INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A brief interpulse interval or "doublet" at the onset of a stimulus train greatly augments the force produced by whole muscles and single motor units (Binder-Macleod and Barker 1991; Cooper and Eccles 1930; Duchateau and Hainaut 1986a; Karu et al. 1995; Macefield et al. 1996). Following this initial short interval (5-15 ms), maximal force is usually obtained with longer interpulse intervals. This can vary for different motor units and with changes in muscle length, however (Burke et al. 1976; Mela et al. 2001; Parmigianni and Stein 1981; Thomas et al. 1999; Zajac and Young 1980). It is unknown whether a similar series of stimuli evokes maximal force from whole human muscles. It is also unclear whether these same stimuli maximize force from muscles that have been paralyzed. The purpose of this study was to determine the pattern of pulses that elicited maximal force from control thenar muscles and those that have been paralyzed chronically by spinal cord injury (SCI). These data become particularly important if functional behaviors are to be restored by electrical stimulation. Skeletal muscles are often fatigable and weak after spinal cord injury (Thomas 1997a,b; Thomas et al. 1997). Thus there is usually a need to evoke as much force as possible and to minimize fatigue in these muscles.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twelve subjects were selected for study. Six able-bodied subjects (mean ± SE age: 27 ± 2 yr) with no reported history of neuromuscular disorder were selected as controls. The other six subjects (33 ± 5 yr) had thenar muscles that were completely and chronically (10 ± 2 yr) paralyzed by cervical SCI. That is, no electromyographic (EMG) activity was produced when the subjects were asked to contract the thenar muscles, and there was no palpable muscle shortening during a manual examination. Injury levels were at C5 (n = 2) or C6 (n = 4), as defined by American Spinal Injury Association criteria (Maynard et al. 1997). These subjects were injured while diving (n = 2), playing sport (n = 1), in a motor vehicle accident (n = 2), or from a fall (n = 1). Each subject population contained five males (83%) because spinal cord injury occurs much more frequently in men (Stover and Fine 1986). The hand (or arm if no sensation was present in either hand) that the SCI subject judged to have the poorest sensation was tested. Each subject gave informed written consent to participate in these experiments. The University of Miami Investigational Review Board approved all of the experimental procedures.

Experimental setup

The experimental setup has been described previously (Thomas 1997a,b). Control subjects sat in a chair. Spinal cord injured subjects remained seated in their wheelchairs. The test forearm was immobilized in a vacuum cast that rested on a tray beside the subject. The hand and fingers were wrapped in therapeutic clay, then held down with a metal plate and Velcro. The thumb was extended and strapped against a force transducer that registered both abduction and flexion forces at right angles to each other. Resultant force was calculated off-line.

Three electrodes made from braided strands of silver-coated copper wire were placed across the thenar muscles to record surface EMG. The common reference electrode lay across the middle of the muscle bellies. The distal and proximal electrodes were placed over the metacarpophalangeal joint and the base of the thumb, respectively. A ground wire was placed along the wrist crease.

Protocol

The best site for stimulating the median nerve just proximal to the wrist was found by delivering single, low-intensity, 50 µs duration pulses at different sites (DS7H Digitimer stimulator). The stimulating electrode was taped to the site at which the largest compound muscle action potential (M-wave) was evoked. The current was then increased in 1-mA steps until there was no further increase in the amplitude of the M-wave. All subsequent stimuli were supramaximal. The M-waves were monitored on an oscilloscope throughout the experiment to ensure that they remained maximal.

The pattern of median nerve stimulation that elicited peak thenar force in a seven-pulse train was determined from six series of pulses (Thomas et al. 1999). Each series consisted of 25 trains of pulses. Each train of the first series consisted of two pulses with the following intervals: 500, 400, 300, 200, 180, 160, 140, 120, 100, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10, and 5 ms. The pair of pulses that produced peak force was selected to begin each train of the second series. That is, the trains of the second series consisted of three pulses, but only the last interval was varied using the intervals just described (Fig. 1). This procedure was repeated for all subsequent series, with the maximal force produced by seven pulses determined in series 6. Trains of seven pulses were used so that the whole muscle data could be compared with that obtained from single thenar motor units (Thomas et al. 1999).

View larger version (77K): [in this window] [in a new window]   Fig. 1. Distal surface electromyograph (EMG; top) and resultant force (bottom) evoked from the paralyzed thenar muscles of a spinal cord-injured (SCI) subject by trains of 2 pulses (A; arrows) and 3 pulses (B) in series 1 and 2, respectively. The last interpulse interval in each series was reduced progressively from 500 to 5 ms (EMG: bottom to top trace; force: right to left). The 1st interpulse interval in series 2 was always 15 ms in this subject because this interval evoked peak force in series 1.

Within any one series, 2-4 s elapsed between trains. Approximately 10-15 min elapsed between each series while the pulse sequence for the next series was programmed on-line for each subject. This between-series rest period prevented progressive force potentiation and muscle fatigue. Following the sixth series, the following stimuli were delivered approximately 1 min apart: 1) five twitches at 1 Hz; 2) three doublets (5-ms interpulse intervals, each doublet 1 s apart); and 3) 1-s trains of 5, 8, 10, 15, 20, 30, 40, and 50 Hz (each train 2-4 s apart) to determine the force produced by different stimulation frequencies. All these stimuli were delivered for comparison to single thenar motor unit data (Thomas et al. 1999).

Data collection and analysis

The proximal and distal surface EMG and abduction and flexion forces were filtered at 30-1,000 Hz and DC-100 Hz, then sampled on-line at 3,200 and 400 Hz, respectively, using SC/Zoom software (Department of Physiology, University of Umeå, Sweden).

All data analyses occurred off-line. For every train in a series, measurements were made of the peak force evoked by the last pulse and the force-time integral (force start to end). The peak force, contraction time (time to peak force) and half-relaxation time (time for force to fall to half the peak value) were only measured for the five twitches and three doublets. Averages were calculated for each twitch and doublet parameter measured. The peak forces evoked by different frequencies of stimulation were measured. The force elicited by 50-Hz stimulation was used as a measure of maximal muscle force.

Peak force of individual trains were normalized to the 50-Hz force to determine what fraction of maximal force was evoked by trains of stimuli consisting of two to seven pulses. The force-time integral for each train was normalized to the maximal force-time integral in series 6 because the duration of the seven-pulse train varied for each subject. Twitch/tetanic, doublet/tetanic, and doublet/twitch force ratios were calculated. The frequency needed to evoke half-maximal force was determined from the linear regression equation fit to the force evoked by stimulation at low frequencies (Thomas et al. 1991).

Statistical analysis

Means ± SE data are given. Two-tailed t-tests were used to compare the measured twitch, doublet, and 50-Hz parameters, the respective force ratios and the frequency required to produce half-maximal tetanic force of paralyzed and control muscles. ANOVAs were used to compare peak forces and the intervals at which these peak forces and force-time integrals occurred across series within and across subject groups. Statistical significance was accepted at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulse pattern for maximal force

Figure 1 shows the distal surface EMG and resultant force recorded from the paralyzed thenar muscles of one SCI subject during the first and second series of stimuli (2-pulse and 3-pulse trains, respectively). When two pulses were delivered at long interpulse intervals (300-500 ms), the twitch forces were unfused (Fig. 1A). With shorter interpulse intervals, the twitches fused progressively, but the EMG was always well maintained. The peak force attained in this record was more than double (2.8-fold) that produced by a single pulse. It was evoked when the pulses were 15 ms apart. Each train in the second series therefore began with a 15-ms interval. The third pulse in the train was delivered at progressively shorter intervals (500 to 5 ms; Fig. 1B). Forces similar to the actual peak force (5% peak force) were attained in series 2 when the last interpulse interval was varied between 5 and 65 ms. The interval midway within this range was chosen as the second interpulse interval in series 3. This same process of fixing another interpulse interval occurred for each subsequent series.

Figure 2 depicts the mean absolute force evoked from all paralyzed (A) and control (B) muscles for each train during all series of stimulation. During series 1, there was a steep rise in force as the interpulse interval was shortened. Similar peak forces were produced in paralyzed (8.6 ± 2.7 N) and control muscles (6.5 ± 0.8 N) during series 1, but the forces evoked from paralyzed muscles were more variable. However, when these data were normalized to the maximal 50-Hz force, two closely spaced pulses produced a significantly higher force in paralyzed (41.4 ± 3.9% maximal, Fig. 2C) versus control muscles (22.7 ± 2.0% maximal, Fig. 2D). Adding one more pulse in series 2-6 resulted in progressively stronger forces. By series 6, both subject groups had increased their force by an additional 34-35%. Thus with seven pulses, paralyzed and control muscles had reached an average force of 15.4 ± 4.2 N and 16.3 ± 1.8 N, respectively. These data represented 76.4 ± 5.6% and 57.0 ± 2.6% maximal force, differences that were significant.

View larger version (31K): [in this window] [in a new window]   Fig. 2. Mean absolute peak force evoked from paralyzed (A) and control muscles (B) for all trains in all series as the last interpulse interval was varied from 500 to 5 ms. C and D: corresponding mean forces normalized to the maximal 50-Hz force for each subject. Series 1-6 included trains of 2-7 pulses, respectively. Dotted lines join data for each series. For clarity, standard error bars have only been included for series 1 and 6.

A similar sequence of seven pulses produced peak force in paralyzed and control muscles (Fig. 3A). The peak force in series 1 occurred at significantly shorter intervals (paralyzed: 8 ± 2 ms; control: 8 ± 2 ms) than in all subsequent series (paralyzed: 35 ± 5 ms, 36 ± 2 ms, 36 ± 3 ms, 34 ± 3 ms, 32 ± 3 ms; control: 43 ± 4 ms, 37 ± 5 ms, 38 ± 4 ms, 28 ± 1 ms, 35 ± 5 ms). There was no obvious interval that produced peak force in series 2-6, however. For different subjects, quite similar forces were evoked when the last interpulse intervals were varied between 5 and 80 ms (Fig. 2). We chose intervals that were intermediate within these ranges. Using this strategy, there was no significant difference in the intervals chosen for series 2-6.

View larger version (16K): [in this window] [in a new window]   Fig. 3. Mean ± SE interpulse intervals at which peak force (A) and force-time integral (B) were achieved in series 1-6 for paralyzed muscles of SCI subjects () and control muscles ().

With the addition of one more pulse in each series, there was a similar and near parallel increase in the relative force evoked in paralyzed and control muscles (35 and 34% maximal, respectively; Fig. 4A). The biggest force increments occurred in series 1, but thereafter declined (Fig. 4B). These data show that the differences in the relative force achieved in paralyzed and control muscles with two to seven pulses resulted from the greater force enhancement in paralyzed muscles with a short interpulse interval during series 1. 

View larger version (18K): [in this window] [in a new window]   Fig. 4. A: mean peak force (±SE) evoked during each series, expressed relative to maximal 50-Hz force. C: mean peak force-time integral (±SE) for each series normalized to the maximal force-time integral in series 6. B and D: the mean relative increases in peak force (B) and force-time integral (D) during each series for paralyzed muscles of SCI subjects () and control muscles ().

Maximal force-time integral

The mean absolute force-time integral for each train during all series are shown for paralyzed and control muscles in Fig. 5, A and B, respectively. Note the general increase in the force-time integral with the reduction in interpulse interval in series 1. The peak force-time integral during series 1 was similar for paralyzed (1.1 ± 0.4 Ns) and control muscles (0.9 ± 0.1 Ns) and occurred at mean intervals of 21 ± 8 ms and 12 ± 3 ms, respectively. Peak force-time integral occurred at much longer intervals in all subsequent series and typically when the force from the last pulse was unfused (Fig. 1). There was no significant difference in the maximal force-time integral achieved by series 6 for paralyzed (4.4 ± 1.4 Ns) and control muscles (4.5 ± 0.6 Ns). Figure 5, C and D, shows the force-time integrals normalized to the maximal values in series 6. Notice that the increment in force-time integral for each extra pulse was almost constant after the larger initial increase in series 1 (Fig. 4, C and D).

View larger version (36K): [in this window] [in a new window]   Fig. 5. Mean absolute force-time integral for paralyzed (A) and control (B) muscles for each train in all series plotted against the last interval in the trains. C and D: the corresponding data expressed relative to the maximal force-time integral for series 6. For clarity, standard error bars are only included for series 1 and 6. The large deflection in series 5 (A) resulted from muscle spasms.

The sequence of intervals that produced the maximal force-time integrals was similar for paralyzed (21 ± 8 ms, 73 ± 6 ms, 76 ± 5 ms, 86 ± 5 ms, 75 ± 5 ms, 74 ± 14 ms) and control muscles (12 ± 3 ms, 82 ± 9 ms, 103 ± 9 ms, 101 ± 10 ms, 93 ± 9 ms, 97 ± 11 ms; Fig. 3B). After the first series, however, the intervals that produced maximal force-time integrals were much longer than the intervals that produced peak force (Fig. 3). It should be noted that this study was designed to optimize peak force rather than peak force-time integral. Thus the values for the maximal force-time integrals in series 2-6 may have been different if this parameter had been used to determine the pulse sequence.

Different pattern of force production in one SCI subject

In one spinal cord-injured subject, there was no rise in peak force at shorter interpulse intervals in series 1 (+, Fig. 6). Rather, the force plateaued when the last interpulse interval was varied between 5 and 60 ms. This behavior was more typical of series 2-6 for all the other paralyzed and control muscles (Figs. 1 and 2). We therefore performed two experiments on this subject to examine how the force changed when the first interpulse interval was varied. The force evoked in the first experiment during series 1-3 (+, Fig. 6) closely matched that produced during the second experiment (, Fig. 6). Yet in the first experiment the initial interpulse interval in series 2-6 was short (+, 10 ms, the data included in the previous results). In the second experiment the first interpulse interval in series 2-6 was longer (, 45 ms). Thus this change in the initial interpulse interval did not alter the evoked force. These data also illustrate the day-to-day repeatability of the experimental setup for this subject. The forces recorded from this subject's thenar muscles were high compared with the means for paralyzed muscles (twitch force: 4.6 N vs. 2.5 ± 0.6 N; doublet force: 11.5 N vs. 6.8 ± 1.7 N; 50 Hz force: 34.7 N vs. 20.8 ± 6.4 N). However, the ratios between the forces for this particular subject were similar to the means calculated for all paralyzed muscles (twitch/tetanic ratio: 0.13 vs. 0.13 ± 0.02; doublet/tetanic ratio: 0.33 vs. 0.35 ± 0.03; doublet/twitch ratio: 2.5 vs. 2.7 ± 0.08).

View larger version (18K): [in this window] [in a new window]   Fig. 6. Force evoked from the paralyzed thenar muscles of 1 SCI subject during series 1-3 of 2 different experiments in which only the initial interpulse interval in series 2-6 was varied (, 45 ms; +, 10 ms).

All the other subjects participated in only one experiment. Thus the day-to-day repeatability of their optimal pulse patterns was not tested. However, the variability in the intervals that produced peak forces for all series was low across subjects (Fig. 3).

Twitch, doublet, and tetanic forces

Typical twitch, doublet, and 50-Hz forces evoked from the thenar muscles of one spinal cord-injured and one control subject are shown in Fig. 7, A and B, respectively. The forces have been normalized to the control tetanic force so that the relative magnitudes of the respective forces can be compared directly. The twitch/tetanic force ratio was the same for these two subjects (0.12), but the doublet/twitch and doublet/tetanic force ratios were much higher for the paralyzed muscles (2.91 and 0.34 vs. 2.06 and 0.25).

View larger version (23K): [in this window] [in a new window]   Fig. 7. Twitch, doublet, and 50-Hz force records from 1 SCI (A) and 1 control (B) subject. C and D: mean ± SE force ratios for paralyzed () and control muscles ().

The absolute mean twitch, doublet and 50-Hz forces were similar for paralyzed (2.5 ± 0.6 N, 6.8 ± 1.7 N, 20.8 ± 6.4 N) and control muscles (3.5 ± 0.4 N, 7.1 ± 0.8 N, 28.5 ± 2.5 N). Force data were more variable for paralyzed compared with control muscles, however. These data resulted in similar twitch/tetanic force ratios for each group (0.13 ± 0.02 vs. 0.12 ± 0.01), but significantly higher doublet/tetanic and doublet/twitch force ratios for paralyzed (0.35 ± 0.03, 2.73 ± 0.08) versus control muscles (0.25 ± 0.01, 2.07 ± 0.07; Fig. 7, C and D). These force ratios were not influenced by muscle fatigue because similar mean twitch and doublet forces were recorded during series 1 and after series 6 for each group of subjects.

The forces produced by paralyzed and control muscles in response to stimulation at different frequencies are depicted in Fig. 8. Although the maximal tetanic force varied widely for paralyzed muscles (Fig. 8A), the data from paralyzed and control muscles were similar when they were normalized to maximal force (Fig. 8B). The frequency required to generate half-maximal force was also comparable for paralyzed (14.3 ± 0.7 Hz) and control muscles (13.8 ± 1.0 Hz) even though twitch and doublet contraction times were significantly shorter for paralyzed (55.0 ± 2.1 ms and 77.6 ± 3.7 ms) versus control muscles (67.0 ± 3.1 ms and 94.1 ± 4.6 ms). There were no significant differences in twitch and doublet half-relaxation times between the subject groups (paralyzed: 52.2 ± 6.9 ms and 64.9 ± 8.0 ms; control: 56.6 ± 4.3 ms and 63.5 ± 4.2 ms).

View larger version (29K): [in this window] [in a new window]   Fig. 8. Absolute (A) and normalized forces (B) recorded from paralyzed muscles of SCI subjects () and control muscles (). Each curve shows data from 1 subject.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study shows that maximal force and force-time integral are elicited from paralyzed and control thenar muscles when a brief (5-15 ms) interpulse interval is followed by longer intervals. The short initial interpulse interval evoked almost twice the force increment in paralyzed muscles compared with control muscles (41 vs. 23% maximal). However, both types of muscle increased their force by an additional 34-35% when there were seven pulses in a train. Thus the force difference between paralyzed and control muscles with a train of pulses resulted from the initial closely spaced pair of pulses.

Pulse patterns for maximal force and force-time integral

After an initial brief interpulse interval, the intervals that generated maximal force in both types of muscles were always shorter than those that produced the maximal force-time integral. Similar results have been obtained for control thenar motor units (Thomas et al. 1999), motor units of cat hindlimb muscles (Burke et al. 1976; Zajac and Young 1980), and various cat, rabbit, and human muscles at optimal or long muscle lengths (Duchateau and Hainaut 1986a; Mela et al. 2001; Parmigianni and Stein 1981). This was the case when more global optimization of interpulse intervals was performed (Zajac and Young 1980) and with the more restrictive optimization performed in the present study. In contrast, short postdoublet intervals were deliberately chosen to reduce ripple in the force that was produced by unfused contractions (Karu et al. 1995) or were needed to maximize the force-time integral of rabbit tibialis anterior at short muscle lengths (Mela et al. 2001). Thus in most situations examined to date, similar pulse patterns maximize the force or force-time integral produced by single motor units and various whole muscles across species. It is interesting that these same patterns maximize force in muscles that have been paralyzed by spinal cord injury. We know that these paralyzed muscles are highly fatigable, are often active spontaneously and are sometimes weak compared with control muscles (Fig. 8) (Thomas 1997a,b). Alterations in synaptic inputs to motoneurons and changes in the intrinsic properties of motoneurons are also possible after spinal cord injury (Calancie 1991; Cope et al. 1986; Mayer et al. 1984; Munson et al. 1986; Zijdewind and Thomas 2001). Yet none of these factors strongly influence the pulse pattern that optimizes muscle force or force-time integral after chronic paralysis.

Maximal evoked force

A similar pattern of seven pulses did evoke different amounts of force from paralyzed and control muscles, however (76 and 57% maximal, respectively). This difference resulted from the initial short interval producing more force in paralyzed versus control muscles (41 and 23% maximal, respectively). Doublet/twitch and doublet/tetanic force ratios were therefore significantly higher for paralyzed compared with control muscles. This greater than linear summation of muscle force in response to a "doublet" has been proposed to involve both enhanced calcium release from the sarcoplasmic reticulum and the mechanical advantage of taking up the slack of the passive elements of the muscle (Duchateau and Hainaut 1986a,b; Hill 1949; Parmigianni and Stein 1981). It is likely that both these factors contribute to the doublet force increases that we observed in control and paralyzed muscles. However, differences in initial muscle stiffness may contribute to the greater relative forces elicited in paralyzed muscles by two closely spaced pulses. After stroke, triceps surae muscle stiffness is higher in the affected leg than in the nonaffected leg (Svantesson et al. 2000). In our experience, muscles paralyzed by spinal cord injury can appear and feel quite stiff compared with muscles of control subjects. Atrophy is relatively common (Thomas 1997a; Thomas et al. 1997). Much of the muscle bulk may be replaced to some extent by connective tissue (Williams and Goldspink 1984). Tendons have often shortened too, particularly those of flexor muscles (Thomas and Westling 1995). This stiffness change may be largely caused by alterations in the passive properties of the muscles, by different motor unit activity, or both (Dietz et al. 1981; Williams et al. 1988; Zijdewind and Thomas 2001). Thus it is reasonable to expect that there may be corresponding changes in the contractile properties of the muscle fibers that become evident when contractions are elicited by electrical stimulation.

Differences in stiffness may also explain why the relative doublet force produced by whole paralyzed muscles (41% maximal) was similar to that of control thenar single motor units (48% maximal) (Thomas et al. 1999). Both of these relative forces were stronger than that produced in control whole muscles in the current study (23% maximal). Since the control thenar units were a representative sample for control thenar muscles (Westling et al. 1990), one would expect similar calcium regulation for control thenar single motor units and whole muscles. Thus the force differences between control single motor units and whole muscles presumably relates to variations in stiffness introduced by the recording conditions (activation of a single motor unit in an otherwise relaxed muscle vs. contraction of the whole muscle).

After the initial brief interpulse interval, the increments in peak force were smaller but of similar magnitude for both control and paralyzed muscles. The more effective force increase from the doublet in paralyzed muscles was therefore maintained with a seven-pulse train even though any slack in the passive muscle elements had presumably been eliminated. This result suggests that doublet force differences between control and paralyzed muscles cannot be attributed to stiffness alone. Consistent with this suggestion is the observation that peak force was evoked in the paralyzed thenar muscles of one spinal cord-injured subject during the first series using a range of intervals (from 5 to 60 ms; Fig. 6) and not just with a short interpulse interval as in all other paralyzed and control muscles. In this particular case, the force enhancement from the initial short interpulse interval must have come from other means. Changes in calcium regulation (Duchateau and Hainaut 1986b; Parmiggiani and Stein 1981) with paralysis may play a role. Calcium uptake is slowed with chronic muscle disuse (Howell et al. 1997). The paralyzed muscles we studied were probably less active than they were prior to injury. Thus the calcium released by the first stimulus may not have been taken up in paralyzed muscles by the time a second stimulus was delivered. Calcium activation curves also differ for fast and slow muscle fibers (Stephenson and Forrest 1980). A slow to fast fiber type conversion has been well documented after human spinal cord injury (e.g., Burnham et al. 1997; Grimby et al. 1976). This shift in fiber type composition with injury may result in different calcium regulation for control and paralyzed muscles. Alterations in muscle-specific tension (force per unit area) may also contribute to the enhanced doublet force of paralyzed muscles. The specific tension of rat soleus muscle doubled following spinal cord transection (Lieber et al. 1986). The specific tension of different muscle fiber types may also differ, although this is controversial (Tötösy de Zepetnek et al. 1992). If this is the case, specific tension may well be different in paralyzed versus control muscles, given their different fiber type compositions.

Functional implications

The present study demonstrates that doublets, while beneficial for immediate, high force generation in control muscles, are even more effective at producing force in paralyzed muscles. Trains of pulses that begin with a doublet may therefore be useful for activating weak paralyzed muscles, particularly when near maximal activation is needed to achieve function. The force advantage that these optimized pulse trains provide over that produced by seven equally spaced pulses still needs to be explored. Furthermore, we do not know whether the excessive fatigability of paralyzed muscles (Lenman et al. 1989; Shields 1995; Stein et al. 1992; Thomas 1997b) can be reduced by the repeated delivery of these optimized pulse trains.

After the initial short interval, peak force and force-time integrals were produced with longer interpulse intervals. The pulse pattern for maximal activation of paralyzed muscles thus seems to closely mimic that used naturally by the CNS (Garland and Griffin 1999). It is also interesting that peak force was obtained with a range of intervals after the initial doublet. These data suggest that the intervals between the potentials of voluntarily activated motor units can vary quite widely (from 5 to 80 ms) without large changes in force. The force generated during voluntary contractions may be smooth if the motor units are activated asynchronously at low rates (Rack and Westbury 1969), even if these long interpulse intervals evoke unfused contractions.