ABSTRACT. Normally, whether evidence of observation e confirms hypothesis h relative to background evidence k is a matter of the logical relations between e, h, and k, and independent of whether h was formulated before or after the discovery of e. This is supported by intuitions about various examples from history of science, and fits in with a Bayesian account of confirmation. There are however abnormal cases where the background evidence has a special character, when it does become relevant when h was formulated and k records that.