1. Spontaneous flexion-extension tremor of the elbow was recorded in normal human subjects while they exerted a flexing force against different loads.2. Increases in the flexing force up to (1/2) or (3/4) of the subject's maximum were accompanied by increases in the amplitude of the tremor. A further increase to very large forces was associated with a decrease in tremor.3. When the subject flexed against a spring, the frequency of the tremor was related to the properties of the spring; with increasingly stiff springs the tremor frequency increased up to about 12 Hz. With the stiffest springs, however, the tremor became irregular and very small in amplitude.4. Weights added to the limb at the wrist reduced the frequency of tremor.5. By using appropriate combinations of spring and mass, the principal tremor frequency could be adjusted between 2 and 12 Hz, higher in some subjects. Within this range the frequency(2) was approximately proportional to the spring stiffness/mass, just as it is in a spring-mass system, though the forearm provided some of this mass, and some of the spring-like resistance was in the flexor muscles.6. When by suitable loading the tremor frequency was brought between 8 and 12 Hz, and the subject forcibly flexed his elbow, the tremor became regular and large in amplitude. This was a form of limited instability in the stretch reflex which arose because a powerful reflex response to extension of the elbow acted back on the flexor muscles after a delay.7. When the limb was so loaded that its natural frequency was outside this 8-12 Hz range, it behaved as a filter, and in response to any disturbing noise it oscillated mainly at frequencies close to the natural frequencies of either the mechanical system or the stretch reflex.8. This filtering function of the limb with its stretch reflex probably plays an important part in the control of normal movement and tremor.9. The large flexing forces that were here associated with vigorous tremor normally occur when a limb supports a large mass. Such a mass would give the mechanical system within the limb a low natural frequency; inconveniently large oscillations in the 8-12 Hz range would therefore seldom occur.