The Kilogram Stopped Being an Object

Until May 2019, the kilogram was a cylinder of platinum-iridium alloy kept in a vault near Paris. Not a representation of the kilogram: the kilogram. Whatever that object massed was one kilogram by definition, everywhere in the world.

The arrangement has an obvious defect once you see it. If the artefact is the definition, the artefact cannot be wrong. Any drift in its mass is undetectable in principle, because there is nothing more authoritative to compare it against. Its official copies, distributed to national laboratories, diverged from it by tens of micrograms over the course of a century, and nobody could say whether the copies had gained or the original had lost. A standard that cannot be checked is a standard nobody can verify, and metrologists spent decades working towards a replacement.

The replacement inverts the relationship between measurement and constant. Planck's constant relates energy to frequency and appears throughout physics, and it had been measured with increasing precision for a century, using, among other things, mass standards. The redefinition fixed its numerical value by international agreement. Instead of measuring the constant using a kilogram, the kilogram is now realised using the exact constant, via a Kibble balance that compares mechanical and electrical power with extraordinary precision.

The practical consequence is that the standard is no longer held anywhere. Any laboratory capable of performing the experiment has it. Nothing depends on an object surviving a fire, a theft or the slow accumulation of surface contamination, and there is no privileged location.

The 2019 change completed a programme running back to the metre, which stopped being a bar in 1960 and was tied to the speed of light in 1983. Every base unit is now defined by a fixed constant or an atomic property. A subtle benefit is that the definitions will not need revising as measurement improves: better experiments improve the realisation, not the definition, so the system is stable in a way the artefact era never was.

Time is the unit realised most precisely by a wide margin. Caesium clocks reach fractional uncertainties around one part in ten to the eighteenth, and optical clocks have pushed further, to the point where a redefinition based on an optical transition is under active discussion. Because the metre is defined via the speed of light and a time interval, improvements in clocks quietly improve the realisation of length as well, which is one reason metrological effort concentrates so heavily on timekeeping.

None of this reaches an ordinary laboratory directly. What reaches it is traceability: an unbroken chain of comparisons linking a working instrument through laboratory standards to a national institute and thence to the definition, with an uncertainty stated at every link. That chain is what makes a calibration certificate mean something, and a break anywhere in it invalidates everything downstream.

It is administrative, unglamorous, and the reason a measurement taken in one country can be trusted in another, which is, in the end, what a system of units is for.