Why this project, and the rules we set before we rolled the dice

We are going to research one French-Canadian family from nothing — one marriage, chosen at random from a parish répertoire, climbed upward on every line to the immigrants — and write down everything as we go: what we tried, what worked, what broke, what we decided and why.

Steve has spent more than twenty years on his own lines (the Oligny and Saurette families of the Richelieu and Laprairie, published at ondits.com). That work had a head start no beginner has: a tree of twenty-two thousand people, a filing system, and a memory full of parishes. The question this project asks is simpler and harder: what does a sound method look like when you have none of that? If the answer is any good, it should work for anyone.

What we are doing

• The case study comes first. One 1820–1850 marriage, both spouses, every line, upward to the New France arrivals. Roughly 60 to 120 ancestral couples. Finishable.
• The framework is extracted from the work, not imposed on it. Records are small text files, one per person, family, source and question; a script checks that every claim cites a source that exists and that no claim is graded higher than its sources allow. Nothing is proven by saying so. METHOD.md holds the rules; they change only with a journal entry saying why.
• This journal is the deliverable for anyone else. Entries are never rewritten after their day. A correction is a new entry that links back.

The rules we carried in

They are Steve’s, earned on his own families, and they are in METHOD.md. The ones that matter most on day one: the primary act governs — a répertoire or a PRDH page is a lead, never a proof, and a claim resting only on compiled sources cannot rise above Probable; identity is a claim — two documents with the same name are two people until a document says otherwise; count before pattern; and check what is already held before calling anything new.

The rule we added because of who Steve is

Steve knows the South Shore. He cannot un-know it. So we made prior knowledge a thing the framework models rather than a leak: anything from memory, family lore, another tree, or his own GEDCOM goes into a prior record as unverified, may direct a search, and may never be cited as a source. His tree is opened once — today, for the surname counts below — and that opening is itself a record (PK-0001). Every beginner has a grandmother’s story to treat the same way.

How we would choose the marriage

A beginner does not choose their ancestors, so we decided to let dice choose ours — through a gate, so a hopeless draw could not cost us a season. The gate, written before the first roll:

1. Married 1820–1850, in a South Shore / Richelieu parish whose répertoire Steve owns (twenty-two volumes). Before 1850 the whole ascent sits inside PRDH’s dense coverage.
2. The entry names both sets of parents — four lines wide from day one.
3. A first marriage for both (no veuf, no veuve). Widowhood is a fine complication for generation two, not for the door.
4. Neither surname is one of Steve’s line surnames, and neither is common in his tree — written as “about ten individuals or fewer”, a guess we knew we might have to revise.
5. Each surname has roughly 5–40 marriages in the same volume — a real local family, neither a lone migrant nor a same-name swamp.
6. A bonus, not a requirement: the parents’ own marriages are in the same volume.

Draws are recorded whether they pass or fail. The failures are part of the method.

← All entries