← All articles
physicsthermometerstemperaturemeasurementSeptember 17, 20264 min read

How Does a Thermometer Work? Measuring Something You Cannot Touch

By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.

Temperature is not a substance and cannot be counted. It is measured indirectly, by finding something that changes reliably when it changes and reading that instead: a liquid expands, a metal's resistance rises, a junction between two metals produces a voltage, a surface emits more infrared. Every thermometer is a device for converting one of those into a number, and each is accurate over a different range.

What temperature is

Heat and temperature are frequently confused and are different quantities. Heat is energy in transit between bodies at different temperatures, measured in joules. Temperature is a measure of the average kinetic energy of the particles in a substance, and its practical meaning is the direction heat will flow: from higher to lower, always. A bathtub of warm water contains far more heat energy than a spark, and the spark has a much higher temperature. That definition explains why a thermometer works at all: placed in contact with something, it exchanges energy until both reach the same temperature, which is thermal equilibrium, and the thermometer then reports its own state. It also explains the main source of error in practice, since the thermometer must be small enough not to change what it is measuring, and must be given time to equilibrate, which is why a reading taken too early is always wrong in the direction the instrument came from.

The main types

Each design exploits a different temperature-dependent property, and the choice follows from range, speed and accuracy:

  • Liquid in glass, relying on thermal expansion, cheap and requiring no power; mercury has an almost linear expansion and a wide liquid range and is being phased out for toxicity, with dyed alcohol used instead at a cost of a lower upper limit
  • Bimetallic strip, two bonded metals with different expansion rates that bend with temperature, used in dial thermometers, ovens and older thermostats, robust and not very precise
  • Thermocouple, exploiting the voltage generated at a junction between two dissimilar metals, which covers an enormous range including very high temperatures, is cheap and rugged, and is less accurate than the alternatives
  • Resistance thermometer, usually platinum, whose electrical resistance rises predictably with temperature, giving high accuracy and stability, which makes it the workhorse of industrial and laboratory measurement
  • Thermistor, a semiconductor whose resistance changes sharply with temperature, very sensitive over a narrow range, which is what most digital clinical thermometers contain
  • Infrared thermometer, measuring the thermal radiation a surface emits, which requires no contact and reads only the surface, and whose accuracy depends on knowing how efficiently that surface radiates

The scales

Three scales are in general use and they are not interchangeable in the way they are usually treated. Fahrenheit, devised in 1724, set its zero at the temperature of a particular freezing brine and produced a scale whose ordinary numbers happen to span the range of human weather comfortably. Celsius, from 1742, was defined by the freezing and boiling points of water at standard pressure, originally with the scale running the other way. Kelvin is the scientific scale and differs in kind, because it is absolute: its zero is the point at which the thermal motion of particles is at the minimum permitted, so there are no negative values, and a doubling of the kelvin value means a genuine doubling of thermal energy in a way that a doubling of Celsius does not. A degree kelvin and a degree Celsius are the same size, which is why conversion is a simple offset. Since 2019 the kelvin has been defined by fixing the value of the Boltzmann constant, which ties temperature to fundamental constants rather than to the properties of water.

Why measuring body temperature is harder than it looks

Human body temperature is not a single number, and the everyday figure of 37 degrees Celsius comes from a study published in 1868 that measured axillary temperatures in a large German sample. Later work has consistently found a lower average, closer to 36.6, and analyses of long-run data suggest a genuine decline over the past century as well as differences in measurement. Temperature varies by site, with rectal, oral, axillary, tympanic and temporal artery readings differing systematically by a degree or more, so a measurement is meaningless without knowing where it was taken. It varies through the day by around half a degree on a circadian cycle, with a minimum in the early morning, and it varies with activity, menstrual cycle, age and ambient conditions. Infrared forehead scanners, which became ubiquitous during the pandemic, measure skin rather than core temperature and are strongly affected by sweat, cosmetics, ambient temperature and distance, which is why their sensitivity for detecting fever in screening settings has been reported as poor.

The takeaway

A thermometer measures temperature indirectly through something that varies with it: a liquid's expansion, a metal's resistance, a junction voltage or emitted infrared. It reports its own temperature after reaching equilibrium, which is why it must be small and given time. Temperature measures average particle kinetic energy and heat is energy in transit, which is why a spark is hotter than a bath containing far more heat. Kelvin is absolute and now defined by fixing the Boltzmann constant.

Practise this

Questions from Heat and Thermodynamics

Reading about something is not the same as being able to recall it. These are real questions from the Heat and Thermodynamics unit in our Physics track, answers and explanations included. The unit has 118 in total across 20 steps.

  • Guess the numberLevel 3

    1. A heat engine takes in 800 J of heat and does 200 J of useful work each cycle. What is its efficiency as a percentage?

    Answer: 25 percent

    Efficiency = useful work / energy input = 200 / 800 = 0.25 = 25 percent.

  • Multiple choiceLevel 2

    2. What does the temperature of an object actually measure?

    • The average kinetic energy of its particlescorrect
    • The total number of particles it contains
    • The total mass of the object
    • How much light the object gives off

    Temperature is a measure of the average kinetic energy of the particles in a substance.

  • Fill the blankLevel 2

    3. According to ____'s law, halving the volume of a gas at constant temperature doubles its pressure.

    • Boylecorrect
    • Charles
    • Newton
    • Ohm

    This inverse pressure-volume relationship is Boyle's law.