Memory for well-known musical phrases was tested first for recognition in the absence of any specific musical context and then for recall given the preceding musical phrase as a contextual cue. Recognition and recall were found to be largely, but not completely, independent. Moreover, there was no evidence of any greater dependency between recognition and recall than that previously observed in the relation between word recognition and recall, as summarized by the Tulving-Wiseman law. These findings significantly extend the range of applicability of this law.It has been claimed that there is an empirical law in the relation between recognition and recall. This law has come to be known as the Tulving-Wiseman law, after those who discovered the regularity that it embodies (Tulving & Wiseman, 1975). Tulving and Wiseman proposed that this regularity could be summarized mathematically by the following equation:where the constant e = .5. In this equation, the relation between recognition and recall is assessed by the conditional probability of recognition (Rn) given recall (Re), and the relation is one in which recognition and recall are largely, but not completely, independent.The claim that this relation between recognition and recall is lawful has been recently discussed by Gardiner (1988Gardiner ( , 1989, Jones and Gardiner (1990), Nilsson, Dinniwell, andTulving (1987), andNilsson, Law, andTulving (1988). But the fullest review has been provided by Gardiner and Nilsson (1990), who state the law as follows:Recognition of a set of to-be-remembered items in the absence of any item-specific context is largely independent of their subsequent recall given item-specific contextual cues-provided these cues are functionally related to, hut not equivalent to, the target items themselves.Gardiner and Nilsson reviewed results from 78 relevant experiments from 42 published articles. These experiments yielded a total of 272 different observations of the probability ofrecognition given recall, and-with few exceptions-these observations all corresponded approximately with the probability of recognition given recall that would be expected, on the basis of Equation 1, from the probability of recognition. The experiments naturallyWe thank Ian Cross for help in computer programming and James Harnpton and Paul Williams for help in recording the music. Requests for reprints should be addressed to lohn Gardiner, Memory