Watch accuracy and chronometer standards

A chronometer is not a style of watch. It is a watch whose movement has passed a defined test at an accredited laboratory and carries a numbered certificate to prove it, and Swiss law reserves the word for exactly that. Everything else printed on a dial, "precision", "superlative", "adjusted to five positions", is the manufacturer's own claim. It may be an excellent claim, but it is not a certification.

The short version. COSC tests uncased movements and allows a mean daily rate of minus 4 to plus 6 seconds a day. METAS tests the finished, cased watch and allows 0 to plus 5 seconds plus resistance to 15,000 gauss. Grand Seiko publishes tighter in-house figures that no third party verifies. And an ordinary quartz watch beats all of them.

Mean daily rate, and the four things that move it

A mechanical watch does not have one rate. It has a different rate in every position, at every temperature and at every state of wind. Every standard therefore measures a mean daily rate: the average gain or loss per day across a defined set of positions and conditions over a defined number of days.

This makes the commonest owner complaint meaningless. "My watch gained eight seconds yesterday" is not a rate; it is one sample of a noisy signal, taken in an unrepeatable mix of positions. Only an average over a week or more can be compared with a specification.

Four behaviors produce that average.

Positional variation. Held dial up, the balance staff pivots rest on their flat ends; held vertical, they rest on their sides, friction changes, and rate changes with it. Standards test positions because this is where watches misbehave.

Isochronism. A balance keeping the same period however far it swings. No real oscillator is perfectly isochronous, so as the mainspring runs down and the swing shrinks, the rate drifts. Hence "at full wind" in almost every specification.

Beat error. The balance should spend as long swinging one way as the other. If the hairspring stud or roller jewel is a fraction off center, tick and tock are unequal. Small beat errors barely affect the rate; large ones hurt self-starting.

Amplitude. How far the balance swings, in degrees. Not a rate specification, but the health signal underneath all of the above: falling amplitude, from aging oil, a tired mainspring or added friction, shows on a timing machine long before the rate goes obviously wrong.

COSC: the Swiss chronometer certificate

The Contrôle Officiel Suisse des Chronomètres is an independent Swiss association, founded in 1973, running laboratories in the watchmaking cantons. It certifies individual movements rather than models, and each passing movement gets its own certificate number. COSC certifies well over a million movements a year, the bulk for a handful of big names.

The procedure follows ISO 3159, the international standard for wrist chronometers with a spring balance. It runs for 15 days, in five positions (dial up, dial down and three vertical crown positions), at three temperatures: 46F, 73F and 100F (8C, 23C and 38C). The movement is wound daily and its rate measured. Broadly, the early days cycle through the positions at 73F (23C), the later days sample the cold and hot points, and the last day returns to 73F (23C) to see whether the movement comes back to where it started.

Seven criteria must all be met. For mechanical movements larger than 20 mm across, they are:

Criteria 2, 3 and 5 are the interesting ones, because they constrain consistency: a movement averaging plus 1 second a day by swinging wildly between positions fails even though its mean looks perfect. Rate resumption compares the end of the test with the beginning. Movements under 20 mm are allowed wider tolerances.

METAS Master Chronometer: testing the whole watch

The Master Chronometer certification is run by METAS, the Swiss Federal Institute of Metrology, and was developed with Omega, which launched it in 2015. It is open to other brands, but Omega remains by far the largest user.

Two things distinguish it. COSC certification of the movement is a prerequisite, so METAS sits on top of the Swiss test rather than replacing it. And all eight tests are run on the fully cased, finished watch, closing the gap COSC leaves open. In substance they are:

1. Function of the movement in a magnetic field of 15,000 gauss.
2. Function of the complete watch in a field of 15,000 gauss.
3. Deviation of the daily rate after exposure to 15,000 gauss.
4. Mean daily rate of 0 to plus 5 seconds a day, over several days, six positions and two states of wind.
5. Rate deviation between full wind and roughly a third of the power reserve.
6. Rate deviation across the tested positions.
7. Water resistance to the stated depth.
8. Power reserve against the stated figure.

The magnetic figure is a real number rather than a slogan: 15,000 gauss is the same order as a clinical MRI magnet, and it is achievable only with non-ferrous hairsprings and escapement parts rather than a soft-iron cage, as our guide to magnetism and watches explains. Note too that the rate specification is one-sided: a Master Chronometer may gain, but not lose.

In-house and regional standards

Grand Seiko

Grand Seiko does not use COSC. It publishes its own GS Standard and inspects to it in-house. For mechanical calibers the published figure is a mean daily rate of minus 3 to plus 5 seconds a day in static testing, over an inspection of around 17 days across six positions and three temperatures. A tighter Special standard applies to selected calibers, published as roughly minus 2 to plus 4 seconds a day, though the wording varies by model, so read the specification sheet for the caliber in front of you.

The static GS figure is genuinely tighter than COSC and uses six positions rather than five, but it is self-certified: Grand Seiko is both manufacturer and examiner. To its credit it also publishes a wider expected rate for normal wear, around minus 1 to plus 8 seconds a day, which is the more useful number for an owner.

The Geneva Seal and Qualité Fleurier

The Poinçon de Genève, or Geneva Seal, dates from 1886 and is administered by Timelab in Geneva. It is primarily a finishing and origin mark: the movement must be assembled, adjusted and cased in the Canton of Geneva and meet a long list of construction and decoration requirements. Since a substantial revision in 2011 it also tests the finished watch, with a rate criterion of no more than about one minute of deviation over seven days, plus water resistance and power reserve checks. Rate is a condition of the seal, not its purpose.

Qualité Fleurier, established in 2004 by Val-de-Travers manufacturers, is the most demanding on paper: COSC certification first, then the Chronofiable aging and shock sequence, then the Fleuritest, in which the cased watch is worn by a robot arm simulating a day of human movement and must return 0 to plus 5 seconds over 24 hours. Very few watches carry it, which is rather the point.

DIN 8319 and the Glashütte tradition

Germany has its own chronometer standard, DIN 8319, comparable to ISO 3159 in structure and tolerance. Testing was revived at the restored Glashütte observatory in Saxony in the mid-2000s with the Thuringian state metrology office, continuing an observatory tradition that ran from the 19th century through the German naval and astronomical trials for marine chronometers and deck watches. Certificates are rare, and testing is on the cased watch.

The certifications compared

How to measure your own watch

The seven-day wrist test

This is the only test that tells you how your watch behaves in your life, and it is free.

1. Set the watch precisely against an internet time source, using the hacking seconds function if the movement has it.
2. Each morning at the same time, to the minute, note the total deviation from that source in seconds. Not the change since yesterday: the total since you set it.
3. Wear and store the watch exactly as you normally would.
4. On day seven, divide the total deviation by seven. That is your mean daily rate.

Two habits make the result meaningful. Read at the same time each day, because a rate measured over 23 hours compared with one measured over 25 is noise. And never judge from a single day: a watch that gains four seconds one day and loses two the next is running at plus 1 second a day, which is excellent. If the weekly average sits inside the maker's specification, the watch is doing its job.

The timing machine

A timegrapher clamps a microphone to the case, listens to the escapement and reports rate, amplitude and beat error for whatever position it is holding the watch in. Used well it finds the cause of a bad rate in half a minute, which is what no amount of wrist testing will do. Used badly it misleads, because it measures an instant in one position while your wrist presents a changing sequence of positions, temperatures and winding states. Diagnose with the machine, judge with the seven-day test. What each reading means, what a healthy amplitude looks like and how a wrong lift angle setting can make a tired movement read as fresh are set out in our guide to timing your own watch.

Regulating versus servicing

Regulation means adjusting the effective length of the hairspring, or the inertia of a free-sprung balance, to shift the rate. It takes minutes, costs very little, and moves a consistently fast or slow watch to near zero.

What it cannot fix is inconsistency. If the rate swings widely between positions, collapses on day two, or comes with low amplitude, the cause is mechanical: worn pivots, dried lubricant, a poorly poised balance, a damaged hairspring or magnetism. Regulating around that changes nothing. A rough triage:

• Consistently fast or slow at healthy amplitude: ask for a regulation.
• Suddenly fast, often by minutes a day: suspect magnetism first, a seconds-long fix.
• Large positional spread or falling amplitude: a service is due.
• Fine when fresh, poor by day two: a torque or mainspring problem, not a regulation problem.

Quartz wins, and it is not close

None of this makes a mechanical watch competitive on raw accuracy. A standard quartz watch typically keeps time to around 15 seconds a month; a thermocompensated quartz movement, which measures its own temperature and corrects for it, is specified in seconds per year. COSC certifies quartz movements too, to a separate standard with a tolerance of a fraction of a second per day.

The best mechanical chronometer in the world is therefore beaten by a cheap quartz watch. Mechanical watches are worth owning for the mechanism, the craft and the longevity, as our quartz versus mechanical guide sets out. Read accuracy standards as evidence of how carefully a movement was built and adjusted, not as a claim to precision in absolute terms.