1. When the dark-adapted eye is exposed to a brief duration (2 msec) luminous line the resulting threshold disturbance is much sharper (decay constant of ca. 10 min arc) than would be expected in a system which is known to integrate the effects of light quanta over a distance of 1 deg or so.2. When the forcing input is a pair of brief duration parallel luminous lines the threshold disturbance falls off sharply at the outsides of the pattern but on the inside a considerable spread of threshold-raising effects may occur unless the lines are sufficiently far apart.3. The threshold disturbance due to a briefly exposed edge shows an overshoot reminiscent of ;lateral inhibition'.4. If the threshold is measured at the centre of a black disk presented in a briefly lit surround then (a) the dependence of threshold on time interval between test and surround suggests that the threshold elevation is due to a non-optical effect which is not ;metacontrast'; (b) the dependence of threshold on black disk diameter is consistent with the notion that the spatial threshold disturbance is progressively sharpened as the separation of luminous edges increases.5. If the threshold is measured at the centre of briefly exposed luminous disks of various diameters one obtains the same evidence for an ;antagonistic centre-surround' system as that produced by other workers (e.g. Westheimer, 1965) for the steadily light-adapted eye.6. The previous paper (Hallett, 1971) showed that brief illumination of the otherwise dark-adapted eye can rapidly and substantially change the extent of spatial integration. The present paper shows that brief illumination leads to substantial ;inhibitory' effects.7. Earlier approaches are reviewed: (a) the linear system signal/noise theory of the time course of threshold disturbances (Hallett, 1969b) is illustrated by the case of a small subtense flash superimposed on a large oscillatory background; (b) the spatial weighting functions of some other authors are given.8. A possible non-linear model is briefly described: the line weighting function for the receptive field centre is taken to be a single Gaussian, as is customary, but the line weighting function for the inhibitory surround is bimodal.