The Conversion Every Physics Lab Runs a Hundred Times a Day
The Kelvin scale is the only temperature scale that starts at a physically meaningful zero. Zero kelvin — absolute zero — is the temperature at which a classical ideal gas would occupy zero volume and possess zero thermal energy. The Celsius scale starts at the freezing point of water at one standard atmosphere — a perfectly useful reference for weather, cooking, and biology, but physically arbitrary. Converting from Kelvin (the scale of physics) to Celsius (the scale of human experience) means shifting the zero point by exactly 273.15. That's it. One subtraction.
°C = K − 273.15
The reverse — K = °C + 273.15 — is the Celsius to Kelvin converter. Add 273.15 instead of subtracting. The two operations are inverses of each other, and both are exact. No rounding. No approximation. The offset 273.15 is a defined constant, established by the 10th General Conference on Weights and Measures (CGPM) in 1954 and untouched by the 2019 SI redefinition. The number itself didn't change. The definition of what a kelvin is — and therefore what the number 273.15 actually means — changed entirely.
May 20, 2019: The Day Temperature Became a Statement About Energy
Before 2019, if you asked a metrologist "what is a kelvin?", the answer involved a very specific cell of water in a laboratory near Paris. The triple point of water — the unique temperature and pressure at which ice, liquid water, and water vapor coexist in equilibrium — was defined as exactly 273.16 K. The kelvin was 1/273.16 of the thermodynamic temperature of the triple point of water. To calibrate a thermometer, you needed a triple-point cell: a sealed glass vessel containing pure water of precisely controlled isotopic composition (Vienna Standard Mean Ocean Water, or VSMOW), cooled until all three phases appeared simultaneously. The triple point is exquisitely reproducible — to within about 0.0001 K across different laboratories — but it's still a physical artifact. And physical artifacts drift, break, and are only available in well-equipped national metrology institutes.
The 2019 redefinition — the most comprehensive overhaul of the International System of Units since its creation in 1960 — threw out every artifact-based definition. The kilogram was severed from the platinum-iridium cylinder in Sèvres. The ampere was severed from the force between two hypothetical infinite wires. And the kelvin was severed from the triple point of water. In its place: the Boltzmann constant.
The Boltzmann constant, k, relates the average kinetic energy of particles in a gas to the temperature of that gas. In the kinetic theory of gases, the average translational kinetic energy of a monatomic ideal gas molecule is (3/2)kT. If you know k exactly, and you can measure energy (in joules), you can determine temperature — no water cell required. On May 20, 2019, the CGPM fixed k at exactly 1.380649 × 10−23 joules per kelvin. This number was chosen to match the best experimental measurements available at the time, ensuring continuity — a thermometer calibrated before the redefinition reads the same after it, to within the uncertainty of the best pre-2019 instruments.
The practical consequence is subtle but profound. Temperature is now a mechanical quantity — a measure of energy per degree of freedom per particle. A kelvin is the temperature change that corresponds to a thermal energy change of 1.380649 × 10−23 joules per degree of freedom. The triple point of water is no longer defined as 273.16 K — it is now measured to be approximately 273.16 K, with an uncertainty that depends on how well you can realize the Boltzmann constant in your experiment. The 2019 redefinition didn't change the number you type into a conversion calculator. It changed what the number means: every kelvin-to-Celsius conversion is now, at its deepest level, a statement about the average kinetic energy of molecules.
Why Instruments Speak Kelvin and Humans Hear Celsius
Walk into any materials science lab, any semiconductor fab, any cryogenics facility, and look at the instrument readouts. The mass spectrometer's ion source temperature: 523 K. The diffusion furnace's process setpoint: 1,373 K. The cryocooler's cold head: 4.2 K. Every one of these numbers is in Kelvin, because the physics equations that govern the process — the Arrhenius rate equation for diffusion, the ideal gas law for the mass analyzer, the BCS theory for the superconductor — require absolute temperature. You cannot plug a Celsius number into PV = nRT and get a meaningful answer.
But the process engineer who reads that 1,373 K furnace setpoint doesn't think in Kelvin. She thinks in Celsius: 1,100°C. The lab notebook entry reads "annealed at 1,100°C for 30 minutes." The equipment spec sheet lists the maximum operating temperature as 1,100°C. The safety interlock trips at 1,150°C. Every number the engineer writes down, types into a report, or communicates to a colleague is in Celsius — the unit that maps to human experience. The instrument outputs Kelvin. The human reads Celsius. The conversion happens somewhere in between — sometimes in the instrument's firmware, sometimes in the data acquisition software, sometimes in the engineer's head. When it happens in someone's head, it had better be right.
This is the defining asymmetry of the Kelvin-to-Celsius conversion. The Celsius-to-Kelvin direction — adding 273.15 — is what you use when setting up an experiment. The Kelvin-to-Celsius direction — subtracting 273.15 — is what you use when reading back the results. Both are trivial arithmetic. Both are exact. But the K-to-°C direction is the one that gets done more often, because instruments measure and humans interpret.
Worked Examples
0 K → −273.15°C
0 − 273.15 = −273.15°C. Absolute zero. No classical thermal energy. Quantum zero-point motion persists — the Heisenberg uncertainty principle forbids a particle from having both precisely zero momentum and a precisely defined position — but no work can be extracted from a system at 0 K. The third law of thermodynamics says you can approach 0 K arbitrarily closely but never reach it. The record, set at MIT in 1999 with sodium Bose-Einstein condensates, is roughly 100 picokelvin — 0.0000000001 K, or −273.1499999999°C.
2.725 K → −270.425°C
2.725 − 273.15 = −270.425°C. The cosmic microwave background radiation. Every cubic centimeter of the universe — even the emptiest intergalactic void — glows at this temperature, the afterglow of the Big Bang, redshifted over 13.8 billion years from roughly 3,000 K to 2.725 K. The CMB was discovered in 1964 by Penzias and Wilson, who initially thought the persistent 3 K noise in their horn antenna was pigeon droppings. It won them the 1978 Nobel Prize. The slight anisotropies in this background — temperature variations of about ±0.0002 K across the sky — encode the seeds of every galaxy cluster and cosmic void we see today.
273.15 K → 0°C
273.15 − 273.15 = 0°C. Water freezes. The most fundamental calibration checkpoint in thermometry. A properly constructed triple-point cell — a sealed Pyrex vessel containing pure VSMOW water under vacuum — realizes 273.16 K (0.01°C) with an uncertainty of ±0.0001 K. National metrology institutes use these cells as their primary temperature reference. Your kitchen thermometer does not.
293.15 K → 20°C
293.15 − 273.15 = 20°C. Room temperature. Standard ambient for electronics testing per IEC 60068-1, for dimensional metrology per ISO 1 (the international reference temperature for length measurements is 20°C), and for most chemical thermodynamic data tables. A 5 K deviation from 20°C changes the length of a 100 mm steel gauge block by roughly 5.5 μm — enough to fail a Class 0 calibration.
310.15 K → 37°C
310.15 − 273.15 = 37°C. Human core body temperature — the number Carl Wunderlich reported in 1851 as the mean of over one million axillary readings in Leipzig. Modern studies place the actual mean oral temperature closer to 36.4°C (309.54 K). A 37°C reading in a 75-year-old can be a low-grade fever. Context matters more than the integer.
373.15 K → 100°C
373.15 − 273.15 = 100°C. Water boils at 1 standard atmosphere. Every steam table in mechanical engineering starts here. At 2,000 meters elevation (≈795 hPa), water boils at approximately 93.4°C (366.55 K). Pressure cookers raise the boiling point by pressurizing — at 2 bar gauge (≈3 bar absolute), water remains liquid to about 134°C (407.15 K).
Common Kelvin to Celsius Conversions
| K | °C | Context |
|---|---|---|
| 0 K | −273.15°C | Absolute zero. Unattainable — third law of thermodynamics. |
| 2.725 K | −270.425°C | Cosmic microwave background. The universe's baseline temperature. |
| 4.2 K | −268.95°C | Liquid helium boils. MRI magnets, CERN magnets, quantum computing. |
| 20.3 K | −252.85°C | Liquid hydrogen boils. Rocket propellant for Centaur upper stage, SLS, Ariane 6. |
| 77 K | −196.15°C | Liquid nitrogen boils. The workhorse cryogen. Cheaper than milk. |
| 90.2 K | −182.95°C | Liquid oxygen boils. Pale blue, strongly paramagnetic, oxidizes everything. |
| 194.7 K | −78.45°C | Dry ice sublimation. The coldest temperature most non-scientists ever handle. |
| 233.15 K | −40°C | °F = °C crossover. Jet A-1 freezing specification. Exposed skin danger zone. |
| 255.37 K | −17.78°C | Fahrenheit's ammonium chloride brine zero. Road salt effectiveness drops sharply. |
| 273.15 K | 0°C | Water freezes. Ice point. The fundamental thermometry anchor for 200+ years. |
| 293.15 K | 20°C | Room temperature. ISO 1 reference for dimensional metrology worldwide. |
| 298.15 K | 25°C | Standard lab conditions. Most thermodynamic tables reference this temperature. |
| 309.54 K | 36.4°C | Actual mean human oral temperature (Mackowiak 1992, Obermeyer 2017). |
| 310.15 K | 37°C | Wunderlich's 1851 mean — still the number on every thermometer box. |
| 373.15 K | 100°C | Water boils at 1 atm. Every steam table, distillation column, and autoclave spec begins here. |
| 933 K | 659.85°C | Aluminum melts. Casting, extrusion, rolling — all happen above this threshold. |
| 1,811 K | 1,537.85°C | Iron-carbon eutectic (mild steel melting point). The backbone of structural engineering. |
| 3,688 K | 3,414.85°C | Tungsten melts. Highest melting point of any element. Incandescent bulb filaments. |
| 5,778 K | 5,504.85°C | Solar photosphere. A blackbody at this temperature peaks in the green — the Sun is white, not yellow. |
Kelvin Temperature Differences vs. Absolute Temperatures: A Common Trap
A temperature difference in Kelvin is identical to a temperature difference in Celsius — because the scales share the same degree size. If a sample was heated from 293.15 K to 393.15 K, the temperature rise is 100 K, which is identically 100°C. No subtraction of 273.15 applies to differences. You only subtract 273.15 when converting an absolute temperature — a single point on the scale, not a span.
This distinction trips people up regularly. A data sheet says "thermal stability: ±0.5 K." That means the temperature stays within ±0.5°C of the setpoint — not ±0.5 K in some special Kelvin sense, but ±0.5°C in the most ordinary Celsius sense. The two degree sizes are identical. The symbol "K" on a stability specification is convention, not conversion. If you see "ΔT = 5 K" in an engineering document, write "ΔT = 5°C" in your report with total confidence — and don't add or subtract 273.15 from it.
The trap is worse in the other direction. A temperature specification that reads "storage temperature: −20°C to +60°C" converted carelessly to Kelvin becomes "253.15 K to 333.15 K" — both readings are absolute temperatures and need the offset. But a specification that reads "temperature rise during test shall not exceed 30°C" converted to Kelvin becomes "30 K" with no offset. Seeing "°C" and thinking "add 273.15" is a reflex that must be suppressed when the number is a difference, not a point. Every experienced engineer has made this mistake at least once. The good ones catch it before the oven melts.
Engineering Context
Kelvin is the only temperature unit that can appear in a denominator. When you see a specification like "thermal coefficient of expansion: 12 × 10⁻⁶ /K" or "temperature coefficient of resistance: 0.00385 /K," the K in the denominator is not a temperature point — it's a per-degree scaling factor, and it is identical to /°C. But in equations where temperature appears in a denominator or an exponent — the Arrhenius equation k = A·exp(−Ea/RT), the Stefan-Boltzmann law P = εσT⁴, the ideal gas law PV = nRT — the T must be in Kelvin. Plugging a Celsius temperature into the Stefan-Boltzmann law doesn't just give you a slightly wrong answer; it gives you nonsense. A surface at 0°C (273.15 K) radiates at 315 W/m². Using 0°C as T gives 0 W/m² — off by infinity percent. Every radiative heat transfer calculation in building energy modeling, satellite thermal design, and furnace engineering uses Kelvin internally precisely because Celsius in a T⁴ term produces results that aren't even wrong. The conversion to Celsius happens at the output layer — reports, dashboards, user interfaces. Never in the computation layer. See also: Celsius to Kelvin for setting up those experiments; Fahrenheit to Kelvin when the input data comes from US instruments; and the Temperature Conversion Guide for the full framework.
More temperature conversions: fahrenheit to celsius · celsius to fahrenheit · celsius to kelvin · fahrenheit to kelvin · kelvin to fahrenheit · Guide
Related Unit Converters
Frequently Asked Questions
What is the difference between Kelvin and Celsius — aren't they the same scale shifted?
Yes — and that's exactly what makes the conversion so clean. Anders Celsius's 1742 scale (as later flipped by Linnaeus) and Lord Kelvin's 1848 absolute scale share the same degree size by design. Kelvin explicitly chose to keep the Celsius degree magnitude because it was already the scientific standard in Europe. The scales differ only in their zero point: Celsius anchors at water's freezing point (a convenient but physically arbitrary reference), Kelvin anchors at absolute zero (the point where ideal-gas volume extrapolates to zero and classical thermal energy vanishes). The difference is 273.15 degrees. No scaling factor. No fraction. Just an offset. This makes Kelvin-to-Celsius the simplest affine conversion in the entire SI system — and one of only two affine temperature conversions most engineers ever use (the other being Fahrenheit to Celsius, which requires both a scaling factor and an offset).
Does the 2019 SI redefinition affect the K to °C conversion formula?
Not at all. The formula °C = K − 273.15 is unchanged. What changed is the chain of reasoning that establishes what "K" means. Before 2019: the triple point of water is exactly 273.16 K → therefore k ≈ 1.380649 × 10⁻²³ J/K (measured). After 2019: k is exactly 1.380649 × 10⁻²³ J/K → therefore the triple point of water is approximately 273.16 K (measured). The 273.15 offset — the distance from absolute zero to water's freezing point — remains a defined constant. The 2019 redefinition inverted the logical dependency (from water-defines-kelvin to kelvin-is-defined-by-k-and-kelvin-defines-water) without changing any numeric values. A thermometer calibrated in 2018 and a thermometer calibrated in 2020 read the same temperature to within the measurement uncertainty of either instrument.
When would I use Kelvin to Celsius instead of Celsius to Kelvin?
K to °C is the readout direction — you use it when an instrument or dataset reports values in Kelvin and you need them in human-readable Celsius. Every cryostat controller, thermocouple logger, PID loop tuning interface, and scientific dataset (NASA planetary science, CERN cryogenics, semiconductor fab process data) outputs Kelvin natively. Converting to Celsius is the last step before the number reaches a human. C to K, the reverse conversion, is the setup direction — you use it when configuring an experiment, setting a furnace program, or entering a temperature value into software that expects Kelvin internally. In practice, the two directions are used symmetrically: K→°C for reading, °C→K for writing.
Can Kelvin be negative?
No. 0 K is absolute zero — the theoretical minimum. Classical thermodynamics prohibits negative Kelvin temperatures because temperature is proportional to average kinetic energy per degree of freedom, and kinetic energy cannot be negative. However, in certain quantum-statistical systems with a bounded energy spectrum and a population inversion — notably laser gain media and some nuclear spin systems — physicists have created negative absolute temperatures on the Kelvin scale. These systems are not "colder than absolute zero." Counterintuitively, they are hotter than any positive temperature — heat always flows from a negative-temperature system to any positive-temperature system. Negative Kelvin temperatures are a laboratory curiosity studied in quantum thermodynamics, not a phenomenon you'll encounter outside a physics department. In every engineering context, K ≥ 0.