Why is freezing 32 degrees Fahrenheit? Daniel Gabriel Fahrenheit set 0 with a colder mixture of ice, water and sal ammoniac (ammonium chloride) or sea salt, not with freezing water. Measured from that zero, ice and water read 32 on his thermometers. His 1724 paper does not explain why that mark is exactly 32.
The boiling point of 212 was not chosen at all. Fahrenheit measured it: rain water rose to that degree on his scale as it boiled. Because the boiling point depends on air pressure, a Royal Society of London committee later set rules for marking it. Its 1777 report records 32 and 212 as the two fixed points makers already used.
The three fixed points Fahrenheit published in 1724
The first published definition of the Fahrenheit scale is in a Latin paper that Fahrenheit published in the Philosophical Transactions in 1724. In this paper on freezing water in a vacuum, he says his weather thermometers run from zero to 96 (scan of Fahrenheit's 1724 paper). The scale rests on three fixed points:
- 0: a mixture of ice, water and sal ammoniac or sea salt. A thermometer placed in it drops to the zero mark. Fahrenheit notes that this works better in winter than in summer.
- 32: ice and water mixed without the salts. He calls it the point where freezing begins (terminus initii congelationis), because in winter still water gets a thin skin of ice when his thermometer reaches this degree.
- 96: a thermometer held in the mouth or armpit of a healthy person until it takes on the full heat of the body.
The same paper describes a separate fever thermometer with its scale extended to 128 or 132 degrees. His thermometer for boiling liquids also starts at zero and runs to about 600 degrees, because mercury itself starts to boil near there.
Fahrenheit does not say in the paper why he picked 32 and 96. The numbers appear only as the marks that build the scale.
Supercooled water and the 32-degree mark
An experiment in the same paper describes water staying liquid in air colder than 32 degrees. On March 2, 1721 (Old Style), Fahrenheit set out a glass globe that had been emptied of air and half filled with rain water. The thermometer beside it showed the air at 15 degrees. The water was still liquid through the night and at five the next morning.
When he broke the tip and let air in, the water filled at once with thin plates of ice. He then broke the globe and dropped some of the ice into a glass of water. A little later that water was also full of ice plates, and a thermometer placed in the mix read 32.
Water that stays liquid below its freezing point is now called supercooled. According to a 2024 release from the American Chemical Society (ACS), ice crystals form more readily on tiny impurities such as soot or bacteria. That is why everyday water usually starts to freeze at 32 °F, while completely pure water can stay liquid below it.
Boiling point 212: a measurement, not a choice
The boiling point of 212 appears in another paper Fahrenheit published the same year, in a table of the heat of boiling liquids. Rain water reached 212 degrees when boiling, and spirit of wine (alcohol) reached 176 (English translation of the paper). Fahrenheit does not give the air pressure for these measurements.
The same paper shows that the boiling tests used the same degree marks. Fahrenheit says he corrected the densities of the liquids to 48 degrees. On his thermometers, he adds, that degree sits midway between the strongest artificial cold and the heat of healthy blood. Those two references are 0 and 96, and 48 is exactly halfway (Latin scan of the paper).
So in 1724, 32 was a mark that built the scale and 212 was a result read from it. The document that sets how to mark 212 came about half a century later.
The origin of 32 and 96: the accounts disagree
How Fahrenheit arrived at 32 and 96 has been argued over for a long time, since his paper gives no reason. In a 1937 paper, J. Newton Friend wrote that numbers as awkward as 32 and 212 had been debated since the scale was accepted (Friend's paper in Nature). Published opinions, he added, differed widely.
According to Friend, Martine, a contemporary of Fahrenheit, thought Fahrenheit built the scale haphazardly once his zero was fixed.
Some sources trace the numbers to the thermometer of the Dane Ole Rømer. The table below sets four sources side by side on 32 and 96 and on the zero point.
| Source | Where 32 and 96 come from | Zero point | 212 |
|---|---|---|---|
| Fahrenheit, 1724 papers (primary) | Given as fixed points, no reason | Ice, water and sal ammoniac or sea salt, no proportions given | Measured in rain water |
| Britannica (January 16, 2026) | First 30 and 90, then 32 and 96 | An equal mixture of ice and salt | One of the two points its definition of the scale uses |
| Ashworth, Linda Hall Library (2016) | Rømer: water 7.5, body 22.5. Splitting each degree interval into four gives 30 and 90, then 96 was preferred | Rømer's mix of salt, water and ice | Measured, "wasn't planned" |
| Machin, NPL (2024) | Three points Fahrenheit "claimed" | Equal parts ice, water and salt | Not discussed |
The fullest version of the Rømer story comes from historian of science William B. Ashworth Jr., writing for the Linda Hall Library.
In his account, Fahrenheit split each degree interval of Rømer's scale into four equal parts. That multiplied Rømer's values by four (the 30 and 90 in the table). He then moved body heat to 96 for what Ashworth calls "complicated reasons" and stretched the scale a little.
Ashworth adds that the boiling point of water came out on this scale without being planned.
Popular accounts add one more explanation: there are 64 degrees between 32 and 96, and 64 is two to the sixth power. In this telling, halving the interval again and again makes the degree lines easier to mark. The paper in which Fahrenheit describes his scale gives no such reason. Forum claims such as "the coldest winter day" or "animal blood" have no support in any source in the table.
The sources also part ways on the zero point. Fahrenheit's Latin text gives no proportions for the mixture. Britannica's entry on the Fahrenheit scale describes an equal ice-salt mixture. Graham Machin of the UK's National Physical Laboratory (NPL) writes of equal parts ice, water and salt in a 2024 paper.
Machin also doubts that this lowest point was reliable and thinks only 32 and 96 were usable fixed points in practice.
The 1777 Royal Society report and how to mark 212
A committee report published by the Royal Society in 1777 records 32 and 212 as the two fixed points then in use on Fahrenheit thermometers. It says that everyone who made and used these thermometers agreed on two numbers. The point in melting ice or snow was to be called 32, and the heat of boiling water 212 (scan of the 1777 report). What the report added were rules for setting these points.
Henry Cavendish was among the report's signers. Written 41 years after Fahrenheit's death, the report never mentions 96, the salt mixture or body heat. Its adjustment instructions cover only the freezing and boiling points.
The committee's real concern was the boiling point. The report says that point sat at least two or three degrees higher on some thermometers than on others. This held even for the best makers. The first cause it names is that nobody had settled at what barometer height the point should be set.
The report shows the effect of air pressure with a rule based on De Luc's experiments. By this rule, the boiling point of water shifted by 1.59 °F as the barometer rose from 29.5 to 30.5 inches.
So the committee recommended marking the boiling point in steam with the barometer at 29.8 inches (29.5 inches if done in a closed vessel). In modern inches, a length converter puts that height at about 757 mm.
According to Britannica, the final form of the scale required 98.6 for normal body temperature instead of 96. By today's formula, 96 °F is about 35.6 °C.
Where the 180 degrees and the 1.8 factor come from
The 180 degrees between freezing and boiling are the gap between the two fixed points: 212 minus 32 leaves 180.
The Celsius scale puts 100 degrees across the same interval, so the two intervals stand in a ratio of 180 to 100.
Reduce that ratio with the ratio calculator and you get 9:5. So a temperature difference of 1 °C equals a difference of 1.8 °F.
A single reading also needs the shifted zero. To go from Fahrenheit to Celsius, subtract 32 and divide the result by 1.8. The temperature converter does this in both directions and shows each step.
What sets 32 °F on today's scale?
Today 32 °F follows from a chain of linked definitions. The National Institute of Standards and Technology (NIST) publishes a guide to the International System of Units (SI). In it, Celsius temperature equals the kelvin temperature minus 273.15, and Fahrenheit is tied to Celsius by a fixed formula (NIST conversion table).
In this chain, 32 °F corresponds exactly to 0 °C and 273.15 kelvins.
The kelvin itself has been defined since 2019 by a fixed value of the Boltzmann constant. The International Bureau of Weights and Measures (BIPM) publishes this definition of the kelvin. By the NIST conversion table, a temperature difference of 1 °F equals 5/9 of a kelvin.
Other units have been tied to metric definitions too. The inch is exactly 2.54 cm by definition, and the pound's kilogram value was fixed by agreement. A similar international decision sets the nautical mile at 1,852 meters.
According to the BIPM guide to the international temperature scale, the 1927 scale used water for two of its defining points. The melting point of ice was 0.000 °C and the boiling point of water was 100.000 °C. In 1954, the triple point of water (where ice, liquid water and water vapor coexist in equilibrium) became the only point defining the kelvin.
The boiling point of water stayed a defining point of the international scale. According to the same guide, the 1968 scale added the freezing point of tin as a permitted alternative to it.
The International Temperature Scale of 1990 (ITS-90) does not list the boiling point of water among its defining points. Water appears only as its triple point at 0.01 °C, which is 32.018 °F.
So today 212 °F is the formula's equivalent of 100 °C. The temperature at which water boils at sea level is a measured value. The table below sums up how the references behind the scale have changed since 1724.
| Year | What happened | Reference near freezing | Role of the boiling point |
|---|---|---|---|
| 1724 | Fahrenheit's papers: 0 for an ice and salt mixture, 96 for a healthy person | Ice water at 32, a fixed point | 212, measured |
| 1777 | Royal Society committee records the points in use and recommends marking boiling at 29.8 inches of barometer | Melting ice or snow at 32 | 212, fixed point |
| 1927 | First international temperature scale | Melting ice at 0.000 °C | 100.000 °C, defining point |
| 1954 | Kelvin defined by the triple point of water | Triple point at 0.01 °C | Stays a defining point, with tin as an alternative from 1968 |
| 1990 | ITS-90 takes effect | Triple point at 0.01 °C | Not among the defining points |
| 2019 | Kelvin defined by the Boltzmann constant | 32 °F = 0 °C = 273.15 K, by formula | 212 °F = 100 °C, by formula |
By the last row, 32 and 212 now come from linked definitions. Both still trace back to Fahrenheit's readings of ice water and boiling water, but water itself no longer defines the scale.






