You need three things, and you almost certainly have two of them already.

  1. The moment you last set the watch to the correct time.
  2. The moment you checked it afterwards.
  3. How many seconds ahead of the clock, or behind it, the watch was when you checked.

The third one is the only one people find fiddly, and it is simpler than it sounds. Put the watch next to a clock you trust. If the clock says 10:00:00 and the watch says 10:00:08, the watch is eight seconds ahead. That is the number.

Step 1
Step 2
Step 3
seconds

If the clock read 10:00:00 while your watch showed 10:00:08, the watch was 8 seconds ahead. Count the difference between the two, not the time on either.

Ahead is not the same as fast

Worth separating, because the two get muddled and this page depends on the difference.

Ahead and behind describe where the hands are at one moment. Your watch is eight seconds ahead of the clock right now. That is what you type in.

Fast and slow describe how the watch runs. It gains four seconds every day. That is what comes out.

Being eight seconds ahead tells you nothing on its own, because it matters enormously whether that took two days or two months. The rate is the thing that travels.

What the answer means

The number is your watch’s rate, in seconds per day, usually written spd. Positive means it gains, negative means it loses. A watch at +4 is running four seconds a day fast, and after a week it will be roughly half a minute ahead.

That is the whole arithmetic, and there is nothing stopping you doing it on paper. Take how far the watch had drifted, divide by how many days it took to drift that far, and you have seconds per day. The calculator is only saving you the division.

If you never got the watch set exactly, that is fine too. Any two moments where you know how far off it was will do; the method does not need a zero to start from, only two known figures and the gap between them. How to measure watch accuracy without a timegrapher covers that version.

Why it refuses under twelve hours

A rate is a comparison between two moments, so it inherits whatever was imprecise about either. Read the dial half an hour after setting it and any small misreading gets divided by a very small number of days, which turns a fraction of a second into a rate of dozens.

Twelve hours is about the point where a daily rate starts to mean something. A day or two is better. For a quartz watch, whose drift is far smaller, a week or two reads truer still.

What it cannot do

This is one pair of readings taken in whatever position the watch happened to be in, against whatever clock you happened to read. Three things limit it, and none of them are the calculator’s fault.

  • It only knows the two moments you gave it. A watch that gained for a day and lost for the next averages out to something that describes neither.
  • It does not know how the watch was resting. Gravity moves the rate, often by more than the rate itself, so a figure from a watch that spent the interval dial up is a figure for dial up.
  • It is only as good as the clock you compared against. Your phone’s displayed clock is close, but the second on screen is not necessarily the true one.

A rate is a reference point, nothing more. It describes how much a watch gained or lost per day across the readings you actually took, and nothing about how it holds up against magnets, water or a knock.

Why your watch runs fast or slow goes through the ordinary reasons behind whatever number you just got, and what a change in it is and is not worth reading into.

Doing this properly

One pair of readings is a data point. What tells you something is a series, meaning the same watch measured repeatedly, in positions you have kept track of, against a clock that is actually right.

That is what Ratekeeper is for. It takes the reading from a five-second tap, compares it to network time rather than your phone’s display, groups readings into runs so a reset does not corrupt the history, and breaks the rate down by the position the watch rested in. It is free for your first watch.