Why you do not need a chronometer watch
When I read the following review about a Chinese watch on Amazon, I became suspicious.
It was an attractive watch but for the price, something had to be wrong with it. Click on the photo to see the watch on Amazon.
The watch arrived, and it performed considerably less accurately than I would expect a Seiko automatic watch to perform on a timing machine. In the chart below (from the internet), the column that matters is the Rated Accuracy Range. Click on the chart if you want to see the full chart.
The Rated Accuracy Range is the maximum variation in timekeeping of the watch on the timing machine in different positions. For example, if the watch gains 35 seconds per day in the crown up position and loses 45 seconds per day in the crown down position, the RAR is 80 seconds, typical for a Seiko, Orient or Citizen. Comparable Swiss watches do somewhat better. Comparable Chinese watches perform less accurately on average. In most watches, the largest variation in timekeeping is crown down because the hairspring stud is on the other side of the movement and the variation is caused by hairspring sag.
For the price, something had to be wrong with it, and there was. The mechanism looked like a Citizen automatic watch, but there were differences. Pulling out the crown to the first position, the date mechanism would not change, so it was necessary to pull the crown out to the second position and move the hands around and around until the date was correct, as you would change the date on many older watches like the Rolex 1570. The other problem with the new Chinese watch was that it was essentially dry: it needed lubrication. Upon taking the watch apart to lubricate the jewels, I could see that the jewels were flat: no oil sink. I therefore used a thicker lubricant than usual on all bearings except the escapement, so that the lubricant would be less likely to drain away. Using a thicker lubricant would cause the timekeeping variation to increase on the timing machine, but I was not worried as this was a lower grade watch that did not even have oil sinks. Furthermore, the watch was difficult to wind before it was lubricated. The Rolex-style back of this watch made the watch difficult to close, so I filed two notches in the side so that I could close the watch with a regular watch case tool.
The mainspring provided plenty of power for the escapement, but it was too powerful for the automatic mechanism to work effectively: the watch stopped after 11 hours on my wrist. This was the biggest problem in this watch, and it can be observed in other watches. The automatic mechanism had small ball bearings under the rotor, similar to Seiko. Either a less powerful mainspring should be installed, or a heavier rotor. An easy solution to this problem is to wind the watch manually: four winds twice a day seem to be enough.
Placing the watch on the timing machine, I found that the watch gained a lot of time in the crown down position, so I adjusted the regulator so that the watch would lose 5 seconds a day in the face down position. I wanted the watch to lose some time, as I will explain shortly. How the watch performs on the timing machine is not important. What matters is how the watch performs on my wrist. I have seen watches gain time on my wrist after adjusting them to zero on the timing machine, like the Omega Speedmaster I once had, and I wanted this Chinese watch to lose a few seconds. My 5 second per day adjustment on the timing machine became 3 seconds in 12 hours in the chart below. However, the performance of a watch over a minute or two on the timing machine will often not be the same as the performance of the same watch over a 12 hour period. This watch had an RAR of 76 seconds in 12 hours in this chart, which is 152 seconds in 24 hours.
The timekeeping in a watch changes as the mainspring unwinds, such that the accuracy during the first 12 hours is much better than the second 12 hours. The hairspring in this Chinese watch proved to be surprisingly consistent in the face down position, showing almost exactly the same accuracy after 24 hours as it did when fully wound, suggesting that the error caused by isochronism was only about 1 second.
Things become more interesting when I compare this Chinese watch to a Rolex 1570 ("Superlative Chronometer Officially Certified"). I adjusted the Rolex so that it would lose 3 seconds a day face down on the timing machine. This watch had an RAR of 25 seconds in 12 hours in this chart, which is 50 seconds in 24 hours.
Adjusting either watch on a daily basis is simple. They lose time on my wrist and in the face down position overnight. When my watch is 15 to 30 seconds behind, I place it crown down overnight so it can gain some time. Repeat as needed so that the average error becomes zero. If you can get a Chinese watch to keep time with an average error of zero, then you do not need a Certified Chronometer watch for daily wear.
I cannot recommend this Chinese watch because of the many problems it had, but one of the Japanese brands would be better: Seiko, Orient and Citizen. Having said that, if you are looking for an expensive, high-status watch for special occasions, it might as well be a Superlative Chronometer. The finest watches have the Geneva Seal on the movement.
Once I could see how my watch performed on my wrist over several days of normal wear, it was time to adjust the regulator with the watch on the timing machine with the watch in the face down position, so that the watch would gain 14 seconds per 12 hours relative to the previous reading, or as close as possible. Afterwards, I hoped the timekeeping on my wrist would be accurate. The data is quite close for all positions, except for crown down.
I also updated the chart for the Rolex 1570 to include more positions for a better comparison.
In a test run, the Chinese watch was on my wrist during the day, and placed in the crown left position overnight, gaining and losing 5 or 10 seconds with an overall loss of 17 seconds in 7 days, or -2.4 seconds per day. I wound the mainspring four times, twice a day, during the test run. Then I placed the watch in the crown down position until the overall loss became zero seconds, after 9 hours. I expect that few watches could match this performance.