How this instrument works
Daniel Gabriel Fahrenheit, a Danzig-born instrument maker working in Amsterdam, published his scale in 1724, having first made mercury thermometers repeatable enough to be worth naming after himself. Zero was the coldest point he could hold steady — ice, water, ammonium chloride in slurry; ice alone melted at 32, blood heat sat near 96. Anders Celsius, an astronomer at Uppsala, proposed a hundred-step scale in 1742 but ran it upside down, with 0 for boiling, 100 for melting ice. Colleagues reversed it soon after his death, then everyone said centigrade until 1948, when the General Conference on Weights and Measures settled on degree Celsius.
That 5⁄9 is counted, never measured. Both scales were pinned to identical phenomena at standard pressure; Fahrenheit laid 180 divisions across that gap where Celsius laid 100, so 100⁄180 reduces to 5⁄9, a plain rational number, while 32 merely shifts an origin. Foundations have moved since: kelvin has been fixed by Boltzmann's constant since 2019, degrees Celsius follow from kelvins as t = T − 273.15, so Fahrenheit now hangs off Celsius by this very relation. No step in that arithmetic is an approximation.
Daily weather still arrives in Fahrenheit across America, plus one short list of places tied closely to it — Belize, Palau, the Bahamas, the Cayman Islands, the Marshall Islands. US ovens, thermostats, pool heaters, food-safety charts all speak it too, which is why recipes at 350 °F want the dial set near 177 °C, or why 165 °F printed on the poultry packet becomes 73.9 °C on probes marked in metric.
- Type your reading into the Fahrenheit (°F) field; 98.6 °F, ordinary body heat, is loaded as a starting point.
- Read your answer off the Celsius (°C) line, which refreshes on each keystroke to six decimal places.
- Negative readings are fine: enter -40, where both scales agree, a crossing that happens exactly once.
- Converting a gap between two temperatures rather than a single reading? Drop 32, scale by 5⁄9 alone.
Worked example — 98.6 °F on a clinical thermometer
An American-made ear thermometer reports 98.6 °F, while a chart in front of you is ruled in Celsius. Put 98.6 into the Fahrenheit (°F) field. By hand it goes: 98.6 − 32 = 66.6, then 66.6 × 5 = 333, then 333 ÷ 9 = 37. Your Celsius (°C) line reads 37.0, with no remainder to round away.
Landing on whole 37 is no accident. That 98.6 was manufactured by converting 37 °C — the average Carl Wunderlich published in 1868 from an enormous series of armpit readings — so pushing it back through this formula merely undoes his arithmetic. Treat that trailing decimal with suspicion: later studies place typical adults nearer 36.6 °C, while any single reading wanders by several tenths depending on hour of day, or on where the probe was placed.
Questions
Is that 5⁄9 factor exact, or has it been rounded?
Exact. It is the ratio of two counted intervals, not some measured constant: 180 Fahrenheit degrees cover precisely what 100 Celsius degrees cover, from ice point up to steam point, so 100⁄180 reduces to 5⁄9. Written out it runs 0.5555… forever, so anyone working on paper should keep that fraction, doing it in two moves — multiply by 5, divide by 9 — instead of typing 0.556 while inheriting an error in the third decimal place.
Why must I subtract 32 before multiplying?
Because these scales disagree about where zero belongs as well as how large a degree is. Taking away 32 lines up two origins, moving your reading onto a count that starts at melting ice; only then does stretching by 5⁄9 mean anything. Reverse that order — multiply first, subtract afterwards — for answers roughly 14 degrees adrift at every point on either scale. This single slip accounts for most mistakes people make doing temperature conversions on paper.
How do I convert temperature differences rather than readings?
Scale by 5⁄9, leaving 32 out. An offset belongs to one point on the scale, never to the gap between two points, so 18 °F of rise means 10 °C of rise, while ±5 °F of tolerance is ±2.78 °C. Pushing differences through the full formula is the classic blunder in translated engineering specifications: narrow acceptance bands emerge enormous, negative, or both, yet reviewers often miss it because every step of arithmetic looks correct.
Does any temperature read identically on both scales?
Yes: −40. Set both sides equal in this formula, giving x = (x − 32) × 5⁄9, which solves to x = −40, so −40 °F names one physical temperature with −40 °C. That crossing is unique, which makes it a quick audit: feed −40 into any converter you are testing, then treat any result other than −40 as proof its arithmetic is broken.
Can I estimate this in my head?
Subtract 30, then halve. For weather it lands close enough to be worth trying: 70 °F estimates as 20 °C against a true 21.1, while 90 °F gives 30 against 32.2. Halving amounts to using 5⁄10 in place of 5⁄9, so a guess always runs low, with that shortfall growing as you climb. Below roughly 0 °F, or above 100 °F, abandon this trick for real arithmetic.
Is Fahrenheit an imperial unit or a US customary one?
Neither, strictly. Imperial units and US customary units cover length, mass, volume; Fahrenheit sits outside both, shared by Britain with America until UK forecasts switched to Celsius during the 1960s. That is why an older British cookbook may quote Fahrenheit alongside gas marks while weighing flour in grams. For absolute work, US engineering uses Rankine, built on a Fahrenheit-sized degree: °R = °F + 459.67.