33.58 turns the proton magnetic moment into a tightening court: with a stable single-particle readout, frozen geometry, and a controllable gradient G whose sign and magnitude are blinded and interleaved, μₚ or gₚ must show a reproducible non-zero linear slope k inside a preregistered small-gradient window, flip sign when G reverses, cancel in sum between +G and −G within uncertainties, weaken strongly in a low-gradient surrogate, disappear in a near-zero-gradient null, and survive dual reversal routes and switching-artifact audits; under V02/V08/V09-compatible tighten, this remains only one proton magnetic-moment gradient lab-boundary ledger rather than a proton micro-ontology or fundamental-force verdict.
Use this section as a compact machine-readable EFT reference.
33.58 turns proton magnetic-moment readout into a gradient court. The admissible claim is not that any slow drift in μₚ or gₚ proves a new ontology, but that a controlled reversible gradient can produce one first-order, odd-symmetry, linear micro-drift under a fully frozen laboratory aperture.
Measurement centers on μₚ or gₚ extracted from a frequency-ratio readout, the signed magnitude of gradient G under a unified geometric model, the fitted slope k and its confidence interval inside a preregistered linear window, and the residual structure that reveals whether quadratic terms dominate. Odd symmetry is then scored by whether the drifts at +G and −G cancel in sum within uncertainties.
The workflow locks the laboratory scene before any verdict. A stable single-particle apparatus keeps the magnetic field and electrode conditions fixed, the gradient is generated by a reversible configuration with multiple magnitudes and both directions, geometry and rejection rules are frozen before acquisition, labels stay blinded during extraction, and interleaved sequences such as +G, −G, 0, +G, −G suppress slow drift under a fixed settling window.
Controls must prove that the court can see real sensitivity and reject apparatus mimicry. A known computable frequency bias verifies the extraction chain, a geometry-matched low-gradient surrogate should strongly weaken the slope, a near-zero-gradient run must erase any linear relation to pseudo-gradient labels, and two independent reversal routes must lead to the same odd-symmetry conclusion if the effect is genuine.
Support requires a stable non-zero linear slope in the preregistered window, reliable odd symmetry under gradient reversal, and clean separation from the low-gradient and near-zero controls across batches or days. Falsification follows from a zero slope, failed odd symmetry, no stable linear window, or controls that produce comparable response. The named adversaries are switching-induced electromagnetic perturbations, thermal or mechanical stress, and extraction bias or blinding leakage.
So the chapter closes only one tightened proton magnetic-moment gradient ledger aligned with the V08/V09 laboratory audit lane. It may not become a proton micro-ontology, a gradient ontology, or a force-law verdict. Its clean onward value is to hand a reversible gradient court to 33.59, 33.60, and 33.62.