ON THE ESTIMATION OF THE PHYSIOLOGICAL COST OF MUSCULAR WORK