Twelve Dollars, One Kinked Spiral, and Fifteen Teeth Untouched

Twelve dollars. That’s what the seller wanted for the pocket watch in the tray of broken clasps and single earrings. The case was scratched, the crystal cracked, and when I shook it, nothing moved inside. “Parts only,” she said, which is antique-market code for this thing is dead and we both know it.

I bought it anyway. Not because I have any romantic attachment to pocket watches — I don’t — but because I’ve spent the last month printing escapements in PLA and wondering what the real thing looks like up close. My orrery ticks. It advances Mars by one simulated day per swing of a plastic pallet. But it’s a loud, clunky mechanism compared to the Swiss lever escapement that’s been standard in watches since 1754. I wanted to see the gold standard, even if the gold was just gilded brass and the standard was broken.

The case back pried off with a fingernail. Inside: a movement covered in fine dust and a faint smell of old oil. The balance wheel sat motionless. No tick, no hum, just the corpse of precision.

Taking It Apart

Watchmakers spend years learning disassembly sequences. I spent forty minutes watching YouTube and taking notes. The order matters: hands first, then dial, then barrel bridge, then the wheel train, then the balance cock (the small plate holding the balance wheel), then the balance itself. Get it wrong and you’ll launch a hairspring across the room.

The hands came off with a hand-puller I borrowed from fountain pen restoration — same principle as the nib extractor, just smaller leverage. The dial lifted out after removing two tiny screws I could barely see and definitely couldn’t have threaded back in without magnification. Under the dial: the motion works, the gears that translate the movement’s rotation into hours and minutes. Nothing obviously broken, which meant the problem was deeper.

Removing the barrel bridge exposed the mainspring. Even wound down, a mainspring stores dangerous energy. Watch repair forums are full of cautionary photos: deep cuts, embedded steel ribbons, blood. This mainspring was fully unwound — had been for years, probably — so I unscrewed the barrel cover without incident. Inside: a coiled steel ribbon about twenty-five centimetres long and maybe 0.1 mm thick. It stores walking-around energy. Wind it tight, and this thin strip of metal powers the movement for thirty-six to forty hours, slowly relaxing through a gear train that converts torque into ticks.

The mainspring was intact. Not the culprit.

Seventeen Jewels, or Maybe Fifteen

The movement’s stamped with “17 JEWELS” in an optimistic font. I counted fifteen visible bearing points with synthetic ruby centres — two for the balance wheel, two for each of the four wheels in the gear train, and two more for the pallet fork. Somewhere in there are supposed to be cap jewels and an impulse jewel to make seventeen, but either I’m missing them or the manufacturer was optimistic with their count.

Jewel bearings were invented in 1704 by three men who shared an English patent, and until Auguste Verneuil figured out how to synthesize corundum in 1902, they were made from real ruby and sapphire. Synthetic or not, they reduce friction from 0.35 (brass on steel) down to about 0.10. That matters when you’re trying to run a gear train from a spring that weighs less than a gram.

The escape wheel has fifteen teeth. I know this because I counted twice after watching a repair video where someone destroyed a movement by gripping the escape wheel wrong and bending a tooth. The teeth interact with ruby pallets at precisely cut angles — eleven to fifteen degrees of “draw” that make the escapement self-locking. Damage a tooth and the timing changes. There is no undo.

I did not touch the escape wheel with tweezers. I did not touch it at all.

The Hairspring Problem

The balance wheel, when I finally got it out, was frozen because its hairspring had kinked. Not the subtle kink of manufacturing tolerance — a hard bend, maybe from a drop, that put a right angle into a spiral that should be smooth. Hairsprings are temperamental. They’re also why watches got accurate: before the balance spring was added around 1657 (by either Robert Hooke or Christiaan Huygens, depending on who you believe), mechanical watches drifted by hours per day. With the spring, error dropped to about ten minutes. Minute hands appeared on watch dials only after this improvement. Before 1680, there was no point displaying minutes because the mechanism couldn’t track them.

I could try to straighten the kink. Watchmakers do, using tweezers and patience and a level of fine motor control I haven’t developed in sixty hobbies. The smart thing is probably to source a donor hairspring — or a whole new balance assembly — and swap it in. But then I’m doing watch repair, not watch understanding, and what I wanted from this twelve-dollar corpse was comprehension.

I spent an hour just looking at the pallet fork under a loupe. Two ruby pallets, entry and exit, mounted on a lever that swings a few degrees with each tick. Every beat — and a watch running at 21,600 beats per hour produces six per second — releases exactly one tooth of the escape wheel while giving the balance wheel a tiny push to keep it swinging. The orrery I printed uses the same principle, but watching the real thing, even a frozen real thing, shows me how much I was approximating. My PLA pallets are chunky, my tolerances measured in tenths of millimetres. This watch measures in hundredths.

The parts sit in a jar lid on my desk now. Barrel, mainspring, wheel train, balance assembly with its kinked spring. The case went back in the drawer. Tomorrow I’ll research hairspring sources, or maybe I’ll find another dead watch to dissect. There’s a satisfaction in taking something apart when you already know you can’t put it back together — no pressure to succeed, just permission to learn.

The twelve-dollar curriculum continues.