The method worked for fifty-one years before anyone knew why

The letter is dated Paris, 30 January 1810, and it is signed Montalivet, Minister of the Interior. It tells Nicolas Appert that the Board of Arts and Manufactures has examined his process and found it good, and that twelve thousand francs are his. Then it names the price.

Appert is to publish. He is to print the method in full, and send the ministry two hundred copies. The transmission of these copies, the minister writes, being the only condition I impose on you for the payment of the 12,000 francs.¹

A government bought a working food technology and paid for it in books, on the explicit condition that nobody would ever be able to patent it. Napoleon does not appear in the document. He is in almost every retelling.

Forty-five years in the cellars

Appert, born around 1749 and dead in 1841, gets described as a confectioner or a brewer depending on who is telling it. He settles it himself in the book's opening pages, counting forty-five years spent in the pantries, the breweries, store-houses, and cellars of Champagne, as well as in the shops, manufactories, and warehouses of confectioners.¹ Both, and a good deal else.

The method he had arrived at by then is four steps long. Enclose the substance in a bottle. Cork it with the greatest care. Submit it to boiling water in a bain-marie for a greater or less length of time, according to their nature. Remove and cool.¹

Read step three again, because it is the whole of process control for the next century. No temperature. No duration. No organism, because none was known. According to their nature, and the judgment of a man who had spent forty-five years watching things spoil.

The reason he gave

The book contains an explanation, and it is the one everyone held at the time: air being a most destructive agent.¹ Spoilage was understood as a chemical assault by oxygen, and the corking and the heat were understood as ways of getting the air out and keeping it out.

This has the pleasing property of being wrong in a way that produces correct behaviour. Exclude the air and you also exclude what arrives with it. Heat the sealed bottle and you kill what is already inside. The procedure is sound. The account of the procedure is not.

The scientific establishment agreed with him, which is the part I find hard to let go of. Joseph Louis Gay-Lussac signed the certificate approving Appert's book in April 1810. That December he read a memoir to the Institute explaining preservation in terms of oxygen.² The same chemist certified the method in the spring and misexplained it before the year was out.

The Englishman who did not invent the tin can

Peter Durand took out a British patent in August 1810, and he is routinely credited with inventing the tin can. His own patent says otherwise on its face. The invention was, in the wording of the enrolled specification, communicated to him by a person residing abroad.³ The specification lists glass first and tin third, as one option among several vessels.³

Who the person abroad was is a real question with a real answer, or at least a good candidate. Work by the canning historian Adrian Cowell points at Philippe de Girard, and cites Charles Blagden's diary at the Royal Society for January 1811 recording that the patent is to be taken out in the name of Durand. That research appeared in a trade magazine rather than a journal,⁴ so treat it as a strong lead rather than a settled fact. What is settled is that Durand's own document says he was not the inventor.

The industry followed anyway. Bryan Donkin and his brother-in-law John Hall bought Durand's patent out, had a canning factory running by 1812, and were supplying the British military by 1813.⁵ Three years from patent to army rations, which is roughly the speed at which a technology moves when nobody is required to understand it.

What happens when the theory is missing

It goes wrong in exactly the way you would expect. If your rule is heat it for a greater or less length of time according to its nature, and you have no way to say how long is enough, then a supplier who cuts the time has broken no specification, because there is no specification.

In February 1852 the House of Commons spent an evening on Stephan Goldner's contract to supply preserved meat to the Royal Navy. Of 6,378 canisters opened, 5,468 were condemned, described in the debate as a perfect mass of putridity.⁶

The number is shocking and the conclusion usually drawn from it is wrong. Later in the same debate a member reads out a different figure: of 35,000 canisters sent out with Captain Austin, eighteen were bad.⁶ One supplier had been using oversized cans that never reached temperature at the centre, and canning as a technology was working fine. Without a theory, there was no way to tell those two facts apart except by opening the tins.

The Franklin story, which is not what you were told

The other thing everyone knows about early cans is that they killed Sir John Franklin's expedition through lead solder. It is a good story and the evidence has been running against it for a decade.

A 2018 study looked at where the lead actually sits in the bone microstructure of the crew, and concluded that the skeletal microstructural Pb distribution data do not support the conclusion that Pb played a pivotal role.¹⁰ The lead is largely in bone laid down before the ships ever sailed, which points at a lifetime of exposure in industrial Britain rather than at the provisions. An earlier reassessment, working from only seven skeletons, had already raised the ships' own lead water pipes as a rival source.¹¹ More recent work puts the weight on thiamine and vitamin C loss instead.¹²

None of that exonerates the tins, and the case is genuinely unsettled. It does mean the confident version of the story is running well ahead of the bone chemistry.

Fifty-one years

Louis Pasteur's work on fermentation and on spontaneous generation ran through the first half of the 1860s.⁷ That is the moment the account of why a sealed heated bottle keeps stops being about oxygen and starts being about organisms, and it lands fifty-one years after Montalivet's letter.

Even that is only the general principle. The first real bacteriology of canning, tracing specific spoilage in canned goods to specific organisms, came from Samuel Prescott and William Underwood at MIT, published in 1897.⁸ Eighty-seven years.

And the number a modern cannery actually runs on, the thermal death time of Clostridium botulinum spores, waited on data published by Esty and Meyer in 1922.⁹ The vocabulary that now surrounds it, the botulinum cook and the twelve-log reduction, came later still, built on their measurements. A hundred and twelve years after the method was bought and printed, somebody finally wrote down how long is long enough.

The order of things

We are used to a story where science arrives first and the kitchen catches up. Preservation ran the other way round for over a century. The technique was correct, industrialised, feeding navies and killing people through bad execution, and entirely unexplained, and the leading chemist of the age had put his signature to it and then published the wrong reason.

There is a version of this in every kitchen. Somebody's grandmother knew exactly how long to boil the jars and could not have told you what a spore was. The knowledge was real, transmissible and correct, and the explanation attached to it was decoration.

What strikes me about Montalivet's letter is that the state seems to have understood this perfectly well. It did not ask Appert for a theory. It asked him for two hundred copies of the instructions.