Wednesday, September 21, 2016

TIME 2 - Time under test

Okay - so previously we took a little look at time and defining it.  I put forward that this was the best I could come up with ...

Mike Talks' Definition Of Time

Time - a phenomenon which allows change to occur.

That is to say, without time, a system will remain in stasis.

Today I want to explore some meaning to that definition.

Measuring time?

Well as I mentioned previously, we can't measure time, but we can measure change.  It seems fair to say that the more change you're observing, the more time has elapsed.

If I went to the bottom of my road and counted cars passing me as my "change" to monitor - is seems fair to say I'd expect it to take longer to count 1000 cars passing me than to count 10.  But to know for sure I'd need to be able to check it against something I could be certain changes consistently and regularly.

Some of you might say "well just check with a stopwatch" - to which I'd remind you from last time, all the stopwatch is doing is counting oscillations in a quartz crystal.  So how can you be sure that's correct?

Synchronizing watches

These days with smartphones which autocorrect the time regularly, we're losing touch with synchronizing our clocks.  Even the best watch in the world can be inaccurate and drift a little - typically the more accurate, the more expensive.

But even this requires us to synchronise our device with a reference clock which we believe to be accurate,
  • When I was young we used to have a series of radio pips on the radio to tell us when the hour was on.
  • We could ring up the speaking clock on the phone
  • Towns in England would use the chimes of a town/church clock or a time ball to make aware the turning of the hour

The faulty clocks conundrum

Consider this as a challenge - you're given two clocks and told that whilst one is accurate, the other isn't correctly keeping time.  How do you work out which of the clocks is faulty?

There isn't actually any way I can see to solve this puzzle - if you leave both clocks running, one will be fast, the other slow.  There's no way to tell which is the accurate and which is the faulty one without somehow having a third reference time which you know to be accurate.

It's possible that one is running so incredibly fast/slow that you can tell by watching that "that's not ticking correctly", but you're not really using the second clock.

By the way - I typically hate these kinds of puzzles. If someone knew one clock was accurate, why ever the hell didn't they label it as such?

How do you actually test a clock?

Actually this has made me want to look a lot into "'how did people test the accuracy of clocks"?

This took me to the story of 18th Century inventor John Harrison, who tried to create an accurate naval clock or chronometer for measuring longitude.  Typically a large clock of his period is designed to be accurate, but completely stationary.  Smaller pocket watches existed, but it wasn't unusual for them to lose as much as fifteen minutes in a single day.

The challenge in a chronometer was to create a device which could move on the rolling, salt heavy, humid conditions of the sea, but retain accuracy of a large stationary clock.

If on a long sea voyage you knew London time, and you measured the time of midday, when the sun's shadow is smallest and points due north (in the Northern Hemisphere anyway), then for every hours difference from midday London, you were 15 degrees of longitude from London.  Read here for more detail.

Through experimentation, he turned a lot of ideas on clocks on their heads.  Many believed clock accuracy was down to using slow moving, heavy pendulums, where John Harrison found smaller, faster moving parts worked better - indeed a longitude watch could keep better time than a clock, as was proved by his experiments.  [His chronometer was noticeably different from normal clocks/watches from it's rapid tick-tick-tick]

John Harrison typically spent over 31 years building and testing a series of devices.  One such test involved sending the H4 watch overseas to Jamaica with his son, where it was found to only lose a few seconds over the 6 week voyage.

What's somewhat frustrating looking through all the information there is on John Harrison - there's a lot documented on every revolutionary feature he incorporated into his device, and how they were made.  But only a few clues as to how they were tested - this was annoying because (a) I'm a tester and (b) the devices can only be as good as the tests used to measure them.

I had a go at trying to reverse engineer how this might have been tested, and came up with the following possibilities,

  • Checking noontime
  • Checking vs accurate land clock
  • Checking vs night sky

Checking noontime

One obvious method for checking a chronometer would be to synchronise the device to twelve noon using a sundial to measure when a shadow is smallest / Due North.

You wait for noon the next day, and see if it's showing 12 o'clock.  The problem with this is it's quite a shallow test - and noon isn't something you can measure "to the second".

You can probably increase the confidence in this by running it daily over months, to see how much it shifts.  This is probably why Harrison spent 31 years testing!

Fundamentally with this, you're using one source of change (the hands on your watch under test), and testing it against another source (the time it takes the Earth to rotate), looking for correlation.

Checking vs accurate clock

This helps to give you a gauge, but as said above, how can you tell which of the clocks were "most accurate".  That said, such clocks run night and day, and someone would notice if the clock was striking noon when the sun was just rising in the morning.

Like the method above, this is another form of correlation testing.

Finding longitude of a known location

There are some records that show John Harrison's son used the H4 watch in Jamaica to predict the longitudinal position, and found after the sea voyage it was correct within a few seconds.  However of course - if the 18th Century lacked any means to accurately get the position of Jamaica (the whole point of the chronometer), how do you know the H4 wasn't measuring it more accurately than had previously been possible?

I suspected that Jamaica's latitude and longitude might have been calculate using the night sky - but again, there's a frustrating lack of information just on the internet about this. 

But which was it?

Don't you hate not knowing?  Thankfully James Bach recommended a book Longitude by Dava Sobel which solved this in.

It seems they'd found an accurate way on land to test time by observing the motion of the moons of Jupiter.  This was originally put forward by Galileo, but the astronomer Cassini perfected this with a series of tables (little did they know, there was an error in the system that Cassini corrected for, which turned out to be due to the speed of light difference due to different relative ranges between the Earth and Jupiter).

But this method of taking time could only be done by a trained mathematician/astronomer on solid land.  Hence knowing Jamaica's position relatively accurately.  Obviously only a few people had the skills to do this calculation, and it was completely unsuitable for use on the rolling sea.

Read more about this method here.


I noticed that in 2015, a replica of the H4 was "tested"and found to keep time accurately.  You can read about it here.

Please notice my use of quotation marks there.  The replica was "tested" at the Greenwich observatory in a stationary location.  Part of the requirements of the chronometer for Harrison was that it to keep accurate time, whilst at sea.  So this isn't really an adequate test in my opinion.  [I'd have wanted to add in some rocking motion to the test]

Do you disagree?  Then comment below ...

Now playing:  "Make Me Smile", Steve Harley & Cockney Rebel

No comments:

Post a Comment