75 per cent ethyl alcohol (filtrate), A; control, O ; 75
percent ethyl alcohol (precipitate), X; 97 per cent ethyl alcohol (precipitate), V.

STIMULATION OF GROWTH BY CROWDING

155

the sea urchin and starfish contain growth promoting substances.
These substances pass out into the surrounding water during seg-
mentation, and later stages of larval development There is

55

50

45

<t, 40

35

^ 30

25

20

20

30

40

Time

50

in Hours

60

Fig. 12. — Figure showing growth in length of plutei in the presence of animal
charcoal (A) and fuller's earth (x) compared with those growing in sea- water (O).
(From Peebles, 1929.)

some experimental evidence in favor of the conclusion that the in-
hibiting substances are associated with the lipoid constituents and
the accelerating factor is contained in the protein molecule

iS6 ANIMAL AGGREGATIONS

The retarding effects of secretions of growing embryos are removed in
the presence of animal charcoal and fuller's earth. The percentage of
normal larvae resulting from eggs grown in the presence of these
adsorbents is greatly increased, while mortality is decreased."

By using hanging-drop cultures, a part of which contained isolated
sea-urchin eggs while the others held small groups of eggs, Frank and
Kurepina (1930) report accelerated growth in the grouped eggs.
These results are particularly noticeable if the temperature is al-
lowed to rise slightly above the normal. A resume of two of their

TABLE VII
A

Number of eggs per drop 1-2 16-20

Number of such drops ^ . . lo 3

8| hours after fertiUzation 75% have 8 blasto- o % at 8 blastomeres

meres 12% past 8 blasto-
25% past 8 blasto- meres

meres 88% at 16 blastomeres

B

Number of eggs per drop 1-2

Number of such drops 10

42:1 hours after fertiliza-
tion /S*: no movement

25% slight move-
ment
10% plainly mov-
ing
0% gastrulae*

* Percentages as reported in original work.

3-6

• ia-20

7

5

0% no movement

0% no movement

26% slight move-

4% slight move-

ment

ment

40% plainly mov-

12% plainly mov-

ing

ing

34% vigorously

60% vigorously

moving

moving

0% gastrulae

24% gastrulae

experiments, showing the type of results obtained under these con-
ditions, is given in Table VII. These results are interpreted by the
experimenters to indicate a stimulating effect of self-radiation as
suggested by Gurwitsch's mitogenetic rays. It is clear that such an
interpretation is far-fetched at present; but the results indicate,
despite careless reporting, that there is an optimum number of eggs
which lies well above the minimum at which, under certain condi-
tions at least, growth is favored as compared with that shown by
eggs isolated into similar amounts of sea-water.

STIMULATION OF GROWTH BY CROWDING 157

HETEROTYPIC CROWDING IN TISSUE CULTURES

We have noted above the case of fibroblasts growing in plasma
only when under the close influence of leucocytes, as an instance of
the direct effect of different kinds of tissues grown together upon the
ability of the one to grow at all ; there is also evidence that differen-
tiation is stimulated or accelerated by the presence of two sorts of
cells in close association. Thus Ebeling and Fischer (1922) combined
a ten-year-old strain of fibroblasts grown in pure culture with a two
months' strain of epithelium which had been similarly grown in pure
culture. After the two had been grown together for some time, the
epithelium became rounded into a sort of epithelial glandular tissue
lying within a supporting network of fibroblast cells. Champy (191 4)
and Drew (1923) have reported somewhat similar results from com-
bining these two kinds of tissue cells into one culture.

GROWTH-PROMOTING SUBSTANCES

The possibility of growth being promoted by small amounts of
obscure chemical substances is indicated by the well-known work
upon vitamins in connection with the growth and well-being of man
and certain other mammals and of birds. The exact application of
the facts developed in connection with work on vitamins with ani-
mals at the level of group life with which we have been dealing is at
present unknown, since practically no work has been done upon the
vitamin relations of the invertebrates and little upon those of the
lower vertebrates. From the work upon the higher vertebrates we
know that vitamins A and B are both growth-promoting substances
whose absence from the diet leads to serious disturbances and finally
to death, and whose presence even in minute amounts promotes the
normal metabolic processes which result in growth.

The possibility of growth-promoting substances being concerned
with the physiological effects of groups of animals upon the individ-
uals composing the group is further indicated by the work on "bios."
The literature on this subject is voluminous and confused. Tanner
(1925) presents an exhaustive review and bibliography of the re-
searches from i860 to 1924. "Bios" is the name provisionally given

158 ANIMAL AGGREGATIONS

by Wildiers (1901) to a mysterious organic substance which he be-
Heved to be necessary for the prohferation of yeast cells. After ap-
proximately a quarter of a century of work upon the subject, Tanner
summarizes the situation regarding bios as follows :

"One group of investigators denies the existence or need on the
part of the yeast plant, of a substance like 'bios.' They feel that
yeasts will grow without this accessory substance.

"Another group believes that 'bios' is necessary for the growth of
yeasts. They are unable to secure growth of yeasts in pure solutions
without it. Certain of these investigators have reported fractiona-
tion of 'bios' into components which are necessary to one another.

"A third group of investigators believe that yeast will grow in
pure nutrient solutions without 'bios' but that the addition of a
'bios' containing substance may cause increased growth. Whether
this acceleration in growth following the addition of a 'bios' con-
taining substance is due to 'bios' or to some other factor in the pre-
parate has not been satisfactorily estabhshed. In this connection it
is well to point out that even a medium such as beer- wort which is
rich in 'bios' may be improved by the addition of other 'bios' con-
taining substances.

"A fourth group may also be recognized including those who have
isolated 'bios' or substances having 'bios' properties."

Throughout his review Tanner's attitude is satisfactorily critical;
and from a study of it, supplemented with certain of the original
research reports, it seems to me that the evidence favoring the view
that there is a growth-promoting substance which markedly stimu-
lates yeast growth is too strong to be disregarded at the present
time. Concerning whether the yeast cells are able to synthesize this
substance from nutrient solutions lacking it, as has been claimed, the
evidence is not yet so clear.

The same problem in a somewhat different guise is met with in the
studies concerning whether inorganic substances taken alone are ade-
quate for the growth of green plants. This question was most re-
cently raised by Bottomley (191 5 and subsequent papers) and
Mockeridge (1920, 1927), who showed that certain complex organic
substances, when partially broken down by bacterial action, stimu-

STIMULATION OF GROWTH BY CROWDING 159

lated the growth of Lemna and other water plants to a marked de-
gree. Bottomley gave the name "auximones" to these substances
which were effective in promoting growth for green plants. It soon
became apparent that the green plant Lemna can grow and multiply
for indefinite periods in a purely inorganic medium (Clark, 1924,
1926; Ashby, 1929), and Wolfe (1926) was led to the point of view
that Bottomley's theory of the need of growth-promoting substances
by green plants was completely refuted. Ashby made a more com-
plete analysis of the problem (1929a) and has demonstrated that
small amounts of organic substance obtained from fresh horse dung
will increase the rate of growth of Lemna if present in only 0.2 parts
per million and that the growth-rate is little affected by additions of
this material beyond 2.0 parts per million.

The duckweed which Ashby used in these experiments had been
growing for 6 months on a purely inorganic medium made up in
glass-distilled water. Environmental conditions such as light, tem-
perature, and pH were adequately controlled. The mean frond
weight remained the same in control and experimental solutions;
the mean frond number increased 42 per cent; the area of the fronds,
33 per cent; and there were roughly 80 per cent more chloroplasts in
the cells of fronds treated with 0.002-0.02 grams per liter of dry ex-
tract, as compared with untreated fronds.

The addition of 0.2 parts per million of organic matter to a solu-
tion containing already 1,210 parts per million of mineral matter
will significantly increase the growth of Lemna. The power of the
extract is not affected by autoclaving; hence the effect is not due to
an enzyme. The ash constituent of the extract does not increase the
growth-rate, and increasing the nitrogen content by adding 0.003
grams per Hter of KNO3 did not affect the growth, while adding one-
hundredth of this amount of nitrogen as organic matter did signifi-
cantly increase the rate of growth. It seems clearly estabhshed by
this work that,, while Bottomley's "auximones" are not essential for
plant growth, the addition of extremely minute amounts of organic
material produces this effect by acting as a catalyzer.

The work with vitamins, tissue extracts, bios, and auximones in-
dicates clearly that the presence of very small amounts of organic

i6o ANIMAL AGGREGATIONS

material may strongly affect the growth processes of organisms pres-
ent. Relatively slight conditioning by the products of plant or ani-
mal metabolism produces marked results, which are not necessarily
increased by increasing the amount of material present. One reason
for the failure to recognize the presence and effectiveness of such
compounds hes in the exceedingly minute minimal quantities neces-
sary to produce maximal effects. Unless especial care is taken, traces
of such materials will be present as contamination and will produce
as great an effect as if more were added.

CHAPTER X

STIMULATING EFFECTS OF CROWDING ON
THE RATE OF REPRODUCTION

In a preceding chapter we have seen that there is much support for
the conclusion that crowding decreases the rate of reproduction
among animals generally, with specific instances among the
Protozoa; the Crustacea, of which Daphnia is an example; in the
insect Drosophila; and among birds. We have also seen that there is
at times an optimum crowding for the growth-rate, which does not
necessarily coincide with the minimum population density. It is now
necessary to examine whether the evidence that has been advanced
demonstrates a similar optimum, at least for certain animals at some
time in their Ufe-cycle, in so far as their rate of reproduction is con-
cerned.

The phenomenon which we are to discuss may deservedly be
called "Robertson's phenomenon," since Robertson was most active
in collecting evidence of its existence. He himself gave it the name of
"allelocatalysis," which he defined (1924a) as meaning "the acceler-
ation of multiplication by the contiguity of a second organism in a
restricted volume of nutrient medium."

His announcement of the existence of this phenomenon naturally
awakened an immediate interest among biologists generally and
among students of the Protozoa in particular, many of whom have
been unable to confirm its existence. Hence it becomes necessary to
examine the development of the problem in an effort to evaluate the
results of work centered on it.

Robertson (1921a, 1923, 1924a, 19246) found that when two in-
fusorians, Enchelys and Colpidium (later identified as Colpoda), are
introduced into the same restricted amount of fresh culture medium,
the early rate of reproduction following a period of readjustment,
called the "lag period," is not merely double that shown when a
single infusorian of the same species is similarly treated, but is some

161

i62 ANIMAL AGGREGATIONS

multiple in excess of this. He reports that he has obtained a rate of
reproduction from two and a half to ten times that which might
otherwise be expected. In his later work Robertson found this effect
to be more marked when the transplants are freed from contamina-
tion with the parent culture medium by repeated washings. The in-
creased rate of reproduction, which Robertson calls the "allelocata-
lytic effect," does not depend upon conjugation, because this does
not occur within the conditions of the experiment. Robertson at-
tributes the stimulation to the diffusion of some agent from the
organisms into the culture medium, which accelerates their repro-
ductive rate. When more than one organism is initially present, the
concentration of this substance within the organism is higher; and
the rate of multiplication is increased as a direct consequence.

Table VIII gives Robertson's allelocatalytic data from his 1921a
paper, showing the effect of placing 2 Enchelys farcimen together in
a single drop (0.08 cc.) of culture medium as compared with isola-
tions of single individuals into the same amount. The figures given
include all cases recorded in this paper, except those which Robert-
son says were run under conditions unfavorable for allelocatalysis
but which were run and described to test out the conditions under
which allelocatalysis might occur. Omitted cases include those in
which the parent culture was over 3 days old ; those in which one of
the two was purposely killed; and those introduced into bacteria-
free media.

Jahn (1929) calls attention to the fact that the averages of the
generation time, during the first observation after isolation, in four
much cited experiments of Robertson's, Nos. 238A, 240A, 23 7 A, and
242A, show a variation of +31 per cent, about a median of 9.1 hours;
and that one pair of experiments, Nos. 310A and 311 A, with i iso-
lated individual and one culture of 2 individuals, shows a similar
variation of + 20 per cent. He reasons that since Robertson has
listed these experiments in several publications they must be typical,
but that the acceleration in experiment No. 31 lA is less than the
variation in i -animal cultures and that therefore all Robertson's
results must be questioned.

However, enough data has been cited in Table VIII to show that

STIMULATING EFFECTS OF CROWDING

163

TABLE VIII

Showing Robertson's Original Allelocatalytic Data

(The Data in the Last Column Are Based on Those of the Preceding Column,

Restated To Show Directly the Effect of the Presence of a Second

Organism in the Same Limited Amount of Medium)

Culture No.

No. of Animals
Introduced

No. after 24
Hours

Ratio

Ratio per Single
Original Animal

Normal Hay Infusion; Parent Culture 24 Hours Old

Normal Hay Infusion; Parent Culture 48 Hours Old

164

ANIMAL AGGREGATIONS

TABLE Will— Continued

Culture No.

No. of Animals
Introduced

No. after 24
Hours

Ratio

Ratio per Single
Original Animal

Normal Hay Infusion; Parent Culture 48 Hours Old — Continued

Normal Hay Infusion; Parent Culture 72 Hours Old

Normal Hay Infusion; Parent Culture Age not Given

224A.

22SA.

24/

1:3

1:1.5

they cannot be so easily dismissed. As will appear shortly, I hold no
brief for the allelocatalysis theory as stated by Robertson; but it is
only fair to note, as Robertson himself states (1927), that in order to
be closely comparable, experiments such as these must have the
same history and be run at the same time, a fact well known to
experimental workers in this and other related fields of physiological
zoology. It is therefore hardly fair criticism to compare a single
allelocatalysis experiment run at one time with another run later.
Fortunately, we do have a statistical method of analysis which takes
into consideration the biological sensitiveness of paired experiments,
that is, "Student's" method (Student, 1925; Fisher, 1925). Applying

STIMULATING EFFECTS OF CROWDING 165

this method to the results listed in Table VIII, we find that there
are 128 chances in 10,000 of random sampling giving as large dif-
ferences in the ratios per individual animals as those given in the
first eight comparisons listed, which is well within the range of
statistical significance. It will be noted that these include all the
experiments cited by Robertson (1921a) that were run in normal hay
infusion with the parent culture 24 hours old. If one extends the
inquiry with Student's method to include all the cases listed in
Table VIII, one finds that the results are almost statistically per-
fect; that under no conditions would random sampling be expected
to give the results shown in that table.

But the fact that the results are statistically significant does not
necessarily mean that they support a given hypothesis. It merely
means that in the case of Robertson's experiments here cited,
Enchelys divided significantly faster when 2 individuals were placed
together in a drop of culture medium, as compared with i individual
of similar history similarly treated.

Robertson thought at first (1922) that old Enchelys culture fluid
contained no substances that were toxic to the Infusoria isolated
into it from young cultures; but as evidence accumulated, he
changed his opinion to conclude (1924) that "the ultimate cessation
of reproduction in old cultures is attributable also to the accumula-
tion of a product of growth, and possibly of the same product that
was originally responsible for the acceleration." The presence of an
accelerator Robertson finds indicated by: {a) the accelerative effect
due to the addition to the culture medium of a small proportion of
culture fluid which was previously inhabited by the infusorian En-
chelys; (b) the increase of reproductive rate when the volume of
fluid is reduced so that less of the accelerator passes into the fluid and
more remains within the organism ; and (c) the fall of the reproduc-
tive rate when one of two recently divided individuals is isolated into
fresh culture medium, in comparison with the reproductive rate of
the other left in the original medium.

Originally Robertson did not believe this phenomenon to be due
to a heavier growth of bacteria, although the results are said to de-
pend on the presence of an optimum amount of bacteria in the cul-

i66 ANIMAL AGGREGATIONS

ture medium. Later, after the evidence furnished by Cutler and
Crump showed that the growth of Colpidium is much more facih-
tated by the "contaminating bacteria" which they found ordinarily
associated with this infusorian than it is by Sarcina, which they
employed as a food organism, Robertson (1927) considered the pos-
sibility that Colpidium and other ciliates are restricted in their food
to certain species of bacteria for which they modify the nutrients
available, thus existing in a sort of symbiosis with their food species.
However, he concludes that whether the allelocatalytic effect origi-
nates with the ciliates or with associated food organisms, it remains a
mutually accelerative effect of contiguous organisms upon their
reproductive rates.

Robertson shows that his results are not due to the carrying-over
of twice the amount of parent culture medium when 2 organisms, in
place of I, are transplanted; for in his experiments infusorians
washed with medium similar to that into which they are subse-
quently subcultured show the effect even better than unwashed ani-
mals. Neither is the effect due to mere numbers, for recently divided
or dividing individuals, if transferred together, act as single animals.

To explain the observed results, Robertson advances the following
hypothesis (1923). During nuclear division each nucleus retains the
charge of autocatalyst with which it was provided, and adds to it
during the course of nuclear synthesis. At each division the auto-
catalyst is shared between the nuclear substance and the surround-
ing medium in a proportion determined by its relative solubility and
by its affinity for chemical substances within the nucleus. The mu-
tually accelerative or allelocatalytic effect of contiguous cells is due
to each cell's losing less of the autocatalyst to the medium because of
the presence of the other. According to this hypothesis, the auto-
catalytic effect of growing colonies, whether attached or detached,
is due to the same cause.

Fischer's work (1923), in which he found that fibroblasts grow in
vitro only when tissue cells are numerous and close together, can be in-
terpreted as giving supporting evidence to the allelocatalytic hypoth-
esis. Burrows (1924), growing cancer cells in vitro, found a similar
stimulation; this and other similar evidence from tissue culture has

STIMULATING EFFECTS OF CROWDING 167

been reviewed at some length in the preceding chapter. Similar evi-
dence from the field of general bacteriology will be presented in
chapter xv under the heading proposed by Churchman (1920),
"Communal Activity of Bacteria."

On the other hand, Cutler and Crump (1923), in cultures of Col-
pidium, failed to find the allelocatalytic effect when the volume of
the medium was reduced; and Greenleaf (1924), using Paramecia
and PleiirotricJia, in a brief note records his failure to confirm Robert-
son's work. Peskett (1924, 1924a) observed such stimulation to divi-
sion in but 3 cultures of yeast out of 128 examined.

Robertson (1924a, 1924^) re-examined the problem in the light of
results published up to that date, and explained the lack of success
of other workers as being principally due to their failure to wash the
organisms before transfer. His reasoning here (1927) is as follows:
Presumably, if contiguous animals can affect each other's growth
without the occurrence of conjugation, they must do so through the
agency of some soluble substance which they emit and which is
transferred from one organism to another through the medium which
they inhabit. Presumably also, this soluble substance must be very
abundant in the thickly inhabited cultures from which subcultures
are usually prepared. This leads, of necessity, if Robertson's reason-
ing is sound, to the removal of the parent culture medium from the
animals by washing before transfer if allelocatalysis is to result.
Robertson reports (1924a) that allelocatalysis increases with pro-
gressive removal of preformed catalyst until a maximal effect is
reached just before its total removal. In a preceding paper (1924) he
presented evidence that washing not only removed adherent auto-
catalyst but also washed out the accumulation of this hypothetical
substance from the living cells themselves.

Robertson also emphasizes (19246) an earlier statement that in
comparing the reproductive rate care must be taken to estimate the
population some time before it has attained maximal density. The
end result of introducing a second individual is to reduce the rate of
reproduction, since the final maximum is the same in all cases and is
independent of the size of the seeding.

Peskett (1925, 1925a) returns to the problem of possible alleloca-

1 68 ANIMAL AGGREGATIONS

talysis in yeast. Apparently he used very careful technique, washing
his transplants and plotting a growth curve based on a number of
examinations at different phases of the culture. He finds that in 46
cultures with synthetic media whose volumes ranged from 2 to
146 cu. mm. he cannot attribute any differences observed to changes
in volumes, as would be required if Robertson's hypothesis of allelo-
catalysis holds for yeast. With 57 cultures, containing bios and vary-
ing from 0.5 to 362 cu. mm., he observed a tendency for more rapid
growth in the cultures with the larger volumes. In only 2 cases out
of 8 near the minimal volume did he find evidence of acceleration in
the cultures. The statistics on mortality of the cultures show that
those started with 2 cells usually have a higher viability but that
the percentage of difference is not great. He concludes that allelo-
catalysis does not occur in the yeast studied.

Cutler and Crump (1925) also repeated their experiments with
Colpidium, thoroughly washing their animals before transfer. They
again failed to obtain evidence of the stimulation demanded by
Robertson's hypothesis. It may be that their technique differs suffi-
ciently from that of Robertson to account for some difference in
results, since the infusorians which Robertson washed were fre-
quently injured in the process while Cutler and Crump found no
deleterious effect from their washings.

Calkins (1926) gives the results of 60-day experiments with
Uroleptus, in which i, 2, 3, and 4 animals were introduced into i
drop of fresh medium, respectively. The division rate was found to
be inversely proportional to the initial number of individuals inocu-
lated.

Returning to this problem in 1927, Robertson states that all of
the data and conclusions concerning allelocatalysis in Infusoria
which have been issued from his laboratory in recent years remain
valid, save that they may apply to the associated food organism and
not to the Infusoria themselves.

Robertson, in further considering the possible reasons for the
discrepancy between his own work and that of his critics, stresses the
need of being sure that the animals isolated to test for allelocatalysis
should have the same division rate and that this effect is shown

STIMULATING EFFECTS OF CROWDING 169

only when more than one Hving cell is present and that the presence
of a second cell which has died does not affect it. He particularly
criticizes the work of Cutler and Crump and of Peskett because of
the high death-rate in their subcultures, not only among the inocu-
lum but also, as Cutler and Crump have shown, among the animals
produced in the subculture itself. He also criticizes the artificial me-
dium used by the latter investigators, and presents evidence to show
that their synthetic medium is not well buffered above pH 7.8, which
is the optimum region for the growth of Enchelys. Data are also
presented showing the great difference in viability of different In-
fusoria from a crowded culture, and he suggests that a part of the
discrepancy between his own work and that of Cutler and Crump is
due to differences in the protozoan-bacteria ratio on account of the
different techniques employed.

Greenleaf (1926) reported in full his experiments along this line,
using Paramecium aurelia, P. caudatum, and Pleurotricha lanceolata,
inoculated in from 2 to 40 drops and Stylonychia pustiilata in 2 and
5 drops of culture medium, to determine the influence of volume of
culture medium on the rate of cell division, and the influence of cell
proximity on this rate, and to investigate the factors concerned with
the lag period. In the experiments covering the first point, Greenleaf
reports the results by means of giving the average per diem rate car-
ried through a period of 5 days. Under these conditions, in 4 experi-
ments only, out of 27, were 2 drops found to be more favorable for
reproduction than 5 drops. Similarly, in only 4 experiments out of
27, did the 5-drop cultures give a higher rate of reproduction than
the 20-drop cultures. In the comparisons between the 20- and the
40-drop cultures, the former gave a higher rate in 8 experiments; the
latter in 9 cases; and i was a tie.

Again, in his investigation of the effect of 2 animals present in a
given volume, 2 or 5 drops, he used the average per diem division
rate based on 5-day periods and found that the single animals divid-
ed the more frequently in both cases, although he did find that
"there was less difference in the division rates of the animal carried
alone and the two carried together in the 5-drop series than in the
2-drop series. This indicates that in the larger volume the depressing

170 ANIMAL AGGREGATIONS

effect of the presence of 2 animals is less marked than in the smaller
volume." He did not attempt to carry this further to find the rela-
tive effect of a further increase in volume.

Greenleaf's work is chiefly to be criticized because of his taking
average division rates for a period of 5 days instead of following the
actual division rate per day — the more so since Robertson had pre-
viously called attention to the necessity of such care and Peskett
(1925) had used particular care to examine his cultures at fairly
frequent intervals in his later studies upon the rate of growth in
yeast cells.

In keeping with his other results, Greenleaf found that isolating
infusorians into 5 drops of culture medium causes a higher rate of
reproduction if the cells are re-isolated after 24 hours than if they are
allowed to stand for 48 hours in the same medium. ,

Myers (1927) worked with washed Paramecium caudatum and
found that ''increasing the numbers of individuals present in a given
volume does not increase the rate of reproduction either from the
beginning of the cultures or after the first fission." In his work he
used 2-16 drops of culture medium, measuring from o.i to 0.8 cc.
His observations were usually made three times per 24 hours, at
intervals of 6, 6, and 12 hours.

Petersen (1929) justly comments on a part of this work of Myers,
showing that his system of taking observations at 6-, 6-, and 12-hour
intervals may mask significant results since an indicated interpreta-
tion of his data shows that he does not regard changes of from 6.0
to 8.4 hours before the first division time as being significant, al-
though the higher number represents an increase of 40 per cent. Of
course, the technique used may require greater differences before
significance is reached.^

' In the light of this criticism supported by tabular evidence, it is hard to understand
Jahn's (1929) diii&culty which causes him to assert that Petersen has selected certain
data from two sets of Myers' experiments. Actually she used all the data published by
Myers concerning i6-drop cultures which bore on this point, and lists all three sets of
these experiments both in her summarizing table of Myers' work and in the accompany-
ing discussion. From other remarks concerning Petersen's work, Jahn seems to have
overlooked the fact that she was interested in contrasting the effects obtained by
testing for aJlelocatalysis in small and in relatively large volumes of medium, from
which came her particular attention to the results reported by Myers from his i6-drop

STIMULATING EFFECTS OF CROWDING 171

Yocum (1928) reported results resembling allele catalysis from his
work with Oxytricha. He isolated unwashed individuals to 4 and to
10 drops of sterile medium when 100 drops equal a cubic centimeter.
After 24 hours he found that when all cases were averaged, the divi-
sion rate in the smaller volume exceeded that in the larger by 14 per
cent, and that the average in the smaller volume in each of his series
was greater than in the larger volume by over 10 per cent. He inter-
preted this as meaning that the smaller volumes reach a high con-
centration of an autocatalyst earlier than do the larger volumes.

The work of Petersen (1929) chronologically belongs at this point;
but it will be examined in more detail shortly, after considering the
work of Jahn (1929), who investigated the relation between density
of population to the growth rate in Eiiglena sp. Euglena was chosen
for its ability to grow in autotropic media. Single isolations were
abandoned on account of excessive evaporation. Of 22 cultures
started with washed sister-cells isolated into 3 or into 6 drops of
medium, 5 cultures were counted at the end of 10 days. Three of
these showed larger numbers in the smaller volume, one showed the
reverse, and the other had about equal numbers. No data are given
for the intervening period.

Thereafter mass cultures were run in which the ratio of the initial
volume of individuals to medium ranged from i : 383,300 to i : 3,833, -
000. Results are recorded as obtained by a series of dilutions of from
0.005 to 0.55, 0.066 to 0.66, 0.089 to 0.0894, 0.081 to 0.73, 0.14 to
0.555, ^i^d 0.42 to 9.5 thousands per cubic centimeter of culture
fluid. Initial counts were made in from i to 4 days in the different
series. In Series 2, 3, 5, and 6' the initial count in the more concen-
trated medium was made approximately 1-3 days before that in the
least crowded series; hence it is hard to judge initial conditions.

Only in Series i do the recorded data show the initial count made
at the same period in the different experimental lines; and here the

cultures. It may be, as Petersen suggests, that the method of recording data chosen by
Myers does not allow his results to be used as evidence for allelocatalysis. If that be
true, there are similar reasons for thinking that, in so far as the i6-drop cultures are
concerned, they cannot be used as evidence against this interpretation.
' The time of making the initial count in Series 4 is not given.

172 ANIMAL AGGREGATIONS

results at the end of the first day's exposure clearly bear out the
author's conclusions that his mass experiments with Euglena offer
no evidence of any allelocatalytic effect. As would be expected, from
all of the work to date, Robertson's included, after the first few days
Jahn found a significant difference in rate of increase, with a more
rapid rate in the less dense populations, which he is inclined to
ascribe to the combined effect of relatively more food and relatively
less wastes. The effect of mass of Euglena, which is capable of
growing in autotropic media, may well differ from that of holozoic
organisms, particularly since their CO2 relations diverge so de-
cidedly.

The work which throws most light on this tangled problem is that
of Petersen (1929). Petersen used Paramecium caudatum in her ex-
periments. They were grown in a boiled-hay medium which was
bacterized 24 hours before using. Animals were isolated by means of
a drawn-out pipette. Any attempted washing was done by re-isola-
tion. Throughout her experiments 24 drops equaled i cc.

The usual method of comparing individuals of the same clone was
not followed in this work because it was found early in the investiga-
tion that there may be a marked difference in the division rate of
animals from the same clone. Even the division rate of sister-cells
showed a difference. Her method, as finally worked out, was to iso-
late 2 animals with approximately the same rate and time of divi-
sion. These were allowed to divide until 4, or 8, or 16 cells were pro-
duced, according to the needs of the experiments, at which time all
the descendants of one animal were subjected to one set of condi-
tions while all from the other were placed under the conditions with
which comparisons were to be made. Because of the re-isolation so
early in the life of the subculture, the lag phenomenon was avoided.

Under these conditions, four sets of experiments were run, as
follows: (i) cell proximity in small volumes of medium; (2) variation
in volume of medium; (3) cell proximity in large volumes of medium;
and (4) conditioning of medium, both large and small volumes.
Petersen's results will be given in some detail because of their im-
portance.

Preliminary experiments in 2 drops of medium gave some slight

STIMULATING EFFECTS OF CROWDING 173

indication of a stimulating effect upon the rate of reproduction when
2 organisms, instead of i , were present ; and two series of experiments
were run to investigate this tendency. In the first series the animals
were washed once; in the second they were isolated unwashed. Ex-
aminations at 12, 18, 24, 36, and 48 hours failed to reveal any indica-
tion of acceleration due to the presence of 2, as compared with i
animal, in the original subcultures, either with the washed or with
the unwashed cultures. Again, contrary to the report of Robertson,
there was no speeding up of the division rate after the first fission.
Further experiments, made with small volumes after Petersen had
found that a dift'erent relation exists when animals are subcultured
by one's, by two's, and by four's in larger volumes, still gave the
same negative effects. With these small volumes the work of Peter-
sen is in agreement with that of Myers and of Cutler and Crump
rather than with that of Robertson.

Because of the evidence of retardation in the early experiments,
Petersen thought that the volumes used might be entirely too small.
Acting on this lead and using 20 drops of culture medium (24 drops
equal i cc), Petersen transferred 1,2, and 4 Paramecia, which had
been washed three times, into this volume of bacterized medium,
with the results shown in Table IX. In these experiments there is
evidently an acceleration in the early division rate of the subcultures,
associated with an increase in the number of animals introduced.
Further experiments shown in Tables X and XI demonstrate that
this is not an accidental result.

The question as to whether the accelerative effect is species-
specific was apparently answered for this particular organism by the
fact that in 5 sets of experiments similar to those reported above
which were found to be noticeably infested with a small ciliate after
the first 24 hours of isolation, the rate of division was nearly or quite
independent of the initial seeding. Similar results were obtained with
the only other set of contaminated cultures entering into this series
of experiments.

So far, these experiments of Petersen's have shown that with
smaU volumes, /^ cc. or less, there is no acceleration with the intro-
duction of a single individual in the initial subculture, but that with

174

ANIMAL AGGREGATIONS

TABLE IX

Showing Division Rate of Sets of 4 Paeamecia Transferred Singly, in Two's,

AND IN Four's, to 20 Drops of Bacterized Medium. Washed

Twice in Bacterized and Once in Alkalinized Medium

(Data from Petersen)

One

Two

Four

Hours

24

48

24

48

24

48

4

5

9

4

5

16

4

8

18

4

4

9

4

5

10

4

5

12

4

5

9

4

6

13

4

6

16

4

4

II

4

5

20

4

7

16

4

4

9

4

6

10

4

6

14

4

8

16

4

8

17

4

8

24

4

6

12

4

8

24

4

8

24

4

7

14

4

8

16

4

8

16

4

7

16

4

14

24

4

14

24

4

9

10

4

8

12

4

10

14

Total

40

59

115

40

73

162

40

80

178

Mean division rate

1-47

2.87

1.85

4 05

2.0

4-45

TABLE X

Showing the Effect of Double Washing and Transfer of Paramecia

to 20 Drops of Alkalinized Bacterized Medium

(From Petersen)

Transfer

One

Two

Four

24

48

24

48

24

48

4

6

21

4

8

30

4

8

38

4

7

12

4

8

16

4

10

3b

4

6

25

4

7

34

4

8

37

4

8

23

4

8

30

4

8

52

4

6

22

4

8

34

4

8

40

4

5

28

4

5

34

4

7

42

4

7

25

4

8

47

4

9

5b

4

6

9

4

8

22

4

8

33

4

7

18

4

8

32

4

8

38

4

6

16

4

7

18

4

8

20

4

6

14

4

8

16

4

8

23

4

5

22

4

8

17

4

8

24

4

5

14

4

6

18

4

8

20

4

6

28

4

7

34

4

7

40

Total

56

86

277

56

104

382

56

"3

499

Mean division rate

1-53

4-94

1.85

6.81

2.01

8.80

STIMULATING EFFECTS OF CROWDING

175

large volumes, f| cc, such acceleration does occur; with still larger
volumes we begin to see that the effect is in reality a relationship be-
tween the mass of animals and the mass of culture media. Inocula-
tions similar to those made above, but made into 40 drops of me-
dium, showed in 48 hours some evidence of acceleration of division
rate with an initial seeding of 4, but with this volume 2 individuals
reacted as did i. Whatever the process of conditioning the medium,

TABLE XI

Showing the Effect of Double Washing and Transfer of

Paramecia to 20 Drops of Unalkalized Bacterized Medium

(From Petersen)

Transfer

One

Two

Hours

24

24

24

4
4
4
4
4
4
4
4
4
4
4
4

14

16

IS
20
18
14
19
21
16
18

19
21

4
4
4
4
4
4
4
4
4
4
4
4

16
16
16

28
16
24
22
26
27

23
24
26

4
4
4
4
4
4
4
4
4
4
4
4

16
16
16

30

28

30
26
26
30
31
26
30

Total

48

211

48

264

48

305

Mean division rate . .

4-39

5-5

6.35

2 were no more effective in 48 hours with 1.67 cc. of medium than
was a single individual, while 4 were significantly more effective than
either. These results are exhibited in detail as Table XII.

When the numbers produced in 20-drop cultures initially contain-
ing I and 4 individuals are examined by Student's method for de-
termining the statistical value of paired experiments, one finds, if all
of Petersen's comparable data are considered, that there are 2
chances in 10,000 of obtaining so great a deviation from the mean by
random sampling. When the similar data for the 40-drop cultures
are similarly considered, the probability, even from this short series

176

ANIMAL AGGREGATIONS

of experiments, is shown to be slightly less than 5 chances in a 100,
which is usually considered to suggest statistical significance.

Petersen approached this problem from another angle. Experi-
ments were carried on to determine whether or not the division of an
animal in a Hmited amount of medium would accelerate the rate of
division of individuals isolated into the same volume. Both washed
and unwashed Paramecia were so tested. First, washed animals were

TABLE XII

Paramecia Washed Three Times and Transferred Singly, in Two's, and in

Four's, to 40 Drops of Alkalinized, Bacterized Medium

(From Petersen)

Transfer

One

Two

Four

24

48

24

48

24

48

4

8

16

4

8

16

4

8

16

4

8

16

4

8

16

4

8

16

4

8

16

4

8

16

4

9

19

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

8

4

8

16

4

8

16

4

12

18

4

8

16

4

8

16

4

10

16

4

8

16

4

9

16

4

10

16

4

8

32

4

8

32

4

16

40

4

8

22

4

8

32

4

8

24

Total

44

76

162

44

77

172

44

93

181

Mean division rate

1-75

3 69

1-75

369

2. II

4. II

isolated into 2 drops of bacterized medium. At the end of 12 hours
the animals were removed, and the slides in which fission had oc-
curred were noted. Five sets, consisting of 8 animals each, each set
being the fission products of i animal, were isolated singly into the
drops in which fission had recently occurred ; and comparable isola-
tions were made into those drops where animals had lived for the
same length of time but without division. There was no significant
difference in division rate at the beginning or at the end of the ex-
periments, on washed or unwashed individuals.

Petersen says: "This is not what would be expected on the basis
of Robertson's work. His reports indicate that acceleration of divi-

STIMULATING EFFECTS OF CROWDING 177

sion rate should appear after the first division in a limited volume,
due to the liberation of an autocatalyst from the nucleus to the
pericellular medium at the time of division.

"Another attempt to accelerate division rate by the use of condi-
tioned medium was made with cultures of large volumes. Cultures of
20 drops in which an animal had lived and divided were inoculated
with isolated unwashed animals and their division compared with
that of other comparable animals also unwashed, which were isolated
to 20 drops each of fresh bacterized medium.

"Animals isolated into the 20 drops in which one individual had
divided and lived (d) and animals isolated into 20 drops of fresh
bacterized medium (n.d.) at the end of 18 hours gave the following
numbers respectively.

d 4,2,4,4,2,2,3,3,2,3,2,2,2,2,2,2—41

n.d I, I, I, 2, 1, I, I, I, I, I, I, I, 2, 1, 1, 2 — 19

"Thus indications are that an animal living and dividing in 20 drops
of medium conditioned it in such a way that an acceleration in the
division rate of other individuals isolated into it is marked.

"Apparently in the small volumes, one individual is able to condi-
tion the medium sufficiently to cause rapid multiplication ; and there-
fore when one animal is transferred to a small volume of conditioned
medium, its rate of division is not faster than that of a similar in-
dividual transferred to unconditioned medium. But in large vol-
umes, one individual is not able to condition the medium immediate-
ly; and when, therefore, one animal is introduced into a large volume
of conditioned medium, division takes place faster than in the case
of a similar individual introduced into unconditioned medium.

"Greenleaf and Myers in work with conditioned medium used
only small volumes, 2-5 drops. Both report a depressed division
rate in the conditioned medium, which they interpret as due to the
increased toxicity of the old conditioned medium."

The importance of Petersen's work lies in the fact that for the
first time in the development of this phase of the problem of the
effect of initial numbers on the rate of reproduction in relatively
small subcultures, she has been able to obtain both positive and

178 ANIMAL AGGREGATIONS

negative results at will, depending on her manipulation of volume
relations. Her work and discussion make it impossible for one to
lump together all volumes in testing for the possible stimulating
effect of the presence of more than i cell in a limited amount of
medium.

Here we shall have to let the discussion of Robertson's phenome-
non rest for the time being, pending the accumulation of further
data. So far, the evidence demonstrates that reproductive rate does
not always depend on the number of the original transplants; that,
in fact, under many conditions and with many organisms it has not
yet been shown to have any positive correlation. All are agreed that
in the end the rate of reproduction falls sooner in small subcultures
seeded with more than one organism and more rapidly than it does
when a single individual is isolated. However, the repeated experi-
ence of Robertson, supported now by Yocum's work in so far as it
covers the same ground, and by the work of Petersen with larger
volumes of culture medium, demonstrates that true acceleration of
the rate of reproduction may occur associated with the introduction
of more than one organism into a limited amount of medium.

When the very establishment of the phenomenon is a matter of
such great difficulty, it is to be expected that the explanation of the
phenomenon, when it does occur, is still uncertain. At present we
cannot accept Robertson's hypothesis of the action of an autocata-
lyst; and while there is evidence for the production of some condi-
tioning agent, some A''-substance, which renders the culture more
favorable to growth, more light is needed, particularly concerning
the role of bacteria in the phenomenon among other possible factors,
before much progress can be made toward a solution of this aspect
of the problem. Perhaps when we have examined the relations of
masses of larger animals to survival under adverse conditions, a sub-
ject to be taken up soon, we can better understand the complexities
of the present problem.

An effect strikingly similar to Robertson's phenomenon has been
described by Chapman (1928), though not with that in mind. In his
work upon the effect of a limited environment, in this case whole
wheat flour, upon the number of confused flour beetles, Triholium

STIMULATING EFFECTS OF CROWDING

179

confusum, that will develop, Chapman found, as was stated earlier
(chap, vii), that with these animals and with this medium it is pos-
sible to demonstrate that the total population per gram of flour
arrives at a constant level of 43.97 ± 2.88. Chapman's table record-
ing the population level attained at successive intervals when 2, 4,
8, 16, 32, and 64 beetles are placed in 32 grams of flour for each lot is
repeated here as Table XIII.

TABLE XIII

Showing the Number of Beetles per Gram of Flour with Initial Population

Varying from 2 to 64 per 32 Grams. The Results Are Given

IN Terms of Beetles per Gram of Flour

(From Chapman)

Days

Grams

32

32

32

32

32

32

0.062
115

2-5

II

28
30
40

45

42

0.125

5-5
12

27
31
44
43
43
47

0.25
9
19

32
30
42
38
39
42

05
16
20
30
29
43
42
50
44

I
27
30
43
46

53
52
50
50

2

II

42
38
55
47
53
44
46

45

2K

^4

69

81

104

123

139

Chapman is interested in showing by these figures, as by those
cited in chapter vii, that, regardless of the initial seeding, a point of
equilibrium is reached after which the population remains relatively
constant. His medium is distinctly advantageous in this respect, for
with the Protozoa which we have just been considering, after the
similarly constant point of equilibrium is reached, the population
declines on account of the changes induced in the culture medium.
In flour beetles, as we have said before, this decline can be controlled
by changing the culture medium.

Our own interest is in examining the effect of numbers in the ini-
tial population upon the early rate of reproduction. The relation-
ships existing for the 11- and 25-day periods shown above are given
in Figure 13, which has been drawn from the data just given. The
figure shows immediately that for these first periods in the develop-

i8o

ANIMAL AGGREGATIONS

ment of the population the greatest increase comes with an initial
population of 4, while the least rapid rate of reproduction comes with
the smallest number of individuals present.

This indicates, as did the data of Petersen, that there is an opti-
mum initial population-medium relationship, which, for certain

8-

1^

-

A^V

/

\,^^

-

/

\^--.

-

/

^^ ^' //days

//

/ 1

1

1 1 1 1

1 -

Z 4 8 16 32 64

Initial population per 52 g/ns. of f/our

Fig. 13. — Showing the relation of initial density of Tribolium population in 32 gm.
of flour to rate of reproduction. Recalculated from data reported by Chapman.

volumes of medium, is different from that given by the smallest
population present. The fact here seems as clear as could be expect-
ed from an experiment not designed to test this particular point.
More recently T. Park, working in this laboratory, has confirmed
Chapman's results (unpublished data). The explanation is even less
apparent than in the case of Robertson's phenomenon in the Pro-
tozoa.

CHAPTER XI

EFFECT OF CROWDING ON SURVIVAL
AND OXYGEN CONSUMPTION

Studies on the causes and effects of animal aggregations begun in
191 2 and carried on intermittently thereafter were summarized
(Allee, 1920) for the water isopod Asellus communis, three species of
land isopods, and the ophiurid starfish Ophiodcrma hrevispina. The
results showed that, in general, bunching in these species is most
prevalent under adverse conditions and when there is no means of
satisfying the normally positive thigmotactic reaction. Of the single
factors tested, the tendency to collect in bunches is most strongly
encouraged in Asellus by the breeding reaction, in the land isopods
by the amount of moisture present, and in Ophiodenna by the
amount of light.

All of these animals had shown in preliminary experiments a
lowered rate of metabolism immediately following the formation of
the aggregation, as measured by their oxygen consumption or car-
bon-dioxide production. When isolated and bunched animals stand
for long periods of time, the effect on the metabolic rate is reversed.
Both come to have lower rates than at the beginning of the experi-
ment, but the decrease is much greater with isolated than with
bunched animals. In the land isopods this decrease is accompanied
by a greater loss of water by the isolated individuals. The conclu-
sion was drawn that ''under laboratory conditions the formation of
aggregations serves to make these animals more quiet and in the
long run proves to be what is usually called an adaptive reaction."
This early statement remains a fair summary of the evidence since
collected in this laboratory.

In an eariier chapter attention was called to the fact that land
isopods aggregate into fairly compact masses in the absence of water
and separate when water is added. Experiments on two species,
Oniscus asellus and Cylisticus convexus, were run to test the effect of

i8i

i82 ANIMAL AGGREGATIONS

bunching on changes in water content as indicated by weight of the
animals (Allee, 1926). When the isopods were placed on a moistened
filter paper with maximum moisture present with which the animals
would remain bunched, the aggregated isopods increased in weight
3.8 per cent from water intake while their isolated fellows were in-
creasing 9.7 per cent.

Similar experiments with bunches and single individuals placed
under desiccating conditions showed that in a typical experiment
isolated isopods lost water three times as fast as did a group of 10
isopods gathered into a close bunch. Further, all 10 of the bunched
animals in this typical experiment were alive with a loss of weight of
less than 16 per cent after 7.45 hours, when the last of the isolated
individuals was found dead. The last 2 isopods to die showed a
water loss at this time of about 44 per cent. Sixty per cent of the
isolated isopods were found dead after 4.38 hours' exposure.

These observations show a definite survival value of a group of
animals even at the low level of social integration existing among
land isopods. It may be remarked again that, outside the breeding
season, as these were, isopods collect in bunches due primarily to
non-social, individual tropistic reactions, and that almost their only
social attribute is tolerance for other animals and their products
within a limited space.

STARFISH AUTOTOMY

The essentials of the situation among the ophiurid starfish are
similar, but the mode of expression differs (Allee, 1927). Ophiurids
have the practice of fragmenting their arms under certain condi-
tions. Early experiments showed that there is a greater tendency to
practice autotomy with isolated than with bunched Ophiodenna.
The tests to be considered were run in connection with respiration
studies to be described later. Two groups of 8 each were placed in
two large bottles each containing 8 liters of sea-water. Eight similar
starfishes were isolated into flasks of about i liter each. The water
was usually changed daily and never left longer than 48 hours. No
attempt was made to supply food.

The following schedule of numerical symbols was adopted : o, per-

EFFECT OF CROWDING ON SURVIVAL

183

feet arms; i, tip gone; 2, end gone; 3, one- third arm gone; 4, one-
half arm gone; 5, two-thirds arm gone; 6, mere stump left. Indi-
viduals were removed from the bottles when they showed frag-
mented arms yielding on the foregoing scale a total value for the five
arms of a single starfish of 15-20, regardless of the distribution of
effects between the different arms. Table XIV gives the mean time
in days before each lot was discarded. The data at hand are faulty

TABLE XIV

Showing the Time in Days in Long Respiration Experiments before

THE Different Lots Were Discontinued Because of Death

or the Fragmentation of Arms

Experiment No.

First
Bunch

Second
Bunch

Mean

Isolated

Difference

Glass*

Duration
in Days

I

2

3

4

c

19.6

13 -5
50
9.0
963

7-75

20.6

13 3
S-O
9 25
956
7-5

20. I

13-4

50

913
9.6

7.63

16.6
12.5
5-75
6.13
6,2
6.5

3-5
0.9

0.75
30
3-4
113

9-75

7-5

21

14

5

8

II

6.

8

Mean

Mean of last
two only . .

10.75

10.87

10.81

8.85

8.69

8.53

8,62

6.35

8.63

* The Ophioderma in this column were isolated into liter flasks containing glass rods bent into various
shapes.

in that, after respiration experiments were stopped, there were fre-
quently some individuals that had not yet reached the degree of
fragmentation necessary for removal. In some cases these were ob-
served, though under altered conditions, until autotomy had pro-
gressed past the arbitrary dead line; but in others this was not done,
and the last day of respiration tests was recorded for use in this
table. This procedure markedly favors the isolated individuals since
there were fewer of them so treated. The length of time the experi-
ment ran depended mainly on the temperature, since fragmentation
proceeds more rapidly at high temperatures.

In every instance except one, the mean survival time was greater
for the bunched animals than for those isolated under similar condi-
tions and at the same time. In four of the six comparisons the dif-

1 84 ANIMAL AGGREGATIONS

ference is large enough to be of significance. Student's method of
statistical evaluation shows that for all six experiments there are 49
chances in 1,000 of random sampling yielding so great a variation
from the mean in either direction.

The one exception where the isolated individuals showed a greater
survival time is instructive. For some reason one or two of the ani-
mals in each of the bunches of this experiment died soon after one of
the daily inspections, disintegrated and polluted the whole liquid,
and caused the death of the remainder. Such an extreme catastro-
phe could not happen with the isolated individuals.

The survival of the starfishes isolated into liter flasks containing
small heaps of variously bent glass rods is also instructive. Here the
starfishes came to rest on or among these glass rods just as in nature
they rest among eelgrass blades. It will be noted that for the two
tests made, the survival is comparable with that of grouped animals
rather than with the starfishes isolated into bare containers. It ap-
pears as though the satisfaction of the thigmotropic appetite has
approximately the same survival value whether the satisfaction
comes from contact with the piles of glass rods or the group of in-
dividuals of the same species.

RESPIRATION STUDIES WITH ISOPODS

The effect of aggregation upon the aggregants can be followed
more closely by studying the effect upon the rate of respiration.
Several such studies have been made with different animals, some of
which will be summarized.

Preliminary experiments upon the two species of land isopods
mentioned above, indicated that soon after these isopods aggregate
their rate of respiration is decreased, as compared with similar ani-
mals isolated for a similar time (Allee, 1926). This tendency appears
after the isopods have been bunched for 5 minutes, and extends at
least through the first hour of bunching. When the bunches and
solitary individuals are compared after standing in the laboratory
for a longer period of time, the isopods taken at random from the
bunches are giving off carbon dioxide the more rapidly.

Oxygen consumption of two other species of land isopods was

EFFECT OF CROWDING ON SURVIVAL 185

tested in manometer respirometers such as were described by Kraj-
nik (1922). The results obtained support the conclusions reached in
earher studies with carbon-dioxide production, that is, that with
land isopods the recently bunched individuals are carrying on respi-
ration at a less rapid rate than when recently isolated. In exact
ratios, the rate of oxygen consumption with recently bunched and
recently isolated Armadillidium in two tests were 1:1.486 and
1 : 1.368, while with Tracheonisciis the same ratios were i : 1.214 and
1:1.344.

The Armadillidium tests allow other comparisons. Since groups of
isolated or bunched individuals were set away in the dark for ap-
proximately 24 hours, one can compare their physiological condition
near the beginning of this period with that at the end and also make
cross-comparisons.

In general, the determinations show that, under the conditions of
the experiments, there is a very marked decrease in oxygen consump-
tion after 24 hours' isolation and starvation — 70 and 65 per cent,
respectively, in two sets of experiments. There is a similar but less
pronounced reduction when the animals are bunched and starved —
in that case 31 and 29 per cent, respectively.

At the end of 24 hours the bunched isopods are uniformly using
more oxygen per unit weight than are the isolated individuals. The
observed ratios were 1:1.64 a-^d 1:1.48. All the differences men-
tioned are statistically significant, since the least difference in means
is still over eleven and a half times their combined probable error.

Experiments extending over 50 hours showed that the bunched
isopods kept their higher rate of respiration during this period as
compared with similar isolated individuals. The aggregated individ-
uals exhibited a greater range in rate of oxygen consumption during
this time, probably due to the greater difference between repose and
occasional activity. They also were much less likely to move about
than were the isolated individuals. In all these experiments neither
set of isopods was fed, but the w^ater content was manipulated so
that there was no essential difference in weight, as shown by random
weighings.

Calibrations show that oxygen consumption for the Armadillidium

1 86 ANIMAL AGGREGATIONS

isolated for about i hour is at the rate of 219.6 cc. per hour and kilo-
gram; for those isolated for about 24 hours it is 77 cc. per hour and
kilogram; for those bunched approximately an hour, 160.5 ^^d for
those bunched for about 24 hours, 130.2 cc. per hour and kilogram.
Similar calculations for the Tracliconiscus show a mean oxygen con-
sumption of 263.5 cc. per hour and kilogram when isolated less than 2
hours, and of 196.7 cc. when bunched about the same time.^

STARFISH RESPIRATION

The marine ophiurid starfish, phylum Echinoderma, stands well
removed from the land isopod, phylum Arthropoda, in the evolution-
ary scale and in habitat. The oxygen consumption of isolated and of
bunched Ophioderma was tested by Winkler's method.

A typical experiment was made on three groups of 8 animals each,
selected so that there would be the same size relations in each group.
The members of each lot were momentarily dried on filter paper and
weighed in a known amount of sea-water. The lot weighing most
was placed together in a bottle holding approximately 8 liters (ac-
tually 8,500 cc); the lot weighing next most was separated, and
each individual was placed in an Erlenmeyer flask of about i liter
capacity (actual average 1,143 cc). The third and lightest group of
8 animals were placed together in a second bottle like the first. Ap-
propriate blanks were run on both sizes. All bottles and flasks were
fitted with rubber stoppers, which carried tubes for drawing titra-
tion samples without contact with air and with a minimum disturb-
ance. The containers were gently reversed twice just before sam-
pling to insure an even mixing of the sample, and other due experi-
mental precautions were taken. Determinations were made 4 hours
after the start of the experiment to determine initial relations. Later,
tests were made at 24-hour intervals, except that during the middle
course of some experiments they were made at 48-hour intervals.

' Hyman (1923) found the oxygen consumption of the anterior pieces of Planaria
dorotocephala to be at the rate of 260 cc. per hour and kilogram. Krogh (1916) lists a
rate in calories equivalent to 393.6 cc. per hour and kilogram for Apis meUifica, and
of 72 and again of 11 5.2 for Musca. The highest rate listed in Krogh's tables for
crustaceans, all gillbreathing, is of 6.528 cc. per hour and kilogram. The values given
above for land isopods are subject to a calibration error of about 5 per cent.

EFFECT OF CROWDING ON SURVIVAL 187

The longer tests were terminated piecemeal, as different individ-
uals fragmented their arms to such an extent that comparisons were
no longer fair. The records show the rate of oxygen consumption, as
well as the total oxygen consumption, of twelve groups of aggregated
Opliiodcrma during their entire starvation period, and of six lots of
similar size and number which had been isolated until they, too, died
of starvation and confinement. In addition to the corrections for
the slight difference in amount of water available for each animal,
the crude data were further corrected for differences in weight, since
tests had shown that, within the size limits used, oxygen consump-
tion is directly related to the size of the animal, although the greatest
observed spread between the total weight of 8 isolated animals and
either of the accompanying bunches was only 0.4 gm.

Six tests were run to determine the initial relations in oxygen con-
sumption of isolated and bunched individuals. When these were
examined at the end of the first 4 hours, it was found that on the
average the groups of larger individuals, totaling 48 in all, had con-
sumed 208 cu. mm. of oxygen per individual of 1.25 gm. moist
weight. The 48 smaller animals making up the second bunch had
consumed 197 cu. mm. when reduced to the same standard weight;
while the 48 isolated starfishes had consumed 379 cu. mm. per stand-
ard individual of 1.25 gm. Here we see again that at the start of the
experiment the isolated individuals are consuming oxygen at a much
higher rate than are their aggregated fellows when the only known
difference between them is that the latter have been allowed to col-
lect in bunches. When these results are examined statistically by
Student's method, we find that there are 66 chances in 10,000 of
random sampling yielding such a wide deviation from the mean.
Needless to say, such results are statistically significant. This initial
relationship continued in most cases past the first 12 hours, and
sometimes past the 24-hour samphng, but had usually disappeared
before 48 hours.

Two of these initial sets of experiments and four other similar
ones were followed through with starving animals until terminated
by death or by fragmentation of the arms. The mean results in
terms of total oxygen consumption of the surviving members of each

ANIMAL AGGREGATIONS

lot are summarized in Figure 14, which in A plots the mean rate per
hour per living individual in the isolated (solid) and aggregated state
(broken line). The broken line is drawn as the mean of the two

Fig. 14. — Showing, A, the mean rate of oxygen consumption per hour at the be-
ginning and during successive fifths of the starvation period; each space on the ordi-
nates represents i cu. mm. oxygen. B, The mean oxygen consumption; each space on
the ordinates represents 100 cu. mm. of oxj-gen. The broken line gives the respiration
of the bunched individuals.

EFFECT OF CROWDING ON SURVIVAL 189

series of bunches which were run simultaneously. The means for
each of these are indicated for each point. The graph starts with
the rate given during the first 4 hours and proceeds with that given
during the different fifths of the experiments. This means of chart-
ing is used because of the varying length of the experiments to be
summarized, all of which were physiologically the same length, in
that they were terminated when the starving individuals being
tested had reached approximately the same degree of depression as
measured by death or by fragmentation of the arms.

Graph B shows similarly the total mean oxygen consumption for
each group. The initial amount indicated is that which would have
been given had the animals continued to respire for a period equaling
the others in length but at the rate given in the first 4 hours of
experimentation. Again the solid line represents the isolated in-
dividuals and the broken line the bunched animals, and again the
spread of the means for each series of bunches is indicated. This
form of presentation shows at a glance the more rapid consumption
of oxygen in initial stages, particularly on the part of the isolated
individuals, in contrast with the reduced rate later. The increase in
the rate of respiration near the close of the experiment is due to
the greater consumption of oxygen in the process of autotomy and
to the inclusion of some cases in which decay of fragmented parts of
arms may have occurred.

The isolated animals have a rate of oxygen consumption 186 per
cent above that of the groups during the first 4 hours. Later this
mean rate of consumption falls for both, but more rapidly for the
isolated than for the bunched individuals, so that, when the entire
course of the experiments is considered, the rate of use of oxygen
by the isolated animals is only 83 per cent of that of the accompany-
ing groups. The differences of these means have about the same
statistical value as that noted for the initial 4 hours of respiration.

The foregoing statement and graph does not show the extent of
the difference between the oxygen consumption of the isolated and
the bunched animals, since, on the average, the former died off" more
rapidly than did the bunched individuals. The results reported here
are in terms of the mean oxygen consumed by the animals still liv-

I go

ANIMAL AGGREGATIONS

2 00

180

160

140

120

100

Fig. 1$. — Showing the total oxj^gen con-
sumption of isolated (solid line) and bunched
starfish (broken lines) in the different fifths
of the starvation experiments. The heavy
broken line gives the mean of the two
bunches shown by the lighter lines. Each
space on the ordinates represents i cc. of
oxygen.

ing; hence, in the later stages
of the tests we are comparing
the mean rates of the most
hardy of the isolated individu-
als with that of a large bunched
group that has been subjected
to less rigid selection.

When the differential autot-
omy is considered in connec-
tion with oxygen consumption,
it is found that the effect of
isolation is much more marked
than when only means of sur-
vivors are compared. These re-
lations are shown in Figure 15,
which gives the total oxygen
consumed, corrected for weight
of the different animal groups.
Again the initial value is that
which would have been con-
sumed had the initial rate held
for a period equaling the others
shown. It is added only for the
purpose of graphic comparison.
The preceding types of analysis
showed that after the initial
period the rate of oxygen con-
sumption remained approxi-
mately constant at their re-
spective levels for both bunched
and isolated individuals. On the
other hand, the complete data
show a decided falling-off in
oxygen consumption near the
end of the experiments, due to
the decrease in numbers of ani-
mals present. The effect of the

EFFECT OF CROWDING ON SURVIVAL 191

differential length of life under experimental conditions is shown in
the earlier and greater reduction in oxygen consumption in final
stages of the isolated rather than of the bunched individuals.

The possibility of consuming a similar amount of oxygen in the
same time was the same for the lot of isolated as for the bunched
Ophioderma, had they been given equal treatment. The conclusion
is obvious that when similar Ophioderma are isolated without food
in clean glass receptacles of approximately equal volume per number
of individuals present, their rate of oxygen consumption, following
a period of initial stimulation, is significantly depressed, and their
expectation of remaining intact, or of living, is less than if they are
allowed to aggregate.

FACTORS CONTRIBUTING TO GROUP PROTECTION

The question immediately arises as to the type of beneficial in-
fluence exerted by the grouped individuals upon each other. There
has not been opportunity to investigate this aspect of the problem
thoroughly. Certain data have been collected that are of decided
interest, and these will be presented not as a final answer but as a
possible guide to the ultimate solution of the problem.

The earlier students of the effects of crowding, who uniformly
found only deleterious effects, supposed that there is some harmful
secretion given off by the crowded animals. Similarly, workers who
have reported beneficial effects of increased numbers of individuals
in relation to volume of the containing water have postulated some
beneficial secretion which is preservative in action. Drzewina and
Bohn, in their series of studies upon this subject, in which they used
a wide range of aquatic organisms, suggested that the observed
beneficial effects are due to a hypothetical chemical which is secreted
by the mass of animals in sufficient amounts to have an autopro-
tective effect. When the crowded conditions produce harmful
results, they postulated another chemical substance which is auto-
destructive.

Inasmuch as these Ophioderma live naturally among eelgrass
where they crawl over the blades, and since the formation of dense
aggregations is greatly retarded if not entirely prevented by the
presence of eelgrass in the stock aquaria, it seemed possible that

ig2

ANIMAL AGGREGATIONS

the observed effects might be due to physical rather to chemical
factors, at least in this case. In other words, it seemed possible
that in the absence of suitable non-ophiurid materials the animals
might themselves serve as substitutes for certain elements of the
normal physical environment.

This possibility was tested by placing glass rods of different
lengths, bent into different shapes, in some of the Erlenmeyer res-
piration flasks and using these for isolation chambers. The glass

TABLE XV

Showing the Initial Relations in Oxygen Consumption of Ophioderma Iso-
lated INTO Flasks Containing Irregular Heaps of Glass Rods, in Com-
parison WITH Bunched Individuals and with Those Isolated into Plain
Flasks

(The Results Are in Terms of Cu. Mm. of Oxygen Consumed
per Mean Individual of 1.25 Gm. Weight)

Experiment No.

Bunches

Mean

Glass

Difference

Isolated

Difference

Temperature
(Centigrade)

I

192
120
226
220
290
166

280

iSS
229
410
269
291

88

35

3

190

— 21

125

421
176
522
397
507
253

141

21

293

-13

238
-38

22-24

23-24

23-22

24-24 . 5

24-24

25-25

2

4

6

Mean. . .

202

272

70

379

107

rods formed a loose irregular pile, over and through which the
starfish could crawl or against which it could come to rest. Four
such flasks, with their accompanying controls, were added to the
series already described. Such experiments were begun only after
the conclusions outlined above were clearly indicated; they com-
prised six sets of 4-hour tests of initial respiration and two of the
long-time respiration experiments. The tests were made similarly in
every respect to those already described ; the flasks containing glass
rods were placed alongside those with isolated and bunched indi-
viduals with the behavior of which they were to be compared.

The results of exposure for 4 hours under these conditions are
shown in Table XV. Here the respiratory rate is calculated in
terms of 1.25 gm. per individual and for the actual amount of water

EFFECT OF CROWDING ON SURVIVAL

193

present after the glass rods were added. The table gives a compari-
son of the initial respiration of 24 Ophioderma isolated into flasks
containing glass rods, with that of 48 isolated into plain flasks, and
with 96 divided among the accompanying bunches.

The Ophioderma associated with the clean glass rods consumed
more oxygen per individual in 5 out of 6 cases than did their asso-
ciated bunches. The mean dift'erence of 70 cu. mm. is beyond the
statistical border of significance, since there are 85 chances in a
1,000 in this case and 112 in the next, that random sampling would
give so great a deviation from the mean. On the other hand, the

TABLE XVI

Showing Final Mean Relations in Oxygen Consumption of Ophioderma

Separated under the Conditions Shown for the

Heading for Table XV

(The Results Are in Cu. Cm. of Oxygen Calculated to
Equal \\^eight for Each Series)

Experiment No.

First Bunch

Second
Bmich

Isolated

Isolated with
Glass Rods

Weight (Gr.)

Temperature
(Centigrade)

.

14.050

7.820

14.103
7.809

II. 914
5.220

17.926
10. 811

1-34
1 .21

21-25
21-24

6

individuals associated with the glass rods consumed less oxygen than
the accompanying animals isolated into plain flasks, in 4 out of the
6 cases, with a mean difference of 107 cu. mm. Thus the initial
effect of the presence of the heap of glass rods with the isolated in-
dividuals was to produce a rate of oxygen consumption intermedi-
ate between that of the bunched and the other isolated individuals.
The position of the mean, nearer to that given by the bunched indi-
viduals than to those isolated in plain flasks, while suggestive, is not
significant statistically.

The results in the two total respiration experiments which con-
tained this feature are given in Table XVI. Here the presence of the
glass rods was accompanied by an even greater total oxygen con-
sumption than that of the bunched individuals and, as was shown
in Table XIV, the autotomy effects are more closely related to those
of the bunched than to those of the other isolated animals.

194 ANIMAL AGGREGATIONS

We have, then, indications that the provision of an opportunity
for physical contact with the Hfeless glass rods produces effects
similar to those given by the bunching of live starfishes. Obviously,
more work is needed at this point before we can come to a definite
conclusion. Other preliminary experiments indicate that the pres-
ence of irregular heaps of paraffined glass rods or of parafiined
rubber tubing tends to prevent autotomy of arms of individuals as
compared with other starfishes isolated into plain dishes.

NATURAL AGGREGATIONS OF ASELLUS

The fortunate discovery of a group of gigantic aggregations of
water isopods, Asellus communis, gave opportunity for further res-
piration experiments on these under natural conditions. An account
of some aspects of these aggregations has already been given, de-
scribing the methods of formation of such groups. For our present
purposes it is necessary to remember that these aggregations oc-
curred at the downstream end of culverts and on the downstream
side of an overflow across a sand roadway which divided an exten-
sive cat-tail swamp at the head waters of Dune Creek, in the Indiana
dunes region. At the lower edge of this overflow, just before it
widened out into the lower swamp, great masses of Asellus collected
in winter and early spring about willow shrubs, old cat-tails, or in
depressions where they might find a lodging.

As was shown in the analysis in a preceding chapter, the forma-
tion of these aggregations was conditioned to a large extent, perhaps
completely, by the interaction of the tropisms of the individuals
with environmental factors. In turn, the accumulation of animals
was sufficient to affect decidedly the water surrounding and pene-
trating them. This effect was shown by the loss of 60 per cent of
the oxygen normally present in the stream. In one observation the
stream above the main bunch had 6.37 cc. of oxygen per liter, while
a collection from the midst of this large aggregation had but 2.61
cc, a loss of 3.76 cc. of oxygen. Similarly the pH of the stream was
lowered as much as 0.2 of a pH unit in extreme cases, and averaged
0.09 of a unit less than normal when all the twenty-three tests were
considered. These changes were brought about when the clusters

EFFECT OF CROWDING ON SURVIVAL 195

were located in a stream 15-18 cm. deep where the surface water
was moving at the rate of 25 cm. a second.

Field tests were run to determine the effect of aggregation upon
the rate of oxygen consumption of the aggregated isopods. For this
purpose, a respiration chamber was made by firmly attaching a
short piece of snugly fitting rubber tubing to a small wide-mouthed
bottle of about 14 cc. capacity. The rubber tubing was long enough
to extend beyond the bottle mouth about the length of the bottle
neck. A second similar bottle inserted into this rubber collar until
the mouths of the two met tightly made a respiration chamber of
27.3-30 cc. capacity. The rubber tubing served as a collar to hold
the two bottles firmly bound together, and came in contact with
the water only to a very slight extent. At the end of the respiration
period the contained isopods were all shaken down into the first
bottle. The other was removed, closed with its glass stopper, and
used as an isopod-free collection sample. The agitation necessary
to shake the isopods all into one bottle must have mixed the con-
tained water fairly thoroughly.

A sample collected in this manner was treated according to the
usual Winkler's method, except that only one-fourth of the usual
amount of the reagents was added. Titration was carried on in the
field, using a standardized sodium-thiosulphate solution of about
0.002 normality. Blanks were run with each set of five respiration
tests. These tests were always run in parallel series — one composed
of animals taken from the aggregations; the other containing the
same number of isopods of similar size that w^ere scattered about
singly when collected, and which had not been aggregated recently
if at all. Ten male isopods, or 20 of the smaller females, were the
standard number used at one time in each respiration chamber.
The respiration tests of lots from the aggregation and those that
before the tests had been scattered singly were, with one exception,
run simultaneously. In this exceptional case, the results obtained
may be affected by the greater amount of oxygen present in the
water when the isolated animals were tested.

Care was taken to collect the water for the tests from the same
depth at the same point in the stream, and in as nearly the same

196

ANIMAL AGGREGATIONS

manner as possible. After the respiration chambers were closed,
they were placed side by side in running water and were thus exposed
to identical conditions during the respiration period, which usually
lasted for i hour. The isopods from each series were brought into
the laboratory, where their moist weights were determined. The
results obtained are summarized in Table XVII. The figures on
oxygen consumption and weight are mean results from the number
of groups indicated in the first column, or from the same number of
isolated individuals as were tested in the groups.

TABLE XVII

Showing the Amount of Oxygen Consumed by Groups of 10 Male Isopods during

Their First Hour after Removal from a Large Bunch, as Compared with

A Similar Number of Animals That Were Isolated When Collected

E^5

BtracHED Isopods

Scattered Isopods

\umber
Groups

Cu. Mm. 0,
Used

Weight
(Grams)

0. Calculated
to Equal
Weight

Cu. Mm. O2

Used

Weight
(Grams)

0, Calculated
to Equal
Weight

6

5
12

9
IS
13*

5-7
28.9
31-4
33-2
64.0

1-338
I - 105s
1-330
1-337
I -245

19.9

42.9
41.9
48.0
80.3

1.326
I .402
I. 187
1.252
I. 16

s

5

4

5

36,6

46.9

52.8

86.2

* Last test ran 2 hours.

Twenty-five comparable tests were run with or without correcting
for weight. The set of lo isopods that had previously been scattered,
in all cases but two, used more oxygen during the respiration period
than did the set of the same number taken from the aggregation and
started at almost the same time. Always the total amount of oxygen
consumed by all the members of one series of animals that had been
scattered was greater than that used by the series from the aggre-
gated isopods run at the same time.

Without correcting for weight, the mean difference in oxygen con-
sumption in these live different series shows that those that were
soHtary when collected used 12.33 cu. mm. more oxygen during the
first hour's exposure in a respiration chamber than those that came
from the aggregation. Apphcation of Student's statistical method
shows that there is less than i chance in 100 of getting so great a
deviation in either direction from the mean in random sampling.

EFFECT OF CROWDING ON SURVIVAL 197

Such results are statistically significant. If observed differences in
weight are considered, the calculated difference in oxygen consump-
tion between the bunched and the isolated isopods is still greater.

A group of tests were run to find the respiration rate of the
bunched isopods as compared with those scattered among the vege-
tation in the upper swamp. These latter animals had not aggre-
gated; nor had they been exposed recently to the stimulus of a strong
current. The tests were made as were those preceding, except that
females were used because they were so much more numerous than
males in the swamp. Since the females are much smaller than the
males, 20 were placed in each respiration chamber, where 10 males
had been used in the other tests. The rate of oxygen consumption
was sHghtly higher in the isopods taken from the aggregations than
that of the similar females from the vegetation of the upper swamp,
even when the observed figures are corrected for differences in
weight; but the results as observed are without statistical signifi-
cance. Before calculation for weight differences there are 8 chances
in 10 of securing so great a difference by random sampling. More
work is needed at this point, but the results indicate that the fe-
males of the aggregations are in more nearly the same physiologi-
cal state as those scattered in the upper swamp than were the males
from the aggregations like those scattered in the current below the
swamp but above the aggregations.

The rate of oxygen consumption of these females taken from the
aggregations approaches that of the males run at approximately the
same temperatures. The males tested on the afternoon of April 15,
at a temperature of 15° C, consumed, on the average, 33.2 cu. mm.
of oxygen per 10 males. The 20 females tested 2 days later at 17° C.
used, on the average, 30.2 cu. mm. of oxygen. They weighed 1.1853
against 1.3765 grams for the males. If they had continued to con-
sume oxygen at the same rate per unit weight and had weighed as
much as the males, they would have used 35.1 cu. mm. of oxygen
in the same time.

If the comparison is carried a step farther, we find that these
females do not have a rate of oxygen consumption comparable with
that of males scattered through the stream just above the aggrega-
tions. Using values calculated from equal weights, we find that the

igS ANIMAL AGGREGATIONS

latter consumed 52.8 cu. mm. of oxygen as against 35.1 for the ag-
gregated females. If the male isopods in the upper swamp have
about the same rate of oxygen consumption as the females there,
a point unfortunately not directly tested, we can reconstruct their
respiration history as follows:

The isopods in the upper swamp, on being swept off their feet,
struggle against the current and are borne along clutching at straws
as they pass. Sometimes an isopod clasps another and both are
carried along head over heels until they are deposited in some de-
pression with a sandy bottom, from which they start upstream only
to be carried down again. Such isopods have their rate of respira-
tion greatly increased as a result of their exertions. If they are
carried on down so that they lodge among the grasses below or
among their fellows who have already lodged there, they again take
on about the same rate of respiration they had shown in the upper
swamp.

COMPARISONS AND CONCLUSIONS

Like the land isopods and the brittle starfish, Asellus is not usually
regarded as a social species. These isopods are frequently found in
greater abundance in one place than another; but outside the breed-
ing season, environmental analyses usually reveal significant differ-
ences which may condition differential distribution; and, as we have
seen, these natural aggregations originate largely through the reac-
tions of individuals to their environment, rather than through social
impulses. Great aggregations of these animals in nature have hith-
erto escaped notice; and, so far as I am aware, no other great natural
aggregation at the low level of group organization here existing has
been analyzed to find the physiological effect of aggregation upon
the aggregants.

The laboratory experience outlined* above had previously shown
that, conditions being otherwise favorable, land isopods and the
brittle starfish, Ophioderma brevispina, have their rate of oxygen
consumption increased as a first effect of aggregation. Later the
rate of oxygen consumption is lowered as compared with individuals
isolated under similar conditions. The respiration relations were
not obtained at the beginning of the aggregations in the case of the

EFFECT OF CROWDING ON SURVIVAL 199

isopods in the field. Neither was it possible to arrange to make
respiration tests without disturbing the aggregations as was done in
the laboratory. One must use care in interpreting results based
upon the oxygen consumption given during the first hour following
the considerable disturbance occasioned by the extraction of 10
isopods from the midst of an aggregation as compared with that of
10 isopods for the first hour that they are placed together in a small
respiration chamber after having been previously scattered.

It must also be noted that the isopods from the aggregations and
those collected while still separate cUd not behave similarly in the
respiration chamber. The former tended to gather in the restricted
space furnished by the necks of the bottles forming the respiration
chamber, where they were well shaded by the rubber tubing that
held the two bottles together. Those collected while scattered tend-
ed to remain so, and were more likely to be found exposed to the
brighter light at the ends of the bottles. Since all lots tried behaved
uniformly in this regard, and since previous location also made a
difference in the oxygen consumption, the differential behavior with-
in the respiratory chamber may be taken as one of the factors in
determining the respiration relations observed. The exact weight
that should be given this factor has not been determined. Such
behavior might result in the shaded, more aggregated isopods using
less oxygen during the respiration period than would the unshaded
scattered individuals, as was found to occur.

As a result of experience with respiration of laboratory aggrega-
tions, it seems probable that the differential respiration observed is
due to a difference in muscular tonus between the aggregated and
the isolated animals. If this idea is applied, we have the reasonable
assumption that isopods taken from the upper swamp far enough
away so that they are undisturbed by the currents set up by the
overflow are under approximately the same muscular tension as are
those taken from the dense aggregations, and that both are under
less muscular tension than are the isolated iospods that have re-
cently been exposed to the full impact of a current which in many
instances has been carrying them hurtling along over a sandy stream
bed. The evidence at hand allows no suggestion as to whether the

200 ANIMAL AGGREGATIONS

respiration results observed are due to muscular tension acting as
a direct result of the previous position of the isopods or whether
they are due to the difference in muscular tension resulting from
the differential behavior, which is in turn conditioned by their previ-
ous position.

The possibility that the increased carbon-dioxide tension and the
decreased oxygen supply within the aggregation may have reduced
the oxygen requirements of these animals is negatived by the fact
that the rate of oxygen consumption of isopods from the center of
the aggregations was slightly, but not significantly, higher than that
of those taken from the upper swamp where the oxygen tension and
pH were approximately that of the main current above the large
clusters.

The results obtained from the study of these enormous aggrega-
tions of isopods in nature demonstrate that such masses can affect
environmental conditions markedly, even in flowing water. Obvi-
ously, such groupings would have still greater effect in quiet water
or on land. As we shall soon see, such aggregations may have survi-
val value in the laboratory, due to the modification they produce
in their environment. No tests were made of the possible survival
value of the isopod aggregations in nature, but the fact that they
do alter their environment to a measurable degree extends the possi-
bility of the application of the laboratory studies on the survival
value of masses of animals. As in the case of the laboratory groups
of land isopods and of Ophioderma, these water isopods in nature
under adverse conditions tend to use other isopods in place of in-
animate elements of their physical environment, when the former are
present in considerable numbers and the latter are lacking in their
usual ratio of abundance in proportion to the numbers of animals.

In conclusion we may recall the statement at the beginning of
this chapter, which these later studies have served to confirm, that
under laboratory conditions the formation of aggregations serves
to make these animals more quiet, and in the long run proves to be
what is usually called an "adaptive reaction"; and, so far as we have
information, the same results come also from aggregations under
natural conditions.

CHAPTER XII
PROTECTION FROM TOXIC REAGENTS

Shortly after the pubUcation of the preliminary announcement
which the preceding chapter restates, Bohn and Drzewina (1920)
published the first of a brilhant series of studies, apparently not yet
concluded, which opened for investigation another phase of the
subject that has yielded much evidence concerning the possible
survival value of groups of animals for the individuals composing
the groups. The general results obtained may be summarized in
their own words (1926):

"Nous avons recherche I'intervention du facteur masse dans les
reponses de divers organismes vis-a-vis de multiples agents nocifs
du milieu exterieur. Dans une masse M d'eau, on introduit une
masse m d'un etre vivant {m etant egal ou inferieur a M/ioo),
cette masse resiste a un agent nocif determine (substance chimique,
par exemple); mais une masse plus petite, m/10 ou m/ioo, ne
resiste pas; tout se passe comme si la masse de matiere vivante
exergait, vis-a-vis d'elle-meme, un eft'et protecteur (auto-protec-
tion)."

MASS PROTECTION FROM COLLOIDAL SILVER

Drzewina and Bohn worked with colloids of the heavy metals,

particularly with colloidal silver. We have spent much time in this

laboratory, also,' studying mass relations of animals when exposed

to colloidal silver;' and if the following report and discussion is

' The stock suspension of colloidal silver was made as follows, from directions fur-
nished by Dr. Terry-McCoy of the Department of Chemistry of the University of
Chicago: Dissolve 4 grams of commercial dextrine in 100 cc. of water and then 4 grams
of pure caustic soda. Dissolve 3 grams of silver nitrate in 20 cc. of water and add to the
dextrine-soda solution. The precipitate of silver oxide that is formed is gradually re-
duced by the dextrine, the color changing to reddish brown. After 20-30 minutes, add
100 cc. of 90 per cent alcohol and stir. Allow the mixture to settle for 15-20 minutes,
and then pour off the liquid from the particles of silver at the bottom. Add about 200
cc. of water and the silver particles will generally disperse; however, if this is not the
case, shake or stir until an even suspension of silver results.

202 ANIMAL AGGREGATIONS

based largely upon our own work (Allee and Schuett, 1927), it is only
because I am more familiar with this, and by no means because it is
more important than work in this field done by Drzewina and Bohn
in following out their original discovery.

It is easy to demonstrate, as Drzewina and Bohn state, that
within limits, the greater the mass of animals, the better the pro-
tection. Thus with Planaria dorotocephala, twelve sets of 2 each,
exposed at room temperature to 10 cc. of water containing 10 drops
of the standard suspension of colloidal silver, showed the beginning
of head disintegration in from 4.5 to 10 hours; while ten similar lots,
each containing from 10 to 75 worms, all lasted over 36 hours in the
same volume and concentration. A species of Cladocera and the
isopod Asellus communis showed similar, but less marked, group
protection.

Similar tests, in which the animals were exposed to the action of
dilute suspensions of colloidal silver for considerable time and were
then washed and transferred to water to which they were accus-
tomed, yielded such results as are listed in Table XVIII. With the
worms, the results are summarized in terms of survival after ex-
posure, with added information concerning the number of worms
that were able to crawl either with or without stimulation with a
camel's hair brush. Other data at hand show that the protective
effect of the group was greater than is indicated by this summary,
but the summarized account is sufficiently conclusive. With the
brittle starfish, Ophioderma, tests were made concerning the righting
ability of the animals after exposure to colloidal silver. In these
tests, after exposure, the washed animals were transferred to sepa-
rate finger bowls each containing 250 cc. of well-aerated sea-water.
It will be noted that the mean righting time of the isolated animals
is given in minutes while that of the animals tested in groups of
5 is given in seconds.

More cases with different concentrations could be given, but
the result is the same in all. This agrees entirely with the experience
of Drzewina and Bohn, who used a wide range of animals; and with
that of Bresslau, using different toxic reagents with ciliates. Yet the
possible menace of crowding, so frequently given as the only result

PROTECTION FROM TOXIC REAGENTS

203

of confining many animals within a small space, may easily be
demonstrated even here. With the Ophioderma, for example, if the
glass finger bowls were covered with glass plates during the exposure
to colloidal silver, thus stopping free gaseous exchange, the grouped
animals were in much worse condition than were their isolated
fellows.

A priori, one would expect exactly such a result, since in each
case there is much less of the toxic substance present per individual

TABLE XVIII

Showing the Results of Exposure to Suspensions of Colloidal Silver
Followed by Transfer to Water Similar to That to
Which the Animals Were Accustomed
All were run at room temperature.

Animal

Cc. of
Water

Drops

Colloidal

Silver

Number
Animals

Time

Exposed to

Colloidal

Silver

(Hours)

Number

That
Sur\'ived

Number
That
Were

Moving

Number
That
Died

Planaria maciilata

15
15

5
5

12
12

I
I

6(3)
i(7S)
5(3)
1(25)

2

2

26

26

II

72

9

25

2

66

23

7
3
6

P. dorotocephala . .

10
10

2
2

40(1)
3(25)

7
7

15

75

I

63

25

Dendrocoelum lac-
teum

5
5

I
I

5^3)
1(24)

26
26

3
18

I
18

12

6

Ophioderma bre-
vispina

50
50

2
2

48(1)
12(5)

14,5-16
14-5-16

Righted

41
59

Mean
Time

30 min.
38 sec.

composing the larger groups than when from i to 3 animals are
placed in the same amount of the same concentration of toxic sub-
stance (Pieron, 1921). Drzewina and Bohn consider this possibility
(1921C, 1928) and dismiss it as inadequate because of their experi-
ments upon the effect of reconditioned toxic solutions and upon the
relation of volume of the toxic material to its effect upon exposed
animals. They summarize their conclusions repeatedly in some
such words as the following (1921J):

204 ANIMAL AGGREGATIONS

"Bref, si chez diverses especes que nous avons examinees jusqu 'id
(Paramecies, Colpodes, Stylonichies, Stentors, Hydres, Convoluta,
Glossiphonies, tetrads de Grenouille) tout se passe comme s'il y
avait emission rapide d'une substance ou de substances assurant une
defense; chez Polycelis c'est le contraire:' on assiste non pas a une
auto-protection mais a une auto-destruction.''^

They do not believe that the greater resistance of the group is
due to a more rapid using-up of the toxic substance, and cite two
types of experiments to support their contention (1921). In the
first type they expose some hundred of Convoluta in one suspension
and 2 each in the accompanying dishes. Even if the former receive
10 drops of colloidal silver to 25 cc. of water and the isolated animals
have only 2 drops for the same volume, they find that the large
group survives after the others are disintegrated. This is a fairer
test than if the same amount of colloidal silver were added in both
cases, but still it must be remarked that the ratio of colloidal silver
per individual is about ten times greater in the case of the 2 individ-
uals, when compared with the larger group.

We have made some exact tests to cover this point, and I present
the summary of our results in Table XIX. The tests reported
show clearly that if a group of 25 Planar ia are placed in the
same volume and the same concentration of colloidal silver as
are isolated individuals, the former survive the experience in
good condition while the majority of the latter succumb, and
almost all the others are severely affected. Further, if similar pla-
narians are placed at the same time in the same volume of water,
to which has been added sufficient colloidal silver to make the
concentration per individual equal to that given the isolated worms,
the bunched animals all succumb to a treatment that left 37.5 per
cent of the isolated individuals alive, even if most of them were
strongly affected. And finally, if the groups are placed in propor-
tionate volume and proportionate concentration of colloidal silver,
their condition approaches that of the isolated individuals. This
evidence shows that the greater protection which the mass furnished
when exposed to colloidal silver was due to the smaller amount of the

' Against KCl (W. C. A.).

PROTECTION FROM TOXIC REAGENTS 205

toxic substance which each individual had to remove from suspen-
sion in order to lower the strength below the threshold of immediate
toxicity. The mechanism for this removal will be discussed later.

Another type of experiment was run with Ophioderma to deter-
mine whether the protective effect of the bunch was due to the fact

TABLE XIX

Animals Used: Planaria dorotocephala, 15-25 Mm. Long, Exposed to Action of
Colloidal Silver for 7 Hours, Then Washed and Transferred to Water
TO Which They Were Acclimated, and Examined on the Day Following
the Exposure
I. 40 planarians isolated into 10 cc. of water plus 2 drops colloidal silver

No. Percentage

Normal, crawling when stimulated i 2.5

Less than half disintegrated 2 5.0

Half and less than two-thirds disintegrated ... 4 10. o

Two-thirds but not all disintegrated 8 20.0

Wholly disintegrated 25 62 . 5

II. 150 planarians in 6 lots each in 250 cc. water plus 50 drops colloidal silver

Normal, crawling when stimulated o o

Less than half disintegrated o o

Half, but less than two-thirds disintegrated. . . 18 12

Two-thirds but not all disintegrated 52 35

Wholly disintegrated 80 53

III. 71 planarians in three lots each in 10 cc. of water plus 2 drops of colloidal

silver

Normal, crawling when stimulated 63 88.8

Normal, not crawling when stimulated i 1.4

Head started to disintegrate 7 9-8

Head completely disintegrated o 0.0

IV. 75 planarians in three lots each in 10 cc. of water plus 50 drops of colloidal

silver
All wholly disintegrated.

that the bunched animals were exposed to a smaller concentration
of colloidal silver than were the isolated animals. Since the area of
the surface would affect the aeration and so affect survival values,
the larger numbers were placed in crystallization dishes with a diam-
eter of 24 cm., while the smaller bunches and the isolated animals
were in the usual finger bowls with a diameter of 10 cm. The essen-
tials of the set-up were:

2o6 ANIMAL AGGREGATIONS

One bunch of lo animals with 25 cc. of sea-water and 1.5 drops of
colloidal silver each, and with a surface exposure of 45 sq. cm. per
animal.

One bunch of 5 animals with 5 cc. of sea-water and 0.3 drops of
colloidal silver each, and with a surface exposure of 15.6 sq. cm.
per animal.

Five isolated animals with 25 cc. of sea-water and 1.5 drops of
colloidal silver each, and with 78 sq. cm. surface exposure.

A second set was exactly similar, except that the animals were
exposed to 2, 0.4, and 2 drops of colloidal silver per individual.

The exposure in a typical experiment lasted 15.5 hours, after
which the animals were washed as usual and transferred to fresh
sea-water for observation for the following 24 hours. At intervals
records of numbers that righted after having been turned over, and
other evidences of activity, such as spontaneous motion of tube feet,
were recorded.

All of the animals bunched in the 10 cm. finger bowls recovered
sufficiently to right themselves within 24 hours, despite the fact
that 4 of the 10 were much corroded by the action of silver that had
settled to the bottom of one of the finger bowls. In both experi-
ments, the isolated animals fared next best, although they made a
much poorer showing than did the bunched individuals that were
exposed to a much lower amount of colloidal silver per animal;
only 6 of this set of 10 righted. The bunched animals, exposed to
the same volume and amount of colloidal silver as those isolated,
ranked a poor third, which may be due to the decreased surface
exposure or to some of the better-known ill effects of crowding.

These experiments indicate either that the decreased amount of
colloidal silver or the reduced volume of sea-water present per
individual, or both, markedly favored the survival of the animals
bunched in the small amount of water present in the finger bowls.
Drzewina and Bohn emphasize the latter factor, and accordingly
experiments were run to test out this point. Contrary to the expe-
rience of Drzewina and Bohn (1921a), the toxicity with Ophioderma
was found to depend on the concentration of the colloidal silver
rather than on the volume to which the animals are exposed.

PROTECTION FROM TOXIC REAGENTS 207

It is worth noting that the protective action of the bunch appears
only in the more toxic suspensions, where apparently the toxic
strength per individual was more nearly reached even with the
bunched animals. Drzewina and Bohn apparently would interpret
such an observation as meaning that the autoprotective substance
is secreted more rapidly under more toxic conditions.

RECONDITIONED SOLUTIONS

The second and more convincing experiment used by Drzewina
and Bohn as evidence in favor of their view that the group pro-
tection is furnished by some sort of autoprotective secretion, rather
than by exhaustion of the toxic substance, is as follows (1921(1):

"Nous decantons la solution oil depuis 24 heures sejournent une
cinquantaine d'embryons (tadpoles of Rana fusca) et dont la teinte
revele la presence du colloide; nous y ajoutons le meme nombre de
gouttes que la veille, i par exemple, et nous y plagons deux em-
bryons neufs du meme age. Ceux-ci survivent, alors que des in-
dividus temoins, places dans une solution neuve a i goutte de
collargol, succombent, comme c'est la regie pour les isoles. II semble
ainsi que, attaquees par le colloide, les larves emettent, rapidement,
une substance (ou des substances) qui a pour effet de les proteger."

Wholly similar experiments were run with the brittle starfish,
Opiiiodcrma, to test for the presence of the postulated autopro-
tective secretion. Colloidal suspensions in which Ophioderma had
been exposed were reconditioned by 3 hours aeration with room
air in order that the new lot might not suffer from low oxygen
tension, and were filled to the original volume with distilled water.
The original suspension had been made with i drop of colloidal
silver for each 25 cc, and the same amount was added to this re-
conditioned water. Tests showed that the aeration of freshly pre-
pared suspensions for this length of time caused a color change
but did not markedly affect the toxicity.

Three lots, each consisting of one group of 5 bunched animals and
4 isolated individuals, were placed severally in 50 cc. of such a re-
conditioned suspension, 9 hours after the previous experiment had
closed. These were run simultaneously with three similar lots in

208

ANIMAL AGGREGATIONS

freshly prepared suspensions of 2 drops of colloidal silver in 50 cc.
of sea-water. All conditions were similar. The temperature was
18.5° C. The results are summarized in Table XX.

If Drzewina and Bohn were correct in thinking that the colloidal
silver is not removed from solution on exposure to the animals,
there should have been an excess accumulation of this substance
in the twice-used water; but after again conditioning the used sub-
stance, it was necessary to add the same amount, i drop of colloidal

TABLE XX

Showing Righting Time of Ophioderma after an Exposure of 17 Hours in

Reconditioned and Fresh Suspensions of Colloidal Silver
All gave the righting reaction.

Bunch

Isolated

Number

Mean Time

Spread

Number

Mean Time

Spread

Reconditioned Suspension

IS

73 sec.

10 sec-
12 min.

12

25 sec.

5- II 8 sec.

New Suspension

IS

20 sec.

7-75 sec.

12

25 sec.

10-65 sec.

silver per 25 cc. to this suspension, in order that it might be as
deeply colored as was a fresh suspension made up with that amount
of the colloidal silver. There was only sufficient colloidal silver left
in this twice-used suspension to discolor the liquid. Either the
animals or the aeration, or both, had removed the greater part of
the silver; and it can be readily demonstrated that the effect was
not wholly due to aeration. The water thus treated was distinctly
sirupy, and evidently held organic matter, received from the two
lots of Ophioderma which it had contained.

The effect of this twice-reconditioned suspension was tested as
before. Comparison tests were run with newly prepared suspen-
sions, and the results are summarized in Table XXI.

There is no evidence here of the presence of an autoprotective

PROTECTION FROM TOXIC REAGENTS

209

secretion in the sea-water, protecting animals from the action of
the colloidal silver; nor is there evidence for the presence of an
active autodestructive agent, which Drzewina and Bohn also postu-
lated to explain certain cases in which the greater mass of animals
exposed to KCl die more rapidly than do the solitary individuals.
In the presence of an active autodestructive secretion one would
expect the mass to die more rapidly than do the isolated individuals,

which is not the case.

TABLE XXI

Showing Righting Time of Ophioderma after an Exposure of 19 Hours in
Twice-Reconditioned and in Fresh Suspensions

Bunch

Isolated

Number

Mean Time

Spread

Number

Mean Time

Spread

Reconditioned Suspension

15

46 sec.

10-2 2 1 sec.

II*

345 hrs.

i7sec.-2 2hrs.

New Suspension

15

34 sec.

10-175 sec.

12

1.5 hrs.

iosec.-i4hrs.

* One failed to give the righting reaction.

There is good evidence that the secretion of slime or other or-
ganic matter into the suspension does remove the colloidal silver
and so render the solution less toxic. It is particularly significant
that, after having been twice used, the suspension required as much
colloidal silver as was used in a fresh suspension to bring it to the
same color. Similar results were obtained from other experiments.

More recent experiments in this laboratory, the results of which
are unpublished as yet, show clearly that groups of goldfish live
longer in a given volume of a given concentration of colloidal silver
than do isolated goldfishes similarly isolated into the same amount
of the same concentration of colloidal silver suspension; and that,
under these conditions, the group removes significantly more silver
from suspension than do the isolated individuals. The experiments
do not yet show conclusively whether this will account for all of
the observed group protection.

2IO

ANIMAL AGGREGATIONS

SPECIFICITY OF MASS PROTECTION AGAINST COLLOIDAL SILVER

There are two types of specificity possible: the protective secre-
tion may have no other function, and so be specific in that sense;
or it may be limited in protection to the species producing it. The
first type of specificity will be discussed later. The latter aspect was
investigated by placing one animal in a restricted volume of water
containing a number of animals of a different species. All our ob-
servations show that the protective action of the mass is not limited

TABLE XXII

Showing the Non-specificity of the Protection against Colloidal Silver

Animals

Many Cladocera, i Asellus

I Asellus

Many Cladocera, i Planaria

I Planaria

50 Asellus, I Planaria

1 Planaria

Desiccated parotid gland, 2 Planaria

2 Planaria

Snail slime, 2 Planaria

2 Planaria

Snail slime, 2 Planaria

2 Planaria

1 Physa, 2 Planaria

2 Planaria

Water (Cc.)

10
10
10

Colloidal
Silver (Drops)

10
10

5
5

5
5
5
5

Time to Death
(Hours)

Over 36

4-5
Over 36

5-5
Over 36

5-5
Over 36

7-S

Over 36
Less than 18
Over 36
Less than 16
Over 36

to a given species. This is what would be expected if the fixing of
the colloidal silver in some manner is the principal element in the
protective action. The results of typical experiments are summa-
rized in Tables XXII and XXIII.

As these tables show, such diverse organisms as Cladocera, Asellus,
pond snails, pond leeches, Dendrocadum, and even pond moss, if
present in quantity, markedly protect planarians from the toxic
action of colloidal silver. Even the actual presence of living organ-
isms is unnecessary; snail slime without the snails protects efficiently
apparently by adsorbing the colloidal silver. The slime becomes
densely colored as the water becomes lighter. Suspensions in
water of desiccated parotid glands of sheep exhibit similar adsorptive
phenomena and have similar protective value.

PROTECTION FROM TOXIC REAGENTS 211

A still more severe test of the specificity of the protection was
made by placing recently killed Asellus in glass dishes containing
10 cc. of water plus 2 drops of colloidal silver, and introducing
with these dead isopods one Planar ia dorotocephala. At the end of
7 hours' exposure, 20 planarians isolated into similar volumes of
the same concentration showed, 6, one-third disintegrated; 13, one-
half disintegrated; and one, wholly so. The 5 worms isolated into
suspensions containing dead isopods showed i intact though bloated;
3, heads disintegrated; and i with both head and posterior end

TABLE XXIII

Showing Community Protection against Colloidal Silver
Two drops of colloidal silver in 5 cc. pond water; exposure 13 hours, then washed and
placed in pond water. Temperature 21° C.

Animals

Crawling

Lived

Died

23 Planaria maculata

10 isolated P. maculata

25 Demiroca'litm, i P. maculata

30 Glossophonia, i P. maculata

26 Asellus, I P. maculata

25 Segmentina, 1 P. maculata

Pond moss, i P. maculata

23

I

I

I

23

9

disintegrated. Thus the presence of the recently killed isopods de-
cidedly protected the otherwise isolated worms from the toxic ac-
tion of the colloidal silver. This, taken with all the other evidence
at hand, furnishes convincing proof that the protective action of
the mass against colloidal silver extends beyond the species limits.
We have shown in the preceding pages that, within limits and
other conditions being equal, there is greater protection the greater
the mass of the animals present when exposed to the same amount
of colloidal silver in the same volume of water. Further, the protec-
tion is largely, and perhaps completely, furnished by the fixation
of the toxic substance by the mass of animals, so that each escapes
receiving a lethal dose; while, with the same concentration, isolated
individuals receive a stronger dose of the toxic substance. The
colloidal silver may be differently fixed in different animals; but
with those that secrete slime, like planarians, the colloidal silver is

212 ANIMAL AGGREGATIONS

adsorbed on the slime. With other animals observed, it may be
removed by adsorption on the surfaces of the animals themselves.
Finally, as would be expected from this mechanism, we have dem-
onstrated that the protection furnished by the mass, is, at least to a
considerable extent, independent of the species present.

Our experiments do not support the hypothesis that group pro-
tection among these aquatic animals is furnished by the rapid pro-
duction in the presence of a toxic agent, such as colloidal silver, of a
more or less mysterious autoprotective secretion. It is true that
the production of slime by planarians actually serves as an auto-
protective agent, not only in fixing colloidal silver in these experi-
ments, but very probably in protecting the planarians from sudden
changes in culture or habitat water. But slime production cannot
be regarded as a specific autoprotective secretion, either in the sense
that it is used for no other purposes, for obviously it plays many
other roles in the economy of slime-producing organisms, or in the
sense that slime is limited in its protecting power to the one
species producing it, since the protection furnished by aggregations
of mixed species is easily demonstrated.

OTHER CASES OF GROUP PROTECTION

Bresslau (1924) found that Protozoa give off a substance which
he calls "tektin," a mucin-like body, which is given off in greater
abundance when the animals are stimulated by heat, pressure, meth-
ylene blue, iodine, etc. The tektin, when given off, takes up water
rapidly and exhibits strong surface activity, adsorbing foreign par-
ticles readily. Bresslau tells of putting 2 cc. of liquid from a culture
of Infusoria (either Colpidium or Paramecium) containing many in-
dividuals in one dish and a similar amount from the same culture,
but with few individuals, into another. Into each he introduced i
cc. of I per cent solution of methylene blue. Both produced the
tektin and adsorbed this poison : the culture with the many animals
so much more completely that the possibility of surviving the toxic
action of the poisonous material was greatly increased.

Carpenter (1927) approached this problem from a wholly different
angle. Without referring to the work of Drzewina and Bohn, she

PROTECTION FROM TOXIC REAGENTS 213

attacked the problem of the toxicity of metallic salts on fishes on
account of her interest in stream pollution. Using lead nitrate, she
found that the fatality curve is described by the equation K = i/t log
i/conc, where iT is a constant dependent upon the toxic substance
employed and / is survival time for any given concentration (cone).

When a number of minnows (Leuciscus) were killed successively
by placing them in the same 500 cc. solution of N/'io lead nitrate, the
survival time was found to be prolonged with each successive fish
until the eighth had a survival time approximately double the stand-
ard figure for that concentration and size of fish. The actual sur-
vival times reported are, in order, 73, 89, 92, 93, 94, 120, and 130
minutes. Carpenter makes the deduction that the solution was
successively weakened by the abstraction from it of a certain pro-
portion of the lethal substance by each fish tested and that the
lethal efficacy was thus progressively reduced.

The lethal efficacy of the solution varied in inverse proportion to
the actual size of the fish exposed and directly according to the
amount of lead (as Pb) required to kill the fishes. If the original solu-
tion be relatively weak, a single fish will remove an important frac-
tion from a small amount of solution, and will therefore show a
longer survival time than a similar fish isolated into a large volume
of solution of similar strength. This is in keeping with Huxley's ob-
servations (1922) on the action of mercuric chloride on the gill tissue
of Mytilus. Carpenter concluded that the lead salt causes death,
due to the formation of a film over the gills and skin of the fish, by
an interaction of mucus and the metallic ion, which causes death by
suffocation. If insufficient lead ions are present, the film is shed; and
the solution being thus freed from its toxic element, recovery en-
sues.

An attempt was made to establish the amount of lead which en-
ters into combination with the mucus, and the amount, if any,
which enters the blood-stream. Two large minnows were placed
together in 1-5 liters of distilled water, to which was added sufficient
lead nitrate to supply 6.21 mg. of lead. Immediately after death,
the bodies were transferred to fairly dilute acetic acid for 4 minutes
until the gills were quite clear. After washing the bodies in distilled

214 ANIMAL AGGREGATIONS

water and combining both liquids, these were found to carry 4.62
mg. of lead. The original solution was found, by similar analysis,
to have 1.8 mg. of lead after the death of the fishes. Adequate
analysis of the fishes' bodies showed no lead had penetrated. Car-
penter points out that the overplus of 0.21 mg. found in the analyses
as compared with the original solution makes the experiment incon-
clusive in some respects, but that it does show clearly that the fishes
remove toxic substances by adsorption, just as do our results with
colloidal silver.

In later work (1930) Carpenter finds that the survival time of a
number of North American fishes is directly influenced by the value
of the ratio of the volume of solution to mass of individual fish so
long as the solution is of constant molar concentration. When the
concentration and volume are constant, the survival varies inversely
according to the mass of the fishes. For equal masses of animals, the
survival varies directly with the ratio of the volume of the solution
to the molar concentration. The results upon survival of exposing
groups of fishes to lethal concentrations and volumes of lead salts as
compared with isolated fishes, are not yet clear, but the later mem-
bers of a processionary series killed in the same solution survive
longer than do the earlier members on account of the using up of the
harmful agent.

Preliminary experiments in our own laboratory in which the sur-
vival of groups of 10 goldfish was compared with that of the same
number of individuals isolated into the same volume and concentra-
tion of lead nitrate have revealed no significant differences when
solutions of from N/io to N/200 were used. The technique used
in these experiments differed from that of Carpenter; hence we are
not ready to criticize her results. But our experiments do show that
the group effects clearly exhibited with colloidal silver are not so
readily demonstrated with lead nitrate.

Pawlow (1925) came at the same general problem from still an-
other angle. Led on by Ostwald's basic work on adsorption of toxic
salts, Pawlow worked out theoretical formulas to express the relation
between toxicity and adsorption on animals. These show that if
the active toxic substance has a direct effect upon the organism,

PROTECTION FROM TOXIC REAGENTS 215

then the survival depends on the volume of the medium for any giv-
en concentration of the toxic agent and also upon the density of the
population. The same relationship holds if the active lethal agent
works through capillary attraction or by division among phase
boundaries. These relationships are recognized as forming a bio-
logical aspect of the law of mass action.

In experiments on the brine shrimp, Artemia salina, exposed to
reduced or to increased salt solution, Pawlow found that the theory
given above was verified. The mean life-duration varied from 32.1 to
6.1 in different salt concentrations when the volume was quad-
rupled, while halving the numbers of animals present decreased the
survival between 6.7 and 15.2 per cent. Similar relationships were
found for the amphipod Gammarus pulex when exposed to certain
concentrations of NaCl.

Stimulated to similar studies by their interest in the implications
of the work of Gurwitsch on mitogenetic rays, Frank and Kurepina
(1930) have demonstrated mass protection for sea-urchin eggs from
the action of KCN, both as regards the rate of development and the
percentage surviving. They report that io~^ N solution exerted the
following effects:

Number of eggs i in each 13 drops 68 in 3 drops

No cleavage 85 % 20%

2 blastomeres 15% 4°%

4-8 blastomeres 0% 40%

Another type of protection has already been shown to be operat-
ing, in at least certain instances, with aquatic forms (chap, vii) and
is of sufficient importance to be repeated here in some slight detail.
The discovery of this factor is a result of the work of Fowler (1931),
who has examined the resistance of Daphnia pulex to a great many
chemicals. Usually he found the groups more resistant than were
isolated individuals run in the same volume of the same solutions.
When he pushed the analysis of this phenomenon, particularly with
CaCL, he found that the group was consuming less oxygen per in-
dividual for a given exposure time than were the accompanying
single individuals. Further, when the experiments were run in such
a way that the CO2 was absorbed as soon as produced, the survival

2i6 ANIMAL AGGREGATIONS

of the animals in the groups became the same as that of the isolated
animals. When he re-examined his original data in the light of
these findings, it was discovered that the cases in his early work in
which the singles lived for a longer time than the group were those
in which the dilution of the chemical was such that acclimatization,
rather than direct resistance, was an important factor. In such
cases, as Child has repeatedly pointed out (e.g. 191 5), the animals
with the higher rate of general metabolism have the greater power of
acclimatization, and hence of survival. On the other hand, whenever
the individuals tend to produce conditions which depress their rate
of metabolism and are exposed to solution strengths lethal within a
relatively short time before acclimatization becomes a factor, such
depression has survival value.

Fowler's work also demonstrated that when Daphnia are exposed
in dilute solutions of sodium or potassium hydroxide, the grouped
individuals lived significantly longer than did similar isolated ani-
mals, and that under these conditions the hydroxide in the solution
surrounding the group was reduced in strength as compared with
that surrounding relatively isolated animals. Eight experiments
with NaOH showed that 2 animals in 20 cc. solution in 3 hours
reduced the hydroxide from a mean concentration of 0.00094 N to
0.000764 N. Under the same conditions 20 Daphnia reduced the
same amount of the same solution to 0.000606 N, a difference of al-
most 30 per cent, with a statistical probability of 0.002. Similarly
significant results were obtained with KOH solutions. Here the car-
bon dioxide given off by the animals reacts with the hydroxides to
form carbonates and water, thus weakening the toxic concentration
of the medium. The groups produce carbon dioxide more rapidly
than do animals isolated into the same volume of medium, and hence
decrease the toxicity more rapidly and survive longer.

The problem of the number of animals present in relation to their
survival when exposed to toxic agents presents a somewhat different
aspect when applied to insects placed in containers into which toxic
gases have been introduced in concentration just sufficient to kill
only a part of the isolated individuals with the exposures employed.
Here the production of autoprotective materials presents a distinctly

PROTECTION FROM TOXIC REAGENTS 217

different problem from that found in aquatic animals. Discussion
of experimentation in this connection will be reserved until chapter
xiv, which deals with aggregations and insect survival.

GROUP PROTECTION FROM HIGH TEMPERATURES

Not all adverse conditions in nature are the result of the presence
of toxic materials in aerial or watery solutions where the types of
protection we have just been considering might operate. One other
adverse condition is that furnished by the physical factor of high
temperature. Concerning the ratio between the mass of exposed
animals in relation to volume of medium with respect to this physi-
cal factor, Robertson (192 1) records observations upon the Aus-
tralian infusorian Enchelys farcimen.

A temperature of 30° C. prevents subcultures of this protozoan
from multiplying, and the isolated individual almost inevitably dies.
But shade temperatures of 30° C. are known in pools of South
Australia in which wild Enchelys live. Wild infusorians brought
into the laboratory are similarly killed by such temperatures if
isolated; but if the culture shdes are populated by 20-30 individuals,
they can successfully resist exposure to temperatures of 33-34° C.
for as long as 7 days in succession without apparent injury or abnor-
mality. Also, a similar number of individuals put into fresh hay
infusion at this temperature survive and multiply, while isolated
individuals perish.

Typhoid bacilli can be rendered abnormally susceptible to hydro-
gen peroxide by a degree of heat which is just sublethal; and the
growth inhibition so caused can be neutralized, as in bacteria natu-
rally susceptible to peroxide, by the accumulation of certain prod-
ucts of bacterial growth (Burnet, 1925a). These peroxide-neutraliz-
ing materials are more rapidly produced the larger the colony of
bacteria and the more vigorous their growth. This subject will re-
ceive further attention in chapter xvi.

Drzewina and Bohn (1926, 1928) report similar results with sper-
matozoa. Although these will be considered at greater length else-
where, it should be stated that both of these observers interpret
the greater protection which they find to be given by the presence

2i8 ANIMAL AGGREGATIONS

of greater numbers, to the conditioning of the water by some sort
of autoprotective exudate. Robertson brings his observations into
Hne with his general theory of the extrusion of an autocatalytic
substance, while the others regard mass protection from high tem-
perature as an excellent example of the effect of an autoprotective
secretion.

PROTECTION FROM ULTRA-VIOLET RADIATION

Another t>pe of experimentation which should yield critical tests
of the efhcacy of group protection and of its mechanism should be
found in treatment with radiant energy applied as ultra-violet
radiation. Hinrichs (1927) has reported that Arbacia sperm are less
affected in concentrated than in dilute suspensions (obviously a
mass protection) ; and Petersen has obtained similar results in this
laboratory in radiating Paramecia (unpublished). Accordingly, tests
(Allee, 1928) were made concerning various aspects of survival
value of groups as compared with isolated planarians, when exposed
to the full spectrum of a quartz mercury- vapor arc, with temperature
controlled during the exposure. The results collected demonstrated
two facts:

In the first place, the presence of products of cytolysis produced
by exposure to ultra-violet radiation are more harmful than bene-
ficial to those worms that have been exposed to the action of ultra-
violet radiation or to those that have not been exposed. Similarly,
w^ater containing products of metabolism or exudates given off by
the animal, either in its usual laboratory culture or when exposed
to ultra-violet radiation, shortens rather than lengthens the survival
of other worms isolated into it.

In the second place, the exposure of a number of worms in a
limited amount of water with limited exposed area gives much
greater protection for the individuals than if they had been exposed
singly or in pairs. Some of the implications of these results deserve
consideration and will be discussed in the order just given.

When the survival of worms in various sorts of worm-conditioned
water is compared with that of similarly treated planarians in
aerated well-water, the worms in the conditioned water are found to

PROTECTION FROM TOXIC REAGENTS 219

die more quickly. The difference in the survival periods is usually
small, and the periods themselves are highly variable for different
lots of worms. Eleven such group comparisons are possible with the
data at hand; and in these, nine cases definitely favor the well-water,
one favors the conditioned water, and the other is recorded as a tie
but vo.th more careful observation it would probably have been
placed definitely in the majority column. The main inquiry was
concerned with the possibility of there being a protective value in
the conditioned water, and there is no doubt of the negative an-
swer given by the experiments. Rather, the converse is indicated
though the evidence is not complete concerning the possibility
that with different water and with smaller proportions of the con-
ditioning matter, such conditioned water may prove advantageous
to the worms placed in it.

With regard to the definite protection furnished when many
worms were exposed to ultra-violet radiation in a limited amount of
water and with a limited surface area, the experimental evidence
indicates that this protection is due to some sort of interference with
the penetration of injurious rays or to some other biophysical effect
of numbers, rather than to the presence of some exudate or exudates
released as a result of the radiation. As has been stated above, water
containing such exudates produced harmful rather than beneficial
results.

In the experiments where the worms were more densely crowded
(60 to I cc. of water), the protection was obviously connected with
the "shading" which Hinrichs mentions in her studies on the radia-
tion of sperm suspensions. In the less dense groups (15 to i cc),
"shading" in the usual sense is less obvious; but exposure at this
density also resulted in definite protection to the group-exposed
worms, which suggests that some other factor or factors may have
been operating. Protection furnished by the presence of so few
worms may be an illustration of the phenomenon called by Drzewina
and Bohn "catalysis by contact." Similar shading would undoubt-
edly have resulted from the exposure of isolated worms in water
conditioned by the pressure of products of cytolysis. The experi-
ment was not tried, since the only information to be gained would

220 ANIMAL AGGREGATIONS

have been the relative value of the protection and the toxicity of
such conditioned water.

It is worth noting that all the members of the group exposed to-
gether were benefited by the fact that they were together. In such
circumstances it might have happened that certain animals at
the surface would take the brunt of the harmful rays. Their pres-
ence might serve to protect the other members, and the group might
thus be of value to the species if not to all the individuals composing
it. In one experiment, where the subsequent history was taken for
all the exposed animals, the interval before the first effects of ex-
posure was observed in the 30 animals exposed as a group ranged
from 27 minutes to 32.8 hours. Their final cytolysis ranged from
8.25 to 106 hours. The same conditions were observed for the 26
worms exposed in pairs in from i minute to 23.5 hours, and from
3.4 to 84 hours, respectively. Thirteen (50 per cent) of those ex-
posed as pairs were visibly affected before the end of the first hour
after exposure, while 6 (20 per cent) of the grouped animals were
similarly affected in the same time. Six of those radiated in pairs
were dead before the first of the group died.

I am not prepared to generalize widely from these experiments
with the relatively large and highly pigmented Planaria doroto-
cephala that mass protection from ultra-violet radiation is always
due solely to some sort of physical interference rather than to the
possibly protective action of some exudate. With small organisms
such as sperm of sea-urchins or with Paramecia, which are more
translucent, the mass protection may be due to the latter factor, as
Hinrichs suggests. Her observations show, however, that even with
such minute organisms, the fact of physical interference is significant.

Single planarians, and to a greater extent massed planarians,
would be more nearly analogous to sperm or protozoan aggregates
than to a suspension in which distribution is fairly equal in three
dimensions and each organism is surrounded by a medium of approx-
imately uniform consistency. In the latter case, it is conceivable
that a closer equilibrium between cells and medium must be main-
tained than with larger forms. This subject is one for experimenta-
tion rather than a priori discussion.

PROTECTION FROM TOXIC REAGENTS 221

In general the experiments on the group resistance of Planaria
to ultra-violet radiation forward our understanding of mass rela-
tions of individuals in two ways: First, the results emphasize the
fact that the phenomena of possible protection of individuals by
chemical exudates, which has been demonstrated for many animals
exposed to different situations, is not universal. Second, they show
that even when the massed animals are known to produce chemical
exudates which are harmful, the massing may still have survival
.value through the changed physical conditions which it produces.

The work reviewed in the present chapter makes clear that group
protection from toxic agents may operate in diverse ways. There
may be group protection solely as a result of the distribution of
toxic material among so many individuals that no one receives a
lethal dose, or the toxic substance may be adsorbed on' the slime
which is often produced in copious quantities under the stimulation
of the abnormal situation. The survival value of the group may be
due to the depressed physiological condition obtaining among its
numbers, which favors increased survival when the animals are
exposed to strongly toxic solutions which kill actively metabolizing
animals more rapidly than those whose rate of metabolism is lower.
The group may act in a purely physical manner by altering the
electrical conditions, or against light and ultra-violet radiations by
a simple shading phenomenon. The evidence to date does not ex-
clude the possibility of protection from active toxic conditions by
the group conditioning the medium through some protective secre-
tion, although this kind of explanation should be adopted only on
positive proof. While it is clear that group protection is a fact, it is
also certain, as might have been expected, that there are many ways
by which groups accomplish this protection, and that more than one
may be acting in the same case.

CHAPTER XIII
RESISTANCE TO HYPOTONIC SEA-WATER

We have just seen that when exposed to a variety of toxic agents,
with the mass of animals in optimal relation with the volume of the
medium and other conditions being favorable, groups of animals-
will survive longer than will equal numbers of single individuals
when each is isolated into the same volume of the same medium to
which the group is exposed. Our experience with colloidal silver
and with several other reagents indicates that such protection is
largely, and at times probably completely, furnished by the fixa-
tion of the toxic substance either directly by the mass, or by shme
and other products given off, so that it is either removed or dis-
tributed among so many individuals that each receives a sublethal
dose. If the animals are closely aggregated, only those on the out-
side receive the full impact of the toxic agent, so that those within
are protected by another type of mechanical action of the mass.
Such an attack on the problem of the protection of the individual
by the mass, while demonstrating the fact of the protection, does
not furnish critical evidence concerning the production of an auto-
protective secretion other than slime, which in these cases presum-
ably owes its protective action to its adsorptive power.

A better opportunity to obtain critical evidence is furnished when
the toxic properties of the solution are due to the absence of easily
measured materials rather than to the presence of some toxic sub-
stance added to the water. Such conditions are fulfilled when marine
animals are placed in hypotonic sea-water. In concluding a note on
the effect of exposing the marine turbellarian Convoluta roscofensis
to hypotonic sea-water, Bohn and Drzewina state (1920):

"Toutes conditions egales d'ailleurs, les individus isoles sont in-
finiment plus sensibles que les individus groupes, comme si le fait
d'etre groupes constituait pour eux une protection. Le contraste
est souvent saisissant. Soient deux verres de montre, dont I'un

RESISTANCE TO HYPOTONIC SEA- WATER 223

contient, dans I'eau deluee a 75 pour 100, quelques individus, et
I'autre plusieurs centaines de Convoluta; les premiers sont cytolyses
en quelques heures, les derniers apres plusiers jours."

Elsewhere they state clearly their belief that the observed pro-
tection is due to the secretion of an autoprotective substance by the
mass in greater quantity in proportion to the available volume of
water than is possible for the isolated animals.

Lapicque (1921), in discussion, made the obvious objection that
the introduction of differing numbers of marine animals into the
same quantity of hypotonic sea-water would have a differential ef-
fect upon the salt content and that the difference in survival might

Fig. 16. — Procerodes wheatlandi, dorsal view. The posterior end forms a muscular
sucker by means of which the animal attaches to the under side of rocks near the low
tide line. These animals are usually found in considerable numbers on a given stone,
if at all.

be due to this direct action of the greater mass of animals. He
rightly thought there should be a quantitative determination of the
salt content at the end of the experiment.

Drzewina and Bohn (192 1&, 1928) replied by calling attention to
the small size of the Convoluta worms, which are about 3 mm. long.
They calculate that the amount of salt such worms would carry
would not affect sensibly the salt concentration of the solution, but
have not refuted this criticism by experiment.

At Woods Hole I have had an opportunity to test this situation
with two main objectives: First, the question of fact involved: Is
there a greater protection furnished by the presence of large num-
bers of animals in hypotonic sea-water when compared with fewer
similar animals in the same volume of water? And, second: What
is the mechanism of the protective action of the group, if it be found?

For these studies a turbellarian, Procerodes wheatlandi (Girard),
Figure 16, was selected as representing a form taxonomically some-

224 ANIMAL AGGREGATIONS

what related to Convoluta. These animals are normally subjected
to hypotonic sea-water when a heavy rain occurs at low tide.

These small worms reach a length of about 5 mm. and are about
I mm. in width. In nature they are found in abundance on the lower
sides of stones in small tide pools, near or below low tide line. They
are not abundant in deeper water. Usually they were taken from
the protected sides of stones that were firmly located on a sandy
substratum; evidently they cannot stand the full sweep of the waves.
They were usually present on a given stone in considerable numbers,
if at all. As the water went stale in the laboratory, if there were
numbers of worms present they would collect on the surface film in
shaded areas in dense aggregations. Here, as in the field, they did
not occupy all the apparently optimal space.

Appropriately safeguarded experiments showed that these worms
will survive exposure to equal amounts of tap-water the better (a)
if they are present in numbers; {b) if isolated into tap-water in
which other living Procerodes have previously been exposed; and
especially (c) if exposed in tap-water in which other Procerodes have
died and disintegrated in whole or in part, even when such a condi-
tioned medium is boiled or filtered. As is to be expected, these
worms live longer if the salinity of the tap-water is increased by as
much as 0.05 per cent above a minimum of that value. With well-
washed worms it is possible to demonstrate protection when the
tap-water contains exudates from living or dead Procerodes in which
the salt concentration, as measured by the amount of chlorides pres-
ent, is not the determining factor. The method used consists in titrat-
ing with N/ioo silver nitrate, using a i per cent solution of potas-
sium chromate as an indicator.

The first experiments (Allee, 1928) demonstrated that the pro-
tection was due neither to sea-water contamination nor to the leach-
ing-out of electrolytes in the proportions in which they exist in sea-
water. They did not exclude the possibility that the protection may
have been given by the leaching-out of electrolytes in some other
proportion from that found in sea-water. Such an explanation of
the observed protection is the most simple and obvious one to be
advanced. The possibility of its operation was tested at the first
opportunity.

RESISTANCE TO HYPOTONIC SEA- WATER 225

For this purpose experiments were repeated, using the technique
suggested but with the additional precaution of checking the electri-
cal resistance of the different solutions to which the worms were
exposed, at the beginning and at intervals during the progress of
the survival tests (AUee, 1929).

This method of determining the amount of electrolytes present
measures all electrolytes, instead of depending on the well-known
relation between the amount of chlorine and the total salt concen-
tration in sea-water. The experiments to be reported fall into two
main groups: those in which the water had an initial resistance of
about 1,900 to 2,500 ohms, and those in which the initial resistance
was 5000 to 6,500 ohms. The experiments reported in 1928, as nearly
as can be told from chlorine titrations and survival values, belong
to the former level. Preliminary tests showed that with hypotonic
sea-water at the greater dilutions, and under the conditions of these
experiments, an initial difference of about 1,000 ohms is needed to
affect significantly the survival time of these worms.

All the comparisons made concerning the effect of homotypically
conditioned medium upon the survival of Procerodes are summarized
in Table XXIV, which gives the results of nine separate sets of ex-
periments, each of which consisted of from three to six independent
sets of tests. In the results down to those of Experiment 71 the
initial resistance of the conditioned water was regulated by the
addition of distilled or of tap-water to the conditioned medium. In
the later experiments the resistivity was controlled by dialysis.

Direct comparisons indicate that dialyzed Procerodes culture me-
dium is less effective than is similar dialyzed culture medium which
also contains the water extract of dead Procerodes worms. These
results are in keeping with those obtained in 1928 with the methods
then in use, and in all probability represent the true state of affairs.'

In all, 640 worms were used, which were divided equally between
Procerodes-conditioned water and fresh water to which enough dilute

' Castle (1928) in his work on the life-history of Planaria velaia similarly found that
in placing 50 small fragments of the flatworm in 10 cc. of distilled water some will sur-
vive. If the same mass of the same number of small fragments is placed in 200 cc. of
water, usually all will die. With small volumes, more than 75 per cent will survive. He
attributes this protective action of the mass primarily to the rapid conditioning of the
medium by the products from disintegrating pieces.

226

ANIMAL AGGREGATIONS

sea-water had been added to bring it to an equal initial resistivity.
These worms showed a mean survival of 18.38 hours longer in the
worm-conditioned water than in the controls. Each group of ex-
periments considered singly gave positive results, although some
individual pairs did not. Despite the variability in procedure used
in the different experiments, the combined results show a statistical
significance of 0.03 when considered as nine paired experiments.
These results are graphically summarized in Figure 17, which
shows two sets of histograms. That marked A gives in black the

TABLE XXIV

Showing the Effect of Procerodes-coKDiTiotiED Water upon Survival
OF Procerodes Isolated into Extremely Hypotonic Sea-Water

Experiment

Resistance in Ohms

Conditioned Water

Start

Late

Dilute Sea-Water

Start

Late

Number
Tested

Survival in Hours

Con-
ditioned
Water

Dilute
Sea-
Water

Differ-
ence

41-43
43-45
46-48

49-51
52-54
71. . .
71a. .
73- ■ •
73a. .

1 ,916
2, 100

2,350
2,400
2,000
5,150
5,150
6,200
6, 200

1,358
1,438
1,780

1,930
1,320
3,550
3,550
4,700
4,700

1,897
2,100
2,350
2,400
2,000
5,150
4,250
6,200
5,100

1,445
1,480
1,760
1,888

1,735
4,000
3,400

4,500
4,000

56
60
60
56
60
58
58
118
114

88.16

59 97

58.1

20.45

10.05

37.18

37.18

12 .06

11.30

13

59-72

22.94

14 05

9.29

2.53
25.21

23 95
4.31
342

percentage of worms surviving at the indicated hourly intervals
when placed in tap-water with an initial resistivity of from 5,000
to 6,550 ohms. Just above, in the shaded blocks, is given the added
percentage of survival resulting from the presence of sufficient sea
salt to decrease the initial resistance to from 1,890 to 2,400 ohms.
The upper clear blocks give the added survival resulting from the
presence of water extract of Procerodes worms with the same initial
resistance as the hypotonic sea-water.

Graph B shows similarly the survival in tap-water with an initial
resistivity of from 5,000 to 6,550 ohms, in hypotonic sea-water made
by adding dilute sea-water to pond or distilled water, bringing it to
a resistivity of from 5,150 to 6,550 ohms, and finally, at the top.

RESISTANCE TO HYPOTONIC SEA-WATER

227

CM rt

2 28 ANIMAL AGGREGATIONS

the survival in dialyzed Procerodes-conditioned water having the
same initial resistivity.

HETEROTYPICALLY CONDITIONED WATER

Having demonstrated in two successive years that Procerodes
may be protected, at least to some extent, from the harmful effect
of extremely hypotonic sea-water by being isolated in water condi-
tioned by exudates or watery extracts of other Procerodes, and that
whatever the protective device may be, it is not a direct result of an
increase in total electrolytes present, it becomes of importance to
inquire carefully concerning the specificity of this protection.
Evidence will be presented here concerning the species specificity
only, since to date no work has been done to attempt the analysis of
possible functional specificity. In the course of this study the effects
of the following kinds of media were tested against very dilute sea-
water with the same resistivity: Paramecimn culture medium, water
extract of Planaria maculata (a fresh-water turbellarian) and of P.
maculata culture media, and water extracts of a marine amphipod.
The effect of each of these will be discussed in the order given.

Paramecium culture medium. — The solution used was the clear
brownish surface liquid from a hay infusion which contained a
vigorously growing and almost pure culture of Paramecium. The
general situation can be presented most clearly by describing one
experiment in some detail and by adding summaries of all. In
Experiment 74, hay-infusion liquid, such as has just been described,
containing many Paramecia was boiled and dialyzed against run-
ning tap-water until it showed a resistance of 6,450 ohms at 20° C.
Tap-water was brought to the same resistivity by adding 0.25 per
cent sea-water. Another lot of tap-water was brought to 5,450 ohms
resistance.

One hundred and fifty Procerodes were isolated, 50 into each of the
modified tap-waters and 50 into the dialyzed Paramecium Q\AtVir&
medium. At the end of 6 hours the water in the more dilute tap series
had changed from 6,450 to 3,350 ohms and that in the culture me-
dium had fallen from the same initial level to 3,550 ohms. Both
were changed to 6,550 ohms in their respective media. After an-

RESISTANCE TO HYPOTONIC SEA-WATER

22g

other 6 hours, this had fallen to 4,950 ohms for the modified tap-
water and to 5,750 ohms for the culture media. Again the liquids
were changed to appropriate solutions, each at 7,400 ohms at 19° C.
After 24 hours the water from worms just beginning disintegra-
tion in the dilute sea-water showed 4,850 ohms resistance, while that
from worms in the same condition in the culture medium showed

TABLE XXV

Showing the Survival Time, in Hours, of Procerodes Isolated into i Ml.

Each of Paramecium Hay Infusion and of Dilute Sea-Water

OF THE Same or Greater Resistivity

Medium

Worm No.

Survival Time in Hours

Resistance

Mean

Maximum

Minimum

Range in
Ohms

Control I

Culture

Control II

I- 9

11-19

21-29

31-40

41-50

51-60

61-70

71-80

81-90

Xi-io

Xii-20

X21-31

X31-40

X41-50

X51-60

49

49

49

18
21

15

12

27
22

17
28

19
16

25
18
II

25

20

15
25
19

05

85

75
70
95
50
00

30
20

85
93
50
40
20
50
20
85
30

36.0

38.5
34
34

58. 5
58.0
40.0
40.0
64.0
34

49-3
41 .0
17.0

58.5
41 .0
40.0

58.5
64.0

3-5
6.0

3-5

3-5

II-5

II .0

8.5
8.5
3-5
7.0
7.0

8.5
4.0

8.5
8.1

3-5
6.0

35

Control I

Culture

Control II

Control I

Culture

Control II

Control I

Culture

Control II

Control I

Culture

Control II

, ^ { Control I

g % { Culture

'"^ [Control II...

7400-3350
7400-3550
6150-3550

4,950 ohms. Again the surviving worms were changed to water
having a resistance of 7,200 ohms for each solution. Ten hours
later both showed a resistance of 4,475 ohms.

The approximate survival time was obtained for 49 worms ex-
posed singly to i ml. each of the two solutions whose history has
just been given. It will be noted that in no case was the resistivity
less in the Paramecium culture medium than in the accompanying
control solution. The survival times are summarized in Table
XXV, in which Control I is used to mean the modified tap-water

230 ANIMAL AGGREGATIONS

with the same resistivity as the culture medium, while Control II
designates the less-dilute salt solution.

The summary shows that the worms in the culture medium lived,
on the average, 25.85 hours, while those in the very dilute sea- water
with the same initial resistivity lived only 15.2 hours. The differ-
ence of 10.65 hours, when examined by Student's method, shows a
statistical probabihty of 0.0714, when considered as 5 pairs of tests
as listed above, but when considered as 49 individual pairs, the
probability becomes 0.00002, which is clearly significant.

As convincing as is this experience that Paramecium culture me-
dium with equal or less total electrolytes than accompanying dilute
sea- water has a survival value for marine Proccrodes isolated into it,
the same series of experiments furnished still further evidence that
such is indeed the case. In the series labeled "Control II" the initial
resistance was 1,000 ohms less, and therefore a less-dilute solution
of sea-water than Control I. This water was renewed each time the
others were changed, and was kept at least 1,000 ohms more con-
centrated at these times. It was never found to be less concen-
trated than was the culture medium, and only once to have the same
concentration. The worms in this less-dilute sea-water lived longer
than did those in Control I by an average time of 4.1 hours, a dif-
ference which is not statistically significant.

The worms in the culture medium showed a mean survival time of
6.55 hours greater than did those in the more concentrated Control
II. When individual pairs are considered, this has a probability of
0.02, and hence is statistically significant. Further experiments
with Paramecium culture medium and with water extracts and
culture medium of Planaria maculata, and water extracts of marine
amphipods found in close association with Procerodes in nature, give
essentially the same results and show that heterotypically condi-
tioned fresh water has protective value for Procerodes isolated into
it as compared with the survival when isolated into hypotonic sea-
water with the same electrical resistivity.

While this general relation holds, there was not necessarily the
same degree of protection from each type of solution. Whether this
is due to the conditions under which the experiments were run, or

RESISTANCE TO HYPOTONIC SEA-WATER 231

whether it is an inherent property of the different heterotypically
conditioned solutions, is not yet apparent. At any'rate, the findings
to date are summarized in Figure 18, which gives the experience
from comparable experiments run simultaneously. The vertical axis
shows the percentage of worms surviving at any given time; the
horizontal axis gives the time elapsing since the beginning of the
experiment. One finds that there is a marked drop in the percentage
of survivals within the first 48 hours and that thereafter the curves
tend to flatten, reaching extinction in from 16 to 20 days.

The mean survival for the whole group is not plotted on this
chart, but practically coincides with the graph for Proccrodes-condi-
tioned w^ater, except that it continues just above the base line until
after the 15-day mark. The graph for hypotonic sea-water having
the same initial resistivity runs below the lowest graph on the chart
at all points, except that at the 2- and 3-day periods it is very slightly
above this lowest graph.

These results show clearly the lack of species specificity in the
protection of Procerodes against the lethal effect of hypotonic sea-
water; and remind one, in this respect at least, of the results ob-
tained by Allee and Schuett (1927) that protection from such toxic
substances as colloidal silver also lacks species specificity. The pro-
tection against colloidal silver appears in a large part to lack func-
tion specificity. Whether or not the present protective mechanism
also has other and general functions is not clear at the present time,
although on general grounds one would be inclined to think that
such would be the case.

POSSIBLE FACTORS CONTRIBUTING TOWARD THE OBSERVED
SURVIVAL VALUE OF CONDITIONED SOLUTIONS

The facts recorded are plain. The lethal effect of the fresh water
is clearly less for solutions that have been conditioned by the pres-
ence of living organisms, when compared with hypotonic sea-water
having equal initial resistivity. The exact source of this condition-
ing is not yet clear. In some of the Procerodes-conditioned media
the survival value is due to exudates from the living worms; in
others, where water extracts were prepared, the survival value may

232

ANIMAL AGGREGATIONS

1

1

.5

a

10

$5

1

t

>

<s

(M

5;

^

.a

f^

^

^

^

1

^,

•5<

■^

^

1

1

1

K^

^-

RESISTANCE TO HYPOTONIC SEA-WATER 233

have been in part due to the action of bacteria, although this is
practically ruled out in those experiments in which the solution was
given a fair pasteurizing treatment in the formation of the extract,
which was later boiled vigorously. The bacterial growth in such
solutions must have been slight, at least in the early history of the
isolations.

With the Paramecium culture medium the role played by the
different elements entering into the conditioning of this medium
could be told only by experimentation. The Paramecia were grow-
ing in a hay infusion which contained the water extracts of the hay
as well as the products of bacteria and of Protozoa other than the
Paramecia. All these must be considered as possible conditioning
factors, since the work above has clearly shown that the protection
is furnished by heterotypic as well as by homotypic conditioning.

It has been possible to begin experimental analysis of the factors
leading to increased survival in these biologically conditioned solu-
tions. To date these indicate that the results are not primarily due
to the depressing action of the conditioned medium, in so far as
these are mimicked by dilute alcohol. They are not due to differ-
ences in pH. Gelatine suspensions do not confer similar benefits.
The colloids present do not mask the amount of electrolytes in
solution.

Preliminary tests have been run to determine whether the protec-
tive substance would be adsorbed on animal charcoal. These gave
results of sufficient interest to deserve being presented in outline.
Animal charcoal was added to the dialyzed culture medium. The
whole was stirred, boiled, filtered twice with suction and once with-
out, and, after redialysis to 6,050 ohms, a standard survival experi-
ment was set up with hypotonic sea-water controls at 6,050 ohms
and 5,050 ohms. The resistance decreased in all solutions, as usual,
during the progress of the experiment. The solutions were changed
twice in the first 20 hours. As usual after standing, the culture water
had less electrolytes than did either of the accompanying solutions.

Under these conditions the 50 Procerodes isolated into the treated
culture medium lived a mean time of 11.03 hours, while those in
hypotonic sea-water with the same initial resistivity lived 9.21 hours.

234 ANIMAL AGGREGATIONS

The difference, 2.72 hours, has a statistical probabiUty of 0.0564,
which is too great for much significance to be attached to the ob-
served difference. These values are to be compared wi^i the results
shown in Table XXV, which were obtained immediately preceding
the present experiment. That test showed a difference in survival
of 10.65 hours, with a probability of 0.00002.

When comparison is made between the survival in the charcoal-
treated medium and the less-dilute sea-water which had an initial
resistivity of 5,050 ohms, the difference in survival time is 2.53 hours,
with a statistical probabihty of 0.072. These values are to be com-
pared with a survival difference of 6.55 hours and a probability of
0.02 in the preceding experiment with the same type of culture
water similarly dialyzed but not treated with charcoal.

These results indicate that the charcoal did remove a part of the
actively protective substance, and suggests the need of further work
at this point. Perhaps more extended adsorption would have a still
greater effect. Another possibility must be mentioned, although
here, too, the evidence available at the present time does not warrant
the drawing of definite conclusions. I refer to the possibility of the
differential leaching of materials in raw and in biologically condi-
tioned water. Early in the analysis of the results here summarized,
indications appeared that worms isolated into water conditioned by
living organisms tended to lose electrolytes less rapidly than did
their controls isolated into hypotonic sea-water with the same initial
resistance. Accordingly, the resistance records were examined for
comparable experiments. The different exposures were for varying
lengths of time, and hence show a marked variation in the change
in resistance from that given at the beginning of the experiment.
The initial resistance of the observed solutions stood at different
levels in the various experiments, but always at the same level in
both the experiment and the control for any given experiment. The
mean difference in the whole series of experiments shows that there
was 4.8 per cent less change in the resistivity of the biologically
conditioned media than in their accompanying hypotonic sea-water
controls. Despite the many known elements of variation, this has
a statistical probability of 0.00068.

RESISTANCE TO HYPOTONIC SEA- WATER 235

One must be cautious in interpreting this lessened change in elec-
trical resistivity in the animal-conditioned media as being the
fundamental cause of the longer survival of the worms isolated into
such media, because in general we have seen that much greater initial
differences in resistivity have relatively little effect upon survival,
and also because gelatine suspensions showed a smaller resistivity
change than did the controls, and yet had no protective action.

There is a possibility that if the worms lose contained electrolytes
less rapidly into the surrounding medium in the presence of materials
from other organisms, organic materials necessary for the continued
well-being of the worms, which are not measured by present meth-
ods, may also escape less rapidly, as Robertson assumes to happen in
the case of allelocatalysis in Protozoa ; and that the observed protec-
tion may be due to the retention within the organism of a certain
substance or substances necessary for continued existence, rather
than to the presence of a definite protective material in the treated
solutions.

We are here dealing with some chemical difference in the medium
unanalyzed as yet, rather than with some sort of physical protection
of which different types have been described (Allee, 1920, 1926,
1927; Drzewina and Bohn, 1928). The material behaves in some
aspects like that which Banta and Brown (1929) find affecting the
percentage of males in crowded cladoceran cultures, which they
attribute to the accumulation of excretory products; in others, like
that which Petersen reports as conditioning Paramecium cultures so
as to allow higher division rates, especially since the latter has been
shown not to be species-specific. We are probably dealing here with
the sort of protection which Drzewina and Bohn originally postu-
lated in their communication of 1920, quoted at the beginning of
this chapter.

CHAPTER XIV

RELATION BETWEEN DENSITY OF POPULATION
AND INSECT SURVIVAL

We have already seen that the insect Drosophila reproduces less
rapidly the greater the population density (chap, vii), and that,
on the other hand, the confused flour beetle {Tribolium) during
initial stages in cultures of flour reproduces more rapidly if the pop-
ulation for given environments be larger than the minimum (chap,
x). In later chapters we shall find that crowding exerts certain
physiological effects upon aggregated insects which are expressed in
morphological changes. These facts concerning mass relations be-
tween non-social insects are the more interesting in view of the high
state of social organization to be found at times in this group.
Unless we can find a strong substratum of generalized co-operative
survival values among the subsocial insects, we shall have difliculty
in demonstrating a connection between the phenomena associated
with relatively slightly integrated animal aggregations and those
connected with the more closely knit social communities.

For the present we desire to consider the relations between num-
bers of non-social insects and their survival in the face of unfavorable
conditions. Three lines of evidence exist upon this point: (i) the
data from laboratory and field experiments concerning the relation
between numbers of insects present and their survival in the pres-
ence of toxic reagents; (2) studies on the relation between population
density and longevity in Drosophila; and (3) the more usual type of
natural-history observations showing that the survival value of
crowding in the presence of predators holds for insects as well as for
the larger animals, for whom it has been more frequently reported.

MASS PROTECTION FOR GRASSHOPPERS

Two workers have independently reported mass-survival values
for insects exposed to poisonous substances. Deere, for the majority

236

DENSITY OF POPULATION AND INSECT SURVIVAL 237

of his experiments in this laboratory, used different species of grass-
hoppers, working upon the meadow grasshopper, Xiphidiuni fascia-
tus, more than on any other single species. Experiments were run
upon the effects of ether, carbon tetrachloride, ethylene chloride,
ethyl alcohol, and hydrocyanic acid. Two types of experiments
were carried out. In one set the insect containers had a direct
volumetric relation to the number of individuals per bottle, so that
if 5 insects were used, the container had five times the volume of
that containing but i insect; and if 10 were used, the container had
ten times the volume of that employed with only i animal. In
such experiments the volume percentage of the toxic gas was kept
constant.

In the other type of experiment, all tests were run in the same
size bottle regardless of the numbers exposed. In these, too, the
volume percentage of the toxic agents was kept constant. The
amount of gas used was calculated to produce narcosis in most cases,
without death necessarily following.

In all experiments the group survived longer; but, as might be
expected, there is relatively greater group resistance when the vol-
umes are identical than when they are proportional to the number of
insects. Similarly, single individuals have been found to have a
somewhat longer survival when placed in a smaller bottle rather
than a large one, when each has the same volume percentage of toxic
gas. The results of one of the constant volume experiments are given
in Table XXVI.

The effect of exposure of these animals may be summarized as
follows: In the group of 30, 82 per cent survived for 24 hours;
in the group of 20, 90 per cent survived; in the group of 10, 90 per
cent survived; while of the 10 singles only 60 per cent survived
for the 24 hours of the experiment. In control lots under similar
conditions, but without the ethylene chloride gas, 97 per cent of
the group of 30 survived, and, all of each of the other lots lived for
the period of the test. Apparently the group of 30 suffered some-
what from the ill effects of crowding, but not enough fully to mask
the survival value of the group when exposed to this amount of
narcotic gas.

238

ANIMAL AGGREGATIONS

The Xiphidium have cannibahstic tendencies, which caused sev-
eral experiments to be discarded. When Melanoplus were used, they
showed greater initial activity but otherwise gave similar results.

Ethyl alcohol differed from the other vapors used in that the
effect on the animals was much more gradual, extended over a
longer period of time, and stimulated them to greater activity.

TABLE XXVI

Showing the Relative Survival of Groups and Isolated Individuals of the

Grasshopper Xiphidium, Exposed to Equal Volumes of 27.5 Volume per Cent

OF Ethylene Gas, August 17, 1928; Barometer 746, Temperature 24° C.

All were placed in 240 cc. bottles supplied with 66 cc. of the gas.

Minutes until
First Effect

Minutes*
UNTIL Quiet

Condition after 12 Hours

Number of Animals

Apparently
Normal

Affected

Apparently
Dead

20

25 (few)

35 (several)

25 (few)

25

24

13

27

29

12

7
18

24

25

50 (3)
45 (few)
45 (few)
37
44
43

24

17

9

I

3

I

2

20

2

I

I

I

I
I

49
63

58

I

I
I

41
60

I

I

* Removed at this time to ordinary air.

The period of recovery was also longer with this reagent. The dif-
ference in recovery between the group and the single individuals
was also greater with alcohol than in the case of the other vapors
used.

The outstanding effect noted in the case of insects exposed to
HCN was the lack of stimulation when the insects were first ex-
posed. Xiphidium in particular remained motionless from the time
the gas reached them. Muscular activity could be induced, if at all,
only by most vigorous shakings. This would loosen their attach-
ment from the bottle or from whatever object to which they might
be cHnging. Those that were fully narcotized would fall to the

DENSITY OF POPULATION AND INSECT SURVIVAL 239

bottom without movement; others might show shght movements
of the feet, particularly if these came into contact with a solid ob-
ject, even when there was no other visible movement. The recovery-
period from HCN was also shorter than from exposure to the other
gases used; but otherwise the results were similar, particularly in
that the group showed greater survival value than isolated indi-
viduals.

Such results from different numbers of insects exposed to the
same volume of the same concentration of toxic gases are according
to expectation from the experience previously recorded with aquatic
organisms exposed to various toxic conditions. The apparent ex-
planation is that, when the toxic gas is distributed among a number
of organisms, no one individual is as likely to receive a dose sufficient
to cause death as if it were exposed singly to the same quantity of
the same gas. The explanation of the findings when groups and
singles are exposed to volumes in proportion to the numbers present,
indicating that there is group protection in proportion to the num-
bers present even in the face of similar conditions for each indi-
vidual, is not clear, and further work is needed at this point.

It seems probable that under such conditions of crowding, even
with equal space and an equal amount of the toxic gas present per
individual, the group may cause a greater concentration of carbon
dioxide in its neighborhood, or by some other effect may very well
cause these animals to have a lowered rate of metabolism. Under
these conditions, and with the concentration of poisonous substance
at the right strength, one would expect the members of the depressed
group to survive longer than would the more active single individ-
uals. This suggestion is in line with the susceptibility work of Child
(191 5) and the resistance of Daphnia to various chemical reagents,
as developed by Fowler and discussed in a previous chapter.

Experimental results similar to those of Deere have been obtained
by Bliss (personal communication) under natural conditions. In his
field studies Bliss found that the camphor scale insect {Pseudoanidia)
has a greater natural death-rate the greater the density of the popu-
lation on a given twig. When exposed to adverse conditions, such as
are represented by spraying with an oil spray, the opposite condi-

240 ANIMAL AGGREGATIONS

tions prevail, and there is decidedly greater survival the more dense
the population. In making these observations, due care was taken to
consider only those insects located on trees in comparable situations
and on comparable parts of the trees. Bliss concludes that death
from these sprays is not due to suffocation, as has been supposed, but
is due to the taking-up of some poisonous material by the insects;
and that when they are present in greater numbers, the toxic mate-
rial is divided between more insects, no one of which is so likely to
receive a lethal dose as if the population were less dense.

POPULATION DENSITY AND LONGEVITY IN DROSOPHILA

The relation between the density of population of laboratory cul-
tures of wild-t>pe Drosophila and their life-duration were less ex-
pected. Pearl and Parker showed in 1922 that statistical analyses
of data accumulated in other studies indicated greatest longevity
from bottles originally stocked with from 35 to 45 of the wild
stock per bottle. In 1923 the same workers reported the results of
an experiment made to test out this question of an optimal popula-
tion. The data on the length of life of 12,382 individuals showed
that the optimal density of population, when longevity is taken as
the criterion, is not found in the minimal populations but lies in the
region of initial densities of from 35 to 55 per i-ounce vial, and the
increase in length of life from the lowest density is at a much more
rapid rate than is the decline of duration of life after the optimal
density is passed.

A more complete analysis of the problem is given by Pearl, Miner,
and Parker (1927). In this experimental work the flies were kept
in I -ounce vials stoppered with cotton plugs and held at 25° C.
The bottles were examined daily, the dead flies were removed and
their age recorded, and the living flies were at the same time trans-
ferred to fresh bottles of newly prepared food. In the first experi-
ments banana agar was used as food, but similar results were ob-
tained with a synthetic medium.

The extent of the experimental data may be visualized from the
following statement of the numbers used in one experiment. One
hundred and fifty vials were started with an initial population of

DENSITY OF POPULATION AND INSECT SURVIVAL 241

I pair each; similarly 80 vials contained originally 2 such pairs;
50 vials contained 3; and 40 vials contained 4 pairs each. Thirty
vials were started with 5, and 30 more with 6 pairs each, and 20
vials contained an initial population of 15 flies, or 7.5 pairs. Ten
vials were started with each of the following populations: 20, 25,
35. 45^ 55» 65, 75, 85, 95, 105, 125, 150, and 200. In another experi-
ment the initial densities per i -ounce vial were: 5, 25, 50, 75, 100,
200, 300, 400, and 500.

The results from the first experiment are given graphically in
Fig. 19, which shows the mean duration of life of wild-type Drosophi-

to
35

1

^r-

^

^\

\

p.

lis

^">

■C

a /O

s

1

^

^<i

c

^

"<.

^

^

—

10 20 30 ^0 SO 60 70 80 90 100 110 120 130 I'M /SO 160 170 ISO 190 200 210
/nitial Density

Fig. 19. — Showing the mean duration of life in Drosophila with relation to the initial
density in i-ounce bottles. (From Pearl, Miner, and Parker.)

la at different densities of population. The second experiment yield-
ed similar results and demonstrated that with the higher densities
there is relatively slight effect upon longevity of marked increases in
original density.

Pearl, Miner, and Parker sum up their experience with this sort
of experiment as follows (1927):

"The rate of mortality of Drosophila is profoundly influenced by
the density of the population, that is, by the number of flies to-
gether occupying a limited universe in which volume of air, volume
of food, and area of food surface are constant.

"There is an optimal density of population for Drosophila under
the conditions of these experiments. This optimal density falls

242 ANIMAL AGGREGATIONS

somewhere in the region of 35-55 flies per one ounce bottle contain-
ing 8 cc. of food substrate. At densities of population above and
below the optimum, the specific death rates are higher at all ages
than they are at the optimum."

Such results as these make one fairly bristle with questions, of
which only a few can as yet be answered. Pearl and his associates
have reported on investigations concerning the effect of changes in
density during the progress of the experiments. Since an initial
population of about 35 flies per xdal was found to lie near the opti-
mum under the conditions used, a number of vials were set up
with this density and apparently' were followed according to the
practice of the preceding experiments until the sixteenth day, at
which time the survivors were etherized and the tips of the wings
of half of them were clipped, after which they were returned to
their proper bottles. Appropriate tests showed that the wing-clip-
ping as practiced did not significantly affect life-duration.

Vials with an initial population of 200 were treated in like manner
up to the sixteenth day, when they were etherized and counted.
In part of such bottles the population was then brought back to its
initial density of 200 by adding marked flies which up to that time
had lived under the optimal conditions furnished by an initial popu-
lation pressure of 35 flies per ounce vial. Others were brought to
the original population density of 200 per ounce bottle by adding
flies surviving from stocks started from that density. The results
of such treatment are shown graphically in Figure 20, which gives
survival curves starting with 100 flies at 16 days of age, and the
effect of previous history of crowding upon the duration of life
after that age.

In commenting upon these results, the experimenters state that
here, as before, flies subject to an initial density of 35 per ounce
bottle survive, on the average, about double the time those live
that are subjected to an initial density of 200 for the same size vial
and with the same food supply. They continue:

"When flies which have lived the first 15 days of their fives under
the conditions implied by an initial density of 35 are on the i6th
day of their age submitted to a density of 200, and live out the

DENSITY OF POPULATION AND INSECT SURVIVAL 243

remainder of their lives under the conditions implied thereby, their
average duration of life is reduced in these experiments from the 34
or 35 days which it would have been had they stayed in the bottles
with an initial density 35, to 22.83+0.19. This result shows that

K Density 35 Throughout
X i^/ings Not Clipped

''rs Density 35 Throughout
\Aj'Vings Clipped

Age in Days

Fig. 20. — Showing survival curves starting with 1,000 individuals of Drosophila
at 16 days of age which had been subjected to different densities of population up to
that time. (From Pearl, Miner, and Parker, 1927.)

crowding produces a heavy increase in mortality even though it
occurs as late as 16 days of age.

"Flies which have lived the first 15 days of their lives under the
conditions implied by an initial density of 200, and then at the age
of 16 days are again subjected to a density of 200, have a significant-

244 ANIMAL AGGREGATIONS

ly shorter duration of life than do their companions in the same
bottles who spent the first 1 5 days of their lives in bottles of initial
density 35. The difference is 22.83 + 0.19 minus 19. 71 + 0. 17, which
is 3.12 + 0.25 days. This difference is more than 12 times the prob-
able error. It may be taken as probable to the point of practical
certainty that excessive crowding in early life deleteriously affects
the survivors of 16 days of age so that they are significantly less
resistant to the effects of heavy crowding again at that time than
are flies which lived at optimal densities in early life."

These Drosophila data indicate that the harmful effect of supra-
optimal densities of population is most marked near the beginning
of adult life. This original sensitivity decreases with age, so that
older flies are less affected either by the original or the average den-
sity of population. Even so, excessive crowding has been shown to
increase the death-rate at later ages of these flies and to produce this
effect almost immediately upon the increase in density. "Further-
more," as Pearl and his collaborators state, *'it appears that the
amount of shortening of life produced by crowding at any age is
influenced by the previous history of the flies relative to density of
the population. This suggests that there is a deleterious effect of
supra-optimal crowding in early life even upon those flies that do
not immediately die as a result of it, and that this effect endures for
at least the first 15 days of life." Unfortunately for us. Pearl's
experiments have not been concentrated at the part of the survival
curve which would show whether suboptimal crowding will have the
same effects that supra-optimal densities have been demonstrated
to have.

These experiments on Drosophila, like those of Chapman on the
confused flour beetle, are distinct from the great majority of crowd-
ing experiments in that they are carried on in a non-liquid environ-
ment, whereby the temptation to postulate the production of some
X-substance responsible for the phenomena observed is much re-
duced. Pearl suggests that the biological explanation for his ob-
servations lies in an entirely different direction, but the matter has
not been discussed publicly, even in general terms, by members of
his laboratory. We know that in the case of effect of crowding on

DENSITY OF POPULATION AND INSECT SURVIVAL 245

egg production in fowls, Pearl and his associates regarded the under-
lying stimulus to lie in the psychological field, and it is possible to
regard the effects recorded here as expressions of some sort of stimu-
lus physiology, such as Goetsch has recorded in his experiments
upon the effect of crowding in retarding growth of rapidly swimming
animals such as young tadpoles.

There are, however, other possibilities which must be considered.
In vials stoppered with cotton there may be an increased CO2 ten-
sion developed which may in part account for the observed effects.
Then, too, the reduction in length of life at suboptimal densities
may be an expression of the inability of the small populations pres-
ent to gain control of "wild" organisms other than the food yeasts
present in the cultures, while the supra-optimal density effects may
be related at least in part to food shortage and to excess of excretion
products. Obviously, the resolution of this situation into causal
factors is not easy; but it is equally obvious that we have here one
of the most suggestive of the phenomena yet presented, indicating
the wide application and fundamental importance of the physiologi-
cal effects of animal aggregations upon the aggregants.

MASS PROTECTION IN NATURE

A different aspect of the effects of numbers on insect survival is
illustrated by the observations of Haviland (1926) upon the protec-
tive value of feeding aggregations of certain chrysomelid beetle
larvae {Coelomera cayennensis) . Haviland says:

"The larvae are thickly hairy and the last segment of the body is
expanded into a strongly chitinized, shovel-shaped flange. The lar-
vae feed in a compact mass on the upper surface of Cecropia leaves.
Their heads are all directed inwards while the caudal expansions
thresh to and fro on the outside of the circle. The chief enemy of
these larvae is a carnivorous Pentatomid bug {Phyllochinis) which
loiters on the outskirts of the throng awaiting the opportunity to
impale a larva with his proboscis and drag it from its fellows. As
long as the circle of shovels is unbroken, the bug stands httle chance,
for his stylets cannot penetrate their polished armor and he cannot
reach the soft bodies beyond. But as the larvae feed they move out-

246 ANIMAL AGGREGATIONS

wards from the original center, the circle becomes wider, and sooner
or later the enemy slips in between the defenses and secures a
victim. According to my observations, however, the circle is not
broken naturally until the larvae are full grown, in which case the
loss of one or two individuals does not signify, for before the bug
has digested his meal, the rest of the brood enter the pupal stage
and so escape."

These observations bring us back again to the familiar biological
level of the struggle for existence between predators and their prey.
They are given here because it frequently appears that such masses
of feeding larvae are more at the mercy of their enemies than if
they fed alone, unless they have developed means of detecting the
approach of enemies and transmitting the information through the
mass or unless they have developed effective means of active group
defense like that of the soldier caste of termites and ants.

CHAPTER XV
COMMUNAL ACTIVITY OF BACTERIA

Many of the relationships which we have found to hold for animals
can also be studied with bacteria. The large amount of work done
upon these micro-organisms and the excellent technique developed
both in culturing and in making isolations cause the results obtained
to be the more significant for our studies. In one way bacteria as
presented here and spermatozoa as discussed in the following chap-
ter may serve as test cases for us. The situation may be stated thus:
We have been presenting certain evidence concerning the role of
numbers of individuals present in relation to the physiology of each
individual. The studies so far have dealt with a wide range of ani-
mals. If we turn to some totally different organisms, such as the
bacteria, shall we find similar relationships prevailing there? If we
do, and if we also find the same conditions holding for the sperma-
tozoa, we shall have the greater reason for regarding the phenomena
we are studying as being of universal biological significance.'

Buchanan (191 8) has summarized the life-cycle of a bacterial
culture into seven periods. These are illustrated in Figure 21, whe