From the theory of generalized motor programs (GMPs), Young and Schmidt (1990, 1991) derived methods to examine motor organization. A unit is defined as a segment of behavior governed by a single GMP, for which relative timing is invariant. If so, the correlations (within subjects, over trials) among the times of various kinematic landmarks within the unit should be close to 1.0; correlations between two landmarks in different units should be low. This observation allows the empirical identification of units, as well as their durations and between-unit boundaries. Using a coincident-timing task, we show here that a backswing consisting of a biphasic movement was organized as a single unit; this unit was followed by a forward swing as a second unit. For this task, increasing the speed of the display decreased the durations, but not the funda- mental unit structure of the actions. The findings add confidence that these methods provide reasonable descriptors of motor organization.