Water is not an element

Two gases burn together in a sealed globe and the walls fog with dew that weighs exactly what the gases weighed. Steam driven through a red-hot iron gun barrel comes out the far end as inflammable air, leaving the iron heavier by the difference. Synthesis and analysis, within two years of each other, and the second of the four classical elements is gone.

1766 – 178716 min

By the end you should be able to

Explain why water was the most convincing of the four elements, and what evidence had to be cleared away first
State what Cavendish measured, and why he read his own result in phlogiston terms
Explain why synthesis and analysis together settle a question that either alone leaves open
Recover the combining ratio of the two gases from Cavendish’s published volumes

Of the four classical elements, water had by far the best case. It is uniform. Rain, spring water, melted snow and the distillate off a still are the same substance, and they behave identically. Nothing else in the eighteenth-century laboratory was that dependable. It survives every operation. Freeze it, boil it, condense it, freeze it again — you get the water back, exactly, every time. That is what an element is supposed to do. It has no parts you can get at. Every decomposition anyone knew ran on heat or on charcoal, and water laughs at both. Heat turns it into steam and the steam turns back into water. And it appeared to be capable of transmutation into earth. Boil water in a glass vessel for long enough and a solid residue appears. Distil that water and the residue appears again. Boyle had reported it, it had been confirmed many times over, and it was still being defended by respectable chemists in Lavoisier’s own lifetime. The natural reading is that water — the simplest thing there is — can be condensed into earth, the other simple thing. That last one had to go first, and getting rid of it was the first serious thing Lavoisier ever did.

The experiment

Water boiled for a hundred and one days

Antoine-Laurent Lavoisier · 1768 – 1770 · Paris

The question: Prolonged boiling turns water into an earthy residue. Is water transmuting into earth, or is the earth coming from somewhere else?
The apparatus: A pelican: a sealed glass vessel whose two curved side-arms return the condensate to the body, so that the water can be boiled indefinitely without any of it escaping. Distilled water sealed inside; the whole apparatus weighed before, and boiled continuously for 101 days.

Theory predicted

If water becomes earth, the water plus its residue should weigh what the water did, and the glass should be unaffected.

They measured

The earthy residue duly appeared. The apparatus as a whole had not changed weight at all. But the glass vessel, weighed separately afterwards, had lost very nearly the weight of the residue that had appeared inside it.

Why it mattered

The earth was the flask. Hot water dissolves glass, slowly, and 101 days is long enough to see it on a good balance. Water does not transmute into anything, and the classic evidence that it does was an artefact of the container. It is also the earliest instance in this course of the move Lavoisier would build a career on: weigh the whole system, including the parts nobody thought to weigh.

Meanwhile there was a gas nobody could place. Drop iron filings or zinc into dilute vitriolic acid and it fizzes off something that is not fixed air and not common air. It is so light that a bladder filled with it rises to the ceiling. It burns instantly, and a mixture of it with air explodes with a sharp bark that will crack a bottle. Inflammable air. Cavendish characterised it properly in 1766 and did the thing nobody else was doing: he measured its density. He made it roughly a fourteenth as heavy as common air — the lightest substance anyone had ever weighed, by a wide margin. And for a while he thought it might be phlogiston itself. That sounds absurd now and it was a good guess then. Phlogiston is what leaves a metal when the metal turns to calx. Here is a metal dissolving in acid and giving off, at that exact moment, a substance that is fantastically light and burns with no residue whatsoever. If you had to nominate a candidate for pure phlogiston, this is what it would look like.

The experiment

Exploding inflammable air with common air

Henry Cavendish · Experiments from about 1781; read to the Royal Society and published in Philosophical Transactions, 1784 · Clapham Common, London

The question: What is left when inflammable air is burned in a closed vessel? Every other combustion produces fixed air, or an acid, or a solid. What does this one produce?
The apparatus: A glass globe of known volume, filled with measured volumes of inflammable air and common air, sealed, and fired by an electrical spark. The globe weighed before and after; the condensate collected, weighed and tasted for acidity.

Theory predicted

On the phlogiston account inflammable air is a phlogiston-rich substance, and burning it should leave whatever the phlogiston was attached to. Nobody predicted the answer.

They measured

The walls fogged with dew. Collected and examined, it was pure water — no acid, no residue, nothing dissolved in it, provided both gases were clean to start with. The two gases were consumed together: 423 measures of inflammable air went with 1000 measures of common air, and about a fifth of the common air disappeared along with nearly all of the inflammable air. The water weighed what the gases had weighed.

How sure could they be? Cavendish was probably the most careful measurer of the century, and this is where it shows. He noticed that when the mixture was not proportioned correctly a trace of acid appeared with the water, tracked it down, and thereby discovered nitric acid — and, in the course of the same work, found that about 1/120 of common air would not react with anything at all. That residue was argon, and it was not identified for another 110 years.

Why it mattered

Water can be made. It is the product of burning one gas in another, and the masses balance exactly. Whatever water is, it is not something that arrives from outside the reaction.

It is worth pausing on how strange that result would have been. Everything else that burns produces something obviously different from what went in: ash, a calx, fixed air, an acid smoke. Here, two invisible elastic fluids are exploded together and what is left is a puddle. The most familiar liquid there is, produced out of two gases, in a sealed globe, from nothing that was wet. And the accounting closes. The globe weighs the same before and after — this is a sealed vessel, and by now you know what that means. The water inside it weighs what the two gases weighed. Nothing entered and nothing left. That is one half of the argument, and by itself it is not quite enough. Someone can always say: you have not shown that this dew is water; you have shown that it is a new liquid that resembles water. The way to close that off is to start from water — ordinary water, out of a bucket — and take it apart.

Iron and a weighed charge of water, sealed in together with no air at all. Heat it, and watch three numbers: what the iron gains, what gas appears, and what the total does.

Loading the balance…

The decomposition, run on the balance. The iron blackens and gains weight; inflammable air appears in its place; the total does not move. Read the note under the water gauge: the pressure never changes, because every molecule of water that goes is replaced by one of inflammable air, so the gas in the vessel is swapped out entirely without the count altering. The real experiment was a flow system — a gun barrel with steam going in one end and the gas collected at the other — precisely so that the hydrogen could be weighed on its own. Sealing it in the same vessel is what makes the null reading visible.

The experiment

Water decomposed by red-hot iron

Antoine-Laurent Lavoisier and Jean-Baptiste Meusnier · 1783 – 1785; demonstrated publicly at the Arsenal in February 1785 · Paris

The question: If water is made of two gases, then starting from water it must be possible to get them back. Can water be taken apart?
The apparatus: An iron gun barrel, laid through a furnace at a slight incline and heated to red heat. Water dripped in at the upper end from a weighed retort, flashing to steam as it entered. The far end led through a cooled tube — to condense anything that came over as liquid — and then under a bell jar, where the gas was collected and measured. The barrel weighed before and after.

Theory predicted

If water is elementary, it goes in as steam and comes out as steam, and the barrel is unchanged. If it is a compound of inflammable air with the same substance the metals take from the atmosphere, then the iron should seize that substance and gain weight, and the inflammable air should be released.

They measured

Inflammable air came off the far end continuously, and it burned. The iron in the barrel was converted to a black scale and had gained weight. The mass the barrel gained plus the mass of the gas collected accounted for the water that had gone in.

How sure could they be? The measurement that carries the argument is the sum, not any single term. Meusnier was a military engineer and the apparatus reflects it: the point of the condenser on the outlet was to make sure that undecomposed water was collected and subtracted rather than quietly written off as gas.

Why it mattered

Water is decomposed. Not altered, not spoiled — separated into two weighable products, one of which is a gas that was not there before and the other of which is the substance the last lesson took out of the air. Combined with the globe, this closes the loop: water can be made and water can be unmade, and both directions balance.

Why it takes both directions

Either half of this on its own leaves a hole. Together they do not. Synthesis alone — burning two gases and getting water — invites the reply that you have produced a new liquid resembling water, or that the water was somehow present in the gases all along and has merely been released. Cavendish himself took the second of those readings, and it is not a foolish one. Analysis alone — driving steam over hot iron and collecting a gas — invites the reply that you have not decomposed the water but reacted with it, producing something new. Which, incidentally, you have: the black scale in the barrel is a real product. Both, with the masses matching in both directions, leaves nowhere to stand. The same two substances, in the same proportions, going in and coming out. Whatever water is, it contains these two things and nothing else, and the quantity of each is fixed. This is the shape of a great many settled questions in science, and it is worth naming: take it apart, put it back together, and weigh everything both times. Lavoisier did it to air in 1777 and to water in 1785, and by 1789 he was able to publish a list of substances that had survived every attempt at decomposition anyone could devise — the first modern table of elements. Thirty-three entries. Twenty-three of them are still on the list.

Henry Cavendish

2010-10-19 09:36:05. Public domain

He was pathologically shy, spoke to almost nobody, communicated with his housekeeper by note, and was among the most exact experimenters who has ever lived. He weighed the Earth, discovered argon without knowing it, and made water out of two gases — and reported the whole thing in the language of a theory he never abandoned.

The water controversy

Who established the composition of water is one of the genuinely unresolved priority disputes in the history of chemistry, and it stayed poisonous for a century. The bare sequence: Cavendish had the experiments by about 1781, showed them to friends, and read his paper to the Royal Society on 15 January 1784. James Watt — the steam-engine Watt — wrote to Priestley in April 1783 proposing that water is composed of dephlogisticated air and phlogiston, from experiments Priestley had described to him. His paper was read to the Royal Society on 22 April 1784, three months after Cavendish’s, although the letter predates Cavendish’s reading. Charles Blagden, Cavendish’s assistant and later Secretary of the Royal Society, was in Paris in June 1783 and told Lavoisier what Cavendish had found. Lavoisier and Laplace burned the two gases together on 24 June 1783, with Blagden present, and reported water. Lavoisier’s published account did not make Blagden’s role especially prominent, and Blagden minded. All three camps, and their partisans for the next hundred years, believed the others had behaved badly. Watt’s supporters produced an entire book of correspondence to prove his priority. There is no clean verdict, and the honest summary is that Cavendish made the measurement, Watt got to a similar idea independently and slightly earlier in writing, and Lavoisier — who was told about it — supplied the only interpretation that survives. Which is the recurring lesson of this act. Making the observation, publishing it first, and understanding what it means are three different achievements, and in the eighteenth century they kept landing on different people.

Cavendish’s own reading of his result is the part that gets skipped, and it is the most interesting thing in the lesson. He did not say that water is a compound of two elements. He said, in effect, that water was there all along. Dephlogisticated air is water deprived of phlogiston; inflammable air is water united with phlogiston; put the two together and the phlogiston finds its place and the water is released. That accounts for every observation he made, including the exact mass balance, and it does so without conceding a thing. Lavoisier’s account is that the water is made — that two elementary substances have combined and that neither of them was water. And then, in 1787, he and Guyton de Morveau, Berthollet and Fourcroy published a new system of chemical names, and inflammable air became hydrogène: from the Greek hydro and genes, the water-former. The substance whose defining property is that it makes water. That is not a label. It is Lavoisier’s entire theory compressed into a word, and once chemists were using it every time they mentioned the gas they were asserting his conclusion whether they meant to or not. The naming reform did as much to end phlogiston as any experiment did — which is worth noticing, and worth being slightly uneasy about.

You might think

Water is H₂O, so it is two parts hydrogen to one part oxygen — twice as much hydrogen as oxygen.

Actually

Twice as many hydrogen atoms, and therefore twice the volume as a gas. By mass it is the other way round and not by a factor of two: water is 11.2% hydrogen and 88.8% oxygen, a ratio of about 1 to 8. An oxygen atom is very nearly sixteen times as heavy as a hydrogen atom, so two hydrogens against one oxygen is 2 against 16. The reason this trips people is that both statements are true of the same substance and they point in opposite directions, and eighteenth-century chemists had exactly the same problem — Cavendish measured volumes because volumes were what a eudiometer gave him, while Lavoisier wanted masses because masses were what a balance gave him, and the two communities took another twenty-five years to work out that the volume ratios were the simpler ones and Avogadro to explain why.

Problem

Cavendish’s ratio

Cavendish reported that 423 measures of inflammable air were consumed by 1000 measures of common air, both being used up together. Taking common air to be 20.95% oxygen by volume, in what ratio — by volume — do the two gases combine?

Inflammable air consumed: 423 measures
Common air consumed: 1000 measures
Oxygen in common air, by volume: 20.95%

By the experiments with the globe it appeared that when inflammable and common air are exploded in a proper proportion, almost all the inflammable air, and near one-fifth of the common air, lose their elasticity, and are condensed into dew.
Henry Cavendish"Experiments on Air", Philosophical Transactions, 1784. "Lose their elasticity" means they stop being gases. The sentence describes the synthesis of water and does not use the word.

1766Cavendish isolates inflammable air and measures its density. He suspects it may be phlogiston itself.
1768–70Lavoisier boils water in a sealed pelican for 101 days and shows the "earth" is dissolved glass.
1781Cavendish has the globe experiments: inflammable air burned in common air gives water.
April 1783Watt writes to Priestley proposing water is dephlogisticated air combined with phlogiston.
June 1783Blagden tells Lavoisier what Cavendish has found. Lavoisier and Laplace repeat it within days.
15 Jan 1784Cavendish’s "Experiments on Air" is read to the Royal Society. Watt’s paper follows in April.
Feb 1785Lavoisier and Meusnier decompose water over hot iron and recompose it, publicly, at the Arsenal.
1787Méthode de nomenclature chimique. Inflammable air becomes hydrogène, the water-former.
1789The Traité prints a list of 33 simple substances. Water is not on it. Neither is air.

Where this leaves us

Two of the four classical elements are gone inside eleven years, and both were dismantled the same way: take the thing apart, weigh the pieces, put it back together, weigh it again. What chemistry has now is a list. Not a philosophical scheme of four principles, but an empirical inventory of substances that have resisted every attempt at decomposition anyone can devise — with the explicit admission, written into the Traité, that anything on it may come off tomorrow if someone finds a way to break it. It also has a conserved quantity, a naming system, and a balance that closes on every reaction anyone can run. The gun barrel above turns 6.00 g of water into 0.671 g of gas and 5.33 g of oxygen in the iron, and those add to 6.00 g, and they will add to 6.00 g every time anyone repeats it anywhere. Which is precisely the thing to be suspicious about next. Every time. Copper carbonate dug out of a Siberian mine and copper carbonate precipitated on a bench in Madrid have identical composition, to the limit of anyone’s ability to measure. And where two different compounds of the same pair of elements exist, the second holds exactly twice the oxygen of the first. Not 1.9. Not 2.1. Nothing about a continuous, infinitely divisible substance explains a ratio of two.