■MfcijiliiigW- ..JBb,

"' ^"'''^"^"'"ivip;''^i.,-i'J^^iJ^i'J::i."'-

iMir'7 Vein r?''!, IPI' ■■". f'"",, — - ■■ ■-j-j -

5{g5iiiiigiaiiii»M».niiiiii.ijiSai58li8s1iliti3i!8fll|i ■

1 1

m

r^-

LX)

I-

r-:^

%=s

O

_J- =r

□

2=

rq

□

m

a

Recent Advances in
Invertebrate Physiology

A Symposium

Sponsored by

The National Science Foundation
The Tektronix Foundation
The University of Oregon

Bradley T. Scheer, Editor

Theodore H. Bullock

Lewis H. Kleinholz

Arthur W. Martin, Associate Editors

UNIVERSITY
OF OREGON
PUBLICATIONS
Eugene, Oregon. 1957

FOREWORD

The idea of a meeting on the Pacific Coast of physiologists interested in
the invertebrates was first conceived by Professor A. W. Martin of the
University of Washington. In December 1954, he asked the members of
the editorial committee for this volume to work with him in organizing a
symposium on recent advances in the physiology of the invertebrates.
When plans were well along, in the early spring of 1955, certain actions of
the administration of the University of Washington were interpreted by
many of the invited speakers and members of the committee as prejudicial
to accepted principles of academic freedom. As a result, the present writer
was asked to accept responsibility for completing the organization so well
begun by Professor Martin, and the meeting was held on the campus of
the University of Oregon in Eugene in September 1955. We are especially
indebted to the National Science Foundation, which provided the bulk
of the funds for travel expenses, board, and lodging for the participants in
the meeting, and underwrote the publication of this volume. The Tektronix
Foundation of Portland, Oregon also contributed generously to the sup-
port of the meeting, and the University of Oregon provided facilities and
secretarial and administrative assistance.

The primary aim of the symposium was to afiford an opportunity for
physiologists interested in the invertebrates to become better acquainted
personally, and to exchange information and ideas. In this aim, the meeting
was eminently successful. Limitations of time and funds made it impossible
to bring together more than a small group ; the present volume is designed
to bring to a wider audience some of the material presented at the sym-
posium.

The committee wished to place as few restrictions as possible on the free
interchange of views. Consequently, no attempt was made to obtain ver-
batim accounts of the formal presentations or of the subsequent discussion.
The papers in this volume were prepared by the authors to cover the same
material as their oral presentations, but are not necessarily identical with
the papers as they were read. It w^ll be obvious to the reader that the
papers are of various types. Some are reviews of a large amount of ma-
terial from an entire field ; others are accounts of personal research in a
more limited field. Two papers presented at the meeting, by C. L. Prosser
and T. H. Waterman, are not included here. B. J. Krijgsman, whose
paper is included, was unable to attend the meeting. It was impossible,
within our limitations of space, time, and funds, to cover the whole vast
subject of invertebrate physiology; the selection of subjects included

[iii]

iv FOREWORD

here, though to some extent arbitrary, may be said to give a fair repre-
sentation of the most active areas of research at the present time.

I should Hke to take this opportunity to express my personal gratitude
to the other members of the committee for their helpful cooperation in
planning the meeting, and in the preparation of this volume ; to Miss
Marjorie Foxworthy (now Mrs. Charles Turbyfill) for her many services
before, during, and after the meeting; to Mr. Robert P. Bronson for his
help with travel arrangements; and to my students, A. S. Hu, R. M.
Myers, J. H. McAlear, and R. V. Crisera, for their help during the meet-
ing. Mr. George N. Belknap, University editor, and Mr. Donald Shep-
ardson, superintendent of the University Press, have been very helpful
in seeing the book through the press.

Bradley T. Scheer
Eugene, Oregon

CONTENTS

Page

Neuronal Integrative Mechanisms 1

Theodore Holmes Bullock, Ph.D., Professor of Zoology, University of
California, Los Angeles, California.

Diurnal Migration of Plankton Crustaceans 21

Edward R. Baylor, Ph.D., Assistant Professor of Zoology, and Fred-
erick E. Smith, Ph.D., Assistant Professor of Zoology, University of
Michigan, Ann Arbor, Michigan.

Prey-Predator Recognition in the Lower Invertebrates Z7

L. M. Passano, Ph.D., Instructor in Zoology, Yale University, New
Haven, Connecticut.

Prey Capture in Mantids 51

HoRST Mittelstaedt, Dr.Phil.Nat., Max-Planck-Institut fiir Verhal-
tensphysiologie, Wilhelmshaven, Deutschland.

Nervous Control of Insect Muscles 7Z

Graham Hoyle, D.Sc, Lecturer in Comparative Physiology, Depart-
ment of Zoology, University of Glasgow, Scotland.

Myogenic Rhythms 99

J. W. S. Pringle, ScD., M.B.E., F.R.S., Lecturer in Zoology, Cam-
bridge University, England.

The Machinery of Insect Flight 117

Edward G. Boettiger, Ph.D., Associate Professor of Zoology, Univer-
sity of Connecticut, Storrs, Connecticut.

Neuromuscular Mechanisms 143

C. A. G. WiERSMA, Ph.D., Professor of Biology, California Institute of
Technology, Pasadena, California.

Neurohormones or Transmitter Agents 161

John H. Welsh, Ph.D., Professor of Zoology, Harvard University,
Cambridge, Massachusetts.

Endocrinology of Invertebrates, Particularly Crustaceans 173

L. H. Kleinholz, Ph.D., Professor of Biology, Reed College, Portland,
Oregon.

Humoral Dependence of Growth and Differentiation in Insects 197

Dietrich Bodenstein, Ph.D., Insect Physiologist, Medical Labora-
tories, Army Chemical Center, Maryland.

The Hormonal Control of Metabolism in Decapod Crustaceans 213

Bradley T. Scheer, Ph.D., Professor of Biology, University of Oregon,
Eugene, Oregon.

[v] n^^'\i:\'^

VI

CONTENTS

Osmotic and Ionic Regulation inAquatic Invertebrates 229

James D. Robertson, Ph.D., F.R.S.E., Senior Lecturer, Department of
Zoology, University of Glasgow, Scotland.

Recent Advances in Knowledge of Invertebrate Renal Function 247

Arthur W. Martin, Ph.D., Professor of Zoology, University of Wash-
ington, Seattle, Washington.

Some Features of the Physiology of the Tunicate Heart 277

B. J. Krijgsman, Ph.D., Research Fellow, and Nel E. Krijgsman,
Division of Comparative Physiology, Department of Zoology, University
of Cape Town, South Africa.

The Rhythmic Nature of Life 287

Frank A. Brown, Jr., Ph.D., Morrison Professor of Biology, North-
western University, Evanston, Illinois.

^^^CAi

NEURONAL INTEGRATIVE MECHANISMS*

Theodore Holmes Bullock
University of California at Los Angeles

Integration is to put parts together into a whole. Such a process occurs
in organisms at many levels, from the subcellular to that of the community.
The levels of interest for the present purpose begin with a whole neuron,
therefore do not embrace an analysis of the mechanisms in the cell or its
membrane, and extend only one or two steps up the hierarchy, through
closely knit groups of neurons to relations between such groups, but not so
far as the level of the whole nervous system of any animal. This limitation
is not one of appropriateness but is imposed by our methods and present
understanding, as physiologists.

Indeed our understanding of the actual mechanism of nervous integra-
tion, our insight into the unit behavior which might account for this subtle
and complex result, is so meager that it may be asked, what can we say ?
This paper makes no pretense of accounting for very much normal be-
havior, and will conclude by emphatically invoking as yet unknown levels of
interaction ; but it makes an effort to say what can be said today about the
properties of neurons which must be involved in, and in certain cases appear
even to account for, the observed integration. It goes little beyond a list of
the properties — each of which provides a degree of freedom or an available
mechanism for altering the input-output function. These properties gen-
erally are additive, so that with only a few it is possible to obtain rather
complex permutations. Still our knowledge permits a very limited foray
into the vast field of higher nervous integration, and I am emboldened for
it only because so few have undertaken to bring together the several mecha-
nisms that are now known (see Fessard, 1954, 1956), while one often still
hears instances of thinking on these problems in which the neurons are
treated as purely digital or are otherwise oversimplified.

One further disclaimer is necessary. We must deal with observed prop-
erties of neural units even though they cannot be explained by current
theories of cellular mechanism. So we are not accounting for the properties ;
rather in enunciating the phenomena which may explain higher levels, we
are formulating the problems awaiting attack at molecular levels.

Let us return to the definition of our problem. Integration, I said, para-
phrasing the dictionary, is to put parts together into a whole.

* Aided by grants from the National Institutes of Health, the National Science
Foundation, and the University of California.

[1]

2 INVERTEBRATE PHYSIOLOGY

Now, what is the whole which is referred to? At the levels below be-
havior, neurophysiologists today regard it as a pattern in time and space of
quantal events, each event brief compared to the events of behavior ; these
are the impulses in the efferent nerve fibers. At the level of the single
neuron we may perhaps best express the whole as the probability of firing
within the next given interval of time, or we may revise "firing" to read
"change of state influencing another neuron," since it seems to me im-
portant to recognize the possibility of subthreshold events as adequate
stimuli even though clear cases are not yet known. We can formulate our
definition in simple terms like these if we but recognize, and then put aside
for the present, those variables which the neuron integrates into its proba-
bility of firing which are not immediately determined by other neurons, e.g.,
general chemical milieu, physical deformation, and temperature. Expressed
in terms of impulses or changes of state effective upon other neurons, in-
tegration at the unit level then becomes in the general case a relation be-
tween input and output which is either more or less than one. Usually this
means the algebraic summing of separate neuronal channels one or more of
which produces more or less than one output pulse for each input pulse ; so
long as this is true, the channels may have equal or different weights in
their effect upon output and the same or opposite sign, i.e., excitation or
inhibition. But we have to admit also the case where only one input channel
reaches the neuron under consideration, for in our type of system a quali-
tative difference between inputs is not an essential condition ; the inte-
grating cell does not know whether the signals come in the same or differ-
ent channels. The essential feature is that the neuron place some value,
other than one, upon at least some of its incoming signals, according to
their intensity, time course, time of arrival, and the locus upon the neuron
where they impinge. This definition of integration at the neuron level will
then include all junctions except those that are purely 1 :1 relays. It will
certainly include many neuroeffector junctions in which therefore the
nonnervous cell is the integrating cell. Sensory neurons certainly integrate
in the broad meaning given first, putting together different quantities in
the milieu into a probability of firing. And they may do this in part by means
or with properties which will help us to illuminate junctions. But if any
should object to the notion that receptors already integrate, they may wish
to exclude receptors on the ground that there is no input from other neurons
— it is not nervous integration. But sometimes there is ! The same cell, the
same terminal ramifications may be transducers of mechanical simuli and
postsynaptic elements under nervous control (e.g., Kuffler and Eyza-
guirre, 1955; Lowenstein, 1956).

You have patiently listened to my definition of integration. We are
supposed to talk about recent advances in invertebrates, and I must accord-

NEURONAL INTEGRATIVE MECHANISMS 3

ingly confide in you my definition of this category. For present purposes
invertebrates shall mean any animals which an invertebrate zoologist finds
interesting.

Some Properties of Units Permitting
OR Influencing Integration

At the level of the single neuron we may first list a number of the prop-
erties or conditions which classical or recent experimental facts indicate
as the probable bases of the ability of the neuron to integrate incoming
signals. Obviously all the static and dynamic characteristics of the cell more
or less directly permit or determine the activity, but we shall enumerate
only some of them, at the same time expressing the hope that extension of
intracellular analysis like those of Hodgkin, Huxley, Katz, Cole, Grund-
fest, Eccles, and others will isolate further factors and show their degree
of lability and variation in junctional membranes.

The resting potential is often thought of as a fixed character which has
only one value which is "normal," its maximum value. The evidence, how-
ever, can be construed to suggest that some synaptic regions normally
have a membrane potential which is less than its maximum, and can be
pushed either way and maintained at new levels. The level of this potential
affects not only the spike height but the excitability and the magnitude and
sign of after-potentials and of subthreshold responses.

Spike threshold and its time course after activity, the excitability cycle,
require no development here beyond the reminder that we have little infor-
mation on these crucial properties in synaptic regions of various prepara-
tions and animals. Accommodation, in particular, has not been examined
comparatively or in junctional regions; and examination is the more
necessary since the recent discovery that the classical rise in threshold
with slowly rising stimuli does not in fact obtain in the frog axon free of
connective tissue. (Tasaki and Sakaguchi, 1950; Diecke, 1954). There
are, however, other efifects of maintained or slowly changing subthreshold
depolarization, as on the form and size of the spike. This is important for
us because of the possibility that the action of terminals, dendrites, and
somata may be similarly influenced (see below under subthreshold lability) .
There may be great differences in the minimal slopes, below which thresh-
old is never reached, in different types of neurons.

Based on the distribution of thresholds or of synaptic endings in a group
of neurons, there can be curves of various shapes relating input (number
of synchronized fibers) to output (number of postunits firing). The
variable relation of input to output becomes of greater potential value as
an integrative mechanism when the output of one group of neurons is in

4 INVERTEBRATE PHYSIOLOGY

turn the input for another order of neurons and both have nonhnear curves.
Addition of output-input curves can in the extreme case readily produce a
step function, representing a kind of labile multiunit threshold providing
stability and discrimination (Fessard, 1954).

Besides the spike threshold we must recognize a separately variable
subthreshold excitability. This is manifested as a nonlinear increase in the
membrane potential with increase in stimulating current below spike
threshold (Fig. 1 ). The active, graded response which does not propagate
does not have a threshold, but it has a very labile excitability.

100

Fig. 1. Stimulus-response relation of the
subthreshold local potential in the third-order
giant fiber of the squid. Stimulus applied to
the stellate ganglion directly ; recording from
the same ganglion. Two experiments are
shown : A. Cathodal stimuli, whose voltages
are shown by the upward deflected square tops,
elicit local responses, whose amplitudes are
shown by the downwards deflected triangular
waves, plus one spike which goes ofif scale and
is seen as a descending phase above the base
line. Plotted as per cent of threshold on ab-
scissa, per cent of maximum recorded local
potential on ordinate (spike ofif scale). B.
Cathodal stimuli (not shown) give responses
above base line, anodal stimuli of same in-
tensity below. Plotted as above but ordinates
are cathodal response minus anodal to show
the development of nonlinear, nonelectrotonic,
"active" local response. (From Bullock, 1948;
reprinted with permission of the Journal of
Neurophysiology. )

Subthreshold activity exhibits lability also in other ways (Bullock, 1948 ;
Bullock and Hagiwara, 1955). Its rate of rise and especially its rate of fall
vary even from moment to moment under repeated low-frequency stimu-
lation. It may hesitate for many milliseconds before growing up into a
spike or starting its fall. Its spatial decrement may vary, possibly as a con-
sequence of change in time course or possibly as a consequence of a labile
decremental propagation. In some conditions, there is a heightened ex-
citability after a subthreshold response, beginning without any refractory
period (e.g., fresh axon of squid). But in other conditions there occurs a
depression after such a response, with a recovery which requires many
milliseconds (e.g., fatigued axon or synapse of giant fibers of squid). This
depression may be followed by a supernormal period. These considerations

NEURONAL INTEGRATIVE MECHANISMS 5

gain in significance if we believe that integrative neurons are typically
under tonic subthreshold influence.

The excitability we have spoken of so far is excitability to artificial stimu-
lation of limited kinds. But the outstanding characteristic of synapses is
their sensitivity to some consequence of activity in other neurons, and we
need not be concerned here whether that is one or more specific chemicals
or nonspecific ions. We do need to note : ( 1 ) that the response may be
excitation or inhibition; (2) that one and the same input citannel (pre-
synaptic fiber) can cause one or the other, depending on the level of the
membrane potential of the postsynaptic neuron; (3) that different pre-
synaptic fibers can cause response of the same sign but different rates of
rise, jacilitation, maximum height, etc., as in crustacean muscle; (4) that
these differences may be discontinuous and unequal in the proportional
role of the several characters measured ; (5) that summation of the differ-
ent kinds of input in the same postjunctional cell may be complex (crusta-
cean muscle, crayfish central giant-to-motoneuron synapse) ; and (6) that
inhibition is not just the reciprocal of excitation as measured by its effect
on various aspects of activity.

A highly variable property of the utmost importance for integration is
the response of the postsynaptic unit to repeated presynaptic impulses. In
some cases (e.g., inhibitory escape in the cardiac ganglion of lobsters) the
initial eft'ect of a sustained barrage gradually diminishes as measured by
the output of the postunit, reaching a plateau at a new level (Fig. 2). In
the example referred to this happens in some seconds at a frequency of
presynaptic activity of 20-40 per sec. In other cases (e.g., the synapse of
pacemaker upon follower cell in the same ganglion) there is actually what
may be called de jacilitation — the postsynaptic potential to the second pre-
synaptic impulse is less than to the first, the difference being proportional
to the frequency. This may be regarded as a consequence of relative refrac-
toriness. Its importance lies in the fact that it happens in a frequency range
within the normal firing range of these ganglion cells. The more familiar
cases are those of jacilitation — which should be distinguished from tem-
poral summation by the criterion that each response or increment is greater
than the last. The magnitude of facilitation and its rate of growth and decay
vary widely. As an example, their consequences can be clearly seen in
Wiersma's comparison (1952b) of the responses to the same average fre-
quency delivered with alternately long and short intervals and delivered
with uniform intervals. Some junctions give the same response (small,
slowly decaying facilitation) , others respond enormously more to the paired
train (large, rapidly decaying facilitation).

Another group of properties with profound influence upon output, es-
pecially in the formation of patterned bursts, is in the domain of ajtcr-

INVERTEBRATE PHYSIOLOGY

Input from
inhibitor axon

^ fcvWon process
Excitability / \ov

(smoothed)

r-^ inhibitory process i

Output in

motor axon —^-^ ^ ' ' ' '

Time in

seconds

Fig. 2. Diagram of the response of a single unit of the cardiac ganglion of a
lobster to stimulation of the inhibitory axon. Each vertical line represents an im-
pulse. Maintained activity in the inhibitor produces, first, deep depression, followed by
partial escape or adaptation. Termination of inhibitor activity produces a postinhibi-
tory excitation after a brief latency or inhibitory after-effect. The observed input
and output are related by an algebraically summing series of processes, including
many of those listed in the text. Just two of these, which are moreover not directly
measured but inferred, are shown here. (After Maynard.)

effects. These may be positive or negative or both in sequence, of various
relative durations and magnitudes. To say the same thing in more familiar
terms, there may be an after-discharge following cessation of presynaptic
excitatory bombardment or an after-inhibition following the end of in-
hibitory influx, and there may be, with or without this positive after-effect,
a rebound effect — postexcitatory depression or postinhibitory excitation.
When phases of opposite sign succeed each other, they may be affected
independently by various factors, as though manifesting separate under-
lying processes.

Recently we have distinguished another property which is of importance
in permitting repetitive firing in response to a single presynaptic impulse
in the cardiac ganglion of lobsters (Hagiwara and Bullock, 1955). This
may be described as a sajcty factor of much less than one so that the post-
synaptic impulse, once initiated in the axon, cannot invade the cell body
(antidromically). This protects the latter from the possibility of loss of
any partial depolarization it may have built up. Its significance depends on
the asymmetrical relation between cell body and axon — the slow synaptic
potential of the soma can spread electronically into the axon with less

NEURONAL INTEGRATIVE MECHANISMS 7

attenuation for any given time constant of the membrane than the brief
spike potential of the axon can spread into the soma ; thus the soma can
excite repetitive discharge of the axon with a single, long-lasting synaptic
potential, while the resulting spikes are seen in the soma as tiny, five
millivolt electrotonic deflections (Fig. 3). This case also illustrates the in-
teraction of several factors in determining when a neuron fires. It is obvious
that a fixed threshold voltage across the membrane of the soma does not
exist ; something else interacts with voltage so that successive spikes occur
at smaller depolarizations.

A somewhat similar situation may account for the activity of dendrites
as analyzed by Clare and Bishop (1955a,b). Dendrites in the vertebrate
cortex appear not to conduct all-or-none impulses toward the cell body,
as has been classically supposed from the law of dynamic polarity of Cajal
(1909). Instead slow potentials generated in dendrites and spread elec-
trotonically may influence the spike-initiating region of the soma. One can
think of two interacting regions of integration, the dendrites and the soma ;
in this way the vertebrate cortex differs from most invertebrate nervous
tissue where the soma probably plays little role and cell-free neuropiles are
responsible for integration (Bullock, 1952).

The last property to which we will allude here is spontaneity. This varies
not only in the degree to which it is developed but also in the form by which
it is manifest, and the question is still unanswered whether these several
forms are different in underlying mechanism. Spontaneous subthreshold
activity seems to be of two kinds but these are possibly basically the same.
In some cases it is quasi-sinusoidal, in others it rises (depolarization) more
or less linearly to a point where it reaches a threshold and initiates a spike
the repolarization of which carries the membrane back to the high level
from which the so-called generator potential can begin again. These two
forms of subthreshold change differ at least superficially as a sine wave
oscillation from a relaxation oscillation ; that is, one form can continue to
go through successive cycles without an all-or-none discharge, the other
requires such a discharge to restart the cycle. ^ The generator potential is
best known in certain sense organs and in the specialized muscle cells
pacing the vertebrate heart, but it is also present in integrative neurons
which control other neurons, i.e., pacemaker interneurons in the lobster
cardiac ganglion.

Spontaneous activity may be relatively rhythmic or nonrhythmic. Even
in the same unit a continuous spectrum may be shown between very small
and very large standard deviations of the intervals between spikes — the

1 We have recently found, in spontaneous ganglion cells of the lobster cardiac
ganglion, that a propagated spike is not necessary to repolarize the soma to a high
level and restart the cycle. An active but graded form of soma potential suffices.

INVERTEBRATE PHYSIOLOGY

B

J

Fig. 3. A. Intracellular potential of the soma of a large neuron of the cardiac
ganglion of a lobster. The activity results from stimulation applied to the ganglion
a few mm. away. The first deflection is an antidromically conducted spike. The large
deflection is interpreted as a synaptic potential resulting from arrival of an impulse
in a presynaptic fiber ; it is all or none by this form of presynaptic stimulation. The
record shows two spikes arising from the synaptic potential, one at its peak and one
on the falling phase ; other records show from one to five, the number apparently de-
termined by the condition of the postsynaptic element, not by a difference in the
number of presynaptic elements. Time marks, 10 msec. ; vertical calibration, 10 mv.

B. Intracellular potential of the soma of a large neuron of the cardiac ganglion
of a lobster. The activity is spontaneous in the ganglion and in the form of bursts,
one for each heart beat. The large deflection is regarded as a synaptic potential result-
ing from arrival of presynaptic impulses, and the smaller slower deflections are
interpreted similarly. The sharp spikes (7-9 per burst) are the impulses in the axon
leaving this soma ; they arise from or near the crest of a synaptic potential. Note
that, as in A and C, they do not have a fixed voltage threshold but a threshold which
depends on other factors, e.g., time, even in naturally occurring repetitive firing. The
spikes between bursts are preceded by a slow depolarization or generator potential ;
these are regarded as spontaneous activity of the cell we are in. The record illustrates
the complexity of the interaction of integrative properties in a simple ganglion. The
record is 3.6 seconds long.

C. Intracellular potentials of a median giant fiber of an earthworm. Two micro-
electrodes are inside the fiber, 11.3 mm. apart, the one farther from the stimulating
electrodes is the lowest beam, the nearer one is the middle beam. The top beam is an
extracellular monitor still farther down the fiber. Two shocks are given, 2 msec, apart.
The first elicits action potentials of 98 and 110 mV. on resting potentials of 72 and
74 mV. The second elicits smaller direct spikes plus complex small potentials in the
nearer penetration only (arrow), regarded as synaptic potentials from small fiber
bombardment eventually leading to a spike which propagates. (There is also a lateral
giant spike on the external leads, in the middle of the sweep, but this has no reflection
in the internal median giant leads.) Like the preceding, the record shows the inter-
action of prepotentials and a complex recovery of excitability in determining firing.
Time in msec. (The experiment and permission to use the record are due to the kind-
ness of C. Y. Kao.)

NEURONAL INTEGRATIVE MECHANISMS 9

former at high average frequency and the latter at low frequency of firing.
But at any given average frequency there are types of neurons which are
markedly regular and others which are markedly irregular in successive
intervals. This suggests intracellular variables of importance in effective
magnitude, such as fluctuating spike threshold, fluctuating amplitude of
sinusoidal potential or of rate of rise of generator potential, fluctuating
area of nonpropagated or decrementally propagated activity, and the pres-
ence of multiple loci of origin of spontaneous subthreshold activity
(Wiersma, 1952a).

Besides these short-term changes in the frequency of firing there are in
some cases long-term changes in average frequency on the scale of minutes.
We will consider their significance below.

Eyzaguirre and Kuffler (1955) have just described a puzzling case of
repetitive firing in the cell body of a neuron which has recently received an
antidromic spike. Whatever the explanation — and these authors propose a
tenable one based on difl^ering local delays or partial blocks in the several
dendrites — it is germane as an indication of the degrees of freedom present,
and our purpose here is only to review the available ways in which the
neuron can exhibit the several variables whose interaction could accom-
plish integration. In our present state of knowledge this means we have to
include some observations whose "explanation" is less obvious than that
of others. Eyzaguirre and Kufiler's interpretation of the intracellular after-
discharge in the stretch receptor cell of the dorsal muscle sense organ of
crayfish may be correct ; but Bullock and Turner (1950) reported a similar
phenomenon in the giant fiber of the earthworm, so the phenomenon does
not depend on the particular anatomy in the crayfish receptor. In the earth-
worm, a spike initiated at stimulating electrodes and conducted a long dis-
tance down the giant fiber arrived at a locus of partial or complete block
( anode of a polarizing circuit) where it hesitated before proceeding or died
out ; after five or more milliseconds a burst of several impulses at high fre-
quency originated at that locus or immediately adjacent to it.

One of the consequences of spontaneity may be sensitivity to zveak elec-
tric fields. At any rate one can control the frequency of discharge of spon-
taneous integrating centers by passing a fraction of a microampere through
a mass of tissue of a few ten-thousands of an ohm resistance, where the volt-
age drop along the length of a single cell must be a fraction of a millivolt
(Bullock, Burr, and Nims, 1943, on Limulus ; Maynard, 1956b, and Terzu-
olo and Bullock, 1956, on lobster cardiac ganglion ; Hagiwara, Oomura, and
Takagi, unpub., on citrated squid axon) . The voltage drop across the mem-
brane must be still smaller. Either an excitable mechanism not familiar to us
is operating or the curve of membrane potential against firing interval is
exceedingly steep, which means the threshold is very critical and constant.

10 INVERTEBRATE PHYSIOLOGY

The evidence against electrical transmission, based on the absence or
minute size of the voltage change across the postsynaptic membrane pro-
duced by the arrival of the presynaptic impulse (del Castillo and Katz,
1954, on muscle; Bullock and Hagiwara, 1955, on squid) may be con-
clusive. But this does not mean, as some have supposed, that weak electric
fields are without influence on poised or already active neurons. The ex-
periments cited on cardiac ganglia, as well as many others of the same sort,
classical and recent, are direct and pertinent and, when considered quanti-
tatively, impressive in the sensitivity they bespeak.

The significance of this sensitivity is the enormous integrating poten-
tiality, in complex centers, of the fields of current interacting among small
and large groups of neurons. Here ordinary synaptic pathways give way
in importance to architectonics. And synchronization and desynchroniza-
tion of graded subthreshold activity of somata and dendrites take on a para-
mount significance both in producing fields effective upon other units and
in sensitizing the somata and dendrites themselves to effects en masse (cf .
Fessard, 1954).

Taken together with our earlier conclusion (see above under sub-
threshold excitability) about tonic subthreshold influence, these consid-
erations also lead us to the suggestion that much of normal nervous func-
tion occurs without impulses but mediated by graded activity, not only as
response but also as stimulus.

Pattern Formation in the Discharge of Groups of Neurons

As a special case of the most general interest we may examine the inte-
grative mechanisms capable of organizing patterned bursts of impulses in
which the serial order is determined centrally and in which several efferent
neurons are coordinated. Since overt behavior consists in just such co-
ordinated bursts of impulses, as far as its neurophysiology is concerned,
this problem is a large segment of the problem of behavior. It is too much
to expect that we can enunciate a satisfactory general solution or even a
complete solution of a single case. But I believe there are some things we
can say which will carry us quite a way in accounting for simple patterns
with only the properties outlined above. Actually there is little difficulty in
drawing hypothetical circuit diagrams of neurons with connections and
properties within known limits which will produce a given pattern of out-
put impulses in space and time. But there has been little effort to discover
what actual neurons and connections are employed in real cases, perhaps
because the enormous neuron pools in the familiar cases are too complex
in sheer number of cells and impulses. A few cases have been studied re-
cently in which a very small number of nerve cells control a large muscula-

NEURONAL INTEGRATIVE MECHANISMS 11

ture and in which physiologically initiated movements or impulse bursts
can be nearly completely accounted for in the neurograms.

In the simplest case there is a cell which has a fixed frequency of firing
and this cell is simply turned on and off by input from the periphery or from
higher centers. This has been found in the control of (neurogenic) sound
production in a cicada (Hagiwara and Watanabe, 1956) and in the control
of electric organ discharge in Torpedo (Albe-Fessard and Szabo, 1954.)
In both the pacemaker cell is an interneuron, not a motoneuron, and the
fixed frequency is high — 200 per second in the former, 100 in the latter.
The frequency or intensity of stimuli to sensory nerves does nothing in
the cicada but determine how long the pacemaker will buzz and how
promptly it will start. The system is like an oscillator controlled by a switch
which can be only on or off but which can be turned on with various speeds
due to the finite distance the switch must be moved before it changes its
state. In Torpedo it does not yet seem clear whether the 100 per second
frequency is independent of the input. In both cases the frequency-deter-
mining interneuron is penultimate — it controls the motoneuron directly,
one motoneuron on each side in the cicada, about 70,000 on each side or
100 for each interneuron in the electric lobe of Torpedo. There is one other
step in the cicada. Whereas the electric-lobe motoneurons follow the in-
terneurons 1 :1 after the first impulse of a series, the cicada motoneurons
follow every other interneuron impulse, therefore firing the muscles at
100 per second. Moreover, the two sides are always 180 degrees out of
phase, so that there must be some reciprocal inhibition of the two sides.

The only other preparation which I will discuss here is the lobster heart
ganglion (Fig. 4), which is somewhat more complicated. This is largely
based on the work of my former associate. Dr. Donald Maynard, but some
aspects have been extended by Dr. Hagiwara and myself (Maynard, 1953
a,b,c, 1956a,b,c; Bullock, Cohen, and Maynard, 1954; Hagiwara and
Bullock, 1955) . Here a pattern is repeated at regular intervals, correspond-
ing to each heart beat ; and normally the heart beat, or as we shall call it the
burst, is paced by the activity of certain of the four small posterior cells.
Here, as in the system we have just examined, each of the follower cells
responds to the pacemaker, or to some other cell triggered in turn by the
pacer, with a train of impulses whose frequency is not the same as that of
any other cell but is peculiar to the cell. But this frequency is not fixed.
It starts high or quickly rises to a maximum and then falls along a curve
characteristic for the cell over some hundreds of heart beats. This fre-
quency/time curve could conceivably be determined entirely by the prop-
erties of the given cell since the cell can respond to a single incoming im-
pulse by a repetitive discharge, as we have seen happen in intracellular
records. A single large, slowly decaying synaptic potential can, by the

12

INVERTEBRATE PHYSIOLOGY
ant.

LNS

Fig. 4. Diagram of the cardiac ganglion
of the spiny lobster, Panulirus intcrruptus.
Not to scale ; the ganglion is about 12 mm.
long, the largest cells about SOju,. Neuro-
piles simplified. LNS, large neuron soma;
i, the single inhibitor fiber ; a, the two
accelerator fibers from the central nervous
system ; np, neuropile ; DA, dendritic ar-
borization, possibly sensory; SNS, small
neuron soma.

post.

mechanism outlined earlier, elicit a short train of up to five impulses. In
a normal burst it is not so simple, however.

The pacer fires repetitively. It seems probable that the pacer fires the
other small cells directly and that some or all of them drive the large cells.
This means there could result a simple "open chain" determination of
pattern, as Maynard calls the type of system proposed by Rijlant for the
Liniulus heart, all the cells being controlled directly or through inter-
neurons by the pacemaker, in direct line or chain of cominand (Fig. 5).

But it seems more likely that in the normal burst there is interaction.
The frequency /time curves strongly suggest that two additional factors
are operating : ( 1 ) great individual dififerences in the time course of ex-
citability, responsiveness, or both (we have seen that these are separate
properties of a cell) to the same input signal, and (2) feedback, at least
positive and possibly also negative.

This feedback is probably critically responsible for normal burst main-
tenance. Once a pacemaker starts activity, followers build it up greatly

NEURONAL INTEGRATIVE MECHANISMS

13

by positive feedback, and so intensify synaptic excitation as to bring on a
compensatory depression terminating the burst. The silent period and
hence burst frequency is thus determined not by any single cell but by the
integrated activity of several, i.e., by the duration and intensity of the
whole burst. The excitability cycle by itself may be sufficient to account
for the intermittent activity patterned out of an apparent natural tendency
to continuous activity, at least whenever we have several interacting units.
When intermittent trains occur in single isolated neurons we must invoke
a new underlying oscillatory process, but such instances do not appear to
be a normal part of the activity in this ganglion.

A A fS^

Fig. 5. Possible connections of cardiac ganglion leading to patterned burst forma-
tion. A. Multiple chain. Top cell is the pacemaker, middle row interneurons, bottom
row, motor neurons as proposed by Rijiant for Limulus. B. Closed chain. There is no
morphological pacemaker. Every cell acts as motor neuron and at times as pace-
maker and as interneuron. Burst formation depends on the properties of interaction
rather than on connections. C. Modified closed chain as in lobsters. Small cell acts as
pacemaker, but there is feedback, and patterning of burst depends primarily upon the
properties of this interaction and its sequelae in the cells. (After Maynard.)

The system just outlined is a "closed chain" in Maynard's terminology
if all cells are equally effective upon each other. The actual observations
are most easily understood by considering the cardiac ganglion to be a
"modified closed chain" (Fig. 5) with some asymmetrical connections and
specialization of function but in which recriprocal interaction and spon-
taneity of all units are dominant features. The largest cells are in all likeli-
hood incapable of triggering a normal burst of the ganglion, although they
are spontaneously active and in special circumstances do pace each other.
Their unique role is that of motoneurons, but they act also as integrative
neurons which formulate an output that is a function of, but quite unlike,
their input, and help to determine burst durations and hence frequency.

One advantage of this system is its lack of dependence on any one critical
value. No one cell fires at the burst frequency, i.e., heartbeat frequency.
The pacemaker fires a train in each burst and is simply the first of probably
four duplicates. Each of these four, having been turned off at the end of a
burst by accumulating hyperpolarization from internal and external
sources, develops a slow generator potential rising a few millivolts in a

14 INVERTEBRATE PHYSIOLOGY

second or two toward a very constant firing threshold. The burst fre-
quency depends then not only on the rate of rise of the generator potential
in the fastest pacemaker unit but also on the degree of depression at the
end of the last burst, which in turn depended on the activity of many units
which participated in the burst, thus on the burst duration and intensity.
We are in the presence here of reverberating circuits, but they serve not
to maintain activity in a continual self-re-excitation but to "provide means
of spontaneously active units to undergo a slower alternating auto-excita-
tion and depression" (Maynard). The system is capable of relative sta-
bility in the sense that removal or silence or hyperactivity of any or even
several units does not greatly upset the pattern.

These are the observed properties and the inferred mechanisms. I must
emphasize that we do not know the connections in detail, and it is still only
the simplest inference that these properties are to be explained in terms
of the properties of units as recorded from single cells in this ganglion.
There is reason to believe that this arrangement is not Hmited to a ganglion
with such a small number of cells but applies also to the large, many-celled
ganglion of the heart of Limulus.

Superimposed upon or perhaps underlying the excitatory state which
plays the role just described is a sensitivity to physiological degrees of
stretch or inflation. We do not know whether this inheres in all the cardiac
ganglion cells or only in some, but it appears to be a direct response of
these cells and not due to influx from some separate sensory neurons. The
effect of stretch is to accelerate or enhance spontaneity.

In addition to integrating stretch, its own spontaneity, and intramural
excitation and possibly inhibition, the cardiac ganglion neurons must also
integrate input they receive from two pairs of extramural acceleratory and
one pair of inhibitory nerve fibers from the central nervous system, each
of whose effects is differential with respect to different aspects of activity
and not simply reciprocal with the other. For example, burst frequency
and number of spikes per burst are decreased by inhibitory stimulation but
spike frequency within a burst may go up. The time courses of develop-
ment, adaptation, and after-effects are slower for acceleration and are not
mirror images of those in inhibition.

We do not understand the factors that determine these effects ; at times
a change in the tonic level of extramural acceleration or inhibition will
change the burst pattern progressively and nearly proportionally for each
parameter, at other times it will affect the large followers primarily and
may stop output to the muscle without gross change in burst pattern, and
at still other times the small pacers are affected more, causing a slowing,
for example, of the burst frequency without corresponding depression of
the followers, which now escape and fire just before each burst.

NEURONAL INTEGRATIVE MECHANISMS 15

Changes in the phasic rather than the tonic level of this extramural in-
fluence can lead to still more complicated differential effects, since now
we have the several successive phases of after-effects, which are different
between large and small ganglion cells and between acceleration and in-
hibition. An example of these effects is the paradoxical driving of nearly
quiescent units by phasic inhibition, via the postinhibitory rebound
(Fig- 2).

Signal Versus Noise and Some
Specifications for a Brain

The common case of continual discharge in sensory nerve fibers may be
partly understood as representing ( 1 ) reception of steady states or/and
(2) a state of poise associated with high sensitivity, and also (3) a pro-
vision for signaling in one line both positive and negative changes in the
stimulating parameter (e.g., increases and decreases in temperature, light,
or stretch, and forwards and backwards movements of hairs in statolith
organs). But it raises a serious problem in any case where rhythmicity is
not perfect. What change in the output of the sensory nerve fiber consti-
tutes a signal of environmental change and not a random fluctuation in
spontaneity ?

This problem has some general significance and not alone in certain
sense organs but probably in the integration of signals between centers in
the brain. Deferring for a moment any contributions to the discrimination
between signal and noise provided by a multiplicity of parallel channels
(afferent nerve fibers), it is necessary to consider first the alternatives
available for analyzing the input in each single channel. Among them we
may recognize certain ones, as follows. The signal could be regarded as :

( 1 ) The actual instantaneous frequency, i.e., the reciprocal of the inter-
val since the last impulse (Fig. 6). If this constitutes the significant signal
for the central analyzing mechanism the threshold of the mechanism would
be very high, since the signal has to be reliably higher than any spon-
taneous fluctuation in interval. In the general case the fluctuations in indi-
vidual intervals are larger than any fluctuations in averaged or integrated
frequency. This possibility therefore seems unlikely or at least maladaptive
and wasteful of information.

(2) Frequency averaged over some period (Fig. 6) . The threshold will
be lower than in ( 1 ) but, if the time constant of integration is short, not
much lower ; and, if the time constant is long, reaction time, detection of
brief stimuli, flicker resolution, and spatial localization of moving stimuli
will suffer. Some compromise seems quite possible but not as advantageous
as the followins:.

16 INVERTEBRATE PHYSIOLOGY

^^:^7^-y\'z-'."^^X'-.''::-^-''^.'!*^y,\l''J'i':^'-^'-'C^^^^

Fig. 6. Frequency-time plots of the activity of a single receptor unit in the infrared
sense organ of the rattlesnake, Cro talus, to three different intensities of physiological
stimulation. Two simultaneous plots of the same activity : the finer spots in a more
nearly smooth curve are from an integrator which averages the frequency with a
time constant of 2.5 seconds. The coarser spots with a wide scatter are from a pulse-
interval-plotter which places the spot higher on the ordinate the shorter the interval
since the last impulse. The ordinate scales are not given here as they are not im-
portant for the present purpose. Time scale at lower right is one second. The case is
chosen because the wide scatter of spots indicates a very arrhythmic activity except
at high frequency under strong stimulation ; this is typical of this sense organ. In A
the stimulus is weak though well above threshold defined as a significant change in
the integrated frequency ; but the unintegrated frequency record signals the environ-
mental event very poorly and unreliably. While the integrated frequency greatly im-
proves the detection, it is still dependent for its usefulness upon the absence of slow
fluctuations in the spontaneous background; this short-term record does not show
such fluctuations. B and C show the disadvantage of integrated frequency as a signal
at high stimulus intensities — its slow response; the unintegrated frequency reliably
reports much sooner. Note partial adaptation and postexcitatory silent period before
spontaneous background returns. (From Bullock and Diecke. )

Instead of the actual frequency we may consider the ratio of, or the
difference between, frequency recently transpiring and a background fre-
quency.

(3) The ratio, frequency-integrated-over-some-short-time- just-past
( Frci ) to frequency -integrated -over -some -longer -period -just -past,
( Frc2 ) • Sensitivity would increase as the time constants RCi and RC2
increase. The greater RCo is, the better (within a limit depending on the
relative role of this channel in change detection and steady-state detection) .
RCo cannot be less than several minutes if it is to avoid loss of information
by an adaptation of the central nervous system more rapid than that of
some receptors. RCi cannot be longer than a few seconds if loss of informa-
tion is to be avoided, because sensory adaptation to weak stimuli already
occurs in that time in many receptors. It should be, in fact, a fraction of a
second in order to keep down reaction time and loss of information on
brief stimuli and flicker.

Steady-state input from nonadapting receptors, like position and some
temperature receptors, could not be processed by such a formula, for long

NEURONAL INTEGRATIVE MECHANISMS 17

maintained high frequency would gradually raise the denominator until
the ratio fell to unity. These modalities or the steady-state aspect of such
input must use (2), above.

A further improvement in the extraction of information from such
fluctuating signals arising from phasic input would result if the receiver
could place a greater value upon changes in the ratio which occur rapidly,
since these are more likely to be real stimulus-signals and less likely to be
spontaneous fluctuations. This likelihood is based on the randomness of the
successive intervals in spontaneous fluctuations ; several short intervals
in succession are highly unlikely. One possible formulation is the following.

(4) [Ratio (3) ] X [Rate of change of ratio (3) integrated over RC3] .
RC3 would probably be intermediate between RCi and RC2. If F suddenly
changes, Frci will be increased soon, the ratio will increase, the rate of
increase of the ratio will be maximal for a short while, and a sudden stimu-
lus will be signalled thereby. The signal meaning "strong stimulus" will
then begin to decline even if the high F is maintained — first, because of the
decrease over time constant RC3 in the extra multiplying factor, and then
because of the gradual fall in the ratio, as the high F gradually raises the
denominator.

The possibility that, instead of a ratio of short-term integrated frequency
to long-term integrated frequency, the latter should simply be subtracted
from the former is available but seems less likely because of its dispropor-
tional sensitivity at low intensities.

It is possible by one further step to extract still more information from
the same pattern of impulses. If the central nervous system can subtract
a fixed frequency from the incoming frequency and then estimate the ratio
of frequency integrated over the short term to that over the longer period,
and increase the value if there has recently been a sudden rise, the etTects
of random fluctuations in successive intervals in the background spon-
taneity could be largely eliminated.

[F^^^^^^^~^n..a]mtegr^tedoyerRC. rRate of change of this T

[Factual — Fnxed] integrated over RC, ^ Lratio during RC3 J

If the fixed frequency is higher than the extreme range of random fluc-
tuations, no ratio is necessary since the numerator is zero, but sensitivity
will not be maximal. Sensitivity will be maximal if only a few spontaneous
impulses exceed the fixed level ; the ratio will be more quickly altered by
a real stimulus and the time constant RCi can now be shorter than in the
last two formulae, improving reaction time, flicker, and brief stimulus de-
tection while still smoothing enough to eliminate nearly all false signals.

Thus far we have considered the desiderata for a valid signal of en-

18 INVERTEBRATE PHYSIOLOGY

vironmental change in the single unit. Superimposed on these is the possi-
bility of gaining reliability of small signals and hence sensitivity to actual
stimuli by summing the activity in independent parallel channels or afferent
fibers. This very probably occurs as an important normal mechanism in
those organs or animals that can afford to have many channels, and per-
mits the central nervous system to demand a high level of significance.

The problem of extracting reliable information from a continuously
active background of fluctuating impulse intervals seems likely to be quite
a general physiological problem. Spontaneous activity has emerged in
recent years as a feature of many sense organs. Even in the best cases of
rhythmicity, the constancy of successive intervals is relative and small
signals look much like noise. The possibilities suggested may be difficult
to test physiologically, but would be extremely interesting if they approxi-
mate the specifications for actual integrative centers. It can be anticipated
that in different animals and in dift'erent modalities the relative roles of
different parameters of the integration will differ significantly. In arthro-
pods, following the line of argument developed a few years ago by Wiersma
(1952b), it may be true for some sense modalities that only a few parallel
channels are available and reliability or sensitivity or resolution may be
sacrificed.

We have not attempted here to extend the consideration to the problem
of resolution ; but it is clear that there must be a somewhat similar integra-
tion in cases like the vertebrate eye, ear, temperature, and tactile senses,
where central discrimination far exceeding that of individual receptor units
is achieved by comparing the firing frequencies of many channels from
overlapping units of slightly different maxima. Little is known about this
process in invertebrates and we shall only pause to note that it appears
quite amenable to explanation in terms of the properties of units.

Expectation of New Levels of Complexity

So far we have attempted to explain or describe integrative phenomena
in terms of familiar properties of units. This effort could be carried a good
deal farther and might well account for a considerable part of the function-
ing of nervous tissue in higher animals. Certainly a treatment of the pres-
ent theme must begin with, and today consists largely of, a list of such fa-
miliar properties.

The question whether the known properties of units suffice, when com-
bined in great and intricate enough permutations, is so risky that few
neurophysiologists will be caught guessing, though some have recently
answered not only "no," but have implicitly or explicitly denied that physio-
logical (matter-energy) mechanisms will be found adequate to account

NEURONAL INTEGRATIVE MECHANISMS 19

for behavior and mental phenomena (Sherrington, -1951 ; Eccles, 1953).
The gap at present is certainly staggering. It must be realized that much
of our knowledge of brain physiology, while contributing to the better
localization and fragmentation of the problem, actually still consists in
specifying the nature of the phenomena to be explained without un-
equivocally helping to decide whether these will be explicable in terms of
the known properties of neural units or of those plus as yet unknow^n
properties or will not be explicable in any physiological terms.

I will confine myself here to the mere statement of my own faith — and
it is just that — that no extension of the known mechanisms will be found
adequate to explain higher activities of central nervous systems, but that
we need not fall back on a dualism. Our position is a little like a meteor-
ologist or oceanographer trying to account for the great events of the ocean
and atmosphere from the known properties of the individual atomic species.
I believe we have yet to discover fundamental new properties and relations
at the level of masses of neurons — emergents in the old sense of inhering
in but not readily predicted from a catalog of unit properties. One example
may be the slowly traveling waves of synchronous, slow, subthreshold po-
tential change in neuron masses.

I believe it will require such emergent mechanisms to understand, for
example, complex integrations like that which must accompany the central
control of afferent influx — now well known for several modalities. There
must be a central correction for the control so that the world is interpreted
reasonably correctly. Or again we may think of the Efferenzkopien of
van Hoist which similarly corrects for the distortions of input caused by our
movements. The plausibly postulated energies, drives, appetitive behavior,
releasers, and other entities of behavioral science and above all the amazing
phenomenon of nearly nonlocalizable, anaesthesia- and shock-proof learn-
ing appear to me, at once, to require such still undiscovered physiological
parameters and to be the stimulus to new levels of search.

REFERENCES

Albe-Fessard, D., and T. Szabo, 1954. fitude microphysiologique du neurone inter-

mediare d'une chaine reflexe disynaptique. Comp. Rend. Soc. Biol. 148, 281-284.
Bullock, T. H., 1948. Properties of a single synapse in the stellate ganglion of squid.

/. Neurophysiol. 11, 343-364.
Bullock, T. H., 1952. The invertebrate neuron junction. Cold Spring Hrbr. S\mp.

Quant. Biol. 17, 267-273.
Bullock, T. H., H. S. Burr, and L. F. Nims, 1943. Electrical polarization of pacemaker

neurons. /. Neurophysiol. 6, 85-98.
Bullock, T. H., M. J. Cohen, and D. M. Maynard, Jr., 1954. Integration and central

synaptic properties of some receptors. Fed. Proc. 13, 20.
Bullock, T. H., and F. Diecke, 1956. Properties of an infrared receptor. /. Physiol, in

press.

20 INVERTEBRATE PHYSIOLOGY

Bullock, T. H., and S. Hagiwara, 1955. Further study of the giant synapse in the stel-
late ganglion of squid. Biol. Bull. 109, 341-342.
Bullock, T. H., and R. S. Turner, 1950. Events associated with conduction failure in

nerve fibers. /. Cell. Comp. Physiol. 36, 59-82.
Cajal, S. R., 1909-1911. Histologic dn systeme nerveux de I'homme et des vertebres.

Trans, by L. Azoulay. Paris.
Clare, M. H., and G. H. Bishop, 1955a. Properties of dendrites ; apical dendrites of

the cat cortex. E.E.G. Clin. Neurophysiol. 7, 85-98.
Clare, M. H., and G. H. Bishop. 1955b. Dendritic circuits : the properties of cortical

paths involving dendrites. Amer. J. Psych. Ill, 818-825.
Del Castillo, J., and B. Katz, 1954. Changes in end-plate activity produced by pre-
synaptic polarization. /. Physiol. 124, 586-604.
Diecke, P., 1954. Die "Akkomodation" des Nervenstammes und des isolierten Ran-

vierschen Schniirringes. Ztschr. Naturf. 96, 713-729.
Eccles, J. C, 1953. The Neurophysiological Basis of Mind. Clarendon Press, Oxford.
Eyzaguirre, C, and S. W. Kuffler, 1955. Further study of soma, dendrite and axon

excitation in single neurons. /. Gen. Physiol. 39, 121-153.
Fessard, A. E., 1954. Mechanisms of nervous integration and conscious experience. In

Brain Mechanisms and Consciousness, a Syniposiu)n. Edited by E. D. Adrian, F.

Bremer and H. H. Jasper. Charles C. Thomas. Springfield, 111.
Fessard, A. E., 1956. Formes et caracteres generaux de I'excitation neuronique. XXe

Congress International de Physiol. 1, 35-58.
Hagiwara, S., and T. H. Bullock, 1955. Study of intracellular potentials in pacemaker

and integrative neurons of the lobster cardiac ganglion. Biol. Bull. 109, 341.
Hagiwara, S., and A. Watanabe, 1956. Discharges in motoneurons of cicada. /. Cell.

Comp. Physiol. 47, 415-428.
Kuffler, S. W., and C. Eyzaguirre, 1955. Synaptic inhibition in an isolated nerve cell.

/. Gen. Physiol. 39, 155-184.
Loewenstein, W. R., 1956. Modulation of cutaneous mechanoreceptors by sympathetic

stimulation. /. Physiol. 132, 40-60.
Maynard, D. M., 1953a. Inhibition in a simple ganglion. Fed. Proc. 12, 95.
Maynard, D. M., 1953b. Activity in a crustacean ganglion. I. Biol. Bidl. 104, 156-170.
Maynard, D. M., 1953c. Integration in the cardiac ganglion of Homarus. Biol. Bull.

105, 367.
Maynard, D. M., 1956a. Activity in a crustacean ganglion II. Pattern and interaction

in burst information. Biol. Bull. 109 420-436.
Maynard, D. M., 1956b. Direct inhibition in the lobster cardiac ganglion. In ms.
Maynard, D. M., 1956c. Effects of inhibition on interaction in the cardiac ganglion of

lobsters. In ms.
Sherrington, C. S., 1951. Man on his Nature. Cambridge Univ. Press. 2nd ed.
Tasaki, I., and M. Sakaguchi, 1950. Electrical excitation of the nerve fiber. Part II.

Excitation by exponentially increasing currents. Jap. J. Physiol. 1, 7-15.
Terzuolo, C. A., and T. H. Bullock, 1956. Measurement of imposed voltage gradient

adequate to modulate neuronal firing. Proc. Nat. Acad. Sci. 42, 687-694.
Wiersma, C. A. G., 1952a. Repetitive discharge of motor fibers caused by a single im-
pulse in giant fibers of the crayfish. /. Cell. Comp. Physiol. 40, 399-420.
Wiersma, C. A. G., 1952b. Neurons of arthropods. Cold Spring Hrbr. Symp. Quant.

Biol. 17, 155-165.

DIURNAL MIGRATION OF PLANKTON CRUSTACEANS*

Edward R. Baylor and Frederick E. Smith
University of Michigan

The behavior exhibited by certain freshwater and marine plankton in
response to changes of single parameters of the environment, such as
polarization, wave length or intensity of radiation, temperature, pressure,
pH and redox potentials, are of increasing interest because they can be
systematized and subjected to physiological analysis. In general the be-
havior in response to these parameters appears to be concerned with
vertical migration or obtaining food. What effect, if any, these parameters
may have on other behavior patterns is beyond the scope of our present
efforts.

The experiments to be described have been performed by the authors
and William E. Hazen with the help of students and colleagues. Our re-
search strategy consisted of two stages of attack. In the first stage, we con-
centrated our efforts entirely on those qualitative observations which we
felt would delineate the general behavior mechanisms under study. Thus
we postponed the development of quantitative techniques, the second stage
of the attack, until we had completed some preliminary analyses of a num-
ber of the behavior-pattern mechanisms. The quantitative techniques now
under development are particularly important for either elegance of de-
scription or accuracy of analysis, since many behavior patterns of zoo-
plankters are actually the end result of algebraic addition of many appar-
ently random locomotion vectors and velocities, such as appear in ortho-
kinesis, klinokinesis (Fraenkel and Gunn, 1938), or color dances (Smith
and Baylor, 1953). The qualitative results reported here may in all cases
be quantified to the following extent — that more than 75% of the popula-
tion (never less than 100 individuals unless otherwise specified) must
execute the same behavior pattern at the same time in response to the
stimulus. The present paper is a summary of certain of our qualitative
observations.

Rather than survey what many animals do in many diverse situations
we thought it better to attempt a synthesis of what one animal does in
many situations. This tentative synthesis is complex ; for the moment its
purpose is to serve as an object of criticism by this audience as a guide to
more fruitful experiments. We are not at all sure that our interpretations

* This study was supported in part through a grant from the Office of Naval Re-
search. Contract No. NONR 1224(05).

[21]

22 INVERTEBRATE PHYSIOLOGY

are correct, but we feel that they are the simplest we can make so far.
Actually, the behavior patterns may turn out to be even more complex.

In general we have concerned ourselves with two types of behavior —
orientation to food and diurnal vertical migration. First, we shall take up
orientation to food, a series of behavior patterns mediated by the com-
pound eye in Daphnia magna, the large water flea, and in Eubranchipus,
the fairy shrimp. We have particularly concerned ourselves with those
zooplankters which filter suspended phytoplankton. I shall digress mo-
mentarily to point out how suspended phytoplankton alter certain im-
portant parameters of the environment. First, a suspension of unicellular
phytoplankton polarizes a vertical beam of light so the light which is
scattered horizontally is strongly polarized in the horizontal plane. We
presume the mechanism operating to polarize the light is a simple reflec-
tion from the many spherical or cylindrical surfaces presented by the
suspended phytoplankton or bacteria. Clearly, the sizes of particles are
wrong to produce polarization by Rayleigh scattering. Second, the pig-
ments present in many such plant cells absorb relatively more of the short
end of the visible spectrum than the long end. Third, the respiration and
photosynthesis of such plant cells cause a change in the pH of the water
suspending them. With this information, we are in a position to consider
the behavior patterns of those animals orienting to such suspensions as
food objects.

Polarized Light Orientation

Daphnids orient at right angles to a vertical beam of plane polarized light
by swimming back and forth in the beam at right angles to the plane of
vibration of the light. When the geometry of the light beam and of the
compound eye are examined, one sees that a vertical beam of light polarized
in the transverse plane of the animal enters only those facets of the com-
pound eye which are directed upward, forward, or backward. Fig. 1 shows
a highly diagrammatic geometrical analysis of this optical situation. Only
a fraction of the normal complement of cone lenses is shown and these
are directed upward in the anterior-posterior, forward in the dorsal-
ventral, and to the right in the lateral planes of the animal. The pigment
mass in which the cone lenses are embedded is not shown. Polarized light
vibrating in the transverse plane of the animal finds parallel internal re-
flecting surfaces only in those cone lenses ( 1 and 3 of Fig. 1 ) lying in or
near the plane defined by the dorsal-ventral-anterior-posterior lines. Note
that in this case the lateral lenses (2 of Fig. 1) do not permit internal re-
flection, since there is no surface parallel to the plane of vibration of the
light. Since polarized light can penetrate to the inner tip of only those cone
lenses presenting an internal surface parallel to the plane of vibration of

DIURNAL MIGRATION OF PLANKTON

23

the light, then clearly the geometry of the compound eye and the plane of
vibration of the light control which lenses shall have light penetrating to

Fig. 1. A sixty-degree projection of only three cone lenses of
a daphnid compound eye showing various light paths. At A light
is nonpolarized. At B the light vibrates in the transverse plane
of the daphnid. At C the light vibrates in the dorsal-ventral plane
of the daphnid. At D a surface parallel to the light vibrating in
the transverse plane of the animal permits reflection of this light
to photoreceptors at the tip of the cone. At E the surface is vir-
tually perpendicular to the plane of vibration of the light and
hence the light is refracted through the lens to the outside again
without reflection to the inner tip of the cone.

the tip of the cone. The important exceptions to this statement are those
lenses directly facing the light source.

For a horizontal beam of polarized light vibrating in the transverse
plane of the eye, the lenses permitting penetration of light to the tip are
those directed anteriorly, dorsally, and ventrally. The cone lenses directed
laterally do not present parallel surfaces for internal reflection and there-
fore are not illuminated by a horizontal beam of polarized light vibrating
in the transverse plane of the animal. Operationally the behavior pattern
in response to polarized light would appear to orient the animal so that
the light can enter the ventral, anterior, and dorsal lenses only and would
overtly exclude light from the lateral lenses.

24 INVERTEBRATE PHYSIOLOGY

If a horizontal beam of plane polarized light vibrating in the vertical
plane is presented the animal so that the light would enter only the lat-
erally directed lenses, the daphnid makes considerable effort to swim on
its side where the light will be excluded from the laterally directed cone
lenses. Further proof may be adduced by experiments described else-
where (Baylor and Smith, 1953).

In the final analysis the response to polarized light in the natural en-
vironment is a relatively simple sort of positive phototaxis brought about
by an increased intensity of the light entering the cone lenses which are
directed anteriorly, posteriorly, dorsally, and ventrally compared to the
intensity of light entering the laterally directed cone lenses. If a daphnid
is presented with 360 degrees of light, i.e., inside an opal glass globe uni-
formly illuminated from the outside, movement appears completely random
horizontally and vertically.

Color Dances

Color dances are statistical behavior leading to food (Smith and Baylor,
1953) and maintaining the animal within a useful range of its food once
such food is found. This behavior is best seen with a light source from
above giving a uniform intensity over the entire aquarium. Under red light
(6,000 A and over) the population appears calm, the individuals dancing
upright in the water, with a small horizontal vector in their locomotion.
The vertical vector is larger and varies somewhat throughout the popula-
tion. The velocity of such locomotion is quite low, the animals occasionally
appearing to be suspended in the water. Under blue light (5,000 A or
shorter) the population is distinctly agitated, the individuals leaning well
forward in their dance and roaming about with a large horizontal vector
to their locomotion. Velocities are estimated at three to five times the
average velocity in red light.

A change of the direction of the light beam from vertical to horizontal
shows that the color dances are oriented, not to gravity, but to the direction
of propagation of the light. The largest vector and hence the direction of
locomotion in the blue dance is always oriented at right angles to the line
of propagation of the light, while the largest vector in the red dance is
always directed parallel to the line of propagation of red light. When blue
light is introduced from the side, a vigorous wandering occurs in all direc-
tions in the vertical plane perpendicular to the light beam ; when red light
is introduced from the side, the red dance remains a quiescent dance with
a predominant vector, if any, in a line parallel to the direction of the light,
the animals swimming slowly away from the red light and returning more
rapidly at irregular intervals.

Under white light from above the dances are discrete ; at any one mo-

DIURNAL MIGRATION OF PLANKTON

25

ment an individual is either red-dancing or blue-dancing. If the proportion
of energy in the short wave lengths is too high, all the animals will blue-
dance all the time. Within a range of proportions specific for each species,
and probably afifected by several environmental factors such as presence
of food or hunger, the individuals will change spontaneously back and
forth from one dance to the other. The total proportion of time spent
doing each of the dances is extremely sensitive to the proportional energies
of short and long wave lengths. A convincing test of this hypothesis may
be performed by placing, over a dish of daphnids, a shallow Incite tray
having two or more compartments, one of which is filled with clear water
and the other with a phytoplankton suspension. If the dish and overlying
tray of water and phytoplankton is illuminated from above, the organisms
rapidly gather under the phytoplankton.

Color dances have been observed in several freshwater Cladocera —
Daphnia magna, Ccriodaphnia, Moina, and Bosmiita — and also in a fairy
shrimp, Eubranchipiis. We have observed color dances in the following
marine zooplankters : Squilla larvae, two pontellid copepods, and a har-
pactacoid copepod.

COMPOUND EYE

YELLOW
RECEPTOR

o

- RECEPTOR

v-

Ul

z

Ul

—1

:j

_J

*- <

z

1—
<

<

PNI

»-

OC a.

5 •-

"■"

I

rf

< 3:

c»

3

Ul

CO a:

_l

^

►- ►-

OU

<

•^ s^

X <
-

u.

Ul UJ

-J |2
z

*- UJ

z

Ul

oc:

)PLAN
,ITY:

Ul

z
<
_l

PHOTOTAXES

SIGN G| VIGOR

POLARIZED
LIGHT

+

t^ ^

V -|

il

1

ORIENTATION

^

1
1 _*

DANCE PATTERNS

"i

ROVING

FEE DING

1*7

•

1 *

g

1-

z

Ul

1-
z
u

CO

to

Ul

Ul

u.

a:

li_

o.

HUNGER. IF PRESENT

u.

_

jr\

Fig. 2. Statistical behavior associated with food.
Arrows indicate relationships.

26 INVERTEBRATE PHYSIOLOGY

If the compound eye of Daphnia magna is denervated and the animal
is allowed to recover from this operation the color dances are completely
missing.

One might be tempted to conclude that Cladocera living in water with
humic stains would never blue-dance and might hence die of starvation by
never reaching food which might be nearby. However, such is not the
case, since the color-dance responses to incident wave lengths are over-
ruled by hunger or by the presence of food. For example, an animal which
is hungry will blue-dance in all wave lengths of the visible spectrum. If
food in the form of suspended photoplankton is now added to the hungry
population, there will be nothing but red-dancing observable no matter
what the incident wave length.

Fig. 2 summarizes the situation with regard to color dances, showing
that these two behavior patterns are mediated by the compound eye which
has a receptor for long and for short wave lengths and showing that
hunger or the chemical stimulation of food can overrule the light stimulus.

Vertical Migration

Vertical migration is essentially a very complex combination of geotaxis
and phototaxis which is influenced by a number of parameters of the en-
vironment. In various marine and freshwater zooplankters it can be in-
duced by light intensity or wave length, pH, redox poising compounds,
temperature, or pressure. Sometimes all of these parameters are effective
in a single species, such as Daphnia magna (see Fig. 3), and sometimes
visible light has little effect, as for example certain marine zooplankters
of the Inland Waterway of Florida. Since gravity is a constant force in the
environment capable of serving as a behavior cue along with radiant en-
ergy which is less constant, it is not wholly unexpected that geotaxic and
phototaxic behavior patterns have evolved in response to those parameters
of the environment which have diffuse vertical gradients, as for example
light, temperature, pressure, pH, and redox potentials. The gravitational
field of force of the earth has been well exploited by the Cladocera and be-
havior responses to nongradient situations are oriented not to the stimulus
per se but to gravity. A nongradient situation has no spatial dimensions for
cueing an oriented response but the stimulus serves to set off a gravita-
tional response. When one considers thermal, chemical, or radiant energy
gradients of the environment in relation to the size of Cladocera, it is clear
that the change in intensity of the gradient over the length of the animal
is too small to be detected and resolved into directional information on
which to base an oriented taxis. Hence, we have behavior patterns like
geotaxes initiated by chemical or radiant energy stimuli. Radiant energy
sources, on the other hand, may be localized if the animal has a receptor

DIURNAL MIGRATION OF PLANKTON

27

PHOTOTAXIS

KINESIS

Vf:-*t

NAUPLIUS ETE

/lOLET RECEPTOR

:mperature ^^

RESSURE

■^

ORIENT

ATION VIGOR

SIGN6-

^>^ red: -

REDOX

<''?-;•

v..

^<^>

'< ^

CHANGES OF TEMP.

& PRESSURE: DOTH
EYES MUST BE STIM

COMPOUND EYE

BLUE
RECEPTOR

YELLOW
RECEPTOR

ANTENNAL

GRAVITY
RECEPTORS

Fig. 3. Behavior patterns mediated by the naupHus eye, the compound eye, and
the antennal gravitational receptors. Arrows indicate relationships. Signs on the
arrows refer to the positive and negative taxes. Words on arrows indicate intensity
of stimulus. Broken lines show relationships present in the Cladocera, absent in
Eubranchipus.

backed up by an opaque curtain. Gravity receptors appear to be localized
in the swimming antennae of Daphnia magna and monitor this force only
when the animal is not swimming. (Grosser, Baylor, and Smith, 1954).
Rose (1925) has obtained results with pH and temperature similar to ours.

The Nauplius Eye and Vertical Migration

The nauplius eye of adult Crustacea is smaller and of less importance
as a taxononiic character than the compound eye. Thus, its functional sig-
nificance has been somewhat neglected. Unpublished data of Lockhead
(personal communication) indicate that it has photosensitivity in the fairy
shrimp. It was completely unanticipated that the nauplius eye of daphnids
would be concerned with responses to pH, pressure, temperature, and
redox potentials as well as being sensitive to ultraviolet and x-rays.

28 INVERTEBRATE PHYSIOLOGY

However, the series of behavior patterns concerned with vertical migra-
tion described below can be shown to be associated wdth the presence of
the nauplius eye in Daphnia magna. Destruction of the nauplius eye by
microsurgery or by needle beam x-rays results in complete disruption of
the behavior pattern. Tests have been carried out in the case of each be-
havior pattern described for Daphnia magna. Only those behavior patterns
associated with vertical migration are disrupted.

Ultraviolet and violet light produce positive geotaxis and a negative
phototaxis which is mediated by the nauplius eye. For example, if a popu-
lation of daphnids is suddenly exposed to a horizontal beam of violet light,
there results an immediate movement of the population down and away
from the light beam. The proportions of the down and away vectors of the
locomotion depend on the w^ave length of the horizontal light beam, its
intensity, and the state of accommodation of the daphnid population to the
intensity of the beam. The down vector is greater the shorter the wave
length and is roughly proportional to the intensity. A cinematographic
analysis of this stimulus-response situation is being carried out. We may
summarize this behavior as follows :

( 1 ) Dimming wave lengths either shorter or longer than 5,000 A can
provoke upswimming.

(2) Brightening wave lengths shorter than 5,000 A can provoke down-
swimming.

(3) Brightening wave lengths longer than 5,000 A cannot provoke
downswimming but may cause a slow spread downward.

(4) Shifting the spectrum of light toward blue without changing the
intensity can provoke downswimming.

(5) Shifting the spectrum of light toward red without a change in in-
tensity can provoke upswimming.

Fairy shrimp show behavior similar to that indicated for daphnids.

Light and Vertical Migration

The vertical response to light stimuli of the inshore plankters of the
Inland Waterway of Florida^ is considerably different from that of the
freshwater plankters. An increase of light intensity may induce down-
swimming and a decrease in intensity may induce upswimming but the
intensity changes required are very much greater. A 95% decrease in in-
tensity (compared to 1-10% for freshwater Cladocera) is required to elicit
a clear, though usually temporary response. Most species having a hori-

1 The marine work reported here was performed at the Marineland Laboratory,
Marineland, Fla.

DIURNAL MIGRATION OF PLANKTON 29

zontal phototaxis show light-induced geotaxis ; however, some species
(medusae, Mnemiopsis, and a pteropod, Creseis) which have no lateral
phototaxis show light-induced geotaxis. Certain marine organisms may-
show a reversal of behavior outlined above. For example, Callinectid
megalops larvae, usually being geopositive to increased intensity, may
sometimes show the reverse pattern and swim upward in response to an
increased intensity. In general, if inshore plankters are geopositive in the
dark ( in response to factors other than light ) , they will upswim to an in-
crease in light intensity, while, if they are geonegative or show neither
tendency in the dark, they will downswim to an increase of light intensity.
In the presence of long wave lengths of light, the addition of short wave
lengths either has no effect at all, or else the organisms downswim with
the addition or upswim with the subtraction of short wave lengths.

Light and Geotaxis

There are two effects of light on geotaxis. One eff'ect is associated with
wave length while the other eft'ect is associated with intensity. We have
not determined the action spectrum of the daphnid eye accurately in the
violet region and hence these effects may be identical. In our tests geotaxis
is distinguished from phototaxis by the use of a horizontal light beam.
Thus phototaxes will be directed toward or from the light source while
geotaxes are directed up or down. The behavior pattern here is readily dis-
tinguished from blue-dancing in a horizontal beam where movement is also
at right angles to the beam.

The effect of an increase of intensity is downswimming (see Fig. 3).
Provided the animal is accommodated to the intensity of the horizontal
beam, the etiect is almost a pure downswimming in the near ultraviolet
with a very small negative phototaxis. The eft'ect of shortening the wave
length of a horizontal beam of violet light from a monochromator is also
downswimming. If one assumes that the peak of the violet action spectrum
for Daphnia magna is 3,000 A or below, then the response to either an in-
crease in intensity with the wave length held constant or a decrease in
wave length is indistinguishable operationally. Both may appear as an
increase of intensity in the experimental animal. Conversely, either
dimming the intensity or increasing the wave length will produce up-
swimming, provided the animal is accommodated to the initial intensity
and wave length. Thus, geotaxis in response to light stimuli appears asso-
ciated with phototaxis in the integrated behavior of Daphna magna. In
the natural environment the two behavior patterns of phototaxis and geo-
taxis operate together and in the same direction in response to the same
stmulus although one is oriented to light and the other to gravitv (see
Fig. 3).

30 INVERTEBRATE PHYSIOLOGY

Temperature and Phototaxis

Since the response to a vertical light beam under various conditions of
temperature is up or downswimming, the following tests were all con-
ducted in a horizontal light beam to eliminate the possibility of a light-in-
duced geotaxis (which is discussed above). There are at least four
separate and independent effects of temperature on the phototaxic re-
sponse of Daphnia magna and Eubranchipus to white light, which are
summarized under "Temperature" and "Changes of Temperature" in
Fig. 3. Daphnids reared at 15° C and exposed to a horizontal light beam
show vigorous positive phototaxis at temperatures of 0° to 5° C. The
organized photopositive response is not a prolonged one at very low tem-
peratures in the presence of light intensities of greater than 10-foot candles,
since it is superseded by a somewhat violent paralytic seizure ending in
death. The paralytic seizure may be induced by similar light intensities at
room temperature in the presence of 1 :10'^ acetylcholine and eserine. On
the other hand the paralytic seizure at low temperatures may be avoided
entirely by making the water in which the daphnids swim 1 :10'^ atropine
(Baylor, 1954). At the opposite end of the temperature scale for the
daphnid, namely 15 degrees above its normal environmental temperature,
there is a vigorous negative phototaxis in response to a horizontal beam
of white light. In view of the fact that the direction of the intensity vectors
of light in the normal environment of such animals is vertical and that ex-
tremes of temperature are hazardous, the positive phototaxis in cold water
and the negative phototaxis in warm water may be considered to have
some adaptive significance.

Small but continuous decrements of temperature can significantly in-
crease the velocity of downswimming in response to any stimulus which
will initiate downswimming. Thus a population of daphnids will continue
to swim downward through a thermal gradient at high and svistained ve-
locities in spite of the fact that the stimulus which initiated the response
may have decreased in intensity. The downswimming pattern performed
through such a thermal gradient will be maintained until the population
leaches 10 to 15 degrees below normal environmental temperature, where
the incipient positive phototaxis in response to cold temperature halts
the downward migration.

Small but continuous increments in temperature can significantly in-
crease the velocity of upswimming, once such a response is initiated by
any of a number of stimuli already discussed. The pattern is now exactly
the converse of that described above for decrements of temperature. Once
the population of daphnids is started swimming upwards through such a
thermal gradient, they will continue to swim upwards till an upper tem-
perature is reached which will halt the pattern.

DIURNAL MIGRATION OF PLANKTON 31

While the adaptive significance is not so obvious here, we hypothesize
that the value may lie in carrying out the pattern of vertical migration (for
whatever value it may have) in spite of the fact of accommodation to the
light stimulus and in spite of the fact that appropriate intensity and wave
lengths decrease rapidly with depth. Conversely, it allows a rapid approach
to the surface waters in spite of rapidly increasing intensities of light.

In the zooplankters of the Inland Waterway of Florida vertical responses
to temperature were also observed. However, the range of useful tempera-
tures was considerably limited. At pH 8.05 a shift of minus 0.5 degrees
from 31° C would induce upswimming, while a shift to 3L5° C would in-
duce downswimming. A shift of as much as 5 degrees would irreversibly
destroy the response to either pH or temperature.

Temperature and Geotaxis

In the dark, positive geotaxis results from immersing Daphnia magna
in water 10 to 15° C above the temperature at which it was reared. This
behavior pattern is a parallel of the phototaxis described in response to
temperature. Small increments or decrements of temperature will cause
the sustained high velocities of locomotion described for phototaxis in re-
sponse to these stimuli for Cladocera only under white light.

Hydrostatic Pressure and Vertical Movement

In general, the effects of pressures on certain marine and freshwater
zooplankters appear to be identical and to fall into three categories.

First, there is the efifect of pressures from 750 to 6,000 psi. Pressures of
these magnitudes cause paralysis followed by death within minutes, par-
ticularly at low temperatures. In Daphnia magna the efifect appears to be
on the choline esterase system, since it can be alleviated by the use of atro-
pine in physiological concentrations. It is not known whether or not the
same protection is afforded marine organisms.

Second, there is the effect of pressure ranging from 30 to 500 psi.
Throughout this range, pressure can cause upswimming depending on the
temperature or pH of the water within the pressure bomb. For example,
the megalops larvae of portunid crabs swim downwards at all tempera-
tures above 30° C when the pH is 8.1, but can be made to swim upwards
by application of pressure. The higher the temperature the more pressure
is required to induce upswimming. At 40° C the response is poor and
heat death ensues in minutes. Death is not particularly delayed by high
pressure.

The third category is small pressure changes in a low pressure range,
from partial vacua to 10 psi. For the Florida Inland Waterway zooplank-
ters such small changes in hydrostatic pressure were found to produce

32 INVERTEBRATE PHYSIOLOGY

vertical movements which compensate by a change in depth for the small
pressure change. Thus, partial vacua of 6 inches of sea water produce a
downward movement of approximately 6 inches. The converse operates
for positive pressures unless the animal is already at the surface, in which
case it still makes an effort to swim upward. The results of small positive
pressure changes are shown below :

PRESSURE INCREASES REQUIRED TO PRODUCE UPWARD
SWIMMING IN VARIOUS ANIMALS

Animals Pressure Increase

(psi)

Small pelagic annelids 2

Hydrozoan medusae 5-10

Temora 2

Mncmiopsis 5

Pteropods 0.1-5.0

Pontella 0.1

Sagitta 9

Pleiirobrachia 9

Paleomonetes larvae 9

Acartia 9

Cenfropages - 9

CalUncctes megalops & zoea 9

Lucifer larvae 9

Pencils larvae 9

The lowest threshold for positive or negative pressures in the Florida
marine forms was 3 inches of sea water. Such a low threshold to pressure
raises the question of mechanism. One possibility is that these organisms
possess small gas bubbles, although none is discernable with high dry
microscopy in the glass-clear medusa Pleurobrachia, or Sagitta. A test of
this hypothesis is available. If the pressure is so reduced that boiling occurs,
any gas bubble present would be expected to expand enormously and float
the organism to the surface. In none of the species tested did this happen.
The pressure decrease accompanying evacuation of a suction flask induces
a strong positive geotaxis, and the populations literally grovel on the
bottom while the water all around them is boiling at room temperature.
Gas bubbles, in the ordinary sense, at least, are not the mechanism. No
fresh-water organisms responded to this test either.

Small continuous increments of pressure significantly increase the ve-
locity of upswimming of the Florida marine plankters until a pressure
of 15 to 30 psi is reached. When the pressure was maintained at these
levels for a few minutes, there resulted a destruction of the pressure-sensi-
tive mechanism. Recovery did not occur within a week in which these
animals were kept alive in the laboratory. Small continuous decrements of
pressure also increase the downswimming velocity until the sea water boils.

DIURNAL MIGRATION OF PLANKTON 33

No destruction of the mechanism results. Hardy and Bainbridge (1951)
noted low-pressure responses in marine plankton.

pH and Phototaxis
The influence of pH, or carbon-dioxide-induced changes in pH, on
daphnid behavior has been observed often since Jacques Loeb ( 1904) re-
ported reversal of phototaxis when he profligately poured beer into his
experimental aquaria of daphnids. In general we have confirmed these ob-
servations ; high pH values induce positive phototaxis while low pH values
induce negative phototaxis. The crossover point from positive to negative
phototaxis appears to depend to some extent on the pH at which the or-
ganisms were reared. Daphnids reared at pH 8.0 become vigorously posi-
tive at pH 8.5 and vigorously negative at pH 7.0. The adults are apt to be
normally photopositive to white light at pH 8.0 when fed on green algae.
Conditions of culturing may change the sign of taxis ; these will be dis-
cussed under redox potentials. Immature forms often exhibit the reverse
photo- and geotaxis of the adults. This fact has often been observed and is
believed by Skadowsky ( 1939) to be a function of metabolic rate.

pH and Geotaxis

Geotaxes in response to pH changes occur in the dark (vertical distri-
bution of a population of daphnids is easily measured by a flash photo-
graph). Results of such studies show that Daphnia magna swims upward
in the dark at pH 9.0 and swims downward at a pH of 7.0 or below. Such
geotaxis may occur in addition to the phototaxis in response to pH since
it takes longer for a population to reach equilibrium position in the dark
than in the light.

Probably the most startling induction of geotaxis by pH is to be observed
in the Inland Waterway of Florida, where small changes of pH are of pri-
mary importance in controlling the vertical migration of the zooplankton.
Special care must be taken in the collection of the plankton sample to avoid
a change of more than 2° C or 0.7 pH units. Either of these will de-
stroy the response to pH or temperature. Vertical-movement tests in re-
sponse to pH were run in the laboratory on three representative species :
a hydrozoan medusa, a pontellid copepod, and blue crab megalops and
zoea larvae. Some efi^ort was made to obtain relatively pure populations
of the three species tested for pH responses ; this was, however, imprac-
tical so that other unidentified organisms were often included in the tests.
The responses of these unidentified organisms were identical with those
of the carefully isolated species studied. When the water temperature was
maintained at 31° C and the pH was varied through the range of 8.0 to
8.1, there was a uniform response of the species tested ; a pH of 8.0 caused
persistent upswimming while changing the pH to 8.1 induced persistent

34 INVERTEBRATE PHYSIOLOGY

downswimming. Light has Httle effect on this reaction. The same responses
could be obtained with more marked shifts of pH, although departures of
more than 0.5 units from the normal environmental pH quickly destroyed
the mechanism. The daytime pH of the Inland Waterway is 8.1, the noc-
turnal pH 8.0. There is sufficient turbulence to prevent a gradient. The
pH values were obtained with a Beckman Model B pH meter equipped
with titrating electrodes. Each reading was preceded by a buffer and
temperature check.

Redox Potentials and Phototaxis

Oxidizing substances to which daphnids are permeable cause positive
phototaxis while reducing substances to which daphnids are permeable
cause negative phototaxis. The redox poising compounds utilized for
such tests are largely those intravitam stains which readily penetrate
daphnids, some of which appear preferentially concentrated in the cells
of the nauplius eye. Additionally, catechol and cysteine were used. Catechol
(10"^M) having an E'o of + 0.33 volts produces strong upswimming
while (10""M) cysteine having an E'o of —0.14 volts produces strong
downswimming (Smith 1954). The mid-point of the range where no
effect occurs for a population of DapJinia reared on green algae appears to
to about -f- 0.045 volts E'o, while the mid-point for a culture reared on bac-
teria appears to be lower. The important point here is that the majority
of organisms reared on algae appear to be photopositive, while those reared
on bacteria having a low redox potential are photonegative. We feel that
these results will largely explain the photopositivity reported for daphnids
by Clarke (1930) and the photonegativity later reported for the same
species by the same author (1932). A summary of our observations on
phototaxis and redox poising compounds is given in Fig. 3.

REFERENCES

Baylor, E. R., 1954. The interaction of light and drugs in the cold narcosis of Daphnia.

Proc. Fed. Soc. for Exp. Biol. 13, 543.
Baylor, E. R., and F. E. Smith, 1953. The orientation of Cladocera to polarized light.

Am..Nat.B7,97-\0\.
Clarke, G. L., 1930. Change of phototropic and geotropic signs in Daphnia induced by

changes of light intensity. /. Exp. Biol. 7, 109-131.
Clarke, G. L., 1932. Quantitative aspects of the change of phototropic sign in Daphnia.

J. Exp. Biol. 9, 180-211.
Fraenkel, G. S., and D. L. Gunn, 1940. The Orientation of Animals. Clarendon Press,

Oxford.
Grosser, B. I., E. R. Baylor, and F. E. Smith, 1953. Analysis of geotactic responses

in Daphnia magna. Ecology 34, 804-805.
Hardy, A. C, and R. Bainbridge, 1951. Vertical migration of plankton animals. Na-
ture 168, 327-328.
Loeb, J., 1904. The control of heliotropic reactions in fresh-water crustaceans by

chemicals, especially CO,. Univ. of Calif. Pub. in Physiol. 2, 1-2.

DIURNAL MIGRATION OF PLANKTON 35

Rose, M., 1925. Contribution a I'etude de la biologic du plankton. Arch de Zoologie
Experimentale et Generale 64, 387-542.

Skadowsky, S. N., 1939. Physiological analysis of phototaxis in daphniae (Daphnia
pulex). Uchenye Zapiski Moskovskyo Gosundarstvcnnjo Universiteva. 33, 237-246.

Smith, F. E., 1954. An analysis of the interaction of pH and redox in the diurnal mi-
gration of Daphnia. Proc. Fed. Soc. for Exp. Biol. Res. 13, 543.

Smith, F. E., and E. R. Baylor, 1953. Color reponses in the Cladocera and their ecol-
ogical significance. Am. Nat. 57, 49-55.

PREY-PREDATOR RECOGNITION IN THE LOWER
INVERTEBRATES*

L. M. Pass A NO
Yale University

Above the broad herbivorous base of the ecological pyramid of animals,
whether one is considering the land or sea, are the lesser numbers of car-
nivores. Most of these animals must be below the apex, must feed on their
animal prey yet in turn be fed upon by other organisms. The success of
these animals means that a judicious balance between recklessness and
care, efrective capture of food versus preventive caution in the presence of
enemies, has been achieved. The organism's behavior patterns are adapted
to meet these dual needs. The animal's morphological and physiological
equipment dictates the form of this behavior, its complexity, and its adapta-
bility. Obviously, the success of this balance can be measured by the abund-
ance of the group under consideration.

Without being able to explain the physiological mechanisms that are
involved, nevertheless we can see in the elaborate and specialized sense
organs, in the many and varied motor responses, and above all in the com-
plexity of the central nervous systems, the physiological and morphological
machinery that determines the success of most carnivores. But, especially
to those interested in comparative physiology, a few groups of relatively
simple metazoan carnivores, such as many of the coelenterates and the
free-living flatworms, pose an immediate problem. These animals do not,
as a rule, possess elaborate sense organs. Their tissues, as judged by such
criteria as regenerative powers or specialization of function, do not appear
to be completely subordinate to the entire organism. Above all, as far as
the coelenterates are concerned, they have no central nervous system. We
are unable to evoke any "black box" of suitable complexity to account for
the behavior pattern that we can so easily observe. With the relative sim-
plicity of both behavior and morphological equipment, it is not surprising
that these organisms constitute an immediate challenge to the physiologist.

As the title of this paper indicates, only one aspect of these analytical
problems will be dealt with here, the recognition and distinction of prey
and predator. We shall try to determine how far, in the present state of
our knowledge, it is possible to explain these behavior patterns in physio-
logical language. Let us admit at the outset that this attempt cannot be too

* This paper is dedicated to Alexander Petrunkevitch in honor of his eightieth
birthday.

[37]

38 INVERTEBRATE PHYSIOLOGY

successful. In spite of the work of such deservedly famous investigators
as Romanes, Loeb, Pearl, Jennings, Parker, Koehler, and Pantin, we still
have very little of the basic physiological information that we need before
we can attempt any reasonably complete explanation. For instance, we
shall see that chemoreception plays an important role in these responses ;
yet the receptor units detecting these environmental clues have not often
been identified. But if any generalization can be made, it is that various
quite simple metazoans code their sensory input, so that relatively weak
stimulation of mechano- and chemoreceptors leads to food-capture activi-
ties, whereas strong stimulation leads to defensive reactions. Even in such
aganglionic animals as sea anemones, there is no single form of stimulation
that invariably leads to withdrawal or defensive responses.

While many of the invertebrate phyla include species which lead an
active, carnivorous existence, hunting their food, it is often impossible to
obtain from the literature sufficient information both on their natural prey
and on the predators which, in turn, feed on them. One might imagine that
the success of certain forms with highly specialized diets, such as the
gastropod Archidoris which feeds on certain sponges, or many of the nudi-
branchs (Hunt, 1925) which feed on hydroids, is determined by the availa-
bility of their food rather than by predation. But certain of the latter group
clearly exhibit withdrawal behavior when they are vigorously prodded,
suggesting that their protective devices of noxious taste, slime, and hydroid
nematocysts may not always be sufficient deterrent to hungry fish or crabs.

A number of the carnivorous mollusks, however — and I am thinking
here of such gastropods as Natica, Murex, Nassa, and Urosalpinx — are
known to constitute an important part of the diet of fish (Hunt, 1925;
Hancock, 1955) and starfish. Most of these gastropods can be distinguished
from their herbivorous cousins by the reduction or loss of the crystalline
style (Yonge, 1930). It is interesting to note that these snails are not
generally very restricted to certain prey ; Urosalpinx, for example, might
as fairly be termed the barnacle-drill or the mussel-drill as the oyster-drill.

From the work of Copeland (1918) it seems quite certain that chemore-
ception is important in guiding these mollusks to their food. He showed
that in Nassa (= Alectrion) the sensory organ, the osphradium, samples
the water which enters the siphon. The animal responds to dilute food ex-
tracts by increasing its rate of locomotion, extending its proboscis, and
orienting the direction of movement towards the increasing chemical con-
centration. In addition to this "nasal" response, moreover, these snails
show "taste" responses to relatively strong food solutions with tentacles,
underpart of head, and anterior end of foot. There is no evidence that the
eyes play any part in food capture.

It would be interesting to know whether or not tactile cues, in the ab-

PREY-PREDATOR RECOGNITION 39

sence of chemoreception, can also be used by these animals. If they can
differentiate, by touch alone, bivalves or barnacle shell from a rough rock,
this would seem to imply a fair degree of central integration or very specific
sense receptors. On the other hand, these animals may be able to find their
prey from the relatively low concentrations of waste products that are being
excreted into the water, although this would imply that they possess the
degree of sensitivity and selectivity that appears to characterize many sym-
bionts (Davenport, 1955). It would seem most unlikely that such stimuli
as were used by Copeland (extracts of fish and oyster meat) would be the
naturally attracting substances for the gastropods which feed on live bi-
valves, or such forms as Scaphander which swallow whole prey (Hunt,
1925).

The best known cases of escape reactions among the invertebrates to be
be considered here involve the reactions of certain moUusks in the presence
of starfish (Bullock, 1953, for review). Nassa, which is probably more of
a scavenger than a carnivore restricted to living prey, shows a classic and
peculiar leaping withdrawal when touched by certain starfish. While this
example has not yet been sufficiently analyzed to preclude a role of tactile
stimulation of the snail by the predator's tube feet (compare Dakin, 1910,
and Hoffmann, 1930) , it seems quite certain that strong chemo-stimulation
by substances on the surface of the echinoderm are the primary cue. Re-
cently Heinsohn ( 1955 ) , looking at this response, reported a violent twist-
ing, followed by rapid locomotion, by the snail Caliostoma responding to an
extract of Pisaster. The total response lasted at least 30 minutes, during
which time the animal moved over 1.6 meters. Interestingly enough, there
seems good reason to suppose that other groups of animals are equally
sensitive to substances on the integument of certain starfishes. Hancock
(1955) mentions the toxicity of Solaster, while I have observed what may
be the same response in several species of ophiuroids to Solaster and Derm-
asterias. The latter star is particularly effective in causing the extraordi-
nary escape response of the sea anemone Stomphia, to be mentioned later.

While this relative wealth of information is available concerning possible
recognition and escape reactions of gastropods to starfish, apparently
nothing is known about similar behavior initiated by cues from fish ; yet it
would seem (Hunt, 1925) that a variety of bottom fish form the chief
predators of these snails. Clearly, however, the general pattern of gastro-
pod activities — movement away from well-lit areas for some species, or
crowding into the intertidal or even supratidal zones, the occupancy of
crevices, and so forth — all tend to aid these mollusks in avoiding their
enemies.

Two complicating factors in any physiological analysis of behavior are
the possibilities that conditioning may occur or that changing "physio-

40 INVERTEBRATE PHYSIOLOGY

logical states" may affect the responses of the organism to specific stimuli.
While I see no sound reason to exclude the possibilities of conditioning, or
learning, in the gastropods, I know of no demonstration of this. But the
efTect of changing states seems to be present. Forty years ago Wenrich
(1916), working with the fresh water mussel Anodonta, clearly demon-
strated that a variety of internal conditions changed their sensitivity to
light. Besides the unexplained but omnipresent "individual differences,"
the presence of eggs or embryos, of foreign material in the mantle chamber,
and whether the animal had been stimulated previously or not — all these
conditions had definite effects. As will be noted later, this modifiability of
responsiveness is to be found throughout all the lower metazoans. For
instance. Gee (1913) maintained that the internal state of leeches —
whether they were hungry or satiated — profoundly influenced their food-
finding behavior.

Gee emphasized the importance of "random movements" in the leech
activity pattern. He postulated that the central nervous system of these
annelids had two main functions: (1) the production of "spontaneity,"
and (2) adaptation to constant external stimuli. It is clear that he assumed
that the central nervous system integrated all sensory input and that the
reaction of the organism was the resultant of these concurrent stimuli.
Perhaps it is even more than this ; it should be remembered that Copeland
(1930) and Copeland and Brown (1934) convincingly demonstrated a
case of conditioning in the polychaete Nereis, in which the normal positive
response to food extracts was duplicated by touching the anterior end of
the worm, usually a withdrawal stimulus.

Copeland and Wieman (1924) earlier had demonstrated the normal
chemokinetic feeding behavior of these sand worms, and it is probable
that this is general in the carnivorous polychaetes. On the other hand, while
defensive (escape) responses may be initiated by strong or specific
chemical stimuli, it is clear that these animals normally depend on sensitive
vibrational or tactile receptors, synapsing with giant fibers for this use.
An easily observable example of these elegant responses can be seen in
the "quick-as-a-wink" withdrawal of the tubeworm filterers such as the
sabellids.

It is an obvious truism, in general resume of the behavior patterns of
these animals, that chemical and vibrational or tactile cues are the im-
portant external stimuli enabling effective prey-predator recognition. How
much the disturbances engendered by prey movement, as against the dem-
onstrated role of weak chemical stimuli, ensures the success of the preda-
cious carnivores has not yet been determined. Clearly, more knowledge is
necessary, both at the behavioristic and physiological levels, before specific
analyses can be made. However, it does seem true of all of the organisms

PREY-PREDATOR RECOGNITION 41

under consideration that visual stimuli do )iot directly guide either their
defensive or predaceous behavior. Rather, those forms which have photo-
reception appear to use this information to direct or modify overall activity
patterns, and to keep them within the generally restricted environmental
niche to which they are adapted.

This sort of resume of the physiological mechanisms guiding such com-
plex organisms as annelids and mollusks seems, and is, very naive. When
one comes to the free-living flatworms, however, such simplifications may
begin to come somewhat closer to validity. Thanks largely to the classic
observations of Pearl ( 1902) and the careful analysis of the role of chemo-
reception by Koehler (1932), a considerable body of information is avail-
able concerning planarians. These carnivores are guided to their prey
mainly by chemoreceptors,. utilizing separate sensory units for weak and
strong stimuli (Wulzen, 1917). As in the gastropod mollusks, these units
appear to be the analogues of vertebrate smell and taste receptors, respec-
tively. If the "physiological condition" of the animal allows it, weak
chemical stimuli will initiate and orientate a positive gliding movement.
In some species this movement must be against the current as well as along
a chemical gradient (Koehler, 1932). Strong illumination may prevent
this positive response. At some point when the animal is close to the source
of the chemical stimulus, other chemoreceptors, generally located on the
proboscis or grasping organs (Redfield, 1915; Wilhelm, 1915) are
sufficiently stimulated to assume control of the behavior pattern, changing
a seeking behavior to a feeding behavior. Positive-orienting reactions may
also be elicited by mild tactile stimuli. The specific feeding reflex may re-
quire both stimuli concurrently, or else a strong chemical stimulus may be
sufficient.

Pearl (1902) emphasized that positive responses to both chemical and
mechanical stimuli were abruptly changed to negative responses when the
strength of the stimuli was increased beyond some arbitrary and variable
point. For instance, fairly strong unilateral tactile stimulation of the an-
terior end of the planarian would cause a turning away from that side,
extension of the stimulated side. This extension, according to Pearl, is due
to contraction of the circular, dorsoventral, and transverse muscles. A
somewhat weaker stimulus, however, causes a turning towards that side,
due to the contraction of the longitudinal muscles. Thus, according to this
worker, neither "weak" nor "strong" stimuli lead to a crossing over of the
reflex to the opposite side. Also, both chemical and tactile stimuli lead to
either of these responses. Pearl's positive and negative responses, depend-
ing on their strength.

In addition to these two groups of receptors on the proboscis and the
anterior lateral regions, there are receptors on the posterior end of these

42 INVERTEBRATE PHYSIOLOGY

planarians. Strong stimuli of either kind cause a reflex "escape" activity
pattern, with the "ghding" movement changing to a "crawHng" movement.
This reflex appears in an all-or-none manner, not showing differential
reflex behavior to weaker stimuli.

It is a common observation that there is considerable variation among
individual worms to standard intensity stimuli. This feature appears to be
a commonplace of planarian behavior; not only do different individuals
vary, but the same animal shows opposite responses on different occasions.
Pearl suggested that this variation was due to varying "physiological or
tonic conditions" of the organism. He believed that the function of the
nervous system is to "preserve tonus," and claimed that if the "tonus" is
gone it is very hard to get positive reactions. This interpretation would give
an inhibitory or depressing role to the central nervous system. Against this
idea is the result of removal of the "cephalic ganglion" of the polyclad
Leptoplana (Hovey, 1929), which causes it to lose all tactile reflexes en-
tirely, becoming less sensitive rather than more sensitive to stimuli. How-
ever, the noticeable differences between triclad and polyclad nervous sys-
tems (Turner, 1946) may make this seeming contradiction meaningless.

As in the case of annelids and mollusks, it seems clear that more must
be learned about the reflex physiology of these organisms before an attempt
can be made to explain their behavior patterns. How much of the be-
havior is dependent on the "cephalic ganglion" and how much is locally or
regionally autonomous? Furthermore, can the modification of stimulus
threshold by changes in the internal conditions, such as previous feeding
(Pearl, 1902) or lack of "symbiont" Hydra nematocysts (Kepner et al.,
1938), be explained without recourse to postulating an integrating cen-
tral nervous system ? By demonstrating a conditioned reflex in Leptoplana,
Hovey (1929) apparently showed that the "brain" of these animals has
the necessary functional complexity. His localization of this conditioning
in the cephalic ganglia is not convincing, however.

For three distinctly different sorts of reasons, the coelenterates differ
from the other phyla that we are considering here. First of all, they have no
well-defined nervous ganglia to exercise overriding control of their
nervous system.^ Secondly, we have reason to feel closer to an analysis of
their behavior in physiological terms than in the case of other phyla ; in
part, this is simply an expression of their apparent simplicity, but it is also
a tribute to the work of many famous biologists, among them being Ro-

1 The marginal sense organs of scychozoans may be synaptic areas of a sort. In a
series of important papers, Horridge (1955a, b) presents evidence for the independence
of two conducting systems in hydromedusae, interconnected with polarized synapses.
He also suggests that ". . . the ring nerves, which have ganglion cells all along their
course..." cause inhibition of swimming activity during feeding activity (1956),
Thus the "ring nerves" may be acting as an integrating center in these jellyfish.

PREY-PREDATOR RECOGNITION

43

manes, Loeb, Parker, and Pantin. Finally, these generally carnivorous
animals are either, for our purposes of the moment, sessile forms, or
passively floating or swimming organisms ; in other words, they do not
"hunt down" their prey. The well-known "fishing" behavior shown by
certain medusae, the trachyline medusa Gonionemus (Yerkes, 1902) for
example, seems more nearly related to the "slowly" waving tentacles of
sea anemone or hydra than to any example of directed hunting.

Pollock (1883), a colleague of Romanes, appears to have been the first
to recognize that anemones respond to weak chemical stimulation. His
observations were confirmed and extended by Nagel (1892), Loeb (1918,
for resume), Jennings (1906), and Batham and Pantin (1950a,b). The

•Q^

f

Fig. 1. Appearance of Mctridium during various phases: a, specimen ingesting
food; b, typical appearance J4-J^ hour after food ingestion; c, d, swaying movements
6 hours after food-extract stimulus ; e, f, g, successive stages in antiperistaltic con-
striction, leading to defecation; and h, "shrivelling." From Batham and Pantin,
1950b.

latter authors, in giving us the most complete available analysis of the
various activities of sea anemones we have, in Metridium, differentiated
different "phases" of "inherent activity"^ that may follow feeding or
stimulation by food extracts without subsequent reward. As a result of

2 By "inherent activity" these authors mean that ". . . the activity is an observed
property of the animals which does not arise directly from external stimuli" (1950a,
p. 299) . It is clearly not a simple chain reflex.

44 INVERTEBRATE PHYSIOLOGY

chemical stimulation, the animal may ( Fig. 1 ) first of all expand the disc,
then sway, then show a lengthy parietal contraction, then distend, then
defecate, and finally "shrivel." Here, then, we find an elaborate sequence
of behavior "phases" which at their start, at any rate, show apparently
purposeful food capture behavior. Is it any wonder that earlier workers
often wished to endow these organisms with rudimentary intellect ?

The actual capture and swallowing of prey is largely effected by the
anemone's so-called (Parker) "independent effectors," the nematocysts,
cilia, and mucus glands. All of these respond to chemical stimuli from the
external environment (Parker, 1917a; Pantin and Pantin, 1943), al-
though it is probable that nematocysts, at any rate, must also be tactually
stimulated. Moreover, weak tactile stimulation of the tentacles may also
be involved in moving food towards the mouth.

Most anemones protect themselves from forms preying on them simply
by contracting into a compact mass, and sometimes by extending nema-
tocyst-armed threads, the acontia. A few anemones will release their foot-
hold and move slowly away if stimulated for a relatively long time by
"strong" stimuli. Clearly, the terribly potent nematocysts are the chief
protection these animals have against their enemies. Yet how is their be-
havior, and how is the discharge of their nematocysts, controlled so as to
differentiate between foe and food? Ewer (1947) demonstrated that those
types of Hydra nematocysts that are used for defense have their tactile
threshold raised by food extracts, whereas the threshold of the discharge-
triggering mechanism of prey-catching and prey-holding nematocysts are
made more sensitive to weak mechanical stimuli by food extracts (see also
Pantin, 1942). "Integration" of the sensory information thus occurs
immediately at the sense cell-effector level.

For sensory input that goes into the anemone's nervous system there
must be some sort of coding device. As in the planarians, "weak" tactile
or vibrational stimuli elicit food capture behavior, "strong" stimuli cause
withdrawal. On the tentacles of Calliactis, for example, very weak stimuli
will cause a slow discharge of action potentials. Decremental conduction
in the tentacle-disc region will prevent many of these spikes from reach-
ing the synapses into the through-conduction-system, since these synapses
require facilitation — that is, two or more impulses must reach the barrier
within a few seconds — only an occasional impulse will be initiated in the
through-conduction-system.

These occasional spikes are enough to initiate phases of "inherent ac-
tivity" (Batham and Pantin, 1954), to start the expansion of the disc, or
the swaying response, for instance. They are not enough to cause the pow-
erful "quick-closure" protective response, since the innervation of these
muscles is also protected by facilitation requirements.

PREY-PREDATOR RECOGNITION 45

Pantin and I (Passano and Pantin, 1955) recently demonstrated the
great sensitivity of the disc and tentacles to a local pressure change. This
is detected by the mechanoreceptors (whatever element they may be).
Stimuli that are insufficiently "strong" or frequent enough to penetrate to
the through-conduction-system do cause waving movements of the tenta-
cles. On the other hand, tactile receptors of the column and foot regions
are adapted for use as danger indicators. They are gastrodermal, not epi-
dermal, and the cartilage-like mesoglea of Calliactis prevents them from
being excited by moderate stimuli. Small crustaceans can often crawl over
the column without causing any response. Strong stimuli, such as a vigor-
ous poke by a large crab, will discharge the receptors. The resultant excita-
tion immediately gets into the through-conduction-system (since there
probably is no facilitation barrier), and, if further action potentials get
into the system within a few seconds, a fast "protective" closure of the
disc spincter muscle occurs.

Chemical stimuli have the same general effects on sea anemones ; but,
since the substances cannot penetrate the column epidermis and mesoglea,
the extra-oral portions of the animal are remarkably insensitive. Only the
most concentrated solutions are sufficiently strong to lead to contraction
of the "withdrawal" muscles. Yet in a recent series of observations Yentsch
and Pierce (1955) showed that Stomphia coccinea responds violently to
a substance from the surface of a number of starfish, including Hippastcrias
md Dennasterius. I recently noted (Passano, 1955) that Epmctis is also
very sensitive to this substance. The tentacles and disc (of both) are most
sensitive, as would be expected, but stimulation of the column also elicits
responses. This, of course, is immediately reminiscent of the sensitivity
shown by mollusks to starfish, but it is not known whether the active agent
is the same. Sfoinphia shows a striking series of motor reflexes that are
initiated by this stimulus — thrashing back and forth, becoming rigidly dis-
tended, and detaching itself from the substrate. Sund (1955) has recently
shown that exactly the same sequence of reflexes can be caused by a quick
series of six or eight electrical stimuli, perhaps one second apart. It thus
seems reasonable to conclude that some strong facilitation barrier must be
broken down by a series of spikes in the through-conduction-system, but
that, after that, one motor activity triggers another in a continual sequence.

Another such specific chemical sensitivity, but this time evoking feeding
reflexes, has recently been demonstrated by Loomis (1955a,b). He has
shown that 10"^I\I solutions of the tripeptide glutathione cause Hydra to
show vigorous and persistent feeding responses and, furthermore, that this
substance leaks out of Daphnia after these Cladocera have been punctured
by nematocysts. There is some evidence that glutathione is also a specific
stimulating substance for other hydrozoans, but a few preliminary trials

46 INVERTEBRATE PHYSIOLOGY

done in the 1955 invertebrate course at Wood's Hole failed to show any
obvious specificity in anthozoans. This difl^erence might possibly be linked
to the apparent differences in the nerve-net properties of Hydra and those
other members of the phylum that have been investigated (see Pantin,
1950) ; on the other hand, it may be that other fresh- water hydrozoans
are also glutathione sensitive.

A number of different coelenterates have "phases" of rhythmic feeding
activities. According to Jennings (1906) :

. . . green Hydra, undisturbed, show rhythmic activity. Every minute or two it con-
tracts and then extends in a new direction. In this way the animal explores the region
about its place of attachment and largely increases its chances of obtaining food. This
motion seems to take place more frequently in hungry individuals, while in well fed
specimens it may not occur.

Also Torrey (1905) and later Parker (1917b) showed that the solitary
hydroid Corymorpha shows a "nodding" behavior every 2^/^ to 3 minutes,
moving the hydranth down to the substrate by bending the stalk and then
puUing the tentacles along the bottom. This activity is suppressed by caus-
ing water to flow over the tentacles, but is continued by both decapitate
stalk and isolated hydranth, although at a shiver rate, when these parts
are isolated. Here it would look as though inherent rhythms of nervous
activity, in each part, reinforce each other in the intact animal, while higher
levels of activity, such as that caused by continual tentacular stimulation,
inhibit the rhythmic pattern. One would expect that it would be valuable
to analyze these cases more closely in the way Batham and Pantin have
investigated M etridimn.

The most pressing general problem that needs to be attacked experi-
mentally and analytically is : How are these "inherent activity" phases
initiated and modified by changes in the environment, both external and
internal? The truth would seem to include both Parker's (1917b) "...
behavior is chiefly determined by their immediate environment," and
Jenning's (1906) "...complex movements and changes in movement
may occur from internal causes, zvithout any change in the environment"
(my italics). Immediate environmental stimuli not only have a direct
effect on actinian behavior but also lead to slow long-lasting changes in
"inherent activity" and in the "phase" of activity. We must find out how
these environmental changes are detected by the organism and what are
the corresponding changes in the level of nervous activity. To put it an-
other way, what slow changes in the "background level" of nerve-net-
activity occur? Some coelenterates may have inherent "built in" back-
ground level maintainers, such as Corymorpha, the green hydra, and
possibly scyphozoans with tentaculocysts ; and other forms may depend on
external stimuli, such as tide-pool currents (Parker, 1917a) for Metridium

PREY-PREDATOR RECOGNITION 47

or the "sympathetic caressing" tactile stimulation of commensal fish de-
scribed by Gohar (1948).

Finally, I would like to suggest that the study of such problems is not
only of interest to those invertebrate physiologists studying coelenterates,
flatworms, annelids, or mollusks, but may be valuable also to the compara-
tive physiologist and the comparative ethologist. In spite of the obvious
dififerences in the physiological mechanisms concerned (Pantin, 1950),
there are striking analogies between the behavior patterns that I have
described and the innate behavior patterns studied by such workers as
Schnierla and Van der Kloot in insects and Tinbergen and Lorenz in
vertebrates. The latter author, in discussing the "innate behavior patterns"
that he has studied so carefully, points out (1950) that, while they are
superficially reflex-like in being triggered by specific stimulus situations,
they are actually indistinguishable from what von Hoist calls "automatic
rhythms." The threshold of a starving sea anemone to external stimuli
capable of initiating "phases" of feeding activity gradually decreases,
eventually to the point at which such phases begin in the apparent absence
of any special stimulus ; this seems strictly comparable to Lorenz' cases
where "Captive animals, deprived of the normal object or releasing situa-
tion . . . will persist in discharging the same sequences of movements at
a very inadequate substitute object or situation."

It will be of the greatest interest if we can further analyze the different
integrating mechanisms presented to us by these animals, which eithei
have no central-nervous-system "black box," or whose "black box" is so
relatively accessible to investigation and so relatively simple.

REFERENCES

Batham, E. J., and C. F. A. Pantin, 1950a. Inherent activity in the sea-anemone, Me-

tridium senile (L.). /. Exp. Biol. 27, 290-301.
Batham, E. J., and C. F. A. Pantin, 1950b. Phases of activity in the sea-anemone,

Metridiion senile (L.) and their relation to external stimuli. /. Exp. Biol. 27, 377-

399.

Batham, E. J., and C. F. A. Pantin, 1954. Slow contraction and its relation to spon-
taneous activity in the sea-anemone Metridium senile (L.). /. Exp. Biol. 31, 84-103.

Bullock, T. H., 1953. Predator recognition and escape responses of some intertidal
gastropods in presence of starfish. Behavior 5, 130-140.

Copeland, M., 1918. The olfactory reactions and organs of the marine snails Alcctrion
obsoleta (Say) and Busy con canaliculatnin (Linn.). /. Exp. Zool. 25, 177-227.

Copeland, M., 1930. An apparent conditioned response in Nereis virens. J. Comp.
Psychol. 10, 339-354.

Copeland, M., and F. A. Brown, Jr., 1934. Modification of behavior in A^ereis virens.
Biol. Bull. 67, 356-364.

Copeland, M., and H. L. Wieman, 1924. The chemical sense and feeding behavior of
Nereis virens Sars. Biol. Bull. 47, 231-238.

48 INVERTEBRATE PHYSIOLOGY

Dakin, W. J., 1910. The visceral ganglion of Pectcn, with some notes on the physi-
ology of the nervous system, and an inquiry into the innervation of the osphradium

in the Lamellibranchiata. Mitt. cool. Sta. Neapel. 20, 1-40.
Davenport, D., 1955. Specificity and behavior in symbiosis. Quart. Rev. Biol. 30, 29-

46.
Ewer, R. F., 1947. On the functions and mode of action of the nematocysts of Hydra.

Proc. Zool. Soc. Land. 117, 365-376.
Gee, W., 1913. The behavior of leeches with especial reference to its modifiability.

Univ. Calif. Piibl. Zool. 11, 197-305.
Gohar, H. A. F., 1948. Commensalism between fish and anemone (with a description

of the eggs of Aniphiprion bicinctus Riippell.) Publ. Mar. Biol. Sta. Ghardaqa. 6,

35-44.
Hancock, D. A., 1955. The feeding behavior of starfish on Essex oyster beds. /. Mar.

Biol. Ass. [7. /?. 34, 313-331.
Heinsohn, G., 1955. Escape reactions of gastropods to starfish. (UnpubHshed student

report, Invert. Physiol. Course, Friday Harbor, Wash.)
Hoffman, H., 1930. tlber den Fluchreflex bei Nassa. Z. vcrgl. Physiol. 11, 662-688.
Horridge, G. A., 1955a. The nerves and muscles of Medusae. ll.Geryonia probosci-

dalis Eschscholtz. /. Exp. Biol. 32, 555-568.

Horridge, G. A., 1955b. The nerves and muscles of Medusae. IV. Inhibition in
Aequorea forskalca. J. Exp. Biol. 32, 642-648.

Hovey, H. B., 1929. Associative hysteresis in marine flatworms. Physiol. Zool. 322-
333.

Hunt, O. D., 1925. The food of the bottom fauna of the Plymouth fishing grounds. J.
Mar. Biol. Ass. U. K. 13, 560-599.

Jennings, H. S., 1906. Behavior of the Loiver Organisms. Columbia Univ. Press, New
York.

Kepner, W. A., W. C. Gregory, and R. J. Porter, 1938. The manipulation of nema-
tocysts of Chrorohydra by Microstomum. Zool. Anz. 121, 114-124.

Koehler, O., 1932. Beitrage* zur Sinnesphvsiologie der Siiswasserplanarien. Z. vergl.
Physiol. 16, 606-756.

Loeb, J., 1918. Forced Movements, Tropisms and Animal Conduct. Lippincott, Phila-
delphia.

Loomis, W. F., 1955a. Specific qualitative microbioassay for reduced glutathione.
Fed. Proc. 14, 247.

Loomis, W. F., 1955b. Glutathione control of the specific feeding reactions of Hydra.
Ann. N. Y. Acad. Sci. 62, 209-228.

Lorenz, K. Z., 1950. The comparative method in studying innate behavior patterns.
Symp. Soc. Exp. Biol. 4, 221-268.

Nagel, W. F., 1892. Der Geschmackssin der Actinien. Zool. An::. 15, 334-338.

Pantin, C. F. A., 1942. The excitation of nematocysts. /. Exp. Biol. 19, 294-310.

Pantin, C. F. A., 1950. Behavior patterns in lower invertebrates. Svnip. Soc. Exp.
Biol. 4, 175-195.

Pantin, C. F. A., and A. M. P. Pantin, 1943. The stimulus to feeding in Anemone sul-
cata. J. Exp. Biol. 20, 6-13.

Parker, G. H., 1917a. Actinian behavior. /. Exp. Zool. 22, 193-229.

Parker, G. H., 1917b. The activities of Corymorpha. J. Exp. Zool. 24, 303-331.

Passano, L. M., 1955. Unpublished observations.

Passano, L. M., and C. F. A. Pantin, 1955. Mechanical stimulation in the sea-anemone
Calliactis parasitica. Proc. Roy. Soc, B 143, 226-238.

PREY-PREDATOR RECOGNITION 49

Pearl, R., 1902. The movements and reactions of fresh-water planarians : a study in

animal behavior. Quart. J. Micr. Sci. n.s. 46, 511-714.
Pollock. W. H., 1883. On indications of the sense of smell in Actinia. J. Linn. Soe.

(Zool.) 16,474-476.
Redfield, E. S. P., 1915. The grasping organ of Dcndrocoehtm lactcitm. J. Anini. Be-

hav. 5, 375-380.
Sund, P. N., 1955. Response to electrical stimulation in Sfoinphia coccinca. (Unpub-
lished student report, Invert. Physiol. Course, Friday Harbor, Wash.)
Torrey, H. B., 1905. The behavior of Corymorpha. Univ. Calif. Publ. Zool. 2, 338-340.
Turner, R. S., 1946. Observations on the central nervous system of Leptoplana acti-

cola. J. Comp. Neurol. 85, 53-62.
Wenrich, D. H., 1916. Notes on the reactions of bivalve mollusks to changes in light

intensity : image formation in pecten. /. Anini. Behav. 6, 297-318.
Wilhelmi, J., 1915. Einige biologische Beobachtungen an Siisswassertricladen. Zool.

Ans. 45, 475-479.
Wulzen, R., 1917. Some chemotropic and feeding reactions of Planaria macidata. Biol.

Bull, 33, 67-69.
Yentsch, C. S., and D. C. Pierce, 1955. "Swimming" anemone from Puget Sound.

Science 122, 1231-1233.
Yerkes, R. M., 1902. A contribution to the physiology of the nervous system of the

medusa Gonioncmus jnurbachii. I. Sensory reactions of Gonionemus. Amer. J.

Physiol. 6, 434-439.
Yonge, C. M., 1930. The crystalline style of Mollusca and a carnivorous habit cannot

normally co-exist. Nature 125, 444-445.

PREY CAPTURE IN MANTIDS*

HORST MiTTELSTAEDT

Wilhelmshaven

The problem of absolute optic localization is one of the earliest discussed
in human psychophysiolog>% and one which was disregarded for the longest
time in the physiology of the invertebrates. The neglect may be merely due
to the fact that there are only a few cases in the invertebrates where that
special question can be asked, one of these being the case of the praying
mantid.

But let me explain the problem in the human case first. If you are to pick
up, say, a pencil lying in front of you on a table and if you have time enough
to do that without a rush, you not only see the pencil but your hand too.
Therefore, you merely have to move your hand in such a way that the
difference in position of these two observed things will disappear. The
situation will change fundamentally if you are not allowed to see your hand
or if movement goes too fast, as, for instance, in playing tennis, hammering
nails, or throwing a ball into a goal. In such a case it normally will not be
possible to correct the movement once started by watching the difference
between its direction and that of the goal. Consequently success here de-
pends upon information about the direction of the goal only. We certainly
may assume our optic centers to be able to transmit a signal pattern rep-
resenting the directional component of the retinal image concerned. Thus
these centers can be expected to provide information about the direction
of the goal relative to the eyeball. But, of course, that does not necessarily
mean information about the direction of the goal relative to the body. For
we can move our eyes and our head. It's fairly clear what should be con-
cluded : the message steering the movement of the hand should contain
information about the position of the eyeball and of the head too. How this
information is gained in fact is the question — discussed in human psy-
chology since the times of Helmholtz (1866) — with which I shall deal in
the related case of the mantid.

Mantids, h'ing in ambush all day, detect their prey by means of their
well-developed compound eyes. The prey is faced by movements of the
head, the eyes being firmly attached as in all insects. If the prey is the
proper distance away, it is captured by a sudden stroke of the forelegs. Two

* This investigation has been carried out at the Max-Planck-Institut fiir Verhaltens-
physiologie in Wilhelmshaven. The author is greatly indebted to Dr. R. Wette for
his advice about statistics, to Frau L. Dinnendahl for the preparation of the figures,
and to Dr. T. H. Bullock and Dr. B. T. Scheer for carefully reading the manuscript.

[51]

52 INVERTEBRATE PHYSIOLOGY

facts should be noted : First, the stroke has a time duration of about 10 to
30 milliseconds. Because of that short interval it should hardly be expected
that the stroke is controlled by watching the difference between its direc-
tion and that of the prey. Later we shall show evidence that, if present at
all, such a control at least is not eft'ective. Second, though the animal tends
to bring its head and its prothorax into one line with the prey, it is able to
hit a prey which has a considerable lateral deviation from the median plane
of the prothorax.

For these reasons the problem seems to be quite similar to the human
case. The direction of the stroke must be determined by a message repre-
senting the direction of the prey relative to the prothorax. Consequently,
that message should not only contain information about the direction of
the prey relative to the head, but of the position of the head too. It may be
supposed that the first information is provided by the compound eyes and
the second by proprioceptive sense organs which are able to present a
message about the position of the head. And indeed mantids possess a well-
developed system of neck receptors, and I shall examine these sense organs
first. But before going into details, I want to make some general remarks.

As already indicated by the way I have introduced the problem, it is the
functional organization of a system as a whole I want to understand. Con-
sequently, though I shall take advantage as far as possible of what is already
known about sense organs, efi^ectors, and nerve cells, the aim of this in-
vestigation is not — or at least not primarily — to learn more about single
elements. In the present case, as we shall see later, we are dealing with four
quite distinct functional units. My task will be to discuss how they act to-
gether and thus to explain the performance of the whole system by the
properties of its parts and their functional interrelations.

This being clear, I should say a few words about the necessary limita-
tions and restrictions. First, I shall deal with the orientation problem only
and omit all ethological questions about — for instance — appetitive behavior,
drives, and releasing factors. Second, though the learning problem will be
touched upon, we shall be concerned with the functioning of the fully de-
veloped mechanism in the adult animal only. Third, perfect localization
includes an indication of direction and of distance as well. I shall deal with
the direction problem only, and only with a special part of it, namely, the
orientation in the main horizontal plane.

First Series

Now we are ready to start with the first experimental series planned
to throw light upon the role played by the proprioceptors. The neck re-
ceptors of mantids are of a well-known type. There are two pairs of hair
plates each studded with from tens to hundreds of hair sensillae ( Fig. 1 ) .

PREY CAPTURE IN MANTIDS

53

The sternocervical plate was found by Pringle (1938), the tergocervical
plate by the author (Mittelstaedt, 1952). There is another very small
group of hairs at the pro-episternal wall of the prothorax which I shall not
mention further, because its functional influence can be neglected.

Fig. 1. Proprioceptors of the neck
region (left side). K = sternocervi-
cal hair plate, situated on the an-
terior end of the laterocervical scler-
ite (L). N = tergocervical plate.
The common afferent nerve of both
organs is cut at the ventral border of
the laterocervical sclerite. (O).

The external physical mechanism of these hair plates is easily under-
stood. In the normal position of the head some of the hair sensillae are
bent down by the posterior wall of the head. If the head is turned hori-
zontally, say to the left, some more hairs on the left-side plates are bent
down, while on the right side a correlated nimiber are set free and erected
elastically.

As Pringle (1938) has demonstrated in Periplaneta, bending down a
hair causes an increase in the impulse frequency of the afferent nerve,
which, after an initial peak rising and falling within a few seconds, remains
constant in time. Thus each position of the head causes a correlated pattern
of nervous activity, and the system is indeed able to take up information
about the position of the head as has been demonstrated by the author in
dragonflies (Mittelstaedt, 1950).

The first experimental series consists of four sets. The performance of the
animal in hitting prey has been examined, ( 1 ) after cutting ofif the sensory
nerves of the hair plates, (2) after giving the head a fixed position relative
to the prothorax, (3) after a combination of (1) and (2), (4) after having
loaded the head of the otherwise undisturbed animal by an extraneous
mechanical force.

In all four cases a statistical method was used to get quantitative results
about the hitting performance. The animals (I have worked mainly with
Parastagmatoptera imipunctata from Argentina) were sitting free on the
gauze ceiling of the cage in which they were normally kept. Then as many
files (I have used Calliphora and Lucilia) were brought in as the experi-
menter and his assistant could easily observe while they wrote down what

54

INVERTEBRATE PHYSIOLOGY

happened. The number and direction of misses, the number of hits and
their bias, that is, whether the prey was hit by the two forelegs or by one
only, were noted. Thus we shall characterize the hitting performance by
two quantities, first the frequency of misses and the second the left or right
tendency calculated from the frequency of biased misses and hits. With
regard to the latter quantity it is important to note that, by definition, left
tendency plus right tendency always equals 100%. Thus, if right tendency
is 100%, left tendency is zero; and if right tendency is 50%, there is no
bias at all. ( I should add here that, in calculating the values of some of the
diagrams — Figs. 2 and 5 — I have used a different statistical procedure than
the usual one. The difference will be less than 1 % if number of strokes is
98 or more and thus will be unimportant in most cases.)

I shall begin with the effects of eliminating the hair plates by nerve
section, and shall confine myself to the results of the bilateral total deaffer-
entation. In experiments carried out with 5 individuals, the frequency of
misses increased from 10-15% in the normals to 70-80% immediately
after this operation. There was only a relatively small improvement in
the following weeks. On an average, this may be even smaller than in the
example shown in Fig. 2, where it is partly due to the chances of an acci-
dental bias. If you are watching these animals you gain the impression that
they only have a chance to hit flies sitting straight ahead of the prothorax.

100%

24 days

Fig. 2. Result of total proprioceptive deafferentation in two individuals. Ordi-
nate : frequency of misses (open circles) and right tendency (solid circles), re-
spectively. Abscissa : days after nerve section ; the performance recorded before
operation is shown left from the 0-Iine. Thin vertical lines : standard error.

PREY CAPTURE IN MANTIDS

55

Normally flies sitting on the right side are missed to the left and vice
versa. I shall discuss these results later.

Next the head was given a fixed position relative to the prothorax by
what may be called a little bridge of balsa wood fastened on both ends with
paraffin, so that the neck region was not touched at all. Fig. 3 shows the
effect of a head deviation of 10 to 30 degrees to the left (in 6 animals).

x>~_.

right tendency

faults

275 strokesj yalue

1 Z 3 I. 5 6 7

Time (days)

Fig. 3. Result of head fixation 10 to 30° to the left (6 individuals). Ordinate:
frequency of misses (solid circles) and right tendency (open circles), respectively.
Abscissa: days after fixation; the performance recorded before head fastenmg i_s
shown left from 0-line. The limits of expectation, each based on a P-level of 0.025,
are plotted in two of the values.

There is a large bias to the right now and a frequency of misses of about
75%, remaining constant for at least a week. There may be a slight im-
provement of about 10% the following week, but in general we get another
proof of the small learning capacity of the mantid. If the head is fastened
near its normal median position the animal has a smaller bias and a better
achievement. In some cases where the exact median position was reached
by chance, the bias was zero and the frequency of misses normal.

In the next set, head fastening and deafferentation were combined. The
result just presented was obtained even in individuals which had under-
gone total proprioceptive deafferentation, suggesting that the effect of the
operation is cancelled if the head is fixed. In order to get a more rigorous
proof of this, the proprioceptors were eliminated on one side only. For in
the controls, without head fastening, this intervention causes not only a

56

INVERTEBRATE PHYSIOLOGY

large decrease in hitting performance but also a strong bias, namely, a fre-
quency of misses and a right tendency of 80-90% within 24 hours after
operation on the left side. The experimentals were first tested with the
head fastened only, and then operated unilaterally leaving the balsa bridge
intact. The result is seen in Fig. 4. There is an effect of the additional op-
eration and in the same direction as in the controls. The effects of the con-
stant head deviation and of the unilateral nerve section are clearly super-
posed. The relative importance of the components can be estimated from
the fact that the bias caused by eliminating one-half of the proprioceptors
is completely compensated for by a mean head deviation of less than 20°
(see left column of Fig. 4).

r- tendency
caused by fixaiion

r- tendency
faults

of sirokes
of animals

20 -i07o

before op. after

30,7 -^(^

i3V — 1,6,8]

202 156

4

-50 7o

before op. after

50,0-^(78))
23,^^718

175 UL

60-80 7o

before op after
(62,5 — 95,0)

197

60

Fig. 4. Result of head fixation combined with unilateral proprioceptive deafferen-
tation classified into three groups according to the bias caused by head fastening only.
A right tendency of 20-40% (left column) corresponds to a mean head deviation of
<20° to the right ; an r-tendency of 50% corresponds to the median head position ; an
r-tendency of 60-80% corresponds to a mean head deviation of <20° to the left. The
mean values obtained with head fastening alone are shown under "before op.," those
obtained within 24 hours after an additional section of the left nerve under "after."
Left nerve section without head fixation (controls — not shown in the table) causes
a frequency of misses and a right tendency of 80-90%.

In the fourth and last set of this series I loaded the head of the other-
wise undisturbed animal by an extraneous mechanical force. A small stick
of balsa wood was gummed on the posterior wall of the head and then a
small ball of plasticine of known weight was fastened on the stick at a
distance of about 10 mm. Of course the animals had to be tested sitting
on the vertical wall of their cage and even then we only counted the strokes
executed while the prothorax had a vertical position.

PREY CAPTURE IN MANTIDS

57

The result is seen in Fig. 5. With the exception of one series carried out
with two individuals which, even in the unloaded state, had a considerable
deviation from the normal frequency of misses, a significant decrease in

20 30 ^0 50 60

loa d [mg cm] — *■

10 20 30 W 50 60
load [mgcmj -»-

Fig. 5. Result of loading the head with an extraneous mechanical force. The
angular momentum caused by the load is so directed as to turn the head to the right
(for method see text). Ordinate: frequency of misses and left tendency, respectively.
Abscissa: angular momentum in mg.-cm. (mg. here is a unit of force, not of mass).
Solid circles, first experiment (2 individuals) ; open circles, second experiment (3
individuals) ; double circles, third experiment (3 individuals). In the last named
the standard error is plotted. Figures at the values: number of strokes (n).

performance is reached not earlier than at a load of 50 mg. cm. That is a
remarkable achievement if we consider that the head of Parastagmatoptera
weighs 25 mg. and hence there is a load of twice the head weight at a
distance of twice the head diameter. Thus we may be certain that loads
which can be expected to occur in normal life (for instance, if the animal
is catching the second fly while eating the first) are perfectly ruled out by
the mechanism at hand.

Introduction of an Hypothesis

Now we have gained sufficient information to put forward an hypothesis
concerning how the mechanism works. If we only had to account for the
first result, of deafiferentation, we would assume that the hair plates pro-
vide for the disputed additional information about head position. If we
knew only the second result, of head fastening, we certainly should be
advised to the contrary, namely to assume that there is no such influence
at all. But it may be concluded from the third that the truth will be some-
where between ; and the last result could give an idea of how this compro-
mise is expected to work.

Evidently the mechanism is very sensitive to or easily upset by changes

^^ INVERTEBRATE PHYSIOLOGY

in head position which cannot be corrected or, as can be said as well to
deviations from the intended head position. Removal of the hair plates
causes a sharp decrease in achievement, but the effect is eliminated or even
reversed simply by fastening the head in a suitable position ; thus it must
be assumed that these organs are involved in a system which controls the
head movement-in a system which provides for the intended head position
to be reached in fact.

To give a precise form to this hypothesis, so that conclusions can be
derived which can be tested further by experiment, I shall" take advantage
of methods and theories recently developed to explain the functioning of
systems like that with which we are dealing. I mean the modern theory of
automatic control and of control systems in general developed during the
last decades to a high level of precision and universality. I shall make an
attempt to apply that concept to the system which in mantids controls the
movements of the head.

If a prey comes into sight, the mantid turns its head to face it Thus
there must be a functional unit which transforms the position of the prey
relative to the compound eyes into a central nervous representation of that
position. The latter must be transmitted to the centers controlling the neck
muscles, which then start a head movement determined bv that message
with regard to direction, amount and/or speed. But as soon as the head
changes itsposition, the optic message is changed too, so that the output
of he physiological pathway must necessarily influence its own input To
make that completely clear, I shall plot it diagrammaticallv (Fig 6a)'

1 shall define all directions or positions by the angular deviations from
the median plane of the organ concerned, the axis of reference for all angles
bemg the vertical axis of the head movement. We have to distinguish ( 1 )
the angle between the prey and the head, that is the optic input ; (2) the
angle between the head and the prothorax ^, that is the neck motor output
and the proprioceptive input as well ; and (3) the angle between the prey
and the prothorax a. Then we should consider the optic control unit con-
verting the angle <^ into a central nervous message <^„ and the neck motor
unit which transforms <f>, into a head deviation ,,. Since

— (T — jj..

Thus, as indicated by the arrows, information is transmitted in one direc-
tion only, and flows within a closed loop, the output being negatively fed
back to the input. » t> j

Now I shall take into account the final steady state of the system only-
that IS, the position after all movements have come to rest, all actions and
forces being in complete equilibrium. If, for the sake of an easier introduc-

/"c-=0c-4

3t=(p^+F6c

Fig. 6. Functional diagram of the mechanism underlying localization in mantids.
The hypothesis is developed by steps from (a) via (b) to (c). (a) Optic feedback
loop only. As indicated by the arrows, information flows from the optic unit (ampli-
fication factor: A (opt)) to the neck motor unit (amplification factor: A (neck))
and again to the optic unit, (b) Optic and proprioceptive feedback loops. The neck
motor unit is controlled by the difference between the optic {<p^) and the propriocep-
tive (3J center messages, (c) Complete hypothesis. A fst^okei amplification factor
of the central unit which determines the direction of the stroke (/c). For full explana-
tion see text.

60 INVERTEBRATE PHYSIOLOGY

tion, we assume that there is a constant hnear proportionahty between <f>
and ^c and between <^c and /x, respectively, and if we define the propor-
tionality factors as A (opt) and A(neck) respectively, then

ii^ = A(opt) X A,„eck) X <^ or

~ ^ A, opt) X -'^(neck)

9

The ratio ix/(f> will be named Ac (total) , the total interior amplification of
the control circuit. You may note that this is a pure number without any
dimension. It may be worth while to give an example of how that works.
Assume the system has an Ac (total) of 4, and a fly comes into sight at
cr = -(- 20°, that is 20° to the right. Then the fixation process will come to
rest exactly at a head deviation /a of -|- 16°, ^ thus being +4°. For only
with that optic input can the neck motor output be + 16°, if we have 4 times
total amplification. Thus in general :

o- 1 + Ac(total)

and the prey will be better centered as Ac (total) increases. The ratio

^— will be named the "fixation-deficit."
a

Now we are able to formulate the hypothesis, namely as follows : It is
assumed that fi is converted by the hair-plate system into a message 8c, and
that the neck muscles are steered by the difiference between the optic center
message 4>c and the proprioceptor center message 8c- Thus there are two
circuits working together (see Fig. 6b).

In order to understand the operation of these circuits, let us watch the
proprioceptive subcircuit acting in isolation. Suppose the head to be thrown
out of its median position by some extraneous influence, say 20° to the
right. Then the hair plates will transmit a message causing a head move-
ment to the left. The system will come to rest at the smallest value of /a
allowed for by the amount of the extraneous influence and the total amplifi-
cation of the proprioceptive circuit. Therefore we may say that this circuit
tends to minimize /x against all extraneous influences. Because the optic
circuit — as we have just learned — tends to minimize (j>, it can easily be seen
that there must be a rivalry between the two systems, except only in the
case that o- = 0, the prey being straight ahead of the prothorax — then
cj} and IX should be zero. In all other prey deviations neither the one nor the
other circuit wifl reach its target. Thus under the conditions adopted the
fixation-deficit must have a finite value, and hence, at equilibrium, there
will be a constant correlation between the proprioceptive input, the optic
input, and the deviation of the prey from the body axis. Consequently the
optic and the proprioceptive center messages (4>c, 8c) must also have a

PREY CAPTURE IN MANTIDS 61

constant correlation to the deviation of the prey from the body axis ; and
this again means that each of them — at equilibrium — contains the informa-
tion which is required to determine the correct direction of the stroke.^

Let us first assume that the direction of the stroke be determined by
the optic center message only (Fig. 6c, without the dotted arrow). It may
be useful to illustrate this by an example. For the sake of simplicity the
direction of the stroke k may be defined as the angle between the endpoint
of the stroke and the median plane of the prothorax, the axis of reference
being the vertical axis of the head movement. Then the prey will be hit
if K is equal to o-. Now consider the system has a fixation-deficit of \0% ;
that is — at equilibrium — the optic input equals -f 1 ° if the prey deviates
+ 10° from the body axis, +2° if the prey deviates +20°, and so on. Evi-
dently the central units involved in the determination of the stroke merely
have to amplify the optic input by a constant factor in order to hit the prey
exactly.

Let us see now how this hypothesis fits the experimental results first
presented.

( 1 ) If the hair plates are eliminated by nerve resection on both sides,
the proprioceptive circuit breaks down. Thus the optic circuit normally
working against it will now be more effective in minimizing (f). Conse-
quently there must be a smaller fixation-deficit than before. Because the
factor by which the optic input is multiplied in order to steer the direction
of the stroke is presumably not changed by the operation, the deviations
of the stroke from the body axis will always be too small, except in the one
case when the prey is sitting straight ahead. As we have learned, this is
just what happens in fact.

(2) If the head is given a fixed position, say 20° to the left, then the two
circuits are both blocked. Thus the performance will depend upon whether
the animal succeeds in centering the fly by means of movements of the legs
and the prothorax. If the fly is at least approximately centered that way,
before the stroke is released, the optic input will be about zero, the stroke
thus going more or less straight ahead. But since the fly then in fact has a
mean deviation of 20° to the left, it normally will be missed to the right,
quite in accordance with the facts. Yet it will be hit frequently, if the head
is fastened in its median position. As we have learned, that is indeed so.

(3) The result of the combination experiment does not agree with the
assumption that the stroke is determined by the optic-center message only.
If the head cannot be moved, the additional elimination of proprioceptors
should then have no additional effect at all. Consequently we must conclude

1 To satisfy the rigorist, it may be added that this is true even if we leave out the
linearity and proportionaHty conditions introduced at the beginning.

62 INVERTEBRATE PHYSIOLOGY

that there is yet another way by which the proprioceptors, at least in that
special situation, can act upon hitting performance. The superposition
effect indicates that the proprioceptive-center message is added to the optic-
center message (see Fig. 6c, dotted arrow). Yet, there is strong evidence
that the proprioceptive influence is smaller than the optic influence. Other-
wise, if it were equal or larger, the bias demonstrated in the head fastening
and in the loading experiments must be zero or even reversed in sign.

(4) Finally the loading experiment gives a good proof of the biological
importance of the proprioceptive subcircuit, namely its ability to rule out
disturbing forces which act upon the neck motor system. According to
theory there should be no bias and no lowering of the hitting performance,
as long as these forces are eliminated completely. As soon as the head
begins to deviate from the intended position, say to the right, there would
occur a left bias — quite in accordance with the facts. It should be noted
that the subcircuit works not only against extraneous influences but also
against changes in the normal state of the muscles, caused for instance by
fatigue or any sort of metabolic disturbance.

Second Series

As the hypothesis put forward can be defined with mathematical pre-
cision, we should now make an attempt not merely to get knowledge about
the connections between the basic units of the system as we have done up
to now, but to determine quantitatively the functional relations predicted
by the theory and, finally, to calculate the constants of the system. I think it
is fairly clear what should be measured, namely the fixation-deficit in
normals and after bilateral nerve cutting.

The plan of the procedure is simple enough ; the prothorax is fastened,
a fly is presented at a measurable direction a, and, if the animal faces it, the
position of the head /x is measured.

I have used two different techniques, one of which is shown in Fig. 7.
The animal is sitting in its normal upside-down posture on a paper dial
hanging by threads of silk. The weight of the animal is compensated by
means of two pulleys and a counterweight not visible in the picture. A stick
of balsa wood gummed on the head projects through a hole, loose enough
to allow for rotation, and carries a pointer. Thus only the head movement
was limited to rotation about the relevant axis, all joints were free, and
the animal seemed to be comfortable even for weeks. In the second
apparatus also the head was completely free, bearing a small thin piece of
paper, which controlled the current of a photocell circuit. Finally, at the
output of the device, the position of the head, of the fly, and of the thorax
was registered automatically and continuously.

It should be mentioned that this experimental series is not yet finished,

PREY CAPTURE IN MANTIDS

63

Fig. 7. Device for measuring the fixation-deficit. The animal
is fixed at the prothorax, the head being free for rotation about
the vertical axis. The deviation of the head and of the prey from
the median plane of the prothorax are measured on the same
dial.

only three animals being investigated in the first and another three in the
second device. But because the result seems to be fairly interesting with
regard to the problem as well as to the concept adopted, I feel I should not
withhold it here.

It turned out that two quite different sorts of head movements occur in
mantids, similar to the relevant eye movements in man. There are quick
saccadic head movements, on the one hand, and smooth continuous move-
ments, on the other. The first are observed if the prey appears and moves
beyond a distance of about 30 mm. — that is, in Parastagmatoptera, nearly
double the reach of the stroke. About 30 mm. distance both types are seen ;
if the prey comes nearer, it is followed by continuous movements only.
One may be certain that, in the continuous movements, the optic control
circuit is working in fact, but it must be doubted whether that be true
within the short intervals of the head jerks. The time duration of the jerks

64

INVERTEBRATE PHYSIOLOGY

as a function of their angular size could not yet be measured exactly be-
cause of the time lag of the recording instruments. Motion-picture photo-
graphs showed a time duration of less than 80 msec, in jerks of 90°. Thus
the duration of jerks smaller than 20° possibly come close to the latency of
the optic circuit. If the jerk is executed without any feedback, whether
optic or proprioceptive in origin, then the amount of the movement must
be based exclusively upon the optic information present before the jerk
starts. Consequently the position of the head, after the jerk has finished,
should depend not only upon the position of the fly but on the initial po-
sition of the head too. Thus, if initial positions are variable, there should be
no constant proportionality between the position of the fly (a) and the
position of the head (^t) at the end of the jerks, unless the jerks always

■T3

a

Y

^jf Fly I Prothorax

Fig. 8. Endpoints of fixation movements of the head plotted against the position of
the fly which is faced. Ordinate : deviation of the head (^u.)- Abscissa : deviation of the
fly (<r), both from prothorax median plane. -|- : deviation to the right; — : deviation
to the left. The experimenter moves the fly in small steps from —25 to -\-2S and return.
The procedure is repeated three times. Small circles : head positions reached by con-
tinuous movements. Large dots: head positions reached by jerks. Dots are grouped
along the lower line, if the fly had reached the position concerned by moving from left
to right ; but they are grouped along the upper line, if the fly had moved from right
to left. Distance from the fly to the head : 29 mm.

PREY CAPTURE IN MANTIDS

65

exactly reached their goal. We shall see immediately that, though the latter
is not the case, there is a proportionality between [x and a, but of a very
astonishing form. In the example shown by Fig. 8, the fly is about 29 mm.
from the mantis. There are both continuous movements and jerks, the end-
points of the jerks being more strongly marked. The continuous move-
ments are weak and ineil^ective ; but the endpoints of the jerks are grouped
along two straight lines nearly equal in slope. The values lie on the lower
line only, if the fly is moved from left to right, and on the upper line only,
if it is moved from right to left. The distance between the two lines with
regard to abscissa is about 9° (see Fig. 9). Nine degrees at a distance of

-^ Fly / Prothorax

Fig. 9. The same experiment as Fig. 8 with head positions reached by jerks only.
The slope of the upper full line (regression coefficient) is 0.85 ± 0.03, that of the lower
full line 0.83± 0.01(6). The abscissa distance between these lines of about 9° cor-
responds to the angular width of the fly presented. Hence the upper and lower broken
lines indicate the positions which the head would have, if the right and left sides of the
fly were fixed exactly. Yet the differences in slope of full and broken lines are con-
siderable (15 and 17%, respectively), and would be expected due to chance with a
probability of less than 0.0005.

29 mm. is about 4% mm. Four and a half mm. is about the mean horizontal
width of the flies presented. Thus the simplest explanation is to assume
that the animal tends to face the left side of the fly while the fly is going
from left to right, and the right side if it is going from right to left. But in
neither case was the side of the fly centered exactly. The slopes are 83 and

66

INVERTEBRATE PHYSIOLOGY

85%, the difference being insignificant. Yet the difference between each
slope and 100% is highly significant. Thus I don't see how to avoid the
assumption that, at the end of a jerk, the head has a position very near to
that which it would have if the circuit had reached its final steady state.
In view of our problem it is particularly interesting what happens if the
prey comes nearer to the reach of the forelegs. Here again, especially at
the beginning of the experiment, a tendency to face the sides of the fly is
observed in some cases. But in the period before the stroke is released, the
picture normally changes (see Fig. 10). There are still two pathways, but

o
o

+10"

g -10'

Y

-m

right side of fly { "7

0'

" 1 1 1 1

'Zf.'J^i^ideoffly

1 1 1 1

-20'

-10° 0° +10'

<^ Fly I Pro: horax

+20'

Fig. 10. Same procedure and same mapping as in Figs. 8 and 9, but the fly is pre-
sented at a distance smaller than 30 mm. The head followed the fly exclusively by
continuous movements. Open circles and full lines : head positions while fly is moving
from left( — ) to right ( + ) ; solid circles and broken lines : head positions while fly is
swinging from right ( + ) to left( — ). The fixation-deficit is equal to 1 minus the slope
of the curve and hence about 15%.

they differ by not more than about 2° and, if the prey moves from left to
right, the head at o- = is about one degree right from the center of the fly
and vice versa, thus being slightly in advance. We get an estimate of the
achievement produced here while considering the width of the fly and the
fact that the best ommatidia of this insect have angular apertures of about
one degree. The lines have slopes of 82 to 88%, in accordance with the
first result. Thus, the fixation-deficit being about 15%, the total circuit
amplification can be calculated to be nearly 6.

PREY CAPTURE IN MANTIDS 67

Unfortunately the experiments planned to give information about the
fixation-deficit after eliminating the hair plates were executed before I
knew the distance rule mentioned and the fact that the jerks normally are
directed to the sides of the prey. For mere technical reasons the fly (in the
first apparatus) was presented at a distance of more than 30 mm. and
hence faced by saccadic movements almost exclusively. The head devia-
tions (fx), if plotted against the fly deviations (a) showed a considerable
spread and, at first sight, it seemed impossible to detect whether the mantis
had faced the left or the right side of the fly. The difficulty was overcome
by considering the way the data were obtained. As the fly normally was
moved from the median position ((7 = 0) to the right or left sides respec-
tively (a = ± 60°) while the reverse occurred very rarely, it should be
assumed that, in the majority of presentations on the right side of the
mantis, the left side of the fly (as seen from the mantis) was faced and vice
versa. Hence the regression function of head deviation /x on fly deviation <t,
if calculated for each side separately, should result in two parallel re-
gression lines separated by an abscissa value corresponding to the width of
the fly. Furthermore their slopes should be larger than the one calculated
from the total set. This was indeed so in 3 out of 4 sets. The result is given
in the following table :

SLOPES OF THE REGRESSION FUNCTIONS OF

HEAD DEVIATION (m) ON FLY DEVIATION ((7)

FOR TWO INDIVIDUAL MANTIDS

Individuals

Total Slope of the Regression
Function of ju. on cr

Mean of the Isolated
Left and Right Slopes

Normal

Deaf-
ferented

P

Normal

Deaf-
ferented

I

0.72
±0.018

0.89
±0.014

< 0.0002

0.72
±0.018

0.99
±0.04

II

0.79
±0.009

0.88
±0.015

<0.0002

0.91
±0.02

1.02
±0.08

It can be seen that

(1) In all cases, the slopes obtained after bilateral proprioceptive de-
afiferentation are larger than those obtained before.

(2) In both individuals tested, the differences between the total slopes
before and after operation are highly significant with P < 0.0002.

(3) The means of the left and right slopes (with one exception probably
due to an unfavorable accumulation of a values around the median po-
sition) are about 10% larger than the totals.

(4) The left and right slopes after deaft'erentation are close to unity.

68 INVERTEBRATE PHYSIOLOGY

Discussion

It has been demonstrated in the second series of experiments that, at
the time before the stroke is released, the prey is faced by a continuous
optic feedback process. At equiUbrium, the fixation hne does not center
the prey but deviates from it by an amount proportional to the angles be-
tween the prey and the median plane of the prothorax. So far the assump-
tions made by our hypothesis have been confirmed. All further conclu-
sions from this series, and their interpretation in relation to the problem,
depend upon whether it can be ascertained that the saccadic head move-
ments are controlled in essentially the same way as the continuous ones
or not.

The first two results (1,2) of the last experiment obviously corroborate
the assumption made before, that the saccadic movements too are con-
trolled by feedback processes. The result not only shows that there is an
efifect of the proprioceptors on head movement, but it shows just that sort
of influence which the theory predicts in the case of two counteracting
control circuits. For, if the proprioceptive circuit be cut ofif, the optic circuit
should be more eft'ective in minimizing the deviation of the prey from the
fixation line (f>, and hence the ratio /x/o- (since <^ = o- — /i) should be larger
after operation than before.

Yet, if we compare the larger spread of the head positions reached by
saccadic movements with the negligible fluctuations of those reached by
continuous ones (cf. Figs. 9 and 10), it is fairly clear that the endpoints
of the jerks are not identical with the values which correspond to the final
steady state of the feedback process. Consequently, at least one restriction
must be made, namely, that in the jerks the feedback is blocked before the
equilibrium is reached. As an alternative hypothesis, it could be assumed
that a jerk is a simple reflex initiated by the deviation of the fly (<^), and
not at all controlled by the effects of its output. But then, as already men-
tioned in the last section, the linear regression functions of /^ on o- obtained
in all these experiments would be difficult to explain. Thus, the data avail-
able at the present state of the analysis are best fitted by the assumption
that both sorts of head movements are controlled by optic and propriocep-
tive feedback, though the dynamic qualities of the two systems may be
fundamentally dififerent.^

2 Future analysis may reveal that the continuous process operates near the theoret-
ical optimum of velocity and dynamic stability, so that any further enlargement of
speed (in the jerks) would cause overshooting and oscillation, were the circuit not
interrupted from time to time. Such a dynamic dichotomy in the ways of fixation
seems to be plausible from a biological point of view ; the pursuit of a moving prey
which is far beyond the reach of the stroke demands quickness but no precision. The
reverse is true if, after a long stalking, the direction of the stroke has to be de-
termined.

PREY CAPTURE IN MANTIDS 69

With these restrictions born in mind, I want to discuss some of the
further consequences of the second experimental series. The result last
presented (4) indicates that the endpoints of the jerks after deafferenta-
tion are distributed at random around the true points of fixation (the left
or right side of the fly). If the same occurs in the continuous movements,
so that the fixation-deficit after deafiferentation is zero, some interesting
conclusions can be drawn on the qualities of the system. For then either the
optic or the neck motor unit must be an "integral action controller," that
is, the output of the system depends on the time integral of its input, and
hence only comes to rest if the input is zero. Evidently the neck motor tmit
operates this way and the optic unit does not ; otherwise there would be
no fixation-deficit in the normal mantis. Consequently, since the input of
the neck motor unit is controlled by the difference between the optic and
the proprioceptive center messages, both must be equal at equilibrium.
And this again means that all disturbing forces acting upon the neck motor
system are completely ruled out within the limitations set by the capacity
of the muscles. It should be noted that this is supported by the result of
the loading experiment (ci. first series). Thus, even if future research
should show that the fixation-deficit after deafferentation slightly differs
from zero, the following statements can be considered as good approxima-
tions, namely :

Ac (total) — ^'°P"

•^(prop)

and hence the function of the deviation of the stroke (k) on the deviation
of the prey (a) at equilibrium is

A (Stroke) ' (l~hF)

' + '

•^ (opt) -^ (prop)

where F is the factor which determines the additional influence of the
proprioceptive center message on the direction of the stroke (cf.
Fig. 6c).

Finally, the efficiency of the functional organization as revealed by the
present analysis may be briefly considered from a biological point of view.
Four main advantages are combined by the system :

( 1 ) The head can be moved and hence the prey can persistently be
faced by the region of the compound eye with the best visual acuity.

(2) Nevertheless the prey can be localized correctly, even if it deviates
from the body axis.

(3) The hitting performance is independent of external and internal
influences which act upon the neck muscles.

70 INVERTEBRATE PHYSIOLOGY

(4) The "calibration" of the mechanism which determines the direction
of the stroke can be based solely upon the amplification factors of the pro-
prioceptive and optic units. Thus it is at least not inconceivable that the
correct adjustment is predetermined by the structure of the sense organs
and the nerve connections involved.

Summary

The sensory-motor coordination which enables mantids to hit their
prey is analyzed by recording the hitting performance under controlled
experimental conditions. It is found that :

(1) Normal mantids (Parastagmatoptera unipunctata) hit about 85%
of the flies they intend to capture. If the proprioceptors of the neck region
are eliminated by nerve section, the hitting performance is irreversibly
reduced to 20-30%.

(2) If the head is rigidly fixed on the prothorax in the median position,
the performance is normal; but it decreases to 25% if the head deviates
from the body axis by 10-30°. The prey is missed to the left if the head
has been turned to the right, and vice versa.

(3) If head fastening and unilateral elimination of the proprioceptors
are combined, the effects of both are superposed. The loss of one-half of
the neck receptors is equivalent to an angular deviation of the head less
than 20°.

(4) If the (free) head is loaded by an extraneous force, the achieve-
ment remains normal until the load surmounts twice the head weight at
twice the head diameter.

It is concluded that the direction of the stroke depends upon feedback
processes which control the position of the head as follows : The fixation
movements of the head, which precede the release of the stroke, are steered
by the difference between the optic-center message (which is a function
of the angle between the prey and the fixation-line) and the proprioceptive-
center message (which is a function of the angle between the head and the
body axis ) . If the fixation movements have come to rest, the direction of
the stroke is determined by the optic and (to a smaller extent) the pro-
prioceptive-center messages, which then both contain the required infor-
mation.

The hypothesis is cross-checked by measuring the position of the head
at the end of the fixation movements. It turns out that the fixation line
does not center the prey, but deviates from it by an amount proportional
to the angle between the prey and the body axis. As predicted by the theory,
this deviation is diminished after total proprioceptii^e deafferentation.

PREY CAPTURE IN MANTIDS 71

REFERENCES

Fisher, R. A., and F. Yates, 1949. Statistical tables. 3rd ed. London.

Helmholtz, H. v., 1866. Handhuch der Physiologischen Optik. 3 Aufl. 1910, III. Bd.
(pp. 203-207). Hamburg and Leipzig. (American edition: Physiological Optics,
1925, Vol. 3. Menasha, Wis.)

Hoist, E. V. u. H. Alittelstaedt, 1950. Das Reafferenzprinzip. Naturzviss. 37, 464.

Ludvigh, E., 1952. Control of ocular movements and visual interpretation of environ-
ment. Arch, of Ophthahn. 48, 436, 442.

Mittelstaedt, H., 1950. Physiologic des Gleichgewichtssinnes bei fliegenden Libellen.
Z. vergl. Physiol. 32, 422.

Mittelstaedt, H., 1952. Uber den Beutefangmechanismus der Mantiden. Verh. Dtsch.

Zool. Ges. 1952, 102.
Mittelstaedt, H., 1954a. Regelung in der Biologic. Regelungstechnik 2, 177.
Mittelstaedt, H., 1954b. Regelung und Steuerung bei der Orientierung der Lebewesen.

Regelungstechnik 2, 226.

Oppelt, W., 1954. Klcincs Handbuch technischer Regelorgdnge. Weinheim-Bergstr.
Pringle, J. W. S., 1938. Proprioception in insects III. The function of the hair sen-

silla at the joints. /. of Exp. Biol. 15, 467.
Siebeck, R., 1954. Wahrnehmungsstorung und Storungsv^-ahrnehmung bei Augen-

muskellahmungen. Graefes Arch. f. Ophthahn. 155, 26.

Sperry, R. W., 1950. Neural basis of the spontaneous optokinetic response produced
by visual inversion. /. of Comp. and Physiol. Psych. 43, 482.

NERVOUS CONTROL OF INSECT MUSCLES*

Graham Hoyle

University of Glasgow

Detailed studies of the nervous control of muscles have heen made
principally on vertebrates and crustaceans. They have been concerned
particularly with the elucidation of the mechanisms of neuromuscular trans-
mission and have left largely unsolved many of the more general problems,
such as the method of maintaining tone and the way antagonist muscles
are used (see Elftman, 1941). Ideally it should be possible to give a com-
plete analysis of an integrated movement in terms of all the events in-
volved (for both the agonists and antagonists concerned) , i.e., motor nerve
impulses, transmission processes, muscle fiber contractions, activation of
proprioceptors, sensory nerve impulses, and central nervous integrative
processes, all stated quantitatively. Perhaps the greatest theoretical interest
lies in the central processes, and it was ably demonstrated by Sherrington
( 1906) half a century ago that the experimentally accessible neuromuscu-
lar apparatus can be used as a window to the functioning of the central
nervous system. An extension of these studies to other classes of animals
may be justified on the grounds of their intrinsic interest and also because
some of them may provide better experimental material for analysis of
some of the general problems.

It seems likely that insects, which have been little studied in regard to
their neuromuscular phenomena, ofi^er excellent material. They certainly
ofifer some challenging problems. In all insects there are functionally im-
portant muscles which are microscopically small, sometimes composed of
no more than a dozen muscle fibers. Yet the joints operated by those
muscles are moved with the precision which characterizes most insect
movements, and the delicacy of action compares favorably with that en-
countered in the highest vertebrates, in which each muscle is composed of
thousands of muscle fibers and is innervated by hundreds of nerve fibers
operated by a central nervous system of immense complexity. The small
size of insect limbs and the simplicity of both muscles and innervation
ofifer peculiar advantages for a complete study of the subtler aspects of
nervous control as well as the special problems of their neuromuscular
transmission.

* I wish to thank Dr. T. D. M. Roberts for his helpful criticism of the first draft of
this paper. The electrical apparatus used in making the original observations reported
here was purchased with the aid of an award from the Grant-in-Aid Fund of the Royal
Society.

[73]

74 INVERTEBRATE PHYSIOLOGY

A complete study of the phenomena of control in insects would shed
light on the functioning of the insect central nervous system, on which
there is at present almost no information, and hence on the properties of
neuropile in general. There seems every reason to hope that a detailed study
of the nervous control of insect muscle should be quite feasible and of gen-
eral value.

It is to be expected that the insect mechanisms may be found to be some-
what similar to those of the crustaceans, many of which have already been
fairly completely elucidated; but there is no longer any need for insect
physiologists to borrow ideas from the crustacean field in interpreting the
insect phenomena. Modern techniques of investigation, particularly the
use of intracellular capillary microelectrodes, make it possible to give
unambiguous information about the functioning of individual muscle fibers.
Techniques are available for the stimulation of single nerve fibers even
when these cannot be prepared separately, and they have already proved
of value in insect work (Hoyle, 1955b, partly based on a method by Kuffler
and Vaughan Williams, 1953).

Although this paper will be concerned with the limb muscles of insects,
occasional reference will be made to thoracic or wing muscles and to the
sound-producing muscles of cicadas, many of which have been evolved
from muscles operating limbs ; so far all the evidence shows that, although
the histology and metabolism of these thoracic muscle fibers has been
greatly altered in some orders, there has been no great change in either
their pattern of innervation or their neuromuscular mechanisms. At the
present stage of the investigation being undertaken by the author, it is not
possible to go much farther than a description of the neuromuscular
mechanisms, although a promising start has been made in the direction of
studying natural nervous control in the body ; some of this unpublished
work will be described briefly. A cursory examination of the problem shows
that tiny muscles cannot be satisfactorily operated along vertebrate lines,
where graded tension is produced by varying the number of units each in
one of two alternative states, i.e., rest or "all-or-nothing" contraction.
There are just not sufficient muscle fibers available for executing smooth
contractions by this method, even if space were available for the large
number of nerve fibers and their cell bodies which would be required to
control them. Hence an elucidation of the anatomy of the innervation must
play just as important a part as a study of the physiology in contributing
to an understanding of the nervous control of insect muscle.

The Innervation of Insect Muscle

Insect muscles are supplied with only a very small number of motor
nerve fibers; Mangold (1905) demonstrated a double nerve-fiber inner-

NERVOUS CONTROL OF INSECT MUSCLES 75

vation of Decticus thoracic and leg muscles and Dytiscus wing muscles ;
Montalenti (1928) described a triplotomic branching in Hydro philus leg.
Pringle (1939), using a physiological method, showed that the extensor
tibiae muscle of the metathoracic leg of Periplaneta receives two motor
axons. The locust metathoracic extensor tibiae muscle receives three axons
(Hoyle, 1955a). The homologous muscles of the pro- and mesothoracic
legs only receive two (Hoyle, unpublished) ; many other locust muscles,
e.g., the retractor unguis, also receive two. The flexor tibiae of several
species, e.g., Romalea microptera (Ripley, 1954), Acanthacris riificornis
and Zonocerus sp. (Ewer, 1954), do however appear to receive four or
even more axons, as indicated by a consideration of the number of steps
which can be obtained in the tension developed by the muscles when a very
carefully graded stimulus is applied to the motor nerve.^ The flexors of the
tibiae of all the legs of the locusts Locusta migratoria and Schistocerca
gregaria also seem to receive several axons. I have obtained graded steps
from these muscles during stimulation of the motor nerve and at the same
time have recorded intracellular action potentials from various muscle
fibers in different parts of the muscles. As the stimulus strength is raised,
groups of fibers in different regions come into twitch activity. The action
potentials in all the fibers are nearly identical electrical responses of the
"fast" type (see below). Evidently the orthopteran flexor tibiae muscles
are composed of four to six motor units, each of which has a separate nerve
supply. From the evidence in the literature we may tentatively regard the
Pm/?/aM^/a flexor trochanteris (Pringle, 1939) and the Z^yfwcM.? extensor
trochanteris as being similarly constructed. Graded contraction in these
muscles could be effected not only by the special arthropod methods to be
described, but also by the vertebrate method of varying the number of
motor units in action at a time. The "fast-fiber motor unit, i.e., the comple-
ment of muscle fibers supplied by a single "fast" axon, of the flexor tibiae
muscles consist of a single, coherent bundle of muscle fibers, in contrast
to the vertebrate motor unit which is probably composed of fibers scattered
throughout the whole muscle (data in Tiegs, 1953). The smallest insect
muscles are formed of single bundles of fibers and the larger ones are con-
structed of several such bundles (termed "muscle units" in Hoyle, 1955a).
The locust and cockroach extensor and flexor tibiae are all composed of
several muscle units ; those of the extensors all receive branches from the
same two (or three) axons, whereas those of the flexors may receive inde-
pendent nerve supplies. Perhaps this complexity of innervation imparts a
greater degree of controllability to the flexors. In the pro- and mesothoracic
legs of locusts and in all the legs of the cockroach, the flexor tibiae play a

1 The same is also true of some coxal muscles of the cockroach (Becht and Dres-
den, 1956).

Id

INVERTEBRATE PHYSIOLOGY

more important role functionally than their antagonists ; the flexors are
the main postural (tonic) as well as the main active muscles, so a high
degree of control may be required for them.

In regard to the "fast" fiber, the insect motor unit may be either the
whole muscle, as in the extensor tibiae, or only one or a few of the com-
ponent muscle units, as in the flexor tibiae. Very recently Tiegs ( 1955) has
demonstrated by staining methods the presence of two axons supplying the
thoracic muscles of several species of insects. The sound muscles of cicadas
are, however, supplied with only a single axon (Hagiwara, 1953 ; Pringle,
1954).

a.

c.

Fig. 1. Diagram to illustrate the three principal types of
muscle organization encountered in insects. They are all shown
here as if doubly innervated, (a) Single-unit type, e.g., retractor
unguis, levator and depressor tarsi, (b) Multiple-unit, common-
innervation type, e.g., extensor tibiae, (c) Multiple-unit, separate-
innervation type, e.g., flexor tibiae, extensor trochanteris.

We may conclude that a double innervation is probably the common
mode ; in a few instances there is an additional axon and some muscles are
suppHed by only one axon. From the point of view of neuromuscular trans-

NERVOUS CONTROL OF INSECT MUSCLES 11

mission the flexor tibiae and similarly constructed muscles need not be re-
garded as having more than the usual double innervation, but only as com-
pound or multiple-unit muscles, the individual units of which conform to
the single- or double-innervation pattern. On this basis insect limb muscles
can be divided into three categories (Fig. 1). These are the single-unit
type (Fig. l,a), e.g., the retractor unguis, levator, and depressor tibiae
muscles; the multiple-unit, common-innervation type (Fig. l,b), e.g., the
extensor tibiae ; and the multiple-unit, separate-innervation type ( Fig. 1 ,c ) ,
e.g., the flexor tibiae and the extensor trochanteris.

Neuromuscular Junctions

A wide variety of endings has been described for the terminations of
motor axons on insect muscles. Filiform branches with no definite junc-
tions have been observed (Montalenti, 1928 ; Morison, 1928) and this is the
kind usually encountered in the Crustacea (Van Harreveld, 1939). The
finer branches have been described as actually entering the muscle sub-
stance (Marcu, 1929; Tiegs, 1955). However, there are strong physio-
logical grounds for not accepting this picture of axons penetrating the
muscle fibers and also for being rather skeptical about the fibrillar type.
This caution is strengthened by the fact that in a few instances end plates
of definite structure have been clearly observed (Tiegs, 1955; Hoyle,
1955a.) There are many early references to endings of end-plate type,
often large enough to warrant description as Doyere eminences. The
histological observations are rendered extremely difficult by the presence
of enormous numbers of fine tracheae, tracheoles, and tracheal end cells.
The tracheoles often penetrate right into the muscle-fiber substance. In this
case there is no doubt about the accuracy of the observations, and these
diffuse surface and penetrating intracellular tracheoles with their asso-
ciated cells could easily be responsible for the reports of fibrillar-type nerve
endings.

The end plates are very refractory to staining. In locusts, where it is just
possible to microdissect down to the level of end plates, to make out their
outline, and even to pull them off the muscle fibers, no technique has been
found which will stain them at all satisfactorily. The fine nerve twiglets
enter the end plates at the immediate point of contact with the muscle fiber.
There the nerve sheath appears to become confluent with the muscle-fiber
sheath, the fine sarcolemma. At the point of contact there is often a large
nucleus or cluster of nuclei of the sheath, and it may be this cluster of nuclei
rather than the end plate proper which gives rise to the appearance of the
Doyere eminences. A fine trachea is also usually associated with this point.
The end plate is composed of about half a dozen fine tongues of granular
cytoplasm. These may be embedded in a matrix of finely granular sub-

78

INVERTEBRATE PHYSIOLOGY

stance. There are numerous large inclusions in the tongues of the end-plate
claws which are syncytial, having numerous small nuclei distributed
through the claws in a random manner. The end-plate claws do not pass
between the myofibrillae, merely resting on the surface of the fiber beneath
the sarcolemma. The attachment must be a very loose one, judging by the
ease with which the whole plate can be pulled off. The close similarity be-
tween locust endings (Hoyle, 1955a) and those of the homopteran
Cyclochila australasiae (Tiegs, 1955) suggest that this may be a common
type of ending in insects ; a generalized ending based principally on the
locust and homopteran endings is illustrated in Fig. 2. This bears a con-
siderable resemblance to the amphibian endings described by Couteaux
(1947). The terminations of the axons within the ending were, however,
not stained.

FATTy
eA/VBLOpB

CBLl-UJ-AR LAYf'K
Of^ SH£ATH

T/ZACHFA

Fig. 2. Drawing to show the structure of a doubly innervat-
ed locust motor end plate. The final branches of the axons did
not stain.

Multiple endings on single fibers were described clearly by Foettinger
(1880). He even obtained fixed specimens showing a series of local con-

NERVOUS CONTROL OF INSECT MUSCLES

79

tractions under the nerve endings. Marcu (1929) counted the endings on
fibers of thoracic muscles and found them at intervals of about SO/a in
Geotrupes, SO^i in Miisca. Weiant (unpublished), found endings at inter-
vals of about 40[x in Pcriplaneta leg muscles though she was unable to obtain
a clear picture of the individual endings. In the leg muscles of Locusta the
endings are about 60/* apart. This kind of innervation may conveniently
be called multiterminal innervation (Ripley, 1954).

The one, two, or three motor axons supplying a muscle unit travel to-
gether, bound by the same sheath right up to the point of contact, i.e., right

cur/CLi

APODBM
"slov^" f=/e/z£

MA//V A^jE-z^l/^ T/Zl/A/K.

Fig. 3. Diagram to illustrate the pattern of innervation of a doubly innervated
insect muscle unit, based on the locust extensor tibae unit.

80 INVERTEBRATE PHYSIOLOGY

into the end plates themselves. This was first demonstrated by Mangold
( 1905) for Decticus and has recently been confirmed for Locusta by Hoyle
(1955a) and Cyclochila, Erythronema, and other species by Tiegs (1955).
Anatomically (Tiegs) and physiologically (Hoyle) it has been shown
that, although one (usually the larger) fiber probably sends a branch to
every end plate, the others (smaller ones) do not. Each end plate on a par-
ticular fiber is, however, probably similarly innervated, though there may
be exceptions to this rule. This means that, whilst every muscle fiber re-
ceives a uniform supply from one axon, only a proportion of the fibers
receive supplies, again of similar kind, from the others. A diagram illustra-
ting the general pattern of the innervation of a two-axon muscle is shown
in Fig. 3. Many thoracic muscles are, however, said to have only a single
ending per muscle fiber, e.g.. Apis (Morison) and Erythroneura (Tiegs).
These will be discussed later.

Mechanical Responses

A consequence of the paucity of motor innervation is that special neuro-
muscular mechanisms must be used in order to efifect graded contraction
of the muscles. In spite of the demonstration of this economy of nerve
supply and the physiological exploration of the rather similar crustacean
system, workers in the insect field did not seem to appreciate this point. It
was not until 1939, when Pringle studied the motor mechanisms of the
cockroach leg, that the presence of special mechanisms in insect muscle was
adequately realized. Earlier workers (especially Kahn, 1916; Friedrich,
1933; Solf, 1931) had treated insect preparations as if they worked like
the frog gastrocnemius preparation. We may now interpret their results
in the light of recent knowledge as if they had stimulated only the "fast"
nerve fibers of their preparations. The adjective "fast" refers to the relative
speed of the resulting contraction and not to the velocity of conduction along
the motor axon. Actually the two fibers of doubly innervated muscles are
sometimes markedly difl:'erent in diameter, e.g., in Geotrupcs (Marcu,
1929) and in Cyclochila (Tiegs, 1955), and so probably have different
conduction velocities, in which the thickest and fastest axon is almost cer-
tainly also associated with the greatest speed of contraction. But the two
motor axons supplying the extensor tibiae of Locusta and Schistocerca
metathoracic legs are very similar in diameter, conduction velocity, thresh-
old, etc. The corresponding fibers of the mesothoracic legs, although pro-
ducing comparable mechanical responses, are different in thickness, 10-1 I/a
for the "fast" and only 6jx for the "slow," with conduction velocities of
2.2 and 1.5 meters per second respectively. Tiegs found some cases, e.g.,
the dorsal longitudinal muscles of Erythroneura, in which the fiber diame-
ters were about equal.

NERVOUS CONTROL OF INSECT MUSCLES 81

The maximum power of the locust jumping muscle is as great as 20 kg.
per gm., nearly ten times better than for mammalian muscle expressed in
the same terms. The only satisfactory way of comparing power develop-
ment for animals as different as insects and vertebrates is to do it in terms
of the mean cross-sectional area of individual muscle fibers. A rough esti-
mate for this value compared with frog sartorius shows that each locust
fiber develops about the same maximum tension as frog fibers. The differ-
ence in length of the fibers accounts for the apparent colossal strength of
the insect muscles.

The summation of successive contractions during an incomplete tetanus
is often very marked in insect muscle. The tetanus-twitch ratio in Decticus
muscles at 20° C was placed at 10/1 by Solf (1931), and it has a similar
value in all locust leg muscles for "fast" fiber stimulation. The ratio is in-
creased by lowering the temperature below about 15° C. The durations of
the twitches of leg muscles of Periplaneta, Dytiscus, Tettigonia, Locusta,
and ScJiistocerca all fall within the limits of 0.1 and 0.2 sec, i.e., of the same
order of magnitude as frog muscle. In Locusta the latent period between
the action potential peak and the onset of contraction in the relatively long-
fibered metathoracic flexor tibiae muscle is 2-3 msec. Peak twitch tension
is reached in 0.04 sec.

The fusion frequency is usually of the same order of magnitude as that
of frog muscle, i.e., about 20 per second, but some very high values have
been recorded. Kraemer (1929) gives a value of over 50 per second for
Dytiscus and Pringle ( 1939) over 70 per second for Periplaneta.

Stimulation of the second or "slow" motor axon was achieved by Pringle
(1939) in the cockroach metathoracic extensor tibiae muscle. After drying
out the metathoracic nerve trunk which he designated 3b and then remoist-
ening it, he found that the response of the muscle to maximal stimulation
of the nerve trunk was markedly changed. Initially there had been a twitch
following each stimulus and now there was no mechanical response at fre-
quencies below 30 per second, but a smooth extension of the tibia occur-
ring at higher frequencies. The response occurred at a precise threshold,
so was almost certainly due to a single axon. The rate of extension and
final tension increased with rate of stimulation up to frequencies of about
300 per second. The results showed clearly that the requirements of tonic
and slow actions could be met by a single axon, but they did not reveal the
mechanism of action of this axon.

"Slow" axons supplying the pro-, meso-, and metathoracic extensor
tibiae muscles of the locusts Locusta migratoria and Scliistocerca gregaria
leave their respective ganglia by separate nerve trunks. Branches contain-
ing the axons meet in the coxae before final branches, carrying only the
extensor tibiae nerve fibers, leave to supply these muscles, as illustrated in

82

INVERTEBRATE PHYSIOLOGY

Fig. 4. In each ganglion the nerve trunks concerned are the branches N3b
and N5. They are probably the homologues of the branches similarly desig-
nated in the metathoracic ganglion of Periplaneta by Pringle (1939). In
Albrecht's (1953) account of the Locusta nervous system the trunks are
designated N2 and N3, the actual second and fourth nerves not being men-
tioned in his description. The pro- and the mesothoracic N3b nerves carry
the single "fast" axons of the extensor tibiae muscles. The metathoracic
N3b carries a single "slow" axon of the metathoracic extensor tibiae, the
Si axon (Hoyle, 1955). The pro- and mesothoracic "slow" axons travel
in the N5 trunks whilst the metathoracic N5 carries the single metathoracic
extensor tibiae "fast" axon. This divided origin of the "slow" and "fast"
axons supplying the same muscle is extremely convenient, for it means that
the neuromuscular mechanisms of the muscles can be studied during sep-
arate stimulation of the innervating axons without the risk of damaging
them which is encountered when axons have to be separated by dissection.
The metathoracic N3b does, however, contain an additional axon supply-
ing the metathoracic extensor tibiae muscle. This axon (So) evokes no
mechanical response when stimulated alone, but efforts must be made to
avoid stimulating it if the effects of Si alone are to be observed.

2 /^P

rvi. e.t

Fig. 4. Diagram of the locust thoracic nerve ganglion chain showing the origins
of the "slow" and "fast" axons supplying the extensor tibiae muscles. F= "fast" axon ;
Sj^, S.„ "slow" axons; N^ a,b,c, N^, nerve trunks, m.e.t. = metathoracic extensor
tibiae, ms.e.t. = mesothoracic extensor tibiae, p.e.t. = prothoracic extensor tibiae.

Stimulation of the metathoracic extensor tibiae "slow" axon at fre-
quencies up to 10 per second produces no mechanical response except in
a few cases where a minute extension of the tibia may be seen to be asso-
ciated with each stimulus. A slow, smooth extension starts at about 15 per
second and increases in the rate of extension follow increasing frequency
of stimulation up to about 150 per second. The final tension produced dur-

NERVOUS CONTROL OF INSECT MUSCLES 83

ing prolonged stimulation increases with increasing frequency up to about
80 per second. These frequencies are about half the corresponding ones for
Periplaneta. The locust prothoracic and mesothoracic extensor tibiae pro-
duce final tensions at the tibial tips of barely 0.5 gm., but movement also
starts at about 15 per second in the absence of load and increases in rate
up to about 150 per second are evident.

The Mechanism of Transmission
Even in the smallest insects, skeletal muscle fibers are seldom less than
20/x in diameter and often are as large as lOO/t or even more; this means
that the intracellular recording technique can be used in studying trans-
mission. With the recent advances in technique a new standard is required
in work of this kind, so it seems desirable to lay down the principles which
should be follow^ed if the results are to contribute fully as comparative data.
They must be regarded as an ideal program rather than as limiting con-
ditions.

Anatomical Features Which Should Be Known

( 1 ) The nature of the muscle being studied, whether multiple or single
unit or diffuse, and whether the units have common or separate inner-
vation.

(2) The number of axons supplying the units being studied ; antidromic
stimulation is helpful in determining this.

(3) The approximate spacing of the end plates along the fibers.

Stimulation

( 1 ) Stimulation of separate axons. Where axons are neither separated
naturally nor separable by microdissection, difiterences in threshold and
conduction velocity should be utilized.

(2) Monitoring of nerve impulses. Tiny electrodes are particularly sub-
ject to polarization and a constant check should be made on the efficacy of
stimulation. Monitoring becomes essential when differences in threshold
or conduction velocity are being used to study separately axons contained
in a common trunk.

Recording
Intracellular recording from as many fibers as possible from all parts
of the muscle under consideration.

The "Fast" System
Rijlant (1932) found action potentials associated with vigorous spon-
taneous mechanical activity in the legs of Dytiscus and Hydrophilus ; the
potentials did not show any facilitation and were probably due to the ac-

84 INVERTEBRATE PHYSIOLOGY

tivity of a "fast" axon. Pringle (1939) found a similar type of nonfacili-
tating response when stimulating the nerve to the cockroach metathoracic
extensor tibiae muscle. The "fast" responses of several muscles of Locusta
migratoria have been studied in a series of papers (del Castillo, Hoyle, and
Machne, 1953; Hoyle, 1955b,c). Similar techniques have recently been
used in studies on Calliphora vomitoriu, Dytiscus marginalis, and Schisto-
cerca gregaria (Hoyle, unpublished). The "fast" responses have also been
studied in cockroach leg muscles (Wilson, 1954; Hoyle, 1955c), the wing
muscles of Locusta migratoria danica, Gampsocleis hurgeri, and Mecopoda
elongata and the sound muscles of the cicadas Graptopsaltria nigrojuscata
and Platy pleura kaenipferi (Hagiwara, 1953; Hagiwara and Watanabe,
1954).

In the extensor tibiae muscles of all the legs of the locusts Locusta
migratoria migratorioides and Schistocerca gregaria a sufficient number
of intracellular insertions has been made to make it possible to state con-
fidently that in them the single "fast" fibers innervate every muscle fiber.
This is probably true of all the other "fast" fiber systems in the locusts and
also applies to each unit of the locust flexor tibiae muscles and no doubt
also to the extensor trochanteris of Dytiscus (Kraemer, 1932) and others.

The responses of each muscle fiber are substantially similar. In some in-
stances a microelectrode has been used to record the responses in different
parts of a single muscle fiber, and these investigations showed that there
is very little difference between the responses recorded at the different
points. The fibers on which these studies have been made are all ones which
receive multiple nerve endings at intervals of less than 0.1 mm. along their
entire length.

The intracellularly recorded resting potentials average 60 mV (50-65
mV) in all Locusta and Schistocerca leg muscles in good condition when
bathed in a saline containing 10 mM K per liter, which is the lower limit
of their haemolymph potassium content. In zero-5 mM K per liter saline
the resting potentials approach 70 mV. In Gampsocleis wing muscles and
Platy pleura sound muscles the average resting potential is 60 mV in 3 mM
K per liter saline, whilst in Mecopoda and Graptopsaltria the recorded
values were only 42 mV (Hagiwara and Watanabe, 1954). In Calliphora
and Dytiscus values of 60 mV are common in 5 mM K per liter saline.
In Periplaneta flexor tibiae Wilson ( 1954) obtained resting potentials in
2.7 mM K per liter saline averaging 45 mV (30-70 mV ) . There are reasons
for being skeptical about the lower means of 42 and 45 niV, since these
result from several very low values of resting potential which were re-
corded and included in the analysis. In most insect haemolymph the
potassium concentration is rather high ; in the omnivorous cockroach it
may reach 30 mM and in the grass-feeding locust nymph 40 mM. Values

NERVOUS CONTROL OF INSECT MUSCLES 85

even twice these have been found in some species. Now the resting poten-
tial is directly related to the potassium gradient across the muscle-fiber
membrane and an increase in external potassium effects a reduction in the
resting potential (Hoyle, 1953b; Hagiwara and Watanabe, 1954). At
30 mM K per liter the mean resting potential of locust muscle fibers is
reduced to 35 mV, half the maximum possible value. Low resting potentials
may therefore be due to inadequate mixing between the haemolymph and
the saline in certain regions of the muscle. Theoretically there should be
little difference between the resting potential values of healthy fibers
bathed in the same saline, and so higher values of the range are probably
more nearly the correct ones. Fibers with a low resting potential may be
aging, fatigued, damaged, or partly depolarized by high local external
potassium. Results obtained from them must be examined in the light of
these possibilities. When Wilson ( 1954) found that the responses obtained
in the low-resting-potential fibers differed in several respects from those
obtained in the higher ones, he claimed that there were two dift'erent sorts
of fiber and that these should be associated with the "slow" and "fast" sys-
tems. He did not attempt to stimulate the "slow" and "fast" nerve fibers
separately, or even let the preparation do this for him, i.e., by leaving the
connections with the ganglion intact and recording during spontaneous ac-
tivity or reflex stimulation. Although he suggested that a high local ex-
ternal potassium concentration might be leading to the low values for rest-
ing potential, he did not carry out the obvious test and raise the potassium
level whilst recording from the large-resting-potential ("fast") fibers.
Had he done so he would have seen that this treatment converted his
"fast" fibers into "slow" ones. In other words, all Wilson's observations
were probably on "fast" fiber responses recorded from both high- and low-
resting-potential muscle fibers.

The "fast" responses of both the locust and cockroach muscles consist
of large depolarizations which often overshoot the zero potential base line.
Overshoots up to 20 mV may have been recorded from locust fibers having
large resting potentials. In many cases, however, the potential fails to over-
shoot or even quite reach the zero level ; this is almost always associated
with a low resting potential and so possibly with poor conditions. Hagiwara
(1953) and Hagiwara and Watanabe (1954) found similar responses in
the wing muscles of Oxya and the sound muscles of Graptopsaltria and
Platypleura, though overshoots were rarely observed except in Platy-
pleura.

W'hen locust muscle is gradually cooled, the time course of the response
lengthens and an obvious step appears in the rising phase. This step is
sometimes noticeable in the rising phase at ordinary temperatures, par-
ticularly with a fast time base. As the temperature drops to about 12° C

86 INVERTEBRATE PHYSIOLOGY

the step is not only very much more marked but the final hump of the action
potential, the part occurring after the step, is greatly reduced. At about
8° C it is completely absent. There remains a potential with a smooth, un-
stepped rise and an exponential decay. This potential is similar in shape to
the vertebrate end-plate potential and may be similarly designated in in-
sects. There is, however, a marked difference between the insect and the
vertebrate muscle. In the former the end-plate potential is not just confined
to a single site as in vertebrate muscle, and the latency of the response does
not differ in different parts of the fiber. These observations reflect the
nature of the innervation, i.e., the distributed end plates of the insect
muscle ; and the logical interpretation is that in the insects end-plate po-
tentials analogous to the vertebrate end-plate potentials are produced
nearly simultaneously in the several end-plate regions of the fiber. The
area involved is so large that depolarization occurs synchronously over the
whole surface of the fiber.

If the membrane potential is raised or lowered by passing polarizing or
depolarizing current across the muscle-fiber membrane through a second
intracellular electrode inserted in the same fiber, the magnitude of the
end-plate potential is correspondingly raised or lowered. There is a simple
linear relationship between end-plate potential and resting potential, the
line passing through the origin. This evidence strongly suggests that the
end-plate potential is due to the formation of a temporary short circuit of
the resting membrane, probably produced by a chemical substance released
under the end plate and increasing the permeability of the membrane to
several ions (cf. Fatt and Katz, 1951 ) . The effect of reducing the tempera-
ture is to show that the normal response has two components, a primary
junctional response or end-plate potential and a secondary, spike-like re-
sponse. The end-plate potential is probably due to a partial short circuit
of the membrane in the several areas on each muscle fiber in the immediate
vicinity of the end plates. If no further response occurred there would be
a simple exponential decay of the end-plate potential after the peak of the
transmitter action. Instead the depolarizing action affects the properties
of the resting membrane. This leads to a transient additional change in po-
tential which sums with the end-plate potential, thus generating the
secondary response. Typical responses from Calliphora, Periplaneta, and
Schist ocerca muscle fibers are illustrated in Fig. 5. The time courses of
pure end-plate potentials are indicated by dotted lines.

The secondary responses can be studied independently of the end-plate
potentials by passing depolarizing current across the muscle-fiber mem-
brane. They appear at and above a critical level and consist either of
oscillatory responses of small amplitude with frequencies from a few to
over 100 per second, or spike-like responses reaching a height up to 25 mV

NERVOUS CONTROL OF INSECT MUSCLES

87

+10 r

a,, h. c.

-10 -

-fcO

Fig. 5. The time course of typical insect muscle action po-
tentials recorded with an intracellular electrode at 20° C during
"fast"-fiber stimulation. Traces taken from fibers with 60 mV
resting potential from: a, Calliphora vomitoria; b, Periplaneta
americana; c, Schistocerca grcgaria. The probable time courses
of the pure end-plate potentials are shown with dotted lines to
emphasize the magnitude of the secondary responses and their
effect on recovery.

with a duration of 1-6 msec. Only the latter are comparable to those evoked
by the natural stimuli, the end-plate potentials. They may be evoked by
depolarization of about 20 mV. The surprising thing is that the spikes are
never larger than about 25 mV as measured from the depolarization
plateau, which is usually about 14 mV depolarization (45 mV membrane
potential level). Also they are very variable in both magnitude and dura-
tion even when recorded from the same site. Theoretically the largest ones
might just be capable of exciting the resting membrane and so setting
up propagation, but electric pulses of similar magnitude and duration to
the spikes rarely give rise to equally large responses. This means that the
insect secondary responses are only local events ; there is no evidence of
their being propagated, as Fatt and Katz ( 1953) have demonstrated propa-

88 INVERTEBRATE PHYSIOLOGY

gation of the somewhat similar though larger responses of some fibers in
Crustacea, and as occurs always in ordinary vertebrate skeletal muscle. In
vertebrate muscle the spike response may reach a height of 70 mV, three
times that of the locust.

Further information on insect transmission can be obtained by studying
the effects of potassium, calcium, and magnesium on the process. Raising
the magnesium or lowering the calcium in the bathing fluid has an effect
similar to cooling, in that the step in the rising phase of the action potential
which is probably due to a delay in the start of the spike responses is
markedly increased. This effect is partly due to a reduction in the magni-
tude of the end-plate potential. If the magnesium is raised or the calcium
lowered sufficiently, only the pure end-plate potential remains since it
becomes too small to evoke any spike response. The pure end-plate po-
tentials show considerable facilitation and summation. Potassium in excess
lowers the magnitudes of both the end-plate potentials and the secondary
responses. Its effect is partly indirect, due to the reduction of the resting
potential.

A typical locust response from a fiber with a 60 mV resting potential has
an overshoot of 13 mV. The total action potential of 7Z mV is composed
of 48 mV junctional response (peak height) and 25 mV spike response.
A similar fiber not showing an overshoot might have an action potential of
58 mV. This would be composed of about 40 mV junctional response and
18 mV spike response. The presence or absence of the overshoot is not an
important matter functionally, for fibers without an overshoot undoubtedly
twitch quite vigorously. When treatment with high magnesium, high
potassium, or calcium-free saline has reduced the response to a pure end-
plate potential of no more than 20 mV, there is still a small twitch contrac-
tion. On the other hand, a single muscle fiber shows a brisk local twitch
when current sufficient to evoke only a local response is passed. Evidently
either the end-plate potential or the local spike can activate the contractile
mechanism. It has not yet proved possible to affect the secondary response
experimentally without at the same time affecting the magnitude of the
end-plate potential, so an exact estimate of the part played by this response
in eliciting contraction is not possible at present. It seems probable that in
arthropods generally the link between membrane depolarization and ten-
sion is direct and progressive. A wide range of tension can be produced in
the same fiber, provided its membrane potential is lowered in graded steps.
Even the large "fast"-fiber action potentials, which are all-or-nothing
events, may nevertheless be required in quick trains before the muscle con-
tracts fully. Otherwise it is difficult to account for the high tetanus/twitch
ratio encountered even in short-fibered insect muscles where there is vir-
tually no connective tissue and the tendons are inelastic.

NERVOUS CONTROL OF INSECT MUSCLES 89

The "Slow" System

So far the "slow" systems have only been demonstrated in the extensor
tibiae muscles of the orthopterans Periplaneta, Locusta, and Schistocerca
(Pringle, 1939; Hoyle, 1953a and 1955b). A detailed investigation has
only been undertaken in the case of the Locusta metathoracic extensor
tibiae and the Schistocerca prothoracic and mesothoracic extensor tibiae
(Hoyle, unpublished). Using external recording, Pringle (1939) showed
that the action potentials of the "slow" system may facilitate considerably,
by as much as sixfold. In the locust metathoracic extensor tibiae the ex-
ternally recorded action potentials are extremely variable, depending on
the electrodes used and on their position (Hoyle, 1955b). Two types of
record may be observed, small spikes and slow shifts of potential. With
balanced input there is seldom any sign of facilitation, but with focal ex-
ternal recording, using a suitably small electrode, small spikes can be ob-
tained from many sites which clearly resemble the characteristic shape of
end-plate potentials. Some of these show considerable facilitation. At fre-
quencies above about 60 per second there is also summation. In other po-
sitions the potentials cannot be regarded as end-plate potentials, and show
neither facilitation nor summation. Under these circumstances intra-
cellular recording offers the only possible method of resolving the situa-
tion. With this technique many surprising features have become apparent
which could not even have been suspected from an analysis of records
from external electrodes.

Since every muscle fiber is innervated by the "fast" fiber, it follows that
"slow" activity must involve contractions of some or all of the same fibers.
The first surprising feature is that there is a response to "slow"-fiber stimu-
lation in only a fraction of the muscle fibers. In the Locusta metathoracic
extensor tibiae the fraction is about 30%. Partial innervation of this kind
has recently been recorded in crustacean muscles (Furshpan, 1955). The
innervation in the Locusta and Schistocerca prothoracic extensor tibiae
muscles is 40-50% and in the mesothoracic muscles about 40%. The second
surprising feature is the extremely wide range of magnitudes of the re-
sponses (Fig. 6). This is the case even when fibers with similar resting
potentials and similar "fast" action potentials are compared. In the Schis-
tocerca mesothoracic extensor tibae the responses vary from 2 to over 50
mV in dift'erent fibers of the same preparation. The smaller responses are
pure end-plate potentials, the larger ones compound. The largest of the
compound responses are almost as big as the "fast" responses of the same
fibers.

The largest responses show virtually no facilitation and do not summate ;
they are so similar to the "fast" responses that the same transmitter sub-

-Si

t6

w

<D

C

tn

c

^

U

n

rt

u

o

r-

n

H

0)

.H

^^

u

£•;:

<u

4S

OJ

OJ

ii>

2
IS

n

•a

a

O.

a!

e

ii

o

n

(U

c

3

s

O

ys

3
O

11)

x;

c

lU

u

&

6

In

3
O

o

_c

u

'•3
c

tn

H

n

n

O.

r^

n

u

b

o

rt

qi;

iJ

^

D

X

rS

fc

<u

<u

."S

lU

n)

(J

bJU

— ^

•r!

^-'

2 o oj rt i;

3 *- r- u =5

J= .id

E -a H 'S-

3

u

c

j_,

rrt

O

O

(U

c

(J

X

aj

!I)

rrt

rrt

u

X

n

en

aj

O

.^

tn

OJ

o

S
o

O

15

u

X3

"a t^ 3

; cnSH

u

^^1

|io

03 ci3 en

^ >~ „-

In *- 43

■fc5 ■=> .2

4-. T3

-M ^ ^

(U tT3 O

. •- 3 —
K '-S 3

§ - B

' *^ Si ><

Co

O tn

O bo 03

< y s

O (U

ho i:; rt 9i

.2 ^^

- o *-•
- .3 tn

(u aj

C •-

C .3 O o a^

NERVOUS CONTROL OF INSECT MUSCLES 91

stance might be involved, just released in smaller quantities. The smaller
responses often show quite marked facilitation. In a few instances this has
been as great as sixfold. The larger end-plate potentials show facilitation of
only a fraction, say 1/3, of the initial magnitude, but this, combined with
variation in magnitude of the secondary response, may lead to a doubling
of the total response, including occasionally an overshooting of the zero
base line. The magnitude of the "slow"-fiber end-plate potential can be
raised or lowered by raising or lowering the resting potential with the aid
of polarizing or depolarizing current. This is consistent with the view that
the end-plate potential is due to the release of a chemical transmitter sub-
stance which raises the permeability of the membrane in the end-plate re-
gions to some or all ions.

The situation in the extensor tibiae of the jumping leg appears to be
considerably more complex. The end-plate potential type of response is
obtained from only about 10% of all the muscle fibers. In about 20% there
is a response of a different kind. The response consists of a slow depolariza-
tion rising to a height of not more than 1 mV in about 50 msec, and declin-
ing in about 800 msec. These potentials show almost no facilitation, but they
summate during repetitive activation to produce a depolarization plateau.
The two types of responses were designed S^a and S^b respectively
(Hoyle, 1955b). The slow depolarizing fibers (S^a) contract smoothly at
all frequencies. Many of the larger end-plate potential-type fibers (S^b)
show small twitches in response to each stimulus and only contract
smoothly at higher frequencies characteristic of tetanus.

Ordinarily the process of mechanical excitation is linked to the multi-
terminal anatomical arrangement, but the local response evoked by
passing current through an intracellular electrode produces quite a marked
local twitch. It remains possible, therefore, that under tetanus even
a single end plate could elicit an appreciable mechanical response from
the whole fiber, especially in short fibers, and so a single end plate
might be the mode of innervation in some short-fibered insect muscles.
Morison (1928) and Tiegs (1955) have claimed that many insect muscles
receive only one end plate, at one end of the fiber ; Tiegs showed that they
may nevertheless receive two axons. Intracellular recordings from fibers
of this kind would be of considerable interest.

These results also raise the question as to whether or not insect muscle
can be excited directly. Many earlier workers, e.g., Heidermanns (1931),
Solf (1931), thought that in their experiments they were exciting the
muscle directly in the same way that frog sartorius or gastrocnemius can
be excited. The effectiveness of direct stimulation in frog muscle is due to
the fact that an adequate depolarizing current readily sets up an active
membrane response which is progagated as a spike of all-or-nothing char-

92 INVERTEBRATE PHYSIOLOGY

acter along the whole length of the muscle fiber. Since no insect muscle
has yet been shown to be capable of producing a propagated spike action
potential, the claims for direct excitation must be viewed with caution.
The presence of nerve branches throughout an insect muscle makes it
likely that external electrodes will excite these nerve fibers at a lower
threshold than that at which the muscle could be excited, and the presence
of simple "steps" in the response records of Solf (1931) and Kraemer
(1932) suggest that they were only stimulating the nerve. Roeder and
Weiant (1950) found that a muscle of Periplaneta became completely in-
excitable after the motor nerve had been cut and allowed to degenerate,
although the muscle fibers appeared to be still in good condition. Several
insect muscles cannot be made to contract unless a deliberate attempt is
made to stimulate them via the nerve (e.g., Tiegs, 1955).

Inhibition

Friedrich (1933) claimed to have demonstrated the presence of an
inhibitory nerve in Dixippiis leg. Actually all he obtained was a slight
"relaxation" of the resting tibia during stimulation below the threshold of
the exciting nerve. Ripley and Ewer (1951) argued that his effect could
have been obtained with a loosely held preparation by contraction of the
coxal muscles. No doubt other explanations are possible ; the phenomena
have certainly never been confirmed. Pringle was unable to find any evi-
dence for peripheral inhibition during his studies on cockroach prepara-
tions. Ripley and Ewer ( 1951 ) described a relaxation of the levator tarsus
muscle of Locust a when they raised the stimulus strength (applied to the
whole nerve trunk in the thorax) by threefold. I have studied the same
preparation in detail, monitoring nerve impulses and using intracellular
electrodes to record from the muscle. All the fibers in the nerve which
supply the tibia are excited between the limits threshold to threshold plus
25%. At around three times threshold an apparent inhibitory effect
appears, but this is due to failure to excite the nerve (Hoyle, 1955b),
probably because of polarization at the stimulating electrodes. These ob-
servations emphasize the need to monitor nerve impulses in this kind of
experiment. It is unlikely that observations on inhibition will be accepted
at the present time unless they are supported by experiments along the
lines indicated earlier in this paper.

There remains, however, the problem of the function of axons like the
third axon (So) supplying the locust jumping muscle (Hoyle, 1955a).
Since this is only about half as thick as the "slow" axon which travels in
the same trunk, difi^erences either in stimulus strength or in conduction
velocity can be utilized to stimulate it separately. At the same time the
"fast" axon can be separately stimulated and to an incomplete extent the

NERVOUS CONTROL OF INSECT MUSCLES 93

effect of interference between the third axon and the "slow" one can also
be studied. So far no combination of stimuli has been found which leads
to an inhibition of either the "slow" or the "fast" systems. The effect of
stimulating S- is to increase rather than to decrease the mechanical re-
sponse. Intracellular recording reveals that stimulation of So eft'ects a
small hyperpolarization in those muscle fibers which have a low resting
potential. The hyperpolarizations summate to raise the resting potential
up to but never beyond the value of 70 mV, which is probably near the
potassium equilibrium potential. Hyperpolarization of the resting mem-
brane is a property of the crustacean inhibitory nerve. The locust hyper-
polarizer axon is only present in the metathoracic leg, and it may perhaps
be an evolutionary relic of a once common inhibitor axon. Now^ its hyper-
polarizing function alone seems to remain ; this could be of value to insects
in which there is a fluctuating, often high, value of haemolymph potassium
as in grass-feeding locusts. The potassium tends to depolarize the muscle
fibers ; and any agent acting against this tendency would ensure a larger
end-plate potential and active membrane response, which both depend on
the resting potential, and in turn greater activation of the contractile
mechanism would follow.

This is merely speculative, but the presence of the locust S- axon is
significant. It must be taken to indicate that somewhere in the class there
may be fibers possessing the property of peripheral inhibition.

Locust Muscle : The Overall Picture

The locust pro- and mesothoracic extensor tibiae nuiscles are small and
functionally nonspecialized. They seldom support the weight of the body,
so are probably not called upon to produce prolonged activity. They are
concerned with providing some of the thrust needed during walking and
running and in checking quick movements of the antagonist flexors (which
normally take the weight of the thorax) . Their "slow" fibers are well suited
to most of these tasks. The graded end-plate potentials in the various fibers
ensure that some muscle fibers are ready with near-maximal twitches for
immediate work against the inertia of the system. Other fibers producing
little tension at the start of a train of excitation come in later as facilita-
tion and summation develop, to reinforce the first-acting fibers. A smooth
development of tension is always assured even when excitation is at high
frequency. Evidently the "slow"-fiber peak tension of 2/5 maximal "fast"-
fiber tension is adequate for all slow movements and tonus. The tension
can always be reinforced by calling in the "fast" system and so utilizing
the other 3/5 of the fibers. Perhaps in many movements the "fast" system
is used alone.

The metathoracic extensor tibiae muscles are specialized for the function

94 INVERTEBRATE PHYSIOLOGY

of jumping. So great is the power developed during a twitch due to a single
"fast"-fiber action that we can say at once that the F axon is used only in
jumping. All ordinary movements must utilize the Si axon alone. Ten per
cent of the available muscle fibers can produce a peak tension of 80 gm. in
the muscle, representing a thrust of 2 gm. at the foot. This is more than
adequate to shift the locust, itself weighing barely 2 gm., and 10% is the
total number of fibers innervated by the Sn, end-plate potential-type end-
ings. What then is the value of the Sia slow depolarizing endings? Do
these endings represent some sort of evolutionary degeneration? This is
not as fantastic as it may seem. The jumping muscles have been evolved
from muscles resembling the prothoracic and mesothoracic extensor tibiae
muscles where the "slow" system utilizes 2/5 of the fibers. Activation
of 2/5 of the jumping muscles produces considerable tension, enough to
give a small leap if developed rapidly, so a still smaller proportion of the
muscles would be quite adequate for the "slow" system. Only 1/5 of the
tension can actually be produced at all rapidly by the Si axon, since the
Sia fibers are activated so slowly ; an efi^ective reduction in innervation
has therefore been achieved. The metathoracic extensor tibiae is an im-
portant tonic muscle, unlike the prothoracic and mesothoracic muscles ; it
supports the climbing and vertical resting locust and raises the abdomen
during walking. The S^a type ending is probably well-suited to the tonic
control of the muscle.

In making these deductions we must realize that the evolution of the
powerful jumping apparatus required not simply an enlargement of the
limb and a habit of tucking the tibia under the femur. It was also necessary
to develop compatible neuromuscular mechanisms from the existing ones,
and the much more subtle but equally important problem of altering the
central nervous synapses appropriately had to be overcome at the same
time. The requirements of tonic and slow activity had to be met at all
stages and effected by a muscle which acquired the functions of its an-
tagonist whilst increasing its power by 20 times.

The Natural Control of Insect Muscle

The study of neuromuscular mechanisms is only a part of the much
broader field of the whole functioning of the motor apparatus, itself a
branch of the physiological study of animal behavior. Consequently, if we
can advance from a purely descriptive study of the events associated with
transmission to a study of the natural functioning of the muscles, we enter
in effect the realm of the central nervous system and make contact with
the automatic control systems associated with proprioceptive feedback.

With a view to obtaining information about the natural functioning of
locust muscles I have developed a technique which I shall describe briefly.

NERVOUS CONTROL OF INSECT MUSCLES 95

It consists of implanting small fixed electrodes through the cuticle, with
fine, insulated copper wires attached. The animal is allowed the freedom
of a 2 ft. sq. "Perspex" double- walled cage within which it trails the leads
along with it. This makes possible the study of the muscle impulses (elec-
tromyogram ) , which are amplified and recorded on moving paper by an
electroencephalograph machine, and observation of the animal and the
electrical activity in a particular muscle or muscles at the same time. Two
pairs of leads are usually possible without seriously disturbing the animal
and recordings can be made from antagonistic muscles.

Records from the metathoracic extensor tibiae muscle are particularly
easy to interpret, since they consist of a single series of action potentials
due to Si, except in the rare event of jumping. They show that impulses
effecting tonic contraction occur at the low rates of 10-20 per second but in
very irregular trains. Bursts of firing at rates up to 50 per second are
associated with maneuvering, and even in vigorous marching they do not
exceed peaks of 100 per second.

Records from the antagonist flexor tibiae muscle are complex, reflecting
the compound nature of this muscle. Activity consists probably of "slow"-
fiber activity in one or more units reinforced by bursts of "fast"-fiber ac-
tivity, again with one or more units in action.

The mesothoracic extensor tibiae muscle shows tonic activity at about
20 per second due to the "slow" fiber and occasionally increases up to about
100 per second during movement. The bursts of high-frequency activity
during movement are usually reinforced by brief bursts of 4-8 impulses
due to the "fast" fiber. The antagonist flexor muscle, like that of the meta-
thorax, reflects its multiple-unit nature by giving records of a similar,
complex nature.

Antagonist muscles are frequently used against each other, as Elftman
(1941) has calculated that they must be used in human walking, in order
effectively to decelerate a rapid movement, and in other ways.

Conclusion

Most of the work described had been done on locust limb muscles, but
evidence has been presented which indicates that the mechanisms utilized
in locusts are possibly present universally in insects in the skeletal muscles.
This is particularly likely to be true of the "fast" system, and there seems
no reason why the locust "slow" systems could not be utilized by other
insects ; in fact it does not seem necessary to require much modification of
the locust machinery to work say a Drosophila leg.

The "slow" system utilizes the same muscle fibers as the "fast" by bring-
ing a proportion of them into contraction, each in a progressive manner.
There is no justification for accepting Wilson's scheme of separate "slow"

96 INVERTEBRATE PHYSIOLOGY

and "fast" muscle fibers having different properties and separate nerve
supplies. The "slow" system works by virtue of the distributed end-plate
innervation which ensures a uniform depolarization of the muscle fiber
and an extensive, though graded, contraction in response to this. It is this
same distributed end-plate supply which enables insect muscle to function
in the extraordinary mineral environments which insect haemolymph often
presents, media in which the mechanism for establishing the vertebrate
propagated action potential would be immediately paralyzed.

Up to the present time no rational explanation has been offered as to
why there are these "slow"- and "fast"-fiber mechanisms in the large
Crustacea. The answer may lie in a consideration of the insects ; for, as has
been pointed out, the highly economical "slow"- and "fast"-fiber system is
admirably suited to the special problem of controlling a very small muscle
composed of few fibers. If we suppose that the ancestral arthropods from
which the Crustacea and insects were derived were very small animals and
that the "slow"/"fast"-fiber mechanisms were evolved to meet their needs,
then it is readily possible to explain their persistence in the larger deca-
pods. On the grounds of performance there is little to choose between
vertebrate and arthropod.

Summary

(1) The mechanism of production of smoothly controlled movement
by the muscles of insects is discussed in relation to their innervation. The
small number of muscle fibers available makes it necessary to seek some
different phenomena from those familiar in vertebrates.

(2) Each muscle in an insect is usually supplied by two axons and
itself constitutes a complete, single motor unit. Some muscles are how-
ever composed of a few units attached to a common apodeme, each unit of
which is served by separate axons.

(3) The axons usually supply several end plates to each muscle fiber,
evenly distributed along its length. In many cases individual end plates re-
ceive branches from each of the two axons.

(4) One of the two axons, the "fast" axon (F), evokes a large, twitch-
like response following each impulse. The other, the "slow" axon (S), pro-
duces a mechanical response only when impulses in excess of about 15 per
second pass down it. "Slow"-fiber responses always consist of smooth and
relatively slow contractions ; there is a very large increase in the rate of
contraction of the muscle with increasing frequency of stimulation of the
nerve and also in the final tension reached.

(5) The "fast" fiber innervates every muscle fiber and produces large
electrical responses. These are formed of large end-plate potentials and

NERVOUS CONTROL OF INSECT MUSCLES 97

brief, local, spike responses. In some fibers there is an overshoot of the
zero base line.

(6) The "slow" fiber innervates only a proportion of the muscle fibers,
sharing end plates with the "fast" fiber. Neuromuscular transmission is
effected by end-plate potentials of various magnitudes showing consid-
erable facilitation and summation.

(7) In the execution of natural movements there occur combinations
of long trains of "slow"-fiber activity with occasional bursts of "fast"-fiber
activity.

REFERENCES

Albrecht, F. O., 1953. The Anatomy of the Migratory Locust. Athlone Press, London.
Becht, G., and D. Dresden, 1956. Physiology of the locomotory muscles in the cock-
roach. Nature 177, 836-837.
del Castillo, J., G. Hoyle, and X. Machne, 1953. Neuromuscular transmission in a

locust. /. Physiol. 121, 539-547.
Couteaux, R., 1947. Contribution a I'etude de la synapse myoneurale, buisson de

Kuhne et plaque motrice. Rev. Canad. Biol. 6, 563-711.
Elftman, H., 1941. The action of muscles in the body. Biol. Synip. 3, 191-209.
Ewer, D. W., 1954. Personal communication.
Fatt, P., and B. Katz, 1951. An analysis of the end-plate potential recorded with an

intra-cellular electrode. /. Physiol. 115, 320-370.
Fatt, P., and B. Katz, 1953. The electrical properties of crustacean muscle fibers. /.

Physiol. 120, 171-204.
Foettinger, A., 1880. Sur les terminations des nerfs dans les muscles des Insectes.

Arch. Biol. 1, 279-304.
Friedrich, H., 1933. Nervenphysiologische Studien an Insekten. L Untersuchungen

iiber das Reizphysiologische Verhalten der Extremitaten von Dixippiis morosus.

Z. vergl. Physiol. 18, 536-561.
Furshpan, E. J., 1955. Studies on certain sensory and motor systems of decapod

crustaceans. Ph.D. thesis, Calif. Inst, of Technology.
Hagiwara, S., 1953. Neuro-muscular transmission in insects. Jap. J . Physiol. 3, 284-

296.
Hagiwara, S., and A. Watanabe, 1954. Action potential of insect muscle examined

with intra-cellular electrode. Jap. J . Physiol. 4, 65-78.
Harreveld, A. van, 1939. Doubly-, triply-, quadruply-, and quintuply-innervated crus-
tacean muscle. /. Conip. Neurol. 70, 285-296.
Heidermanns, C, 1932. Reizphysiologische Untersuchungen an der Flugmuskulatur

von Aeschna coerulca. Zool. Jahrb. 50, 1-31.
Hoyle, G., 1953a. "Slow" and "fast" nerve fibers in locusts. Nature 172, 165.
Hoyle, G., 1953b. Potassium ions and insect nerve muscle. /. Exp. Biol. 30, 121-135.
Hoyle, G., 1955a. The anatomy and innervation of locust skeletal muscle. Proc. Roy.

Soc.B 143,281-292.
Hoyle, G., 1955b. Neuromuscular mechanisms of a locust skeletal muscle. Proc. Roy.

Soc. B 143, 343-367.
Hoyle, G., 1955c. The efi'ects of some common cations on neuromuscular transmissioa

in insects. /. Phvsiol. 127, 90-103.

98 INVERTEBRATE PHYSIOLOGY

Kahn, R. H., 1916. Zur Physiologic der Insektenmuskeln. Arch, des Physiol. 165,

285-336.
Kraemer, F. K., 1932. Reizphysiologische Untersuchungen an Coleopteren-Muskula-

tur. Zool. Jahrb. 51, 321-396.
Kuffler, S. W., and E. M. Vaughan Williams, 1953. Small-nerve junction potentials.

The distribution of small motor nerves to frog skeletal muscle, and the membrane

characteristics of the fibers they innervate. /. Physiol. 121, 289-317.
Mangold, E., 1905. Untersuchungen iiber die Endigungen der Nerven in den quergest-

reiften Muskeln der Arthropoden. Z. allg. Physiol. 51, 135-205.
Marcu, O., 1929. Nervenendigungen an der Muskelfasern von Insekten. Anat. Ans.

67, 369-380.
Montalenti, G., 1928. Osservazioni sulle terminazioni delle trachee e dei nervi nella

fibra muscolare degli arthropodi. Boll. 1st. Zool. Univ. Roma 4, 133-150.
Morison, G. D., 1928. The muscles of the adult honey-bee (. 4 />!> mcUijcra L.). Quart.

J. Micr.Sci. 71, 395-463.
Pringle, J. W. S., 1939. The motor mechanism of the insect leg. /. Exp. Biol. 16, 220-

231.
Pringle, J. W. S., 1954. The mechanism of the myogenic rhythm of certain insect

striated muscles. /. Physiol. 124, 269-291.
Rijlant, P. Les manifestations electriques du tonus et des contractions voluntaires et

reflexes chez les Arthropodes. Compt. rend. Soc. Biol. Ill, 631-639.
Ripley, S. H., 1954. Neuromuscular mechanisms of the grasshopper, Romalea microp-

tera (Beauv). Ph.D. thesis, California Inst, of Technology.
Ripley, S. H., and D. W. Ewer, 1951. Peripheral inhiibtion in the locust. Nature 167,

106.
Roeder, K. S., and E. A. Weiant, 1950. The electrical and mechanical events of neuro-
muscular transmission in the cockroach, Pcriplaneta americana (L.). /. Exp. Biol.

27, 1-13.
Sherrington, C, 1906. The Integrative Action of the Nervous System. Yale Univer-
sity Press, New Haven.
Solf, v., 1932. Reizphysiologische Untersuchungen an Orthopterenmuskulatur. Zool.

Jahrb. 50, 175-264.
Tiegs, O. W., 1953. Innervation of voluntary muscle. Physiol. Rev. 33, 90-134.
Tiegs, O. W., 1955. The flight muscles of insects — their anatomy and histology ;

with some observations on the structure of striated muscle in general. Phil. Trans.

Roy. Soc. B 238, 221-348.
Wilson, V. J., 1954. Slow and fast responses in cockroach leg muscle. /. Exp. Biol. 31,

280-291.

MYOGENIC RHYTHMS

J. W. S. Pringle
Cambridge University

Rhythmic muscular activity is the most characteristic effector action in
higher animals. Historically, concentration of attention on the limb muscles
of land vertebrates and on other easily isolated muscles in the limbs of
arthropods has tended to produce a picture of muscle physiology which
may not be valid for the animal kingdom as a whole. This review, while
concentrating mainly on insect fibrillar muscle, will concern itself also with
other mechanisms by which rhythms of activity can be generated in muscu-
lar tissues.

The most convenient physiological classification of muscles is one based
on the ideas of Bozler (1948), who first distinguished "long-fibered"
muscles, in which each fiber is innervated by branches of a controlling
nerve fiber, and "short-fibered" muscles with more diffuse innervation
and the possibility of conduction of excitation from muscle fiber to muscle
fiber. Within each class there occur striated and unstriated examples,
correlated in general with speed of action. Thus, the somatic muscles of
most higher animals fall in the first class, whether they are striated as in
vertebrates and arthropods or unstriated as in the byssus retractors of
lamellibranch mollusks, and vertebrate visceral muscles fall in the second
class, whether they are striated as in the heart or unstriated as in the ureter.

Until recently the occurrence of a myogenic rhythm of activity (that is,
a rhythm in which nervous tissues play no essential role) was recognized
only for the class of short-fibered muscles, and in particular for the verte-
brate heart. In 1949 I called attention (Pringle, 1949) to the unusual
neuromuscular mechanism which appears to be present in the indirect
flight and haltere muscles of the higher Diptera, in which a high frequency
of muscular contraction is not accompanied by the synchronous muscle
potentials which are characteristic of the activity of other somatic striated
muscles. The evidence for the existence here of a novel type of rhythmic
mechanism was much extended by Roeder ( 1951), and at the same time
Boettiger and Furshpan ( 1950, 1952) were demonstrating that the skeletal
mechanical system of the fly thorax has by no means the simple lever action
described in most entomological textbooks. By the time of the appearance
of Chadwick's (1953) chapters on insect flight in Roeder's textbook, it
was clear that there was something very unusual in the biophysics of the
flight motor in certain higher insects.

[99]

100 INVERTEBRATE PHYSIOLOGY

Attempts in Cambridge to isolate an indirect flight muscle in order to
elucidate the mechanism of the rhythmic activity were unsuccessful, and
attention was therefore directed to the sound-producing muscle (the
tymbal muscle) of cicadas, which appeared to be histologically similar to
the indirect flight muscles. The results of this work (which was done in
Ceylon on Platypleura capitata) have now been published (Pringle
1954a,b). The hypothesis there put forward is somewhat different to that
suggested by the flight-muscle work in 1949 ; it is that these fibrillar muscles
do not differ from ordinary striated muscle in the way in which they de-
velop tension and can shorten after the arrival of motor nerve impulses,
but that, when they are connected to a nonlinear elastic system in the
skeleton which produces sudden shortening at a critical tension, a process
takes place in the contractile machinery ("de-activation by release") which
changes its properties for a short time to those of the unexcited muscle.
In this "de-activated" condition the muscle can be passively extended by
small forces such as the residual elasticity of the tymbal skeleton, and the
redevelopment of tension after the deactivation interval (aided perhaps
by a sudden stretch at the end of the interval) continues the activity in a
myogenic rhythm. It was suggested that Boettiger's click mechanism in
the dipteran thorax makes this explanation adequate also for the myogenic
rhythm of the indirect flight muscles.

It may perhaps be useful to discuss these ideas in more detail in order to
clarify some of the terminology used. Fig. 1 shows diagrammatically the
course of events which are supposed to take place when a long-fibered
muscle is excited through its motor nerve. The impulse travels along the
nerve fiber (1) to the neuromuscular junction (2) where it initiates the
junctional phenomena ; we are not here concerned with the nature of neuro-
muscular transmission. The junctional processes lead to a local depolariza-
tion of the surface membrane of the muscle fiber (3), which in turn may
produce a propagated action potential. In arthropods this appears to be
rare, but instead a multiplicity of nerve endings on each muscle fiber pro-
duce local depolarizations over a wide enough area of surface to have much
the same effect ; in any case the depolarization appears to be the event which
initiates further changes inside the muscle fiber. These surface changes
lead to events of largely unknown nature (4) which finally induce activity
in the contractile machinery (5) and may move the skeletal elements (6)
and the external environment.

In this complicated and only partially understood sequence of phe-
nomena it is easy to introduce confusion by inexact terminology. I would
like to suggest that we should agree on a definition of words. Excitation, by
analogy with the similar phenomenon occurring in nerve fibers, should be
limited to the processes taking place in the muscle surface membrane ;

MYOGENIC RHYTHMS

101

stimulation is already, by custom, the thing done to the excitable tissue
and excitation is its response. Excitation of the muscle fiber leads to an
active state in the muscle fiber as a whole ; this is the unknown process
labelled (4) in Fig. 1. Normally the active state produces, in turn, an

Fig. 1. Hypothetical course of events in ex-
citation of long-fibered muscle through its motor
nerve.

activation of the contractile mechanism (5), with development of tension
and/or contraction depending on the nature of the load. A corollary of
these definitions would be that tetanus means a maintained contraction or
a maintained tension and not a maintained active state which must be re-
ferred to as such. It is unfortunate that the terms used for processes (4)
and (5) should be so similar, and process (4) might perhaps better be
called priming of the contractile machinery by some form of coupling to
the surface events. The original terms as defined above will, however, be
used in this review. Clearly the distinction between active state and activa-
tion is only essential at present for insect fibrillar muscle, in which deacti-
vation by release aft'ects the contractile mechanism but not processes (3)
and (4). It may well prove, however, that such a distinction would be use-
ful also for other muscles.

102 INVERTEBRATE PHYSIOLOGY

The Evolution of the Myogenic Property
OF Insect Fibrillar Muscle

Boettiger (this symposium) has evidence that the myogenic type of
flight motor occurs in Diptera, Hymenoptera, and Coleoptera and in the
smaller Homoptera (but not in cicada flight muscle) ; it also occurs in the
tymbal muscle of cicadas of the genus Platypleura. The flight muscles of
other insects show the normal 1 :1 ratio between nerve impulses and muscu-
lar contractions (Roeder 1951). Hagiwara, Uchiyama, and Watanabe
( 1954) and Hagiwara ( 1956) have shown that a 1 :1 system occurs in the
tymbal muscles of certain cicadas, and in a survey of some of the larger
Japanese species they have found the fully developed myogenic system only
in Platypleura kaeuipjeri. In my paper on the physiology of sound produc-
tion (Pringle, 1954b), I suggested that the tymbal muscle might be an
evolutionary development of the metathoracic flight muscles, but I have
more recently made a detailed investigation of the anatomy of the sound-
production system in the Homoptera Auchenorrhyncha (Pringle, 1956)
and am forced to revise this opinion. There is a primitive genus of cicadas,
Tettigarcta, the two species of w4iich are found in the mountains of Aus-
tralia, and I have been able to obtain well-preserved specimens of both
sexes of each species. In the anatomy of its sound-production system,
Tettigarcta is intermediate between true cicadas and cercopids, which
have been well discribed by Ossiannilsson (1949). The whole sound-pro-
duction system appears to be morphologically in the first abdominal seg-
ment, as correctly stated by Myers ( 1928) ; and I have now concluded that
sound production has evolved not from the flight system but from a move-
ment made during copulation, which has become converted into a noncon-
tact communication system by the development first of a click system in the
first abdominal tergum (cercopids and Tettigarcta) and then by the
addition of a sound-frequency carrier as the damping of the clicks is re-
duced by the presence of internal resonant air sacs (true cicadas). There
is a correlated development of the receptor organ in these insects, Tetti-
garcta having chordotonal organs, probably sensitive to the click vibrations,
in the same morphological position as the tympanal chordotonal organs of
true cicadas.

This re-interpretation of the homologies, coupled with Hagiwara's evi-
dence that the myogenic system is present in the tymbal muscles of only
some cicada genera, makes it clear that the development of the myogenic
rhythm property in connection with sound production is an independent
evolution from its appearance in the flight motor system. Probably here
also it has evolved independently in the four insect orders in which it
occurs. This at once poses two questions :

MYOGENIC RHYTHMS 103

(1) How is functional continuity preserved in the change over from
a 1 : 1 to a myogenic system in flight and sound production ?

(2) Is deactivation by release a potential property of all muscle, re-
alizable whenever a suitable arrangement is developed in its skeletal attach-
ments, or is there a fundamental difference between the contractile mech-
anisms of fibrillar and nonfibrillar muscles ?

Let us examine the first question, dealing first with sound-production.
Functionally, it is not difficult to suggest reasons why a higher pulse fre-
quency is advantageous in the cicada system. Insect ears respond to sound
pulses with impulses at the pulse frequency ; and, if excitation of the female
through this nerve has a sexually exciting function, more intense excite-
ment should be produced by a higher pulse frequency in the male song.
Hagiwara (1953) and Hagiwara and Watanabe (1956) have described
an alternative method of at least doubling the pulse frequency in the song
of Graptopsalfna nigrojuscata, where the tymbals on opposite sides click
exactly in antiphase, the motor nerve discharge being driven by a gang-
lionic pacemaker. Physiologically, the requirement for effective deactiva-
tion by release appears to be a sufficiently rapid movement of the tymbal
at the in click ; since there is no antagonistic muscle whose contractions
must be coordinated with those of the tymbal muscle, extra clicks above
the 1 :1 ratio with nerve impulses introduce no difficulties in the mechanism
as the system evolves toward greater speed of tymbal movement.

With the flight motor the position is different. The orthopteran case may
be taken as typical of the primitive flight systems in insects, and fortunately
we have a considerable knowledge of the dynamics and physiology of flight
in locusts from the work of Ewer and Ripley (1953) and of Weis-Fogh
and Martin Jensen (1956), and in Periplaneta from Roeder (1951)
and Sotavalta (1954). The physiological picture of locust flight muscles
has been much confused by the experiments of Voskresenkaya (1947),
discussed by Chadwick (1953). This worker gave electrical stimuli
to a locust thorax in which the motor nerves were still attached to the
intact ganglion, and reported various types of rhythmic activity from
the flight muscles, not always correlated with the stimuli. Ewer and Ripley
(1953) were able to show that some of Voskresenkaya's results were due
to after-discharge from the ganglion, but they confirmed that there is not
an exact correlation between stimulus and contraction frequencies even in
the isolated nerve muscle preparation. Electrical recording from the tergo-
sternal muscle showed, however, that the most important phenomenon is a
long relative refractory period in the motor nerve so that, with near-thres-
old intensities of stimulation through platinum wire electrodes, impulses
may be generated at only every second or third stimulus as the stimulus

104 INVERTEBRATE PHYSIOLOGY

frequency is increased, and the frequency of impulses actually reaching the
muscle is always sufficiently low to produce discrete contractions. Only
with stimulus intensities well above threshold could the nerve-impulse fre-
quency be raised to the level which produces a smooth tetanus in the muscle.
There is thus no evidence in the locust tergosternal muscle of any inherent
muscular rhythmicity.

It is probably useful to try to start right from the beginning in a con-
sideration of the evolution of insect flight. There is fossil evidence that the
first winged insects were large creatures with well-developed locomotor
habits and that their wings first arose as lateral expansions of the terga of
the thoracic segments. Wings thus first appeared in insects as new struc-
tures and not as a modification in function of an existing appendage as in
the birds. At first the lateral thoracic expansions of insects must have pro-
vided lift as fixed aerofoils, and a control of their incidence is therefore the
first requirement in a system of control. Chadwick ( 1953) has argued and
Weis-Fogh (1956, Pt. IV) has now demonstrated in the locust that the
main control of lift in flight is by a reflex system adjusting wing incidence in
certain portions of the stroke. This then may be thought to be the primitive
control mechanism, deriving from the time when the wings were fixed
aerofoils and preserved in the make-up of insects ever since. Weis-Fogh
has pointed out to me that a wing-folding mechanism may have been the
second feature to appear, since large lateral expansions must have been
difficult to manage under all conditions ; this also is preserved in modern
flapping forms. Only later, perhaps, did the flapping machinery arise.
Birds, by contrast, probably flapped their wings from the start, using the
locomotor muscles of the fore limbs of their running reptilian ancestors.

The next problem concerns the nature of the rhythmic mechanism. The
more primitive insects have flight muscles which give a normal single
twitch for a single motor nerve impulse (Roeder's 1 :1 muscle), and if
the early insects were large there is initially no requirement for a very
high frequency of wing beat and for a very short muscle twitch. A number
of workers have studied the efl^ect on wing-beat frequency of alterations in
the loading and inertia of the wings ; in 1:1 systems these also provide evi-
dence about the nature of the neurogenic rhythmic mechanism, since the
frequency of motor nerve impulses corresponds to that of the wing beat.
The problem is whether the rhythm-generating mechanism is innate in the
ganglion or involves reflexes from wing sense organs. Roeder (1951) re-
ports that in Periplaneta and in A gratis (Lepidoptera) amputation of the
wings produces little change or a decrease in frequency. Sotavalta ( 1954)
confirms that there is no change in Periplaneta, but records a slight in-
crease (up to 20%) in Lepidoptera. Tiegs (1955) also finds an in-
crease in various noctuids, Neuroptera, and Isoptera (all 1:1 systems).

MYOGENIC RHYTHMS 105

Weis-Fogh (unpublished) tells me that in Schist ocerca the usual result of
amputation is a pronounced irregularity in the wing beat, but that some
rhythmic activity can still occur when the wings are reduced to mere
stumps. One difficulty in intepreting results of this sort on otherwise intact
insects is that we now know^ that the incidence-control reflex must have
been operating ; and, since loading as well as inertia must aflfect the magni-
tude of the forces at the wing base to which sense organs must respond, it
is not easy to predict the nature of the change produced by amputation on
the form and intensity of sensory excitation. It appears likely, however,
that there is, at least in some of the 1 :1 groups, a reflex effect from the
wing sense organs on the rhythmic mechanism, but that something in the
nature of a pacemaker is present in the ganglion capable of discharging
rhythmic bursts of motor nerve impulses even in the absence of sensory
feedback. It is relevant to note here that in Diptera (with a myogenic
rhythm in their indirect flight muscles) amputation produces a fall in the
frequency of motor nerve impulses (Roeder, 1951), although it leads to
a large rise in wing-beat frequency. In all 1:1 insects it is necessary to
suppose that there is a considerable measure of inherent organization in the
flight motor centers in the thoracic ganglion producing coordination of
activity in the various muscles which move the wings; Tiegs (1955) has
emphasized that the evolution of the flight mechanism has resulted in a
reduction rather than an increase in the number of muscles involved.

The first rhythmic mechanism in the insect wing system was thus
probably of a nature similar to that found in the rhythms of swimming and
locomotion, where there is an inherent pattern of coordination in the central
nervous system, but where reflex feedback is always very important and
may sometimes be essential (toad; Gray and Lissmann, 1946). As some
insects got smaller, the frequency of action demanded from this system
increased to the point where it became difficult for the flight muscles to
perform one complete contraction and relaxation during a single beat. This
seems to be the present condition in some Orthoptera, Lepidoptera, and
Odonata. Tiegs (1955) has recorded a normal wing-beat frequency of
57/sec. in the hawkmoth Hippotion. By direct observation he finds that
electrical stimulation at 30/sec. produces a partial tetanus, although com-
plete fusion of twitches is found only at 70/sec. Heidermans (1931) re-
ported a similar condition of partial tetanus in Odonata. Weis-Fogh (un-
published) tells me that in Schistoccrca stimulation experiments suggest
a similar result, but that this is misleading, since there is a rise of tempera-
ture of about 8° C in the thorax of a flying locust, and that at the real tem-
perature of the muscles during flight there is time for a complete contrac-
tion and relaxation ; twitch duration is known to have a high temperature
coefficient in frog muscle (Hill, 1951).

106 INVERTEBRATE PHYSIOLOGY

It is probably at this point in the story that the click mechanism, so beau-
tifully demonstrated by Boettiger in Diptera, begins to be important.
Weis-Fogh (unpublished) has recently obtained some results which help
greatly for an understanding of the origin of this peculiar feature of the
wing articulation. He has shown that in the thorax of Schistocerca the re-
lationship between applied force and wing displacement is not linear in
either wing, and that in the hind wing there is a range of instability which
amounts to a click mechanism ; the fore wing shows what may well be an
earlier evolutionary stage. This demonstration of a click mechanism in an
insect whose flight muscles are of the 1 :1 type suggests that even here we
may have some measure of deactivation by release in the flight muscles.
The instability has the obvious function of increasing the velocity of wing
movement above that obtainable with a simple lever action, but it may also
assist muscular relaxation. If it could be confirmed that there is a similar
instability in the lepidopteran wing articulation, the difficulty of a partial
tetanus in the flight muscles would be resolved ; for the deactivation pro-
cess could remove the tetanic tension during the phase of elongation of the
muscle, and the redeveloped tension once deactivation had worn off would
assist the true twitch tension in the next stroke.

The evolutionary picture is thus of an increase in the neurogenic fre-
quency of wing beat beyond the limits set by tetanic fusion in the muscles,
with partial deactivation by release ensuring separate contractions. From
this to a myogenic rhythm is a small step. Once the myogenic alternation is
assured, the motor nerve impulse frequency can drop back to a low level,
with a continued evolution of high-frequency muscular activity. The articu-
lation mechanics now continue to evolve in some orders to allow increas-
ingly isometric contraction of the flight muscles, apparently a necessity for
the very high-frequency wing beats of the smaller Diptera and Hymenop-
tera. In the higher Diptera the indirect flight muscles do not shorten when
detached from the exoskeleton (Tiegs, 1955), and give the appearance of
being inexcitable by electrical stimuli applied in the mutilated thorax whose
elastic properties have been disturbed. The Homoptera have not evolved
so far in the direction of isometry ; Tiegs (1955) finds that in the jassid
Enrymela faradic stimulation produces an easily visible shortening of the
tergosternal flight muscles, although they have the typical fibrillar struc-
ture of myogenically rhythmic muscles.

Parenthetically it is interesting to consider the cicadas, whose flight
muscles, unlike most of the Homoptera, are of the 1 :1 type. Tiegs (1955)
emphasizes that in Cyclochila (Cicadidae), as in Siphanta (Flatidae), the
flight muscles are intermediate in histological structure between normal
insect muscles and the fibrillar type. Cicadas and other Homoptera are
known (Snodgrass, 1927) to have a peculiar anatomical arrangement of

MYOGENIC RHYTHMS 107

their mesothoracic tergal muscles. Part of this muscle has the normal
longitudinal arrangement and is functionally a wing depressor. But an-
other part, more laterally situated, has an oblique orientation due to the
carrying down of its posterior attachment on the very well-developed
mesothoracic phragma ; this muscle is the main wing levator. Such a re-
versal of the function of a muscle would be almost impossible to understand
if functional continuity is to be preserved in a 1 :1 excitation mechanism ; it
would imply a change over of ganglionic connections of an unprecedented
nature. If, however, the cicadas have passed in their evolution through a
myogenically rhythmic stage, there is no difficulty in understanding such
a change in the timing of the muscular contractions ; in a myogenic system
the timing of contractions is determined by the mechanical conditions, not
by the ganglionic connections of the motor nerves. The fact that Homoptera
have not evolved very far towards an isometric contraction of their flight
muscles has thus allowed the cicada muscles to revert to the 1 : 1 excitatory
mechanism after the oblique tergal muscle had assumed its modern orienta-
tion and role.

Tiegs (1955) has described and illustrated some further features of the
histolog}^ of fibrillar flight muscle which are relevant to this evolutionary
story. He has resolved the problem of the sarcolemma, described by recent
workers as being absent in the bee and in Drosophila. It is, in fact, present
in all cases and is a true cell membrane ; the discrepancy has arisen from
the fact that many apparent "cells" in the transverse section are merely
areas delimited by intracellular tracheae within a very large cell. He has
established that the fibrils which can be isolated even from fresh muscle
are bundles of myofibrils for which he proposes the term "sarcostyle" ; each
sarcostyle in Diptera is formed ontogenetically by the incorporation of a
nucleated myoblast into the syncytial muscle cell. Finally he has shown
that there is an important difference in the histology of the motor nerve
ending as between normal insect muscle and the dipteran fibrillar type ; in
normal muscle the motor nerve ends in "Doyere cones" (apparently ana-
logous with the vertebrate end plates) on the surface of the fiber, but in
dipteran fibrillar muscle the nerve actually penetrates the giant muscle cells
and there are no end plates. How this last observation is to be correlated
with the muscle spikes recorded by Pringle (1949) and Roeder (1951) is
not clear, but it may represent yet another peculiar feature of these very
remarkable tissues.

Myogenic Rhythms in General

The occurrence of a myogenic rhythmic system in many different orders
of insects suggests, as has already been stated, that it has evolved many
times. It is always, so far as is known, correlated with the presence of large

108 INVERTEBRATE PHYSIOLOGY

sarcostyles, a sarcoplasm of low viscosity and with nvimerous large sarco-
somes. Perhaps the most striking case, for which we have at present only
histological evidence, is in the delphacid Perkinsiclla (Tiegs, 1955), where
a tergal abdominal muscle appears to have been drawn into the complex of
flight muscles as a wing levator and to have become fibrillar in structure.
It seems clear that in insects, at least, there is a potential deactivation-by-
release mechanism present in all striated muscle. Is this true of all muscles
in the animal kingdom ?

It is important to emphasize first of all that this type of myogenic rhythm
has little in common with the well-known myogenic rhythm of the verte-
brate heart. In insect fibrillar muscle, the rhythm derives from a property
of the contractile fibrils in the interior of the muscle fiber, and there is little
or no synchronous change in the permeability properties of the external
membrane as measured by electrical potentials across the cell surface. In
heart muscle, on the other hand, the rhythm of mechanical activity is
accompanied by a rhythmic change in membrane potential, and it seems to
be established beyond reasonable doubt that the rhythm is determined by
these membrane properties. Del Castillo and Katz (1955) and Hutter and
Trautwein (1955) have independently demonstrated that the cardio-
inhibitor and cardio-accelerator fibers in the vagus and sympathetic nerves
produce their effect by modification of the membrane potential of the heart
muscle fibers in the usual directions ; the spikes of activity originate at a
critical level of spontaneous slow depolarization during diastole. At least
in the mammalian heart it is also clear that the mechanical stimulus of
stretch produced by an increased venous return does not affect the fre-
quency of beat in the denervated heart ; control of frequency of beat
is a reflex phenomenon from mechanoreceptors in the great veins
and aorta.

In some vertebrate smooth (short-fibered) muscles we also have a myo-
genic rhythm (Bozler, 1948), and here mechanical conditions do affect
the frequency of contractions. Our knowledge of the physiology of these
tissues has lagged seriously behind that for striated muscle, but it seems
probable that here, as in the heart, the rhythmicity derives ultimately from
rhythmic properties of the cell membranes ; the form of action potentials is
similar to those of heart muscle. Conduction of excitation can take place
through the "muscle net" without the involvement of nerves, but is not all-
or-none as in the heart ; whether the effect of mechanical stimuli is mainly
on the membrane properties of the individual muscle cells, or on the inter-
cellular conduction mechanism, remains to be elucidated.

Myogenic rhythms also occur in the hearts of some arthropods (Krijgs-
man, 1952) and moUusks (Krijgsman and Divaris, 1955) ; we have no
knowledge of the membrane potential changes in these tissues, but their

MYOGENIC RHYTHMS 109

rhythmicity is affected by stretch. Pharmacological evidence suggests that
the rhythmicity originates in the surface membrane.

A somewhat different type of myogenic rhythm occurs in the embryonic
skeletal muscle fibers of various lower vertebrates ( Harris and Whiting,
1954; Harris, 1955). In the embryo dogfish the myotome muscle fibers
show rhythmic contractions before motor or sensory nerve fibers reach
them. The contractions are regular and synchronous down the whole
column of one side, but there is no correlation in the timing of contractions
on the two sides. At this stage acetylcholine accelerates the rhythm. When
nerve fipers first reach the muscles they have what Harris calls a "neuro-
cratic" action ; the frequency of the myogenic rhythm is increased, still
without bilateral coordination. Finally true coordinated movements occur
at about the time the sensory nerve supply develops. Electrical recordings
have not been made from these delicate tissues, but Whiting (personal
communication) reports that they are unexpectedly inexcitable by elec-
trical stimuli. Were it not for this observation one would be tempted to
conclude that here again the rhythmicity comes basically from the cell
membranes ; the influence of acetylcholine and the exact synchrony of con-
tractions down the whole column of one side are difficult to explain except
in terms of an impulse mechanism, propagated from muscle cell to muscle
cell ; there is a significant overlap of fibers across the myotome boundaries
at the myogenic rhythm stage, which disappears later.

Is there, then, anywhere in the animal kingdom other than the insects
any sign of a rhythmically contractile tissue in which the rhythm does not
derive from properties of the cell surface ? Rhythmic movement in the lower
animals is usually required for swimming, and Gray (1953) has recently
reviewed his work on this type of propulsion. We know virtually nothing
about the neuromuscular physiolog}' of these movements in any inverte-
brate and it is dangerous to argue too closely by analog}^ from fish and
snakes. Gray (1951, 1955) has shown that even in bacteria and sperma-
tozoa there is a great similarity in the dynamics of swimming to the move-
ments of higher animals, and there is thus probably a strong functionally
conditioned isomorphism between all propulsive mechanisms of this type.
In bacteria and spermatozoa it is hardly possible to conceive of anything
akin to a reflex responsible for rhythmicity or coordination, and here at
least there is a strong prima facie case for looking in the mechanism of con-
traction for the origin of the rhythm. There is no device here for any sudden
release of tension in an active tissue, and if the rhythm is myogenic the
effective mechanical stimulus must be a relatively slow one. The possibility
of a nonnervous origin for the rhythm of swimming should not auto-
matically be excluded even for Metazoa ; a lot could be achieved by a myo-
genic rhythm with "neurocratic" control.

110 INVERTEBRATE PHYSIOLOGY

The Mechanism of Deactivation

In Pringle (1954a) I discussed briefly the possible relationship of the
phenomenon of deactivation by release in insect fibrillar muscle to the me-
chanical properties of vertebrate striated muscle. It is well known that a
sudden release given to an excited muscle leads, if its amplitude of release
is sufficient, to the complete disappearance of tension, followed by its re-
development at a rate comparable to that at the beginning of excitation.
A. V. Hill has interpreted this and other results in terms of a "series elastic
component" in the muscle which goes slack when there is a sudden release.
In the experiments of Gasser and Hill ( 1924) on the frog sartorius muscle,
a release of 10% of the resting length was required to produce a tension
drop to zero, but Hill ( 1950) states that a smaller figure would have been
obtained if due allowance had been made for the elasticity of the suspen-
sion. This disappearance of tension on quick release resembles that found
with the cicada tymbal muscle, but the fact that the tymbal muscle is there-
upon re-extended to its initial length without immediate reappearance of
the original tension makes it impossible to explain the result simply in
terms of a series elastic component. The deactivation process here occurs
with a release of 1.5% of the resting length, showing that any series elastic
component present is very small, a result perhaps understandable in terms
of the different histology of the muscle attachments in insects and verte-
brates.

Any hypothesis about the nature of deactivation by release must take
into account the normal features of muscular contraction (Wilkie, 1954),
the extensive studies of glycerinated muscle fibers (Weber and Portzehl,
1954; Morales, Botts, Blum, and Hill, 1955), and the pecularities of
fibrillar muscle (Tiegs, 1955), including the enzymatic properties of the
large sarcosomes (Watanabe and Williams, 1951). In the absence of any
general agreement on the nature of the biochemical and biophysical ma-
chinery involved in contraction, in spite of intensive current study, hypo-
theses are particularly vulnerable but may nevertheless be attempted in
the hope that the properties of this tissue may throw light on the general
problem.

A generalized if somewhat naive view would be that contraction and
tension development occur in a muscle because excitation in some way
allows access to certain sites in the actomyosin complex of a high-energy
substance liberated by the sarcosomes. It is already clear that maintenance
of tension in striated muscle involves the continuous expenditure of energy,
and the sites are therefore presumably occupied and vacated cyclically, with
breakdown of the high-energy substance. The fact that the additional heat
of shortening is proportional to the distance shortened could be explained

MYOGENIC RHYTHMS 111

if the rate of vacation of the sites is directly proportional to the distance
through which the muscle has shortened ; in addition to the number re-
quired in any unit of time to maintain tension, a further number of high-
energy molecules would then be needed to achieve shortening.

One hypothesis to account for deactivation by release can be stated as
follows. If a sudden shortening occurs in a muscle due to the sudden re-
lease of tension, a large number of sites on the actomyosin complex will
be suddenly vacated. Supply of high-energy molecules by diffusion from
the sarcosomes may be inadequate to provide for their re-occupation imme-
diately following this large demand, and the state of the muscle would then
become temporarily similar to the inexcited condition. This would produce
deactivation by release. Whether or not the deactivation phenomenon will
occur depends, according to this hypothesis, on the relative rates of vaca-
tion of active sites by rapid shortening and of resupply of high-energy
molecules by diffusion from the sarcosomes. If the maximum possible rate
of vacation is low and the diffusion pathway small, complete deactivation
can never occur. If the rate of vacation is large and the diffusion pathway
relatively long, deactivation may last for sufficiently long to allow the
myofibrils to be re-extended to their initial length and a myogenic rhythm
of activity becomes a possibility. A rough calculation of probable ATP
diffusion rates over the distances involved shows that a time lag of a few
milliseconds in re-establishing the local concentration in the myofibrils is
a possibility.

Consistent with this hypothesis is the histological observation that the
"fibrils" are thick in insect muscles showing a myogenic rhythm. Tiegs
( 1955) has shown that the fibrils which can be isolated from these fibrillar
muscle are, in fact, sarcostyles composed of several myofibrils, but this does
not alter the fact that the large sarcosomes are situated between the sarco-
styles with a relatively long diff'usion pathway. The diameter of the sarco-
styles in fibrillar muscles of Hymenoptera, Coleoptera, and jassids is from
3 to 5/x, by contrast with values of less than 1/a for orthopteran wing
muscles, insect limb muscles, and vertebrate striated muscle. This differ-
ence and the large sarcosomes are the only known histological features
consistently associated with the myogenic rhythm property.

A hypothesis of this sort demands no sudden change in the basic mech-
anism of contractility in the change from the 1 :1 state to the myogenic
rh}-thm, and there is thus no difficulty in the occurrence of rhythms in many
different evolutionary lines in the insects. It makes no suggestion about
the nature of the active state in the muscle fiber during which access to
the sites on the actomyosin complex is possible by the high-energy mole-
cules. Pringle ( 1954a) showed that the duration of the active state follow-
ing the arrival of a single nerve impulse is not markedly different, whether

112 INVERTEBRATE PHYSIOLOGY

the mechanical response is a single twitch (isolated muscle) or a train of
myogenic contractions.

This first hypothesis has been stated in terms of a complex between
actomyosin and the high-energy molecule leading to contraction or the
development of tension. According to Weber and Portzehl ( 1954) the evi-
dence from the glycerinated muscle model is that tension is produced when
ATP bound in the fibers is split ; ATP whose splitting is prevented by
— SH poisons is the most effective plasticizing agent for the fiber model.
The resting state is with ATP bound to the actomyosin and the fibrils
fully plastic and extensible. In the fiber model ATP splitting is inhibited
at the concentrations above a certain value but still in the physiological
range. Weber supposes that, when the muscle is brought into the active
state, ATP splitting starts because of a shift in this critical inhibitory con-
centration and tension then develops.

A second hypothesis may now be stated consistent with the view that
tension development accompanies the splitting and not the binding of
ATP. Again it must be supposed that the rate of splitting depends on the
rate of shortening. A sudden shortening on quick release therefore produces
a large and nearly synchronous splitting of ATP, and the re-occupation of
the sites gives an initial plasticizing efTect ; the muscle therefore behaves
as if it was relaxed for a short interval of time until splitting again occurs.
Possibly the sudden stretch of the out click of the tymbal or the opposite
stroke of the wing again initiates splitting and the development of tension.
It is necessary in order that this cycle of events shall occur that the splitting
of ATP with the development of tension does not immediately follow the
binding of ATP on the actomysin complex, so that the re-occupation of the
sites by ATP shall have time to produce a plasticizing effect before splitting
has again proceeded far enough to produce an appreciable tension. There is
here no diffusion lag producing the oscillation ; the necessary condition is
that the splitting shall follow the binding of ATP with a finite time lag
even in the active state.

A lag between binding of ATP and splitting with development of ten-
sion is not inconsistent with the result of experiments on normal striated
muscle. When a vertebrate striated muscle is excited by stimulation of its
motor nerve, the twitch tension does not appear for several milliseconds.
Hill (1949) has explained this lag in terms of the stretching of the series
elastic component, and has shown that, if quick stretches are given in
addition to stimulation, a change in the mechanical properties of the muscle
can be detected much earlier ; he concludes that activation starts almost
immediately upon excitation. It would be equally possible to interpret this
experiment as indicating a normal slow development of activity in the con-
tractile mechanism, but that quick stretch accelerated the activation. There

MYOGENIC RHYTHMS 113

is, so far as I am aware, no other evidence that activation occurs in verte-
brate muscle more rapidly than is indicated by the development of tension.
Experiments by Weber and others on the glycerinated fiber model are
unable to produce evidence about the initial rate of splitting of ATP and
the development of tension since the time course of events is here always
limited by inward diffusion of ATP into the model.

The structural peculiarities of fibrillar muscle would be correlated, on
this type of hypothesis, with a higher degree of lateral association of muscle
elements than in normal muscle, so that plastic behavior is reduced to a
minimum ; mechanical events are thus transmitted more completely to the
contractile machinery and synchronization of the energy-yielding cycles is
more perfect after quick release. Boettiger's evidence ( this symposium) of
a high coefficient of elasticity in passive flight muscle may point to the
correctness of this correlation. Clarification of the biochemical nature of
the deactivation cycle must, however, await further quantitative studies.

Summary

( 1 ) Myogenic rhythms of activity have been described in the heart and
certain visceral short-fibered muscles of vertebrates, in the hearts of
moUusks and some arthropods, in the striated trunk muscles of the dogfish
embryo, and in the indirect flight and tymbal muscles of certain insects
(fibrillar muscles). They may also occur in the swimming movements of
micro-organisms and spermatozoa.

(2) In vertebrate short-fibered muscles the rhythmic property resides
essentially in the surface membrane of the fibers ; potential changes are
observed synchronous with the contractions. In insect fibrillar muscles
myogenic rhythmicity normally occurs in the presence of a nonlinear elas-
tic ("click") mechanism in the muscular coupling, and depends on deacti-
vation of the contractile fibrils, with no synchronous potential changes in
the fiber membrane. The mechanism of rhythmicity is unknown in dogfish
embryo muscle and in the contractile filaments of micro-organisms and
spermatozoa.

(3) Fibrillar structure is correlated with myogenic rhythmicity in in-
sects, but has evolved independently several times in the motor systems for
flight and sound production. In Homoptera, whose sound-production
mechanism in the first abdominal segment may be derived ultimately from
movements during copulation, the course of the evolutionary change from a
neurogenic to a myogenic rhythm can be understood without difficulty. In
the flight system it is suggested that the deactivation phenomenon was first
useful as a method of achieving a high-frequency rhythm of neurogenic
wing beat and later led to a myogenic rhythm in several different orders.

114 INVERTEBRATE PHYSIOLOGY

In cicadas the flight muscles have possibly evolved through a myogenic
stage back to a 1 :1 system.

(4) Two tentative hypotheses are put forward to explain the phe-
nomenon of deactivation by release. One depends on a longer diffusion
pathway between sarcosomes and the interior of the sarcostyles in fibrillar
muscle; the other is based on Weber's hypothesis that the splitting of
bound ATP provides the immediate energy source for the contraction of
the actomyosin complex in the myofibril, and postulates a lag between ATP
binding and ATP splitting to account for the deactivation interval.

REFERENCES

Boettiger, E. G., and E. Furshpan. 1950. Observations on the flight motor of Diptera.
Biol. Bull. 99, 346-347.

Boettiger, E. G., and E. Furshpan, 1952. The mechanics of flight movements in Dip-
tera. Biol. Bull. 102, 200-211.

Bozler, E., 1948. Conduction, automaticity and tonus of visceral muscle. Experientia
4,213-218.

del Castillo, J., and B. Katz, 1955. Effects of vagal and sympathetic nerve impulses on
the membrane potential of the frog's heart. /. Physiol. 129, 48-49.

Chadwick, L. E., 1953. The motion of the wings. The flight muscles and their control.
In Roeder, R. D., Insect Physiology. New York.

Ewer, D. W., and S. H. Ripley, 1953. On certain properties of the flight muscles of
Orthoptera. /. Exp. Biol. 30, 170-177.

Gasser, H. S., and A. V. Hill, 1924. The dynamics of muscular contraction. Proc.
Roy. Soc. (London) B 96, 398-437.

Gray, J., 1951. Undulatory propulsion in small organisms. Nature 168, 929-933.

Gray, J., 1953. Undulatory propulsion. Quart. J. Micr. Sci. 94, 551-578.

Gray, J., 1955. The movement of sea-urchin spermatozoa. /. Exp. Biol. 32, 775-801.

Gray, J., and H. W. Lissmann, 1946. Further observations on the effect of de-afferen-
tation on the locomotor activity of amphibian limbs. /. Exp. Biol. 23, 121-132.

Hagiwara, S., 1953. Activity of the main sound muscle of cicada. Kagaku 23, 145-146
(in Japanese).

Hagiwara, S., 1956. Neuromuscular mechanism of sound production in the cicada.
Physiol. Comp. Occol. 4, 142-145.

Hagiwara, S., H. Uchiyama, and A. Watanabe, 1954. The mechanism of sound pro-
duction in certain cicadas with special reference to the myogenic rhythm in insect
muscles. Bull. Tokyo Med. Dent. Univ. 1, 113-124.

Hagiwara, S., and A. Watanabe, 1956. Discharge of motoneurons in cicada. /. Cell.
Comp. Physiol. 47, 415-428.

Harris, J. E., 1955. The development of swimming movements in the embryo of the
dogfish, ScvUorhinus canicula. Ann Acad. Science. Fenn. Ser. A, IV (Biologica),
No. 29.

Harris, J. E., and H. P. Whiting, 1954. Structure and function in the locomotory sys-
tem of the dogfish embryo. The myogenic stage of movement. /. Exp. Biol. 31,
501-524.

Heidermans, C, 1931. Reizphysiologische Untersuchungen an der Flugmuskulatur
von Aeschna caerulea. Zool. Jb., Allg. Zool. Physiol. 50, 1-31.

MYOGENIC RHYTHMS 115

Hill, A. v., 1949. The abrupt transition from rest to activity in muscle. Proc. Roy. Soc.
{London) B 136, 399-420.

Hill, A. v., 1950. The series elastic component of muscle. Proc. Roy. Soc. (London)
5 137,274-280.

Hill, A. v., 1951. The influence of temperature on the tension developed in an isometric
twitch. Proc. Roy. Soc. (London) B 138, 349-354.

Hutter, O. F., and W. Trautwein, 1955. Vagal effects on the sinus venosus of the
frog's heart. /. Physiol. 129, 48.

Krijgsman, B. J., 1952. Contractile and pacemaker mechanisms of the heart of arthro-
pods. Biol. Rev. 27, 320-346.

Krijgsman, B. J., and G. A. Divaris, 1955. Contractile and pacemaker mechanisms of
the heart of mollusks. Biol. Rev. 30, 1-39.

Morales, M. F., J. Botts, J. J. Blum, and T. L. Hill, 1955. Elementary processes in
muscle action : an examination of current concepts. Physiol. Rev. 35, 475-505.

Myers, J. G., 1928. The morphology of the Cicadidae'. Proc. Zool. Soc. Loud. 365-472.

Ossiannilsson, F., 1949. Insect dntmmers. Opusc. ent. Suppl. 10.

Pringle, J. W. S., 1949. The excitation and contraction of the flight muscles of in-
sects. /. Physiol. 108, 226-232.

Pringle, J. W. S., 1954a. The mechanism of the myogenic rhythm of certain insect
striated muscles. /. Physiol. 124, 269-291.

Pringle, J. W. S., 1954b. A physiological analysis of cicada song. /. Exp. Biol. 31,
525-560.

Pringle, J. W. S., 1956. The structure and evolution of the organs of sound produc-
tion in cicadas. Proc. Linn. Soc. Lond. (in press) .

Roeder, K. D., 1951. Movements of the thorax and potential changes in the thoracic
muscles of insects during flight. Biol. Bull. Woods Hole 100, 95-106.

Snodgrass, R. E., 1927. Morphology and mechanism of the insect thorax. Smithson.
Misc. Coll. 80, no. 1.

Sotavalta, O., 1954. The effect of wing inertia on the wing-stroke frequency of moths,
dragonflies and cockroaches. Ann. Ent. Fenn. 20, 93-101.

Tiegs, O. W., 1955. The flight muscles of insects — their anatomy and histology ; with
some observations on the structures of striated muscle in general. Phil. Trans. Roy.
Soc. (London) 238,221-359.

Voskresenkaya, A. K., 1947. Functional peculiarities of the neuromuscular apparatus
of the wings of insects. /. Physiol. USSR 33, 381-392.

Watanabe, M. I., and C. M. Williams, 1951. Mitochondria in the flight muscles of in-
sects. I. Chemical composition and enzymatic control. /. Gen. Physiol. 34, 675-689.

Weber, H. H., and H. Portzehl, 1954. The transference of muscle energy in the con-
traction cycle. Progress in Biophysics 4, 60-111.

Weis-Fogh, T., and Martin Jensen, 1956. Biology and physics of locust flight. I, II, III,
IV. Phil. Trans. Roy. Soc. (London) 239, 415-585.

Wilkie, D. R., 1954. Facts and theories about muscle. Progress in Biophysics 4, 288-
324.

THE MACHINERY OF INSECT FLIGHT*

Edward G. Boettiger
University of Connecticut

Nature, in the course of evolution, has experimented with the problems
of heavier-than-air flight. Three different solutions have proven success-
ful, and today birds and insects share with man the control of the air. Once
in possession of a supplementary power source, man required only knowl-
edge of the principles of flight to develop the airplane. The flying machinery
of birds and insects could furnish necessary clues, and so during the latter
half of the nineteenth century, many studies of animal flight were under-
taken. These provide the background of our present knowledge.

Among birds, the mechanical aspects of the flight mechanism are quite
similar, while insects exhibit a variety that is obvious even to the casual
observer. The presence of an exoskeleton makes possible different me-
chanical arrangements to couple the driving muscles to the wings and gives
to insects the remarkable features of their flight : rapid starts and changes
in direction, hovering, and in some cases even backward flight. Insect
flight machinery includes many tiny structures moving in intricate designs,
and the muscles that move them. Our knowledge of this complex mech-
anism is still only fragmentary, but already new and exciting physiological
adaptations have been found. The intent of this paper is to consider in
some detail the most interesting of these — after a brief introduction to the
general features of insect flight. The appearance in recent years of an ex-
cellent review by Chadwick (1953) relieves me of the necessity of dealing
with the earlier work. As a source of information on power output of in-
sects in flight, one should study the beautiful experiments of Hocking
(1953) and of Weis-Fogh (unpublished) ; and for flight muscle histology,
morphology, and development, the recent work of Tiegs (1955).

Classification and Distribution of Flight Mechanisms

To sustain an insect in flight at its normal cruising speed requires the
consumption of relatively large quantities of fuel by the driving muscles.
Only 3-5% of the food energy can be used to give momentum to the air
flowing through the wings (Hocking, 1953). That the air flow can be

* The original work discussed here was begun and carried out through several
summers in collaboration with Dr. Edwin Furshpan. Other students who have par-
ticipated in various phases of the work include Frances McCann, Richard Baranowski,
William McEnroe, and Rudolph Pipa. The generous financial assistance of the Na-
tional Institutes of Health made the study possible.

[117 J

118 INVERTEBRATE PHYSIOLOGY

maintained during 85% of the wing cycle (Williams and Galambos, 1950)
results from the very complex pathway the wings traverse, as was noted
by many earlier investigators. The movement cycle, producing the neces-
sary precise changes in angle of attack, depends upon the proper and con-
stantly changing relations among the structures of the articulation in their
horizontal and vertical movements. The mechanical features are intimately
related to the physiology of the flight muscles ; for they determine the limits
of length change and to some degree the tension and the rate of change of
tension in the muscles. Each type of flight machinery encountered in insects
is therefore an integrated, well-adapted system.

The following components constitute the flight machinery : ( 1 ) the
wings; (2) the wing articulation, including the direct muscles that con-
trol the setting of the articulating parts, the base of the wing, and the struc-
tures that relate it to the major portion of the thorax; (3) the thoracic
component or special parts of the thorax that serve to couple the driving
(indirect) muscles to the articulation ; and (4) the driving muscles. Since
it is not possible to treat here all of these adecjuately even for one type, the
wings and the aerodynamic problems of insect flight will not be discussed.

On the basis of the presence or absence of the thoracic component, the
flight mechanisms may be divided into two types. In the more primitive
type, the muscles that furnish the power to move the wings are attached
directly to the articulation, as in the Odonata or dragonflies, where each of
the four wings possesses a set of elevators and depressors ( Sargent, 195 1 ) .
Although it is known that the fore and hind wings move in opposite phase
(Chadwick, 1940), there is no information on the nervous control of the
muscles or on the details of their movement. The muscle is reported to be
in incomplete tetanus when stimulated at flight frequency, indicating that
only part of the possible tension change is useful (Tiegs, 1955). The ten-
sion in a tetanus is little more than in a twitch. As found in a number of
other insect muscles, the protofibrils are organized into sheets or lamellae
separated by mitochondria and sarcoplasm. No true fibrils are present and
so these muscles are termed lamellar muscles.

All other flying insects possess the thoracic component, coupling special
indirect muscles to the wing articulation. These muscles, especially in the
higher insects, almost completely fill the thorax and may be of three gen-
eral types: (1) lamellar; (2) microfibrillar in which the protofibrils are
organized into fibrils which have a diameter of about 1.5 ft in the freshly
teased preparation; and (3) fibrillar with large fibrils averaging about
3.0 IX in the fresh state (Pipa, 1955). This classification is certainly not a
rigid one but is useful for purposes of discussion. In some flight muscles,
in fact, the protofibrils appear to be organized into both lamellae and micro-
fibrils in the same cells (Tiegs, 1955).

MACHINERY OF INSECT FLIGHT 119

Lamellar muscle is found in Odonata and certain Orthoptera ( Blattidae,
Mantidae) which do not have typical longitudinal muscles attached to a
well-developed phragma. This muscle type apparently appeared early in the
insects. In many softer-bodied insects, as the Lepidoptera, and in those with
two sets of longitudinal muscles, as the Ephemeroptera and Locustidae,
microfibrillar muscle is found. The insects with the most spectacular flight
ability have harder cuticles and fibrillar muscle (Hymenoptera, Diptera,
Coleoptera, Hemiptera, and many Homoptera).

The present evidence supports the idea that two very dififerent flight
mechanisms have evolved among the insects, one associated with presence
of microfibrillar or lamellar muscles and one with fibrillar muscle. In the
former type, little is known of the mechanical aspects of flight or of the
physiology of the driving muscles. Where studies have been made (Roeder,
1951 ) , it is evident that the wing stroke is synchronous with nerve stimula-
tion of the muscle. This stimulation is often a single impulse but may be
multiple (McCann, unpublished). The flight muscles show little summa-
tion and each stroke is either a twitch or brief tetanus. This mechanism we
have termed the synchronous type.

The existence in insects of some sort of peripheral control of wing move-
ment was suggested by the experiments of de Geer ( 1776), who found that
on removing the wings of some insects, thereby unloading the muscles,
the wing beat frequency greatly increased. Pringle (1949) demonstrated
that the action potentials appearing in the thorax of flies during flight were
not correlated with the wing movements. By simultaneous recording of
these potentials and of thoracic movements, Roeder (1951) confirmed
this observation for several species of Diptera and Hymenoptera. More
recently this behavior has been found in Coleoptera, Homoptera, and
Hemiptera. This second t}'pe of flight mechanism we term asynochronous.

Only in the Homoptera is there evidence for the existence of both syn-
chronous and asynchronous types in one order. The extensive study of
this group by Tiegs (1955) suggests a step in the evolution of the asyn-
chronous mechanism. He finds that in the cicadas the synchronous flight
muscle arises by the multiplication of a few rudimentary muscle fibers. In
the other Homoptera studied, formerly free myoblasts become incorporated
into the young fibers. The myoblasts extend along the growing fiber, add-
ing new fibrils. In jassids each myoblast adds one new fibril. In cercopids
the muscle starts to develop by cleavage of functional nymphal fibers or of
rudimentary fibers arising in the early instars. Then the free myoblasts
become incorporated into these young fibers.

With this information it is tempting to speculate that it is only from the
free myoblasts that fibrillar muscle can be formed. The apparent presence
of both mechanisms in the cercopids might then result from the fact that

120 INVERTEBRATE PHYSIOLOGY

this group shows an intermediate condition with respect to the formation
of the muscle, the flight muscles developing from both preformed fibers
and free myoblasts.

The asynchronous system presents a basic problem in muscle physiology,
for excitation and contraction appear to be dissociated. The key to the solu-
tion of this problem lies in the peculiar mechanical system in which the
muscle operates. We must therefore begin our analysis with the mechanics
of the flight.

The Mechanics of Flight

When flies are placed in a jar containing CCI4 vapors, flight movements
are induced at a certain level of anaesthesia. A number of other volatile
compounds will produce a similar result, but only in CCU does the flight
tone sharply increase in frequency. At a certain level the movements be-
come erratic and sputtering, suddenly coming to a stop with the wings
either up, as at the end of a normal up stroke, or down, as at the end of a
down stroke. All intermediate positions are unstable. Considerable resist-
ance is encountered if one attempts to move the wings from the stable po-
sitions. However, when a critical point is reached, the wings snap without
further aid to the end of the stroke (Boettiger and Furshpan, 1952). Be-
cause of its similarity to the action of a common noise maker this was called
the click mechanism (Boettiger and Furshpan, 1950).

Movements of the wings under CCI4 can best be achieved by pressure
upon the scutellum. The remarkable feature of these wing movements is
that the wings, on being snapped up and down, appear to follow the normal
flight path with proper changes in angle of attack and in direction during
the stroke. Evidence for a similar snap action during flight was obtained,
and so it was suggested that CCI4 by its effect on the flight muscles, direct,
indirect, or both, sets the articulation as in normal flight. Therefore, if the
positions of the structural components are studied in CCl4-treated flies
with the wings in the up and in the down positions, the mechanics of flight
may be worked out (Boettiger and Furshpan, 1952).

The mechanical system moving the wings in flight is composed of the
articulation and the thoracic component. As these are bilateral structures,
their movements will be described on only one side. The thoracic com-
ponent shown in Fig. 1 consists of the scutellar lever and the tergopleural
elements that move it, the anterior notal process with the parascutum to
which it is attached, and the mesopleural process.

Scutellar Lever : The tergum or upper part of the thorax is hinged
at j-k and rests at its posterior end on the joint at i, Fig. lA. On the down
stroke contraction of the longitudinal indirect muscle (5, Fig. 2) pulls the
articulation i of the scutellum a forward and slides the tergum posteriorly,

MACHINERY OF INSECT FLIGHT

L

121

A

Fig. 1. A. Left side of thorax of Sarcophaga bullata. B. Ventral view of scutellar
lever dissected free from the thorax. C. Detail of lever arm and attachment to scutel-
lum, outside above and inside view below : a, scutellum ; b, lever arm ; c, process
articulating with the first axillary sclerite ; d, anterior parascutum ; e, posterior
parascutum ; f, junction of prescutal ridge, transverse ridge and parascutal hinge;
g, end of parascutal hinge ; h, point in line of attachment of notum with lever arm ;
i, point of rotation of the scutellar lever on the postnotum ; j-k, fulcrum of the notum
lever in the action of vertical muscles ; 1-m, line of attachment of notum with lever
arm ; n-o, lateral vertical cleft ; p, anterior notal process ; q, transverse ridge ; r, arti-
culating groove for attachment of postnotum ; s, scutellar bridge ; t, triangular struc-
ture supporting articulation of lever and postnotum ; u, groove for a process of first
axillary sclerite.

thus rotating the scutellum downward. The forward-projecting arms b of
the scutellum act upon the wing articulation to force the wings down. The
shortening of the vertical muscles ( 1, 2, 3, Fig. 2) pulls the tergum down
and forward, forcing joint i of the scutellum posteriorly. By this action
the longitudinal muscles are restretched, the scutellum is rotated upward,
and the arms of the scutellum move down, thereby forcing the wings up.

122 INVERTEBRATE PHYSIOLOGY

This lever-like structure, moved by the action of the indirect muscles on
the tergum and on the joint of the scutellum, is termed the scutellar lever.
By the action of this lever, the wing-tip movement is amplified about 20
times. The two wings move together since the lateral arms are quite rigidly
connected through the scutellum. Reducing this structural rigidity, by
simply removing the soft cuticle forming the top of the scutellum, destroys
the ability for sustained flight.

Fig. 2. Horizontal section through
the notum. Structures labeled as in
Fig. 1, with the following additions :
V, prescutal ridge ; w, chitinous sup-
X"-^ 3 — ^^ :^-^ m ~z: lH-M^ porting structure ; X, anterior hard-
ened plate; 1, first dorsoventral
muscles ; 2, second dorsoventral
muscles ; 3, oblique dorsal muscles ;
4, tergal remoter muscles ; 5, longi-
tudinal muscles.

Anterior Notal Process: The anterior notal process (/?, Figs. 1, 2)
in flies is supported by and moves in conjunction with the parascutum
(d and e, Fig. 2) and so these must be treated together. The anterior notal
process is strongly supported by the termination, /, of the prescutal and
transverse ridges. The parascutum is hinged to the rest of the tergum be-
tween / and g so that the anterior notal process can be freely rotated up
down but cannot twist. The presence of this hinge makes it impossible for
the tergum to produce up and down movements of the anterior notal pro-
cess except through the scutellar lever.

Mesopleural Process : The mesopleural process is a pleural deriva-
tive to which is attached one end of the second axillary sclerite (Fig. 3) . It
serves as a fulcrum for the rotation of the sclerite. At the end of each stroke
it moves in closer to the tergum, thereby increasing the wing-stroke
amplitude.

The second mechanical component is the wing articulation, consisting
of the 1st and 2nd axillary sclerites, a number of other structures of
secondary importance for wing movement, and the direct muscles of flight.
The articulation contains elements for folding the wings as well as those
concerned in the wing cycle. The basic features of its operation, however,
depend upon the movements of the 1st and 2nd axillary sclerites and their
relations to the elements of the thoracic component described above. These
relations are illustrated in Fig. 3. The space between the hinge h, connect-
ing the parascutum to the lateral border of the tergum, and the mesopleural
process a (Fig. 3) is bridged by the parascutum (including the anterior
notal process) and the 2nd axillary sclerite. At k the radial vein of the

MACHINERY OF INSECT FLIGHT

123

wing is tied in with the anterior notal process and the 1st axillary e. The
wing cycle reflects primarily the movement of this point of union. On the
down stroke, k is moved up by the action of the scutellar lever on the 1st
axillary sclerite, and moved posteriorly, due to the movement of the tergum
transmitted through the anterior notal process and the parascutum. As a re-
sult the wings move down and forward with proper changes in tilt. On the
return stroke the reverse action of the parts brings the wings back to the up
position. The union point k moves in a cycle as it follows different paths
on the up and down strokes.

A

B

Fig. 3. Cross-sectional view of thorax showing details of the articulation of the
right wing. A. Wing in up position, anterior view. B. Wing in the down position,
anterior view. C. Posterior view of the axillary sclerites of right wing showing their
relation to the mesopleural process, the lever arm, and the anterior parascutum : a,
mesopleural process ; b, pleural apophysis ; c, anterior pleurosternal muscle ; d, an-
terior parascutum ; e, first axillary sclerite ; f, second axillary sclerite ; g, base of
radial vein; h, hinge; i, radial vein; j, hook articulation; k, point of articulation of
anterior notal process, first axillary sclerite, base of radial vein, and second axillary
sclerite ; 1, end of the lever arm.

When the articulation is set for flight or by the action of CCI4, the hinge
of the parascutum and the mesopleural process are brought closer together.
This results from an outward thrust at / (Fig. 1), the junction of the pre-
scutal and transverse ridges, produced by tension in the indirect muscles,

124 INVERTEBRATE PHYSIOLOGY

and an inward force on the mesopleural process resulting from its struc-
tural connections and the contraction of the pleurosternal muscle. At the
beginning of the next stroke, whether an up or a down stroke, k is moved
by the action of the scutellar lever acting through the 1st axillary sclerite.
The hinge of the parascutum and the mesopleural process are forced apart
to accommodate the full width of the parascutum and 2nd axillary, storing
potential energy by straining the tergum and the mesopleural process. At
the critical point, where k moves past the direct line between the hinge h
and the hook articulation of the 2nd axillary sclerite on the mesopleural
process, the union point k is forced strongly toward its extreme position
by the recoil of the strained elements, releasing the energy stored at the be-
ginning of the stroke. By the wing articulation and the thoracic component,
therefore, changes in length of the indirect muscles are magnified 400-600
times and the muscle can operate almost isometrically.

The foregoing is a simplified and brief statement of the mechanical
events of the wing cycle of Sarcophaga hullata Parker as analyzed in CCI4-
treated flies. Other factors, such as the secondary wing articulation, the
posterior notal process, the basalar and the subalar, may also play im-
portant roles.

To obtain further information and to check the conclusions made on
CCU-treated flies, studies were made of the flight mechanism in action. If
the description given above holds for normal flight, the following can be
expected : ( 1 ) the anterior notal process must move in and out for each
up stroke or down stroke ; (2) the mesopleural process must also move in
and out each stroke ; ( 3 ) the movements of the scutellum should accurately
reflect, as a built-in isotonic lever, the changes in length of the driving in-
direct muscles during flight.

Tiny pieces of mirror silver were fastened to the thorax of flies with
wax, one on the side of the mesopleural process, one on the tergum just
above the parascutum, to indicate the in and out movement of the anterior
notal process, and one on the side of the scutellum to show wing position.
A light beam, reflected by the three mirrors, was brought to focus on
moving film in a Grass kymograph camera. By careful adjustment the
three spots could be brought close together, but not in a vertical line.
Therefore, some means of obtaining simultaneous ordinates in the three
records was necessary. A two-bladed fan driven at high speed was mounted
to cut the light beam before it reached the mirrors. Blanks appearing in
the records could then be lined up and the instanteous position of each
structure during a cycle determined. A typical result is shown in Fig. 4A.
The dotted lines indicate the positions of the blanks made by one blade of
the fan on the three records. It is apparent that not only are there in and
out movements of the tergum and mesopleural process but also anterior-

MACHINERY OF INSECT FLIGHT

125

posterior rotations evident in the anterior notal process record, where for
part of the cycle the light is moving forward faster than the film.

Fig. 4. Movements of the thoracic structures. A, fly, Sarcophaga ; B, wasp, Po-
listes ; C, fly, Sarcophaga. Records of optical levers from mirrors attached to the
scutellum (and moving with the scutellar levers ScL), to the mesopleural process
(MP), and to the tergum just above the parascutum where in-and-out movements of
the anterior notal process (ANP) are effective. In A and C the original records are
traced and movements are tilting the mirrors not only in the vertical axis but to
various degrees horizontally, with or against the time axis. In B the record has been
redrawn to eliminate anterior-posterior movements and the three beams shifted so
that simultaneous movements of time are vertically under each other. In A and C
the time correspondence is indicated by the dotted segments. The terms up and down
refer to wing position ; in and out to lateral movement of the part relative to the
thorax.

From the static analysis (Fig. 3) based on CCl^-treated flies, it was
predicted that, at the beginning of a stroke, the anterior notal process moves
in (light beam down) while the mesopleural process moves out (light beam
down). After the critical period, the mesopleural process moves in (light
beam up) and the anterior notal process out (light beam up). The record

126

INVERTEBRATE PHYSIOLOGY

(Fig. 4A) shows that these movements actually do occur during tethered
flight in each stroke, up or down. The relative amount of movement is seen
to be different in the two strokes, showing that they are not symmetrical.
For the wings to operate always at the best angle of attack, the up and
down movements of the wings must take different paths. These paths are
determined by the setting of the articulation. The setting cannot be altered
during the course of a single cycle, for the direct muscles cannot sig-
nificantly change their tensions within the very brief cycle of operation of
the indirect muscles, 6 msec, in Sarcophaga.

The agreement between the movements predicted from a study of CCI4-
treated flies and those found in flight are sufficiently good to give strong
support to the above analysis of the mechanics of flight in Diptera. As will
be indicated later, it is probably that all insects with the asynchronous mech-
anism have some sort of snap action at the articulation, for it plays an im-
portant role in assuring full amplitude of wing movement. In Fig. 4B are
presented results obtained on the wasp PoUstes sp. The record has been re-
drawn to eliminate anterior-posterior movements and the three records

3 30Aec.

Fig. 5. Movements of tlie scutellar lever recorded by optical levers from mirrors
on the scutellum. All records on fly, Sarcophaga bullata. A, normal ; B, wings re-
moved ; C, wings removed and articulation damaged. A, B, and C are taken on the
same fly, with the same magnification and time scale. Wing frequency is stated on the
record. D, erratic flight movements in CCl^-treated fly. E and F, parts of the same
record showing the development of a fast stop on the down stroke (E) and on the
up stroke (F).

MACHINERY OF INSECT FLIGHT 127

shifted to the same axis. The complete cycle took 8 msec. The anterior notal
process does not show double movement each stroke, but is rigidly con-
nected to the terguni and consequently moves with it, there being no
hinged parascutum. Removal of the wings in wasps reduces the amplitude
of muscle movement to about one-half, while in flies there may be httle
or no change (Fig. 5B). This suggests that wing inertia plays a role in
maintaining wing amplitude in wasps, while in flies the snap action in the
articulation is more important.

If the various articulating parts are not held in the proper relation,
erratic flight movements result. The sputtering flight under CCI.1 already
referred to is an example. Frequently a mounted fly will show similar be-
havior, apparently trying difi^erent settings in an effort to attain free flight.
In Fig. 4C movements of the scutellum and anterior notal process were
caught in a moment when the articulation was not properly adjusted, and
so the relation between these erratic movements and the mechanics of
normal flight can be determined. In this case the scutellum moved little at
the beginning of the down stroke. Not until the anterior notal process
moves inward, to allow the union k of this process and the 2nd axillary
sclerite to attain the critical point, does the scutellum move rapidly as the
result of the recoil of the strained elastic elements. The fast movement
begins at the instant the anterior notal process reverses direction. When
the inhibition to movement is somewhat greater, the wings may be brought
to a sudden stop, as shown by the scutellar movement record of Fig. 5E,F.
The stop may last the duration of several cycles. The inhibition to move-
ment develops gradually, being greater each cycle until more force is re-
quired to move the articulation past the critical point than is generated
by the indirect muscles. Movement then stops until balance is again
achieved. On the down stroke the stop appears at a different point than on
the up stroke. These unusual movements can, therefore, be readily under-
stood with our information on the mechanics.

One of the simplest ways to reveal the snap action is to study scutellar
movements with the wings removed. After the removal of the wings, the
only resistance to movement is that in the articulation. Once this is over-
come at the beginning of each stroke, the movement accelerates to the end,
driven by the stored elastic energy- (Boettiger and Furshpan, 1951).
Normally the wing acts as a governor to smooth out the stroke. The accel-
erating movement, beginning at the critical period, is brought to a sudden
halt by mechanical stops. The articulation allows only limited movement
of the scutellar lever and consequently of the driving muscles. Mechanical
limits to the movement in such a vibrating system may be necessary to pre-
vent the tearing of the muscle by the build up of amplitude due to inertial
forces. That the setting of the articulation may be altered by injury to

128

INVERTEBRATE PHYSIOLOGY

secondary elements of the articulation is shown in Fig. 5C, taken on the
same animal as records A and B. Sensory elements of the base of the wing
which regulate loading in the articulation may have been injured. In one
case the articulation was placed into a unique state by CCI4 (Fig. 5D) ; for,
after a delay on the up stroke, the critical point was passed but the moving
parts apparently hit an elastic stop, rebounding back past the critical
point immediately, without locking into the up position. The mechanical
action here is similar to that which accounts for the ringing of the tymbal
in the cicadas studied by Pringle ( 1954) .

To complete this survey of the kinds of movements that may be brought
about by the vmique mechanical system and its equally unique driving
muscles, a few normal starts and stops are illustrated in the records of
Fig. 6. In A the fly was making a series of starts and stops but there was

A

I I I

Fig. 6. Scutellar movements during starts and stops recorded on fly, Sarcophaga
bullata, with optical levers. A, normal ; B, wings removed ; C, wings loaded with wax.
Time scale applies to all records.

Still evidence of a continuous rhythm of very small amplitude between one
stop and the next start. The articulation is set very rapidly in a normal
start, for full amplitude is attained almost immediately. The slight vibra-
tions in the record suggest that the indirect fibrillar muscles are excited
before the articulation is set. A stop with wings removed, B, is contrasted

MACHINERY OF INSECT FLIGHT 129

with one occurring when the wings are loaded with wax. The movements
decrease after the cessation of excitation much more rapidly in the fly
without wings. The slow decrease found when the wings are loaded shows
the effect of inertia. The muscle loses its tension relatively slowly after
excitation is stopped, but will continue to vibrate only with an inertial
load (see below).

In addition to improving our understanding of the mechanism moving
the wings, the foregoing examples give much information on how the
driving muscles must operate : ( 1 ) they shorten very little because of the
large mechanical amplification factor; (2) vibrations of high frequency
with practically no amplitude are possible; (3) the snap action is not
necessary for the operation of the asynchronous mechanism; and (4) a
shortened muscle may be easily relengthened for some time after it has
shortened.

The Physiology of Fibrillar Muscle

Having discussed in some detail the wing articulation and the thoracic
component of the flight machinery in one group of insects with the
asynchronous system, we may now turn to a consideration of the driving
muscles. These consist of two sets acting in opposition upon the thoracic
component, as described above. The evidence in the literature suggests
that these fibrillar muscles have unique physiological properties. They are
capable of very high frequency operation, up to 2,200 cycles per second
(Sotavalta, 1953). In flies with thorax open, electrical stimulation pro-
duces no visible shortening. That the action potentials from the thorax do
not correlate with wing movements has already been noted, as has the in-
crease in frequency when the wings are removed. Do these special proper-
ties of fibrillar muscle represent a modification of the excitatory or of the
contractile mechanism ?

If the excitatory process is basically modified in fibrillar muscle, the
electrical properties of the membrane might show differences when com-
pared with other muscles. To test this possibility, microelectrodes were
inserted into the fibers of the dorsal longitudinal muscle of flies and both
resting and action potentials recorded by the method of Nastuk and
Hodgkin (1950). The resting potential was usually about 60 mV and the
action potential 80-100 mV, although values as high as 120 mV were ob-
served (Boettiger and McCann, 1953). The form of the action potential
is shown in Fig. 7D. Several species of v/asp gave very similar action
potentials. From the fibrillar muscle of Coleoptera, however, large action
potentials were not easy to obtain and did not show clear evidence of over-
shoot. In the best preparations, resting and acting potentials of 60 mV
were found. The microfibrillar muscle of a few species of moths pro-

130

INVERTEBRATE PHYSIOLOGY

duced action and resting potentials similar to that found in beetles, though
usually smaller. Qualitatively it appears that the electrical manifestation
of the excitatory processes of fibrillar muscle fibers are similar to those of
ether muscles. Different groups of insects have fibrillar muscle with quanti-
tatively difterent electrical properties. These variations may eventually be
correlated with the different patterns of nerve innervation of fibrillar
muscles indicated by the work of Tiegs ( 1955) .

n o V e. m e riir

loo/s

ec.

Fig. 7. Action potentials of the indirect flight muscles of the fly, Sarcophaga bul-
lata. A, Normal discharge recorded with external leads ; B, during stimulation of
the ganglion, the action potentials of all fibers recruited to stimulus frequency, flight
continuing ; C, same after cessation of flight movements. Time scale applies to records
A, B, and C. D, single-fiber action potential of longitudinal flight muscle recorded
with internal microelectrode on stimulation through leads directly on the muscle;
+ and — refer to the sign of the internal electrode. Stimulus artifact shows at the
beginning of the record. E, single-fiber action potentials recorded with internal elec-
trode during spontaneous flight movement compared with thoracic movements. Small
movements in the action potential record are probably action potentials in adjacent
fibers.

When transmeml)rane potentials of a muscle fiber are recorded during
flight, one observes a beautifully regular discharge, apparently in all fibers.
Identical discharge patterns can be elicited by the vapors of ether and a
number of other compounds. With CCI4 the discharge frequency rises

MACHINERY OF INSECT FLIGHT 131

rapidly, the spike height decreases, and the membrane does not completely
recharge between responses. The regular discharge patterns arise in the
thoracic ganglion. A record of the action potentials of a single muscle
fiber and of thoracic movements is included in Fig. 7E, to show beyond
all doubt that there is no phasic relation between the membrane events and
the mechanical movements produced by the muscles.

Response to Electrical Stimulation .

It is possible to drive the flight mechanism of a fly through stimulating
electrodes pushed into the vicinity of the thoracic ganglion (Boettiger,
1951). In the best cases a series of wing beats follows each stimulus,
making the sound of a short buzz. With increasing frequency of stimula-
tion, these fuse into continuous movements, though the buzz is louder and
higher in pitch immediately following each stimulus. At 15-20 stimuli per
second in Sarcophaga, the movements are of normal amplitude and fre-
quency w^ithout modulation. As the frequency of stimulation is increased
still further, the flight tone sputters and suddenly the wings stop in either
the up or down position. The fly now shows the same responses as one
treated with CCl4.

A record of the thoracic potentials with external leads is shown in Fig.
7 before (A) and during (B) driven flight. The normal irregular firing of
the muscle fibers gradually changes to a synchronous one as the muscle
fibers are recruited to the driving stimulus frequency. During the recruit-
ment there was no noticeable change in flight movements. After a short
period of driven flight the fly stopped and folded its wings back but the
action potentials remained (C).

These experiments with driven flight prove that, even with simultaneous
excitation of the muscle fibers of both sets of antagonistic muscles, the
shortening and lengthening cycle of the contractile elements are not altered.
Also we may conclude that the high-frequency behavior of these muscles
is not the result of an alteration in the number of fibers responding, since
the excitatory processes of all fibers were synchronized with the stimuli.

Under the conditions of driven flight the direct muscles must also be
excited. In certain cases it seems that the indirect driving muscles are acti-
vated while the direct muscles are not. In Fig. 6A the small vibrations in-
dicate activity of the fibrillar muscle, the rapid rise in amplitude at the
start resulting from contraction of the direct muscles. Movements of the
scutellum and muscle potentials were recorded in an unusual start under
external stimulation (Fig. 8). The stimulus is shown to control the firing
of the indirect muscle fibers, for only one large action potential follows the
stimulus artifact. At the beginning of the record, each stimulus produced
only a small slow movement of the scutellum due to its action on some direct

132

INVERTEBRATE PHYSIOLOGY

Fig. 8. A start during stimulation of the thoracic ganglion of the fly, Sarcophaga
bullata. AP is a record of the action potentials obtained with external leads, each
potential preceded by a shock artifact. SC is the simultaneous record of movements
of the scutellum recorded by an optical lever.

muscle. After one of these movements, a very slight vibration is noted. The
next stimulus increased the oscillatory response, which, with continuing
stimulation, gradually built up to normal amplitude. This start contrasts
with normal starts (Fig. 6). The wings were held straight out and the vi-
brations began from this position. Although some of the direct muscles
must have been excited, the articulation was apparently not set. The medial
positions of the wing are unstable in an articulation set for flight. The be-
ginning of the vibrations may be correlated with the shortening of the
direct muscles, principally the pleurosternal, but under conditions where
the indirect muscles were already under tension and resisting the shorten-
ing. Excitation of the opposing fibrillar muscles without the proper me-
chnical situation results in each muscle preventing the shortening of the
other. To get out of this impasse, movement must be initiated by an acces-
sory agent, in this case the direct muscle. Once one muscle is shortened
slightly, and its antagonist necessarily lengthened, vibrations begin and
with proper setting of the articulation grow to full amplitude. The smallest
vibrations in the record of Fig. 8 are about one-twentieth normal amplitude
but at very nearly normal frequency. As the shortening of these muscles
during flight is 0.02-0.04 mm., the shortening in the vibration is less than
2 /x or 0.047o of muscle length. In wasps, similar vibrations have been
noted. In one case the ring stand to which the wasp was fastened could be
felt to vibrate while the animal appeared to be perfectly still.

In these very small vibrations one cannot believe that any snap action
occurs. All that is necessary is for the opposing excited fibrillar muscles to
be moved in opposite phase, and this movement need not be very fast. The
system having a negative resistance, once started, will then tend to build
up. For the attainment of full amplitude with a damped load, such as the
wings moving through the air, the snap action may be necessary.

Although much useful information can be obtained by the study of in-
tact insects, the correctness of our conclusions and the details of the

MACHINERY OF INSECT FLIGHT 133

physiology of the fibrillar muscle must be studied in an isolated muscle
prep^iration by the classical methods.

The Meclianical Properties of Fibrillar Muscles

The discovery of the snap action in the articulation and the demonstra-
tion of its importance in normal flight suggested to us the early experi-
ments of Gasser and Hill (1924) on the responses of the frog muscle to
quick changes in length. The theory was proposed by Boettiger and Fursh-
pan (1950) that the opposing longitudinal and vertical indirect flight
muscles were both in complete tetanus. By this it was not meant that the
tension is smoothly maintained, but that the muscle is kept in the active
state as described by Hill (1949). In such a tetanus or maintained active
state, tension is a function of length and of the velocity of shortening. The
opposing muscles cannot usually neutralize each other, since the position at
which they would have the same length is unstable ; because of the setting
of the articulation, one muscle would be lengthened and the other short-
ened when flight begins. In the shortening muscle, tension would be ex-
pected to fall rapidly after the critical period. On the return stroke the
antagonist would quickly lengthen the muscle. To obtain work from such
a system the reappearance of tension during the lengthening must be
delayed, so that at each length the muscle has greater tension while short-
ening than lengthening.

The first step in testing this theory was to show that fibrillar muscle
could be put into typical isometric tetanus. Since we already had a body
of information on flies, the first experiments were performed on one of the
anterior vertical muscles of Sarcophaga. McEnroe (1952) reported that,
with one end of the muscle detached and coupled with a recording lever, a
steady tension of 250 mg. could be obtained during tethered flight. The
tension increased before the start, was continued during flight, and slowly
disappeared at the end of flight. Upon stimulation the muscle went into
tetanus. Much larger isometric tensions were later obtained in large Taba-
nids (McEnroe, 1954).

The original theory was therefore substantiated to some extent. As far
as the excitatory process was concerned, the muscle was in a maintained
active state similar to tetanus during flight. The changes in tension neces-
sary for the production of the wing cycle must result from the changes in
length, in the speed of shortening, and in the speed of lengthening. Pringle
(1954) found similar responses to stimulation in the tymbal muscles of
those species of cicadas possessing fibrillar muscle to produce high-fre-
quency tones. He accounted for the relengthening of the muscle without
the usual rise in tension by assuming a deactivation by release. The de-
activated state was considered to last a short time, requiring immediate

134 INVERTEBRATE PHYSIOLOGY

restretching of the muscle for the muscle to take advantage of the deactiva-
tion. The possibility of reactivation on stretch was also suggested. Our re-
sults with intact flies (Fig. 5E,F) show that after a fast stop the shortened
muscle may remain short for the duration of several cycles without the
return of tension. When the inhibition to movement is removed, the short-
ened muscle can be as rapidly lengthened by its antagonist as in the normal
stroke. Lengthening, therefore, must be as important for the redevelopment
of tension as shortening is to the fall of tension.

Since the fly preparation was technically difficult to handle because of the
very small movements possible, the study was continued with large bumble
bees. Upon removal of the head and abdomen, the thorax was impaled on
two needles pushed into a small mounting board. A third needle, inserted
into the cuticle at right angles to the other two, firmly anchored the thorax
with the posterior end oriented upward. The phragma to which the longi-
tudinal muscle is attached was exposed, cut from its connections with the
articulation, and fastened with a double hook to an RCA mechanotrans-
ducer for recording muscle tension. The transducer could be raised and
lowered precise amounts in order to study tension at different muscle
lengths.

Fibrillar muscle of the bumble bee exhibits marked summation to a series
of maximum stimuli, as does that of the fly and of the tymbal muscle. The
response of the bee preparation to stimulation at 8 per second is shown in
Fig. 11 A. The single isometric twitch is quite small as compared to the
complete tetanus attained at about 40 stimuli per second. Relaxation is
very slow, taking one-half to one second for the tension to drop to zero
after the cessation of stimulation. The slow relaxation explains the stops
noted in Fig. 6B,C.

A typical tension-length diagram of the muscle in maximum tetanus is
shown in Fig. 9. The maximum amount the muscle can shorten is only
about 12% of its rest length, or 0.9 mm. With the arms of the phragma
attached to the articulation, the muscle is held in the thorax at the length
at which it develops maximum isometric tension, and shortening is limited
to 0.1-0.2 mm. The curve of passive tension is also shown. This limited
movement means that only a portion of the tension-length curve is used
in flight. Since the isometric tension varies little over this range (Fig. 9),
the fall in tension necessary for the work cycle must result from the rapid
shortening. In these isometric contractions there was no evidence of
oscillatory behavior, the muscle acting as other skeletal muscle in all
regards.

The relation between isometric and isotonic contraction is shown in Fig.
10. For these experiments the mounting board with the bee thorax prepara-
tion was fastened to a thin piece of metal hinged to a support. To load the

MACHINERY OF INSECT FLIGHT 135

muscle, weights were attached to the piece of metal. Changes in length
were recorded by a second transducer coupled to the metal strip by a light
spring. The tension recorder was connected to the oscilloscope to give
vertical deflections of the beam and the length recorder to produce hori-
zontal deflections. The tension-length relations of the muscle were there-
fore drawn out instantaneously on the screen.

?"

V]S.

0.5 0.7 0.9

LENGTH '«'"

Fig. 9. Tension-length relations of stimulated (active) and unstimulated (passive)
longitudinal flight muscle of bumble bee, Bombus. The muscle was stimulated to
maximum complete tetanus. The active tension curve was obtained by allowing the
muscle to shorten isotonically without a weight and then to build up tension at the
shorter length isometrically. The absence of the weight explains the difference be-
tween Fig. 9 and Fig. 10. The passive curve was obtained by stretching the unstimu-
lated muscle.

An example of an isotonic response is illustrated in Fig. 10, recon-
structed from observations made on a number of preparations. The un-
stimulated muscle was loaded with 14 grams (c). On stimulation the
muscle shortened to b isotonically, and then began to oscillate with increas-
ing amplitude to a maximum c-c\ Upon the cessation of stimulation the
muscle lengthened and the oscillations decreased as shown by the envelope
c-a, c'-a. If the oscillations are prevented, the muscle shortens to d, a point
on the active tension-length curve. By jarring the preparation the oscilla-
tions again appear, the muscle lengthening and the amplitude increasing

136

INVERTEBRATE PHYSIOLOGY

again to c-c' as indicated by the envelope d-c, d-c' . The direction of move-
ment around the tension-length loop can be found by dimming the trace
through the Z-axis by the output of either the length or the tension re-
corder. By noting whether the dimmer trace is above or below the brighter,
one can immediately determine whether the tension is higher or lower on
muscle shortening and so whether the system is doing work on the muscle
or the muscle on the system. In the case where the muscle is vibrating the

40-

30-

o

to

uj20-
10-

0.1

c'

0.3 * 0^5
LENGTH

Fig. 10. Behavior of the longitudinal flight muscle of the bumble bee, Bombus, in
isotonic contractions with a weight. The loops are not actual records but are drawn
from observations on many preparations. The records were obtained with strain
gages giving instantaneous measure of tension and length. Tension was applied to
the horizontal plates and length to the vertical plates of a cathode-ray oscillograph
to give the dynamic tension-length diagrams shown. Muscles stimulated at 60 per
sec. to complete tetanus. See text for explanation of details.

platform, as in Fig. 10, the tensions are higher on shortening since the
muscle is doing work. The area enclosed in the loop represents the work
done by the muscle against damping forces in the mechanical system. By
introducing more frictional damping, larger loops were obtained. Many
dififerent loops are possible within the area enclosed by the curves of active
and passive tension, depending upon the load, the damping forces, and the
characteristics of the contractile element of the muscle.

MACHINERY OF INSECT FLIGHT 137

The contractile element is in series with a very stiff spring. To describe
the theoretical implications of the tension-length loops in terms of these
two elements, a simple model can be used. In this model the elastic element
is represented by a spring and the contractile element by one's arm muscles.
The muscles support the spring, to the end of which a weight is attached.
If the weight is set in motion by an external force its movements will soon
be damped out. One is instructed, however, to keep the weight in motion
at the same amplitude and therefore must contrive to put into the system
just enough energy to overcome the damping forces. One can do this most
efficiently by shortening and lengthening his muscle in the same sinusoidal
movement as the spring and weight. If the muscle moves in phase with the
spring, no energy is transferred. But by moving in the same rhythm, and
slightly out of phase with the spring, one can maintain the motion. The
movements of the muscle must be slightly ahead of the movements of the
spring. This is only possible because of the inertia of the weight. If the
movement of the weight is more heavily damped, one's movements must be
more out of phase with the spring to maintain the system in motion. The
tension-length diagrams of the above model would show loops similar to
those of Fig. 10. The area of the loop would be greater with greater phase
shifts. Tension would be greater during shortening and the movement of
the beam around the loop would be counterclockwise, the area representing
work the muscle does against damping forces.

With spontaneous vibrations the loops are always counterclockwise.
When, however, a mechanical system is used to make the excited muscle
shorten and lengthen at different frequencies, the movement is either
clockwise, the system doing work on the muscle, or counterclockwise, the
muscle doing work on the system. The phase angle is a function of fre-
quency and shifts from a minus value, length reaching its maximum before
tension, to a plus value, tension reaching its maximum before length. In
the former case, area represents work done by the muscle, and in the latter,
work done on the muscle.

The significance of the loop can be summarized as follows : ( 1 ) the
slope of the major axis of the loop is determined by the compliance of the
excited muscle, (2) the area of the loop is a measure of the work done by
the muscle against external damping forces and is a fvmction of the maxi-
mum tension, the maximum length, and the phase angle between tension
and length. (3) the position of the loop in the tension-length area depends
upon the load and the change in the force the muscle can exert as velocity
increases. When velocity is zero the muscle shortens until it attains the
minimum length at which it can just sustain the load. In vibration as
velocity increases, the muscle, because of the force-velocity relation, cannot
exert at this short length a tension equal to the load. The muscle must

138 INVERTEBRATE PHYSIOLOGY

therefore lengthen, as illustrated in Fig. 10, until it can exert an average
force, in the dynamic state, equal to the load.

The large loop h-i (Fig. 10) represents a possible tension-length loop
of the muscle operating in normal flight. For sinusoidal motion the area of
this loop can be calculated from the following relation

work per cycle = ir PqXo Sin $

where Po is one-half the maximum tension change, and Xo is one-half the
length change. Using the data — Po = 20 grams, Xq = .0075 cm., and 6 =
30° — the work is 225 ergs per cycle. If the antagonist muscle does the
same work and the frequency is 100 cycles per second, the power output
per second is 45,000 ergs. Since only about two-thirds of this can be con-
verted into usable work by the wings (Hocking, 1953), only some 30,000
ergs are available to move the bee. This is probably about one-half that
necessary. For the muscle operating in the insect one or more of the follow-
ing must be greater than the values used above : the maximum tension, the
maximum length, or the phase angle.

It has not yet been possible to load the muscle properly ; consequently,
loops of this size have not been experimentally obtained. The action of the
articulation, the air resistance, and the wing inertia of the intact animal
cannot be imitated easily. Were this possible, we have every reason to
believe that the preparation would be able to do the amount of work re-
quired of it by the insect in flight.

Some additional information can be obtained from the response of the
muscle to transient rapid changes in length (Fig. 11) (Boettiger and
Furshpan, 1954a,b). For these experiments the platform on which the
preparation was mounted was attached to a small rod running through
a bearing and fastened to the center of the diaphragm of an earphone. By
an on-and-off switch the current through the earphone coil could be con-
trolled to produce small changes in length of the muscle, 0.05-0.2 mm.
Tension and length were recorded as a function of time.

Two kinds of controls were used. In B the muscle was passively stretched
to about 30 grams and then subjected to changes in length. The tension in
this stretched unstimulated muscle followed very closely the changes in
length. This result demonstrated that our experimental setup was adequate
and that the unstimulated muscle behaved as a simple physical system. The
same muscle stimulated to produce isometrically the same tension, and sub-
jected to the same length changes, gave the response shown in C. Upon
lengthening the muscle after a rapid shortening, the full length was attained
before the full tension. The muscle produced a lower tension at each length
when being stretched than it had while shortening.

A second control is shown in D, where the stimulated longitudinal micro-

MACHINERY OF INSECT FLIGHT

139

fibrillar muscle of a moth was subjected to the rapid changes in length.
At the completion of lengthening the tension was above the initial tension
at this length, while in the bee muscle it was bejow. Fibrillar and micro-
fibrillar muscle differ in their response to changes in length.

T
L

I ■! 1 1-

r

< — • — I — I — • — • — I — I — t-j

Fig. 11. The effect of rapid transient changes in length. T is muscle tension and
L is muscle length. All marks on the length record show the instant of stimulation
at 60 per sec. A, isometric myogram of bumble-bee muscle, stimulation at 8 per sec.
to show summation of contraction ; B, response of unstimulated bee muscle stretched
to give about 40 gms passive tension ; C, same muscle stimulated to give active tension
of about 40 gm. ; D, flight muscle of a moth showing behavior of nonfibrillar muscle ;
E, bumble bee, successive records showing effect of decreasing interval between
release and stretch; F, wasp, Sphecius; G, bumble bee, Bombus.

In E is a series of records in which the lengthening of the muscle is
delayed a variable time after the imposed shortening. At the completion of
.shortening, the tension falls slightly for about 10 msec and then rises, if it
is not relengthened, to a tension characteristic of the new shorter length.
This rise in tension after quick release is typical of muscle and even of the
glycerinated muscle model. Relengthening the muscle during the 10 msec.
before the tension starts to rise results in a somewhat greater active tension

140 INVERTEBRATE PHYSIOLOGY

rise after the completion of the lengthening than if relengthening is de-
layed. A record on a large wasp is shown in F, since it gave quite a large
fall in tension following shortening. These drops in tension do not appear
to result from mechanical factors in the experimental setup but are rather
the result of some lengthening of the contractile elements.

If as in E ( 1 ) the muscle is rapidly lengthened after the initiation of
stimulation, but before the tension has increased greatly, one may some-
times see the same effect as found when the muscle is lengthened after a
rapid shortening. The tension continues to rise after the completion of
the stretch.

A very rapid stretch after a shortening results in a sharp rise in tension
followed by a precipitous fall after the completion of the movement, and
then a rise to the isometric level, G. In the bee during flight one stroke lasts
about 5 msec. In this experiment the muscle has been stretched in 1 msec.
This viscous-like behavior of the muscle must set an upper limit to the
speed of the system.

These transient responses of the muscle show that, following a rapid
shortening, the contractile elements lengthen slightly and the tension falls.
During the subsequent rapid relengthening, only a portion of the tension
has returned by the end of the movement, as little as 30% in the best cases.
The tension continues to rise very rapidly after the muscle has attained
its initial length and may overshoot the isometric tension at which shorten-
ing began. The rise is due to the shortening of the contractile elements.

Many of the characteristics of the flight machinery found in the study
of intact insects now find explanation in the physiology of fibrillar muscle.
The muscle is basically an elastic system through which chemical energy
is furnished by the contractile elements as necessary to overcome the damp-
ing forces tending to halt motion. Since the snap action, acting in the same
manner as inertia, tends to resist movement, work must be done against
it as well as against air damping at the beginning of the stroke. The elastic
energy so stored maintains movement toward the end of the stroke when
muscle tension is falling.

If we now place the muscle we have studied back into the insect from
which it was isolated, coupling it to the mechanical system described, and
to its antagonist, we can give it the proper label and file it away. At some
future time, however, as we remove it to study its structure we may per-
haps see something more than its external morphology, in fact, one of na-
ture's most successful solutions of heavier-than-air flight.

Summary
Insects, making use of the special mechanical properties of an exoskele-
ton, have evolved two principal types of flight mechanisms. In the syn-

MACHINERY OF INSECT FLIGHT 141

chronous type the usual direct correlation exists between the nerve im-
pulses and the muscle response. In those insects possessing fibrillar flight
muscle the mechanism operates in a different manner. During the active
state initiated by each nerve impulse a variable number of contraction cycles
may occur. Studies of this mechanism were made on the large fly Sarco-
phaga bullata. Recordings of action potentials from the muscle with ex-
ternal and internal leads show that processes involved in building up the
active state are the same in fibrillar muscle as in many other striated
muscles.

The special properties of the asynchronous mechanism result from the
nature of fibrillar muscle and the mechanical system with which it functions.
A snap action is demonstrated in the articulation that is shown to depend
upon the relations of the first two axillary sclerites with three components
of the basic thoracic structure. Records of movements of these components
during flight reveal their action under a variety of conditions and features
of the physiolog}' of fibrillar muscle as well.

Using the longitudinal flight muscle of the bumble bee, Bombiis, it was
possible to study the mechanical properties. Isometric twitch is small. A
smooth tetanus is obtained at 40-60 stimuli per second. Tension-length
relations of maximal isometric contractions are typical but show that
shortening is limited to 10% of muscle length. In situ the shortening is
mechanically limited to 1%.

Only when the active muscle is allowed to shorten with an inertial load,
or is rapidly stretched after a rapid shortening, are its special properties
revealed. In the first case the dynamic tension-length relation is a loop,
the area of which represents the work done in one cycle against viscous
damping forces such as the movement of the wings through the air. The
response to rapid stretch when the muscle has less than the tension charac-
teristic of its length (as after rapid shortening) , in contrast to that of other
muscles, shows that the muscle may be stretched to full length while ten-
sion rises to only a fraction of its initial value. Following the attainment of
full length the tension rises rapidly.

The presence of special mechanical properties in the mechanism moving
the wings is correlated with the operation of fibrillar muscle. Together
these features constitute the principle machinery to move the wings.

REFERENCES

Boettiger, E. G., 1951. Stimulation of the flight muscles of the fly. Anat. Rec. Ill, 443.
Boettiger, E. G., and E. Furshpan, 1950. Observations on the flight motor of Diptera.

Biol. Bull. 99, 346.
Boettiger, E. G., and E. Furshpan, 1950. Observations on the flight motor of Diptera.

Fed. Proc. 10, 17.

142 INVERTEBRATE PHYSIOLOGY

Boettiger, E. G., and E. Furshpan, 1952. The mechanics of flight of Diptera. Biol. Bull.
102,200-211.

Boettiger, E. G., and E. Furshpan, 1954a. The response of fibrillar muscle to rapid re-
lease and stretch. Biol. Bull. 107, 305.

Boettiger, E. G., and E. Furshpan, 1954b. Mechanical properties of insect flight
muscle. /. Cell. Comp. Physiol. 44, 340.

Boettiger, E. G., and F. McCann, 1953. Single fiber action potentials in insect fibril-
lar muscle. Fed. Proc. 12, 17.

Chadwick, L. E., 1940. The wing motion of the dragonfly. Bull. Brklyn. Ent. Soc. 35,
109-112.

Chadwick, L. E., 1953. The motion of the wings ; aerodynamics and flight metabo-
lism ; the flight muscles and their control. In K. D. Roeder, Insect Physiology, New
York, 577-655.

de Geer, C., 1776. Mcmoire pour seri'ir a I'historie des insects. Stockholm (quoted
from Chadwick 1953).

Gasser, H. S., and A. V. Hill, 1924. The dynamics of muscular contraction. Proc.
Royal Soc. {London) B 96, 398-437.

Hill, A. v., 1949-50. The abrupt transition from rest to activity in muscle. Proc. Roy.
Soc. {London) B 136, 399-420.

Hocking, B., 1953. The intrinsic range and speed of flight of insects. Trans. Roy. Ent.
Soc. {London) 104,223-345.

McEnroe, W., 1952. Tension-length curves of insect fibrillar muscle. Fed. Proc. 11,
104.

McEnroe, W., 1954. Insect fibrillar indirect flight muscle. Master's thesis, Univ. of
Conn.

Nastuk, W. L., and A. L. Hodgkin, 1950. The electrical activity of single muscle
fibers, /. Cell. Comp. Physiol. 35, 397.

Pipa, R. L., 1955. A comparative histological study of the indirect flight muscles of
various insect orders. Master's thesis, Univ. of Conn.

Pringle, J. W. S., 1949. The excitation and contraction of the flight muscles of in-
sects. /. Physiol. 108, 226-232.

Pringle, J. W. S., 1954. The mechanism of the myogenic rhythm of certain insect
striated muscles. /. Physiol. 124, 269-291.

Roeder, K. D., 1951. Movements of the thorax and potential changes in the thoracic
muscles of insects during flight. Biol. Bull. 100, 95-106.

Sargent, W. D., 1951. The flight of the dragonfly. Biol. Rev. C.C.N.Y. 13, 8-10.

Sotavalta, O., 1953. Recordings of high wing-stroke and thoracic vibration frequency
in some midges. Biol. Bull. 104, 439-444.

Tiegs, O. W., 1955. The flight muscles of insects — their anatomy and histology ; with
some observations on the structure of striated muscle in general. Philos. Trans.
Roy. Soc. {London) 238,221-359.

Williams, C. M., and R. Galambos, 1950. Oscillographic and stroboscopic analysis of
the flight sounds of Drosophilia. Biol. Bull. 99, 300-307.

NEUROMUSCULAR MECHANISMS*

C. A. G. WiERSMA
California Institute of Technology

It has become well established that the control of muscular contraction
by the nervous system in the typical vertebrate striated muscle is a spe-
cialized case, and that other types of control are present. From a com-
parative viewpoint this is a logical development. For the nervous control
of muscular action must be a secondarily evolved process, since the con-
tractile elements can be brought into action without nervous structures in
protozoa, and most likely also in sponges. Therefore, it is likely that the
original function of the nerves was regulation of myogenic contractions
and that only subsequently the nerves obtained more complete control.
It may well be that in, e.g., the smooth muscles of the digestive tract in all
phyla the regulatory function of the nervous system has been maintained.
However, for the body musculature, especially the parts involved in quick
withdrawal reactions, nervous activation became necessarv at an early
stage, whereas slower and more tonic muscles may have kept a greater
independence. But again, for quick reaction more than one method of acti-
vation is used. In the vertebrate striated muscle fiber it is the development
of a conducted muscle action potential which spreads the excitation from
the end plate to the rest of the fiber, but in many arthropods conduction
along branches of the nerve fibers is certainly mainly responsible for this
spread. It will be the task of future investigations to discover to what ex-
tent the possibilities roughly outlined above are adequate for the explana-
tion of the control of the muscles in the different phyla, how far as yet un-
known mechanisms have been realized, and also how far the different
muscles of one animal are under a similar type of control.

Muscles with fast and with slow contraction speeds are widespread. It
is interesting to note that in protozoa contractile fibrils differing vastly in
this respect are present. The types of innervation of muscle fibers of one
muscle are not always similar, as has been demonstrated in the frog
muscles, many of which have two types of muscle libers each with its own
type of innervation. Kufffer (1955) has reviewed the work which estab-
lished this. The "slow" or tonic system shows, in contrast to the conven-
tional fast system, multiple nerve endings on the muscle fibers and an ab-

* This review will be largely limited to those newer papers, which have appeared
since the author's previous reviews (Wiersma, 1952-53) which dealt partially with
the same topic.

[143]

144 INVERTEBRATE PHYSIOLOGY

sence of twitches and of conducted muscle action potentials. Instead,
only slow contractions and facilitating junctional potentials are present.
The two nerve-muscular systems are here very independent. In arthropods,
on the other hand, all indications are that the same muscle fibers are in-
volved in the various contraction types (see below), and that the type of
innervation is much the same. In other phyla fast and slow contractions
are sometimes definitely due to two different types of muscle fibers ; but
it will be again necessary to obtain more information before meaningful
comparisons with the better-known mechanisms can be made.

The use of intracellular electrodes may help much for the rapid increase
of our knowledge about these questions. However, the results with decapod
crustaceans to be described may be a warning that, even with these
methods, the finer details may be so complex that conclusions as to types
may be difficult to draw.

One factor influencing the effect produced by a nerve impulse arriving at
the muscle fiber is undoubtedly the state of the fiber. Not only may there
be previous facilitation and fatigue, but the actual state of stretch of the
muscle fiber may have a considerable influence on the outcome. In verte-
brate striated muscle it is well known that the heat produced during isomet-
ric and isotonic twitches is quite different. Ralston and Libet ( 1953) found
in addition that the mere stretching of a vertebrate motor end plate may
result in a conducted muscle potential when this did not occur in the relaxed
muscle fiber— which brings to mind what von Uexkiill (1904) long ago
showed, that it is often easier to make stretched muscles contract. He found
that brittle star arms would invariably contract upward when hanging
down, whether or not the stimulation was nearer to the stretched side of
the arm. In cases of myogenic contractions, stretch is also known to be
an important factor, as in vertebrate intestine and in the heart of Helix
(Willems, 1932). It seems likely that at least in some of these instances
reflex activity is not present.

These considerations may have made it clear that a great variety of
conditions exist, and that the conclusions which can be drawn from the
study of any one preparation or even of the muscles of one group may have
only a limited applicability to other systems. With this reservation in mind
the following review of recent investigations may be undertaken.

Investigations in Arthropods

Among the invertebrates, the arthropods still remain by far the most
favorable for neuromuscular studies. Their muscle fibers have lent them-
selves readily to the application of intracellular electrodes. The results thus
obtained with insects and the intriguing studies of the relations in indirect

NEUROMUSCULAR MECHANISMS 145

flight muscle of higher insects are treated elsewhere in this volume, and
except for one reference will not be considered.

In decapod Crustacea the use of internal electrodes has given some re-
markable results. Studying the electrical properties of the muscle mem-
branes as such, Fatt and Katz (1953a) have shown that they differ
markedly in several respects from those of vertebrate muscle fiber. These
properties of the membrane may well have a considerable influence on its
functional relations to contraction and inhibition. But, since at present
these relations are not clear, the results of their studies will not be dis-
cussed further.

With the use of internal electrodes, Fatt and Katz ( 1953b) and Fursh-
pan and Wiersma (1954) were able to show that an impaled muscle fiber
of a doubly motor-innervated muscle would react on stimulation of either
axon. The potentials obtained depend on the axon stimulated; in "slow-
axon" stimulation considerable facilitation may be needed before the de-
polarization is evident, while single impulses in the "fast" axon usually
give rise to a clearly observable deflection. These results were obtained
with many fibers of a wide array of muscles and represent undoubtedly the
normal efTect. But the experiments do not necessarily prove that all muscle
fibers of a doubly motor-innervated muscle receive innervation from both
axons ( see below) . Fatt and Katz ( 1953b) have studied the distribution of
the potentials along the fiber, impaling one at a number of loci, and have
very often found insignificant dift'erences in amplitude and time relations.
These results are thus in sharp contrast to those obtained with the focal
end-plate potential of mammalian muscle fibers, and make it certain that a
decapod crustacean muscle fiber normally receives its excitation at the many
nerve endings which each axon has along the length of the fiber.

Since there exi'sts a good deal of variation in type of innervation and
effects of stimulation in difiierent muscles, a new terminology, with some
terms already used for other preparations, has been proposed by Furshpan
(1955). With a slight variation in definition these terms will be presented
here. Instead of multiple innervation, an older term which covered at the
same time the fact that many nerve endings are present and that more
than one axon makes connection with a muscle fiber, the term multiterminal
innervation will be used to describe the fact that one axon has a consid-
erable number of endings on a muscle fiber. To indicate that more than one
motor fiber innervates a muscle, the term polyneiironal motor innervation
will be used (subdivisions like dineuronal motor innervation, etc. can be
derived from this). But every decapod crustacean muscle fiber may have
at least a dineuronal innervation, since in all known instances there is at
least one inhibitor, and sometimes two, present in addition to the motor
fiber or fibers (Wiersma, 1941 ). The types of potentials which can be ob-

146 INVERTEBRATE PHYSIOLOGY

tained from muscle fibers show differences that make a classification neces-
sary. Instead of end-plate potentials the presumably purely local effects
around the nerve endings will be called junctional potentials as was done
by Kuffler and Gerard (1947) for the comparable potentials in the slow
muscle fibers of the frog. The secondary response which arises from these
potentials after they surpass a certain value will be called a spike. This will
not imply that the process is conducted over the whole membrane ; when
the latter takes place it will be called a conducted spike. The conducted
spike always shows, as far as known, an overshoot of the membrane po-
tential, whereas spikes may vary from just visible enhancements of the
junctional potentials to the maximum level with overshoot. Since there are
strong indications that the junctional potentials (but not the spikes) of the
different axons innervating the muscle fiber have different mechanical
results, which do not depend on the shape or size of the potential, they must
also be named with regard to the axon bringing them about. Therefore, in
a double motor innervated muscle it will be necessary to distinguish be-
tween "slow" junctional potential and "fast" junctional potential.

The potentials of the contractile part of the muscle-stretch receptor
organs have been studied by Kuffler (1954) and Furshpan (1955), using
one or two nerve impulses only, which will presumably make the contribu-
tion of any "slow" motor axon, if present, negligible. These structures are
favorable for this type of work because they constitute thin isolated strands,
which we consider as single muscle fibers. They offer a clear picture of the
motor innervation, the motor fiber (s) running along the length of the
structure, giving off branches into it at many points (Alexandrowicz,
1951). Both Kuffler and Alexandrowicz are inclined to consider them as
consisting of bundles of muscle fibers rather than single units. However,
Furshpan, using two microelectrodes, failed to find any sign of high-re-
sistance membranes between them, indicating the absence of charged
membranes. The only exception found was when one electrode was in the
anterior muscular segment, the other in the posterior one, separated by the
intercalated region in which the sense cell has its endings.

Using the anterior muscular section, Kuffler and Furshpan have both
observed spikes which were not conducted. Kuffler found that, when no
spikes were present at a given time, stretching the organ by pulling it to
one side would bring one about, but this spike was confined to the stretched
region. Furshpan, using two internal electrodes, found that one locus
might spike at a time that the other gave only a large junctional potential.
When a second impulse was delivered shortly afterwards, the other locus
would also spike. He could show that the speed of spread of the spike,
when spiking took place all along the structure, would be the same as that
of the junctional potential, and concluded that the nerve fiber plays the

NEUROMUSCULAR MECHANISMS 147

primary role in the distribution of the excitation along the muscle fiber
and that the spike is not the indication of conduction.

While it is thus quite possible that under normal conditions different
parts of one fiber react difi^erently to stimulation, it may be considered
certain that dift'erent muscle fibers of one muscle react rather independently.
In the closer muscles of crabs it was regularly observed that a single im-
pulse in the "fast" axon would give a large spike in part of the muscle fibers,
in others an abortive spike or large junctional potential, while in a different
part of the muscle only small junctional potentials would be present in the
fibers (Furshpan and Wiersma, 1954). Sampling in these cases has not
yet been extensive enough to make certain that there is a constant gradient
in this respect from the back to the front of the muscle, although this was
often observed. Fatt and Katz (1953b) also report large differences be-
tween the responses of muscle fibers of one muscle.

Polyneuronal innervation is, according to the results of Furshpan
(1955), also variable from muscle fiber to muscle fiber in one muscle.
In many muscles with double motor innervation the indications are at
present that the great majority of fibers, perhaps all, do indeed receive
branches from both axons. But in the main flexor of the rock lobster, which
has four motor axons supplying it, it was found that, when single penetra-
tions of numerous muscle fibers were used, only a relatively small per-
centage (7%) of the muscle fibers responded definitely to all four motor
axons. Response to three axons was obtained in 29%, to two in 26%, and
to only one axon in 38% of the fibers. In the last case it is known that, of
these, 90% responded solely to stimulation of the particular axon, which
elicits the fastest contraction. Since none of the fibers tested failed to re-
spond to at least one motor axon, these figures may represent a fair approxi-
mation of the distribution. A further significant observation of Furshpan
is that it is possible to obtain spike potentials by combining the stimulation
of two axons, when each gives only a junctional potential by itself. This
proves that the junctional potentials caused by any of the motor axons
must be considered equivalent with regard to their relation to the mem-
brane changes.

In insects a rather similar picture has been reported by Hoyle (1955
and this volume) for locust muscles. Here only certain of the muscle fibers
of a given muscle receive a polyneuronal motor innervation, the majority
being supplied by a single axon.

A functional differentiation between parts of a crustacean muscle, which
must be due to unequal distribution of axon types or number of their end-
ings, has been reported (Wiersma and Ripley, 1954). It was observed
that, in contrast with the great majority of leg joints, one joint in the
walking legs of a hermit and of a dromid crab can rotate as well as bend at

148 INVERTEBRATE PHYSIOLOGY

the joint. It was found that, of the two motor fibers innervating these
muscles, the "fast" axon causes rotation in one direction, the "slow" in the
opposite.

Comparing muscles as a whole, differences in the process of their neuro-
muscular transmission can be demonstrated, not only between species and
between muscles with different location, but also between homologous
muscles. In Homarus it was observed (Wiersma, 1951) that the closer
muscles of its three types of claws show a remarkable difference in their
mechanical reaction to single and double nerve impulses in the "fast" axon.
In a more detailed study (Wiersma, 1955) of these effects, it was found
that in this species there is a strong correlation between the appearance of
diphasic action potentials and twitches. In general many crustacean muscles
show monophasic potentials with outside leads, even when twitches are
present (Wiersma and Van Harreveld, 1938a; Wiersma and Wright,
1947) . But in Homarus it was found that single impulses in the "fast" axon
of the small claws of the second and third leg result in a clearly diphasic
deflection, accompanied by a twitch. In the cutter claw of the first thoracic
legs, which shows at best a slight movement of the tip on a single impulse,
the deflection is largely monophasic, while in its partner, the crusher claw,
which does not move at all, it is completely so. When two impulses at a
short interval are given, the mechanical effect in the cutter claw is dramatic,
as it closes completely and with considerable force. A nearly maximal
diphasic deflection precedes this contraction. In the crusher claw no me-
chanical effect is obtained, and a summation of the monophasic deflections
takes place. In the small claws the second deflection is only somewhat larger
than the first, and the summated contraction is still not strong enough to
close the claw. For this animal it seems certain that conducted spikes are
responsible for the diphasic deflection. The very strong faciliation effect
which a single impulse in the "fast" axon of the cutter claw exhibits, in con-
trast to the much weaker one of the other muscles, is a good demonstration
how slight differences in properties can make crustacean muscles especially
well adapted for certain special functions.

In general, the relations between junctional potentials, spikes, con-
ducted spikes, and the resulting contractions remain uncertain. The great-
est problem in this regard remains the one which has been named the
paradox (Wiersma and Van Harreveld, 1938b). There can be little doubt
that under given circumstances, especially on low-frequency stimulation,
stimulation of the "fast" axon will result in large junctional potentials,
which do not elicit a contraction, while similar stimulation of the "slow"
axon will result in small junctional potentials, accompanied by a slow con-
traction. How this is possible when both potentials occur in the same mem-
brane is completely unknown. This quandary is, of course, caused by our

NEUROMUSCULAR MECHANISMS 149

lack of knowledge of the transfer from electrical to mechanical effects. In
this connection an observation on the spread of contraction in a living
muscle fiber is interesting. Matthaei and Tiegs (1955) photographed a
slowly spreading contraction wave in a slightly damaged spider muscle
fiber. The contraction originated from under an end-plate structure. It
first spread across the muscle fiber before it went in two directions away
from this region.

Further clues with regard to this problem should be obtainable from the
effects of the inhibitory impulses. Fatt and Katz (1953c), studying es-
pecially the inhibition in the opener muscle of the hermit crab, have come
to a number of interesting conclusions. Depending on the magnitude of the
membrane potential they found that inhibitory impulses could have either
no electrical effect at all (which was the condition when the membrane
potential was "normal"), cause a hyperpolarization when the membrane
potential was low, or give a depolarization when it was higher than normal.
The time course of these inhibitory polarization effects was of the same
order, but slightly longer than that of the excitatory junctional potentials.

Reductions of the excitatory junctional potential up to 90^ of its value
could be obtained when the inhibitory impulse preceded the excitatory one.
When an inhibitory stimulus was given during the course of an excitatory
junctional potential change, its decay time was speeded up. These results
certainly go far in explaining the inhibition of the junctional potentials by
the postulation of one inhibitor-receptor reaction, which changes the ion
permeability of the muscle-fiber membrane and competes with the action of
an excitatory transmitter. However, they offer no ready explanation for
mechanical inhibition which occurs without even a reduction of the facili-
tation of the junctional potentials ( Marmont and Wiersma, 1938 ; Wiersma
and Ellis, 1942). It is difficult to believe that the membrane could change
its electrical properties without at the same time influencing the facilita-
tion process. Hence it still seems likely that the main effect of inhibitory
stimulation is on a transmission process between the membrane changes
and the contractile process.

It has been shown that spacing of the impulses in inhibitory stimulation
can have an effect similar to that of excitatory stimulation. Ripley and
Wiersma ( 1953) found that the same number of inhibitory impulses, when
given in pairs at a short interval, gave a more pronounced inhibition in the
opener muscle of the claw of the crayfish than when they were given all at
equal time intervals.

That transmitters are involved seems quite certain. Concerning their
nature little is known as yet, which may well be due to the way in which the
nerve fibers end, sublemnally in the muscle substance. The endings may
thus be well protected from the direct influence of drugs. In accord with this

150 INVERTEBRATE PHYSIOLOGY

concept is the fact that, when a drug is found effective in a nerve-muscle
preparation, it can be shown that in most instances it has a similar effect
when applied to the axon alone (Ellis, Thienes, and Wiersma, 1942).
According to Florey and Florey (1954) there is evidence that, in double
motor-innervated muscles of the crayfish, acetylcholine is the transmitter
of the fast system, while 5-hydroxytryptamine would bring the slow sys-
tems into action since these drugs caused, on perfusion, contractions of
different types. It was, however, not shown that these effects were not due
to stimulations of the axon branches outside the muscle fibers.

Since it is well known that both fast and slow systems can be inhibited
by the same inhibitory fibers and that the same holds true for muscles with
four motor fibers, a very interesting field of research offers itself.

Florey (1954) has since withdrawn the claim that 5-hydroxytrypta-
mine is the transmitting substance for the slow contractions. This sub-
stance causes stimulation of sensory end organs and their sensory nerves,
which in turn stimulate in some way the motor nerves. The present re-
viewer considers it likely that this stimulation is due to an ephaptic trans-
mission process, which takes place in a region where the slow motor axons
are hyperexcitable because of an existing demarcation potential. This will
be normally near their cut end.

Investigations in Molluscs

The physiology of neuromuscular transmission in molluscs still suffers
from the uncertainty caused by the fact that histological methods have not
yet shown conclusively where peripheral ganglion cells are present, inter-
calated between the motor fibers of the main nerves and their endings. A
very recent preliminary publication (Bowden and Lowy, 1955) reports the
presence of nerve cells in all lamellibranch muscles so far examined, among
which are muscles previously believed to be free from such cells. It will
have to be proven that these nerve cells and their synapses are situated in
the motor pathway ; but it will be necessary to consider this possibility in
evaluating the physiological data, until a definite answer is obtained.

In the adductor muscles of Anodonta, Barnes ( 1955) has found evidence
that the fast part of the muscles is innervated by nerve fibers which behave
as typical motor axons, causing a relatively quick contraction and relaxa-
tion (it is, however, of importance to keep in mind that van Overbeek,
1931, observed that the fast part of this muscle could be brought to con-
traction by quick stretch after its isolation from the ganglia) . Although the
slow part of the muscle could be made to contract by nerve impulses, its
relaxation would depend on an active process through the medium of in-
hibitory fibers. This view corresponds well with one held for the slow part
of the adductor muscle of Pecten by Benson et al. (1942), who showed

NEUROMUSCULAR MECHANISMS 151

that, by stimulation of certain parts of the nerve bands going to the muscle,
relaxation was considerably speeded up — indicating the existence of in-
hibitory axons.

In the adductor muscle of Mytilus there is no obvious difference between
fast and slow muscle fibers. In contrast to Anodonta and Pcctcn, in which
the fast fibers are striated, here all fibers seem to be smooth (Lowy, 1955).
Using the animal's naturally occurring contractions and relaxations, Lowy
(1953) found action potentials present in this muscle, even during the
long-lasting tonic states, in which they were infrequent. Higher-frequency
bursts occurred at the onset of contraction, but also at the onset of relaxa-
tion. The latter are rather unexpected, but might come about when both
excitatory and inhibitory impulses were reaching the muscle, when the
inhibitory ones would inhibit the contraction but not the concurrent muscle
action potentials. Lowy (1955) now reports that muscle action potentials
can also be obtained from muscles, isolated from the ganglia. He proposes
that these may be due either to the presence of nerve impulses in the
peripheral nerve-ganglion system or be of myogenic nature ; he favors the
former view. With Barnes one may still have doubt whether only one type
of muscle fiber is present in this muscle ; for, though all fibers may be
smooth, this would not be a certain demonstration that some of them
cannot contract much faster than others. -^

Another instance in which a molluscan muscle consists of two types of
muscle fibers, but this time in series with each other, has been reported by
ten Gate and Verleur (1952) — the retractor muscle of the main tentacle
of the snail Helix pomatia. The distal part is dark in color and hollow and
gives much more phasic contractions than the tonic light-colored proximal
part.

From these observations one may be inclined to believe that in the
molluscs as in the amphibians two separate neuromuscular systems are
present, of which the slower one may well be mainly under local nervous
control in contrast to the situation in amphibians. Under the circumstances
it would be futile to speculate about the types of innervation of the muscle
fibers as such. Inhibition, for instance, may well be an effect wholly located
in the peripheral nervous system, and therefore not comparable to that of
the arthropods. One would like to have at hand more data about the high-
est forms in this phylum, such as the squid. But here the investigations of
Prosser and Young (1937), which showed that a single impulse in the
giant motor fibers would lead to a maximal contraction of the whole part
of the mantle, whereas smaller axons seemed to give similar but more re-
stricted twitches, are still the only indication that polyneuronal motor
innervation may be a possibility in molluscs.

In the long-fibered anterior byssus retractor of Mytilus, whose fibers

152 INVERTEBRATE PHYSIOLOGY

may run the length of the whole muscle, Twarog ( 1954) found that applied
acetylcholine depolarizes the membranes and instigates contraction. When
the drug is subsequently washed out, the membranes repolarize but con-
traction remains. Relaxation of this tonic part of the contraction can be
obtained by the application of 5 -hydroxy tryptamine (this is also the case
when contraction is caused by other means). This drug does not cause a
noticeable change in polarization and is inefifective in preventing the onset
of the contraction caused by acetylcholine. In bioassays, the normal pres-
ence of these substances in this muscle was observed.

From these interesting observations it would seem that 5-hydroxytryp-
tamine may be considered rather a relaxing than an inhibiting substance,
and that active relaxation is necessary to release the "catch" mechanism,
provided that the substance works directly on the muscle fibers and that
only one type of muscle fiber is present.

In a very recent paper (not considered in the original manuscript)
based on independently performed experiments, Hoyle and Lowy ( 1956)
come to the conclusion that inhibitory nerves are present in the anterior
byssus retractor muscle of Mytilus edulis. Furthermore, this muscle
would have a built-in system which can fire automatically, this system
consisting either in peripheral nerve cells or the spontaneous activity in
muscle fibers themselves. They find again that during prolonged tonic
contractions action potentials are present, although these occur only in
certain areas of the muscle and not everywhere. They conclude that tonic
contractions are based on the tetanic activity of such parts. They confirm
that 5-hydroxytryptamine abolishes tonic responses, and that this sub-
stance does not materially afifect the phasic ones. Because the muscle is
unable to destroy added 5-hydroxytryptamine they consider it unlikely
that it is a natural transmitter. This reviewer is less inclined than the
writers to accept on this evidence the absence of a "catch" mechanism.
Twarog's (1954) findings with acetylcholine, in which the fiber stays
contracted even after the acetylcholine is washed out, indicates to her
that such a mechanism may be present. One would want to correlate the
total electrical activity of the muscle and the contraction in order to evalu-
ate the significance of the action potentials present during rest. The pres-
ence or absence of such correlation would go far to prove or disprove the
contentions of this paper. It is interesting to note that, as in crustacean
muscle, the independent reactions of the muscle fibers making up the
muscle are a great hindrance to this approach.

Investigations in Echinoderms

The presence of phasic and tonic muscle fibers in echinoderms was con-
vincingly demonstrated by von Uexkiill for the muscles moving the spines

NEUROMUSCULAR MECHANISMS 153

of sea urchins (1900). He found that the muscles consist of two rings of
radial muscle fibers, the outer ring for the relatively quick movements of
the spine, for which relaxation is also fast, while the inner muscles can
lock the spine so strongly into position that, when forced, it may break
rather than give, or the tonic ring may be torn, so that only the phasic
muscle fibers remain active. Two types of contractions have recently been
reported for the pharyngeal retractor muscles of the sea cucumber, Cucii-
marm, on stimulation of the radial nerve (Pople and Ewer,. 1954, 1955).
The quick response did not show facilitation, but the slow one showed a
marked and prolonged type of facilitation. These findings are in accord
with the two types of action potentials previously obtained from the retrac-
tor muscles of the pharynx of Thyone (Prosser, Curtis, and Travis,
1951). Pople and Ewer argue, however, that the different contractions
found are not due to two types of muscle fibers but to neuro-neural facilita-
tion in the ganglion-cell complex, described by Smith (1950) and called
the "motor complex." It seems certain that this complex must play some
part in the reactions of these preparations but whether it can completely
explain the different types of contraction may be doubted.

Prosser (1954) has made an electrophysiological and histological in-
vestigation of the long retractors of the body wall of the sea cucumber,
Thyone. This short-fibered smooth muscle gives only one type of contrac-
tion, which may be considered as of the fast type (Prosser, Curtis, and
Travis, 1951). On "direct" stimulation the spread of the muscle action
potentials is very restricted. The reason for failure of the propagation is
ascribed to the fact that many small nerve fibers pass from the radial nerve
to the muscle. These behave as separate units and not as branches of one
or a few axons. From these results the conclusions are drawn that the
muscle is not syncytial in structure and that many unconnected motor
nerve fibers are involved in the innervation of this muscle. The latter con-
clusion is in accord with the results of Pople and Ewer (1954) and agrees
too with all previous work on echinoderm musculature.

It would seem at present that the neuromuscular systems of echinoderms
may show a greater similarity to those of amphibians than do those of any
other phylum, since two different neuromuscular systems controlled by
many motor neurons seem to be present. However, this resemblance may
still be only superficial. For it will have'fo be shown of what importance
the peripheral ganglion cells are in the motor chain before any conclusions
can be drawn.

Investigations in Annelids and Other "Worms"

With one notable exception no recent research seems to have been per-
formed on this group. This is regrettable, since it would be of special im-

154 INVERTEBRATE PHYSIOLOGY

portance to compare the annelid and the arthropod neuromuscular
systems. Prosser and Melton (1954) have analyzed the proboscis re-
tractor muscle of Phascolosoma (Sipunculoidea) with electrophysiologi-
cal and histological methods. They found both fast and slow contrac-
tions present and two types of action potentials. The muscle fibers are all
smooth and short. Many parallel nerve fibers innervate the muscle and the
axons can be divided into two classes, thicker (2/x) and thinner (below
IjLi). Both types of action potentials fail to be conducted after nerve de-
generation. It is therefore concluded that conduction is always by nerve
fibers and not by protoplasmic bridges between muscle fibers. Whether
the muscle fibers also are of two types or whether dineuronic motor inner-
vation occurs has not yet been decided. Multiterminal endings may well be
present. The fast potentials have the properties of spikes and do not facili-
tate, the slow ones those of junctional potentials with facilitation. On fa-
tigue the spike shows a prespike potential, but a similar phenomenon was
observed for the slow potentials. It seems therefore premature to identify
these phenomena with the spike and junctional potentials of the Crustacea.
There is, however, a similarity in the shape of the action potentials, which
are different in both cases from the conducted potentials of vertebrate
muscle.

Thus in Phascolosoma the evidence is that relatively small muscle areas
are governed by specific nerve fibers ; but it should be kept in mind that
in the sabellid annelids M^JjfzVo/a (Nicol, 1948) smd Branchiomma (Nicol,
1951 ) the giant fibers in the central nervous system innervate a very large
part of the longitudinal musculature directly. These giant fibers consist of
a conglomeration of many cells, and one may, therefore, consider that in
principle the muscles would still be innervated by many axons forming a
single unit. If this were the only innervation, these muscles, which form
the greatest part of the whole musculature, could function only in the with-
drawal reflex. For the earthworm, preparations have been described in
the older literature which, with modern techniques, might yield significant
results concerning this problem (e.g., Garrey and Moore, 1915).

Investigations in Coelenterates

In the "lowest" phylum in which neuromuscular transmission is pres-
ent, investigation might well be very difficult. It is therefore gratifying that
certain facts have gradually come to light. A most welcome contribution
has been the observation of Horridge (1953, 1954) that motor-nerve im-
pulses can be obtained from single axons, a single impulse accompanying
each beat of the subumbrella in the jellyfish Aurellia. These results make it
much more likely that the through-conducting system of Pantin (1952) in
sea anemones functions indeed as he pictured it, namely by axons con-

NEUROMUSCULAR MECHANISMS 155

nected by synapses with two-way conduction and with no noticeable
synaptic delay, making it possible to consider these systems as simple
motor-nerve fibers. As a consequence the facilitations observed have then
to be located at the myoneural junctions.

From anatomical evidence of Semal-van Gansen (1952a,b) it seems
somewhat doubtful whether in Hydra the same differentiation is respon-
sible for the difference in reactions shown by the longitudinal and circular
muscles, which are here ecto- and endodermal, respectively. The muscle
fibril bundles are found to be located in the basal parts of the cells. The
ectodermal bundles are in juxtaposition with each other, and more than
one bundle may belong to one ectoderm cell. In the transverse fibers of the
endodermal cells there is only one muscle fibril bundle per cell and these
are usually well separated from each other. The nervous system consists
in the ectoderm of multipolar cells, which may fuse, while in the endoderm
the nerve cells are bipolar without visible connections. There may thus be a
relation between these anatomical findings and the rather quick contrac-
tion of the longitudinal, ectodermal layer and the much slower stepwise
contractions of the circular, endodermal fibers.

In sea anemones, Batham and Pantin (1954) and Passano and Pantin
(1955) have found further confirmation of the differences in reactions of
the different muscle layers. It has especially become clearer that the slow
contractions and with them the slow movements of the animal as a whole
may be based on spontaneous activity in the muscles.

Investigations in Tunicates

Tunicates have been investigated by Hoyle ( 1952) with methods similar
to those used by Pantin on sea anemones, and some remarkable resem-
blances were shown to exist. But, as Hoyle has pointed out, until one can
be more certain of the alleged underlying factors, it must necessarily re-
main doubtful how far the phenomena observed find their origin in the
neuromuscular transmission. It is nevertheless remarkable that animals
which are phylogenetically presumably so far apart seem to have many
similarities in mechanisms. The sessile mode of living may be part of the
explanation, but other sessile animals clearly have different mechanisms,
or at least still show in part of their behavior properties which the free-
living related forms possess (e.g., barnacles).

Conclusion

Some very gratifying advances have been made in the area covered by
this review. In one respect, though, the recent results may be somewhat
discouraging, since they have shown that slight changes in the properties
of neuromuscular transmission result in great differences in readily ob-

156 INVERTEBRATE PHYSIOLOGY

servable reactions to stimuli and that these shght changes actually differ-
entiate muscles and even the muscle fibers functionally. Hence a welter of
phenomena is encountered which makes analysis much more difficult. Of
course, this very same differentiation is in itself a highly interesting prob-
lem with many connotations, especially for such fields as transmission by
synapses and integration. But it is no longer possible to think of a given
special mechanism as typical for a group or phylum ; instead, the whole
array of variations has to be considered. We have as yet no good indica-
tion of how widely varied the mechanisms are in the arthropods nor how
much of this variability is present in other phyla. It is obvious that a much
more intensive study of all kinds of neuromuscular transmissions will have
to be made before it will be possible to compare intelligently the different
phyla with each other. It can only be hoped that the use of impaled muscle
fibers will provide a way of gathering data at a much increased rate. How-
ever, the mechanical aspects of contractions should not be neglected for a
comprehensive picture. There is here a field wide open for investigation,
in which much more research is necessary than is at present being per-
formed. Let us hope that this research will be done ; for it will be a re-
warding task, not only for its own sake but also for a better understanding
of problems of wider scope.

Since the foregoing paragraphs were written, a number of papers about
Coelenterates have appeared. These can be only summarily discussed. Hor-
ridge (1955 a,b,c ; 1956) continued his investigations in different medusae,
using neuromuscular preparations and histological investigations, but
without the study of the electrical phenomena. He found that in hydro-
medusae Geryonia proboscidalis and Aequoria forskalca the swimming
movements are governed by a circular through-conducting nervous sys-
tem. In Geryonia this system is not influenced by the radial system, which
is responsible for the movement of the manubrium by tonic contractions
of its radial musculature towards a place stimulated on the bell. But in
Aequoria contraction of the radial musculature has an inhibiting effect
on the through-conducting system as in a number of other hydromedusae.
In the scyphomedusa Rkizostoma pulmo repetitive stimulation causes a
shortening of the refractory period in a part of the muscle fibers of the
circular muscle of the bell. In the scyphomedusae Cyanea and Cassiopea
the compensatory contractions made to keep the animals in a vertical po-
sition during swimming were studied. It was found that tonic contractions
are caused by a second nervous net, the diffuse net, which acts locally and
delays the relaxation time. In order to explain the all-or-none contraction
of swimming due to the through-conducting nerve ring and the tonic con-
tractions of the compensatory contractions, Horridge postulates the pres-
ence of double motor innervation of the muscle fibers. Ross (1955) has

NEUROMUSCULAR MECHANISMS 157

studied the effect of temperature on the mechanical effects of single and
double stimuli in the sea anemone Calliactis. At higher temperatures,
single impulses start to give visible contractions of the longitudinal mus-
cles, but at the same time the summation period for two impulses de-
creases considerably. From these results he draws further conclusions
regarding the processes of excitation and facilitation.

REFERENCES

Alexandrowicz, J. S., 1951. Muscle receptor organs in the abdomen of Homarus
vulgaris and Palinums vulgaris. Quart. J . Micr. Sci. 92, 163-199.

Barnes, G. E., 1955. The behavior of Anodonta cygnea L. and its neurophysiological
basis. /. Exp. Biol. 32, 158-174.

Batham, E. J., and C. F. A. Pantin, 1954. Slow contraction and its relation to spon-
taneous activity in the sea-anemone Metridium senile (L). /. Exp. Biol. 31, 84-
103.

Benson, A. A., J. T. Hays, and R. N. Lewis, 1942. Inhibition in the slow muscle of
the scallop, Pecten circularis aequisulcafus Carpenter. Proc. Soc. Exp. Biol.
AT.F. 49, 289-291.

Bowden, J., and J. Lowy, 1955. The Lamellibranch muscle. Innervation. Nature
176, 346-347.

Cate, J. ten, and J. D. Verleur, 1952. Recherches sur la fonction du M. retracteur
du tentacule majeur d' Helix pomatia (L). Physiol. Coinp. et Oecol. 2, 346-349.

Ellis, C. H., C. H. Thienes, and C. A. G. Wiersma, 1942. The influence of certain
drugs on the crustacean nerve-muscle system. Biol. Bull. 83, 334-352.

Fatt, P., and B. Katz, 1953a. The electrical properties of crustacean muscle fibres.
J.Physiol. 120,171-204.

Fatt, P., and B. Katz, 1953b. Distributed "end-plate potentials" of crustacean
muscle fibres. /. Exp. Biol. 39, 433-439.

Fatt, P. and B. Katz, 1953c. The efTect of inhibitory nerve impulses on a crusta-
cean muscle fibre. /. Physiol. 121, 374-389.

Florey, E., 1954. Uber die Wirkung von 5-Oxytryptamin (Enteramin) in der Krebs-
schere. Ztsch. f. Naturf. 9b, 540-545.

Florey, E., and E. Florey, 1954. Uber die mogliche Bedeutung von Enteramin
(5-Oxy-Tryptamin) als nervoser Aktionssubstanz bei Cephalopoden und deka-
poden Crustaceen. Ztsch. f. Naturf. 9b, 58-68.

Furshpan, E. J., 1955. Studies on certain sensory and motor systems of decapod
crustaceans. Thesis, California Institute of Technology.

Furshpan, E. J., and C. A. G. Wiersma, 1954. Local and spike potentials of impaled
crustacean muscle fibers on stimulation of single axons. Fed. Proc. 13, 51.

Garrey, W. E., and A. R. Moore, 1915. Peristalsis and coordination in the earthworm.
Amer. J. Physiol. 39, 139-148.

Horridge, A., 1953. An action potential from the motor nerves of the jellyfish,
Aurellia aurita Lamarck. Nature 171, 400.

Horridge, A., 1954. The nerves and muscles of Medusae. I. Conduction in the nerv-
ous system of Aurellia aurita Lamarck. /. E.rp. Biol. 31, 594-600.

Horridge, A., 1955a II. Geryonia proboscidalis Eschscholtz. /. Exp. Biol. 32, 555-568.

Horridge, A. 1955b. III. A decrease in the refractory period following repeated stimu-
lation to the muscle of Rhicostoma puhuo. J. Exp. Biol. 32, 636-641.

Horridge, A. 1955c. IV. Inhibition in Aequorea forskalea. J. Exp. Biol. 32, 642-648.

158 INVERTEBRATE PHYSIOLOGY

'Horridge, A. 1956. V. Double innervation in Scyphozoa. /. Exp. Biol. 33, 366-383.
Hoyle, G., 1952. The response mechanism in Ascidians. /. Mar. Biol. Ass. U.K. 31,

287-305.
Hoyle, G., 1955. Neuromuscular mechanisms of a locust skeletal muscle. Proc. Roy.

Soc. (London) B 143, 343-367.
Hoyle, G. and J. Lowy, 1956. The paradox of Mvtilus muscle. A new interpretation.

/.ii.r/',5jo/. 33,295-310.
Kuffler, S. W., 1954. Mechanisms of activation and motor control of stretch recep-
tors in lobster and crayfish. /. Neurophysiol. 17, 558-574.
Kuffler, S. W., 1955. Contracture at the nerve-muscle junction: the slow muscle

fiber system. Amer. J. Physical Med. 34, 161-171.
Kuffler, S. W., and R. W. Gerard, 1947. The small-nerve motor system to skeletal

muscles. /. Neurophysiol. 10, 383-394.
Lowy, J., 1953. Contraction and relaxation in the adductor muscles of Mytilus

edulis. J. Physiol. 120, 129-140.
Lowy, J., 1955. The lamellibranch muscle. Contractile mechanism. Nature 175, 345-

346.
Marmont, G., and C. A. G. Wiersma, 1938. On the mechanism of inhibition and

excitation of crayfish muscle. /. Physiol. 93, 173-193.
Matthaei, E., and O. W. Tiegs, 1955. The path of the slow contractile wave in

arthropod muscle fibre. Phil. Trans. Roy. Soc. (London) B 238, 349-359.
Nicol, J. A. C, 1948. The giant nerve-fibres in the central nervous system of Myxi-

cola (Polychaeta, Sabellidae). Quart. J. Micr. Sci. 89, 1-45.
Nicol, J. A. C, 1951. Giant axons and synergic contractions in Branchiomma vesiculo-

sum. J. Exp. Biol. 28, 22-31.
Overbeek, J. van, 1931. Uber die Tonuserzeugung unter dem Einfluss von Muskel-

dehnung (bei Anodonta cygnca). Ztsch. j. vcrgl. Physiol. 15, 784-797.
Pantin, C. F. A., 1952. The elementary nervous system. Proc. Roy. Soc. (London)

B 140, 147-168.
Passano, L. M., and C. F. A. Pantin, 1955. Mechanical stimulation in the sea-ane-
mone CaUiactis parasitica. Proc. Roy. Soc. (London) B 143,226-238.
Pople, W., and D. W. Ewer, 1954. Studies on the motoneural physiology of Echino-
dermata. I. The pharyngeal retractor muscle of Cucmnaria. J. Exp. Biol. 31, 114-

126.

Pople, W., and D. W. Ewer, 1955. II. Circumoral conduction in Cucumaria. J. Exp.
Biol. 32, 59-69.

Prosser, C. L., 1954. Activation of a non-propagating muscle in Thyone. J. Cell, and
Cornp. Physiol. 44, 247-254.

Prosser, C. L., H. J. Curtis, and D. M. Travis, 1951. Action potentials from some
invertebrate non-striated muscles. /. Cell, and Comp. Physiol. 38, 299-319.

Prosser, C. L., and C. E. Melton, Jr., 1954. Nervous conduction in smooth muscle
of Phascolosoma proboscis retractors. /. Cell and Comp. Physiol. 44, 255-276.

Prosser, C. L., and J. Z. Young, 1937. Responses of muscles of the squid to repeti-
tive stimulation of the giant nerve fibers. Biol. Bull. 73, 237-241.

Ralston, H. J., and B. Libet, 1953. The effect of stretch on action potential of volun-
tary muscle. Amer. J. Physiol. 173, 449-455.

Ripley, S. H., and C. A. G. Wiersma, 1953. The effect of spaced stimulation of excita-
tory and inhibitory axons of the crayfish. Physiol. Comp. et Oecol. 3, 1-17.

Ross, D. M., 1955. Facilitation in sea anemones. IV. The quick response of CaUiactis
parasitica at high temperatures. /. Exp. Biol. 32, 815-821.

NEUROMUSCULAR MECHANISMS 159

Semal-van Gansen, P., 1952a. fitude du systeme musculaire chez Hydra attcnuata

(Pallas), /icac?. Roy. Belg. 38, 642-665.
Semal-van Gansen, P., 1952b. Note sur le systeme nerveux de VHvdra. Acad. Roy.

Belg. 38, 718-735.
Smith, J. E., 1950. The motor nervous system of the starfish, Astropccten irregularis

(Pennant) with special refrence to the innervation of the tube feet and ampullae.

Phil. Trans. Roy. Soc. {London) B 234, 521-558.
Twarog, B. M., 1954. Responses of a moUuscan smooth muscle to acetylcholine and

5-hydroxytryptamine. /. Cell, and Comp. Physiol. 44, 141-164.
Uexkiill, J. von, 1900. Die Physiologic des Seeigelstachels. Ztschr. f. Biol. 37, 334-

403.

Uexkiill, J. von. 1904. Die ersten Ursachen des Rhythmus in der Tierreihe. Ergeb. d.
Physiol. 3-11, 1-U.

Wiersma, C. A. G., 1941. The inhibitory nerve supply of the leg muscles of different
decapod crustaceans. /. Comp. Neurol. 74, 63-79.

Wiersma, C. A. G., 1951. On the innervation of the muscles in the leg of the lobster
Homarus vulgaris. Arch. Neerl. Zool. 8, 384-392.

Wiersma, C. A. G., 1952. Comparative physiology of invertebrate muscle. Ann. Rev.

Physiol. 14, 159-176.
Wiersma, C. A. G., 1953. Neural transmission in invertebrates. Physiol. Rev. 33,

326-355.
Wiersma, C. A. G., 1955. An analysis of the functional differences between the

contractions of the adductor muscles in the thoracic legs of the lobster, Homarus

vulgaris. Arch. Nccrl. Zool. 11, 1-13.
Wiersma, C. A. G., and C. H. Ellis, 1942. A comparative study of peripheral inhi-
bition in decapod crustaceans. /. Exp. Biol. 18, 223-236.
Wiersma, C. A. G., and A. Van Harreveld, 1938a. A comparative study of the double

motor innervation in marine crustaceans. /. Exp. Biol. 15, 18-31.
Wiersma, C. A. G., and A. Van Harreveld, 1938b. The influence of the frequency

of stimulation on the slow and the fast contraction in crustacean muscle. Physiol.

Zool. 11,75-81.

Wiersma, C. A. G., and S. H. Ripley, 1954. Further functional differences between
fast and slow contractions in certain crustacean muscles. Physiol. Comp. et Oecol.
3, 327-336.

Wiersma, C. A. G., and E. B. Wright, 1947. The nature of the action potentials of
crustacean muscles. /. Exp. Biol. 23, 205-212.

Willems, H. P. A. 1932. Uber die Herzbewegungen bei der Weinbergschnecke
(Helix pomatiaL.) Z. vcrgl. Physiol. 17, 1-100.

NEUROHORMONES OR TRANSMITTER AGENTS

John H. Welsh

Harvard University

During the past two decades studies on invertebrate nervous systems
have contributed greatly to our understanding of the basic mode of op-
eration of nerve cells. Giant nerve fibers of the squid and crustacean leg
nerve fibers have provided axons of conveniently large size for detailed
study of the phenomena associated with conduction of the nerve impulse.
Various other invertebrate preparations have furnished highly suitable
systems for studies of the transmission process. Thus observations on the
isolated heart of Venus mercenaria have told us much concerning the
importance of molecular configuration in the reaction of acetylcholine
with its receptor substance and revealed for the first time the role of
5-hydroxytryptamine as a neurohormone. Insect and crustacean neuro-
secretory systems have shown how specialized certain neurons can be,
for purposes of producing quantities of transmitter agents.

The results of these studies make it increasingly obvious that the most
characteristic feature of a neuron is its ability to release, at its termina-
tions, a substance which acts on an adjacent neuron or an effector cell,
or which is carried via the circulation to a distant part of the organism.
This paper will summarize a small portion of the evidence for the im-
portant role of neurohormones.

It has been suggested elsewhere (Welsh, 1955) that the term neuro-
hormone be defined as an organic compound produced by neurons and
released at their endings to act as a chemical messenger or hormone, either
locally or at a distance. Included among the neurohormones would be the
neurohumors such as acetylcholine, nor-adrenaline, and 5-hydroxytrypta-
mine, which often act over rather short distances, and the neurosecretory
substances such as vertebrate oxytocin and vasopressin and the products
of the neurosecretory systems of insects and crustaceans (see Scharrer,
1955), which normally act at some distance from the point of
release.

One purpose of this discussion will be to show how much the various
neurohormones have in common, especially in regard to their transport,
storage, and release. A second purpose wall be to review briefly our
knowledge of the chemical nature and distribution of the neurohormones.
One other aim will be to point out how incomplete is our understanding
of the basic mechanism of action of the individual neurohormones.

[161]

162 INVERTEBRATE PHYSIOLOGY

Synthesis, Axonal Transport, Storage, and Release
OF Neurohormones

As long ago as 1914, J. F. Gaskell observed the exciter action of adrena-
line on the pulsating blood vessels of the leech. Associating this effect with
the presence of chromaffin cells in the nervous system of the leech, Gaskell
said, "It is just possible that in the case of the leech the adrenaline passes
from the cell to the periphery by way of the motor nerve itself." While
this view of axonal transport of materials has since been expressed by a
number of workers, there has been a rather general reluctance to accept
the evidence as truly convincing. Studies of neurosecretory cells by Ernst
and Berta Scharrer and others (see Scharrer and Scharrer, 1954a,b),
provide evidence that synthesis of their products takes place largely in
the cell body. They are then carried inside the axon to the terminals,
which are often modified for storage, whence they are released on appro-
priate stimulation of the neurosecretory cell. Perhaps the most convincing
evidence for proximodistal movement of materials in axons of inverte-
brate neurons comes from cutting and ligation experiments such as those
of B. Scharrer (1952) on Lciicophaea and E. Thomsen (1954) on Calli-
phora. Other workers have done somewhat similar experiments in cer-
tain crustaceans (e.g.. Bliss and Welsh, 1952; Passano, 1953). When
neurosecretory tracts are cut or blocked, there is an accumulation of
secretory material proximal to the point of interruption. Normally, this
material with characteristic staining properties is found in far greater
concentration in the terminals of neurosecretory cells than elsewhere in
such neurons.

Those who have been interested in neurohormones have given much
thought to the question of the state in which these highly active molecules
exist while yet within the neuron. Are they bound to a carrier protein
or lipoprotein? Do they exist as an inactive precursor homologue re-
quiring chemical transformation? Is the characteristic stainable material
of a neurosecretory cell a carrier substance or the neurohormone ? Potter
(1954) has described several tinctorial types of nerve endings in the so-
called sinus gland of the blue crab, Callinectes. Does this indicate that
each type contains a different neurohormone ?

There is a recent and very interesting development that helps answer
some of these questions and that may prove to be one of the most im-
portant steps in our understanding of neurons. In the spring of 1954,
several groups of electron microscopists (De Robertis and Bennett,
Palade, Palay, Robertson) reported that axon terminals from a variety
of animals, including the crayfish and earthworm, contained collections
of mitochondria and small vesicles. De Robertis and Bennett (1955) de-

NEUROHORMONES 163

scribe these vesicles in the neuropile of the earthworm nerve cord as oval
or spherical bodies between 200 and 400 A in diameter, with a dense
periphery and a lighter center. They find them in large numbers in pre-
synaptic terminals and essentially lacking in postsynaptic processes. They
also find them apparently lying between pre- and postsynaptic mem-
branes and suggest that the vesicle may penetrate the presynaptic mem-
brane and discharge its contents or possibly enter the postsynaptic
cytoplasm.

There has been a growing body of evidence that acetylcholine, adrena-
line, and the neurohormones of the posterior pituitary are contained in
cell particles with many of the properties of mitochondria. Some of this
evidence has been summarized elsewhere (Welsh, 1955). It would now
appear that these mitochondria-like particles are the "vesicles" and
"granules" now seen in a variety of nerve endings. It remains to be de-
termined whether the "spheroid systems" seen by Passano (1953) in the
cell body, axons, and terminals of neurosecretory cells of the marsh crab,
Sesarma reticulata, are aggregates of smaller vesicles which are formed
in the cell body and move down the axon for storage in the axon terminals.

Crustacean sinus glands, long thought to be more or less typical en-
docrine organs, were first recognized to be groups of nerve endings by
Passano (1951) and Bliss (1951). These endings are modified for stor-
age and release of neurosecretory substances into the circulation. Other
examples of nerve endings in crustaceans that release substances directly
into the circulation are the pericardial organs described by Alexandro-
wicz (1953) and the "sinus plates" of the prawn, Lcandcr, described by
Knowles (1953). The pericardial organs are neuropile-like networks of
nerve fibers that spread across the openings through which blood enters
the pericardial cavity. They release a heart-accelerating substance into the
blood which is one of the means of regulating the heart rate ( Alexandro-
wicz and Carlisle, 1953). Sinus plates consist of nerve endings which
contain a chromatophore-activating substance (Knowles, 1953). Carlisle
and Knowles (1953) suggested that groups of nerve endings that do not
innervate a structure but instead store and release active materials di-
rectly into the circulation, be called "neurohaemal organs." In the insects
the corpora cardiaca are regions where most of the axons from neuro-
secretory cells in the brain terminate and release their neurosecretory
substance (Scharrer and Scharrer, 1944, 1954b). Therefore, "sinus
glands," corpora cardiaca, pericardial organs, and the vertebrate posterior
pituitary, all analogous structures, are examples of neurohaemal organs.
Neurohumors, such as acetylcholine and nor-adrenaline, are probably
stored and released in a manner similar to that described for the neuro-
secretory substances. Some of the evidence to support this view, gained

164 INVERTEBRATE PHYSIOLOGY

from observations on vertebrates, has been summarized elsewhere (Welsh,
1955).

One may picture a typical neuron, whether vertebrate or invertebrate,
as an elongated cell whose cell body is its main synthetic center for the
production of a specific transmitter agent or neurohormone. Lipid-coated
packets of the neurohormone flow with the axoplasm to axonal endings,
where they form a reserve. On the arrival of a nerve impulse, resulting
in an increase in membrane permeability, a certain number of vesicles (or
their contents) are released. The neurohormone may act at close range
to excite an adjacent neuron or efifector cell, or it may be carried in the
circulation to regulate a more lengthy process, such as the activation of
insect thoracic glands, which in turn produce a molting hormone.

No longer is it possible to encompass all neurons under the headings
"cholinergic" and "adrenergic." Instead we must be prepared to accept
a terminology that will recognize a variety of chemical transmitters or
neurohormones.

The Chemical Nature and Identification of Neurohormones

In the vertebrates, acetylcholine, adrenaline, and nor-adrenaline have
been isolated from the nervous system and chemically identified. There is
adequate physiological evidence that these three substances act as chemical
transmitters. From the vertebrate posterior pituitary gland the neuro-
secretory substances, oxytocin and vasopressin, have been isolated and
identified as polypeptides, and their detailed structure is known.

All too often in the invertebrates the only procedure used in attempt-
ing to identify a neurohormone has been to compare the physiological
effects of a nerve extract or of nerve stimulation with the effects pro-
duced by the application of a series of known candidate compounds. One
has often been forced to follow such a procedure, because of the very
small amounts of tissue available for chemical study. By use of extraction
and bioassay, with other common pharmacological procedures, acetylcho-
line has been reported to be present in representatives of most of the major
phyla of animals (see Prosser, 1946). Such methods do not always give
the true chemical identity of a substance and should, where possible, be
supplemented by other means of identification. This appears especially
desirable in connection with the identification of acetylcholine, since there
are other members of this class of compounds known to occur in the in-
vertebrates (Erspamer and Benati, 1953; Whittaker and Michaelson,
1954; Augustinsson and Grahn, 1954).

Fortunately, by means of basically simple chromatographic and electro-
phoretic procedures, it is now possible to identify many naturally occurring
organic compounds even though they are available in very small amounts.

NEUROHORMONES 165

Such methods have been appHed recently in identifying choline esters and
catechol and indole amines in certain invertebrates. Three examples will
be cited.

Ostlund (1954) has cleared up the uncertainty over the question of the
occurrence of adrenaline in insects. By means of chromatographic separa-
tion, followed by elution and bioassay, he has found adrenaline, nor-ad-
renaline, and dopamine in a variety of insects. Adrenaline is present in
least amount, while dopamine is most abundant of the three. In two lots of
whole mealworms (Tenebrio larvae) adrenaline was present in amounts of
0.021 and 0.061 fig/gm., nor-adrenaline in amounts of 1.3 and 2.2 /^g/gm.,
while dopamine values were between 10-15 /i,g/gm. in both lots. Ostlund
suggests that the presence of relatively large amounts of dopamine may
indicate that it is the precursor of nor-adrenaline. A similar origin of nor-
adrenaline has been proposed in the vertebrates.

A second example also comes from recent studies on insects. Various
earlier workers (e.g., Corteggiani and Serfaty, 1939; Mikalonis and
Brown, 1941) reported large amounts of acetylcholine in certain insect
nervous systems. Recently some question had arisen regarding the true
identity of the acetylcholine-like substance in insects. Now, however,
Augustinsson and Grahn (1954), using chromatography, have found
acetylcholine in the head of the honeybee. They also have evidence for the
presence of one or two other unidentified esters of choline.

In another invertebrate phylum, paper chromatography has been suc-
cessfully applied in the identification of a biologically active substance
present in nerve tissue. Earlier observations had suggested that 5-hy-
droxytryptamine might be a mediator of the cardiac accelerator neurons to
the heart of Venus mercenaria (Welsh, 1953). Chromatography revealed
the presence of this indole amine in nerve ganglia of Venus. We now have
evidence for its occurrence in nerve tissue of the gastropod, Busycon, the
lamellibranchs, Venus, Mactra, and Ensis, and the cephalopods, Loligo
and Octopus (two species).

At the present time it appears that the neurohumors of the invertebrates
are, perhaps, only slightly more varied than those of the vertebrates. In
addition to acetylcholine, there may be certain other choline esters or,
possibly, simpler quaternary ammonium compounds acting as acetylcho-
line-like agents. Much of the evidence for the presence and normal action
of catechol amines in the invertebrates requires confirmation. The occur-
rence of the indole amine, 5-hydroxytryptamine, which often has an ad-
renaline-like activity and which originates in chromaffin cells, as does
adrenaline, makes it imperative that the identity of the biologically active
amines in the invertebrates be given closer scrutiny.

We know next to nothing regarding the chemical nature of the neuro-

166 INVERTEBRATE PHYSIOLOGY

secretory substances of the invertebrates. Representatives of most of the
invertebrate phyla, from flatworms through protochordates, have been
shown to have certain speciaHzed neurons that give histological signs of
secretory activity (see E. Scharrer and B. Scharrer, 1954a,b ; B. Scharrer,
1955). The products of such cells help to regulate a variety of processes,
such as chromatophore activity, reproductive activity, and growth phe-
nomena, including molting of arthropods. In fact, most physiological pro-
cesses in a group like the Crustacea appear to be under primary or second-
ary control of substances released from the nerve endings constituting
the "sinus glands." It might be added, at this point, that the discovery of
organ Y in the Crustacea by Gabe (1953) and demonstration of its role
in molt control in the green crab, Carcinits (Echalier, 1954, 1955), makes
it appear probable that some postulated actions of neurosecretory sub-
Stances in the Crustacea are actually performed by a hormone from organ
Y. This gland, however, may be controlled by a neurosecretory substance.
By means of electrophoresis, three different "chromactive" (chromato-
phore-activating) substances have been obtained from insects and Crusta-
cea (Carlisle, Dupont-Raabe, and Knowles, 1955). In no case has the
chemical identity of a neurosecretory substance in an invertebrate yet been
determined, although such a step forward would appear imminent.

Modes of Action of the Neurohormones

Little is known concerning the basic mechanism of action of the neuro-
hormones in either the vertebrates or the invertebrates. There is evidence
that acetylcholine and adrenaline act at the surface of some cells to pro-
duce changes in ion permeability and in resting potential of the cell mem-
brane, but the exact series of chemical and physical events is not known.
In this section a few examples will be given of recent progress in our un-
derstanding of the modes of action of invertebrate neurohormones.

It was pointed out earlier that neurohormones appear to be of two main
types, the neurohumors, which act mostly at short range and for relatively
brief duration, and the neurosecretory substances, which may act at some
distance from the point of release and for relatively long periods of time.
Acetylcholine normally has a very short life after leaving a neuron, be-
cause of the abundance of cholinesterase waiting to hyrolyze it. The rate
of destruction of the catechol and indole amines when injected or applied
is often slower than the destruction of acetylcholine, but amine oxidase is
of common occurrence in the invertebrates (Blaschko, 1952). Acetyl-
choline and certain of the amines act as transmitters of processes where
rapid onset of effect and rapid recovery are required. Often they have
opposing actions ; nowhere is this better illustrated than in many mollusc
hearts, where acetylcholine has a depressor action while 5-hydroxytryp-

NEUROHORMONES 167

tamine is excitatory. That nature seldom sticks to a set pattern, however,
is seen in the exception to this rule provided by the heart of Mytilus cali-
fornianiis. Here, both acetylcholine and 5-hydroxytryptamine are excit-
tory, as R. B. Wait (personal commvmication ) recently observed while
working at the Marine Field Laboratories of the University of Washing-
ton.

Most moUuscan smooth muscle fibers are inconveniently small for use
of internal electrodes in recording membrane potentials. However, the
very long fibers of the anterior byssus retractor muscles of Mytilus edulis
permit measuring of a demarcation potential in a manner similar to that
long used in studying nerves. Taking advantage of this anatomical situa-
tion, Twarog (1954) finds that acetylcholine causes depolarization and a
tonic contraction of the byssus retractor muscle, while 5-hydroxytrypta-
mine relaxes tonic contractions. These observations suggest that this
muscle is doubly innervated and that opposing neurohumors mediate be-
tween nerves and muscle fibers.

By a quantitative measure of the relative activities of a wide range of
acetylcholine analogues, Welsh and Taub (1948, 1950, 1951, 1953) were
able to show a relationship between molecular structvire and biological
activity on the Venus heart. Certain deductions could be made concerning
the so-called acetylcholine receptive substance. In many respects, the
patterns of pharmacological action of acetylcholine antagonists on the
Venus heart resemble those seen in vertebrate autonomic ganglia. How-
ever, important dififerences provide further evidence that acetylcholine
receptors which have a common basic configuration may nevertheless
differ in details.

We now have considerable knowledge of the pharmacology of 5-hy-
droxytryptamine analogues on the Venus heart. The extraordinarily per-
sistent excitatory action of certain of the ergot alkaloids and of lysergic
acid diethylamide (LSD) appears due to the presence of the 5-hydroxy-
tryptamine structure in lysergic acid and the added stable nature and
stickiness of the ergot derivatives of lysergic acid (Welsh and Taub, 1948;
Welsh, unpublished).

These examples of recent attempts to learn more concerning the struc-
ture-activity relations of neurohumors and their analogues, if discussed
in greater detail, would show how favorable certain invertebrate prepara-
tions can be in such studies. As certain invertebrate nerve fibers have been
useful in gaining further insight into the conduction process, so may other
properly chosen invertebrate preparations tell us much concerning the
details of the transmission process.

Since we know more concerning neurosecretory systems and the action
of their products in insects and crustaceans, these groups will be used to

168 INVERTEBRATE PHYSIOLOGY

illustrate the type of action characterizing certain neurosecretory sub-
stances. Certain insects have groups of neurosecretory cells in the brain
whose axons end largely in the corpora cardiaca. These cells produce a
neurohormone that exerts a trophic influence on the thoracic glands. The
thoracic glands, in turn, produce a molting hormone that has recently
been isolated and crystallized, and its empirical formula determined
(Butenandt and Karlson, 1954). The molting hormone influences a num-
ber of physiological processes associated with molting and subsequent
growth. In this case we have a neurosecretory substance that may have to
exert an influence over a considerable period of time. Also it may act at
some distance from the point of release.

The neurosecretory system of decapod crustaceans is anatomically
more complex than that of the insects (Bliss and Welsh, 1952; Bliss,
Durand, and Welsh, 1954). Likewise its physiological role appears more
involved. Products of this system are employed in bringing about color
changes, retinal pigment movements, gonad development, and molting
with its many attendant processes. Chromatophores and retinal pigments
are directly controlled by neurosecretory substances carried in the blood.
Molting in crustaceans appears to be controlled by a hormone from organ
Y, the production and release of which during the intermolt period are
probably prevented by a neurosecretory substance. Again we find the
neurosecretory substances acting at a distance and over considerable
periods of time. Until we know more concerning the chemical nature of
the invertebrate neurosecretory substances we cannot hope to under-
stand in full detail their mechanisms of action. It is possible that they act
in a manner similar to that of the neurohumors but are more stable and
tend to form a more lasting complex with cellular components.

In conclusion, it may be said that the invertebrates are providing useful
information toward a better understanding of the neurohormones and
their modes of action.

Summary

The term neurohormone (or transmitter agent) is here used to designate
any organic compound that is released from neuronal endings and serves
to convey a message to other cells, tissues, or organs. One type of neuro-
hormone, which we might continue to call a neurohumor, is produced by
neurons that are in close association with other neurons or with efifectors.
Acetylcholine, adrenaline, nor-adrenaline, and 5-hydroxytryptamine are
neurohumors known to occur in certain invertebrates. A second type of
neurohormone consists of the neurosecretory materials released from
neurons which often end on blood spaces and which are sometimes highly
modified for the production, storage, and release of transmitter agents.

NEUROHORMONES 169

The neurohumors act mainly at short range and for brief duration, while
the neurosecretory materials may be carried via the circulation to distant
parts of the organism where they may act over relatively long periods
of time. The neurosecretory cells are sometimes organized in systems, as
for example in the decapod crustaceans.

There is increasing evidence in both vertebrates and invertebrates that
neurohormones are synthesized, transported, stored, and released accord-
ing to a common pattern. The invertebrates provide good examples of
grouped neuronal endings that are modified for storage of neurohormones.
Some of these grouped endings are analogous to the vertebrate neuro-
hypophysis.

Paper chromatography has been successfully applied in the identifica-
tion of the neurohumors in certain groups of invertebrates. Examples
are cited. None of the neurosecretory substances of invertebrates has yet
been chemically defined although work in this direction is now going on.

We know little concerning the basic mode of action of the neurohor-
mones in the invertebrates. Some of the doubly innervated organs, such
as the moUuscan heart, furnish favorable material for the study of certain
phases of this problem. Here, and in some other places, there is an indi-
cation that the neurohumors may act at cell surfaces to alter permeability.

REFERENCES

Alexandrovvicz, J. S., 1953. Nervous organs in the pericardial cavity of the decapod

Crustacea. /. Mar. Biol. Assoc. 31, 563-580.
Alexandrowicz, J. S., and D. B. Carlisle, 1953. Some experiments on the function of

the pericardial organs in Crustacea. /. Mar. Biol. Assoc. 32, 175-192.
Augustinsson, K.-B., and M. Grahn, 1954. The occurrence of choline esters in the

honey-bee. Acta Physiol. Scand. 32, 174-190.
Blaschko, H., 1952. Amine oxidase and amine metabolism. Pharmacol. Rev. 4, 415-458.
Bliss, Dorothy E., 1951. Metabolic effects of sinus gland or eyestalk removal in the

land crab, Gecarcinns lateralis. Anat. Rcc. Ill, 86.
Bliss, Dorothv E., and J. H. Welsh, 1952. The neurosecretory system of brachyuran

Crustacea. 5/o/. Bull. 103, 157-169.
Bliss, Dorothy E., J. B. Durand, and J. H. Welsh, 1954. Neurosecretory systems in

decapod Crustacea. Zeitschr. f. Zellforsch. 39, 520-536.
Butenandt, A., and P. Karlson, 1954. tJber die Isolierung eines Metamorphose-Hor-

mons der Insekten in Kristallisierter Form. Zeitschr. Naturforsch. 9b, 389-293.
Carlisle, D. B., and F. G. W. Knowles, 1953. Neurohaemal organs in crustaceans.

Nature 172, 404.
Carlisle, D. B., M. DuPont-Raabe, and F. G. W. Knowles, 1955. Recherches pre-

liminaires relatives a la separation et a la comparaison des substances chrom-

actives des Crustaces et des Insectes. C. R. Acad. Sci., Paris 240, 665-667.
Corteggiani, E., and A. Serfaty, 1939. Acetylcholine et cholinesterase chez les In-
sectes et les Arachnides. C. R. Soc. Biol., Paris 131, 1124-1126.
De Robertis, E. D. P., and H. S. Bennett, 1954. Submicroscopic vesicular component

in the synapse. Fed. Proc. 13, 35.

170 INVERTEBRATE PHYSIOLOGY

De Robertis, E. D. P., and H. S. Bennett, 1955. Some features of the submicroscopic

morphology of synapses in frog and earthworm. J. Biophys. Biochem. Cytol. 1,

47-58.
Echalier, G., 1954. Recherches experimentales sur le role de "I'organe Y" dans la mue

de Carcinus moenas (L). Crustaces Decapodes. C. R. Acad. Sci., Paris 238, 523-

525.
Echalier, G., 1955. Role de I'organe Y dans le determinisme de la mue de Carcinides

(Carcinus) moenas L. (Crustaces Decajodes) ; Experiences d'implantation. C. R.

Acad. Sci., Paris 240, 1581-1583.
Erspamer, V., and O. Benati, 1953. Identification of murexine as /3-[imidazolyl-(4)-

acryl-choline]. Science 117, 161-162.
Gabe, M., 1953. Sur I'existence, chez quelques Crustaces Malacostraces, d'un organe

comparable a la glande de la mue des Insectes. C. R. Acad. Set., Paris 237, 1111-

1113.
Gaskell, J. P., 1914. The chromaffine system of annelids and the relation of the system

to the contractile vascular system in the leech, Hirudo medicinalis. Philos. Trans.

5 205,153-211.
Knowles, F. G. W., 1953. Endocrine activity in the crustacean nervous system. Proc.

Roy. Soc. Lond. B. 141, 248-267.
Mikalonis, S. J., and R. H. Brown, 1941. Acetylcholine and choline-esterase in insect

central nervous system. /. Cell. Comp. Physiol. 18, 401-403.
Ostlund, E., 1954. The distribution of catechol amines in lower animals and their

effect on the heart. Acta Physiol. Scand. 31, Supp. 112.
Palade, G. E., 1954. Electron microscope observations of interneuronal and neuro-
muscular synapes. Anat. Rec. 118, 335.
Palay, S. L., 1954. Electron microscope study of the cytoplasm of neurons. Anat. Rec.

118,336.
Passano, L. M., 1951. The X organ-sinus gland neurosecretory system in crabs.

Anat. Rec. 111,86.
Passano, L. M., 1953. Neurosecretory control of molting in crabs by the X-organ

sinus gland complex. Physiol. Comp. et Oecol. 3, 155-189.
Potter, D. D., 1954. Histology of the neurosecretory system of the blue crab,

Callinectes sapidiis. Anat. Rec. 120, 716.
Prosser, C. L., 1946. The physiology of nervous systems of invertebrate animals.

Physiol. Rev. 26, 337-382.
Robertson, J. D., 1954. Electron microscope observations on a reptilian myoneural

junction. Anat. Rec. 118, 346.
Scharrer, B., 1952. Neurosecretion XL The effect of nerve section on the intercebra-

lis-cardiacum-allatum system of the insect Leucophaca maderae. Biol. Bull. 102,

261-271.
Scharrer, B., 1955. Hormones in invertebrates. The Hormones 3, 57-95. (Academic

Press, New York.)
Scharrer, B., and E. Scharrer, 1944. Neurosecretion. VI. A comparison between the

intercerebralis-cardiacum-allatum system of the insects and the hypothalamo-

hypophyseal system of the vertebrates. Biol Bull. 87, 242-251.
Scharrer, E., and B. Scharrer, 1954a. Neurosekretion. W. Mollendorff, Handbuch der

mikroskopischen Anatomic des Menschcn 6(5), 953-1066.
Scharrer, E., and B. Scharrer, 1954b. Hormones produced by neurosecretory cells.

Recent Progress in Hormone Research 10, 183-240.
Thomsen, E., 1954. Studies on the transport of neurosecretory material in Calliphora

erythrocephala by means of ligaturing experiments. /. Exp. Biol. 31, 322-330.

NEUROHORMONES 171

Twarog, Betty, 1954. Response of a molluscan smooth muscle to acetylcholine and

5-hydroxytryptamine. /. Cell. Com p. Physiol. 44, 141-164.
Welsh, J. H., 1953. Excitation of the heart of Venus mercenaria. Arch. Exp. Path. u.

Pharmakol. 219, 23-29.
Welsh, J. H., 1955. Neurohormones. The Hormones 3, 97-151. (Academic Press, New

York.)

Welsh, J. H., and Rae Taub, 1948. The action of choline and related compounds on the

heart of Venus mercenaria. Biol. Bull. 95, 346-353.
Welsh, J. H., and Rae Taub, 1950. Structure-activity relationships of acetylcholine

and quarternary ammonium ions. /. Pharmacol. Exp. Therap. 99, 334-342.
Welsh, J. H., and Rae Taub, 1951. The significance of the carbonyl group and ether

oxygen in the reaction of acetylcholine with receptor substance. /. Pharmacol.

Exp. Therap. 103, 62-73.
Welsh, J. H., and Rae Taub, 1953. The action of acetylcholine antagonists on the

heart of Venus mercenaria. Brit. J . Pharmacol. 8, 227-iii.

Whittaker, V. P., and I. A. Michaelson, 1954. Studies on urocanylcholine. Biol. Bull.
107, 304.

ENDOCRINOLOGY OF INVERTEBRATES, PARTICULARLY
OF CRUSTACEANS*

L. H. Kleinholz
Reed College

Many of the early studies in comparative endocrinology were under-
taken to seek among the invertebrates endocrine functions analogous to
those known for the vertebrates ; but few substantial contributions re-
sulted, probably because the normal physiolog}^ of a particular process
among invertebrates was inadequately established. The decade between
1920 and 1930 saw demonstrations of hormonal factors in physiological
processes of invertebrates in which the pattern and direction of research
for a number of years to come was indicated. The first of these was Kopec's
(1922) report that removal of the brain from the last instar larva of
Lyniantria resulted in a failure of pupation to occur ; when the brain was
reimplanted into the abdomen, pupation was initiated. Perkins (1928)
and Koller (1928) almost simultaneously demonstrated that the chro-
matophores of crustaceans were regulated by a blood-borne substance
originating in the eyestalks, instead of by nerves, as had been postulated
up to that time.

The lively activity set off by these early studies has resulted in such a
large and specialized body of literature in the area of invertebrate endo-
crinology that it would be impractical for one person to attempt a critical
review and survey of this field. This subject has consequently been divided
into a number of topics which can conveniently be reviewed by the speak-
ers and participants of this symposium. I shall discuss the general endo-
crinology of invertebrates, particularly of crustaceans, Bodenstein will
survey the endocrine basis of growth and development in insects, Welsh
will analyze the function of neurohormones in invertebrates, and Scheer
will discuss the metabolic aspects of molting in crustaceans.

The classical criteria in investigations of endocrine problems of verte-
brates (i.e., removal of the gland suspected of endocrine function, ob-
serving the interference with a normal physiological process as a conse-
quence of such gland removal ; obtaining a restoration of the normal pro-
cess by the implantation of such glands, or by the injection of extracts or
separated fractions of extracts prepared from these glands ; demonstra-
tion of the effective substance in the blood) can be expected to apply in

* This manuscript was prepared during the tenure of a grant, G-1395, from the
National Science Foundation.

[173]

174 INVERTEBRATE PHYSIOLOGY

similar studies among invertebrates. But application of these criteria has
not always been feasible with the invertebrates, so that conclusive proof
of endocrine function in certain physiological processes among these
groups has not always been possible. For example, one of the major ob-
stacles in the early studies dealing with endocrine regulation of chromato-
phores in crustaceans was the fact that the source of the chromatophore-
activating hormone was unknown, other than that it occurred in the eye-
stalk ; ablation of the eyestalks to remove the source of the hormone at the
same time removed the retina, which was necessary as a photoreceptor in
in the normal physiology of color change ; the limitation imposed on the
early investigators of this subject permitted at best the presentation of
strong presumptive evidence for endocrine participation in the regulation
of color change, through relatively gross deficiency and replacement experi-
ments. An additional criticism that could be directed against those early
studies which were Hmited to testing the efifects of injected tissue extracts
was that of specificity of the prepared extract, and the difficulty in dis-
tinguishing between pharmacological and physiological effects. In other
words, the extent to which tissue extracts were reproducing the normal
physiological process could not be readily determined.

Hanstrom and his collaborators (Hanstrom, 1934, 1935, 1937; Carlson,
1935 ; Sjogren, 1934) soon placed crustacean endocrinology upon a more
substantial morphological basis by describing two apparently secretory
structures that occurred in the eyestalks of a number of species of higher
crustaceans. These structures were the X-organ and the sinus gland (first
called the "blood gland" by Hanstrom). Experimental attempts at localiz-
ing the source of the chromatophorotropic hormones led Hanstrom and
his colleagues to favor the view that the sinus gland was the origin, al-
though their experiments did not conclusively exclude the X-organ from
some role in color changes.

Almost simultaneously in these same years appeared a number of
accounts which indicated that a number of physiological processes were
regulated or influenced by hormones originating in the eyestalk. In addition
to color change, which probably has been the most studied of the various
endocrine-influenced functions, those of retinal pigment migration, molt-
ing and growth, general metabolism, and some phases of reproductive
physiology have been the areas most closely investigated. But, despite the
impressive literature that has been built up in the past two decades, our
knowledge of invertebrate endocrinology is still fragmentary and incom-
plete compared to that of the vertebrates ; basic discoveries are still being
made ; basic questions are still unanswered.

Instead of a detailed survey of the literature in this area, an attempt
will be made here to examine the present trends in crustacean endocrin-

ENDOCRINOLOGY OF CRUSTACEANS 175

ology. The reader is referred to recent reviews by Brown (1952) and by
Scharrer (1952). Three of the older review papers (Kleinholz, 1942;
Brown, 1944; Panouse, 1947) examined the problems of crustacean en-
docrinology. Summaries of many of the papers given at the Symposium
on Neurosecretion at Naples in 1953 are contained in the supplement to
Vol. 24 of the Puhblicazioni della Stazione Zoologica di Napoli.

Molting Hormones

Among crustaceans, as in arthropods generally, growth is a discon-
tinuous process ; the external skeleton is shed, and a new skeleton formed ;
increase in size is restricted to the period between the casting of the old
skeleton and the secretion and hardening of the new one. Between ecdyses
may be an interval (of varying duration in different species) of lack of
growth usually designated as the intermolt interval. Molting may be
seasonal or continuous. As might be expected, the casting of the external
skeleton in molting is only an outward superficial indication of a veritable
metabolic upheaval that occurs at this time. A subsequent paper in this
symposium will discuss these metabolic features in more detail. (Scheer).

The rediscovery (Abramowitz and Abramowitz, 1938, 1940; Brown
and Cunningham, 1939; Smith, 1940; Kleinholz and Bourquin, 1941) of
observations made earlier by Zeleny ( 1905) and by INIegusar ( 1912), that
eyestalk removal shortened the intermolt period in crustaceans, was the
stimulus for the considerable number of recent studies inquiring into the
physiology of this process and the possibility of its control by hormones
originating in the eyestalk. While most of the studies cited above con-
cerned themselves chiefly with the effects of eyestalk removal on molting.
Brown and Cunningham were the first to present evidence that the accel-
erated molt of eyestalkless animals was due to the removal of a molt-in-
hibiting hormone that apparently occurred in the sinus glands ; these
authors found that, when the sinus gland was implanted into the body of
eyestalkless crayfish, the usual accelerated molt was delayed.

Drach (1939, 1944) described a series of morphological stages that
occurred in the interval between two molts of crustaceans, and in his later
study investigated the effects on molting in Leander when the eyestalks
were removed during most of these stages. Drach's observations are worth
summarizing here because they established criteria for subsequent investi-
gations of this process and indicated that it was a matter of considerable
importance at what stage during the intermolt period eyestalk removal
was done. Drach found that the normal molt cycle could be divided into a
number of stages based on morphological characteristics, which he desig-
nated by the letters A, B, C, and D. In stage A the exoskeleton was very
soft, in stage B the branchiostegites were supple ; stage C could be divided

176 INVERTEBRATE PHYSIOLOGY

into two substages characterized by increased rigidity of the branchio-
stegites and the condition of the sensory hairs along the margins of the
body; stage D was divided into 4 subdivisions, Di (divided into three
subgroups), Da, Dg, and D4, culminating in ecdysis. In a group of 340
Leander measuring 25-50 mm., sampled in October, the percentage dis-
tribution of the various stages was : A and B, 2.6^0 ; Ca, 16.7% ; Cb,
21.1% ; D/, 21.7% ; W, 14.1% ; D/", 6.4% ; D. and D3, 17.0%. Drach
found that, if the eyestalks were removed in either of the C stages or in
stage Di', a significant shortening of the interval between the operation
and the ensuing molt of these animals occurred as compared with unop-
erated control animals in comparable stages. After removal of the eye-
stalks in the subsequent Di stages or in D, or D3, molting in the operated
animals was not significantly accelerated over that in the unoperated
controls.

The regulation of the molt cycle described above could not be clearly
attributed to the sinus gland if judged by the criteria used in endocrino-
logical analyses ; it was generally agreed that eyestalk removal accelerated
molting, and that implantation of sinus glands into eyestalkless animals
delayed or prevented this accelerated molt, but the additional demonstra-
tion was lacking that removal of the sinus glands alone would efifect the
same acceleration of molting as was accomplished by removal of the eye-
stalks. A series of three independent studies published simultaneously
(BHss, 1951; Havel and Kleinholz, 1951; Passano, 1951b) showed that
careful surgical removal of the sinus glands, leaving the rest of the eye-
stalk undamaged, had no accelerating efifect on molting or on some of the
metabolic processes associated with molting. These results were in striking
contrast to the accelerated molt obtained with removal of the entire eye-
stalk. An explanation of these differences was offered by the studies of
Passano (1953), who questioned the acceptance of the specific endocrine
function of the sinus gland in accelerating molting; he postulated that,
since removal of the sinus gland itself had no effect on inducing precocious
molting, the delay in precocious molting seen after sinus-gland implanta-
tion might have been due to a nonspecific chemical effect rather than to
hormonal action. By a series of localization experiments Passano demon-
strated that removal of the X-organ induced accelerated molting in the
crabs Uca, Callinectes, and Sesarma, and that implantation of the medulla
terminalis, which includes the X-organ, prevented or delayed induced
molting.

The role of the X-organ in crustacean endocrinology had been obscure
for the twenty years following its discovery by Hanstrom. The investiga-
tions described above not only revealed the physiological activity of this
organ in the molting process, but Bliss and Passano proposed that the

ENDOCRINOLOGY OF CRUSTACEANS 177

X-organ and the sinus gland constitute an anatomical complex, connected
to each other by way of the so-called sinus-gland nerve (Bliss and Welsh,
1952; Passano, 1951a; but see also Enami, 1951, and Gabe, 1954). Accord-
ing to the hypothesis elaborated by Passano, the fibers of the sinus-gland
nerve are axons arising from the neurons that make up the X-organ, and
the secretory products of the X-organ cells are transmitted along the fibers
of this nerve to the sinus gland ; the sinus gland is considered to consist
principally of the free axon endings which have been distended by the
accumulation of the neurosecretory product.

While the studies cited above have resolved some of the apparent con-
tradictions of the earlier investigations of the hormonal basis of molting in
crustaceans, even more recent experiments have indicated that additional
hormonal mechanisms may be involved. Carlisle and Dohrn (1953) postu-
lated the existence within the eyestalk of a molt-accelerating hormone for
which they proposed the name of growth hormone or somatotrophin. Their
conclusions were drawn from experiments which consisted of the injection
of a variety of extracts into Lysmata during the winter when the normal
molting rate of this animal is low. The highest molt rates (and also mor-
tality among injected animals) were shown by Lysmata injected with
human chorionic gonadotrophin, mammalian posterior pituitary extract,
and extract prepared from the eyestalks of female Lysmata collected in the
summer. The effective molt-inducing properties of the remaining injected
substances were, in descending order: extracts prepared from the eye-
stalks of female Palaemon collected in summer ; uninjected controls ; boiled
extract prepared from eyestalks of female Lysmata collected in summer ;
extracts prepared from eyestalks of male Lysmata collected in summer ;
extracts from eyestalks of Lysmata collected in winter ; and acidulated
distilled water — but the differences among these last five groups are
probably not significant. Because of the high mortality among their ani-
mals, Carlisle and Dohrn subjected their results to a statistical probit
analysis for determining the significance of the data from their experi-
mental and control groups. The possibility exists, however, that they may
have been measuring a nonspecific chemical efi^ect of injected substances.
In subsequent studies, Carlisle (1953a) reported that the molt-accelerat-
ing effect could be obtained by feeding animals eyestalks from donor ani-
mals and by a single injection of extract (equivalent to three eyestalks).
Some unanswered questions arise concerning the dift'erences in effective-
ness of extracts prepared from the eyestalks of males as against those pre-
pared from females, and whether implants of equivalent amounts of eye-
stalk tissue might not have given more striking results than a single injec-
tion of eyestalk extract.

Two later reports by Carlisle (1953c and 1954) are in striking con-

178 INVERTEBRATE PHYSIOLOGY

trast with those of other investigators of crustacean molting. In the first
of these CarHsle found that eyestalk removal in Leandcr showed no evi-
dence of a molt-inhibiting hormone, and that the evidence pointed rather
to the existence of a molt-accelerating hormone in the eyestalk. In this
particular regard Carlisle seems to have overlooked Drach's (1944) study
with the same species, described earlier in this section. The conclusions of
these two investigators are diametrically opposed. An explanation for
these striking differences is not readily apparent ; Carlisle used only female
Leander, 55-70 mm. in total length (rostrum-telson), at a temperature of
13.5 ± 1°C, in nonrunning water, the animals being fed twice weekly, and
the average intermolt period being about 35 days ; Drach's experimental
and control animals (apparently) were of both sexes, were 25-50 mm. in
total length, were maintained at an average temperature of 14° C in the
month of October, were kept in running water, the animals were fed daily,
and the average intermolt period was about 20 days. Apart from the
possible difference of distribution between the two sexes among the ani-
mals used in the experiments of the two investigators, other differences
between the two studies may have been important ; Drach used smaller
animals than those used by Carlisle, and thus had predictably shorter
intermolt periods ; Carlisle apparently paid no attention to the stage of the
intermolt cycle in which the eyestalks of his animals were ablated, a factor
which Drach had shown was of considerable consequence in the subse-
quent molt ; for significant shortening of the intermolt period occurred
only if the eyestalks were ablated in stage C or in the first part of stage D^.
It can be seen from Drach's data that, in a population of 340 Leandcr of
the size class studied, only 60% were in these stages C and early D^ of the
intermolt cycle. An additional factor that might have contributed to these
differences may be an unrecognized artifact in the experimental conditions.
Scheer and Scheer (1954) observed no increase of molts in Leander after
eyestalk removal and interpreted ^heir results as suggesting a molt-accel-
erating factor in the eyestalks, similar to that proposed by Carlisle and
Dohrn. Carlisle ( 1954) found similarly that eyestalk removal in Carcinides
during the molting season gave no evidence of a molt-inhibiting hormone
from the eyestalks, when comparisons with control animals were m.ade.
Carlisle's observations and conclusions are at variance with those of other
investigators, but the resolution of these differences is not readily appar-
ent. Scheer (this symposium) has explained the difference in results be-
tween Drach's, Carlisle's, and his studies as possibly due to distinct races
of the Leander serratus studied (CarHsle, 1955). Most of the studies of
the effects of eyestalk removal on molting have been made on crabs and
crayfish ; the need for a wider exploration with other species of prawns
seems indicated to clarify this point.

ENDOCRINOLOGY OF CRUSTACEANS 179

A series of three recent short papers has presented evidence for another
hormone which is concerned with molt. A paper by Gabe (1953) reported
the presence in a variety of crustacean species of a structure which is named
the Y-organ, located in the antennary or maxillary metamere ; this struc-
ture has a highly secretory appearance during stage D of the molt cycle,
and Gabe therefore postulated that it might be concerned with the molt
process. Echalier (1954, 1955) undertook an experimental study of the
role of the Y-organ in the molt cycle of Carciniis, and his first reports indi-
cate that removal of the Y-organ results in a block in the development of
the usual sequence of stages of the intermolt period : 50 young animals
serving as controls had molted once within a two-months period and either
had molted a second time or were close to the second molt ; of 90 experi-
mental animals from vvhich the Y-organ had been surgically removed, 68
had not molted ; of the 22 operated animals which did undergo molt, 10
had been operated in stage Do, very close to the approaching normal molt,
but were blocked in stage C of the following intermolt period. When 3-4
pairs of Y-organs were removed from donors and were implanted into
Carcinus which had been without Y-organs, 6 survived the implantation
and two of these resumed the normal intermolt cycle and molted 40 days
after the implantations. While the number of experimental animals cited
in this second report is small, and may represent only a preliminary study,
the evidence is qualitatively such that the presence of an accelerating hor-
mone from the Y-organ may be added to the inhibiting hormone from the
X-organ-sinus gland complex to constitute the crustacean endocrine arma-
mentarium for molting.

How such a double set of hormones for the regulation of molting is
physiologically employed is still unknown. Our knowledge of the coordi-
nation of physiological mechanisms in molting, despite the wealth of mor-
phological and physiological details of the process, is still amazingly super-
ficial. The trigger mechanism to molt is unknown. We have only some
slight indication that feeding or nutritional state, daily photoperiod
(Stephens, 1955), and temperature may play a role in the process; but
how these internal and external environmental conditions, along with the
more complex metabolic phenomena of molt, may interplay with the hor-
mones involved, and how the initiation and cessation of secretion of these
hormones may be regulated are still largely obscure. We might be justified
in predicting that secretion of the molt-inhibiting and the molt-accelerating
substances must be interrelated ; for it would otherwise seem physio-
logically uneconomical to have two opposing regulatory devices partici-
pating in so slow a physiological process as molting.

The neurosecretory complex of sinus gland-X-organ will undoubtedly
be the subject of additional study, since it is apparently the basis not only

180

INVERTEBRATE PHYSIOLOGY

of molting but of most of the other known endocrine processes of crusta-
ceans. The elaboration of neurosecretory droplets, and their transmission
through the axoplasm of the fibers constituting the sinus-gland tract, prob-
ably could be studied histologically, but the stimulus and mechanism of
release of the active material from the sinus gland may be more difficult to
explain. Are the fibers of the sinus-gland tract capable of transmitting
nervous impulses, and thus of participating in the release mechanism, or is
release effected by a mechanism as yet unknown?

Retinal Pigment

The compound eyes of the higher crustaceans, which constitute the prin-
cipal photoreceptors of these animals, are composed of a large number of
ommatidial units. Each ommatidium is usually equipped with three sets
of retinal pigments, which have a somewhat varied terminology in the

—DP

"RP

^S-PP-

Fig. 1. Ommatidia from the retinas of Palaemonetes vulgaris,
showing the structure and the positions of the retinal pigments
in L, light-adapted eyes ; D, dark-adapted eyes ; E, eyes from
dark-adapted animals injected with eyestalk extract; C, cornea;
DP, distal pigment; PP, proximal pigment; RP, reflecting pig-
ment; BM, basement membrane; RH, rhabdome.

ENDOCRINOLOGY OF CRUSTACEANS 181

older literature but have in the more recent studies been designated as the
distal retinal pigment, the proximal retinal pigment, and the reflecting
retinal pigment ( Fig. 1 ) .

These retinal pigments may undergo movements in response to light
and to darkness, but the numbers of retinal pigments showing such move-
ments and the extent of their movement in response to light or darkness
may be a species characteristic. Thus, for example, in Palaemonetes all
three retinal pigments undergo such movements ; in Astacus the reflecting
pigment of the retina is fixed in position above the basement membrane,
while the distal and the proximal retinal pigments do undergo photo-
mechanical movements ; in Homarus the distal and the reflecting pigments
are fixed in position and only the proximal retinal pigment moves in re-
sponse to light and to darkness.

The movements of the distal and proximal pigments have been ex-
plained as functioning to screen the sensory component of the ommatidium,
the rhabdome, in bright light, and to uncover this rhabdome in low light
intensity and in darkness. Thus, in bright illumination the proximal pig-
ment moves above the basement membrane of the eye and the distal pig-
ment migrates centrally, so that the two sets of pigments form a collar
around the rhabdome ; in this condition light entering the ommatidium di-
rectly will stimulate the rhabdome, and light rays which enter obliquely
from adjacent ommatidia will be screened out by the collar of light-absorb-
ing black pigment, presumably melanins, which are contained in the distal
and proximal cells. In darkness or in dim light the distal and the proximal
pigments move away from the rhabdome so as to leave it relatively un-
screened, and dim light which enters the eye may pass readily through
several adjacent ommatidia to stimulate the sensory receptors of several
units, and are thus more effective in stimulation than would be the case
were the rhabdomes screened by the pigments. In those species where it
undergoes photomechanical movements, the reflecting pigment lies above
the basement membrane in darkness, and below the basement membrane
of the retina in light. These white pigment granules, which appear to be a
mixture of purines and pterins (Kleinholz and Henwood, 1953; Klein-
holz, 1955), are believed to reflect dim light that enters the eye over sev-
eral adjacent receptors, thus increasing the effectiveness of the dim light
as ,a stimulus.

The first experimental evidence that the movement of these retinal pig-
ments might be under hormonal regulation was shown by Kleinholz ( 1934,
1936), who found that injection of extracts prepared from the eyestalks of
a variety of crustaceans into dark-adapted Palaemonetes used as the test
animals brought about light adaptation of the distal and the reflecting
retinal pigments (see Fig. 1) ; the proximal pigment was apparently un-

182 INVERTEBRATE PHYSIOLOGY

affected by these extracts. These observations were confirmed by Welsh
(1939), who investigated retinal pigment movement in Cainbarus, and
found, not only that injected extract was effective on the distal pigment of
Cambarus, but that use of more concentrated extracts also brought about
light adaptation of the proximal pigment as well ; the reflecting pigment
does not move in Cambarus. The refractoriness of the proximal pigment
of Palaemonetes to injected eyestalk extract (Kleinholz, 1936) may have
been due to inadequate concentration of the eyestalk extract, as compared
with Welsh's observation on the proximal pigment of the crayfish retina,
but the possibility should not be overlooked that the differences in response
of the proximal pigments of Palaemonetes and of the crayfish may be due
to different physiological mechanisms of regulation (see below). A sub-
sequent study by Welsh (1941) pointed to the sinus gland of the eye-
stalk as the presumptive source of the eft'ective substance in Cainbarus.
Additional supporting evidence for the participation of an endocrine factor
in retinal pigment migration was presented by Kleinholz and Knowles
(1938), Kleinholz (1938), and Sandeen and Brown (1952). In the
first study it was found that movement of the distal retinal pigment in
Leander was not an all-or-nothing response to illumination, but that the
movement of this pigment could be graded between the extremes of light
and dark adaptation by varying the intensity of illumination. Sandeen and
Brown reported a similar situation for Palaemonetes. Kleinholz (1938)
found further that the amount of migration of the distal retinal pigment
in Leander was proportional to the concentration of the injected extract,
and that the graded responses observed by Kleinholz and Knowles might
therefore be mediated by the amount of hormone released into the circula-
tion. Smith (1948) extended the possibility of humoral activation of the
retinal pigment to brachyuran crustaceans when he found that injection of
extracts of sinus glands, optic ganglia, and other portions of the central
nervous system brought about varying degrees of reduction of the glow
observable at night in dark-adapted crabs ; which particular retinal pig-
ments were affected by these injections were, however, not investi-
gated.

Critical proof of endocrine regulation of crustacean retinal pigments
was incomplete. The evidence described above came predominantly from
injection experiments, and could be subject to the reservations raised
against evidence exclusively of this nature. In an attempt at localizing the
source of the retinal pigment hormone, Welsh (1941) found in Cambarus
that extracts of the sinus gland and extracts of medulla terminalis ( seat of
the X-organ) were active but those of cerebral ganglia were not. He con-
cluded that, although some activity was shown by extracts of medulla
terminalis, this might be due to residual tissue from the sinus gland or to

ENDOCRINOLOGY OF CRUSTACEANS 183

material which had escaped from the sinus gland during removal, and that
the sinus gland might therefore be the source of this principle.

Following Welsh's study came a series of reports which, while failing
to settle entirely the question of hormonal regulation with experiments in-
volving removal of the sinus glands, at the same time gave indication that
the regulation of the retinal pigments was probably more complex than
hitherto believed. Kleinholz (1948a,b) reported that sinus-gland removal
from one eye of the crayfish, followed by ablation of the other eyestalk, had
no effect on the ability of the proximal retinal pigment to adapt to light
and to darkness when such operated animals were placed under appro-
priate conditions of lighting; of 5 such animals placed in darkness, the
proximal pigment of all was in the dark-adapted position, while, of 21
similarly operated animals kept in an illuminated environment, the proxi-
mal pigment of 20 was in the typical light-adapted position and that of one
animal was irregularly light adapted. No reliable observations of the effects
on the distal pigment of the crayfish were available from Kleinholz's study
because of the damage caused to the retinal cones in the process of histo-
logical sectioning. Similar effects of sinus-gland removal were found by
Smith (1948) in his study of glow in the retinas of the crabs Hemigrapsus
and Pachygrapsus; glow, which is generally associated with the dark-
adapted retina, did not appear during daylight as a result of this operation,
while it was generally present at night, but was variable in degree. Klein-
holz (1949) found that in isolated eyestalks of the crayfish Astacus the
proximal retinal pigment could become light- or dark-adapted when placed
in moist chambers under appropriate conditions of illumination ; under the
conditions of these experiments the eyestalks were isolated from the cen-
tral nervous system (except for the optic ganglia contained within the
eyestalk) and from the circulatory system. The result of these three studies
do not support the view that a hormone from the sinus gland is solely re-
sponsible for causing the proximal pigment of the crayfish (and of the
crab?) to move into the light-adapted position and indicate the possibility
that the proximal retinal pigment cells of the crayfish retina may respond to
light and to darkness as independent effectors. Knowles ( 1950) found that
the proximal pigment cells of Leander from which the sinus glands had been
removed respond in normal fashion to light and to darkness ; he, too, in-
clined toward the interpretation that these proximal retinal pigment cells
were behaving as independent eff'ectors. In none of the studies cited
has the independent-eft'ector hypothesis been critically demonstrated to
the exclusion of other possibilities. One such possibility which has been
mentioned (Kleinholz, 1948a), and which may be indicated in the experi-
ment of Knowles, is a source of retinal pigment hormone outside of the
eyestalk. Conclusive evidence for such sources, beyond that resulting only

184 INVERTEBRATE PHYSIOLOGY

from injection experiments, would present serious limitations to the
classical methodology of demonstrating endocrine regulation of these
effectors.

The evidence for the sinus gland as the source of a hormone causing
light adaptation of the distal retinal pigment in the Natantia seems more
favorable. In addition to the evidence from the injection experiments of
Kleinholz (1936, 1938), Knowles (1950) found that the distal retinal
pigment of Leander, from w^hich sinus glands had been removed, attained
maximal dark adaptation and w^as not affected by changes of illumination ;
these results, however, were not unequivocal, for in a few such operated
individuals the distal retinal pigment underwent a slight proximal migra-
tion (toward light adaptation). The latter responses might have been due
to slight injury to the optic ganglia, caused during removal of the sinus
glands, a condition which Smith ( 1948) found to result in varying degrees
of light adaptation in the retina of crabs, or they might have been due to
some of the other physiological possibilities mentioned above.

In recent years it has been proposed (Brown, Fingerman, and Hines,
1952; Brown, Hines, and Fingerman, 1952; Brown, Webb, and Sandeen,
1953) that, in addition to a hormone which brings about light adaptation
of the distal retinal pigment of Palaemonetcs, an antagonistic hormone,
which causes dark adaptation of the distal retinal pigment, may be pres-
ent. The basis for this hypothesis lies in two kinds of observations ;
Palaemonetes from which one eyestalk has been removed show less light
adaptation than normal animals, while similar one-eyed animals show the
same dark adaptation as normal animals ; the other kind of observation
resulted from studies of the kinetics of light and dark adaptation of the
distal retinal pigment of animals which had been successively dark-adapted
for varying periods, given light stimuli of different durations, and then re-
turned to darkness. It was found that the ensuing variations in the rate
of readaptation to darkness could be explained in terms of a hormone that
causes dark adaptation of the distal retinal pigment cells. By way of experi-
mental test of this hypothesis, these authors studied the effects on the
kinetics of dark adaptation of injecting — into previously dark-adapted
animals given a light stimulus and then returned to darkness — extracts of
eyestalks, of central nervous system, of tritocerebral commissures, and of
sea water. It was found that the rate of subsequent readaptation to dark-
ness was greater after injection of the eyestalk extract and of the central-
nervous-system extract than with the sea-water control, and that Palae-
monetes injected with extracts of tritocerebral commissure showed less
light adaptation than the controls. These authors had no success in in-
ducing dark adaptation of the distal retinal pigment in light-adapted ani-
mals, and point out that the only condition under which it was possible in

ENDOCRINOLOGY OF CRUSTACEANS 185

their study to demonstrate the presence of a dark-adapting hormone was
under the environmental condition of complete darkness.

ClIROMATOPIIORES

The first systematic study of a physiological process in crustaceans
which gave evidence of endocrine regulation was that of color change, the
effectors in this case being the pigment cells located within the hypodermis
and on the deeper-lying organs of the body. The readiness with which these
chromatophores may be discerned depends, among various crustacean
species, on the degree of transparency of the overlying tissues and exo-
skeleton. The chromatophoral systems of different species may range
through a complexity of colors, morphology, and distribution over the
body surface ; the physiological responses are effected by centrifugal or
centripetal streaming of the chromatophoral cytoplasm in which the pig-
ment granules are carried, resulting in the dark or colored phase of the
animal when the pigment is dispersed through the interlacing processes of
the chromatophores, and in the light phase of the animal's color change
when the pigment granules are withdrawn from the chromatophoral pro-
cesses and are concentrated near the center of the cell body ; intermediate
conditions between the extremes of response are also possible. The colored
appearance of an animal may be determined by the abundance and distri-
bution of a particular type of chromatophore ; where there are physio-
logically responsive chromatophores containing an assortment of pigments
in the system of an animal, the animal may be able to assume a variety of
colors, depending on which chromatophore components have their pigment
granules dispersed and which concentrated. In some species color changes
may occur in adaptation to the color of the background, in others to change
in light intensity. In investigations of color change, the most obvious com-
ponent of the variegated chromatophore system has been the one usually
studied, while those chromatophores which are less abundant have been
relatively slighted.

The studies establishing an endocrine basis for color change in crus-
taceans, the subsequent observations on the diversity of response of differ-
ent chromatophore systems in a variety of crustacean species, and some of
the attendant problems that arose from these studies have been amply
leviewed by the authors mentioned in the introductory section of this
paper. In recent years a large proportion of the studies of color change
has been concerned with the demonstration by extraction and injection
methods of the presence of different active principles from the central
nervous system of crustaceans. Many of these studies were prompted by
an effort to resolve one of the early problems in this field ; i.e., whether
the regulation of the various types of chromatophores in a color-changing

186 INVERTEBRATE PHYSIOLOGY

crustacean could be explained on the basis of one hormone or was due to
a number of hormones. In brief summary, Brown and a number of collabo-
rators (see Brown, 1952, for specific citations) have reported three chro-
matophorotropic principles from the sinus gland of crustaceans, the dis-
tinction being based to some extent on dififerent solubilities in ethanol and
on the response of particular chromatophore types in different crustacean
species : ( 1 ) a principle which causes concentration of the red chromato-
phores of Palaemoneies, (2) a principle which causes dispersion of the
black chromatophores of Uca, and (3) a principle which causes the black
pigment in the chromatophores of the telson and uropod of Crago to be-
come concentrated. From the central nervous system of a variety of crus-
taceans two additional principles were adduced: (4) one which disperses
the melanophore pigment of Crago both in the body and in the "tail," and
(5) one which concentrates the melanophore pigment of the body of Crago
but not of the "tail" of Crago.

Most of the evidence presented in attempting to settle the problem of
localization of the chromatophorotropic hormones of the eyestalk and of
the central nervous system has consisted of histological demonstration of
apparent neurosecretory structures in the central nervous system and of
injection experiments involving extracts prepared from the central ner-
vous system. The recent explanations of the part played in molting and its
associated metabolic processes by the X-organ were facilitated by de-
ficiency experiments involving removal of the sinus gland; the advisa-
bility of similar deficiency experiments in studies of color change would
seem apparent.

As long ago as 1940 Brown reported that most of the chromatophoro-
tropic effect of whole eyestalks of Palaemoneies and Uca was contained in
the sinus glands. A study by Brown, Ederstrom, and Scudamore (1939)
neatly complemented the early injection experiments by examining the
effects of surgical removal of the sinus glands from the eyestalk in Palae-
moneies on the animal's subsequent color behavior. These authors found
that the glands could be removed without apparent serious disturbance to
the rest of the eyestalk. In such Palaemoneies it is to be expected that the
erythrophores will become dispersed and the animal will remain in the
dark phase of its color range if the sinus gland is the chief source of a
chromatophore-concentrating principle; in addition the animal will be
unable to adapt its erythrophores to an illuminated white background.
When 16 such operated animals were tested on an illuminated white back-
ground, the erythrophore responses were such that the animals fell into
three groups : 5 showed no concentration of the red pigment, remaining
permanently dark ; 4 showed a weak concentration of the red pigment ;
and 7 underwent strong concentration of the erythrophores. Thus, only

ENDOCRINOLOGY OF CRUSTACEANS 187

about one-third of the operated animals showed the expected chromato-
phore behavior. But, to test further the completeness of sinus-gland re-
moval, eyestalks from each of these three groups were extracted and
the chromatophorotropic activity determined quantitatively by injection
into Palaemonetes and Uca employed as test animals. A direct rela-
tionship was found between the activity of these eyestalk extracts on the
test animals and the degree of response to white background shown by the
erythrophores of the operated animals whose eyestalks were used in
making the extracts. Such physiological testing of the efficacy of complete
sinus-gland removal is highly desirable as a critical approach to the ques-
tion of localization of the origin of the chromatophorotropic hormone and
indicates that there is no erythrophore response to illuminated white back-
ground on total removal of the sinus gland.

Somewhat similar results were obtained by Panouse ( 1946) with sur-
gical removal of the sinus gland in Leander, although the proportion of
successfully operated animals was slightly higher than that obtained by
Brown and his co-workers ; in Panouse's study 10 of 20 sinus-glandless
prawns failed to show any white-background response. Here the success
of the surgery was checked by inspection after the subsequent molt, when
the scar at the site of the operation became transparent like the rest of the
body ; if a fragment of the sinus gland had escaped removal, it was readily
visible and its volume could be estimated. Knowles (1950) in his study
of the control of retinal pigment migration in Leander also used the im-
mobility of the dark chromatophores of animals kept for 10 days under
various conditions of illumination and background as an indication of
successful sinus-gland removal ; about one-third of the operations under-
taken were successful.

The observations cited above complement the evidence from injection
experiments in pointing to the sinus gland as the source of chromatopho-
rotropic hormones. Evidence of a like nature for the origin of such hor-
mones in the central nervous system would be technically much more
difficult, if not impossible, to obtain. But even the existing evidence pre-
sents us with certain anomalies, at least with regard to the question as to
whether secretion of chromatophorotropic hormone from the central ner-
vous system can occur as a normal physiological process in color change.
If surgical removal of the sinus glands can be accomplished without dam-
age to the photoreceptor apparatus of the eyestalk, and if the pathways for
visual reflexes remain intact after such surgery, physiological secretion
by the central nervous system should enable such animals to adapt to an
illuminated white background ; the above investigators do not report such
observations except those white-background responses which they explain
as due to incomplete removal of the sinus glands. Thus, while there is some

188 INVERTEBRATE PHYSIOLOGY

evidence, predominantly from injection experiments but also, to a lesser
extent, from observations of chromatophoral behavior under other experi-
mental conditions, for sources of hormone outside the eyestalk, there is in-
sufficient evidence for the participation of these extra-eyestalk sources of
hormone in the normal physiology of color change. Panouse (1946, 1947)
has expressed some reservations as to the specificity of extracts of the
central nervous system in the activation of chromatophores.

Carlisle, Dupont-Raabe, and Knowles (1955), using the methods of
paper electrophoresis, have undertaken separation of substances which
affect various components of the chromatophore system from extracts of
sinus glands and postcommissural organs of Leander. From extracts of
sinus glands and of postcommissural organs they obtained a substance A
which on injection caused concentration of the red and yellow pigments
of the large and the small chromatophores. Substance B, having different
migration properties and separable only from postcommissural organs,
upon injection, concentrated the large red chromatophores of the body
but expanded the red pigment of the small chromatophores of the body
and of the uropods. Other substances showing more marked electropho-
retic mobilities than A or B, but acting only on a single pigment, were
separable from postcommissural extracts when the latter were allowed to
remain a certain time at laboratory temperature (time and temperatures
not stated) ; the concentration of A and B in these extracts was diminished.
It is also reported that fresh extracts of the X-organ have no effect on the
chromatophores, but that the same extracts after boiling cause concentra-
tion of the dark pigments. These experiments indicate the possibility of
substances which by various treatments can be altered or broken down
into chromatophorotropically active components.

Reproductive Systems

The question of the existence of sex hormones in crustaceans which
maintain secondary sex characters has been under discussion by zoologists
for many years. The basis for this discussion has been the observable
change in the secondary sex characters of a number of species, either asso-
ciated with parasitic castration or induced experimentally by x-ray or
radium irradiation which destroyed the gonads. This subject has been
reviewed by a number of the authors mentioned in the introductory sec-
tion ; there seems to be general agreement that evidence for the secretion
of hormones from the gonads to maintain secondary sex characters is not
completely satisfactory.

An attempt at experimental approach to this problem along the lines of
conventional endocrine surgery by Takewaki and Nakamura (1944) has
not been particularly fruitful in presenting evidence in favor of secretion

ENDOCRINOLOGY OF CRUSTACEANS 189

of hormones from the gonad. These authors were able to perform surgical
castration upon male and female isopods, and found no consequent modifi-
cation of the permanent sex characters. Oostegites, which constitute a
brood pouch for developing young, formed in more than 90% of castrated
females and are thus independent of the gonad. Since earlier investigators
had reported that radiation castration was followed by failure of the
oostegites to develop (hence the argument for a hormone from the ovary
to maintain the oostegites), it seems unjustifiable from the results of
Takewaki and Nakamura to explain such changes in terms of destruction
of the gonad.

Reinhard (1950) proposed that differentiation of the inner rami of the
pleopods of Callinectes, similarly associated with care of the developing
young, was under the influence of an ovarian hormone. It may be unwise
to extend the conclusions from one order of crustaceans, the isopods, to
another, the decapods ; but, in the absence of any more positive evidence
for ovarian hormones, the results of the Japanese workers remain without
substantial challenge.

The cement glands in the female crayfish are a secondary sex character
for which the possibility of nonovarian endocrine control factors have been
indicated by Stephens ( 1953). This author reports a factor in the eyestalks
which inhibits development of the cement glands in the mature female ; a
possible role of neurosecretion from the central nervous system in regu-
lating these structures is suggested by implantation experiments, but
Stephens believes further experimental verification is desirable.

Studies reporting endocrine effects upon the gonads themselves seem
to be based on more substantial evidence. Panouse ( 1946) was the first to
report a marked effect of eyestalk removal in Leander serratus on the
ovaries. He found that by 45 days after eyestalk removal the wet weight
of the ovary was 13 times that of unoperated control animals, and the dry
weight more than 30 times that of controls. These results occurred in ex-
periments execvited well before the normal breeding season ; the egg lay-
ing which follows such ablation experiments appeared normal. To charac-
terize these results more adequately as a possible endocrine effect, surgical
removal of the sinus glands was done on one group of animals and was
found to give similar results, but not as marked as those obtained with eye-
stalk removal, in explanation of which Panouse suggested the possibility
of incomplete sinus-gland removal. Reciprocal experiments involving
weekly implantations of 2 sinus glands into the abdomens of animals with-
out eyestalks not only prevented the rapid growth of the ovary, but actually
depressed ovarian weight below that of normal unoperated animals. The
evidence from these experiments suggests an inhibitory hormone from
the sinus gland, but the question arises as to whether the inhibitory effect

190 INVERTEBRATE PHYSIOLOGY

is on the ovary itself or on some other organ which normally secretes an
accelerating" or gonadotropic hormone. Experimental procedures to an-
swer this question faced technical difficulties ; but, in the light of some re-
ports of neurosecretory structures associated with the cerebral ganglia,
Panouse undertook implantations of such tissues into a small number of
animals. No significant effect on ovarian growth was detectable, but
Panouse himself recognized that the small number of animals (26) in-
volved and the possibility of insufficient implantations make repetition of
this experiment desirable.

Panouse's results were confirmed on a number of other crustaceans by
Brown and Jones (1947, 1948) and by Carlisle (1953b). A study of the
reproductive cycle in the female crayfish by Stephens (1952) postulates
the participation of a number of hormones, two from the sinus gland and
two from the cerebral ganglia, but these proposals are still speculative and
require experimental verification.

On the other hand, results reported by Arvy, Echalier, and Gabe ( 1954)
present more adequate evidence for an additional source of an endocrine
substance which is gonadotropic in function. These authors find that bi-
lateral removal of the Y-organ (described above in the section on molting)
in sexually immature Carcinides results in ovaries in which oogonia and
mitoses are rare, follicle cells and vitellogenesis of the oocytes are reduced,
and cytolysis of oocytes has occurred. Comparable anomalies appear in
similarly operated males. Removal of the Y-organ in sexually mature males
and females has no observable effect, such operated animals not being
different from normal controls given blank operations. At this stage the
authors are unable to decide whether the effect on the gonads is a general
metabolic one, because of the arrest of the molting rhythm, or whether it
is a specific gonadotropic effect limited to gametogenesis. The Y-organ
may be a structure which is physiologically in balance with the sinus gland
in regulating ovarian growth, a relationship that would be worth closer
examination.

Other Activities and Other Phyla

In very few animal phyla other than the arthropods have endocrine pro-
cesses been demonstrated in much detail. For the most part, studies among
the other phyla consist of scattered observations, which have been sum-
marized in some of the early reviews, particularly that of Hanstrom
( 1939). On the other hand, there have been a number of studies of neuro-
humoral activity among invertebrates, many of which are reviewed in this
symposium by Welsh. Recent studies of the distribution of catechol amines
in invertebrates (Ostlund, 1954), reviews of the pharmacology of indo-
lealkylamines, particularly of 5-hydroxytryptamine (Erspamer, 1954),

ENDOCRINOLOGY OF CRUSTACEANS 191

and studies of activity of crustacean pericardial organs on the crustacean
heart (Alexandrowicz and Cadisle, 1953) all give indications of endocrine
characteristics and the possibility of function as a neurohumor in normal
physiological processes. Further studies will be able to distinguish between
physiological and pharmacological effects of these substances.

Laviolette (1950) has verified reports by earlier investigators of hor-
monal influence of the gonad on secondary sexual structures of the genital
tract in gastropod mollusks. Fragments of the hermaphroditic gonad of
Mesarion, from which all spermatozoa were absent and containing only
fully grown oocytes, were implanted into young Arion or into individuals
with infantile genital tracts. The hosts were sacrificed one month later and
the genital tract accessories (albumen gland, ovospermiduct, copulatory
pouch, and the calcareous glands of the neck of the genital atrium) were
found considerably modified in comparison with the controls : while the
host had not increased appreciably in size, the genital tract had trebled ;
the genital tract in controls of the same age remained infantile. When a
fragment of ovospermiduct from a young Arion was recovered from the
general cavity of an adult Koheltia in which it had been implanted for five
weeks, the fragment had appreciably increased in size and had differen-
tiated histologically toward the type of structure found in the adult. Simi-
lar results were obtained on homotransplantation of the albumen gland be-
tween two individuals of Limax of different ages. Castration was success-
fully performed on adult Limax at an age of 10 months, when the genital
tract is fully developed. Three months afterward, appreciable regression
of different parts of the genital tract, particularly of the albumen gland,
had occurred ; regression of the penis was less marked.

One additional area of possible endocrine function should be mentioned
here, that of the tunicates among which Carlisle has made some observa-
tions. An old report of Hogg (1937) on the occurrence of a gonadotropin
in the neural gland complex of tunicates when tested against the mouse
had not been widely accepted, probably because of the small number of
test animals involved ; Carlisle ( 1950) repeated these and additional tests
on male toads, also with small numbers of test animals, with results indi-
cating gonadotropic activity. More appropriately related to tunicate physi-
ology is an hypothesis proposed by Carlisle (1951) that the gonadotropic
hormone from the tunicate neural gland constitutes the afferent portion
of a gametokinetic reflex. The basis of this hypothesis is that injection of
human chorionic gonadotropin into Ciona and into Phallusia caused re-
lease of gametes from the gonads : injection of extract prepared from the
neural complex of 1,000 Ciona into 9 Phallusia provoked a similar response
in 6 out of 9 individuals. From the results of additional experiments involv-
ing the nervous system of tunicates Carlisle proposed that a hormone from

192 INVERTEBRATE PHYSIOLOGY

the neural gland passes to the central nervous system (the neural ganglion)
by a nonvascular route, and that the efferent pathway of the gametokinetic
reflex consists of a nervous discharge along nerves to the gonads, effecting
release of gametes. The high concentration of the Ciona extract (1,000
glands) injected into 9 animals may raise some question as to whether this
is a true physiological hormone effect ; the nervous efferent portion of the
proposed gametokinetic reflex could be tested by electrical stimulation of
the effector nerve from the ganglion to the region of the gonad.

Summary

Hormonal factors have been indicated in the regulation of some physi-
ological processes among crustaceans : molting and associated metabolic
phenomena ; retinal pigment migration ; chromatophoral behavior in color
change ; and some aspects of reproductive physiology. The application of
critical standards of experimental methodology^ reveals that, despite an
abundant literature in the areas mentioned, conclusive proof of endocrine
intervention in some of these processes is lacking.

Molting among crabs and crayfish seems to be influenced by a molt-
inhibiting hormone originating in the sinus gland-X-organ complex, al-
though some investigators, studying Mediterranean prawns, fail to reveal
evidence for a molt-inhibiting hormone and postulate instead a molt-
accelerating hormone. On the other hand, preliminary evidence for the
origin of a molt-accelerating hormone in a newly described structure, the
Y-organ, has been presented for the crab, Carcinus. It is possible that
these two structures may be interrelated in controlling the molt cycle of
crustaceans.

Most of the studies of endocrine factors in the regulation of the retinal
pigments of crustaceans have been made on crayfish and prawns. For the
distal retinal pigment, there is direct evidence, although not unequivocal,
that movement of this pigment into the light-adapted position in prawns
is mediated by a hormone from the sinus gland ; studies of the kinetics of
Hglit and of dark adaptation in Palacmonetes point to the possibility of a
hormone that causes dark adaptation of the distal retinal pigment. Hor-
monal regulation of the proximal and the reflecting retinal pigments has
been less certainly demonstrated.

Deficiency experiments and injection experiments point to the sinus
gland as the source of chromatophorotropins. The evidence for a source of
chromatophore-activating hormone outside the crustacean eyestalk is less
satisfactory, being adduced predominantly from histological studies and
the injection of extracts of central-nervous-system tissues; technical diffi-
culties make deficiency experiments less feasible than with the sharply
circumscribed sinus gland.

ENDOCRINOLOGY OF CRUSTACEANS 193

No convincing evidence has been presented for the maintenance in
crustaceans of secondary sex characters by hormones from the gonads,
although it has been indicated that the cement glands of the female cray-
fish may be regulated by nonovarian hormones. Two apparent sources
of gonadotropic hormones have been described : one originating in the
sinus gland is inhibitory, since removal of the sinus glands leads to hyper-
trophy of the ovary; the other, whose specific function is less certain,
originates in the Y-organ ; removal of the Y-organ in sexually immature
Carcinides results in degenerative changes in the ovaries and testes, but
removal in mature animals has no observable effect.

A few studies are described where endocrine factors have been proposed
for secondary sex characters in gastropod mollusks, and as part of a
gametokinetic reflex in tunicates,

REFERENCES

Abramowitz, A. A., and R. K. Abramowitz, 1938. On the specificity and related prop-
erties of the crustacean chromatophorotropic hormone. Biol. Bull. 74, 278-296.
Abramowitz, R. K., and A. A. Abramowitz, 1940. Molting, growth and survival after

eyestalk removal in Uca pugthfor. Biol. Bull. 78, 179-188.
Alexandrowicz, J. S., and D. B. Carlisle, 1953. Some experiments on the function

of the pericardial organs in Crustacea. /. Mar. Biol. Assoc. U. K. 32, 175-192.
Arvy, L., G. Echalier, and M. Gabe, 1954. Modifications de la gonade de Carcinides

(Carcinus) moenas L., [Crustace decapode], apres ablation bilaterale de I'organe

Y. C. R. Acad. Sci., Paris 239, 1853-1855.
Bliss, D. E., 1951. Metabolic efifects of sinus gland or eyestalk removal in the land

crab, Gccarcinus lateralis. Anat. Rec. Ill, 502-503.
Bliss, D. E., and J. H. Welsh, 1952. The neurosecretory system of brachyuran Crus-
tacea. Biol. Bull. 103, 157-169.
Brown, F. A., Jr., 1940. The crustacean sinus gland and chromatophore activation.

Physiol. Zool. 13, 343-355.
Brown, F. A., Jr., 1944. Hormones in the Crustacea. Quart. Rev. Biol. 19, 32-46, 118-

143.
Brown, F. A., Jr., 1952. Hormones in Crustaceans. In The Actions of Hormones in

Plants and Invertebrates. New York.
Brown, F. A., Jr., and O. Cunningham, 1939. Influence of the sinus gland of crusta-
ceans on normal viability and ecdysis. Biol. Bull. 77, 104-114.
Brown, F. A., Jr., H. E. Ederstrom, and H. H. Scudamore, 1939. Sinus-glandectomy

in crustaceans without blinding. Anat. Rec. 75, suppl. 129-130.
Brown, F. A., Jr., M. N. Hines, and M. Fingerman, 1952. Hormonal regulation of the

distal retinal pigment of Palaemonetes. Biol. Bull. 102, 212-225.
Brown, F. A., Jr., M. Fingerman, and M. N. Hines, 1952. Alterations in the capacity

for light and dark adaptation of the distal retinal pigment of Palaemonetes. Physiol.

Zool. 25, 230-239.
Brown, F. A., Jr., and G. M. Jones, 1947. Hormonal inhibition of ovarian growth in

the crayfish, Cambarus. Anat. Rec. 99, 657.
Brown, F. A., Jr., and G. M. Jones, 1948. Ovarian inhibition by a suius gland principle

in the fiddler crab. Biol. Bull. 96, 228-232.

194 INVERTEBRATE PHYSIOLOGY

Brown, F. A., Jr., H. M. Webb, and M. Sandeen, 1953. Differential production of two
retinal pigment hormones in Palaenwnefes by light flashes. /. Cell. Coinp. Physiol.
41, 123-144.

Carlisle, D. B., 1950. Gonadotrophin from the neural region of ascidians. Nature
166, 737.

Carlisle, D. B., 1951. On the hormonal and neural control of the release of gametes in
ascidians. /. Exp. Biol. 28, 463-472.

Carlisle, D. B., 1953a. Studies on Lysmafa seticaudata Risso (Crustacea Decapoda).
III. On the activity of the molt-accelerating principle when administered by the
oral route. IV. On the site of origin of the molt-accelerating principle — experi-
mental evidence. Pubbl. Stas. Zool. Napoli 24, 279-292.

Carlisle, D. B., 1953b. Studies on Lysmata seticaudata Risso (Crustacea Decapoda).
V. The ovarian inhibiting hormone and the hormonal inhibition of sex reversal.
Pubbl. Stas. Zool. Napoli 24, 355-372.

Carlisle, D. B., 19S3c. Molting hormones in Leander (Crustacea Decapoda). /. Mar.
Biol. Assoc. U. K. 32, 289-296.

Carlisle, D. B., 1954. On the hormonal inhibition of molting in decapod Crustacea.
/. Mar. Biol. A.^soc. U. K. 33, 61-63.

Carlisle, D. B., 1955. Local variations in the color pattern of the prawn Leander ser-
rattis Pennant. /. Mar. Biol. Assoc. U. K. 34, 559-563.

Carlisle, D. B., and P. F. R. Dohrn, 1953. Studies on Lysmata seticaudata Risso
(Crustacea Decapoda). II. Experimental evidence for a growth- and molt-
accelerating factor obtainable from eyestalks. Pubbl. Stas. Zool. Napoli 24, 69-83.

Carlisle, D. B., M. Dupont-Raabe, and F. G. W. Knowles, 1935. Recherches prelim-
inaires relatives a la separation et a la comparaison des substances chromactives des
Crustaces et des Insectes. C. R. Acad. Sci., Paris 240, 665-667.

Carlson, S. P., 1935. The color changes in Uca pugilator. Proc. Nat. Acad. Sci. 21,
549-551.

Drach, P., 1939. Mue et cycle d'intermue chez les crustaces decapodes. Ann Inst, ocean-
ogr. Monaco 19,103-391.

Drach, P., 1944. fitude preliminaire sur le cycle d'intermue et son conditionnement
hormonal chez Leander serratus (Pennant). Bull Biol. France Belg. 78, 40-62.

Echalier, G., 1954. Recherches experimentales sur le role de "I'organe Y" dans la mue
de Carcinus moenas (L.) Crustace Decapode. C. R. Acad. Sci., Paris 238, 523-525.

Echalier, G., 1955. Role de I'organe Y dans la determinisme de la mue de Carcinides
(Carcinus) moenas L. (Crustaces Decapodes) ; Experiences d'implantation. C.
R. Acad. Sci., Paris 240, 1581-1583.

Enami, M., 1951. The sources and activities of two chromatophorotropic hormones in
crabs of the genus Sesarma. II. Histology of the incretory elements. Biol. Bull.
101, 241-258.

Erspamer, V., 1954. Pharmacology of indolealkylamines. Pharmacol. Rev. 6, 425-487.

Gabe, M., 1953. Sur I'existence chez quelques crustaces malacostraces d'un organ com-
parable a la glande de la mue des insectes. C. R. Acad. Sci., Paris 237, 1111-1113.

Gabe, M., 1954. La neurosecretion chez les invertebres. Ann. Biol. 30, 5-62.

Hanstrom, B., 1934. Neue Untersuchungen iiber Sinnesorgane und Nervensystem der
Crustaceen. III. Zool. Jahrb. (Abt. Auat.) 58, 101-144.

Hanstrom, B., 1935. Preliminary report on the probable connection between the blood
gland and the chromatophore activator in decapod crustaceans. Proc. Nat. Acad.
Sci. 21, 584-585.

Hanstrom, B., 1937. Die Sinusdriise und der hormonal bedingte Farbwechsel der
Crustaceen. K. svenska Vetensk. Akad. Handl. IIL 16, 1-99.

ENDOCRINOLOGY OF CRUSTACEANS 195

Hanstrom, B., 1939. Hormones in Invertebrates. Oxford.

Havel, V. J., and L. H. Kleinholz, 1951. Effect of seasonal variation, sinus gland re-
moval, and eyestalk removal on concentration of blood calcium in Astacus. Anat.
Rec. 111,571-572.

Hogg, B. M., 1937. Subneural gland of ascidian (Polycarpa tecta) : an ovarian stim-
ulating action in immature mice. Proc. Soc. E.vptl. Biol. Med. 35, 616-618.

Kleinholz, L. H., 1934. Eye-stalk hormone and the movement of the distal retinal
pigment in Palacmonetes. Proc. Nat. Acad. Sci. 20, 659-661.

Kleinholz, L. H., 1936. Crustacean eye-stalk hormone and retinal pigment migration.
Biol. Bull. 70, 159-184.

Kleinholz, L. H., 1938. Studies in the pigmentary system of Crustacea. IV. The uni-
tary versus the multiple hormone hypothesis of control. Biol. Bull. 75, 510-532.

Kleinholz, L. H., 1942. Hormones in Crustacea. Biol. Rev. 17, 91-119.

Kleinholz, L. H., 1948a. Migrations of the retinal pigments and their regulation by
the sinus gland. Conference scientifiqiie internationale sur I'endocrinologie des
arthropodes. Bull. Biol. France Belg. suppl. 33, 127-138.

Kleinholz, L. H., 1948b. Factors controlling the migration of the proximal pigment
of the crustacean retina. Anat. Rec. 101(4), 15.

Kleinholz, L. H., 1949. Responses of the proximal retinal pigment of the isolated
crustacean eyestalk to light and to darkness. Proc. Nat. Acad. Sci. 35, 215-218.

Kleinholz, L. H., 1955. The nature of the reflecting pigment in the arthropod eye.
Biol. Bull. 109(3), 362.

Kleinholz, L. H., and E. Bourquin, 1941. Effects of eye-stalk removal on decapod
crustaceans. Proc. Nat. Acad. Sci. 27, 145-149.

Kleinholz, L. H., and W. Henwood, 1953. The nature of the retinal reflecting pigment
in macruran crustaceans. Anat. Rec. 117, 6Z7.

Kleinholz, L. H., and F. G. W. Knowles, 1938. Studies in the pigmentary system of
Crustacea. III. Light intensity and the position of the distal retinal pigment in
Leander adspersus. Biol. Bull. 75, 266-273.

Knowles, F. G. W., 1950. The control of retinal pigment migration in Leander ser-
ratics. Biol. Bull. 98, 66-80.

Koller, G., 1928. Versuche iiber die inkretorischen Vorgange beim Garneelenfarb-
wechsel. Z. vergl. Physiol. 8, 601-612.

Kopec, S., 1922. Studies on the necessity of the brain for the inception of insect
metamorphosis. Biol. Bull. 42, 323-342.

Laviolette, P., 1950. Role de la gonade dans la morphogenese du tractus genital chez
quelques Mollusques Limacidae et Arionidae. C. R. Acad. Sci., Paris 231, 1567-
1569.

Megusar, F., 1912. Experimente iiber den Farbwechsel der Crustaceen. Arch. Entzv.
Mech. Org. 33, 462-665.

Ostlund, E., 1954. The distribution of catechol amines in lower animals and their
effect on the heart. Acta, physiol. scand. 31, suppl. 112, 1-67.

Panouse, J. B., 1946. Recherches sur les phenomenes humoraux chez les crustaces.
Ann. Instit. occanogr. Monaco 23, 65-147.

Panouse, J. B., 1947. Les correlations humorales chez les crustaces. Ann. Biol. 23,
33-70.

Passano, L. M., 1951a. The X organ-sinus gland neurosecretory system in crabs. Anat.
Rec. \\l,S02.

Passano, L. M., 1951b. The X organ, a neurosecretory gland controlling molting in
crabs. Anat. Rec. Ill, 559.

196 INVERTEBRATE PHYSIOLOGY

Passano, L. M., 1953. Neurosecretory control of molting in crabs by the X-organ

sinus gland complex. Physiol. Comp. Oecol. 3, 155-189.
Perkins, E. B., 1928. Color changes in crustaceans, especially in Palaemonetes. J. Exp.

Zool. 50, 71-105.
Reinhard, E. G., 1950. An analysis of the effects of a sacculinid parasite on the

external morphology of Callinectes sapidiis Rathbun. Biol. Bull. 98, 277-288.
Sandeen, M. I., and F. A. Brown, Jr. 1952. Responses of the distal retinal pigment of

Palaemonetes to illumination. Physiol. Zool. 25, 222-230.
Scharrer, B., 1941. Endocrines in invertebrates. Physiol. Rev. 21, 383-409.
Scharrer, B., 1952. Hormones in insects. In The Actions of Hormones in Plants and

Invertebrates. New York.
Scharrer, B., 1953. Comparative physiology of invertebrate endocrines. Ann. Rev.

Physiol. 15, AS7-A72.
Scheer, B. T., and M. A. R. Scheer, 1954. The hormonal control of metabolism in

crustacans. VII. Molting and color change in the prawn Leander serratus. Pubbl.

Stas. Zool. Napoli 25, 397-418.
Sjogren, S., 1934. Die Blutdriise und ihre Ausbildung bei den Dekapoden. Zool.

Jahrb. (Abt. Anat.) 58, 145-170.
Smith, R. I., 1940. Studies on the effects of eyestalk removal upon young crayfish

{Cambariis clarkii Girard). Biol. Bull. 79, 145-152.
Smith, R. I., 1948. The role of the sinus glands in retinal pigment migration in

grapsoid crabs. Biol. Bull. 95, 169-185.
Stephens, G. C, 1953. The control of cement gland development in the crayfish,

Cambarus. Biol. Bull. 103, 242-258.
Stephens, G. C, 1955. Induction of molting in the crayfish, Cambarus, by modification

of daily photoperiod. Biol. Bull. 108, 235-241.
Stephens, G. J., 1952. Mechanisms regulating the reproductive cycle in the crayfish. I.

The female cycle. Physiol. Zool. 25, 70-84.
Takewaki, K., and N. Nakamura, 1944. The effects of gonadectomy on the sex char-
acters of Arniadillidiutn indgare, an isopod crustacean. Jour. Far. Sci. Imp. Univ.

Tokyo (Sec. 4) 6,369-382
Welsh, J. H., 1939. The action of eye-stalk extracts on retinal pigment migration in

the crayfish, Cambarus bartoni. Biol. Bull. 77, 119-125.
Welsh, J. H., 1941. The sinus glands and 24-hour cycles of retinal pigment migration

in the crayfish. /. E.rp. Zool. 86, 35-49.
Zeleny, C, 1905. Compensatory regulation. /. E.rp. Zool. 2, 1-102.

HUMORAL DEPENDENCE OF GROWTH AND
DIFFERENTIATION IN INSECTS

Dietrich Bodenstein
Medical Laboratories, Army Chemical Center, Marj'land

The postembryonic life of insects presents a series of developmental
steps interrupted by molts, by means of which the immature organism
gradually attains its adult form. The insect undergoes a larval or nymphal
molt when it retains its juvenile characteristics. It metamorphoses when
adult structures occur after the molt. Molting is usually accompanied by
growth, but it also always involves differentiation. Whenever the animal
passes from one stage to the next, morphogenetic events of great com-
plexity take place. These lead in the case of the skin, for instance, to the
deposition of an entirely new cuticle with all its often very complicated
structural elements. These processes of growth and differentiation are
under the control of hormones. The humoral situation prevailing at any
given stage regulates and guides the expression of the developmental char-
acters. Obviously the manifestation of these developmental events depends
not only on hormones but also on the target organs that respond to these
humoral stimuli. In the present account, special emphasis will be given to
a discussion of the responsive behavior of the reacting tissue. Those in-
terested in other aspects and further details of insect endocrinology may
consult the recent reviews by Wigglesworth (1954) and by Bodenstein
(1953b and 1954).

The Humoral Cycle

Although this paper will concern itself mainly with an analysis of the
target material, it is necessary for background information to give a short
account of the humoral cycle that triggers and controls the target re-
sponses. The humoral mechanism which causes the insect to undergo a
larval molt or to metamorphose can be briefly outlined as follows. Prior
to each molt, a humoral cycle is set into motion by impulses which in cer-
tain cases are known to be nervous in nature. They provoke the secretion
of a hormone from special cells in the brain, the so-called neurosecretory
cells. Under the influence of this brain hormone, the prothoracic glands
become activated and produce hormone. This hormone apparently acts
directly on the target organs. Its initial activity causes a wave of prolifera-
tion in the epidermal cells and thus sets the stage for molting. Because of
this growth-promoting ability, the prothoracic-gland hormone has been

[197]

198 INVERTEBRATE PHYSIOLOGY

called a growth hormone. As the titer of the hormone gradually rises,
under the influence of this hormone alone, the target organs respond by
differentiation of imaginal structures. For example, the epidermis will lay
down the complicated imaginal cuticular pattern under the influence of a
sufficient hormone titer. Since it inaugurates imaginal differentiation,
the prothoracic-gland hormone has also been called a differentiation hor-
mone. Thus growth and imaginal differentiation are caused by the same
hormone.

The sequence of events is different when the production of the pro-
thoracic-gland hormone is followed by the release of a second hormone
from the corpus allatum. In the presence of the allatum hormone, the re-
sponse of the targets is modified. After the mitotic wave induced by the
prothoracic-gland hormone has run its course, the allatum hormone causes
the targets to form larval structures ; thus the allatum hormone is respon-
sible for larval differentiation and in its presence a larval molt ensues.
Since the juvenile, or larval, characters are maintained by the presence
of the allatum hormone, the latter has been named the juvenile hormone.
During the entire larval life of the insect, the allatum hormone is present
in sufficient titer to cause larval molts. Only in the last larval stage is the
allatum hormone titer too low in relation to that of the prothoracic-gland
hormone to make its effects felt ; under these conditions, the animal meta-
morphoses, that is, the targets differentiate into imaginal structures. One
will notice that the development of larval features is controlled by the com-
bined action of these two hormones. As a matter of fact, the allatum hor-
mone can make its effects felt only in the presence of the prothoracic-gland
hormone, for it is the latter that sets the molting process into motion. The
special humoral balance prevailing at any time and the state of responsive-
ness of the target organ to this balanced hormone system together deter-
mine the features of the insect characteristic of any stage.

Growth

Growth in insects is often cyclic. At definite time intervals the immature
insect molts. Molting can be regarded as the visible expression of growth.
By the term "growth" we mean cell multiplication, unless otherwise indi-
cated. Now it has been known for a long time that the first perceptible indi-
cation of molting is the occurrence of a mitotic wave in the epidermal tar-
get, which is followed by differentiation. At each molt, therefore, the target
exhibits two main developmental reactions, namely growth and differen-
tiation. Only for didactic reasons will these two developmental events be
treated separately. Actually, both are closely related and often occur
simultaneously. We assume that cell multiplication can only occur in a
hormone-conditioned environment, which at the molting time reaches its

INSECT GROWTH AND DIFFERENTIATION 199

effective threshhold ; during the nonmitotic intermolt periods there is
apparently too little growth hormone in the system to be effective. The
experimental evidence for the above assertions is as follows.

Cyclic Growth

The necessity of the prothoracic-gland hormone for growth is well
illustrated in experiments on Drosophila (Bodenstein, 1943). Imaginal
discs transplanted into the abdominal cavity of adult male f^ies do not
grow. But growth in these organs can be induced by the simultaneous
transplantation of ring glands, which in these animals produce the pro-
thoracic-gland hormone. Thus only in the presence of the hormone is
mitotic activity possible. Similarly, in tissue culture experiments ( Schmidt
and Williams, 1953) one finds that spermatogonia of Platysamia cecropia,
isolated in a hanging drop of caterpillar blood, divide only when blood
containing prothoracic-gland hormone is used ; no growth occurs in sper-
matogonia isolated in blood containing no prothoracic-gland hormone.
The humoral dependence of the cyclic mitotic wave in the cells of the epi-
dermis is also well documented. Mitosis in the epidermis takes place only
when prothoracic-gland hormone titer has reached a certain threshold
shortly before molting (Wigglesworth, 1934; Kiihn and Piepho, 1938).
Mitosis never occurs when the hormone concentration is prevented experi-
mentally from reaching the effective level. In the other hand, whenever a
molt is induced experimentally, mitotic activity in the epidermis is also
induced. No mitosis is seen in the epidermis during the intermolt period,
when the prothoracic-gland hormone titer is expected to be very low.

The cyclic rise of the hormone level and the associated induction of
mitosis are also evident in wound healing. If small wounds (needle pricks)
are made in the ventral coxal skin of the cockroach leg, they are closed by
a migration of the epidermis cells over the wound surface. This depletes
the area surrounding the wound of cells ; thus it lacks a normal cell density.
This situation is regulated during the next molt when, under the influence
of the normal prothoracic-gland hormone level, the mitotic wave starts in
anticipation of this molt. At this time, increased cell division at the depleted
areas brings back the normal cell density (Bodenstein, unpublished). In
similar experiments on the cuticle of P. cecropia, much the same observa-
tions were made, for here too it is at the time of the occurrence of the
mitotic wave and not before that the deficient cell number is restored
(Smith and Schneiderman, 1954).

Continuous Grozvth

Quite different is the course of events in the growth behavior of the
imaginal discs of Drosophila larvae. During the entire larval period these

200 INVERTEBRATE PHYSIOLOGY

discs grow at a rather constant rate ; at least, there seems to be no peak
of growth associated with the molting times (Enzmann and Haskins,
1938). Growth of these discs seems independent of the cyclic increase of
hormone. Indeed, much growth occurs in these organs during the inter-
molt periods when the hormone level must be subthreshold. Does this in-
dicate that the discs do not depend on the prothoracic-gland hormone for
their growth? It does not, for, as mentioned before, these discs trans-
planted into an adult male host only grow in the presence of hormone
supplied by the ring gland. Growth of these discs is therefore only possible
in an environment properly conditioned by prothoracic-gland hormone.
The fact that these discs do grow during the larval intermolt period indi-
cates the presence of the hormone in the larval system in a titer sufficient
for the growth of the discs but inadequate for the induction of molting.
The same is apparently true for the imaginal discs of lepidopte