An Electromyographic Analysis of Forelimb Muscles During Overground Stepping in the Cat

Most previous studies of locomotion in the cat have been concerned with the function of the hind limb and its musculature. Studies of overground stepping movements (Goslow, Reinking & Stuart, 1973 ; Wretzel, Atwater & Stuart, 1976) and electromyographic (EMG) activity (Engberg & Lundberg, 1969) of the hindlimb of intact cats, decerebrate cats stepping on a treadmill (e.g. Shik, Severin & Orlovsky, 1966) and in intact dogs (Tokuriki, 1973a, b, 1974; Wentink, 1976) are available but only the work of Miller and colleagues (Miller & van der Meché 1975; Miller, Reitsma & van der Meché, 1973;,Miller, van der Burg & van der Meché, 1975; Halbertsma, Miller & van der Meché,1976) has considered the role played by the forelimbs in the overall planning and execution of stepping. Although such studies have suggested interesting comparisons to studies of hind limb function, they have, for the most part, been restricted to cats stepping on a treadmill belt. Further, reports of the pattern of activity of forelimb muscles have been confined to brief descriptions of only a few muscles.

The present paper describes the patterns of activity of forelimb muscles during overground stepping in cats, correlates these patterns with observed overground stepping movements, and discusses the roles played by individual muscles and muscle groups in the forelimb step cycle. A preliminary report of this work has been presented (English, 1977), and the pattern of activity and movements of the shoulder girdle complex are described in detail elsewhere (English, 1978).

Experiments were conducted on eight adult (B.W. 2·5–3·5 kg) cats, judged to be co-operative, with a total of 107 repeated trials. Before each experiment the cat was anaesthetized with ketamine HC1 (Vetalar Parke-Davis), 40 mg/lb (18mg/kg). Bipolar fine wire (50μm diameter) electrodes, bared 0·5 1·0 mm at each tip, were inserted into the centre of each of three individual muscles by means of a hypodermic needle using the method outlined by Basmajian (1973). The electrode leads were then connected to miniature dual FET source followers (Basmajian, 1973) mounted on the animal’s back. Inter-electrode resistance at the source followers was 35–50 kΩ The cat was connected to the recording equipment with a lightweight ribbon cable. Comparisons of stepping patterns of cats with and without the mounted equipment revealed no appreciable differences. Signals were differentially amplified (gain = 1000, common mode rejection 80 dB, linearity continuous between 20 Hz and 1 kHz) and high pass filtered (12 dB/octave, − 3 dB point at 80 Hz) to attenuate any line current or low frequency signals generated by movement of the electrode wires. This preparation resulted in stable EMG signals for period of at least 2–3 h, although recording sessions usually lasted only about 1 h, commencing after the animal recovered from anaesthesia. Electromyograms from the three muscles sampled were recorded on FM tape for later analysis.

Sites of electrode insertion were checked in three ways. The most common method of checking was careful palpation of the muscles in question and brief, low intensity stimulation of each through the electrode wires. In the case of deep muscles, this method was supplemented by using surgically implanted electrodes (see for example Cohen & Gans, 1975, for technique). Both cranial and caudal parts of serratus ventralis were implanted in three animals and antebranchial muscles were similarly prepared in three other animals (nine muscles). No readily observable differences in limb movements between animals with implanted electrodes and those with percutaneously placed electrodes were noted. Consistency of results between implanted and non implanted experiments was used as a second criterion of assurance of electrode placement. Finally, electrode placement was checked in four cats by dissection of animals dispatched immediately after an experiment.

Simultaneous to EMG recordings limb movements were monitored. This was either via slow-motion cinematography (57 experiments, c. 2000 steps) or videotaping (50 experiments, c. 2000 steps). In all steps so monitored, the time of placement and removal of the foot were noted. More detailed movement analysis was conducted on 200 steps selected to include each of the cats and to be representative of walking, trotting and galloping steps. Motion picture images were obtained at 64 frames s⁻¹ with a Bolex H-16 motion picture camera equipped with a Switar 10 mm F2·8 wide angle lens. The camera was placed in the same place in each trial, 5·4 m away from the cat, and remained stationary throughout. Single frame analysis of these records was performed using a digitizing motion analyser interfaced to a minicomputer (PDP 8e, Digital Equipment Corporation). Only steps found in the centre of the motion picture field were selected, to reduce possible parallax errors. The position of the metacarpophalangeal (MP), wrist, elbow and shoulder joints and the intersection of the spine and vertebral border of the scapula could be visualized easily beneath the skin of the cats, which were closely shaved, in each frame. It has been reported (Miller & van der Meché, 1975) that in treadmill-stepping cats, forelimb movements analysed in this manner compare favourably to observations of joint positions made by cineradiography. In each frame, each of the above-mentioned points and a marked reference point near the thoraco-lumbar junction were assigned Cartesian co-ordinates. From these data points, the angles at the forelimb joints (see Fig. 1 and Miller & van der Meché 1975, for descriptions) in each frame were calculated using the arctangent. Scapular angle is defined as the angle between the spinous process and the horizontal. Combining these measures with an indicator of foot placement, the angular movements throughout each step cycle (here defined as the interval between consecutive removals of a forelimb) were monitored. The same original data points also were used to examine the relative horizontal and vertical displacements of the segments of the forelimb during the step cycle. The Cartesian co-ordinates of the centre of the hand, forearm, arm and scapula were calculated and the rectilinear co-ordinates of the reference point subtracted from each. The result is the displacement of each limb segment in relation to the animal’s body. Movement analysis was confined to two dimensions (sagittal plane) because previous reports (e.g. Jenkins, 1971) have suggested that the range of mediolateral forelimb movements in stepping cats is small.

Results from several steps of the same duration (number of motion picture frames) were grouped, and averages of angles and displacements in each frame and the velocity of stepping in each step computed. Averaging was performed within the limits of using only the same cat during the same locomotor trial and care was taken to assure that each of the steps included were from steps of the same gait (i.e. walking, trotting or galloping). Although this method necessarily combines variation in measurements due to technique and that due to variations in limb movement, it is considered an acceptable sampling technique. Variation due to measurement technique is assumed to be small since comparison of the same data sampled on different occasions and periodic checks during collection (every tenth frame was digitized twice as a check) never resulted in differences in excess of 10%. Even so, no effort is made to assess the variance in limb movements which arises from either source.

Motion picture and EMG records were made synchronous by means of a hand operated device which placed simultaneous pulse signals on the tape and film records. Videotape records offered additional synchronization when a special effects generator was used to place the EMG trace, as seen on an oscilloscope, in the same field as the stepping cat. From print-outs of recorded EMG activity displayed on a Grass oscillo graph chart recorder, synchronous records of the onset, duration and relative intensity of activity of each of the muscles sampled could be correlated with footfalls and thus with the recognizable divisions of the step cycle. No effort was made to quantify the intensity of EMG activity, such as integration of rectified signals (see, for example, Betts et al. 1976, for technique). The intensity of EMG activity of a particular muscle, as used in this paper, refers to the amplitude of its spike components only and does not imply any relationship to force or any other parameter.

The EMG figures shown below were constructed from data from several different experiments which employed different cats. Each of the patterns of activity depicted was chosen from several similar examples. Samples of activity of each of the forelimb muscles were obtained from six of the cats in at least one stepping trial and in four of the cats on at least two separate occasions. In the remaining cats each of the muscles of a particular region (shoulder or antebrachium) were sampled at least twice. Thus the pattern of activity chosen for illustration is thought to be representative.

Angular movements of the forelimb joints are shown in Fig. 1 for each of the characteristic gaits used by cats -walk, trot and gallop, as defined by Stuart et al. (1973) Each plot is typical of a particular gait, and represents an average of five steps of the same duration. Gallops are shown divided into two types: those where the analysed forelimb leads (i.e. is placed down last) and those where it trails. Between these types may be seen differences in the movements of the forelimb, especially in the duration of individual epochs. Despite fairly substantial differences in gait, in step duration and in stepping speed, the angular movements of the forelimb display a common pattern, described by a modification of the Philippson step cycle (Philipp-son, 1905; Miller & van der Meché 1975). The F or flexion epoch begins as the toe is lifted from the ground. Flexion occurs at all of the forelimb joints, as well as between the scapula and the lateral thorax. Flexion of the scapula is defined as rotation of the vertebral border caudal or of the glenoid fossa cranial on the lateral thorax and is represented by decreasing joint angles in Fig. 1. Flexion at the other forelimb joints is represented similarly (see also Miller & van der Mèche, 1975). During the latter part of F, dorsiflexion (extension) of the wrist joint begins and the F epoch ends as the elbow and the shoulder joints begin to extend and the first extension epoch (E¹) begins. The scapula continues to be flexed on the thorax throughout E¹. At forelimb placement the second extension epoch (E²) begins. Dorsiflexion at the wrist joint continues, the elbow and shoulder joints flex and scapula is extended. Scapular extension is defined as the opposite of flexion and is represented by increasing angles in Fig. 1. The flexion occurring at the wrist, elbow and shoulder joints is said to be in response to the loads imposed on them by weight-bearing since forelimb stepping movements produced by cats not subjected to such loads (suspended decerebrate and swimming cats) show no E² flexion at these joints (Miller & van der Meché, 1975). In cat hind-limbs, comparable E² flexion movements have been noted despite activity in extensor muscles (Engberg & Lundberg, 1969) so that this epoch has been referred to as the yield. The E² epoch ends as the elbow and shoulder joints begin to extend and E³ begins. The wrist joint generally begins to be palmar flexed a short time (15–30 ms) later. The scapula is extended throughout E³. The epoch and step end as the forelimb is lifted from the ground.

These results are very similar to those reported by Miller & van der Meché (1975) for cats stepping on a treadmill belt, both in the range of angular movements used and in the actual angles. Differences in the temporal spacing of step cycle elements between treadmill and overground stepping patterns were also not substantial. However, no data are avilable as to the variance in patterns of angular movements in either study, so the significance of any differences between treadmill and overground stepping movements cannot be evaluated.

Fig. 2 shows the horizontal displacements of the anatomical centre of the hand, forearm, arm and scapula occurring during the step cycle, with respect to a reference point near the centre of the cat. Each plot represents the average of the same five steps of the same duration as Fig. 1.

During the walk, each of the segments moves craniad, with respect to the animal’s body, in the swing phase, and caudad in the stance phase. The magnitude of this displacement is greatest at the most distal segment (the hand) and least at the most proximal (the scapula). The anatomical centre of each segment is generally most craniad at limb placement and most caudad at lift-off. During slow trots (roughly 1·5–2·0 m s⁻¹) (not shown in Fig. 2), a nearly identical pattern is seen, except that the duration of the step is shortened. However, during fast trots (c. 2·0–3·0 m s⁻¹) the cranial-most position of some of the segments occurs earlier in the step so that the hand and forearm are moved caudad prior to placement. A less-marked caudad displacement of the arm is noted prior to placement and the relative displacement of the scapula on the lateral thorax is much as described at slower stepping speeds. Relative displacements during the rest of the step cycle are similar to those found during walking and slow trotting. During galloping, a similar type of displacement is found in both leading and trailing limbs. The hand, forearm and arm and, at very fast speeds, the scapula also, are all moved caudad prior to placement. Their cranial extreme occurs one or two frames (15·6–31·2 ms) earlier. Despite this change in relative displacement of limb segments, angular movements of forelimb joints of the same steps are not markedly different than at slower speeds (Fig. 1).

Of special interest in the context of displacements of limb segments are the trans-latory movements of the scapula on the lateral thorax which occur during each step. Fig. 2 shows that the centre of the scapula, and thus the entire forelimb, moves craniad during the swing phase and caudad during the stance phase. The overall magnitude of this displacement ranges from a few mm at slow stepping speeds to a few cm at fast speeds. These scapular translatory movements and the activity of muscles producing them have been described in detail elsewhere (English, 1978).

Fig. 3 shows the patterns of activity of shoulder muscles during walking, trotting and galloping steps. They have been described in detail elsewhere (English, 1978) so they will be described only briefly here. In Fig. 3 and the two following figures, each set of recordings is a composite of several experiments showing typical EMG ‘waveforms’ recorded from the individual muscles. The recordings shown for galloping cats are those of the trailing forelimb only. Records of muscular activity of the leading forelimb bear the same temporal relationships to limb placement and removal as those recorded in the trailing forelimb and thus were not shown.

In general, all of the shoulder muscles show patterns of activity of one of three types. Flexor muscles begin activity just (c. 20–30 ms) prior to forelimb lift-off and continue until the end of the flexion (F) epoch or slightly into E¹. These muscles include the brachiocephalicus and atlantoscapularis. Extensor muscles are mono-phasically active during the extension epochs. They begin activity during the first extension epoch (E¹), and cease activity just before lift-off. Substantial variability in the time of onset of activity is noted. Some muscles begin activity after the onset of shoulder and elbow joint extension whereas others begin before the onset of E¹. Most extensor muscles cease activity at or just prior to the end of E³. Among these muscles are latissimus dorsi, pectoralis minor, serratus ventralis, supraspinatus, infraspinatus and subscapularis. The remainder of the shoulder muscles show a biphasic pattern of activity-they display activity both during F and the extension epochs. Most show more intense activity during one of their two periods of activity but, as with all assessments of intensity of EMG activity in these muscles, variability in activity is extensive. Some of these muscles are not truly biphasic in their pattern of activity (e.g. rhomboideus capitis et cervicis and acromiotrapezius in walk, pectoralis major in walk and trot) since their flexion and extension periods of activity run together but their activity during both flexion and extension periods distinguishes them from flexors and extensors.

At different stepping speeds, the patterns of activity of the individual muscles remain fairly constant with respect to their temporal relationship to the step cycle epochs (Fig. 3). The duration of step cycle elements shortens differentially with increases in stepping speeds (Miller & van der Meché 1975). Since the onset and cessation of activity of the muscles of the shoulder region occur in roughly the same relationship to the onset of the step cycle elements at all stepping speeds, the duration of activity of muscles active during the different epochs shortens with increasing speed in like manner. Any observation as to the relative intensity of activity at different stepping speeds must be made with caution since no quantitative analysis of the activity was made.

The patterns of activity of flexor and extensor muscles of the elbow joint during walking, trotting and galloping are shown in Fig. 4. Among the extensors, recordings are shown for the long, lateral and medial head of the triceps brachii muscle. Recordings from accessory and intermediate portions of the medial head were similar. No records were attempted from the anconeus muscle. Its pattern of activity during stepping was assumed to resemble that of the rest of the triceps complex. The flexor muscles included the biceps brachii, brachialis and brachioradialis muscles.

Activity patterns of all of the extensors are similar in terms of onset and duration. All begin activity during E¹, usually after the onset of elbow extension, continue activity across limb placement and cease activity just prior to limb lift-off. Although the intensity of activity throughout each period of activity is somewhat variable within each muscle, all show most intense activity during E² and E³ and least activity during E¹. Activity in the lateral head is most intense during E² while the long and medial heads are most active during the early parts of E³. As noted in the case of shoulder muscles, the duration of extensor activity shortens with increases in stepping speed while the onset remains in fairly consistent temporal relationship to the step cycle epochs.

Activity in all of the flexors of the elbow is also similar. It generally begins just prior to limb lift-off (F onset) and ceases at or near the onset of E¹. In biceps brachii and brachialis, intensity of activity is fairly constant throughout but intense activity in brachioradialis is not usually noted until a short time after the onset of elbow flexion. Intense early activity is noted among all three flexors during gallops especially prior to elbow flexion movements, but less so during walks and trots. Activity ceases with elbow extension in biceps brachii and brachialis at slow stepping speeds (walks and trots) but continues into E¹ in galloping animals. Activity in brachioradialis continues into E¹ at all stepping speeds noted. A second burst of activity in biceps brachii and brachialis, usually of low intensity and occurring either during E² or across limb placement, was seen occasionally,

Fig. 5 shows the patterns of activity of muscles of the antebrachium during walking, trotting and galloping movements in cats. Both named extensor muscles (anatomically dorsiflexors) and named flexors (anatomically palmar flexors) show a fairly common pattern of activity. With few exceptions, activity begins during the latter portion of F, continues throughout the three extension epochs and ceases near the onset of the following flexion epoch. The intensity of EMG activity generally is low during the swing phase and greatest during E² and E³. Extensor muscles (palmar flexors) tend to display more overall activity during F than flexor muscles (dorsiflexors), but no quantitative data are available. Similar temporal relationships of the onset and cessa, tion of activity to step cycle elements are found at all stepping speeds.

In contrast to the majority of antebrachial muscles, the activity patterns in the supinator do not fit the overall pattern described above. The supinator displays a pattern of activity beginning in F and usually ending before the onset of E³. It maintains the same temporal relationship to the step cycle elements at all stepping speeds noted and maintains a fairly constant level of intensity throughout its period of activity. Pronator teres, the antagonist of supinator, has a pattern of activity at all stepping speeds sampled which resembles that noted in the palmar flexors of the wrist and digits except that a second burst of activity, confined to the F epoch, was noted consistently.

In their study of electromyographic activity of cat hind limb muscles, Engberg & Lundberg (1969) noted that, in general, all of the muscles examined showed one of three basic patterns of activity. Hind limb extensors begin activity during the first extension epoch (E¹), 20–40 ms after the joint about which they exert their main effort begins to extend, and continue activity until just prior to removal of the foot from the ground. Rectified and integrated activity is moderately intense during E¹, less so at placement, is most intense during E² and declines gradually during E³. Flexor muscles begin activity late in E³, about 30 ms before foot lift-off and attain peak activity during the F epoch. Flexors of the knee and ankle also display a lower intensity burst of activity at foot placement. Flexors of the hip show no such secondary burst of activity but maintain F epoch activity for a variable duration of E¹. Other muscles have been termed ‘bifunctional’ by Engberg & Lundberg since their patterns of activity make them difficult to interpret as either flexors or extensors. Similar results have been obtained in hind limb EMG studies of dogs stepping on a treadmill (Tokuriki, 1973a, b, 1974; Wentink, 1976) and in decerebrate cats induced to step on a treadmill by stimulation of the mesencephalic ‘locomotor region’ (e.g. Shik et al. 1966) (see also Wetzel et al. 1976, for recent review). Also observed by Engberg & Lundberg was a substantial amount of variability in the actual pattern of activity (waveform) of the individual muscles. Although many muscles correspond generally in their pattern of activity to one of the three types, each has its own individual waveform, both in terms of the relationship of onset and duration to step cycle element and in the relative intensity of activity at different times during the step. This individuality of waveforms was especially notable in flexor muscle activity.

The findings of the present study of forelimb EMG activity during unrestrained stepping are similar to those of Engberg & Lundberg for hind limbs. Many muscles show a pattern of activity similar to that described for hind limb extensors, activity lasting from E¹ to late in E³. The muscles of the triceps complex, many of the shoulder muscles and most of the antebrachial muscles might be considered to show this pattern. Others are active in a pattern similar to that of hind limb flexor muscles. Among these are the elbow flexors, brachiocephalicus and atlantoscapularis. Still others display patterns of activity during the step cycle which make them difficult to describe as flexor or extensor. Many of these have two phases of activity in each step and they correspond to the ‘bifunctional’ group of Engberg & Lundberg. Among them are the trapezius and rhomboid muscles of the shoulder region.

Also similar to the findings of Engberg & Lundberg is the variability in overall EMG waveform both within and between individual muscles. In almost every forelimb region studied, individual muscles displayed characteristic patterns of activity within each of the larger three groupings. For example, the three heads of triceps brachii all have patterns of activity which make them easily recognized as extensors but each muscle has a slightly different characteristic waveform (see Fig. 4). If several consecutive steps are analysed, this same waveform exists but its individual spike components vary (Fig. 6). Although no superimpositions of consecutive rectified and integrated EMG trains such as that used by Engberg & Lundberg was performed, this observation suggests the same sort of variability in pattern of activity within each forelimb muscle as was observed for hind limb muscles.

Thus the results of EMG and movement analysis of unrestrained overground locomotion in both fore-and hind limbs of cats might be interpreted to suggest that the limb movements used during stepping are the result of a central locomotor programme which conveys specific information to each muscle. Most of the variations in the characteristic waveforms of individual muscles are attributed to differences in muscular activity associated with adjustment to environmental perturbations encountered in different steps. One possible consequence of this model is that specific muscles or muscle groups can be interpreted as playing specific functions during stepping. The remainder of this discussion considers the roles played by different forelimb muscles during stepping in this context.

Alexander (1974) has recently speculated that the limb muscles of dogs can play three roles during stepping: they can produce stepping movements and thereby impart kinetic energy to the bones to which they attach; they can resist or prevent movements and thus absorb energy; and they can function as elastic bodies, absorbing energy in one part of the step and imparting it at a later point. He also proposed that at each limb, the large, proximal muscles probably function mainly in imparting energy while the smaller, more distal muscles act mainly in absorbing energy. The initial portion of this proposition also supports the earlier postulates of other authors (e.g. Gray, 1944; Gambaryan, 1972), especially at fast stepping speeds.

The results of EMG analysis of the large, parallel-fibred muscles connecting the body and the shoulder region are consistent with their use mainly as energy-imparting devices. Although many of these muscles probably play a role in transferring the portion of the weight of the body supported by the limb to the limb, since the shoulder girdle of cats is attached to the body entirely by muscle (see English, 1978, for discussion), their activity during non-weight-bearing and during portions of the step when weight-bearing might be presumed to decrease (E³) suggest that they act also in producing shoulder movements. During the swing phase, the scapula is moved craniad on the lateral thorax and is flexed, while during the stance phase, it is extended and pulled caudad. This combination of rotatory and translatory scapular movements has been interpreted primarily as a means of enhancing step length, especially if the movements of the two forelimbs occur out of phase (English, 1978). Extension and caudal translation of one scapula would then occur coincident with flexion and cranial translation of the other. The results of EMG experiments suggest that latissimus dorsi and pectoralis minor extend the scapula during the stance phase and contribute to its caudal translation. Activity in muscles of the rotator cuff-supraspinatus, infraspinatus and subscapularis-is proposed to maintain a firm attachment between scapula and humerus so that these large muscles can produce scapular movement even though they have no scapular attachments. Scapular flexion and cranial translation during the swing phase are similarly interpreted as being produced by atlantoscapular is, brachiocephalicus and pectoralis major (consult vector diagram, Fig. 7). The serratus ventralis muscle, whose cranial and caudal components are co-active through much of the step, may also play a role in the co-ordination of rotatory and translatory movements of the shoulder girdle. During the stance phase, the caudal portion of serratus ventralis could act primarily to pull the scapula caudad while the cranial portion of the muscle acts to produce scapular extension. Complementary activity in the trapezius and rhomboid musculature suggests that resolution of the force vectors associated with the shoulder girdle (see Fig. 7) results in placement of the fulcrum for scapular rotation near the caudal portion of the scapula. Under these circumstances, the caudal portion of serratus ventralis would act about a small moment arm for scapular extension but still be capable of producing caudal scapular translation. The cranial portion would act about a large moment arm and could produce powerful scapular extension without necessarily contributing a large cranial translatory effort. The reverse of this mechanism might be expected in the production of flexion and cranial scapular translation during the swing phase. Thus EMG activity in muscles of the shoulder region are consistent with the use of long, parallel-fibred muscles to move the limb craniad on the body during the swing phase of stepping and pull the body past the forelimbs during the stance phase.

The movements of the humero-ulnar articulation are confined to simple flexion extension alternations during stepping, although Jenkins (1973) has pointed out that additional conjunct movements occur at all mammalian elbow joints. The patterns of activity of the muscles of the elbow region during stepping generally reflect these simple movements. All three heads of triceps brachii display patterns of activity during stepping which are consistent with their acting mainly as extensors of the humero-ulnar joint. Brachialis, biceps brachii and brachioradialis all show activity patterns which suggest that they merely flex the joint during the F epoch (Fig. 2).

The latency between the onset of elbow extension and EMG activity indicates that early E¹ extension is produced passively, due to gravity. The pattern of activation of the triceps suggests that they function in forearm extension during E¹. During the stance phase these muscles undergo what has been described by Goslow et al. (1973), as an elastic stretch-shorten sequence. The muscles are lengthened while contracting during E² and are said to store energy in series-elastic elements. Subsequent shortening during E³ recovers the stored energy and adds energy of contractile origin. A similar phenomenon may occur in limb flexion. Flexor activity begins prior to the onset of the F epoch so that the flexor muscles may undergo lengthening contractions during late E³. Energy may be stored in series-elastic elements and be recovered once the F epoch begins. Thus, both the flexors and extensors of the elbow might be considered in the role of energy-imparting devices and elastic bodies. They produce active bony movements and absorb energy at one point in the step and return it at another point. However, some differences exist between the patterns of EMG activity in components of each group. Among the extensors, the medial head of triceps brachii shows more intense activity during E¹ than the lateral and long heads and less activity early in E². The long head shows most activity during E³, except at fast stepping speeds, whereas the lateral head is intensely active for much of the stance phase. All three muscles act to extend the elbow joint but the differences in the observed patterns of EMG activity indicate that each may play a more significant role than the others at different times during the step cycle. The same might be said of the elbow flexors, although the differences in EMG waveform noted are more subtle. Betts et al. (1976) have suggested that differences in the histochemical composition of the lateral head and the intermediate portion of the medial head of triceps brachii in the cat also may be of importance in assessing the locomotor functions of these muscles. The lateral head of triceps brachii contains mostly type FO and FOG fibres and the intermediate portion of the medial head mostly type SO. The former fibre types are associated with short twitch contraction times and forceful but phasic activities, the latter with slower contraction but more tonic use (see for example Burke & Edgerton, 1975). Although it is tempting to speculate as to the specific individual roles of these in-parallel synergists during stepping, on the basis of histochemical and EMG studies, clearer elucidation of this interesting topic must properly await further experimental, kinematic and kinetic studies.

The wrist joint and the digits of the forelimb are also moved through a relatively simple dorsiflexion-palmar flexion alternation during each step cycle (Miller & van der Meché, 1975, and above). Dorsiflexion begins during the late portion of the F epoch, c. 15–30 ms prior to the onset of extension at the elbow and shoulder joints. This movement continues through limb placement, is exaggerated during the E³ yield, and ends during the E³ epoch. Palmar flexion then continues past limb lift-off and into the following F epoch. Activity patterns in muscles whose lines of action indicate that they dorsiflex the wrist and digits are generally consistent with this simple alternation. Activity begins at or just prior to the onset of E¹, continues throughout the stance phase and terminates or is greatly attenuated by the end of the step. However, muscles whose direction of pull suggests that they palmar flex the wrist and digits are active at the same time. Many of these muscles begin activity before and cease activity after the dorsiflexors, suggesting as noted, for example, by Willis et al. (1966), that muscles acting on the wrist and digits play roles other than simple dorsiflexion and palmar flexion.

One such role is indicated in the pattern of activity of muscles producing palmar flexion, such as the relatively large palmaria longus. After beginning activity late in F, these muscles are stretched, first actively by the action of the dorsiflexors during E¹ and then passively by the weight of the body (though dorsiflexors continue activity) during E². Co-activation during E¹ might also be interpreted to produce fine control of hand posture prior to placement. The co-activation of pronator teres and supinator in control of rotary movements may be especially significant in this regard. The lengthening contractions resulting from this stretch might be expected to store mechanical energy in elastic elements in series with the contractile elements of the muscle, such as their long tendons. During the following E³ epoch, and in connexion with the resulting diminution in intensity of activity in most dorsiflexors, the stored energy might be recovered during shortening contractions and used in palmar flexion. Although electrical activity in palmar flexors ceases at or near limb lift-off, palmar flexion continues through much of the following F epoch. This continuance of mechanical activity after the end of electrical activity can be explained partly by the predicted period of persistence of mechanical activity (30–40 ms, according to Grillner, 1972, 1973), partly by the continued recovery of stored elastic energy and partly by the effect of gravity on the wrist joint.

Alexander (1974) has suggested a similar function for ankle extensors and ante-branchial muscles on the basis of kinetic analysis of jumping in the dog. He suggested that the long tendons lying in series with these muscles and their pennate fascicle architecture made them especially suitable for acting as passive elastic bodies, ‘probably absorbing about as much energy as they perform’ (p. 571). A similar role has been proposed for tendons in human walking (e.g. Cavagna, 1970). Many muscles of the wrist and digits in cats are also pennate and contain tendons, in series, whose lengths are probably substantially longer than that of the muscle fibres (for an elegant recent analysis, see Gonyea & Ericson, 1977). If energy is stored in those tendons during E¹ and E² and released during E³ then it is not inconceivable that the muscles acting on the wrist and digits during stepping in cats act primarily as elastic bodies, dissipating the kinetic energy of the body during E² and adding to it during the subsequent E³ epoch.

This work was completed with the co-operation of Dr J. V. Basmajian and the staff of the Regional Rehabilitation Research and Training Center of Emory University. Special thanks are due to Mr Harold Clifford who wrote many of the programs used in the computer reduction of the data. Financial support was provided by the McCandless fund of Emory University.