Figure 1 shows typical recordings of electomyographic potentials (EMG) recorded from the forearm and somatosensory potentials (SEP) recorded from the scalp in response to magnetic stimulation of the upper arm with a circular coil. Figure 2 shows boxcar filtered data from the same subject where the M/S ratios (normalized EMG/normalized SEP) for both current directions closely match the mean ratios of all 20 subjects. Figure 3 shows a different subject where a larger difference between the ratios was observed. Note that in both cases, CW (clockwise) current produced a larger EMG than did CCW (counter-clockwise) current in the coil, while CCW stimulating current produced a larger SEP than did CW current.

All 20 of the subjects had more EMG activity and less SEP activity when stimulating with a CW current direction compared to the CCW current direction (p 1). The magnitude of this difference was significant for 18 of the 20 subjects (p

As for the effectiveness of the coils on the subjects, coil 3 (Table 2) was the most effective. In 13 of the subjects, it produced the largest contrast between the ratios. Coil 2 was responsible for producing the largest contrast between ratios in 6 of the subjects while coil 1 was best in only one subject.

The mean of the M/S ratios for all subjects was 4.5 times larger for CW stimulation than for CCW stimulation (Table 1). This value is an underestimate of the actual magnitude of the difference, since the ratio of the two values is bounded on the low end by 0, but unbounded on the upper end. If there was no effect due to current direction, or current direction affected EMG and SEP equally (e.g., one direction was more effective in eliciting neural responses than the other), the mean M/S ratio would be the same (specifically, it would have a value of 1) for both current directions. To provide a symmetrical distribution around this mean value of 1, and to allow statistical analysis of the data, the ratios were log transformed. Note that, since the data were log transformed, one needs to look at the ratio of the CW to CCW M/S ratios, not the difference between the two.

The log transformed data showed that the mean M/S ratio for CW stimulation was 5.25 that of the ratio for CCW stimulation, with a 99% confidence interval of 4.19 to 6.59 (Tukey's Honestly Significant Difference). A multi-way ANOVA test on the log transformed data from all 20 subjects gave a p