The purpose of the study was to determine which biomechanical factors changed and how they changed during the process of skill acquisition. A novel task was selected that could be learned during one testing session. Each of the 25 male subjects threw a ball from a cup 40 times at a graded target while biomechanical data were collected on-line to an A to D converter by a program in a PDP-11/34 computer. The cup was attached to the end of a hinged metal frame into which a subject’s dominant arm was fastened. This limited the throwing motion to forearm flexion and extension only, as the elbow joint center coincided with the hinge of the frame.

Muscle activity in the biceps brachii, brachialis, brachioradialis, and triceps brachii was monitored via EMG, using surface electrodes. The following events were calculated by a computer program during the counter-movement and extension phases of each throw:​ the duration of high activity, low activity (over 75 percent and under 50 percent, respectively, of activity levels during throws for maximal distance), and any activity, plus the amount of muscle activity.

Elbow displacement data were obtained from a potentiometer at the hinge and later smoothed and differentiated to calculate velocity. Accelerations were calculated from a piezoelectric accelerometer and a pulse generated by the release of a contact switch by the ball indicated the end of the throw. A computer program was also used to calculate the maximum flexion angle, angle at release, velocity at release, peak positive acceleration, and peak negative acceleration.

Since the subjects acquired the task within the first 10 trials, group analysis was done on these throws. Only the mechanical factors monitored showed changes as skill was acquired, but elbow angle and elbow velocity at release were the best predictors of score. (The multiple correlation coefficient accounted for 99 percent of the variance in score.) With practice, the maximum flexion angle and release angle decreased, release velocity increased, peak positive acceleration increased, and peak negative acceleration decreased.

A regression equation was built predicting the horizontal ball angle from the horizontal cup velocity at release for a perfect score. It was found that to obtain the desired performance outcome each subject selected his own limited range of ball angles and cup velocities at release. This was his strategy and, as the ball angle decreased, velocity increased.

The values obtained for biomechanical factors were the same for the subjects divided into three groups according to the rate at which they acquired the skill. The factors included in individual regression equations to predict score were also the same for the three groups. In addition, no differences were found in the strategies used by the members of each group in order to obtain a perfect score.