As part of a series of Ethnomethodological Studies of Work, this paper focusses upon a short stretch of a final concert performance of the Saint-Saens Septet by a set of amateur musicians in which timing errors occur but in response to which various manoeuvres "successfully' restore synchrony. I set out to demonstrate that these afford a strategic access for ethnomethodologists to sets of musicians' practices whereby musical synchrony is ongoingly accomplished. The central curiosity of this study is the set of distinctive practices whereby a musical text set out in a spatial, linear sequence of markings is realized in a collective performance that unfolds not only in precise time values but also in synchrony.This paper analyzes three errors in a short passage and the ways in which they are "corrected' either by the errant musician himself or by the other musician in response. In these cases, the restoration of synchrony and the routine performance of the piece thereafter are accomplished successfully in the sense that those listening to the tape of it generally fail to notice any problem! This contrasts to my detection of problems during both my participation in the performance and repeated listening of the resulting audio-tape, I argue that it is this systematic difference in possible hearings that musicians are oriented to when attempting to "fake it'. Thus, the essential interest is in members' methods for maintaining as well as restoring synchrony.What can we retrieve of members' in situ, embodied practices where talk is absent from a close description of the resultant order, the details of the musical performance? I claim that Schegloff's stipulated requirement that CA ground its claims in the details of the talk is not appropriate for the study of music. In contrast with conversation in which we are competent participants, one requires musicians' insider knowledge in order to recover just what members are doing. Repeated listening is not enough.