Recognition has traditionally been regarded as a more sensitive measure of retention than recall.1 Recent experiments have suggested, however, that when the number of alternatives on the recognitionand recall-tests2 and the similarity between correct and incorrect alternatives on the recognitiontest are equated,3 the apparently greater sensitivity of recognition is reduced, if not eliminated. Furthermore, the temporal sequence of recognition and recall appears to modify the effect that these two methods of measuring retention have upon each other. Recognition has been found to be poorer after recall than before recall, while on the other hand, recall is better after recognition than before.4 This difference was attributed by the authors to the dropping out of weak associations during an initial recall test, thus lowering subsequent recognition performance, and to the rehearsal function of the recognition test which serves as a final learning trial for later recall.5 Tests of recognition in all of the studies cited above have been of the discriminative variety, with the Ss required to select the correct from among incorrect alternatives. Discrimination of the correct response from a display of alternatives would generally be expected a somewhat easier task than 'absolute' recall which requires total reconstruction of the item. Furthermore, the Ss would seem to benefit more from partial learning6 on this test of recognition than on recall. On the other hand, a binary 'yes-no' test with