A redundant robot manipulator has several certain or expected advantages over a nonredundant one. It is expected, among other capabilities, that the joints vary with constant velocities during the execution of those tasks which in a nonredundant manipulator require variable joint velocities. In this way, motion becomes more precise because of the elimination of errors associated with velocity change in joints. In this paper, it is shown that this expected advantage is not possible for all the joints, and that only as many joints as the degree of redundancy can have constant velocities.