It can be shown that following an angular velocity step stimulus delivered in darkness, the nystagmic responses can be effectively 'dumped' after any interval in time by the application of an appropriate step decrement in velocity. In practise the null velocity is bracketed between those step decrements inducing just detectable nystagmus to left and right and can be determined within a range of +/- 1.5 degrees s-1. With test stimuli of 22, 44, and 64 degrees s-1 the 'dump' velocities have been established at varying intervals in time on four normal subjects. Contrary to expectations the dump velocity/time relations for all three test stimuli follow a convergent linear course. The dump velocities are unaffected by fixation suppression of the nystagmus induced by the test stimuli. The seeming irrelevance of nystagmus generation to dump velocity values is confirmed by the good correspondence with the results of a separate study using the oculogyral illusion as a guide in place of nystagmus. These findings are difficult to relate to conventional concepts of cupular dynamics.