Cymatics

What it is: the study of what happens to loose material on a surface that is vibrating. Sand on a plate driven at one of its resonant frequencies gathers into a geometric figure. The figure is reproducible, it is different at every resonance, and it has been computable since 1850.

What it is usually taken for: evidence that frequency "creates form," and through that, evidence for a great many claims about healing, consciousness and ancient knowledge. Most of those claims are not supported by the physics, and the physics is considerably more interesting than they are.

---

The one thing most explanations get backwards

The sand does not collect where the energy is. It collects where the energy is not.

A plate driven at a resonance does not move uniformly. It settles into a standing wave with lines that stay still — nodes — and regions between them that move a great deal — antinodes. Loose material is thrown off the moving regions and comes to rest on the still ones. So the bright figure in every cymatics photograph is a map of the parts of the plate that are doing nothing.

This matters for reading the rest of the subject. A pattern appearing is not energy taking a shape. It is material being swept out of everywhere else.

---

The record

Date	What	Tier
1787	Ernst Chladni publishes Entdeckungen über die Theorie des Klanges in Leipzig, with eleven copper plates of figures produced by drawing a violin bow across the rim of a sanded plate	EMPIRICAL
1808–09	Chladni demonstrates the figures in Paris; at Napoleon's instigation the Academy of Sciences announces a 3,000-franc prize for a mathematical explanation	DOCUMENTED
1816	Sophie Germain wins it on her third submission, with a fourth-order equation for vibrating elastic surfaces. Barred as a woman from the university and the Academy that trained and judged her competitors, she entered anonymously — and in the first round hers was the only entry submitted	DOCUMENTED
1850	Gustav Kirchhoff publishes the correct boundary conditions and the biharmonic plate equation that still bears his name	EMPIRICAL
1967	Hans Jenny publishes Kymatik, coining "cymatics" and extending the technique to liquids and pastes. Self-published through Basilius Press — frequently described as peer-reviewed and never was	DOCUMENTED
1996	Jahn, Devereux and Ibison measure six ancient structures in England and Ireland and find every one resonating between 95 and 120 Hz, in the Journal of the Acoustical Society of America	EMPIRICAL
1999	Watson and Keating document standing waves inside megalithic monuments in Antiquity	EMPIRICAL

---

The mathematics, stated plainly

A thin plate in free vibration obeys the biharmonic equation

∇⁴w = (ρh/D)·∂²w/∂t², where D = Eh³ / 12(1−ν²)

On a disc the separable solutions carry both an ordinary and a modified Bessel function — w(r,θ) = [A·Jₘ(kr) + B·Iₘ(kr)]·cos(mθ) — and the allowed values of k come from imposing zero bending moment and zero shear at a free rim. Those allowed values are the plate's modes. Everything a Chladni plate does follows from them.

Two consequences that are rarely stated and are the whole point:

1. A plate driven between its modes shows nothing at all. Not a fainter figure — no figure. The sand lies where it fell. Most of any frequency sweep is empty, and a simulation that draws a crisp pattern at every frequency is teaching the opposite of what the instrument does.
2. The frequencies belong to the plate, not to the numbers. Change the thickness, the radius or the metal and every mode moves. A frequency quoted without the plate it was measured on carries no information.

---

Where this connects to architecture, and where it does not

The same eigenvalue problem, one dimension up, describes the air in a room. Both ask which standing patterns a bounded domain allows. Sand on a plate collects on the nodes; in a room, a listener stands in them. For a rectangular enclosure the answer is exact and has been since Rayleigh:

f = (c/2)·√[ (nₓ/L)² + (n_y/W)² + (n_z/H)² ]

This is the honest version of the claim that "a cathedral is a Chladni plate you can walk inside." It is a statement about the mathematics, and it holds.

The limit, which is usually skipped: a frequency does not imply a size until you say which wave in which medium. At 110 Hz a sound wave in air is about 3.1 metres long, so a chamber needs roughly 1.5 metres in some direction to hold one. A bending wave at 110 Hz in a one-millimetre steel plate is about 30 centimetres long, because bending waves in thin plates are dispersive and far slower. A hand-sized plate and a room-sized chamber can therefore both ring at 110 Hz with no contradiction at all. Treatments that miss this either conclude that the chambers prove something mystical or that the plate comparison is nonsense. Both errors come from the same omission.

---

What the archaeology actually found

Six ancient structures, differing in shape and size, each sustaining a strong resonance between 95 and 120 Hz — the lower baritone of a male voice. Published in a top-tier acoustics journal in 1996.

The uniformity is the finding. Differently shaped rooms of different sizes have no obvious reason to agree that closely. What the paper does not claim, and what this archive does not claim, is that the builders intended it. Six is a small sample chosen for accessibility rather than at random, and a stone room of roughly human scale will tend to produce a resonance in that region whether or not anyone planned it.

What would settle it: measure thirty or more chambers chosen at random, and model the measured geometries to see how much of the band falls out of the dimensions alone. If the clustering survives both tests, it is a real result about how these rooms were built. If it does not, it was selection, and this archive will say so.

---

Claims carried below the line

• 528 Hz repairs DNA. No controlled study measures DNA repair at any audio frequency. The hexagonal figure a plate makes near 528 Hz is real plate physics and bears on nothing else. The number itself comes from a 1974 digit reduction published in 1999, not from any older source.
• 110 Hz alters brain activity. One laboratory, thirty subjects, a humanities-adjacent journal in 2008, and no independent replication since. It is the study most often used to argue the chambers were built deliberately, and it is the weakest link in that argument.
• Rosslyn Chapel's ceiling cubes encode a score in Chladni figures. An 1860s restoration replaced a large number of the cubes, and lithographs from 1837 show cubes already missing beforehand. The surface being read is substantially Victorian.
• "The universe is not made of atoms, it is made of vibration" — Pythagoras. He did not say it. What he did was find that consonant intervals correspond to whole-number ratios of a string's length, which is the first recorded discovery that a perception obeys arithmetic — a stronger claim than the invented quote.

---

Sources

Source	Type	What it supports
Chladni, Entdeckungen über die Theorie des Klanges (1787)	Primary	The figures, and the method
Kirchhoff (1850); Leissa, Vibration of Plates, NASA SP-160	Academic	The plate equation and the tabulated eigenvalues the Library's solver is validated against
Jahn, Devereux & Ibison, J. Acoust. Soc. Am. 99(2), Feb 1996, p. 649	Journal	Six structures, 95–120 Hz
Watson & Keating, Antiquity 73:280, June 1999	Journal	Standing waves at Camster Round
Declercq et al., J. Acoust. Soc. Am. (2004)	Journal	The El Castillo staircase chirp as Bragg scattering
Katz et al., Notre-Dame acoustic campaign (2015)	Institutional	~6 s reverberation; the measurements now guiding the restoration
Cook et al., Time and Mind (2008)	Academic	The 110 Hz EEG pilot, carried at HYPOTHESIS

---

In the Library

The interactive version is the Cymatics exhibit, where the plate is solved rather than drawn: the eigenvalues come from the free-edge boundary conditions and are checked against Leissa's published table on every run of the validation script, so the figure and the frequency cannot disagree. The chamber tab runs the same problem in three dimensions against dimensions you choose.

Note, October 2026: until this rebuild, the exhibit printed a frequency beside every figure that had no arithmetic relationship to it — the mode numbers were typed in per preset and were wrong by factors between seventeen and two hundred. The fault is recorded on the exhibit rather than quietly repaired.