Error/Failed report on Equation 601

This deposition archives the complete record of a pre-registered Bayesian model comparison test of a parameter-free prediction from Quantum Traction Theory (QTT) regarding a putative 5.9-year modulation in laboratory determinations of Newton's gravitational constant G.

**Outcome.** The pre-registered analysis returned ln B(H_QTT, H_0) = -0.142, well below the pre-registered threshold of +3 for a moderate-claim publication. The pre-registered null branch was therefore triggered, and the corresponding paper draft (titled "A conditional parameter-free prediction of the 5.9-year modulation in laboratory measurements of Newton's gravitational constant from a two-clock cosmological framework") was closed without submission to any peer-reviewed journal.

**Methodology.** The test was conducted on the n=13 Schlamminger 2015 compilation of post-1980 precision-G measurements (Phys. Rev. D 91, 121101(R)) using the IERS EOP 14 C04 length-of-day series as the candidate driver signal. The QTT-locked model (H_QTT) had two free global parameters (an overall amplitude and a single phase lag) regardless of dataset size; the period was fixed at 5.9 years and per-laboratory phases and amplitudes were determined by the framework's geometry rules. The null model (H_0) was constant G with per-method extra Gaussian noise; the alternative (H_1) was a free sinusoid with extra noise. Bayesian evidence was computed by dynamic nested sampling with priors frozen in advance.

**Pre-registration.** The pre-registration commit hash f646db2e90c30587d30149d58e6992ee83bf944e was recorded at 2026-05-02T15:30:18Z, before the analysis run. This commit contains both the priors YAML and the populated data table; the substantive content of the priors YAML, however, is byte-identical to the version distributed in the analysis-starter pack approximately 24 hours earlier and was not modified at any point during the analysis. The data table was populated by Codex / Claude Code from independent published sources (Schlamminger 2015 Table I and the cited individual G-measurement papers), not from the priors. The single-commit ordering is procedurally weaker than a multi-commit trail (priors-then-data-then-analysis in separate commits) but is substantively clean: the priors' content has a verifiable provenance independent of the data, and no priors-tuning post-data-inspection occurred.

**Scope of the null result.** This outcome does not invalidate the QTT framework as a whole, nor any of its other such as The companion paper "A conditional parameter-free prediction of the neutrino mass-squared ratio from a two-clock cosmological framework" (Zenodo DOI 10.5281/zenodo.19960814), which uses the same Axiom 1 (two clocks) but a different observable, is currently under review at the European Physical Journal C and is unaffected by the present null. The framework manuscript itself (Zenodo DOI 10.5281/zenodo.17527179) is also unaffected; an internal review concurrent with the closure of this paper has identified that Section 22, equation (601), of the framework manuscript may require re-derivation for full consistency with axioms A1-A7. That re-derivation is undertaken as a separate framework-correction task, decoupled from any paper submission, and is documented in `framework_correction_brief.md` within this deposit.

**Why this deposit exists.** Pre-registered, parameter-free, falsifiable tests of non-standard theoretical frameworks that return null and are reported as such are rare in physics. The author's view is that the act of honourably closing a paper on a pre-registered null result has scientific value independent of any successful prediction, and that the public record of such closure is itself worth depositing. This deposition documents the methodology, the priors, the analysis, the result, and the closure decision.

**Contents.** The deposit includes: (i) the closed paper draft in its final state, (ii) the pre-registered priors YAML and pre-registration hash, (iii) the populated G-measurement data table with derived QTT structural quantities, (iv) the Codex/Claude Code analysis outputs (Bayesian evidence, BIC values, posterior samples, diagnostic plots), (v) the closure document RESULTS_FAILED.md, (vi) the framework-correction brief, and (vii) an operational closure checklist. A README file orients the reader to the deposit.