Why Are There Different Temperature Scales? Fixed Points and Accidents of History

Measuring temperature requires something that changes reliably with it, such as the expansion of a liquid, and turning that into numbers requires at least two reproducible reference conditions to anchor the scale. Any two reproducible conditions will do, and everything else follows from that choice, which is why scales differ. The freezing and boiling points of water are attractive because they are easy to reproduce, though both depend on pressure and on purity, which caused real problems before those dependencies were understood. Human body temperature was used and turned out to be less constant than assumed. Mixtures of ice and salt give a low reference point. Whatever is chosen, the interval between the two points is then divided into a number of degrees, and that number is as arbitrary as the points themselves.

Each makes different choices and they are related by simple conversions:

• •Fahrenheit, with a zero based on a freezing brine mixture, water freezing at 32 and boiling at 212
• •Celsius, with zero at the freezing point of water and 100 at its boiling point at standard pressure
• •Kelvin, with the same sized degree as Celsius but zero at the coldest possible temperature
• •A Celsius degree is 1.8 times the size of a Fahrenheit degree
• •The two scales cross at minus forty, where both read the same number
• •Kelvin requires no degree symbol, since it measures from an absolute zero rather than a chosen reference

The absolute scale differs in kind rather than merely in calibration. Cooling a gas reduces its pressure and volume in a way that, extrapolated, would reach zero at a definite temperature, and that convergence for all gases pointed to a real lower limit rather than a convention. Absolute zero corresponds to the minimum possible thermal energy, and no system can be cooled to it, which is a consequence of thermodynamics rather than a technical limitation. Because the zero is real, ratios on the scale are meaningful, so a doubling of absolute temperature means a genuine doubling of thermal energy, which is not true on a scale with a chosen zero. That is why every physical equation involving temperature uses the absolute scale, and why using Celsius in such an equation gives nonsense.

Liquid expansion is only one way of sensing temperature and most modern measurement uses something else. A thermocouple exploits the small voltage produced where two different metals join, which varies with temperature and covers an enormous range. Resistance thermometers use the change in electrical resistance of a metal, usually platinum, and are the standard for accurate work. Thermistors do the same with semiconductor materials and respond quickly. Infrared instruments read the radiation an object emits and require no contact at all, which is why they can measure a furnace interior or a distant surface, though they need to know how efficiently the surface radiates or the reading is wrong. Thermal imaging extends this to a picture. And at extremes, neither expansion nor resistance works, so very low temperatures are measured by magnetic properties and very high ones by the spectrum of emitted light.

The definitions have been tightened repeatedly and the current arrangement abandons water entirely. Celsius was originally defined by water's freezing and boiling points, then redefined in terms of a single point where ice, water and vapour coexist, which is more reproducible than boiling because it does not depend on pressure. Since 2019 the kelvin has been defined by fixing the value of a fundamental constant relating energy to temperature, which means the scale no longer depends on any substance at all and can be realised by several independent methods. Celsius is now defined as a fixed offset from that. Fahrenheit is defined by conversion from Celsius. Practical measurement uses a documented scale of reference points, including the freezing and triple points of specified pure substances, which laboratories reproduce to calibrate instruments.

A scale needs two reproducible reference conditions, and everything else follows from which two were chosen, which is why scales differ. The absolute scale is different in kind because its zero is a real physical limit rather than a convention, which makes ratios meaningful and is why physical equations require it. Since 2019 the definition fixes a fundamental constant and no longer depends on water.