How do they know the accuracy of the clock? Eg: 2s of drift in 65M years.
Naively, measuring that seems more difficult than the clock. Wouldn’t you need a perfect clock as a baseline? There must be something really clever here.
By adding up the uncertainties of all intermediate measurements they made.
As a parallel, for a regular clock, you'd measure and compute the uncertainties of:
- The distance between the axis of rotation and the center of mass of the pendulum,
- The local acceleration of gravity.
And deduce the pendulum period and its uncertainty by calculus.
In that case I'm not sure you can build (and measure) a pendulum that's more precise than a previous realization of the second (because the meter is defined from the speed of light and the second, and gravimeters probably depend on the definition of the second too).
That's why the definition of base units is carefully chosen to not depend on previous realization. IIRC time is the most fundamental of base units, because other base units depend on it.
They measured the stability by comparing the nuclear clocks with more stable conventional atomic clocks. In Vienna, they used a commercial single-ion optical clock [1], and in Beijing they used a hydrogen maser (a type of microwave atomic clocks) that was calibrated to the International Atomic Time scale via GNSS satellites.
Your intuition is not completely wrong. We are currently in process of redefining the fundamental SI second because our new clocks are more precise than the definition. It is quite mind-bending stuff.
There are already many years since the best optical clocks have exceeded by many orders of magnitude the precision of those on which the SI definition is based.
There are 2 reasons why the redefinition has been postponed by a couple of decades.
One reason is that the best optical clocks are very complex and it can be difficult to make them operate continuously for a long time. Until more reliable clocks will be made, the only solution for maintaining a continuous time scale would be to keep in each laboratory a relatively large number of optical clocks in operation, so that every time when one or more are halted for maintenance there would remain enough clocks still running. This is feasible, but expensive.
Besides waiting for more reliable or cheaper optical clocks, there is one more fundamental reason why the redefinition has been postponed.
When the cesium clock was chosen for the SI definition, there was no alternative with comparable precision, so the choice was obvious. Now there are about a dozen of different kinds of optical clocks with similar performances and it is not clear which will become eventually the best.
Therefore there exists a proposal with which I agree, that the future SI definition should no longer be based on a single frequency standard, but the unit of time should be derived from a frequency that would be the weighted geometric mean of the frequencies of many kinds of optical clocks, perhaps 5 or more of them.
Such a definition would also give the possibility that depending on the evolution of clock technology some standard frequencies could be added to or deleted from the set of frequencies used to compute the mean, or their weights could be altered.
The fact that thorium 229 can be used to make a nuclear clock has been discovered many years ago.
Nonetheless, this has become possible only today, after many years of researches. There is no doubt that future thorium 229 clocks will be better by several orders of magnitude than these first 2 prototypes, so it is likely that the corresponding clock frequency will be added to the set of frequencies on which the future SI definition will be based.
Naively, measuring that seems more difficult than the clock. Wouldn’t you need a perfect clock as a baseline? There must be something really clever here.