The seed-amplitude dependence of the winding density changes sign after the linear stage in an undamped two-dimensional Abelian-Higgs quench at critical coupling

We study a strictly two-dimensional classical Abelian-Higgs model at critical coupling, with undamped Hamiltonian dynamics, quenched by a linear mass ramp from cold random-phase seeds of amplitude σ₈. The dependence of the winding density on σ₈ changes sign after the linear stage: at its end the density rises weakly with the seed, while 4.5-6.5 Kibble-Zurek times t̂ later, in windows aligned at each run's own end of the linear stage, it falls, 2.8-fold over 4.5 decades in raw plaquette windings (individual vortex cores are not resolved there). This seed scan is a report-only replication, on fresh seeds, of a reversal seen in an earlier study, and its detailed statistics are post hoc. In seed-paired runs bitwise identical through the linear stage, changing only the ramp afterwards moved the next burst of winding births in the direction of a sealed calibration, by 1.7-2.4 times its size. In tests planned before their runs but with these effects known, the S4-to-S2 early-rise/late-fall sign pair (σ₈ = 10⁻⁴ to 10⁻²) survives a factor-two joint refinement of lattice spacing and time step. In 32 and 128 t̂ boxes, the late-fall slope and the four ramp-contrast sign tests retain their registered directions; the 16 t̂ box exposes a finite-volume limit. At three other couplings, only selected signs were tested, and they hold. The late state is a transient on a running ramp; we claim nothing about asymptotic counts, damped dynamics or superconducting films.