This is the sequel to The History of the Barometer (1965) and prequel to the more ambitious Invention of the Meteorological Instruments (1969). Middleton has a very particular area of expertise and knows it very well. He seems to have examined every thermometer that was produced before the year 1800 which still exists and to have read almost every text that references them, in the original language.

Why was I interested in such a particular book? Middleton tells the reason in his Preface: unlike barometers, where almost all the progress since the 1600s has been technical, the history of thermometers is as much about what we think ‘temperature’ is as it is about the device itself. My interest is in how these philosophical questions about temperature were asked and answered.

---

Prerequisites: Less than none: Check your idea of temperature at the door.

Originally Written: September 2021.

Confidence Level: Not my ideas.

All these pictures except one come from Middleton’s book. I reference what figure they are in the book.

---

---

What is a Thermometer?

We experience ‘hot’ and ‘cold’ as raw sense data. There are particular nerve endings associated with each. Imprecise ideas of temperature predate humanity.

Thermometers provide a precise idea of temperature.

For something to be considered a thermometer, it must have two things:

• Some experimental apparatus that does something in response to being hot or cold.
• A temperature scale to distinguish different degrees of hot and cold.

These two things must be combined to form a measuring device.

The Ancients

The ancients had both parts of a thermometer separately, but did not put them together.

Philo of Byzantium and Hero of Alexandria both did experiments that showed the expansion and contraction of air when it is hot and cold, respectively. A closed container of air was connected by a pipe to a pitcher of water. When the air was hot, it expanded, and the surface of the water rose. When the air was cold, it contracted, and the surface of the water fell.

Galen introduced the first temperature scale, which ranged from the hottest substance, boiling water, to the coldest substance, ice. An equal mixture of the two was a neutral temperature. There were four degrees of heat and four degrees of cold (in opposite directions from neutral), which could be produced by mixing different quantities of boiling water and ice.

Independent Inventions

The thermometer was invented at least 3 separate times in between 1590 and 1630. It is hard to know the exact date or even who was first because some of the earliest references assume that the reader is already familiar with them. We can determine that thermometers were first invented either in the Netherlands or in northeastern Italy.

Like the experiments of Philo and Hero, the first thermometers used the expansion and contraction of air.

Italy

The invention of the thermometer in northeastern Italy was by either Santorio Santorre or Galileo Galilei in either Padua or Venice by 1612 at the latest.

In that year, we have letters between Sagredo and Galileo talking about thermometers. Sagredo initially credits Santorio, but Galileo corrects him and claims to have invented it himself. Subsequent Italian scientists credit Galileo as the inventor of the thermometer.

This same year, Santorio published a book called Commentaries on Galen which includes a description of a thermometer and its use in medicine. Santorio references Hero as his inspiration for this device.

The Italian story gets further complicated because of a book written in Rome in 1611 by Telioux. His description of how it works is complete nonsense and the picture he draws could not be constructed. Even though this is earlier, he obviously did not invent it.

Since Telioux wrote from Rome, thermometers had to have already spread across northern Italy by then. Middleton (the author) thinks that the thermometer was most likely invented in the 1590s or 1600s while Santorio and Galileo were at the University of Padua together. However, since Teloiux’s diagram is significantly different from the Paduan thermometers, it is possible that it is a crude representation of another independent invention from central Italy.

Netherlands

It is even more difficult to determine who invented the thermometer in the Netherlands. They are referenced as well known devices by Thomas Tymme in 1612, Salomon de Caus in 1615, Francis Bacon in 1620, and J. B. Van Helmont in 1624. By 1625, thermometers had spread across England and Germany.

Dutch scientists credit Cornelius Drebble with the invention of the thermometer. We do not have any direct evidence that he built a thermometer. We do have a patent of his from 1598 that uses the expansion and contraction of air to power a self-winding clock, so Drebbel understood the principles behind the air thermometer.

Drebbel also invented navigable submarines, the compound microscope, several other optic devices, a brighter and more durable red dye, several explosives, air conditioning, an automatic harpsichord, and a chicken incubator with a feedback mechanism to keep it at a constant temperature. He was certainly capable of inventing the thermometer.

The Dutch thermometers consisted of two bulbs, connected by a J-shaped tube. The lower bulb contained water and was open to the atmosphere, while the upper bulb was closed and contained air.

In the 1600s, there was little scientific communication across the Alps. The first comparison between Dutch and Italian thermometers was made by Jean Leurechon in 1626. He also coined the term ‘thermometer’.

While we’re in this part of the world, we should also mention Robert Fludd, a Welsh Rosicrucian mystic who practiced medicine in London. He designed a thermometer in 1626 based on Philo’s experiment. Thermometers certainly existed in England beforehand, but it’s not clear how familiar Fludd was with them. He did not use it as a scientific instrument – Fludd believed that all knowledge comes directly through revelation, arguing against Kepler and other natural philosophers.

France

The third independent invention occurred in Le Bugue in southwestern France. In 1632, Jean Rey had another different type of thermometer.

Rey had published a book in 1630, where he argued that metals’ weight increases when heated because the air condenses and adheres to the metal. This is correct, but not because air gets denser when heated, as Rey claimed. This triggered a scholarly exchange of letters. Air thermometers were used as a counterargument. Rey’s response was that he had never seen an air thermometer, but his own thermometer behaved differently.

Rey’s thermometer used the expansion and contraction of water. The water level went up as it was heated, not down like an air thermometer. This was the first liquid-in-glass thermometer, although it was not sealed on top to make it into a permanent device.

An Idea Whose Time Has Come

Over a 40 year period, the thermometer was invented independently 3-5 times: in the Netherlands (Drebble?), in Padua/Venice (Santorio/Galileo), in France (Rey), maybe near Rome (Teloiux’s source), and maybe in London (Fludd). This is clearly an idea whose time has come.

I’m not entirely sure why this idea occurred at this time. What is surprising is that it occurred at almost the same time in places that were not in close communication.

This is at the very beginning of the Scientific Revolution. It predates Nova Organum by Francis Bacon and (probably) Galileo’s astronomical investigations. Perhaps the question of why people began building thermometers is the same as the question of why empiricism began.

What do Thermometers Measure?

By 1630, thermometers had spread across most of western Europe. But even though the device existed, it was not always clear what it measured.

Weird Interpretations

Some of the interpretations of thermometers are extremely weird.

Thermometers reliably go up and down over the course of the day. Do you know what else goes up and down during the day? Tides.

Some early thermometers, especially those that contained water in some way, were thought to be tide simulators by those who made them. This should have been dismissed as nonsense, even then.

The connection between tides and the moon had been noted by some ancient writers, including Pliny the Elder, Seneca, and Claudius Ptolemy, but was dismissed by others, most notably Aristotle. Almost all medieval scholars, both Muslim and Christian, believed that the moon caused the tides. This was one of the ancient dogmas challenged by the scientific revolution. For example, Galileo tried to use tides as evidence for the motion of the Earth. The expansion and contraction of seawater as it warmed and cooled (like a liquid-in-glass thermometer) was another proposed mechanism. The correct theory of tides had to wait until Newton’s theory of gravitation in 1687.

There is an even more ridiculous interpretation of a thermometer. Otto Guericke, mayor of Magdeburg, Germany, made a 10 foot tall air thermometer for the town square in 1660. There was a float on the surface of the liquid, attached to pulley system, that made a small statue of an angel move up and down.

He thought of it as a perpetual motion machine.

It was more impressive to Guericke that the little angel kept moving than that the angel provided information about how hot or cold it is.

Air Pressure

In the 1640s, air pressure was discovered. This is a problem because air thermometers measured both the temperature of the air and the pressure on the surface of the water. Since air pressure fluctuates, the readings on the thermometer are not consistent. To deal with this, new thermometers had to be designed that were isolated from the outside air.

The key person in the development of sealed thermometers was the Grand Duke of Tuscany, Ferdinand Medici II, who was a much better scientist than a politician.

His first sealed thermometer was in 1641. It was a glass tube mostly filled with alcohol (‘spirit of wine’). Inside were small glass balls full of air. Different balls had different thickness of glass and so different average density. When the thermometer changed temperature, the alcohol would expand or contract. The temperature could be determined by counting the number of glass balls at the top and bottom.

In 1654, Ferdinand II presented another sealed thermometer. This is the one everyone is familiar with today. It contained a glass bulb at the bottom and a thin glass tube going up which was sealed at the top. A liquid (alcohol again) filled the bulb and part of the tube. The temperature is determined by how far up the tube the liquid extends.

The Grand Duke created an Academy of Experiments which, among other things, produced many of these Florentine thermometers. The thermometers were sent to other intellectuals around Italy to form the first meteorological network. We still have one of these records, from the Monastery of Angels in Florence. They recorded the temperature every hour of the day, from December 15, 1654 to March 31, 1670.

Sun and Shade

Thermometers show a hotter temperature in the sun than in the shade. This agrees with our experience – we feel hotter in the sun than in the shade too.

This is good if the goal is for the thermometer to measure what we experience, but it is a challenge for using thermometers for meteorological observations. How do we arrange our thermometer so that it measures the temperature of the air and nothing else?

In order to resolve this question, we need to understand heat transport. So this problem wasn’t solved until the 1800s. Beforehand, ad hoc methods were used, like putting the thermometer on the north side of a building or under a cover in the middle of a field.

In a liquid, the primary heat transport mechanisms are convection and conduction. Radiation is negligible because most liquids (including water and alcohol) are opaque to infrared light. A thermometer in a liquid quickly equilibrates with its surroundings.

In air, conduction is almost zero. Convection is still important. Air is transparent to infrared light, so radiation is also important. Convection makes the thermometer the same temperature as the air, but radiation makes the thermometer read some average of the temperatures of everything within line of sight. If the thermometer is in sunlight, then the surface of the sun is included in the average, which makes it way hotter. Reflected sunlight can also raise the temperature.

To measure air temperature, you need to either decrease radiation or increase convection.

The more common strategy was to build a shed to block radiation from the sun, while still allowing enough air flow to keep the entire shed from heating up. Many of these sheds were designed by British or Canadian meteorologists.

The French had a different strategy. They put the thermometer in a sling and spun it around really fast – so the bulb moves at over 20 mph (10 m/s). This does increase convection, but it is a bit dangerous for the thermometer.

Eventually, electric fans were invented and everyone started putting thermometers in sheds with fans. This both decreases radiation and increases convection. The thermometer then measures the temperature of the air instead of anything else.

The True Temperature

Thermometers suggest that temperature is an objective thing that can be represented by a single number.

Philosophical Issues

There are multiple things that impact our experience of hot and cold. A humid day with still air will feel different from a dry day with a breeze, even if the thermometers read the same thing. Today, these effects are included in weather reports as the ‘feels like’ temperature.

Different people experience temperature differently. Some people consistently feel colder than others, even when all of the external conditions are the same. The experiences of hot and cold also depend on your past, as George Berkeley pointed out in 1713. A bowl of lukewarm water feels warmer if your hand has recently been in an icebox than if your hand has recently been near an oven.

Before the invention of thermometers, people thought that the temperature of well water and the inside of deep caves is warmer in winter than in summer. Thermometers showed that the temperature of well water and deep caves remain close to the same throughout the year. Is this something new that we’ve learned about our world or is this an indication that thermometers don’t really capture ‘hot’ and ‘cold’?

A thermometer extrapolates an objective reality from highly varied personal experience.

The Relationship between Temperature and Volume

From a more technical perspective, different thermometers do not measure temperature in the same way. Whatever the true temperature $T$ might be, we actually measure the volume of some substance. We believe that there is a well defined relationship between this volume and the true temperature.

What kind of function is the relationship?

Ideally, the relationship between temperature and volume would be linear: $V = \alpha T$.

This relationship between temperature and volume certainly does not have to be linear.

For water, it isn’t even monotonic. Water reaches a minimum volume a few degrees above freezing and then expands again. The relationship is always highly nonlinear near phase transitions, so thermometers are not reliable when the liquid inside is close to freezing or boiling.

Alcohol has a highly nonlinear relationship. This was problematic because alcohol was initially the most commonly used liquid in thermometers. Marks were placed evenly on the thermometers, which did not correspond to uniform degrees of temperature.

Mercury has a more linear relationship between temperature and volume. A thermometer with mercury will not agree with a thermometer with alcohol, even if both have the same two fixed points and the same number of degrees before these points. Which one accurately represented the true temperature could not be determined at the time.

One proposal for resolving this issue dates all the way back to Galen, and was re-proposed by Renaldini in 1694. Mixing equal amounts of boiling water and ice water should result in water with temperature halfway between the two. The markings on the thermometer should be determined by mixing different amounts of ice water and boiling water together. They should not just be evenly spaced – both because the expansion might be nonlinear and because the glass tube might not be perfectly uniform.

The question of the ‘true temperature’ was not fully answered until almost 1850. A notion of temperature was needed that did not depend on the expansion and contraction of different materials. Carnot showed that the efficiency of an ideal heat engine depends only on the hot and cold temperature the engine used. William Thomson (later Lord Kelvin) proposed that thermodynamic temperature be used as the true temperature. Thermometers made from hydrogen gas were shown to be the most linear.^[1]By this point, people had invented good air thermometers isolated from atmospheric pressure. Other thermometers could mark degrees that represented the true temperature, even though they weren’t spaced uniformly along the thermometer’s tube.

The thermodynamic temperature was later related to the average energy of the atoms in the gas or liquid, our modern understanding of temperature.^[2]Actually, our modern understanding of temperature is more subtle than this. The definition of temperature is the derivative of the internal energy with respect to the entropy. This is related to the … Continue reading

But let’s go back to where we were chronologically.

Standardized Temperature Scales

By the late 1600s, there were many thermometers at many locations across western Europe. People wanted to compare the measurements made by these different thermometers. The easiest way to do this is with a standardized temperature scale.

There are two strategies you can use to create a standardized temperature scale. The first is that everyone uses thermometers made by the same person. The second is to have a clear published procedure that everyone can follow when making their thermometers.

The first strategy is easier. The act of producing a thermometer is imperfectly captured by any written procedure. The most consistent set of thermometers are often all made by the same person or workshop. We’ve already seen one example of this with the Florentine thermometers. The biggest advantage of the second strategy is that it scales.^[3]Pun intended. A single individual or workshop can only make so many thermometers. To make more thermometers than this, other people need to be able to produce them too.

The next phase in the history of thermometers is characterized by competitions over which temperature scale would become standard. During this process, thermometers were often made with many scales.

Questions

There are many questions that need to be answered when creating a temperature scale.

The first question is: Which way is up? Are you measuring degrees of heat, so hotter temperatures correspond to larger numbers, or are you measuring degrees of cold, so colder temperatures correspond to larger numbers?

The second question is: What expands or contracts to indicate temperature? Air thermometers had fallen out of fashion and thermometers that use the expansion of metals had not been invented yet. So it was going to be some kind of fluid inside the thermometer.

But which one?

It needs to remain liquid for all relevant temperatures. So water is eliminated – especially because it expands as it freezes. Initially, alcohol was the most popular choice. It expands and contracts a lot as the temperature changes. Mercury only became popular later. It doesn’t expand or contract as much, but it could be used to make even more precise thermometers.

The third question is how many fixed points to use.

To determine a temperature scale, you need to pick where the zero is and how big each degree is. There are two ways of doing this.

The first way was first developed by Robert Hooke in 1665. Hooke decided to use alcohol (‘spirit of wine’) as his liquid and have larger numbers be hotter. He chose a single fixed point: water that is just starting to freeze. Each degree is defined as the temperature change that causes the alcohol to expand by 1/1000th of its initial volume.

The other way uses two fixed points and a choice of the number of degrees between these two points.

Sebastiano Bartolo was the first person to use two fixed points, in 1672, using melting snow and boiling water with $18^\circ$ in between. He was ignored because his description was confusing: he described an air thermometer while the figure was of a liquid-in-glass thermometer. In 1680, Francesco Eschinardi did something similar with a liquid-in-glass thermometer. His work was read to the Royal Society of London, but Hooke chose to ignore it. This idea is usually ascribed to Carlo Renaldini in 1694. He had been at the original Academy of Experiments in Florence, and had a very clear explanation.

Boiling water and freezing water / melting ice were common choices for the fixed points, but people also used the freezing point of an ice/salt mixture, melting butter, and the cellars of the Observatory of Paris.

Fahrenheit Scale

In 1702, in Copenhagen, Denmark, Ole Rømer produced a set of 5 high quality thermometers. They were still well calibrated 40 years later. For his temperature scale, he used $60^\circ$ for boiling water and $0^\circ$ for the coldest temperature it gets in Denmark in the winter. This puts freezing water at $7 \tfrac{1}{2}^\circ$. He used these thermometers to record the daily temperature in Copenhagen.

Daniel Gabriel Fahrenheit was born in Danzig, but spent most of his life in the Netherlands. Fahrenheit visited Rømer in 1708 and decided that he was going to make thermometers.

Fahrenheit changed Rømer’s scale in two ways: he adjusted it so that freezing is at $8^\circ$ instead of $7 \tfrac{1}{2}^\circ$ and he multiplied all of the numbers by four. Fahrenheit spent most of the rest of his career making thermometers.

The mercury thermometers made by Fahrenheit were incredibly accurate. They agreed with each other to 1/4th of a degree. They were clearly the best thermometers available at the time and were sold widely. A few of these thermometers still survive.

Fahrenheit was a craftsman, not a natural philosopher. He was interested in selling thermometers, not in publishing how anyone could make thermometers.

In 1724, Fahrenheit was elected to the Royal Society of London. That year, he published 5 short papers, his only published work. One of these was a short and vague description about how he made thermometers.

Fahrenheit listed four fixed points for his thermometers. The zero point was determined by a mixture of water, ice, and either sea-salt or sal-ammoniac. This works better in winter. The temperature when water just starts to freeze is $32^\circ$. Put the thermometer in someone’s mouth to fix $96^\circ$. The boiling temperature of mercury is set at $600^\circ$.

Of these four points, only one of is actually reasonable for a fixed point: freezing water. Fahrenheit’s procedure could not have produced thermometers which agreed to within 1/4th of a degree. We should be skeptical that this publication actually describes how he built thermometers.

Fahrenheit definitely did not use the boiling point of water as a fixed point. He discovered that the temperature which water boils changes with pressure. He even invented a barometer which determined the air pressure by measuring the temperature at which water boils. He did mention that, at a particular pressure, water boils at $212^\circ$. People copying Fahrenheit’s thermometers adopted this as a fixed point.

Fahrenheit’s workshop was inherited by Hendrik Prins, who continued to produce quality thermometers.

Réaumur Scale

While Fahrenheit’s thermometers were becoming the standard in the Netherlands and England, a type of thermometer due to Réaumur in 1730 became dominant in France and much of the continent.

Réaumur was highly respected by scientists across the continent, but Middleton (the author) thinks he’s overrated. Réaumur thought he was being original, even though his procedure is very similar to what Hooke did 70 years earlier.

Réaumur used alcohol and wrote at great length about how to prepare his ‘spirit of wine’. This thermometer had a single fixed point: freezing water. Each degree is when the volume of the alcohol increases by 1/1000th of its original.

Réaumur’s thermometers typically ranged from $1000^\circ$ (freezing) to $1080^\circ$, although the upper point was not any special temperature. He also mentioned that the temperature of the cellars of the Observatory of Paris was $1010 \tfrac{1}{4}^\circ$.

These thermometers were precise, but this was mostly because Réaumur made them multiple feet tall.

As Réaumur’s books spread around the continent, people misunderstood them to create a different “Réaumur” temperature scale. This is a two point temperature scale, with $0^\circ$ for freezing water and $80^\circ$ for boiling water. These thermometers used mercury instead of alcohol.

Celsius Scale

Both Réaumur and Peder Horrebow suggested making thermometers with $100^\circ$ between freezing and boiling water, but neither made any of these thermometers.

Joseph Nicolas Delisle was a French astronomer at the court of Peter the Great in St. Petersburg who started making thermometers in 1732. He uses mercury for his fluid and a single fixed point: boiling water. This was $0^\circ$. He chose the spacing of the degrees so freezing would occur at $150^\circ$. Delisle made many copies of these thermometers and sent them to scientists across Europe, encouraging them to compare them with other thermometers.

The freezing point of mercury was discovered in St. Petersburg using one of Delisle’s thermometers. Since mercury contracts while freezing, these thermometers would give some fantastically cold numbers. Why this discovery took place in Russia is left as an exercise to the reader.

One of Delisle’s thermometers was sent to the Swedish physicist Anders Celsius. Celsius was impressed by the thermometer and added another scale across from Delisle’s scale. His scale had $100^\circ$ instead of $150^\circ$, but was still oriented in the same direction. Celsius soon began making these (inverted) centigrad thermometers.

Shortly after Celsius’s death in 1744, Swedish scientists reversed his scale to make “Celsius” thermometers with $0^\circ$ at freezing and $100^\circ$ at boiling. It’s not clear who did this first, but the most likely candidates are Daniel Ekström or Märten Strömer. Another candidate is Carl Linnaeus, the great botanist and creator of the Linnaean classification system. Like everything Linnaeus did, there was a connection to plants: Linnaeus wanted to accurately measure the temperature of his greenhouses.

These “Celcius” thermometers became the norm in Scandinavia and gained some followers on the Continent.

In 1799, the French Revolutionary Government decided that the Réaumur Scale was terrible (fair) and destroyed the last of Réaumur’s original thermometers in the Observatory of Paris (not fair). By this point, the only other scales that were still in widespread use were the Fahrenheit scale, in England & the Netherlands, and the Celsius scale, mostly in Scandinavia. Copying the English was Not An Option, so they adopted the “Celsius” scale as part of the metric system. Napoleon’s conquests spread the metric system across continental Europe.

Summary

I have made a table of all of the temperature scales I’ve described, along with a few more from the book:

Scale	Strategy	Up Is	Liquid	# Fixed Points	Zero Point	$|T_B – T_F|$	Date	Country
Hooke	Procedure	Hot	Alcohol	One	Freezing Water	???	1665	England
Newton	Procedure	Hot	Linseed Oil	Two	Melting Snow	$34^\circ$	1701	England
Hauksbee	One Producer	Cold	Alcohol	???	Blood?	$\sim 175^\circ$	1723	England
Rømer	One Producer	Hot	Mercury	Two	Denmark Winter	$52\tfrac{1}{2}^\circ$	1702	Denmark
Fahrenheit	One Producer	Hot	Mercury	Four?	Salt Mixture?	$180^\circ$	1708+	Netherlands
Réaumur	Procedure	Hot	Alcohol	One	Freezing Water	$\sim 100^\circ$	1730	France
“Réaumur”	Procedure	Hot	Mercury	Two	Freezing Water	$80^\circ$	1730+	Continent
Delisle	One Producer	Cold	Mercury	One	Boiling Water	$150^\circ$	1732	Russia
Celsius	Procedure	Cold	Mercury	Two	Boiling Water	$100^\circ$	1741	Sweden
“Celsius”	Procedure	Hot	Mercury	Two	Melting Snow	$100^\circ$	1744+	Sweden
Christin	Procedure	Hot	Mercury	One	Pounded Ice	$100^\circ$	1743	France

The Strategy column is whether a single producer made all of these thermometers or if they wrote a procedure for other people to copy. If there is a single producer, we often don’t know all of the details of how they were made.

The $|T_B – T_F|$ column is the number of degrees between boiling and freezing, which I include even if those thermometers did not extend all the way to boiling. In these cases, the number is approximate.

This is still only a small fraction of the temperature scales in use in the 1700s. The French chemist Antoine Lavoisier collected thermometers and had thermometers with 38 different scales.

Other Kinds of Thermometers

There is still about half the book left and a great many more interesting instruments. Since I expect that you are less interested in thermometers than Middleton is, I will only mention a few.

There are thermometers meant to measure different things about the temperature and there are thermometers which use different materials to measure temperature.

Other Temperatures to Measure

Most thermometers show the temperature now. There are some thermometers which are designed to show the maximum or minimum temperature or both since the thermometers were set.

The first of these was designed by Jean Bernoulli^[4]Not to be confused with the many other mathematicians in the Bernoulli family. in 1698. This was an air thermometer, with extensions off of the tube to trap some of the fluid at the maximum or minimum temperatures.

This was just the first of many ingenious designs.

There are also thermometers which are designed to record the temperature continually. The first of these was also designed in the era of air thermometer: by Christopher Wren in 1663. Practical recording thermometers were not produced until the 1800s.

One of the first uses of the camera was to record meteorological observations. This was done in 1839, only a few months after the camera was invented.

Metal Thermometers

Solids also expand and contract as the temperature changes. Metals were the most commonly used solid. These thermometers are more durable than the common liquid-in-glass thermometers.

If the entire thermometer is made of the same material, then the scale will also expand. The thermometer has to be made of (at least) two materials that expand differently.

Two different types of metal are layered on top of each other. They are then cut into a thin strip and rolled into a coil. As the temperature changes, the layers expand or contract relative to each other, causing the coil to wrap tighter or more loosely. A needle attached to the end of the coil shows the temperature.

Electrical Thermometers

The first electrical thermometers were designed by T. J. Seebek in 1822. He showed that when two different metals at different temperatures are in contact, a current will flow. Which direction the current flows is determined by the thermoelectromotive series. These thermometers respond quickly, so they can be used to measure very rapid fluctuations. They are also better at measuring small differences in temperature. These thermometers have found a few specialized niches, like measuring how the temperature varies going into the ground, but did not spread widely.

In 1821, Humphry Davy discovered that the resistance of metals to electrical currents changes with temperature. Carl Wilhelm Siemens proposed that this could be used for a thermometer in 1871.

Detailed comparisons between resistance thermometers and air thermometers were made by Hugh Longbourne Callendar in 1887. The resistance is well described by a quadratic function of the temperature all the way up to $600^\circ$C. This allowed them to be easily calibrated for a wide range of temperatures.

There is a particularly clever way to turn a mercury liquid-in-glass thermometer into an electrical thermometer due to Berthold in 1889. The mercury conducts electricity well and so can be used as part of a circuit. Berthold ran a thin platinum wire through the middle of thermometer. As the level of the mercury rises and falls, how much of the wire that is in the mercury changes. The resistance of the circuit is proportional to the length of wire that is not in the mercury.

As electronics have improved, it has become much easier to measure the resistance of circuits with excellent precision. They can now be made more precise, more durable, and smaller than even the best liquid-in-glass thermometers. Most thermometers in the world are now derived from the work of Siemens and Callendar.

Conclusion

We use thermometers without a second thought. The thermometer tells us a number and that is the temperature.

But what is temperature?

We know from ample personal experience that temperature is only loosely related to our experiences of ‘hot’ and ‘cold’. These experiences are highly subjective, even compared to other sense data. People are more likely to disagree about the temperature than about what color something is, for example. Even a single person will experience the same temperature differently in different circumstances.

Temperature is also distantly related to the underlying physical phenomenon. We know that temperature is a measurement of the average energy of the atoms in a material, but we don’t often think about it that way. Our thermometers do not measure the average atomic energy directly. Instead, they measure the volume of some liquid or the electrical resistance of some circuit. The connection between the underlying physical phenomenon and what we measure is complicated.

Temperature exists about halfway between raw sense data and the underlying physical phenomenon. This is what makes its history interesting. We began building thermometers long before we understood the connection in either direction.

The history of the thermometer is the history of trying to design and improve a measuring device without knowing exactly what it is we are trying to measure.