Journal of Neurology, Neurosurgery, and Psychiatry 1990;53:208-214

Pathological stretch reflexes on the "good" side ofhemiparetic patients

A F Thilmann, S J Fellows, E Garms

Department ofNeurology andClinicalNeurophysiology,Alfried KruppKrankenhaus, Essen,Federal Republic ofGermanyA F ThilmannS J FellowsE GarmsCorrespondence to:Dr A F Thilmann,Department of Neurologyand ClinicalNeurophysiology, AlfredKrupp Krankenhaus, Alfred-Krupp-Strasse 21, D-4300Essen 1, Federal Republic ofGermany.Received 24 January 1988and in revised form 12 May1989.Accepted 6 September 1989

AbstractThe reflex EMG responses from a tendontap or an imposed, medium amplitude(300), stretch at a range of stretchvelocities have been recorded from thetriceps and biceps muscles of normalhuman subjects and in both the affectedand "unaffected" arms of hemipareticpatients under relaxed conditions. In thehemiparetic arm, exaggerated tendonjerks were, as expected, observed in bothmuscles. The response of the biceps toelbow extension was also exaggeratedcompared with normal values and dis-played both an additional, earlier com-ponent and a much reduced velocitythreshold. The triceps, in contrast,showed depressed responses to elbowflexion, with a much higher velocitythreshold than normal subjects. Further-more, on the supposedly "unaffected"side of the hemiparetic subjects, thereciprocal pattern was seen, with depres-sion ofthe biceps response and a raising ofits threshold, along with considerablyexaggerated responses in the tricepsincluding earlier components not seen inthe normal subjects. The increased ex-citability ofthe flexor musculature on thespastic side may be paralleled by in-creases in activity in the segmental path-ways responsible for modulation ofagon-ist/antagonist activity in the ipsi and con-tralateral limb, leading to an inhibition ofthe ipsilateral extensors and contra-lateral flexors and excitatory input to thecontralateral extensors. Thus the "good"side of hemiparetic patients also receivespathological changes, and studies of themechanisms of spasticity should avoidthe use of the "unaffected" side of hemi-paretic subjects as a control for monitor-ing pathological reflexes.

Spasticity is a motor disorder which occursafter CNS lesions affecting the pyramidal tractat a variety of locations. Clinically it is definedby exaggerated tendon jerk reflexes, increasedmuscle tone (velocity-dependent increases inthe resistance to movement), muscle weaknessand, in some cases, clonus.'

If the cause of the spasticity is an ischaemiclesion in the vicinity ofa middle cerebral artery,the disability is visible clinically only on thecontralateral side of the body. Given the over-whelmingly one-sided nature of this deficit, it

is perhaps not surprising that little study hasbeen made of the "unaffected" side ofhemiparetic patients. Indeed, it has becomecommon practice to use the "good" side as a

fully matched control for the description ofchanges on the affected side.23 This studyconcerns the responses to mechanical distur-bance of antagonist muscle groups at the elbowon both sides of hemiparetic subjects and thecomparison of these responses to those of a

group of normal subjects. The results indicatethat the so-called "unaffected" arm of thehemiparetic subjects does not respond with thenormal pattern, and that pathophysiologicalchanges have occurred on both sides ofhemiparetic subjects.

MethodsExperiments were performed on a group of tenhemiparetic patients (for clinical details, seetable 1) and a group of ten age-matched normalsubjects, all of whom gave their informedconsent to the procedures employed. Allprocedures were approved by the local ethicalcommittee. The subjects were seated side on

(see fig la) in an adjustable chair and the armunder study was placed in a moulded plasticcast (c) which restrained, by means ofa series ofstraps, the forearm and hand. The cast was inturn attached to an aluminium support moun-ted on a bearing (b), its position being adjustedso that the elbow joint lay directly over the axisof rotation. The height of the chair was adjus-ted to bring the elbow and shoulder joint to thesame level, with the subject's back and shoul-der supported by adjustable rests. Dis-placements were applied by a powerful electricmotor (Servalco, MC24P(3000W)) with 4-quadrant servo control system (Seidel, series5000), via a rigid connecting rod (r) attached tothe cast at the level of the wrist. The motor wasstrong enough to impose fully stereotypedmovements, irrespective ofarm inertia or resis-tance generated in the subject's arm. Thus, fora given set of motor parameters, an identicaldisplacement was applied to all subjects. Dis-placement amplitude was 30°. Extending stret-ches moved the arm from 75° flexion to 105°flexion, flexing stretches from 105° to 75°, 90°flexion corresponding to a right angle at theelbow joint. Displacement velocities up to2620/s were applied, under the control of alaboratory computer (DEC, PDP1 1/73). Posi-tion and velocity curves for the extendingstretches are shown in fig lb. All velocitycurves were calculated by the computer direc-

208

Pathological stretch reflexes on the "good" side of hemiparetic patients

Table 1 Clinical details of hemiparetic subjects

Case Age Localisation Imaging Duration ofno Sex (yr) of ischaemia' method' spasticity Paresis3 Tone4

1 f 19 l.mca CT, MRI, A 7 weeks 0 02 m 32 r.mca CT, A 15 months 3 33 m 40 l.mca CT, A 4 weeks 0 14 m 67 l.mca CT, A 4 weeks 0 05 f 40 r.mca CT, A 6 weeks 0 06 m 56 l.mca CT, A 3 months 2 37 f 49 r.mca CT 4 weeks 1 28 m 35 r.mca CT, A 8 months 3 39 f 24 r.mca CT, A 16 months 4 210 f 41 r.mca CT 9 months 0 1

'l.mca = left middle cerebral artery. r.mca = right middle cerebral artery.'Imaging method(s) used to establish lesion localisation: CT = computed tomography; MRI = magnetic resonance imaging;A = angiography.3Paresis of the hemiparetic arm, expressed according to the British MRC scale: 0 = no voluntary power; I = visible contractionwithout effect; 2 = movement without the influence of gravity; 3 = movemnent against gravity; 4 = movement against resistance;5 = normal.'Tone of the hemiparetic arm, expressed on the Ashworth scale: 0 = normal; 1 = slight increase in tone, giving a "catch" whenthe limb is moved; 2 = more marked increase in tone, but limb easily flexed; 3 = considerable increase in tone-passive movementdifficult; 4 = limb rigid in flexion or extension.

tly from the position records. Due to themechanical construction of the equipment, thesame motor pulses produced slightly higherpeak velocities during extending stretches thanduring flexing stretches. Ten stretches wereapplied at each velocity, at random intervalsbetween 5 and 8 seconds. The subjects wereinstructed to remain fully relaxed and not toreact to the displacement.Displacement was monitored using a light

sensitive detector (United Detector Tech-nology, LSC5D), which followed themovements of a light source rigidly attached tothe drive shaft. This detector yielded a signalwhich, after suitable processing, could resolvemovements of <0.06° at the elbow, with nodetectable phase shift. Electromyograms(EMG's) were recorded from the biceps andtriceps muscles using custom built surfaceelectrodes which contained a small pre-amplifier and were designed to reducemovement artifacts to a minimum. The signalsobtained were amplified and filtered (band-width 20Hz to 1KHz), then passed, along withthe position signal, to the ADC boards of thecomputer, which sampled at a rate of 1KHz foreach channel. Both EMG's were then rectified

and averages constructed from the ten dis-placements applied at each stretch velocity.Position, raw EMG data and the motor controlpulses were also recorded on tape (Racal,Store7D).At the start of each experimental series,

tendon jerk reflexes were obtained from thebiceps and triceps muscles, using a reflexhammer containing a microswitch which trig-gered computer sampling. Five responses wereobtained from each muscle and peakamplitude, onset latency and duration sub-sequently determined.For the displacement experiments, each

averaged record was analysed offline to yieldthe peak EMG amplitude, onset latency andduration of the responses. Peak EMGamplitude was chosen as a measure of responsemagnitude after trial measurements showedthat this value was strongly correlated to theintegrated area of the EMG response. Onsetlatencies of the responses were measured fromthe onset of movement, not from the controlpulse to the motor, which preceded movementonset by 10 ms. The statistical significance ofdifferences between the groups was assessed byvariance analysis.

Figure 1 (A)Displacement apparatus.For explanation ofsymbols, see text.(B) Position (lower) andvelocity (upper) curvesforthe five rates of extendingdisplacement applied.

Q

209

Thilmann, Fellows, Garms

ResultsElbow ExtensionIn the normal subjects, under relaxed condi-tions, elbow extension at velocities greater than200'/s resulted in a burst of activity in thebiceps EMG in eight out of 10 cases. In theremaining two subjects no response wasobtained, even with the maximum rate ofstretch applied (262'/s). Figure 2a shows theresponses of a single subject over a range ofstretch velocities. The response was generallysmall (average 38 yV, SE13). No correlationexisted between its magnitude and the mag-nitude of a particular subject's biceps tendontap response. The response occurred with alatency of between 40 and 70 ms (average56 ms, SE3) and lasted for between 25 and45 ms (average 30 ms, SE3). As the stretchvelocity was lowered, the response disap-peared, usually when stretch velocity fell below200'/s. In the subject illustrated, a small res-ponse remained at 172'/s, but in no case was aresponse seen if stretch velocity was slowerthan 165'/s.On the affected side of the hemiparetic

patients, as would be expected from clinicalexperience, the tendon tap reflex was exag-gerated in both the elbow extensor and flexormuscles, in the latter to a significantly higherlevel (p < 0 001) than that seen in the normalsubjects. In addition, it was apparent that theresponse to rapid elbow extension was alsoexaggerated. In all of the patients studied, theEMG of both the elbow flexors and extensorsshowed no sign of background activity at theonset of the displacements and no subject whoexhibited contractures was selected for thisstudy. Figure 2b shows the response in thebiceps EMG of a typical patient to the samerange of stretch velocities applied to the normalsubject in fig 2a. The shaded lines represent thestandard error bars of the group average foronset latency and duration of the normal

subjects. It can be seen that the response is notonly larger, but much more complex in form:while a large part of the response (averageamplitude at 262'/s: 82 ,uV, SE22) occurswithin the time period of the normal response,activity is apparent both before and after thisperiod. The late EMG activity, occurring after90-100 ms, cannot be assumed to be purelyreflex in nature, and its origin and significancewill be considered in a later paper. The earlieractivity (average amplitude at 262°/s: 60 yV,SE15), appearing 25-35 ms (average 27 ms,SEI) after the onset of movement, was notobserved in a normal subject, but was seen inthe biceps EMG of the affected side in sevenout of 10 of the hemiparetic subjects. As thestretch velocity was lowered, the mediumlatency activity, as in normals, quickly disap-peared, never being observed with stretchvelocities below 200'/s. The early activity,however, persisted and appeared followingstretches as slow as 117'/s, well below thethreshold of the most sensitive of the normalsubjects (165'/s).A similar range of displacement was applied

to the "unaffected" arm of these hemipareticpatients. Somewhat surprisingly, in all but twoof the subjects, no response to extendingstretch was seen. Similarly, the biceps tendontap reflex was significantly smaller (p < 0-001)than that of the normal subjects. Figure 3shows the average EMG amplitude followingbiceps stretch plotted against stretch velocityfor the normal subject (n) and for the spastic (s)and "good" (g) arm of the hemiparetic, withthe corresponding tendon jerks. For the affec-ted side of the hemiparetic, the early (sI) andmedium (sII) latency activity is shownseparately. It can be seen that while the affectedarm of the spastic subjects responds moremarkedly (p < 0-01) and at lower velocities ofthe stretch than the normal arm, the so called"unaffected" side shows significantly less

Figure 2 The response ofthe relaxed biceps muscleto imposed extension of theelbow at a variety ofstretch velocities (showncentre) in a typical normalsubject (A) and on theaffected side of ahemiparetic subject(B). Each EMG trace isthe rectified and averagedresponse to 10displacements. The shadedlines in (B) represent themean onset and offsetlatency of the responses ofthe group of normalsubjects (SE).Displacement amplitude300.

210

211Pathological stretch reflexes on the "good" side of hemiparetic patients

Figure 3 Group averagesof the EMG amplitude(with SE bars) of thebiceps tendon reflex(BTR) and of theresponse of the relaxedbiceps to imposed extensionat a range of stretchvelocities for the normalsubjects (n) andfor thespastic (s) and "good"(g) side of thehemiparetic. Stretchvelocity is shown beneatheach bar. On the affectedside of the hemiparetic,amplitude of response isgiven for activity, s(II),occurring within theaverage normal latencies(see legend to fig 2) andfor the earlier activity,s(I).

>

-0

Ql.O)CO

o In s g

BTR

activity (p < 0-001) than the affected side, butalso less than the normal subjects (p < 0 005),indicating that changes from the normal statehave also occurred in this arm.

ElbowflexionIn the normal subjects, elbow flexion producedmore considerably EMG responses in thetriceps than elbow extension did in the biceps.Only one subject showed no response to stretchand the responses persisted at lower stretchvelocities. Figure 4a shows the triceps EMGresponses of a representative normal subjectover a range of stretch velocities. At the mostrapid rate of stretch, a burst of activity can beseen (average amplitude 89 ,uV: SE27), appear-ing with an average latency of 58 ms (SE6) andaverage duration of 36 ms (SE3). As stretchvelocity was reduced, this componentgradually reduced in both amplitude and dura-tion, but was not abolished until stretchvelocity fell below 100°/s.

172 117 80 262 234 172 117 80 262 234 172 117 80

s (I) s (11) 9

On the affected side of the hemipareticsubjects, stretch of the triceps, even at the mostrapid stretch velocities applied, produced a

reflex response in the triceps EMG in only50O/, of the cases. In contrast, the tendon jerkstended to be exaggerated over normal values,although not significantly (fig 5). In thosesubjects who did show a response to imposeddisplacement, it was significantly smaller(p < 0-025) than in the normal responses

(average amplitude at 252'/s: 23 ,uV: SE 12), butoccurred with similar latency and duration.The threshold of this response, however, was

considerably higher than normal values, withno responses being seen for stretch velocitiesbelow 150°/s and only one subject showing a

response at velocities of stretch lower than200'/s.On the "unaffected" side of the hemiparetic

subjects, the pattern of response was, as for thebiceps, far from the norm. Figure 4b shows theresponses of a typical "good" side at a range of

Figure 4 The response ofthe relaxed triceps muscleto imposedflexion of theelbow at a variety ofstretch velocities (showncentre),for a typicalnormal subject (A) andfor the "unaffected" sideof a hemiparetic subject(B). Each EMG trace isthe rectified and averagedresponse to tendisplacements. The shadedlines in (B) represent themean onset and offsetlatency of the responses ofthe normal subjects (SE).Displacement amplitude30°.

Ir I

I. 25 :n0 5 I )Oc 125;.--- -r-(.,C.r'. I i I

.1 2 n, U

Thilmann, Fellows, Garms

Figure 5 Group averagesof the EMG amplitude ofthe Triceps tendon reflex( TTR) and of theresponse of the relaxedtriceps to imposedflexionat a range of stretchvelocities for the normalsubjects (n) andfor thespastic (s) and "good"(g) sides of thehemiparetic. On the"unaffected" side of thehemiparetic, amplitude ofresponse is given foractivity occurring withinthe average normallatencies, g (II), andforthe earlier activity, g(I).

n s gTTR

stretch velocities. As in fig 2, the shaded linesrepresent the standard error of the groupaverage for onset latency and duration of thenormal subjects. It shows that the dominantpart of the response occurs within this timeperiod. Magnitude and duration of this com-ponent do not differ from normal values. Inaddition, however, it may be seen that shorterlatency activity is present (average latency at252'/s: 29 ms SEl; duration: 27 ms SE1). Suchactivity was present on the "unaffected" side inhalf the hemiparetic subjects, but unlike thesituation in the flexor musculature on theaffected side, it was never accompanied by late,tonic activity. Figure 5 shows the triceps EMGamplitudes of the spastic (s) and "good" (g)sides of the hemiparetic subjects and those ofthe normal subjects (n). As can be seen, thethreshold of the activity on the "good" side didnot differ significantly from that of the normalresponse. The threshold of the response on theaffected side, however, was considerably raisedand its amplitude significantly smaller than inthe normal subjects (p < 0025) and on theunaffected side of the patients (p < 0005).

If figs 3 and 5 are compared, a clear picture ofinteraction between the two sides of the hemi-paretic subjects becomes apparent: on thespastic side, the magnitude of the flexor res-ponse is increased relative to normal values andits threshold lowered, while the response of theextensor musculature is depressed and itsthreshold raised. On the "unaffected" side ofthe spastics, the reciprocal pattern is seen, withdepression of the flexor and increase in theextensor response.

DiscussionOn the affected side of the hemipareticsubjects, as would be expected from typicaldescriptions of spasticity, the response of thebiceps to stretch was strongly exaggerated.

252 215 164 113 72 252 215 164 113 72

g (1) g (11)

This EMG activity was also more complex inform, resembling the modulated response thatwould, in normal subjects, be obtained onlyfrom the stretch of a voluntarily activatedmuscle.4 In all cases, however, the spasticsubjects were relaxed at the time of dis-placement, with their muscles electricallysilent. Furthermore, this pattern was alsoobtained in subjects who had no voluntarypower at the arm, suggesting that the patterndoes not depend on descending voluntarydrive. The occurrence of this more complexpattern in the relaxed arm seems to represent adifference from the situation at the hemipareticankle, where it was uncommon to find morethan a single component in the relaxed state.5Nevertheless, to discuss the pattern observedin this study in terms of the M1/M2 pattern(described solely for voluntary activated mus-cle) and the changes that may occur in it due tosupraspinal lesions would seem unlikely toclarify our findings, given the many differencesbetween the motor system in the relaxed andthe voluntarily activated state. A breakdown inthe phasic nature of the agonist burst in a rapidvoluntary elbow flexion movement, leading to aprolongation of the burst has recently beenreported in hemiparetic subjects6 and it couldbe argued that the current changes are moreanalogous to this loss of modulation than theincidence of new reflex components. An objec-tion to this argument is that the EMG responseon the hemiparetic side is not only prolonged,but contains a component occurring with aconsiderably shorter latency. This would sug-gest that, somewhere in the reflex pathway, thethreshold at which a component begins torespond has fallen and/or the magnitude of itsresponse considerably increased. Given that noevidence has yet been found for increasedsensitivity of muscle spindles in spasticity,7 itwould seem more likely that these changes haveoccurred centrally, probably at a spinal inter-

212

s

Pathological stretch reflexes on the "good" side of hemiparetic patients

neuronal level. This idea is supported by thedifferences in the behaviour of the tendon jerkresponses and the response to imposed dis-placement in the hemiparetic arm: while thetendon jerks of both biceps and triceps areexaggerated, the response to displacement isenhanced in the biceps but strongly depressedin the triceps. As the tendon jerk is known to bemediated predominantly (but not exclusively)over a monosynaptic pathway,8 it seems clearthat the modulation of the responses toimposed stretch, which is not reflected in asimilar modulation of the tendon jerks, must beoccurring at a higher level than the motorneuron pool. More precise location ofthe site ofthis action is not possible from the presentresults. One conclusion to be drawn from thediffering behaviour of the tendon jerk and theresponses to imposed stretch is that the tendonjerk response does not reveal the full complex-ity of the changes in the reflex system inspasticity and may not provide a true reflectionof the likely magnitude of the reflex response toa more physiological stimulus.

In the triceps of the affected arm, in contrastto the enhanced responses seen in the biceps,the reflex response to imposed displacement isboth smaller to that of the normal subjects anddisplays a much higher velocity threshold.Such a modulation of excitability in anantagonist muscle group in spasticity is wellknown from studies of the lower limb, wherethe relative weakness of the pretibial musclesand hyperactivity of the ankle extensors is animportant contributing factor to the impair-ment of gait in spasticity.9 " This imbalance hasbeen attributed, in the relaxed state, toincreased reciprocal inhibition of the pretibialmuscles by the ankle extensors." It is reason-able to assume that a similar mechanism is inoperation in the muscles of the upper arm,given the general principle that an increase inexcitation in a muscle group is accompanied byparallel changes in the activity of its associatedsegmental pathways, leading to increasingactivity in excitatory pathways to synergistsand in inhibitory pathways to antagonists.'2The appearance of the reverse pattern of

changes on the supposedly "unaffected" side ofthe hemiparetic subjects has not, to ourknowledge, previously been reported. All of thesubjects studied were selected because theirlesions were clearly localised by CT scanningand additionally, in most cases, by NMRtechniques and angiography and in all cases thesymptoms of spasticity were one-sided. Never-theless, a marked depression of responses in thebiceps and exaggerated responses in the tricepswas seen on the "unaffected" side of all of thehemiparetic subjects. As this pattern wasclearly reciprocal in the hemiparetic arm, it istempting to attribute these changes to inter-limb innervation at a spinal level, specificallyover pathways which normally deal with inter-limb coordination in complex movement pat-terns,'2 with the increased excitability of theaffected biceps leading to inhibition of thecontralateral flexors and excitation of theextensors. Our results do not, however, allowthe exclusion of other mechanisms, including

interactions at supraspinal levels. Such alter-natives cannot be excluded, as it is known that,while the greater part of pyramidal tract fibrescross to the contralateral side, around 10%oremain on the ipsilateral side of the supraspinallesion,'3 and may thus directly influence the"unaffected" side. Brodal," has suggested thatdisturbances of fine motorcontrol on the"unaffected" side of a hemiparetic subject mayarise from disruption of cortico-cerebellarpathways with bilateral actions on motorneurons: again, a possible contribution fromthis source cannot be excluded.The changes in the supposedly "good" side

of a hemiparetic subject are of significance forneurophysiological studies of spasticity, whichhave, in the past, commonly utilised the "unaf-fected" side of hemiparetic subjects as a con-venient, matched control for the description ofthe pathophysiological changes that occur inthe upper motor neuron syndrome. Thepresent findings indicate strongly that thispractice should be discontinued. The role ofthese changes in the disability of the patientshas still to be clarified. The patients themselvesreported no problems on their "unaffected"side, but given the severity of the disability onthe hemiparetic side, such judgements of nor-mality must clearly be viewed as subjective. Itseems desirable to investigate the extent ofreflex dysfunction on the "unaffected" side, toestablish the extent to which "hemiparesis" ismerely a term for naming the most severelyaffected side following a localised supraspinalinsult.A final minor point to be considered is that,

in the relaxed state, the triceps muscle of thenormal subjects responded more markedly,and at lower velocities of stretch, than did thebiceps. This is, at first sight, somewhat sur-prising: from the principle of physiologicalextensor dominance, proposed by Sherrington,it is to be expected that the biceps, which is themajor anti-gravity muscle at this joint, wouldshow the most marked reflex responses. If theamplitude of the reflex responses was the solebasis for comparison, it could be argued thatthis discrepancy could be attributed to the useof surface electrodes for EMG recording. Thetriceps, however, also responds at much lowerstretch velocities, suggesting that this dif-ference in reflex excitability cannot be accoun-ted for entirely on these grounds. A possibleexplanation could lie in the different mor-phology of these muscles: the biceps hasproportionally, much longer tendons than thetriceps,'5 which will tend, in the relaxed state,to reduce the transmission of the imposedstretch to the muscle receptors, and hence themagnitude of the reflex response.16 Thus if, dueto its longer tendon, more of the early part ofthe movement was lost in taking up tendonslack, the biceps muscle spindles might beexposed to lower acceleration rates than thoseof the triceps, despite the attainment of thesame peak stretch velocity in the two move-ments. The presence of such tendon com-pliance might also explain the relatively longlatencies of the responses, although it would beexpected from work on the relaxed triceps

213

Thilmann, Fellows, Garms

surae,'7 that a greater difference in latencybetween the biceps and triceps response wouldhave been observed.

We thank Professor J Noth for his helpfulcomments on the manuscript, U Arenz fortechnical assistance and H F Ross and NSchaller for software support. This work wassupported by a grant from the Deutsche Fors-chungsgemeinschaft. SJF is in receipt of aresearch fellowship from the Alexander vonHumboldt Stiftung.

1 Lance JW. Pathophysiology of spasticity and clinicalexperience with Baclofen. In: Feldman RG, Young RRand Koella KP, eds. Spasticity: Disordered motor control.Chicago: Year Book medical publishers, 1980:185-203.

2 Becker WJ, Hayashi R, Lee RG, White D. Modulation ofreflex and voluntary EMG activity in wrist flexors bystimulation of digital nerves in hemiplegic humans.Electroencephalog Clin Neurophysiol 1987;67:452-62.

3 Lee WA, Boughton A, Rymer WZ. Absence of stretch reflexgain enhancement in voluntarily activated spastic muscle.Exper Neurol 1987;98:317-35.

4 Lee RG, Tatton WG. Motor responses to sudden limbdisplacements in primates with specific CNS lesions andin human patients with motor system disorders. Can JNeurol Sci 1975;2:285-93.

5 Berardelli A, Sabra AF, Hallett M, Berenberg W, Simon SR.Stetch reflexes of triceps surae in patients with upper

motor neuron syndromes. J Neurol Neurosurg Psychiatry1983;46:54-60.

6 Fagioli S, Berardelli A, Hallett M, Accornero N, ManfrediM. The first agonist and antagonist burst in patients withan upper motor neuron syndrome. Movement Disorders1988;3: 126-32.

7 Burke D. Critical examination of the case for or againstfusimotor involvement in disorders of muscletone. In:Desmedt JE, ed. Motor control mechanisms in health anddisease. New York: Raven Press, 1983:133-50.

8 Burke D, McKeon B, Skuse NF. Dependence ofthe Achillestendon jerk on the excitability of spinal reflex pathways.Ann Neurol 1981;10:551-6.

9 Dietz V, Berger W. Interlimb coordination of posture inpatients with spastic paresis. Brain 1984;107:965-78.

10 Corcos DM, Gottlieb GL, Penn RD, Myklebust B, AgarwalGC. Movement deficits caused by hyperexcitable stretchreflexes in spastic humans. Brain 1986;109:1043-58.

11 Yanagisawa Y, Tanaka R, Ito Z. Reciprocal inhibition inspastic hemiplegia ofman. Brain 1976;99:555-74.

12 Baldiserra F, Hultborn H, Illert M. Integration in spinalneuronal systems. In: Brooks VB, ed. Handbook ofPhysiology, sec 1. Vol 2. Pt 1. Bethesda: AmericanPhysiological Society, 1981:509-95.

13 Flechsig P. Anatomie des menschlichen Gehirns undRfickenmarkes auf mylogenetisches Grundlage. Leipzig:Georg Thieme, 1920.

14 Brodal A. Self-observations and Neuro-anatomical con-siderations after a stroke. Brain 1973;96:675-94.

15 Fellows SJ, Rack PMH. Changes in the length of the humanBiceps brachii muscle during elbow movements. JPhysiol1987;383:405-12.

16 Rack PMH. Stretch reflexes in man: the significance oftendon compliance. In: Barnes WJP, Gladden MH, eds.Feedback and motor control in invertebrates and vertebrates.London: Croon Helm, 1985:217-29.

17 Fellows SJ, Noth J, Prisett M, Thilmann AF. Stretch reflexlatency at the relaxed normal human ankle is dependent ontendon compliance. J Physiol 1987;394: 101

214