You’re reading Age of Invention, my newsletter on the causes of the British Industrial Revolution and the history of innovation. This free edition went out to over 20,000 people. To support the newsletter you can become a patron by upgrading your subscription here:
The standard pre-history of the steam engine goes a little like this:
There were a few basic steam-using devices designed by the ancients, like Hero of Alexandria’s spinning aeolipile, which are often regarded as essentially toys.
Fast forward to the 1640s and Evangelista Torricelli, one of Galileo’s disciples, demonstrates that vacuums are possible and the atmosphere has a weight.
The city leader of Magdeburg, Otto von Guericke, c.1650 creates vacuums using a mechanical air pump, and is soon using atmospheric pressure to lift extraordinary weights. This sets off a spate of experimentation by the likes of Robert Boyle, Robert Hooke, Christiaan Huygens, and Denis Papin, to create vacuums under pistons.
As a result of the new science of vacuums, by the 1690s and 1700s the mysterious Thomas Savery and especially the Devon-based ironmonger Thomas Newcomen are able to develop the first commercially practical engines using atmospheric pressure. Steam engine development continued from there.
This is the narrative that had become set in the 1820s, if not earlier, and has been repeated with many of the same names and dates by book after book after book ever since. It’s a narrative that I have even repeated myself.
But, as I only recently discovered, atmospheric pressure and vacuums were actually being exploited long before Torricelli was even born, by people who believed that vacuums were impossible and had no concept of atmospheric pressure. Devices very much like Savery’s, which exploited both the pushing force of expanding hot steam and the sucking effect of condensing it with cold — what we now know to be caused by atmospheric pressure — were being developed far earlier.
I began to give a more accurate account of the development of the atmospheric engine in a detailed three-part series on why the steam engine wasn’t invented earlier (see parts I, II, III, which give more detail and the references). But I haven’t put it all together in one easily digestible place, and since writing I’ve continued to discover even more. So here’s a rough sketch summarising what really happened, based on everything I’ve found so far:
From as early as the third century BC, Philo of Byzantium described an experiment to raise water using both the expansion of air and water vapour through heat and their contraction from cooling. We don’t know if he applied this more widely, but Hero of Alexandria in the first century described applying exactly the same principles to a self-replenishing dripping fountain and to opening temple doors — far more intricate and significant than his famous rotating aeolipile, which used only the expansive force of steam.
By the fifteenth, sixteenth, and seventeenth centuries — and probably earlier — metallurgists and alchemists were using a stationary aeolipile for finer tasks like bending glass pipes or in metalwork. This was essentially just a retort-shaped metal vessel with a long, narrowing spout, which when directed at a flame would have a sort of blow-torch effect. There were various suggestions to point these at small turbines, including in Leonardo da Vinci’s notebooks, in a 1551 Ottoman manuscript by Taqi ad-Din, and in a 1629 work by Giovanni Branca. And the expansive force of heating air or water is applied in more direct ways by the French inventors Marin Bourgeois and David Rivault c.1605-7 to power guns and cannon, and by the Spanish military engineer Jerónimo de Ayanz y Beaumont c.1606 to drive up the water flooding mines.
By the end of the sixteenth century, if not much earlier, there was a knack to refilling the alchemist’s aeolipile that exploited atmospheric pressure, more or less copying Philo’s and Hero’s basic experiment: heat the vessel when empty, put the narrow spout in some water, and watch the water be sucked up and into it as the vessel cools. A version of this using a long-necked glass flask appears in print as early as 1558 thanks to Giovanni Battista della Porta, and by the 1580s is being used as a sort of demonstration of how the wind might work. (Hero’s full manuscript is meanwhile first printed in Latin in 1575.)
By 1598, the Dutch alchemist Cornelis Drebbel takes the wind microcosm further, turning it into a machine that can harness the perpetual motion of the universe. He notices that the water in the inverted flask, if left alone, will continually rise and fall from changes in the weather, and uses this to periodically reset spring-driven clockwork. In c.1606 Drebbel presents an intricate “perpetual motion” on this principle to King James I of England, which is kept at Eltham Palace and becomes famous all over Europe. The perpetual resetting mechanism indicates the time, day, months, and years, the phases of the moon, and has water rising and falling to either side of a glass ring, ostensibly in keeping with the ebb and flow of the tides.
Things then move more quickly, as people all over Europe try to decipher the workings of Drebbel’s perpetual motion and recognise in it the ancient inverted flask experiment. Drebbel himself applies the same principles to invent the self-regulating oven — perhaps the first thermostatic feedback control mechanism — which is used for artificially incubating eggs. And either he, or someone else, adds a scale to the inverted flask experiment, reinterpreting it as a device capable of measuring both temperature and atmospheric pressure. This becomes known in England as a weather-glass and in the Dutch Republic as a thunder-glass or Drebbelian instrument, while in Italy, by 1612, Santorio Santorio arrives at the same improvement, his version becoming known as the thermometer. (Though it’s not until the 1640s — possibly thanks to Duke Ferdinand II of Tuscany — that the complicating influence of atmospheric pressure is removed to make it a pure thermometer)
Drebbel’s colleague at the English court, the French engineer Salomon de Caus, in 1615 publishes even more devices using the same principles: a hint of how Drebbel was resetting his perpetual motion’s clockwork, a statue that would sound a pipe when the sun shone at a specific time (Drebbel reportedly constructed these too), and what is essentially a more elaborate version of Hero’s solar-powered dripper fountain, which pushes some water up during the heat of the day and replenishes itself by sucking up more water when it cools. In a 1644 edition of de Caus’s book, his brother Isaac also added a self-replenishing device on the same principles to trigger a water-powered barrel organ that would play a song when the sun shone on it at noon.
In 1622, the Dutch scientist Isaac Beeckman notes in his diary how the underlying principles of Drebbel’s machine could be used to raise water to a reservoir through the suction effect cooling air and water, before raising it even higher by heating it up again. Beeckman suggests using the raised water to drive a small waterwheel, and considers developing it into a self-replenishing water clock.
In the 1630s, with the inverted flask experiment in mind, the German or Danish gunmaker Kaspar Kalthoff experiments at Vauxhall near London with making a water-raising device on a much larger scale. At first he tries using the suction effect from condensing steam, but discovers that this effect alone cannot raise water any further than about 10 metres (34 feet). He then switches his attention to harnessing expansive force of steam, much like de Ayanz had in c.1606, before being forced to abandon his machines because of the English Civil War. Kalthoff tries again at Dordrecht in the Netherlands in 1649, but the machine is burned down by farmers before being completed. It is unclear whether Kalthoff’s 1640s experiments tried to combine steam’s expansion and condensation, or if he abandoned condensation completely. In 1663 his former employer the Marquess of Worcester describes an experiment using only the expansive force of steam.
It’s only now that we start to get step two in the traditional narrative, as the reasons for the 34-foot limit to suction start to get more attention. In 1630 Giovanni Battista Baliani writes to Galileo about his failure to get a copper siphon to hold water above that level, leaving an inexplicable gap above the water, which by 1638 Galileo believes to be a vacuum — something traditionally considered impossible. In 1643 Gasparo Berti tests this with water in a long lead pipe and by using a glass sealed to the top, discovers that there really is an observable gap above the water. In 1644 Evangelista Torricelli and Vincenzo Viviani repeat the experiment in miniature by using a much denser liquid, mercury. No matter the size of the space left above the liquid, it always falls to the same level. Torricelli reasons that instead of suction being caused by the impossibility of a vacuum, it is actually from the weight of the atmosphere pushing a column of liquid up until their weights balance out. (And by creating what was essentially a weather-glass with a vacuum instead of trapped air, invents the barometer.)
In 1649, William Petty — later famous for his works on economics and statistics — experiments with raising water from the suction effect of condensing steam. Like Kalthoff, he hits the same limit to suction of about 34 feet. But despite having been made aware of Torricelli’s experiment just a year previously, he seemingly does not yet believe that there can be a vacuum and is still trying to find a way around the limit. For reasons unknown to us, Petty does not pursue his experiments further.
This brings us pretty much up to the point where the traditional onus on Torricelli’s experiments may start to influence the trajectory of further development. Torricelli’s hypothesis on the weight of the atmosphere became more widely adopted, and the experiments of von Guericke, Boyle, Huygens, Hooke, Papin, and others had them explicitly in mind. Thomas Savery, for his widely-adopted 1698 engine using both the expansion and condensation of steam, explicitly described the effects of condensation in terms of vacuums and atmospheric pressure.
It already appears that believing in Torricelli’s hypothesis was entirely unnecessary to arriving at a Savery-like engine. Sucking up water by cooling air or steam had a long history, with many of the above even using alternating condensation and expansion in much the same way, from Hero’s dripping fountain to Beeckman’s waterwheel. There’s even evidence to suggest that Savery himself was originally inspired by the ancient inverted flask experiment. But as if all that was not enough, I’ve now discovered a design for a suction device that explicitly says so.
In 1653, Father Jean François, a mathematics and science teacher at various Jesuit colleges throughout France — speculated to be Descartes’s maths teacher — published La Science des Eaux, or the Science of Water, which contained a section on the art of raising water above its source, including a chapter on the use of heat. François then described the inverted flask experiment in great detail, explaining its effects in terms of the impossibility of a vacuum, before passing “from this pleasant invention to a profitable one”: using the exact same principles on a larger scale by having the heat of a chimney or kitchen stove create steam or expand air within a vessel, before cooling it to raise water from a well below. Here’s his diagram (G is the inverted flask):
François goes into a great deal of detail about how to get it to work, strongly suggesting that he had worked on at least a prototype. And — here’s the crucial bit — he then notes that it can also be explained using Torricelli’s hypothesis: “I gave the reasons for these effects, following the old way of reasoning about the void that nature flees and abhors. But those who are attached to new experiments maintain that … the water rises up the pipe AB by being pressed by another column of air, which is on the surface of the water surrounding this pipe.” It didn’t matter what one believed, however, as the device would still work: “This being the case, the effect of our invention remains even more assured, provided that the height of the pipe does not exceed about thirty feet.” From the perspective of inventing an operational atmospheric engine, François was clear: the main thing that Torricelli’s theory told inventors was simply not to bother trying to get suction to work above 34 feet — something that Kalthoff and Petty had apparently discovered for themselves anyway, even if they had no way to explain it.
The development of the atmospheric engine was thus significantly longer and more complicated than the traditional narrative suggests. Far from being an invention that appeared from out of the blue, unlocked by the latest scientific advancements, it started to take shape from decades and centuries of experiments and marginal improvements from a whole host of inventors, active in many different countries. It’s a pattern that I’ve seen again and again and again: if an invention appears to be from out of the blue, chances are that you just haven’t seen the full story. Progress does not come in leaps. It is the product of dozens or even hundreds of accumulated, marginal steps.
This is, unfortunately, a pretty common phenomenon in the history of technological innovation. A few people get the credit for inventing a new technology or coming up with the original idea, but it was actually due to the hard work of hundreds of people.
The famous inventors remembered by history deserve their fame, but it is too bad that all the other people's hard work has vanished into the mists of history.
Fascinating stuff as always. Why do you suppose we didn't get more useful machines earlier? Even something as useful as the Savery pump? Let alone the Newcomen engine?
It would be fantastic to see some estimates of torque and power output from each of these early devices, even very rough ones