The Same Person, Twice

This check began as a Žubrinić-only test and that was the mistake. Run across all 1833 published people it finds the same disease in four families — and the first version of it was badly wrong, in a way worth showing rather than quietly fixing.

Comparing names is harder than it looks

Three further corrections were needed, and two of them came from outside the data — one from a photograph of a grave, one from noticing what was in the father's slot.

And the last was forced by reading the rejections. Going through all thirty-two by hand — which had never been done — showed the parent test failing on the commonest thing in these registers. Mile Blažević is in the tree three times, and the three entries give his parents as Antona Blažević & Marija Sojat, Antone Blažević & Mare Sojat, and Anton Kljapo Blažević & Marija Šojat. That is one couple written three ways, and the check called it three families.

Mare is not a variant spelling of Marija. It is a different word for the same woman — and this archive has a whole page about exactly that. The same knowledge now sits in the tool: a short list of the short forms these registers actually use, plus a rule that collapses doubled letters, because Phillipus and Philippus are two substitutions apart and no single-edit test will ever join them. Five groups came back, including a Lucia and two Magdalenas.

And one was forced by a South Australian birth index. This archive held Kate Zubrinich and Katalyn Zubrinich as two women — identical in birth date, birth place, death date, death place and both parents — and this check walked straight past them, because “kate” and “katalyn” are disjoint tokens rather than one inside the other. A FindMyPast search of the South Australian births settled it: one registration in 1886, at Port Pirie, for a Kate, and no Katalyn anywhere in 176 records of that surname. One woman.

The last of those was forced by a gravestone. A photograph of a grave at Klenovica names Lovro Kalanj 1906–1965, Ana Kalanj 1919–1971, Ivan Kalanj 1942–1996, Marijan Kalanj 1944–2009 — one family, four people. The tree holds all four twice, and because the duplicated parents are spelled Lovro in one copy and Lovre in the other, the parent test rejected the children as different people. The stone said otherwise and the stone was right.

Which is the useful lesson: a parent-agreement test fails precisely when the parents are themselves duplicated. It is a good test and it is not a complete one.

What survives the tests

What this does not do

• Nothing is merged. This archive does not edit the tree. It reports what the tree says, and here the tree says the same child twice.
• The rejected 33 are not cleared. "Rejected" means the evidence for a duplicate failed, not that two people are proved distinct. Where the parents are themselves duplicated — as the Kalanj grave showed — a rejection can be wrong.
• Two of the rejections are the other way round, and they are the firm ones. One person is born once and dies once. Two firm death years rejects a necronym — a child given a dead sibling's name, which shares a surname, a given name, a birth year and a village and is still two children. And since 20 September 2026 two full birth days does the same and does it harder: Magdalena Kalanj is 8 June 1829 at Klenovica in one record and 19 February 1829 at Ledenice in the other. A different day, a different month and a different village — and she was being carried as a live duplicate because the tool had only ever rejected on death years.
• The weakest tier should probably not be believed. Same name, same year, nowhere named. It is published because leaving it out would be a choice about what counts, made silently.

What cannot be true → The Žubrinić fragments → Spelling is geography →