Have you ever wondered what the ultimate authority is regarding pitch calibration? You can check your tuning fork against someone else's (more expensive) tuning fork. They can check their fork against an electronic tuning device. The maker of the electronic tuning device can check his device against a laboratory frequency standard. So where does this chain end? It all comes down to the exact definition of a second.
In 1964, the Twelfth General Conference of Weights and Measures defined one second to be the time equal to 9,192,631,770 cycles of oscillation of the fundamental state 2 S(1/2) of an atom of cesium-133 when transitioning between two hyperfine levels F=4, mF=0 and F=3, mF=0 undisturbed by external magnetic fields. I have absolutely no idea what that means, but if someone asks, that's the official definition of a second. The exact number of cycles was chosen so that the second would agree as closely as possible to the common understanding of a second, which is supposed to provide 60x60x24 seconds as the mean time from one sunrise to the next. And in 1964 it did just that. Today, due to the slight slowing of the earth's rotation, the definition of a second using the cesium atom (so-called atomic time) is a little too fast for the earth rotation. That is why we often have a leap-second added on to the end of December or June, which makes 61 seconds in the last minute of the months that are so affected.
In the US, standard time
and frequency are maintained by the National Institute of Standards
and Technology (NIST). Old-timers may remember this agency as
the National Bureau of Standards. They have a bank of
cesium-stabilized oscillators in Colorado designed to measure time to
an accuracy of one part in 2,000,000,000,000,000. In terms of
musical accuracy, that's .00000000000087 cents!
To see how you can make practical use of the NIST services, see Calibration Using NIST.