Where does $\alpha_{\mathrm{EM}} = 1/137$ come from?

The old mystery

$\alpha_{\mathrm{EM}} \approx 1/137.036$ controls the strength of the electromagnetic coupling. It is one of the best-measured constants in physics, and one of the worst-explained. Feynman wrote: “It’s one of the greatest damn mysteries of physics: a magic number that comes to us with no understanding by man.”

PT proposes a computable answer. Not a heuristic, not numerology: an explicit three-factor product.

Step 1 — the bare product

At scale $\mu^* = 15$, three holonomy angles are computed on the $q_+ = 1 - 2/\mu^*$ branch:

(These come from $\sin^2\theta_p = \delta_p (2 - \delta_p)$, theorem T6.)

Their product:

This is bridge axiom BA5 (which is in fact a derived theorem). We are already within 0.5% of the experimental value. Without any fit.

Step 2 — the dressing

$1/136.28$ is not $1/137.036$. The difference comes from the inactive primes $\{11, 13\}$: they do not contribute dynamically as primaries, but they leave an echo polarization that dresses the coupling at very short distance.

The dressing is a cascade of three corrections (R51, monograph chapter 10):

The $\Delta_i$ are loop-order corrections (1, 2, 3) computable from $\gamma_p$ and $\sin^2\theta_p$. We obtain:

Final precision: better than $10^{-4}$ relative to the CODATA value. No parameter fitted at any step.

Why a product, not a sum?

This product structure is not aesthetic. It comes from the Chinese remainder theorem applied to the three circles $\mathbb{Z}/3\mathbb{Z}$, $\mathbb{Z}/5\mathbb{Z}$, $\mathbb{Z}/7\mathbb{Z}$.

These three circles are orthogonal in the cube $\mathbb{T}^3 = \mathbb{Z}/(3 \cdot 5 \cdot 7)\mathbb{Z}$. The total transition amplitude between two states of the cube factorizes into a product of per-circle amplitudes. And $\sin^2\theta_p$ is precisely the transition amplitude around circle $p$.

So $\prod_p \sin^2\theta_p$ is not a heuristic formula: it is the direct application of the Pontryagin principle on the torus $\mathbb{T}^3$. It is forced by the structure.

What about the $q_-$ branch?

$\alpha_{\mathrm{EM}}$ lives on the $q_+$ branch (couplings). If we computed the same product on the $q_- = e^{-1/\mu^*}$ branch (geometry), we would not get $\alpha_{\mathrm{EM}}$ — we would get geometric observables (Newton’s constant, quark masses, CKM mixing). This bifurcation is seeded by the parity operator $p = 2$, then frozen at $\mu^* = 15$; it separates the lepton/vertex sector from the quark/propagator sector.

What is not in the calculation

• No standard QED vacuum polarization. PT replaces loop diagrams with sieve corrections $\Delta_i$.
• No GUT unification: no large broken gauge symmetry needed. Couplings $\alpha_s$, $\alpha_{\mathrm{EM}}$, $\sin^2\theta_W$ come independently from the same cascade, on the same active primes.
• No QFT running: scale dependence is arithmetic (changing $\mu$), not dynamical.

It is a minimal frame producing a maximal number. If the measured $\alpha_{\mathrm{EM}}$ shifted significantly at a different $\mu$, we would know the PT identification has a flaw. As of now, at $\mu^* = 15$, the count is right to $10^{-4}$.

That is the value of the number 137: a product, a dressing, and zero parameter.

← All essays