Fig. 1. Experimental setup. (A) In the free-coxa preparation, the dorsal surface of Blaberus was attached using epoxy glue to 0.95 cm thick Plexiglas^TM and the tarsus removed (broken outline). A servo-motor applied a dorsally directed impulse to the tibia, 1 mm from the distal tip. The free response of the leg was recorded at 1000 frames s^–1 and the position of a marker 1 mm from the distal tip of the tibia was digitized. The abdomen was pulled dorsally slightly and held so that it did not interfere with the free response of the leg. The fixed-coxa preparation was identical except the leg was removed from the body and glued to the Plexiglas^TM at the coxa. In all instances the femur–tibia joint was rigidly fixed by cyanoacrylate, and in half the trials the C-Tr-Fe joint was also rigidly fixed. (B) The joint axes of rotation of the distal joints are parallel with the applied impulse direction (into the page). The body coxa joint's primary axis of rotation runs medio-laterally, so it is free to rotate as a result of the dorsally directed impulse perturbation. The body–coxa joint also has secondary joint axes parallel to the applied impulse. (C) Video frames show the time sequence of the left metathoracic leg's response to an impulse perturbation. The animals's dorsal surface is towards the left of the image while anterior is towards the bottom of the image.