The Fine Structure Constant and the Geometry of the Poincaré Homology Sphere

This paper derives that the fine structure constant α ≈ 1/137.036 arises as a canonical invariant of the Poincaré homology sphere S³/2I (where 2I is the binary icosahedral group) through the algebraic geometry of the E₈ singularity. We demonstrate that every structural element of the golden-ratio formula 1/α = 360φ⁻² − 2φ⁻³ + (3φ)⁻⁵ — its coefficients, exponents, signs, and the appearance of φ — is determined by the McKay correspondence linking icosahedral symmetry to E₈. The leading coefficient 360 = V×E is uniquely characterized as the Regularized Spectral Width of 2I via the Two‑Lock Theorem, and in the icosahedral case this spectral invariant matches the fundamental invariant-degree data {12,20,30}. The exponents {2,3,5} coincide with the E₈ Brieskorn singularity x² + y³ + z⁵ = 0, with degree consistency 2×30 = 3×20 = 5×12 = 60. The golden ratio enters through a Molien transparency: at t = φ⁻¹, det(I − g·φ⁻¹) = 1 exactly on one pentagonal conjugacy class (the 2π/5 class), uniquely selecting φ as the evaluation point. We further exhibit a "golden cancellation": the spectral width factorizes as 𝒲(2I) = 36·(4+2φ)(3−φ) = 360, with all φ‑dependence cancelling algebraically — a property exclusive to the icosahedral case, as tetrahedral and octahedral controls fail with large discrepancies. The product V·E·χ = 720 encodes the SU(2) spinor rotation cycle, with χ = 2 measuring the topological cost of the double cover. Finally, we establish a structural obstruction (Meta‑Theorem 10.4) showing that 14 standard mathematical frameworks — including spectral geometry, index theory, and TQFT — cannot produce the required three‑term polynomial type, and we define a stratified orbifold functional that reproduces 1/α from canonical data with zero adjustable parameters. Historically, the two‑term golden‑ratio approximation was identified by Heyrovská and extended to three terms by Sherbon. The present work provides evidence that this structure is not numerical coincidence but a consequence of the E₈/icosahedral correspondence.