The Bench · N° 01

What the jewels actually do

Seventeen, twenty-one, twenty-five. The number on the movement is not a quality score, and an international standard exists because countries once charged import duty by the stone.

Hen & Mills Editorial · 9 min read · 9 August 2026

Turn a mechanical watch over and somewhere on the movement, usually in small print near the balance, you will find a number followed by the word jewels. Seventeen. Twenty-one. Twenty-five. Occasionally something faintly absurd, like eighty-three.

Nearly everyone reads that number as a quality score. Nearly everyone is wrong about how it works, and the real story involves a friend of Isaac Newton, a French chemist with a blowtorch, and an international standard written because countries were charging import duty by the stone.

Not a ruby, a hardness rating

Start with the definition, because it is stranger than you would expect.

ISO 1112, the international standard covering horological jewels, does not define a jewel as a ruby. It defines it as a "non-metallic movement component whose hardness is higher than HV 1 200". HV is Vickers hardness. The standard sets a floor and says nothing at all about what the stone is made of.

In practice it is almost always synthetic corundum, the mineral that gives us both ruby and sapphire. Corundum sits at 9 on the Mohs scale and measures roughly 1,800 to 2,200 on the Vickers scale, so it clears the ISO threshold with room to spare. Ruby and sapphire are the same mineral. The red one has chromium in it.

What the jewel does is narrower than most people assume. ISO 1112 says a functional jewel "serves to stabilize friction and reduce the wear rate of contacting surfaces". Not remove friction. Stabilise it.

That distinction is the whole engineering case. A steel pivot turning in a brass plate does not simply wear the hole oval over twenty years. It changes how much it drags as it wears, and a balance that drags differently this year from last year keeps different time. A jewel gives the pivot a bearing surface that barely changes across the life of the watch, so the watch you regulate at one year old behaves much the same at fifteen.

Made in a furnace, not dug out of a mine

The rubies in your watch were grown.

In 1902 the French chemist Auguste Verneuil announced a method for producing synthetic corundum. Finely ground alumina with a little chromium oxide is dropped through an oxyhydrogen flame burning above 2,000 degrees, and recrystallises below it into a single mass called a boule. He published the full details in 1904. The process is still in use, and it is regarded as the founding step of modern industrial crystal growth.

It is also cheap, which is why a modern movement can carry twenty-five stones without the stones being the expensive part. The cost sits in cutting, drilling and setting them to a few microns, not in the material going in.

So anyone telling you a jewel count reflects the value of the gems is describing a watch made before 1902. For two hundred years before that they had a point.

The problem Newton's friend solved

Jewelled bearings arrived in 1704, and the story is better than it needs to be.

Nicolas Fatio de Duillier, a Swiss mathematician and a close friend of Isaac Newton, worked out that friction between metal parts was what wore movements out and robbed them of precision. With the London watchmakers Peter and Jacob Debaufre he patented a way of piercing rubies so a steel pivot could turn inside one. The technique held the stone in a treadle lathe with sealing wax while a diamond-tipped tool drilled it.

The first bearings they jewelled were the balance staff, which is exactly where you would start. The balance is the part whose freedom to swing decides whether the watch keeps time at all.

English makers then kept the method to themselves for decades. Ferdinand Berthoud is recorded as the first watchmaker on the continent to take it up, in 1768. Sixty-four years is a long time to sit on an idea, and it is a useful reminder that the Swiss did not invent everything they are now famous for.

Where the seventeen actually go

Seventeen became the number that meant "fully jewelled" for a simple hand-wound watch. No standard mandates the arrangement. It is convention, and it goes like this.

• Balance staff: two hole jewels and two cap jewels, so four
• Roller jewel on the balance: one
• Pallet fork: two hole jewels and two pallet stones, so four
• Escape wheel: two
• Fourth wheel: two
• Third wheel: two
• Centre wheel: two

Seventeen. Every pivot in the going train that turns fast enough or carries enough load to wear, plus the stones the escapement uses to shove the balance along.

Look at what is missing. The barrel, which turns roughly once every few hours. The hour wheel. The ratchet wheel. The winding shaft. Slow or rarely moving parts do not wear in a way a jewel would fix, so putting one there achieves nothing except a bigger number on the dial.

Three of them are not bearings at all

Three of the seventeen are not bearings at all.

The two pallet stones sit in the fork and are the surfaces the escape wheel teeth actually strike. The roller jewel, often called the impulse jewel, is a small pin standing proud of the balance that the fork shoves against on every swing. Neither one is holding a pivot. Both are transmitting force, which is why ISO 1112 gives them a clause of their own covering "jewels contributing to the transmission of a force or movement", with pallets named as the example.

They take a beating for it. At 4Hz, a common modern rate, the escapement locks and releases 28,800 times an hour, every hour, for as long as the watch runs. Steel against steel would not survive that for decades without changing shape, and a changed shape means changed timing. Ruby does survive it, which is why those three surfaces have been stone since the eighteenth century.

Why the balance gets four and everything else gets two

This is the part almost nobody explains, and it is the most satisfying answer in the whole subject.

A hole jewel is not simply a ring with a hole through it. On the pivots that matter, the bore is olive-shaped, curved rather than straight-sided, so the pivot touches it along a narrow band instead of down the whole length of the hole. Less contact means less friction. Olive holes are used where it counts most: the balance staff, the escape wheel, the pallet arbor.

Then there is the oil. A watch runs on films of lubricant measured in fractions of a millimetre, and the hard problem is not applying the oil, it is keeping it where you put it.

The cap jewel solves that. It is a flat stone with no hole, sitting on top of the hole jewel and closing the end of the bore. The hole jewel is convex on the side facing it, so the two are only truly close near the centre, leaving a gap of roughly two to three hundredths of a millimetre. That gap is an oil reservoir, and capillary action, the tendency of liquid to creep into very narrow spaces, holds the oil around the pivot instead of letting it wander across the plate.

The cap jewel also takes the end-thrust of the pivot, which is why the balance staff needs one at each end while a train wheel does not. That is four stones on the balance, doing two different jobs, and both of them are necessary.

Why twenty-five is not better than seventeen

Add automatic winding and the count climbs, legitimately. A rotor needs a bearing, and the reversing and reduction wheels that turn its swinging into winding are in almost constant motion while the watch is on a wrist.

An automatic in the low twenties is therefore honest. Those extra stones are doing work a hand-wound movement never asks for.

What the number does not do is scale. Twenty-five is not forty-seven percent better than seventeen. Once every surface that needs a jewel has one, additional stones sit in places where wear was never going to be the problem. Past that point you are counting components.

One detail in ISO 1112 is worth knowing here. A jewelled unit built from several stones, such as a ball bearing carrying a rotor, "should be counted as a single functional jewel". A ball race with seven ruby balls is one jewel on the dial, not seven.

The customs loophole that needed a standard

Why does an international standard exist for this at all?

ISO 1112 answers that in its own introduction, and it is unusually blunt for a standards document. The number of functional jewels is a sign of quality, it says, and then: "The watch manufacturers use this fact to promote their products and some watch-importing countries use it as a basis for defining custom duties."

There was money in the number. Import duty keyed to jewel count, and marketing keyed to it as well. So stones went into movements with nothing to do, and dials started carrying counts in the high double figures.

Clause 5 lists what may not be counted. Jewels for ornament. Jewels capping a jewel hole without acting as a bearing, which the standard describes as oil chambers or dust protection. Jewels supporting the hour wheel, ratchet wheel, transmission wheel or winding shaft. Jewels holding date and calendar discs in place.

Clause 6 is a single sentence: "In sales literature and general advertising, and in marking the instrument, only the number of functional jewels and functional jewelled units shall be mentioned when describing the characteristics of the timekeeping instrument."

That sentence is what killed the hundred-jewel movement. If you see one on a vintage dial, you are looking at a watch made before the rules caught up, or one made somewhere the rules were ignored.

So what is the number actually worth?

Less than the dial implies, and it is worth being straight about that.

A count tells you the movement is jewelled where a movement ought to be. It cannot tell you how well the jewels are set, how round and how well polished the holes are, how the pivots are finished, whether the oil was applied properly, or whether the watch keeps time. Those are the things that separate a good movement from an adequate one, and none of them appear in the number.

Where it earns its keep is as a rough sort. Fewer than seventeen in a mechanical watch means somebody left a pivot running in brass. Seventeen to twenty-one hand-wound, or low-to-mid twenties automatic, is normal and tells you only that nothing is obviously wrong. A number far above that on a cheap watch deserves a second look, because ISO 1112 says every stone counted has to be doing something, and it is fair to wonder what.

Treat it as a checkbox. It confirms a movement is built the way movements are built. It was never able to tell you which of two watches is better, and the makers who lean hardest on it are usually the ones with least else to say.