Analysis of the Mechanical Responses of Metathoracic Extensor Tibiae Muscles of Free-Walking Locusts

Contractions of the metathoracic extensor tibiae muscle (METM) of the locust are controlled by two excitatory neurones, designated ‘fast’ and ‘slow’ respectively (Hoyle, 1953 a) and an inhibitory neurone (Usherwood & Grundfest, 1964, 1965). The ‘fast’ excitatory neurone is used infrequently, i.e. only during rather specialized activities such as jumping and some forms of kicking. During other activities such as walking, standing or climbing the insect uses only the ‘slow’ excitatory and inhibitory neurones for controlling the mechanical responses of the METM (Runion & Usherwood, 1968). Recent investigations of the nervous control of the METM in the locust, Schistocerca gregaria have shown that the level of activity of the ‘slow’ excitatory neurone is influenced by information from two major sensory systems located in the metathoracic leg, a phasically and tonically responsive chordotonal organ found in the femoral segment (Runion & Usherwood, 1966 a; Usherwood, Runion & Campbell, 1968) and phasically responsive sensilla trichodea located on the tarsal segments (Galloway, Runion & Usherwood, 1966; Runion & Usherwood, 1968). The phasic input from the tarsal sensilla appears to be important in determining the level of activity of the inhibitory neurone, and to a lesser extent, the ‘slow’ excitatory neurone. The phasic and tonic inputs from the chordotonal organ both influence the activity of the ‘slow’ excitatory neurone but they have little influence on the activity of the inhibitory neurone.

In the METM of Schistocerca the inhibitory innervation is mainly restricted to a small bundle of fibres at the proximal end of the muscle (Usherwood & Grundfest, 1965). Usherwood (1967) and Cochrane, Elder & Usherwood (1969) have shown that these fibres are tonically responsive whereas the rest of the fibres in the METM are phasically responsive. When the insect is standing still, the ‘slow’ excitatory and inhibitory neurones are both active and presumably control the tonic state of the METM at this time. The inhibitory neurone under these conditions could have three functions. It could attenuate the contractions of the METM that are evoked by the ‘slow’ excitatory neurone. It could accelerate the relaxation of the METM following a ‘slow’ contraction and it could reduce the resting tension of the tonically responsive fibres of the METM. In performing the latter function it could re-prime the contractile system of the tonically responsive fibres (Usherwood, 1967).

During walking, activity of the METM alternates with that of the antagonistic flexor tibiae muscle, bursts of ‘slow’ excitatory impulses to the METM occurring in the periods between tibial flexion. The inhibitory neurone to the METM is inactive except just prior to flexion of the tibiae when it fires 1−4 times. Most of this information has been obtained using free-walking locust preparations (Runion & Usherwood, 1966 b) which can give information on the activity both of the nerves and of the muscles of this insect. Studies with these preparations have been mainly directed towards evaluating the neuronal mechanisms underlying the activities of the femoral muscles of the metathoracic leg of the locust although investigations of the meso-thoracic and prothoracic legs are currently in progress. Since the overt behaviour of the locust is to a great extent an expression of the forces developed by its muscles it would be unwise to explain the behaviour of this insect solely in terms of the frequency of motor and sensory nerve impulses and of the electrical responses recorded from its muscles. The mechanical responses of the METM, with its complex innervation and two types of muscle fibres, certainly could not be predicted on the basis of its nervous input. It is thus unfortunate that it has not yet been possible to devise a suitable method for recording the changes in tension and/or length of skeletal muscles in free-walking preparations without drastically altering the behaviour of the locust by loading the appendages with a mechanical recording system. Attempts have been made to circumvent this problem by using indirect methods of assessing the mechanical responses of the METM of free-walking locusts which do not involve loading the appendages. The results of studies using these techniques are made the subject of this publication and provide further insight into the functional role of the inhibitory innervation of the METM of the locust.

Both intact restrained preparations (Usherwood, Runion & Campbell, 1968) and free-walking adult preparations (Runion & Usherwood, 1966b) of the locust, Schisto-cerca gregaria were used in this investigation. The locusts were supplied as last instar nymphs by the Anti-Locust Research Centre, London and maintained at 32°C on a diet of Bemax and fresh washed grass. All the electrophysiological studies were made at 22°C.

Three to eight pairs of recording electrodes were implanted into one side of the metathorax and into the metathoracic leg of that side of the insect. Electrodes implanted into the tibial segment of the leg monitored the activity of the tarsal sensilla trichodea, the tarsal campaniform sensilla (Pringle, 1938), the tarsal chordotonal organ and the tibial spines (Runion & Usherwood, 1968). Electrodes placed in the metathorax on nerve 3b monitored the activity of the slow excitatory neurone and the inhibitory neurone innervating the METM. Electrodes on nerve 5 recorded afferent signals from leg receptors including the femoral chordotonal organ and the tarsal receptors. The electrical activity of the femoral muscles was monitored by extracellular electrodes implanted in the flexor tibiae muscle and the METM.

The electrical information obtained using these recording electrodes was amplified and displayed on a Tektronix 565 oscilloscope and recorded photographically using a Grass C4 camera. The insect was photographed throughout the recording period by a second Grass C4 camera as described by Galloway, Runion & Usherwood (1966). The photographic and electrophysiological recording systems were synchronized as illustrated in Fig. 1.

The largest efferent potentials recorded from nerve 3b were from the ‘slow’ excitatory and inhibitory neurones which innervate the METM. Since the excitatory potentials (c. 50−60 μV) are usually larger than the inhibitory potentials (c. 20−25 μV), and since the two types of potentials have different time courses, it is relatively easy to separate them from each other. However, to eliminate any possibility of confusing the two types of nerve potentials, the following procedure was adopted at the end of each free-walking experiment. The free-walking preparation was fixed on its back and the METM was exposed without altering the positions of the recording electrodes in the metathorax. Further recordings were then obtained from nerve 3b, the ‘slow’ excitatory and inhibitory neurones being excited reflexly by touching the tarsi on either the ipsilateral or contralateral metathoracic legs. At the same time intracellular recordings from fibres of the exposed METM were obtained using glass intracellular microelectrodes filled with either 3M potassium chloride or potassium propionate. The inhibitory postsynaptic potentials recorded from the METM are normally hyperpolarizing responses (Usherwood, 1968) whereas the ‘slow’ excitatory potentials are normally depolarizing responses. Therefore, by comparing the intracellular records from the METM with the extracellular records from nerve 3b, it was possible to distinguish clearly between the excitatory and inhibitory nerve impulses.

The ‘slow’ excitatory and inhibitory data could then easily be abstracted from the neurograms obtained from nerve 3b. It is estimated that in most experiments up to 90% of the ‘slow’ excitatory and inhibitory information was successfully retrieved from the neurograms obtained from the free-walking preparations. It was only during periods of high-frequency discharge that a significant amount of information might have been lost.

Details of the activities of the ‘slow’ excitatory and inhibitory neurones were abstracted from the electrical recordings from nerve 3b of free-walking preparations (hereafter termed ‘donor’ preparations) as described above. A dual track tape programme was then prepared from this abstracted information, with the ‘slow’ excitatory information on one track of the magnetic tape and the inhibitory information on the other track, the temporal relationships between the two sets of events being carefully maintained. The magnetic tape was programmed by manual manipulation of a tape recording head. The tape head was fixed to a Perspex block designed to travel on a U-shaped track over instrument-grade magnetic recording tape. The magnetic tape was held under the U-track in a guide (tolerance ± 0·05 mm.) 250 mm. in length. The electrical activity of nerve 3b in the donor preparation was recorded photographically at a film speed of 250 mm./sec. It was therefore convenient to transfer inhibitory and ‘slow’ excitatory information which was recorded during a period of one second directly from the film records to the magnetic tape. The excitatory and inhibitory information was transferred to the magnetic tape by pulsing the appropriate tape head with a brief 10 KHz sine wave. The ends of the programmed tape were then spliced together making a continuous loop so that the programme could be repeated cyclically (estimated error ±0·1 msec.).

Programmed tapes were used to stimulate the ‘slow’ excitatory and inhibitory neurones of restrained locust preparations (hereafter termed ‘recipient ’ preparations) (Fig. 3). Ideally, the same locust should be used as donor and recipient but attempts to do this were not very successful mainly due to the long time taken in processing and analysing the data obtained from the donor insect and in transferring the relevant information to magnetic tape. Therefore all the results described in this publication were obtained using two different insects, one a donor the other a recipient. Every effort was made to prepare the recipient locust in exactly the same way as the donor locust, i.e. with a thoracic harness, and with recording electrodes implanted in the appropriate regions of the metathorax and metathoracic leg. The recipient preparations did, however, have an additional set of electrodes placed on nerve 3c for ‘reflex’ stimulation of the inhibitory neurone to the METM (Usherwood & Grundfest, 1965).

A recipient insect prepared in this way was fixed down on its back with Tackiwax with only the tibial segments of the leg free to move. The tibial segment was then attached by a metal shaft to the anode pin of an RCA 5734 mechano-electronic transducer. Finally, nerve 5 of the recipient preparation was cut to prevent autotomy of the leg during stimulation with the programme on the magnetic tape.

The tape programme was played on a Roberts 997 tape deck and the two outputs (inhibitory and excitatory) were processed through a pair of micrologic Schmitt triggers and a pair of micrologic stimulators (Runion, H. I. to be published) to convert them to pulses suitable for neural stimulation. These pulses were applied to nerves 3b and 3c respectively of the recipient preparations via stimulating electrodes (Fig. 3). The duration (1 μsec. −100 μsec.) and amplitude (1−9 mV) of the pulses from the stimulators could be varied independently. During stimulation of the recipient preparations with the tape programmes the outputs of the stimulators were displayed on a multibeam oscilloscope together with the output (electrical correlate of muscle tension) from the mechano-electronic transducer. Although this is an indirect method for investigating tension changes in the METM of ‘free-walking’ locust preparations reasonably accurate estimates of the mechanical responses of the METM can be obtained. The technique has been of particular value in determining the effects of peripheral inhibition on the ‘slow ’ mechanical responses of the METM.

The METM of a recipient preparation was stimulated in three different ways: (a) with the entire programme on the tape loop; (b) with only the inhibitory programme; and (c) with only the ‘slow’ excitatory programme. Errors resulting from possible delays in the central nervous system during ‘reflex’ stimulation of the inhibitory neurone (Usherwood & Grundfest, 1965) are not considered important in this type of analysis, since other experimental errors introduced by the donor-recipient type of analysis are probably of far greater magnitude. Also if the inhibitory neurone is a branched structure with one branch in nerve 3 b and another in nerve 3 c as in the cockroach (Bergman & Pearson, 1968) there would in any case be no central synaptic delay resulting from stimulation of the inhibitory neurone via nerve 3c.

At the end of each experiment a check was made, using the intracellular recording technique described above, to determine whether the ‘slow ’ excitatory and inhibitory neurones of the recipient preparations responded exactly in accordance with the commands on the tape programme. Since this involved the removal of the flexor ibiae muscle the mechanical properties of the system were altered and the preparations could not be used again for testing other programmes.

There are, of course, a number of problems associated with the donor-recipient technique. For example the responses of different locust preparations to the same programme often varied. Also the ‘slow’ excitatory and inhibitory neurones in the recipient insect sometimes fired spontaneously during programmed stimulation. These additional responses, especially when they were excitatory, were not always detected until the photographic records of the mechanical responses of the METM had been examined. It was usually then too late to repeat the programme on that particular insect preparation. The difficulties of this technique must not be overemphasized, however, for some sort of success was achieved with at least 50% of the recipient preparations. The technique is, nevertheless, very time-consuming and for this reason an alternative quicker method of estimating the mechanical responses of the METM of the free-walking preparations was devised which did not involve using recipient preparations. This method, which is described below, is much less accurate, however, than the donor-recipient technique.

An electronic simulator (Fig. 4) was designed to reproduce the electrical and mechanical responses of the METM that occur during stimulation of the ‘slow ’ excitatory and inhibitory neurones. Its operation can be described briefly as follows. The ‘excitatory’ and ‘inhibitory’ information from the programmed tape loops was converted to pulses of constant amplitude by two Schmitt triggers and then injected via diodes into two separate monostable multivibrators. The monostable circuits were used to set the durations of the inhibitory and ‘excitatory’ pulses so that they were comparable to the ‘slow ’ excitatory and inhibitory postsynaptic potentials recorded from fibres of the METM with intracellular microelectrodes, i.e. 40−300 msec, for the inhibitory pulses and 15−150 msec, for the excitatory pulses (Usherwood & Grundfest, 1965). The slopes of the simulated excitatory postsynaptic potentials (EPSPs) and inhibitory postsynaptic potentials (IPSPs) were determined by passive integration of the outputs of the monostable circuits through resistance-capacitance networks. The facilitatory properties of the EPSPs and IPSPs were simulated by varying the pulse currents transferred to the capacitance of these networks by means of 10K variable resistors and silicone diodes selected for their forward conductance properties. Summation of the simulated inhibitory and excitatory potentials was achieved by varying the time controls of the integrators whereas interaction between the inhibitory and excitatory potentials was simulated using an operational amplifier in a ‘subtractor’ configuration. However, before this could be done accurately it was necessary to correct for the differences in amplitudes which normally occur between EPSPs and IPSPs in the locust METM, since the amplitudes of the outputs from the integrator circuits of the simulator are roughly identical (±0·1 mV.) This was achieved by placing voltagedivider networks in the emitter circuits of buffer amplifiers separating the passive resistance-capacitance networks and the operational amplifiers. The buffer amplifiers also isolated the operational amplifiers (100 Ω) input impedance) from the integrator circuits (5 Ω) input impedance). To obtain simulated mechanical responses operational amplifiers were coupled to the outputs of the ‘potential’ simulator, and were used as imperfect integrators.

The simulated electrical and mechanical responses are in many respects similar to those recorded from the METM of the locust (Fig. 5). For example, when the simulator is activated with information contained on a programmed tape loop the outputs from the ‘potential’ and ‘tension’ circuits are very similar to those obtained from the recipient preparations. The major deficiency of the simulator is that the output from the inhibitory circuit is always a simulated relaxation whereas relaxation of the METM during inhibitory stimulation alone occurred in only about 30 % of the recipient preparations. Nevertheless the simulator provides a convenient means of estimating the electrical and mechanical properties of the METM during different behavioural activities.

In Fig. 6 the mechanical responses recorded from the METM’s of two recipient preparations during stimulation of the ‘slow’ excitatory and inhibitory neurones with two different programmes are illustrated. The tape programmes were prepared, in this instance, using data obtained from a donor insect when it was standing still. At this time the mechanical state of the METM is controlled by the interplay of the ‘slow ’ excitatory and inhibitory neurones. By comparing the tension induced in the recipient METM by both the combined ‘slow ’ excitatory and inhibitory components of a tape programme with that induced by the ‘slow ’ excitatory component alone it was possible to gauge the effect of peripheral inhibition on the ‘slow’ contractions of the METM. In all cases this effect was a reduction in the magnitude of the ‘slow’ mechanical response, though this reduction rarely exceeded 10%. This is not particularly surprising, however, for only a small minority of the METM fibres innervated by the ‘slow’ excitatory neurone receive inhibitory endings (Usherwood & Grundfest, 1965). It would be of interest to repeat these experiments using the American lubber grass-hopper Romalea microptera since in the METM of this insect all the fibres innervated by the ‘slow’ excitatory axon receive inhibitory endings (Usherwood & Grundfest, 1965). However, in both Romalea and Schistocerca all the tonically responsive fibres are innervated by the inhibitory neurone and since tension changes rather than length changes are usually involved during standing behaviour, it is probable that these fibres are very important at this time. In the majority of the recipient preparations the ‘resting’ tension of the METM was apparently unaltered when the insect was stimulated with only the inhibitory component of the tape programme, although in a few preparations a slight fall in the ‘resting’ tension of the METM occurred. Possibly the potassium content of the haemolymph of the preparations which relaxed during inhibitory stimulation was artificially raised by damaging some of the fibres of the femoral muscles during implantation of the myographic electrode. Alternatively, it is possible that variations in haemolymph-potassium levels normally occur in Schisto-cerca as suggested by Hoyle (1953 b). In any event, if the tonic fibres of the METM are slightly contracted due to potassium depolarization then when the METM is excited by the inhibitory neurone these fibres will be hyperpolarized and will relax (Usherwood, 1968). The inhibitory activity could also have the added effect of re-priming the contractile system of these tonic fibres since there is evidence that they cannot remain contracted indefinitely during a sustained potassium depolarization (Usherwood. 1968).

When data obtained from donor preparations during regular walking were used for programmed stimulation of recipient preparations, the functions of the inhibitory neurone during walking were clearly indicated. Relaxation of the METM, which precedes flexion of the tibia, is accelerated by the 1−4 impulses from the inhibitory neurone which occur at this time (Fig. 7). This inhibitory action presumably enables the flexor tibiae muscle to propel the tibia forward to the flexed position without having first to overcome any residual tension in the METM. It has been suggested that this could be one mechanism for maximising the stepping or walking rate whilst maintaining the dual function of the METM as a phasic and tonic muscle (Runion & Usherwood, 1968).

When tape programmes of inhibitory and ‘slow’ excitatory data (Fig. 8), abstracted from neurograms of donor preparations in which both metathoracic legs had been progressively deafferented, were used to stimulate a recipient preparation, the mech-anical responses of the METM of the recipient locust were abnormal. In Fig. 9 each record contains a number of superimposed tension records from a recipient METM during stimulation with a tape programme obtained from a donor locust which was walking in a regular manner. Each trace represents a single step response (flexion plus extension of the tibia) of the metathoracic leg of the donor insect, and each ‘step’ on the programme was repeated three times as follows. Initially, the complete programme, i.e. both inhibitory and ‘slow’ excitatory components; then only the inhibitory component; and finally only the excitatory component. Normally, activity of the METM alternates with that of the antagonistic flexor tibiae muscle during regular walking, and the mechanical responses of the METM are partly inhibited just prior to tibial flexion (Fig. 9a). When data recorded from a donor preparation, in which the ipsilateral (left) metathoracic tarsus was removed, were used to stimulate a recipient preparation it was observed that the number of inhibitory impulses preceding flexion of the tibia was reduced from four to two per step. (Fig. 8) This meant that tension in the METM no longer fell to near zero prior to tibial flexion (Fig. 9b). It was also noted that during tibial extension the tension developed by the METM was greater than normal. Although the duration of the activity of the flexor tibiae muscle was not increased by this operation, there was a tendency for the ‘slow’ excitatory axon to fire during flexion of the tibia. This rarely occurred in the intact insect. When the contralateral metathoracic tarsal segments were also removed, the number of inhibitory impulses preceding flexion of the tibia fell further to one per step, the duration of the activity of the flexor tibiae muscle increased by a factor of two, and the ‘slow’ excitatory neurone fired throughout flexion. The period of activity of the flexor tibiae muscle was further increased by macerating the contents of the ipsilateral tibial segment an operation which destroyed most of the tibial receptors including probably, the subgenual organ. However, the activity of the inhibitory neurone during walking did not fall to zero until the contents of the contralateral metathoracic tibia were also macerated. This operation also further increased the period of activity of the flexor tibiae muscle.

The effects of this progressive deafferentation on the mechanical activity of the METM during regular walking were also studied using the electronic simulator (Fig. 10). This method broadly confirmed the changes in tension described above and illustrated how the stepping rate is reduced following deafferentation of the metathoracic legs.

The donor-recipient technique and the electronic simulator were also used to investigate the mechanical responses of the METM during jumping and kicking. The METM is involved in two types of kicking responses. The first of these we shall call the ‘defensive kick’ since it is used to fend off other locusts which approach too closely from the rear. During defensive kicking the inhibitory neurone becomes inactive just prior to extension of the tibial segment and does not fire again until the kick is completed (Fig. 11). The cessation of inhibitory activity if followed by activity of the flexor tibiae muscle and a marked increase in the activity of the ‘slow ’ excitatory neurone to the METM. This causes a build-up of tension in the METM (Fig. 12). Presumably, the contractions of the flexor tibiae are used to prevent METM from shortening until its tension has reached a maximum. When this occurs the flexor muscle relaxes and the tibial segment is vigorously extended through shortening of the METM. In this respect the behaviour of the flexor tibiae muscle and the METM is similar to that which occurs during jumping (Figs 13,14) except that, during jumping, the METM is activated by both its ‘fast’ and ‘slow’ excitatory axons (Brown, 1967). At the end of the defensive kick the flexor tibiae muscle is activated once again to return the tibia back to the position it normally occupies during standing (Usherwood, Runion & Campbell, 1968).

The second type of kicking response occurs during courtship behaviour and we shall therefore refer to this as the ‘courtship kick’. Both males and females exhibit this type of behaviour, the males when confronted by a rival male and the females when they are not receptive to the advances of a male. The courtship kick differs from the defensive kick in that it involves activity of the ‘fast’ excitatory neurone to the METM. Otherwise, the mechanisms underlying the two responses are similar. It is interesting to note that during both forms of kicking the input from the tarsal receptors remains at a high level even though the tarsal segments are off the ground during extension of the tibia. This observation was a little disconcerting in view of our previous remarks on the activity of the tarsal receptors (Runion & Usherwood, 1968). However, photographic records of the metathoracic leg, taken during these kicks, indicate that the tarsal segments actively move with respect to each other and the tibia. These movements could undoubtedly activate the tarsal campaniform sensillae and tarsal chordotonal organ and could account for the high level of activity recorded from the tarsal afferents.

It has been demonstrated that during some forms of walking and during kicking the inhibitory neurone to the METM is inactive. This is also true during the jump response. The discharge of the inhibitory neurone falls to zero just prior to the jump (Fig. 13.) The ‘slow’ excitatory neurone then becomes very active and at the same time the flexor tibiae muscle contracts. When the discharge frequency of the ‘slow’ excitatory neurone reaches a maximum the flexor discharge ceases and the fast excitatory neurone to the METM fires. This causes a vigorous extension of the tibia and the insect is propelled off the ground (Fig. 14).

In view of the interest in central oscillators as possible mechanisms for the neuronal control of insect muscles during stereotyped behavioural responses (e.g. von Holst, 1948; Hoyle, 1964; Huber, 1962; Hughes, 1958; Wendler, 1966; Wilson, 1965, 1966), it is perhaps somewhat surprising that the main conclusion of this study and of our related studies on the control of locust leg muscles is that sensory information is of considerable importance in determining the activities of the motor neurones of the metathoracic leg muscles during standing, walking, kicking and jumping.

Investigations of the locust neuromuscular system have revealed that there are at least two sensory systems, the femoral chordotonal organ and the tarsal sensilla, which influence the activity of the ‘slow’ excitatory and inhibitory neurones of the METM. The metathoracic femoral chordotonal organ has been shown to provide information on the position, velocity and direction of movement of the tibia (Runion & Usherwood, 19666; Usherwood, Runion & Campbell, 1968). The phasically responsive tarsal sensilla, on the other hand, give qualitative and quantitative information on the relationship between the tarsal segments and the substratum and of changes in the load on the metathoracic leg (Runion & Usherwood, 1968). During walking, these receptors could provide cues for timing the contractions of not only the metathoracic leg muscles but also the muscles of the prothoracic and mesothoracic legs via the intersegmental commissures.

It appears that the inhibitory neurone which innervates the METM is used to vary the force developed by the tonic muscle fibres in the METM. When the insect is standing the inhibitory neurone often discharges at times when tension in the METM is falling, thereby accelerating the relaxation of the tonic fibres of the METM. As mentioned earlier, relaxation of these tonic fibres is also accelerated during walking, for the inhibitory neurone fires 1–4 times just prior to tibial flexion. It has been suggested that by increasing the rate of relaxation of the tonically responsive fibres of the METM an increase in the stepping frequency could be achieved. Recent studies on the ionically responsive fibres of the METM have shown that they give a measurable contraction to a single ‘slow’ excitatory impulse but that the duration of this response is about fourteen times larger than that of the phasically responsive fibres in this muscle (Cochrane, Elder & Usherwood, 1969). It is perhaps significant that during slow walking when the rate of relaxation of the METM may not be so critical, the inhibitory neurone is often completely inactive. The activity of the inhibitory neurone during walking is presumably determined to some extent by the input from the tarsal receptors and possibly also the subgenual organ in the tibial segment since on removal of these receptors the number of inhibitory potentials occurring just prior to flexion falls to only one per step. At the same time there is a transitory fall in the stepping frequency although after a few hours this returns to the normal rate of 4-5 steps/sec. We have not yet investigated whether the activity of inhibitory neurone also returns to normal although this is obviously an important point.

The problem of the relative roles of peripheral and central processes in controlling the activities of the leg muscles of the locust has not yet been completely resolved. One way of assessing the importance of the femoral chordotonal organ and the tarsal sensilla in controlling and co-ordinating the contractions of the femoral muscles would be to investigate the output of these muscles during programmed stimulation of the afferents from the chordotonal organ and tarsal sensilla. The afferent fibres of these leg receptors could be stimulated using impulse sequences abstracted from neurograms obtained from walking donor preparations. The mechanical responses of the METM of a recipient preparation during this programme could then be compared with those obtained in response to a programme based on data obtained from nerve 3 b of the donor locust at the same time as the data used to compile the afferent programme was obtained. If the responses of the METM to these two forms of stimulation were markedly different then it seems reasonable to assume that in the regular walking locust the information from the leg receptors does not directly determine the discharge patterns of the neurones to the METM. If, on the other hand, the two sets of responses are identical it could be assumed that central oscillators do not completely control the METM during walking. There are, of course, innumerable difficulties associated with this form of approach. For example, there is the difficult problem of the complexity of the afferent system. Programmed stimulation of the afferent nerves from the chordotonal organ and tarsal receptors would involve excitation of all the afferent fibres in these nerves simultaneously. This obviously bears little or no relation to what occurs in vivo. Also, cross reflexes and intersegmental reflexes are probably important during walking, so that differences in the responses to afferent and efferent programmes could be accounted for by the absence of these additional inputs.

On the basis of the results presented in this paper and in preceding papers (Runion & Usherwood, 1968; Usherwood, Runion & Campbell, 1968), it seems reasonable to assume that in the locust peripheral mechanisms are more important in determining the activities of the leg muscles during walking than they appear to be in determining the activities of the flight muscles during flight activity (Wilson, 1965). Central oscillators could be involved in walking in the insect but the output of these oscillators, if they are present, must be highly dependent on the inputs from the leg receptors. Recent studies on the walking mechanism of the cockroach have led F. Delcomyn (personal communication) to a similar conclusion, i.e. that ‘co-ordination’ of leg movements is probably the result of the interaction of a central signal controlling movements of the leg and peripheral feedback onto this signal for determination of the timing of the movement.

This work was supported by Grant NB–05 626–03 from the United States Public Health Service and by a Grant from the Science Research Council to P. N. R. Usherwood.