Endometriosis affects approximately 190 million women of reproductive age worldwide. Standard management — hormonal suppression, surgical excision, and combined oral contraceptives — controls symptoms but suppresses the reproductive axis, preventing pregnancy as a mechanism of treatment. The gynecological literature has empirically converged on combination approaches pairing injury-rate reduction with antifibrotic therapy, yet no formal account exists of why this combination is non-redundant — recovering lesions neither monotherapy can — rather than merely additive. This paper formalises endometriosis as a candidate instance of Mode II compensation failure: a regime in which a compensatory response (fibrotic wound healing) fires appropriately but its consolidation operator is misaligned toward a pathological attractor. From a cycle-averaged differential equation incorporating two documented nonlinearities — self-amplifying extracellular matrix (ECM) stiffening and resolution saturation — we derive a closed-form locking threshold ε* = ρ / [ρ + (√α + √(θ/ψ))²]: the misalignment above which, for a lesion at its current state, the fibrotic attractor is the only stable state. We prove that neither nonlinearity alone produces bistability; both are required (necessity), and sufficiency is established separately. Three testable predictions follow. (i) Anti-platelet modulation of wound-initiation rate raises ε*, slowing progression in early-stage disease. (ii) TGF-β3 restoration of resolution efficiency expands the recoverable basin. (iii) The combination is non-redundant and, in the fragile regime (ε* < 1/2), super-additive on the locking threshold — recovering lesions neither monotherapy reaches. As a structural consequence, because none of the intervention targets involves the hypothalamic–pituitary–ovarian axis, the combination preserves ovulation where hormonal suppression does not. A worked example illustrates how the ε* structure suggests separating dienogest's fibrotic benefit from its fertility cost via direct myofibroblast apoptosis induction combined with anti-platelet therapy. All predictions are falsifiable with existing or obtainable data. This paper is part of a program on formal frameworks for biological compensation. The companion framework paper is posted at DOI: 10.5281/zenodo.20515948.