The Third Selection That Isn't: The Squared Share's Repair, Killed by the Read-Back Geometry

The Squared Share (batch 1) implied eta_B = 3.53e-10, 42.1% below the measured 6.1e-10; its hopeful reading was a third selection raising epsilon by the needed factor 1.7265. This paper tests every address the read-back geometry owns, in exact fractions: the crossing levies exactly nothing (toll-free by theorem, paper #8); the K-orbit quotient identifies the pair's members and zeroes the share difference (factor 0); face counting (m+1 faces vs m cells) gives the largest forced factor, 1.0196, leaving eta_B = 3.6023e-10 still 40.9% short; the end-face dress returns exactly 1; a second-sever selection is unowned geometry, evaluated at two sentences and not used; the census of owned near-unity factors finds none in [1.6, 1.9], and the toll itself (2.055) overshoots and would double-levy the crossing that levies nothing. The repair is dead; the construction stands with its flank, graded by PD-889 alone. A negative result against our own construction, printed at the same size as the construction.