COSC vs METAS vs in-house accuracy standards

COSC certifies a bare movement to a mean daily rate between 4 seconds slow and 6 seconds fast. METAS certifies the finished, cased watch to an average daily precision of 0 to 5 seconds fast, and adds magnetism, water resistance and power reserve checks. Brand standards are the manufacturer testing its own assembled watches, often to a tighter figure than either, on its own authority.

The number is the part everybody quotes and the least interesting part of the story. What matters is what was inside the timing machine when the number was measured.

What the COSC test actually involves

COSC applies the test regime defined in ISO 3159, and it is more thorough than the single headline number implies. Each movement runs for 15 days, in 5 positions, at 3 temperatures: 8, 23 and 38 degrees Celsius (roughly 46, 73 and 100 Fahrenheit). Rates are read daily and seven separate criteria are computed.

The famous one is the mean daily rate, which must fall between minus 4 and plus 6 seconds a day. The other six are about consistency rather than average, and they are the reason certification is not trivial to pass:

• Mean variation in rates across positions, capped at 2 seconds a day.
• Greatest variation in rates between any two consecutive days in one position, capped at 5 seconds.
• Difference between horizontal and vertical rates, which must fall between minus 6 and plus 8 seconds.
• Largest deviation of any single measured rate from the mean, capped at 10 seconds.
• Thermal variation, capped at 0.6 seconds a day per degree Celsius.
• Rate resumption, comparing the final rate to the early average, capped at 5 seconds.

A movement can therefore have an excellent average and still fail, because it is inconsistent. That consistency requirement is what a chronometer certificate is really buying you, and it is worth more day to day than the average is.

Two limitations follow from the design. The test is run on the movement, held in a test fitting, before it meets its final case, dial and hands. And in Switzerland the word chronometer is legally protected, so a watch cannot carry it without having passed an official test, which is why the term means something specific rather than being marketing language.

What METAS adds, and why it is a different kind of claim

Master Chronometer certification is run by the Swiss Federal Institute of Metrology, the national metrology body, and it was introduced in 2015. It begins where COSC ends: the movement must already be COSC certified, and then the finished watch goes through a further set of tests over ten days.

The headline criterion is an average daily precision of 0 to plus 5 seconds a day, measured on the cased watch across six positions and at two states of wind, typically full and around a third remaining. That second condition is a real-world detail COSC does not test: it checks that the watch still keeps time when the mainspring is well down, which is where mechanical watches usually time worst.

The magnetic test is the part that separates it from everything else. The watch must function and keep to specification after and during exposure to a field of 15,000 gauss, which is 1.5 tesla. For scale, the long-standing antimagnetic threshold in ISO 764 is 4,800 amperes per meter, about 60 gauss. The METAS requirement is not a marginal improvement on that, it is a different order of magnitude, and reaching it requires silicon or other non-ferrous hairsprings and escapement components rather than a soft-iron shield. Our guide to magnetism and watches covers why that matters in a world full of magnetic clasps and speaker magnets.

METAS also verifies the water resistance and the power reserve of the finished watch, so the certificate covers claims that are otherwise made purely by the manufacturer. The certification is open to any brand, not just the group that developed it, though uptake outside that group remains limited.

Brand standards: better than they sound, worse than they look

House standards get dismissed as marketing, which is lazy. Several of them are genuinely stricter than COSC and, crucially, are measured on the assembled watch.

Rolex certifies its movements with COSC and then runs its own test on the cased watch, publishing a Superlative Chronometer tolerance of minus 2 to plus 2 seconds a day. That is a tighter window than COSC, on a harder test subject, which is a real engineering claim.

Grand Seiko tests to its own published standard, over a longer period and in more positions than COSC uses, and quotes a static mean daily rate for its standard mechanical calibers, with tighter figures for its higher grades. It also does something unusually honest: it publishes a separate expected rate for the watch when worn, which is wider than the static figure, because a watch on a moving wrist behaves differently from one in a timing machine.

Patek Philippe abandoned external certification for its own seal, which covers finishing, materials and long-term servicing alongside a rate tolerance measured on the cased watch.

The weakness is structural rather than technical. These tests are run by the party with the commercial interest in the result. There is no external audit, no published pass rate, and no independent recourse if your individual watch does not meet the figure. You are trusting the brand's reputation, which in these cases is generally well earned, but trust is what it is.

What none of them promise

Every one of these standards measures a watch under controlled conditions in a laboratory. None of them measures a watch on a wrist, and the difference is not small.

On the wrist, the watch sits at skin temperature for most of the day and room temperature at night. It spends time in positions the test never weighted the same way, gets shocked when you close a car door, and runs at whatever state of wind your activity produces. The distribution of positions overnight, which is entirely down to how you leave the watch on the nightstand, can shift a week's total by many seconds on its own.

Certification is also a moment in time. It says nothing about the watch three years later, after the lubricants have aged or after a magnet has done its work. A certified chronometer that now gains 30 seconds a day is not a fraud, it is a watch that has been magnetized or needs service. Our guide to timing your own watch explains how to tell the difference at home, and the accuracy calculator will turn a week of observations into a daily rate.

How to use this when buying

Ask three questions in order.

Was the watch tested, or the movement? A cased-watch figure is a stronger claim than a movement figure at the same number of seconds, because it accounts for everything that happens during final assembly.

Who ran the test? COSC and METAS are independent of the seller. A brand standard is not, which does not make it false but does change what the claim is.

What else does it cover? COSC covers rate only. METAS covers rate, magnetism, water resistance and power reserve. Brand programs vary and you have to read what they actually say.

Then weigh it against the premium. A COSC-certified version of a movement costs a little more than the uncertified version. Master Chronometer certification and the components that make it possible cost more again. If precision alone is the goal, note the uncomfortable arithmetic: an ordinary quartz watch at any price is typically within 15 seconds a month, which comfortably beats every standard on this page.

The mistake people make

The mistake is comparing the seconds figures directly, as though minus 2 to plus 2 were simply better than minus 4 to plus 6 in the same units. They are not the same measurement. One is a cased watch, one is a loose movement, and the tolerance windows are computed differently on top of that.

A related error is treating certification as a guarantee of daily performance. It is a factory test result, valid at the moment it was taken, and it comes with no warranty about your wrist. Plenty of certified watches drift outside their certified window in normal ownership, and plenty of uncertified ones sit within a couple of seconds a day for years.

The last mistake is paying a large premium for a certificate on a watch you will wear casually. If your watch is off your wrist half the week, position and winding effects will swamp the difference between these standards long before the standards themselves matter. Buy the certificate if you want the engineering it reflects, or if the magnetic resistance solves a real problem in your life. Do not buy it expecting your watch to be right at the end of a month.