PROPRIOCEPTION IN INSECTS · Proprioception in Insects 119 The tarsal groups (7-11) occur on the distal ends of tarsal segments 1-5. There are no campaniform sensilla associated with

PROPRIOCEPTION IN INSECTS

II. THE ACTION OF THE CAMPANIFORM SENSILLA ON THE LEGS

BY J. W. S. PRINGLE, B.A.

From the Department of Zoology, Cambridge

(Received 14 March 1937)

(With Thirteen Text-figures)

INTRODUCTIONTHE previous paper described the action of the campaniform sensilla on the palpsof Periplaneta as stress receptors responding to strains in the cuticular skeleton.These sense organs are widely distributed in insects, and we shall here considertheir mode of action in the legs, where, as has been shown by Mclndoo (1914-20),they occur throughout the Class. As in the previous paper, the American cockroach,Periplaneta americana, has been used for most of the experiments, and the methodsemployed are identical to those already described; sensory impulses in the nervesfrom the sensilla are recorded by means of an amplifier and Matthews (1928)oscillograph.

A considerable amount of descriptive morphology is included in the followingaccount, as it is necessary for a clear understanding of the working of the sensilla.In the latter part of the paper previous theories of their function are discussed, andfinally an attempt is made to interpret their mode of action in terms of their knownhistological structure.

RESULTSGeneral morphology of the cockroach leg

The legs of the Periplaneta are all alike except for minor details of the size ofthe segments. The Ilnd leg is illustrated in Fig. 1, from which the general structurecan be seen.

The coxa is articulated at the base to the pleuron by an extrinsic joint (pl.c.)with sclerotized condyles, and to the trochanter, a flexible projection of the pleuron,by a complicated hinge joint (h.). It can therefore pivot about the line joining thesetwo, and can also rotate about the pleural attachment as far as the flexibility of thetrochanter will allow. Its muscles are complicated, and need not be described.

The trochanter articulates to the coxa by two extrinsic condyles, which limitits movement rigidly to one plane. The intersegmental membrane is here com-pletely flexible and offers no resistance to movement over a wide range. The linethrough the two condyles is vertical when the coxa is folded flat against the body,but rotation of the coxa about its hinge alters this until at the extreme limit of coxalmovement it is nearly horizontal and the leg moves in a vertical plane. The normal

Proprioception in Insects 115

standing position (Fig. 2) is between these two extremes, so that the coxo-tro-chanteral hinge is not quite vertical. The muscles moving the trochanter are aflexor and an extensor (Fig. 3). These are termed levator and depressor respectivelyby Snodgrass (1935), but it seems simpler to describe them by their action on thejoint and not by their supposed effect in moving the animal as a whole. The flexor

Fig. I. The Ilnd right leg of Periplaneta americana. Ventral view showing the types of articulationof the segments b ball and socket joint between tibia and tarsus; ex. coxa; fm. femur; h. hingejoint between coxa and trochantin ; pl.c. condylic joint between coxa and pleuron; pts. pretarsus;tb tibia; tn. trochantin, tr. trochanter; tt. tarsus

is attached by a single chitinous apodeme to the base of the trochanter round theinner side of the joint; the extensor has one main apodeme attaching to the outerside of the trochanter and two accessory apodemes (Fig. 4, ext' Jr.). These last arewell developed only in the second and third pairs of legs, the extensor trochanterisof which is used in the running movements of the insect. The trochanter contains

Fig 2 Fig 3

Fig 2. Pertplaneta americana, seen from below in the normal standing position.Fig. 3. Trochanter of the Illrd right leg of Pertplaneta, dorsal view. c. coxo-trochanteral condyle;ex. coxa; ext.tr. apodeme of the extensor trochanteris muscle; fl tr. apodeme of flexor trochantens;fm. femur, h hairs on the trochanter; tr. trochanter; i, 5, groups of campaniform sensilla.

but a single muscle, the reductor femoris, inserted along the dorsal side of the base ofthe femur.

The femur is firmly attached to the trochanter by an intrinsic hinge (Fig. 4, h.)on the ventral side. The cuticle forming this hinge is very tough and the pull of thereductor muscle is balanced by the elastic restoring force of the hinge. The hinge

8-3

n 6 J. W. S. PRINGLE

line, also, is not straight, so that any movement results in a distortion of the cuticleof the whole segment, which further resists movement. In the normal standingposition the whole weight of the body is taken on the elasticity of this hinge, con-traction of the reductor femoris serving only to lower the insect further on to theground and, if strong enough, to break the hinge and amputate the limb.

The tibia articulates with the femur by another pair of extrinsic condyles,giving a single plane of movement. The femur is slightly twisted, but the plane ofthe femoror-tibial hinge is also nearly vertical in the standing position. The twomuscles, flexor and extensor (depressor and levator respectively of Snodgrass) actantagonistically, the flexor being the larger of the two.

The tarsus provides an example of extrinsic monocondylic articulation. Thetibio-tarsal joint is of ball and socket type (Fig. i, b.), the two muscles being bothdepressors and attached on either side; thus besides the ordinary movement of

fi ir

Fig. 4. Trochanter of the 111 rd right leg of Penplaneta, ventral view. c. coxo-trochanteral condyles;ext.tr. extensor trochantens; ext.tr accessory apodemes of extensor trochanteris j fl.tr. flexortrochantens; h. trochantero-femoral hinge joint; 2, 3, 4, groups of campaniform sensilla.

depression, a sideways movement could be produced by relatively stronger con-traction of one muscle or the other.

The tarsal segments and the pretarsus articulate by intrinsic hinges with oneside toughened as in the palps. But from the end joint inwards there is a gradualincrease of sclerotization and infolding of the cutical, so that the joint between thefirst and second tarsal segments is practically also a ball and socket joint. There is asmall depressor pretarsi, with a long thread-like apodeme passing back to thefemur, where lies most of its muscle.

Sensory arrangement

Campaniform sensilla occur on the legs in the following positions: four groupson the trochanter, one on the base of the femur on the dorsal side, one on the outsideof the base of the tibia, and a group on each of the segments of the tarsus (Figs. 3,4, 5 and 6).

The numbers of sensilla in the groups are shown in Table I, compiled from asingle fair-sized insect.


3rd Is

2nd Is

Fig. 6

Fig. 5. Femero-tibial joint of the Illrd left leg of Penplaneta, dorsal view. c. condyle; ext.tb. extensortibiae; fl.tb. flexor tibae; 6, group of campamform sensilla.Fig. 6. Tarsus of the Illrd leg of Penplaneta. Dorsal view to show the position of the campaniformsensilla. ind U , 3rd tt. second and third tarsal segments; 8, 9, groups of campaniform sensilla

Table I. Number of campaniform sensilla on the legs of Periplaneta americana.The right 1st leg had only four tarsal joints

Sensilla group

Leg I R.L.

II R.L.

Ill R.L.

1

20

18

19202018

2

1920

191920

19

3

IS1818181919

4

131314131515

5

10

97787

6

887711

11

7

21

1

11

1

8

222221

9

2

22

2

2

IO

2I

II

II

II

II

IIII

Trochanter and femoral groups

If the trochanter and femur are treated with boiling potash to remove all but thecuticle, and are then cleared in clove oil, the sensilla are immediately obvious undera magnification of 50 x and have the appearance of pores, their thin coveringmembrane being almost transparent. The detail is shown in Fig. 7. The sensillavary in shape and size, but are rarely quite circular and sometimes long and narrow.Their length varies from 6 to 24p. They are arranged more or less regularly inlines, with the long diameters of the sensilla diagonal to the direction of the lines.Sometimes there is indication (Fig. 7, 3) of a second alignment, at an angle to thefirst. When this is apparent, the direction of the long diameters of the sensillabisects the angle between the two lines.

With intra-vitam staining with a methylene blue, the large sense cells of thesensilla can be seen. Those of group 5 are always visible through the more trans-parent cuticle, but the trochanteral cells lie deeper, and dissection is necessary tofind them. Each sensillum has one cell. Cross sections of the nerve that supplies

n8 J. W. S. PRINGLE

the groups of sensilla on the ventral side of the trochanter of the Ilird leg showthat there are here 52 fibres, varying in diameter from 6 to 15/x. Although twodifferent specimens were concerned, this compares well with the 53 sensilla ingroups 2, 3 and 4 of the Ilird leg, as shown in Table I. The sections were cut justbefore this nerve joins the main trunk of the leg, and evidently at this point there

»* 1

# #

Fig. 7. Details of the orientation of the groups of campaniform sensilla on the Ilird legs of Pert-planeta. The numbering corresponds to that in Table I, but the drawings were made fromdifferent specimens from those used for Figs. 3, 4 and 5

has been no joining up of fibres (compare the sensilla on the palp; Pringle, 1937).Higher up the nerve the number of fibres is so large that it is impossible withoutserial sectioning to tell which comes from which sense organ.

Tibial and tarsal groups

Group 6 occurs on the tibia just proximal to the first large tibial spine (Fig. 5).The cuticle is thick here. Unlike the other groups on the leg, the sensilla do not allpoint in the same direction, but two subgroups can be distinguished, each with aconstant orientation (Fig. 7, 6). The possible significance of this is discussed later.


The tarsal groups (7-11) occur on the distal ends of tarsal segments 1-5. Thereare no campaniform sensilla associated with the tibio-tarsal joint. Fig. 6 shows thesensilla on the tarsal segments 2 and 3. They lie on the overlapped part of thesegment proximal to the joint, and their long diameters are always longitudinal tothe limb. The long apodeme of the depressor pretarsi runs along the opposite,i.e. the under, side of the tarsal joints.

Physiology

The fibres from the leg campaniform sensilla all run in the larger of the twonerves to the leg, and impulses from them can be recorded in it. It has not beenpossible to study the organs singly, and all records therefore show the combined

Fig 8. Discharge in the large nerve from the Illrd leg of Penplaneta following light (A) and strong(B) pressure on the side of the trochanter. Time in I/IO sees.; recording from the base of the coxa.

effect of impulses in many fibres. The types of response shown in Fig. 8 are obtainedby pressure with a glass needle on the side of the trochanter. This distorts the cuticleand will set up large forces in its surface. Evidently the sensilla here are reactinglike those on the palp to stresses in the cuticle (Pringle, 1937). Sometimes, with lessstrong pressure, a regular rhythm is observable (Fig. 8 A), though the large variationin impulse size shows that more than one fibre is active. A regular adaption canbe observed in this record and is difficult to explain except as the action of a singlefibre with other impulses superimposed; in which case the total number of activefibres in the nerve at this point (the base of the coxa) cannot be very large.

All the groups of sensilla can be excited in this way by pressure on the cuticlenear them, but the normal stimulating conditions are less easy to determine. Whenthe nerve is cut in order to record the impulses from its distal end, the motor

120 J. W. S. PRINGLE

supply to the leg is removed and the muscles relax. The central end shows a largespontaneous discharge, which is likely to include the normal mechanism for main-taining tone. With the muscles relaxed, the trochanter and the leg distal to it maybe moved on the coxa as far as the dicondylic hinge will allow, and very little activityis observed in the nerve during this movement; there are certainly none of the largeimpulses from the campaniform sensilla.

Any movement of the femur on the trochanter, however, gives rise to a dischargesimilar to that obtained by pressure on the trochanter. This is easily understood fromthe fact that the trochantero-femoral hinge is not straight but curved, and cannottherefore be bent without distorting the whole of the cuticle in the neighbourhood.This will produce the necessary strains for stimulation of the campaniform sensilla.It has already been explained that in normal standing the weight of the body istaken on the elasticity of this hinge; the intensity of excitation of these sensilla willthus be dependent on the vertical pressure on the legs.

Passive bending of the dicondylic femoro-tibial joint also produces a slightdischarge of impulses in the nerve, but it is difficult to be sure that some of the longspines are not being stimulated by this movement, and these are known to give anincompletely adapting discharge (Pumphrey, 1936). The campaniform sensilla can,however, be strongly excited by any sideways bending of the joint out of the normalplane of movement.

The response of the tarsal groups has not been studied. Any touching of thetarsus produces such a large discharge of impulses from the tactile hairs thatfurther activity in the nerve would not be noticeable. Neither has it been possibleto investigate the effect of active contraction of the muscles, though stresses wouldcertainly be set up in the cuticle if such contractions were resisted by an outsideforce. But before discussing in general the function of these leg sensilla, it will benecessary to consider more closely the nature of the stimuli to which they respond.This involves a consideration of the structure of the sensillum, and a review ofprevious theories of its function.

The function of the campaniform sensilla

Terminology. The term "sensilla campaniformia" was proposed by Berlese(1909) to include the whole class of insect sensilla that resembles in general charactersthose first described by Hicks (1857). Minor structural variations had produced inthe literature a wealth of different names, some of which, particularly "Grubenohne Kegel", "Sinneskuppeln", and "Papillen" of German writers, still persist.But Berlese's term is now in general use (Snodgrass, 1935).

Structure. The morphology has been extensively examined and is now relativelyclear (Fig. 9). It may be summarized as follows:

(1) A canal penetrating the whole thickness of the cuticle, covered at the top bya domed cap, and lined with some substance with different staining reactions to therest of the cuticle (Newton, 1931; " Polstermasse " of Vogel, 1911, and Sihler, 1924).

(2) The top of the canal, and the depression or swelling that the cap makes inthe surface of the cuticle, sometimes circular, but more usually elongated in one


direction; with the cap membrane thickened along the long diameter of the oval(sensilla on the halteres of Diptera, (Pflugstaedt, 1912); and on the appendages ofPeriplanetd).

(3) The cap membrane usually domed above the surface of the cuticle, butsometimes recessed so that its top is level with the surface ("scapalen Papillen" ofthe halteres, Pflugstaedt, 1912; Fig. 9 D).

(4) Sensilla usually occurring in groups with a definite arrangement and orien-tation (Mclndoo, 1914).

(5) The single sense cell large, with a distal process inserted on the under sideof the middle of the domed cap by a highly refracting body, which in oval sensillamay also be elongated on the long diameter. It is this body that makes the dome

Fig. 9. Types of campaniform sensilla A. Diagrammatic. B, C. Sections through the long andshort diameters of a sensdlum on the haltere of Calliphora. D. Sensillum on haltere of Syrpkus.E. From the cercus of Penplancta. F. From the labium of Dytiscus G. From the mandible ofDytiscus. a, b, outer and inner lamellae of cap membrane; c, cuticular connexion of d, the distalprocess of the sense cell. (From various authors, after Snodgrass, 1935.)

appear to be a pore when viewed with certain arrangements of transmitted light, aswrongly described by Mclndoo (1914) and Newton (1931).

(6) Other modified hypodermis cells usually absent, but apparently chitogenouscells occur in some cases (Sihler, 1924).

The sensilla have been described from a variety of positions in many insects.Snodgrass (1935) lists: "head, thorax, abdomen, the antennae, mouthparts, legs,wing bases, cerci, and ovipositor of adult insects,... also... on the larvae of somespecies". In addition, the halteres of Diptera and Strepsiptera are richly providedat their base (Pflugstaedt, 1912; Ulrich, 1930). Mclndoo, in a series of papers from1914 to 1920, gives tables of the number of sensilla in various positions in differentinsects, but usually only as totals of a number of groups. His paper of 1914 givesa more detailed description of the main groups on the legs, wings, and sting ofApis. Unfortunately he is not always accurate in his descriptions and seems tohave confused several types of sensillum. For instance, he states (Mclndoo, 1920)that there are 77 "pores" on the maxillary palps of Periplaneta americana. We haveseen (Pringle, 1937) that the total number of campaniform sensilla on both sides isactually 22 in a typical specimen. This disagreement is too great to be explained asindividual variation, and he must have been including either gland openings or


more likely sockets where hairs have been shed; these can resemble campaniformsensilla very closely in untreated material, but after clearing with clove oil thecharacteristic structure of the sensillum is at once obvious. Mclndoo's results arethus likely to be unreliable as to the number of sensilla, but his papers do show thatthese occur very generally thoughout insects.

Function. Mclndoo alone among investigators of the campaniform sensillaregards them as olfactory in function. His work must be examined in some detail.

He described their histology and published figures (Mclndoo, 1914) showingthe protoplasm of the distal sensory process freely exposed to the exterior. Newton(1931) reinvestigated the question, and established that there is a definite cuticularcovering over the cell process, and that the pore of Mclndoo was an artefact due totreatment of the preparation. The existence of this covering does not, however,exclude the possibility of the sensillum having a chemoreceptive function, as anumber of workers have shown that chitin, when unimpregnated with other sub-stances, is highly permeable (Wigglesworth, 1933). Mclndoo's other evidencemust therefore be considered.

In many of his papers he describes experiments purporting to show that the"olfactory pores" really have this function. All the experiments are similar inform. The insect is placed in a gauze cage, and left until it becomes stationary.The odour under test is then brought up under the cage and any reaction on thepart of the insect noted. Any movement is taken as a positive result, and thereaction time is measured with a stop-watch. By measuring the reaction time ofindividuals with various numbers of their "pores" removed or covered up withsome substance he finds a rough relation between average reaction time and numberof pores left. The odours tested were usually strong essential oils such as pepper-mint, wintergreen, bergamot, etc.

The method is crude and liable to a number of objections. No attempt is madeto exclude other stimuli, such as light intensity and air vibration which mightstimulate the insect to movement. The effective concentration of the odorouschemical at the surface of the sense organs will be affected by draughts and will thusnot be constant throughout the series of experiments. And the reaction timedepends on other things besides the intensity of excitation of one type of senseorgan; it is affected by the whole state of the central nervous system, which will beseriously upset by amputation of any part of the body. Even the amount of handlingto which Mclndoo's insects were subjected would be enough to slow down theirreactions.

But in order to test the olfactory theory in a final and definite way, severalstrongly smelling substances, including some of those Mclndoo used, were appliedto the campaniform sensilla on the trochanter of Periplaneta, both as a gas byholding them some distance away, and by the drastic method of painting theliquid on the sensilla. Oils of peppermint and wintergreen, and amyl acetatewere tested in this way, and no increase in the number of impulses in the leg nervecould be detected by the loud-speaker noise or on a viewing screen. That the sen-silla of this preparation were healthy and normal could be shown by the usual


response obtained to pressure on the trochanter, and the experiment suggestsstrongly that the campaniform sensilla have nothing whatever to do with olfaction.

The alternative view, that they are mechanical receptors of some type, has beenput forward by several authors; and Weber (1933) quotes Demoll as suggestingmore definitely that they might serve a proprioceptive function by responding tobending of the cuticle. No possible mechanism, however, has been described.

The mechanism of the campaniform sensilla

The experiments already described suggest that the campaniform sensilla aremechanical receptors, responding to stresses in the cuticle. This behaviour has beenverified experimentally for sensilla on the palps and legs of Periplaneta. It remainsto show that such a function is possible for all such organs wherever they occur, andto define more accurately the adequate stimulus.

With the exception of some of Mclndoo's account, which have been shown tobe unreliable, there is no case known of a campaniform sensillum occurring in apart of the cuticle which would not be liable to strains.

The main groups are located at the limb joints, where stresses are set up bybending and muscular contractions, and at the wing and haltere roots where areconcentrated the forces that move the organ. Other well-established positions areat the base of the barbs of aculeate stings (Mclndoo, 1914), on the mandibles ofvarious biting insects (Dytiscus, Hochreuter, 1912; bees, Janet, 1911), and on theoscillatory gills of ephemerid nymphs (Eastham, 1936). In all these cases stresseswill occur in the cuticle under normal conditions. And though the sensilla describedby Sihler (1924) from the cerci of Orthoptera would seem at first to be an exceptionas the cercus as a whole should not be liable to strains, it is clear from his descriptionsthat the sensilla occur only in close connexion with structures at the base of thehairs, and on portions of the cuticle which are likely to be strained when the hair isbent.

There is nothing then in previous publications that contradicts the view.

Primitive type

A simple type of sensillum is found on the segments of the thoracic legs of thecaterpillar of Smerinthus ocellatus (Lepidoptera, Sphingidae). It is circular inappearance, and in cleared whole preparations the only structure that can bedistinguished is a slightly heavier pigmentation of the sides of the canal. The wholeof the cuticle is very elastic and flexible except for the claws at the end of the leg.A slowly adapting discharge of impulses can be detected from a few fibres in thenerves to the legs when the cuticle is distorted, and the behaviour thus seems tobe similar to that of other sensilla. It can be supposed that the normal adequatestimulus is resistance to the contracting of the leg muscles by contact of the clawwith an outside object.

With such an elastic cuticle it is probable that the distortion from any appliedforce will be rather large, and a small degree of differential elasticity between parts

124 J- W. S. PRINGLE

of the sensillum could produce enough local movement to excite the end of thenerve. The difference in appearance of the lining of the canal may give a clue tothis.

Adult type

In adult insects the cuticle is often greatly toughened by deposition in the chitinof other substances (Wigglesworth, 1934); actual bending of the skeleton is reducedto a minimum, and it is necessary to consider how the sensilla can be excited bymechanical stimuli, situated as they often are in the thickest parts of the cuticle. Todo this we must first discuss some of the general mechanical properties of surfacesand hollow tubes.

A thin, fiat surface can be subjected to two types of strain; to bend by forcesat right angles to its plane, and to shear by forces in its plane. Illustrative models

Fig. 10. Diagram to illustrate the resolution of shear forces into compression and extension com-ponents. A, a flat surface; B, C, a hollow cylinder subjected to twist and bend. Continuous lines,extension. Dashed lines, compression. For further explanation see text.Fig. 11. Diagram of models illustrating the action of the campaniform sensillum. A. To representthe actual sensillum. B Model constructed in a circular framework for measurement of the effectof compression in different directions. See text.

can be made out of sheet rubber, which is sufficiently elastic to allow of visiblemovements. Fig. 10 A shows a diagram of such a model, made by stretching apiece of 1/16 in. rubber over a rectangular metal frame ABCD. A shearing forcecan be set up in the rubber by distortion of the rectangle into a parallelogram(A'B'C'D'). The distance AC is lengthened and BD shortened. The surface materialis in tension in the direction AC, and in compression in the direction BD. Unlessthere was an initial stretch in the rubber, creases will appear parallel to AC. Ashearing force thus consists of tension in one direction and compression in another.It should be noted that the lines of compression and extension need not intersectat right angles.

A hollow cylindrical tube has shearing forces set up in it under any appliedforce (Fig. 10 B, C). With a twist, the lines of compression and extension run in a


spiral manner round the tube in opposite directions; with a bending moment, theouter side has extension longitudinal and compression transverse, the inner sideextension transverse and compression longitudinal, and the sides a normal diagonalshear. Again it must be noted that the lines do not necessarily intersect at rightangles, and with a bending force they are not even parallel. All this applies, ofcourse, only to a homogeneous surface; any variation in thickness or compositionalters the position and direction of the compression and extension lines. It assumesalso that the forces are applied evenly round the ends of the tube.

Insect limb segments are never articulated evenly all round their base; if theywere the articulation would be rigid. Nor are they perfect homogeneous cylinders.The force lines will therefore not be regular. But the important point is that in ahollow structure such as the insect skeleton, all stresses can be expressed as shearingstresses, and resolved into compression and extension components.

The probable mode of action of the campaniform sensilla in registering thesestresses can best be explained by reference to a model which shows many of theproperties of the sensillum. Fig. 11 A shows one variation of this model. A rubbermembrane stretched over a framework is pierced by a circular or oval hole, andacross the long diameter of the oval is glued a strip of paper, slightly domed out ofthe plane of the diaphragm.

The hole in the rubber represents the canal through the cuticle, and the stripof paper the longitudinal thickening of the cap membrane or of the attachment ofthe nerve process. The rest of the cap is assumed to be relatively more elastic thanthis thickening and is omitted. This differential elasticity of two parts of the skeletalstructure is an essential assumption. It need not be a sudden change of propertyalong a definite line; it can in fact be a gradual grading; but it must exist for theproper functioning of the sensillum according to the present theory. If opacity canbe taken as a measure of the thickness of the cuticle or of its degree of impregnationwith other substances, and this as the cause or constant accompaniment of increasedtoughness, then there is good evidence for its existence. For in all adult sensillaexamined the surface view in transmitted light shows a clear line down the longdiameter of the cap (Fig. 12).

Slight distortion of model 1 from the rectangular shape produces in onedirection a lengthening of the hole, and in the other a shortening. This causes thepaper strip to be more or less humped, and its middle point moves on a line atright angles to the plane of the diaphragm. If the end of the nerve process is imaginedas being inserted on the under side of the middle of the paper strip, it will bestretched by one distortion and compressed by the other. One or both of thesemovements could lead to excitation of the ending.

The magnification of movement provided by this arrangement is very largeindeed; the more so if the dome be nearly flat. In order to measure it accurately,model 2 was constructed (Fig. 11 B). This is similar to model 1 except that twostrips of paper are attached across the hole, one on each side and with equal amountsof doming. The movement between the middle points of the two strips will bedouble the stretching of a fibre attached from one to a point fixed with respect to the

126 J. W. S. P R I N G L E

diaphragm. The model was fixed horizontal and a small electrolytic cell constructedon the lower of the two strips. This consisted of a short length of glass tubing fixedvertically with one electrode inserted through the paper from below. The otherelectrode was mounted on a straight piece of wire from the upper strip and was freeto move inside the glass tube. Both electrodes had circular disks of platinum foil astheir main surfaces. With the tube filled with an electrolyte (dilute salt solution) acell is thus made with the interelectrode distance variable; provided the upperelectrode moves vertically, the resistance of such a cell is roughly proportional tothe distance between the electrodes.

Model 2 was made in a circular frame consisting of a pair of wooden embroideryrings. This made possible the determination of the effect of distortion in variousdirections. With the arrangement shown in Fig. 11 B any required distorting force

Fig 12. Photomicrograph of the sensilla of group 4 on the Illrd leg of Periplaneta, showing theappearance in transmitted light Note the doming and longitudinal thickening of the cap mem-brane. From a potashed preparation cleared in clove oil.

can be applied in any direction. The model has the disadvantage that in it thecompression and extension lines always cross at right angles, and it is therefore notpossible to reproduce the diagonal intersections that occur in the insect cuticle.Owing, also, to the rigidity of the oak rings of the framework, large forces arerequired to produce a distortion, and only a small proportion of the restoring forcewill be derived from the elasticity of the rubber diaphragm.

Fig. 13 shows the relation between the direction of the applied force and theresistance of the cell for various intensities of compression. The zero angle is theline through the strip of paper. The resting resistance of this model was 92 (arbitraryunits), and the curve for the least applied force (750 g.) crosses this line at a pointcorresponding to an angle of 450. At this angle there is no movement. For thehigher values of distorting force, owing to the fact that the paper strip has to beglued on to the rubber over a definite area, there is an appreciable twisting of thepaper at any angle not o or 900, and this produces a sideways movement of the upper


electrode which will introduce an error into the readings. This factor would notappear in the actual sensillum, in which forces with 45 ° orientation should produceno movement.

There is thus a critical angle below which there is an increase and above whichthere is a decrease of doming. If it be assumed for the moment that excitation onlyoccurs with increased doming, then the sensillum will be excited when the com-pression component of the shear force makes an angle of less than a certain anglewith the direction of thickening of its cap membrane; for force lines intersectingat right angles the critical angle will be 45 °.

•a

|"aa

+90°

+60°

+30°

0°

-K)°

-60°

-90°

Si*

\

-

-

/. / 1

225(

]

\

/

II

Iii

1 \1 '

1

1

NX

J /

// /

_L1\\

i

i

50 92 100Resistance (arbitrary units)

15*1

Fig. 13 Graph obtained from model 2, showing the relation between the angle of t̂ ie appliedcompression force and the resistance of the cell, for various values of the compression force.

The sensitivity of the sensillum will be dependent among other things on theabsolute length of the long diameter of the cap. For a sensillum of given area, agreater sensitivity will thus result from an oval shape. In the various groups ofsensilla on the legs of Periplaneta there is often a considerable variation in length.This, if other things are equal, should give a variation in sensitivity among theindividuals of a group.

Since over any given small area of the cuticle the compression lines will alwaysbe running approximately parallel, the important rule may thus be laid down, thatsensilla with parallel orientation will respond to the same type of shear force. Each


group of such sensilla will act as a unit, and if innervated from the same fibre will beextending the range of that fibre quantitatively but not qualitatively. This conditionwe have already seen in the sensilla of the palps of Periplaneta (Pringle, 1937), andeven without proof of a sensory anastomosis in the legs, it simplifies very con-siderably the interpretation of the behaviour of these sensilla also.

Unidirectional sensitivity

The assumption made above that each sensillum reacts only to. the compressioncomponent of the shear force must now be justified. Owing to the irregularities ofthe cuticle of the legs it has not been possible to prove this objectively, and itrests only on a number of circumstantial considerations, none of which is con-clusive. The point is therefore not finally settled, but the following may be con-sidered as evidence:

(1) In each of the leg groups in Periplaneta except group 6, the sensilla are allorientated in the same direction. If the sensillum were equally sensitive to thecompression and extension components of shear, an equally efficient system wouldresult if some were at an angle of 900 to the others (for force lines intersecting atright angles). The fact that this arrangement is not found suggests that only onecomponent is effective.

(2) Group 6 on the tibia of Periplaneta is exceptional in that two directions oforientation are apparent. In some insects this group is represented by two groups{Apis, Mclndoo, 1914). If each sensillum reacts only to one component, then thefunctional duplicity is preserved in Periplaneta, though all the sensilla are groupedtogether.

(3) It is very noticeable with many hair sensilla that only one direction ofmovement is effective. This was noted by Pumphrey (1936) for the tibial spines ofPeriplaneta, and is also true of certain hairs on the metathoracic spiracle (author'sobservation of the sensory impulses). Histological studies have shown that thenerve fibre is inserted on one side at the base of insect hairs in very much the sameway as under the dome of the campaniform sensilla. The same behaviour shouldtherefore be shown by both types of sensillum.

(4) The sensilla on the tarsus of Periplaneta are situated on the upper surfaceand have their long diameters parallel to the length of the leg. When the tarsus ispressed against the ground by contraction of the depressor metatarsi, the compressionlines in the cuticle of the upper surface of the segments will also be longitudinal.If only one component is effective, then it must be the compression component, forunder no normal circumstances can the reverse condition (tension lines longitudinal)be produced in the living insect.

(5) In discussing the response of the sensilla on the palp to free movement of thesegment by its muscle it was necessary to assume that the sensilla reacted only tocompression down the length of the palp (Pringle, 1937). The sensilla of the palpjoints are all orientated with their long diameters parallel to the length of the palp,and it is difficult to avoid the conclusion here that the tension component acting inthis direction is ineffective.


Discussion of the function of the campamform sensilla on the legsof Periplaneta

With the above theoretical background we can now return to the question of thefunction of the sensilla on the legs of Periplaneta. Each group must be consideredin turn.

The tarsal sensilla present no difficulty; as stated above, they are orientatedparallel to the only direction in which a compression component of shear cannormally be set up, and they should react to the pressure of the tarsal segments onthe ground. But the others are more complicated.

Group 6 of the Ilnd and Illrd legs is situated in a region where the cuticle isirregular, owing to the proximity of the thickened edge round the base of the firsttibial spine. Its two subgroups are orientated almost at right angles to each other,and in general direction the larger group is at an angle of about 700 to the lengthof the tibia. This makes its function difficult to decide. In the 1st leg the cuticleis more regular and the orientation is almost exactly longitudinal and transverse,which would make it respond to the bending forces produced by resistance to thecontractions of the extensor and flexor tibiae. This may also be the function ofgroup 6 in the Ilnd and Illrd legs; but there is also the possibility here, owing tothe fact that in normal standing the femoro-tibial hinge of these legs is nearlyvertical, that it may respond rather to the sideways bending that must occur underthese conditions. That this movement is certainly effective was shown experimentallyfor these legs, but it was not excluded that resistance to straight extension orflexion might not also excite. In the 1st legs, the coxa is normally rotated until thishinge is nearly horizontal, so that the distinction does not arise, but in the hind twopairs it becomes important to distinguish between the sideways forces present in thestanding position and those produced by the movements of running. Possibly boththese excite group 6.

In the case of the trochanteral and femoral groups deductions from anatomybecome even more difficult owing to the great divergence of this region from thecylindrical shape. It can only be pointed out again thar in the normal standingposition there will be a force tending to bend the trochantero-femoral hinge, andthat it has been shown experimentally that this excites some of the groups of sen-silla.

When the insect is standing on the ground, or when it is hanging inverted, therewill, then, be excitation of at least some of the groups of campaniform sensilla(certainly 6-11, and some or all of 1-4). This behaviour provides a factual basis fora sense that has many times been postulated from experiments on the generalbehaviour of insects. Some of these may be considered.

In the cockroach itself, Hoffmann (1933) has described a righting reflex which isinitiated by the absence of contact of the legs with the ground. Hoffmann wasunable to decide between tactile and pressure stimuli as the normal means ofinhibiting this, but the following experiment shows that the latter are responsible.If all but the first tarsal segments of the legs of a cockroach are amputated, glass

JEB'XVl 9


rods can be inserted in the cut ends of the legs in such a way that there is nostimulation of any tactile hairs. An insect so treated appears to be perfectly normalin its standing and locomotion. If placed on its back, the movements of the rightingreflex at once appear, and can be inhibited again, as shown by Hoffmann, bylowering a glass plate on to the insect. This inhibition appears when there is nocontact with the plate except by the tips of the glass rods in the legs, the pawingmovements ceasing in any leg of which the rod can be pressed against the plate. Itseems evident that it is excitation of the campaniform sensilla that is normallyproducing the inhibition of the reflex.

Fraenckel (1932) and Hoist (1935) describe further effects due to contact of thelegs with the ground. Fraenckel relates the onset of flight to the removal of thisstimulation, while Hoist assigns to pressure stimuli on the legs an important rolein the co-ordination of walking movements, both in insects of all sorts, and inmyriapods, and Crustacea. Campaniform sensilla do not, indeed, occur in some ofthe animals he mentions, but it is possible, nevertheless, that some similar cuticularstrain sense is at work throughout. It may, perhaps, be found to be a general rulefor arthropods that here, in direct contrast to the vertebrate plan, the force exertedon the limbs by the weight of the body and the contraction of muscles is measured,not by the tension in the tendons of the effectors, but as compression in the skeletalsupports. If this be so, it must modify to a large extent our conception of the role ofproprioception in this phylum. The conclusions of Crozier & Stier (1928-9), whofind in a differential force sense in the legs a mechanism for explaining geotropicorientation, must be reconsidered if this sense does not, as they thought, provide aseparate measure of the state of each muscle. The campaniform sensilla may not bethe sole proprioceptive sense organs in the leg (Slifer (1936) has recently shown thatchordotonal organs are much more widely distributed than had been thought), butthe large size of the fibres supplying the campaniform sensilla (cp. Barnes, 1930),and the type of stimulus to which they respond, suggest that they are an importantpart of the reflex system. They may provide the only force sense in the legs.

We have seen, both in the palps and the legs of Periplaneta, how these sensillaserve as proprioceptors not for any one muscle, but for the appendage as a whole.This provides another contrast to the vertebrate arrangement, for it means that inthe insect the behaviour of the limb is automatically integrated before beingreported back to the central nervous system. It is tempting to correlate this smalldegree of qualitative perception with the instinctive nature of most insect move-ments. Nicholas & Barron (1935) have claimed recently for Amphibia, and it iswidely accepted for the higher vertebrates, that the proprioceptive impulses playa large part in the building up of the complex behaviour patterns of these animals;it would be impossible for much variety of movement to develop in this way in ananimal where proprioceptors can only function correctly with one type of co-ordination of the muscles of each limb segment.


SUMMARY1. The campaniform sensilla on the legs of Periplaneta are similar in action to

those on the palps, and respond to strains in the cuticle.2. They are arranged in groups at the joints, with parallel orientation of the

sensilla of a group.3. Tests with various chemical substances show a complete absence of sensitivity

to olfactory stimuli.4. A theory is given of the mode of action of the sensilla in terms of a mechanical

model based on their observed structure. Each group of parallel sensilla should actas a unit, responding to those forces which have a compression component of shearin the direction of their long diameters.

5. This theory makes it possible to predict the behaviour of the sensilla fromtheir anatomical arrangement. Most if not all the groups on the legs are so arrangedas to be sensitive to the forces present when the insect is standing on the ground.

6. The sensilla probably provide the basis for the sense of contact pressurepostulated by Hoist (1935), Hoffmann (1933), Crozier & Stier (1928-9), Fraenkel(1932) and others.

7. Comparison of this propnoceptive mechanism with that of the vertebratelimb reveals an absence of qualitative sensitivity that may have an importantbearing on the question of the evolution of behaviour.

REFERENCESB\RNES, T C. (1930) C R Soc. Bwl., Paris, 105, 385.BERLESE, A. (1909) Git Insctti. Milan.CROZIER, W. J. & STIER, T . J. B. (1928-9). J. gen. Pkynol 12, 675.EASTHAM, L. E. S. (1936). Trans. R. ent. Soc. Land. 85, 401FRAEJIKEL, G. (1932) Z. vergl. Pkynol. 16, 371.HICKS, J. B. (1857) J Linn. Soc. (Zool.), 1, 367.HOCHREUTER, R. (19:2). Z tvui. Zool. 103, 1.HOFFMANN, R. W. (1933). Z. vergl. Phystol. 18, 740.HOLST, E. V. (1935). Bwl. Rev. 10, 234.JANET, C H (191 I ) . Apiculteur, 55e Annee, No 3, Mars, p. 107MCINDOO, N. E. (1914). J exp. Zool. 16, 265.

(1915) Bwl. Bull. Wood's Hole, 28, 407.(1917). J. comp. Neurol. 29, 457.(1918) J.Morph 31, 113.(1920) J. comp. Neurol. 31, 405.

MATTHEWS, B. C. H. (1928). J. Pkynol. 65, 225.NEWTON, H. C. F. (1931). Quart. J. rmcr. Set. 74, 647.NICHOLAS, J. S. & BARRON, D H. (1935). J. comp. Neurol. 61, 413.PFLUGSTAEDT, H. (1912). Z. wiss. Zool. 100, 1.PRINGLE, J W. S. (1937). J. exp. Bwl. 15, 101.PUMPHREY, R J. (1936). J. Phystol. 87, 6 P .SIHLER, H (1924). Zool. jfb., Abt. Anat. 48, 519.SLIFER, E H (1936). Ent. News, 47, 174.SNODGRASS, R. E (1935). Principles of Insect Morphology. New York.ULRICH, W. (1930) Z Morph. Okol. Tiere, 17, 552.VOGEL, R. (1911). Z. ivtss. Zool. 98, 68.WEBER, H. (1933). Lekrbuch der Entomologte. Jena.WIGGLESWORTH, V. B. (1933). J. exp. Biol. 10, 1.

(i934)- Insect Phynology. London.

9-3

PROPRIOCEPTION IN INSECTS · Proprioception in Insects 119 The tarsal groups (7-11) occur on the distal ends of tarsal segments 1-5. There are no campaniform sensilla associated with

Documents