The motions of Newton's cradle, consisting of several steel balls hanging side-by-side, have been analysed in terms of a sound pulse that travels via points of contact among the balls. This presupposes a focused energy beam. When the pulse reaches the fifth and final ball, the energy disperses and dislocates the ball with a trajectory equivalent to that of the first ball after it was released. The pulse passes unchanged through the internal balls without, therefore, causing movement of these balls. Lack of movement can be affirmed by immobilising one or more of the balls, thereby disproving both the gap and vibrating lattice models. This also contrasts with previous mechanisms that postulate complete energy dispersal within a ball prior to transferring the energy to another ball. Inserting an inelastic barrier between the second and third balls disrupts the pulse such that it spreads out to reach regions that are not in contact with another ball. As a result, the normally stationary third ball is forced into a forward motion, thereby pushing the fourth and fifth ball with it as a single unit. The model is valuable in explaining a fact that has puzzled physicists for generations: why is only one motional mode observed among a multitude of motions that maintain a constant momentum and kinetic energy as required by the laws of physics? The answer lies in the fact that all motions, except the one that is actually observed, require a rebound in one or more of the balls. Since the energy beam formed upon impact is unidirectional, reverse motions are not accommodated.