Research status
A measured corpus, with its limits attached.
This page separates the current export from the evidence-backed recovery minimum. It also distinguishes rows with saved candidate viewer metadata from rows with none; neither state is an original-image reading. The numbers describe work states, not records, people, or conclusions.
Working-corpus baseline
| Measure | Count | What it means |
|---|---|---|
| Distinct legacy row keys in current export | 626 | The inherited export retains 626 distinct old row keys. Its legacy key omitted fields that distinguish at least one row preserved in the saved HTML. |
| Evidence-backed recovery minimum | ≥627 | At least 627 distinct normalized index-row tuples are recoverable from the saved HTML. This is not a count of proven documentary records, occurrences, or people. |
| Countable core-cluster tuples in current export | 623 | The current-export count of countable core-cluster index-row tuples. |
| Evidence-backed countable recovery minimum | ≥624 | The minimum after restoring the tuple lost by the old key. This is not a record or person count. |
| Current-export date span | 1636–1948 | The earliest and latest years carried by the retained index rows, not verified surname endpoints or a completely searched period. |
| Current-export rows dated before 1700 | 84 | Retained countable early leads; this is not the recovery-adjusted minimum. |
| Recovery minimum dated before 1700 | ≥85 | The omitted recoverable tuple is dated 1696. These remain derivative index-row tuples, not verified occurrences. |
| Geocoded rows in current export | 621 | Rows with working place coordinates; geocoding does not verify the name reading. |
| Parish-originator labels in current export | 56 | Distinct labels across all retained rows; 54 occur among the 623 retained countable rows. |
| Retained countable rows with saved candidate viewer metadata | 83 | The intended image has not yet been proved unique for every row. These are not resolved locators or original-image readings. |
| Retained countable rows with no saved candidate viewer metadata | 540 | These remain derivative index leads. Together with the 83 metadata rows, they account for the 623 retained countable rows. |
| Original images independently read and recorded | 0 | First diplomatic transcription, blind second reading, and adjudication counts are also zero within this legacy corpus. |
What this release does not claim
This release does not treat either the 626 current-export row keys or the recovery minimum of at least 627 tuples as people, documentary records, verified occurrences, or original-image readings. It does not claim a single lineage, shared origin, earliest bearer, or complete world coverage. Within this legacy corpus, zero original images have been independently read and recorded. Similar or repeated strings are not merged into one person, and a catalog span is not treated as proof that every locality appears in every year.
Current research priorities
Resolve early entries to exact source images
Bind each high-value index lead to its repository, book, image or folio, record boundary, and access/rights state.
Obtain independent readings
For load-bearing records before 1800, retain two qualified readings or a documented adjudication before a field is called verified.
Audit coverage, not just hits
Record the complete source-unit denominator, searched range, gaps, unreadable regions, negative searches, and remaining leaves.
Expand sideways at documentary floors
Study spouses, children, siblings, sponsors, witnesses, neighbors, households, occupations, estates, and migration associates as separate observations. Relationship claims still require their own proof.
Keep literal streams separate
Each source spelling stays visible and separately countable. No public normalization or pooled total is used while the name-relationship policy remains unadopted.
From research infrastructure to named evidence
R2 now publishes six record-centered dossiers with actual names and facts. They are not biographies or tree profiles: each begins with a specific record, cites every displayed claim to a stable source ID, separates source statements from identity conclusions, exposes conflicts, and withholds names that lack an individual publication decision.
Browse the released records or see how identity and relationships are evaluated.