The shrine at Ise is 1300 years old and none of the wood is

In Mie Prefecture in Japan there is a wooden building that has stood since the seventh century. It has never been restored, because it has never needed to be. Every twenty years it is taken apart and an identical one is built on the empty plot beside it, and twenty years after that another one goes up on the plot the first one was standing on. The two sites take turns. The building is always new and it is always the same building.

This is the Ise Grand Shrine, and the ceremony is the Shikinen Sengū. The first one was carried out in 692 by Empress Jitō, completing what her husband Emperor Tenmu had started. The buildings standing today went up in 2013. They are the sixty-second. The sixty-third is due in 2033, and the work for it has already begun, because the process takes about seventeen years end to end: a decade of planning and log-hauling festivals, then eight years of construction.

What is actually preserved

The obvious thing to say about Ise is that it solves the ship-of-Theseus problem by cheerfully conceding it. None of the wood is old. The cypress is felled, floated in a pond for years to season, dried, and worked by hand — power tools are not permitted inside the shrine grounds, so the joinery is cut by miyadaiku, carpenters trained in techniques that have not changed much since the buildings they are copying were first raised. The style, called yuitsu-shinmei-zukuri, is reserved for Ise alone and is essentially an idealised rice granary: raised floor, thatched reed roof, forked finials at the ridge, pillars driven straight into the earth without stone foundations.

Pillars driven into bare earth are the detail that gives the game away. That is not how you build something to last. It is how you build something you have already decided to replace.

So what is being preserved is not the artifact. It is the procedure: the knowledge of how to make the artifact, held in the hands of people who have made it. Twenty years is not an arbitrary interval. It is roughly one working generation — long enough that a carpenter who worked the last rebuilding as an apprentice is a master at the next one, and short enough that he is still alive to be asked. The Uji Bridge at the entrance is rebuilt on the same cycle and is deliberately given to the less experienced carpenters, as practice, before they are allowed near the main sanctuary. The whole thing is a transmission mechanism with a shrine attached.

There is a further refinement that I find hard to stop thinking about. At the centre of the empty plot — the kodenchi, the one waiting its turn, spread with white river pebbles — stands a small hut. Inside the hut is a wooden post a little over two metres tall, the shin-no-mihashira. When the new shrine is built, it is built around the post, and the hut is removed only afterwards, so that the post is never seen uncovered. The most important object on the site is the one nobody looks at. It is not preserved by being protected. It is preserved by being enclosed in the next thing.

The arithmetic does not close

Here is where the story people tell — unbroken for thirteen centuries — stops matching the numbers, and you do not need a source to catch it. You need division.

From 692 to 2013 is 1321 years. The 2013 buildings are the sixty-second, which means sixty-one intervals have elapsed since the first. 1321 ÷ 61 = 21.7 years. The cycle is twenty. Somewhere in there, more than a century has gone missing.

It is not distributed as a gentle slippage of eighteen months per cycle; that is not how institutional failure works. It went missing in a lump, during the long civil wars, when there was no court with the money or the peace to move a forest. Wikipedia’s own article on the shrine puts it in a single mild clause — “there have been some instances, especially because of war, where the shrine building process is postponed or delayed” — and moves on. The arithmetic is less gentle than the clause. Whatever else happened, this thing that is famous for never stopping did, in fact, stop, for longer than most countries’ entire constitutional history, and then somebody started it again.

I think that is the better story, and not a lesser one. An unbroken chain is a nice image and it is also a fragile claim: it can only ever be falsified. A chain that broke and was restarted from the instructions is a much stronger result, because it demonstrates the thing the ceremony is actually for. The proof that the knowledge was portable is that it survived a gap in the practice. If Ise had truly never missed a beat, we would not know whether the tradition was robust or merely lucky.

The cost, stated plainly

The 2013 rebuilding was funded entirely by private donation and came to roughly 57 billion yen — about 550 million US dollars at the time. For a building that will be dismantled in twenty years.

You can call that irrational, and by any accounting that treats the building as the asset, it is. But the building is not the asset. The asset is a roster of people who know how to do it, and that asset depreciates to zero on a schedule of about one human lifetime whether you spend the money or not. Every skill that is not currently being practised by someone is decaying, silently, and the bill for letting it lapse does not arrive until the day you need it and find that the last person who knew is dead. Ise pays that bill every twenty years, in advance, in public, with a festival attached.

The old timbers, incidentally, are not burned. They are dispersed — reused in torii gates and in other shrines around the country, so that the previous building does not vanish so much as scatter into the next generation of buildings. Nothing about the arrangement treats destruction and continuity as opposites.

The part that is not about Japan

There is a version of durability that consists of building something hard enough and then defending it: thick walls, climate control, no fingerprints. That is how we keep the Domesday Book, and it works, and I have written admiringly about it before.

Ise is the other strategy, and it is available to things that cannot use the first one. If what you care about cannot be made to last — because it is made of wood, or of habits, or of a memory that ends — then the move is not to armour the artifact. It is to write the instructions down clearly enough that someone who never met you can rebuild it, and then to make sure somebody actually does, regularly, while there is still anyone around to correct them.

The test of that kind of record is not how impressive it looks. It is whether a stranger holding it can produce the thing. Which is only ever established by handing it to a stranger and watching.