Normal subjects, with closed eyes, attempted to keep constant either the force exerted at the wrist or the position of the wrist against an elastic load. The load was attached to the wrist 275 mm from the axis of rotation of the elbow joint. During recording, the far end of the elastic load was displaced slowly enough that it was not immediately perceived but far enough for perception to occur before its completion. 2. The over-all relation between wrist force and position for the two conditions was approximately linear and could be described in terms of effective stiffness. The effective stiffness for the constant-position task averaged 2.8 N/mm (210 N m/rad), while for the constant-force task the mean effective stiffness was -0.028 N/mm (-2.1 N m/rad), indicative of slight over-compensation. 3. Averaging the performance at the onset of the imposed disturbance indicated that the subjects' behaviour consisted of two parts: an initial, small-range response followed by a second phase over the remainder of the displacement. The transition corresponded to the subjects' threshold for detection of the disturbance. 4. The stiffness measured for the response prior to perception was taken as a measure which included the tonic stretch reflex. The stiffness was altered appropriately for the two tasks, being lower when the subjects tried to maintain the force exerted constant (average 1.1 N/mm, 83 N m/rad) than when they attempted to keep the position constant (average 2.3 N/mm, 170 N m/rad). A small degree of co-contraction occurred but could be dissociated from the stiffness changes. 5. Scaling the results allowed comparison of the initial stiffness with values for the decerebrate cat. When analysed in this way, the values recorded in man during the constant-position task were similar to those reported for short-range stiffness in the decerebrate cat. 6. The thresholds for detection of the disturbance were much lower than those reported for subjects with relaxed muscles. 7. The stretch reflex in man has a direct role in compensating for small disturbances during motor tasks. It may also function to improve detection of applied disturbances by magnifying the corresponding force change. Once the stimulus is perceived and voluntary intervention is possible, a greater contrast is seen between the subjects' performance of the two tasks.
