Kolers and Duchnicky (1985) examined the acquisition of cognitive skills in a task involving the reading of inverted letters. In 5 experiments they found gradual learning between pages, but not within pages. This difference between within- and between-pages learning was interpreted as corroboration of a theory of discontinuous skill expression. In this article, we argue that the difference between within- and between-pages learning is not caused by a discontinuity of any kind, but by an artifactual data analysis. A formal analysis of Kolers and Duchnicky's data shows that learning occurs continuously.