Black plate (219,1)

Science Progress (2008), 91(3), 219–240 doi: 10.3184/003685008X360632

The myth of the boiling point

HASOK CHANG

ABSTRACT

Around 1800, many reputable scientists reported significant variations in the

temperature of pure water boiling under normal atmospheric pressure. The
reported variations included a difference of over 1 C between boiling in
metallic and glass vessels (Gay-Lussac), and ‘‘superheating’’ up to 112 C on
extracting dissolved air out of water (De Luc). I have confirmed most of these
observations in my own experiments, many of which are described in this paper.
Water boils at the ‘‘boiling point’’ only under very particular circumstances.
Our common-sense intuition about the fixedness of the boiling point is only
sustained by our limited experience.

Keywords: boiling point, replications of historical experiments, super-

heating, water, De Luc, Gay-Lussac

Hasok Chang is Reader in Philosophy of Science at the Department of Science

and Technology Studies, University College London (Gower Street, London
WC1E 6BT, UK; E-mail [email protected]). He received his PhD at Stanford
University in 1993, after a bachelor’s degree at the California Institute of
Technology.
www.scilet.com 219
 Black plate (220,1)

Introduction
We all learn at school that pure water always boils at 100 C
(212 F), under normal atmospheric pressure. Like surprisingly
many things that ‘‘everybody knows’’, this is a myth. We ought
to stop perpetuating this myth in schools and universities and in
everyday life: not only is it incorrect, but it also conveys misleading
ideas about the nature of scientific knowledge. And unlike some
other myths, it does not serve sufficiently useful functions.
There are actually all sorts of variations in the boiling temperature
of water. For example, there are differences of several degrees
depending on the material of the container in which the boiling
takes place. Removing dissolved air from water can easily raise its
boiling temperature by about 10 degrees centigrade.
The fickleness of the boiling point is something that was once
widely known among scientists. It is quite easy to verify, as I have
learned in the simple experiments that I discuss below. And it is
still known by many of today’s experts. So actually the strange
thing is: why don’t we all hear about it? Not only that, but why do
most people believe the opposite of what is the case, and maintain it
with such confidence? How has a clear falsehood become scientific
common sense?

History
History of science provides the most convenient entry to the
constellation of issues discussed in this paper. The old thermometer
shown in Figure 1 is emblematic. This instrument, dating from the
1750s, is preserved at the Science Museum in London; the glass
stems have broken off, so all we have is the frame, which shows
four different scales on it. The third one is the familiar Fahrenheit
scale. (The second one, due to Delisle, is ‘‘upside down’’, with 0
at the boiling point and increasing numbers as it gets colder1.)
There are two boiling points marked on this thermometer. At the
familiar 212 F it says ‘‘water boyles vehemently’’. Down at about
204 F it says ‘‘begins to boyle’’. What is going on here? You may
think that the artisan who made this thermometer must have been
pretty incompetent on scientific matters. But it turns out that this
thermometer was the work of George Adams, official scientific
instrument-maker to King George III. And the idea of two boiling
points actually came straight from Isaac Newton, whose temperature
scale published in 1701 was indeed the first of Adams’s four
scales2.

220 Hasok Chang

 Black plate (221,1)

Fig. 1. George Adams’s thermometric scale, showing two boiling points (inventory
no. 1927 – 1745). Science MuseumyScience & Society Picture Library.

Stimulated by such oddities from the 18th century, I looked more

deeply into the history, to see what people really knew and thought
about the boiling point in those early days. In fact there was so
much uncertainty about it that in 1776 the Royal Society of London
appointed a special committee charged with making definite

www.scilet.com The myth of the boiling point 221

 Black plate (222,1)

recommendations about the ‘‘fixed points’’ of thermometers. The

Royal Society Committee recorded various types of variations in the
boiling temperature of water3. Henry Cavendish, who chaired the
committee, left us a rather enigmatic statement in one of his
unpublished manuscripts: ‘‘The excess of the heat of water above
the boiling point is influenced by a great variety of circum-
stances’’ 4.

Fig. 2. Henry Cavendish (1731 – 1810).

222 Hasok Chang

 Black plate (223,1)

Another key member of the committee was Jean-André De Luc,

Genevan geologist, physicist, meteorologist, theologian and busi-
nessman. By this time he was living in England, installed in
Windsor as ‘‘Reader’’ to Queen Charlotte. Around 1770 De Luc
had made extensive investigations into boiling5. He reasoned that in
an ordinary boiling situation the layer of water touching the heated
surface, where the vapour bubbles form, must be much hotter than
the rest of the water. He wanted to find out the temperature of that

Fig. 3. Jean-André De Luc (1727 – 1817). Geneva, Bibliothèque publique et

universitaire, Collections iconographiques.

www.scilet.com The myth of the boiling point 223

 Black plate (224,1)

‘‘first layer’’, which would be the temperature of ‘‘true ebullition’’.

So he put water into a narrow-necked flask and heated it in a bath
of oil (rather than on an open flame), trying to bring the whole body
of the water to the same temperature by slow heating with minimal
loss of heat at the open surface. But when he did this, De Luc found
that the water would not boil normally at all. The bubbles were
infrequent and very large, sometimes explosive; the temperature
was high and unsteady, sometimes reaching up to 103 C.
A further puzzle awaited De Luc. He noticed that the presence
of dissolved air in water induced what seemed like premature
boiling. He tried to take the air out by various methods, but
concluded that for sufficient results he needed to put previously
boiled water in a sealed container and shake it for a long time
(remember how shaking a bottle of fizzy drink releases bubbles of
gas). He reported: ‘‘This operation lasted four weeks, during
which I hardly ever put down my flask, except to sleep, to do
business in town, and to do things that required both hands. I ate,
I read, I wrote, I saw my friends, I took my walks, all the while
shaking my water.’’ Four mad weeks of shaking had its rewards.
De Luc’s precious airless water reached 112.2 C before boiling off
explosively.
In the course of the 19th century, further study revealed boiling
to be an even more complex and unruly phenomenon than De Luc
had glimpsed. For example, in the 1810s Joseph-Louis Gay-Lussac
in Paris reported that water boiled at 101.2 C in a glass vessel,
while it boiled at exactly 100 C in a metallic vessel6,7. This result
became fairly well known, but there was no definitive explanation
of it available for a long while. In 1842 François Marcet in Geneva
extended Gay-Lussac’s work and reported that water could reach
over 105 C in a glass vessel treated with hot sulfuric acid8.
Superheating became a clearly recognized object of study after
Marcet’s work, stimulating a string of virtuoso experimental
performances vying for record temperatures. François Marie
Louis Donny, chemist at the University of Ghent, combined
Marcet’s ideas with a revival of De Luc’s ideas about the role of
air, and produced a stunning 137 C using airless water in his own
special instrument9. Donny declared:
‘‘The faculty to produce ordinary ebullition cannot in
reality be considered as an inherent property of liquids,
because they show it only when they contain a gaseous
substance in solution, which is to say only when they
are not in a state of purity.’’

224 Hasok Chang

 Black plate (225,1)

Fig. 4. Joseph-Louis Gay-Lussac (1778 – 1850).

www.scilet.com The myth of the boiling point 225

 Black plate (226,1)

In 1861 the work of Louis Dufour, professor of physics at the

Academy of Lausanne, added yet another major factor for consid-
eration10. He argued that contact with a solid surface was the crucial
factor in the production of ebullition, and demonstrated the sound-
ness of his idea by bringing drops of water floating in other liquids
up to 178 C, without even purging the air out of the water. Even
Dufour was outdone, when Georg Krebs in 1869 achieved an
estimated 200 C with an improvement of Donny’s technique11.

Experiments
I was very surprised to read these reports. Having put a detailed
account of them in the first chapter of my book on the history and
philosophy of thermometry1, I was still left with a problem of
incredulity. Were the 18th- and 19th-century scientists right? Or
was this an error like the infamous recent case of ‘‘cold fusion’’, or
the older case of ‘‘N-rays’’? I decided that there was only one way
to find out: see for myself, in the lab.
In six sets of experiments, I confirmed and extended the
seemingly anomalous results that I found in the scientific literature
from the late 18th century and the early 19th century. (Video clips
showing some highlights of these experiments are available
online12.) These experiments were initially carried out in the
summer of 2004, and repeated with very similar results in the
summer of 2007, in the Graham Laboratory at the Department of
Chemistry at University College London. The basic setup was very
simple: distilled water boiled in various containers, with various
heat sources. The temperature was monitored by three types of
thermometers: (a) ordinary mercury thermometers, graduated down
to one degree (centigrade); (b) Beckmann thermometers, which are
very large mercury thermometers with a very fine scale (graduated
down to one-hundredth of a degree), whose zero can be set at any
point in a wide temperature range; (c) a platinum electric-resistance
thermometer, with a digital display, reading down to one-tenth of a
degree.

Experiment 1. The indefiniteness of the boiling point

It is easily seen that ordinary boiling is a very complicated and quite
indefinite phenomenon, as witnessed by the Adams thermometer
discussed above. If water is heated in an ordinary glass beaker with
a Bunsen burner, vapour bubbles can be seen forming at the base of
the beaker from very early on in the process. When the temperature

226 Hasok Chang

 Black plate (227,1)

of the main body of the water is much below 100 C, these bubbles
get collapsed before making it through to the surface. This is called
‘‘subcooled’’ boiling by modern engineers. In older terminology,
what we have is ‘‘hissing’’ (sifflement in French), in which the
vapour bubbles are collapsed, with a characteristic noise, before
reaching the surface of the water. It was also known as the ‘‘singing
of the kettle’’ to serious tea-drinkers – the peculiar noise heard just
before full boiling sets in.
What exactly is boiling, then? If one considers that boiling begins
when the bubbles start breaking the surface, then it seems that water
‘‘begins to boyle’’ around 96 C (204.8 F), in fact just around where
Adams had marked it (Figure 1)! There is quite active boiling from
around 98 C. After ‘‘vehement boiling’’ starts, the temperature does
seem to settle around 100 C (with expected slight variations linked
to atmospheric pressure). But in fact the temperature creeps up
slowly, often reaching around 101 C, consistent with Gay-Lussac’s
report. The maximum attainable temperature after prolonged vehe-
ment boiling is quite a stable point in each instance, but it is not
constant across different circumstances, even under fixed pressure.

Experiment 2. Different temperatures in different vessels

After observing the very ordinary boiling of water with some care,
the next obvious thing to do was to test Gay-Lussac’s claim that the
boiling temperature was affected by the material of the vessel in
which the boiling took place. In the glass vessels I used (Pyrex), the
temperature of prolonged vehement boiling was always over the
standard boiling point, easily reaching 101 C. Even higher tempera-
tures, up to 102.4 C, were observed in a beaker previously treated
with high heat. In metallic vessels, the temperature of vehement
boiling can be as low as 99 C. It was always lower than in glass
vessels, whenever back-to-back trials were made. Among metallic
vessels, great variation was observed depending on the quality of
the inner surface: the temperature can be near that in glass beakers
in a smooth-surfaced stainless pot, but usually lower in a rough-
surfaced stainless pot; the temperature was much lower in a rough-
surfaced aluminium pot (maximum of 99.3 C); Teflon-coated pots
produced the lowest boiling temperatures. The results of my
experiment are easily consistent with the glass-versus-metal differ-
ence of 1.232 C reported by Gay-Lussac.
Generally, it seems quite clear that the maximum attainable
temperature of vehement boiling is different in different types of
vessels. I have easily observed differences of over 4 C in the

www.scilet.com The myth of the boiling point 227

 Black plate (228,1)

temperature of boiling water in quite ordinary glass, metal and

ceramic vessels. The temperature variations do not seem to be a
straightforward function of vessel size and shape (except as the
shape affects the surface area of the water, as discussed in
Experiment 5 below). Rather, it seems that a crucial factor is the
smoothness of the surface, as predicted by the standard modern
theory of bubble-nucleation, discussed below. The differences
observed in different vessels are not only in temperature, but also
in the shape, size and number of bubbles. The variability of boiling
behaviour is illustrated most strikingly in a comparison between
ceramic mugs and Teflon-coated pots. On a Teflon-coated surface,
bubbles form very eagerly and can be seen sticking to the surface
from very low temperatures; the temperature of both the onset and
the peak of boiling is significantly lower compared to boiling in a
glass beaker. In the opposite direction, the trial with some ordinary
ceramic mugs showed very high temperatures, sometimes exceeding
102 C, with bubbles forming and detaching themselves with great
difficulty and a peculiar noise. With bubbles not forming at a high
enough rate, the water cannot lose heat quickly enough, and ends
up in a superheated state. In one case (observed on 2 August 2004)
the water temperature reached 106 C without boiling, at which
point I terminated the heating for fear of explosion. When boiling
takes place in a ceramic container, the bubbles are clearly larger and
less frequent than in other types of vessels, with some completely
quiet periods.

Experiment 3. Lower temperature in a hydrophobic vessel

Gay-Lussac’s explanation for the effect of surface quality was that
the boiling temperature was higher when water adhered to the
surface of the vessel more strongly. When Marcet developed Gay-
Lussac’s work further, he noted that a glass beaker covered in a
layer of sulfur, which was water-repellent, showed a boiling
temperature of 99.7 C. I tried a modern version of the experiment
suggested by Andrea Sella, in which the inside of an ordinary glass
beaker was made hydrophobic by treatment with Me3SiCl
(trimethylsilyl chloride). The boiling temperature in this ‘‘sila-
nized’’ beaker was much lower than in an untreated beaker; in
the silanized beaker, good boiling was possible from around 99 C,
and the maximum temperature attainable was just around 100 C.
The qualitative behaviour was very similar to boiling in the Teflon-
covered pot, with numerous bubbles forming easily from a very low
temperature and crowding around at the bottom surface. To confirm

228 Hasok Chang

 Black plate (229,1)

that nothing had happened to the water chemically, I poured out the
same water to an ordinary beaker and boiled it again; normal
boiling behaviour was observed, with a higher temperature and
only a few spots of active bubble-formation at the bottom surface.

Experiment 4. The action of boiling chips

Gay-Lussac also reported that throwing in metal filings or even
powdered glass into water boiling in a glass vessel lowered the
temperature, bringing it closer to 100 C. In fact this idea was later
developed into the use of ‘‘boiling chips’’ to avoid superheating
and facilitate smooth boiling, a common practice in chemistry labs
to this day. Initially I experimented with relatively old-style boiling
chips, namely ‘‘anti-bumping granules’’ made of marble. These
granules clearly lower the temperature of vehement boiling in glass
and ceramic vessels and bring it close to 100 C, also making the
bubbles smaller and more frequent. In metallic vessels the granules
had much less effect; in one case (2 August 2004), the addition of
the granules to a Teflon-coated pot was even seen to raise the
temperature by about 0.3 C.
With the use of PTFE (Teflon) ‘‘boiling stones’’, the lowering of
the boiling point is more significant. Inserting these boiling stones
into water boiling in a glass beaker at about 101 C produces
extremely vigorous bubbling. The surface of the water breaks
violently, and the boiling stones are continually covered in a film
of vapour. This vigorous boiling brings the temperature down
below 100 C. In trials in ceramic mugs, the boiling temperature
can be brought down by over 3 C. If the boiling stones are placed
in the water before heating (in glass beakers), very active bubbling
starts from around 91 C, and vehement boiling from around 98 C.
The maximum attainable temperature is only around 99 C.

Experiment 5. Superheated boiling by slow heating

I carried out a long series of experiments in an attempt to replicate
De Luc’s work published in 1772, to examine what happens when
water is boiled using a gentle source of heat while minimizing heat
loss at the surface. De Luc had used a round flask with a long, thin
neck for this purpose, and I was able to approximate that quite
easily by using chemists’ volumetric flasks. For the source of heat,
it was convenient to use a hotplate, which is very hot but still much
gentler than a naked flame. The boiling behaviour in this setup is
very different from boiling driven by an open flame. As the

www.scilet.com The myth of the boiling point 229

 Black plate (230,1)

temperature approaches 100 C, the water starts to boil in a normal

way. As boiling continues, however, the temperature continues to
rise, easily exceeding 102 C, while the bubbles increase in size and
decrease in number and frequency. During this phase the bubbles
tend to come in bursts, making what 19th-century observers termed
‘‘bumping’’ (which is prevented by the use of boiling chips). Later
in the process we can observe the ‘‘puffing’’ behaviour: long quiet
periods punctuated by isolated large bubbles; sometimes the puffs
are explosive, throwing some water out of the flask. The tempera-
ture can easily exceed 104 C during puffing. After trying De Luc’s
method of gentle heating with a bath of hot oil, I switched to a bath
of graphite heated on a hotplate for safety and convenience.
The boiling behaviour in a graphite or oil bath is quite similar to
that seen in the hotplate setting. Sometimes there is one spot (or
two) on the inner surface of the flask, or on the thermometer bulb,
from which a continuous stream of bubbles arise (I will call this
‘‘streaming’’). The boiling temperature can be quite steady if there
is continual streaming, and tends to get only up to 102 – 103 C. If
there is no streaming and the boiling is unsteady, the temperature is
quite variable. During bumping, the temperature dips visibly with
the production of a burst of bubbles, and climbs slowly back up
during periods of no activity. During puffing, the temperature
usually creeps up in the long run (despite noticeable dips when
large bubbles are produced), going up to 104 – 105 C. When water
is boiled in this way, even the use of anti-bumping granules fails to
bring down the temperature near 100 C. The granules do facilitate
smooth boiling rather than bumping or puffing; however, even
while smooth boiling takes place, the temperature creeps up
appreciably.
I must emphasise that the superheating observed here is not a
state of unstable equilibrium that is reduced to the ‘‘normal’’ stable
equilibrium on some stimulus. Boiling by streaming can be
sustained, apparently indefinitely, at 102 – 103 C. During bumping
and puffing, the production of a burst of bubbles or a large puff
does bring down the temperature, but never close to 100 C. There
are various stable temperatures depending on the particular circum-
stances, and from my experiments there is no evidence that 100 C
is generally speaking any more stable than any other point.

Experiment 6. Superheating facilitated by de-gassing

Finally, I tried to replicate De Luc’s result on taking dissolved air
out of the water. But unlike De Luc I was not willing to commit to

230 Hasok Chang

 Black plate (231,1)

four weeks of shaking, so I found an alternative method. My

procedure begins with the recognition that heating water to 100 C
already removes much of the dissolved air, since the solubility of
air in water decreases sharply with temperature. And then the water
is boiled for a long time in a loosely covered pot, to sweep out as
much of the remaining air as possible through the bubbling process.
Then the boiled water is poured carefully into a long-necked flask,
and placed on a hotplate. Boiling in this partially degassed water is
very bumpy, and the temperature goes well beyond 100 C, almost
certainly resulting in further degassing (I say ‘‘almost certainly’’
because I have not been able to find data for the solubility of
atmospheric gases for water beyond 100 C). After a while the flask
is removed from the hotplate, and allowed to cool slightly. And
then it is inserted into a graphite bath for gentler heating, with the
graphite temperature at only about 250 C.
At high degrees of superheating the insertion of an ordinary
mercury thermometer excites violent boiling, as the roughness at
the tip of the thermometer serves as a site for nucleation (bubble-
formation). Therefore the temperature of the water in the graphite
bath can only be monitored intermittently. For the most part the
water is absolutely still, although its temperature is very high.
Inserting the thermometer prompts very active boiling, bringing
the temperature down; temperatures of 107 – 109 C are easily
recorded. At high degrees of superheating, the water will
explode on contact with the thermometer, or sometimes sponta-
neously. The Beckmann thermometer, whose bulb is very large
and smooth, does not provide effective nucleation sites, so that a
continuous monitoring of temperature is possible. However, as the
instrument I was using only had a 6-degree scale, and the zero
was calibrated near 100 C, the water temperature went well off the
scale at higher degrees of superheating. Using a Beckmann
thermometer in the degassed water, the result is often quite
remarkable: nothing visible at all happens for a long while, and
then suddenly the water explodes; this happens at estimated
temperatures of 105 – 108 C. In either setup, after an explosion
the water may settle into bumpy boiling at a relatively low
temperature (still well over 100 C), or return to a still state at a
higher temperature punctuated with puffing and explosions. The
latter is the likely outcome when the initial de-gassing is sufficient.
If the surface area of the water is relatively large (which happens
when the water level goes down sufficiently below the neck of the
flask), the fast evaporation that happens at the surface of the
superheated water must cause a rate of heat loss matching the rate

www.scilet.com The myth of the boiling point 231

 Black plate (232,1)

of heat input from the graphite. In that equilibrium situation, it is

easy to have water superheated to 105 – 106 C sitting there
indefinitely with no bubbling.
The following (27 August 2004) is a typical run of this
experiment, with fairly good de-gassing, with a Beckmann thermo-
meter: initial explosion at 105.50 C, about 4 minutes after passing
100 C; post-explosion temperature down to 103.4 C; afterwards
temperature climbing back up quite steadily, despite a couple of
small sharp puffs; 4 minutes after the initial explosion, another
explosion, at about 106.3 C; for 2.5 minutes afterwards, puffing at
around 105.5 C; thermometer removed for the next 4.5 minutes,
resulting in no significant difference in the behaviour of the water;
on re-insertion of the thermometer, temperature found to be steadily
over 106 C, with occasional puffs accompanied by a hissing noise.
All in all, De Luc’s reports on the boiling behaviour of de-gassed
water are amply confirmed, although I have never been able to
record temperatures reaching 112 C as he reported in one trial.

If the boiling point is indefinite, how can

thermometers be made?
The results of these experiments leave a puzzle regarding the very
possibility of thermometry: if there were such unmanageable and
ill-understood variations in the temperatures of boiling water, how
could the boiling point have served as a fixed point of thermometry
at all? It seems that the variations would have threatened the notion
of a definite ‘‘boiling point’’, but the very thermometers used for
the investigation of these variations in the 18th century were
graduated with sharp boiling points. We can only conjecture that
there must have been an identifiable class of boiling phenomena
with sufficiently stable and uniform temperatures, which allowed
the calibration of thermometers with which scientists could then go
on to study the more exotic instances. Happily, a closer look at the
history bears out that conjecture. There were three main factors that
allowed the boiling point to be used as a fixed point despite its
vagaries.
First of all, an immediate relief comes in realizing the difference
between the temperature that water can withstand without boiling,
and the temperature that water maintains while boiling. All
observers of superheating from De Luc onward had noted that the
temperature of superheated water went down as soon as steady
boiling was induced, or each time a large bubble was released.
Extreme temperatures were reached only before boiling set in, and

232 Hasok Chang

 Black plate (233,1)

could be disregarded for the purpose of fixing the boiling point. De

Luc got as far as 112 C without boiling, but the highest temperature
he recorded while the water was boiling was 103 C (a figure quite
consistent with my own results). Still, the latter is 3 C higher than
the ‘‘normal’’ boiling temperature, and there was also Gay-Lussac’s
observation that the temperature of boiling water was over 101 C in
a glass vessel. Marcet investigated this question with more care than
anyone else. In ordinary glass vessels, he observed the temperature
of boiling water to range from 100.4 to 101.25 C. In glass treated
with hot sulfuric acid, the temperature while boiling went easily up
to 103 or 104 C, and was very unsteady in each case due to
bumping8.
The second factor helping to stabilize the boiling point was in
fact a whole set of miscellaneous factors, which might cause
embarrassment to misguided purists. The spirit here was to do
whatever happened to prevent superheating. The Royal Society
Committee avoided superheating by using metallic vessels instead
of glass. Gay-Lussac showed how to prevent superheating in glass
vessels by throwing in metal chippings or filings (or even powdered
glass). Other investigators found other methods, including the
insertion of solid objects (especially porous things like charcoal
and chalk), sudden localized heating, and mechanical shocks. But in
many practical situations the prevention of superheating simply
came down to not bothering too much. If one left naturally
occurring water in its usual state full of dissolved air (rather than
taking the trouble to purge air out of it), and if one left the container
vessels just slightly rough (instead of smoothing it off with some-
thing like hot sulfuric acid), and if one did not do anything else
strange like isolating the water from solid surfaces, then common
boiling did take place. Serious theoretical arguments about the
factors that facilitate ebullition continued into the 20th century,
but all investigators agreed sufficiently on how to break super-
heating and prevent bumping in practice. Marcel Verdet observed
that under ‘‘ordinary conditions,’’ there would be dissolved air in
the water and the water would be in contact with solid walls, and
hence boiling would ‘‘normally’’ set in at the normal boiling
point11. It was a great blessing for early thermometry that the
temperature of boiling was quite fixed under the sort of circum-
stances in which water tended to be boiled by humans living in
ordinary European-civilization conditions near the surface of the
earth without overly advanced purification technologies.
However, happy-go-lucky sloppiness is not the most robust
strategy of building scientific knowledge in the end, as the Royal

www.scilet.com The myth of the boiling point 233

 Black plate (234,1)

Society Committee realized quite well. The committee’s lasting

contribution, the last of our three factors contributing to the fixity of
the boiling point, was to find one clear method of reducing the
variations of the boiling point due to miscellaneous causes: ‘‘The
most accurate way of adjusting the boiling point is, not to dip the
thermometer into the water, but to expose it only to the steam’’ 3.
Somehow, using the boiled-off steam rather than the boiling water
itself seemed to eliminate many of the most intractable variations in
the temperature:
‘‘The heat of the steam therefore appears to be not
sensibly different in different parts of the same pot;
neither does there appear to be any sensible difference
in its heat, whether the water boil fast or slow; whether
there be a greater or less depth of water in the pot; or
whether there be a greater or less distance between the
surface of the water and the top of the pot; so that the
height of a thermometer tried in steam, in vessels
properly closed, seems to be scarce sensibly affected
by the different manner of trying the experiment’’ 3.
The recommendation to use steam came most strongly from
Cavendish, who had already made the same proposal in his
review of the instruments used at the Royal Society. Marcet later
confirmed that even steam resulting from highly superheated boiling
was only a fraction of a degree above 100 C8.

Why don’t we all know about the variations?

Even after we understand how the accurate graduation of thermo-
meters was possible despite the variability of the boiling point, a
question remains as to why most people are not aware of that
variability, even though most of us boil water on a daily basis.
The main reason is that in everyday life we tend to boil water
always in the same sort of manner: in broad vessels with heating
from the bottom, which requires intense heat sources because of
the high rate at which heat is lost at the open surface and the
sides. (In the graphite bath used in Experiment 6, I could not
make water boil at all in any wide-open vessels, despite urging
the graphite temperature to over 250 C.) In order to see that the
‘‘standard’’ act of boiling could be different for people with
different ways of life, consider the following three situations,
one fictitious and two real.

234 Hasok Chang

 Black plate (235,1)

(a) Imagine a civilization with no access to open flames, but only to hot
stones or sand; there, boiling would be done in narrow-necked
flasks, routinely producing the kind of superheated boiling that I
have observed in volumetric flasks heated with a graphite bath or a
hotplate. Such people would not dream of drawing a sharp line
between the ‘‘liquid’’ and ‘‘gas’’ regions in a temperature-pressure
phase diagram.
(b) Coming back to real life in the 20thy21st century, note widespread
reports of the superheating of water in microwave ovens, which heat
water directly and evenly, not by means of a hot solid surface. This
phenomenon is often noticed because the superheated water is apt
to boil over violently when instant coffee is dropped into it. (Try
putting ‘‘superheated water’’ into an internet search engine; Joe
Wolfe of the University of New South Wales has a particularly good
online discussion13.) Some recent researchers on boiling have
employed laser pulses as a heating mechanism.
(c) In the 19th century, the age of the steam engine, boiler explosions
resulting from superheating created a real concern, and various
methods of preventing superheating were devised successfully. I
have not seen a good theoretical or historical account of this
business, but I suspect that the water in the steam-engine setup is
prone to de-gassing, and that the explosive force would have been
quite considerable in high-pressure engines. In the 20th century,
mechanical and chemical engineers encountered a wide variety of
situations in which boiling takes place, and have been accumulating
experimental and theoretical knowledge of various different types of
boiling that take place in different situations, as I will explain briefly
in the next section14,15,16.

Modern theories of boiling

In the traditional theory of boiling in physics and physical
chemistry, the boundary between the liquid state and the gas state
is sharply defined as the line on which the vapor pressure (as a
function of temperature) reaches the level of the external pressure
This is as shown in a typical phase diagram (Figure 5). The
assumption of the sharpness of the liquid-gas boundary theoretically
precludes any variability in the boiling temperature under fixed
external pressure; this means that there is no obvious way of
accommodating the observed variations within the traditional
physical theory. In modern treatises on boiling in mechanical and
chemical engineering, we do not find the standard thermodynamic

www.scilet.com The myth of the boiling point 235

 Black plate (236,1)

Fig. 5. A standard phase diagram.

phase diagrams. Instead, the engineers’ paradigmatic representation

of boiling is the ‘‘boiling curve’’, which plots the rate of heat
transfer against the degree of the ‘‘surface superheat’’ or the
‘‘excess temperature’’. Figure 6 shows a typical boiling curve.
The contrast between the phase diagram and the boiling curve
reveals a fundamental difference in physicists’ and engineers’
understanding of boiling. In fact it is impossible to translate
straightforwardly between the two discourses, so the situation
here warrants the use of Thomas Kuhn’s concept of incommensur-
ability17.
The main independent variable in the engineering discourse is
how much the temperature of the heating element exceeds the
‘‘normal’’ boiling point. I assume that the water in immediate
contact with the heating element (what De Luc called the ‘‘first
layer’’ of water) is also heated beyond the normal boiling point –
by how much, we cannot really say. It would be extremely difficult
to measure such a thing and, presumably, the engineers are more
interested in variables that they can measure and control, like the
temperature of the heating element. Therefore, in the best modern
theory of boiling we have, the temperature of the water itself has no
role to play! And if we do assume that there is some degree of
superheating in the first layer of water, and seek to say something
about the effect of that superheating, we find that there is no theory
that can be applied easily. The question cannot even be articulated
in standard thermodynamic discourse, because the basic theory

236 Hasok Chang

 Black plate (237,1)

Fig. 6. A standard boiling curve, from Incropera and DeWitt (1996), p. 502 (figure
10.4). Courtesy of John Wiley and Sons.

there rests on the idealized assumption that superheating never

occurs. Instead, we would need to give a microphysical treatment of
the chemical bonds among the water molecules, and the interaction
of water molecules with solid surfaces (and with molecules of
dissolved gases).
The other main thing to note about the engineers’ boiling curve is
that the main dependent variable is the rate of heat transfer. These
engineers are mainly interested in boiling as a method of carrying
heat away from hot places (one can easily imagine the conse-
quences of not understanding this correctly, in trying to keep a
nuclear reactor from overheating, for example). In that context, the
temperature of the liquid water, especially well above the first layer,
is distinctly of secondary interest, and is freely admitted to be quite
variable depending of the situation. The engineering treatises on
heat transfer give a detailed classification of boiling behaviour,
largely determined by the degree of surface superheat and the
configuration of the boiling setup. A great deal of experimental
work is also going on.

www.scilet.com The myth of the boiling point 237

 Black plate (238,1)

Particularly pertinent for current purposes is the modern theory

of nucleation (bubble-formation), which gives excellent and
detailed explanations of the effect of vessel-surface quality on
boiling behaviour (shown in Experiment 2 and Experiment 4).
Surface tension emerges as the basic reason why water is prone to
superheating; in order for boiling to take place, vapour bubbles
need to form within the body of the water, and grow sufficiently
to be visible as they come up through the water. Now, the basic
condition for a bubble to sustain itself is that the vapour pressure
should match (or exceed) the external pressure. However, there is
a complication here because the water molecules forming the
surface of the vapour bubble attract each other. This attraction
manifests itself in the form of surface tension, which tends to
close up the bubble. Therefore an additional force of vapour is
required to sustain the bubble, which means the water temperature
has to be higher than the boiling point indicated by the simple
balance between the vapour pressure and the external atmospheric
pressure.
Standard theory says that the additional pressure created by
surface tension is inversely proportional to the radius of the
bubble. In other words, the additional pressure to be overcome
becomes infinite when the radius is 0, which means that it would be
impossible to grow vapour bubbles without starting from finite-
sized spaces. This is why the precise quality of the solid surface
becomes so important for the facilitation of boiling. If the surface
has micro-pores and it is sufficiently water-repellent, then there
would be pockets of vacuum or trapped air that can serve as the site
of bubble-formation (nucleation). This is why bubbles only arise
from specific places in ordinary boiling situations, especially as the
water gets more and more de-gassed with the progress of boiling.
Some questions arising from my experiments are more difficult,
and they are not satisfactorily resolved by an elementary knowledge
of the modern engineering theory of boiling that I have so far
acquired. First, it is difficult to understand the role of dissolved air
in facilitating boiling; this may require some detailed molecular
modelling, which the engineering theory does not provide. Second,
the lowering of the boiling temperature below the thermodynami-
cally defined boiling point is difficult to understand. The only
explanation I can currently offer is that the ‘‘first layer’’ of water
in those situations must be heated well beyond the normal boiling
point although the main body of the water is much cooler, and that
the bubbles rise to the surface before they have enough time to be
collapsed in coming through the cooler water.

238 Hasok Chang

 Black plate (239,1)

Concluding remarks
My investigation has revealed some significant gaps in the common
knowledge of boiling in standard physics and chemistry, especially
in the way these subjects are taught, even in higher education.
These gaps exist not because science is incapable of filling them,
but because science needs to set aside many questions and facts in
order to allow its focus on the current cutting-edge of research.
History and philosophy of science can serve the function of
preserving and developing aspects of scientific knowledge that are
lost and neglected in the process of scientific progress. I would have
not learned all the good things about boiling that I have presented in
this paper, if I had not started by learning from historical sources.
And I would not have looked into that history if I had not been
investigating philosophical questions about how we can know
whether our thermometers are reliable. Using history and philo-
sophy of science to improve our knowledge of nature is a program
of research that I call ‘‘complementary science’’ because it supple-
ments current specialist science without disputing its legitimacy1. I
hope that this brief presentation of a concrete question has given
you a glimpse of the potential of this research programme.

Acknowledgements
I would like to thank colleagues in the UCL Department of
Chemistry, especially Andrea Sella and Crosby Medley, for their
help with the experiments. I thank the Leverhulme Trust for
providing funding for the experimental work. Thanks are due to
many more people, who are listed in the online paper accompanying
the video clips.

References
1. Chang, H. (2004) Inventing temperature: measurement and scientific progress.
Oxford University Press, New York.
2. Newton, I. (1701) Scala Graduum Caloris. Calorum Descriptiones et Signa.
Phil. Trans. Roy. Soc. Lond., 22, 824 – 829.
3. Cavendish, H., Heberden, W., Aubert, A., De Luc, J.-A., Maskelyne, N.,
Horsley, S. and Planta, J. (1777) The Report of the Committee appointed by
the Royal Society to consider of the best method of adjusting the fixed points of
thermometers; and of the precautions necessary to be used in making experi-
ments with those instruments. Phil. Trans. Roy. Soc. Lond., 67, 816 – 857.
4. Cavendish, H. (1921) Theory of boiling. The scientific papers of the Honour-
able Henry Cavendish, F.R.S., Vol. 2, Chemical and dynamical, Edward Thorpe
(ed.), pp. 354 – 362. Cambridge: Cambridge University Press.

www.scilet.com The myth of the boiling point 239

 Black plate (240,1)

5. De Luc, J.-A. (1772) Recherches sur les modifications de l’atmosphère, Vol. 2,

pp. 227 – 438. Geneva.
6. Gay-Lussac, J.-L. (1818) Notice respecting the fixedness of the boiling point of
fluids. Annls. Philos., 12, 129 – 131.
7. Biot, J.-B. (1816) Traité de physique expérimentale et mathématique, vol. 1, pp.
41 – 43. Deterville, Paris.
8. Marcet, F. (1842) Recherches sur certaines circonstances qui influent sur la
température du point d’ébullition des liquides. Biblioth. Univers., New Ser., 38,
388 – 411.
9. Donny, F. (1846) Mémoire sur la cohésion des liquides, et sur leur adhérence
aux corps solides. Annls. Chim. Phys., 3rd Ser., 16, 167 – 190.
10. Dufour, L. (1861) Recherches sur l’ébullition des liquids. Arch. Sci. Phys.
Natur., 12, 210 – 266.
11. Gernez, D. (1875) Recherches sur l’ébullition. Annls. Chim. Phys., 5th Ser., 4,
335 – 401.
12. http://www.ucl.ac.uk/sts/chang/boiling
13. http://www.phys.unsw.edu.au/~jw/superheating.html
14. Incropera, F. P. and DeWitt, D. P. (1996) Fundamentals of heat and mass
transfer, 4th edn. John Wiley and Sons, New York.
15. Hewitt, G. F., Shires, G. L. and Polezhaev, Y. V. (eds.), (1997) International
encyclopedia of heat and mass transfer. CRC Press, Boca Raton.
16. Kandlikar, S. G. (ed.), (1999) Handbook of phase change: boiling and
condensation. Taylor & Francis, London.
17. Kuhn, T. S. (1962) The structure of scientific revolutions. University of Chicago
Press, Chicago.

240 Hasok Chang