Schwarzschild spectral ladders on the negative imaginary axis: Endpoint nonselection, branch-cut phase, and Jost classification
Compactified spectral discretisations may yield stable negative-imaginary-axis (NIA) eigenvalue ladders, but finite-matrix convergence does not establish quasinormal poles. For axial Schwarzschild perturbations, ingoing--outgoing factorisation leaves an unwanted horizon solution behaving as $t^{4\alpha}$ at $\Omega=-i\alpha$, and a $C^\infty$-flat unwanted infinity solution. Thus endpoint $C^k$ regularity is nonselective for $4\alpha>k$, so the $C^2$ continuum problem cannot have a discrete NIA spectrum over $\alpha\simeq15$--32. Exact Chebyshev-grid formulas give roots-grid maximum radius $O(n^{2p})$ for $x\sim C(1-y)^{-p}$, explaining C1, $x=2/(1-y)$, versus C2, $x=4/(1-y)^2$. Arbitrary-precision pencils nevertheless reveal a reproducible 68-point C1 ladder, while C2 reorganises it. Both physical lateral Jost determinants remain stably nonzero at every C1 frequency. A 0.05-spaced on-cut scan finds no zero, and refined argument-principle calculations give zero winding in both continuation strips. This rejects all 68 candidates, although the mesh and strip evidence is not interval-certified. Same-damping controls recover Schwarzschild QNMs $n=60,100,130$, with $|\widehat{\mathcal D}|=10^{-48}$--$10^{-56}$. The ladder has surface-gravity quarter spacing, follows the parameter-free Casals--Ottewill branch-cut-strength phase, and pairs with damping projections of QNMs $n=62,\ldots,129$. At its first member, a finite-frequency calculation finds a branch-strength zero at $\alpha_q=15.07832396512359$, between the asymptotic prediction and the C1 root, supporting, but not proving, sequence-wide cut-phase locking. Thus accurate finite-pencil eigenvalues can fail the invariant Jost/Evans pole criterion. Whether representation-dependent nodes with Keldysh weights converge collectively to the Schwarzschild cut response and Price tail remains open.