With baseline shifts within 0.03 mag, fixed 15 day, 0.25 mag dips at 104 or 107.2 days need errors over 0.10 mag on at least 20 ZTF nights in 2022–2025 at J0139+5245

We test how many separate nights need large errors to admit stable dimming at the white dwarf ZTF J0139+5245. We use 424 selected g/r exposures on 226 nights in 2022–2025 from public ZTF DR24. A local correction uses 115 comparison streams. The model has a 15 day rectangular dip of at least 0.25 mag on every cycle, in both bands. Its normal magnitude stays within 0.03 mag of each earlier stream reference. We count nights with an in-box exposure that must have an error toward higher brightness of more than 0.10 mag. At every continuous phase, the count is at least 20 nights at a fixed 104 day period and 24 nights at 107.2 days. Thus a model with at most 19 nights outside the error bound cannot fit these selected data at either period. The proof allows arbitrary error distributions and correlations. It does not estimate an error probability or prove that the error and baseline bounds hold. A separate residual-injection check adds 0.25 mag dips to individual late measurements and gives at least 20 and 23 deep nightly values for the same periods. Earlier SCI work already reports the quiet record, timing coverage, and model limits under Gaussian errors. This contribution adds a deterministic tolerance to bad nights, independent calibration, and an audited public-data package. Changing shapes, missing cycles, detection selection, and other periods remain outside the main test. The package preserves source queries, frozen inputs, code, figures, numerical results, and two independent arithmetic audits.

white dwarf · ZTF J0139+5245 · planetary debris · ZTF DR24 · bounded errors · transit search · public data · robustness