The basic truth that there was an order for substantial quantities of iron pipe for Paris is thus not in doubt (and never was); but this has not prevented repetition of the story of illicit export of "pipes" for use as gun-metal. W.H.Chaloner, remarks on this in John Wilkinson, Ironmaster, in History Today, Vol 1, May 1951, pp 66-7. In a subsequent paper in Business History, 1964, he blames Randall for the perpetuation of the stories [p 79], but also cites a contemporary accusation from March 1782 that John Wilkinson exported 24,000 tons of iron in six weeks, consisting of cannon balls and cylinders of unusual thickness [1964, pp 91-2] - clearly nonsense.

Nonetheless, it can be shown that Wilkinson was actively conniving at evading export controls on iron, lead and copper for the same scheme, and indeed proposed casting the lead as thick pipes to that end [17 May 1779, below]. If such intentions had become known, it is not surprising that adverse stories circulated. The events of this episode occurred mostly between 1777 and 1782 (peace with France only followed in the interval 1783-1793). During this period the Channel was a battleground, from the day when the Arethusa challenged the Belle Poule, 17 June 1778. War with France had been brewing since the American declaration of independence in 1776; American privateers had already become a major nuisance by the summer of 1778. There was a considerable impact on Wilkinson's business in general, as his coast-wise transport was disrupted, by shortage of seamen, from the activities of the press, and action by privateers in the Irish Sea and Channel. Surprising as these exports may seem after experience of total war in this century, Wilkinson and Boulton and Watt obtained legal passports from governments for the export of engineering components for the Paris scheme. The mechanism for so doing was known to them, though it apparently took much longer than the corresponding French passport, to Périer's irritation. Letters took very little time to cross the Channel: one letter dated 1 September 1779 from Boulton and Watt, received a reply dated 9 September. It was not just letters and materials that crossed the Channel: Périer himself was openly travelling in Britain in late 1778. William Wilkinson was in France throughout this period.

It has long seemed to me that the easiest way to demonstrate the truth of the matter might be to establish what length of water pipe was actually laid in Paris. As an aside, that basic information might eventually lead to examples being recovered. (There is a good chance that some of them are still in use, even.) While I am aware that there are potentially important sources in England and in France that I have not seen, it does begin to look as though the forty miles of pipe may be a myth: six miles might be nearer the mark, in the event. That does not detract from what has been a fascinating search, yielding fruit of a different kind.

As far as I am aware, little attempt had ever been made to pursue the topic of this Parisian work in English, before Chaloner's paper of 1964, which is primarily concerned with the business aspects of the iron trade during the war. Dickinson's two paragraphs were as much as appeared substantively, outside Muirhead's biography of Watt. Thus:

This paper is an attempt to describe the background to the scheme, and will show that Wilkinson's pipes were laid throughout the heart of old Paris. The scheme itself, in which the Wilkinsons were shareholders, does not appear to have been a financial success, and it was overtaken by the French Revolution of 1789. The story will be told as far as it can be taken without protracted work on the Boulton and Watt archive, and research in France: the bones of the matter do however emerge from published material available in England. Most references to Watt's correspondence in this paper come from published sources, excepting a few kindly supplied by Douglas Braid, or by Ann Williams at Bersham. They are cited by source, date and correspondents in the text. A key to these is given in the Appendix. It should be noted that each published collection is a mere selection, with surprisingly little overlap. Muirhead for example did not transcribe the whole of each letter [Vol I, p xiii]; there is probably more detail available to anyone with opportunity to access the archive directly.

By the late 18th century the water supply problems of Paris were critical - London was not much better served, nor any large conurbation of the time, of course. The average supply in Paris at this time was variously estimated at 5-10 litres per head per day, probably less than half that of London [Goubert, p 51]. The population had risen to approximately 800,000 [Vachette, p 2], but the sources of drinking water were effectively those of the mediaeval town. Even from these wells and aqueducts, "drinking water" was probably a euphemism, by Victorian standards, let alone those of the EC, backed by undreamed-of analyses. Such water was distributed at public fountains, and by carrier, at a price. The Seine itself provided much of the volume in daily use, from a variety of hydraulic machines powered mostly by the flow of the river itself. Unfortunately, much of this came from downstream of the outfalls of major sewers of the city. (A later estimate is that 2-3% of water abstracted from the Seine was sewage). One of the pressing problems was for water for fire-fighting: indeed the Périer scheme followed shortly after an event which caused great concern, when a hospital for contagious diseases (Hôtel Dieu) was burned down in 1773.

The great scientists and engineers of France, such as Perronet, Chezy and Bossut were engaged on schemes to solve the problems, by bringing clean water from afar. At this juncture, the streams of the Bièvre and Yvette were to be tapped, and brought to Paris by aqueduct. The difficulty of course was the projected cost, which the administration of the town could not finance. (A channel was to be made 17,352 toises [33.82 km] long, to deliver 1500 pouces de fontenier [say 30.3 megalitres per day], at an estimated cost of 7,862,000 livres [about £344,000])

Pouce de fontenier, incidentally, is an old measure, developed from fountain design: the discharge of a circular orifice of one pouce diameter, immersed by one ligne. There is some difficulty in reconciling a modern calculation of that discharge with contemporary definitions. Not only is such a slight immersion non-standard, but units such as livre and pouce varied locally at the time. I have used a figure of 3.087 Imperial gallons per minute; the unit is certainly close to 3 gpm. Fontenier (spelling varies) is of course the name of the turn-cock who operated fountains, and also the valves of the city water supplies, the modern English waterman.

Périer's scheme was only one of several alternatives afoot in the late 1770's, but all of these were recognised as stop-gaps at the time. Neither in volume nor water quality could they compete with the grand schemes projected from the 1760's. The quality aspect may have lain directly behind the financial failure of the Périer scheme, together with a gross under-estimate of the real costs of establishing a water supply across a city, to achieve the promise of adequate water, on demand, in private houses.

There are two contemporary sources which I have been able to locate and use. The Périer Prospectus is dated 1781 and was thus issued to attract customers shortly before the scheme started to deliver water [Prospectus de la furniture et distribution des eaux de la Seine á Paris, par les machines à feu, Paris, 1781. British Library ref RB 23a 1748; also in B&W collection]. There was a shorter Prospectus at an earlier stage. A longer pamphlet was issued by a displaced competitor, the Vachettes, and provides an alternative view of matters [Vachette, Precis historiques sur l'établissement des pompes à feu, etc, par les Srs Vachette, frères, Paris, 1791 - British Library ref F27*(18)]. It is perfectly clear that neither source is trustworthy in isolation, though there is engineering information in both which is probably not contentious. Between them, they provide considerable material about how such schemes functionned. Any water supply scheme is specific to its sources and supply area, but the London schemes which Périer used as an exemplar (at least in his Prospectus) and many later schemes (Périer was expanding to Lyons concurrently, for example), would be essentially similar.

We know many of the details of the Périer scheme through the efforts of Girard, who undertook the management of the Paris waterworks from 1807, including the construction of the Canal de L'Ourcq, which fulfilled for a while the promise of the Yvette scheme, supplying two-thirds of Paris' needs in 1860. [Pierre Simon Girard, Recherches sur les Eaux Publiques de Paris ....., Paris, 1812; summarised in William Humber, The Water Supply of Cities and Towns, London, 1876, pp 7-8]. Girard also wrote studies of the Canal de L'Ourcq, and its distribution, dated 1808, 1819, 1831 and 1845. By 1812 he was Chief Engineer of the Ponts et Chaussées, as well as Director of the Canal de L'Ourcq, and of the Eaux de Paris.

Further details of the Paris supply systems are available in Burnell's paper of 1866, which describes the increasing demands for purity (no longer met by the open reservoirs and navigable channels of the L'Ourcq scheme) and volume, and the progression to longer closed aqueducts and higher standards familiar in Victorian Britain too. [G.R.Burnell, On the Water Supply of the City of Paris, in Minutes of Proceedings of the Institution of Civil Engineers, Vol XXV, London, 1866]. Even in the new schemes about 1866, none of the Parisian water was filtered, other than in very small quantities, commercially, or domestically: no wonder the French are so fond of bottled water.

The Périer scheme (actually that of the brothers Jacques-Constantin and Augustin-Charles Périer des Garennes, mécaniciens - references are invariably to Jacques) was in direct competition with two others. M.Capron had previously proposed a very similar scheme, but dependent on river-driven hydraulic machines. M d'Auxiron was claiming priority as the first to suggest the use of pompes à feu to raise water from the Seine. The Secretary of State, Amelot, referred the whole issue to the Bureau de la Ville on 17 August 1776. A reply was made by the Prévot des Marchands on 25 October 1776. This declared that the city was still unable to undertake even the 450 pouce scheme from the Bièvre river which it preferred, and could not renounce as an option for happier times. It demanded an indemnity against any claims by the successful entrepreneur, should the Yvette river scheme ever be implemented, too. Capron's scheme was not preferred, because of the obstruction to navigation which would be caused by his mill-driven pumps. Neither d'Auxiron nor Périer could claim the invention of the fire-engine pump; and they determined to recommend the scheme that gave most benefit to the city. Périer won easily, because he proposed to bear all the costs, while d'Auxiron demanded that the materials of the existing pump at the Pont Notre Dame and all the distribution pipes, which the Bureau valued at 1,500,000 livres, should be surrendered to him.

Matters now proceeded to Letters Patent, issued on 7 February 1777, which were registered with Parliament on 16 July 1778. The terms of the arrêt are given in full by Girard [1812, pp 255-7], and allow the formation of a company to execute the works, with certain conditions attached. In particular, Périer obtained the 15 year privilege he sought, to exclude all other schemes employing steam pumps. He could not prevent any other scheme, nor had he any recourse if the Bièvre or Yvette schemes were resurrected in competition.

There were other schemes promoted at this time. Desforges proposed in 1777 to establish a supply of "pure and salubrious" water directly from the Seine adjacent the Arsenal (i.e. from upstream of most of the existing supplies), using hydraulic machines, placed on a new stone bridge, in place of those of Notre Dame. He proposed in addition to replace the ancient lead distribution pipes with cast iron; all at an estimated cost of 6,000,000 livres. He intended that the sale of the water at public fountains at 3 deniers la voie should contribute to the cost of a hospital, but this did not impress the Bureau, who pointed out that the new site offered little advantage, and while the scheme was cheaper than that of the Yvette, it would produce only one tenth of the flow.

Another private scheme of 1780, rejected by the Lieutenant of Police as an obstruction to navigation, perhaps indicates the desperate state of water supplies in Paris: it was for pumps à chapelet - chain pumps, to be operated by men, to supply the Place de la Bastille.

The more interesting proposal was made in 1781, by an engineer, Bourbon de Charancourt. This was for 18 fontaines épuratoires to be installed on the banks of the river, supplied by boat-mills, and he required exclusive rights of sale of filtered water, at 3 deniers la voie. This was initially rejected, because of the impact on both navigation and road traffic at so many heavily frequented points, and because Charancourt had made a secret of his process, which could not be judged. He soon resubmitted it, with proof of the process attested by Lassone at Versailles (.."the most miry [fangeuse] water made limpid"..), and now demanded 30 years privilege for his filters. It is perhaps significant, in relation to rising expectations and to Vachette's subsequent complaints about the Périer scheme, that the Bureau felt obliged to allow a trial, on this new evidence; but they limited him to six filter "fountains" at three sites; granted 17 May 1782.

To put the question of filtration into perspective, it was not actually a new idea. There were attempts to introduce filtration for military purposes at least as early as 1685, by Luc Antonio Porzio; in the 18th century, the Frenchman Joseph Amy described such semi-portable systems. In England, a Patent was issued in 1791 to J.Peacock, and filters were constructed for the public supply to Paisley in 1804. Chelsea Waterworks followed in 1827, and all surface waters for public supply were sand-filtered from the 1840's, after legislation. In fact Goubert gives further information (unsourced) about the early use of sand filters in Paris: two fountains were placed on the banks of the Seine in 1771, but had a capacity of only 12 cubic metres per day. He also remarks that Admiralties were experimenting from about 1780 with early charcoal filters for supplies to ships [pp 39, 53].

Vachette refers to a number of other early schemes, intending to belittle the novelty of the Périer scheme, commencing with that of Orry in 1740, to import English steam pumps for a site at l'Estrapade, which was thwarted by the outbreak of war in 1741. D'Auxiron is said to have published a pamphlet in 1765, Sur les eaux de Paris, and to have refuted the benefits of Parcieux's Yvette scheme in 1769, proposing instead a fire-engine located upstream of Paris. (Aqueducts were considered relatively vulnerable to prolonged drought or frost, and to the need for repairs, in Périer's view). Berthier proposed in 1769 a hydraulic machine at the Ile Saint-Loius with a lift of 100 pieds to distribute to the whole of Paris.

The brothers proposed to form a company to undertake the works proposed, in return for the right to sell water both at public fountains, and to private consumers. The arrêt repeats the request of the brothers for powers to carry out the works, and adds clauses of its own. Thus while the Périers wanted power to lay pipes under the pavé, manholes, and other things necessary to establish the steam pumps, the grant is additionally for suction points, valves, and all other sorts of construction necessary for the perfection of the establishment. Perhaps Watt's view of the Périers' scheme and knowledge of water works was justified [2 May 1777, see below].

The decree granted an exclusive privilege for the use of steam pumps for 15 years, but no other commercial protection at all; and stipulated that if a flow of 150 pouces was not being delivered within three years from the date of the decree (7 February 1777), the whole was null and void. It repeats that the whole was to be at the undertakers' expense: and at "their risk, peril and fortune". Another critical factor was that the locations of pumps and pipes was to be agreed with the Prévôt: this was to be a major cause of delay, apparently.

The Company was formed on 27 August 1778, in fact, shortly after the decree was formally registered in parliament. It divided a share capital of 1,440,000 livres between 1200 shares, reserving the right to issue more capital, for anticipated extensions after a few years operation. The Périers were to receive 20,000 livres a year, as "perpetual administrators", and five other shareholders were to direct the operations of the Company (the Compagnie des Eaux de Paris).

Watt has a slightly less enthusiastic account of events at this period, and clearly did not have a high opinion of Périer.

Watt to Boulton, 2 May 1777 [Muirhead].

(Clearly, the illicit sales of Watt engine parts by Wilkinson, which provoked an inquiry by 1794 [Chaloner, 1951, p 69] had not then commenced.)

Watt to Dr Black, 13 January 1779 [Muirhead].

Watt to Boulton, 4 May 1780 [Muirhead].

In the aftermath, and after a visit to Paris, in connection with an engine for Marly, Watt wrote to Dr Roebuck, on 3 February 1787 [Muirhead]:

There were actually considerable delays in commencing work, while the Company argued into 1780 with the Prévôt about who should determine matters concerning the distribution pipes and maintenance of the public fountains; and who should resolve complaints about the quality of the water. (The penalty clause must have been relaxed accordingly: no water was delivered until July 1782.)

The works constructed started at Chaillot, where a canal 7 pieds wide was formed under the Versailles road leading to the pumps. There were two engines here, rated in Périer's Prospectus at 48,600 muids per 24 hours [13.35 megalitres per day], delivered to four reservoirs at 110 pieds above the Seine. From these, it was distributed in a cast iron pipe of one foot diameter along the faubourg Saint Honoré, and then branched out along the principal roads. The pumping machinery features in Prony's Nouvelle Architecture Hydraulique of 1790/6 [Vol II, Plates 38/9]. These drawings, several views of the pumping stations, a portrait of Périer, and a model of his wooden pipe boring engine from Chaillot, are reproduced in J.Payen, Capital et machines à vapeur au XVIII^e siècle, Paris 1969.

The distribution system spread successively from a fountain at the Porte Saint Honoré in July 1782, to others at Chaussée d'Antin, Porte Saint Denis, and the entry to the Rue du Temple. Another pump was installed on the opposite bank at Gros-Caillou to serve the faubourg Saint Germain; then a third and fourth above Paris, near the ancient Gare de l'Hôpital, and opposite it at the Arsenal.

The Company was obliged to erect public fountains, where carriers drew their supplies too, and also made private house connections. The terms were for 3, 6 or 9 years, at 50 livres per annum for each muid supplied in 24 hours. The Company provided the necessary pipes in return for a double premium for the first year's supply. However, to implement all these connections, the Company had to purchase and lay pipes of both wood and lead, to extend the main cast iron network, and it seems that the costs may have been under-estimated.

Certainly the share capital had to be dramatically increased as the system was extended. In December 1781, another 600 shares of 1200 livres; in August 1784, 2200 shares of 1200 livres; and another 1000 shares of 4000 livres each in July 1786. By this time the issue had reached 8,800,000 livres, and questions were being asked about how the situation had arisen. A pamphlet war ensued, in which one of the principal opponents was Vachette. Vachette had obtained the privilege of building fountains in 1771, using horse-pumps à manége, but was in direct competition with the new Company. The richer new company bought the Vachette company for 150,000 livres in 1785, as both companies found it a nuisance to be in competition [Vachette].

The peak returns for the Company are said to have been reached in 1786, when 45,883 livres were received from sales to some 20,000 private houses, and another 66,278 livres from sales at public fountains. It is suggested that the projectors had over-estimated the income to be expected from private connections. It is at least a possibility that the quality of water was by then a greater issue: Charancourt's scheme must have raised awareness of the problem; though if Vachette is to be believed, the problems were acutely self-evident.

The Company was in a state of financial discredit by 1787, and had indeed raised 100 new shares in January 1787 to pay for dividends. By early 1788, one banker held most of the stock, at a loss, and an attempt was made to sell the undertaking to the city, sanctionned by a general meeting of the shareholders. In the meantime, the city had however been making moves to implement the Yvette/Bièvres schemes. On 18 April 1788, a new administration was in place, called the Administration Royale des Eaux de Paris.

Chaillot, the first site to be constructed, had two Watt engines, imported from England. These were named Augustine and Clémentine and ran until 1852/3 [Payen pp 107, 135]. They had pump cylinders of 26 pouces diameter, nominally [27.7 inches: in fact 28 inches] and a stroke of 8.88 feet, rated at 10 strokes per minute. There is creative arithmetic in some of Périer's statistics at this point, and Vachette claims that even the Watt pumps (which he indicates may have been altered by Périer) never achieved their rating. In summer they only reached six strokes per minute, and in winter never more than eight. The fuel cost was vastly greater than anticipated (but only half that of the subsequent Périer engines). These pumps could in necessity operate together via a 24 inch delivery main. They drew their water from a basin lined with cut-stone, dug three feet below low water level, and fed from the river via a short canal. Périer makes a point of stating that contrary to rumour the intake was well above a noxious drain at Chaillot, and well below the drains of the Place de Louis XV [pp 9-10]. Vachette is not impressed, and even more concerned that it only drew from the side of the river: in low flow conditions only the centre of the river was salubrious [p 26]. When the river was low, the basin acted partly as a well, and the hardness of water supplied increased markedly.

The four reservoirs were each 58.4 x 19.5 x 2.9 metres, holding in total the daily rated flow of 48,600 muids. Périer emphasises that the intention was for one to be filling, one settling, and one in service, with one in reserve at any time, and that the whole could be bypassed if necessary. He had them placed in cascade in 2 foot steps to assist in cleaning. There are contradictions here, and Vachette indicates that in the summer it was necessary to clean the lot every two weeks, and that this could only be done by flushing from the highest to the lowest, and draining the sludge down the rising main. The final 58 metres of the pump branch could not be drained, and all the sludge that entered it was simply pumped back to the reservoirs. The basin collected 3 or 4 feet of silt every year, and was cleaned with great difficulty and expense, requiring low river conditions and cessation of pumping.

The distribution mains were of 12 inch bore, according to English sources, but Périer claimed one pied [12.789 inches]. This discrepancy will only be resolved with a sample from Paris. There were smaller iron pipes to feed particular areas, possibly 6 inch bore. It is clear that in practice much of the smaller pipework was in wood - Vachette even says that was the original intention - with individual service pipes in lead. This use of wood may explain why Périer only offered 12-15 feet pressure at pavement level in most areas. As in London, the "service" was intermittent, under the control of turn-cocks, who opened valves to small areas in turn, for a matter of minutes, to refill private reservoirs, on every second day. This was said to allow the company two days to repair any burst pipe, without loss of supply to customers.

The Company provided a tariff for internal lead pipework; for reservoirs (sized for twice the daily subscription) of either simple wood, or lead-lined; and if the customer wished to avoid the inconvenience of overflows inside the house, float-operated valves in their reservoirs. They could also purchase filtering tanks for that part of the supply intended for the table; and [p 18]

This last might relate to an early form of hydraulic ram, though not at this date a self-acting one. Such devices were introduced in 1775 by Whitehurst; but I know of no references for this period to domestic use of rams, or of water-pressure engines, or force-pumps driven by water wheels, and so cannot identify the type. (Payen notes [p 38] the existence of a MS by Périer on the hydraulic ram, dated 1810). If the tanks were filled in minutes, as claimed, and without excessive waste, then the quantity raised must have been very small, with such a low driving head, and manual operation of a ram.

For large quantities of supply, and for customers close to the main iron network, a "high service" could be provided to reach the top of any house.

Such was the profusion of water anticipated by Périer that he proposed to provide at intervals in all streets served, in recesses in walls, a valve and leather hose for fire-fighting, capable of producing a 40-50 foot jet in most parts. He also proposed to sprinkle the streets in dry dusty periods, and flush snow and slush into the drains, whenever the Government considered it necessary. All in all, he would eliminate the foul air of Paris, and throat infections with it.

The reality appears to have been rather less splendid. Whenever the river was low, and temperatures were above 10-12 degrees (scale not stated) for a few days, the whole system was totally infested with insects. The water-carriers were forced to employ cloth filters at the public fountains, changed several times a day.

Vachette says that the situation at the second installation was even worse. The first pumps were abandoned, because of complaints about the water quality; and the replacements at Gros-Caillou (opposite Chaillot) drew from such a bad point of the river that the receiving reservoir had to be drained and flushed three or four times a day - and only refilled when the river water was not disturbed by nearby drains, washerwomen, or trains of timber being dragged past.

The third installation near the Arsenal was much better even by Vachette's admission, with no settling basin, and a long suction pipe, laid eighty feet into the river.

One further point of interest emerges from Vachette's complaints about the Périer system [p 41]. He indicates that there were 200 faulty wooden valves in the system. Périer proposed brass [cuivre] replacements, at twice the cost of wooden valves, but Vachette claimed to have successfully experimented with valves of cast iron and brass, at one tenth of the cost of those of brass, and that this discovery was of great significance for all hydraulic establishments.

Payen reproduces a report by Coulomb for the Academie des Sciences in 1783 [pp 267-274], which provides a little extra information. The steam-pipe pressure was measured as 32 inches of mercury; the condenser vacuum at 28.8 inches on his visit. He indicates that the piston was hanging at the top of its stroke for well over a second, and considered that the vacuum was not formed fast enough in the cylinder. He produces limited consumption data - the machines never ran continuously for 24 hours - which indicates that the machines were working at about four times the efficiency of Newcomen engines, whatever their other shortcomings.

Perhaps more surprisingly, Coulomb's report emphasises that the efficiency of the machinery depended on Périer's use of a large air vessel (4.5 pieds diameter, 17.5 high):

Périer felt that the air vessel should be mostly full of "condensed" air, and he arranged to replenish this continuously by providing a small hole in the pump cylinder, above low water level; and provided two tell-tale taps in the air vessel.

Payen gives a lengthy account of the Périer factory at Chaillot, established on surplus ground of the Company at Chaillot from October 1778. There was a small engine of 22½ pouces by 4 pieds stroke, driving a pump of 22½ pouces on a lift of 15-17 pieds to drive a water wheel and a series of machines; and a forge bellows directly. This engine was evidently supplied by Wilkinson, as below, but was not operational before March 1783. One of the machines was for boring wooden pipes (a model exists). There was a group of four reverbatory furnaces for casting iron, said to be the first in France, and each capable of 5 milliers (say 2½ tons) in three hours - it is not difficult to see who Périer was copying. Périer eventually cast cylinders to 2 metres diameter and 3.25 metres length at Chaillot. Watt's favourable opinion of the factory and its products has been given above [3 February 1787]. He also refers to it in 1785 in a letter to Boulton [11 August, Muirhead]:

The pumps for Gros-Caillou and Gare were all built at Chaillot [Payen, p 139]; and also many of the components of the Chaillot engines; and probably the boilers. It seems likely that only the major castings and more specialised components came from Wilkinson and Watt; and at least the early pipes for the scheme. Payen only refers to wooden pipes from Chaillot, but it is clear that much of the main-laying in Paris was done after the Chaillot factory was operational, and indeed long after any Wilkinson deliveries of which we have certain knowledge.

Vachette tells us [p 22] that to set up his scheme, Périer was obliged to travel abroad, to see similar schemes, and foundries, in Flanders; and then to London, to "take lessons from Wilkenson, founder". How literally, one wonders ? It is known that he visited Wilkinson in Broseley before 2 May 1777 [Watt, 2 May 1777, in Muirhead, as above; and Payen p 103]. Vachette [p 6] says that a visit to England at the end of 1777 was paid for by the Company. He was in England from December 1778-January 1779; and Payen cites Périer's own later statement that he had been to England five times [p 104]. Apparently his first contact with Boulton was a letter dated from London 24 November 1778 [Payen], when he had discovered that Boulton was actually in Cornwall on business. He was in England to deal with Wilkinson about pipes for his scheme, which had already commenced; and was very anxious to negotiate for two Watt engines before he left. In fact he only met Watt and Mrs Boulton, and left proposals with them: he wanted the rights to two 63 pouce engines, in exchange for shares in his company, and the value of one third of the calculated fuel saving. Payen gives letters of the 7 and 12 January 1779, written from London. Périer was constrained by the powers delegated by his co-directors; Boulton and Watt by the limited protection they enjoyed abroad, as we have seen.

The formal agreement is dated 12 February 1779, and is given in full by Payen. The engines are here 63 inches, pumps are 28 inches with a 9 foot stroke, 10 strokes per minute, and work to 110 French feet, transcribed as 170 English feet - a manifest error. Vachette says that there was an error in the Academy calculation of fuel consumption, on which the saving is calculated, too. The agreement actually waives the usual payments for fuel savings, for such a "noble undertaking", and Boulton and Watt were to provide full drawings for Périer to erect the engines, for a payment (or shares, declined) of 24,000 livres tournois. (The account would not be fully settled until late 1786 - Périer was full of regret, but quite unable to pay [Payen, various, to 23 September 1786, and see below]). At that point it was not determined how much of the engines would be purchased from England, excepting the cylinders.

In the meantime, Périer had been chasing Wilkinson [JW to MB, from Bersham, 23 December 1778]:

On 10 March 1779 [Payen], Périer wrote to B&W touching on what would be made where, and that they had asked Wilkinson to obtain passports for the materials - the French passports were being sought. If neutral shipping was used, it would have to go via Ostend, not direct to Rouen, greatly increasing the cost. On 28 March 1779 [Payen], they were pressing for details of the balance beam, but the chains were already forged in Paris, and the piston rods were to be made in a large anchor forge. The building work for engines and reservoirs would be finished in under three months. French passports for engines and pipes had been obtained. Périer wanted a shipment of steels (a tonne), with the first cargo.

Périer wrote to B&W on 2 May 1779 [Payen], with a reply for Wilkinson, who was absent at this time, and anticipating that one complete engine, and the air vessel would be ready, and that the first cargo would be made up with 24 and 12 inch pipes. Passports to protect the ship against seizure by French or English, and for Havre de Grâce would be sent shortly. The Captain was to sail as for Ostend, with secret instructions:

It is not clear whether Chepstow was an expensive port to ship from, or whether it was simply the cost of the extra passage from Ostend to Havre/Rouen.

Périer goes on to ask for immediate shipping of 10 tons of copper cake and 20 tons of pipe-lead ex-Bristol; and presses for two or three ships to be employed to avoid delay -

- when would the rest of the pipes be shipped ? He wanted 2,000 livres of different steels of which B&W had spoken to him, and

He would be similarly obliged if Wilkinson would load

Perhaps we see here another seed for the future rumours about gun-metal. Indeed, Watt advised Wilkinson on 13 May 1779 [Chaloner, p 84] that the iron should be sent as easily moulded parts such as gudgeon beds, grate bars, etc. Wilkinson in fact replied on 17 May that not only would there be no room for the iron on the next voyage, but that French metal was most excellent anyway.

On 24 June 1779 [Payen], Périer wrote that he had heard that two shipments were ready, and had a French passport for the brigantine Severn which would be used in addition to the Mary. ....They had not written to Wilkinson, not knowing where he was, but sent thanks for his diligence. Captain Williams was to use a named pilot from Havre to Rouen, who would have the address of a house in Rouen for all necessary facilities. They approved the pipework changes made by Wilkinson [see below]; and begged for the drawings still to be sent by B&W.

On 2 August 1779 [Payen], Périer had heard that Boulton had been unable to obtain the English passport for the Mary: the delay was most disagreeable, and he was considering the use of neutral shipping, to avoid the slow and difficult process of obtaining passports. (He may have underestimated the difficulty of finding ships that could both go to Rouen, Chester, etc and yet load bulky and heavy pipes and cylinders; whence Wilkinson's habitual use of few ships, presumably). On 4 July 1779, Wilkinson had been able to consult Captain Williams about the passage, and was clearly concerned about the answers. The ship Mary could only reach Rouen after discharging half her cargo at Havre, and by incurring a two month delay, with £5 per ton extra freight costs. Just as bad, Périer was to be advised that the two halves of his air vessel each weighed 6 tons, and there were no cranes at Havre capable of lifting them out [to Watt, Chaloner p 85].

In September 1779, Périer was in Havre, to arrange for the Mary's arrival. Payen states that deliveries for the Watt engine were made in September-October 1779, but also that Périer fell into the water during the unloading of steam engines on 6 November 1779; the source is not explicit, but it may have been related to the inadequate tackle. In fact the Mary had been discharged at Honfleur in October [Chaloner p 87].

Wilkinson's progress in this period is mentioned a few times in the B&W archive. The Périer engine was being made at Bersham in March [Wilkinson, from Brosley, 23rd, B&W], and on 1 April 1779 he reported from Bersham some detail dimensions of a cylinder to B&W:

And on 10 May, also from Bersham:

To Watt, 17 May 1779, from Bersham:

On 6 June 1779 [B&W], Wilkinson noted from Brosley that 300 12-inch pipes were ready, and another 300 would follow in one month, sufficient with the engine and large pipes to load the two ships. There is also a letter from Pyrke at Newnham dated 10 June, accepting on a trial basis a freighting of the Severn with 12 inch pipes (fearing the 24 inch too large and weighty), from Chepstow, "if the pipes can be shipped there without the use of a crane". This was for five weeks ahead.

Boulton and Watt had seen the pipes being manufactured at Willey at the end of May. One 24 inch and three 12 inch pipes were being cast each day at Willey, and many more at Bradley. It was hoped to increase production to 300 per month [B&W to Périer, 5 June 1779, Chaloner p 81].

A letter from Wilkinson, at Broseley, to Watt of 8 June 1779 [B&W] speaks of warranties for cargoes, and for ships ("I shall be obliged to do that for the vessel in order to keep her insurance down") and Périer's London agent, Motteux:

Wilkinson commented on 18 June 1779 [B&W]:

On 2 July, Wilkinson commented in despair to Watt that the cursed war would be the ruin of the engine business in France [Chaloner p 83]. There was no immediate improvement: Périer wrote on 2 August 1779 [Payen], saying that since Boulton had still not been able to obtain a passport, he was again considering the use of neutral shipping, despite the expense, to avoid the slow and difficult process of obtaining passports - this presumably relates to future shipments. He actually refers to not being able to predict when the embargo would be raised, but the letter to which he was replying is lost, and the meaning is unclear. On 3 August 1779 [B&W] Wilkinson wrote:

There were other transport problems, too. Engine parts despatched from Soho almost missed shipment. They reached Stourport on 30 June, were not loaded until 24 July, left there only on 17 August, and did not arrive in Chepstow on 29 August as expected. The Mary had finally sailed from Chester for Chepstow on 29 August, having lost three successive crews to the press-gangs [Watt, 4 September; Chaloner p 86]

On 18 April 1780 [B&W], Wilkinson had just dispatched Périer's nozzle via Stourport. It appears that a ship was sent out from Chester in 1781 on a British passport dated 28 December 1780, carrying some material re-shipped from Chepstow to Chester. However, this was on a French passport for Nantes (and may have been primarily material for an engine for Jary), and was only the second passport obtained for the Mary [Chaloner p 89]. It is not all clear that the Severn ever received a passport. It is possible that cargoes were none the less sent by neutral shipping.

From about August 1779 there is a dirth of information from any published source. It is however from this period that the stories of pipes piled up at Caughley Dingle (Randall) and Chepstow arise. Certainly there had been two full cargoes waiting at Chepstow by the autumn of 1779. John Byng saw many pipes there in June 1781 [Torrington Diaries, ed C.Bruyn Andrews, London 1934, p 26]:

In view of the shipping problems noted until a few months previously, and later evidence, this 21 miles must be suspect, though it might relate to the intended total order. Chaloner considers it small wonder, all in all, that little more effort was made to make a second shipment direct to Périer. That however leaves a problem as to who actually made the parts for the second engine at Chaillot, among other things, and how Wilkinson survived what must have been a vast loss if so few pipes were sent of those made.

For whatever reason, the first machine was not tried until the summer of 1781, and had not exceeded six strokes per minute by 17 September 1781 [Payen]: it never exceeded eight per minute. In view of the fact that work was still proceeding in Paris on the initial distribution system in 1781/2, a reference of 26 March 1781 [B&W] may relate to this scheme: Wilkinson wrote from Castlehead that he needed some 6 inch pipes there and:

There are then more letters about non-payment of Watt's premium than anything else. On 20 September 1780, Watt wrote to Wilkinson asking how Périer's account stood with him: the reply on 26 September was that even after having had to take some of his money as extra shares, he was still owed £9,000 - "...I have replied to this with as much Temper as an Englishman is possessed of...". He had always suspected that the "ship [Périer's scheme] hung rather heavy.... and your fixing upon money rather then shares was a fault he could not forgive" [Chaloner p 90]. John learned more of the situation from William, on a joint visit to (neutral) Brussels and Liege in January 1782, and wrote to Boulton "We have too many Dukes and courtiers concerned in this enterprise, which will in the end ruin it" [Chaloner p 91].

In a similar vein, Wilkinson commented to Watt on 11 July 1780 [Chaloner p 87] that Thomas Vickers, a Bersham millwright-erector had been sent to France on the Mary, but was not allowed to land, probably by Périer:

Plate IV of Girard's work of 1812 has a detailed representation of the principal water supply systems, ancient and recent. Two of the Périer pumping stations are clearly marked on it, together with the reservoir[s], fountains, and the runs of main pipework. This totals, as drawn, some 17.3 miles, but does not specify the size of pipework, nor indeed its material [see Figure]. The supposition must be that the smaller branch systems are omitted. Girard does not show some of the pumping stations referred to by Vachette [p 32-4]; at the Rue de Bourgogne, already abandoned by 1791 in favour of Gros-Caillou; near the Arsenal, for St Antoine; and near the "gare" for l'Estrapade. No associated pipework is shown in these areas either in his plan.

Unfortunately, Vachette produces a statement [p 21], said to be from a report by a commission of inquiry in 1785, that only 5.88 miles of iron pipe had actually been laid at that time. A further 14.5 miles of wood pipes was in use by the company, not necessarily all laid by Périer (despite his wood-boring engine at Chaillot), since by this time he had taken over the Vachette system.

When we read of up to 600 12-inch pipes being readied for despatch, we have to recall that this is barely one mile when laid. The rate of production might be inferred to be half a mile per month. In 1781 there was an order outstanding for a little over two miles of 6 inch pipe, though not necessarily for Paris [B&W 26 March 1781].

The only other source for quantities of pipes is John Byng who refers to about 21 miles of pipe. (They must have been 12 inch diameter, and the weight and length cited are equivalent to a wall thickness of 0.7 inches). There is not the slightest evidence that Wilkinson was seeking shipping on anything like this scale: over 12,000 pipes (40 months production in two years ?), and some 5,000 tons. The Marycannot have loaded much over 200 tons weight on each voyage. Furthermore, it seems that payments were not coming through from Périer.

If these pipes had been made, but were never delivered because of problems either of shipping during a war, or lack of progress by Périer, then Wilkinson must have suffered very severe losses. It seems more likely that production was suspended at a much earlier stage, possibly of an original order of 21 miles. Clearly, more evidence is required to resolve the problem.

Chaloner does cite a reference for a second order secured by Wilkinson from the Paris Company in 1788: for 40 miles of pipe [D.McPherson: Annals of Commerce, 1805, IV, p 176]. I suppose that this is the 40 miles quoted by Dickinson.

This paper will conclude with a brief note on pipe joints: it has become clear not only that the subject requires further research, but that the material warrants a further paper in its own right. The use of iron pipes was not new in the 1770's, though Wilkinson typically increased the scale of production dramatically. Cast iron pipes are recorded from as early as 1455, at Dillenburg in Germany. About 1720, there were apparently moves to produce pipes in lieu of ordnance in the Russian Baltic ironworks [D.Braid, Transactions of the Newcomen Society, Vol 59, 1987-8, p 11]. The Marly scheme to supply the fountains at Versailles included some 15 miles of iron pipes in sizes up to 20 inches bore, from 1725. These were flanged pipes, in short lengths of about one metre. Interestingly, Diderot's Plate IV/3rd/IX (Plate 94 of the facsimile) shows such a pipe, but with hexagonal flanges. In England, cast iron was used for public water supply (in trunk mains, not distribution) from 1746 at Chelsea, where some 1500 yards of 12 inch flanged pipe were laid; and at various other sites in London at a similar date, with the advent of steam engines. At Edinburgh, 6 miles of 7-9 inch pipes were laid in 1790. Few details seem to have survived about any of such installations.

Socket joints are alleged to have been invented (for cast iron pipes only, note) by Thomas Simpson in 1784, for repairs to flanged pipes, which regularly tore at intervals of about 300 yards, from seasonal thermal movements. They were used only two or three times up to 1791, when Simpson used the run-lead socket joint throughout on a new main for the Chelsea Waterworks.

What has brought the present paper to a halt is a single phrase in a letter from Périer to Watt, cited by Payen [24 June 1779; p 260]:

What, then, was the astragal ? For which pipes were the flanges substituted ? By far the majority of pipes on the orders were for laying in the streets, but some were obviously for use in the pumping plant. It is most likely that flanges would be retained for the machinery, not least to contain thrusts, as now. Did Wilkinson originally offer flanged pipes throughout in accordance with Périer's request and common practice, but actually substitute an alternative, when faced with the manufacture of "forty miles" of pipe, in lengths of nine feet ? Certainly, flanged pipes are impractical for laying in the ground, especially in the streets of a city. There are too many forced changes of direction at inconvenient angles, both horizontally and vertically, and too great a probability of settlement and concentrated imposed loads.

It should perhaps be said that Payen comments [p 120] as follows:

Thus Payen seems to imply a continuous flange in place of the hexagonal form shown by Diderot, rather than a more fundamental change in the type of joint ?

The term astragal, usually a semi-circular moulding on a column or gun-barrel, is apparently otherwise unknown in the context of water pipes. It is my belief that it must relate, by analogy, to the enlargement of the end of a pipe to form a socket joint. (There are typically very small "astragals" at about three-foot centres on examples of pipes at Bersham, but these hardly substituted for flanges. Their function is not clear, but they may relate to feed points ?). Astragal could possibly mean a thickening of the barrel behind a flange, in preference to reinforcing webs: again a small fillet is to be seen at the root on examples at Bersham, but nothing convincing as an "astragal".

Since there was a general consensus among London water engineers in 1821 that Thomas Simpson invented the run-lead socket joint in 1784, what exactly was Wilkinson's new joint in 1779 ? There is one distinct possibility: the turned and bored socket joint. This seems to appear from nowhere, and has equally disappeared almost without trace. All references in the 1821 Minutes for London waterworks are to run-lead sockets. However, during most of the 19th century it was very commonly laid by British engineers worldwide, in parallel, tapered or hemispherical versions. The first explicit reference I have was 1844, but it was clearly not a new form. Curiously it was regarded as a more secure joint than the run-lead socket. It was metal to metal, and although latterly at least painted, it presumably rusted up rapidly. It became a recognised practice to provide a run-lead joint every tenth pipe to provide for thermal movement in the ground. There is also a history of development of run-lead sockets to prevent the lead being progressively extruded under cyclic movement.

There are a number of tantalising references to the boring of pipes, and turning, by Wilkinson; one of his boring machines incorporated special beds for boring pipes, and what might have been a facing lathe for pipes. Wilkinson revolutionised boring of guns and then cylinders, to high standards; it seems clear that most of his pipes were internally bored, too. Did Wilkinson develop the turned and bored joint, then ? If not, what was the astragal that was substituted for the flanges ? Why have we apparently no examples of anything other than flanged, or occasionally plain spigot, joints on his pipes that survive ? Are any Wilkinson pipes going to be uncovered by developments in Paris in the foreseeable future, to provide a sample ? The material has suddenly become more interesting than the Paris scheme itself, returning us to Bersham and Willey.

B&W Archive, Birmingham - from Bersham transcripts, directly from Ann Williams, or via Douglas Braid. The transcripts are not all in full, and contain many evident errors. They do not, apparently, contain all the papers printed by Payen, and this reinforces my supposition that a prolonged search in the archives would produce more information. There are certainly long gaps in the published record at periods when great efforts were being made on the Paris scheme.

Muirhead, J.P., The origin and progress of the mechanical inventions of James Watt, 3 Vols, London 1854.

Payen, Jaques, op.cit.. He appears to indicate that all his Périer correspondence came from the B&W archive; but there is no overlap between these letters and other selections. He says [p 20] that he wrote for photostats of anything referring to Périer.

The Coulomb report is in the archives of the Academie des Sciences, Paris.

W.H.Chaloner, Hazards of trade with France in time of war, 1776-1783, in Business History, Vol VI, June 1964, pp 79-92.

Périer, op.cit.

Vachette, op.cit.

Goubert, J-P, The Conquest of Water, (English translation) Cambridge 1989.

J.Bouchary. L'eau à paris à la fin du XVIII siècle, Paris 1946 (not seen).

J.C.Périer, Sur les machines à vapeur, Paris 1810.

These have caused some problems: there were local variations in France at that period: the unit intended is not always specified. Some are not even precise units.

24 livres tournois = 1 guinea [February 1779 - Agreement]

1 livre = 20 sols = 240 deniers

1 livre = 1.10254 pounds avoirdupois

1 toise = 6 pieds royale = 6.39459 English feet

1 pied = 12 pouces = 144 lignes

35 pintes = 1 pied cube and weighs 70 livres [Belidor]

1 muid = 8 pied cube = 274.704 litres = 60.43 Imp.gallons

thus = 280 pintes, but Périer gives 250 pintes [Prospectus]

Goubert, in translation, p 173, gives 238 litres, a "hogshead".

voie d'eau [water-carrier's load, in two buckets/barrels]:-

dictionary: about 30 litres

Périer: about 30 pintes, thus 8-9 voies per muid; voie = 30.5-34.3 litres

pouce d'eau/fontenier [discharge of defined orifice - see text]:-

Belidor: 14 pintes/28 livres per minute = 3.087 Igpm

Coulomb: implies 3.022 IGPM [Report on Chaillot]