Saramsha

No. 012  ·  28 September 2026

Sand Finds the Silence

Sprinkle sand on a metal plate and bow its edge. The grains are thrown off every part that is moving and come to rest on the lines that are not — so the sand draws the shape of the note. Slide the pitch until the plate answers. Drag the bow anywhere on it.

Drag the bow

About this piece

The figures that cost Napoleon 3,000 francs

In 1680 Robert Hooke drew a bow along the edge of a glass plate dusted with flour and saw the flour arrange itself into patterns. A century later Ernst Chladni made a life's work of it. His Entdeckungen über die Theorie des Klanges of 1787 catalogues plate after plate of brass, strewn with sand, bowed at the rim: for each pitch the sand abandons most of the plate and gathers into a figure of startling regularity — crosses, stars, grids, rosettes. He toured Europe demonstrating them, and in 1809 he performed for Napoleon, who understood at once that nobody could explain them. He put 3,000 francs before the Institut de France as a prize for a mathematical theory of vibrating elastic surfaces.

It was won, eventually, by Sophie Germain. Barred as a woman from attending the Academy's sessions or studying at the École Polytechnique, she entered anonymously in 1811 as the only competitor; the judges found her analysis wrong. She entered again in 1813 and got an honourable mention. On her third attempt, in 1816, she won — with the right governing equation, though her boundary conditions were still not sound. Gustav Kirchhoff settled those in 1850. And in 1909 a young Swiss physicist, Walther Ritz, invented the method now taught as Rayleigh–Ritz — the ancestor of every finite-element solver in use today — for the express purpose of computing the Chladni figures of a free square plate. The sand was setting mathematics homework for a hundred and thirty years.

Why the sand moves at all is the nicest part, and it is not what the finished picture suggests. A grain sitting on a violently moving patch of metal is thrown into the air and lands a little way off, at random. A grain sitting on a line that is not moving is not thrown anywhere. So the sand takes a random walk whose step length is the local loudness, and such a walk drains downhill in loudness and pools where the plate is still. That is exactly what the grains on this page are doing: each frame every grain samples how hard the metal beneath it is shaking, hops that far in a random direction, with a bias down the slope, and freezes when the shaking stops. Nobody draws the lines. Michael Faraday found the exception in 1831: very fine powder, lycopodium spores, does the opposite and collects at the loudest places, because grains that light are steered not by the plate but by the swirls of air above it.

The plate has a set of shapes it likes — its modes — and a pitch belonging to each. Feed it something in between and it barely stirs; the curve above the pitch slider is that indifference drawn out, and the peaks are the pitches it cannot refuse. A stiff plate resists bending rather than stretching, so its pitches climb with the square of how finely the mode is folded: mode after mode is spaced further apart than on a guitar string, which is why a plate never sounds like a note and a bowed one sounds like a shriek.

Two details here are worth hunting for. First, the bow chooses the figure. On a square, the mode folded three ways east–west and five ways north–south has exactly the same pitch as its mirror image, five ways east–west and three north–south. The plate is free to vibrate in any blend of the two, and what settles it is where you happen to be bowing. Hold the pitch and drag the bow: the figure reorganises without the note changing at all. Second, a bow on a still line is a bow on nothing. Put it at the dead centre of the round plate and every mode with a nodal diameter through the middle vanishes from the spectrum, leaving only the rings. The curve on the slider redraws itself as you drag, so you can see modes being switched off.

Violin makers still do this. Tapping and bowing an unglued top or back plate, with glitter or tea leaves on it, and reading the figures, is how the thickness is graduated — the practice was formalised by Carleen Hutchins in the 1960s and the first few modes of a free plate have standard numbers among makers to this day. A two-hundred-year-old parlour trick is a working instrument in a workshop.

The square plate uses Rayleigh's classic approximation for a free edge, a sum of products of cosines, driven at a point; the round one uses Bessel mode shapes with a fixed rim, and both are given the stiff plate's frequency law so that the modes stay far enough apart to hunt for. Forced response, nodal field, gradient and eight to fourteen thousand grains are all recomputed in your browser every frame; nothing is uploaded and no sound is recorded. Tab to the plate, then arrow keys move the bow, [ and ] step between modes, S is sound, W the wave underneath, R fresh sand.