So you have finally found ‘the one’ for sale – the watch you have been searching high and low for months. You have studied the photos online and determined that the case is unpolished with some light scratches consumate with its age. It has the rare dial you have been looking for and you have done your homework and know that this was made for one year only and the serial number is consistent with that year.
You’ve shopped the seller by checking the forums for negative reviews. You have even emailed the seller a couple of reasonable questions and requested a hi-res photo just to check they have the watch in hand and the seller proves responsive. The only nagging question on your mind is how to interpret that timegrapher screenshot the seller posted. What do those numbers and the graph mean? Is the watch a good runner? Has it really been serviced as the seller claims and if so, how well?
The Timegrapher is Your Friend
Hopefully you have already seen the tell-tale blue* screen shot of a timegrapher while browsing vintage watch ads. More and more sellers are starting to provide timegrapher results with their listings which, believe me, is a good thing. A timegrapher is a piece of horological equipment that can measure the vital statistics of a watch’s movement by listening to the sounds made by the movement as it runs.
* Not all timegraphers have blue screen but almost all the budget ones that are becoming popular do. Since that is also the model I have, these will be the screenshots used for the article.
The humble timegrapher has the ability to cut through all of the seller’s bullshit and really reveal the internal state of that unpolished, rare one year only watch you are seriously considering. They are not infallible and some deep-seated issues may be hidden or mis-diagnosed by the timegraphers, however for most watches, not much can hide behind that little blue screen. If the watch been correctly serviced, it should be very apparent. If it needs service, that will also be visible and if it’s a complete basket case, that will normally be very obvious.
At its core, a timegrapher is a sensitive microphone connected to an accurate internal clock and some logic to measure and compare the faint tick, tick, tick from your escapement against the internal clock signal and display the differences as a graph. Given a few more parameters it can also calculate how far your balance wheel is swinging and if the escapement lever is in the correct place.
Performing the Measurements
Measuring is straightforward: place the watch on the combined stand/microphone and the machine will start measuring. For a few seconds the timegrapher only listens, calculating the initial readings from the noise of the escapement. Opening the watch is not necessary since the timegrapher is able to hear the internals if the watch through the case. It is best to avoid a lot of background noise while using the timegrapher otherwise it may not be able to differentiate between watch and background. A cloth draped over the watch almost always attenuates the background noise enough to get a good reading. Once the beat rate of the movement has been calculated, the machine starts sampling properly.
Gimme the Numbers
The top line of the time grapher provides the numerical data. The timegrapher provides readings in numerical form as shown below, and also in a graphic, acoustical trace that often gives a better idea of the movement’s health. After all, as they say, a picture paints a thousand words.
From left to right, the first number is the current rate error expressed in seconds per day. The figure of +1s/d above indicates that the watch is currently running slightly fast and is expected to gain 1 second in the next 24 hours. Alternatively, a figure of -30s would indicate that a watch is running slow and will likely lose 30 seconds in the next 24 hours. A well-regulated watch should be within a few seconds a day and watchmakers typically set a watch to gain time when in error rather than lose it. An unregulated watch may be out by many seconds per day, while a watch that needs service or restoration may be out by many minutes per day, if it can even run that long.
Some positional variation is normal even on a well-regulated watch, especially a vintage watches that can have loosened up a little after many years of use. Positional variation means that the rate error will change depending on whether the watch is dial up, dial down or in a vertical orientation. Consequently, a watch that is regulated to multiple positions means that the rate error has been measured and minimized in some of these different positions. A good watchmaker should ask if you wear your watch on the right or left wrist since that will affect how they regulate your watch in the different positions. For example, a watch with the crown at 3, when worn on the left wrist will, on average, spend a significant period of time crown-down but almost no time crown-up.
The next number is the amplitude which is a measure of how far in degrees the balance wheel is rotating back and forth. The amplitude will be dependent on the type of movement as some movements are naturally high amplitude (low-300s) and some are low amplitude (mid-200s). For example many Swiss movements should have an amplitude above 300° when running well while a modern Seiko may be as low as 230° from the factory. In order to make a judgement based on the amplitude measurement, it is important to know what is ‘normal’ for the movement under test.
The amplitude will also be greater at full-wind than it will be after the watch has been running for hours. However, generally speaking, if you know the normal range for the movement, the amplitude reading can tell you how much torque is getting from the main spring to the balance and, therefore, provide a qualitative indication of the friction in the movement. A recently serviced movement will have a high amplitude for its range while a watch that needs lubrication will have a pitifully low amplitude that could be as low as 140° meaning the balance wheel is barely swinging at all (just 70°, less than a quarter turn, in each direction).
The number to the right of the amplitude is the beat error. The is the difference in milliseconds between the the time the escapement lever spends to the left compared to the time the escapement lever spends to the right. In other words it is the time difference between the ‘tick’ and the ‘tock’ of the movement. Ideally it should be 0.0ms meaning that the balance wheel swings exactly as far clockwise as it does counter-(anti-)clockwise. It is common to have a reading that is not quite zero which does not have much effect on the time-keeping. However, a larger number, for example more than 0.4ms, could indicate a poorly regulated watch and anything more than 0.8ms really could indicate a problem with the balance. A higher beat error will increase the difficulty in regulating the watch and will also likely give more positional errors – a big difference between dial up and dial down for example.
The final number at the right hand side commonly alternates between two parameters. The first is the beat rate of the watch in vibrations per hour and is normally calculated automatically by the timegrapher. Older vintage watches will typically be 18000 vph while modern watches will typically be 28800 vph. Lower end Seiko watches will be 21600 vph and hi beat movements will be 36000 vph. The beat rate can also be set manually if needed, but if it is needed, it generally means there is a major issue with the watch because the ticking is not regular enough for the timegrapher to establish the pattern.
The second number displayed is the lift angle which is an input parameter that the operator sets and indicates the angle between the banking pins between which the escapement lever travels. The value is dependent on the geometry of the movement and does not vary but is used to calculate the amplitude. In order to measure the amplitude accurately, the lift angle must be known and correctly set. For most vintage Seiko movements, it will be 54.5°, for many ETA and Sellita movements it will be 50°. Failure to set the correct lift angle can give a 10%-20% error on the amplitude and so can be used by unscrupulous sellers to bump up an otherwise low amplitude a little.
The Freshly Serviced Vintage Watch
Below is the trace from my recently serviced 1962 King Seiko at full wind. The numbers are indicating that it is running with a very low rate error of between 0 and +2 seconds per day. Not bad for a 59 year old watch! The amplitude is a very healthy 288° and the beat rate is measured to be 0.0ms. The numbers indicate that this is a very healthy movement.
The impression of a freshly serviced watch given by the numbers is reinforced when we look at the graphical trace. The line we see is generally level and and free of scattered points as each point lies on the line. The slight sawtooth appearance to the line indicates the beat error is not quite 0 but very close to it and is emphasized by the low resolution of the timegrapher screen. With a graph like this, we don’t really need to look at the numbers to get a good impression of this watch: it is good to go for several years with no reason to think about service right now.
A Recently Purchased Sellita
Below is the trace from my 2015 Squale Originale Opaco with its Sellita SW200 movement at full wind. The rate error is reported as 0s per day which certainly looks very impressive and the amplitude is also fine but look at the beat error.
The timegrapher is reporting a 0.4ms error with the watch in the dial up position. This is in itself is not a problem and it might indicate simply that the watch was not regulated from the factory. Alternatively, it may indicate the watch might have some positional stability problems which is more serious. In the Squale’s case, the error rate is within 2 seconds per day in all positions, which incidentally is better than COSC, even though the SW200 in the Squale is not the COSC-certified variant. In this case, the movement has never been serviced and was simply not regulated from the factory, so this beat error has probably always been there and can be adjusted out when the watch is finally serviced.
When we look at the graph we see a level trace indicating that the watch is keeping excellent time but now we see two traces whereas with the King Seiko above, we only had one. This is because of the beat error. Now that the there is an imbalance between the clockwise rotation and the counter clockwise rotation of the balance, we see the acoustic trace displayed as two lines. The greater the beat error, the greater the distance between the two lines displayed on the timegrapher screen. Overall, we can see that this movement is also in good health – accurate with good amplitude – and while the beat error is a little high, it is not affecting the performance of the watch.
So we have seen two good timegrapher traces, but what about a downright terrible one? My 1960s LeGant skin diver with an A. Schild Swiss movement was described by the seller as a good runner. Guess what? It isn’t, as we can see from the snow storm of its timegrapher trace:
Rather than being a good runner, this watch is basically not running at all. The trace is very noisy with dots all over the place meaning that there is a lot of extraneous noise coming from that movement. The watch is running so badly that the timegrapher cannot even calculate the rate error. The 0.0 beat error is not to be believed, the timegrapher simply has no idea of the correct beat of the watch so it cannot calculate the error of that beat. This watch badly needs a service and probably some new parts as well. Needless to say, if you ever see a trace like this in a for sale ad, run away.
The Modern Seiko
Below is the trace from a brand new Seiko 4R36 movement at full wind. The numbers are typical of a low-end, mass produced, unregulated mechanical movement. The +15 seconds per day error is within the manufacturers specification, which for the Seiko 4R36, is between +45 and -35 seconds per day. There is plenty of beat error as well with a calculated 1.1ms error. Amplitude is good, which is to be expected from a brand new watch. Remember Seiko’s 21600 vph movements tend to run with less amplitude than their Swiss counterparts – above 250° is good.
However, there is something the generally disappointing numbers do not tell us. Note how the the two lines of the Seiko trace look quite smooth and regular. That is a general indicator that the Seiko 4R36 is running well but simply needs regulation. As with most Seiko movements, the basics are very good and with a little regulation, this basic 4R36 can match the accuracy of movements many times more expensive. Pretty, the 4R36 is not; accurate, it can be, with regulation. Here is the same movement after ten minutes regulation, much better:
The eBay Vintage Buy
We have all been there I suspect… late at night, with maybe a drink in hand, we saw something interesting, the price was right and the Buy It Now button was calling our name. The description contained either the doubtful ‘it’s a good runner’, or perhaps the ever-popular ‘it keeps great vintage time’, or the downright suspect ‘it seems to run, but I have not timed it’. Despite the warnings signs, you clicked the button, and the watch is now winging its way to you.
Fact: I have never bought a vintage watch from eBay that did not need servicing… and that includes watches that were sold as having just been serviced!
So what can you expect from such a buy? Well, poor amplitude due to a long overdue service is probably a given. Hopefully, the watch will run for at least 24 hours between winds. Beyond that, you probably can’t really assess the quality of your purchase by simply wearing it. Step forward the timegrapher. Below is the trace from a Seiko 6139 ‘Pogue’ that I had persuaded myself, on the basis of photos alone, late at night, was the proverbial ‘diamond in the rough’
So what do we have? Well, the amplitude is low but not terrible, and the beat error is bad but again, not terrible. The watch is gaining quite a lot of time indicated by the steep angle of the trace but the lines are nice and parallel meaning there is not much wrong with this movement. The high positive error tells us the watch is running fast and that tells us the balance is swinging slightly too quickly backwards and forwards.
If you remember experimenting with pendulums as a child, you’ll remember that a short pendulum swings more quickly. So in this case, the effective length of the balance spring is probably a little too short. This could be because the watch needs regulation. Part of regulating watches without a freely spring balance (like the 6139 Pogue) is adjusting the effective length of the balance hairspring by moving the curb pin left or right. Another possibility is that some of the coils of the hairspring are conjoined either by old grease and dirt or by magnetism, both of which are reasonably easy to remedy, either by cleaning or by demagnetizing.
Some of you may be thinking that the low amplitude might have an effect on the error rate, but the rate is generally not affected until the amplitude is very, very low, which is not the case here. So I would first try regulating the watch and then if that proved difficult, I’d take a close look at the hairspring. However, the low amplitude indicates it’s time for a service anyway, since a freshly serviced 6139 movement should be in the 220°-240° range.
Real World Examples
So let’s have a look at some traces from real watch ads and work out what they tell us about the watch. The first image is for a Seiko 6139 from a seller on Chrono24. The watch is priced at a premium with box and papers, so what does the trace tell us? Well the beat rate of the movement is correct so that is a good start. The beat error is 0 and the accuracy is good, running a few seconds fast.
The only very small concerns are the curve of the line indicating that this movement has some wear and the amplitude of 204° is a pretty low when a freshly serviced 6139 should be at least 220°. The watch is sold as serviced and the lack of beat error and tight accuracy would lead me to believe it has been but not particularly well. So overall, maybe not serviced to the highest standards, given the amplitude, but it should be a reliable vintage buy. In this case, you are paying for the mint condition, included box and papers with an adequate service thrown in.
The next example is a vintage Omega Seamaster from another seller on Chrono24. The timegrapher is reporting 0 error which is a bit of a surprise, and perhaps a little suspicious, given that the trace is not flat. The beat error is quite large and I while I am not familiar with the movement, being a vintage Swiss movement, I would expect amplitude of more than 240°.
There is quite a lot of noise on the movement and a few random dots. My conclusion here is that the movement is basically ok for the age (from 1961) but the watch needs a service to clean and lubricate it. I doubt there is much wrong with this movement. To his credit, the seller is not claiming the watch has been recently serviced and I think from the timegrapher result, we can agree that it has not.
Our final example comes from a forum where a poster is asking whether he should get his 15-year old Rolex serviced based on the timegrapher results he posted:
Here we can see a fairly flat trace with very little noise which is always a good sign. There is some separation between the lines indicating some small beat error. The numbers confirm what the graph is telling us. The error rate is varying but is always close to +5 s/d which might even still be in spec for this model of Rolex. The overall beat error is calculated at a totally acceptable 0.2ms. Finally, the amplitude is calculated as 266 degrees which while lower than it would be after servicing, is still OK for a 15-year old watch.
There are two schools of thought on Rolex servicing… follow the recommended schedule or recognize that the movements tend to be workhorses and leave them alone something eventually fails. In this case, the watch is 15-years old and so will contain good quality modern synthetic oils. Such oils should hold their properties for much longer than the watch oils of the past. Therefore, given the numbers are good, if not very good, for a 15-year old movement and it will contain good quality oil, I would say servicing is not yet needed for this watch.
I have hopefully shown in this article how timegrapher results can be interpreted to help you make a more informed vintage watch purchase. I may have even encouraged you to invest $150 in a budget timegrapher of your own to start measuring your own collection. Models are available on eBay and Amazon, as well as watch parts suppliers.
Timegrapher results are not infallible and they can be fooled by seriously sick movements, but in the vast majority of cases, a basic timegrapher shows exactly the health of the movement. The actual numbers you see can vary with a lot of environmental factors so don’t necessarily expect to see exactly the same rate error, beat error or amplitude as the seller posted in the ad, however the graph will always give you a good qualitative assessment of the movement.