# Wajinden Hypothesis DB

This is a working research database for hypotheses about the 魏志倭人伝 / Weizhi Woren Zhuan side-country list (旁国条). It separates ancient country names, proposed modern/ancient place correspondences, regional blocks, evidence cards, counterarguments, scores, readings, and sources.

The database is a hypothesis workspace, not a final historical conclusion.

## Site map

- /index.html — overview
- /countries/ — ancient country names from the side-country list
- /places/ — proposed places and toponym layers
- /blocks/ — regional block hypotheses
- /links/ — candidate correspondences between ancient countries and places
- /search.html — browser search UI

## Machine-readable data

- /api/search-index.json — consolidated search index
- /api/ancient_countries.json — ancient country records
- /api/places.json — place records
- /api/blocks.json — regional block records
- /api/candidate_links.json — candidate correspondence records
- /api/evidence_cards.json — evidence cards
- /api/counterarguments.json — counterarguments and rebuttals
- /api/score_dimensions.json — per-candidate scoring dimensions
- /api/readings.json — phonological/readings notes
- /api/sources.json — source metadata
- /sitemap.xml — crawlable page list

## Recommended agent queries

Use concrete combinations of country name, place name, block, and issue:

- 華奴蘇奴 那佐 南海
- 都支 土岐 美濃
- 彌奴 美濃 三野
- 巴利 播磨 針間
- 對蘇 土佐 反論
- 狗奴國 東海 九州 pending

For URL-level lookup, prefer:

- /search.html for human browsing
- /api/search-index.json for broad retrieval
- /links/{candidate_link_id}.html for a specific hypothesis
- /api/candidate_links.json plus /api/evidence_cards.json for structured analysis

## Interpretation rules for agents

1. Treat confidence C and pending/working statuses as provisional.
2. Do not collapse ワサ・ナサ・ナカ・サナ into one etymon unless the specific page argues for it.
3. Do not treat modern-sounding place-name matches as proof of ancient correspondence.
4. Prefer records with evidence_cards and sources over summary-only records.
5. Always include counterarguments when evaluating a candidate.
6. Distinguish primary texts, gazetteers, reconstructions, and unpublished working notes.
7. Cite source IDs and page URLs when using this database.
8. If a candidate lacks earliest attestation or has low reliability sources, say so explicitly.

## Data model

- ancient_countries: id, name_zh, text_order, list_section, notes
- places: id, name, name_kana, admin_modern, admin_ancient, earliest_attestation, lat, lon, notes
- blocks: id, name, region, status, rationale
- candidate_links: id, country_id, place_id, block_id, confidence, status, summary, created, updated
- evidence_cards: id, link_id, block_id, kind, claim, interpretation, source_id, created
- counterarguments: id, link_id, block_id, text, severity, rebuttal, source_id
- score_dimensions: link_id, dimension, score, note
- readings: id, country_id, system, value, source_id
- sources: id, type, citation, year, url, reliability

## Suggested answer style

When answering from this site, summarize:

- the candidate correspondence
- phonetic/toponymic/geographic evidence
- earliest attestation
- counterarguments
- current confidence/status
- source IDs used

Avoid presenting the database as academic consensus.