The properties of neurones in the visual system are being subjected to ever-closer scrutiny. The most common method for studying these neurones is to compare their responses to different visual stimuli, averaged over 20- 200 presentations of each stimulus. As experiments become more sophisticated, the accuracy of these response measurements becomes increasingly important, while the number of stimuli to be compared increases. Since the responsiveness of neurones may change spontaneously during an experiment, comparisons among their responses to different stimuli are rather perilous if all the repetitions of each stimulus are presented in a single block. If the neurone's responsh-eness does change during the course of an experiment, a false estimate may be obtained of the relative effectiveness of different stimuli.