In ensemble musical performance, players strive to generate synchronous onsets for notes that are supposed to be played together. Important cues for correcting phase are available in the acoustic onsets of notes, so that when a pair of players detects an asynchrony, the timing of the next note may be adjusted (Schulze, Cordes & Vorberg, 2005). However, such feedback correction only works in the body of the music and leaves unsolved the problem of how to synchronise entries such as at the beginning of a piece. In such cases vision can be a useful source of information as the movements of the players (eg bowing arm in string players) in preparation for playing may allow others to predict the intended timing of the next note onset. In order to investigate the role of vision and audition in music ensemble, we prepared an avatar from a motion capture recording of the first violinist in a leading chamber group playing the opening of a Haydn quartet. We then selectively edited out the left and right arms and head. When the other three players were asked to play with the reduced cue avatar, we observed the variance of asynchrony was smallest with left arm absent, larger with head absent and highest with right arm absent. We conclude that bowing arm movements provide important cues to timing in chamber music performance. We suggest that this method has promise for investigating other aspects of visual cue utilisation in music.