Celsius, Fahrenheit and the delta trap nobody warns you about
Nine out of ten temperature conversions online use the wrong formula for one specific — very common — case. Here is the case, and the fix.
The formula everyone learns for converting Celsius to Fahrenheit is `F = C × 9/5 + 32`. It works perfectly for a single temperature reading — outside is 20 °C, that is 68 °F, done. It gives the wrong answer for a temperature *difference*, and "a temperature difference" is what half of all real conversions actually are.
The two rules, spelled out
A *point* on the scale converts with an offset: F = C × 9/5 + 32 > A *difference* on the scale converts without an offset: ΔF = ΔC × 9/5
"The reactor climbed by 20 °C" is a difference. Convert it with `ΔF = 20 × 9/5 = 36 °F`. Not 68. The 32 in the point formula shifts between the two zero points (0 °C = the freezing point of water, 0 °F ≈ the coldest a saltwater mix will freeze), and there is no zero-point shift when you are talking about a change. If you leave the 32 in for a delta you are effectively adding the freezing point of water to your temperature change, which is obviously wrong once you write it out.
Where this bites
**Cooking.** "Reduce the oven by 20 °C" translates to "reduce by 36 °F", not "reduce to 68 °F." Google search results for this exact phrasing routinely get it wrong.
**HVAC.** A datasheet claiming a "20 °C rise across the coil" means a 36 °F rise. The correct absolute inlet/outlet numbers depend on the intake temperature.
**Engineering specs.** A material with thermal expansion of 12 µm/m per °C has thermal expansion of 6.667 µm/m per °F — divide by 1.8, do not add 32.
**Weather forecasts.** "Temperatures will fall by 15 degrees tonight" is a delta. If the source is Fahrenheit, that is a 15 °F drop = an 8.3 °C drop.
Rate units follow the delta rule
Anything with °C or °F in the denominator — thermal expansion, specific heat capacity, heat-transfer coefficient, temperature coefficient of resistance — uses only the ×9/5 factor with no offset. Specific heat of water is 4.186 J/g·°C = 2.326 J/g·°F. Convert by dividing by 1.8, never by adding 32.
And then there's Kelvin
Kelvin is the SI unit of temperature and it is a *thermodynamic* scale — zero really is zero energy, not the freezing point of water. `K = C + 273.15` exactly. Deltas are the same in Kelvin and Celsius: `ΔK = ΔC`. This is why physics equations use Kelvin: absolute-zero-referenced values behave nicely under multiplication and division, and there is no offset to worry about.
There is a fourth scale, **Rankine (°R)**, which is to Fahrenheit what Kelvin is to Celsius — 0 °R = absolute zero, size of degree same as Fahrenheit. American thermodynamics textbooks use it. `°R = °F + 459.67`.
Five-second rule
For a single reading, use the offset formula. For a change, difference, rate, or anything with °C in the denominator, use only ×9/5 (or ÷1.8 going the other way). If you are unsure whether a value is a point or a delta, ask what happens when the input is zero — a 0 °C point is 32 °F; a 0 °C change is 0 °F.
Try the converters mentioned in this article
FAQ
How do I convert a 20 °C temperature rise to Fahrenheit?+
20 × 9/5 = 36 °F rise. Do not add 32 — that offset only applies to converting a temperature reading, not a difference.
Is a 1 °C change the same size as 1 K?+
Yes, exactly. Celsius and Kelvin have identically sized degrees; only their zero points differ.
What is absolute zero in Fahrenheit?+
-459.67 °F, or 0 °R (Rankine).
Why is the freezing point of water not zero in Fahrenheit?+
Fahrenheit set his zero at the coldest temperature he could reliably reproduce — a brine of ice, water and ammonium chloride. Water freezing was defined as 32 °F later.
This article was written by The Turbo Unit Converter engineering desk (Metrology & tooling) and last reviewed on 2026-06-10 against NIST SP 811 and the BIPM SI Brochure. Read our full editorial policy.