14

Knowledge is Power

I should not have neer so high a value as I now cherish for Physiology, if I thought it could onely teach a Man to discourse of Nature, but not at all to master Her; and served onely, with pleasing Speculations, to entertain his Understanding without at all increasing his Power.

– Robert Boyle, Some Considerations (1663)¹

§ 1

What is the relationship between the Scientific Revolution and the Industrial Revolution, between the mathematicians’ revolution and the mechanical revolution? The claim with which this book opened, that the Scientific Revolution is the most important event since the Neolithic Revolution, depends on our answer to this question; for if the Scientific Revolution was merely an event in the world of ideas, its importance is relatively limited, while if it opened the way to a new control over nature, then the Industrial Revolution can be seen as merely an extension of the Scientific Revolution, the extension of the procedures, language and culture of the new science to a wider social stratum of technicians and engineers. There is no doubt that Bacon and his followers aspired to transform the world through their new science. In the mid-eighteenth century Birch’s History of the Royal Society carried as its epigraph a quotation from Bacon: ‘Natural philosophy as I understand it does not slip away into sublime and subtle speculations, but is applied effectively to relieve the inconveniences of the human condition.’² The motto of the French Academy of Sciences, founded in 1666, was naturae investigandae et perficiendis artibus (‘the investigation of nature and the improvement of technology’) changed in 1699 to the snappier invenit et perficit (‘progress through discovery’).

It is easy now to find some of the first scientists’ expressions of enthusiasm naïve: thus Ambroise Sarrotti, who had come to England to accompany his father, Paolo, the Venetian ambassador (1675–81), returned home to organize a scientific society which conducted experiments with vacuums.ii At the end of the first year he proudly announced to his colleagues: ‘If, since the beginning of the world, all mankind united together could have done every year, as much as you alone have done this year last past, they would live now as happy in this world, as in a terrestrial paradise.’³ This, despite the fact that they had discovered nothing of any use whatsoever. It is not surprising then that not everyone was convinced of the practical utility of the new science.⁴ Jonathan Swift wrote Book 3 of Gulliver’s Travels (1726) with the sole aim of denying it. Yet his attack suggests that he hesitated as he tried to define the nature of the enemy. Laputa is an airborne island, ruled by scientists who are so obsessed with matters mathematical that they are unable to pay any attention to the world around them: they rely on the services of flappers, who strike their ears and mouths with inflated bladders in order to remind them when to listen and when to speak. But down below in Balnibarbi, the colony over which they rule, an academy has been formed, in imitation of Laputa, where scientists pursue practical objectives in the most impractical ways, making sunbeams from cucumbers, and thread from spider’s webs. The governor general, who alone disapproves of the new inventions, tells Gulliver:

That he had a very convenient Mill within Half a Mile of his House, turned by a Current from a large River, and sufficient for his own Family, as well as a great Number of his Tenants. That about seven Years ago, a Club of those Projectors came to him with Proposals to destroy this Mill, and build another on the Side of that Mountain, on the long Ridge whereof a long Canal must be cut, for a Repository of Water, to be conveyed up by Pipes and Engines to supply the Mill: Because the Wind and Air upon a Height agitated the Water, and thereby made it fitter for Motion: And because the Water, descending down a Declivity, would turn the Mill with half the Current of a River whose Course is more upon a Level. He said, that being then not very well with the Court, and pressed by many of his Friends, he complyed with the Proposal; and after employing an Hundred Men for two Years, the work miscarryed, the Projectors went off, laying the Blame intirely upon him, railing at him ever since, and putting others upon the same Experiment, with equal Assurance of Success, as well as equal Disappointment.⁵

He does not say what sort of ‘engines’ were employed, but Swift surely had in mind the first steam engines, which were generally used for raising water. So on Swift’s account the new science is both totally impractical and at the same time obsessed with practicality. This is not an impossible combination – indeed, it seems to describe Sarrotti rather well – but it is certainly a puzzling one.

Historians of science have not advanced our understanding of the relationship between the new science and technological progress much beyond Swift. Naturally, Marxist historians have wanted to argue that the new science was the result of new social relations. As the Russian Boris Hessen (who was executed in 1936, an early victim of Stalin’s Great Purge) put it in 1931, ‘Step by step, science flourished along with the bourgeoisie. In order to develop its industry, the bourgeoisie required a science that would investigate the properties of material bodies and the manifestations of the forces of nature.’ But Marxists were not alone in assuming that the new science was motivated by its possible practical applications: Robert K. Merton in his classic study of 1938, Science, Technology and Society in Seventeenth-century England, in which he emphasized the role of Puritanism in encouraging useful knowledge, followed Hessen in arguing that seventeenth-century science was indeed intended, through and through, to have practical applications, despite his own rejection of Hessen’s Marxist assumptions.⁶

A series of studies, however (those of Alfred Rupert Hall being particularly influential), have claimed to show that, whatever the intentions of scientists may have been, in practice, the new science had virtually no influence on technological progress. A key case-study was provided by Watt’s steam engine (1765). Watt developed his new engine in Glasgow, where Joseph Black had proposed the concept of latent heat (c.1750). Black later collaborated with Watt and invested in his new engine. Was Watt familiar with the concept of latent heat when he devised his new engine, and did the new theory inform his new technology? He insisted that he was not, and historians came (almost reluctantly) to take him at his word.⁷ Lawrence Joseph Henderson is frequently quoted as saying (apparently in 1917), ‘Science owes more to the steam engine than the steam engine owes to science.’⁸ After all, Sadi Carnot finally produced a satisfactory theory of the steam engine only in 1824, more than a hundred years after Newcomen’s first engine, and sixty years after Watt’s. Hall thought it was ‘not quite’ but very nearly true to say that ‘engineering owed nothing to science’ until very late in the eighteenth century. Thomas Kuhn thought that science and technology were antithetical to each other, at least until the 1870s.⁹

One might think that the historians of technology would have wanted to question this disjuncture between theory and practice – but at first they were the same people as the historians of science.¹⁰ The major attack on the established orthodoxy has come only very recently, and from an unexpected quarter: the new economic historians of the Industrial Revolution, who emphasize the importance of skills and technical innovation, of what they call ‘the knowledge economy’.¹¹

On this question the new economic historians are (as will become apparent) in the right. But those who argue that science played a key role in the Industrial Revolution need to have an answer to a simple and by now classic question: What role did science play in the invention of the steam engine? Before tackling this problem, however, we need to unpack the apparently straightforward notion of practical knowledge. The key issue here is one of timescale: How long should one wait before dismissing a theoretical achievement or a technological advance as having little or no practical relevance? Does, as Hall assumed, the new science have to be contemporary with the technology that derives from it?¹²

Take ballistics: Galileo initially hoped that his discovery of the law (as we say) of fall, and with it of the parabolic path of projectiles, would revolutionize gunnery. When his disciple Torricelli entered into practical tests to see if Galileo’s theory described how cannon balls actually fly, he discovered that it didn’t: he insisted that the theory remained sound, even though it could not be applied to fast-moving projectiles because the effects of air resistance were not adequately understood (it could, it turned out, be applied to mortar shells fired short distances at low speeds).¹³ Ballistics was finally revolutionized between 1742 and 1753 by Robins and Euler, with the discovery of the sound barrier and an understanding of the effects of rotation in flight (deliberately induced, of course, by rifling; but Torricelli’s cannon balls were tumbling as they flew), and, as a result, the production of equations for reliably calculating trajectories. Galileo’s physics was intended to be practical but turned out to be of little practical use in its most obvious field of application. Nevertheless, his idealized parabolic trajectory in a vacuum was an essential precondition for the much more sophisticated analysis by Robins and Euler of actual trajectories. Galileo’s theory was practical; it just took a full century to pay off. For the young Napoleon, who was exceptionally good at mathematics, the problems which had defeated the great Torricelli were by the 1780s mere school exercises – the school, of course, being the École Militaire.¹⁴

Or take the challenge that preoccupied Galileo through a large part of his working life: that of establishing longitude at sea. Degrees north and south (latitude) are easy to calculate, providing one knows the date, from the height of the sun at midday; degrees east and west (longitude) are much harder to establish, since there is no obvious reference point that one can use. Galileo theorized that one could use eclipses of the moons of Jupiter (which he had discovered in 1610) as a sort of universal clock. With reliable tables predicting future eclipses, one ought to be able to tell the exact time wherever one was in the world; and, if one knew the local time (time elapsed since midday, for example) one could then compare local time with the time at the place for which the tables were calculated, and then easily calculate degrees west or east of the reference point. The theory was fine. Calculating the movement of the moons was less straightforward, but Galileo and his associates made strenuous efforts, and Galileo even constructed a little mechanical model, the Jovilabe, which enabled him to work out the location of the moons without doing complex calculations; he would have done better, of course, if he had known that it was also necessary to make an allowance for the speed of light, since the time at which an eclipse seems to take place varies depending on how far away Jupiter is from the Earth.

The central problem, however, was simple: How could one look through a powerful telescope at a tiny, distant object and make reliable observations while on a ship bouncing about on the waves? Galileo devised a pair of powerful binoculars which were clamped to the head, since it was hard to hold a telescope still enough on a moving boat, in what amounted to a gimballed chair in which one could sit to observe (compasses were already mounted in gimbals). Solving the problem of longitude was a universally recognized challenge: indeed, governments had promised enormous rewards for anyone who could succeed. Galileo hoped to establish his immortal fame by this discovery more than any other: he tried to claim the reward offered by the Spanish government, but failed (his pupil Castelli went to sea, but became hopelessly seasick); and in his last years he entered into clandestine negotiations with the Dutch government in the hope that they would take up his ideas and make them work in practice, but he failed.¹⁵

Or did he? By 1679 the Cassini family (who had emigrated from Italy to France, where they became famous as astronomers and cartographers) were using the moons of Jupiter to calculate longitude, if not at sea then at least on dry land. Such measurements guided their recalculation of the size of France (France turned out to be a full 20 per cent smaller than previously believed), and their calculation of the shape of the globe (which turned out to be good news for Newtonians, and a devastating blow for Cartesians). Galileo was right: the moons of Jupiter were a promising way of measuring longitude. It just took sixty years to make his proposal work in practice, and then it worked only if one had one’s feet on solid ground.¹⁶

There were alternative projects for calculating longitude. For a long time it was hoped that measuring compass deviation and dip would enable sailors to establish their coordinates. Despite generations of effort, this proved illusory, because dip and deviation change unpredictably over time.¹⁷ The simplest scheme in the end turned out to be the best: all one needed to do was to take a reliable timepiece on the journey and use it to measure the difference between local time (local noon, for example) and the time at the reference point (the Greenwich meridian, for example).

Galileo believed that he had shown that pendulums tell perfect time, and devised a pendulum clock (though he did not build it; by the time he turned his attention to the question he was blind, and his son, who tried to help him, lacked the necessary manual skills). Huygens, without knowing of Galileo’s work, went on to build the first pendulum clock (1656) and to refine the law of the pendulum (1673). Meanwhile, Robert Hooke, Huygens and Jean de Hautefeuille devised between 1658 and 1674 ways of controlling a balance wheel (which had been invented in the fourteenth century, and was more stable than a pendulum for a travelling timepiece) with a spring so that small clocks and watches would tell time reliably. Still, the task of making a seagoing clock or watch was far from solved: such a timepiece had to remain accurate despite changes in temperature and humidity, and despite the movement of the waves. The problem was not solved until John Harrison produced the first reliable marine chronometer in 1735.¹⁸ Were the discoveries of Galileo, Hooke and Huygens irrelevant? Certainly not, but they were insufficient. It took more than a century to solve the problem, but in the course of that century steady progress was made towards a solution.

Clockwork, of course, was not a seventeenth-century innovation. As we have seen, the first mechanical clocks date to the late thirteenth century, and their geared machinery derived from water-wheels and windmills. Water-wheels were known to the ancient Greeks and Romans but were far from common; thanks to an early medieval proto-Industrial Revolution, they quickly became widespread around the end of the first millennium CE. The Domesday Book records more than six thousand mills driven by water-wheels in England in 1086. Vertical windmills followed quickly: the first securely dated one was in Weedley, Yorkshire, in 1185. Given that the greatest concentration of medieval watermills was in England, it is surely not a coincidence that it was in England that we find both the first recorded vertical windmill and the first recorded clock. Steam did not overtake water and wind as a source of power until after 1830;¹⁹ in Swift’s Laputa, as in eighteenth-century England, steam power did not replace water power but supplemented it.

Halley’s isogonic map of magnetic variation, published in 1701. Each line on the map is like a contour line but, instead of marking a uniform measurement of height, it marks a uniform measurement of magnetic variation. Halley had conducted two expeditions to make the measurements on which the map was based, and the hope was that this would open the way to using magnetic variation to measure longitude.

Nevertheless, it has been claimed that the innovations of Galileo, Hooke and Huygens made possible the geared machinery of the Industrial Revolution.²⁰ Before the mid-seventeenth century, gears were laid out and cut by hand; Hooke designed the first machine to produce identical gears, thus making possible the mass production of machinery. Inevitably, eighteenth- and nineteenth-century engineers turned to clockmakers to build their machines (Richard Arkwright, for example, worked with the clockmaker John Kay to produce the spinning frame in 1769), and the quality of what they could accomplish had been greatly enhanced as a result of the revolution in clockmaking that had taken place in the years after 1656.²¹

Mechanical clocks give us a valuable chance to think comparatively, because we can see how other cultures responded when they were introduced to them by European travellers in the sixteenth century. The Japanese were soon manufacturing their own clocks (just as they were quick to manufacture their own guns); while the Chinese showed little interest in using clocks for telling time, nor in making their own, despite the fact that Su Song had produced a sophisticated water-driven clock for astronomical purposes in the eleventh century. As far as they were concerned, clocks were merely delightful but useless luxury objects – rather like musical boxes. (The Chinese were similarly slow to adopt the technology of the military revolution, despite the fact that gunpowder originated in China.) There was thus nothing automatic about the widespread adoption of the clock which took place in medieval Europe.²²

Yet clocks spread rapidly in the fourteenth and fifteenth centuries: first, because Europeans were already mechanically minded (all those water-wheels and windmills); second, because their circular geared movements reflected in miniature the movements of the Ptolemaic heavens (early clocks often measured astronomical time – the phases of the moon, the signs of the zodiac – as well as diurnal time); and third, because clocks provided an impersonal mechanism for the coordination of community activities (the saying of the offices in monasteries and cathedrals, the opening and closing of markets in towns and cities). Egalitarian communities (cities, monasteries and cathedral chapters all chose their leaders through elections) are governed by the clock, while despotisms are not; clocks were given prominent, public places in monasteries, cathedrals and town halls, but they were slower to establish themselves in royal palaces. (Even now, my university campus, built in the 1960s, is dominated by a clock tower which is there not to tell the time but to convey the impression that ours is a disciplined, egalitarian community.) These factors – cultural, technological, conceptual, political – were absent in China, and hence the Chinese admired clockwork but had no use for it.

Clockwork obviously fostered the notion that the universe could be understood as a complex mechanism, and Copernicans were committed to the view that the same physical principles were at work in the heavens and on Earth. Thus Kepler could write in 1605, inspired by his reading of Gilbert on magnetism:

My aim is this, to show that the celestial machine is not like a divine creature, but like a clock (he who believes the clock to be animate assigns the glory of the artificer to the work), insofar as nearly all the diversity of motions are caused by a simple, magnetic and corporeal force, just as all the motions of a clock are caused by a most simple weight.iiii I will also show how this physical account is to be brought under mathematics and geometry.²³

But medieval and Renaissance clocks were so imperfect that not only did the weight which drove them have to be raised every day, the time also had to be corrected, and it was only the improved clocks of Huygens which made it possible to think of the universe as a perfect, clock-like mechanism which required no tending by the divine clockmaker. We find the new post-Huygens imagery grafted on to Descartes’ imagery of ‘automata’ as early as 1662, six years after the first pendulum clock, in this passage by Simon Patrick, an advocate of the new science:

Then certainly it must be the Office of Philosophy to find out the process of this Divine Art in the great automaton of the world, by observing how one part moves another, and how these motions are varied by the severall magnitudes, figures, positions of each part, from the first springs …iiiiii²⁴

Clockwork, by providing a fruitful metaphor, encouraged the Scientific Revolution, and, by fostering the development of sophisticated, geared machinery, it facilitated the Industrial Revolution, but it was not itself the product of either of those revolutions, nor was it a necessary precondition for either one, for there were other types of geared machinery.

There is another important example of the delayed pay-off in technological progress. Hydraulic engineering was a major concern for the first engineers, such as Leonardo, and consequently an immediate one for Galileo and his disciples. Galileo advised on drainage projects; his pupil Castelli advised the papacy on the management of rivers and published a major treatise on the subject (Della misura delle acque correnti, 1628); his pupil Torricelli made a major theoretical breakthrough when he formulated what is now known as Torricelli’s law (1643), which enables one to establish the velocity of flow given a particular head of water (or the head given a known velocity of flow), and he also did practical work on the flow of the River Chiana, a tributary of the Arno; his pupil Famiano Michelini, who was also his successor as philosopher to the Grand Duke, also published on hydraulics (Trattato della direzioni de’ fiumi, 1664).²⁵

Yet a hundred years went by before John Smeaton in England, relying on Torricelli’s work, set out to perform a systematic programme of experiments with model water-wheels in order to establish which designs were the most efficient (comparing undershot and overshot) and how each design could be made to work best: How big should the wheel be, and how fast should it revolve for maximum efficiency? How deep into the water should the paddles of an undershot wheel go? Smeaton discovered, to his surprise, that overshot wheels (where the water enters the wheel at the top) were twice as efficient as undershot wheels (where the water flows along the bottom of the wheel): theory had led him to expect them to perform equally (though Desaguliers had rightly suspected that, in practice, overshot wheels were superior),²⁶ and he had difficulty explaining why they performed so differently. Smeaton thus developed a series of practical rules of thumb to guide water-wheel construction, and was widely influential in instigating a shift from undershot wheels to overshot or, where this was not practical, to breast wheels (where the water strikes the wheel halfway up). It is at this point – and at only this point – that we can say that the work of Galileo and his pupils on water flow had finally paid off by facilitating a markedly improved practical technology.²⁷

John Smeaton’s model water-wheel in ‘undershot’ configuration: the wheel is two feet in diameter (from An Experimental Enquiry, 1760).

The case of water-wheels is a particularly interesting one because the technology had developed extremely slowly for almost a thousand years. Millwrights had learnt by trial and error what worked and what did not, but rapid advance required systematic experimentation, and this only occurred after the experimental method had been elevated to a new intellectual status. Smeaton himself had trained in the law and then apprenticed as a machine-maker before becoming an engineer (he was the first to call himself a ‘civil engineer’ – as opposed to a military engineer – and he went on to found a society of civil engineers)²⁸ and a Fellow of the Royal Society. He combined practical and theoretical knowledge, as Hooke had in watchmaking. And, of course, he was responding to an economic situation where the demand for power was growing rapidly. He went on to build steam engines, harbours, bridges and canals (including the Calder Navigation, a set of cuts and locks which made and still make the River Calder navigable).

What was the obstacle to performing Smeaton’s experiments in the 1680s or even the 1580s?iviv Smeaton’s work depended on two intellectual preconditions being satisfied. First, it was well known that working with scale models could often be misleading, because full-size machines often performed quite differently. The conceptual apparatus for thinking about this problem had been provided by Galileo in his Two New Sciences, and Smeaton addressed one aspect of it, the fact that friction tends to be greater in scale models than in full-size machines, by cleverly measuring the amount of friction generated in his models and then compensating for it. Second, Smeaton’s work depended on the systematic application of Torricelli’s law. We might want to add a third precondition: in calculating the efficiency of a water-wheel by comparing the output of the wheel with the input of the stream, Smeaton was assuming a Newtonian law of the conservation of energy. In that sense his work was post-Newtonian. But he could have compared the output from different types of water-wheel without having an absolute measure of efficiency. Moreover, in defining force Smeaton steered clear of the conflict between the followers of Newton and the followers of Leibniz over the definition of ‘force’ (a conflict now resolved by distinguishing momentum from kinetic energy): this conflict did not have to be resolved for his work to succeed.

It would thus seem clear that it would have been impossible to carry out Smeaton’s experiments in the 1580s but perfectly possible in the 1650s, and straightforward once the arguments of Newton’s Principia (1687) began to be widely understood. Nor was there anything new about working with models: Desaguliers was building model steam engines by the 1720s, and he was surely not the first. Yet it was not until the middle of the eighteenth century that Smeaton and Watt both used models to work out how to transform the efficiency of powered machinery. Those who think that modern science derives from the empirical experimental enquiries of craftsmen and artisans need to account for the extraordinarily slow evolution of water-wheel technology prior to the introduction of Smeaton’s scientific method. To employ the experimental method systematically and self-consciously, as Smeaton and Watt did, one needed both a certain amount of sound theory and a confidence that experimentation, though it might be laborious, presented an excellent prospect of making major progress. The theory was not new in the 1750s, but the confidence was. The source of that confidence was a sustained programme of advertising the new science through public lectures and books conducted by the disciples of Newton, above all by Desaguliers.²⁹

In the end, early modern science defeated two of the most difficult practical problems it had set itself: the calculation of the path of projectiles under real-life conditions, and the measurement of longitude. If seventeenth-century scientists did not see these problems solved, nevertheless they prepared the ground for their eighteenth-century successors, who did. In addition, in the mid-century, Smeaton and Watt transformed the efficiency with which water and steam power were harnessed to drive machinery; in the short term, Smeaton’s achievement was the more important; in the long term, Watt’s. In 1726, when these practical problems remained as yet unsolved, Swift’s case against the utility of science seemed sound; it would have been much harder to make the case against in 1780, or even in 1750. Strangely, historians have remained stuck in Swift’s world, and when they read texts such as Smeaton’s they read them naïvely, as if they simply reflect a programme of tinkering around with models, and as if all the terminology used is commonsensical; they remain oblivious to the fact that it was the new science that discovered the relationship between the head of water and the speed of the stream.

§ 2

The first great, practical achievement of the new science was Newcomen’s steam engine of 1712 – the very engine Swift was presumably mocking when he complained about mills being built where there were no rivers. It is important to put Newcomen’s achievement into perspective. By 1800 only 2,200 steam engines had been built in Britain, some two thirds of which were Newcomen engines, and a quarter Boulton and Watt engines.³⁰ Between 1760 and 1800 almost twice as much new water power (much of it the result of Smeaton’s work) as steam power became available.³¹ The great age of steam still lay ahead: when Mary Shelley published Frankenstein in 1818 her vision of the horrifying power of the new science scarcely included steam (a single reference to ‘the wonderful effects of steam’ was added, probably by Percy Shelley, as the book went to press), although Blake was already writing about ‘dark satanic mills’ in 1804 (he probably had in mind the Albion Flour Mills, the first large factory in London, powered by a Boulton and Watt steam engine and built in 1786).³² In 1807 Fulton’s steamboat began a regular passenger service between New York City and Albany, the state capital; in 1819 the SS Savannah, a ship that combined sail and steam, crossed the Atlantic; Stephenson’s Rocket rattled along the rails in 1829. By 1836 it was possible to describe steam as marking ‘a new era in the history of the world’. It had multiplied the powers of mankind ‘beyond calculation’.³³

In 1712 the Industrial Revolution and the age of steam lay far in the future; by 1836 they were a reality. They had been summoned into existence by a new culture of technical expertise, by men like Watt and Smeaton and by England’s high wages (for many of the new inventions were profitable only in a high-wage economy).³⁴ The steam engine did not make the Industrial Revolution inevitable, but it did make it possible. There had been high-wage economies before (after the Black Death, for example), but no Industrial Revolution. It is true that many of the new inventions that were central to the Industrial Revolution – Arkwright’s spinning frame, for example – owed nothing to science; but without Smeaton’s improved water-wheels and Boulton’s and Watt’s improved steam engines, the factories in which they were made could never have been powered.

In order to understand steam engines it may be helpful to think about methods of making coffee. Some people make it by dripping water through a filter containing coffee grounds: they are relying on gravity. Others use an espresso pot, which uses steam to drive water up through the grounds: the pot is a pressurized steam system, which is why it requires a safety valve. And some use the vacuum method, where water is driven up into a higher container by steam (at low pressure, for it has only to overcome the weight of the water), but then, when the heat is removed and the steam condenses, it creates a vacuum, sucking the water back down through the grounds. The vacuum method relies on atmospheric pressure.

The steam engine was the product of seventeenth-century science, which had experimented with vacuums, and with air and steam pressure.³⁵ A simple example of air pressure is the air gun, called in the seventeenth century ‘the wind-gun’. Mersenne described one in 1644, which is also the first year in which one is referred to in English; Boyle published a design for one in 1682.³⁶ It worked by compressing air into a container with a bellows and using the compressed air to power a dart or pellet. Steam in a confined space could also be used to create pressure. This principle was employed by della Porta in 1606, and in 1625 Salomon de Caus devised a steam fountain. It worked just like a stove-top espresso machine: the pressure of steam in a chamber from which there was only one exit drove the water in the container upwards and out. Boyle’s law provided a theoretical account of how pressure could be used to produce powerful force, if one could only find a way of harnessing that force for a useful purpose.

But there was an alternative to constructing some sort of high-pressure mechanism. That alternative – a low-pressure mechanism – derives from von Guericke’s work with his air pump. Von Guericke had shown that, if one pumped the air out of a cylinder, the pressure of the atmosphere would drive a piston down in the cylinder and the force would be such that even a team of strong men would be unable to resist it.³⁷ In 1680 Huygens came up with an alternative way of harnessing atmospheric pressure. He used an explosion to drive the air out of a cylinder through a valve; then, when the hot gases cooled, a piston was sucked downwards, raising a weight.

This idea was taken up by Denis Papin, a medical doctor who had started his scientific career as an assistant to Huygens, performing air-pump experiments. He had then moved to England: Papin was a Protestant and life was becoming increasingly uncomfortable for Protestants in France. Here, he worked as an assistant to Boyle; on Boyle’s own testimony, Papin devised many of the experiments published in Boyle’s A Continuation of New Experiments (Latin, 1680; English, 1682) and performed all of them. Indeed, the book was not really by Boyle at all, since it was written by Papin.³⁸ Papin was elected a Fellow of the Royal Society in 1680 (his social status was quite different from that of a mere technical assistant), but his financial position was precarious (he was exempted from paying fees); from 1681 to 1684 he was employed in Venice and, although he returned to England, he left again in 1687, first becoming a professor of mathematics in Marburg (where he fell out with his fellow academics, who saw no need for a mathematics professor, and his co-religionists, who excommunicated him), and then from 1695 working as an engineer advising the landgrave (or count) of Hesse in Kassel. There he successfully tested a primitive submarine in the River Fulda.³⁹

Giovanni Battista della Porta’s steam pressure pump, from Tre libri de’ spiritali (1606).

De Caus’s steam-powered fountain, from La Raison des forces mouvantes (1615).

Papin advanced Huygens’ idea a further step. He constructed a cylinder containing a small amount of water, which he heated over a flame. The water turned to steam, drove out the air and forced the piston to the top of the cylinder, where a spring engaged a bolt. The heat was then removed, the steam condensed and the piston charged; as soon as the bolt was pulled it would be driven down the length of the cylinder by the pressure of the air. This was, in effect a wind gun driven by atmospheric pressure and where the piston replaced the bullet. Papin went on to imagine a series of such pistons turning gears to drive a boat and so saving on the cost of oarsmen (galleys were still in widespread use, particularly in the Mediterranean and on rivers), and he thought an engine of this sort could be used to pump water out of a mine if there were no river to drive a water-wheel nearby.⁴⁰ Unfortunately, he had no mechanism for rapidly charging and discharging the cylinders, or (at this point) for getting them to discharge in an orderly fashion.

During these years he engaged in a number of different steam-engine experiments, the high point of which was the construction of a steam-powered carriage which ran around the floor of his parlour.⁴¹ He even looked forward to the time when steam-driven armoured cars would travel faster than the cavalry. His enemies, mocking him, put it about that he was working on a flying machine, and indeed he admitted that the thought had crossed his mind.⁴² He designed, as his personal contribution to the war against Louis XIV (who had driven Protestants, including Papin, out of France), a mortar to throw grenades 90 yards at the rate of two hundred an hour (or even five hundred, he later claimed). The design was straightforward: by pulling on a lever, a piston was drawn down in a cylinder, creating a vacuum; when the piston was released it was driven up the cylinder, creating the propulsive force to lob the mortar towards the enemy. This was, in other words, an adaptation of his atmospheric steam engine, or rather a reversion to his earlier plan for a wind-gun powered by atmospheric pressure.⁴³

Papin was still working on his atmospheric steam engine in March of 1704. How much progress did he make? The answer to this question is to be found in a notebook belonging to an English lawyer, musician and literary figure, Roger North.⁴⁴ There North described and sketched a two-cylinder atmospheric steam engine which he said he had seen ‘onely in modell’. In this period the word ‘model’ is ambiguous: it can carry the modern meaning but more often it refers to a graphic representation, a plan or drawing.⁴⁵ The phrase ‘in model’ is extremely rare, but a single sheet published in 1651 is described in its title as a summary of Christian doctrine ‘in model’: it is a wall-chart.⁴⁶ So North probably saw not a working model, or even a maquette, but a drawing – hence his insistence that he had only seen it in model. When he saw this drawing we cannot be sure: an earlier entry was written in 1701, which gives us an approximate date. This, presumably, is the engine which powered the little steam carriage which ran around Papin’s parlour.

The engine sketched by North is a development of Papin’s atmospheric engine; now the cylinders have automatic valve gear and operate reciprocally (in 1676 Papin had designed an air pump with precisely these features). The drive mechanism is evidently very similar to the one Papin had illustrated when he published an account of his steam-engine experiments in French in 1695, and which he claimed was modelled on mechanisms to be found in watches, although now the pinion moves away from the rack when the drive stroke is completed, rather than the other way round: the rack-and-pinion with a ratchet mechanism is distinctive because it is far from the best solution to the problem of how to drive a wheel by a piston (a crank being far better). Boyle’s first air pump had used a rack-and-pinion mechanism to drive the piston in the pump (the opposite of its use here, where the piston is driving the rack-and-pinion mechanism), but there was no ratchet to allow the rack to draw back without turning the pinion. It is possible it is the work of someone following in Papin’s footsteps, but it seems much more likely to be the work of Papin himself; evidently, he had sent a drawing of his latest engine to one of his friends in England, and this drawing had been shown to North. But there is no sign of Papin working on an atmospheric steam engine after 1704, or of news of the version of his engine recorded by North being disseminated. The advances Papin had made between 1695 and 1704 had no influence, and but for North’s sketch we would have no knowledge of them. Papin’s real contribution, as we shall see, lay elsewhere.

Roger North’s notebook entry showing his drawing of a two-cylinder steam engine and the rack-and-pinion mechanism by which the pistons turn an axle. (From British Library Add. MS 32504.)

§ 3

In 1698 Thomas Savery, a military engineer and Fellow of the Royal Society, obtained a patent for a steam-driven pump that used both atmospheric and steam pressure to raise water (some suspected he had simply copied an earlier design by Edward Somerset, the Marquess of Worcester (d.1667), who had devised a steam-powered pump).⁴⁷ Steam was introduced into a cylinder, which was then cooled by having water sprayed on to it. The steam then condensed, drawing water up a pipe into the cylinder. A valve was closed, the water was heated, and the steam generated drove the water upwards out of the cylinder. Thus Savery’s engine sucked and blew, just like a bellows, but the sucking was caused by condensing steam and the blowing was caused by expanding steam. Apart from the valves, this engine had no moving parts. Because the suck was driven by atmospheric pressure it could raise water no more than 30 feet or so, while the blow could push water upwards any distance, providing the pressure in the cylinder was high enough. Savery thus proposed mounting his device near the bottom of a mine and using it to pump water to the surface. In practice, the engine was used to power ornamental fountains, but not to pump water out of mines, because Savery could not build boilers and cylinders which would sustain a sufficiently high pressure.⁴⁸

Papin’s 1695 illustration of various pneumatic engines. The system on the left uses a water-wheel to drive pistons which pump air, which, by driving a second set of pistons, raises and lowers a bucket. At the centre top there is a representation of Papin’s piston powered by the pressure of the atmosphere: once the steam has condensed, removal of the pin labelled E causes the piston to descend. On the right there are two images of his rack-and-pinion ratchet mechanism.

News of Savery’s engine reached the landgrave of Hesse, and Papin was put to work devising a high-pressure steam pump. His initial efforts were apparently not very successful, and Savery was consulted on how to improve his design. In time, Papin successfully used a steam engine to pump water for an ornamental fountain (Louis XIV’s waterworks at Versailles had made ornamental fountains a highly competitive field for rulers and aristocrats). One of his engines blew up (despite the fact that Papin had invented the first safety valve), almost killing the landgrave, while the boiler of another had burst when it froze in the winter. Papin’s pump is often, and with good reason, described as a modification of Savery’s, though Papin claimed he had devised it independently.⁴⁹ It was significantly different from Savery’s pump in that it raised water only on the blowing cycle, and it separated the water used to power the system (which was being turned into steam and then condensed) from the water being pumped by using a float (which looked rather like a piston, but was not used to drive machinery), the idea being that this would prevent heat being wasted in warming the water being pumped through the engine. Moreover, absent from Papin’s design was a simple device employed by Savery: the use of a spray of water over the cylinder to cool it in order to speed the condensation of the steam.⁵⁰

Papin’s steam pump, from Nouvelle manière pour élever l’eau par la force du feu (1707). The boiler is on the left and the tank that is being filled on the right; there needs to be a steady supply of water into the hopper labelled G. Figure 2 is concerned with the design of the water-wheel that the pump is intended to drive. The pump is a modification of Savery’s engine, with a float introduced to separate the steam from the water being pumped. It is fitted with two of Papin’s safety valves.

Papin became increasingly discontented in Hesse, where the landgrave failed to give his researches the support he felt they warranted, and so he determined to return to England. It is commonly said that he built a steam-powered boat on to which he loaded all his belongings. He set out from Kassel on the River Fulda bound, eventually, for England. Unfortunately, he had gone 15 miles when he came to the junction of the Weser, and so to a stretch of river over which a guild of boatmen held a monopoly. He had made efforts without success to obtain an official exemption. The boatmen, determined to enforce their rights, seized his craft and destroyed it. And this was the end of steam-powered transportation for almost a century.

But the story of the steam-powered boat is based on a misunderstanding: Papin had built a boat powered neither by sail nor oars, and the boat was indeed destroyed. But it was not (as is clear from his correspondence) powered by a functioning steam engine. Papin had built a paddleboat (and not the first paddleboat – here, too, Savery was ahead of him), not a steamboat: the paddles were powered by cranks turned by hand.⁵¹ It is puzzling that this story is so often repeated without any sign of scepticism: after all, if it was possible to build a working steamboat in 1707, why would it take a further century to establish reliable steam propulsion on water? One author has not hesitated to reach the obvious conclusion that a dastardly conspiracy must have been at work. Yet the evidence to refute this myth was printed as long ago as 1880.⁵²

Papin, having lost most of his belongings with the wreck of his boat, and having been separated from his wife, finally arrived in England in 1707 and proposed to the Royal Society that they fund him to build his steam-powered craft. The Society submitted his proposal to Savery, who was not only the leading expert in the field but also held a patent which was so broadly worded that it covered any steam-powered engine. Savery insisted that the float/piston would produce too much friction to be workable. Newton, as President of the Society, dismissed the whole project as far too expensive.⁵³ Of course, Newton may have been biased against Papin because Papin was a friend of Leibniz, with whom Newton was increasingly coming into conflict. The Royal Society, after years of decline, in which it was short of funds and performing few experiments, was showing signs of a new enthusiasm for experimental science, but Papin did not benefit.⁵⁴

Certainly, Newton was right: Papin’s scheme was hopelessly expensive. The reason is apparent from the illustrations of Papin’s engine. It requires a supply of water into the pumping mechanism, and the source of that supply must be higher than the top of the cylinder.⁵⁵ If the engine were to be installed in a boat, and the water taken from the river or sea, then the whole of the engine would have to be below the waterline, which requires a very large boat with a very deep draught.⁵⁶ Papin was well aware of this: he proposed to the Royal Society a ship of eighty tons, perhaps 100 feet long, costing ‘but four hundred pounds’ to build.⁵⁷ How much was £400 worth? Fifty thousand pounds in modern money, using a retail price index, but £725,000 in modern money using an average wage multiplier. A more helpful measurement perhaps is that it was four times the salary of the Lucasian professor of mathematics at Cambridge: so let’s say £400,000.vv

Nineteenth-century illustrations of Papin sailing his steam-powered boat are thus completely misleading in that they show an engine mounted on top of the deck of a small craft, not a large sea-going vessel with an engine below decks. It is impossible to get round the problem that Papin’s steam-pressure engine could not be made to work to power a boat unless it was implemented on a large scale.vivi The scheme is simply impractical. Papin, who was constantly coming up with new schemes (he, like so many others, thought he could make a clock accurate enough to measure longitude) could find no one to back him. His final years were ones of failure and poverty. We last hear of him on 23 January 1712. ‘I am,’ he writes, ‘in a sad case.’⁵⁸ We do not know where, when, or how this great engineer-scientist died.⁵⁹

§ 4

Only five years after Papin’s failure, Newcomen produced the first commercially viable steam engine. The great merit of Newcomen’s engine was its simplicity of conception and modesty of ambition. It consisted of a single piston, powered by the pressure of the atmosphere. When the piston is driven downwards it pulls on a large beam which works a force pump. The weight of the pump mechanism ensures that the piston rests in the up position. Air is driven out of the cylinder by filling it with steam; the steam is then condensed by injecting water into the cylinder (Newcomen discovered how to do this by accident), and the atmosphere pushes the piston down; steam is then reintroduced into the cylinder at atmospheric pressure; this releases the piston, which is raised by the weight of the pump. The engine ran slowly, at about fifteen cycles per minute. The engine is simple because, like Papin’s first steam engine of 1690, it consists of a single cylinder powered by atmospheric pressure only. Savery’s engine, and Papin’s second engine, needed to build up a high pressure in order to be effective but, in practice, boilers and cylinders could not be made that withstood such pressure. On the other hand, Newcomen, unlike Savery, had to construct a moving piston, with all the difficulties of potential friction and leaking that entailed.

Newcomen had little formal education. Born in 1664, he was an ironmonger in Dartmouth, Devon, and an elder in the local Baptist Church. Yet, almost singlehandedly (we know of one assistant, Mr Cawley, a glazier), he brought a new technology into existence. How was this possible? It puzzled contemporaries just as it puzzles us. The first possibility is that he worked in complete isolation, without knowledge of anything that had gone before. One only has to formulate this possibility to see that it must be wrong. For a start, Newcomen could not have devised his engine without knowing about the pressure of the atmosphere, since this provides its driving force. It is true that knowledge of air pressure was widespread by 1712, and any explanation of the workings of a barometer would have conveyed to Newcomen the discoveries of Torricelli and Pascal. But he would have needed this as an absolute minimum.

We have almost no direct information about Newcomen prior to 1712 but, from what he told his associates in later life, two things seem evident. First, he began work on his steam engine around the same time as Savery began work on his, so no later than 1698. Second, he worked in complete independence of Savery.⁶⁰ Nevertheless, some scholars have felt this makes no sense. Newcomen must have benefitted from either Savery’s or Papin’s expertise. One scholar has boldly, and against all the evidence, claimed that Newcomen was simply an employee of Savery.⁶¹ Another, going as he himself puts it, ‘in the face of all the evidence’, has suggested that Newcomen and Savery could have met in January 1707 or soon after, when we know that Savery went to Dartmouth – but this is far too late (and so is soon altered by a sleight of hand into ‘by 1705’, which is still too late).⁶² A late-eighteenth-century scholar ‘solved’ the problem by claiming that Hooke (who died in 1703) had written to Newcomen describing Papin’s first steam engine. This story still gets repeated, despite the fact that the documents which are supposed to support it do not exist, and have been known not to exist since 1936.⁶³ Another scholar says that ‘Thomas Newcomen surely must have seen Papin’s sketches of his models of proto-engines and pumps, published in various issues of Philosophical Transactions between 1685 and 1700’ – glossing over the fact that none of Papin’s publications in the Transactions dealt with steam power; they all relied on water-power or man-power.⁶⁴

The Newcomen Engine, as illustrated in John Theophilus Desaguliers, A Course of Experimental Philosophy (1734–44; taken from the 1763 reprinting). The boiler is on the left, with the piston rising vertically out of it and connected to the rocking arm.

Papin’s engines were the closest in conception to Newcomen’s. The great historian Joseph Needham said, very sensibly, ‘I find it almost impossible to believe that Newcomen did not know of Papin’s steam cylinder.’⁶⁵ But Papin had made and operated his first engine in Germany. No Englishman, as far as we know, had ever seen it. He described it several times in print, in Latin and in French, but never in English. A single paragraph appeared describing it in English in a review of one of Papin’s publications which appeared in the Philosophical Transactions for 1697:

The fourth letter shows a method of draining mines, where you have not the conveniency of a near river to play the aforesaid [pumping] engine [by means of a water-wheel]; where having touched on the inconveniency of making a vacuum in the cylinder for this purpose with gunpowder [as Huygens had done], he proposes the alternately turning a small surface of water into vapour, by fire applied to the bottom of the cylinder that contains it, which vapour forces up the plug [i.e. piston] in the cylinder to a considerable height, and which (as the vapour condenses as the water cools when taken from the fire) descends again by the air’s pressure, and is applied to raise the water out of the mine.⁶⁶

It is highly unlikely that Newcomen ever had access to the Philosophical Transactions, but even if he did this single paragraph, without any supporting illustration, would have left him with a tremendous amount of work to do. As for the more advanced Papin engine sketched by North, this would obviously have been of great interest to Newcomen had he known of it, but it probably postdates the beginning of Newcomen’s programme of experimentation, and its design is much more complicated than Newcomen’s – indeed, we may doubt that Papin ever got it to work properly.

It may be helpful to list some of the things that Newcomen would have had to invent to make a successful steam engine, or which he would have needed, even before that, to conduct a programme of experiments. Much of what he needed was straightforward. The pump bucket and pump handle, for example, were straightforward applications of existing technologies, and the boiler was basically a large brewer’s copper. But other things were far from straightforward. First, though the idea of using a cylinder and piston went back to Guericke, there was no recent experience of combining this with steam in England. Second, Newcomen needed a means of making the piston air-tight. He sealed his piston with a leather washer and a layer of water injected into the cylinder. (John Morland had designed pumps using pistons in the 1680s – his seal was quite different.)⁶⁷ Third, it would have been nice to have a pressure gauge: the barometer is the first pressure gauge, but Boyle and Papin had described a sophisticated pressure gauge in the Continuation of New Experiments of 1682. Crucially, it was essential to have a safety valve, which is in itself a form of pressure gauge: Papin had invented one, and had incorporated one in his 1707 design (though perhaps not on the version which had exploded). Newcomen used a version of Papin’s safety valve (called the Puppet Clack) in his engine.⁶⁸ In addition, he needed a technique for having the valves in the piston open and close by the action of the machine itself.⁶⁹

Lastly, there is a further prerequisite. It is a feature of Savery’s engine that it runs better on a small scale than when scaled up: as the cylinders get bigger, the volume of the cylinders increases more rapidly than their surface area, and so cooling becomes less efficient. So when Savery built a model he would have been misled into thinking he had made a breakthrough. Newcomen’s engine is the opposite: the ratio of power to friction is most unfavourable on a small scale, and becomes more favourable as the engine is scaled up because the volume of the cylinder (which determines the power of the engine) increases more rapidly than the piston’s circumference (which determines the amount of friction).⁷⁰ Desaguliers and a friend later built models of both the Savery and the Newcomen engines: despite his extraordinary expertise, Desaguliers was plainly taken aback to see the Savery engine outperform the Newcomen engine.⁷¹ Newcomen must have understood this scaling issue from the beginning, otherwise he would never have persisted when his first models performed (as they must have) very poorly. He must have obtained this knowledge from some source.

Of course, Newcomen could have invented all this and more; after all, he worked on his new engine for some fourteen years before he was ready to put it into public operation. But it is worth knowing that only a few years ago an attempt to build a one-third-scale replica Newcomen engine ran into a great deal of difficulty. Even with good plans, even with plenty of technical expertise, even with a knowledge of what the end product was supposed to be and an absolute certainty that it could be made to work, it turned out that it took many months of tinkering and fiddling to get the engine to run properly.⁷² Ideally, Newcomen needed a source of information which would have supplied him with all sorts of bits and pieces of key information, so that he could then concentrate on working out how to assemble a working machine. That would have been quite enough to keep him busy in his spare time for a decade or more.

§ 5

There was indeed such a source, one that Newcomen is much more likely to have come across than the Philosophical Transactions. It is a source that historians of the steam engine have overlooked because it does not discuss steam engines. Indeed, it has been generally overlooked: there is not a single citation of it in Google Scholar, or in Thomson Reuters Web of Science. One could easily form the impression that no one has read it in the past century, despite the fact that its author is well known and, judging by the number of surviving copies, the book sold well when first published. I refer to A Continuation of the New Digester of Bones, published by Denis Papin in 1687.⁷³

Papin published the first account of his New Digester in 1681. It was, quite simply, the first pressure cooker, a sealed bain-marie. Because the pressure cooker turns water into steam under pressure it cooks at a higher heat than normal boiling water, and so much more quickly, or (in the case of digesting bones) it cooks until hard materials have been reduced to soft pulp. (Papin’s Digester has a special place in history because in 1761 or 1762 Watt carried out his first experiments with steam by attaching a syringe to the safety valve of a Papin Digester, thus making a primitive steam engine.) The New Digester and the Continuation were often bound together, and it is easy to imagine Newcomen acquiring either the Continuation alone or both volumes together in 1687 or at any time in the decade before he began work on his steam engine. His motive would have been straightforward: there was a possible profit to be made making and selling Papin’s device and, since Papin had insisted that anyone was free to copy it, and had not protected it with any patent, there was no obstacle to Newcomen trying to make money out of it.

The short title is, however, a poor guide to the contents of Papin’s book. The full title is more informative: ‘A continuation of the new digester of bones: its improvements, and new uses it hath been applyed to both for sea and land: together with some improvements and new uses of the air-pump, tryed both in England and Italy’. This is, in part, a book about the air pump (although historians of the air pump and of vacuum experiments have failed to read it),⁷⁴ and it provides an illustration and description of Papin’s most recent (and last) model.⁷⁵ Papin’s pump consists of a cylinder with a piston; the piston is sealed by a layer of water, and Papin describes with care how to achieve this.⁷⁶ The method used corresponds to the method initially used by Newcomen – though he later found a better one.⁷⁷ The cylinder, like the piston of Newcomen’s steam engine, has a number of valves and inlets which open and close with the action of the piston. (Papin was the first to make an air-pump in which the valve action was automatic.) There is one valve closed by a weight, although it is not in this case a safety valve; however, Papin describes the operation of such a valve. Thus the basic technology of the steam engine’s piston is laid out because that technology overlaps with the technology of the air pump – it is precisely because of these overlaps that Papin could go on three years later to build the first steam engine.viivii

Papin’s 1687 air pump, from A Continuation of the New Digester.

But the Continuation does more than that. It provides the reader with the line of thinking that led Papin to the invention of the steam engine. Here is what he says:

I might also reckon among the uses of this Engine [the air-pump] the strength it can afford to produce great effects without the encumbrance of great weights: For a tube very even and well workt may be made very light and yet being emptyed of Air it will endure the pressure of the Atmosphere: Nevertheless a plug very exact at one end of that tube would be pressed towards the other with a very great strength, at least if the tube was of a pretty great Diameter: for example if it was a foot Diameter the plug would be press’t with the strength of about 1800 pounds. The Renouned Mr Guerike hath the first tryed to apply that strength to shoot leaden bullet’s [sic] through a gun, as may be seen in the description he hath given of it in his book of the Pneumatick engine: I have also endeavoured since to add some thing to his invention, as may be seen in the Philosophical Transactions for the mounth of January 1686: since that I have calculated that a bullet of lead of an Inche Diameter being thus shot through a barrel 4 foot Long should acquire the swiftness to fly about 128 feet in a second; but if the same swiftness was to be given to a bullet of a foot diameter it should be made of Iron hollow within, so that it should weigh but about 37 pound and half: For if it was made solid of Lead, it would weigh about 450 pounds, and so in passing the length of about 4 foot in the barrel, it would acquire but the swiftness of 32 feet in a second … The end of the barrel through which the bullet passeth must be stop’t with something strong enough to uphold the pressure of the Atmosphere, and that being met in the way of the bullet takes away also some of its strength.⁷⁸

What Papin is describing here is a wind-gun powered by atmospheric pressure; but it is clear that the device, which requires the bullet to punch a hole to escape from the barrel, is not fit for any practical purpose.

What he is also describing is a piston powered by atmospheric pressure; he was on the point of inventing the atmospheric steam engine, but here the vacuum is created by his pump, not by the condensing of steam. He does, however, repeatedly describe how to use water applied to the outside of a container full of steam in order to bring about rapid condensation (although he never used this technique in his own steam engines) and so a vacuum.⁷⁹ If he read him, Newcomen had only to put two and two together in exactly the way Papin was to go on to do to have the basic design of a steam engine. If Papin could do it, why could Newcomen not do it, too? Moreover, Papin here introduces the reader to the problem of scale: as you scale up the gun, it becomes less efficient, because the weight of the bullet increases faster than the surface area of its end. Someone who thought hard about this might grasp that as the diameter of the tube increased, the increase in the weight of the bullet would be partially compensated for by a decrease in the proportion of energy lost to friction.

Newcomen’s steam engine is a bit like a locked-room plot in a detective story. Here is a dead body in a locked room: How did the murderer get in and out, and what did he use as a weapon? Our puzzle is that we have Newcomen in Dartmouth in or around 1698, and we can see no way in which knowledge of the steam engine can reach him. As with the locked-room mystery, if we can find one solution, then we have found the solution. Of course, we cannot exclude the possibility that Newcomen went to London and met Papin in 1687; Papin indeed advertized that he would be available at a certain time each week to demonstrate his digester, although in fact he soon left the country. But we do not need to imagine such a meeting. With a copy of the Continuation in his hands, Newcomen would have known almost everything that Papin knew about how to harness atmospheric pressure to build an engine. All the bits and pieces were there; all he had to do was recognize how they could be assembled to serve a new purpose, not to create a gun but to power a pump. And the Continuation, with its instructions on how to build a revised model of the Papin Digester, is precisely the sort of book that a provincial ironmonger and small-time manufacturer would have been looking out for. The last thing Newcomen would have expected to find in it, the last thing he would have been looking for, is a description of a new type of power capable of producing great effects without the encumbrance of great weights. From this unintended encounter, I believe, the steam engine was born.

Desaguliers, in the first major study of the steam engine, insisted that all the great advances in steam-engine design had been made by chance:

If the Reader is not acquainted with the History of the several Improvements of the Fire-Engine since Mr Newcomen and Mr Cawley first made it go with a Piston, he will imagine that it must be owing to great Sagacity and a thorough Knowledge of Philosophy, that such proper Remedies for the Inconveniencies and difficult Cases mention’d were thought of: But here has been no such thing; almost every Improvement has been owing to Chance … ⁸⁰

Desaguliers chose his words with care. He said that the improvements were owing to chance, but he left his readers to make up their own minds as to whether the action of first making the steam engine go with a piston required a thorough knowledge of philosophy or not. Certainly it required some philosophy, and some inherited technology. Both, I suggest, were provided by Papin’s Continuation.

Indeed, when Desaguliers explains the workings of Newcomen’s engine, he makes a remarkable move. He asks you to imagine an engine in which ‘a philosopher’ uses an air pump to create a vacuum in a piston, before going on to describe Newcomen’s actual design, in which steam is condensed in a piston to make a vacuum. This philosopher is certainly not Newcomen; but Desaguliers had, I think, correctly intuited the only plausible route to the invention of the Newcomen engine. Where Newcomen had put two and two together to make the steam engine, Desaguliers, in order to explain its working, has taken them apart again, reinventing Papin’s atmospheric wind-gun.⁸¹

Historians have long debated the extent to which science contributed to the Industrial Revolution. The answer is: far more than they have been prepared to acknowledge. Papin had worked with two of the greatest scientists of the day, Huygens and Boyle. He was a Fellow of the Royal Society and a professor of mathematics. In the twenty years between 1687 and 1707 he worked towards the construction of a viable steam engine, but in the end he failed. Newcomen picked up, I suggest, not where Papin ended, with his modified Savery engine, but where Papin began. In doing so he inherited some of the most advanced theories and some of the most sophisticated technology produced in the seventeenth century. It was this that made the Industrial Revolution possible. First came the science, then came the technology.viiiviii