12-20,u as the motor fibres. They also showed that the afferent fibres from skin

Transcription

1 436 J. Physiol. (1956) I3I, I THE RELATIVE EXCITABILITY AND CONDUCTION VELOCITY OF SENSORY AND MOTOR NERVE FIBRES IN MAN BY G. D. DAWSON From the Medical Research Council, Neurological Research Unit, The National Hospital, Queen Square, London (Received 18 August 1955) Eccles & Sherrington established in 1930 that in the hindlimb of the cat the larger afferent nerve fibres from muscle fell into the same group of fibre size, 12-20,u as the motor fibres. They also showed that the afferent fibres from skin in a dorsal digital nerve lay in a group between 8 and 12ju in size. Although the relative size of motor, muscle afferent and skin afferent nerve fibres has not been studied in man in an experimental situation comparable with that used by Eccles & Sherrington in the cat, there is evidence suggesting that their findings are not applicable to man. Sunderland, Lavarack & Ray (1949) have shown that cutaneous nerves in the arm in man contain many large fibres; in one superficial radial nerve 15 % of the fibres were found to be between 20 and 23,u in size. In addition to this, Kugelberg (1944) observed that a strength of electrical stimulation can be found which, when it is applied to the median or ulnar nerves near the wrist, will produce sensations referred to the peripheral distribution of these nerves without causing a detectable twitch in the muscles of the hand. This observation suggests that the sensory afferent nerve fibres from the fingers may be as large as, or larger than, the largest motor fibres to the small muscles of the hand. If the afferent fibres from the small muscles of the hand are in the same size group as the motor fibres, then they also will be no larger than the afferent fibres from the fingers. Kugelberg's observation may easily be repeated, but for reasons which will be considered later it must be interpreted with caution. The purpose of this paper is to present further evidence, obtained by examining the relative electrical excitability and conduction velocity of sensory and motor nerve fibres in man. This evidence supports that interpretation of Kugelberg's findings which suggests that the afferent nerve fibres from the fingers in man are as large as the largest motor or muscle afferent fibres from the small muscles of the hand. This fact must be considered in interpreting the observed effects of ischaemia, injury

2 NERVE CONDUCTION VELOCITY IN MAN 437 or toxic substances on the function of peripheral nerve in man and in the analysis of the central effects of electrical stimulation of peripheral nerve. METHODS The methods used for stimulating and for recording nerve action potentials through the skin were, with minor modifications, those described previously (Dawson & Scott, 1949). The arrangements of the stimulating and recording electrodes for examining the median nerve are shown in Fig. 1. When measuring the afferent conduction time between wrist and elbow the stimulus was applied to electrodes 1 and 2 on the index finger, and the ascending volley in the median nerve was recorded from the two pairs of electrodes, 5 and 6, and 7 and 8, over the nerve trunk. An earth connexion was made to a plate, E 1, on the dorsum of the hand. For measuring motor fibre conduction time from elbow to wrist the stimulus was applied first to electrodes 7 and 8, then to electrodes 5 and 6, and the action potentials in some of the small muscles of the hand supplied by the median nerve, usually in the thenar eminence, were recorded from electrodes 3 and 4 for each position of the stimulus Fig. 1. Diagram of the electrode arrangements for stimulating and recording. The continuous lines indicate that when an electrode connected to one was used for stimulating, it was the cathode of the pair; when the electrode was used for recording, relative negativity of that electrode caused an upward deflexion in the record. To compare the excitabilities of the sensory and motor nerve fibres at the wrist a stimulus was applied to electrodes 5 and 6, either alone or followed by another applied to electrodes 1 and 2, the muscle action potentials were recorded from 3 and 4 and the ascending nerve action potentials from 7 and 8. In this situation the earth connexion was made to a plate, E 2, on the flexor aspect of the forearm. Cerebral action potentials evoked by the afferent volley were recorded from a pair of scalp electrodes, 9 and 10, 5 cm apart over the contralateral sensorimotor area using the averaging methods which have been described elsewhere (Dawson, 1953, 1954). For examination of the ulnar nerve, electrodes 1 and 2 were put on the little finger and 3 and 4 on the hypothenar eminence over the abductor of the little finger. Electrodes 5 and 6, and 7 and 8 were also placed over the ulnar trunk just above the wrist and elbow. Electrodes 1 and 2 were silver strips 10 mm wide, coated with silver chloride, covered with lint and soaked with brine. They were bent to encircle the finger and stimulate the digital nerves. Electrodes 3and 4 for recording the muscle action potentialswere eithersmallchloride-coated silver plates 16 by 24 mm, held on with adhesive tape and making contact through electrode paste, or brine-soaked pads 15 mm in diameter. For picking up the nerve action potentials electrodes 6 and 8 were of the strip or T type previously described, but 5 and 7, since they had also to be used as stimulus cathodes, were replaced by spherical pads 15 mm in diameter to reduce the area of contact and increase the density of the stimulating current. Sometimes it was necessary to scratch the skin lightly under electrodes 5 and 6 to minimize the resistance between them as this helped to reduce the direct pick-up of the stimulating current from electrodes 1 and 2. Also, when the stimulus was being applied to a finger the nerve action potential which could be recorded was

3 438 G. D. DA WSON considerably smaller than that produced by a stimulus to a nerve trunk at the wrist; very good relaxation was therefore necessary to get useful records unobscured by muscle action potentials. For this reason the subjects were examined lying supine on a comfortable couch with the arm slightly abducted and supported at three or four places by broad webbing straps hung from a beam. Except where stated, all the subjects examined were healthy adults. All measurements have been made from records of at least twenty traces superimposed or, in the case of the cerebral evoked potentials, from a record of the average of as many responses as were necessary to give repeatable results at the prevailing level of spontaneous brain activity. For reasons which will be discussed later, no control of limb temperature was carried out. RESULTS Relative excitability of sensory and motor fibres Peripheral effects of a single stimulus. When a brief electrical stimulus was applied to the median or ulnar nerve trunk at the wrist it was found that many subjects reported paraesthesiae radiating into the fingers when no sign could be found of a muscle twitch in the hand. This was entirely in accord with Kugelberg's (1944) observation, and he quotes Erb as having made the same observation in In some instances, however, a few motor units were excited by the stimulus when it was well below threshold for the great majority of motor fibres and after the subject had ceased to report sensations in the fingers. Also, it was found that a stimulus which was just above threshold for sensation with the cathode in one position over the nerve, might stimulate only motor fibres when the cathode position was changed slightly. These effects are probably partly due to the vagaries of the current distribution in the relatively large and heterogeneous mass of tissue round the nerve. They may also be due to the separation into fascicles by fibrous tissue of the nerve fibres from different parts of the hand; the subject has been investigated and reviewed by Sunderland (1945). For these reasons the stimulating cathodes over the nerve trunk at 5 and 7 (Fig. 1) were placed so as to give the best possible stimulation of the motor fibres to the muscles from which the action potentials were being recorded. This decreased the chance of the sensory fibres appearing falsely to have a lower threshold than the motor fibres, but increased the number of instances where some motor excitation was found with stimuli below threshold for sensation. Stimuli about threshold only allow comparison of the excitabilities of the most sensitive, or most accessible, motor and sensory fibres. A clearer picture may be obtained if a wider range of stimulus strengths is used, from threshold to maximal for motor fibres, and the variation in the number of fibres excited is estimated. This may be done by stimulating at the wrist, recording the muscle action potentials in the hand and the ascending nerve volley above the elbow. Sets of records taken in this way from the ulnar nerve are shown in Fig. 2. Records A to D are from the ulnar nerve in a healthy adult, and E to H from the ulnar nerve in a child with a severe sensory neuronopathy. The upper

4 NERVE CONDUCTION VELOCITY IN MAN 439 record of each pair shows the muscle action potentials, from electrodes 3 and 4, due to the nerve volley descending from the site of stimulation; and the lower record, from electrodes 7 and 8, is of the ascending nerve volley. With a stimulus maximal for motor fibres, such as that in Fig. 2 A, the ascending volley will consist of a mixture of antidromic impulses in the motor fibres and orthodromic impulses in any sensory fibres of threshold the same A lb D mv E F ~~~~~G H E2 7 8 Fig. 2. The upper trace shows superimposed records of the muscle action potentials in abductor minimi digiti produced by a stimulus to electrodes 5-6 over the ulnar nerve at the wrist. The lower trace, on a faster time base, shows the centripetal nerve volleyrecorded above the elbow. Records A to D are from a healthy subject, and E to H from a patient with no histologically demonstrable large afferent fibres. Stimuli reduced from motor maximal in A and E to motor threshold in D and H. The top time scale for the muscle action potentials shows 1 and 5 msec. The bottom time scale for the nerve action potentials shows 0-1 and 1 msec. The notch in the top time scale in this and later records shows the relative duration and timing of the top and bottom time scales. as or lower than that of the motor fibres. When the stimulus was reduced to about threshold for the motor fibres (Fig. 2 D) in the healthy subject an ascending action potential was recorded some 25 % of the amplitude of that found with the stimulus maximal for motor fibres. This indicates that not less than 25% of the amplitude of the ascending action potential in Fig. 2 A was

5 440 G. D. DAWSON due to nerve fibres more easily stimulated than the motor fibres. Since the cathode was placed to give the best possible stimulation of the motor fibres to the abductor minimi digiti, from which the muscle action potentials were being recorded, it is unlikely that this result, which is typical of those obtained in these experiments, was due to all the motor fibres being relatively inaccessible to the stimulus. In addition, when the stimulus was just threshold for the abductor minimi digiti none of the other muscles in the hand supplied by the ulnar nerve was found to be twitching. This suggests that the low threshold fibres excited by the stimuli at and below motor threshold, and apparently making up about a quarter of the ascending volley in Fig. 2 A, and all the ascending volley in Fig. 2 D, are afferent in function. The records in Fig. 2 E to H, made under the same conditions as those in Fig. 2 A to D, but from the child with the severe sensory neuronopathy, support this suggestion. The clinical picture presented by this child is to be described fully elsewhere. He was apparently insensitive to all forms of skin or deep stimuli, whether of a type which might be expected to produce pain or otherwise. No reflexes could be elicited in the trunk or limbs and there was evidence, particularly in the fingers and hands, of multiple injuries with delayed healing and malformation. Biopsy of a minor peripheral cutaneous nerve in the foot showed a complete absence of large myelinated nerve fibres. Cutaneous stimulation produced no flare, and it was concluded that the disease process affected the posterior root ganglia, leaving the motor nerve fibres intact. This was supported by the fact that there was no muscular wasting and power was good although the child, at the age of 3- years when these records were made, was unable to walk. Also, the general level of excitability of the motor nerve fibres and the range between threshold and maximal stimuli was normal. In this case the ascending nerve volley which could be recorded was probably made up entirely of antidromic impulses in motor fibres. When an electrical stimulus was applied to the ulnar nerve and its strength was decreased from maximal for motor fibres (Fig. 2 E) to threshold (Fig. 2 H), the size of the nerve action potential fell at the same rate as the muscle action potential. At this last strength nothing was recorded from over the ulnar nerve and there was no evidence of activity in any nerve fibres with a threshold lower than that of the motor fibres. Results such as these, obtained when the cathode was placed to stimulate motor fibres preferentially, indicate the presence in the normal ulnar nerve of afferent fibres with a low electrical threshold. They do not, however, show whether these low threshold afferents come entirely from small muscles in the hand, from the skin and fingers, or from both. When such experiments were repeated, but using the median nerve instead of the ulnar, then it was found that with the stimulus at motor threshold the ascending volley in the median nerve was relatively about twice as big as it was in the ulnar nerve, also with the stimulus at motor threshold. Since the median nerve supplies much more

6 NERVE CONDUCTION VELOCITY IN MAN 441 skin and fewer muscles in the hand than the ulnar nerve, it seems more likely that the larger relative size of the afferent volley in the median is due to stimulation of fibres from the skin and fingers rather than afferent fibres from muscles. A direct comparison of the thresholds of excitability of afferent fibres solely from a finger, and motor fibres to the hand muscles, is difficult. This is so because of the great differences in size of the digital nerves and the nerve trunks at the wrist and in the amounts of tissue over them. Indirect evidence may, however, be obtained in the following ways. Central effects of the centripetal volley. With a stimulus applied to a nerve at the wrist it is possible, in addition to recording the peripheral muscle action potentials and the centripetal volley in the nerve, to record the cerebral potentials evoked by it. From such records the size of the cerebral response can be related to the size of the centripetal nerve volley, and the excitability of the fibres giving rise to the cerebral response can be compared with that of the motor fibres in the nerve. The results of an experiment of this type are shown in Fig. 3. Each record in the top line shows the average ofthe cerebral responses to 220 stimuli of one strength, the middle line shows superimposed records of the centripetal nerve action potentials produced by the first 55 of the same 220 stimuli and the bottom line shows the muscle action potentials due to the first 55 stimuli. The stimulus used in Fig. 3 A was maximal for motor fibres, and it was reduced through B and C to just above threshold for motor fibres in D. In E and F the stimulus was reduced further, and in G is shown a control record of the same number of sweeps with no stimulus. The scalp electrodes were connected in such a way that when the back one of the pair, which was over the gensorimotor area, became positive with respect to the front one an upward deflexion resulted in the record. The first significant deflexion in the cerebral records in the top trace (Fig. 3) is the downward one, starting 20 msec after the stimulus. This deflexion varies in size with the position of the electrode over the sensorimotor area and is more commonly seen when stimulating the median nerve; it is frequently absent when the stimulus is applied to the ulnar nerve. Following this an upward deflexion begins at msec and reaches its peak at msec. During this deflexion the sensorimotor area opposite to the arm being stimulated becomes positive with respect to the rest of the scalp. For any one set of stimulating and recording conditions the form of these two initial phases of the response is highly repeatable; the later phases vary more with changes of attention or wakefulness of the subject. The short latency of these initial phases of the response and the constancy of their form suggest that they represent events occurring at, or shortly after, the arrival of the sensory volley at the cortex. It is the relation of the size of these first phases of the cortical evoked potential to the strength of the stimulus and the size of the afferent volley which will be considered. In the records in Fig. 3 the size of the initial phases of the cerebral response fell off more slowly with reduction of

7 442 G. D. DAWSON stimulus strength than did the size of the muscle action potential. In Fig. 3 D, where the stimulus was just above threshold for motor fibres, the muscle action potentials from the thenar eminence were approximately 5 % of their maximal size, shown in Fig. 3 A. In contrast to this the cerebral response in Fig. 3D was 85% of its size in Fig. 3 A due to a stimulus which was maximal for motor fibres. In Fig. 3F, where the stimuli were just big enough to produce easily detectable action potentials in the median nerve at the elbow and were well below motor threshold, they still evoked cerebral potentials the initial phases of which were over 35% of those in Fig. 3 A. Stimuli applied to the ulnar nerve gave rise to the same sequence of changes, though the difference in the rate of fall of the cerebral evoked potential and the muscle action potential is less. These records, and others made in comparable conditions, show first that the cerebral responses may be evoked by stimulation of fibres with a threshold 3_ 4.. N~.%.25F.LV 20UV -~~ I mv E2 Fig. 3. The top record shows the average of 220 cerebral responses to stimuli to the median nerve at electrodes 5-6. Stimulus strength was decreased from maximal for motor fibres in A to zero in G. The middle record shows superimposed the centripetal nerve action potentials and the bottom record the thenar muscle action potentials due to the first fifty-five of the 220 stimuli. The top time scale for the cerebral responses shows intervals of 5 and 20 msec. The bottom time scale, for the nerve and muscle action potentials, shows 1 and 5 msec. The time of the stimulus is shown by the spike at the start of the notch in the top time scale.

8 NERVE CONDUCTION VELOCITY IN MAN 443 lower than that of the motor fibres. Secondly, they do not show evidence for the existence of any considerable number of afferent fibres with a threshold lower than that of those giving rise to the cerebral responses. Had there been such a group of fibres it might have been expected that the size of the cerebral responses would have fallen more rapidly with decreasing stimulus strength than the size of the centripetal nerve volley. In fact, in the records in Fig. 3, after the centripetal nerve volley had lost the component due to the antidromic impulses in motor fibres, the nerve volley and the cerebral response fell at about the same rate and disappeared at the same shock strength. Interaction of volleys from finger and wrist. A stimulus applied to electrodes 5 and 6 (Fig. 1) will produce in the sensory nerve fibres in the nerve trunk both an ascending orthodromic volley and a descending antidromic volley which will travel into the finger. A second stimulus applied to electrodes 1 and 2 on the finger can be timed to produce an ascending volley that will meet the descending volley from the wrist. If the stimulus at the wrist is strong enough, the descending volley it produces will occupy all the afferent fibres in which an ascending volley is also produced by the stimulus to the finger. The ascending volley due to the finger stimulus will therefore be blocked and no action potential due to it will be recorded at the elbow. If the muscle action potentials also are recorded from electrodes 3 and 4, then the number of motor fibres which are stimulated by a shock at the wrist just strong enough to block the volley from the finger may be estimated. The excitability at the wrist level of afferent fibres from the finger may therefore be compared with the excitability of the motor fibres at the same level in the nerve. The results from an experiment of this type are shown in Fig. 4. The upper trace records the action potentials from electrodes 3 and 4, over the thenar muscles. The lower trace records the action potentials from electrodes 7 and 8 over the median nerve at the elbow. When a stimulus S1 was applied to electrodes 5 and 6 at the wrist an ascending nerve action potential A1 (Fig. 4A) was recorded. Since the stimulus was below threshold for motor fibres no muscle action potential appeared in the upper trace from electrodes 3 and 4. When a stimulus S2 was applied 2 msec after the start of the lower trace to electrodes 1 and 2 on the finger, a smaller action potential A2 (Fig. 4 B) with a longer latency was recorded. So long as the wrist stimulus S, was kept below a critical strength the two stimuli S, and S2 applied together produced both action potentials A1 and A2 (Fig. 4C). As S, was increased above this critical size the action potential A1 continued to increase in size, but A2 became smaller and finally disappeared, even though the finger stimulus was kept constant. This is shown in Fig. 4D and E; in D the strength of the stimulus Sl, although it was sufficient to cause a considerable reduction in the amplitude of the action potential A2, was only just above threshold for the motor fibres. In Fig. 4 E, where S, was sufficiently strong to reduce the volley A2 from the finger below the limit of detectability the muscle

9 444 G. D. DAWSON action potential it produced in the upper trace was still only just over onethird of the size of that produced by a maximal stimulus and shown in Fig. 4F. It seems, therefore, that the more excitable fibres from the fingers have an electrical threshold at the wrist as low as that of the most excitable motor fibres at the same level in the nerve and lower than that of most of the motor fibres. A B C S1 S2 Sl S2 5smv 3-4 _ 7-8 1(.1 A~~~~~ and A2 due tosiuu At lcrds12 2 cosan thoghu, /inrasn fo Fig. 4. The top record shows the thenar muscle action potentials due to the stimulus S, at electrodes 5-6. The bottom record shows the centripetal nerve action potentials, Al due to stimulus S, and A2 due to stimulus S2 to electrodes 1-2. S2 constant throughout, S, increasing from threshold for motor fibres in C to maximal in F. Top time scale for the muscle action potentials shows 1, 5 and 20 msec and bottom time scale for the nerve action potentials shows 1 and 5 msec. Relative conduction velocity in sensory and motor fibres A major source of error in the measurement of conduction rate was demonstrated by Helmholtz & Baxt (1868, 1870) when they showed that in arm nerves in man the velocity of conduction in motor nerve fibres varied widely with temperature. That this variation may be serious under normal laboratory conditions can be seen from the fact that in uncovered limbs of healthy subjects at rest motor fibre conduction velocities from 70 down to 40 m/sec have been recorded. On account of the gradient of temperature along the axis of the

10 NERVE CONDUCTION VELOCITY IN MAN 445 limb it is difficult to apply measurements of temperature made at only one or two points near the nerve to correct these results. Therefore, to avoid the inconvenience of recording nerve action potentials with the limb in a constant temperature bath, no attempt has been made to measure absolute values of sensory or motor fibre conduction velocity. Instead, measurements have been made of the conduction times between the same pairs of electrodes along sensory and motor fibres in the same stretch of nerve; this allows the relative conduction velocity of the two sets of fibres to be estimated. The interval of time between the two sets of measurements was made sufficiently short for any significant change of temperature to be unlikely, or alternatively the sensory conduction time was measured both before and after the measurement of motor conduction time. The chief remaining source of error lies in the uncertainty about the position along the nerve trunk where excitation actually takes place when a stimulus is applied to the overlying skin. If a stimulus at the wrist or elbow is increased slowly from below threshold for motor fibres, and a record is made at high amplification from over the muscles supplied by the nerve, it is sometimes found that the first units which fire in the muscle are not those with the shortest latency. As the strength of the stimulus is increased the latency of the muscle action potential may diminish in small discrete steps as new units are added. The muscle action potential then begins to increase in size with little or no change in latency until it is between half maximal and maximal. At this strength a continuous decrease in latency may begin and go on with stronger stimuli after the muscle action potential has ceased to increase in size. The stimulus chosen for measuring the motor fibre conduction time was strong enough to stimulate the units with the shortest latency, but not so strong as to reduce the latency by spread of the stimulus from the cathode. The latency was measured from the time of stimulation to the first detectable rise of the muscle action potential, and the motor conduction time was taken as the difference between the latencies in successive records with the stimulus at elbow and wrist. Records of nerve action potentials made through the skin are usually triphasic: but the size of the initial phase, which has been recorded as a downward deflexion and in which the electrode nearer to the approaching action potential becomes positive with respect to the more remote one, is very variable and it may be almost absent. In arranging the electrodes at the start of an experiment small adjustments were made to their positions until the nerve action potentials had as nearly as possible the same shape whether recorded at the wrist or elbow. When closely similar shapes could not be obtained the records were not used for purposes of measurements. Afferent conduction times were taken as the difference in time between corresponding points on the action potentials recorded at the wrist and elbow. Two measurements were made, the first to the bottom of the initial positive, downward deflexion or,

11 446 G. D. DAWSON where that was too small, to the first detectable rise above the base-line, and the second to the negative peak of the action potential. The measurements to the first trough and to the first peak gave different values for the conduction times as the action potentials from the elbow were usually slightly different in duration from those picked up at the wrist, probably owing to dispersion over the longer conduction distance. An example of the type of record from which measurements were made is shown in Fig. 5. In A are shown a pair of records A ~~B 'C ' VI~~~~~~~2V 7-8\ ^ Stim. 5-6 Stim. 7-8 Stim Fig. 5. The records in A show the conduction time in the median nerve between electrodes 5-6 and 7-8 of an afferent volley due to stimulation at electrodes 1-2 on the index finger. B shows the latency of the muscle action potential at electrodes 3-4 after stimulating at electrodes 5-6 and C after stimulating at electrodes 7-8. The time scales show 1 and 5 msec. made simultaneously from electrodes over the median nerve at the wrist, upper trace, and at the elbow, lower trace, with the stimulus applied to electrodes 1 and 2 on the index finger. In B and C are successive sets of records of the muscle action potentials from over the thenar muscles with the stimulus at electrodes 5 and 6 at the wrist in B and at electrodes 7 and 8 at the elbow in C. Fourteen subjects were used, and twenty experiments were carried out, ten on the ulnar nerve and ten on the median. No subject was used twice in either group, and the results obtained are shown in Table 1, in which the measurements are given to the nearest 0'05 msec. All the twenty experiments taken together gave a mean motor conduction time of 5-12 msec, and a mean sensory conduction time of 4-73 msec measured from the starts of the sensory action potentials and of 4-88 msec measured from the peaks. The means of the differences between motor and sensory conduction times were 0393 and 0243 msec. Analysis of the results in Table 1 shows that if, in fact, the motor and sensory conduction velocities were the same, then differences of the sizes found

12 NERVE CONDUCTION VELOCITY IN MAN 447 might be expected to occur by chance less than once in every 1000 samples of this size for the measurements to the starts, or less than once in every 100 samples for the measurements to the peaks. The figures for the ten median nerve experiments taken alone also give differences which might be expected to occur by chance less than once in every 100 samples of that size, but the figures from the ten ulnar nerve experiments taken alone are less significant. TABLE 1. Conduction times between wrist and elbow in motor and sensory fibres in median and ulnar nerves. Fourteen subjects, ten median and ten ulnar experiments. Tm =motor time, T. =sensory time from starts of action potentials, and T,, =sensory time from peaks (msec.) Subject Tm TM, T8p T"TM Tm TeS I.S. 4* * P.N *90 4* R.G. 6*15 5*85 5* L *70 4*75 -O H. 4'45 4* T.W. 4* M.H. 4* T.S J.C J.D * Mean 4* (median t = 3-43 t = 3-73 alone) P <0.01 P <001 R.G. 5*20 5* J.P * M.H G.D * X35 P.N. 4* I.S * T.W * *85 T.S *25 5* J.C V.E * Mean (ulnar t=3-19 t=2-15 alone) P=0012 P=0-06 Mean (combined) t=4-17 t=2-99 P <0o001 P <0o01 Here the difference between motor conduction time and the sensory conduction time measured from the starts of the action potentials might be expected by chance once in every eighty-three samples and the figures derived from the peaks give a difference which might occur by chance once in every sixteen or seventeen samples. A wider separation between the electrode positions was found to be an advantage when recording from the ulnar nerve. This may account entirely for the longer conduction times obtained from the ulnar nerve experiments.

13 448 G. D. DA WSON DISCUSSION One factor which must be considered in interpreting the observations on excitability in the nerve trunks at the wrist is fibre branching. If a majority of the motor fibres to the small muscles of the hand had already branched at the level of the stimulating electrode at the wrist, and had possibly become smaller as a result, this might account for their relatively higher threshold. Eccles & Sherrington (1930) found that branching of motor fibres was greatest at short distances from a muscle and was uncommon beyond 1 cmfromit. Inthe experiments described in this paper the stimulating point was not less than 5 cm from the motor point of the muscle being examined. The evidence of Eccles and Sherrington may not be applicable to man, but other experiments with stimuli applied higher up the nerve and farther from the muscles also suggest that branching is not important. Stimuli applied at elbow level, where branching is probably confined to a small minority of the fibres, still show a lower threshold for the afferent fibres from the fingers than for the motor fibres to the small muscles of the hand. At this level in the nerve trunk, however, the fibres with the lowest threshold are often those from parts of the limb proximal to the hand, which were incorporated in the nerve nearer to the elbow and are therefore probably more superficial in the trunk at the point of stimulation. Cerebral potentials such as those recorded in the experiment illustrated in Fig. 3 might have been evoked either directly by the volley in afferent fibres ascending from the point of stimulation, or they might have been secondary to movements in the periphery resulting from a motor volley descending the nerve. In the second case they would be of little value for estimating the threshold of any sensory fibres being stimulated. Some of the later parts of the cerebral response are probably produced indirectly through the periphery (Bates, 1951), but it has been shown that the latency of the initial phases of the responses in the present experiments was progressively reduced when the stimulating point was moved centrally along the nerve. Also, the stimuli continued to evoke cerebral potentials when they were considerably below the minimum strength at which any movement could be observed or action potentials be detected in the hand muscles. The experiments, therefore, show that the cerebral responses were produced directly by stimulation of afferent fibres, and the greater part of them by fibres which have a lower threshold than that of the motor fibres to the small muscles of the hand. That these low threshold afferent fibres, which are stimulated at the wrist, certainly come in part from the fingers is shown by the experiments on the interaction of volleys from finger and wrist. They might also come in part from muscles; no satisfactory experiment has yet excluded this possibility in man, though the observations of Lloyd & McIntyre (1950) and of Mountcastle, Covian & Harrison (1952)

14 NERVE CONDUCTION VELOCITY IN MAN 449 in the cat show that if the largest muscle afferent fibres have any access at all to the cerebral cortex, it is much less direct than that of cutaneous or other afferents. No cerebral responses were detected with stimuli just too small to produce paraesthesiae in the fingers, but stimuli which could just be felt produced detectable cerebral responses in some subjects. It seems to be clear, therefore, that if any muscle afferents do contribute to the cerebral responses they have not got a lower threshold than that of the afferent fibres from the fingers. In addition, when the stimulus was reduced to near the threshold for sensation and for the cerebral response, nerve action potentials were barely detectable at the elbow. This indicates that if there are any sensory afferent fibres from muscles in the median or ulnar nerves with a threshold lower than that of the afferent fibres from the fingers, they are few in number. The motor conduction time derived from the first rise of the muscle action potentials is probably that of the fastest conducting motor fibres. If this is so, a comparison of the motor conduction time with the sensory conduction time derived from the starts of the sensory action potentials should give the relative conduction velocity of the fastest fibres of both types. From such a comparison it is clear that the fastest conducting sensory fibres from the fingers have a higher conduction velocity than that of the fastest conducting motor fibres to the small muscles of the hand, a difference consistent with the lower threshold of the sensory fibres to electrical stimulation. The measurements of sensory conduction time made between the peaks of the action potentials gave on the average a longer time than the measurements between the starts. This was evidently due to broadening of the action potentials recorded near the elbow caused by dispersion of impulses in fibres with different conduction velocities over the longer conduction distance. The figures of peak to peak times therefore probably represent the conduction time, not of the fastest fibres alone, as do the figures taken from the starts, but also that of a considerable group of slower fibres. In the median nerve experiments it is therefore clear that some of the more slowly conducting sensory fibres, as well as the fastest ones, have a shorter conduction time and a higher conduction velocity than that of the fastest motor fibres. In the experiments on the ulnar nerve also the fastest sensory fibres showed a conduction time significantly shorter than that of the fastest motor fibres in the nerve, but the difference was less than in the case of the median nerve. The slower sensory fibres in the ulnar nerve, represented by the measurements between peaks, showed a conduction time less than that of the fastest motor fibres, but not by a significant amount. In general, the sensory action potentials recorded from the ulnar nerve when the little finger was stimulated were smaller and less easy to measure than those recorded from the median nerve when the index finger was stimulated. How far the difference between the median and ulnar figures was due to a real difference between the two nerves, and how far possibly to less accuracy of the 29 PHYSIO. CXXXI

15 450 G. D. DA WSON ulnar measurements, cannot be decided on the present evidence. Although on the average the sensory conduction velocities were faster than the motor conduction velocities, some individual variations occurred in the fourteen subjects examined. A few showed afferent conduction velocities slightly lower and a few considerably higher than their motor conduction velocities. If the evidence of Sunderland et al. (1949) about the variations between individuals, and between opposite sides in the same individual, in the number of large fibres in cutaneous nerves can be applied also to mixed nerves, as seems likely, then the individual variations in relative conduction velocities found in the present experiments are no greater than might be expected. If lower threshold to electrical stimulation and higher conduction velocity can be accepted in any one species as being associated with larger fibre size, then conditions such as pressure and ischaemia, which affect larger fibres first, may be expected to affect sensory fibres from the fingers before they affect motor fibres to the hand. In this connexion it is interesting to note that Gilliatt & Wilson (1954) found that disturbance of sensation produced by a pneumatic cuff around the upper arm progressed more rapidly in the distribution of the median nerve than in that of the ulnar nerve. This may suggest that the larger differences between sensory and motor conduction velocities in median than in ulnar nerve may reflect a larger size for the finger afferents in the median nerve, rather than just greater difficulty in measu'ring the ulnar action potentials. It may be concluded that interpretation of the central effects in man of peripheral electrical stimulation must be based on the fact that sensory afferent fibres from the fingers have a lower threshold and higher conduction velocity than the motor fibres to the small muscles of the hand. They also have at least as low a threshold as that of the afferent fibres from these muscles. It is therefore probable that the finger afferents are larger than the motor fibres and as large as the muscle afferents; no opportunity has arisen of checking this directly under experimental conditions comparable with those used by Eccles & Sherrington (1930) for their measurements in the cat. SUMMARY 1. The relative electrical excitability at the wrist, and the relative conduction velocity between wrist and elbow, of sensory and motor fibres in the median and ulnar nerves have been examined in man. 2. It is shown that the more excitable sensory afferent fibres from the fingers have a lower threshold to electrical stimulation than that of the motor fibres to the small muscles of the hand. 3. The cerebral potentials evoked by stimulating median or ulnar nerve trunks are produced largely by afferent fibres with a threshold lower than that of motor fibres.

16 NERVE CONDUCTION VELOCITY IN MAN No group of fibres has been found in median or ulnar nerves with a threshold lower than that of those producing the cerebral responses. 5. The sensory afferent nerve fibres from the fingers are shown to have a higher maximum conduction velocity than that of the motor fibres to the small muscles of the hand. 6. It is concluded that, if greater fibre size is associated with lower electrical threshold and greater conduction velocity, the sensory afferent fibres from the fingers in man have a larger maximum size than the motor fibres to the small muscles of the hand and are not smaller than the- afferent fibres from these muscles. I wish to thank Dr E. A. Carmichael for his encouragement throughout this work. REFERENCES BATES, J. A. V. (1951). Electrical activity of the cortex accompanying voluntary movement. J. Phy8iol. 113, DAWSON, G. D. (1953). Autocorrelation and automatic integration. Symposia of IlIrd International EEG Congress. Electroenceph. clin. Neurophy8iol. (Suppl.), 4, DAWSON, G. D. (1954). A summation technique for the detection of small evoked potentials. Electroenceph. clin. Neurophysiol. 6, DAWSON, G. D. & SCOTT, J. W. (1949). The recording of nerve action potentials through skin in man. J. Neurol. 12, ECCLES, J. C. & SHERRINGTON, C. S. (1930). Numbers and contraction values of individual motor units examined in some muscles of the limb. Proc. Roy. Soc. B, 106, GILLIATT, R. W. & WILSON, T. G. (1954). Ischaemic sensory loss in patients with peripheral nerve lesions. J. Neurol. 17, HELMHOLTZ, H. & BAIT, N. (1868). Versuche uber die Fortpflanzungsgeschwindigkeit der Reizung in den motorischen Nerven des Menschen. Mber. k. preuss. Akad. Wiss. (Berlin 1868), pp HELMHOLTZ, H. & BAIT, N. (1870). Neue Versuche uber die Fortpflanzungsgeschwindigkeit der Reizung in den motorischen Nerven des Menschen. Ber. d. Berl. Akad. (1870), pp KUGELBERG, E. (1944). Accommodation in human nerves. Acta physiol. scand. 8 Suppl. 24, 48. LLOYD, D. P. C. & MCINTYRE, A. (1950). Dorsal column conduction of Group I muscle afferent impulses and their relay through Clarke's column. J. Neurophysiol. 13, MOUNTCASTLE, V. B., COVIAN, M. R. & HARRISON, C. R. (1952). The central representation of some forms of deep sensibility. Proc. Ass. Res. nerv. ment. Dis. 30, SUNDERLAND, S. (1945). The intraneural topography of the radial, median and ulnar nerves. Brain, 68, SUNDERLAND, S., LAVARACK, J. 0. & RAY, L. J. (1949). The calibre of nerve fibres in human cutaneous nerves. J. comp. Neurol. 91,