1# Verification checks (P4)21. Clean re-run: deleted results.json, analytic.csv, exclusions.csv; re-ran analysis.py; SHA-256 identical (results.json 78303c9f…). PASS32. Independent spot recomputation of g for 2014-06, 2016-07, 2020-05, 2022-06, 2026-09, 2009-07 directly from vintage files: matches results to 1e-10. PASS43. n_months_growth + n_months_missing_growth = 309 (window months); analytic.csv rows = n_analysed = 308. PASS54. Exclusion ledger: 309 -> 309 -> 308 -> 308, each n_before − n_removed = n_after, chain consistent. PASS65. Weighted sub-period means (156, 152 months) reproduce the overall mean 2.5741201901. PASS76. growth_12m in analytic.csv equals 100(cpi/cpi_lag12 − 1) on every row. PASS87. vintage_used = 2026-10-28 for all months ≥2014-01, 2016-07-15 before (information rule). PASS98. max/min and their months match analytic.csv. 2025-10 absent (no imputation); no month beyond 2026-09. PASS109. Magnitudes: overall mean 2.57% close to 2001–2013 baseline 2.36%; extremes (−1.96% 2009-07, 8.98% 2022-06) plausible for US CPI. PASS11No uncertainty intervals are produced (deterministic arithmetic on published data), so no CI checks apply.12