Precision Instruments and the Demonstrative Order of Proof in Lavoisier's Chemistry Author(S): Jan Golinski Reviewed Work(S): Source: Osiris, 2Nd Series, Vol

Precision Instruments and the Demonstrative Order of Proof in Lavoisier's Chemistry Author(s): Jan Golinski Reviewed work(s): Source: Osiris, 2nd Series, Vol. 9, Instruments (1994), pp. 30-47 Published by: The University of Chicago Press on behalf of The History of Science Society Stable URL: http://www.jstor.org/stable/301997 . Accessed: 20/06/2012 14:38

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press and The History of Science Society are collaborating with JSTOR to digitize, preserve and extend access to Osiris.

http://www.jstor.org Precision Instruments and the Demonstrative Order of Proof in Lavoisier's Chemistry

By Jan Golinski*

If it is true that a controversy approaches its conclusion by the accumulation of facts that impinge upon it, it is only so pro- vided these "facts" are without ambiguity in their implications. For otherwise, twisted by the rival hypotheses, and sometimes with so many more words that they convey less sense, these "facts" so multiply the extraneous questions that controversies become endless. Thus prejudice and imagination freely hold sway and logic is replaced by fashion. -Jean-Andr6 Deluc (1790)1

HAT,I AFTERMORE THAN A DECADE OF DEBATEabout the fundamen- tals of chemical theory,Jean-Andre Deluc should express frustrationis not reallysurprising. As an upholderof the traditionaltheory of phlogistonand an oppo- nent of the new theory of Antoine-LaurentLavoisier, Deluc fearedthat the controversy would never end. Each new fact could be interpretedin differentways by the two sides and, ratherthan resolving the debate,seemed to bringever more subjects into doubt.Reason, supposedly the securepath to scientifictruth, seemed incapable of deciding the issue. The same sentimentwas voiced nearlysimultaneously by the English chemistJames Keir, who tried to curbthe hopes of his fellow phlogistonist JosephPriestley, who looked forwardto an imminentcompromise. Keir cautioned that"there are wonderfulresources in the disputeabout phlogiston, by which either party can evade, so that I am less sanguine than you are in my hopes of seeing it terminated."2 Historiansare interested in controversiesfor muchthe same reasonsthat historical participantslike Deluc and Keir found them so frustrating.As "facts"accumulate on each side, less and less appearsto be certain.Instead, debate ramifies across an ever wider range of questions.Phenomena, methods, apparatus,personal compe- tence, assumptionsand principles-all may become issues in dispute.Hence if con- troversiesbecome prolonged,more and morebackground assumptions and practices

* Departmentof History,University of New Hampshire,Durham, New Hampshire03824-3586. l "Lettrede M. DeLuc a M. De La MWtherie,sur la naturede 1eau, du phlogistique,des acides & des airs,"Observations et MWmoiressur la Physique, sur 1'Histoire Naturelle, et sur les Arts et Me- tiers, 1790, 36:144-154, on p. 153. 2 James Keir to Joseph Priestley,[n.d., 1789?], in A ScientificAutobiography of Joseph Priestley (1733-1804), ed. RobertE. Schofield (Cambridge,Mass.: MIT Press, 1966), pp. 252-253, on p. 253.

OSIRIS1994, 9: 30-47 30 PROOF IN LAVOISIER'SCHEMISTRY 31

are exposed to view. Much recent historicalwork has shown the value of disputes as sites for examining scientific practice as a social activity.In controversiesit is particularlyclear how manyelements of historicalcontext shape rival interpretations of natureand how many "wonderfulresources" are availableto those tryingto close the issue3 Lavoisier's"chemical revolution" presents itself as an underexploitedfield for such study.There has been relativelylittle workon the dynamicsof the controversy, which ebbed and flowed throughoutthe 1780s and into the following decade. Per- haps historianshave been too concernedwith trying to grasp in essentialistterms the real natureof Lavoisier'sachievement or assessing whetherit deservesthe label "revolution."The processof persuasionundertaken by Lavoisierand his allies in the 1780s tends to be regardedas an aftermathto the main events. And yet what Carl Perrincalled the "'triumphof the antiphlogistians"was no walkover,but a lengthy process that deserves detailed investigation.Controversy ranged over numerousis- sues of fact and swelled to embracemethodological, linguistic, and social questions. Lavoisier'ssystem as a whole was articulatedin the contextof this debate.The map- ping of the strugglein its temporal,geographical, and social dimensionsis a large- scale task, but one that promises considerablerewards in historicalunderstanding of the processes of science.4 Such a mappingcannot be attemptedhere, thougha step can be takentowards it by surveyingthe role of instrumentsin the controversy.Trevor Levere has recently remindedus of the importanceof Lavoisier'snovel instruments,including the calo- rimeterand the balance,and of theirrole in the campaignagainst phlogiston. Fred- eric L. Holmes has pointed out how radicala breakthis apparatusmarked with the "longueduree" of the eighteenth-centurychemical laboratory. And ArthurDonovan has arguedthat Lavoisier's instrumentation signals his transferinto chemistryof the methods of the more mathematizedphysical sciences.5In this articleI build upon this work to place Lavoisier'sapparatus against the backgroundof the controversy surroundinghis new chemistry.My aim is to use the circumstancesof dispute to expose the assumptionsand practicesgoverning his deploymentof this particular technology.

'For sociological work on controversies,see H. M. Collins, Changing Order: Replication and Induction in Scientific Practice (London/BeverlyHills: Sage, 1985); and Collins, ed., Knowledge and Controversy:Studies of Modern Natural Science, special issue of Social Studies of Science, 1981, 11(1). Historical studies include Steven Shapin and Simon Schaffer,Leviathan and the Air- Pump: Hobbes, Boyle and the ExperimentalLife (Princeton:Princeton Univ. Press, 1985); Martin J. S. Rudwick, The GreatDevonian Controversy:The Shaping of ScientificKnowledge among Gen- tlemanly Specialists (Chicago: Univ. Chicago Press, 1985); and James A. Secord, Controversyin VictorianGeology: The Cambrian-SilurianDispute (Princeton:Princeton Univ. Press, 1986). 4Carleton Perrin, "The Triumphof the Antiphlogistians$'in The Analytic Spirit: Essays in the History of Science in Honor of Henry Guerlac, ed. HarryWoolf (Ithaca,N.Y.: Cornell Univ. Press, 1981), pp. 40-63. Other work on the controversyincludes Karl Hufbauer,The Formationof the German Chemical Community(1720-1795) (Berkeley/LosAngeles: Univ. CaliforniaPress, 1982); JohnG. McEvoy,"The Enlightenmentand the ChemicalRevolution" in Metaphysicsand Philosophy of Science in the Seventeenthand Eighteenth Centuries:Essays in Honour of Gerd Buchdahl, ed. R. S. Woolhouse (Dordrecht:Kluwer Academic, 1988), pp. 307-325; and some of the essays in ArthurDonovan, ed., The ChemicalRevolution: Essays in Reinterpretation,Osiris, 2nd ser., 1988, 4. 5Trevor H. Levere, "Lavoisier:Language, Instruments,and the Chemical Revolution"in Nature Experimentand the Sciences, ed. Levere and W. R. Shea (Dordrecht:Kluwer Academic 1990), pp. 207-233; FredericLawrence Holmes, Eighteenth-CenturyChemistry as an InvestigativeEnterprise (Berkeley: Office for History of Science and Technology,Univ. California, 1989), esp. Ch. 5; and ArthurDonovan, "Lavoisierand the Origins of ModernChemistry," Osiris, 1988, 4:214-231. 32 JAN GOLINSKI

Controversyenables us to see how materialapparatus is embeddedin specific settingsof practicethat enable it to functionas a tool of investigationand persuasion. We shall see that Lavoisierhad to mobilize particularpersonnel and their skills to craft and use his instruments.He forged links with practitionersof the exact sciences, trainedin the French mathematicalengineering tradition, and with skilled instrumentmakers.6 He expendedsubstantial financial resources on the construction of his apparatus.He mastereddifficult techniques of measurementand calculation- in calibration,for example. He also constructedsocial and literary"technologies," managingthe audiencesat set-pieceexperimental demonstrations and conveying the results in a writtenform that stressedthe accuracyof the proceduresand the high standardof proof therebyachieved. In the ongoing controversy,many aspects of this form of practicewere made explicit in the courseof challengesto, and defenses of, Lavoisier'sclaims. Outside Lavoisier'sown setting, his instrumentsdid not always convey their hoped-forpersuasive potential. Many resources enabled opponents such as Priestley andKeir to evadethe purportedimplications of his experiments.Priestley articulated a radically different model of scientific practice and condemnedLavoisier's sup- posed accuracyas the spuriousresult of excessively complex experimentalcontriv- ances. For Priestley,his own inability to replicate the French experimentswas a reason not to trust them. Thus discussion of instrumentswas implicatedin wider debates about the way science should be practiced.Arguably, the controversywas not just about the facts of chemical phenomenabut about how science should be carriedon. In the face of such radicaldisagreement, Lavoisier's instruments simply could not convey theirmeaning unequivocally. Nonetheless, the controversywas eventuallybrought to a close, albeit in a prolonged and confused way thatdeserves further investigation. It seems clear thatthe extension of Lavoisier'spractices of instrumentaluse played a partin this process. His victory (to a certainextent a posthumousone) reliedupon transmitting a culture of experimentalpractice to supportdiffusion of his instrumentsand replicationof the phenomenathey produced.

I. LAVOISIER'SINSTRUMENTAL STRATEGY By the late 1770s, Lavoisierhad convincedhimself of the need to reformchemical theory fundamentallyand to dispense with the notion of phlogiston.In his "crucial year"of 1772 he had studiedthe calcinationof metalsand the combustionof sulphur and phosphorusand had establishedthe fixationof air in these processes. He then followed up Priestley'sisolation of "dephlogisticatedair," repeating the operation

6 The classic studies of Lavoisier'sinstruments are MauriceDaumas, Lavoisier: Theoricienet ex- perimentateur(Paris: Presses Universitairesde France, 1955), esp. Ch. 6; Daumas, "Les appareils d'experimentationde Lavoisier,"Chymia 1950, 3:45-62; and Daumas, "Precisionof Measurement and Physical and Chemical Research in the EighteenthCentury," in Scientific Change: Historical Studies, ed. A. C. Crombie(London: Heinemann, 1963), pp. 418-430. On the Frenchmathematical engineeringtradition see C. StewartGillmor, Coulomb and the Evolutionof Physics and Engineering in Eighteenth-CenturyFrance (Princeton: Princeton Univ. Press, 1971), esp. Ch. 1; and CharlesCoul- ston Gillispie, Science and Polity in Franceat the End of the Old Regime (Princeton:Princeton Univ. Press, 1980), pp. 506-552. 7For this use of the termtechnologies see Steven Shapin,"Pump and Circumstance: Robert Boyle's LiteraryTechnology," Soc. Stud. Sci., 1984, 14:481-520. PROOF IN LAVOISIER'SCHEMISTRY 33 for producingit by heatingred mercurycalx. In 1776-1777 he determinedthat this "purestpart of the air" was what combined with solid substancesin the course of their calcinationor combustion.He disclosed its role as the portion of the atmo- sphere consumed in respirationand ascribed to it the power to make substances acidic that was to give it its name, oxygen (the acid generator).8 Lavoisier'sreadiness to deploy new apparatus,borrowing it from disciplinesusu- ally consideredbeyond the boundsof chemistry,had been characteristicof his work since his early researchesin mineralogyand geology. Alreadyin the 1760s he had been using thermometricand barometricmeasurements in geological surveysand developinghygrometric methods for analyzingmineral water. It was in the 1780s, however,that he began to exploit physicalinstrumentation with greaterconsistency and to deploy it in his campaignagainst traditional chemical theory.His collabora- tion with the mathematicalphysicist Pierre-Simonde Laplace in 1782-1783 has been illuminatedby a classic study by Henry Guerlacand in a stimulatingrecent paperby Lissa Roberts.9The two collaboratorsdesigned and used a new instrument, which they introducedin a jointly written "Memoiresur la chaleur"in 1783. As Robertspoints out, the (initiallyunnamed) machine was presentedas a purportedly unproblematicmeasuring device for heat exchanges in reactions,the authorspro- fessing that it had no particularimplications as to the natureof heat. The naming of the device as a calorimeteroccurred subsequently in the context of Lavoisier's systematicreconstruction of the disciplinaryprofile of chemistryin his Trait e'le'- mentairede chimie (1789). Roberts also shows that other experimenters,such as JosiahWedgwood and Adair Crawford, experienced difficulties in replicatingLavoi- sier andLaplace's experiments and disputedthe workingof theirmachine. An anon- ymous writerappears to havereflected the generalappraisal, when he wrotein 1797 that, "littlereliance ... can be placed on the accuracyof this much-boastedprocess of the Frenchchemists," although their results had been presented"with all the precision of the new school."10 In the case of the calorimeter,an initial attemptto build a consensus aroundthe supposedlytheory-neutral use of a measuringmachine was succeeded by a more explicitly theoreticaldeployment of the instrument.Lavoisier was also, by the mid 1780s, makinguse of othernew apparatusto try to secureacceptance of his theories of combustion,acidity, and the compositionof water.In this period Lavoisier'sin- strumentswere just as much at issue as his substantivetheoretical claims. At the beginningof the decade he had no allies among leading chemists. Most remained convinced that phlogistonwas a materialentity released from burningbodies. In- deed, phlogiston acquireda new lease on life, in the view of many,when it was identifiedby the Irishchemist Richard Kirwan as the basis of "inflammableair" (the

I Henry Guerlac,Lavoisier-The Crucial Year:The Backgroundand Origin of His First Experi- ments on Combustionin 1772 (Ithaca, N.Y.: Cornell Univ. Press, 1961); Guerlac,Antoine-Laurent Lavoisier: Chemistand Revolutionary(New York:Scribners, 1975); and FredericL. Holmes, Lavoi- sier and the Chemistryof Life: An Explorationof Scientific Creativity(Madison: Univ. Wisconsin Press, 1985). 9 HenryGuerlac, "Chemistry as a Branchof Physics: Laplace'sCollaboration with Lavoisier,"His- torical Studiesin the Physical Sciences, 1976, 7:193-276; andLissa Roberts,"A Word and the World: The Significance of Naming the Calorimeter,"Isis, 1991, 82:198-222. '0 T. H. Lodwig and W. A. Smeaton, "The Ice Calorimeterof Lavoisierand Laplace and Some of Its Critics,"Annals of Science, 1974, 31:1-18; and CriticalExamination of the First Part of Lavoisier's Elementsof Chemistry(London, 1797), pp. 20-21. 34 JAN GOLINSKI

gas Lavoisierwas to call "hydrogen").Kirwan articulated a theory of combustion thatwon considerablesupport. In his view the phlogistonreleased by a burningbody combinedwith dephlogisticatedair to form fixed air, which then unitedchemically with the residueof the solid to form a calx or acid. Kirwansaccount had the appeal of accommodatingthe weight gain thatwas agreedto occurin instancesof combustion and calcinationwhile maintainingthe existence of phlogiston." In the face of this alternativeto his theory of combustion,Lavoisier's fortunes turnedon a new issue introducedinto the debatein the early 1780s:the composition of water.In 1781 HenryCavendish produced water from a mixtureof inflammable anddephlogisticated airs ignited by an electricspark. The experimentemerged from the traditionof eudiometry,in which the "goodness"of a sample air was measured by phlogistication(in this case by sparkingwith inflammableair) andmeasuring the diminutionin volume. The productionof water in the reactionwas a quite unex- pected result. Cavendishcanvassed a couple of explanations,suggesting that the more likely one was that water was part of the compositionof both airs and was releasedon theircombination by a kind of condensationreaction.'2 Lavoisierseized on this result,repeating Cavendish's experiment before its long- delayedpublication. In June 1783, with the assistanceof Laplaceand in the presence of CharlesBlagden (Cavendish'sassistant) and witnesses from the Academie des Sciences, Lavoisierignited jets of the two airs over mercuryin a sealed glass vessel. The experimentmade use of two pneumaticchests that he had recently had con- structedfor storingthe gases. Lavoisierimmediately announced a new interpretation of the reaction,stating that water was the sole productof combinationof the two gases and hence was not an element,as Cavendishand all otherchemists had maintained, but a compound.'3This interpretationexplained two classes of phenomena that had previouslyconstituted troublesome anomalies for his theory.The inflammableair generatedby metals when they dissolvedin acids could now be explained as a productof the decompositionof water,while the reductionof lead calx and othercalxes by inflammableair could be understoodin termsof the combinationof the gas with oxygen from the calx (or oxide) to synthesizewater. Lavoisier'sinterpretation of the reaction was not, however, accepted by other chemists. Cavendish,when he finally publishedthe accountof his own experiment in 1784, referredto Lavoisier'santiphlogistic explanation but professedhimself un- convinced:'As the commonlyreceived principle of phlogistonexplains all phenomena, at least as well as Mr.LAVOISIER'S, I have adheredto that."14 JamesWatt, the Birminghamsteam-engine manufacturer and a friend of Priestley,made a similar distinctionbetween the facts reportedby the Frenchexperimenters and the interpretive gloss they had laid over them. Wattwrote to Deluc that he had no reason to doubt the credibility of the factual reportthat water was the sole productof the reactionand its weight equal to thatof the two gases: "Fromthe characteryou give

" RichardKirwan, "Remarks on Mr.Cavendish's Experiments on Air,"Philosophical Transactions of the Royal Society, 1784, 74:154-169; and Michael Donovan, "BiographicalAccount of the Late RichardKirwan, Esq.," Proceedings of the Royal Irish Academy, 1850, 4: lxxxi-cxviii. 12 Henry Cavendish,"Experiments on Air,"Phil. Trans.,1784, 74:119-153. 13 A. L. Lavoisier,"M6moire dans lequel on a pour objet de prouver que leau n'est point une substancesimple," Oeuvres de Lavoisier ed. J. B. Dumas and EdouardGrimaux, 6 vols. (Paris:Im- primerieNationale, 1864-1893), Vol. II, pp. 334-359. 14 Cavendish,"Experiments on Air" (cit. n. 12), p. 152. PROOF IN LAVOISIER'SCHEMISTRY 35 me of the gentlemenwho made it, thereis no reasonto doubtof its being made with all necessaryprecautions and accuracy."He was, however,not convinced that the experimentwas demonstrativeof the compoundnature of wateror the nonexistence of phlogiston.Alternative ways of "solvingthe phenomena,"which were "as plausi- ble as any other conjectureswhich have been formed on the subject,' remained open.'5Watt's conjectured explanation was very like Cavendish's.Dephlogisticated air was waterdeprived of phlogistonand with its latentheat bound;inflammable air was phlogistonplus a little water and latent heat. When the two airs united, water was releasedalong with heat. Lavoisierwas thus madeaware that a morepersuasive proof than the Junedemon- strationwas needed if his contentionof the compoundnature of water were to be accepted.In the autumnand winterof 1783/84 he laboredto providesuch a proof. His approachwas to attemptmore accuratemeasurement of the quantitiesof reactantsand products, following the lead of GaspardMonge, instructorin experimental physics at the military engineering academy,the Ecole Royale du Genie, at Mezieres.In Juneand July 1783 Monge had conductedhis own experimentson the synthesisof waterindependently of Lavoisier's,measuring the volumesand specific weights of the reactantgases and thus establishingtheir (almost exact) equality to the weight of water produced.Such a quantitativeapproach appealed to Lavoisier because it seemed to offer the rigor of a geometricalstandard of proof, since "it is no less true in physics than in geometrythat the whole is equal to its parts.?16 To repeatthe quantifiedsynthesis experiment,Lavoisier would requirenew vessels capable of measuringthe volumes of the gases used; the pneumaticchests he hademployed with Laplacewere not adequatefor this purpose.He thereforeenlisted the help of Jean-BaptisteMeusnier, a formerpupil of Monge'sat Mezieres,who set to designing appropriatevessels and having them constructedby the instrument maker Pierre Megnie. Meanwhile Lavoisier and Meusnier worked on an experiment to demonstratethe compoundnature of waterby decomposingit into its con- stituentgases. They passed steam througha red-hotiron gun barrel.The steam was taken to be decomposed,its oxygen uniting with the iron to form an oxide and its inflammableair emergingfrom the pipe to be collected along with undecomposed water. Lavoisierreported the success to the Academie in April 1784 and subsequently published the account. Although he admittedthat the proportionsof the constituentsof water could not yet be calculatedwith "mathematicalprecision," since the gun barrelhad also undergoneoxidation on the external surface while being heated, he nonethelessproposed the experimentas a "demonstrativeproof" that waterwas a compound.'7 Again, however,dissension continued. Kirwan and Priestleydenied to Lavoisier's experimentsthe implicationtheir author sought to give them. Both insisted that the proposedanalysis of water was no such thing. What had happenedwas that phlogiston (inflammableair) had been displaced from iron by combinationof water with the metal. In February1785 Priestleydescribed to the Royal Society his own

15 JamesWatt, "Thoughts on the ConstituentParts of Water,"Phil. Trans.,1784, 74:329-353, esp. pp. 329, 333. 16 Lavoisier,"M6moire dans lequel on a pour objet"(cit. n. 13), p. 339. 17 A. L. Lavoisierand J. B. Meusnier, "M6moireoi l'on prouve, par la decomposition de 'eau, que ce fluide n'estpoint une substancesimple," Oeuvres de Lavoisier(cit. n. 13), Vol. II, pp. 360-373, esp. p. 371. 36 JAN GOLINSKI replicationof Lavoisierand Meusnier'sexperiment, in which he used measurements of weights of reactantsto show thatthe source of the inflammableair was the iron, not the water.As had Cavendishand Watt,Priestley charged Lavoisier with trans- gressingthe conventionthat experimental philosophers should simply describewhat they observedand not go beyondthat to impose hypotheticalrationalizations on the phenomena:"Whilst philosophers are faithful narratorsof what they observe, no person can justly complainof being misled by them; for to reason from the facts with which they are suppliedis no more the provinceof the person who discovers them, thanof him to whom they are discovered."18 Facing this persistentopposition to his claim that waterwas a compoundof two gases, Lavoisiercontinued to seek a more stringentand compellingproof, to push back the boundarythat his criticshad erectedbetween the "facts"of experimentand what they insisted could only be an interpretationor "hypothesis."He worked to make the compoundnature of water into a fact-a direct, unmediatedinference from experiment,permitting no possibility of doubt. This was to be done by em- ploying new, more refinedapparatus to yield quantitativeweight measurementsof an unprecedentedaccuracy. His effortsculminated in a large-scaleset-piece demon- strationof the analysis and synthesis of water in the Paris Arsenal on 27 and 28 February1785. On that occasion Lavoisierassembled all the elements of his form of experimentalpractice to conveya demonstrativeproof of his claims;his apparatus was deployedin the full setting designedto maximizeits persuasiveefficacy. 19 The analyticpart of the 1785 experimentwas relativelylittle changedfrom what Lavoisierand Meusnierhad accomplishedthe previousyear. But the operationto synthesizewater from its componentgases was performedwith unprecedentedcare and very sophisticatednew apparatus.Preparing for the experiment,Lavoisier fur- therexploited his links with personneltrained in the mathematicalengineering tradition and with the skilled instrumentmakers who served it. He continuedto work with Laplace and Meusnierand recruitedMonge to help with the synthesis. The instrumentmaker Megnie producedthe new pneumaticvessels designedby Meus- nier towardsthe end of 1783 and was paid 338 livres for them. He also built two new balancesfor Lavoisier,using novel techniquesto suspendthe beams and damp their oscillations.The largerof these two was estimatedto be capableof weighing one pound with an accuracyof about 1 in 100,000. For all his work for Lavoisier, Megnie was paid 1,814 livres duringthe years 1783-1785. No more than400 livres of this came from the Academie, the remainderfrom Lavoisier'spersonal wealth. In straightforwardfinancial terms, Lavoisierwas investingsubstantial resources in apparatusthat would serve his purposes.20 The new pneumaticvessels were describedin a paper publishedby Meusnier. Like the calorimeter,the instrumentLavoisier was subsequentlyto name the "gasometer"was introducedinitially as an anonymousappareil or machine.It was, said

18 JosephPriestley, "Experiments and ObservationsRelating to Air and Water,"Phil. Trans.,1785, 75:279-309, esp. p. 280. 19Maurice Daumas and Denis Duveen, "Lavoisier'sRelatively Unknown Large-Scale Decomposi- tion and Synthesis of Water,February 27 and 28, 1785," Chymia, 1959, 5:113-129; and Holmes, Lavoisierand the Chemistryof Life (cit. n. 8), pp. 237-238. 20 Daumas, Lavoisier (cit. n. 6), p. 149. Comparethe 600 livres that Lavoisierpaid the tinplate workerNaudin for the two calorimetershe constructed.Roberts, "Word and World"(cit. n. 9), makes the telling comparisonwith the averagedaily wages of a skilled worker:1 /2-2/2 livres. PROOF IN LAVOISIER'SCHEMISTRY 37

Meusnier,a "universalinstrument" to manipulatevolumes of gases, "by a perfectly uniformflow, variable at will, and giving, at each instant,the measureof the quantity of air used with all the precisionthat one can desire.21The device was based on the principleof the pneumatictrough, used by numerouseighteenth-century chemists to store gases over water (see Figure 1). An upper tank, open at the bottom, was suspendedfrom a counterweightedbeam so that it could move up and down within a lower tankcontaining water. Pipes enteredthe tank to introducethe gas as it was preparedand to let it out as required.The crossbeamhad arcs of circles mountedon each end to equalizefrictional resistance to motionat all positionsof the uppertank, which was suspendedby a chain designed not to elongate undertension. Meusnier'smajor design innovationwas a means of ensuringa constantflow of gas out of the apparatus.To achieve this, a constantpressure had to be maintained. But as the upper tank descended it would displace water and therebyreduce the pressureon the gas. To compensatefor this Archimedeanthrust, Meusnier designed an ingenious modificationto the beam arm that suspendedthe counterweight.The end of the arm was displacedparallel to the remainderof it, so that the momentof the counterweightaround the fulcrumwould be differentfor differentpositions of the beam. The linearvariation in effective counterweightcompensated for the linear variationin pressureof the gas due to Archimedes'principle. A long screw con- nected the displaced part of the arm to the rest of the beam, so that the degree of displacementcould be adjustedto set a particularpressure. A scale mountedbeside the screw enabled this to be measured.Another scale, on the arc of the arm sus- pending the tank, was read against a pointer mounted in a fixed position on the fulcrumpillar. This scale could be calibratedto give the volume of gas containedin the vessel. And the pressure of the gas could be read from a water manometer mountedon the outside of the lower tank and connectedwith the interior. Two gasometerswere requiredfor the synthesisexperiment: one each for the oxygen and the hydrogen(see Figure2). Towardsthe end of December 1784 Lavoisier and his collaboratorsbegan operationsto calibratethem. For each instrumentthis was a two-stageprocess, requiringseveral days' work.First, the screw of the coun- terweightarm was adjustedwhile the pressureof gas in the vessel was observedon the water manometer.Appropriate positions on the scale were noted to maintain differentset pressures:one inch (of water,above atmosphericpressure), two inches, three inches, and so on. Each setting would producea different,but constant,speed of gas flow. In the second stage the volumes of gas in the vessel, correspondingto differentpositions of the pointeragainst the otherscale, were determined.This was done by filling bottles of known capacity with air drawnfrom the gasometerand noting the change in the pointerposition. This was done severaltimes, with bottles of different sizes, to overcome inequalities in the width of the tank at different heights. When the scale had been calibratedin terms of volume, the conversion could be made to weight by consultingprepared tables of the densities of the two gases undervarious conditions of temperatureand pressure.22

21 J. B. Meusnier, "Descriptiond'un appareilpropre 'a manoeuvrerdiffdrentes especes d'air,"in Oeuvresde Lavoisier (cit. n. 13), Vol. II, pp. 432-440. 22 This accountis drawnfrom Daumas and Duveen, "Lavoisier'sLarge-Scale Decomposition" (cit. n. 19); Meusnier,"Description" (cit. n. 21); and A. L. Lavoisier,TraitW e1ementaire de chimie, pre- sente dans un ordre nouveau, et d'apres les dicouvertes modernes, 2 vols. (Paris, 1789), Vol. II, pp. 346-360. 38 JAN GOLINSKI

The success of the calibrationsdepended, of course, on the variousskills of the experimenters,not least theirtacit knowledgeof the apparatusthey were handling. Accuracywas strivenfor in all the measurementstaken. The position of the pointer againstthe limb scale was readto two places of decimalsof a degree of arc, by use of an attached vernier.Verniers were apparentlyalso fitted to the thermometer, which was read to one decimal place of a degree Reaumur,and to the barometer, which was read to one decimal place of a line of mercury.(One line is V/I2 of an inch, approximately0.225 cm.) Lavoisierhad been interestedin the accuracyof thermometersand barometerssince his early mineralogicalwork, and he now put to use the most advancedprecision versions of these instruments.23 The calibrationscompleted, Lavoisierinvited about thirty savants,including a dozen witnesses nominatedby the Academie, to attend the demonstration.The course of the experimentshas been well describedby Daumasand Duveen.24The first analysiswas begun on the morningof 27 February,and more thanten bell jars were filled with hydrogengas and measured.This hydrogenwas used in the first synthesisexperiment later the same day.One of the two gasometerswas loadedwith the gas, the otherhaving been filled with oxygen obtainedfrom heating red mercury calx. A jet of hydrogenwas led into the reactionvessel, which was filled with oxygen and connectedwith the oxygen-holdinggasometer. After a few failures,the jet was ignited by a sparkfrom an electricalmachine owned by Lavoisier.While the constantflow of gases from the gasometerscontinued, the combustionwas maintained for about three hours. The following morning,both gasometerswere refilled and the combustionresumed, while a second analysis was performedto producemore hydrogen.This had in turnbeen consumedin the synthesisexperiment by the end of the eveningof 28 February.Further work was done on the following days, though it was less carefullywitnessed and recorded.The waterproduced by the synthesis reactionwas carefullyweighed and analyzed,as was the residualgas in the reaction vessel. Finally,the weights of reactantsand products in the two analysesand the one (discontinuous)synthesis were calculated. The circumstancesand results of the demonstrationmade it convincingto many of those who took part.Lavoisier had not so much mounteda show as providedan opportunityfor his colleagues to participatein a well-organizedteam effort. The academiciansinspected the apparatus,took measurements,and signed their names to the recordsof the results.The proceduresof witnessing,recording, and certifying both guaranteedthe authenticityof the results and gave the participantsa stake in theirvalidity. Monge and many of the mathematiciansand physicistsin the Acade- mie confirmedtheir support for Lavoisier'sdoctrine. The crucialconversion was that of the chemist Claude-LouisBerthollet, who wrote enthusiasticallyto Blagden about the demonstrationand, in a paper read to the Academie on 6 April 1785, announcedthat he had been convinced by "the beautiful experiment"that water

23 Estimatesof accuracyof measurementare drawn from examinationof notes from the experiment surviving in the LavoisierMSS, History of Science Collection, Cornell University Library,Ithaca, N.Y, MSS 3.04, 9.02a-d, 9.03a-d, 9.04a-c, 9.05, 9.06, 9.07a-b, 9.08, 9.09, 9.10, and 9.11a-c. On the accuracyof thermometersand barometerssee TheodoreS. Feldman,"Late Enlightenment Mete- orology,"in The QuantifyingSpirit in the EighteenthCentury, ed. Tore Fringsmyr,J. L. Heilbron, and Robin E. Rider (Berkeley/LosAngeles: Univ. CaliforniaPress, 1990), pp. 143-177, on pp. 156- 157, 166. 24 Daumas and Duveen, "Lavoisier'sLarge-Scale Decomposition" (cit. n. 19), pp. 123-126...... ~...... ' v - Jf 21Z iX " 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~t1A

CVALUMSAU A GA Ov sv ' A

FiguJ 1J pu M n g ere , i Pe M I F JO d o p8 esA fiwing p.;SLCII1 44. Courtesy1W qf t/e History of Science Collections, Cornell Uniw r'atv Library

F+igure 1. Jean-Baptiste Meuxnier's gasometer design, built by Pierre Me'gnie'. From Journal Polytypc des Sciences et des Arts, / 786, 1, plate 2, fiw(ing 1). 44. Courtes y oJfthe Hi.story oJ S iecn ewCollec tion~s, Cornell Univesrsity Libraryir. _1* , *~~~~~~~~~~~'WO: f A ce

1.1 =w mile S

...s...... -

Figure 2. Top:Lavoisiers apparatusfor synthesizing water, used in the experimentsof February1785. Bottom:The apparatusforanalyzing water FromOeuvres de Lavoisier(Paris, 1864-1893), Vol.11, plate 5. By permissionof the Syndicsof the CambridgeUniversity Library. PROOF IN LAVOISIER'SCHEMISTRY 41 was indeed a compound.25Not everyonewas so convinced:the chemistsBalthazar- Georges Sage and Antoine Baume, who were presentat the Arsenal,remained op- posed to Lavoisier'stheory.26 But in generalit appearsthat Lavoisierhad managed the social technology of convincing his audience at the demonstrationalmost as successfully as he had manipulatedthe materialtechnology of his apparatus. Persuadinga wider audience was however a differentmatter. On this occasion, Lavoisier'sliterary technology untypically let him down. A brief accountof the experiment,probably based on draftsby Lavoisier,appeared under Meusnier's name in the Journal Polytype des Sciences et des Arts in February 1786.27 Meusnier was apparentlyalso charged with producing a comprehensivereport, but this never emerged. In 1789 Bertholletwas still apologizingfor the fact that Meusnier'sab- sence from Paris on militaryduties had preventedhim from completingthe work.28 The accountthat did appeargave only an incompleteimpression of the sophistica- tion of the apparatusand the precautionstaken in its use. No descriptionwas given of the calibrationprocedures for the gasometersor the meticulous care taken to ensurethe accuracyof the measurements.Somewhat out of the blue, Meusnieran- nounced a rounded-offfigure for the proportionsof the componentsof water:85 percent oxygen and 15 percent hydrogen,by weight; a figure that was within the rangeof the resultsobtained but not unambiguouslyproven. The ratherthin account was buttressedby a somewhat dogmatic and aggressive rhetoric.Readers were bluntlytold thatthe descriptionwas "morethan sufficient to lay hold of the certainty of the proposition"that waterhad just this composition.And the paper concluded by promisingthat methods of precisionmeasurement offered the prospectof uniting chemistrywith the other physical sciences and advancingto make discoveriesof unprecedentedcertainty.29

II. THE CRITICS OF THE STRATEGY Perhapsunsurprisingly, many chemists, particularlyoutside France,were not convinced. In Britain,Kirwan and Priestley continued to voice objectionsto the composition of waterand gained significantsupport. In France,Jean-Claude de Lametherie kept up a steadybarrage of criticismfor the remainderof the 1780s. In the Observa- tions et MWmoiressur la Physique, of which he assumed the editorship in 1785, Lametheriecoordinated the anti-Lavoisianforces, publishingthe views of such critics as Sage and Deluc, translatingpapers by Priestleyand Keir,and launchingregu- lar attacksin his "Discourspreliminaire" at the frontof each annualvolume.30 Variousalternative interpretations of Lavoisier'sprized analysisand synthesisre- actions were advanced.In his Essay on Phlogiston (1787) Kirwanproposed that

25 C.-L. Berthollet, "Memoire sur l'acide marin dephlogistique,"Observations sur la Physique, 1785, 26:321-325, esp. p. 324; and H. E. LeGrand,"The 'Conversion'of C.-L. Bertholletto Lavoi- sier'sChemistry," Ambix, 1975, 22:58-70, esp. pp. 67-68. 26 Perin, "Triumphof Antiphlogistians"(cit. n. 4), pp. 49, 55-56, 62. 27 A. L. Lavoisierand J. B. Meusnier,"D6veloppement des dermie'rsexperiences sur la decomposition et la recompositionde l'eau,"in Oeuvresde Lavoisier (cit. n. 13), Vol. V, pp. 320-334. 28 C.-L. Berthollet,"Consid6rations sur les exp6riencesde M. Priestley,"Annales de Chimie, 1789, 3:63-114, on p. 70. 29 Lavoisierand Meusnier,"D6veloppement" (cit. n. 27), pp. 205-209; cf. Holmes, Lavoisier (cit. n. 8), p. 237. 30 See, e.g., J. C. De Lam6therie,"Discours preliminaire,"Obs. Phys., 1787, 30:3-45, esp. pp. 29-45; and Lam6therie,"Discours pr6liminaire," Obs. Phys., 1789, 34:3-55, esp. pp. 26-29. 42 JAN GOLINSKI inflammableand dephlogisticatedairs would combine to form water only in the conditions of extreme heat that Lavoisier had used. At lower temperaturesthey would form fixed air,as in normalprocesses of combustion.Of the analysisexperi- ment, Kirwanreiterated the traditionalview thatthe inflammableair emergingfrom the gun barrelwas phlogistondisplaced from the ironby water.3'Priestley concurred with this, but his view of the supposedsynthesis was differentfrom Kirwans.When he had repeatedCavendish's original experiment, he had alreadyconcluded that the waterproduced by the reactionof the two gases was presentin theircomposition in the gaseous state. The experimentshowed only that all "airs"contained some pro- portionof water.There was, however,another product, namely "nitrousacid" (later called nitricacid), which was producedby the combinationof the ponderablebases of the gases. Until the early 1790s Priestleycontinued to performthe experiment with what he still called "inflammable"and "dephlogisticated"airs and to record both water and nitrousacid as the products.Although Berthollet and other Lavoi- sians triedto cast doubton Priestley'smethods, asserting that the acid was the result of contaminationof his oxygen by atmosphericnitrogen, Priestley consistently denied such contamination.His position was endorsedby Keir, Deluc, Lametherie, and Watt,among others.32 As well as insisting upon these alternativeexperimental facts, Lavoisier'scritics teased apartthe rhetoricby which he constructedhis claims. Lavoisierhad sought to have his claims accepted as facts by assertingthat they followed directly from precise quantitativemeasurements. In relationto the 1785 demonstrationhe wrote, accordingto Kirwan'stranslation:

This double experiment... may be regardedas a demonstration,. . . if in any case the word Demonstration may be employed in natural philosophy and chemistry. ... The proofs which we have given of the decomposition and recomposition of water being of the demonstrative order, it is by experiments of the same order, that is to say by demonstrative experiments, which they ought to be attacked.33

Ratherthan acceptingthe challenge to confronthim on his own ground,however, Lavoisier'sopponents used variousstrategies to disconnectwhat were agreedto be the facts from theirpurported implications. Kirwanand William Nicholson, the editorof the second editionof Kirwan'sEssay on Phlogiston (1789), focused on the assumed connectionbetween precision and demonstration.Kirwan praised Lavoisier as "the first that introducedan almost mathematicalprecision into experimentalphilosophy," but denied thatone or a few accurateexperiments could overturndecades of chemical experimentationsupport- ing the phlogistontheory. There was no possibility that a single experimentcould

31 Richard Kirwan,An Essay on Phlogiston and the Compositionof Acids, 2nd ed., ed. William Nicholson (London, 1789), pp. 42-44. 32 Priestley,"Experiments and Observations"(cit. n. 18), pp. 282, 294-295; JosephPriestley, "Fur- ther ExperimentsRelating to the Decomposition of Dephlogisticatedand InflammableAir," Phil. Trans.,1791, 81:213-222; J[ames] K[eir], TheFirst Part of a Dictionaryof Chemistry(Birmingham, 1789), pp. 118-119; Deluc, "Lettrea M. De La MWtherie"(cit. n. 1), pp. 145-146; J. C. de Lamdthe- rie, "Mdmoiresur l'air phlogistiqu6(ou impur) obtenue par la combustion de l'air inflammable& de l'air pur,"Obs. Phys., 1789, 34:227-228; andJames Watt to JosephBlack, 8 June 1788, in Partners in Science: Lettersof James Wattand JosephBlack, ed. Eric Robinsonand Douglas McKie (London: Constable, 1970), pp. 166-167. 33 Lavoisier,as quoted in Kirwan,Essay on Phlogiston (cit. n. 31), pp. 59-61. PROOF IN LAVOISIER'SCHEMISTRY 43 be "demonstrative"of a conclusion as revolutionaryas the notion that waterwas a compound.Kirwan insisted that "thebook of natureshould be interpretedlike other books, the sense of which must be collected ... from an attentiveconsideration of the whole."When this was done, contradictionsbetween Lavoisier's claims andother well-establishedexperimental results were not hardto find. Kirwanpointed out that the decompositionof water was supposedlyeffected by iron but not by charcoal, while otherfindings showed charcoal had a greateraffinity for oxygen thandid iron. The doctrineof affinitiesyielded severalother problems for the antiphlogistictheory, as Lavoisierwas obliged to acknowledge.34 Nicholson took a slightly differenttack, in a sophisticatedcritique of Lavoisierfor making excessive claims for accuracyof measurement.After discussing the likely experimentalerrors in measurementsof weights and volumes,Nicholson concluded that Lavoisier'sfigures, with their long stringsof decimals, "exhibitan unwarrant- able pretensionto accuracy."That being so, the rhetoricallink from precision of measurementto certaintyof conclusion was broken.Measurements that showed a quite spuriousprecision could not be taken as proof of what was asserted.Taking up Lavoisier'sterminology of a "demonstrativeorder" of proof, Nicholson wrote: "Whenthe real degree of accuracyin experimentsis thus hiddenfrom our contem- plation, we are somewhatdisposed to doubt whetherthe exactitudescrupuleuse of the experimentsbe indeed such as to renderthe proofs de 1'ordre demonstratif"35 In an earlierwork Nicholson had alreadydiscussed and dismissedthe possibility that experimentalknowledge could attainthe certaintyof demonstration.Drawing upon the resources of British empiricistphilosophy, Nicholson had arguedin his Introduction to Natural Philosophy (1782) that experiments could not convey facts to the mind with sufficientimmediacy to give them the intuitivecertainty of mathe- maticaltruths. "The great perspicuity and certaintyof mathematicalknowledge," he wrote, "arisesfrom the simplicity of the ideas employed, and their not depending on any externalbeing." In experimentaloperations, however, it was not possible to presentall the relevantaspects of the situationto the mind directly and simultane- ously. Hence, "in general, we must be contentedwith less proof than demonstration."36The results of experimentscould not, therefore,be treatedlike the axioms of geometry-demonstrative certaintydid not inhere in the results of experiments, as Lavoisiermaintained. This appearsto have been the generalview among British naturalphilosophers, who sharedthe empiricistposition that the senses could not immediatelyperceive all the elements of a complex experimentalsituation. The English chemist Thomas Beddoes used the same terms in his discussionof the de- monstrativestatus of the watercomposition experiments, in his Observationson the Nature of Demonstrative Evidence (1793):

What for instance is it, that preventsme from being as certain,that waterconsists of hydrogeneand oxygene airs, as of any propositionin Euclid?-nothing surely but the incompetencyof my senses. ... Now if I could perceive the small quantityof azotic air present separatelyuniting with a certainportion of the oxygene air to form acid, while the hydrogeneair unites with the rest to form water;if I could see that the airs

3 Kirwan,Essay on Phlogiston (cit. n. 31), pp. 7, 304, 317. 35 Ibid., pp. viii, xi. 36 William Nicholson, An Introductionto Natural Philosophy, 2 vols. (London, 1782), Vol. I, pp. 1-6, quotationson p. 4. 44 JAN GOLINSKI

previouslycontain only a little or no waterbeforehand, and if there was no heat and light, I shouldhave demonstrativeevidence.

Following a relatedline of critique,Priestley also took aim at the meansby which Lavoisierhad soughtto demonstratethe factualityof his claims. Priestleyidentified andbrought into questionseveral elements of Lavoisier'smaterial, social, andrhetor- ical techniques,subjecting to scrutinythe validityand reliability of instruments,and the formof practicewithin which they were put to use. Priestley'sown epistemology stressedthe autonomyof individualjudgment and the equalityof all observers,and in his view Lavoisierthreatened to use his privileged access to instrumentalre- sources to impose his own authority.38Priestley refused to submit his interpretive judgmentto what he took to be such a nakedassertion of power.He chargedLavoi- sier with using apparatusso complexthat his experimentswere liable to errorand so expensivethat they were impossibleto replicate.The apparatusused by the French academiciansin February1785 was, he wrote, "extremelycomplex, as a view of theirplates will shew, and mine was perfectly simple, so that nothingcan be imag- ined to be less liable to be a source of error."For Lavoisierto use privateresources to develop specially refinedinstrumentation seemed to Priestleyto indicatehis re- fusal to submitto the social validationof widespreadreplication. The synthesisex- perimentrequired, Priestley noted, "so difficultand expensive an apparatus,and so many precautionsin the use of it, that the frequentrepetition of the experiment cannotbe expected; and in these circumstancesthe practisedexperimenter cannot help suspectingthe certaintyof the conclusion."In 1796 he was still insisting that the criticalexperiment on compositionof waterhad "notbeen sufficientlyrepeated." Summarizinghis position in 1800, he maintained:"Till the French chemists can make theirexperiments in a mannerless operose and expensive . .. I shall continue to thinkmy resultsmore to be dependedupon thantheirs."39 Priestleyand the other critics showed that to disputeLavoisier's claims required analysisof the meansby which those claims had been renderedas facts. Instruments that Lavoisierproposed as accurateand refinedwere portrayedby his opponentsas unnecessarilycomplex and liable to error.His mobilizationof specialist skills and investmentof substantialfinancial resources in his apparatuswere said to point to- wards an illegitimateconcentration of instrumentalpower. This would deny other investigatorsthe rightto contributetheir own observationsor to replicatehis experiments. His use of methodsof precisionmeasurement was also denouncedas claim- ing an unjustifieddegree of accuracy,and its purportedconnection with a "demonstrativeorder" of proof was denied. In these respectsLavoisier's opponents pointed to the fragilityof his experimentalpractice; they denied thatthe phenomenahe had producedcould be reproducedin othersettings.

37 Thomas Beddoes, Observationson the Nature of DemonstrativeEvidence (London, 1793), pp. 108-109. 38 Cf. McEvoy,"Enlightenment and ChemicalRevolution" (cit. n. 4), pp. 205-209. 39 Joseph Priestley,Considerations on the Doctrine of Phlogistonand the Decompositionof Water (Philadelphia,1796), ed. William Foster (Princeton:Princeton Univ. Press, 1929), pp. 17, 34, 41; and Priestley,The Doctrine of Phlogiston Establishedand that of the Compositionof WaterRefuted (Northumberland,Pa., 1800), pp. xi, 48, 50, 76-77. PROOF IN LAVOISIER'SCHEMISTRY 45

III. REPRODUCINGLAVOISIER'S TECHNOLOGY Althoughthe controversycontinued for severalyears, Lavoisier's instruments even- tually proved themselves relatively robust tools for extending acceptance of his claims.He showedthat it was possible to replicatehis crucialexperiments, including those on the compositionof water.This requiredreproducing certain features of the setting of the original experimentsat other sites. A certainamount of redesign of materialtechnology and its supportingpractices was also called for. To the extent thatthis succeededand the samephenomena were takento be reproducedelsewhere, the controversywas steadily closed in Lavoisier'sfavor. One relatively well documentedepisode in this process is that concerningthe Dutch chemist and experimentalphilosopher Martinus van Marum.Van Marum traveled to Paris in July 1785, a few months after the dramatic analysis-and- synthesisdemonstration. Although unable to witness thatevent, he had a brief meet- ing with Lavoisierand several conversationswith Monge regardingthe new anti- phlogistic theory.Returning to the Netherlands,he workedto replicateLavoisier's experimentsand, in February1787, wrote to Lavoisierand Monge declaringhis allegianceto the new chemistry.40He then devotedhimself to convertingother Dutch chemists. The criticalnecessity, as he realized, was for convincingand widespread replicationsof the relevant experiments,of which that on the synthesis of water seemed most important.This experiment,he noted, "hadnot previouslybeen performed outside Paris."Dutch chemists had remainedskeptical because "they had had no opportunityto see or to repeatexperiments, the resultsof which formedthe basic principlesof the new chemicaltheory. Indeed, the necessaryapparatus as made by the generousLavoisier at his own expense could hardlybe obtained,owing to its expensivenessand to the difficulty of constructingit with the precision required'" In particular,the gasometersrequired for the synthesis experimentwere prohibi- tively costly.4' Using the resourcesof Teyler'sMuseum in Haarlem,van Marumsolved the prob- lem by significantlysimplifying Meusnier'sdesign. Insteadof a moving tank sus- pended by a counterweightedbeam, van Marum'sapparatus used two vessels: one containingthe gas over water,the other a constanthead of water(maintained by an adjustabletap on a feeder vessel) to keep up a steadypressure and regulatethe gas flow.Linear scales attachedto the side of the gas-containingvessel enabledthe water level, andhence the gas volume,to be measured.With this simplifiedapparatus, van Marumfirst performed the synthesisexperiment in Haarlemin 1791, "beforeall the

40 Martinusvan Marumto Lavoisier,26 Feb. 1787; and van Marumto Monge, 26 Feb. 1787, in Martinusvan Marum:Life and Work,ed. R. J. Forbes, E. LeFebvre,and J. G. Bruijn,6 vols. (Haar- lem: Tjeenk, Willink & Zoon, 1969-1976), Vol. I, pp. 193-194, 255-256. See also T. H. Levere, "Martinusvan Marumand the Introductionof Lavoisier'sChemistry in the Netherlands,"ibid., pp. 158-286; and H. A. M. Snelders,"The New Chemistryin the Netherlands,"Osiris, 1988, 4:121-145, esp. pp. 127-130. 41 Martinusvan Marum,"Lettre 'a M. Berthollet,contenant la descriptiond'un gazometreconstruit d'une manierediff6rente de celui de MM. Lavoisier& Meusnier,"Ann. Chimie, 1792, 12:113-140, trans. in Martinus van Marum,ed. Forbes, LeFebvre, and Bruijn (cit. n. 40), Vol. V, pp. 245-259, 241-242 (quotations).A parallelmay be the slow replicationoutside France of Coulomb'sdetermina- tion of the law of electrostatic force, which was also embedded in the local practices of French engineering physics. See J. L. Heilbron, Electricityin the Seventeenthand EighteenthCenturies: A Studyof Early ModernPhysics (Berkeley:Univ. CaliforniaPress, 1979), pp. 475-476. 46 JAN GOLINSKI devotees of Physics or Chemistry who desired to be present." The result was a sig- nificant success for the Lavoisian theory. Van Marum recorded: "The simple and less costly gasometers I had used for this experiment were ordered here and imitated elsewhere, in order to repeat it in several places."42 Notwithstanding van Marum's claim that his apparatus allowed the synthesis experiment to be performed "with all the accuracy that can be desired," it seems clear that replicability was bought at the cost of some degree of precision. Van Marum declined to give any figures for the results of his experiments, saying only that they were "perfectly in agreement" with those of Lavoisier and his allies.43 Nor did he give an account of the calibration of his gasometers or the calculations to reduce gas volumes to weights. Perhaps a display of precise measurement and calculation of the kind Lavoisier had mounted in 1785 was not necessary in the context in which van Marum performed his demonstrations. More important for winning over a more extensive audience was a version of the experiment that could readily be replicated in a large number of locations. To this extent van Marum had succeeded by departing from Lavoisier's own rather uncompromising attitude to replication of his apparatus. In the Traite Lavoisier had written:

In the presentadvanced state of chemistry,very expensive and complicatedinstru- ments are become indispensablynecessary for ascertainingthe analysis and synthesis of bodies with the requisiteprecision as to quantityand proportion; it is certainlyproper to endeavourto simplify these, and to renderthem less costly; but this ought by no means to be attemptedat the expense of their conveniencyof application,and much less of theiraccuracy.44

In the event, simplifying and cheapening instruments did usually mean sacrificing their accuracy. But to extend acceptance of Lavoisier's chemical theories, the price was worth paying.

IV. CONCLUSION

Historians have noted how the "Chemical Revolution" brought with it new experimental apparatus and new standards of accurate measurement. What might not have been appreciated and certainly deserves further investigation is how integral these developments were to the achievement of Lavoisier and his allies. The lengthy and wide-ranging controversy he provoked inevitably raised numerous issues about the nature of scientific practice. The process of reaching a decision as to the facts of chemical phenomena was at the same time the formation of a consensus concerning how chemistry should be done. The program comprised a nomenclature and a text- book rhetoric; it was promulgated through social structures of communication and discipline; and it was embodied in new instrumentation and skills. Having tri- umphed, the revolution determined how history should be written: van Marum's

42 Van Marum,"Letter to M. Berthollet"(cit, n. 41), p. 242. 43 Ibid., pp. 250, 251, 255. 44 Lavoisier,Traits 6ljmentaire II, pp. 359-360, trans.Robert Kerr, in Elementsof Chemistryin a New SystematicOrder, Containing all the ModernDiscoveries (Edinburgh:William Creech, 1790), p. 319. PROOF IN LAVOISIER'SCHEMISTRY 47 experimentswere warrantedas valid replicationsof Lavoisier's,while Priestley's were condemnedas crudeand contaminated. Clearly,one task that historiansshould set themselvesis to overcomethis retro- spective shapingof events. Anothershould be to situatematerial technology in such a reconstructedhistory. Study of instrumentationcan lead to more than sterile anti- quarianism;it can open the door to a broaderappreciation of all the dimensionsof scientific practice.A focus on controversiesenables us to grasp the importanceof the materialculture of science, because when apparatusis disputedthe connections with specific forms of practiceand discourseare exposed.The links betweeninstru- ments and their usage and interpretationare explicatedin the course of attackor defense. We learn both that science is embodiedin firmlymaterial things and that it is nonethelesssocially negotiatedand historicallyvariable.