About
An evidence-first archive of the D'Arcys of Brisbane, kept to the same method as five sibling archives — and the rules it will not break to make a better story.
This archive exists because a family had a tree that reached the kings of Denmark and nobody had ever asked it a hard question. It is one of six, kept to the same method and by the same hand. The other five →
What it is willing to say, and what it is not
- A family tree is not a source. It is a set of assertions. Of 922 entries in the register, 449 rest on a document; the rest rest on somebody's tree, and every page says which is which. The register →
- A null result is a result. Every source read and found empty is published, so the next person does not repeat the search. What has been read →
- Errors stay on the page where they were made. 56 so far, each with what caused it. Corrections are never made silently. The errors →
- Living people are not published as people. No person page, no birth year, no birthplace, and no name where it would appear as somebody's spouse, parent or child — the rule withholds 2,034 people across the tree. 6 living people are named on this site on purpose, as the authors of a eulogy and as the reason this archive exists, each with the owner's recorded confirmation. Until 21 September 2026 this line read “not published at all — absent from the build”, which was a stronger claim than the archive keeps. How the rule works →
- A better story is not a reason. Two aristocratic descents have been taken off this family — one unevidenced, one disproved — and both are still set out in full rather than deleted. Both grafts →
What it holds
| people published | 473 | of a tree that also holds 2,034 living people, 6 of whom are named here on purpose |
| named in records | 228 | including 172 who are not relatives at all |
| register entries | 922 | 449 from a document, 473 from the tree |
| sources read | 467 | 269 yields, 106 nulls, 46 unreachable |
| errors corrected in public | 56 | plus 13 inherited from the tree |
| generations it will defend | ≈7 | not the twelve the tree claims |
This archive had drifted to the wrong side of the world
It exists to test a British ancestry: a Gloucestershire yeoman, a Staffordshire weaver, a Royal Engineer at Chatham. When this was first counted the archive held 95 Australian records against 70 British — 57% of its evidence from the wrong side of the world.
| Where the evidence comes from | Records | Share |
|---|---|---|
| Australian — Queensland registration, QSA, NAA, AWM, Trove, the Blue Books | 126 | 28% |
| British — FreeREG, FreeBMD, the censuses, TNA, Gloucestershire Archives, the Army Lists, the London Gazette, the Royal Military Calendar, Smyth's survey of the Hundred of Berkeley | 274 | 61% |
| Elsewhere in the empire — Barbados, Madeira, Minorca | 8 | 2% |
The result is that the best-evidenced part of a British family history is its emigration. That is worth knowing, and worth correcting.
It has been corrected. Since the audit was published, eighty-seven British records have been added and one Australian one — the Berkeley estate deeds, the Army List of 1778, a Welsh will, a Bristol census household, nine medieval deeds naming people of Saniger, Bigland's monumental inscriptions, two Gloucestershire wills, a Hanley census, and a line in the London Gazette of 1813.
On 13 September 2026 the archive crossed over. British evidence now outnumbers Australian — 274 to 126, 61% against 28%, having started at 42% against 57%, with 8 records from elsewhere in the empire. That is not a change of subject; it is the same family, evidenced from the country it came from. These figures are computed by tools/audit.py on every build, because the last time they were not, they drifted. That error →
Every source, and which country it is in →
How it is built
A MyHeritage GEDCOM export is parsed in Python, living people are removed by a graph-wide rule rather than a flag, and the result is rendered as a static site with Astro and published to GitHub Pages. Nothing is generated at request time and there is no database — everything you can read here is in the repository, including the tools that built it.