Temperature
Celsius, Fahrenheit, Kelvin
Celsius and Fahrenheit agree exactly once, at −40. Set the two scales equal and solve: C = 1.8C + 32 gives −0.8C = 32, so C = −40. It is the only temperature you never have to convert, which is why aviation and refrigeration specifications quote it without a unit.
This converter carries four scales. Celsius and Fahrenheit are interval scales with arbitrary zeros; Kelvin and Rankine are absolute, starting at the point where thermal motion stops. Type into any of the four boxes and the other three update, with everything routed through kelvin internally and displayed to four decimal places.
The Fahrenheit box opens at 72, a US thermostat setting, which is 22.2222 °C. If you have ever wondered why the United States kept a scale nobody else uses, see Why does the US still use Fahrenheit?
Fixed points worth knowing by heart
| Reference | °C | °F | K |
|---|---|---|---|
| Absolute zero | −273.15 | −459.67 | 0 |
| The point where the scales agree | −40 | −40 | 233.15 |
| Dry ice sublimes | −78.5 | −109.3 | 194.65 |
| Water freezes | 0 | 32 | 273.15 |
| Food-safe fridge maximum | 4 | 39.2 | 277.15 |
| Clinical fever threshold | 38.0 | 100.4 | 311.15 |
| Water boils at 101.325 kPa | 100 | 212 | 373.15 |
| Water boils in Denver (1,609 m) | ~94 | ~201 | ~367 |
| Bread oven | 190 | 374 | 463.15 |
The fever row is the one people convert most often and get wrong: 100.4 °F is not a rounded threshold, it is exactly 38.0 °C, chosen in Celsius and converted for American charts.
The conversions, written out
°F = °C × 9/5 + 32°C = (°F − 32) × 5/9K = °C + 273.15°R = °F + 459.67, and°R = K × 9/5
The 9/5 comes from the degree sizes: 180 Fahrenheit degrees span the same interval as 100 Celsius degrees between water’s freezing and boiling points. The 32 is the offset between the two zeros.
For mental conversion, "double it and add 30" is fast and drifts fast: 20 °C gives 70 against a true 68, but 30 °C gives 90 against a true 86. The exact version costs one extra step because 1.8 is 2 minus 10 percent — double, subtract a tenth of the doubled figure, add 32. For 30 °C: 60, minus 6, plus 32, equals 86 exactly.
Interval scales, absolute scales, and why differences convert differently
Celsius and Fahrenheit place their zeros by convention, so their numbers support addition but not ratios. Twenty degrees Celsius is not twice as hot as ten. Three hundred kelvin genuinely is twice the thermodynamic temperature of 150 K, which is why every gas law, radiation calculation and efficiency formula demands an absolute scale.
The practical consequence catches people in HVAC, brewing and lab work: a temperature difference converts with the ratio alone, never with the offset. A rise of 10 °C is a rise of 18 °F, not 50 °F. A thermostat deadband of 2 °F is 1.11 °C. Subtract the 32 only when converting a position on the scale, never when converting a gap between two positions.
Notation: kelvin takes no degree sign and no capital when spelled out — "300 K", "three hundred kelvin", not "300 °K". The degree sign was dropped by the 13th General Conference on Weights and Measures in 1967. Rankine keeps its degree symbol and its Fahrenheit-sized steps, and survives mainly in American thermodynamics and combustion engineering.
An oven, a fever, and a temperature gap
A British recipe. "180 °C" converts to 180 × 1.8 + 32 = 356 °F. American ovens are dialled in 25-degree steps, so you set 350 °F and accept the six degrees. The usual ladder: 160 °C is 320 (set 325), 200 °C is 392 (set 400), 220 °C is 428 (set 425), 230 °C is 446 (set 450). If the recipe specifies fan or convection, US convection guidance is to reduce a conventional temperature by about 25 °F, so a fan setting is not directly comparable to a conventional one.
A thermometer reading. A child at 39.4 °C converts to 39.4 × 1.8 + 32 = 102.9 °F. Against the 38.0 °C / 100.4 °F threshold that is a fever of 1.4 Celsius degrees, or 2.5 Fahrenheit degrees — the same gap expressed twice.
A gap, not a position. A fermentation protocol calling for a swing of no more than 3 °C over 24 hours allows 3 × 1.8 = 5.4 °F. Running it through the full formula would give 37.4 °F, which is a temperature, not a tolerance.
Where the textbook fixed points are not quite fixed
The kelvin no longer depends on water. Since 20 May 2019 the kelvin has been defined by fixing the Boltzmann constant at exactly 1.380649 × 10^−23 J/K. The triple point of water, which had defined it as exactly 273.16 K since 1954, is now a measured quantity with an uncertainty attached. The everyday numbers did not move; the definition underneath them did.
Water does not reliably boil at 100 °C. That figure assumes a pressure of exactly 101.325 kPa, and on the ITS-90 temperature scale the boiling point at that pressure is 99.974 °C. Altitude matters far more: Denver at 1,609 m sits near 83 kPa, where water boils around 94 °C, which is why boxed cake mixes carry separate high-altitude instructions and why pasta takes longer in the mountains.
98.6 °F is a rounding artefact. It comes from Carl Wunderlich’s 19th-century average of 37.0 °C, converted to one decimal place, which gave the American figure a false air of precision. Large modern datasets put the mean adult temperature closer to 36.6 °C, about 97.9 °F, and individual baselines vary by roughly half a degree Celsius across the day.
Display rounding. Results are shown to four decimal places, so a value such as 5/9 of an odd Fahrenheit reading can round-trip back a ten-thousandth away from where it started.
Which scale to report in
Weather and cooking follow the audience: Celsius for most of the world, Fahrenheit for a US reader. Anything thermodynamic — ideal gas calculations, Stefan-Boltzmann radiation, Carnot efficiency — needs kelvin, because those formulas divide by temperature and an arbitrary zero would make the ratios meaningless. Rankine appears where an American engineering text wants absolute temperature without leaving Fahrenheit-sized degrees.
Scientific papers and datasets should carry kelvin or Celsius with the unit spelled out; a bare number is ambiguous anywhere near room temperature, where 20 could plausibly be either scale. Below −40 and above 100 the two are far enough apart that context usually disambiguates.
Frequently asked questions
Why is -40 the same in Celsius and Fahrenheit?
Setting C = 1.8C + 32 and solving gives C = -40, the single crossing point of the two scales. Below it Fahrenheit readings are lower than Celsius ones; above it they are higher.
Is it 300 K or 300 degrees K?
300 K. The degree sign was dropped from the kelvin by the 13th General Conference on Weights and Measures in 1967, and the unit name is written in lower case even though the symbol is capital K.
How do I convert a temperature difference rather than a temperature?
Multiply by 9/5 or 5/9 only, with no offset. A rise of 10 degrees Celsius is a rise of 18 degrees Fahrenheit; adding 32 would turn a gap into a position on the scale.
What oven setting should I use for a 180 C recipe?
180 C is 356 F, so set 350 F on an oven dialled in 25-degree steps. If the recipe specifies a fan or convection oven, US convection guidance runs about 25 F below the conventional temperature.
Why does water boil below 100 C where I live?
Boiling point falls with air pressure. At 1,609 m in Denver, around 83 kPa, water boils near 94 C. The 100 C figure assumes a pressure of exactly 101.325 kPa at sea level.
What is Rankine used for?
It is an absolute scale with Fahrenheit-sized degrees, so 0 R is absolute zero and R = F + 459.67. It survives in US thermodynamics, combustion and aerospace work where formulas need an absolute temperature but the surrounding numbers are in Fahrenheit.