We investigate the global spectral architecture of the prime number system through two complementary lenses: the logarithmic phase dynamics of the Riemann zero sequence, and the spectral power of the von Mangoldt arithmetic carrier field at the Riemann zero frequencies. Both analyses are grounded in verified Odlyzko zero tables and subjected to systematic falsification campaigns against local, surrogate, and randomised baselines. The two investigations are shown to be dual manifestations of a single underlying phenomenon. The carrier field E (u) = (ψ (eᵘ) − eᵘ) ·e^−u/2 is the Fourier transform of the zero distribution; the log-phase sequence θₙ = log (γₙ) mod 2π samples that transform at the zero ordinates themselves. Together they constitute a self-dual spectral system. Six principal results are established. (1) Normalised log-phase increments exhibit a sharp transition from strong scale-dependence (KS statistic D = 0. 642 between low and mid-height zero blocks) to asymptotic stability (D = 0. 007 between independent deep blocks). (2) The limiting increment distribution is extremely close to but not identical with the GUE Wigner surmise (KS = 0. 024, Earth Mover Distance = 0. 014), with residual deviations predicted to decay as O (1/log γ). (3) Spectral Form Factor correlations converge monotonically (r = 0. 875 → 0. 902) as blocks ascend the zero spectrum — identified here as arithmetic asymptotic freeness. (4) The arithmetic carrier concentrates power at true zero frequencies with ratio Pₜrue/Pₙull ≥ 5. 1× (p = 0. 0007) against 10, 000 randomised constellations. (5) A robust attenuation law P (γ) ∝ γ^−α with α = 1. 93 ± 0. 03 is identified as a finite-scale approximation to the theoretical γ^−2 decay forced by the explicit formula, with the deviation Δα = 0. 07 constituting a measurable finite-scale correction. (6) In both frameworks, all local statistical tests fail (p ≈ 0. 22–0. 63) while all global spectral tests survive — a unified local/global bifurcation that is the arithmetic analogue of spectral complementarity. Three falsifiable predictions follow: the GUE deviation KS (γ) ∼ C/log (γ) should reach approximately 0. 017 at γ ∼ 10⁷; the attenuation exponent α (xₘax) should converge monotonically toward 2 from below; and the SFF monotone convergence should continue at a rate consistent with 1−r ∼ C/log (γₘin). All three are testable with the extended Odlyzko zero tables.
Timothy Desmond (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: