The Shōsōin in Nara has a wooden floor raised two point seven meters off the ground, and that gap of air is the only reason ten thousand documents from the eighth century still exist.
And it has never burned. Not once in twelve hundred years.
Daniel's asking about the difference between a record we dug up and a record somebody never stopped keeping. He wants to know the oldest things humanity has held onto on purpose, not the oldest things we happened to find. The Vatican, the Shōsōin, the Torah scribal tradition. And then the harder distinction underneath it, which is that keeping the object and keeping the words are two completely different jobs.
Which they are. And the Shōsōin is the cleanest case of the first one, because it is a building that has been doing the same job since the eighth century with essentially the same method.
Same method being what, exactly? Keeping the door shut?
Partly, yes. The North Section was sealed under the imperial seal and basically stayed sealed. But the building itself is the trick. It's an azekura, a log-cabin construction, square timbers stacked and locked at the corners without bolts or nails. No fasteners means nothing to shear when the ground moves. It flexes. And the floor is raised two point seven meters, which lets air move underneath and keeps damp out of the storerooms. Everything inside sits in cedar chests, and cedar does two things at once: it buffers humidity swings and it blocks light and pollutants.
So the building is a climate control system that predates electricity by a thousand years.
Basically. And they've measured it. The repository's own account of its survival credits the limited access, the structure, and the surrounding environment. Those three things, in that order.
What's actually in there?
Over ten thousand hand-written documents, all from the Nara period, seven hundred ten to seven hundred eighty-four. Plus around nine thousand objects. It started as a dedication. In 756, Empress Kōmyō gave over six hundred items to the Great Buddha at Tōdai-ji, forty-nine days after Emperor Shōmu died. It wasn't one donation, it was five separate occasions. And then the institution just kept going.
Twelve hundred years. What actually tried to kill it?
The Siege of Nara in 1180. It's the only building to survive that. A lightning strike hit the North Section in 1254. The Great Buddha Hall burned in 1567, and the Shōsōin didn't. Custody moved from Tōdai-ji to the national government after the Meiji Restoration, and since the Second World War it's been with the Imperial Household Agency. The treasures themselves now sit in climate-controlled concrete repositories built in 1953 and 1962. The wooden building was designated a National Treasure in 1997 and a UNESCO World Heritage site in 1998.
So the original building is now a monument and the actual storage is concrete.
The original building did its job for twelve hundred years and got promoted to artifact.
That's the least sentimental thing I've ever heard a building do.
There's a wrinkle, though. The documents were rediscovered in the 1830s by an antiquarian named Hoida Tadatomo, and he cut them apart and reassembled them to study them. And then it happened again in the modern era. So the physical objects survived perfectly and the original order in which they were stored did not.
He had custody and he used it.
He did. Well-meaning, and the arrangement is gone.
Which is the whole problem in miniature, isn't it? The building never lost anything. The archive did.
And that's the distinction Daniel's pointing at. The Vatican is the other half of it.
The Vatican kept the institution and lost the paper.
Badly. The Church started assembling records in the first century. It was called the Holy Scrinium, or the Chartarium, and it travelled with the reigning pope. By 649 there was a permanent home at the Lateran Palace. By the eleventh century the holdings were split across three sites: the Lateran, St. Peter's, and the Palatine Palace. The oldest indirect evidence of papal registers, meaning bound tablets, goes back to the pontificate of Leo the Great, 440 to 461.
That's a continuous institution running for sixteen hundred years. And the records?
Complete papal registers survive only from Innocent III, who reigned from 1198 to 1216. Everything earlier survives in fragments or in later copies. The Catholic Encyclopedia's line is that nearly all the volumes of the papal regesta up to the end of the twelfth century have disappeared. Local warfare and fires in Rome.
So the papacy is older than most of its own archive.
Considerably. And the losses aren't abstract. When the popes moved to Avignon, moving the archive took twenty years, and the Ghibellines sacked it three times during that period, 1314, 1319, 1320. In the 1405 sack of the Vatican, papal registers were thrown into the streets. Then Napoleon ordered the whole archive transferred to Paris in 1809. By 1813 over three thousand crates had shipped, and one scholar's estimate is that between a quarter and a third of what went to Paris never came back.
A quarter to a third.
That's the estimate.
So we know the exact number of crates and not the number of documents, because the documents are gone.
That's what a loss looks like. You keep the shipping manifest.
What did the modern archive actually become?
Founded in 1612 by Paul the Fifth, and closed to outsiders until Leo the Thirteenth opened it in 1881. His line, from 1884, was go to the sources, that is why I have opened the archives to you, we are not afraid of people publishing documents out of them. Francis renamed it in 2019, from the Vatican Secret Archive to the Vatican Apostolic Archive. Today it's eighty-five kilometers of shelving. Thirty-five thousand volumes in the selective catalogue. About sixteen million pages from the Pius the Twelfth era being organized. And as of 2018, a hundred and eighty terabytes of digital storage, seven million images scanned.
Eighty-five kilometers of shelf and the earliest complete register is from 1198.
That's the honest shape of it. Institutional continuity is not document continuity. The Vatican never stopped being the Vatican. It stopped having most of its own paper.
Which raises the obvious question about access. If you seal something, does it survive better or does it just survive less visibly?
That's the Geniza question, and it's a third model entirely.
Tell me about the storeroom.
The Cairo Geniza. Around four hundred thousand Jewish manuscript fragments and Fatimid administrative documents, kept in the storeroom of the Ben Ezra Synagogue in Fustat, which is Old Cairo. Spanning the sixth to the nineteenth century. It's the largest and most diverse collection of medieval manuscripts anywhere.
And the mechanism was religious.
Hebrew script was considered sacred, so texts couldn't be destroyed even after they'd been used. They were deposited instead of discarded. Nobody was trying to build an archive. They were refusing to throw things away, for a thousand years, for a reason that had nothing to do with preservation.
That's the best-designed archive in history and nobody designed it.
It's the richest accidental archive there is. What's in it is the part that gets me. It includes the largest collection of early-medieval Islamic-world day-wage records. And evidence of double-entry bookkeeping in Old Cairo from the eleventh century, which predates the known Italian usage.
Double-entry in Cairo before Italy.
By the record. And Shelomo Dov Goitein indexed it. Roughly thirty-five thousand individuals, four hundred fifty professions, four hundred fifty goods, for the period 950 to 1250.
That's not a storeroom. That's a census.
It is, and nobody meant it to be. The Geniza stayed in use until Western scholars emptied it in the 1890s, and now it's scattered. Cambridge's Taylor-Schechter collection has nearly a hundred ninety-three thousand fragments. The Jewish Theological Seminary has about forty-three thousand. The Bodleian about twenty-five thousand. John Rylands, eleven thousand plus.
So the sacred prohibition kept the paper safe for a millennium and the moment it became valuable it got dispersed.
That is exactly what happened.
There's a pattern forming. The Shōsōin kept the building and lost the order. The Vatican kept the office and lost the paper. The Geniza kept the paper and lost the place.
And the Torah tradition is the fourth one, because it doesn't keep the object at all.
Keep the text, lose the scroll.
That's the whole strategy. The Masoretic tradition preserves information by copying it, over and over, with an increasingly strict set of rules about how. Emanuel Tov's argument is that the Torah's sanctity drove the scribal rules to get more and more stringent over time, but he also says something that cuts against the usual telling. The further back you go, the less the sanctity of the Torah influenced its textual transmission. The meticulous regime everybody pictures is a later development, finalized after the Talmud, in tractate Soferim.
So the famous precision was itself a historical event.
A late one. And we know that because of the Dead Sea Scrolls. Third century BCE to first century CE, found from 1946 onward. They give you the physical control. And what they show is textual plurality. Multiple versions of the same text circulating in the Second Temple period, before the text stabilized into what we'd recognize.
So the scrolls are the evidence that the standard wasn't standard yet.
Exactly that. The tradition that guarantees perfect transmission was a response to a real problem, which is that transmission wasn't perfect.
At what point does the copying tradition beat the storage tradition?
On information, almost immediately. The Masoretic tradition preserves the text to a level of precision the Shōsōin could never match, because every copy is a fresh chance to correct drift. The Shōsōin preserves the objects, and the objects carry things a copy can't: handwriting, paper, ink, the physical trace of who touched it. Neither one is superior. They fail differently. Copying fails by introducing error that nobody catches. Storage fails by catastrophe, and by scholars with good intentions and scissors.
The 1830s scissors. You know what I like about the copying model. It doesn't require anybody to survive. The building can burn. The institution can fold. The next copy doesn't care.
It's fully distributed. And that's why it beats everything else over the very long run.
All right, the actively useful ones. Because that's the part of Daniel's question I don't think gets asked enough.
Domesday Book. 1086. It's the survey William the Conqueror ordered, held in the royal Treasury at Winchester, then moved to Westminster, now at the National Archives at Kew. It was cited in the law courts in the Middle Ages and it still is. 1960, foreshore rights. 2010, sporting rights. 2019, a market.
A court in 2019 cited a survey from 1086.
To settle a market.
That's nine hundred thirty-three years.
And the medieval description of it is remarkable, because they already understood what it was for. Richard FitzNeal, writing around 1179 in the Dialogus de Scaccario, says that just as no judgment of that final severe and terrible trial can be evaded by any subterfuge, so when any controversy arises in the kingdom concerning the matters contained in the book, its word cannot be denied or set aside without penalty.
They treated it as a last judgment you could subpoena.
H. C. Darby called it the oldest public record in England and probably the most remarkable statistical document in the history of Europe. And it nearly didn't make it. It was evacuated to Bodmin Prison in the First World War and Shepton Mallet Prison in the Second. It's been rebound in 1320, in the Tudor period, 1819, 1869, 1952, and 1986.
Six rebinds. The book has had more repairs than most cars.
Every one of those rebinds is a decision to keep it usable rather than keep it untouched. That's the trade.
Finland's the other one I want, because it's the least glamorous and the most convincing.
The Finnish cadastre. The oldest Finnish land books date to 1539, in Tavastia. The origin is a 1524 Swedish parliament decision to create a land book, for fair tax collection. Farms were numbered by village from 1696. The register was completed in 1904 with a hundred twenty-one thousand six hundred eighty-four registered farms. Computerization began in 1968 and finished on the seventeenth of October, 1994. Seven hundred person-years and over fifteen million euros. Today it holds two point one million valid register units.
Four hundred seventy years of paper and then twenty-six years of punching it in.
And the point is that it never stopped being used. Nobody preserved it. They just kept needing it.
Cologne had something similar.
The Schreinsbücher. Legal transactions on houses and land, documented systematically from the first half of the twelfth century, kept in lockable chests they called shrines. The Cologne city archive still holds eighty-one Schreinskarten and five hundred thirty-five Schreinsbücher. And Austria's Franciscean cadastre was ordered by Emperor Franz the First on the twenty-third of December, 1817, and it's underpinned Austrian land registers for over two hundred years. The bicentenary was marked in 2019.
There's a theme in these that has nothing to do with preservation. Somebody's charging tax.
Property and tax. That's what keeps a record alive. Domesday was a tax survey. The Finnish cadastre was for fair tax collection. The Schreinsbücher were property transactions. They survive because people keep needing to know who owns what, and the state never loses interest in that question.
Iceland's the opposite case. Nobody needed it and it survived anyway.
The Arnamagnæan Collection, built by Árni Magnússon, who lived 1663 to 1730. Around three thousand items and fourteen thousand charters. It went onto UNESCO's Memory of the World Register in 2009. And the Flateyjarbók is the one people know. Written 1387, completed 1394. It survived the Copenhagen Fire of 1728 and the Second Battle of Copenhagen in 1807, and it was repatriated to Iceland in 1971. Under a 1965 Danish parliamentary ruling, one thousand six hundred sixty-six manuscripts plus Icelandic charters went to Iceland between 1973 and 1997.
Repatriated.
Which is the thing none of the other cases have to deal with. Every one of these archives eventually becomes a question of who it belongs to.
So we've got four models and none of them is clean. The Shōsōin has the strongest claim to unbroken physical custody of eighth-century documents and its documents were cut apart by a scholar who was trying to help. The Vatican has the longest institutional run and the thinnest early holdings. The Geniza kept four hundred thousand fragments by forbidding their destruction and got scattered the moment they mattered. And the copying tradition keeps the text perfectly and loses every original.
And there's no uncontested single answer to Daniel's question. The Vatican's continuity is real, its oldest holdings are fragments. The Shōsōin is the best claim we've got and it has a caveat.
What about the Chinese imperial archives? I'd expect a Ming and Qing continuity there.
I couldn't confirm that. I'm not going to assert it. The thread didn't get verified and I'd rather say that than hand you something I can't back.
Fair. Say nothing, then.
Nothing.
So what does a digital archivist take from any of this?
The one that keeps landing for me is that every long-surviving record was copied, rebound, evacuated, or digitized, repeatedly. There isn't a single case in any of these where the answer was durability. The Domesday Book has been rebound six times. The Finnish cadastre was computerized over twenty-six years. The Shōsōin's objects now sit in concrete repositories built in 1953 and 1962. The lesson is redundancy and migration, not a better medium.
Write once, store forever is a fantasy.
It's the fantasy that keeps selling, though. Every generation has a storage format that's supposed to end the problem.
And every generation's answer has a reader that goes extinct before the media does.
Which is the digital dark age problem. The bits survive and the machine that reads them doesn't. The Shōsōin's documents are ink on paper, and paper doesn't need a driver.
That's the thing the analog cases have that we don't. A ninth-century scribe could pick up a scroll from the eighth century and read it. Nobody needs a translator for the format.
Access is the other half of it, and it cuts both ways. The Shōsōin's sealed North Section protected the material and also hid it. The Vatican's rule kept records closed for seventy years after a pontificate, which protects people named in them, and also means the history gets written late, by whoever's still alive to write it. And the Geniza's prohibition on destruction was never intended to preserve anything. It accidentally produced the richest medieval archive in the world.
So the two things that saved the most material were a ban on throwing things away and a rule about sacred script.
Not a single professional preservation decision in either one.
I want to know what the Shōsōin's actual condition is. Twelve hundred years of cedar chests and two point seven meters of air.
Documents are fragile. The objects have held up remarkably. But the repository doesn't claim a secret. Its own account attributes survival to limited access and careful handling through history, the structure, and the surrounding environment.
Limited access and careful handling. That's it. That's the twelve-hundred-year technology.
The building is the technology.
What did the Vatican do that the Shōsōin couldn't?
Kept the institution running through total political collapse. The papacy survived the sack of Rome, the Avignon exile, Napoleon, two world wars. The archive didn't, but the office did, and the office is what eventually rebuilt the archive. That's the difference between an institution and a building. The institution can lose everything and start again.
Because the institution is the record.
In a sense, yes. The papacy is a continuously maintained thing, and the paper is downstream of it.
Herman, what do you actually think is the oldest continuously maintained record on Earth?
Honest answer is I don't know, and I don't think anybody does. It depends on what you count. If you count physical objects in unbroken custody, it's the Shōsōin. If you count information transmitted continuously, it's the Torah, and the chain is older than every object that carries it. If you count a document people still use in a court of law, it might be Domesday.
Three different answers to one question.
Because custody, transmission, and utility are three different things and Daniel's asking about all of them at once. That's what makes it a good question. You can't answer it without deciding which one you mean.
I'll take the copying tradition. Not because it's older, because it doesn't need anything to survive. The building can fall down.
What's your problem with the building?
Nothing. I have a problem with anything that requires somebody to keep showing up.
That's a strange position for the mammal that sleeps eighteen hours a day.
I show up eventually.
Corn, when you said the records survive because somebody keeps needing to know who owns what, how long does a land dispute have to sit before it stops being a dispute?
Legally? Depends on the jurisdiction. Practically, apparently nine hundred years, if it's a market in England.
I've got four of them in a cupboard in the hallway. Cadastral maps. My family's plot outside Tralee, my mother's people's land in County Clare, two more from a solicitor who died in the middle of a sale. Twelve sheets. The Tralee one has a boundary marked in red ink and a note in the margin from 1953 that says the stream moved.
The stream moved.
Four meters east. So the boundary on the map and the boundary on the ground stopped agreeing at some point in the fifties and nobody's fixed it since, because nobody's needed to. The land's been in the family and it's not being sold, so the map sits in the cupboard and the stream does what it likes.
So the record and the reality have been out of sync for seventy years.
That's right. There's a man in the Land Registry in Dublin named Fintan who I've been writing to for eleven months about it. He keeps sending back the same form. I keep sending it in with the note about the stream. He's been very patient about it. He sends a form, I send a form, and neither of us has moved the stream.
You've got a correspondence going.
Eleven months of it. I've had longer. The last letter he sent had a sticky note on it that said, and I'm quoting, the map is the record.
That could be on a wall.
It's on a wall. In my hallway. I put it there.
What are you actually going to do with the twelve sheets?
I'm going to get them photographed properly. There's a man in Kimmage with a copy stand and a light box who does manuscript work for a couple of small libraries, and he charges forty euro a sheet, which is too much, and he knows it's too much, which is why he does the last one free. I'm taking them over on Thursday. Twelve sheets. He asked me what they were for, and I said the same reason you'd photograph anything, so that when the paper goes the record doesn't.
Eleven months of forms and the actual answer was a man in Kimmage.
Well, Fintan's not going to photograph them. When I was doing the secondhand books I had a customer walk in with a ledger from a butter market that ran from 1901 to 1974 and his grandmother had written every entry by hand, and I sat and looked at it for about ten minutes and I said to him, this is a very good record, and he said, I know, and neither of us bought or sold anything. I still think about that ledger. When the paper goes, the record doesn't.
That's the whole thing, though — the photograph is a copy, and a copy is exactly what the Masoretic tradition decided was enough.
A copy's fine. The stream keeps moving.
Hilbert, you've had those maps in the hallway the entire time we've been talking about archives and you didn't mention it once.
I was waiting to see if you'd get to the stream.
The Shōsōin has been getting two point seven meters of airflow for twelve hundred years and your boundary has been wrong since 1953, and both of those are archives that work.
One of them's going to Kimmage on Thursday.
So we've got a man photographing land maps from Kerry at forty euro a sheet, and we've been discussing the Masoretic transmission of the Torah for twenty minutes. Those are the same problem.
Every archive I've heard about tonight is a stack of quiet disagreements about which version is the real one.
And the record is the version everybody agreed to stop arguing about.
Which is all a land registry is. A place where the arguing stopped.
The stream didn't stop, though.
The stream never stops. That's why the map's in the hallway and not in the field.
The one thing I'd say about Thursday, Hilbert — if you're going to photograph them, shade the room. Light is the thing that does the damage. The Shōsōin's cedar chests were doing that job for twelve hundred years.
The light box is on the camera side. The sheet never sees it.
That's right. For a man who keeps maps in a cupboard, you've thought about this.
I've had a long time to think about it.
So Daniel's question doesn't have one answer, and I think that's the honest landing. There's no single oldest continuously maintained record. There's a building in Nara that has never burned, a papacy that lost most of its own paper and kept going, a storeroom in Cairo that saved four hundred thousand fragments by refusing to throw them out, and a copying tradition older than every object that carries it.
And the thing all four have in common isn't a medium. It's somebody deciding, at every point where it could have stopped, not to stop. The Shōsōin's North Section stayed sealed for centuries because a succession of people agreed to keep it sealed. Domesday got rebound six times by people who chose to keep it usable. Finland spent seven hundred person-years typing a land register in because nobody was willing to be the generation that dropped it.
Hilbert photographed his just in case. That's the same decision at small scale.
And the copy outlives the paper. That's the Masoretic bet, a thousand years old, still paying.
The one thing I'd want a digital archivist to take from tonight is that the record that survives isn't the one that was stored well. It's the one that kept getting used. Every single case we've talked about, somebody needed it. Tax, land, liturgy, a court case about a market. The Neanderthal cave painting nobody needs doesn't survive. The land map that decides who owns the field does.
And the moment the need stops, the archive stops. That's the risk nobody wants to say out loud. A digital repository with no users isn't an archive. It's a shrine with a bad power supply.
Not the image I expected at the end of this, but I'll take it.
Before we go, the producing on this one was Hilbert Flumingtop, and we'll leave it there.
If you enjoyed this one, try episode ten thirty-two, Ancient Backups; episode eleven seventy-seven, The Race Against the Digital Dark Age; and episode eleven seventy-six, Can AI Resurrect the Digital Tombstones in Our Archives. This has been My Weird Prompts. Send us your own prompt on Telegram at t dot me slash MWP listener bot, or find the whole archive at my weird prompts dot com. We'll be back soon.
See you tomorrow.
