Elements and Atoms: Chapter 6
Water is not an element: Lavoisier

Of the four elements of the ancients, water is the only one which is a pure chemical substance, albeit a compound and not an element. It should not be surprising, then, that it was the last of the four to be shown not to be elementary. After all, water is quite stable thermodynamically, and therefore rather difficult to decompose. And water is so common that its formation in many processes would be quite easy to overlook, especially if its formation was unanticipated or if other possible sources of moisture were present.

The discovery that water is a compound of hydrogen and oxygen (or, in the terminology of the day, of inflammable air and dephlogisticated air) was made in the early 1780s. Credit for the discovery has been given to or claimed on behalf of no less than four individuals: Henry Cavendish (View portrait at the Edgar Fahs Smith collection, University of Pennsylvania.) [Cavendish 1784], Antoine Lavoisier [la Place & Lavoisier 1781], Gaspard Monge (View portrait Centre International de Mathématiques Gaspard Monge.) [Monge 1783], and James Watt (View portrait at the Edgar Fahs Smith collection, University of Pennsylvania.) [Watt 1784][1]. The "water controversy" has engaged historians of chemistry and partisans of the protagonists since the early 19th century. Sidney Edelstein makes a persuasive case for Watt's priority [Edelstein 1948]; however, despite the title of his article "Priestley Settles the Water Controversy," the controversy apparently has not been settled, nor will I attempt to do so.

Even though it is apparent that Lavoisier was not the first to realize the compound nature of water, and was indeed aware of the work of his English contemporaries, I have chosen a selection from his work for the present chapter. The selection, a report of a paper (not even his first on the subject) read to the French Academy late in 1783, has several advantages over those of Watt and Cavendish: it is focused on the nature of water; it addresses experiments beyond the burning of hydrogen to produce water; and it is not couched in the terminology of the phlogiston theory (except in calling oxygen "dephlogisticated air").

Report of a memoir read by M. Lavoisier at the public session of the Royal Academy of Sciences of November 12, on the nature of water and on experiments which appear to prove that this substance is not strictly speaking an element but that it is susceptible of decomposition and recomposition

Observations sur la Physique 23, 452-455 (1783);[2] translation by Carmen Giunta

From the year 1777, M. Lavoisier and M. Bucquet, in a series of experiments carried out jointly, noticed that burning large amounts of inflammable air, obtained from the dissolution of iron in vitriolic acid, formed not the slightest amount either of fixed air nor of any other acid whatsoever.[3]

M. Cavendish made the same observation in England. Furthermore, he observed that if one operates in dry vessels a discernible quantity of moisture is deposited on the inner walls.[4]

Since the verification of this fact was of great significance to chemical theory, M. Lavoisier and M. de la Place[5] proposed to confirm it in a large-scale experiment; and in order to give it greater authority, they engaged several Members of the Academy to be present at it.[6] They prepared a sort of double-tubed lamp for inflammable air, one tube carrying inflammable air and the other dephlogisticated air. The two orifices through which the airs passed were severely restricted, to make the combustion very slow, and they were proportioned in such a way as to supply the amounts of the respective airs needed for combustion. The glass bell into which the double tube led was immersed in mercury, and had no communication with the exterior air.[7] Last July or August M. Lavoisier gave the Academy a detailed description of this apparatus. The quantity of inflammable air burned in this experiment was about thirty pintes and that of dephlogisticated air from fifteen to eighteen.[8]

As soon as the two airs had been lit, the wall of the vessel in which the combustion took place visibly darkened and became covered by a large number of droplets of water. Little by little the drops grew in volume. Many coalesced together and collected in the bottom of the apparatus, where they formed a layer on the surface of the mercury.

After the experiment, nearly all the water was collected by means of a funnel, and its weight was found to be about 5 gros, which corresponded fairly closely to the weight of the two airs combined.[9] This water was as pure as distilled water.

A short time later, M. Monge addressed to the Academy the result of a similar combustion, carried out at Mézières, with a totally different apparatus and which was perhaps more accurate. He determined with great care the weight of the two airs, and he likewise found that in burning large quantities of inflammable air and dephlogisticated air one obtains very pure water and that its weight very nearly approximates the weight of the two airs used. Finally, it was reported in a letter written from London by M. Blagden to M. Bertholet, that M. Cavendish recently repeated the same experiment by different means and that when the quantity of the two airs had been well proportioned, he consistently obtained the same result.

It is difficult to refuse to recognize that in this experiment, water is made artificially and from scratch, and consequently that the constituent parts of this fluid are inflammable air and dephlogisticated air, less the portion of fire[10] that is released during the combustion.

Meanwhile, before admitting a consequence so remote from all received ideas, M. Lavoisier thought it necessary to multiply the proofs and above all, after having established by means of composition the nature of the constituent parts of water, to set himself to the task of regenerating them by means of decomposition.[11]

With this purpose, he filled a crystal bowl with mercury, inverted it in a vessel filled with mercury, and introduced a small portion of water and of iron filings, very pure and not rusted.[12] From the first day, the iron began to lose a part of its metallic luster; it was calcined and converted in part to rust.[13] At the same time it released a quantity of inflammable air in proportion to the quantity of dephlogisticated air which had been absorbed by the iron, as judged by the increase in weight which the filings had aquired after being dried. Thus water, in this experiment, is decomposed into two distinct substances, dephlogisticated air which unites with the iron and converts it to a calx, and inflammable air which remains separate.[14] On the other hand, when one reunites and recombines these same two substances, one recomposes water. Thus one is led still more nearly inevitably to conclude that water is not a simple substance at all, not properly called an element, as had always been thought.

It is easy to imagine that this discovery must have opened to M. Lavoisier a vast field of experiments[15], and they led him to believe that a great number of phenomena which were attributed to the decomposition of bodies were due to that of water. The dissolution of metals in acids supply striking examples. In almost all these operations, the metal begins to be calcined before dissolving; that is to say, that it combines with a certain quantity of dephlogisticated air, a different amount according to the nature of the metal.[16] He maintains he obtained proof through these experiments, many of which he performed jointly with M. de la Place, that in all the dissolutions of metal in vitriolic acid, the dephlogisticated air needed for the calcination of the metal is not supplied by the acid but by the water, and that at the same time inflammable air, which is one of its constituents, becomes freed and is released in its aeriform state.

In contrast, in the dissolutions of metals in nitrous acid[17], the greater part of the dephlogisticated air is supplied by the acid, and the water only contributes a small portion. He reports that he has not yet attempted any research on dissolutions by marine acid, because of some difficulties which attend this kind of combination, of which he promises to give an account.

After having followed the effects of the decomposition of water in the dissolution and calcination of metals, M. Lavoisier gave an account of several experiments which he undertook with the same aim on the fermentation of wine. Although it happens that he has not yet obtained an absolutely decisive result, he thinks it correct meanwhile to suspect and even to believe that the formation of the vinous ingredient is due to the decomposition of water. In this operation, the dephlogisticated air from water unites with the carbonaceous part of the sugary substance and forms fixed air, which is released thoughout the duration of the fermentation.[18] At the same time, inflammable air, modified and combined with another portion of water by means of an intermediate as yet unknown forms the spiritous part. Likewise, in following the operation of plants, he appears to be brought to believe that the formation of combustible plant material is due to the inflammable air contained in water.[19] Doubtless these assertions appear perhaps[20] hazarded at first glance; however, M. Lavoisier promised further detail beyond the evidence contained in this first Memoir. He finished his Memoir with this modest conclusion: that if the decomposition of water in a multitude of operations of Nature and Art is not rigorously demonstrated, it is at least infinitely probable.


[1]These four figures serve as an excellent illustration of the fact that science in the late 18th century was not pursued by specialized professionals. As the chief figure in the "Chemical Revolution," Lavoisier is identified most strongly with chemistry. His interests and work, however, extended to such diverse areas of application as agriculture and munitions. Along with Lavoisier, Cavendish is a familiar figure to chemists acquainted with the history of their discipline. The most important chemical work of Cavendish was in the study of gases. He also made significant contributions in such physical fields as electricity and gravity. Watt and especially Monge are less familiar to chemists. Watt is best remembered as an engineer for his work on the steam engine and Monge as a mathematician, the founder of descriptive geometry.

[2]Rozier's Observations sur la Physique printed many papers read before the French Academy of sciences, or abstracts of such papers, before the Academy printed the official and often revised versions. The final form of this article appears in la Place & Lavoisier 1781.

[3]The first clue Lavoisier mentions is a negative one: no acid obtained upon combustion. Burning carbon-containing compounds produced carbon dioxide ("fixed air"), while burning nitrogen-containing compounds produced nitrogen oxides; both of these substances dissolve in water to form acidic solutions. But burning hydrogen ("inflammable air") produced no acid.

[4]This is an important observation, but does not form a sufficient basis for the conclusion that water is the product of a reaction between hydrogen and oxygen. In terms of a conventional description of the scientific method, this is an observation which inspires a fairly obvious hypothesis (namely that water is formed upon burning hydrogen). The next step in a proper investigation would be to design a careful experiment to test the hypothesis. For instance, such a test would at least make the observation of moisture after combustion unequivocal, and would take pains to exclude other possible sources of water. Lavoisier goes on to describe his own test of the hypothesis, and then mentions similar experiments by other investigators (including Cavendish).

[5]M. de la Place is Pierre-Simon Laplace, better known as a mathematician, astronomer, and physicist.

[6]This is a sort of peer review, but in a form not common today. By the late 18th century, peer review in the form of reporting observations and explanations to other interested and competent persons was well established; indeed, royal academies of science in France and England had regular meetings at which researchers reported observations and experiments and their interpretations.

[7]So the apparatus was designed to slowly burn a mixture of hydrogen and oxygen in a closed container. View a diagram of the apparatus from Lavoisier's Oeuvres, Vol. V.

[8]Lavoisier was not primarily interested in the proportions required for complete reaction between oxygen and hydrogen. Still an observation of a combining ratio of roughly two hydrogen to one oxygen by volume occurs even at this early date and in an experiment not primarily concerned with stoichiometry. We will see a collection of observations about combining ratios of gases a quarter-century later by Joseph-Louis Gay-Lussac [Gay-Lussac 1809].

[9]Here is an example of Lavoisier's signature measurement, the comparison of the weight of the product of the reaction to the weight of the reactants. By accounting for most, if not all, of the weight of the reactants in the products, Lavoisier provided good evidence that he had accounted for all the reactants and products. That is, he could say not only that the combination of oxygen and hydrogen produced water; he effectively ruled out any other reactant or major product in the process.

To take another example from elsewhere in Lavoisier's work [Lavoisier 1775], the fact that the red calx of mercury (mercuric oxide) could be turned into mercury metal was well known to chemists, as was the fact that the mercury product weighed less than the original calx. Lavoisier found that the mass of the products roughly balanced that of the original calx when he collected and accounted for the gas (oxygen) that was also produced in the process.

Lavoisier's reported measurements for burning hydrogen hold up reasonably well. He does not state error limits; however, it is reasonable to infer that his 5 gros means between 4.5 and 5.5 gros. I will translate his observation into modern units, carrying more significant figures than are warranted and then accounting at the end for the inferred uncertainty of roughly ±10%. A gros was a unit of mass defined as 1/8 of a French ounce (once); 1 gros = 3.82 grams. Thus Lavoisier reports collecting about 19.1 grams of water or 1.06 moles. A pinte was a volume measurement equal to just over two English pints; 1 pinte = 0.953 liters. So Lavoisier burned about 28.6 liters of hydrogen. Assuming atmospheric pressure and a temperature of about 20°C, this corresponds to 1.26 moles. Since the reaction is:

2 H2 + O2 --> 2 H2O ,
the moles of water produced are equal to the moles of hydrogen consumed. The quantities just computed from the reported measurements agree reasonably well, given the uncertainty in the reported measurements and in the unreported temperature of the hydrogen; the mean of the two figures is within 10% of each of them.

[10]See the previous chapter for Lavoisier's conception of "matter of fire." He considered the flame and heat evolved in a combustion to be a very light form of matter.

[11]Note the value of multiple independent lines of evidence to support a hypothesis. There was no obvious flaw in reasoning or technique in the experiment Lavoisier just described; however, if he could come to the same conclusion by some other means, the conclusion could still stand even if flaws were found in the experiment. Multiple independent measurements are useful even in routine investigations; in fact, we will see how they led from routine measurement to an unexpected discovery in chapter 14. But multiple lines of evidence are especially helpful in establishing a result as surprising as this was. (After all, water was supposed to be an element.)

Finally, Lavoisier mentions the pair of complementary approaches that was particularly useful in chemistry: synthesis and analysis. The evidence presented to this point came from synthesis: putting together hydrogen and oxygen to get water. Lavoisier would now turn his attention to taking water apart to see if it can be broken down into those same components.

[12]The significance of this arrangement is that the filings were not in contact with the atmosphere, which of course contains oxygen.

[13]By this time, Lavoisier had already explained calcination of a metal as the reaction of a metal with oxygen. (See previous chapter.) The product was then usually called a calx or earth; now we call it a metal oxide. Rust is simply the calx or oxide of iron, Fe2O3. So Lavoisier knew that if rust was formed, the iron picked up oxygen from somewhere.

[14]Lavoisier described elsewhere another experiment in which water was decomposed by contact with iron. View a diagram of the apparatus at Les Amis de Lavoisier. Here water vapor was placed in contact with a hot iron gun barrel.

[15]This is a common occurence in science: once a new idea has become established (even if only in the mind of the investigator and not yet in the scientific community as a whole), it is natural and often fruitful to look for other occurences of it. Lavoisier has just seen water decomposed, so he looks for other circumstances in which water might decompose. In chapter 15 we will see William Ramsay, after discovering a gas with novel properties, begin to search for other new gases with similar properties.

In this case, Lavoisier seems to find the phenomenon where it actually does not occur. False leads in circumstances such as this are also not uncommon, as we shall see in chapter 17 on the discovery of radioactivity. Preconceptions and expectations can lead to fruitful investigations, but they can also color interpretations and even observations. We have already seen Priestley candidly admit to missing unexpected phenomena (chapter 4). Now that the previously unexpected decomposition of water has become evident to Lavoisier, he finds it even in some circumstances where it does not occur.

[16]With our present knowledge we can see that this is not quite correct: the dissolution of metals in strong acids does not involve hydrolysis (the decomposition of water). The process involves a transfer of electrons from the dissolving metal to the hydrogen ions (H+) that are plentiful in strong acid solutions. The acquisition of an electron leads to neutral hydrogen atoms that combine to yield gaseous hydrogen (H2). The loss of electrons by the metal leaves a positively charged metal ion, similar to its condition in a calx or in a salt. The change in surface luster of a metal dissolving in acid is actually the formation of a salt, not a calx: there is no oxygenation involved.

Actually, Lavoisier's chemistry did not make the distinction between water-soluble anhydrous oxides and their corresponding acids. For example, sulfur trioxide gas, SO3, and sulfuric acid, H2SO4, would both be sulfuric acid (or vitriolic acid) to Lavoisier. One can imagine construing a phrase like "dissolving zinc in dilute sulphuric acid" as the reaction

SO3(g) + H2O(l) + Zn(s) --> H2(g) + Zn2+(aq) + SO42-(aq) ,
in which case water is indeed decomposed in the process. The more natural reading, however, is to take the dilute acid to be a solution to which the zinc is added; in that case, what dissociates is not water but H2SO4, and the dissociation (but not the release of H2(g)) occurs even before the zinc is added.

[17]Nitrous acid is now known as nitric acid, HNO3, a different substance from the one now called nitrous acid, HNO2. Lavoisier was correct in distinguishing between dissolutions in sulfuric and nitric acids. The reaction in sulfuric acid (and in "marine" acid, now known as hydrochloric acid, for that matter) is as described in the previous note. In nitric acid, however, the nitrogen receives the electrons given up by the dissolving metals.

[18]The "vinous" or "spiritous" part of wine is ethanol (also known as ethyl alcohol, C2H5OH). Lavoisier was mistaken here as well, however, for the process of fermentation does not involve hydrolysis. [Giunta 2001] Lavoisier was misled by the production of carbon dioxide into thinking that oxygen participated in the reaction; in fact, fermentation is a microbially assisted breakdown of sugar into alcohol and carbon dioxide.

[19]The process ultimately responsible for the combustible matter in plants is photosynthesis, the light-induced production of complex carbon-containing compounds and oxygen from water and carbon dioxide. Although his evidence was no better here than in the case of fermentation, Lavoisier happened to be correct that water is taken apart in this process.

[20]Translator's note: "Doubtless ... perhaps" is awkward, but faithful to the text ("sans doute ... peut-être").


Back to the top of the table of contents of Elements and Atoms.
Back to the top of Classic Chemistry.