SUMMARY1. The effects of graded transcranial magnetic and anodal electrical stimulation of the human motor cortex were compared in human subjects undergoing orthopaedic operations on the spine, before and after withdrawal of volatile anaesthesia.Corticospinal volleys were recorded from the spinal cord in the low-cervical and lowthoracic regions (six subjects) or the mid-thoracic region (two subjects) using bipolar electrodes inserted into the epidural space.2. Electrical stimuli were delivered using anode at the vertex and cathode 7 cm laterally. The corticospinal volley at threshold consisted of a single deflection with a mean latency to peak of 4*17 ms at the rostral recording site. With further increases in stimulus strength the latency of this D wave shortened in two steps, first by 0-89 ms (seven subjects) and then by a further 0-8 ms (two subjects), indicating that the site of activation of some corticospinal neurones had shifted to deep subcortical sites.3. When volatile anaesthetics were given, a corticospinal volley could not be defined in three subjects with magnetic stimuli of 70, 80 and 100% maximal stimulator output with the coil at the vertex (Novametrix Magstim 200, round coil, external diameter 14 cm). In the remaining five subjects, the component of lowest threshold was a D wave recorded at the rostral site at 4 0 ms when stimulus intensity was, on average, 70 %. With stimuli of 90-100 % a total of five small I waves could be defined in the five subjects (i.e. on average one I wave per subject).4. After cessation of volatile anaesthetics in seven subjects, the thresholds for D and I waves were lower and their amplitudes were greater. The D wave remained the component of lowest threshold in all subjects, appearing at the low-cervical level with magnetic stimuli of 50 %. However, in three subjects I waves also appeared at D wave threshold, and the D wave was smaller than with electrical stimulation at I wave threshold. There was no consistent change in latency of the magnetic D wave as stimulus intensity was increased to 100 %.5. These findings suggest that the previously reported difference in latency of the

1. The motor cortex was stimulated electrically (vertex anode; cathode 6 cm lateral) in neurologically normal subjects undergoing surgery for scoliosis, and the evoked corticospinal volleys were recorded from the spinal cord using epidural electrodes. 2. Stimuli >330 V produced a complex D-wave volley containing three separate peaks, with high-threshold components, 0-8 ms (D2) and 1P6 ms (D3), in advance of the lowestthreshold component (D1

SUMMARY1. In fifteen neurologically normal subjects, corticospinal volleys evoked by transcranial stimulation of the motor cortex were recorded from the spinal cord using epidural electrodes in the high-thoracic and low-thoracic regions during surgery to correct scoliosis.2. Anodal stimulation at the vertex produced complex corticospinal volleys that could be recorded at both sites, with multiple waves analogous to the D and I waves documented in animal experiments. These volleys were of higher amplitude when the cathode was 7 cm lateral to the vertex rather than 7 cm anterior. There were no differences in conduction time between the two recording sites for D and I waves, when these waves could be identified at the low-thoracic site.3. Anodal stimuli of 150 V commonly produced a descending volley containing a single peak at both recording sites. Modest increases in stimulus intensity to 225-375 V produced a peak 0-8 ms in advance of the wave of lowest threshold in thirteen subjects and, in seven subjects, further increases produced an additional peak 1-7 ms in advance of the first-recruited wave. The early peaks increased in size with stimulus intensity, replacing the first-recruited wave. These results suggest that the site of impulse initiation with electrical stimulation of the motor cortex shifts from superficial cortex to deep structures, approximately 5 and 10-11 cm below the cortex. These sites are probably the internal capsule and the cerebral peduncle.4. With cathode at the vertex and anode over the 'hand area', the response of lowest threshold occurred at the latency of the anodal D wave but could not be recorded at the low-thoracic site, suggesting that it was generated by the anode over the 'hand area'. Slightly higher intensities induced a 'cathodal D wave' and still higher intensities produced late peaks at latencies of anodal I waves. These cathodal D and I waves involved axons innervating lumbar segments. There was no evidence that cathodal stimulation preferentially produced I waves. Cathodal stimulation at the vertex with the anode 7 cm anteriorly produced similar results: D waves were produced at relatively low intensities, but I waves appeared at relatively high stimulus intensities, if at all.