Does spatial separation incur a processing time requirement before precise alignment judgements can occur? Alignment thresholds for separated lines are measured for exposure durations from 27 to 500 ms, with and without post-masks. The effect of masks on visibility is controlled. Unlike without a post-mask, with an effective post-mask, alignment thresholds improve substantially with time, i.e. in square-root fashion. Alignment across space may be important for further shape analysis. Threshold improvement is probably not explained by a spatial scale shift of visual analysis over time. A higher-order collection stage appears to refine relative position information for up to 200 ms.