I'll preface this by saying two years ago I bought a Timegrapher. I have found it to be invaluable at understanding the 'health' of a watch movement. I highly recommend it for anyone who seriously collects watches. Its akin to an engine tester for a car. I got mine on ebay on sale for $130.

It measures static things like time accuracy in seconds/day, beat error, and amplitude. You can do this in 6 positions easily. It also visually shows you whats happening over time....ie "timegraphing". What it can't do is measure dynamic accuracy like when it is on your wrist day to day, but it does correlate well...ie better static accuracy relates to better dynamic accuracy.

What this led me to do is to define what level of accuracy I am willing to accept.

I used to accept as much as 30 seconds/day for a vintage watch(before relenting to a service) but now that I have the timegrapher and developed some basic skills to regulate a movement, I no longer have to accept that.

At this point, if I'm gaining or losing 10seconds/day, its annoying and I'll try to bring it in closer. I'm also finding that the watches I wear the most now are the most accurate ones vs the pretty ones.

What do you think is an intolerable level of accuracy?

 

I’m a little more liberal with gaining a few seconds per day than I am with losing an equivalent amount. My Globemaster is about +.5 spd, my SMPc is about -1 spd and my Speedy Pro is about +1.2 spd, all well within spec. If any of them were gaining or losing more than 5 seconds per day, that would get my attention, but losing time would bother me more.

 

I work for the government, so plus or minus 5 minutes a day is good by me.

 

that would involve me wearing the same watch more then 2 days...

 

I have a few watches that I privately think of as my"travel grade watches" that keep time to within a minute per week. I consider that to be quite good and choose one of them when going on longer trips.

Otherwise I'm annoyed only if a watch is off by a minute or more a day.

30 seconds per day? I'm like Foo2rama. I scarcely ever wear a watch more than a day or two, before rotating into wearing different one.

Here at the Home For Wayward Watches I'd have to leave some watches behind, never wearing them if I required 10 seconds per day out of them all. I love vintage watches. I enjoyed keeping up with their service and having them properly maintained. I enjoy not being OCD about watches' timekeeping capabilities though. I love to not to sweat the small stuff more than I love precision accuracy in every vintage watch on hand. So much of life has proven to be small stuff.

Now a rifle's relative accuracy at the range? That's different matter entirely.

 

As do I, and I have reappropriated your unused 5 minutes.

 

If a watch is in good condition I'm satisfied with 5 seconds +/- per day, but only for awhile. Sooner or later I just have to try to improve on that.
I got my old NOS Pobeda down to -2 seconds per seven day week, confirmed by a 28 day test. Since then anything less will get irritating after awhile.
If its a watch I wear infrequently for no more than a day at a time 10-15 seconds per day is okay. One I wear daily for weeks at a time and keep wound and ready while wearing something else gets checked against NIST every evening and more than 5 seconds off in 24 hrs is just not good enough.
Lately I've been experimenting with using positional error to bring the timing back in line overnight. It works fine with Swiss movements,like the 2453 of my Midland which loses several seconds when worn during the day but gains them back if left dial side up over night, but is not so consistent with vintage Citizen or Seiko hand winder designs (these being licensed copies in a Sona and a Paul Peugeot).
Perhaps its the relative position of the Balance wheel.
Temperature also appears to be a factor.
If a watch is kept in the same position at an even temperature and not worn then it can be adjusted to near perfection but then when worn outside in excessive heat or cold it will likely be off a bit.

 

Depends what i'm doing, at work I need more accuracy but if i'm lounging 5 mins per day is fine.

 

I’m with the OP on this one, more than 30 seconds per day I get a bit grumbly.

I currently have 7 vintage watches manufactured from about 1951 to about 1969. Between them, accuracy ranges from roughly 45 seconds per day to fewer than 3 seconds per day. Side note: I’ve been using the iPhone Timegrapher app with great success.

 

I can live with a couple of minutes a day. I don't own a watch for accuracy.

 

A great way to cost yourself a bunch of money in search of seconds per day accuracy is to buy yourself a timing machine. A watch that might average out to plus or minus 30 seconds per week and which would normally please you, suddenly become a source of aggravation once you see it on a timing machine. My best timekeeper is a 30 year old Rolex Date-Just with the 3035 movement which is within 15-seconds per month. I rarely wear it.

 

This just isn’t an obsession for me. If I begin to notice, its a problem.

 

Accuracy?

 

I too love my timegrapher though I find more interesting the ways it can indicate the health of a movement through amplitude and beat error. My threshold depends on the age of the watch - a minute per day is fine in a 1930s manual wind but not in a 1990s Rolex!

 

I love it when my frequently worn watches are within +/- 2 sec/day, making it off by only a minute per month (if I wear the watch for several days in a row or keep it on the winder to wear again within a few days). I really don't like to lose more than that, but if it's gaining less than 5 sec/day I can live with it.

Infrequently worn watches, that get rotated out after a day or two, can be within +/- 10 sec a day before it drives me nuts.

 

Speaking for vintage, under 1min a week I can live with for really old or worn watches, if it is under 30secs I'm happy.

Be advised that any old watch shown on a timegrapher can be a theoretical horror show and be consistently under 1min a week in real life... the timegrapher shows a performance snapshot, the actual timekeeping may be better. Of course, if it looks good on the timegrapher it usually is accurate, but you may be surprised sometimes.

Any new/modern watch over 5spd gets a ticket to the trash can.

 

When I consider the complexity of what goes into the design and manufacture of a mechanical watch I'm frankly amazed how accurate they can be. My daily wearer Speedy gains about 6 seconds a day which is what, one second every four hours, from a movement that was designed more than half a century ago? If I want total accuracy I just grab my phone.

Now accuracy on public transport, planes, trains, etc. - that's where it's important and I might get a touch cranky at times

 

Since all my watches are vintage, I'm just glad they run at all.

 

I have a friend who is an audio engineer (designs and builds amps & preamps). For fun we put a Dynaco ST70 (bargain kit tube amp back the day) that he had completely rebuilt and redesigned, on a scope to check its performance. Sine waves were decent to a point but the square waves were horrible as soon as it hit 12khz- just started rounding off (which was common in those amps regardless of how well overhauled).
I asked him what it meant and he said that in theory it means the amp should have horrible transient response, rolled off high end and be distorted at the top. But when lisenting it was superb, extended, shimmering on the top end.
He said he had tested amps that scoped perfectly but would sound lifeless and sterile- almost put you to sleep.
I know whatches and audio are apples and oranges, but sometimes just because it doesn’t test well doesn’t mean it won’t keep “decent” time and it isn’t enjoyable.

 

Audio is subjective to the listener, time is not. I have been an audiophile and EE for 30yrs. Worked for the best of them. The best I could ever advise someone is buy what sounds good to you. That's the bottom line... Test results be damned.

Time though is absolute. It is either right, or it is not.