Proprioceptive Reflexes In The Legs Of Carcinus Maenas (L.)

The preceding paper dealt with the reflex responses to two tactile stimuli of the motor and inhibitor axons innervating the claw-opener muscle of the crab, Carcinus maenas (L.) (Bush, 1962). Peripheral inhibition, through the specific opener inhibitor, was shown to be primarily responsible for the observed reflex inhibition of claw opening. During these experiments it was noticed that if the dactylus was passively moved in the direction of opening, a burst of inhibitor impulses occurred, indicating a strong proprioceptive reflex.

Burke (1954) described a receptor (the ‘PD-organ’, after Alexandrowicz) which responded to movements of the dactylus in the walking leg of Carcinus. This consists of an elastic strand attached between dactylus and ‘closer’ apodeme, with sensory terminals lying in the strand (see Text-fig. 1). Both ‘tonic’ and ‘phasic’ afferent responses, to fixed positions or movements of the dactylus in either direction, were found. Recording from single or small groups of afferent fibres, Wiersma & Boettiger (1959) discovered a majority of the larger axons to respond to unidirectional movement with a non-adapting discharge at a fixed frequency for any velocity above threshold. A wide range of threshold velocities was covered, since the various fibres possessed different thresholds. Fibres were also found whose response varied with the velocity of movement, and with the position of the dactylus. Wiersma (1959) studied the afferent responses from a pair of similar receptors in each of the carpopropodite and mero-carpopodite joints of the leg (Text-fig. 1). ‘CPU, in the former joint, is attached to the ‘bender’ apodeme and responds mainly to stretching; ‘CP2’ attaches to the ‘stretcher’ tendon and responds almost exclusively to bending.

Eckert (1959) studied two proprioceptive reflexes in the claw of Astacus, though he was unaware of the nature of the receptor involved. He found that passively opening the claw evoked a discharge in its opener inhibitor and only a small one in the opener motor axon, whereas passive closing elicited more impulses in the motor axon than in the inhibitor. Prosser (1935) recorded reflex discharges in the leg nerves of the crayfish to ‘flexion’ or ‘extension’ of other limbs, the responses being strongest in the anterior homolateral limbs. He did not record in the stimulated limb. The microanatomy of the coxal proprioceptors has been studied by Alexandrowicz & Whitear (1957), and the fine structure of the sensory terminals of the more distal joint receptors by Whitear (1960).

The present paper represents an investigation of the reflex efferent responses in the same limb to passive movements of the dactylus and propus. The effects on these responses of different velocities of movement and positions of the moving joint have been analysed. An insight into the function of the specific opener and stretchei inhibitors has been obtained from experiments on the interaction of the reflex responses to manipulation of the two joints (see Text-fig. 1). The results described here deal exclusively with the common British shore crab Carcinus maenas (L.).

The efferent responses to dactylus movement were initially recorded in claws using methods similar to those of the preceding study (Bush, 1962). In most experiments, however, walking legs were used, these being more suitable for three reasons. First, the efferent axons to the four distal muscles of a limb (Text-fig. 1) were more readily prepared for recording in walking legs; secondly, the walking-leg responses would be of more general significance; and thirdly, the afferent responses to joint movements had been recorded in walking legs (Wiersma & Boettiger, 1959). In fact the reflex responses in the walking legs proved to be essentially similar to those observed in claws. All the experiments on reflex responses to propus movement were performed on walking legs for the same reasons.

The intact animal was strapped with rubber bands on to a Perspex plate, the appropriate limb being raised on a plasticene platform in a suitable position for preparing the required axons. The apodemes of the coxo-basal levator muscles, normally used to effect limb autotomy, were usually severed at their articulations to prevent autotomy during the experiment, though for walking legs this was not always necessary. The animal was immersed in filtered sea water, which served as a physiological saline, and then cooled in a refrigerator to ‘anaesthetize’ it and to reduce the blood flow before operating. The required axons were exposed and isolated for recording near the base of the propus or carpus, at a point near their separation from the main nerve to enter the muscle.

Recordings were usually made simultaneously from the axons supplying two muscles, commonly the opener and closer muscles or the stretcher and bender muscles (Text-fig. 1). Since the response of the muscle itself or of the complete loop was not required in this investigation, recording was simplified by cutting the axons distally. The central end of the bundle was then raised above the fluid surface on a hooked platinum wire electrode held in a micromanipulator. The two axon bundles were connected through separate pre-amplifiers to the two beams of a Cossor 1049 oscilloscope, the ‘indifferent’ electrode being represented by the grounded saline bath.

Movement of the required joint was produced by a simple Palmer rack-and-pinion arranged to have a maximum displacement of 7–12 mm. This was connected to the moving limb segment by a straight stiff wire, whose free end pivoted with a wire pin looped firmly round the segment. The displacement of the rack-and-pinion, and hence of the moving segment, was recorded by means of a simple lever interrupting a light beam which impinged upon a phototransistor. The output of the phototransistor was led to one channel of the oscilloscope, together with the response of one of the two axon bundles. With a little experience fairly smooth movements of the limb segment could be achieved with this device.

In the walking legs the dactylus moves through an arc of about 120° from the fully ‘closed’ to the fully ‘open’ positions (Text-fig. 1). In the experiments, however, the dactylus was usually moved through only about 100° of this arc. No essential part of the reflex responses to movement was thereby omitted, and this precaution minimized ‘fatigue’ of the receptor endings. The total movement arc of the cheliped dactylus is smaller than that of the walking legs, being about 80–90°. The propus in the walking legs moves through an arc of about 70°, 35° to either side of the midposition in line with the carpus (Text-fig. 1).

In both chelipeds and walking legs, passive opening (extension of the dactylus), by hand or with the rack-and-pinion device, elicits a strong discharge in the specific opener inhibitor as well as in the ‘slow’ closer motor axon (Pl. 1, record A). Passive closing (flexion) of the dactylus, on the other hand, elicits a discharge in the opener motor axon only. In fresh preparations rapid opening movements often also produce a response in the ‘fast’ closer motor axon, though of lower and less constant frequency than in the slow axon (B, C). The response of the fast fibre can readily be distinguished from that of the slow fibre by the former’s larger action potentials (it is the larger axon), by its adaptation and much earlier ‘fatigue’ with repeated movements, and by its less consistent response, usually only to fast movements.

These are the main and most common responses to these movements. Occasionally other, secondary responses appear; for instance, the opener motor axon may respond to passive opening. These secondary responses depend upon several variable factors, including the freshness of the preparation, and the position of the propodite with respect to the carpopodite. The latter factor will be considered later. Most responses referred to in this section are with the propus in its mid-position, in line with the carpus.

After movements applied manually, using a metal probe, the dactylus tended to spring back to its half-open mid-position, owing to the elasticity of its joint. The ‘rebound’ opening following a closing movement elicited a strong inhibitor discharge, which then continued at reduced frequency but with little or no adaptation for several seconds (e.g. 10 sec.), or until a slight forced opening was applied. In contrast, no such rebound occurred following an imposed opening movement. The tonic discharge of the inhibitor was presumably due to a continuing but indiscernible opening following the initial rapid rebound. Under these conditions the most sensitive of the afferent fibres which respond to opening movement continue to fire (Wiersma, personal communication). Closing movement fibres, however, none of which are as sensitive as the lowest threshold opening fibres, evidently do not respond for any length of time during rebound closing.

The extreme sensitivity of these reflex responses to dactylus movement is also well shown in records in which ‘stepwise’ movements, or small visible movements, were applied (Pl. 1, records B–G). Even during the smaller opening movements high-frequency discharges occurred in the slow-closer and opener inhibitor, though the latter axon manifested relatively low frequencies at the lower speeds of movement. Furthermore, in the stationary positions between stepwise movements, as well as in the fully open position, the slow-closer and to a lesser extent the opener inhibitor commonly continued to respond ionically though at reduced frequency. Between closing movements, however, and even in the fully closed position, the opener motoneurone seldom responded. Usually this fibre was active only during closing movements.

Two main parameters influence the frequency of these various responses to dactylus movement. These are the position of the dactylus within its arc of movement and the velocity of the movement. They will be considered in the following sections. It will be seen that the different axons are affected by these parameters to different degrees. (The fast-closer motor axon will not be considered in any detail since its response is usually fatigued at an early stage.)

The preceding qualitative observations have been made both on chelipeds and on walking legs. For the reasons already given, further analysis of these responses has been made on walking legs only. In general, however, most of the results to be described probably apply broadly to the chelipeds also.

During steady closing at low to moderate speeds, the frequency of the opener motor discharge in most preparations increases gradually (Pl. 1, record A). Thus the position of the dactylus within its arc of movement clearly influences the response of this neurone to dactylus movement. Such a ‘position effect’ is seldom seen in the responses of the opener inhibitor or slow-closer motor axon to opening. Instead, these neurones discharge at fairly constant frequencies during constant rates of opening, though occasionally there may be a slight reduction in frequency around the middle of the movement arc. In the fast-closer motor response to opening, however, a slight position effect is sometimes apparent, its frequency increasing somewhat from closed to open.

These results are represented graphically in Text-fig. 2 for the two opener axons. The smooth curve, representing the general ‘trend’ of the instantaneous frequencies (reciprocals of the intervals between impulses), shows that in the motor axon the frequency increases from the fully open to the fully closed position, while the inhibitor frequency is greater both at the beginning and at the end of the movement than in the middle. However, there was in this record a significant variation in the instantaneous velocity of movement (broken line), this being unavoidable with the manually operated rack-and-pinion movement device used. For this reason the ratio of instantaneous frequency to instantaneous velocity of movement was plotted as a function of time, showing that while the opener motor response to closing is distinctly dependent upon dactylus position, the inhibitor response during opening is not.

Even in the opener motor axon, however, the position effect is obscured at higher velocities, as can be seen in the lower two curves of Text-fig. 3. This gives the number of impulses (or average frequency) per quarter of the total movement arc, plotted as a function of the mid-position of each quarter, for four different velocities of opening and closing from one continuous record. The opener inhibitor frequency is almost constant throughout the arc for each of the four velocities. The motor frequency, on the other hand, increases considerably from open to closed at the lowest velocity of closing, less so at the next lowest, but scarcely varies at all at the two higher velocities. This result was typical, although in a few preparations there appeared to be little effect of position on the opener motor response even at the lowest velocities. No clear evidence of a position effect in either the opener inhibitor or the slow closer responses has been found, down to rates of movement of ca. 10° /sec., which is about the minimum velocity that can be achieved with reasonable smoothness with the device employed. Perhaps such an effect would become evident at still lower velocities.

The effect of joint position is also well shown in experiments in which the joint was moved in a ‘stepwise’ manner (Pl. 1, records B, D, E). However many stepwise movements of equal extent and velocity are made in one complete arc, the frequency and approximate number of impulses per opening movement in the opener inhibitor and in the slow closer is similar for all these movements. In the opener motor axon, on the other hand, the frequency and hence the number of impulses per closing movement increases with successive movements from the open to the closed positions. In Text-fig. 4 this result is represented graphically for four records of this kind (from three preparations), with from four to six equal movements per complete arc. In all the opener motor axon responses to closing there is a clear position effect, with increasing response frequency from open to closed, while there is no consistent effect in either of the other two axons. The apparent position effect in the responses of these two axons to the sequence of six stepwise movements (represented by H ) was from an older preparation, which when fresh had yielded typical responses in all three axons, with a definite position effect in the opener motor response but not in either of the other two.

A different kind of position effect is sometimes apparent in the activity of the opener and closer axons during the pauses between stepwise movements. Occasionally during the later pauses between successive closing movements, particularly the last pause, there is a slow discharge in the opener motor axon, which resumes at a slightly higher frequency in the fully closed position. This usually occurs only with the propus in a ‘bent’ position, the significance of which will be discussed later. More commonly, the slow-closer and the opener inhibitor discharge during the pauses between opening movements. Sometimes this discharge of the inhibitor, and even that of the slow closer, show an influence of dactylus position, the frequency being greater in the more open positions and greatest when fully open, though still lower than during movement (Pl. 1, record B). There may also be a reduction in frequency of the stationary position discharge in about the mid-position of the opening arc (D, E). In addition, when the movement steps are small, both these axons, while ceasing to fire immediately closing begins, may resume discharge during the first one or two pauses after full opening, though at a lower frequency than in the latter position or during the preceding pauses (B, D).

All three of the axons considered here respond with increasing frequency to increasing velocity of movement of the dactylus (Pl. 1, record A). The opener motor fibre is apparently more ‘sensitive’ to velocity of closing in the more open positions than in the more closed positions, where the basic response frequency is higher (Textfig. 3). In the other two axons the velocity sensitivity is independent of the portion of the arc considered. In order to compare the velocity sensitivity of the three axons, therefore, the average frequency for the whole arc was determined. Single or sets of three successive 100° arc movement sequences, opening—pause—closing—pause, at velocities of 10–5oo° /sec., were applied to the dactylus at 30–60 sec. intervals. The average response frequency for the whole arc was then plotted as a function of velocity of movement. Movements which were not smooth were omitted.

The results of two fairly typical experiments plotted in this way are shown in Text-fig. 5. In one of these experiments (circles) only the opener motor and inhibitor responses were recorded; in the other (squares) the slow closer response was also obtained. In the latter the opener motor response frequency, which ranged from 20 to 24 impulses/sec., bore no relation to the velocity of closing, and was therefore omitted from the graph to avoid additional complexity. The same happened in several experiments in which the closer axons were isolated for recording, a procedure which involves separating them from the accompanying bundle of afferent fibres from the PD organ. In experiments in which only the opener axons were prepared, however, the opener motor response usually varied with velocity as indicated in Text-fig. 5. The absence of a frequency/velocity relationship may therefore have been due to injury, during preparation, to certain of the PD afferents responding to closing. In general, however, the frequency/velocity curves for these three axons manifest the approximate form and interrelationships represented in Text-fig. 5.

It is of interest to derive some quantitative estimate of the relative ‘sensitivities’ of the three axons to velocity of dactylus movement, in particular for a physiologically significant range of velocities. Visual observation of a crab walking about at different speeds suggests that the most commonly occurring velocities of dactylus movement during locomotion lie between about 50° and 200° /sec., though faster movements no doubt occur during rapid walking or ‘running’, while lower velocities certainly obtain in slow or ‘postural’ movements. Wiersma & Boettiger (1959) found afferent movement fibres with threshold velocities ranging from ca. 1° /sec. upwards, and with saturation velocities of 120° /sec. or more.

Considering now only the definitely functionally significant range of velocities of 50–150° /sec., a simple figure for the ‘velocity sensitivity’ of each fibre is given by the ratio of the difference between the response frequencies at 150° and 5o° /sec. to the frequency at 100° /sec. The values thus obtained from the four curves in Text-fig. 5 are: circles, 0·5 for the opener inhibitor and 0·4 for the opener motor axon, and in the second experiment (squares), 0·3 for the slow closer and 0·2 for the opener inhibitor. The marked difference between the two experiments in these values, and specifically in the values for the opener inhibitor, could have been due to one or more of a number of factors. It has already been suggested that some of the afferents responding to dactylus movement may have been impaired during separation of the PD afferent Bundle from the closer motor axons. The resulting reduction in afferent input could have lowered the efferent response frequencies, perhaps particularly at the higher velocities. However, the relatively high response frequencies at the lower velocities in all three axons in the second preparation (squares) suggest a generally higher ‘excitability’ of the central synapses. This, in combination with a synaptic ‘saturation’ frequency, at the higher velocities, similar to that of the first preparation (circles), might result in the differences noted in Text-fig. 5. Alternatively, these differences might be a consequence of a difference between the two preparations in the numerical distribution and range of the velocity thresholds of the unidirectional movement fibres of Wiersma & Boettiger (1959). Perhaps each of these factors contributed to the velocity sensitivity differences in question.

In any event, the important result is the form of the frequency/velocity curves, and the relative difference in the velocity sensitivities of the three efferent neurones, rather than their absolute values. In general, then, in any given preparation the series —opener motor, opener inhibitor, slow closer—is in order of increasing velocity sensitivity. As was seen in the previous section, this is also the order of decreasing position sensitivity. The place in this series of the fast closer motoneurone is doubtful.

These responses (Pl. 1, records H–L) are essentially analogous to those of the opener and closer axons to dactylus movement (A–G). Passive ‘stretching’ (re-duction of the propus) elicits a strong response in the slow-bender motor axon and in the (specific) stretcher inhibitor (see Text-fig. 1). Passive ‘bending’ (pro-duction) elicits a response in the stretcher motor axon, similar to the response to closing in the opener motor axon (that is, the same neurone). The fast-bender motor fibre, like the fast closer, only responds during rapid stretching movements, and with a less regular and lower frequency than the slow fibre. In general, too, the effects of variation in position and velocity of movement of the propodite on the responses of these axons are similar to those produced by variation in position and velocity of movement of the dactylus on the opener and closer axon responses.

The response of the stretcher-opener motoneurone to bending was, however, less dependent on propus position than was its response to closing on dactylus position. Although the frequency of its response to bending did usually increase slightly from the stretched to the bent position, this ‘position effect’ was smaller than in the response to closing. This might be related partly to the smaller arc of movement of the propus (ca. 70°) than of the dactylus (100–120°). Another possible interpretation will be discussed later. Notwithstanding this qualitative difference, a position effect was still clearly evident in the response to very slow bending, either steady or stepwise, whereas such an effect was distinctly absent from the responses to stretching (K, L).

A second type of position effect was sometimes apparent during stepwise stretching in small steps. This consisted of a brief reduction or even cessation of the response of the stretcher inhibitor and often also of the slow closer, in about the mid-position of the stretching movement arc (Pl. 1, records K, L). This ‘mid-position effect’ may be related to the presence of two CP receptors, in contrast to the single PD receptor ; and to Wiersma’s (1959) finding that position-sensitive fibres from the CP receptors responding in the middle portion of the arc were rare. A similar but less-marked¹ effect was occasionally observed in the opener-inhibitor and slow-closer responses to slow, stepwise opening: mid-position afferents from the PD receptor were also rare or absent.

There is presumably considerable interaction between the reflex responses to movement and position of the different joints of a decapod limb. For instance, the reflex response of an efferent fibre to movement of any one joint may vary somewhat with the position of one or more of the other joints. In particular, such interaction might occur between opener/closer and stretcher/bender responses, since the opener and stretcher muscles are supplied by the same motoneurone, and can therefore act independently only by invoking peripheral inhibition through their specific inhibitors.

Interaction between these two joints has been observed in two experiments in which responses to opening and closing were recorded for different positions of the propus, recording from the opener-stretcher axon bundle and the bender bundle in the carpus. In these experiments the propus was held by the rack-and-pinion device in each of three positions—nearly fully stretched, nearly fully bent, and in its midposition (in line with the carpus). The dactylus was moved with a metal probe, in a fairly steady motion lasting 0·6–1·0 sec., from its mid-position (ca. 45° to the long axis of the leg) to its fully open or closed positions, that is with a velocity of approximately 8o° /sec. It was then allowed to spring back again to its mid-position by its own elasticity. Responses occur during movements of this kind which are essentially similar to those produced by the rack-and-pinion, the elastic rebound eliciting a reflex efferent response similar to that evoked by an imposed movement in the same direction.

The average frequencies for two successive responses in each position is represented graphically in Text-fig. 6. Since the opener inhibitor response during opening was similar in the median and the stretched positions of the propus, the values for these two positions were averaged, and are shown in Text-fig. 6 by broken lines. Again, the opener-stretcher motor response to closing was similar for all three propus positions, so that these values were also averaged. The effect of the position of the propus joint can now readily be seen.

In the first place, while the dactylus is stationary (in its resting, mid-position) between closing and opening movements, the opener inhibitor remains silent with the propus in its mid-position, but discharges steadily while the propus is in its bent position. The significance of this seems clear, for the stretcher-opener motoneurone (but not the stretcher inhibitor) is also active while the propus is bent, this tending to resist the bent position. The accompanying opener inhibitor activity would thus prevent simultaneous excitation of the opener muscle by the activity of the motoneurone. Especially in the stretched position and also in the median propus position, on the other hand, the stretcher inhibitor is active, overriding the motor response which follows opening, and thus resisting stretching. During opening movements the most noticeable response in the present context is the particularly strong opener inhibitor discharge in the bent position. This would ensure complete suppression of any possible opener response which might arise from the activity of the opener stretcher motoneurone, which itself tends to resist the bent position. Conversely, during closing movements the stretcher inhibitor responds exceptionally strongly with the propus in the stretched position, thus tending to resist the stretching by completely overriding the activity of the opener-stretcher motor axon, which itself resists the imposed closing movement.

At a later stage in each of the preceding two experiments, an attempt was also made to test the effect of dactylus position on the stretcher-opener axon responses to stretching and bending movements. No effect was apparent. However, in one of these experiments the response of the opener axons to dactylus movement was shortly afterwards found to have disappeared, indicating that some or all of the PD-organ receptor cells had ‘fatigued’, or been injured or otherwise inactivated. It is conceivable too, that in the other experiment the afferent response to dactylus position had been impaired without any apparent effect on the response to dactylus movement which was still ‘normal’. Thus it remains possible that dactylus position does influence the reflex response to propus movement.

Although there could be little doubt that the proprioceptors referred to in the introduction, PD, CP₁ and CP₂, were the ones responsible for the reflex responses in the present experiments, it was interesting to confirm this. Thus in a few experiments, after recording the normal efferent responses to passive joint movements, the appropriate receptors were progressively eliminated.

In one of two such experiments on the reflex responses to stretching and bending, CP₂ was first detached distally by drilling through the point of its insertion on the integument. This diminished the efferent response to bending but did not eliminate it altogether, perhaps because the receptor still remained loosely attached by fine connective tissue strands running from the main band. The receptor was then exposed at its proximal end by dissection and its origin on the stretcher apodeme was severed. This eliminated most of the efferent response, and what little remained was completely abolished by transecting the sensory axon bundle leaving the receptor. The response to stretching was still apparently unaffected. The insertion of the receptor CP 1 was now excised, after which the efferent response to stretching was considerably reduced. Finally, this receptor was completely eliminated functionally by severing its sensory axon bundle, after which no efferent response could be obtained either to stretching or to bending. That the efferent axons involved were still reflexly responsive was evident from their more or less normal response to dactylus movement. In the second of these experiments, CP₁ was eliminated first, abolishing the reflex responses to stretching; and the subsequent inactivation of CP₂ removed all reflex response to bending.

Since the PD organ signals all proprioceptive sensation from the propo-dactylopodite joint (Wiersma & Boettiger, 1959), it is not surprising that transection of all its sensory axons just proximal to the organ abolished all reflex response to opening and closing. However, cutting through the receptor strand itself at any point along its length, or severing either its insertion or even its origin on the closer tendon, always left some response to opening and closing. This was evidently due to the presence of several fine connective tissue strands attached to the main strand and running obliquely distally and proximally. As with the CP receptors, these auxiliary strands no doubt help to determine the normal response of the receptor, probably by maintaining the organ in its correct alignment and ensuring appropriate distribution of tension along its length.

Wiersma & Boettiger describe the PD sensory cells responding to opening as lying near the base of the organ on its dorsal aspect, and those responding to closing as ventral and slightly more distal. Splitting the organ lengthwise into dorsal and ventral halves, however, failed to separate the efferent responses to opening and closing, since section of either half partly eliminated both responses. Nevertheless, the sensory axons leaving the organ often do so in two broadly distinguishable groups, one mainly from the ventral (closing-responsive) region and the other from the dorsal (opening) region of the organ. In two experiments cutting the dorsally originating group of axons near the organ eliminated most of the reflex response to opening, while section of the group of ventral origin removed the response to closing. What little ‘overlap’ there was in these tests was very likely due to crossing of afferents as they left the organ. Thus these results on the efferent responses to dactylus movement, like those to propus movement, are in general agreement with the observations of Wiersma & Boettiger (1959) and Wiersma (1959) on the corresponding afferent responses.

All the primary reflex responses in these experiments can be considered as resisting the passive joint movements eliciting them. For instance, the closer motor response to passive opening resists this imposed movement. Similarly, the observed reflex responses of the two specific inhibitors reinforce the ‘resistance responses’ of the antagonistic motoneurones. Presumably similar passive joint movements occurring in the freely living animal elicit similar proprioceptive reflex responses, as for example when a dactylus is accidentally displaced by brushing against a rock or by a sudden water current. However, whether similar proprioceptive reflexes are evoked by active movements of these joints, through contraction of the appropriate muscles, is questionable and of considerable importance and awaits investigation. Burke’s (1954) finding that the total afferent response from the PD organ is greater during active than during passive closing, and that a fairly strong response occurred even with isometric contraction of the closer, supports the possibility of there being a significant difference, even in the reflex efferent responses.

Nevertheless, theoretically it seems likely that the closer tendon would move in a similar manner during passive and active opening. Therefore the afferent responses and also the efferent reflex responses from the PD organ are probably similar during passive and active opening. The same may hold for the reflex responses to passive and active stretching and bending. On the other hand the responses to passive and active closing may well differ significantly. For, during the latter half of a passive (but not active) full-arc closing movement the closer tendon buckles, just distal to the PD attachment, at its secondary articulation between tendon proper and the small platelet lying between this and the primary articulation with the dactylus (M. D. Rayner, personal communication). What effect might this buckling have on the reflex response to closing? The weakness of the passive closing response compared to the responses to opening, particularly in the first half of a closing movement, cannot be due to the buckling since it does not occur in the first half of this movement. Its occurrence in the second half of closing, however, might enhance (or perhaps diminish) the ‘position effect’ in the opener motor response. Finally, since buckling presumably occurs at any speed of passive closing, it is unlikely to affect significantly the ‘velocity sensitivity’ of the reflex response.

Assuming for the present that the experimentally observed reflex responses to passive joint movements are similar to the responses to the corresponding active movements, it is of interest to consider the possible functional significance of their relative strength, position effect, and velocity sensitivity. The closer muscles of the walking legs, since they support the animal’s weight and exert power in locomotor movements—and also those of the claws, in grasping prey—function primarily under load and at relatively high tensions. In order for the reflexes resisting imposed opening to serve adequately in a negative-feedback, servo-control capacity, therefore, they should be strong and highly sensitive to dactylus movement and velocity, in all positions of the joint. In contrast, the opener muscles of both walking legs and claws probably do not normally sustain much tension, but instead may serve primarily to restore the dactylus to a more suitable position, in particular to its resting midposition. For this purpose a resistance reflex of the opener muscle dominated by the position effect described would be best suited. The antagonistic stretcher and bender muscles of the walking legs are more alike in size and effect, and probably serve mainly to maintain a certain stiffness in the CP joint, its movement being brought about by alteration in their relative tensions. For these muscles strong resistance responses, moderately sensitive to velocity of joint movement and perhaps also somewhat to position, would be appropriate. In their general features the experimentally observed reflexes in these two joints are in accord with these theoretical considerations, in particular in the dominance of velocity over position sensitivity in the resistance response to imposed opening, and the reverse in the response to closing.

The question arises as to what extent these negative-feedback, proprioceptive reflexes have a co-ordinating function, either within each limb alone or also between different limbs. Evidence on this general problem in arthropods is scanty, though certain experiments of Hughes (1957) on locomotion in the cockroach, and of Hughes & Wiersma (1960) on the co-ordination of swimmeret movements in the crayfish, indicate that such a function is involved. Some support for this view comes from the present experiments, in which propus position was shown to influence the reflex efferent responses to dactylus movement. It may be significant that, according to the evidence thus far obtained, whereas proximal joint position influences the reflex responses to movement of the distal joint, distal joint position apparently does not affect the reflex responses to proximal joint movement. If this is borne out by further experimentation, a general principle concerning co-ordination of the limb joints may be involved.

It has been seen that the specific opener and stretcher inhibitors play an integral part in these proprioceptive reflexes. The observed variations in their responses to dactylus movement with the position of the propus are in accordance with the theory that these inhibitors serve to allow independent action of these two muscles despite their motor coupling. This represents the first experimental evidence for this theory proposed initially by Wiersma (1941). In none of the experiments reported here was any activity attributable to the common inhibitor observed. It is yet to be demonstrated whether this neurone may be active during normal locomotory movements.

The remarkable specificity and consistency of the proprioceptive reflex responses observed in this study, particularly in comparison with the responses to tactile stimulation reported in the preceding paper (Bush, 1962), suggests that thty may represent monosynaptic, afferent-efferent reflexes. This recalls the partially analogous twoneurone muscle-spindle reflexes of vertebrates. By contrast, the corresponding reflexes in Astacus appear to be much less ‘specific’ (Bush, unpublished). This problem of reflex specificity clearly warrants further study, particularly from the point of view of reflex integration and limb co-ordination.

The afferent responses to the two directions of propus movement ‘overlap’ somewhat, in that CP₁ whose predominant response is to stretching has some fibres which respond to bending, and conversely CP₂ has a few stretching fibres (Wiersma, 1959). No such overlap was apparent in the reflex efferent responses recorded in this study, since elimination of CP₁ completely abolished the reflex response to stretching, while CP₂ elimination abolished the response to bending. The apparent ‘redundancy’ (cf. Wiersma, 1958) thus implied might be due, peripherally, to the extreme sensitivity of these receptor endings and, centrally, to an inability of the overlapping afferents alone to produce sufficient synaptic facilitation to evoke a reflex response. Alternatively, these afferents may make inhibitory connexions with antagonistic efferent neurones, though this seems less likely since no antagonistic reflexes became apparent after elimination of either of the CP receptors. A further possibility is that these overlapping afferents have inhibitory or excitatory synapses with other efferents and are involved in reflex co-ordination. Thus they may yet be of functional importance in the normal animal.

Finally, it is important to consider what types of proprioceptor afferents are involved in the reflex responses described. The ‘pure’ unidirectional movement fibres of Wiersma & Boettiger (1959) are the afferents most likely to be concerned in eliciting the reflex responses of the slow-closer and opener inhibitor to imposed opening, and also those of the slow-bender and stretcher inhibitor to stretching. The increasing frequency of the reflex response with increasing velocity of movement would then arise from recruitment of afferents, as the velocity rises above their thresholds. The same may also be true of the reflex response to bending. The opener motor response to passive closing, on the other hand, might be largely or even entirely due to the position- and velocity-sensitive PD movement fibres, at least at the lower experimental velocities. The comparative weakness of this reflex response to closing might, however, result from the relative paucity of sensitive movement fibres responding to closing compared to the sensitive opening afferents (Wiersma, personal communication). Lastly, movement-sensitive position fibres, particularly perhaps the ‘asymmetric’ response fibres in this category, are probably involved in the asymmetric reflex response in stationary positions between stepwise movements. However, these considerations remain speculative pending further experimentation on the efferent responses to selective afferent stimulation.

I wish to thank Prof. J. W. S. Pringle, F.R.S., for his stimulating interest in this work, and Prof. C. A. G. Wiersma for his helpful criticism of the manuscript. My gratitude is also due to Dr J. S. Alexandrowicz and to the Director and staff of the Marine Biological Association’s Laboratory at Plymouth, where some of this work was done, for their hospitality and assistance. A Research Bursary awarded by the South African Council for Scientific and Industrial Research is gratefully acknowledged.