[JP (OP)]

How is the second defined?  What physical process(es) are we measuring?

(This was brought up in the thread here: http://www.thenakedscientists.com/forum/index.php?topic=29238.0;topicseen
I wasn't quite following the discussion in that other thread, so I thought to make the point clearer we could discuss the point here.)

[Soul Surfer]

The historical definition of the second came as a specific fraction of the 24 hour mean solar day.  (day lengths do actually vary throughout the year by a small amount because the earth's orbit is elliptical and the earth's rotation is very nearly constant).  This was reinterpreted to be a particular pendulum but as measurements became more precise it was reinterpreted as a specific number of cycles of a particularly stable spectrum line.

Wikipedia gives this information on the Second   among a lot of more detailed descriptions.

"The second is the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom."

but there is a more stable frequency source that is being considered as an alternative

Nowadays this is more precise and stable than the rotation of the earth and so normal clock time is adjusted by small amounts to keep it in step  it used to be "leap seconds" every few years but now it is smaller and more frequent.

[LeeE]

As Farsight said in that thread...

Since 1967, the second has been defined to be the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom.

That atomic clock employs microwave radiation, which is essentially light. You count 9,192,631,770 microwave peaks going past and call it a second.

[JP (OP)]

Ok.  That makes sense.  So is there a difference in the result you get between counting the number of waves going past and the number of transitions in the cesium atom?  (There would be, I suppose, if you let the atom move at relativistic speeds with respect to you, but isn't that all controlled for in the definition of measuring the second?  The atom is at rest, 0 K, etc.?)

[PhysBang]

Technically, the definition rests on the processes of the electron, not on the motion of the light. The light is how one measures the transitions, but the timing of the transitions is determined by the transition of the electron. The transition from one state to another is not a transition of motion but of energy.

The reliance on light is a practical restriction. The definition of a second as the transition is a practical definition, too. It is something that we can use as a standard of accuracy.

[Geezer]

Technically, the definition rests on the processes of the electron, not on the motion of the light. The light is how one measures the transitions, but the timing of the transitions is determined by the transition of the electron. The transition from one state to another is not a transition of motion but of energy.

The reliance on light is a practical restriction. The definition of a second as the transition is a practical definition, too. It is something that we can use as a standard of accuracy.

Well said Physbang. Ultimately we are using the cesium atom as an oscillator, as we do with all timing devices. The transitions in a cesium atom happen to provide a very stable timing reference. EM Radiation (it's a long way from being visible light) is merely the means used to detect those transitions.

Atomic clocks are rather sophisticated clocks that happen to use atoms as oscillators, but they are still clocks nonetheless.

[syhprum]

There seem to be two basic forms of oscillator that are used in 'clocks' those that depend on gravitational forces i.e the rotation of bodies, pendulums etc and those that depend on inter atomic electromagnetic forces i.e springs, crystals etc.
could atomic clocks be said to form a third class ?.
"(day lengths do actually vary throughout the year by a small amount because the earth's orbit is elliptical and the earth's rotation is very nearly constant)"
My sundial was 20 minutes out at Xmas!

[Geezer]

My sundial was 20 minutes out at Xmas!

That's nothing. I have an Atmos. I never have to wind it up, but it's 20 minutes out every month!

[Farsight]

It's important to remember that you are not counting the caesium hyperfine transitions, you're counting the passing microwave peaks that result from it. The hyperfine transition is a spin flip, like the hydrogen depiction below, and you aren't counting spin flips:

[JP (OP)]

Ok, Farsight, but as I understand it light is just the tool we use to extract information about the atomic transitions, and we make all sorts of requirements to keep the light from influencing the measurement.  Is there any practical difference in this case between the number of atomic transitions and the number of peaks of the light that pass by?

[LeeE]

I agree with JP here; although it is the light that is detected and counted, it is the transitions that create the light that act as the timing mechanism.  The emitted light just acts as a 'messenger' to bring us the information.

[Geezer]

From Wiki - Atomic Clocks.

"Since 1967, the International System of Units (SI) has defined the second as the duration of 9,192,631,770 cycles of radiation corresponding to the transition between two energy levels of the caesium-133 atom."

Even the definition says it's an atomic standard.

If we used light to measure the speed of light, I'm pretty sure we would soon disappear up our own .......

[Farsight]

JP: yes.

Lee: yes, the transitions create the light that acts as a timing mechanism, and light does act as a messenger. But the duration between hyperfine transitions isn't defining the second. The hyperfine transition is something like the pluck of a guitar string. The nature of the guitar string determines the emitted sound, just as the nature of the atom determines the emitted microwaves.  

[Geezer]

Farsight,

Can you explain your last post for me please.

Are you saying that atomic clocks keep time based on properties of atoms, or not?

And if you are saying that they don't keep time based on properties of atoms, please explain what they are using to keep time.

[Geezer]

Cesium fountain atomic clock.

http://tf.nist.gov/cesium/fountain.htm

As it employs a fountain, would that mean we are really measuring time with water?

Please note:

"Eventually, a microwave frequency is found that alters the states of most of the cesium atoms and maximizes their fluorescence. This frequency is the natural resonance frequency of the cesium atom (9,192,631,770 Hz), or the frequency used to define the second."

[lightarrow]

What is measured is a property of an em radiation (the frequency). The fact that radiation is emitted by an atom is not relevant here, but the practical fact this radiation is particularly stable, when emitted by that atom with that electronic transition.

[lightarrow]

Ok, Farsight, but as I understand it light is just the tool we use to extract information about the atomic transitions, and we make all sorts of requirements to keep the light from influencing the measurement.  Is there any practical difference in this case between the number of atomic transitions and the number of peaks of the light that pass by?

Farsight has already answered, but I want to remark that there is absolutely no relation at all between the two things; in theory you could get that radiation with a completely different mechanism, not involving atoms or electrons at all.

[Geezer]

Farsight has already answered, but I want to remark that there is absolutely no relation at all between the two things; in theory you could get that radiation with a completely different mechanism, not involving atoms or electrons at all.

Indeed you could. There are many ways to produce radiation, but they would not produce a very accurate clock. The point you might be missing though is that the Cesium atoms in an atomic clock are controlling the frequency of the radiation. There is a rather complicated feedback mechanism involved to achieve that, but it is there, nonetheless.

I think the NIST page makes that rather clear. It's a bit more difficult to identify it on the Wiki page on atomic clocks, but it does mention it.

[Farsight]

Farsight,

Can you explain your last post for me please.

Are you saying that atomic clocks keep time based on properties of atoms, or not?

And if you are saying that they don't keep time based on properties of atoms, please explain what they are using to keep time.

Yes, of course atomic clocks keep time based on the properties of atoms, just as a guitar note depends on the properties of the guitar string. An atom undergoes a hyperfine transition and emits a photon, then can absorb a photon and undergo the hyperfine transition again. The frequency of that photon depends on the atom you use, the particular hyperfine transition, and other factors such as temperature and gravitational time dilation. But note that frequency is measured in hertz, which is defined as cycles per second, so we have to count cycles to define the second.

[Geezer]

But note that frequency is measured in hertz, which is defined as cycles per second, so we have to count cycles to define the second.

The clock is not measuring frequency which is what you seem to be implying. It's simply counting events. The time between those events happens to be  1/9,192,631,770 seconds. It's really no different from any other clock that uses a pendulum, spring, or whatever.

We don't count cycles to "define" the second. We count cycles to measure a second, or any other interval of time we choose.

You continually imply that there is something circular and unreliable about this process. There isn't. It's a clock.