# Data audit (P1)

Schema: plain text; lines 1-9 provenance header, line 10 separator, then release narrative with ## section headings (also ### sub-headings in the Note to readers). No tabular data; "rows" = lines. Units: rates in %, changes in persons and percent, seasonally adjusted unless stated. Date: reference month September 2026 (reference week Sept 13-19, 2026), released 2026-10-09.

Relevant lines:

• L16 ## Highlights, L17: "Employment declined by 68,000 (-0.3%) ... The unemployment rate increased 0.1 percentage points to 6.5%."
• L23 ## Employment falls for second consecutive month, L24: "employment declined by 68,000 (-0.3%)" (consistent with L17).
• L42 ## Unemployment rate increases to 6.5%, L43: "The unemployment rate rose 0.1 percentage points to 6.5% in September." (consistent with L17).
• L73 ## Unemployment rate up for core-aged women, L74: women 25-54 "rose 0.3 percentage points to 5.3%"; L75: "the unemployment rate for core-aged men fell 0.2 percentage points to 5.8%".

Anomalies / issues:

1. README "Rights" line mentions "RESEA Tables 2-3, London Table 3 header", and the extract header (L9) mentions regression tables; neither exists in this extract — boilerplate from other bundles. No effect on numbers.
2. Text box "The impact of population aging..." (L56-60) and layoff/job-finding/participation paragraphs (L53-55) have no own ## heading and fall under ## Infographic 2 in the flat heading structure. Not relevant to the requested items.
3. L138 "...measure the following goals:" is followed by no list (content not reproduced). Not relevant.
4. "Note to readers", "Seasonal adjustment", "Next release" use ### not ##. Not relevant. No missing values / duplicates in the sense of tabular data; headline figures are printed identically in two places each (no internal inconsistency).

Availability rules: the only input is the admissible extract; no violation. Nothing excluded.