# Extended fact sources — retrieved 2026-11-06 evening, Europe/London

## Wikidata structured metadata (CC0)
Endpoint: https://query.wikidata.org/sparql . All exact queries are saved in `(local build workspace: facts/queries/)`; successful raw result JSON is in `(local build workspace: facts/raw/)`. `(local build script) build_facts.py` uses an identifying User-Agent, sequential batches of up to 15 areas, a 1.1-second pause, and a 50-second request timeout. The query cache is invalidated when the query text changes; `--refresh` explicitly refreshes it. Source QIDs, property IDs and statement URIs are retained on retrieved facts. This is a sourced prototype, not a guarantee every Wikidata statement is correct.

County matching: retain one record for each of the **91 OS Boundary-Line ceremonial/lieutenancy/preserved county shapes**. Wikidata direct instance types Q180673 (ceremonial county of England), Q202435 (lieutenancy area of Scotland), Q630638 (preserved county of Wales) are required. Matching is exact English label/alias, with explicit documented OS name variants. Ambiguous/unmatched records remain unmatched — no council-area or historic-county substitution. **80 matched, 10 unmatched, 1 ambiguous** in this snapshot.

Parish subset: civil-parish type Q1115575; explicit P131 hierarchy to a matched county; exact parish-name match with the OS `CP` suffix removed; P625 coordinate inside the corresponding OS parish polygon, with a 0.002-degree edge allowance because the OS output is generalized. At most 15 alphabetical verified examples for each of Nottinghamshire, Derbyshire, Leicestershire, Lincolnshire, Rutland, Northamptonshire. No claim all parishes or all parish facts are covered.

People: six East Midlands counties only. Direct writer/poet/playwright occupations Q36180/Q49757/Q482980, explicit birth place P19 with P131 hierarchy to county. Ranked by Wikidata sitelinks (a transparent prominence proxy, not a biographical judgment), maximum eight distinct people per county. Dates are sourced P569/P570 values; hierarchy evidence is in saved queries/results. No fictional biographies.

Literature: P50 author connection for these people; a bounded set of directly typed novels, poems and literary works; up to three rows per person. A work is associated through its author's county birth connection — **not** a claim it was written in or about that county. Titles, creator, bibliographic dates, types and URLs only. **No book/poem full text or excerpts; no public-domain status claimed for modern works.** Bibliographic facts are not literary text. No Wikipedia prose was copied. Wikipedia links are navigation only, not a CC BY-SA text import.

Area metadata: labels, short Wikidata descriptions, P36 administrative centre, P571 inception, P576 dissolution, P138 named-after, P94 coat-of-arms image URL, P610 highest point, P1589 lowest point, P1056 produced material/product, P452 industry, P163 flag. P1082 population is included only for preferred statements, with P585 date qualifier where returned. Coat-of-arms/flag **image URLs are references only**; image licences are NOT inherited from Wikidata CC0 and must be checked before display/downloading.

Ecology: direct P6803 means taxon reported at the exact location, **not native status**. The supplementary endemic sampler uses explicit taxon P183 (endemic to an exact country entity), classified by P171 ancestry into Q729 animals/Q756 plants, with P141 conservation-status values where asserted. Endemic sampling is limited to seven requested English-speaking country/territory entities, at most 12 high-sitelink species each (P105 species rank and no other P183 area in the snapshot), not a flora/fauna census. A Wikidata endemism claim is not an independently verified range; higher taxa and multi-area assertions are excluded. Extinct species may occur in the sample with their stated status. Wikidata conservation-status statements are not independently refreshed IUCN assessments. Empty arrays mean not collected, never absence/zero. No symbolic national animal/flower is silently treated as native.

## Official site-specific ecology facts
Editable seed: `Facts/official_ecology_seed.json`. Short factual metadata in our own wording, not copied descriptive paragraphs or photographs. Site scope is retained on each observation. Sources opened during this run:

- Forestry England, Oxclose Wood, Mansfield Woodhouse: https://www.forestryengland.uk/oxclose-wood — skylark, kestrel, buzzard, orchids (species not specified).
- Forestry England, Birklands Sherwood Forest: https://www.forestryengland.uk/birklands-sherwood-forest — nightjars and glow worms; historic timber/grazing/royal-hunting uses. **Not export statistics.**
- Natural England, Leicestershire's NNRs, updated 30 July 2026: https://www.gov.uk/government/publications/leicestershires-national-nature-reserves/leicestershires-national-nature-reserves — Muston Meadows green-winged orchids, great crested newts, named meadow/hedgerow birds and voles; traditional hay/grazing management; Bradgate Park/Swithland Wood and Charnwood Lodge as conservation/geological landmarks. Natural England/GOV.UK content OGL v3.0 except identified third-party material; none copied here.
- Natural England, Dr Carlos Bedson, 28 March 2025: https://naturalengland.blog.gov.uk/2025/03/28/mountain-hares-in-decline-new-research-reveals-diminishing-numbers-in-the-peak-district/ — mountain hares introduced to the cross-county Peak District in the 1870s. The observation is cross-county and is explicitly **not** a Derbyshire population estimate or a claim they are native there.
- Peak District National Park Authority: https://www.peakdistrict.gov.uk/visiting/places-to-visit/stanage-and-north-lees/ring-ouzels — ring ouzels at Stanage/Burbage and national red-list concern as described by this source. UK national red list is not IUCN global status.

Source website copyright/licence governs their prose/images; only attributed factual observations and our own short metadata are included. No images or long source text are redistributed. The attempted Peak District habitat report page could not be fetched (server/TLS error) and is **not used** to fill flora facts.

## HMRC international trade in goods (OGL v3.0)
Source: **HM Revenue & Customs**. Release published **17 September 2026**, second quarter April–June 2026; current source page opened 2026-11-06 Europe/London:
https://www.gov.uk/government/statistics/uk-regional-trade-in-goods-statistics-second-quarter-april-to-june-2026
https://www.uktradeinfo.com/trade-data/regional/2026/
Original workbook:
https://www.uktradeinfo.com/media/bfohjlem/uk_regional_trade_statistics_second_quarter_2026_accompanying_tables.ods

`(local build script) build_trade.py` extracts **2026 Q2** only, SITC section-level goods imports/exports, retaining EU and Non-EU components separately. Every value includes the original worksheet and cell, unit GBP million, provisional flag and retrieval/release date. Original file retained under `facts/raw/`, plus a separate normalized ZIP-path derivative because the publisher's ODS contains Windows-backslash member paths. This is an OpenDocument file, not a Microsoft/Windows database.

**16 area/group records and 640 commodity/flow/partner observations.** Areas include UK, nine English ITL1 regions, England, Scotland, Wales, Northern Ireland and two unallocated categories. Trade excludes services, non-monetary gold and estimates for late response; 2026 figures are provisional. See the original workbook Notes for EU data-source breaks, regional allocation and unallocated trade. There is NO county/parish allocation or invented import/export figure. Counties only link to relevant broader regional context; ceremonial Lincolnshire straddles East Midlands and Yorkshire and The Humber. Other Commonwealth countries' trade remains uncollected in this first pass.

## Final ecology validation refinement
Inspection found that even sole-area Wikidata P183 statements can overstate exclusivity (e.g. plants occurring in both Canada and the USA). The 40 world species records therefore deliberately assert only **reported country association**, NOT exclusive endemism or independently verified native range. The raw property assertion is retained with an explicit caveat, and no distribution map is inferred. This also avoids presenting higher taxa as species. Official county observations remain the stronger, site-specific ecology seed.