Listen — the corpus as a scale

Every hyperbolic knot has an intrinsic tone. The lengths of the closed geodesics in the knot complement form a true invariant — independent of how the knot is drawn — and we map them to pitch: the shortest geodesic (the systole) becomes the fundamental, the longer ones its overtones (f = 150 × length Hz, an absolute scale, no per-knot normalization). So two different knots genuinely sound at different pitches. Below are the 110 distinct intrinsic tones of the corpus, sorted low to high — spanning 25–389 Hz (about 4.0 octaves).

A finding: the pitch tracks the hyperbolic volume, not the crossing number — the geometry, not the combinatorics, decides what a knot sings. Mirror images and machine re-parametrizations share a tone, so they appear once. (Torus knots and connected sums have no length-spectrum tone and are not listed.) See Method for how the tones are made.

— or click tones to build a
25 Hz
T_13_3_2_6
25 Hz
T_11_9_2_6
26 Hz
pz_3_7_11
26 Hz
T_13_7_2_6
26 Hz
T_13_7_2_8
27 Hz
pz_3_9_11
27 Hz
T_9_7_2_6
29 Hz
T_13_5_2_6
30 Hz
T_11_9_2_8
30 Hz
T_7_5_2_6
31 Hz
T_11_9_2_5
31 Hz
pz_5_7_11
31 Hz
T_13_9_2_5
32 Hz
sd_p3_l11
34 Hz
T_13_8_2_5
40 Hz
T_11_3_2_9
40 Hz
T_11_3_2_8
40 Hz
T_11_3_2_7
41 Hz
T_11_3_2_6
42 Hz
T_7_5_2_5
44 Hz
pz_5_7_9
46 Hz
T_9_7_2_10
46 Hz
T_13_8_2_4
46 Hz
T_9_7_2_9
46 Hz
T_9_7_2_8
47 Hz
T_9_7_2_7
60 Hz
pz_3_7_13
65 Hz
T_11_5_2_8
65 Hz
T_11_5_2_6
65 Hz
T_9_5_2_7
65 Hz
T_9_5_2_8
66 Hz
T_9_5_2_9
69 Hz
wz_sqrt31
72 Hz
sd_p3_l7
101 Hz
pz_3_5_13
165 Hz
kf329
227 Hz
spiral
246 Hz
ping-pong
252 Hz
sundown
278 Hz
double-6
326 Hz
kf370
334 Hz
kf365
339 Hz
cosmic-love
351 Hz
double-3
384 Hz
nitro

Recordings

These are synthesised from the knots themselves — additive synthesis from their overtones. No sampled instrument, no MIDI. The melodies are in the public domain — O’Carolan (d. 1738), Beethoven (d. 1827), traditional tunes — with the exception of the two covers at the end, which are credited to their authors.

The scale above, as one recording

The same 110 tones you can click on, played in order from the lowest to the highest. It was not restored from an old file — it is generated from exactly the data shown above, so what you hear is tile for tile what you see (bau_tonleiter.py, three guards: lowest partial is the fundamental · ascending · every fundamental measured back out of the finished audio).

The corpus, low to high

Two sculptures, one knot

Trainstation stands in steel; Sundown is a drawing. The pipeline finds them to be the same knot — and because the tone comes from the geometry, not from the drawing, the two recordings came out byte-for-byte identical. Nothing was matched by hand. Do not take our word for it: both lossless originals are one click away (wav), so you can run a checksum over them yourself.

Trainstation invariant wav
Sundown invariant wav

Each sculpture in its own voice

One tune — Sí Bheag Sí Mhór by Turlough O'Carolan (d. 1738) — played on each sculpture. The melody is the same throughout; what changes is the timbre, and that is the knot itself.

Trainstation invariant
Nitro invariant
Ping Pong invariant
Cosmic Love invariant
Double 3 invariant
Double 6 invariant
Spiral invariant
Proxima B invariant

The corpus as an orchestra

Every voice is a different knot; no sampled instrument is involved anywhere. The House of the Rising Sun (traditional, author unknown) and Beethoven's Ode to Joy (d. 1827) are scored for nine knots each, all nine carrying a true length-spectrum tone.

The House of the Rising Sun — nine knots
Ode to Joy — nine knots
Short piece — 62 entries, 46 knots

What is the knot, and what is only the drawing

An honest pair. The same sculpture, twice: once from its length spectrum — a true invariant, the same for every drawing of this knot — and once from the Fourier coefficients of this particular centre line, which is the sound of this curve, not of the knot. Both are ours; only the first one is an invariant.

Trainstation — from the invariant
Trainstation — from this curve
… and the same tune on the invariant
Trainstation — the bare spectrum
Nitro — the bare spectrum
Trainstation — the overtones alone
Celtic Trefoil — the overtones alone
Knots of Love — the overtones alone
Nitro — full voice
Ping Pong — full voice

Two covers, played the same way

Not everything here is ours. These two are cover versions — the melodies belong to their authors: Nothing Else Matters (Hetfield/Ulrich, 1991) and Road to Nowhere (David Byrne, 1985). Only the sound is ours: the same additive synthesis from knot overtones, no sampled instrument. They are early attempts from July and we have not re-checked their tuning.

Nothing Else Matters — Hetfield/Ulrich, 1991
Road to Nowhere — David Byrne, 1985

Tones derived from the hyperbolic length spectrum · an absolute, parametrization-independent invariant. Knot data CC BY-NC 4.0 · W.A.Stanggassinger in collaboration with Claude Code (Anthropic).