The Man Who Spent Two and a Half Years Taking a Watch Apart

A wristwatch has been taken apart and stopped before any of it could land. The dial floats near the top of a block of clear epoxy. Under it the chapter ring, the date ring, the two halves of the case, then a descending column of bridges, wheels, springs and screws, each hanging where the assembled watch would put it, none of them touching. It is an exploded diagram you can pick up.

Fredrik Flornes Ellertsen built it. It took him two and a half years.

Why nobody had built one

Bartosz Ciechanowski's illustrated explanation of how a mechanical movement works is what pulled Ellertsen into watchmaking, and what he wanted afterward was that picture in his hand. He looked for one. What exists is resin cubes of watch parts sold as steampunk art, which is not a diagram, because a diagram's whole content is where the parts sit relative to each other.

Two things stood in the way. Building it takes a watchmaker, and a watchmaker is trained to make watches run. This is a one-way trip.

And it is punishingly hard. Fifty to a hundred parts, some under a millimeter, have to hang in mid-air in exactly the positions the assembled watch would give them, then survive a resin pour and having the air pulled back out under vacuum. Nothing may sag or rotate while the glue sets. Four prototypes failed. Each attempt costs about eighteen hours of tweezer work, and dropping one part means redoing everything above it.

Fishing line

The answer is 0.7mm monofilament fly-fishing leader, chosen because its refractive index is close to epoxy's, so it all but vanishes once cast. Nylon remembers its spool, so he bakes it across the grill pan of his oven at 150°C for an hour to straighten it.

Parts go on with cyanoacrylate, picked up on the flat of a sewing needle and left to wick into the joint. He points out this is the same motion as oiling a pivot, done with the exact opposite of a lubricant. Bridges hang from their own screws. Cap jewels and cover plates each get a short stud of line, stood upright on its square-cut end.

The failures taught more than the successes. A borosilicate cylinder refracted everything behind it, then shattered a week later, because resin shrinks as it cures. A mold a fraction too tall for the vacuum chamber left one model leaning, and that same cast dissolved the paint off the date ring and dragged it through the block as milky streaks. Clear lacquer sealed the paint but went yellow. He shipped the yellow.

Where it stops being a diagram

The balance wheel is not glued. It hangs free from its own hairspring, stretched out to show its shape, which meant casting the whole assembly upside down so gravity pulled the spring open. Everything else sits where the engineering says. The balance is placed to say this is what the rest of the watch exists to serve.

The movement is a PT5000, the Chinese clone of the ETA 2824-2. Ellertsen is direct about why: he could not justify €300 on a genuine caliber to guarantee it would never run again. That is the constraint on the whole genre. An exploded Datograph would cost a house deposit and show you nothing the clone does not.

The habit behind the object

Ellertsen is not a watchmaker by trade and is plain about it. "I've been learning to repair mechanical watches for a few years now," he writes, and keeps a public logbook as he goes.

The watches are recent. The habit is old. In March 2016 he got annoyed at the game 2048 and wondered how hard an algorithm to solve it could be. His verdict: "It turns out, very." Four days later came an alarm clock with nixie tubes and a daylight wake light, which he says cost him a great many good nights of sleep. Then an electrooculography rig reading eye movement to catch REM sleep and trigger a lucid dream with light, worked on until 2020 and left up unfinished. Then a device in the lid of a snus tin that gradually cuts your consumption, which is what you build when the problem is your own.

Then a gap, and in 2024 a coffee mug repaired with kintsugi, the Japanese practice of mending a break with gold instead of hiding it. Swap the gold for fishing line and the mug for a movement and you have the exploded view.

He offered the casting to Ciechanowski first, on the grounds it would never have existed without that blog post. It went to Adam Savage instead, with a note suggesting he mill the faces square and make it shine. Savage did, and found polishing epoxy takes far more work than it looks like it should.

So the picture all of us learned from is now an object you can turn over in your hands. It waited this long because it needed a watchmaker's hands, an engineer's tolerance for failed prototypes, and no professional stake in keeping the movement alive.