rable to that obtained using the microneurographic technique. 4, Being less affected by the conduction time of the afferent somatosensory fiber and by central processing, SuCV may be a useful parameter with which to evaluate efferent sudomotor function. Denis Λlic Λet al., however, noted that it is not based on the activity of the same unmyelinated axons and sweat glands. 1 Elie et al., who tried to determine SuCV in the lower extremities of normal subjects, failed to obtain SSRs at the proximal site (inner side of the thigh) in 13 of 30 subjects. 4 Further study is needed to determine the meaning of the decrease in SuCV.
The SSR amplitude varies greatly even in normal subjects; therefore, some authors have suggested that amplitude is an unreliable parameter for SSR, 3,9 but others have reported the SSR amplitude is important clinically. We agree with the opinion expressed by Denis Λlic Λet al. that the presence or absence of SSR is only one aspect of the overall considerations. The problem of variability may be countered, to some extent, by using an amplitude parameter with reproducibility. Although all SSR amplitude parameters in our study showed considerable inter-and intrasubject variability, the maximum amplitude gave the best results in terms of reproducibility (Fig. ). High intersubject variability, shown as a large standard deviation (SD) value or coefficient of variation across subjects (CVAS), makes it difficult to set a normative range of SSR amplitudes. For example, we found that the mean maximum amplitude (n = 35) minus 2 SD gives a value below 0 mV. We think a definition of normative range based on percentile would be practical. The log transformation also is used to achieve normality. Latency showed less variation than the SSR amplitude. In our study the mean value (n = 35) of the mean latencies of 20 SSRs in each subject was 7.1% of the CVAS, which corresponds to the value for minimum F latency (6.1%, median nerve and 7.0%, tibial nerve) 7 obtained for 45 normal subjects. In terms of the other reported SSR latency CVAS (4.7% to 12.9%), 4, our 7.1% is an intermediate value. If the normative range is defined as the mean Β± 2.5 SD (range 1.16-1.66 s), only one measurement in the 82 total test sessions, including the follow-up examination (n = 35), would be beyond this range.