certain the freezing of Well, sort of. The definition of a metre is actually 1/299792458 of the distance travelled by light in vacuo in one second. Wavelengths of the orange-red line in the spectrum of Krypton-86 went out long enough ago to qualify for this list at least! The definition of the kilogram is _still_ afaik "the mass of the international prototype kilogram" (or kilogramme, I suppose). And FWIW sea level pressure is 1013 millibars. 760 millimetres of mercury. But kilogram = mass of 1 cubic decimetre of pure water is probably accurate enough for most purposes as long as you don't have too much deuterium or oxygen-18. BTW Celsius isn't an SI unit. SI unit of temperature is the [degree] Kelvin (K, written without the degree sign), defined as 1/273.16 of the absolute temperature of the triple point of, oh no! It's water again! Again, specify hydrogen-1 and oxygen-16 and it's probably accurate enough... See. Not as easy as it sounds, but still possible. If you allow non-SI definitions, start with: Second = 1/86400 mean solar day (that changes only by a couple of percent every million years - enough that dinosaurs had a 400 day year, and ancient Babylonians observed eclipses a few tens of miles from where modern astronomers predicted them or at the wrong time of day) Metre = 1/40000000 circumference of Earth. Originally 1E-7 distance from N Pole to Equator through (I think) Paris. (Longer than SI metre by 0.02%) Then proceed as above with Celsius and volumes of water... Philip.