[alancalverd]

Nobody claims that a cesium clock is perfect, only that they are very very accurate

On the contrary, time is defined by the frequency of a cesium clock, which makes it perfect by definition.

[Janus]

Nobody claims that a cesium clock is perfect, only that they are very very accurate

On the contrary, time is defined by the frequency of a cesium clock, which makes it perfect by definition.

The second is defined by the unperturbed ground-state hyperfine transition frequency of the caesium 133 atom.  But there is a world of difference between this standard and using it in a practical clock.  We can measure that frequency through the electromagnetic radiation emitted by caesium atoms.  However since those atoms ( not being at absolute zero), will be moving at various directions and speeds, that radiation will be Doppler shifted to a certain degree leading to a "smearing" of the measured frequencies.   There are means to reduce the effect(and others) this has but none of them is perfect.   Thus a practical caesium clock will not be 100% accurate. We have been able to push the uncertainty of these clocks to an extremely small value, but not down to zero.

[alancalverd]

We don't measure the frequency of the Cs transition! We use it to measure the frequency of an auxiliary generator because it is the SI definition of frequency. Admittedly there are wobbles, but whilst the kilogram is defined in terms of h, there are experimental uncertainties in every Kibble balance measurement: realisation of a base unit is not the same as definition.

Thanks to the electric wireless telegraph thingy, the cesium clock is entirely practical. I even have a tiny receiver in my car instead of a map. And the magical GPS system makes all the relativistic corrections for me!

[yor_on]

NIST has always impressed me. It seems that the next generation optical clocks will be so accurate that it demands a new survey of Earths gravitational field. https://www.nist.gov/programs-projects/ion-optical-clocks-and-precision-measurements

" There’s just one problem with this plan, and it’s Albert Einstein’s fault. In 1915, Einstein published his famous theory of general relativity, which says objects that have mass actually distort space, making it curve. General relativity also says that space and time are linked, which means that distorting space also distorts time. Specifically, the stronger the force of gravity in a place, the more time slows down.

Normally, this isn’t a problem for life on Earth. Gravity doesn’t change much as you move around the planet, and even going high into the atmosphere on a plane won’t speed up time by any noticeable amount. But for the most accurate atomic clock ever made, small changes in gravity matter a great deal.

In order for NIST’s scientists to accurately know what time it is, they have to know how strong the force of gravity is at the location of their clock. Because the Earth isn’t a perfect sphere, gravity can change by small amounts from inch to inch and foot to foot. Instruments on satellites have measured the force of gravity all over the Earth to an accuracy of a few feet, which has been enough for atomic clocks in the past. But NIST’s new clock is so accurate it needs to know the force of gravity every single centimeter, which is currently beyond the ability of science." https://www.popularmechanics.com/science/a25362045/nist-atomic-clock-gravity/

[alancalverd]

That's one of the limitations of the Kibble balance (Ampere, 1830) too. You need to measure the local value of g, which is most accurately done by Galileo's method as improved by Torricelli (1640): drop a weight down an evacuated tube and time it. It seems that physics really hasn't advanced much, but instrument engineering is several orders of magnitude better!

[evan_au]

I heard the head of NIST Time & Frequency division talking about clock accuracy.
- She says that optical clocks have a similar "Q" factor to the more familiar Cesium clocks
- But because the optical clocks work at a much higher frequency, it is possible to count the frequency much more accurately
- She expects that the international standard of time will switch from Cesium clocks to optical clocks in 5-10 years

Listen (15 minutes): https://www.sciencefriday.com/articles/scifri-extra-about-time/