How We Measure Heat: Fixed Points, and the Constant That Replaced Them

The official definition of the kelvin now reads, in full: the Boltzmann constant is 1.380 649 x 10^-23 joules per kelvin. That is the whole thing. No water, no ice, no reference to any substance at all — and no instruction on how anyone is supposed to use it.

Julian Ashworth begins with that gap. A definition of a unit is not a method of measuring it, and the distance between the two is where thermometry actually lives.

The story runs from the practical problem of getting two thermometers to agree, through the international scale that solved it — a ladder of assigned values at reproducible phase transitions, with specified instruments to interpolate between them — to the day in 2019 when the unit stopped depending on the triple point of water and started depending on a number that had been voted exact.

The chapters in between take the apparatus apart. What is inside the sealed cell of pure water, and why a phase transition holds still while two phases coexist. Why platinum, of all things, is written into the definition of the scale. What a radiation thermometer silently assumes about the surface it is pointed at, and how wrong that assumption usually is. And how three unrelated experiments — sound in a gas-filled cavity, electrical noise in a resistor, the broadening of a spectral line — were pushed until they agreed closely enough for a committee to fix the constant for good.

Two chapters deal with temperatures taken from people. The famous 37 degrees comes from one physician's nineteenth-century series, and a study of three cohorts spanning 1860 to 2017 found mean body temperature falling steadily by about 0.03 degrees per birth decade — a real historical trend, not an artefact of old instruments. The thermal cameras deployed at building entrances during the pandemic detect roughly four in five febrile people under favourable conditions, which is a different claim from the one their presence implies.

Sources are named in the text and listed at the back: the BIPM brochures and realisation guides, NIST's calibration documentation, the Metrologia papers behind the redefinition, and the clinical literature. Where a familiar number turned out to rest on secondary sources, it is flagged rather than repeated with confidence it has not earned. Twenty-four diagrams, including the fixed-point ladder and what interpolates between each pair.