Further observations on the facilitation of muscle responses to cortical stimulation by voluntary contraction

Electroencephalography and clinical Neurophysiology, 81 (1991 ) 397- 402

397

© 1991 Elsevier Scientific Publishers Ireland, Ltd. 0924-980X/91/$03.50 ADONIS 0924980X9100100Y

ELMOCO 90595

Further observations on the facilitation of muscle responses to cortical stimulation by voluntary contraction P.D. Thompson, B.L. Day, J.C. Rothwell, D. Dressier *, A. Maertens de Noordhout ** and C.D. Marsden MRC Human Movement and Balance Unit, Institute of Neurology and The National Hospital for Nervous Diseases, Queen Square, London WC1N 3BG (U.K.) (Accepted for publication: 15 January 1991)

Summary The effect of voluntary contraction on the discharge of single motor units following electrical and magnetic stimulation of the motor cortex was examined using the post-stimulus time histogram (PSTH) technique. The latencies of responses in single motor units of the first dorsal interosseous muscle to cortical stimulation were 2-4 msec shorter when the muscle was contracting than when at rest in 9 of 10 units studied. These latency differences are comparable with those recorded by surface electromyography for compound muscle action potentials following cortical stimulation in relaxed and active muscles. The new findings are that the intensity of cortical stimulation required to discharge a resting motor unit to produce a single PSTH peak produced multiple PSTH peaks when the same unit was contracting. The timing of the PSTH peak of relaxed motor unit discharge corresponded to one of the later PSTH peaks (usually the second) when the motor unit was voluntarily activated. These findings are in keeping with our previous suggestions that the longer latency of responses in relaxed muscles is due to the time taken for temporal summation of multiple descending corticospinal volleys at the corticomotoneurone synapse. Facilitation produced by voluntary contraction occurs at least in part at the level of the spinal cord by lowering motoneurone threshold to enable discharge on the initial descending volley. The higher threshold of relaxed muscles is related to the higher intensities of stimulation needed to recruit multiple descending volleys and discharge resting motoneurones. Key words: Cortical stimulation; Motor unit discharge; Voluntary facilitation

Electrical or magnetic stimulation of the motor cortex in man produces electromyographic (EMG) responses in limb muscles at latencies compatible with conduction in the largest diameter fast conducting monosynaptic component of the corticospinal tract (Marsden et al. 1982). Two fundamental observations on the behaviour of these surface recorded E M G responses are: (i) The latency to onset of the potentials is shorter when the target muscle is voluntarily made to contract before each cortical stimulus, than when the same muscle is relaxed (Rothwell et al. 1987). This difference may vary from 2 to 6 msec for anodal electrical stimulation and approximately 1-3 msec for magnetic stimulation (Day et al. 1986, 1987a; Hess et al. 1987; Rothwell et al. 1987). (ii) The threshold for

* Present address: Department of Neurology, University of Erlangen-Nuremberg, Erlangen-Nuremberg, F.R.G. * * Present address: University Department of Neurology, H6pital de la Citadelle, Liege, Belgium.

Correspondence to: Dr. P.D. Thompson, MRC Human Movement and Balance Unit, Institute of Neurology, Queen Square, London WC1N 3BG (U.K.).

eliciting E M G responses is lower when the muscle is contracting (Marsden et al. 1982; Hess et al. 1987; Rothwell et al. 1987). These findings also have been shown to apply to single units (Calancie et al. 1987). Previously, we have suggested that a single electric cortical stimulus can give rise to multiple descending volleys (D and I waves) in the pyramidal tract of man (Day et al. 1987a,b), as in the monkey (Patton and Amassian 1954). The longer latency of responses in relaxed muscles can be explained by the extra time taken for temporal summation of successive EPSPs released by D and I waves to bring resting spinal motoneurones to discharge. The threshold for discharge of relaxed muscles is higher because higher intensities of stimulation are required in order for the brain stimulus to recruit I waves. Only when I waves have been recruited will there be sufficient multiple descending volleys to release a series of EPSPs at the corticomotoneurone synapses and temporally summate to discharge the motoneurones. The present paper presents new data in support of this hypothesis from a study of single motor unit behaviour, based upon a comparison of PSTHs constructed from the discharge patterns of relaxed and

398

active motor units. Previously, we have examined the PSTHs of single motor units after cortical stimulation when the units were tonically active (Day et al. 1987a,b, 1989). It will be shown that in relaxed muscles, motor units usually cannot be discharged unless the intensity of the cortical stimulus is high enough to produce more than one phase of excitatory input to the spinal motoneurones, and that the timing of the relaxed discharge corresponds in most cases to the timing of the second (or later) excitatory input produced by the stimulus.

Methods

Experiments were conducted on 6 subjects aged between 28 and 38 years. All gave informed consent for the procedures and the studies had been approved by the local Ethical Committee. The brain was stimulated electrically with a prototype of the commercially available Digitimer D180 stimulator built by Mr. H.B. Morton (Merton and Morton 1980). This stimulator gave a peak output of 750 V with a peak current of approximately 1 A and a time constant of 50 ~sec. Stimuli were delivered to the scalp by two Ag-AgC1 electrodes, 9 mm in diameter, placed at the vertex (cathode) and 7 cm laterally on a line joining the vertex and the external auditory meatus (anode) and fixed to the scalp with collodion glue. For magnetic stimulation a stimulator built by Dr. A. Barker, Mr. R. Jalinous and Dr. I.L. Freeston of the University of Sheffield (Barker et al. 1985) was used. A 9 cm diameter coil was centred on the vertex, with the inducing current flowing in the coil so as to preferentially activate muscles of the right hand. Stimuli were delivered randomly every 4 - 1 0 sec. The intensities used ranged from 40 to 100% of the maximal output of each device and varied according to the subject under study. Single motor units were recorded in the right first dorsal interosseous muscle (FDI) with concentric needle electrodes (Dantec type 13L58). The timing of motor unit discharge after each cortical stimulus was analysed by constructing a post-stimulus time histogram (PSTH) the details of which have been described in previous publications (Day et al. 1987a,b). Signals were pre-amplified (Devices type 3160) and amplified (Devices type 3120) with filters set 3 dB down at 80 Hz and 2.5 kHz. Briefly, the subject was first instructed to recruit a single motor unit with the assistance of audio and visual feedback, then maintain the firing rate of that unit at approximately 10 Hz. Random cortical stimuli were given when the unit discharge was stable: A pulse height window discriminator designed by Mr. H.C. Bertoya detected the timing of motor unit discharge, by producing a 0.25 msec pulse when the height of the motor unit potential fell

P.D. THOMPSON ET AL.

within an adjustable voltage window. The marker pulse was produced after a fixed delay of 2 msec, which has been compensated for in the figures and text. The response to each cortical stimulus was viewed on a digital storage oscilloscope to ensure that additional units were not recruited or that spurious triggering had not occurred. Each single motor unit from e a c h trial was carefully inspected both on-line and off-line and compared with a template motor unit which was held on the oscilloscope screen throughout the compilation of the PSTH. Trials in which additional motor units obscured the template unit or spurious triggering occurred were excluded from the PSTH. Only when one stable and clearly discernible unit was activated by the cortical stimulus either at rest or during voluntary contraction did we proceed with the experiment. PSTHs were collected with bin widths of 0.25 msec and contained responses from 100 trials in 27 of the 29 PSTHs collected in this study (the remaining 2 were constructed from 50 trials). After successfully maintaining the motor unit discharge for the construction of a PSTH at a stimulus intensity sufficient to produce one or more peaks in the PSTH, subjects then relaxed and without any change in the position of the needle electrode the procedure was repeated. All information was recorded on magnetic tape for latex analysis using a Racal 7DSFM recorder with a frequency response from DC to 2.5 kHz.

Results

"Fen units in the first dorsal interosseous muscle of 4 individuals were successfully studied. Two criteria were used for inclusion in the study: (i) the unit could be recruited by a cortical shock in the relaxed state: (ii) the same unit could be studied using the same or similar stimulus intensities in both relaxed and active conditions. Several units failed the latter criterion because stimulation at intensities sufficient to produce a discharge in relaxed muscles often recruited several motor units in the active state, making it impossible to maintain single unit analysis. In 2 other individuals we were unable to isolate a unit which fulfilled these criteria: the stimulus intensity at which a motor umt could be recruited in the relaxed muscle was too high for a comparable intensity of stimulation to be used to study a single unit when contracting. In two subjects it was possible to study the responses of one unit to magnetic and electrical stimulation when relaxed and active. A typical example of the behaviour of a single motor unit m the first dorsal interosseous muscle following electrical cortical stimulation when active and relaxed is illustrated in Fig. 1 and is listed in Table 1 (unit 5). During tonic voluntary activation of the motor unit.

SINGLE MOTOR UNITS AND CORTICAL STIMULI

Relaxed

399

intensity of 100% the unit fired with a peak in the PSTH at a latency of about 30 msec. This timing corresponded to the latency of the second peak during voluntary contraction. A different unit from a different subject is illustrated in Figs. 2 and 3. In Fig. 2 (unit 6 in Table I), magnetic stimulation at 32-34% was used in both the relaxed and active records. When active, stimulation produced 3 peaks of increased firing probability at 28.5, 31.75 and 33.75 msec. When the subject was at rest, the unit discharged only at latencies corre-

100%

. . . . . . . . . . .

i. . . . .

MAGNETIC STIMULATION

Active

(32-34%) Active 7 0 %

• . I,I ,I !11 I . . . . . . . . .

, ,,,,.,I 1,11_,1 I

I

[ Active 100% , 10

,

,

L I,II1,11 . . . . . . 20

• 30

Relaxed (32-34%)

5 counts II . . . . 40ms

Fig. 1. Post-stimulus time histogram (PSTH) of a single motor unit in the first dorsal interosseous muscle (FDD to anodal cortical stimulation (unit 5 in Table I). Stimuli were delivered 10 msec before the start of the sweep. PSTHs were constructed from 100 responses (0.25 msec bin width) to stimuli when the subject was maintaining a constant discharge rate of the unit at stimulus intensities of 100% maximum (lower trace) and 70% maximum (middle trace), and from 50 stimuli at 100% maximum when the subject was resting (upper trace). When contracting, a stimulus intensity of 70% produced a single peak (at 25.5 msec) and as the intensity was increased a second peak emerged at an interval of some 4.5 msec later. At rest, a single peak only was evident at a latency of 30 msec, corresponding to the second peak at 100% when contracting. Note that the unit discharged on only one-third of the trials when relaxed in contrast to two-thirds of stimuli when voluntarily activated.

electrical stimulation at 70% produced a single peak of increased firing probability at 25.5 msec after the stimulus. When the intensity of stimulation was increased to 100% a second peak appeared 4.5 msec later. When the subject was totally relaxed, stimulation at 70% failed to evoke any discharge (not illustrated). At an

. . . . . . . . . . . 10

20

II .... 30

40ms

I 25

J 40ms

Fig. 2. Left-hand panel: post-stimulus time histogram (PSTH) of single motor unit in the right FDI to magnetic cortical stimulation (at an intensity 32-34% of maximum; unit 6 in Table I). Responses to 100 stimuli are shown (bin width 0.25 msec), when the subject was maintaining constant discharge of the unit (upper trace) and when at rest (lower trace). Stimuli were given 10 msec before the start of the histogram. The first PSTH peak (latency 28.5 msec) occurred earlier when the muscle was contracting than when relaxed and more peaks were evident when contracting. The latency of the first peak when relaxed (31.25 msec) corresponded to the second peak in the contracting PSTH (31.75 msec). Right-hand panel: raster plots of the single motor unit discharge after cortical stimulation (traces begin 25 msec after the stimulus) from which the above PSTH was constructed. The histogram was constructed from the larger unit shown on these traces. The calibration bar between the two histograms represents 5 counts.

400

P.D. - I ' I t O M P S O N E ' I AI,.

TABLE 1

Latency (msec) of PS'FH p e a k s for relaxed and contracting single motor units (in first dorsal i n t e r o s s c o u s muscle) to electrical a n d / o r m a g n e t i c brain stimulation in 4 subjects. A total of 10 units were successfully studied. .

Motor unit 1~ 2 "

3 ~' 4 ' 5 ~

6 ~' 7 ~' •

t; d 11) ~

Magnetic stimulation (intensity) Active Relaxed Active Active Relaxed Relaxed Active Relaxed Active Relaxed Active Active Relaxed Relaxed Active Relaxed Active Relaxed Active Active Relaxed Relaxed Active Relaxed Active Relaxed

311 2t~

28.6

25.8

28.5

25.7 25.5

NE NE 31.5 31.5 31.5 31.5 27.6 27.S 27 27.4 NE NE NE NE 31.75 31.25 NE NE 27.7

23

28.2 28.6

24.2 24.4

22.2 24.4

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(55~i) 170<~ ) (55~:;~) (70%) (91)G) 191)~ ) 181)<;~ ) (809>)

26.8 26.6 NE NE NE NE N[ NlN NI

No response 33.75

132-34<~ ) (32 34%)

26.75

3( ) 31,5 31,25

22 28.7 28.7 28.2 25

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(81)C~)

25.5

28.2

22.4

.

Electrical stimulatkm (intensity)

(46(4) (49'74) (49%) (51)'~) (38-40%) (38-409;) 138-431~,~ ) (43f~)

33

(7O%) ( I 0()'.Z~) ( 71)~;. ) (I00~;) (556;) (559;>) (55<~) (55(;)

NE NE NE NE 21.4

24.~ NE NE

138<:; ) (40G:)

,,t,,~',d refer to the 4 different subjects in w h o m the m o t o r units were studied. N E = not examined.

sponding to the second and third 'active' peaks. The same unit was examined with anodal stimulation (Fig. 3). When contracting the first PSTH peak had a latency of 26.75 msec and was followed by a second peak at 31.5 msec. The latency of the latter peak corresponded to that of the first peak in the PSTH when the subject was relaxed. Note also that when the unit was active, the timing of the earliest peak to anodal stimulation was some 1.75 msec shorter than the first peak to magnetic stimulation. However, the following peaks had similar latencies for both forms of stimulation. The latency of the initial peaks when the muscles were relaxed also were the same for both forms of stimulation. Similar behaviour was observed in 9 of the 10 units. The shortest latency at which a unit discharged was greater when the subject was relaxed than when contracting, For electrical stimulation, the mean ( _+ 1 S.D.) latency difference was 3.64 + 0.3 msec (4 units). With magnetic stimulation the mean latency difference was 2 _+ 0.6 msec (5 units). These latency differences are very similar to those described for surface E M G potentials (Day et al. 1986, 1987a; Hess et al. 1987). Whenever brain stimulation discharged a relaxed unit, the

same intensity of stimulation produced multiple PSTH peaks when the same unit was contracting. In these 9 units, the timing of the earliest PSTH peak when relaxed corresponded to the latency of the second PSTH peak when the same unit was active (at comparable intensities of stimulation). The probability of a unit discharging in response to a cortical stimulus was greater when the unit was contracting than when relaxed. Units were discharged by the cortical stimulus on 30-90% of trials when contracting compared with 10-50% when relaxed. This can be seen in the difference in size of the 'relaxed' and 'active' peaks in the PSTHs in Figs. 1 and 2. The widths of the PSTH peaks in the relaxed state were a fraction of a millisecond shorter than the peaks when active for both forms of stimulation. This suggests greater variability in the precise timing of motoneurone discharge (over an interval of 1 1.5 msec) when contracting within the preferred intervals following the cortical stimulus. One of the 10 units studied (unit 7 in the table) proved an exception to this general pattern. Anodal electric stimulation at an intensity of 55%. discharged this unit at the same latency when relaxed and active.

SINGLE MOTOR UNITS AND CORTICAL STIMULI

401

ANODAL STIMULATION

AclJve

(55*/o)

• pinni nln.n.nn.l .

10

20

.

.

30

.

.

.

.

40ms

i

J

2S

40ms

Relaxed (55%)

. . . . . . . . . . . . 10

20

30

h....

I

4~ms

i

.. . i 40me

Fig. 3. Same unit as shown in Fig. 2 (unit 6 in Table I). Left-hand panel shows PSTH to anodal stimulation when active (upper panel) and when the subject was relaxed (lower panel) at stimulus intensity of 55% maximum. The first PSTH peak when active (latency 26.75 msec) occurred some 4 msec earlier than when relaxed (latency 31.5 msec). The timing of the second peak in the active PSTH (31.5 msec) corresponded to that of the first peak in the relaxed PSTH. Righthand panels show motor unit discharges from which the histograms were constructed. Note also that the timings of the first peak in the histograms for this unit when relaxed were virtually identical for both anodal (31.25 msec) and magnetic (31.75 msec) stimulation (compare Figs. 2 and 3). When contracting the latency of the first peak to anodal stimulation was about 1 msec shorter than the first peak to magnetic stimulation, but the timing to both forms of stimuli the second peaks remained comparable. The calibration bar between the two histograms represents 5 counts.

However, the probability of discharge (i.e., the size of the peak) was 50% smaller when the muscle was relaxed.

Discussion The present study has confirmed that the latency of a single motor unit discharge in response to electrical or magnetic stimulation of the brain usually is shorter

when the motor unit is active during voluntary contraction than when at rest. This latency difference is of the order of 1-2 msec for magnetic stimulation and 3 - 4 msec for electrical stimulation. These values are comparable to the difference in latency of surface recorded E M G potentials following brain stimulation. For electric stimulation, surface responses recorded when contracting are 2 - 6 msec (mean 3.7 msec) shorter than when the same muscle is relaxed (Day et al. 1986). The difference with magnetic stimulation has ranged from 1 to 2 msec (Day et al. 1986) to 1-5 msec (mean 3.3 msec) (Hess at al. 1987). One possible explanation for this phenomenon is that larger more rapidly conducting peripheral motor units are recruited by voluntary contraction. The results showing the similar behaviour of single motor units and surface E M G responses indicate that this is not the case. Threshold and latency differences between responses in relaxed and active muscles may be explained on the basis that a single cortical shock may generate several descending waves of excitation, similar to the D and 1 waves described in the monkey (Patton and Amassian 1954). The results from 9 of the 10 units fit with this explanation in 2 respects: (i) stimulation of the brain when the muscle was active, at the same intensity that produced a response in the relaxed state, gave rise to multiple peaks of increased firing probability in the PSTH; (ii) the latency of the second of these peaks was the same as the latency of the earliest peak in the relaxed muscle. Multiple peaks in the PSTH probably are produced by the arrival of several EPSPs at the spinal motoneurone (Day et al. 1987a,b). Our hypothesis is that a comparable number of EPSPs are generated when the subject is relaxed, but that units fail to discharge on receipt of the first EPSP because their resting potential is well below threshold. Temporal summation with a second EPSP is needed to raise the neurone above threshold and produce a discharge. Stimuli insufficient to give rise to multiple peaks in the active PSTH usually fail to discharge the same unit in the relaxed muscle. Enhancement of spinal motoneurone excitability during voluntary contraction enables some motoneurones to discharge on arrival of the first descending volley. When relaxed, the time taken for temporal summation of two or more descending volleys corresponds to the latency difference. The novel finding in this study is that the timing of the first peak in the PSTH of a relaxed motor unit is the same as the second peak of the same motor unit when contracting. The result from the anomalous unit is in accord with this framework. In this unit, the first EPSP produced by the stimulus must have been of sufficient size to bring the resting motoneurone to threshold. Since we have no way of knowing how far below threshold a 'resting' motoneurone might be, it is conceivable that a subject who is not totally relaxed may have a propor-

4112

tion of silent units which are close to threshold and easily discharged by any excitatory input. An alternative explanation for the earlier discharge of an active muscle is that voluntary contraction facilitates the cortical response to the brain stimulus, and an additional descending volley (that is not present at rest) is issued at an earlier timing. There was no evidence for such an additional earlier volley in a previous study using electrical stimulation in man nor was there any evidence of a change in size of the initial descending volley (Cowan et al. 1986; Day et al. 1987b). A change in size of the initial descending volley to either an electrical or magnetic stimulus would not explain the latency shift with contraction since even the smallest volley would be expected to discharge some motoneurones when the muscle was active. The threshold and latency differences when relaxed or active are then accounted for by changes in spinal excitability. However, an additional contribution from cortical facilitation to the size of later descending volleys to electrical stimulation when contracting cannot be ruled out on the basis of the present experiments. In animal experiments repetitive intracortical stimulation increases the size of synaptically activated I waves (Jankowska et al. 1975). If the multiple descending volleys set up by each electric cortical stimulus in man are analogous to D (direct) and I (indirect) waves travelling in the same pyramidal tract neurone, cortical facilitation of later 1 waves by summation with voluntary effort and other afferent inputs to the motor cortex might be obscured in the PSTH by the 'shadow' of the initial D wave to electrical stimulation. Similarly, the I waves to magnetic stimulation also might be subject to cortical facilitation. It is not possible to comment on whether the initial volley to magnetic stimulation changes in size since it is not effective in discharging motoneurones at rest and therefore is not seen in the PSTH. Direct recordings of the descending volleys would be necessary to further explore this possibility. Other possible explanations for the earlier responses during voluntary contraction include the recruiting of different regions of the brain during voluntary contraction. Latency differences of 2 - 6 msec would require the spread of current by several centimetres (Day et al. 1987a). Furthermore, higher intensities of stimulation are often required to activate the resting muscle, in comparison with the active muscle. If current spread to deeper parts of the brain were responsible for any latency difference, the higher intensity stimulus required to elicit a response when relaxed might be expected to spread further in the brain and result in a shorter latency; in fact, the reverse is true.

P.D. THOMPSON ET AL.

Finally, these results reaffirm previous observations on the differences between anodal and magnetic stimulation (Day et al. 1986, 1987a, 1989). Latency differences are most evident in the contracting muscle. The latency of the response when relaxed is similar for both techniques. The single unit illustrated in Fig. 3 shows that the timing of the first discharge when contracting is earlier for anodal stimulation, but the following discharge occurs at similar timings. When relaxed the unit discharges at the timing of this second peak for both forms of stimulation.

References Barker, A.T_ Jalinous, R. and Freeston, I.L. Non-invasive magnetic stimulation of the human motor cortex. Lancet, 1985, ii: 110611(17. Calancie, B., Nordin, M., Wallin, U. and Hagbarth, K.E. Motor-unit responses in human wrist flexor and extensor muscles to transcranial cortical stimulation. J. Neurophysiol., 1987, 58: 1168-1185. Cowan, J.M.A., Day, B.L.. Marsden, C.D. and Rothwell, J.C. The effect of percutaneous motor cortex stimulation on H-reflexes in the muscles of the arm and leg in man. J. Physiol. (Lond.), 1986, 377: 333-347. Day, B.L., Dick, J.P.R., Marsden, C.D. and Thompson. P.D. Differences between electrical and magnetic stimulation of the human brain. J. Physiol. (Lond.), 1986, 378: 36P. Day, B.L., Thompson, P.D., Dick, J.P.R,, Nakashima, K. and Marsden, ('.D. Different sites of action of electrical and magnetic stimulation of the human brain. Neurosci. Lett., 1987a, 75:1111 I (16.

Day, B.L., Rothwell. J.~... Fhompson, P.D.. Dick. J.P.R., ('()wan, J.M.A.. Berardelli. A. and Marsden, C.D. Motor cortex stimulation in intact man. II. Multiple descending volleys. Brain. 1987b. 110: 1191-1209. Day, B.L. Dressier. D.. Maertens de Noordhout. A., Marsden, C.D.. Nakashima. K.. Rothwell. J.C and Thompson, P.D. Electric and magnetic stimulation of human motor cortex: surface EMG and single motor unit responses. J. Physiol. (Lond.I. 1989. 4t2: 449473. Hess, C.W.. Mills. K.R. and Murray, N.M.F. Responses m small hand muscles from magnetic stimulation of the human brain. J. Physiol. (Lond.). 1987. 388: 397-419. Jankowska. E.. Padel. A and Tanaka. R. The mode of activation of pyramidal tract cells by intracortical stimuli. J. Physiol. (Lond.). 1975. 249: 617-636. Marsden. C.D.. Merton. P.A. and Morton. H.B. Direct electrical stimulation of corticospinal pathways through the intact scalp in human subjects. Adv. Neurol.. 1982. 339: 387-391. Merton. P.A. and Morton. ll.B. Stimulation of the cerebral cortex in the intact human subject. Nature. 1980, 285: 227. Patton. H.D. and Amassian. V.E. Single and multiple unit analysis of cortical stage of pyramidal tract activation. J Neurophysiol.. 1954. 17: 345-363. Rothwell. J.C.. Thompson. P.D.. Day, B.L.. Dick. J.P.R.. Kachi. T.. Cowan. J.M.A. and Marsden. C.D. Motor cortical stimulation in intact man. I General characteristics of EMG responses in different muscles. Brain. 1987, I10: 1173-1190.