Introduction — Annex F to BKT-32 Annex F formulates the central validation protocol for the BKT-32 module. Its purpose is to translate the relational USC hypothesis into a clear numerical parameter understandable within standard physics and statistics: the covariance-weighted predictive gain outside the calibration sample. This parameter is denoted as (G_^ USC, OOS). A positive value means that the frozen USC residual reduces the reduced (²) error of the null model on validation data. A negative value means that the USC component worsens prediction, while a value close to zero remains inconclusive. The annex does not reduce the USC theory to a single scalar. On the contrary, it treats the scalar as a secondary projection of a relational channel, required for external quantitative evaluation. The internal USC description remains channel-based and includes compatibility, tolerance, closure, scalar-projection defect and residual operator. However, for reviewable validation, a simple and measurable question is required: does this apparatus improve prediction relative to the best available null model, after parameter freezing and outside the calibration sample? The annex defines the observables space, the covariance-weighted scalar product, the orthogonality condition of the USC residual relative to the local tangent space of the null model, the tangent absorption coefficient (ₓ₀₍), the residual-direction alignment coefficient (ₑ₄ₒ), the (BIC10) criterion, the full Bayes factor (BF10), and the procedure for moving from aggregated data to a full test on raw Jefferson Lab data. The annex also contains a demonstrative EHM block, in which a simple scalar closure of three dressed masses leaves a measurable closure defect relative to the proton mass. A model with an explicit closure-defect operator yields strong support in a demonstrative Bayesian test, but this result is not identified with full validation of the CLAS12/KaonLT sector. Its significance lies in showing that the USC apparatus can be written as a falsifiable quantitative procedure. Full PASS status requires applying the same protocol to raw binned data, with a complete experimental covariance matrix, frozen USC parameters after calibration, reproducibility in at least two independent validation blocks and simultaneous fulfilment of the following conditions: positive (G_^ USC, OOS), preference for the USC-residual model after complexity penalty, positive Bayes factor and low absorption by the null model. Annex F is therefore a standalone methodological-numerical module. It can be used not only for BKT-32, but also as a common validation protocol for subsequent PJM–GTWSSF–USC–GTCW / LOM–GTSFC–USC–GTCW modules, in which the hypothesis of a relational USC layer is tested against well-defined null models. Keywords: USC; GTSFC; BKT-32; Annex F; Jefferson Lab; EHM; CLAS12; KaonLT; null model; predictive residual; (G_^ USC, OOS) ; (²) ; covariance matrix; Mahalanobis metric; (ₓ₀₍) ; (ₑ₄ₒ) ; BIC; Bayes factor; PCA; ICA; FastICA; out-of-sample validation; PASS/FAIL.
Robert Kupski (Thu,) studied this question.