Dial up anyone studying radioactive decay and ask them how their experiment is going. Reset the clocks based on rate of decay since the last “known” time.
This isn’t additional slippage, but a different kind of slippage. Leap days (added in “leap years”) are added so that the calendar year stays synchronized with the solar year. It’s related to the time it takes the earth to revolve around the sun once.
Leap seconds are added so that the calendar day stays synchronized with the solar day. It’s related to the time it takes the earth to rotate once (plus or minus a bit) on its axis so that the same spot on the earth is pointed directly at the sun.
They’re not added or subtracted to stay synchronized with any larger celestial events or some concept of sidereal time.
Furthermore, leap seconds aren’t predictable the way leap days are, because the rotation speed of the earth varies irregularly due to its molten layers.
I think your conclusion is still valid despite my points, but by looking more at the sun than the stars (to determine the second or fraction of second). It’s possible we could get closer using other astronomical information, but I bet the Earth’s rotation (relative to the Sun or other stars) would be the most accurately observable phenomenon.
To get super technical, the question raises another question, “Relative to which timescale?”
We have a number of them and they don’t match exactly, for a number of technical reasons. For more info and a bit about how NTP handles leap seconds, see The NTP Timescale and Leap Seconds . The author, Dave Mills, is the guy who invented NTP, which is the protocol that computers use to distribute time. If you’re interested in time and how it’s measured and how it’s been represented across the centuries, check his other papers. I’m not sure whether he has the one posted online that gives the in-depth background that I read on his UDEL site some years ago, but I bet you can find it among the NPT white papers.
Well, every city could start out using the sun to find local noon, and then eventually forming regional time zones for simplicity. Would we still have all of our atlasas and maps and globes and whatnot showing the time zones as they are currently, or would we need a modern equivalent of William Lambert or Sandford Fleming?
In which case I’m not riding any trains soon.
The OP implies that only our memories of what time it is will be wiped out, along with stopping clocks; thus it should be very easy to get back on track. After all, people write down or otherwise record the time and date all the time; all I have to do is look at the files on my computer for the most recent modified/created dates (assuming that all of this happens in an instant and no weird time warping occurs). I’d also know that it is late afternoon right now from the position of the sun, and sometime during the winter even without looking at the time/date. Note also that if clocks are simply stopped, not reset or anything (not sure which one is meant, the title says reset but the OP says stopped), that will tell you the time right there; people will quickly realize that they aren’t running, and for people who think it is an earlier year, what’s up with that calendar showing 2013?
For this to really work, the entire concept of time would have to be erased from our memories; for example, the definition of a second, minute, hour, week, Thursday, January, etc; and/erased from everything that shows a time/date (e.g. calendars become blank, not that a calendar would be useful if we had no idea what they were for).
Well, if they don’t become blank, a lot of calendars show the phases of the moon, and that connection would become apparent pretty quickly, I bet.
Of course, if we lost all clocks, including electronic and computerized clocks, our communication infrastructure collapses immediately since we have no way to compute latency and optimize efficiency. We might have to regress to relay-based dedicated connections, like telephone service in the 1950s.
Pulsars should be all you need. We know of thousands of them; a given phase configuration will uniquely identify a time to very high precision, even considering their occasional glitches. You could then use averaging techniques to give a very high resolution (sub microsecond wouldn’t surprise me).
Oh no doubt, pulsars, much less a bunch of them used together, could make for some pretty handy timekeeping devices.
But to figure out from scratch what the exact time “it is now” that matches as close as possible the previous “time it was before God played a trick on us” requires a bit more.
You need to figure out WHICH pulsar pulse you are on. Not any old pulse. The 50 gazillion one since it was discovered say.
To do that you have to work your way through processes that have progessively shorter and shorter periods, using each one to make sure you are at the right “pulse” before you drop down to the next short period process.
Yeah that’s what I was thinking, too short a period isn’t useful either.
But that’s not a problem if you have enough of them. Let’s say that you have 3 pulsars, and say they spin at 9157, 9479, and 9719 microseconds/revolution. And let’s say that they all lined up at some known moment in the past. Well, because the numbers are relatively prime, we know it’ll be 915794799719 microseconds, or 9.76 days, before they line up again. So we’ve made a long-period pulsar out of a few short-period ones.
If you add a fourth pulsar, say with a period of 9767 us/rev, then we now have a period of 261 years. You can add more to both extend the timeframe and accuracy.
I chose relatively prime rates to make the math more obvious, but in practice that’s not necessary, because any two pulsars will have totally uncorrelated rates (it would be surprising indeed to find two pulsars with rates of a small integer fraction of each other unless they were gravitationally coupled or some such).
There are thousands of pulsars on tap, so it’s guaranteed that there are many combinations that will uniquely identify a particular time. And of course we don’t have to actually wait for them to line up; we can just use their relative phase to perform the same computation.
That is what I was thinking when I first suggested pulsars, but from the Wikipedia page, their rate has a randomness to it. I looked at the source referenced, but all it said was "Outside the steady decrease in all pulsars exhibit to some extent rotational irregularities known as timing noise, which is observable as random wandering in either the pulse phase or its frequency. " No indication of how much noise or wandering.
There would be a coherence time, similar to lasers, beyond which you can’t use the relative phase of multiple pulsars to determine the time. You’d still have the rate of change of their periods, and could use that to estimate the time from any one of them, and with many of them, you’d be more insensitive to noise and measurement error, and could detect which pulsars had a glitch in the intervening time.
How exactly would one go about “stopping” an atomic clock, anyway? Some sort of destruction of the actual machinery?
For the purposes of the OP, a minor EMP that scrambled its internal notion of the current time would be sufficient. It doesn’t have to actually stop; just lose track of things temporarily.
Yup. I’ve always read that pulsars are “more accurate than atomic clocks”, but without much notion for what this means exactly. My intuition is that the coherence time is quite long, especially if you accounted for steady spindown, and that prediction over hundreds of years would be no problem. But I wouldn’t mind having that confirmed by an actual expert.
Presumably television hasn’t been interrupted so it would be a trivial matter to ask a TV network’s Playout facility to add up the material that’s been played since the last known time. I’ve worked in such facilities and, especially if you have no live events, every frame can be accounted for. So that gets you to small fractions of a second at least.
While you could in principle get all the information you need from pulsars, even with some unknown number of glitches in the data, in practice it would be a monumentally huge data analysis problem. It’d be much, much easier to use other methods to get down to a second or so, and then just use the pulsars to refine it from there.