When the human ability to measure things becomes iteratively better over time, it can lead to some counter-intuitive changes in how certain everyday things are defined. Think about one “second”: it doesn’t take too much to work out how long it takes for the Sun to move from noon one day to the next, and then divide that up into 24 hours, and each of those hours into 60 minutes, each of 60 seconds. That’s a second – 1/86400th of a day.
Except, of course, it isn’t that easy, because the Earth has the frustrating habit of not spinning at a constant speed. Day-to-day, the Earth’s spin is affected by a range of factors – from the angle of rotation to its distance from the Sun – which can cause a “day” to be as much as 20 seconds longer or shorter than other days in the year. The most obvious solution would be to take an average of the length of all of the days over the course of a year, and that’s how the second was defined at a meeting of the International Committee for Weights and Measures (CIPM) in 1956. The second, the meeting concluded, was equal to 1/86400th of the average tropical (as in, on the equator) day, as measured in 1900.
Problem solved? No, because the Earth’s rotation is still slowing down over the long term at an unpredictable rate – thanks to tidal forces from the Moon and Sun, and internal geological activity like earthquakes – meaning that the average day of today is almost two milliseconds longer than the average day of 1900. This was apparent decades ago, which is why the CIPM met in 1965 to redefine the second again, in terms that didn’t rely upon an unchanging Earth. Instead, a second became “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”.
(Put more simply: whenever an electron orbiting the nucleus of an atom drops down to a lower orbit, it releases electromagnetic energy which, like a radio wave, resonates at a specific frequency. This is true of all atoms – it’s what radioactivity is, as unstable atoms rapidly decay – but caesium is particularly good for timekeeping because of its stability and reliability, so measuring this energy release anywhere in the universe should, in theory, give the same result every time. And the CIPM decided that a second would be defined as 9,192,631,770 cycles of this energy.)
Defining a second with this kind of accuracy is absolutely necessary to almost every aspect of modern technologically-aided human existence. Both the SI units for metre and ampere are defined with reference to the second, and there are innumerable scientific, engineering, business and cultural projects which rely on precise timekeeping to work worldwide. This is what makes leap seconds so tricky.