233 treasure legends, 126 newspaper reports, one map

Every region has its buried treasure stories. Upper Austria has hundreds: the chest in the castle ruin, the abbey gold walled in by blindfolded masons, the three tubs buried in the wood “in wartime, a hundred years ago”.

The question I wanted answered is a dry one. Do these stories sit on anything? Not “where is the gold”, but: if you put the legends, the old newspaper reports of actual finds and the official registers of known sites on one map, do they line up more often than chance, and what happens to a good story when you check it?

This is desk research with agents. Nobody dug anywhere, and further down is the paragraph of law that says why nobody will.

Three kinds of evidence, kept apart

The map has three kinds of pins, and the whole point is that they never get mixed:

kind	what it is	where it comes from
documented	a site or find an authority has on record	open registers: the heritage office’s lists, the province’s history service (castles, vanished seats, earthworks, Roman sites)
reported	a find somebody wrote about at the time	digitised newspapers, 1850 to 1938
told	a legend	printed collections, mostly one from 1932

After two rounds it holds about 5,500 documented pins, 126 newspaper reports and 233 legends. Later layers add events (battles, fires, floods), the places where money moved (tolls, post stations, ferries, mints), Roman roads, and about a thousand place, farm and house names, most of them recorded around 1830, that remember something: Maut, Burgstall, Urfahr.

A fourth view, which I call the puzzle, lists only the places where a story or a report lies close to a documented site of a fitting kind. A treasure legend at a ruin that is in the register. A coin report next to a Roman road.

What the layers are worth

It would be easy to make this look better than it is, so here are the limits as they stand:

Newspaper text is machine-read Fraktur. It is good enough to find an article and bad at exactly the things that matter: names and numbers. Of 138 reports, 64 rest on the search snippet only. Every figure that counts has to be read on the page image.
Legends are placed by a reader. About a hundred sit at the main village of their municipality because the story says no more than that.
Most legends are stock motifs. Of the 233, only 29 name a time at all. Thirteen are hoards buried in a war.
The roads are conjecture. All 24 Roman road segments in the source are rated that way.
A name is a memory, not a find. Farms called Schatz or Gold usually had an owner by that name.
The rules for what counts as a “fitting” site are my first guess and were tuned against nothing.

So the map does not prove anything. It is a tool for choosing which story is worth an evening.

The best story, checked

The one I liked most: an abbey’s treasure, walled up by masons who were led in blindfolded. The prelate who knew the place died. In the version I first read, the wall was in a house in the provincial capital, and a newspaper of 1931 reports a jug of sixteen gold coins found inside a wall in the old town there. A legend and a find in the same town.

It took one night to take apart.

The first print says something else. The story was first printed in 1864. There, the wall is in the abbey. The capital only appears in a retelling of 1967. The match with the 1931 jug rested on a rewording a hundred years younger than the story.
The parish register settles the death. The retelling has the prelate die in the capital. His death entry of 1803 is in the scanned parish register. He died at the abbey.
The only clue left is a sound. One mason is said to have heard a clock going above him. No abbot of the period fits the whole story.

My estimate afterwards: about 85 % that nothing was ever lost, 10 % that something was hidden from the state and quietly spent a few years later, a few percent that it is still in the abbey. The house in the capital: zero.

A legend that does sit on a find

A ruin in the Salzkammergut has the standard chest-of-gold tale. In July 1881 a local paper reported that two sedan-chair carriers had found 36 Roman coins there, close under the surface. A second paper, four weeks later, has a conservator describing some of them, from Trajan to a son of Constantine, and supposes a Roman fort under the castle. (That second report I have only as machine-read text so far, not checked on the page.) The heritage register lists Iron Age and Roman finds for the site.

So this legend is next to a real find. But the find was made 145 years ago, it was common coinage, the site is a protected monument, and I do not yet know whether the tale is older than the find or grew out of it. That order is the whole question, and it is still open.

Where real hoards came from

Reading all the reports against all the legends, one pattern was hard to miss. The hoards that really turned up did not come out of castles and abbeys. They came out of ordinary houses and farmyards, by accident.

One example, ten kilometres from Linz: in October 1926 a pot came to light in a farmyard, 25 to 30 centimetres down, and broke. Seven gold ducats and 65 silver thalers, the latest coin from 1625. In December the coins were still with the finder. Where they are today, I have not found written down anywhere.

That is what a real one looks like: no story attached, a plain date, a depth, a coin list.

What the method learned

Go to the first print before anything else. Later collections shorten a tale and move it. The 1932 collection even marks the tales its editor only knew from print.
A legend near a find proves nothing until you know which came first.
The second newspaper report is the good one. The first says “coins were found”. The one that follows a few weeks later often has the list. Search six weeks past the first.
Two readers for every handwritten source, and a crop for every number. How that works is in Eight models read a letter from 1726.
A bot check is a stop sign. Two of the digital libraries answer scripted requests with a challenge. The agent does not work around it. I open those pages myself, or we find another copy of the same volume.
Negative results stay on the map. The abbey pin now carries a note: what the check found. A story that has been tested is worth more than one that has not, whatever the answer was.

The law, which decides the scope

In Austria this is federal law, the Monument Protection Act (Denkmalschutzgesetz), in the version in force since September 2024. It was checked in the federal legal database on the day the project started. In short:

Searching on the spot needs a permit from the federal heritage office: digging, on land or under water, metal detectors, systematic measurement (§ 10).
The permit is only given if a person with a relevant university degree leads the work. A hobbyist cannot hold it alone.
Chance finds must be reported without delay, and the spot left as it is for five working days (§§ 8 and 9).
Searching without a permit costs up to 25,400 euros, and the finds can be forfeited (§ 37).
The civil code gives finder and landowner half each (§ 399 ABGB), and takes the finder’s half away if the search was unlawful (§ 400).

Reading archives, newspapers, old maps and published laser terrain models at a desk is none of that. So that is where this project stays. If a case ever ends with a place worth looking at, the route is a dossier to the heritage office or to an archaeologist who can apply, not a weekend with a detector.

This is my reading of the statute, not legal advice, and it covers Austria only.

Why the map is not public

Most of what is on it comes from open data. I still keep it behind a login. A map that puts legends next to find spots is an invitation, and publishing the exact place of an unrecorded site is how sites get emptied. The same rule as in my shipwreck project: the method is open, positions are not.

If you need access to the map, for research or because you work on one of these places, ask me by e-mail: info@joseftot.com.

What I can say in public is the result so far. The stories are mostly stories. The finds are mostly accidents. And the interesting work is in the small set of places where a dated report, a register entry and the lie of the land point the same way.