The project gathered data on a number of experiments using a variety of tasks. The purpose was to show that there are conditions relevant to the differential composition o f a task and other conditions which are irrelevant. Invariances in level, slope, and regularing of pattern of intertrial correlations were found for cranking, a repetitive work task. Learning trials with very simple rotary or linear positioning tasks did not show the superdiagonal form of intertrial correlation matrix found with complex tracking tasks. Experimental variables such as the precision required and the directions given to the subject did affect the differential composition of a task. The general conclusions are that some experimental variables do alter an individual's performance and the processes involved in learning can be manipulated and analyzed. Implications of this study for standardized tests and education programs are discussed. (NS)