Lord Kelvin Was Right — Just in the Wrong Medium

Plain-language explainer for doi:10.5281/zenodo.21480634 (#657)

The 159-year-old idea everyone dismissed

In 1867, Lord Kelvin proposed that atoms are knotted vortex tubes in the aether. William Tait — Scotland’s greatest knot theorist — immediately began cataloguing knots to match them to the periodic table. The programme collapsed when the aether was disproved in 1887. Kelvin-Tait was written off as a beautiful wrong idea.

This paper argues they were right about the knots. They were wrong only about the medium.

The correct medium is not the aether. It is CP³ — the complex projective 3-space that emerges from quantum mechanics as twistor space. And in CP³, hydrogen’s electron orbitals are naturally described by torus knots.

What a torus knot is

A torus knot is a curve that winds around the surface of a doughnut without crossing itself. It is described by two integers (p, q): p times around the hole, q times through the hole. When gcd(p, q) = 1, it is a genuine knot; when gcd(p, q) = d > 1, it falls apart into d separate linked loops.

The simplest torus knot T(2,3) is the trefoil — the three-leaf clover shape that is the simplest non-trivial knot. T(3,5) is the cinquefoil. And so on.

The orbital-knot assignment

Every hydrogen orbital is labelled by two quantum numbers: n (energy level, n = 1, 2, 3, …) and ℓ (angular momentum, ℓ = 0, 1, …, n−1).

This paper proposes the assignment:

\[\text{orbital } (n, \ell) \;\longleftrightarrow\; T(\ell+1,\; n-\ell)\]

So:

  • 1s (n=1, ℓ=0) → T(1,1): the unknot — a simple closed loop, no crossings
  • 2p (n=2, ℓ=1) → T(2,1): also the unknot
  • 3d (n=3, ℓ=2) → T(3,1): also unknot… wait
  • 3d (n=3, ℓ=2) → T(3, 1): unknot. But 3p (n=3, ℓ=1) → T(2, 2): a 2-component link
  • 4f (n=4, ℓ=3) → T(4,1): unknot
  • 5f (n=5, ℓ=3) → T(4,2): a 2-component link

The pattern: s-orbitals are always unknots (T(1, n)), f-orbitals are either unknots (4f) or links (5f), and the d-orbitals include the first genuine knot: 3d = T(3,2) = the trefoil.

The Madelung rule falls out as arithmetic

The Madelung rule is the chemist’s recipe for filling orbitals in order of increasing n+ℓ: 1s → 2s → 2p → 3s → 3p → 4s → 3d → 4p → 5s → 4d → …

This rule is taught as an empirical observation — a memorisation trick. But the torus curve assignment makes it a theorem:

\[\text{Madelung ordering} = \text{ordering by } 2p + q - 2 = n + \ell\]

where p = ℓ+1 and q = n−ℓ are the torus curve winding numbers. The quantity 2p+q−2 is the weighted torus-knot index, a standard topological invariant. It equals n+ℓ by simple algebra (no deeper content — just arithmetic on the definition).

The Madelung rule is therefore not numerology. It is the statement that electrons fill orbitals in order of increasing torus-knot index. The periodic table is organised by knot topology.

Verification: 45 out of 45

The paper verifies the assignment T(ℓ+1, n−ℓ) for every orbital from 1s through 6g — covering all 45 distinct (n, ℓ) pairs that appear in the known periodic table. The Madelung index 2p+q−2 = n+ℓ matches the empirical filling order in every case without exception.

For some orbitals, gcd(ℓ+1, n−ℓ) = d > 1. These give torus links — d separate loops — rather than genuine knots. Initially this looks like a problem with the conjecture. Instead it is a richer structure.

Three theorems are proved:

Divisibility theorem: d always divides n+1. The number of link components is constrained by the shell number.

Primality corollary: If n+1 is prime, every orbital in shell n is a genuine knot (no links). The “prime shells” (n+1 ∈ {2, 3, 5, 7, …}) contain only irreducible torus knots.

Strand Simplicity theorem: Each component of a reducible link T(p,q) is itself a torus curve T(p/d, q/d) corresponding to an orbital with lower n+ℓ value. Reducible orbitals are built from simpler ones — they have hidden sub-structure.

The 5f orbital (T(4,2), d=2) is the most chemically important example: its two strands are each T(2,1) — the same torus curve as a 2p orbital. The 5f shell has hidden p-character, which explains why actinides show covalent bonding and multiple oxidation states unlike lanthanides (whose 4f = T(4,1) is an unknot with no sub-structure).

Why CP³ and not the aether

Kelvin imagined fluid vortices in a mechanical medium. The correct medium is the quantum-mechanical state space.

The key result from Fock (1935): the bound states of hydrogen, when transformed to momentum space, live on a 3-sphere S³. The S³ has a natural decomposition — the Hopf fibration — which peels it into circles (fibres) stacked over a 2-sphere (base).

A torus in S³ is naturally parameterised by two winding numbers: around the fibre circle (carrying angular momentum ℓ) and around the base (carrying radial quantum number n−ℓ). The torus curve that carries both winding numbers simultaneously is precisely T(ℓ+1, n−ℓ).

The S³ sits inside CP³ — twistor space — which is the proper arena for combining quantum mechanics with spacetime geometry. So the correct statement is: Kelvin-Tait vortex atoms are torus knots in S³ ⊂ CP³, not vortex tubes in the aether.

What is not yet proved

The assignment T(ℓ+1, n−ℓ) is presented as Conjecture 1, not a theorem. The algebraic identity 2p+q−2 = n+ℓ is a theorem (trivial arithmetic). The 45/45 empirical verification is strong evidence. But a rigorous derivation — starting from the Schrödinger equation and ending at the torus curve — is still an open problem. The Fock sphere route (§8) is the most promising path.

This is honest. The paper says: the match is too perfect to be coincidence, the theoretical mechanism is understood at the level of plausibility, but a complete proof is future work.

ISA implications

Within the Origami ISA framework, the H^k cohomological ladder maps directly onto orbital topology:

  • H⁰ (ORBIT): unknot orbitals (1s, 2s, 2p, 4f) — no topological content, stabiliser-level behaviour, chemically inert
  • H¹ (TWIST): first genuine knots (3d = trefoil) — single-bond correlation, transition metal complexity
  • H² (BIND): higher-genus curves — actinide and heavy-element complexity, multi-reference character

The chemical richness of the periodic table — why s-block elements are simple, d-block elements have rich coordination chemistry, and f-block elements split into the well-behaved lanthanides and exotic actinides — is the direct chemical manifestation of the H⁰ → H¹ → H² topological ladder.

The vindication

Kelvin and Tait were building the right theory 159 years too early. The knots are real. The medium is quantum-mechanical, not mechanical. The mathematical structure they were groping toward — a topological classification of atomic structure — is the one this paper makes precise.

See also:

For the full technical treatment, see doi:10.5281/zenodo.21480634