The Co-Ordination of Insect Movements: II. The Effect of Limb Amputation and the Cutting of Commissures in the Cockroach (Blatta Orientalis)

The normal rhythms of walking in the cockroach and other insects have been described in a previous paper (Hughes, 1952a). It was shown that movements of the six legs are co-ordinated in a very precise way and that two basic rules are obeyed in the insects studied. Alterations in this pattern can be readily detected and hence provide good material for a study of the role of the central and peripheral nervous systems in the co-ordination of locomotory movements. In the present paper experiments are described in which the removal of legs or the severing of thoracic commissures directly or indirectly prevents sensory impulses from reaching parts of the central nervous system. Removal of a leg also results in the remaining legs having to function under altered conditions. The modifications in walking so produced provide useful information about the mechanism of normal co-ordination.

A great deal of work has already been done in this field, but few investigators have made use of cinematography to analyse the effects of operations. For this reason they have usually failed to detect alterations in the locomotory movements which occur when relatively small parts of the neuromuscular mechanism have been removed. Detailed analyses are given from films of cockroaches with a single leg or any two legs removed. These have revealed certain deficiencies in earlier accounts of alterations in rhythm and have emphasized significant changes in posture of the legs. It is the purpose of this paper to draw attention to some of these and to show how they are related to the altered mechanical conditions and to the normal reflex mechanism of the leg.

The cockroach used throughout these experiments was Blatta orientals. Females were preferred because the absence of wings made their leg movements more visible and also facilitated the fixation of steel wires, small weights, etc., to the thorax. The normal walking of most insects was filmed before any experiment was performed but only slight individual variations were found. A Sinclair ciné camera was used at speeds of 24–32 frames/sec., with lighting provided by two Photofloods or two 100 W. lamps and a spotlight. The insects walked at room temperature (18–20°C.) over a card marked with a grid.

The films were analysed by superimposing the body outline in successive frames and plotting the positions of the limb tips. In this way the position and length of the stride with respect to the body was obtained, and this gives a useful indication of the altered posture of the legs. As the results of this method were not published for the normal insect, examples from two different insects are shown in Fig. 1. The first is from the same shot as the sequence of stills in plate 12 (Hughes, 1952 a),* while the second is from the insect whose middle legs were later amputated in the present work (Fig. 2).

Successive positions of each limb tip are marked and when joined together they give the course of that limb tip relative to the body. Retraction is shown by a heavier line, and of course during this part of the cycle the limb is fixed relative to the ground. Only the frames at which a given leg is maximally protracted or retracted are numbered, but from these the position of the limb tips for any other frame can be determined. Then, by joining together the points for the different legs on the same frame, it is possible to construct the areas of support throughout the whole cycle and to see their relationship to the centre of gravity, which is usually situated about the level of the first abdominal segment. This has been done for frame 3 in Fig 1 B and it can be seen that during this phase the body is supported by L₁,R₁, L₂, R₃.

The tracings below each of these diagrams were obtained from the same strips of film, but here the position of each limb tip is plotted relative to the antero-posterior axis of the body. The points for each frame are plotted on vertical lines which are separated by a standard distance in order to introduce the time co-ordinate. The individual points for a given leg are omitted, but when joined up they form the curves as shown. The thick parts of the curves indicate the time when the leg is fixed on the ground and dashed lines are used for right legs. The points correspond to the intersection of the curves and lines drawn vertically for each frame. Three such lines are drawn, the interval between them being 0·25 sec. The two horizontal lines show the position of head and tip of the abdomen. The order of lifting is given at the bottom of each figure.

The analyses presented were selected to show the most typical movements. There are certainly variations from the patterns shown between both individuals and different runs of the same individual, but space does not permit a detailed description of these.

Carlet (1888) was the first worker to observe the changed rhythm of movements when the two middle legs of an insect are amputated. Later, Buddenbrock (1921) made a systematic series of experiments using the stick insect, Dixippus, which was excellent material for this purpose because of its slow speed. He found no perceptible change when a single leg was removed but consistent alterations in the rhythm followed the amputation of two legs. In all the latter cases diagonal legs moved almost simultaneously, the most extreme alteration occurring when the two middle legs were amputated. In such an insect the fore- and hindlegs of the same side now alternated whereas they had moved simultaneously in the intact insect. Bethe & Woitas (1930) confirmed these observations in their work on the walking of Geotrupes, Dytiscus and Hydrophilus. In the locust, ten Cate (1936) drew attention to the distinct delay between movement of a foreleg and the contralateral hindleg following amputation of the two middle legs. He also observed this tetrapod rhythm in the cockroach, Periplaneta americana (ten Cate, 1941). About the same time as most of the experiments described below were being carried out, similar amputation experiments on Blatta were done by Estartús & Ponz (1951). They did not use cinematography and their observations agree substantially with those of Buddenbrock.

In the present work, the legs were amputated at the coxo-trochanteral joint and the insects photographed immediately after the operation. They were then allowed to walk about for at least 3 days before the same insects were filmed again. No significant difference could be detected between these two sets of photographs. Twelve different combinations of leg amputations were carried out, but most attention was directed towards the amputation of middle legs, as it is in this case that previous workers have thought the most marked changes occur. The principles observed in these experiments are applicable to other amputations which are described later and in less detail.

Apart from a tendency for the body axis to be inclined with the head slightly to the left of the direction of movement, these insects appeared to walk normally. However, the following modifications were found in analyses of photographs and tracings of leg movements of cockroaches from which the right middle leg had been removed (Figs. 2A,,3).

• (i) The right foreleg is retracted much further relative to the head before it is protracted.

• (ii) The right hindleg is protracted much further forward and not retracted so far back relative to the head.

• (iii) The legs on the right side are extended more lateral to the body than normally, while the left legs are held closer to the body axis. Both of these features are most marked in the hindlegs. On the injured side, these modifications will produce an increase in the lateral component of the strut effect which will help to compensate for the loss of the middle leg, which is normally very important in the maintenance of equilibrium about the longitudinal axis of the body.

• (iv) Table 1 shows that the stride length (distance moved by the limb tip relative to the head between complete retraction and protraction) is increased in both the hind- and forelegs on the side from which the middle leg has been removed. All figures in this table relate to the same insect.

• (v) A constant feature is the change in timing of the two legs of the intact pro- and meta-thoracic pairs. R₁ tends to move later than normal with respect to L₁ whereas protraction of R₃ takes place earlier in the cycle of L₃ than normally (Fig. 4).

• (vi) The rhythm of leg movements (Figs. 2 A, 3) varies in detail and appears to depend on the speed of the insect. When moving slowly (2 A) R₁ is protracted as soon as R₃ is in the supporting position. The three legs of the left side show the rhythm L₃, L₃, L₁. Protraction of L₁ is delayed until is placed on the ground. The same insect at a slightly faster speed (Fig. 3 B) uses an almost identical rhythm, but there is a distinct time lag between the instants at which R₃ obtains its point d’appui and when R₁ is protracted. This results in the two right legs alternating at nearly equal intervals of time after one another. At a still greater speed (Fig. 3 A), the insect uses the rhythm of a normal insect, but protraction of R₃ is usually delayed until R₁ is on the ground. There are occasional cases when both R₁ and R₃ are being protracted and there is no apparent support on the right side. This is not true, however, as R₃ is not lifted completely off the ground during protraction, but the body is displaced to the right because of the absence of any lateral thrust on the body by a leg on that side. While such differences in rhythm are found at different speeds it should be noticed that each rhythm remains fairly constant for a given run of at least 5 sec.

A more quantitative analysis of the different rhythms was obtained by considering the movements of the foreleg on the injured side in relation to the activity of the other foreleg and the ipsilateral hindleg. Normally R₁ alternates with L₁ so that protraction of occurs half-way between protractor movements of L₁. Thus in Fig. 4 A protraction of R₁ occurs between values of 0·4 and 0·6 (mean = 0·475) the ratio x/y. Similar values were also found for the hindlegs (Fig. 4B). Also in the intact insect we have the rule that R₁ protracts soon after R₂ is placed on the ground. In the absence of R₂, the movement of R₁ is modified and it is interesting to see how its activity is related to these two features of the normal co-ordination. From the curves showing the leg movements relative to the head it was possible to determine the instant at which R₁ was lifted with respect to (a) the interval between successive protractions of L₁, and (b) the interval between two instants at which R₃ is placed on the ground. The time of protraction of R₃ with respect to L₃ was likewise investigated. The measurements were made on seven different shots of the same insect running at different speeds along a straight path. The results (Fig. 4D–F) show that for a given run the phase relationship remains relatively constant but that there is some variation between different runs. The co-ordination of R₁ with L₁ is more constant than with R₃ which in turn is more co-ordinated with L₃. It can be seen that tends to be protracted later than half-way (i.e. x/y ⩾ 0·5) through the cycle of L₁ in twenty-two of the thirty measurements. In the normal insect R₂ supports the right side of the body while R₁ protracts, and hence this delay in protraction of R₁ will help to compensate for the absence of R₂. The earlier protraction of R₃(x/y ⩽ 0·5) which is clear from the histograms would also have this effect.

The distribution of protraction of R₁ with respect to R₃ (Fig. 4F) shows a large peak between 0–0·1. These are cases where R₁ protracted within one-tenth of a cycle of R₃ being placed on the ground. The absence of protraction within the next tenth of a cycle is interesting in that it may represent the time during which R₂ is normally protracted. The even distribution between 0·4 and 0·7 covers the range found for the intact cockroach (Fig. 4C). The minor peak at 0·8 followed by only a single case in the next category, shows that protraction of R₃ is delayed until R₁ is on the ground, and that it is rare for R₁ to be lifted during the brief period when R₃ is off the ground. In half of the cases examined protraction of R₁ or R₃ took place almost immediately the other right leg was placed on the ground. It is clear that the segmental co-ordination of R₁ and L₁ (and of R₃ and L₃) is more constant than the co-ordination of R₁ and R₃, but in all three cases it is less rigid than in the normal insect.

Such changes which follow the amputation of a single middle leg are related to a loss of the support which it normally provides. Modifications in the movements of the fore- and hindleg on the injured side enable them to replace the amputated leg functionally. This is most apparent at slow speeds, but even at the fast speeds both the position of the points d’appui and the length of the stride are altered. Static stability throughout the cycle is maintained as in the intact animal, but is reduced if the insect moves very rapidly.

It is possible to interpret the changed phase relations of some of the legs in terms of the known segmental reflexes. Pringle (1940) has confirmed Rijlants (1932) finding that an increase in the resistance to depression of a leg produces augmentation of the electrical activity in the depressor muscles. During retraction of in a cockroach without R₁ the proportion of the body weight supported by R₂, tends to be greater than normal ; its protraction relative to L₁ will be delayed, and consequently it will have a longer stride length and move further back relative to the head. At slower speeds, when static stability is rigidly maintained, R₁ does not protract until R₃ has taken up its supporting position. When this happens there will be a sudden decrease in the proportion of body weight supported by R₁ which will excite the levator reflex (Pringle, 1940). The way in which R₃ sometimes protracts as soon as R₁ is placed on the ground, suggests that the consequent decrease in its supporting function might also play a part in bringing about earlier protraction in this leg. It is also probable that some ipsilateral descending pathway is excited when R₁ is placed on the ground. These factors all contribute to the earlier protraction of R₃ relative to L₃. The alterations in timing of R₁ and R₃ relative to the contralateral leg of the same segment have functional advantages with respect to the transverse strut effects. It has been pointed out (Hughes, 1952a) that during retraction this component increases in a foreleg but decreases in a hindleg. Normally the contralateral middle leg antagonizes these effects, and in the absence of R₂, R₁ and R₃ take over this function. Later protraction of R₁ and earlier protraction of R₃ both have the result that their maximum transverse components will be exerted to oppose the action of L₁ and L₃. These increases in both the supporting function and lateral strut action of R₁ and R₃ probably produce an exaggerated ‘rebound ‘of the levator and protractor muscles which will give an increased stride length.

Further evidence for this sort of interpretation is given by a cockroach which had R₂ less well developed than normally. The rhythm showed some of the features noted above, particularly (i), (ii) and (v).

This operation had a much greater effect than the removal of only a single leg. The insects moved more slowly and there was a tendency for the body to touch the ground between the two remaining pairs of legs.

The limb movements are modified in the following ways:

• (i) The stride length of all legs is increased, so that observations (i) and (ii) of the five-legged insect apply to both sides of the body. This increase in stride length is most noticeable in the hindlegs which are protracted much farther forward. Their points d’appui are more lateral to the body than in the normal insect.

• (ii) The two legs of a segment are protracted at nearly equal intervals of time after one another.

• (iii) The rhythm of leg movements is changed. The graphs (Fig. 2B) show that the order of protraction is R₃, R₁, L₃, L₁, R₃, but this was not invariable. This rhythm is identical with that seen in a walking tetrapod.

Previous workers have rightly emphasized the immediate nature of these changes, but at first there is a lack of precision about them which disappears after a few days. This subsequent increase in efficiency is a subsidiary phenomenon, however, for the change in rhythm is instantaneous as is emphasized by the following observation.

If the two middle legs were amputated from a cockroach which was freely suspended from a steel wire attached to the thoracic tergites the remaining legs immediately performed active movements. Analysis of films of simultaneous side and dorsal views showed that the legs are often co-ordinated in the diagonal rhythm described above. The forelimbs were easily followed but the hindlimb movements were made up of the following phases :

• (i) Levation and movements forwards.

• (ii) Movement backwards.

• (iii) Depression.

• (iv) Flexion of the femoro-tibial joint.

The forelegs alternate normally and are related to movements of the hindlegs in that they commence to move forwards when the ipsilateral hindleg has completed phase (i). The rhythm of movements relative to the head is very similar to that described in the four-legged insect when walking on the ground. This co-ordination persists for about io sec., after which the movements become more spasmodic and apparently disco-ordinated. These experiments indicate some measure of intersegmental co-ordination in the absence of drag or changes in the distribution of weight.

One of the reasons for paying particular attention to the amputation of both middle legs was because previous workers maintained that the greatest alteration in rhythm followed this operation. This conclusion was based, however, on the view that the two tripods of support L₁, R₂, L₃and R₁, L₂, R₃ were the basic units of the normal rhythm, so that and L₃ moved simultaneously in one phase and R₁ and R₃ in the other. More detailed study has shown (Hughes, 1952 a) that, in the cockroach at least, the two sequences R₃, R₂, R₁ and L₃, L₂, L₁ are best considered as the basic units which overlap one another and only at the very fastest speeds does a foreleg move almost simultaneously with the hindleg on the same side. The fourlegged cockroach only moves at speeds comparable with the slower speeds of a normal insect and hence is best considered in relation to these. It is obvious that the rhythm R₃, R_1,L₃, L₁, R₃ results when the two middle legs are omitted from the rhythm R₃, R₂, R_1,L₃, L₂, L_1,R₃ which is the basic rhythm at very slow speeds.

The new rhythm obeys the same rules as the rhythm in the intact insect. In fact we shall find that the rule that no leg is lifted before the leg behind has taken up its supporting position is obeyed in all the other amputation experiments. The only exceptions found were the few cases mentioned above when the insect moved rapidly following the removal of a single middle leg. A number of exceptions were found to the other rule after some of the amputations described below, the two legs of a segment sometimes moving together or very nearly so.

These insects appeared to move in a fairly normal way but the body axis was usually inclined with the head to the left of the direction of movement (Fig. 5 A).

• (i) R₂ and R₃ are protracted farther forwards than normally and have their stride lengths increased relative to the contralateral legs.

• (ii) Legs on the right side, particularly R₃, are placed more laterally.

• (iii) Left legs move nearer to the body axis, especially the foreleg which often retracts beneath the prothorax. Stride of L₁ is usually increased.

• (iv) L₂ also quite often protracted far forwards and overlaps L₁. This is necessary to support the body when L₁ is lifted.

• (v) The rhythm of leg movements retains the basic components L₃, L₂, L₁ and R_3,R₂ with a gap left for R₁. The precise timing of these two basic rhythms varies, for example, L₁ may move before or after R₃. There is a modification, however, in so far as R₂ does not protract until L₁ has been placed on the ground and consequently R₂ and L₃ usually move simultaneously. This persists at quite fast speeds when normally the three legs of the tripod L₁, R₂, L₃ would move very soon after one another. This modification together with the altered position of the intact foreleg ensures that there is adequate support at the anterior end of the body when R₂ is protracted. If the insect used the normal rhythm, as supposed by previous authors, there would be an unstable phase when both L₁ and R₂ were off the ground.

These insects slipped on the ground a great deal but frequently showed a diagonal rhythm when walking slowly. More often the middle legs protracted together and the animal fell forwards. At fast speeds the diagonal legs moved synchronously and the head was pushed along on the ground (Fig. 5 B).

• (i) The two middle legs have a slightly increased stride and they are both protracted farther forwards than in the normal insect.

• (ii) The two hindlegs are also protracted farther forwards.

• (iii) All four legs are spread out much farther from the body axis. This general impression is emphasized by the relatively large distance separating the strides of legs on the same side.

• (iv) The rhythm is made up of the two basic components R₃, R₂ and L₃, L₂, but these may vary in their phasing with one another.

The two legs of a segment do not move at equal times after one another and in many cases the middle legs move almost simultaneously. As a result progression is jerky and the front end of the insect is alternately raised and lowered. In the prints this latter feature is very clear because the shadow cast from a light source behind the insect increases as both middle legs retract simultaneously and then decreases as first one and then the other protracts.

In some cases the diagonal rhythm R₂, L₃, L₂, R₃, R₂, L₃ is found but this is certainly not the most common. Protraction of the four legs is not evenly distributed throughout the cycle. This is even more marked at slower speeds. The importance of (i) and (ii) is well shown in Fig. 5 B. If R₂ and L₃ were not so far forwards when L₂ is protracting, the centre of gravity would fall outside the triangle of support as happens when the stride lengths shorten at faster speeds.

These insects were able to move very fast and seemed to be using the normal rhythm. However, the impression was given that L₃ was being dragged and that it exerted little propulsive thrust (Fig. 6A).

• (i) R₁ and R₂ are retracted farther back than normal.

• (ii) Both middle legs tend to be more lateral to the body.

• (iii) L₃ has a relatively short stride and is placed much nearer to the body axis.

• (iv) The basic units of the rhythm, L₃, L₂, L₁ and R₂, R₁, are maintained. In the rhythm shown in Fig. 6 A, protraction of R₂ is delayed until L₃ has taken up its supporting position. This results in the two middle legs often being moved very soon after one another, if not simultaneously.

This modification to the rhythm is again adaptive since L₃ is able to support the hind end of the body while R₂ is protracted. If it did not occur there would be a phase of the cycle when only R_1,L₁ and L₂ were supporting the body. The altered position of L₃ ensures that the centre of gravity falls within the triangle of support R_1,L_2,L₃ as shown in Fig. 6 A.

As was emphasized when considering the amputation of a single middle leg the precise rhythm varies, and another interesting rhythm is shown in Fig. 8 A. Here L₃ is not protracted until R₃ is on the ground. Thus instead of L₂ and R₃ moving together, L₁ and R₃ protract at the same time which is more like the normal rhythm. When R₃ is lifted it is interesting to see how L₃ pushes the body farther away from the point d’appui, presumably because of the sudden decrease in the tranverse strut effect of R₃.

Of all the four-legged cockroaches examined those with the two hindlegs removed were the least affected, in that they moved actively along a straight path and with no awkwardness about their limb posture and movement. However, they dragged the abdomen along the ground for most of the cycle. It is evident from Fig. 6B that during the protraction of a middle leg the centre of gravity falls well behind the area of support.

• (i) The stride length of all the legs is increased. This is especially true of the middle legs which are protracted farther forwards and retracted more than normally. Their position of maximum protraction is anterior to the position of the fully retracted forelegs.

• (ii) The action of the forelegs tends to draw the body towards the point d’appui so that the path of the limb tip relative to the body is not parallel to the body axis. This feature is also found with only a single hindleg amputated, and probably indicates that the forelegs are exerting more of a pull than normally. The impression is given that both middle and forelegs exert more of a pulling than a pushing action, i.e. they seem to lever the body forwards.

• (iii) The rhythm of the four legs is not always the diagonal rhythm R₁, L₂, L_1,R₂, R_1,etc., described by previous authors although this is certainly the most common, and the two sequences L₂, L_1,and R₂, R₁ are always present. At faster speeds it is usual for the two pairs of diagonal legs to alternate with one another as shown at the beginning of the rhythm in Fig. 6B.

Sometimes these insects did not seem to walk very well, the torsi being oriented abnormally with respect to the tibia. However, the legs were co-ordinated in a good diagonal rhythm (Fig. 7B).

• (i) The stride length of L_1,L₃ and R₂ is increased.

• (ii) R₂ and L₃ are protracted much farther forwards than in the normal insect.

• (iii) The point d’appui of L₃ is much more lateral to the body axis, and R₃ moves closer to the body. During retraction L₁ is also held close to the body.

• (iv) The typical rhythm is L₃, L_1,R₃, R₂, L₃.

It is rare for both pairs of diagonal legs to alternate with one another although quite often either and R₃, or R₂ and L₃ protract together at faster speeds giving the rhythms or ⁠.

These animals combine the modifications shown by insects with a single foreleg and a single middle leg amputated. The importance of these modifications is that they again ensure that the centre of gravity falls within the triangle of support throughout the whole cycle. Thus at frame 13 the anterior border of the triangle of support would be behind the centre of gravity if L₃ had not been protracted so far forwards. It is also interesting to notice how the path of R₂ changes direction after this phase, suggesting that the leg begins to exert an extensor thrust along its axis in place of a tractive function which is taken over by L₁ There can be little doubt that R₂ is replacing R₁ functionally.

The tracing and rhythms shown in Fig. 8 are of the same insect before and after amputation of L₁ in addition to R₃. Even after this operation the insect moves quite rapidly, but a tendency was noticed for it to fall forwards as the middle legs protracted simultaneously.

• (i) The most marked change takes place in the movement of L₂ which is protracted much farther forwards and has a greatly increased stride length.

• (ii) L₃ does not move quite so close to the body axis.

• (iii) R₁ and R₂ are more spread out and R₂ is not retracted so far as when only R₃ is amputated.

• (iv) The rhythm is usually R₁, L₂, R₂, L₃, R_2,L₂, etc. This is obviously not the diagonal rhythm.

There are moments of instability when only two legs are definitely on the ground. This takes place when R₁ is protracted and the body is supported chiefly by the two middle legs which are being retracted almost synchronously. The shadow on the photographs showed that the anterior part of the body is lifted up during this phase, and that shortly afterwards it falls (the photograph is blurred) as R₁ is placed on the ground. Frame 5 in Fig. 8 B shows this very well as in this case is not quite in the supporting portion as it usually is at this phase of the cycle. In the next frame (6) all the legs are on the ground and the insect extremely stable.

Except for some apparent dragging of L₃, these insects moved quite well and showed a good diagonal rhythm (Fig. 7 A).

• (i) L₁ is not protracted so far forwards and is retracted farther than normal. This leg is also more lateral to the body, particularly towards the end of retraction.

• (ii) Both R₁ and R₂ are retracted farther than in the intact insect. The stride length of R₂ is increased and it plays an important part in supporting the hind end of the body while L₃ is being lifted forwards.

• (iii) L₃ moves fairly normally but in some cases it is protracted a little farther forwards than normal. Immediately after the operation this leg was scarcely active.

• (iv) The rhythm is nearly always R₂, R_1,L₃, L_1,R₂ etc., with the legs protracting at more or less equal intervals after one another. But quite often there is a tendency for a given pair of diagonal legs to move almost simultaneously, e.g. L₁ and R₂ in the rhythm illustrated. At other times the opposite diagonal pair are more synchronized.

The significance of these changes can be seen at frames 6 – 7 when L₃ is being protracted. The centre of gravity only just comes within the triangle of support which would not happen unless features (i) and (ii) were present.

In these insects only the foreleg supported the body on the right side, and consequently the tip of the abdomen tended to touch the ground on that side. This was particularly noticeable during protraction of R₁. When moving rapidly the body axis was inclined with the head to the right of the line of progression (Fig. 9 A).

• (i) Without ciné analyses it is clear that the right foreleg retracted farther backwards than normally, and that the fore- and hindlegs on the intact side move closer to the body axis. In some cases the action of the single hindleg (L₃) is almost directly behind the abdomen (Fig. 9 A).

• (ii) Protraction of R₁ is delayed so that it is not lifted until after half-way through the cycle of L₁.The stride of R₁ is generally increased.

• (iii) Usually the sequence L₃, L₂, L₁ is shown, but at faster speeds L₃ sometimes moves at half the frequency of the other three legs.

• (iv) L₂ usually protracts as soon as R₁ is placed on the ground, but sometimes it is lifted while is off the ground.

The posterior displacement of the strides of R₁ and L₃, together with their positions being respectively farther from and closer to the body axis, ensure that the Une joining the points d’appuis of these two legs will pass behind the centre of gravity. This is particularly important shortly after these two legs are placed on the ground (Fig. 9A, frame 5).

Immediately after these amputations the insect still moved rapidly and did not appear badly affected as it scuttled along. But at slower speeds the movement was much more jerky and it fell to the right during protraction of R₂. As in other insects with only a single leg on one side, the head was inclined to the injured side of the direction of movement. These insects tired rapidly and then moved in small clockwise circles because activity of the single right leg was greatly reduced (Fig. 9B).

• Posture of the legs is modified so that the three left legs are close to the body, whereas the right middle leg is extended laterally. In fact L₁ tends to move under the body during retraction and this makes it difficult to plot out its movements from the film.

• The sequence L₃, L₂, L₁ is always present.

• The timing of R₂ with respect to the other legs varies but it usually takes place shortly after L₂ is placed on the ground, and hence almost simultaneously with protraction of L₁. Thus the rhythm is usually either L₃, L₂, R₂, L₁, L₃ or L₃, L₂, L₁, R₂,L₃.

• The centre of gravity falls within the area of support except for the phase just before and when R₂ is protracted.

Again these insects moved quite rapidly and the body axis was inclined with the head to the right of the line of progression. They tended to fall just before R₃ was protracted (Fig. 10A).

• (i) Ra was extended laterally and protracted farther forwards but did not retract so far as in the normal insect. The stride length of this leg was often greater than that of L₃.

• (ii) All three legs on the left side were held closer to the body. In addition L₂ is protracted farther forwards than normally.

• (iii) The sequence L₃,L₂,L₁ was well defined. R₃ often protracted simultaneously, or shortly before, L₁ giving the order of lifting L₃, L₂, R₃, L₁, L₃, etc.

Analyses of superimposed frames showed that R₃ protracted as soon as the centre of gravity passed in front of the line between the points d’appui of L₁ and R₃, which is the instant at which the insect falls to the right. When R₃ is placed on the ground again the area of support is bounded anteriorly by the line L₂, R₃, which passes just in front of the centre of gravity, because these two legs are protracted farther forwards than normally. Once again, these features suggest that mechanical factors are concerned in the modifications in posture and rhythm.

The operations were carried out on anaesthetized cockroaches pinned ventral side uppermost. The thoracic ganglia are connected together by pairs of commissures which are visible through the thinner regions of the sternum. A small incision was made in this cuticle and a fine glass hook inserted beneath one of the commissures which was pulled out and severed with a fine pair of scissors. Body fluid was lost during the operation but the insects recovered within a few hours in the majority of cases. Altogether about thirty cockroaches were operated upon in this way. Films were taken at least 24 hr. after the operation.

In any experiment involving operations such as these it is often difficult to distinguish the effects produced by the specific operation from those resulting from the general injury. The results described below are the most common and are considered to represent the result of the interruption of central nervous pathways.

These operations were nearly always successful and when observed the insects appeared to move normally. However, the films revealed changes in the rhythm.

For instance, in a male Blatta with the right pro-mesothoracic commissure cut, the legs of the uninjured side showed a perfect rhythm L₃, L₂, L₁ and this was true of the right side to some extent, but sometimes R₁ fell out of the rhythm. Even when both sides showed their normal rhythms (R₃, R₂, R_1,L₃, L₂, L₁) R₁ was again slightly out of time in that it protracted a little early and therefore moved before L₃, instead of after it as in the normal rhythm (Fig. 10B, ii).

In insects with a single meso-metathoracic commissure cut the hindleg of the operated side fell out of rhythm. In an insect with the right commissure cut the sequence L₃, L₂, L₁ was present throughout, but the legs of the right side, although observing the rule that no leg is moved before the one behind has taken up its supporting position, showed an altered rhythm as R₃ was very often protracted before R₁ (Fig. 10 B, i). Other preparations have the hindleg very inactive unless the insect walks on sandpaper when it moves in a rhythm of its own and not very well co-ordinated with the other legs (Fig. 10 B, iii).

This experiment is analogous to those carried out by Friedlander on earthworms which were cut in half, the two halves being connected by pieces of cotton. He found that the mechanical drag imparted to the posterior half by the anterior region was sufficient to enable the peristaltic wave to cross the gap. Similarly, activity in the two halves was co-ordinated when the nerve cord was the only connexion between the two. He concluded that both nervous and mechanical stimulation were important in bringing about the passage of the peristaltic wave.

In the cockroach the co-ordination of movement is much more seriously affected by the cutting of two as compared with the cutting of a single commissure. If both meso-metathoracic commissures are severed there is often a loss of depressor tone and the hindlegs do not touch the ground, presumably due to loss of connexion with the head. In favourable preparations, when the legs touch the ground, the hind pair of legs alternate in a slower rhythm than the other four legs, which continue in the normal rhythm. No definite co-ordination between the legs in front of the operation and those behind has been observed. Sometimes, however, when one of the hindlegs becomes excited (‘spontaneously’, by being touched with a brush, or by pressure on the abdomen) it pushes the body forwards. When this happens the three legs of that side show the rhythm—hind, middle, foreleg. This confirms the suggestion that when a leg takes up its supporting position the leg in front is protracted, because there will be a decrease in the proportion of the body weight which it supports. The conditions are similar to those operating when Pringle (1940) suddenly decreased the resistance to depression, the depressor muscles becoming less active.

These results are in agreement with those of Baldi (1924) and Ponz & Estartús (1951) who found no co-ordination between the hind pair of legs and the anterior ones when both commissures were severed. Ten Cate (1928, 1936), however, claimed that there was some co-ordination under these conditions. A possible explanation of these differences may be in the completeness of the separation effected between the two ganglia. There are, in addition to the two main commissures, two thinner ones which definitely contain large nerve fibres connecting the two ganglia, as has been observed in horizontal sections of the nerve cord. Hence, it is possible that only the two main commissures were cut in the experiments of ten Cate. Buddenbrock (1921) using Dixippus, and Roeder (1937) with Mantis, were unable to find any co-ordination when both commissures were severed, and concluded that mechanical drag was insufficient to bring about coordination in these insects.

Ten Cate (1941) carried out an interesting series of experiments involving the cutting of various combinations of commissures in cockroaches from which both middle legs had been removed. He found that the diagonal rhythm of movements persisted so long as there was a nervous path connecting the prothoracic and metathoracic ganglia. There appeared to be no specificity concerning this path as the same rhythm remained if both commissures on the right or left side were cut, or if the right pro-mesothoracic and left meso-methoracic commissures were severed. If both commissures between any two ganglia were cut there was little co-ordination between the two pairs of legs, even when the insect was held in contact with a rotating drum. Some of these striking observations have been confirmed.

These experiments upon four-legged insects emphasize the importance of nonspecific pathways within the thoracic chain, and are in contrast to the results of the observations reported above on six-legged insects. In the latter experiments the maimer in which the fore- or hindleg on the same side as the cut commissure protracted a little in advance, suggests that definite pathways have been cut which are normally of importance in regulating the timing of the leg movements. It must be remembered, however, that the change in the six-legged insect is a relatively slight one and that a small alteration in the timing of the leg movements of a fourlegged cockroach would not be so noticeable because of the fewer legs. The two sets of observations are by no means inconsistent and, in fact, give an indication of the type of co-ordinating mechanism at work.

The general significance of modifications in locomotory movements following the amputation of legs was emphasized by Bethe (1931) in his ‘Plasticity theory of the central nervous system’. In this theory he drew attention to the ability of the central nervous system to produce new co-ordination patterns appropriate to the altered conditions. As the transpositions take place almost immediately, it is improbable that learning is involved and suggests that the neurological basis of the new patterns existed before the operation or were produced as a direct result of it. Bethe dismissed the first possibility because of the very large number of preformed patterns which appeared to be necessary to account for co-ordination following the removal of different combinations of legs. This is certainly true if each type of co-ordination has a completely different neurological basis, as the number would be 2ⁿ – 2, where n is the number of legs (Bethe 1930). Instead he suggested that the central nervous system was able to produce suitable co-ordination patterns within itself as a result of the operation. This was a radical change of outlook when compared with the then accepted chain reflex theories which regarded co-ordinated movements as a rigid sequence, each movement being produced reflexly by stimulation coming directly from the previously-acting leg. Although Bethe did not suggest a mechanism whereby such changes might take place, he pointed out that a mechanistic explanation in terms of ‘a general regularity of the distribution of stimuli did not appear to be excluded’. Greater knowledge of proprioceptive mechanisms now makes such an explanation probable. It is not necessary to assume a different wiring diagram for each amputation since variations in activity of the remaining components is sufficient to account for the observed modifications. Thus, it would seem better to consider these plasticity changes as due to the operation of the same reflex mechanisms which produce co-ordinated movements in the normal insect. The altered mechanical conditions following the amputation of legs will change the sensory inflow, not only because of its absence from a given leg but also because of alterations in the excitation of sense organs in the remaining legs and other parts of the body, such as the intersegmental membranes (Diakonoff, 1936).

The results presented in this paper fit in well with such an interpretation since, in contrast to previous work, it has been shown that modifications take place in both posture and rhythm when only a single leg is removed. Examples have been given to show how some of these can be interpreted in terms of the known segmental reflexes. Furthermore, as each amputation produces a unique situation it is to be expected that the co-ordination and posture will likewise be unique, as has been found. Fig. 11 summarizes the results of the twelve different combinations in which one or two legs have been amputated from a cockroach. The altered path of the limb tips indicates how the posture of the legs is modified. In all cases there is a tendency for the remaining legs to take over the supporting function of the amputated legs. The legs adjacent to those amputated are usually the most affected, but changes in posture and rhythm occur to some extent in all of them. The total range in posture of a given leg is quite remarkable and must involve adjustments in the activity of many muscles. Similar changes in limb posture were described in the centipede Scutigera by Lissmann (1935).

Any interpretation of these modifications is bound to be incomplete until a great deal more is known of the mechanical conditions operating, and the precise mode of functioning of the different groups of campaniform sensilla described by Pringle (1938b). The reflex effects of stimulation of the hair plates (Pringle, 1938c) and of the widespread chordotonal sensilla which can have a proprioceptive function (Hughes, 1953) also remain to be discovered. However, in a general way it can be seen how the reflex mechanisms of the normal insect operating under changed conditions would produce the observed modifications in posture.

It is impossible to make any generalizations which will always enable one to forecast the rhythm adopted by the remaining legs when one or more legs are amputated, but certain features are usually recognizable. With extremely few exceptions it has been found that, as in the normal insect, no leg is lifted before the leg behind it on the same side is placed on the ground. The delay between these two events varies, however, and the phase relationship of the sequences on the two sides of the body are also variable. Thus two legs of a segment usually alternate but they may move almost simultaneously. Most of the rhythms can be produced by varying one or more parameters of the normal rhythm (e.g. ⁠, phasing of two legs of a segment, etc.). For instance, the often-quoted transformation to the diagonal rhythm following the removal of both middle legs results if these legs are omitted from the normal rhythm in a slowly-moving cockroach, where p/r is about ⁠.

Previous workers, without the use of ciné-analyses, have held that the diagonal order of lifting is always found in a four-legged insect. In many cases this has been confirmed for the cockroach, but it is certainly not invariably true nor is the rhythm quite the same for any two amputations. Sometimes for instance, the four legs do not move at equal intervals of time after one another, one pair of diagonal legs often tending to move synchronously but not the other pair. Furthermore, the diagonal order of lifting (e.g. R₁,L₃, L₂, R₂, R₁) is not always the most common after a particular amputation and even when most prevalent it is often interrupted by other rhythms. In some cases, of course, the order of lifting in four-legged insects remains the same as in the intact insect, yet produces a diagonal rhythm (e.g. amputation of R₁ and L₁,or R₃ and L₃). Yet in some of these cases also, other rhythms are found which suggest that the normal function of the legs removed and the position of the centre of gravity with respect to the remaining legs are important factors which must be taken into consideration, and these will vary according to the amputation.

It is difficult, therefore, to agree with generalizations which suggest that the diagonal rhythm is invariably found in all instances when two legs are removed from an insect, but it is certainly the most common. The same is true of insects in which it is normal for only four legs to be used in walking. Thus a notable exception has been described by La Greca (1947) in the Acridiid, Tropidopola cylindrica which at slow speeds uses the rhythm etc., or etc., and at faster speeds limb pairs move synchronously in the rhythm ⁠. The tip of the abdomen provides an additional point of support during these movements which is not surprising as the centre of gravity must be quite far back in these insects. The importance of the tip of the abdomen for support was also pointed out by Wille (1924) in Rhipipteryx chopardi which uses only the front four legs in walking. In this insect the legs were lifted in a diagonal rhythm, i.e. R₁, L₂, L_1,R₂, R₁ etc., but usually only two legs were supporting at any one time and the triangle of support was completed by the tip of the abdomen.

The significance of the sequence in which the four legs are lifted in a tetrapod has been considered in detail by Gray (1944). He showed that, of the six possible sequences, the typical rhythm is the only one in which static stability is maintained throughout the cycle. This is only true when the four legs are distributed evenly about the centre of gravity. It is of interest, therefore, that many of the modifications in posture described above will tend to satisfy this condition. Furthermore, in those amputations where an even distribution cannot be achieved, it is to be expected that rhythms other than the diagonal pattern will be present. An example of this is shown by cockroaches without the two hindlegs, where rhythms similar to those described for Tropidopola have been observed. The relationship between rhythm, posture and the position of the centre of gravity with respect to the changing area of support (found in analyses of films), indicates that static stability tends to be maintained throughout the cycle as a result of modifications following the amputation of legs. That this is not always achieved is because it is sometimes physically impossible. It appears, then, that modifications of rhythm, as of posture, can be interpreted in terms of the altered mechanical conditions resulting from a given amputation.

Von Buddenbrock (1921) considered this explanation but rejected it in the light of an experiment with a four-legged Dixippus, in which the middle leg stumps were allowed to rest upon a small platform attached beneath the thorax. The rhythm of leg movements in such an insect was the same as in the normal insect, i.e. it did not show the diagonal sequence. It is important to notice that in this experiment the middle legs were amputated in the middle of the femur. The trochanter remained intact and inspection of this joint has confirmed the expectation that it would be provided with groups of campaniform sensilla and hair plates. Hence, the proprioceptive inflow from these legs could still take place, at least qualitatively, as in the normal insect. The timing of this inflow would be normal as the stumps tend to move in the normal sequence. This interpretation received support from experiments on Dixippus in which the middle legs were amputated proximally to the trochanter. Despite their small size it was possible to arrange for the coxae to move on the platform, and in all cases the remaining legs showed a diagonal rhythm. In his recent textbook von Buddenbrock (1953) describes a further experiment by Schaller (as yet unpublished) in which an insect holding a piece of paper with the two middle legs also showed the diagonal sequence. Here the normal timing of impulses from the middle legs would not be present, as is also true if the middle legs are rendered immobile by anaesthesia or by fixation of the joints. Under these conditions the insect moves as though the legs were removed, as was also shown in crabs by Kühl (1931).

Thus, it would seem that so long as there are only four legs in contact with the substratum the integration of their proprioceptive inflow tends to produce the diagonal rhythm, provided the legs are distributed evenly about the centre of gravity.

In addition to the evidence discussed above, the following observations on normal insects also indicate the importance of mechanical factors in co-ordination.

• (i) If a small weight is placed eccentrically on the metathorax of a cockroach it is found that the middle and particularly the hindleg on that side are displaced more laterally to the body. This effectively increases the lever arm of these legs so that equilibrium in roll will be maintained without any disproportionate increase in the vertical force operating at the tarsi of these legs. These changes in posture are similar to those which occur when a single middle leg is removed from the insect.

• (ii) The paths of the limb tips relative to the head in a normal insect are parallel to the axis of progression, and this also indicates a mechanism balancing the distribution of forces exerted by the legs on the ground. Some preliminary measurements of the vertical forces operating at the tarsi during normal movement confirm this conclusion.

• (iii) When studying the geotactic responses of Tetraopes and other insects, Crozier & Stier (1929) found that the angle at which they walked up an inclined plane was precisely related to its angle of inclination. They suggested that this was brought by some mechanism whereby ‘the tension excitations due to the pull of the animal’s weight are the same, within a threshold difference on the two sides of the body’.

These observations suggest that during movement the forces at the tarsi tend to be equalized on the two sides of the body. The campaniform sensilla are probably of great importance in this mechanism since their distribution will enable them to detect such forces (Pringle, 19386). The hypothesis is suggested that the altered postures and movements described in this paper tend to produce an equal inflow of impulses from these and other proprioceptors on the two sides of the body. It is reasonable to suppose that the central nervous system has a mechanism for comparing the inflow from the two sides, and in these cases produces compensating movements which tend to equalize the feedback for these sense organs.

From the considerations presented in this paper it is suggested that the following factors are of importance in determining the co-ordination of insect movements.

1. The mechanical drag imparted by the other legs. The importance of this factor was emphasized by ten Cate (1928, 1936), but that it is insufficient by itself is indicated by the absence of co-ordination when both meso-metathoracic commisures were severed.

2. The proportion of the body weight which a leg supports. This may be decreased by the leg behind taking up a supporting position or increased by the protraction of adjacent legs. The presence of the sequences L₃, L₂, L₁ and R₃, R₂, R₁ will be maintained by this factor.

3. It is suggested that there is an overall co-ordination between thoracic ganglia regulating movements of all legs, such that feedbacks indicating components of forces acting at the tarsi will be balanced on the two sides of the body.

4. Segmental reflexes objectively demonstrated by Pringle (1940) are of basic importance in all these reactions.

5. Intersegmental effects, whose exact nature is undefined, are indicated by several experiments such as when a single commissure is cut in the intact insect and distinct pathways appear to be disrupted. Intersegmental reflexes have not been demonstrated objectively, although some indication of their presence was found in experiments involving stimulation of cockroach ganglia with direct currents (Hughes, 1952b). Some recent observations by D. M. Maynard (personal communication) have also indicated the presence of such pathways. Thus, electrical stimulation of sensory nerve fibres entering the mesothoracic ganglion produces a response in the muscles of the foreleg on the same side more regularly than in the contralateral foreleg. Such pathways would facilitate the sequence hind-, middle, fore- of ipsilateral legs.

6. The supra- and sub-oesophageal ganglia have an overall effect upon the thoracic ganglia since co-ordinated movements in the strict sense are not present when they are extirpated (Baldi, 1924).

In the light of this discussion insect movement becomes an extremely complex interaction of reflexes most of which reinforce one another. This stress on the role of peripheral sense organs has been justified in the locomotion of both amphibia and fish (Gray 1950, Lissman, 1946a, b), but other workers (see S.E.B. Symposium, 1950) have emphasized the importance of intrinsic factors within the central nervous system in determing the patterns of movement independently of the peripheral inflow. Bethe’s original theory attributed considerable executive properties to the central nervous system, though few people would now accept the suggestion that it was able to ‘invent’ new patterns so quickly. Weiss (1950) has criticized this aspect of Bethe’s theory, but is prepared to accept the presence of pre-existing mechanisms capable of producing the new co-ordination patterns. He accepts plasticity on the ‘intermember’ but not on the ‘intramember’ level, but we have seen that there are adaptive changes in the degree of activity of different muscles within a limb as well as changes in the rhythm of leg movements. However, we must obviously accept the existence of certain preformed tracts and synaptic connexions within the central nervous system, but this does not necessarily mean that these pathways determine the pattern of movements; rather, they are the means by which patterns of peripheral stimulation entering from different sense organs become integrated with one another. The relevance of such observations to current concepts of instinctive behaviour has been considered by Thorpe (1948), and the results described in this paper together with unpublished observations on the swimming of Dytiscus, support his conclusions in that they throw some doubt on the physiological basis of these concepts as applied to insect locomotion.

I wish to thank Prof. Sir James Gray, F.R.S., for suggesting this problem, and him and other members of the Department of Zoology, Cambridge, for their helpful comments and criticisms.