The Entropy-Export Law Is Blind to Levitation and Reads -2.64 J/K on De-Alloying: a pre-registered copper-zinc composition split that determines the excess of export over production from one assay

A levitation observable carries no system entropy at all, and a de-alloying observable carries -2.6396 J/K per 100 g of Al-4wt%Cu, so of the two readings of one contested metallurgical record only the sorting reading lies inside the entropy-export law's currency. That law's left-hand side is the rate of change of the system's own entropy. Raising a rigid body reversibly changes that quantity by exactly zero, at every mass and every height. Withdrawing copper from substitutional solid solution changes it by the configurational entropy of the solution, 0.7289769 J/(mol K) carried on 3.620935 mol. The force reading is not refuted here. It lies outside the law's reach, which is a division of labor. On the sorting reading the law does more than permit the observation. The ratio of exported entropy to produced entropy is one plus the operational ratio of information change to entropy change, so a de-alloying run with a measured entropy production returns that ratio with no coefficient inserted anywhere. The whole experimental ask is the entropy production, one number the existing record does not contain. This paper pre-registers the measurement. An initially homogeneous equimolar copper-zinc rod is exposed, and the copper-to-zinc molar ratio is read at the center and at the edge. A center-minus-edge mole-fraction split of 0.05 is -0.0103974 J/(mol K), a 22.2% difference in the measured ratio, and 267 times the design's three-sigma entropy floor. The claim ends if exposed and sham rods agree to within a split of 0.0031. No rate is predicted, because the law's coefficient is supplied from outside it. The instrument that would perform the exposure is not specified here; what is fixed is the sample, the assay, the statistic, the threshold and the sham control. A script recomputes all 48 numbers.