For my post #1,000 I'd like to share what I found to be some surprising metrics on the accuracy of automatics. I know -- data. Boring. But bear with me for a moment, maybe just long enough to take in the highlights if not the details of what I found.
I set out to take a quasi-scientific approach to gauging the performance of my collection, in the process acquiring multiple 24 hour samples taken on the wrist, at rest, and on winders over a five month period. While the sample of ten watches is small, it is diverse: micros like Aegir and majors like Panerai, Rolex and Omega are represented; six of the ten are COSC certified; some have been recently serviced, some never serviced, others are quite new; and movements are a mix of in house, standard ETA, clone ETA and tweaked ETA hybrids.
So what does this prove? One could argue very little, in that it is a very small sample. The optimist in me is pleased to note, however, that as an average automatics not only do better than expected in real-world timekeeping chores, but can produce stunning accuracy with movements that would be considered modest by most WIS.
I set out to take a quasi-scientific approach to gauging the performance of my collection, in the process acquiring multiple 24 hour samples taken on the wrist, at rest, and on winders over a five month period. While the sample of ten watches is small, it is diverse: micros like Aegir and majors like Panerai, Rolex and Omega are represented; six of the ten are COSC certified; some have been recently serviced, some never serviced, others are quite new; and movements are a mix of in house, standard ETA, clone ETA and tweaked ETA hybrids.
- In general, manufacturers' accuracy claims appear to be conservative. Average on-wrist accuracy was a little over 3 seconds per day, worn 24/7. Half of the watches measured 2 sec/day or less, and two -- the Aegir with a stock Soprod A10 and the Archimede with a standard ETA 2824-2 -- showed no distinguishable variance when compared to a GPS-synced control source.
- "New" may be more important than COSC when it comes to accuracy. The two best on-wrist performances were turned in by new, non-COSC watches. When both wrist and at-rest performances are averaged together, three of the four non-COSC watches finished in the top five.
- Service history is not directly correlated to accuracy. Even though they were very recently serviced and regulated, the two Rolexes performed only marginally better than a sixteen year old, never serviced PAM027 and were bettered by three of the four non-COSC movements.
- Watches are most accurate when worn - and run fast, not slow. Only two watches (Rolex GMT-II and the Ball) kept better time at rest than on the wrist. In all other cases the best times were recorded on the wrist. In only one instance -- when the Ball was timed at rest -- did a watch run slow. In all other cases, they ran exactly on time or fast.
- Winders skew results and may not be good for movements. Results are not shown below, but tests on a relatively cheap multi-watch winder gave wildly varying results. One excellent timekeeper, the Ball, refused to keep running on any winder setting. A Miyota 9015 movement on a watch not listed here kept excellent time on the wrist (+4 sec/day) but was worse by a factor of five when rotated clockwise per the manufacturer's recommendation. Careless use of a winder -- setting it to wind in the wrong direction, for example -- in several cases appeared to cause over-winding that could lead to premature wear on the movement. I'll buy a good winder for the Oris Moonphase, but the other watches will sit quietly from now on when not worn.
So what does this prove? One could argue very little, in that it is a very small sample. The optimist in me is pleased to note, however, that as an average automatics not only do better than expected in real-world timekeeping chores, but can produce stunning accuracy with movements that would be considered modest by most WIS.
| Watch | Movement | COSC? | Age | Service? | At Rest (sec/da) | On Wrist (sec/da) |
| Omega SMP | Cal. 2500 | Yes | 2013 | BNIB 04/15 | 2.50 | 1.50 |
| Rolex GMT-II | Cal. 3186 | Yes | 2005 | 12/14 | 3.00 | 6.00 |
| Oris Artelier | ETA 2688/2671 (Oris 581) | No | 2013 | No | 8.30 | 5.60 |
| Archimede Pilot | ETA 2824-2 | No | 2014 | BNIB 11/14 | 4.00 | 0.00 |
| Aegir CD-2 | Soprod A10 | No | 2014 | BNIB 02/15 | 7.00 | 0.00 |
| Panerai PAM027 | ETA 2892-A2/Soprod 9040 | Yes | 1999 | Unknown | 8.30 | 7.00 |
| Rolex 14060M | Cal. 3130 | Yes | 2006 | 07/15 | 5.80 | 5.26 |
| Panerai PAM523 | P.9000 | Yes | 2014 | No | 5.00 | 2.00 |
| Omega Speedy | Valjoux 7751 (Cal. 3606) | Yes | 2009 | No | 3.00 | 1.00 |
| Ball Conductor | ETA 2892-A2 | No | 2012 | LNIB 11/14 | 1.50 | 4.27 |
| Average | 4.24 | 3.26 |
