3 groups of subjects learned a list of 10 2-syllable adjectives, followed by a second list composed of synonyms of these words. 3 positions of interpolated activity were studied: immediately after original learning, halfway between original learning and recall, and just prior to recall of the original list. The interval from learning to recall was 96 hours. The position halfway between learning and recall of the original list gave the least amount of retroaction. The introduction of a second list just prior to recall yielded less retroaction than when given just after learning of the first list. The results are explained in terms of confusion between constellations of associations and memory traces. (PsycINFO Database Record (c) 2006 APA, all rights reserved)