The Gaiad Genealogy

Individuals

107,317

Families

56,697

Encoding

UTF-8

Licence

Free to use

What this is

The Gaiad is a synoptic ancestry. Where the biblical model descends purely through the father, this one deliberately integrates different kinds of ancestry — Greek, Near Eastern, Egyptian, Trojan, Chinese, Mongol, Indian, Japanese and biblical lines converging into a single descent. Japan is the clearest case: Chinese ancestry arrives through Jimmu and the Kings of Wu on the father's side, Korean and Indian ancestry through Kanmu's mother Takano no Niigasa. Two traditions, two parents, one person.

That integration is the point of the project, and it is what the file is for. Places where traditions usually kept separate turn out to meet are not accidents of import. They were put there.

Reading it well

Every record keeps its identifier

Each individual carries a REFN holding its wikibase QID, so any person can be traced back to the source record. Where a record is matched to Wikidata, that identifier travels alongside it in a _WIKIDATA tag.

A few people have no name

716 individuals appear with an identifier but no name. They are real people in the genealogy — someone's recorded parent, child or spouse — whose own record carries no label. They are exported rather than dropped, because the relationships around them are real even where the name is lost.

Deep time does not fit

GEDCOM years run from 1 to 9999. Aster, at the root, is dated thirteen billion years back. Rather than drop those dates, 944 of them are preserved verbatim in a note attached to the event, so your software shows an undated birth with the raw timestamp beside it.

Dates carry their real precision

A date known only to the year is written as a year, not as the 1 January its underlying timestamp happens to store. Decade and century precision is written with ABT. This matters more than it sounds: 60% of the recorded births are known only to the year or coarser — 12,554 of 21,004 — so rendering them from the raw timestamp would invent twelve thousand birthdays that nothing in the sources claims.

Some people have more than one set of parents

This is deliberate, not corruption. A man born into one house and adopted into another really does have two fathers, and so does a person whose parentage the traditions genuinely disagree about. Both appear as multiple FAMC links, which is legal GEDCOM; most software will show them all and let you choose a preferred line.

A few loops remain

Three ancestry cycles survive in the source data, trapping about thirty records — roughly one person in two thousand. They came in through merges made upstream in Geni and Wikidata, and each is resolved against an external source rather than cut arbitrarily: the ones already closed turned on a marriage contract from 1037, a queen's coffin, and a Roman adoption recorded in a cognomen. The three still open are waiting on a Welsh pedigree, a Servilian stemma, and the Livro de Linhagens. If your software walks ancestors without a visited-set, those thirty records can send it round in circles. Everything else is a clean acyclic tree.