The formula only applies to points, not to changes
°F = °C × 9/5 + 32 converts a temperature reading. It does not convert a temperature change. This single distinction accounts for more real-world conversion errors than any other rule on this site, because the formula looks like it should apply universally, and it very deliberately does not.
Worked example: the interval error
A recipe instruction says "increase the oven by 10°C." Fed naively through the point formula, 10 × 9/5 + 32 = 50, so someone might turn the dial up by 50°F. The correct adjustment is 10 × 9/5 = 18°F — no +32, because the offset only exists to align the two scales' zero points, and a change has no zero point to align. Get this wrong on an oven already at 350°F and you have asked for 400°F when you meant 368°F.
The rule that actually generalises: for a temperature difference ΔT, ΔF = ΔC × 9/5. For an absolute reading, °F = °C × 9/5 + 32. Two different operations, one shared coefficient.
Kelvin and Rankine: the two scales with no negative numbers
Celsius and Fahrenheit both have arbitrary zero points. 0°C is the freezing point of water at standard pressure; 0°F was Daniel Fahrenheit's best approximation, around 1724, of the coldest stable brine mixture he could produce in his laboratory. Neither zero means "no thermal energy." Kelvin and Rankine fix that: both set zero at absolute zero, the theoretical point of minimum possible thermal energy, and neither scale allows negative values in ordinary physical use.
The two absolute scales differ only in step size. A kelvin is exactly the same size as a degree Celsius; a degree Rankine is exactly the same size as a degree Fahrenheit. So K = °C + 273.15, and °R = °F + 459.67 — the same structure, different constants, because each is built on top of a different-sized degree.
Worked example: the absolute-scale error
A materials datasheet specifies a process temperature of 300°C. Someone working in a Kelvin-only calculation converts by simply relabelling: "300, but Kelvin." That is wrong by 273.15 — the correct value is 573.15 K. This is a common early-career engineering mistake, and a persistent one, because 300 and 573 are close enough that a downstream result does not look obviously broken; the error tends to surface only later, as an unexplained factor-of-two-ish discrepancy in a final answer.
Worked example: the formula-direction error
Converting Fahrenheit back to Celsius is not "undo the multiplication, then undo the addition" applied in the same order as the forward conversion. The forward formula multiplies then adds; the reverse formula subtracts then multiplies: °C = (°F − 32) × 5/9. Applying the forward formula's order in reverse — subtracting 32 after multiplying by 5/9 instead of before — silently produces a different, wrong number rather than an obvious error, because both orders of operations return plausible-looking temperatures for everyday values.
Where Rankine actually gets used
Rankine survives almost exclusively in US thermodynamics and aerospace engineering, where source data (chamber temperatures, ambient conditions) arrives in Fahrenheit and the calculation, a Carnot efficiency or a gas-law computation, requires an absolute scale. Converting Fahrenheit readings to Kelvin for a US-sourced calculation and then converting the result back to Fahrenheit at the end adds two unnecessary conversions and two more chances for a units mistake; staying in Rankine throughout removes both.
Worked example: the wrong-scale-in-a-ratio error
Take a textbook problem asking for the theoretical maximum efficiency of a heat engine running between a hot reservoir at 500°F and a cold reservoir at 100°F, using the Carnot formula efficiency = 1 − Tcold/Thot. Plugging the Fahrenheit values in directly gives 1 − 100/500 = 80%, a plausible-looking number that is physically meaningless, because Carnot efficiency is only defined for absolute temperatures. Converting first to Rankine (959.67°R and 559.67°R) gives the correct answer: 1 − 559.67/959.67 ≈ 41.7%. The gap between the wrong answer (80%) and the correct one (41.7%) is large enough that this particular error usually gets caught, but only because efficiency has an intuitive plausible range; the same mistake buried inside a longer calculation chain often is not.
ITS-90: what a "calibrated" thermometer is actually calibrated against
True thermodynamic temperature, the quantity Kelvin formally measures, cannot be read directly off an instrument; it has to be inferred from a physical law (gas pressure, radiation spectrum, electrical noise) with its own measurement uncertainty. Since 1990, calibration laboratories worldwide have instead used the International Temperature Scale of 1990 (ITS-90): a set of defined fixed points (the triple point of argon, the freezing point of aluminium, and others) plus specified interpolating instruments, mostly platinum resistance thermometers, that let two labs on opposite sides of the world agree on "100.00°C" to within a few thousandths of a degree without either one measuring true thermodynamic temperature from scratch. ITS-90 replaced the earlier IPTS-68, and it is itself an approximation to thermodynamic temperature, accurate to within its stated uncertainty at each fixed point rather than exact by definition. For any conversion this site performs, the difference between ITS-90 and true thermodynamic temperature is far too small to matter; it matters in metrology laboratories measuring to five or six significant figures.
A working checklist
- Converting a reading — weather, an oven setting, body temperature? Use the full formula, offset included.
- Converting a change or a difference? Drop the offset, keep only the multiplier.
- Feeding a temperature into a formula that requires an absolute scale — any efficiency, gas-law, or radiation calculation? Convert to Kelvin or Rankine first, never Celsius or Fahrenheit.
- Working entirely with US-sourced Fahrenheit data inside an absolute-scale formula? Rankine avoids two redundant conversions that routing through Kelvin would require.