Celsius, Fahrenheit, Kelvin — three zeros, three worlds

Zero degrees is not one temperature. On the Celsius scale it is where black ice begins. On the Fahrenheit scale it is the kind of cold that freezes your eyelashes. On the Kelvin scale it is the moment when not one scrap of heat can be squeezed out of matter. Someone chose each of those zeros: once from physics, once from a craftsman's pragmatism, once from a calculation at the edge of what could be known. This is the history of three decisions that still split the world into people who check the weather in two digits and people who check it in three.

Before anyone measured heat

For most of history, hot and cold were sensations — things you felt, not things you counted. The first instruments, thermoscopes, appeared around the turn of the seventeenth century in the work of Galileo, Robert Fludd and Santorio Santorio. They were open glass tubes holding a liquid; the air trapped inside expanded and contracted with temperature. The trouble was that it also responded to atmospheric pressure. The same instrument read differently before a storm and after it.

The breakthrough came in 1654, when Grand Duke Ferdinando II de' Medici of Tuscany built the first sealed liquid-in-glass thermometer, freeing the reading from the barometer. Yet for another half-century every physicist calibrated his own instrument his own way: against the coldest day of the previous winter, the temperature of the laboratory cellar, or the warmth of freshly drawn milk. Until two scholars could reproduce the same reference point independently, no number meant anything outside the room where it was taken.

Fahrenheit: a zero from a bucket of salt

Daniel Gabriel Fahrenheit was born in 1686 in Gdańsk. When he was fifteen, both his parents died on the same day — 14 August 1701 — most likely after eating poisonous mushrooms. His guardians packed the orphaned boy off to Amsterdam for a merchant's apprenticeship and a training in bookkeeping. Fahrenheit preferred glassblowing and physics. According to surviving accounts his legal guardians took this so badly that they had an arrest warrant issued against him, intending to press the young man into service on the ships of the Dutch East India Company and send him to Asia.

So for several years Fahrenheit moved around Germany, Sweden and Denmark, learning his craft and staying ahead of the warrant. In Copenhagen in 1708 he met Ole Rømer — the astronomer who first measured the speed of light, and who was then working on a thermometric scale of his own. Rømer had set 0 at the freezing point of brine, 7.5 at the melting of pure ice, 22.5 at "blood heat" and 60 at the boiling of water. Fahrenheit saw what a scale built on reproducible points looked like, and set about building his own. The warrant was soon withdrawn; he settled in Amsterdam and opened a workshop.

Rømer's fractions annoyed him, so he multiplied everything by four: ice melted at 30, body heat sat at 90. Then he shifted the reference points once more, this time with calibration in mind.

He defined zero as the lowest temperature he could reliably reproduce on his own bench: a mixture of ice, water and ammonium chloride (sal ammoniac) in a 1:1:1 ratio by mass. That is a eutectic system — while it melts it holds its own temperature steady, at what we now call roughly −17.78 °C. There is also an anecdote that the zero was meant to be the lowest reading of the bitter winter of 1708–09 in his native Gdańsk, and that Fahrenheit only later devised the brine mixture in order to reproduce that cold on demand. The mixture is documented; the rest is a good story.

Why 32 and 96, of all numbers

The second point was a mixture of water and ice without salt: 32 °F. The third was the temperature of a healthy human body, estimated at 96 °F. (A popular anecdote has Fahrenheit taking that reading under his wife's armpit; Fahrenheit never married, and the sources simply describe a measurement in the mouth or under the arm of a healthy person.)

The choice looks capricious. It is superb engineering. Exactly 64 degrees fit between 32 and 96, and 64 is 2⁶. An eighteenth-century craftsman marking a graduation onto a glass capillary had nothing to calculate: he took a pair of dividers and halved the physical distance between the two control points — into 32 parts, then 16, then 8, 4, 2. The Fahrenheit scale was designed not for a physicist at a blackboard but for a hand at a workbench.

Only after its creator's death did his successors redefine the scale around the boiling point of pure water at standard pressure: 212 °F. That left exactly 180 degrees between freezing and boiling — but it nudged the nominal body temperature from 96 °F to 98.6 °F. That last figure is nothing other than exactly 37.0 °C; it reached us from Carl Wunderlich's nineteenth-century measurements and is a rounding rather than a biological constant — modern studies point closer to 36.6 °C.

Why America stayed with Fahrenheit

The United States is now one of the last countries using Fahrenheit day to day. Not out of stubbornness, but out of arithmetic. The Mendenhall Order of 1893 already defined US customary units legally in terms of the metric standards — formally, the country has stood on the metric system for over a century. But the metric conversion act of 1975 made the switch to SI voluntary, and where a switch is voluntary, the cost of replacing infrastructure wins.

There is, however, an argument Americans make that holds up rather well. The range from 0 °F to 100 °F covers almost exactly the temperatures a human being lives through in a temperate climate: 0 °F is about −18 °C (dangerous cold), 100 °F is about 38 °C (dangerous heat). The scale works like an intuitive percentage of human comfort: 50 °F (10 °C) means chilly, bring a coat; below 30 °F means bring an ice scraper. On top of that, a Fahrenheit degree is smaller — it equals 5/9 of a Celsius degree — so a thermostat can be set precisely using whole numbers.

That is an argument about resolution and range, not about physics. Physics chose otherwise.

Celsius: the scale that ran backwards for three years

Anders Celsius was born in 1701 in Uppsala, into a family in which his father and both grandfathers were professors of the exact sciences. He took the chair of astronomy at twenty-eight. He belongs to history for more than thermometry: watching a compass needle during an aurora, he was among the first to connect the phenomenon with fluctuations in the Earth's magnetic field.

His greatest achievement was taking part in the French geodetic expedition to Lapland (1736–1737) under Pierre-Louis Maupertuis. The team measured an arc of the meridian near the pole and compared it with measurements from Peru. The result confirmed Newton's hypothesis: the Earth is not a sphere but an ellipsoid flattened at the poles. A grateful government funded Celsius's observatory in Uppsala in 1741.

There, between meteorological observations, Celsius spent two years testing mercury thermometers in search of genuinely fixed points. He showed that the melting of ice and snow is practically independent of latitude and pressure, whereas the boiling of water depends strongly on pressure — so the boiling point has to be defined at one agreed atmosphere. That observation is Celsius's real contribution, and it matters more than the numbers themselves.

The numbers he proposed in 1742, in a paper titled Observationer om twänne beständiga Grader på en Thermometer. He divided the scale into 100 degrees — but the other way round from today:

• 0 °C marked the boiling of water,
• 100 °C marked the melting of ice.

This was no whim. In the Swedish climate winter temperatures routinely drop below freezing, and negative values in an observation log are an invitation to a printing error and a transcription mistake. By inverting the scale, Celsius got almost nothing but positive numbers in everyday meteorology. The price was that a rise in temperature meant a fall in the recorded value.

That version survived three years. Shortly after Celsius died in 1744, the scale was turned around. In Sweden this was the work of Carl Linnaeus, who needed a more intuitive instrument to control the climate of the orangery in the Uppsala botanical garden; he ordered from the instrument maker Daniel Ekström a thermometer on which 0 was freezing and 100 boiling. The first Swedish document with readings on the new scale is Hortus Upsaliensis, dated 16 December 1745. Independently and earlier — already in 1743 — the same inversion was carried out in Lyon by Jean-Pierre Christin, who designed the "Lyon thermometer" built by Pierre Casati.

For the next two centuries people spoke of the centigrade scale. The name "degree Celsius" was adopted officially only in 1948, at the 9th General Conference on Weights and Measures — which also removed the clash with the French grade, one hundredth of a right angle.

Why the rest of the world chose water

Celsius won because his reference points are the ones a person bumps into daily. 0 °C is a message to the farmer (frost), to the road crew (black ice) and to the pedestrian (ice on the pavement). 100 °C is cooking and sterilization. Water is everywhere, it is free, and its phase changes mark the boundaries of the biosphere.

A second argument followed: coherence with the metric system. The metre, kilogram and litre were designed as one decimal whole, in which the mass of a litre of water is tied to the kilogram — originally (1795) at the melting point of ice, and from 1799 onward at the temperature of maximum density of water, about 4 °C. The link was never perfectly exact, but it was elegant enough that a scale built on water slotted neatly into a system built on water.

Kelvin: the zero with nothing beneath it

By the mid-nineteenth century thermodynamics had exposed the weakness of both scales: both are relative, anchored to an arbitrary substance. They are useless in equations.

William Thomson, later Lord Kelvin, described in On an Absolute Thermometric Scale (1848) a scale independent of any material whatsoever. His starting point was Sadi Carnot's ideal heat engine, whose efficiency depends only on the temperature difference between source and sink. Thomson noticed that there exists a sink temperature at which efficiency would theoretically reach 100% — meaning no heat could be drawn out of the system at all. That is absolute zero.

The gas laws of Charles and Gay-Lussac led to the same point: the volume and pressure of an ideal gas fall linearly with temperature, so a simple extrapolation yields a temperature of zero volume and zero pressure. Thomson computed it from the coefficient of thermal expansion of air (α ≈ 0.00366 per degree): the reciprocal, 1/0.00366 ≈ 273, giving about −273 °C. Today's value is −273.15 °C — exactly, by definition.

The kelvin is an SI base unit, and since 2019 it has been defined by the Boltzmann constant, fixed as exactly k = 1.380649 × 10⁻²³ J/K. The earlier definition rested on the triple point of water; the new one refers to no substance at all. And one detail that trips up half the internet: the kelvin has no degree. Not "degrees Kelvin" but "kelvins" — the term "degree Kelvin" was abolished in 1968.

What happens just above absolute zero

The textbook definition says temperature measures the average kinetic energy of molecules, so at absolute zero all motion should cease. Quantum mechanics says: not quite.

Under Heisenberg's uncertainty principle, the position and momentum of a particle cannot both be pinned down arbitrarily well. An atom frozen in absolute stillness would have both sharply defined — and that is forbidden. So even at 0 K matter retains a minimal vibration, known as zero-point energy. No heat can be recovered from it; the entropy of the system reaches the lowest value it can take, but matter is not paralysed outright.

Just above zero, matter starts behaving quantum-mechanically on a macroscopic scale: superconductivity appears (electrical resistance vanishes), and superfluidity (zero viscosity — liquid helium can climb the walls of its container), while sufficiently cold atoms may fall into a single shared quantum state, forming a Bose–Einstein condensate.

For the equations of physics an absolute scale is not a convenience but a condition of meaning. The gas laws, energy distributions and radiation laws only work when temperature is positive. Feeding −20 °C into them would yield negative pressure and negative volume. At the same time the kelvin interval is identical to the Celsius degree, so temperature differences carry over between the two scales one to one — converting amounts to a shift of 273.15.

Five scales on one axis

The last two scales are museum pieces. Rankine still hangs on in American engineering thermodynamics — it is absolute like Kelvin but carries a Fahrenheit-sized degree, so it fits the rest of the local arithmetic. Réaumur, for a century and a half the standard of continental Europe and Russia, now survives only in novels and in the recipes of a few cheese dairies.

One more curiosity falls out of the table, and it is easy to check by hand: −40 °C is exactly −40 °F. There is only one temperature at which the two scales show the same number, and it happens to lie where few people care to check.

Three zeros, three questions

Each of these scales answers a different question, and each answers it well.

Fahrenheit asked: how do I build a thermometer that can be reproduced in a workshop? He answered with a slurry of salt and ice, and a graduation that halves cleanly all the way down. Celsius asked: what in the surrounding world is genuinely constant? He answered with water — and, more importantly, with the observation that one of its two fixed points needs a caveat about pressure. Thomson asked: where does heat end? He answered with a limit that cannot be crossed, and in doing so freed temperature from matter altogether.

That is why a thermostat in Ohio reads 72, a forecast in Warsaw reads 22, and a chart from the Webb telescope reads 45 K. These are not three versions of the same number. They are three different answers to the question of where to start counting.

Further reading

• BIPM, The International System of Units (SI Brochure), 9th ed. (2019) — the definition of the kelvin via the Boltzmann constant.
• W. Thomson, On an Absolute Thermometric Scale founded on Carnot's Theory of the Motive Power of Heat (1848) — the source paper; available in the public domain.
• A. Celsius, Observationer om twänne beständiga Grader på en Thermometer (1742) — the original, inverted scale.
• Physics Today (AIP Publishing), Temperature scales: Celsius, Fahrenheit, Kelvin, Réaumur and Rømer — a concise comparison of the historical scales.
• NIST, How Low Can Temperature Go? Lord Kelvin and the Science of Absolute Zero — an accessible account of absolute zero.
• Uppsala universitet, Anders Celsius — a pioneer in investigating the Earth and its changes — a biography from his own university.