Labels Where the Father Should Be

A sorting bucket is an entry like Unknown Sorting Working Blažević or Brothers [of Josephus] Žubrinić. No priest wrote those words and nobody was ever called them. They are labels somebody typed into the father's slot while working through a pile of records — and they are still there, still doing the job of a parent.

The biggest ones

Do the labels encode anything real?

Somebody grouped these children for a reason. Krivi Put 136 For Sorting Perpić carries a house number; 1800-1830 Birth — Selce Papić (Papa) carries a village, a date range and a nadimak. The question is whether the reason survives well enough to place a person by — and it can be tested, because some of the children under each label have a place of their own on record.

It splits sharply, and on one thing: how specific the label is.

So: 124 published people with no place of their own sit under a label that names one — and only 19 of those sit under a label worth believing. The rest cannot be placed this way, and this archive will not place them.

Which is a result worth having either way. The two specific labels are worth taking to a registrar as a question; the vague ones should be treated as what they are, a note to self by someone half way through a sorting job.

Delete the labels and two-thirds of this archive floats free

The section above asks whether a label is accurate. This asks something harder. If every sorting label were removed from the tree, how many of the people published here would still be connected to Hedviga at all — by any path, however long, of parents, children and spouses?

537 of 1708. 31%.

What this does and does not mean. It does not mean these people are invented, or that the research was careless — the register is full of real entries about real people, and this archive publishes by surname group, never having claimed each person is a proven relative. What it means is narrower and still worth saying out loud: for most of the names here, the link to Hedviga is currently made by a sorting note rather than by a document. A reader is entitled to know which two-thirds.

It also reorders the work. Reading one Otočac marriage that joins a Žubrinić fragment to the line is worth more than reading fifty pages that confirm dates already held.

It is not 1171 loose people. It is 95 islands.

“Two-thirds of the archive reaches Hedviga only through a label” sounds like eight hundred separate problems, and it is not. Take the labels away and the published people do not scatter — they fall into 95 coherent families that are simply not joined to hers. Most are internally complete: parents, children, marriages, house numbers. They are whole households missing one link at the top.

And the distribution is kind. Hedviga's own island holds 560 people. The next eight hold 782 between them. So the gap is not eight hundred documents — eight marriage entries would close most of it.

The candidate is the earliest person in each island who still carries a birth year — because the join is missing at the old end, and that is where a marriage entry would attach it. Ties go to whoever has most children, since joining them moves the most people.

Which reorders the research completely. This archive has spent its effort confirming — finding the register entry behind a date already held. That work adds nothing to the number above. One marriage at Otočac would add two hundred and seventeen people. What to ask for →

What follows

• Un-publishing is not removing. A record hidden from the register goes on acting as a parent in every tool that asks whether someone has one. That is how the fragment count was wrong by 25 for months.
• Nothing here has been edited. The buckets remain in the tree exactly as they are. Method →
• Sixty-two of the seventy-four are not Žubrinić. Thirty Blažević, twenty-two Prpić, ten Papić and four Perpić have never been counted as loose ends anywhere on this site. They are, and now they are listed.

The Žubrinić fragments → The same person, twice → What cannot be true →

15 September 2026 — the alphabet, found because a reader asked

The detector on this page looks for the words a researcher writes when they know they are guessing: sorting, working, unknown, brothers of. It could not see a drawer labelled “M”.

Someone looking at a Žubrinić page asked what all the single-letter Žubrinići were, and whether the site was saying they were related. It was. Fifteen nodes are named with one capital letter and a surname — A · B · C · F · G · H · J · L · M · N · P · R · S · T · W — every one of them “X Žubrinić Xubrinich Zubrinic”, with no birth, no death, no place and no source.

Each is entered as the parent of every Žubrinić whose name begins with that letter. M is the father of Mathæus, Michael and Mate; T of three men called Thomas. And they are entered as one another's siblings, so the tree asserted fifteen brothers and sisters called A through W, hanging from nothing, with 75 real people beneath them across 285 relationship links.

The giveaway is the sorting. Jure / Georgius / Juraj is filed under G, not J — the letter follows the Latin form, not the name the page shows. That is a card index, not a household. One drawer even breaks its own rule: Nicolai Žubrinić is filed under M.

What changed. Both the detector here and the site's own junk filter now test for it — one capital letter, a surname, and no date of any kind, so that a real person recorded only by an initial is not mistaken for a drawer. The fifteen no longer appear as anybody's parent, sibling or child, and they no longer have pages. That moved the count on this page from 21 buckets to 37, and the people standing on one from 218 to 286.

And the gap is now stated instead of left blank. A person whose only parent was a drawer used to show an empty space where a father should be. Their page now says so, and names the label. 144 people carry that notice.

The tree itself is untouched. This is a change to what the archive publishes, not to the GEDCOM — the separation the whole project rests on. The drawers are still in MyHeritage, doing what drawers do.