A rattle drum looks like something you shake. It is not. The small two-sided drum sits on a handle, and from opposite sides of the rim hang two short cords, each ending in a hard bead. To make it sound you roll the handle between your palms, back and forth. The drum turns; the beads, being reluctant to change what they are doing, lag behind. The cords swing out, and the beads sweep across the skins and strike them. Nothing is shaken. The sound is made by a delay.

That delay is worth following closely, because it explains nearly every dimension of the toy. At any instant the drum body has been turned to a new angle while the bead is still travelling on the old one. A cord can only pull, so the bead is dragged outwards until the cord is taut, then swept around until it meets the drumhead. The beat arrives a fraction of a turn after the hand that caused it. A faster twist gives a shorter delay and a harder hit; a slow one lets the bead barely reach the skin at all.

Why the cord has the length it has

The cord is a timing device, not a tether. Make it too short and the bead never gets far enough from the rim to build up speed; it scrapes the edge instead of striking the middle. Make it too long and the bead arrives late, lands near the rim on its way back, and eventually wraps itself around the handle. On a small hand drum the cord is usually a few centimetres, roughly comparable to the radius of the head. That is the range where the bead can complete most of a swing in the time the hand takes to reverse.

The bead has to be light and hard. Too heavy and its inertia drags the whole drum off axis, so the instrument wobbles in the hand; too light and there is not enough momentum to make a taut skin speak. A hard wood or moulded bead delivers a short, sharp contact, which is what a tight membrane needs. A soft bead absorbs the impact and produces a dull thud. Neither the cord nor the bead is adjustable on most examples, because the maker has already balanced them against the size of the head.

The skins themselves are part of the timing. A tightly stretched head returns quickly after being struck, so it is ready for the next beat; a slack one keeps moving and smears successive hits into each other. The two heads share the same air inside the drum, yet they are rarely tuned identically, which is why the result is a lively irregular string of taps rather than a mechanical click track.

Why two beads, on opposite sides

With one bead there would be a gap. The handle is turned one way, the bead swings and hits; the hand reverses, and for a good part of that return stroke the single bead is travelling away from the skin with nothing to strike. Put a second bead half a turn around the rim and the two take turns: while one is recovering, the other is arriving. One full back-and-forth of the wrist produces hits from both sides, and the ear hears an almost continuous rattle instead of separate taps.

Placing them opposite each other also keeps the drum balanced. Two beads 180 degrees apart pull on the rim symmetrically, so the handle does not tug to one side as it spins. A pair mounted close together would make the toy cant steadily towards them, and the beats would bunch into one part of the cycle. The arrangement is doing three jobs at once: filling the silence, balancing the rotating mass, and making the rhythm independent of which direction the hand happens to be turning.

It is the same trick as a spinning top, read backwards. A top is stable because its mass keeps its axis; here the beads are useful precisely because they refuse to follow the axis. And a diabolo is built to keep its orientation once spinning, while this drum is built so that one small part deliberately does not keep up with the rest.

What the handle actually decides

Spinning the handle continuously in one direction is a poor way to play it. Once the rotation is steady, centrifugal effect holds the beads out at a roughly constant radius and they simply orbit the drum without ever falling in to strike. The sound depends on the reversal. Each time the hand changes direction the drum slows, stops, and starts back while the beads are still moving; that is the moment the cords fold in and the heads are hit. The player is not driving a rotation, they are repeatedly creating a mismatch.

This is why the instrument rewards a loose wrist rather than strength. A quick small twist produces a dense run of taps; a slow wide roll separates them into individual beats that a child can count. There is a simple test: untie one bead and roll the handle exactly as before. The gaps appear at once, always in the same half of the stroke, and the continuous rattle collapses into a series of pairs. Nothing about the drum or the remaining bead has changed.

The whole mechanism is therefore in the object, not in any instruction that comes with it. There is no rule to learn and no setting to adjust: a cord of the right length, a bead of the right weight, and a head tight enough to answer. That is also why children's games change so slowly. Once a shape has solved its problem this completely, there is very little left for later makers to improve.