This paper presents the second empirical validation of Eormen and its first validation on real accelerometer data with physical structural perturbation. Applied to the CWRU Bearing Data Centre dataset without modification, tuning, or domain-specific calibration, the framework shows 100% baseline integrity, sensitivity across all four observational lenses, and a temporal onset cascade in which structural change becomes observable at distinct timescales from cycle 25 to cycle 699, with zero false onset in normal files. It also reports monotone calibration across all five testable bearing families and identifies an eigenvalue regime bifurcation linked to sensor channel architecture. Eormen is positioned here as a deterministic, non-interfering observational instrument, not a predictive or classification system.