[thedoc (OP)]

The date has been set for the next leap second – June 30th 2012.  Leap seconds help to keep our incredibly accurate atomic clocks in line with the varying length of the Earth day.  But there is debate around whether we need them at all…

Read the whole story on our website by clicking here

 [chapter podcast=3826 track=12.01.25/Naked_Astronomy_12.01.25_9632.mp3] or Listen to the Story[/chapter] or [download as MP3]

[CliffordK]

Interesting podcast.

Perhaps the problem is that the clocks that are accurate to a second every 300 million years aren't in fact as accurate as the designers are saying they are.

If we are both adding and subtracting leap seconds, then one could stop the process all together. 

However, if on average we have to ADD 3 seconds every decade, then that would indicate that on the average the clocks are merely running 3 seconds per decade slow.  One can blame Earthquakes as much as one wants, but either the Earth slowed down by 3 seconds per decade since the invention of the clocks, or the clocks were poorly calibrated.

I suppose it is all a question of precision vs accuracy.
The clocks are precise to 1 second every 300 million years (hit the same place on the target all the time).
But they are not very accurate (miss the bulls-eye).

It would cause a little bit of chaos, but now that the Atomic clocks have been evaluated for 50 years, it is time to evaluate how many ticks long a second is.  The clocks should at least be accurate to a second a century or so.

[Bored chemist]

It has been a long time since the definition of the second was anything to do with the rate of spin of the earth.

Lots of varieties of clocks all agree with each other and they all say that the earth speeds up and slows down a bit.
Over a long enough time scale the earth's rate of rotation will not be constant, but the rate of ticking of the clock will be.

[syhprum]

We need leap seconds to cope with the slighly varibale rate at which the earth rotates but I dont think we should have them too often, about once per hundred years shoild be enough.

[CliffordK]

But, if we are always adding leap seconds.

Then that means that our atomic clocks are running slower than the spinning of the Earth, whether or not the Earth changes slightly on a day-to-day basis.  On a long term basis, one can synchronize the seconds to represent the rotation of the Earth on a decade or century basis.

I misspoke above.  If the clocks are running slow (on average), then the Earth is running fast (on average).  That can not be accounted for by the predicted Earth's long-term rate of rotation slowing down over time.  However, perhaps in a million years, the clocks will be more predictive of the actual diurnal cycle.

However, redefining the second would also mean redefining the speed of light, and the meter.  Not a big deal for most of us though, as our tape measures won't change.

[Bored chemist]

For historical reasons the SI definition of the second relates to the length of the day around the year 1800.
If we were to "redefine" it now in terms of the current rate of the Earth's spin, in addition to being a total pain in the neck, it would still be wrong again by next year.
We wouldn't need to add as many leap seconds, but the rates would still slip slowly out of sync.

[David Cooper]

If they want a measure of time that isn't strictly based on the Earth's rotation, why don't they go the whole hog and find something fundamental to base it on instead of using a historical second? Their argument seems to be based primarily on the idea that leap seconds provide opportunities for computer systems to crash, but at least at the moment these systems are tested from time to time and they are programmed to be able to cope. It might be silly to try to eliminate all these leap seconds only to have to have a leap hour at some time in the future which could really throw everything into chaos. But then again, artificial intelligence will be arround long before that time and will be able to make sure there are no bugs in any systems, so maybe it won't matter after all. The reality is that human-level artificial intelligence will be with us long before there's a need to adjust by so much as ten seconds (at which time the leap second will be reintroduced), so the mismatch will never be allowed to grow big enough to matter to anyone other than astronomers.

[Bored chemist]

"If they want a measure of time that isn't strictly based on the Earth's rotation, why don't they go the whole hog and find something fundamental to base it on instead of using a historical second?"
They did.
They chose the speed of light.
There's a conversion factor- it's about 300,000,000 which they cold have set to any number.
The number they chose was the one that meant they didn't have to alter any previous measured values and all the clocks.

[Geezer]

The Earth is not a very reliable clock. There are much better clocks, so we should use them.

[imatfaal]

The Earth is not a very reliable clock. There are much better clocks, so we should use them.

  Yep - agree entirely.

I have linked below a couple of blogs by Tom Swanson a physicist who designs/builds atomic clocks for the US Naval Observatory

http://blogs.scienceforums.net/swansont/archives/10790
http://blogs.scienceforums.net/swansont/archives/10829