By the time Saturday was finally winding down, it was already well past 22:00.
The New Zealand Sauvignon Blanc I had unexpectedly found in a Moldovan supermarket earlier that evening was now open beside me. I had a computer in front of me, a nice bottle of wine within reach and, more importantly, the entire Internet available for entertainment.
There are probably more sensible ways to spend a Saturday night.
I opened my family-tree database.
That was the end of any realistic prospect of going to bed at a reasonable hour.
The doctor with three wives
One of the important unresolved points in the German side of my family tree concerns my great-grandfather — the father of my German grandfather.
He was a medical doctor and, over the course of his life, had three wives.
That much I knew.
What happened around those marriages, however, had become obscured by the usual family fog that accumulates when information is passed verbally from one generation to another.
My grandfather came from one of those marriages. Many years ago my father had also given me fragments of information about the other two wives. But they really were fragments: some first names, approximate dates, a few family names and relationships, and little more.
Enough to know that there was a story there.
Not enough to reconstruct it easily.
I had already tried using ChatGPT to locate the missing connections. It could help me organise what I knew and suggest approaches, but searching from those sparse details produced nothing convincing.
There was only one thing left to do.
Start digging manually.
Trying combinations until something moved
I began working through the genealogy and heritage sites, changing combinations of names, places and dates.
A surname with one approximate year.
The same surname with another wife's first name.
A location.
A variation in spelling.
A different date range.
Then another combination.
This sort of research is difficult to automate completely because much of it depends on recognising something when it appears. A search result may be technically imperfect and still contain one name, one place or one relationship that suddenly makes the rest interesting.
After perhaps thirty minutes of trying different combinations, I found an entry that looked promising.
I opened it.
Bingo.
It appeared to be my great-grandfather.
The names fitted. The dates fitted. The family relationships began to make sense.
More importantly, that first result gave me something I had not had before: new names and dates with which to search.
And once a genealogical search acquires another reliable-looking name, everything changes.
Three separate trees
I kept going.
Eventually I found all three wives — but not neatly assembled in one convenient family tree.
They appeared in different family-tree constructions created by different researchers.
That was one of the most interesting parts of the discovery.
Each branch seemed to contain only part of the story. The trees I found did not show the complete sequence of three marriages that I had inherited through my own family information.
One researcher had one branch. Somebody else had another. Elsewhere was the third.
I was apparently looking at three pieces of the same family history without the people who had created those trees necessarily having connected them to one another.
I could.
My father's old information provided part of the bridge. The online trees supplied additional names, dates and relationships. Those details led to other records, which in turn produced more names and places.
The three branches began to converge on the same man.
The overlap was substantial: names, approximate dates, spouses, children, locations and successive generations all began fitting together.
That does not yet make every connection proven. Online family trees have to be treated carefully. Several apparently independent trees can ultimately derive from the same erroneous source, and repetition does not transform an error into evidence.
But as a research hypothesis, what I was seeing was becoming extremely persuasive.
Then the tree suddenly jumped backwards
Once the three marriages were connected, the search expanded rapidly.
One of the branches extended backwards generation after generation until I reached a person recorded as having been born in 1699.
That certainly got my attention.
My German grandfather's family was Prussian. I use that word deliberately rather than simply describing the family as German, because the historical geography matters when tracing people backwards through this period.
Germany as a unified empire did not exist until 1871, while Prussia had its own much longer and increasingly complicated political history. At different periods its territories extended across large parts of northern and eastern Central Europe. Borders, administrations and even the country under which a particular town appears in modern databases can therefore be misleading when researching older generations.
Once I began looking at the places connected with the newly discovered branch, the geography was encouraging.
The line was running through northern European territory between Denmark and Prussia.
That made historical sense.
The Danish-Prussian borderlands are particularly awkward for anyone who wants neat modern national labels. Schleswig and neighbouring territories changed political control, and families could live in places that appear Danish in one period and Prussian or German in another without having moved anywhere.
So finding generations moving through that broad Denmark-to-Prussia world did not look out of place at all.
The names fitted.
The dates fitted.
The locations fitted.
And the relationships fitted the fragments of information I already had.
There were simply too many connections for me to dismiss what I had found as a random coincidence.
It had moved from “interesting search result” to “very strong research lead”.
Now came the slower job of proving it properly.
Why I do not keep my family tree on a website
This is also where my way of maintaining the family tree becomes important.
I do not keep my master family tree as a public tree on one of the genealogy websites.
My working database is in MacFamilyTree, which I have been using since version 5. I am now using version 11.
The genealogy websites are research tools for me. They are places where I can search indexes, discover other people's work, locate digitised records and find clues.
They are not my database.
This distinction means that finding something interesting on the Internet is only the first part of the job.
I need to bring the evidence home.
If I find a useful page, I normally preserve it as a PDF. If the page contains historical images or documents, I obtain local copies where this is permitted and practical. I record the URL, the source, names of contributors or contacts where relevant, and notes explaining why the material matters.
I want my research to remain intelligible even years later when a website has changed, a subscription has expired, a user has deleted a tree or a link simply no longer works.
There is also a practical consequence to working this way.
Properly recording everything takes time.
A lot of time.
Capture first, organise later
When I am in the middle of a productive search, I normally use a two-stage process.
The first stage is deliberately fast.
I record the person's name in MacFamilyTree, add enough information to reconnect that person to the developing branch, and attach a note containing the page or source where I found the information.
Then I move on.
I do this because stopping after every discovery to enter every date, event, citation, variation of a place name and supporting document can completely destroy the momentum of a search.
The second stage comes later.
I return to each person, inspect the source properly, collect the relevant documents, standardise names and places, add dates, build the source citations, compare conflicting information and decide which relationships are sufficiently supported to retain.
It is effectively a research queue.
Search ↓ Potential person found ↓ Record name + relationship ↓ Save source URL / note ↓ Continue searching ↓ Later: documents + dates + places + citations + verification
That Saturday night, however, there was a problem with the second stage.
I was far too excited to stop.
Just one more generation
Anyone who has seriously researched a family tree will probably recognise the trap.
You find one person.
That person gives you two parents.
One parent produces another marriage.
The marriage record gives you a place.
The place leads to another database.
That database contains another generation.
And suddenly the perfectly reasonable thought of “I should probably go to bed now” is replaced by:
Just one more.
By that point the bottle of New Zealand wine was beside me, historical records were accumulating on the computer, three previously disconnected family branches were beginning to form one structure and a line that had been largely hidden from me a few hours earlier was disappearing backwards towards the end of the seventeenth century.
Going to bed seemed completely unreasonable.
So I continued downloading and preserving digitised historical documents, correcting names, checking locations, recording sources and trying to understand what I had found.
Eventually I looked at the clock.
02:00.
So much for an early night.
The difference between finding and knowing
The work is not finished.
What I discovered that night gave me a structure that appears to connect three marriages, three previously separate branches and generations extending towards an ancestor recorded as born in 1699.
That is exciting.
But genealogy becomes dangerous when excitement is allowed to turn a plausible connection into an unquestioned fact.
Professional genealogical standards make essentially the same point: a sound conclusion requires broad research, accurate source citations, analysis and correlation of the evidence, resolution of contradictory evidence and a reasoned conclusion based on the strongest available material.
In other words, my job now is not to find another hundred names.
It is to verify the names I have already found.
The online trees have shown me where to look. The old information from my father gives me an independent thread connecting parts of the story. The historical documents I am collecting can now be used to test whether the marriages, births, deaths and parent-child relationships really connect in the way the trees suggest.
That is the less spectacular side of genealogy.
It is also the part that matters most.
Still, there are evenings when the search suddenly works — when people who have been nothing more than disconnected names begin acquiring parents, spouses, places and lives.
Saturday was one of those evenings.
At 22:00 I had a glass of New Zealand wine, a computer and a frustrating gap in my family tree.
At 02:00 I had three marriages connected, several new generations to investigate, a growing collection of historical records and one branch apparently reaching back to 1699.
Not bad for a quiet Saturday night.
AI did not establish the family relationships described here. Those remain part of my ongoing genealogical research and must be evaluated against the historical records and sources I am collecting.
Comments
Post a Comment