[mjevans]

So strange, I remember the Fahrenheit (32, 212) and Celsius (0, 100) numbers, but the precise Kelvin number (which is hard-required for all the good physics equations I've forgotten), that is how much lower than 0C absolute zero is, I only vaguely remember as 272 something. However that's just a well known constant. Very reasonable to include it in the reference dataset.

[Dylan16807]

273 something.

For a good while it was defined as exactly 273.15, but as of a couple years ago it's now based on a specific number of Joules.

[exmadscientist]

The difference between Kelvin and Celsius is still the same. But the Kelvin scale itself has shifted a bit, taking the other scales with it.

[Dylan16807]

Well, okay, they changed Kelvin and Celsius at the same time, so that 273.15/273.16 are preserved, with Celsius based on Kelvin instead of the other way round.

But when you're dealing with the freezing point of water and the old version of Celsius for simplicity, then those numbers aren't as correct as they used to be.

[exmadscientist]

They're not far off, especially near the triple point of water: https://www.bipm.org/documents/20126/2619564/Estimates_Diffe...

See also section 2 of the more-or-less authoritative parent document "Mise en pratique for the definition of the kelvin in the SI", available at https://www.bipm.org/documents/20126/41489682/SI-App2-kelvin...