Francis Hauksbee was active in London between 1699 and 1713. During those years he built scientific instruments, gave public lectures on natural philosophy and worked as a curator of experiments for the Royal Society. His most celebrated instrument is the double-barrelled air pump, which represents the ‘state of the art’ of eighteenth-century vacuum technology in Britain. Based on original texts and an examination of extant pumps of this design, this article offers a description of the air pump and an account of some of the experiments performed with it. In addition, notes on existing Hauksbee pumps to be found in modern museums and collections are provided.
The double-barrelled air pump of Francis Hauksbee (1660–1713), first introduced in 1705, is well known to modern historians of scientific instruments. It is a masterpiece, a stunning construction with ornaments and beautiful woodcarvings, and it represents the ‘state of the art’ of British air-pump technology of the period (figure 1). However, the pump first presented at a meeting of the Royal Society on 15 December 1703 was a smaller, single-barrelled version that had been available to customers since 1702. This pump was a further development of a ‘combined engine’, a large syringe that could be used as an air pump, as a condenser for compressing air, and as a syringe for injecting air, wax or mercury into pathological specimens. This device was based in turn on technologies derived from small cupping pumps, developed by Hauksbee between 1699 and 1702. In an earlier study I gave a comprehensive description of this advance; here, by contrast, I focus on Hauksbee's classic double-barrelled air pump.1
Owing to the remarkable qualities of this pump and the suddenness of Hauksbee's appearance in the records of the Royal Society, there has been much speculation about his background. The commonest theory is that he might have been one of Robert Boyle's (1627–91) assistants, and that the air pump was simply an improvement of Denis Papin's (1647–1712) foot-operated pump with two barrels and self-closing valves.2 A further complication and source of misunderstandings is that he had a nephew with the same name who also made air pumps: Francis Hauksbee the younger (1687–1763). Hauksbee certainly knew about Papin's pump and Boyle's experiments, but the leaps from the experimental practices and pump technology of Papin and Boyle to Hauksbee's are big and discontinuous. My firm contention is that Hauksbee's success had its roots in his experience with cupping pumps.
Hauksbee's engagement as a weekly demonstrator in the Royal Society has often been seen as a facet of Newton's initiative to restore the Society after a period of decline. It is true that his introductory performance on 15 December 1703 took place at the first meeting at which Newton presided as the newly elected president. But he must have caught the attention of the Royal Society long before that, and without Newton's necessarily being involved.3 Hauksbee advertised his cupping utensils from at least January 1699, and air pumps from 1702, and he must have been well known to the London public.4 He often claimed, for example, that his cupping devices were appreciated by ‘physicians and eminent surgeons of this City’.5 As several members of the Royal Society belonged to these professions, it is very probable that some of the same medical men were among Hauksbee's early customers, thus establishing contacts even before air pumps became standard commodities.6 Moreover, I have found nothing to connect Hauksbee with either Robert Boyle or Robert Hooke.
Among the few details we know about Hauksbee are that he was born in Colchester, Essex, in 1660 and apprenticed as a draper to his elder brother John, and that he belonged to the Drapers' Company in London. He lived in Queen's Head Court in Giltspur Street, London, but moved to Wine-Office Court in 1707 and to Hind Court in 1710, both addresses off Fleet Street.7 Further, he was engaged by the Royal Society to perform experiments during its weekly meetings, was elected a Fellow on 30 November 1705, and wrote 50 articles for Philosophical Transactions. He performed about 120 different experiments, 75% of them incorporating the use of air pumps, and published a book, Physico-Mechanical Experiments on Various Subjects.8 He also gave public lectures on natural philosophy, manufactured and sold scientific instruments, and had several apprentices and associates.
His product range embraced utensils for bloodletting and cupping, including fleams and scarificators, cupping glasses, and hand-held syringes (small air pumps) for evacuating the glasses. For anatomists he offered tools for preparing pathological specimens; these tools included large syringes and siphons for injecting various materials into the samples. In addition, he supplied condensing engines for experiments in pressurized air, as well as single-barrelled and double-barrelled air pumps with a series of accessories suitable for experiments in ‘Vacuo Boyliano’.9 He also made barometers. All these devices were made known to the public through advertisements in newspapers and almanacs, and some can be found in legal proceedings at the Old Bailey.10
With regard to his family life, we know that he was married and had several children, one of whom, Ann, married one of his apprentices, Richard Bridger. A low moment in his life must have been the day in the spring of 1707 when he was commanded to demonstrate some experiments on light emitted from mercury for the Queen and the Prince—but without success. He explains the ‘crisis he then was in’ by the ‘unfriendlyness of the air’, it being too humid.11 He died in London in April 1713.
Traditionally, Hauksbee's double-barrelled air pump has been regarded as his only model. However, the pump that Hauksbee presented to the Royal Society on 15 December 1703 was of a different type: horizontally mounted and single-barrelled.12 It was this model that received glowing plaudits from Hauksbee's contemporaries, such as ‘the best air-pump I ever have seen’ (Harris),13 a pump that exceeded others in ‘ease in being used’ (William Derham),14 and ‘the improved one’ (James Hodgson).15 In the belief that these appraisals reflect judgements on the double-barrelled pump, and given the scarcity of information on Hauksbee's activities before December 1703, modern historians have until now had difficulty in explaining the qualities and the sudden appearance of the classic version.
The single-barrelled pump was introduced in October 1702, about a year after the ‘combined engine’ was put on the market (see above). When the ‘engine’ was used as an air pump, it was described as a device capable of making experiments in a vacuum with ‘ease and exactness’ and without the use of ‘cement and stopcocks’. The first pump was characterized in a similar way; it also came with ‘Apparatus for making all manner of Experiments proper to be made in Vacuo Boyliano’.16 There was a self-closing valve in the piston and one in the bottom of the cylinder, and these opened and closed when the plunger was pulled out and pushed in, respectively, thus performing a pumping action. This version was available from Hauksbee's workshop throughout his career, as well as from later makers of pneumatic instruments during the eighteenth century. The point is important, because it indicates the versatility of the device as a simpler and cheaper alternative to the double-barrelled model.17
It is unclear how many of these single-barrelled air pumps exist today. Compared with Hauksbee's classic pump, the devices were much smaller and anonymous, lacking its ornaments and status, and are much less easy to identify as being made by Hauksbee. It is possible, however, that some still exist, although only an extensive search in museums and collections will allow us to know.
The classic double-barrelled air pump
The first sign of an air pump with two barrels (cylinders) is found in an announcement in The Post Man on 1 May 1705. Called ‘Hauksbee's new Pneumatick Engin’,18 the pump was described in the same way as its predecessor (the single-barrelled one), as facilitating experiments in Vacuo Boyliano ‘with ease and exactness, and without the use of cement and stopcocks’. It also came with pieces of apparatus for specific experiments. There was one important difference, however. When approaching a vacuum with a single-barrelled pump, the increasing pressure difference over the piston (vacuum versus atmospheric pressure) made it harder and harder to pull the plunger, something that rendered pumping rather exhausting. This was not true of the novel pump, which consisted of a receiver (the glass vessel for experiments) on a plate on top of the pump structure, and a gearwheel and two racks, each connected to a plunger in a barrel. Pumping was executed by turning the gearwheel back and forth with the crank, causing up-and-down movements of the pistons (figure 2). The cylinders and pistons were supplied with one-way valves, which opened and closed automatically. There was some water in the barrels, enough to cover the piston in its lowest position, thereby keeping the air out. During pumping, the forces on the two pistons would balance each other out; when approaching the lowest pressure, the only resistance to overcome was not much more than the friction of the mechanical parts of the engine (figure 3). The pump was supplied with a mercury gauge (barometer) for monitoring the pressure inside the recipient. It also had a specially designed air-inlet screw, to allow the readmission of air into the receiver after an experiment was over. These elements were joined with a rectangular piece of brass, described by Hauksbee as the ‘cross-piece’ and placed directly under the pump plate.19
A fully equipped pump with about 20 accessories for various experiments sold for £25. When a new pump was shipped to a customer out of town, it came in a wooden crate filled with shavings; the thin glass tube for the gauge was packed separately. One end was to be attached to the cross-piece by a brass socket cemented to the tube, and the other went into a mercury-filled reservoir at the bottom of the pump. There was also a spanner for fastening the socket to the cross-piece, as well as some cement for gluing the gauge tube to the socket. In addition, a spare tube and some cement for fixing the gauge were provided, in case of breakage. Some of the apparatus for the experiments arrived partly disassembled and had to be put together by the purchaser. Detailed assembly instructions came with the shipment, and certain components had notes attached to them explaining where and how they should be mounted. Also included with the shipment were directions on how to perform experiments with the pump, stated to be very easy to understand.20
In his advertisements Hauksbee repeatedly used terms such as ‘ease and exactness’, ‘expedition’ and ‘without the use of Cement and Stopcocks’ to characterize his pumps. The term ‘ease’ indicates simply that Hauksbee claimed that his pumps were easy and convenient to operate, advantages that he seems to have contrasted with the disadvantages of Robert Boyle's air pumps.
One element contributing to ease of operation was Hauksbee's easy and convenient way of avoiding leaks. Unwanted air seeping into the pumps had been a major concern for Boyle. It had been such a significant issue that it even challenged the credibility of the instrument's role in producing new knowledge.21 To keep the problem at bay, Boyle had used a thick paste of sealing wax or cement on the junction between the pump plate and the receiver rim, as well as on the other joints, always applied with a hot iron.22 With Hauksbee's pumps, however, a distinctive characteristic was the absence of such remedies as cement and sealing wax, a feature made possible by the use of gaskets made of wet leather. The receiver was placed on the pump plate, on a leather seal in a brass dish containing water, the purpose of which was to keep the leather moist (figure 2). It was also important for the edge of the glass (receiver rim) to be ‘truly grounded’. With these precautions in place, as the process of exhausting the receiver proceeded, the external air pressure would force the smooth glass rim down and into the wet leather seal, making the system airtight, thus eliminating the need for sealing wax. Hauksbee claimed that his sealing methods were ‘beyond Cement whatsoever’, and that experiments now could be made without ‘any Daubing or Difficulty’.23 Because water was easier to handle than sealing wax, the entire task of performing air-pump experiments thereby became much more convenient and less time-consuming. The mixing, handling and application of sealing wax had become unnecessary, and the time taken to prepare a ‘pump-down’ was reduced or eliminated.
Another feature that made for easy pumping was the absence of stopcocks. Instead of using manually operated taps, Hauksbee had supplied his pumps with inbuilt self-closing valves, a further development of those used in his cupping devices (see above). The phrase ‘without the use of stopcocks’ must again be a reference to Boyle's practice. Pumping with Boyle's first two machines had been a time-consuming and complex operation. A procedure consisting of the opening and closing of stopcocks and the removal and insertion of stoppers, as well as the back-and-forth movement of the crank, was laborious, often requiring two persons.24 With Hauksbee's model the operator had only to turn the crank, and the automatic valves in the barrels and the pistons did the rest, keeping the achieved vacuum inside the receiver. The nature of the valves was crucial. They consisted of a flexible material covering an orifice, a concept that Hauksbee developed when he improved his cupping utensils.25 Self-closing valves were nothing new, however, the principle having been used in water-pumps since antiquity. Hauksbee's ingenuity lay in his being able to manufacture such valves for use with air and in commercial products. These were reliable and tiny, ‘dry’ valves, in contrast with the coarse closing mechanisms for water. In former case even small leaks can be fatal, ‘destroying’ the vacuum, whereas in water-pumps, which in normal use transport the medium from one place to another, some spill can be tolerated. The medium, water, will itself also act as a sealing agent. In air pumps, however, where the intention is to remove the air totally, the medium has no such function.
A problem encountered when working with air pumps is the difficulty of removing the receiver and other parts of the pump after an experiment is complete. This is due to the ambient air pressure of about 1 kg cm−2 on all evacuated components. As an example, the total ambient force on a glass receiver 25 cm in diameter with vacuum inside is almost 500 kg, effectively clamping the receiver to the pump-board. The remedy is to equalize the pressure difference between the inside and outside of the pump in a controlled way. In the double-barrelled pump, this was conveniently achieved with a separate, specially made air-inlet valve with vanes, screwed into the manifold (the ‘cross-piece’) under the receiver plate (figure 2). Atmospheric air was let into the pump by turning the vanes, opening a passage between the threads of the screw and the mating ones in the pump plate.
Another expression occurring in Hauksbee's pump advertisements is ‘exactness’. This alludes to the inbuilt mercurial gauge, which he described as another Excellency: ‘shewing at all times the Degrees of Rarefaction to the greatest nicety’.26 Earlier, Robert Boyle had experimented with barometers in his pumps, placed inside the receiver. In one experiment he had studied the behaviour of a barometer in vacuum;27 later he used a barometer to check the pressure when trying to detect the ether by dropping feathers in a tall receiver.28 To keep the arrangements leak-tight, he had to rely on cumbersome ad hoc solutions requiring a lot of sealing wax.
Hauksbee's gauge, however, was an integrated measuring device, consisting of a long glass tube located under the pump board and connected to the receiver space through its top. The lower end was placed in a cistern filled with mercury. At atmospheric pressure, the mercury would stay in the cistern. However, as the receiver was gradually emptied of air, the mercury would rise in the tube, approaching the 29.5-inch mark on a ruler as the pressure reached its minimum, forced upwards by the pressure difference between the ambient air and the vacuum in the receiver. The ruler was marked every quarter of an inch from the surface of the mercury to a height of 28 inches. Above that, it was graduated in tenths of an inch.29 To obtain accurate readings, it was placed on a piece of cork floating on the mercury in the cistern. By means of two brass loops, it could slide freely up and down along the gauge tube, causing the zero point on the ruler to be always ‘locked’ to the mercury surface. Thus, the actual pressure, represented by the distance between the top of the mercury column and the surface in the cistern, would invariably be measured correctly.30 With this arrangement, variations in atmospheric pressure would influence the mercury level in the tube. Hauksbee therefore recommended that the operator check the ambient pressure with a weatherglass (barometer) when measuring the receiver pressure, and to compensate for it when reading the gauge.31
Soon Hauksbee's idea of continuously monitoring the vacuum became standard practice in pneumatics.32 The idea of a gauge as an integrated measuring device on his pumps, with all its attendant conveniences, was a great step forward in the process of turning the air pump into a laboratory instrument.
How fast could the operator prepare for a new experiment? Preparation included letting air into the receiver after an experiment was over, lifting off the receiver and removing the equipment used, arranging the new apparatus on the pump plate, putting the receiver back again, checking for leaks, and then pumping down to a suitable pressure. Hauksbee described this as ‘Expedition’, asserting that it was now possible to perform several experiments with his new pump ‘in the same time as they could formerly make one’.33 The rapid performance of a series of experiments depended on factors described in the paragraphs above, as well as on pumping speed.
Pumping speed is determined essentially by the ‘pump-capacity’, which denotes the amount of air that can be removed from the receiver in each pumping stroke. Pumping speed is then a measure of the time needed to evacuate a volume of a given size.34 Each barrel on Hauksbee's pump had a volume of 620 cm3. One stroke consisting of a back-and-forth movement of the crank, causing first one cylinder and then the other to pump, yielded 1240 cm3.35 If the operator could maintain one stroke every two seconds, a pumping speed of 620 cm3 per second could be obtained. Hauksbee's contemporary Roger Cotes (1682–1716) of Cambridge University published a textbook that included instructions on how to calculate the number of strokes needed to achieve a certain pressure with Hauksbee's double-barrelled air pump.36 Following his method by using the tables in figure 4, 39 strokes were needed to reduce the pressure in a 12.4-litre receiver (10 times the total cylinder volume) from atmospheric pressure to ca. 0.75 inch of mercury, or 25 mbar (1/40 of atmospheric pressure). Calculations based on modern formulae give the same results. With one stroke every two seconds, the operator could arrive at this pressure in a little more than a minute.
This high experiment rate became possible through a combination of the sealing method, continuous operation without manual stopcocks, the simple air-admission facility, and the high pumping speed resulting from the two interconnected barrels. Together, these factors made the preparations for air-pump experiments much less time-consuming than had been the case with Boyle's pumps.
At this early stage in the development of air-pump technology, a short pump-down time was seen as more important than obtaining a low pressure.37 There are several factors influencing the minimum pressure in this kind of air pump, such as the leaking in of air and evaporation from substances inside the receiver, as well as constructional details of the valves and cylinders.
An apparent and obvious element was the problem of leaks. Despite Hauksbee's advanced sealing methods, there would always be occasions when air entered the pump along imperfectly sealed joints and fissures. In the evacuated receiver, this influx would contribute to a continuous rise in the pressure. The way to tackle this is of course to pump it away. In this way, the minimum pressure in an air pump is reached and maintained when the pumping speed (the rate of removal of air) is equal to the influx of air through the leaks, plus the evaporation (outgassing) from components inside the pump itself. In practice, after a receiver had been evacuated and an experiment was in progress, the operator had to make some strokes with the crank from time to time to keep the pressure at the intended level.
The volume of the cavity remaining under the piston in its lower position in the cylinder, called the ‘dead space’, also limits the final pressure.38 During operation, the pump pressure can never be lower than the pressure obtained when the amount of air kept there becomes expanded into the whole barrel by an outward stroke (figure 2). Despite being designed to be as small as possible, the space cannot be eliminated fully because some room is required for free movements of the leather flap of the valve in the bottom of the cylinder. On some occasions, Hauksbee had water in the barrels to obtain better sealing, which would contribute to filling the dead space and make it smaller. In the double-barrel pump, each barrel had a volume of ca. 620 cm3; if we estimate the dead space to be 5 cm3, the minimum pressure that could be obtained in the receiver would be about 8 mbar, corresponding to a quarter of an inch of mercury. With the given relationship between the dead space and barrel volume, the pressure in the receiver could never become lower than this.
Another limiting factor is the shape of the valve in the bottom of the barrel, which is designed to open and close as a result of the pressure difference over the flap. This valve connects the receiver to the pump barrel and opens during an outward stroke, closing when the piston is pushed back. This works well at the start of a pumping task, when the pressure on one side of the flap is much greater than on the other and is sufficient to force it open. But as pumping diminishes the pressure in the receiver, the pressure difference over the flap decreases. Some force is always required to overcome the tension in the valve material that keeps it closed. At some point, the pressure difference would be too small lift the flap and the valve would stay closed; no further pumping would be possible (valve equilibrium).39 If we estimate this minimum pressure difference to be 4 mbar and the pressure in the cylinder caused by the dead space to be 8 mbar, a pressure of 12 mbar in the receiver was the best the operator could hope for.
Finally, there is yet another issue concerning the lowest attainable pressure in air pumps, which neither Hauksbee nor his contemporaries seem to have been aware of, namely ‘vapour pressure’.40 When the pump was operated at room temperature (20°C) and the pressure approached ca. 26 mbar, the water used as a sealant would start to boil, filling the receiver with water vapour.41 In the end, therefore, it was the crucial sealing agent—water—that paradoxically constituted the limitation on Hauksbee's pumps, giving a theoretical final pressure of about 26 mbar (ca. 0.75 inch of mercury).
The purchaser of an air pump would also receive glassware with which to perform the most basic experiments for displaying the properties of the air, and so on. In a delivery to Samuel Molyneux in April or May 1707, the pump came with a receiver to ‘include a cat or any other animal of a larger sort’, a tall receiver about 18–20 inches high for the simultaneous dropping of lead and a feather in an airless space, and a device for firing gunpowder on an incandescent piece of iron in vacuum. Hauksbee also included apparatus to demonstrate how high the spring of the air in a bottle could elevate mercury in an open, tall glass tube. This apparatus came in several parts ‘for the greater security in the Carriage’. There was a small receiver, open at both ends, intended for ‘those who denied that the air had weight’. When placed on the pump plate with the broad end down, a person laying his hand on top of it would soon ‘be sensible of something that held his hand so fast that it was impossible to remove it’, before air was let into the pump. A bell to be suspended in the receiver was also part of the delivery. This was for demonstrating that ‘air was a medium of sound’. No, or very little sound, should be heard when the receiver was shaken, so as to cause the clapper to strike in vacuum. A series of square glass phials used for showing the spring and pressure of the air likewise came with the pump. The phials were supplied with brass heads containing a one-way valve.42 When placed under the receiver, the air would be sucked out of the bottles during pumping. When air was let in again, the valve would close, keeping a vacuum inside. Approaching atmospheric pressure in the receiver, the phials would burst inwards due to the pressure difference over the flat glass walls. An even more spectacular test was to block the valves keeping air inside the phials, and then to put them in the receiver. Pumping now caused the phials to burst outwards with ‘great violence’, owing to the lack of counterbalancing air on the outside.
Experiments were on Hauksbee's agenda even before he became engaged by the Royal Society in December 1703. The first indication of his future career as a unique experimenter is found in the advertisement of his first air pump, in October 1702.43 There, in addition to air pumps, he offered a range of accessories for making experiments in Vacuo Boyliano. Throughout his career, his focus was on experiments and results, as well as on the special equipment and accessories that were used with the pump. Despite its successful use in the experiments, the pump had no central place in his texts, except for a four-page description in his book Physico Mechanical Experiments. Instead he made it clear that ‘I shall forbear taking any further notice of it, saving what immediately relates to the following Experiments’.44 Whereas Boyle's air pump, also described as ‘Machina Boyliana’, had been an object for development and study, Hauksbee's air pump was a ‘black-boxed’, accepted and efficient instrument that needed little further attention.45 For example, in the ‘Journal Book’, which contains minutes for every meeting that Hauksbee attended in the period 1703–13, the pump is discussed only once, when it is compared with a new model contrived by the Musschenbroek workshop in Leiden, said to be ‘inferior’ to Hauksbee's design.46
Hauksbee's air-pump experiments can roughly be divided into two groups: ‘research experiments’, undertaken for enquiries into particular topics (production of knowledge), and experiments for the demonstration of known facts. The overall purpose of the research experiments was to test hypotheses and solve debates. In an article concerning some characteristics of the air's pressure, Hauksbee wrote: ‘Since the greatest Satisfaction and Demonstrations that can be given for the Credit of any Hypothesis, is that the Experiments made to prove it, agree with it in all respects, without force’.47 Most of his experiments stemmed from discussions in the Royal Society. Through about 220 entries in the ‘Journal Book’ we learn how he was ‘directed’ to plan and perform experiments by Isaac Newton, how he was desired to work more on particular experiments before giving a report to the Society, and how he came up with ideas for improving experimental procedures and equipment.48 One thing that most of the experiments (about three-quarters of them) had in common was the use of the air pump.49 In the period 1703–13 he performed about 120 different research experiments in a variety of fields: electricity and light, capillarity, sound, life experiments, experiments on air and fluids, magnetism, optics, and studies of specific weights. Many of these are well known and have been discussed by various modern authors. What follows is an overview.50
In his experiments on light and electricity, Hauksbee used the air pump to evacuate rotatable glass globes and cylinders, managing to create a flickering, bluish light inside them by rubbing the outside surface.51 The light would disappear gradually when air was let into the vessels, but at atmospheric pressure, rubbing (and hence electrifying) the glass would enable it to attract woollen threads, etc. The equipment used in such experiments, consisting of a glass globe mounted on a wooden stand, with a large wheel, is probably the best-known of Hauksbee's contrivances after the air pump.52
Hauksbee also performed a series of experiments in optics, some of them concerning the refraction indices for a variety of substances, as measured by Snell's law. In one such experiment he determined the indices for common air, vacuum and compressed air, using a hollow prism as well as an air pump and a condensing engine. He also developed a special instrument, made up of the prism, a stopcock, a gauge and a telescope.53
Starting in 1706, Hauksbee studied the phenomenon of capillarity. Here he showed that a liquid rises to a certain level in a thin tube, and that a film of liquid between two vertical, wedged plates forms a hyperbola. This was independent of whether the tests were performed in open air or in a receiver exhausted by an air pump.54
Further, in his experiments on sound he performed a series of tests with a bell ringing in a vacuum as well as in ‘condensed’ (compressed) air. He reported that almost no sound could be heard through the vacuum, whereas the bell sounded even more strongly through compressed air than in normal air. To explore this last phenomenon properly, he felt it necessary to avoid the disturbing noises from the city, and once took the equipment to an open field very early in the morning, performing the tests there.55
Life experiments, such as observing the behaviour of small animals in vacuum, were common in the period. However, Hauksbee also performed a series of more advanced tests on the effects produced in air sucked into an evacuated receiver through red-hot substances such as coal and brass. These tests were undertaken for research into the qualities of air for breathing and combustion, tested by putting cats and birds, and also burning candles, into the receiver.56
Many of the air-pump accessories described above were made for demonstrations of known facts. These experiments had first been undertaken, in many cases, by Robert Boyle a generation earlier, and then as part of the investigation of new phenomena. Now, however, the same experiments served as a means of demonstrating and teaching natural philosophy in public lectures and private homes. One among many examples is the ‘drop’ experiment, first performed by Robert Boyle for dropping leaves in an evacuated tall receiver, in an attempt to detect the ‘ether’.57 Hauksbee supplied equipment for a similar test, although now for showing that all objects fall with the same speed in vacuum.
A good overview of air-pump experiments for teaching and rational entertainment appears in the syllabuses of the public lectures given in London in the early eighteenth century. Around 1708, Hauksbee and James Hodgson (1672–1755) issued a pamphlet with a list of 45 pneumatic experiments to be performed at a course given at Hauksbee's house.58 Air is an invisible substance, but the lecturers used the air pump and a series of devices to make its properties undeniable. The syllabus lists experiments for showing ‘the pressure of the air in general’, for showing ‘the pressure of the air in opposition to suction’ and for showing ‘the pressure of air in opposition to the funicular hypothesis’, among others. Further, various experiments for showing ‘the spring of uncompressed air’ were included, as well as some showing ‘that air is not the cause of gravity’ (the drop experiment), and experiments for demonstrating ‘that air is necessary for conservation of life’.
At least eight pumps of Hauksbee design have survived to this day in museums and collections. The indications are that they were made in the period 1705–45, ranging from the elder Hauksbee's first pump in 1705 to those produced at the end of the Hauksbee period by his associates and Hauksbee the younger around 1745. However, as late as 1766 the standing Hauksbee pump was still available for ‘those who would go for the price of it’, although by this time smaller and cheaper pumps had become available (figure 5).59
The air pump now on display in the Royal Society is probably the best-known exemplar. It contains many original components such as the ‘cross-piece’ under the pump, the air-inlet screw and the gauge. In 1926, it was examined in some detail by the scientific instrument maker Robert Paul (1869–1943). After replacing the broken receiver, the gauge tube and the leathers (the valve flaps), Paul reported that it worked properly and gave a vacuum of about 1 inch (33 mbar).60 The same pump has been studied and discussed by others as well. In 1849 George Wilson (1818–59) published an account on ‘the history of the English air pump’, discussing whether or not the pump in the possession of the Royal Society really was made by Hauksbee or whether it might possibly be a ‘relic’ of Robert Hooke.61 Later, in 1940, H. G. Lyons (1864–1944) wrote that the pump, then exhibited in the library of the Society, was made by Francis Hauksbee the younger. It was ordered by Francis Aston (1645–1715) for himself, and purchased by the Society after his death.62 This pump was for a long period on loan to the Science Museum in London, where it was displayed together with the George III Collection of scientific instruments.63
Another well-known exemplar is found in the Museum of the History of Science in Oxford.64 This is a late version, dated to about 1740 on stylistic grounds. Some parts are not original, and it bears a brass plate on the crankcase inscribed repair'd by edwd nairne london.65 The characteristic ‘cross-piece’ has been replaced with a hand-operated valve, probably a three-way switch for connecting the receiver to the pump barrels during pumping or for closing off the receiver for keeping the vacuum, and for letting air into the pump after the conclusion of an experiment. Furthermore, the two valve seats for the self-closing Hauksbee-type valves in the bottom of the cylinders have been replaced with some of a new design, consisting of a hexagonal pattern for allowing a higher throughput of air.
Two other surviving Hauksbee pumps contain a modification contrived by William Vream, one of the associates of Hauksbee the elder. In 1717 he described an air pump with a mechanism that replaced the back-and-forth movement of the crank with a circular movement, causing the two pistons to rise and fall alternately.66 There is one such pump in the Musée des Arts et Métiers in Paris; the other is found in the National Museum of Scotland in Edinburgh. However, the importance of this amendment, made intentionally to increase pumping speed and lessen the shaking of the pump assembly during pumping, is uncertain. On the one in Edinburgh this mechanism has been removed at a later stage and replaced with gears for the traditional back-and-forth movement.67 The appearance of these two pumps, including their woodcarvings, is very similar. Moreover, the one in Paris has an original gauge scale, graduated in inches from 1 to 28 inches, and in tenths of an inch from 28 to 31 inches. Regarding the Edinburgh instrument, there have been unsuccessful attempts to associate it with either William Cullen (1710–90) or Joseph Black (1728–99). Another possible candidate might be the anatomist Alexander Monro (1697–1767). It is conceivable that Monro had learnt about air pumps generally in Francis Hauksbee the younger's and William Whiston's (1667–1752) lectures in London, had studied anatomical preparation techniques by vacuum during his visit to Leiden and Paris in 1718–19, and procured the pump in London in the early 1720s.68
Yet another Hauksbee pump is at Longleat House near Bristol, belonging to the Marquis of Bath.69 It was made by Hauksbee the elder and was acquired by an earlier marquis around 1708. It seems to be in pristine condition, complete with the manometer with mercury in the reservoir, and with a liquid-filled glass vessel for an experiment under the receiver. According to the records, it must at some point have undergone maintenance, because some of the leather gaskets, the valve flaps, the gauge tube and the receiver have been replaced. Such components, however, were regarded as consumables, and certain spare parts came with the delivery of the pump.70
There is also a Hauksbee pump in another private collection. Not much about the background of this one is known, except that it came from a country house in the southwest of England. It is made of walnut, most probably by Hauksbee the elder. It is in excellent condition, having the characteristic ‘cross-piece’ under the receiver plate. Its appearance and physical dimensions are almost identical to those of the pump at Longleat.
Two more pumps are on display in museums: one in the Deutsches Museum in Munich, and the other in the Museo di Storia della Fisica in Padua. There is little information on the German example, although we know that it came to the museum in the early twentieth century from the Royal Prussian Mining Inspectorate of Clausthal.71 It is of a Hauksbee type with the characteristic cross-piece under the pump plate, but it has an additional open/close valve inserted into the connecting tube between the barrels and the receiver. It lacks the traditional wood ornaments, and the gauge is a modern replacement.72
The pump in Padua was bought by the Republic of Venice in 1738–39 for the new chair of experimental philosophy at the university. It had belonged to Cristino Martinelli, a Venetian Patrician, but the maker of the pump and its earlier history are uncertain. It is described as a ‘pneumatic machine in the manner of Hauksbee’, made from walnut and with two bronze cylinders. It clearly resembles a Hauksbee pump, but the wood carvings are of a slightly different style from most of the others. It was modified at some time between 1740 and 1778, because the crosspiece has been replaced with a three-way valve. There are also signs that the cylinders themselves have been changed. The pump was used by Giovanni Poleni (1683–1761) for his new lectures on experimental philosophy.73
In addition to this information about extant pumps, contemporary sources record several buyers and users of Hauksbee's larger pumps, all of them now lost or in unknown locations. How these pumps came to be lost is unclear, and we know little about them. It is probable, however, that some of them were identical to those mentioned above.
Hauksbee the elder made several mentions of a Dr Pratt, who bought a fully equipped air pump for £25 at some date before January 1706/7.74 Later Hauksbee complained that Dr Pratt had not been ‘so good a husband in keeping it as he ought to be’, indicating that dust and neglect had deteriorated both the appearance and the functions of the pump.75
From a letter from Stephen Gray in Philosophical Transactions we know that a larger Hauksbee pump was in the possession of a Mr Wheeler in the 1730s. In the letter, Gray describes a series of electrical experiments, some of them on the attraction of bodies in a vacuum.76 He also records that Mr Wheeler had attempted to perform some experiments on electricity in a vacuum himself, and that to prevent ‘steams’ from the wet leather gasket from reducing the attractive forces, he had sealed the receiver with cement.
In addition, ‘a large double barrelled Air Pump, with a Mercurial Gage’, possibly used by William Cullen, is mentioned in the Glasgow University inventory of 1727.77
We also know that a Hauksbee pump was procured by Erik Benzelius the younger (1675–1743) for the University Library in Uppsala in 1725. This was prompted by recommendations from Emanuel Swedenborg (1688–1772), who had met Hauksbee the elder in London in 1712.78 Another air pump appears in the background of a portrait of Mårten Triewald (1691–1747).79 In the portrait, Triewald points to the bird in the receiver on top of the pump structure, which resembles Hauksbee's model, having similar ornaments and columns. However, the top plate seems in this case to be rectangular, whereas the top plate of Hauksbee's pump was oval. The pump in the painting might have been the one that Benzelius bought, although it could equally have been another one, as Triewald took a series of instruments with him from England in 1726.80
From letters between Francis Hauksbee and Samuel Molyneux we know quite a lot about an air pump that was sent to Dublin in March 1706/7.81 It was described as a double-barrelled air pump of the ‘larger sort’, with a receiver that could ‘hold a cat or any animal of a larger sort’. The pump came with about 20 different attachments for various air-pump experiments, as well as instructions for unpacking, assembly and use. The price, including the apparatus, spare parts and packing and shipping to Dublin, is given as £26 8s.
One Hauksbee pump, or part of it, was immortalized by Joseph Wright of Derby. In his famous painting An Experiment on a Bird in the Air Pump of 1768, the air pump consists of two parts.82 The spherical receiver on top of the column, holding the bird, is modelled on Boyle's first air pump. However, the section with the pump itself, with the two brass barrels, cranks and ornaments, is clearly taken from Hauksbee's classic, double-barrelled air pump.83
In most historical accounts, Hauksbee's double-barrelled air pump has been primarily, and quite understandably, studied and appreciated for its appearance. But if the pump structure with its beautiful ornaments and shining brass parts necessarily catches the attention, so also should the technology. Once the technical details are entered into, a very sophisticated and well-thought-out design becomes evident. Another aspect that emerges is how ‘modern’ the pump is. If we view Hauksbee's double-barrelled air pump not simply as a pump, but also as a system consisting of several components (that is, as a vacuum system), we find four parts in addition to the wooden structure: the air pump itself (the two cylinders with pistons, valves and the crank), the receiver, the measuring gauge, and the air-inlet function. Today, all these elements are found in various forms in modern vacuum systems.84 One difference is that modern systems normally have a valve inserted between the pump and the vacuum chamber, making it possible to isolate the chamber from the pump and ‘save’ the vacuum. However, in some Hauksbee pumps such a function was added at a later stage, for example to the pump now in the Deutsches Museum.
Hauksbee's system made air-pump experiments a one-man task. It also reduced the pumping time considerably, since no pauses were required between the plunger strokes to operate the stopcocks. This decrease in the need for human labour and time must clearly be one of the ‘many conveniences’ that Harris and other contemporaries referred to.
The air pump was Hauksbee's most important instrument and crucial for his research and activity as a lecturer. Although most of his experiments were in pneumatics, he also investigated phenomena that required an air pump only in particular phases of the experiment. This was a great leap forward from the earlier generation of British air pumps. In the case of Boyle, people at distant locations who had been able only to read his publications never succeeded in replicating his experiments; there was no alternative to first watching the experiments being conducted with Boyle's original equipment.85 With Hauksbee, the technology had become refined, and most of the skills and manipulation techniques that had once been indispensable for the operation of an air pump had now been ‘built into’ the machine.
I thank friends and colleagues at the Museum of the History of Science in Oxford, UK, and the Department of History and Religious Studies, University of Tromsø, Norway, for many interesting discussions and helpful suggestions during this work. Special thanks are due to Richard Holt for reading and checking the manuscript. This study was made possible by private funding, and support from the Scientific Instrument Society (SIS).
↵1 T. Brundtland, ‘From medicine to natural philosophy: Francis Hauksbee's way to the air-pump’, Br. J. Hist. Sci. 41, 209–240 (2008).
↵2 H. Guerlac, ‘Sir Isaac and the ingenious Mr. Hauksbee’, in Mélanges Alexander Koyré: l'aventure de la science (ed. B. Cohen and R. Taton), pp. 229–253 (Hermann, Paris, 1964).
↵3 Royal Society Journal Book, JBC 10, 15 December 1703 (London, 1702–14).
↵4 The Post Man, 31 Jan 1699 and 27 Oct 1702. All newspapers were published in London.
↵5 The Flying Post, 9 May 1699.
↵6 In the period 1690–1700, about 20% of the elected members belonged to this group. See M. Hunter, The Royal Society and its Fellows, 1660–1700 (British Society for the History of Science, Oxford, 1994).
↵7 S. Pumfrey, ‘Hauksbee, Francis (bap. 1660, d. 1713)’ in Oxford dictionary of national biography (ed. H. Matthew et al.) (Oxford University Press, 2004).
↵8 F. Hauksbee, Physico-Mechanical Experiments on Various Subjects (Senex & Taylor, London, 1709; 2nd edn 1719).
↵9 Robert Boyle had recognized that the vacuum created in air pumps was not perfect. To distinguish this vacuum from the philosophical idea of the void, he defined it as a space ‘almost totally devoid of air’. ‘Vacuo Boyliano’ was thus ‘absence of common air’, the condition produced in an air pump. See The works of Robert Boyle (ed. M. Hunter and E. B. Davis) vol. 1, p. 163, and vol. 6, p. 216 (Pickering & Chatto, London, 1999).
↵10 G. Kepar, The Gardener's Almanack for the Year of our Lord 1702 (The Company of Stationers, London, 1702); G. Parker, A Double Ephemeris for Year of our Lord 1703 (The Company of Stationers, London, 1703); Old Bailey Proceedings advertisements (London, 24 May 1699 and 15 January 1700).
↵11 Letter from F. Hauksbee to S. Molyneux (London, 31 July 1707), in ‘Letter Book of Samuel Molyneux’, Southampton Archives Services D/M 1/2, Southampton.
↵12 Brundtland, op. cit. (note 1).
↵13 J. Harris, ‘Air-pumps’, Lexicon Technicum, 1st edn (Dan. Brown, London, 1704).
↵14 W. Derham, ‘Experiments about the Motion of Pendulums in Vacuo’, Phil. Trans. R. Soc. Lond. 24, 1785–1789 (1704–05), at p. 1786.
↵15 The Daily Courant, 9 December 1704.
↵16 The Post Man, 27 October 1702.
↵17 Harris, op. cit. (note 13), ‘Plate with air-pumps’.
↵18 The Post Man, 1 May 1705.
↵19 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux, ‘A Description of the Air Pump’ (London, spring 1707).
↵20 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux (London, 27 February 1706/7).
↵21 S. Shapin and S. Schaffer, Leviathan and the air-pump (Princeton University Press, 1985), pp. 116–117.
↵22 Boyle, op. cit. (note 9), vol. 6, pp. 115, 133 and 141.
↵23 Hauksbee, op. cit. (note 8), p. 3.
↵24 Boyle, op. cit. (note 9), vol. 1, p. 157 (the first air-pump); and vol. 6, p. 36 (the second pump). With Boyle's second pump, a deviation from the right sequence would result in the receiver's being filled with water.
↵25 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux, ‘A Description of the Air Pump’ (London, spring 1707).
↵26 F. Hauksbee, The Post Man, 27 September 1705.
↵27 Boyle, op. cit. (note 9), vol. 1, pp. 192–201. See J. B. Conant, ‘Robert Boyle's experiments in pneumatics’, in Harvard case histories in experimental science (Harvard University Press, Cambridge, MA, 1957), vol. 1, pp. 17–30.
↵28 Boyle, op. cit. (note 9), vol. 6, pp. 135–136.
↵29 Hauksbee, op. cit. (note 8), p. 4.
↵30 According to K. Middleton, the idea of applying a floating index to keep the mercury surface at the zero point was first suggested in the 1770s. See W. Middleton, The history of the barometer (Baros Books, Baltimore, MD, 1964), p. 203.
↵31 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux (London, 27 February 1706/7).
↵32 The usefulness of Hauksbee's built-in barometer was confirmed by Jan van Musschenbroek in Holland in 1711. According to Z. Uffenbach, Musschenbroek planned to supply his own pumps with such ‘tubo mercurii’. Z. Uffenbach, Merkwürdige Reisen durch Niedersachsen Hoölland und Engelland (Ulm, 1754), p. 430.
↵33 Hauksbee, op. cit. (note 8), p. 3.
↵34 The concepts of pumping capacity and pumping speed are here based on manually driven piston pumps only.
↵35 A barrel of this width (5 cm) was too wide to be pulled directly, because 20 kg would have been required to overcome the external air pressure. However, because the two barrels were interconnected, the atmospheric pressure on the pistons would balance each other out.
↵36 R. Cotes, Hydrostatical and Pneumatical lectures (ed. R. Smith) (J. Bentham, Cambridge, 1747), pp. 173–178. Written between 1713 and 1716.
↵37 A. C. van Helden, ‘The age of the air-pump’, Tractrix 3, 149–172 (1991), at p. 169.
↵38 Hauksbee did not mention this problem explicitly, but the corresponding issue with condensing pumps, in which the problem is about the maximum pressure that can be obtained in a closed container, became thoroughly discussed. See F. Hauksbee, ‘An account of an Experiment … upon the Propagation of Sound in Condensed Air’, Phil. Trans. R. Soc. Lond. 24, 1902–1904 (1704–05).
↵39 Hauksbee never mentioned this issue in his writings, but in a sales advertisement by his wife Mary after his death she claimed that Richer Bridger, Hauksbee's apprentice (and her son-in- law) built pumps with a mechanism that overcame this particular problem. It is not known what this contrivance consisted of. See The Englishman, 21 January 1714.
↵40 A good description of the growing attention to this phenomenon is found in C. Hutton, A Mathematical and Philosophical Dictionary, vol. 1, pp. 56–57 (J. Johnson, London, 1796).
↵41 A. Roth, Vacuum technology (North-Holland, Amsterdam, 1976), p. 19.
↵42 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux, ‘A Description of the Air Pump’ (London, spring 1707).
↵43 The Post Man, 27 October 1702.
↵44 Hauksbee, op. cit. (note 8), p. 21.
↵45 According to T. Pinch, black-boxing of instruments occurs when ‘the social struggle over a piece of knowledge has become embedded in a piece of apparatus’. One consequence of black-boxing is simply a lower price and an increased availability of the instrument. See T. Pinch, Confronting nature (D. Reidel, Dordrecht, 1986), p. 214.
↵46 Royal Society, op. cit. (note 3), 22 May 1705. See also P. de Clercq, At the sign of the oriental lamp. The Musschenbroek workshop in Leiden, 1660–1750 (Erasmus Publishing, Rotterdam, 1997), p. 257.
↵47 F. Hauksbee, ‘An Account of an Experiment … Touching the Difficulty of separating two Hemispheres’, Phil. Trans. R. Soc. Lond. 25, 2415–2417 (1706–07).
↵48 Royal Society, op. cit. (note 3), 3 April 1712, 24 April 1712 and 5 April 1711.
↵49 Based on a survey of his articles and book, and relevant entries in the Journal Book.
↵50 T. Brundtland, ‘After Boyle and the Leviathan: the second generation of British air pumps’, Ann. Sci. 68, 93–124 (2011); G. Freudenthal, ‘Electricity between chemistry and physics: the simultaneous itineraries of Francis Hauksbee, Samuel Wall, and Pierre Polinière’, Hist. Stud. Phys. Sci. 11, 203–229 (1981); J. Heilbron, Electricity in the 17th and 18th centuries: a study of early modern physics (Dover, New York, 1979), pp. 229–239; W. Hackmann, Electricity from glass (Sijthoff & Noordhoff, Alphen aan der Rijn, 1978), pp. 29–39; H. Guerlac, ‘Newton's optical aether’, Notes Rec. R. Soc. Lond. 22, 45–57 (1967); R. W. Home, ‘Francis Hauksbee's theory of electricity’, Arch. Hist. Exact Sci. 4, 203–217 (1967).
↵51 Hauksbee, op. cit. (note 8), pp. 45 and 65.
↵52 Brundtland, op. cit. (note 50), pp. 115–117. A replica of Hauksbee's rotating globe machine, made by the author, is now on display in the Library of the Royal Society.
↵53 Hauksbee, op. cit. (note 8), pp. 225–230; Brundtland, op. cit. (note 50), pp. 117–118.
↵54 Hauksbee, op. cit. (note 8), pp. 177–181.
↵55 Hauksbee, op. cit. (note 38). The experiment was performed at the White Conduit in Islington, London.
↵56 Hauksbee, op. cit. (note 8), pp. 282–288.
↵57 Boyle, op. cit. (note 9), vol. 6, pp. 135–136.
↵58 F. Hauksbee Sr and J. Hodgson, Hydrostatical and Pneumatical experiments (London, n.d. (ca. 1708)).
↵59 B. Martin, The Description and Use of a New, Portable, Table Air-pump (the author, London, 1766), pp. 1–2.
↵60 E. N. Andrade, ‘A lecture with experiments on various subjects’, J. Scient. Instrum. 4, 137 (February 1927).
↵61 G. Wilson, ‘On the Early History of the Air-pumps in England’, Edinb. New Phil. J. 46, 330–354 (1849).
↵62 H. G. Lyons, ‘Francis Aston (1645–1715)’, Notes Rec. R. Soc. Lond. 3, 88–92 (1940).
↵63 The pump was inventoried as part of the Science Museum's Vacuum Technology collection. See A. Morton and J. Wess, Public and private science: the King George III Collection (Oxford University Press, 1993), p. 43, n. 19.
↵64 Inventory number 60664.
↵65 Edward Nairne (1726–1806), English instrument maker, London.
↵66 W. Vream, A Description of the Air-Pump (J. H., London, 1717). An illustration of this mechanism is given in J. Desaguliers, A Course of Experimental Philosophy, vol. 2, pl. XXIV, fig. 2 (Innys & Manby, London, 1744).
↵67 R. G. W. Anderson, The Playfair Collection (Royal Scottish Museum, Edinburgh, 1979), pp. 67–70.
↵68 For Monro's use of air pumps, see Brundtland, op. cit. (note 50), pp. 111–112.
↵69 Collection no. 998, Longleat Historic Collection, Longleat, Warminster, Wiltshire, UK. Information from K. Harris (personal communication, 10 February 2001). See also M. Holbrook et al., Science preserved (Science Museum, London, 1992), p. 148.
↵70 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux, 9 January 1706/7.
↵71 Königliche Preussische Berginspection, Clausthal, Germany.
↵72 Inventory no. 3799.
↵73 Museo di Storia della Fisica, Padua, Italy. Information from S. Talas (personal communication, 21 October 2011).
↵74 Presumably to be identified as Benjamin Pratt (1669–1721), one of the founders of the Dublin Philosophical Society in 1683. See K. T. Hoppen, ‘The Royal Society of Ireland. II’, Notes Rec. R. Soc. Lond. 20, 78–99 (London, 1965), at pp. 78, 81 and 88.
↵75 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux, 31 July 1707.
↵76 S. Gray, ‘Two Letters from Mr. Stephen Gray’, Phil. Trans. R. Soc. Lond. 37, 397–407 (1731–32).
↵77 Glasgow University Archives, MS 5291. See Anderson, op. cit. (note 67), pp. 69–70.
↵78 S. Lindqvist, Technology on Trial (Almqvist & Wiksell, Uppsala, 1984), p. 162.
↵79 Mårten Triewald. Painting attributed to Georg E. Schröder. Swedish Academy of Science, Stockholm.
↵80 Lindqvist, op. cit. (note 78), p. 212.
↵81 Hauksbee, op. cit. (note 11), Hauksbee to Molyneux (London, 27 February 1706/7).
↵82 National Gallery, London, NG725.
↵83 Brundtland, op. cit. (note 50), p. 108, n. 50.
↵84 A. Roth, op. cit. (note 41), p. 122.
↵85 Shapin and Schaffer, op. cit. (note 21), pp. 229–230.
- This Journal is © 2012 The Royal Society