How Temperature Scales Were Invented: From Galileo's Thermoscope to Modern Standards

Galileo's Thermoscope (1593): No Scale, Just Direction

The ancestor of the thermometer is the thermoscope, and its invention is conventionally attributed to Galileo Galilei around 1593. Galileo's device was a glass tube with a bulb at one end, partially filled with water or wine and inverted in a vessel of liquid. As temperature rose, the air in the bulb expanded and pushed the liquid down the tube; as temperature fell, the liquid rose. The device clearly indicated temperature change but had two fundamental problems: it responded to atmospheric pressure as well as temperature, and it had no scale — you could say it was hotter or colder, but not by how much.

Galileo's thermoscope was open to the atmosphere. The Dutch inventor Cornelius Drebbel made closed thermoscopes around 1620, and the Florentine Accademia del Cimento (founded 1657) produced sealed alcohol thermometers in the 1650s. Ferdinand II de' Medici, Grand Duke of Tuscany and patron of the Accademia, sponsored the manufacture of dozens of sealed alcohol thermometers, but each instrument maker calibrated them differently. Without reproducible reference points, there was no way to compare readings between instruments.

The 17th century produced many proposed scale systems, none of which gained broad adoption. The fundamental problem was the same in every case: reference points varied between inventors, fluids varied (alcohol, linseed oil, water), and calibration procedures were not communicated in enough detail to be reproduced by others.

Early Scales: Rømer, Réaumur, Delisle

Three early scales predate or compete with Fahrenheit and deserve mention because they reveal how arbitrary the choice of reference points was.

Ole Rømer (1701): The Danish astronomer Ole Rømer, famous for the first measurement of the speed of light in 1676, developed a scale around 1701 using two fixed points: 7.5° for the freezing point of water (he called it "the beginning of winter") and 60° for the boiling point of water. Body temperature was approximately 22.5° on this scale. Rømer's scale influenced Fahrenheit directly — Fahrenheit visited Rømer in 1708 and used Rømer's two-fixed-point approach, though he changed both reference points and the degree size.

René Antoine Ferchault de Réaumur (1730): The French natural philosopher Réaumur proposed a scale in 1730 with 0° at the freezing point of water and 80° at the boiling point. The 80-degree range was chosen because Réaumur used an alcohol thermometer and the alcohol he measured expanded by 80/1000 of its volume over this range — the degree size was defined by the expansion of a specific alcohol sample, not by an abstract division of the range. The Réaumur scale was used extensively in France, Germany, and Russia well into the 19th century, and Russian recipes from the 18th and 19th centuries use Réaumur degrees. To convert: °C = °Ré × 1.25 and °Ré = °C × 0.8.

Joseph-Nicolas Delisle (1732): The French astronomer Delisle, working in Russia at the invitation of Peter the Great, proposed a scale in 1732 that ran backward: 0° was the boiling point of water, and the scale increased as temperature decreased. Freezing water was 150°. The scale was used in Russia for several decades before being replaced by Celsius. To convert to Celsius: °C = 100 − (°De × 2/3).

These competing scales make clear that there was nothing inevitable about any particular choice of fixed points or degree size. Every scale was a human convention.

Fahrenheit's Mercury Thermometer (1724)

Daniel Gabriel Fahrenheit solved two distinct problems simultaneously in his 1724 paper to the Royal Society: he replaced alcohol with mercury, and he established a reproducible three-point calibration system.

Mercury offered significant advantages over alcohol: it expands more uniformly across a wide temperature range, it does not evaporate through glass (as alcohol can), its higher boiling point (356.7°C / 674°F) allowed measurement of temperatures far beyond alcohol thermometers, and its opacity makes it easy to read. Mercury thermometers remained the standard for precision laboratory and medical measurement until mercury began to be phased out for safety reasons in the 1990s and 2000s.

Fahrenheit's three reference points were:

• 0°F: The temperature of a brine solution (ice + water + ammonium chloride, NH₄Cl). Fahrenheit considered this the coldest reliably reproducible temperature achievable in a laboratory.
• 32°F: The freezing point of pure water.
• 96°F: Human body temperature (measured in the mouth or armpit). Fahrenheit described this as "the temperature of the blood of a healthy man."

The 96° body temperature reference was chosen because 96 is evenly divisible by 2, 3, 4, 6, 8, 12, 16, 24, 32, 48 — convenient for physically subdividing the scale on a rule. Once the scale was later refined with more accurate calibration equipment, the body reference shifted to 98.6°F (37°C), and the boiling point of water at 1 atm fell at 211.97°F — rounded to 212°F in the standard definition.

The modern definition of the Fahrenheit scale uses two fixed points: the ice point at 32°F and the steam point at 212°F (at 1 standard atmosphere). The degree size is defined so that there are exactly 180°F between these two points. Conversion to Celsius: °C = (°F − 32) × 5/9. Conversion from Celsius: °F = (°C × 9/5) + 32.

Celsius and the Centigrade Scale (1742)

Anders Celsius (1701–1744) was a Swedish astronomer at Uppsala University, best known scientifically for his 1736 participation in a French expedition to Lapland to measure the shape of the Earth. His temperature scale was proposed in a 1742 paper titled "Observations of two persistent degrees on a thermometer."

Celsius proposed a 100-degree scale between two fixed points: 0° for the boiling point of water and 100° for the freezing point of water — the opposite of today's convention. The inverted orientation was standard in his paper. The reversed scale (0° = freezing, 100° = boiling) was introduced by either Carl Linnaeus or Martin Strömer, both at Uppsala, around 1745, and quickly became the standard orientation.

The scale was called "centigrade" (from the Latin for "one hundred steps") until 1948, when the 9th General Conference on Weights and Measures renamed it "Celsius" to honor its inventor and to avoid confusion with the angular unit "grade" (1/400 of a circle, also called a "gradian"). The renaming also standardized that the "degree" symbol is retained: °C, not C°.

The scientific advantage of the Celsius scale is the logical anchor of its fixed points. Water's freezing point at 0°C and boiling point at 100°C (at 1 standard atmosphere, 101.325 kPa) are phenomena reproducible anywhere with pure water. The 1968 International Temperature Scale (ITS-68) and its successor the 1990 International Temperature Scale (ITS-90) define temperature in terms of fixed points of pure substances across a wide range, with water's triple point (273.16 K, 0.01°C) as one of the defining points.

Lord Kelvin's Absolute Zero (1848)

William Thomson, later Baron Kelvin of Largs (1824–1907), proposed the absolute temperature scale in an 1848 paper "On an Absolute Thermometric Scale." Thomson observed that the coefficient of thermal expansion of an ideal gas implies a lower limit on temperature: if pressure decreases linearly with temperature at constant volume, extrapolating to zero pressure gives a minimum temperature. From Carnot's theorem on the maximum efficiency of heat engines, Thomson recognized that this lower limit — absolute zero — was a fundamental thermodynamic boundary, not merely an artifact of gases.

The Kelvin scale uses the same degree size as the Celsius scale (1 K = 1°C increment), but its zero point is absolute zero: the temperature at which a system has minimum possible thermal energy. Absolute zero is −273.15°C (−459.67°F). The Kelvin scale has no negative temperatures; it starts at 0 K and increases without upper limit.

Key conversions: - K = °C + 273.15 - °C = K − 273.15 - K = (°F + 459.67) × 5/9 - °F = K × 9/5 − 459.67

The Kelvin scale is essential for thermodynamic calculations because gas laws (PV = nRT), Planck's radiation law, the Boltzmann distribution, and virtually all equations of statistical mechanics require temperature expressed as an absolute value. A gas at 300 K has exactly twice the thermal energy per molecule as a gas at 150 K; a gas at 27°C does not have twice the thermal energy of a gas at 13.5°C (the ratio 300 K / 150 K = 2, but 27°C/13.5°C = 2 only by coincidence of scale).

Since the 2019 SI redefinition, the kelvin is defined by fixing the Boltzmann constant k at exactly 1.380 649 × 10⁻²³ J/K.

Rankine: Fahrenheit's Absolute Scale (1859)

William John Macquorn Rankine (1820–1872), a Scottish engineer and physicist, proposed an absolute temperature scale based on Fahrenheit degrees in 1859. The Rankine scale starts at absolute zero (0°R = 0 K = −459.67°F) and uses Fahrenheit-sized degree increments.

Key conversions: - °R = °F + 459.67 - °F = °R − 459.67 - °R = K × 9/5 - K = °R × 5/9

Water freezes at 491.67°R and boils at 671.67°R. The Rankine scale is used in some branches of American engineering, particularly in thermodynamics, steam tables, and aerospace engineering when calculations require an absolute scale but historical data is in Fahrenheit. Its use has declined substantially; Kelvin is the SI standard and has largely displaced Rankine even in American engineering practice.

Absolute zero in both absolute scales: 0 K = 0°R = −273.15°C = −459.67°F.

Modern Temperature Measurement (RTDs, Thermocouples, IR)

Historical mercury and alcohol thermometers have been largely replaced by three classes of electrical temperature sensors:

Resistance Temperature Detectors (RTDs): Electrical resistance in metals increases with temperature in a predictable, highly repeatable manner. Platinum RTDs (Pt100, Pt1000) are the most common standard; a Pt100 sensor has 100 ohms resistance at 0°C and approximately 138.5 ohms at 100°C. The Callendar-Van Dusen equation characterizes the relationship: R(T) = R₀(1 + AT + BT² + C(T−100)T³) for T below 0°C. RTDs are accurate to ±0.1°C in industrial applications and to ±0.01°C in laboratory standard platinum resistance thermometers (SPRTs).

Thermocouples: Two dissimilar metals joined at one end generate a small thermoelectric voltage (the Seebeck effect) that is a function of the temperature difference between the junction and the reference end. Type K thermocouples (chromel-alumel) cover −200°C to +1350°C; Type S (platinum-rhodium/platinum) extends to 1768°C and is used to define the ITS-90 scale between 961.78°C and 1084.62°C. Thermocouples are rugged, inexpensive, and cover a wide range but require cold junction compensation and are less accurate than RTDs (typically ±0.5–2°C for industrial types).

Infrared (non-contact) thermometers: All objects above absolute zero emit thermal radiation. By measuring the intensity and wavelength distribution of this radiation (governed by Planck's law and the Stefan-Boltzmann law: P = εσT⁴, where σ = 5.670 374 419 × 10⁻⁸ W·m⁻²·K⁻⁴), temperature can be inferred without contact. Medical infrared thermometers (tympanic or forehead) use this principle. Industrial pyrometers measure temperatures from −50°C to over 3000°C in furnaces, molten metal, and aerospace testing. The primary source of error is uncertainty in emissivity (ε), the fraction of blackbody emission an object actually emits (ε = 1 for a perfect blackbody; real surfaces range from ~0.02 for polished metal to ~0.95 for blackened surfaces).

All five historical scales remain in current use in different contexts: Celsius for everyday and scientific measurement worldwide, Fahrenheit for consumer use in the US, Kelvin for scientific and SI-standard thermodynamics, Rankine for some US engineering applications, and Réaumur in a few traditional European culinary contexts. For conversions between all five, the temperature converter handles every combination with exact conversion formulas.