Comparison of electromyogram spectra with force spectra during human elbow tremor.

1. The tremor that develops when the elbow is flexed against a spring attached at the wrist has been analysed by determining the 'power' spectrum of the demodulated surface e.m.g. recorded from two of the active muscles, biceps and brachioradialis. This was compared with the corresponding force spectrum obtained by analysis of the force developed at the fixed end of the spring (normally one of stiffness 2.8 N/mm); this force is directly proportional to the movement at the wrist. 2. When the subject was maintaining a high target force (100-160 N), with the aid of a visual display, the tremor was large with a large sharply tuned peak in the force spectrum and there was a clear peak at the same frequency in the e.m.g. spectrum. The coherence (gamma) between the force peak and the corresponding e.m.g. peak typically had a value of 0.95 or above, indicating a high degree of correlation. 3. On developing the same target force against a rigid restraint (70 N/mm) the peak in the force spectrum was absent or very much smaller and less sharply tuned. More particularly, the tremor-related peak seen in the e.m.g. spectrum under compliant conditions was no longer present under rigid conditions. 4. At low target forces (20-40 N) with compliant loading there was a small peak in the force spectrum but no peak could be detected in the e.m.g. spectrum. With increasing target force the mechanical tremor increased considerably and a peak progressively emerged from above the background level in the e.m.g. spectrum, accompanied by the development of a corresponding peak in the coherence spectrum. Thus the difference in detectability of peaks in the e.m.g. and force spectra might simply result from differences in the background 'noise' level of the two types of spectra. 5. Changing the spring stiffness in the range 0.7-12.5 N/mm altered the frequency of the mechanically recorded tremor by 1.5-2 Hz and the peak in the e.m.g. spectrum shifted in approximate correspondence. 6. The findings support the view that the tremor seen with compliant loading of the arm is due to the stretch reflex. In addition, this work should help define the conditions under which spectral analysis of the gross e.m.g. is of practical utility.