The effects of amputation of a rear leg on the pattern of motor activity in the legs of freely walking cockroaches (Periplaneta americana L.) were studied. 2. Amputation affected both the frequency and the timing (phase) of motor bursts during a stepping cycle. Bursts in the stump of an amputated rear leg and in the contralateral (intact) rear leg often occurred at two or three times the frequency of bursts in the other legs. The remaining legs also showed multiple bursting during some steps. 3. Amputation affected the phase of motor bursts in two different ways. First, for every leg pair, phase was more variable after amputation, whether or not the mean phase was affected. Second, for some leg pairs, the mean phase itself was altered. During most steps, the timing of motor bursts in the stump of the amputated leg was walking-speed-dependent relative to bursts in the anterior legs. In contrast, the timing of bursts in the stump relative to bursts in the legs across the body from it showed no such speed-dependent timing. Timing between bursts in pairs of intact legs also showed either speed-dependent or speed-independent effects, depending on the pair under consideration. 4. The effects of amputation were not consistent. After loss of a leg, bursts in some leg pairs occurred synchronously in some insects and alternately in others. Even in single insects there were cases in which the timing between bursts in two legs switched from one value to another during walking. 5. These effects of amputation were manifest during slow walking only. At higher speeds, the timing of motor bursts in different pairs of legs was consistently closer to that seen during walking in intact insects. 6. Three conclusions are drawn from these results. (i) During slow walking, sensory feedback from the legs helps maintain the timing of adjacent ipsilateral leg pairs, but has little influence on contralateral pairs. (ii) During slow walking, either sensory input is quite variable, or it has variable effects on the motor pattern. (iii) During fast walking, sensory input from the legs seems to play a minimal role, if any, in the timing of the motor pattern of walking.
