Breaking the Landau-fit degeneracy in ferroelectric domain-wall tension metrology: a registered, sign-definite correction prediction

The 180° domain-wall tension σ of ferroelectric BaTiO3 is a key parameter for domain-wall device modeling, yet reported values span a factor of forty (1.5–62 mJ/m²) depending on the extraction method. We show that this scatter reflects an attribution degeneracy—each method probes a different direction in the space of bulk Landau parameters, leaving no method-independent observable—and we register a falsifiable route out of it: within a minimal framework of elastic-origin discrete roughness corrections organized by q^{2k} (q=1/5), the wall-tension and wall-width anomalies are locked in the ratio δlnξ/δlnσ=+1, the unique direction of the gradient channel, with positive sign, a derived floor of +2%, and a swept window of +2% to +18%. We specify the demixing statistic that separates this signal from Landau-fit artifacts, pre-registered acceptance criteria, a detection-sensitivity grid in the κ-prior uncertainty, and a feasibility analysis with signal-to-noise ratios of 10 or better in four standard channels; null outcomes inside the declared precision yield pre-registered constraints on the correction coefficients.