I’m looking at my “SkyScan” clock next to my monitor, one of those that claims to synchronize with satellite signals to show an accurate time, and I’m imagining that all satellite times are probably synced to a time originated by either NIST (the old NBS), or Greenwich, or both.
But then I began wondering: when the Greenwich “master clock” was first set, how did they know what time to set it to? How did they know the “real” time in Greenwich at that moment?
A bit of research turned up this at the GMT “official” site: an agreement signed in 1884 by the signatory countries provided, among other things:
Fine…but how did they know when “true midnight” occurred? My fifth-grade education tells me that “midnight” is not an astronomical or otherwise natural event, so what scheme did they use to decide, amongst the thousands of clocks in Greenwich that day, which was correct?
Why isn’t midnight a “real event”? Solar midnight is the moment when the sun is at its lowest elevation with respect to the horizon, just as solar noon is when it’s at its highest elevation. Granted, it’s a little hard to observe…
Solar noon, for any particular spot on the earth between the antarctic and arctic circles, is a well-defined event - it occurs when the line of longittude that passes through that spot, projected outwards into space as a plane, sweeps through the center of the sun as the earth rotates.
Midnight can then be defined, for any particular spot on the earth, as occuring when the opposite line of longitude from the one that passes through your spot, projected outwards, passes through the center of the sun.
Now, of course, measuring this somewhat abstracted event as of the Greenwish observatory can get somewhat tricky as a practical thing, but I don’t really see what the problem is.
hi mike!
The United States considers the time signal from the United States Naval Observatory, to be the official legal time. Whatever they say is, as a matter of law, correct. The National Institute of Standards and Technology maintains clocks in a number of locations, and keeps very close track of the variations in the time kept by those clocks. It advises the USNO when their clock is approaching an unacceptable level of error.
The actual rotational orientation of the Earth is no longer the controlling factor. In fact, the only adjustment made is to add a second or subtract a second every few years to keep local time more or less in time with the special count of “seconds since” method that USNO uses to keep track of time. Modern clock mechanisms are far more regular that the lumbering wobble that is the Earth’s rotation.
So, if you live in the US, what time it is, is whatever the USNO says it is. If you live elsewhere, you may use a different standard.
Tris
“It was a woman drove me to drink and I didn’t even have the decency to thank her.” ~ W.C. Fields
Before atomic clocks became commonplace, observatories had special telescopes (transit telescope) that recorded when a given star passed through the meridian of the observatory. If you knew the positions of the stars in the celestial sphere, you could use them as a source of time.
The drafting of the 1884 agreement is rather abstract, in the sense that the concepts it uses are not immediately observable. They could be derived from empirical obserations, but this was a rather convoluted matter in practice. Though this was sufficiently routine and standardised that these details didn’t need to be defined in the agreement.
What was being observed at Greenwich were solar transits (or stellar transits as a proxy and crosscheck). Each day at “midday”, a south facing telescope was used to establish when the Sun passed across this meridian. That in itself is a non-trivial matter when it has to be done to the required accuracy, not least because the Sun is disk whose size varies slightly over the course of the year.
By definition, this establishes a local noon at Greenwich. But there isn’t a constant interval between these noons from day to day; the variable motion of the Earth around the Sun is a large effect and so this timing varies. A day defined in this way isn’t a constant 24 hours. Hence the notion of a Mean Solar Day. Averaged over the course of an orbit, then you do get something that can be taken as 24 hours. Similarly, you can define a Mean Noon - roughly, the average around which the observed noons systematically move.
A day of 24 hours is then a Mean Solar Day. Mean Midnight is just 12 of these hours away from Mean Noon.
There were endless technical complications, but that was the basic idea.
Incidentally, there was a significant debate in 1884 about whether to draft that clause in terms of Mean Midnight or Mean Noon. Astronomers had been in the habit of using noon for these purposes, for several different reasons. The noon transits were the observable events and so Mean Noon was conceptually closer to the data. It was also traditional for them to define a “day” in such a way that it started at noon, since that meant an entire observing night fell on the same date. Naval traditions had globally tended to follow the same pattern. But what prevailed was the everyday notion that daylight constituted a single day.
Current arrangements, based on atomic clocks, are largely only historically connected to the 1884 version.
In practice, almost everyone uses UTC, which is maintained by the International Earth Rotation Service (IERS). Atomic time (TAI) is a statistical product of atomic clocks in research institutions all over the world, including many at the USNO. The Navstar GPS system has its own time scale, which is related to TAI.
Thanks to all for the replies, but especially Bonzer that nailed it perfectly, in that the explanation makes sense for how they did it back then.
For the others deriding my fifth-grade education: well, I was operating under the assumption that when midnight rolls around tonight, it merely means 12:00 (or 0000, if you prefer), and that midnight - or noon - is not tied to a specific astronomical event. I stand illuminated.
If I remember correctly atomic clocks lose time when travelling by air so wouldn’t this mean the atomic clocks at various locations around the world would have to be built on site or was the lost time accounted for in an adjustment when they arrived at their final destination?
The relativistic effect of air travel is relatively small in comparison to other errors. An atomic clock is a very stable oscillator (cesium or rubidium) that drives a set of electronic circuits that count the oscillations. Its function is similar to the pendulum in a mechanical clock or the quartz crystal in a digital watch. Once an atomic clock is purchased, about $50K last time I checked, and installed, you need to set its time. This can be done by comparing its time to time signals broadcast by NIST (WWV, WWVB, WWVH), LORAN or GPS. If the location of the transmitter is known, the propagation delay can be estimated and compensated for. Currently, GPS is the cheapest and most accurate way to set a clock. In the old days, people used to fly around the world with portable atomic clocks in order to set and check the time at remote locations with atomic clocks. They would buy two airplane tickets, one for the engineer and one for the atomic clock.
Gravity also affects the frequency of an atomic clock, so if you need extreme accuracy, you have to consider the gravitational field strength at where the clock was installed. This varies over the surface of the Earth and changes with altitude.
Has there ever been a court case where this was used? For example, payment had to made by 5PM but one party’s clock was off? I’m especially interested in cases where one party is a governmental body.
If you had to take departure on a railroad train 50/60 years ago you would have to know down to a second to avoid missing the train.
Train crews took great pride in ‘running’ on time.
Today an hour or so maybe to catch a plane. If you are an hour or so late you may have to wait longer due to one delay or another.
When I was a boy I had a Sears & Roebuck ‘Railroad’ pocket watch.
It had a big locomotive embossed on the back. Kept pretty good time, like a minute variation per day.
A bit larger than the real precision Elgin & Waltham R.R. watches the trainmen carried.
Built like an alarm clock. Made by one of the New England clock companies.
I never “missed an appointment” as long as I had it. (Had very few appointments.)
It was much more than a matter of pride, it was a matter of safety. Before good radio communications and in busy areas running on time was vital. I remember reading about a train accident at a busy junction in England some years ago. I don’t recall the exact number, but there were well over 100 trains/hour through that junction at the busy times of the day. If they aren’t on time it creates on hell of a mess.
Well, I don’t know any existing case law about variations in time standards. But contracts that fail to state an authority would default to the USNO in the US, but the party seeking damages would have to prove his own accuracy, and the inaccuracy of the defaulting party, and do so in a manner that would convince a jury that it was enough time to cause an actual material failure to fulfill the intent of the contract.
Since the difference between one authority on time and another is likely to be a matter of milliseconds at the most extreme, that seems really unlikely to me that a contract could be judged to be in default over that small a difference in time of completion.