Three experiments are reported where a cue that was varied in time indicated a letter pair (or pairs) in a circular display with six pairs. The S had either to report letters from a pair (Exp. 1) or to decide about the equality of the letter pair (Exp. 2) or to decide about the presence of a target letter given with various delays (Exp. 3). Exp. 1 shows a short-lived partial report superiority, the loss being primarily due to adjacency errors. In Exp. 2 a short loss in the correct same decisions, but almost no loss in the correct "differents" was observed. In spite of its search task character, Exp. 3 showed the same loss as Exp. 1, 2. In all experiments performance recovered with the latest ISI (1 sec). The results of Exp. 1 can be explained by post-categorical accounts of the partial report (PR-) effect (loss of positional information), those of Exp. 2 by visual confusion, i. e. a precategorical account, those of Exp. 3 by neither. The results suggest that the PR-effect might be due to non-visible persistence rather but to visible persistence. A theory of early visual processing which would also explain the PR-effect is still lacking.