How a Thermometer Works: From Liquid to Digital

Every thermometer, regardless of era or technology, does the same fundamental job: it converts a physical property that changes with temperature into a number you can read. In a classic liquid-in-glass thermometer, that property is the expansion of a fluid. In a digital thermometer, it is the changing electrical resistance of a tiny semiconductor. In an infrared gun, it is the thermal radiation streaming off a surface. The methods differ enormously, but the underlying logic is the same, and understanding each type helps you know which to trust and when.

Liquid-in-Glass Thermometers and Thermal Expansion

The oldest thermometers you would recognize work on a principle you can see with your eyes: liquids expand when they get warmer and contract when they cool. Seal a liquid inside a narrow glass tube with a bulb at one end, heat the bulb, and the liquid climbs. Mark the tube with a scale, and you have a thermometer. The narrower the tube relative to the bulb, the farther a small temperature change pushes the column, making the reading easier to see.

Early versions in the 1600s were open to the air, which meant atmospheric pressure changes could shift the reading and make the whole device unreliable. That problem was solved in 1654 when Ferdinando II de’ Medici, Grand Duke of Tuscany, introduced sealed glass tubes partly filled with alcohol. Florentine glassblowers turned these into remarkably sensitive, uniform instruments, and the basic design persisted for centuries.¹

Mercury eventually became the liquid of choice for most professional and clinical thermometers. It stays liquid over a wide range, expands very evenly with temperature, and does not wet the glass (so it does not leave a film behind as it drops). Alcohol-filled thermometers, often dyed red or blue, remained popular for everyday use and for measuring temperatures below mercury’s freezing point.

Why Mercury Thermometers Disappeared

If you grew up shaking down a silver-tipped thermometer under your tongue, you might wonder where they all went. Mercury is toxic. A single broken thermometer releases a small amount of the metal, but in hospitals and schools those breakages add up, and the environmental cost of mercury disposal became untenable. Most countries began phasing out mercury thermometers for clinical use in the 2000s.

The main liquid replacement is galinstan, a silvery alloy of gallium, indium, and tin that looks and behaves much like mercury in a glass tube but is non-toxic. In a study comparing mercury, galinstan, and standard digital thermometers in children, researchers found no statistically significant difference in mean temperature between the mercury and galinstan readings. Digital thermometers, however, showed significant differences from both, and their sensitivity for detecting fever was only about 68 percent. When the child’s actual temperature was 39 °C or higher, the digital thermometer gave a false-negative reading roughly 65 percent of the time, compared with about 31 percent for the galinstan device.² That does not mean every digital thermometer is unreliable, but it shows that the technology inside a thermometer matters as much as the number on its screen.

How Digital Thermometers Sense Temperature

A digital thermometer does not have a column of liquid to watch. Instead, it relies on a small electronic component whose electrical resistance changes predictably with temperature. The most common version is a thermistor, a bead of semiconductor material typically just a few millimeters across. In a negative-temperature-coefficient (NTC) thermistor, resistance drops as temperature rises; the relationship is steep enough that even small temperature changes produce a measurable shift in resistance.³

A microprocessor inside the thermometer reads that resistance, converts it to a temperature using a calibration equation stored in its memory, and displays the result on a screen. The whole process can happen in seconds, which is why digital thermometers beep so quickly compared with the several minutes a glass thermometer needs. The trade-off is that cheap consumer-grade thermistors can drift slightly over time or vary between units, which partly explains the accuracy gap some studies have found between digital and liquid-column thermometers.

More expensive digital instruments use platinum resistance sensors, sometimes called RTDs, which are extremely stable and linear. These are the workhorses of industrial and laboratory measurement but are overkill for a bedside fever check.

Infrared Thermometers and Measuring Without Contact

Every object above absolute zero emits thermal radiation. Hotter objects emit more radiation and shift the peak of that radiation toward shorter wavelengths. An infrared thermometer captures a portion of that radiation with a lens, focuses it onto a sensor (usually a thermopile, which is a stack of tiny thermocouples), and uses the signal to calculate a temperature. The calculation draws on the physics of thermal radiation, specifically integrating the relationship between radiance and temperature described by Planck’s Law.⁴

The appeal is obvious: you never touch the thing you are measuring. That made infrared thermometers popular in food safety, industrial maintenance, and especially pandemic-era fever screening. But the technology has a built-in weakness. The calculation assumes a certain emissivity for the surface being measured, essentially how efficiently that surface emits radiation compared with an ideal “blackbody.” Human skin has high emissivity, around 0.98, so forehead thermometers work reasonably well. Shiny metal surfaces, on the other hand, have low emissivity and can fool an infrared gun into reading far too low. Research on radiation thermometry has shown that lower-wavelength measurement systems produce less temperature error from emissivity uncertainty, but errors can still be meaningful if the surface properties are unknown.⁵

Tympanic (ear) thermometers are also infrared devices. They measure the thermal radiation from the eardrum, which shares a blood supply with the brain’s temperature-regulation center, making it a useful proxy for core body temperature. Temporal artery thermometers scan across the forehead and use an algorithm to estimate core temperature from skin readings.

How Accurate Are Clinical Infrared Thermometers, Really?

The short answer: they are convenient, but they are not perfect. A study in an African outpatient clinic compared tympanic, temporal artery, and contactless infrared thermometers against a standard oral electronic thermometer. The tympanic and temporal artery models both tended to read slightly higher than the oral reference, by about 0.24 and 0.23 °C respectively, while the contactless forehead device read about 0.06 °C lower.⁶ These biases are small on paper, but they matter when a fever cutoff is a hard line for sending a child home from school or admitting a patient.

Similar findings come from other clinical settings. In febrile outpatients, average differences between ear and forehead thermometers and an oral reference were roughly 0.09 °C in either direction, with standard deviations wide enough that individual readings could be off by more.⁷ In intensive care patients, tympanic readings from both ears showed good agreement with a central nasopharyngeal measurement, reaching about 88 percent concordance, and forehead, oral, and axillary methods all showed acceptable precision.⁸

The practical takeaway: if you are screening large groups quickly, infrared thermometers are valuable for catching most fevers. If you need a precise reading to guide medical decisions, a well-placed oral or rectal thermometer remains the more reliable choice. Ear thermometers split the difference, offering decent accuracy when positioned correctly, but a bad seal or a waxy ear canal can throw the reading off.

Why Some Thermometers Are Faster Than Others

You have probably noticed that a glass thermometer takes a few minutes to settle, while an infrared gun gives a reading almost instantly. The difference comes down to thermal inertia and what the device actually needs to reach equilibrium with.

A liquid-in-glass thermometer has to absorb heat from its surroundings until the bulb reaches the same temperature as the thing being measured. That process follows Newton’s law of cooling, and the time it takes depends heavily on how well heat transfers between the medium and the bulb. In still air, a glass thermometer can take well over a minute; in stirred water, it settles much faster because the liquid delivers heat more efficiently.⁹ This is why oral thermometers ask you to keep your mouth closed: your tongue and saliva are the “stirred water” that speeds things up.

Digital thermistors are smaller and have less thermal mass than a glass bulb, so they equilibrate faster, but they still need physical contact. Infrared sensors skip equilibrium entirely. Because they detect radiation rather than conducting heat, they respond in fractions of a second. Optimized thermopile designs have achieved response times under a millisecond in laboratory settings, far faster than any clinical use requires.¹⁰

Gas Thermometers and the Search for Absolute Accuracy

All the thermometers discussed so far need to be calibrated against something. That “something” ultimately traces back to the fundamental definition of temperature, which is where gas thermometry enters the picture. A constant-volume gas thermometer works on the principle that the pressure of a fixed volume of gas rises in direct proportion to its absolute temperature. By carefully measuring pressure changes in a known quantity of gas, researchers can determine thermodynamic temperature with extraordinary precision.

These instruments are not something you would find in a kitchen drawer. They are large, delicate, and slow, designed for national metrology laboratories. One well-known implementation measured thermodynamic temperatures at 72 points between roughly 14 K and 287 K using helium-4, providing reference data that helps anchor the temperature scales everyone else relies on.¹¹

The International Temperature Scale of 1990 (ITS-90) is the practical temperature scale used worldwide. It defines fixed reference points, like the triple point of water and the freezing points of various metals, and specifies which types of thermometers should be used in which temperature ranges. ITS-90 exists because no single thermometer type works well across the full span from near absolute zero to thousands of degrees.¹² Everyday thermometers are calibrated, directly or indirectly, against these reference standards.

When Your Thermometer Lies to You

Knowing how different thermometers work helps you understand when they might give misleading readings. Here are the most common sources of error across thermometer types:

• Parallax in glass thermometers: If you read the mercury or alcohol column from an angle instead of straight on, you can easily misread by a degree or more. Always read at eye level.
• Poor contact with digital probes: A thermistor in an oral thermometer needs good contact with tissue. If the probe sits loosely under the tongue or the patient has just been drinking cold water, the reading will be artificially low.
• Surface conditions for infrared guns: Sweat, sunscreen, or a cold breeze across the forehead can all change how much radiation reaches the sensor. Earwax or a poorly aimed ear probe creates similar problems with tympanic thermometers.
• Ambient temperature extremes: Some consumer digital thermometers are rated to work between about 10 °C and 40 °C ambient. Using one outdoors on a freezing day can skew the electronics themselves.
• Low batteries: A dying battery in a digital thermometer can affect the voltage reference the microprocessor uses to interpret resistance, producing readings that drift.

None of these errors are dramatic in everyday life, but they can stack. A cheap digital thermometer with low batteries, used on a sweaty child who just ran inside, might read a full degree below the child’s actual core temperature. That is enough to miss a significant fever.

Thermocouples and Industrial Measurement

Outside the home, one of the most widely used temperature sensors is the thermocouple. It consists of two wires made from different metals, joined at one end. When that junction is heated, a small voltage appears across the other ends of the wires. The voltage depends on the temperature difference between the junction and a reference point. By measuring that voltage and knowing the characteristics of the metal pair, you get a temperature.

Thermocouples are popular in industry because they are cheap, rugged, and cover an enormous range, from cryogenic temperatures up to well over 1,000 °C depending on the metals used. They are what monitor the temperature inside a kiln, a jet engine exhaust, or a chemical reactor. Their accuracy is typically not as fine as a platinum resistance sensor, but they are hard to break and fast to respond, which matters more in many industrial environments.

Measuring Temperature Inside a Living Cell

At the far frontier of thermometry, researchers have figured out how to measure temperature at scales smaller than a single cell. One technique exploits tiny defects in the crystal structure of diamond, known as nitrogen-vacancy (NV) centers. These defects behave like atomic-scale sensors: their electronic spin properties shift with temperature, and those shifts can be read out using laser light and microwave pulses.

Using diamond nanocrystals injected into living cells, one research team directly measured the local thermal environment at length scales as short as 200 nanometers, essentially mapping temperature variations within different parts of a single cell.¹³ Another group demonstrated temperature precision down to about 5 millikelvins per root-hertz in bulk diamond sensors, and accuracies as fine as 1 millikelvin using nanodiamonds just tens of nanometers across.¹⁴

Why would anyone need to take a cell’s temperature? Because temperature is not uniform inside a cell. Mitochondria generate heat as they produce energy, and local temperature spikes may play a role in how cells signal, divide, or respond to drugs. Researchers have also developed all-optical approaches to NV thermometry, which simplify the hardware needed and could eventually make subcellular temperature mapping a routine tool in biomedical research.¹⁵

Picking the Right Thermometer for the Job

The range of available thermometers is wider than most people realize, and the best choice depends entirely on what you are measuring and how much accuracy you need. For checking a child’s fever at home, a well-made digital oral or rectal thermometer remains the most reliable option for its price. Ear thermometers are a solid second choice, especially for kids who will not sit still with a probe under their tongue, but you should expect occasional readings that are off by a few tenths of a degree. Forehead infrared thermometers are the fastest and least invasive, but their readings are the most sensitive to technique and environmental conditions.

For cooking, a thermocouple-based instant-read probe gives you speed and range that a glass thermometer cannot match. For monitoring a room or a refrigerator, a basic digital sensor with a thermistor is cheap and more than accurate enough. And if you need to know the temperature of a blast furnace or a cryogenic dewar, you are in the territory of specialized industrial sensors calibrated against national standards that ultimately trace their authority to gas thermometers and the fixed points of ITS-90.

The progression from a sealed tube of alcohol in 1654 to diamond nanocrystals measuring heat inside a living cell spans nearly four centuries of ingenuity. The core challenge never changed: temperature is invisible, so you have to find something visible, or at least measurable, that tracks it faithfully. Every thermometer you have ever used is just a clever answer to that same challenge.