Luister naar de muziek van het universum hier
Het artikel onderzoekt een vraag uit de Vacuum.Net-theorie: welk deel van de punten op een driehoeksrooster blijft eindeloos “draaien” in plaats van naar nul te gaan.
Met een computerassistentie bewijst de auteur dat dit aandeel geen limiet heeft, maar schommelt tussen ongeveer 29,6% en 32,0%.
De schommeling herhaalt zich exact telkens als de schaal met een factor 3 toeneemt.
Die factor 3 is de tritave uit de Bohlen–Pierce-muziekschaal, en de schommeling heeft de structuur van een Shepard–Risset-toon.
De boventonen van die schommeling leveren een nieuw soort “timbre” op dat muzikaal bespeelbaar is.
Zo vormt het artikel een nog niet eerder beschreven brug tussen getaltheorie en muziek.

Take a point of the triangular lattice. Subtract a digit −1, 0 or +1 and divide by 1 − ω, where ω is a cube root of one. Repeat. Some points reach 0. The rest end in one of two points that turn forever. The Vacuum.Net theory asked what share of points turns. We show that the share has no limit. It swings between about 29.6 % and 32.0 %. The swing repeats exactly each time the size grows by a factor 3. We prove this with a computer-assisted argument. We measure the shape of the swing and split it into overtones. The repeat ratio 3:1 is the tritave of the Bohlen–Pierce scale. The swing has the structure of a Shepard–Risset tone. Its overtones give a timbre that can be played. The essay argues that this is a meeting point of number theory and music that has not been described before. It treats the musical side at length.
1. Why this question matters
Mathematics often advances until an operation can no longer be undone, then steps back and adds a new object to cross the barrier. Division by zero, irrational numbers, imaginary numbers, non-Euclidean geometries: each was once a wall and became a door. A companion essay, Barriers and Back-Steps, lists fourteen such barriers and predicts that the next answers will often be rhythms rather than single numbers [3]. This essay provides a concrete, proved case of that prediction.
The question at stake is small enough to state in one breath. The answer is not a number. It is a rhythm, and the rhythm can be heard.
The essay has two halves. Sections 2 to 6 are mathematics: the question, the measurement, the proof, and the spectrum. Sections 7 to 12 are music: ratios and the comma, the tritave, endless tones, spectral composition, chord geometry, and what is genuinely new. Section 13 states frankly what has been proved and what has not.
2. The Vacuum.Net theory in brief
The Vacuum.Net theory begins from movement rather than numbers [1]. There is one strand. It moves and returns. Every distinction is a loop of this strand. Its second axiom, N2 (“closure”), says that a loop persists only if it closes; what does not close passes on as tension.
Each elementary step of the strand is a trit: forward (+1), standing still (0), or backward (−1). Trits combine into ladders. Two ladders are basic: doubling (×2) and tripling (×3). Powers of 2 and powers of 3 never meet exactly. The theory calls their near-meetings closure depths; they run 1, 2, 5, 12, 41, 53, 306, 665, and so on. At depth 12 the miss is the Pythagorean comma, 3¹²/2¹⁹ ≈ 1.0136, an object musicians have known for over two thousand years. Section 7 returns to this.
A companion book derives a full mathematics from these rules in 22 rungs [2]. Rung 10 places the trits in the plane, where they become the Eisenstein integers z = a + bω, with ω = (−1 + i√3)/2. These form the triangular lattice. The base of the number system is β = 1 − ω. Dividing by β shrinks distances by |β| = √3 and rotates the plane by 30 degrees.
That book left one question open, labelled IX.3: does the share of points that keep turning settle to a fixed value? This essay answers it.
3. The number system
For background on why number systems work this way: in ordinary base 10, dividing by 10 and taking a remainder digit is the familiar expansion algorithm; here exactly the same idea is run with the complex number β on the triangular lattice.
Every Eisenstein integer z leaves a remainder −1, 0 or +1 after division by β. Call this the digit d. The step function is
T(z) = (z − d) / β.
A worked example. Start with z = 2. Its digit is −1, so T(2) = (2 + 1)/β = 3/β = 2 + ω. Now 2 + ω has digit 0, so T(2 + ω) = (2 + ω)/β = 1 + ω. But 1 + ω maps to itself: it is a fixed point that turns forever. (Indeed, (1 + ω) − 1 = ω = β(1 + ω), so subtracting the digit returns exactly the same point.)
Every orbit therefore ends in one of three places: 0, the point 1 + ω, or the point −(1 + ω). A point that reaches 0 has a finite expansion
z = d₀ + d₁β + d₂β² + … + dₖβᵏ,
and we call it finite. The others we call turning. By the symmetry z ↦ −z, exactly half of the turning points flow to each of the two fixed points.
Is this system known? Yes, in part. It is a symmetric canonical number system in the sense of Akiyama and Scheicher [4]. Their Theorem 2.2 decides, for quadratic bases x² + Ax + B, whether every integer has a finite expansion: it requires |A| < 1 + B/2 with |B| ≥ 2, or A = 1 + B/2 with |B| > 2. Our base satisfies x² − 3x + 3 = 0, so A = −3 and B = 3, and neither condition holds. That turning points exist is therefore already known. The wider literature on shift radix systems studies which individual orbits are finite or periodic [5][6]. What was not studied, as far as our search reaches, is the share of finite points — and no one reported that this share fails to settle.
4. The measurement
Two independent programs were written [18]:
- Backward program. Start from each point and apply T repeatedly until the orbit ends.
- Forward program. Enumerate all finite expansions with up to 16 digits — 3¹⁶ = 43,046,721 of them, all distinct — and count how many lie inside a disc.
Both use exact integer arithmetic. The 15-digit and 16-digit lists agree in all counts up to norm 1.1 million, so the counts are complete there. The two programs agree to four decimals.
The norm of a + bω is N(z) = a² − ab + b², the squared distance from 0. The table shows the share of turning points in discs of growing norm:
| Norm N | Share turning | Note |
|---|---|---|
| 72,900 | 0.3076 | low phase |
| 126,266 | 0.3102 | high phase |
| 218,700 | 0.3076 | N × 3: same value |
| 378,799 | 0.3102 | N × 3: same value |
| 656,100 | 0.3076 | N × 3: same value |
| 1,136,398 | 0.3102 | N × 3: same value |
Table 1. Share of points that never reach 0, in discs of norm N. Multiplying the norm by 3 returns the same value.
The share does not drift toward a limit. It moves up and down. Each factor 3 in the norm brings it back to where it was. On a fine grid it ranges from 29.6 % to 32.0 %. Counting hexagons instead of discs gives different absolute values but the same swing. Figure 1 of the original essay shows one full period measured three times, with the three curves lying exactly on top of each other.
The exact repetition under ×3 is the central observed fact. Everything that follows builds on it.
5. The proof
The argument has four steps and uses one external theorem and one computer certificate.
5.1 The tile
Let F be the set of all infinite sums d₁β⁻¹ + d₂β⁻² + … with digits −1, 0, +1. This is a compact fractal set, the tile. It obeys the self-similarity equation
βF = (F − 1) ∪ F ∪ (F + 1).
Let U be the set of finite points, and let W be the union of the translated tiles u + F over all u in U.
5.2 W is self-similar
For u in U and any digit d, the point βu + d is again finite, because T maps it back to u; and every finite point arises this way. Combining this with the tile equation gives
βW = W.
So W looks exactly the same after scaling by √3 and rotating by 30 degrees. Let G(R) be the fraction of a disc of radius R that W covers. Discs do not care about rotation, hence
G(√3 · R) = G(R) exactly.
This is the rigorous form of the observed period 3 (in norm, R² scales by 3).
5.3 Counting equals area
Lagarias and Wang proved that tiles of this kind tile the plane by a lattice [7]: for an irreducible base polynomial and a complete set of digit residues, the tiling lattice is the smallest invariant one containing the digit differences [8]. Here that is the Eisenstein lattice itself. So the tiles u + F do not overlap except on a set of zero area, and each has the area of one lattice cell, √3/2.
Every point of F lies within ρ = 1/(√3 − 1) ≈ 1.366 of 0. Let A(r) be the number of finite points in a disc of radius r. Then the area of W inside a disc of radius R lies between (√3/2)·A(R − ρ) and (√3/2)·A(R + ρ). As R grows, the share of finite points therefore differs from G(R) by at most a constant times 1/R — a vanishing boundary error.
5.4 The certificate
It remains to show that G is not constant. We bound G at two radii with exact counts, and the derivation that the counts are complete is short: every finite point within radius R + ρ needs at most 15 digits, because one step reduces |z| to at most (|z| + 1)/√3, so after 13 steps the point lies within √3 of 0; there only 0 and the six units remain, four of which reach 0 within two steps, while the other two are the turning points. Hence the 16-digit list is complete for the radii used.
| Radius R | Lower bound of G(R) | Upper bound of G(R) |
|---|---|---|
| 389.73 (high phase of turning) | 0.67540 | 0.68513 |
| 321.57 (low phase of turning) | 0.69797 | 0.70998 |
Table 2. Certified bounds on the covered fraction G at two radii. The intervals do not overlap: the margin is 0.0128. Two further pairs of radii give margins 0.0130 and 0.0116.
The upper bound at the first radius sits below the lower bound at the second. So G is not constant. Since G repeats exactly under scaling by √3, and the share of finite points differs from G only by O(1/R), the share takes values at most 0.6851 and at least 0.6980 infinitely often.
Theorem. For the base 1 − ω with digits −1, 0, +1, the share of Eisenstein integers in a disc that never reach 0 has no limit. Its lower limit is at most 0.302 and its upper limit at least 0.315.
Status. This is a computer-assisted proof. It relies on the Lagarias–Wang tiling theorem and on exact integer counts. The numbers in Table 2 can be recomputed with the published scripts [18]. An independent check by a second party would be welcome.
6. The spectrum
Because the swing repeats exactly whenever the norm grows by a factor 3, it can be decomposed into a mean and overtones, exactly like a periodic sound. We sampled one period at 243 points, repeated over three periods, and took the Fourier transform:
| Component | Amplitude | Phase |
|---|---|---|
| Mean (fundamental level) | 30.87 % | — |
| Overtone 1 | 0.90 % | 23.9° |
| Overtone 2 | 0.38 % | 131.0° |
| Overtone 3 | 0.12 % | −4.7° |
| Overtone 4 | 0.03 % | 148.6° |
| Overtone 5 | 0.05 % | −46.8° |
| Overtones 6–8 | < 0.03 % each | — |
Table 3. Decomposition of the swing. The values agree over three periods to 0.00001 in amplitude.
The mean also has meaning of its own: weighted by 1/|z|², the finite points have a logarithmic density, computed directly as 0.3089 turning. So a mean exists even though a plain limit does not — a distinction worth pausing on.
Lapidus and van Frankenhuijsen built a theory for precisely this kind of oscillation [9]. In their language, a fractal has complex dimensions whose imaginary parts set the frequencies of its oscillations; one can literally “hear the shape of a fractal” [10]. Here the complex dimensions lie at 1 + 2πik/ln 3 in the norm, for k = 1, 2, 3, …. Table 3 gives their strengths. In plain words: the trit tile has a fundamental and overtones.
7. Music I: the two ladders and the comma
Music is the oldest applied mathematics of ratios. Two strings sound well together when their lengths have a simple ratio: the octave is 2:1 and the fifth is 3:2. These are exactly the two ladders of the Vacuum.Net theory: doubling and tripling.
A circle of twelve fifths almost returns to the start, overshooting seven octaves by the Pythagorean comma, about 23.5 cents. The theory proves that the miss never vanishes [1]: no power of 3 equals a power of 2. In the language of the net, this is a closure that fails; the loop does not close, and what does not close passes on as tension. Musicians met this barrier two thousand years before number theorists named it, and their resolutions follow the same pattern as the barriers of mathematics themselves: step back, add a new object, pay with a lost property.
| Barrier in music | Resolution | Property lost | Link to the net |
|---|---|---|---|
| Twelve fifths do not close on the octave (the comma) | Temperament: spread the comma | Pure fifths | Closure fails (T16–T18) |
| Spreading the comma evenly requires 2^(1/12) | Equal temperament (Zhu Zaiyu 1584; Stevin) | Rational ratios | Closure depth 12 |
| Twelve steps give poor thirds | Finer divisions: 41 and 53 steps (Jing Fang; Mercator; Holder) | Simplicity of the keyboard | Closure depths 41 and 53 |
| The octave itself as frame | Tritave 3:1 (Bohlen–Pierce, 1970s) | Octave equivalence | The ×3 ladder alone |
| Pitch as a fixed height | Circular pitch (Shepard 1964; Risset) | A top and a bottom | Log-periodicity |
| The note as the unit | The spectrum as material (Grisey, Murail, 1970s) | The note | Spectrum of the tile |
| Chords as lists | Chords as points of an orbifold (Tymoczko 2006) | Flat chord space | The node is an orbifold |
Table 4. Barriers in the history of tuning and composition, read with the barrier pattern of [3]. The last column is a reading, not a theorem.
The closure depths of the theory — 12, 41, 53 — are exactly the historically important equal divisions: twelve-tone equal temperament was computed by Zhu Zaiyu in 1584 and independently by Simon Stevin [15]; the 53-step division goes back to Jing Fang, was quantified by Mercator, and was shown by Holder in 1694 to give good thirds as well [16]. This correspondence is not new — it is simply the continued fraction of log₂3 — but it shows that the net and music share their first barrier.
8. Music II: the tritave
Most music repeats at the octave: a note and its double are heard as “the same note, higher.” The Bohlen–Pierce scale breaks this habit. It repeats at 3:1 — an interval called the tritave, equal to an octave plus a fifth — and divides it into 13 equal steps of ratio 3^(1/13) [11]. It was discovered independently by Heinz Bohlen, Kees van Prooijen and John Pierce, and instruments exist for it, including a family of clarinets.
The trit tile repeats at 3:1 in the norm. One extra trit digit multiplies the norm by 3. So the natural period of the tile is a tritave, not an octave — a precise fact, not an image: Table 1 shows it. The ×3 ladder of the net is the backbone of the tritave, just as the ×2 ladder is the backbone of the octave.
The reading is this: the octave belongs to the binary ladder; the tritave belongs to the trit ladder. A music built on trits would be a tritave music, and the Bohlen–Pierce scale is such a music. It was found by ear and by acoustics, not by number theory. The tile now gives it a second, arithmetical foundation.
9. Music III: endless tones
In 1964 Roger Shepard built tones from partials spaced one octave apart, keeping the overall loudness envelope fixed while shifting the partials upward [12]. Listeners heard a scale that rises forever but never gets higher. Jean-Claude Risset made a continuous glide with the same effect. The trick works because the sound is identical after an octave of shift: mathematically, it is a function repeating after a fixed ratio — a log-periodic function.
The share of turning points is exactly such a function. It keeps changing as the disc grows; after one tritave it is back where it started; it never settles and never runs away. The tile behaves like a Shepard–Risset tone built in the tritave instead of the octave.
There is one profound difference. Shepard designed his envelope by hand. The envelope of the tile is not designed: it follows necessarily from the arithmetic of 1 − ω and the digits −1, 0, +1. Figure 1 of the original essay is that envelope; its asymmetric shape, a slow fall and a fast rise, comes from the boundary of the tile near 0.
10. Music IV: an arithmetical timbre
In the 1970s, Gérard Grisey and Tristan Murail made the spectrum of a sound into the material of composition — spectral music. Grisey’s Partiels (1975) begins from the measured spectrum of a low trombone note and lets an ensemble rebuild it [13]. Its material came from acoustics: instruments, bells, voices.
Table 3 is a spectrum that comes from arithmetic. Its overtones fall roughly as 1 : 0.42 : 0.13 : 0.04 : 0.05, giving a tone that is soft and round with a clear second harmonic. Three short sound files were produced (v1.0) [18]:
- Tile timbre. One period of the swing is used as the waveform of a 220 Hz tone. What you hear is the trit tile itself.
- Bohlen–Pierce ladder. The thirteen steps of one tritave, played with the tile timbre.
- Risset glide in the tritave. Pure partials spaced 3:1 under a fixed envelope, shifted endlessly upward; it rises without end and returns after each tritave.
This offers spectral music a new source of material. Any number system that fails to terminate produces a log-periodic share like ours, and each has its own spectrum. A composer could take these spectra as material, as Grisey took the trombone. The prediction is testable: different bases give different timbres, and both the computation and the audition are within reach.
11. Music V: orbifolds
Dmitri Tymoczko showed that chords live in curved spaces called orbifolds: a chord is a point, a chord progression is a path, and the space folds onto itself because octaves and the order of notes do not matter [14]. For two-note chords the space is a Möbius strip whose edges act as mirrors. His paper was the first on music theory ever published in Science.
The node of the Vacuum.Net theory (Rung 11) is also an orbifold: the modular group acting on the upper half-plane, with two special points of order 2 and order 3 where the space folds — the two turns of the node [2]. Tymoczko’s chord spaces fold by permutation and reflection; the node folds by the turns of order 2 and 3.
The trit tile adds a third element to this family. Tymoczko’s spaces describe where chords are; Shepard’s tones describe how pitch can circle; the tile describes how a quantity circles under scaling by a fixed ratio. A reading, not a proof: the net contains all three. Its orbifold is the node. Its circle is the tritave. Its timbre is the spectrum of the tile.
12. What is new
We claim the following, and no more:
- New (as far as our literature search reaches): for the base 1 − ω with digits −1, 0, +1, the share of non-terminating points has no limit; it repeats with period 3 in the norm. This is proved with a computer certificate, and its mean and overtones are measured.
- Known: that the system does not terminate everywhere (Akiyama–Scheicher); that the tile tiles the plane (Lagarias–Wang); the theory of complex dimensions (Lapidus); the tritave, Shepard tones, spectral music, and chord orbifolds.
- New as a connection: the failure of a number system to terminate produces a log-periodic share whose period is the base ratio — here, the tritave — and that share’s spectrum is an arithmetical timbre that can be played. We have not found this connection described before.
Why this may matter. Music theory studied ratios, circles of pitch, and spectra separately. Number theory studied which expansions terminate. Here all three musical themes appear inside one number-theoretic object — the plane of Rung 10, where the trits become complex numbers. If the companion essay is right that the next mathematics answers with rhythms [3], this is one of its first examples. It is also, literally, a sound.
13. Status and open points
- Proved (computer-assisted): the share has no limit (Section 5). It relies on the Lagarias–Wang tiling theorem and exact counts; independent verification is welcome.
- Measured, not proved: the exact range 29.6–32.0 %, the mean 30.87 %, and the overtone amplitudes.
- Open: a closed formula for the mean and the overtones; the same analysis for other bases; a formal proof in Agda or Lean.
- Reading, not proof: every link to the Vacuum.Net theory and to music in Sections 7–11 beyond the stated facts.
14. Conclusion
A small question from the Vacuum.Net theory has a precise answer. The share of turning points does not settle. It circles with a fixed period of one trit digit, and that period is a tritave. The circle has a mean, a shape, and overtones, and it is proved that the circle is real. Music has known each of these ingredients for decades — the tritave, the endless tone, the spectrum as material, the folded space of chords — but they appear here together inside a single number system. The trit tile has a sound. It can be computed, proved, and played.
Annotated references
[1] Konstapel, J. (2026). Vacuum.Net Theory, Ninth Edition: Foundations 9.2. Constable Research, Leiden.
Source of the axioms N1–N5 and of theorems T16–T18 on the two ladders ×2 and ×3, the closure depths, and the Pythagorean comma. Sections 2 and 8 rest on it.
[2] Konstapel, J. (2026). Mathematical Theory Generated with the Vacuum.Net Theory, version 7.0 (19 November 2026). Constable Research, Leiden.
Rung 10 introduces the Eisenstein plane with base 1 − ω; Rung 11 introduces the node as an orbifold. Open point IX.3 asks whether the share of turning elements settles. This essay answers that question.
[3] Konstapel, J. (2026). Barriers and Back-Steps, version 1.0 (21 November 2026), and The Next Mathematics. constable.blog (21 November 2026).
The companion essays. They argue that the next barrier of mathematics is logical and that “rhythms are answers.” The present essay supplies a proved example of that pattern. Reference [17] (the K-theoretic counterexamples of Burklund, Hahn, Levy and Schlank, arXiv:2310.17459) is the latest barrier in their chain, cited for context only.
[4] Akiyama, S. and Scheicher, K. Symmetric shift radix systems and finite expansions. Preprint, University of Tsukuba (www.math.tsukuba.ac.jp/~akiyama/papers/centralsrs6.pdf).
Defines symmetric canonical number systems with digits centred on 0. Their Theorem 2.2 decides, for quadratic bases, when every integer has a finite expansion; it shows that our base x² − 3x + 3 fails this test, but it gives no density. The starting point of the whole question.
[5] Akiyama, S., Borbély, T., Brunotte, H., Pethő, A. and Thuswaldner, J. M. (2005). Generalized radix representations and dynamical systems I. Acta Mathematica Hungarica 108, 207–238.
Introduces shift radix systems, the common frame for canonical number systems and beta-expansions. The best entry point into the literature on finite expansions.
[6] Kirschenhofer, P., Pethő, A., Surer, P. and Thuswaldner, J. (2010). Finite and periodic orbits of shift radix systems. Journal de Théorie des Nombres de Bordeaux 22, 421–448.
Studies which orbits end in 0 and which become periodic, treating sets of parameters rather than the proportion of starting points — the gap this essay fills.
[7] Lagarias, J. C. and Wang, Y. (1997). Integral self-affine tiles in Rⁿ, part II: lattice tilings. Journal of Fourier Analysis and Applications 3, 83–102.
Proves that every standard digit set yields a self-affine tile that tiles the plane by a lattice. This is the external theorem on which the proof in Section 5 relies.
[8] Steiner, W. and Thuswaldner, J. M. (2012). Rational self-affine tiles. arXiv:1203.0758.
Its introduction states the Lagarias–Wang result in exactly the form used here: for an irreducible characteristic polynomial and a complete residue system of digits, the tile gives a lattice tiling with respect to the smallest invariant lattice containing the digit differences.
[9] Lapidus, M. L. and van Frankenhuijsen, M. (2013). Fractal Geometry, Complex Dimensions and Zeta Functions, 2nd ed. Springer.
The theory of complex dimensions: oscillations in the geometry of a fractal and in its spectrum are described by the same complex numbers. This is the language of Section 6, where the complex dimensions sit at 1 + 2πik/ln 3.
[10] Lapidus, M. L. (2015). The sound of fractal strings and the Riemann hypothesis. arXiv:1505.01548.
An accessible survey explaining how one can “hear” a fractal through its complex dimensions, with connections to the Riemann hypothesis. Recommended first reading for the fractal-oscillation side.
[11] Mathews, M. V., Pierce, J. R., Reeves, A. and Roberts, L. A. (1988). Theoretical and experimental explorations of the Bohlen–Pierce scale. Journal of the Acoustical Society of America 84, 1214–1222. Müller, N.-L. (2019). The Bohlen–Pierce clarinet: exploring a new tonality. The Clarinet 46/3.
The scale that repeats at 3:1 (the tritave) instead of 2:1 (the octave), divided into 13 equal steps. Müller documents the instruments built for it, including the clarinet family.
[12] Shepard, R. N. (1964). Circularity in judgments of relative pitch. Journal of the Acoustical Society of America 36, 2346–2353. Deutsch, D. (2010). The paradox of pitch circularity. Acoustics Today 6(3), 8–14.
Shepard’s tones are built from octave-spaced partials under a fixed spectral envelope; shifting them in log frequency gives the endlessly rising scale. Risset made the continuous glide. Deutsch’s review is the modern overview of the phenomenon.
[13] Cagney, L. (2023). Gérard Grisey and Spectral Music: Composition in the Information Age. Cambridge University Press. Grisey, G. (1975). Partiels, for 18 instruments.
Spectral music takes the measured spectrum of a sound as compositional material; Partiels famously starts from a low trombone note. Section 10 proposes an arithmetical spectrum as a new material of the same kind.
[14] Tymoczko, D. (2006). The geometry of musical chords. Science 313, 72–74. Tymoczko, D. (2011). A Geometry of Music. Oxford University Press.
Chords as points in orbifolds: spaces folded by octave equivalence, permutation, and inversion. The first music-theory article in Science and the basis of the comparison in Section 11.
[15] Rehding, A. Fine-tuning a global history of music theory: divergences, Zhu Zaiyu, and music-theoretical instruments. Music Theory Spectrum. Liu, Y. (2013). Equal temperament circa 1600 in China and Europe. Canadian Mathematical Society, Winter Meeting.
Twelve-tone equal temperament was computed by Zhu Zaiyu in 1584 and, independently, by Simon Stevin shortly after; both sources discuss the mathematics and the priority question.
[16] Barbour, J. M. (1951). Tuning and Temperament: A Historical Survey. Michigan State College Press.
The standard history of tuning, covering Pythagorean tuning, the comma, meantone temperament, equal temperament, and the 53-step division known since Jing Fang and computed by Mercator and Holder.
[17] Burklund, R., Hahn, J., Levy, I. and Schlank, T. M. (2023). K-theoretic counterexamples to Ravenel’s telescope conjecture. arXiv:2310.17459.
The latest barrier in the chain of Reference [3]. Cited for context only; not used technically in this essay.
[18] Konstapel, J. (2026). Research notes “Hogere windingen”, versions 1.8–2.2. Python scripts (controle/*.py) and audio files (1_tile_timbre, 2_bohlen_pierce, 3_risset_tritave), v1.0. Constable Research, Leiden (unpublished).
Contains every computation of Sections 4–6 and the certificate of Section 5, with code that reproduces them, plus the three sound files of Section 10. The place to begin for independent verification.
