Philosophers of Experiment

Author(s): Ian Hacking

Source: PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association,
Vol. 1988, Volume Two: Symposia and Invited Papers (1988), pp. 147-156
Published by: The University of Chicago Press on behalf of the Philosophy of Science Association
Stable URL: http://www.jstor.org/stable/192879 .
Accessed: 08/05/2014 18:20

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 support@jstor.org.

The University of Chicago Press and Philosophy of Science Association are collaborating with JSTOR to
digitize, preserve and extend access to PSA: Proceedings of the Biennial Meeting of the Philosophy of Science
Association.

http://www.jstor.org

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
 Philosophers of Experiment

Ian Hacking

University of Toronto

The Neglect of Experiment: thatis the title of Alan Franklin's(1986). He did not
mean to imply that scientists were neglecting experiments,spinningwell financed cob-
webs of theories while laboratoriesdecayed for lack of funds. He meant that historians
and philosophersneglected the experimentalside of science. That was true, and is no
longer so. Although his title was fine when he was writing, the times have passed it by.
A decade before there had been almost no reflective philosophy of experiment. What
little had been published was not seen as writing aboutexperiment-that was not some-
thing to write about-but as discussion of the theory/observationdistinction,or the
impossibility of eliminating a theory by crucial experiment,etc. The even-handed
Dictionary of Scientific Biography discreetly cut articles on experimentersand expanded
those on theorists. ThaddeusTrenn's(1977) on the experimentaldiscovery of isotopes
was poorly received. The principleof not opening old wounds preventsme from quoting
here remarksin conversationmade by some of our most distinguishedhistorians,a pro-
pos of that 'tedious recountingof test tubes andjottings'.
The contrastwith the past 3 or 4 years is extreme. There have been historico-philo-
sophical internationalconferences devoted to experiment. They have dutifully produced
volumes of collected papers,just as if experimentwere a legitimate subdiscipline (Batens
and van Bendegem, 1988, Gooding et. al. 1988). The practiceis so new that I think that
it began with the three 'experimental'papersby Peter Galison, J.S. Rigden and Roger
Steuwer in Achinstein and Hannaway(1985).
There has been a growing numberof books. Often, as in the case of Galison (1987),
they present a rich tapestrywoven from incidents in the historyof science, but glowing
with philosophical colours. As I write, the currentissue of Isis (79, 1988, no. 3) is dedi-
cated to our topic. Several of the paperscontributedto PSA 1988 are about the philoso-
phy of experiment(Baird 1988, Stump 1988). Nor should we become fixated on such
local events; we must also turnto fundamentalstudies of experimentand technology that
find some of theirroots in the work of Habermas,e.g. Radder(1988).
So intense and continuing has been this activity thatI shall present a highly selective
retrospectiveexhibit of ten years of collective thinkingaboutthe laboratorysciences.
What has been the interestof experimentfor philosophers?Partof the answer is that we
have been addressingold questions in new ways: fact, fiction, forecast,rationality,justifi-

PSA 1988, Volume 2, pp. 147-156

Copyright? 1989 by the Philosophy of Science Association

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
148
cation, irrealism,demarcation,Duhem's thesis, and so forth. But I shall also develop new
themes, and show how thinkingaboutexperimentbearson philosophicalissues in ways
that have not yet been much noticed.
In no way do I wantto makeexperimentmoreimportantthantheory. One of my mes-
sages is the richness,complexityandvarietyof the scientificlife. In particular,theoryis not
one thing butmany,andexperimentnot one thingbut many. Philosophyof science has been
impoverishednot only by its obsessionwith theory,butalso with its complacentdoctrinethat
thereareone or at most two kindsof theories(e.g. real theoriesandbridgeprinciples). If we
may for a momentspeakof the experimentalliberationmovement,one of its aims has been
notjust to elaboratethe life of experiment,butalso to improvethe qualityof life for theo-
ries-along with makingthe theory/experiment distinctionnot obsoletebutmultifaceted.
In presentingthe work of others I should mention the obvious, thatthere is a division
between sceptics and admirersof experiment. That is nothing new. Philosophy of sci-
ence has always been riven by thatdifferencein instincts. The adulationof science was
characteristicof Popperand Carnap,however heatedtheir superficialdifferences of opin-
ion. Feyerabendwas sceptic, Lakatosadmirer,and Kuhn, seen as sceptic, probablyisn't.
It is a misfortuneof this session that both Galison and I respect science too much. The
chairman,Andrew Pickering,provides some valuablecounterweight.
In the late 1970's historians,philosophers,sociologists, anthropologists,and retiring
scientists began to write, in a reflective mode, aboutexperiments. On the surface,their
endeavourswere largely unrelated,althoughthis shift from theory to experimentas object
of enquirydoes betoken a largermovement in contemporaryculture. Once people did
begin to think aboutexperiment,those conductingsocial studies of science got there first,
startingwith Ravetz (1971). The sceptics among them riled admirersof science and pro-
voked heateddiscussions of evidence and rationality.
What data, what experiments,furnishsignificantevidence? Gilbertand Mulkay
(1984) interviewedinvestigatorsin bioenergeticsto see what results were regarded,not as
'crucial', but at least as 'key' to the recent developmentof theirfield. They reportedon
the basis of interviews thatthe "selectionof certainexperimentsas key did not depend on
any clearly identifiablequalities of the experimentsthemselves, nor even on the reception
reportedlygiven to the experimentsby contemporaryresearchers"(p.122). They doubted
thatthere is "any kind of data thatcan be used to provide a firm bedrockfor historical
descriptionand analysis"(p.124). They were writingfor historians,but the message for
philosopherswas manifest. This was not the old rationalismof Popperor Lakatos,teach-
ing that there are no crucial experiments,on the groundthat theoryis paramount. It is the
far more radicalclaim that as a matterof brutefact, there are no criteriafor distinguishing
key data. There is a blooming confusion. It may end in consensus, but not because the
communityis constrainedby the evidence.
This sceptical stance begged for counter-attack.According to Joseph Robinson
(1986), Gilbertand Mulkay lacked "anunderstandingof the particularscientific issues
involved in the specific cases" (p.52). To follow the discussion seriously,we have to
look up e.g. a 1966 NationalAcademy of Science paper"ATPFormationCausedby
Acid-Bath Transitionof Spinach Chloroplasts."These debates can be conductedonly by
bringingout what is specifically at issue in the laboratoryand its environs.
This tiny fracas illustratesliterallydozens of ratherdetailed controversiesbetween
sceptics and admirersof experimentalscience. Both partiesuse a detailed case, but want
the readerto infer, "that'show it is, everywhere".
Such debates have shiftedphilosophicalquestionsaboutevidence. Probabalistslike
Keynes and HaroldJeffreys,and laterCarnapand to some extent the studentsof subjec-

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
 149

tive, personalor judgmentalprobability,began with a clear problem. What is the relation

between the evidence (or, the availableevidence) and an hypothesisof interest?Carnap's
symbol 'c(h,e)' epitomized the conceptionof inductiverationalityand evidence. Later,as
the programmedegenerated,the questionbecame, how does new evidence modify my
structureof beliefs or judgements?Social studies of science, in the vein of Mulkayand
many others, put the very idea of "the"evidence (or, "the availableevidence") in question.
The doctrine of the strongprogrammein the sociology of knowledge is betterknown
than individualdisputes about spinachchloroplastsor whatever. The explanationof the
acceptanceof a propositionor a practiceshall not include "andthe propositionis true and
supportedby evidence, that's why people believed it" or "andthe practice works and
attainsits ends, that's why people adoptedit." Above all, don't say that a belief was
accepted because it was reasonableto do so! That is the strongestof attackson a timeless
concept of rationality(like the one epitomized as 'c(h,e)'. The idea is that what is ratio-
nal, and what counts as working, is determinedin an historicaland social setting. There
is no such thing as a good reason, tout court.
I find spinoffs of the strongprogrammemore fascinatingthanits original tenets. The
programmehas nothing in particularto do with experiments,but inevitably it has applica-
tion to them. I think of the exceptionally original work of Schafferand Shapin (1986).
Their "Hobbes, Boyle and the ExperimentalLife" is the most importantcontributionto
discussion of what A.C. Crombie (1981) called styles of scientific reasoning in the
Europeantradition. The book is about the introductionof a new style of reasoning, the
laboratorydemonstrationand the probingof natureby instruments. It is an epic whose
protagonistsare Hobbes and Boyle, with the latteras victor.
ErnanMcMullin(1988a) discusses the shapingof scientificrationality,of "whatcounts
as a good reason in a scientific argument,"of "second-leveldisagreements"(p.3).
McMullinrunsthrougha familiargamutof theory-obsessedcontributorssuch as Kuhnand
Lakatos. The lectureon which his paperis based was given just as the book by Shapinand
Schafferwas published. The book, with its imaginativeand controversialdiscussion of
how an experimentalstyle of reasoningwas put in place, transformsthe level of debate. A
new generationis speaking,one informedby carefulreflectionon experiment,ratherthan
theory. We can now considerin detail what it is for one style of reasoningto replaceanoth-
er (as the sceptic would put it) or how styles of reasoningevolve (as Crombiewould put it.)
This, I suggest, is how people in the futurewill addressMcMullin'stheme, "the shapingof
scientific rationality."Notice how we have moved on from the stridenttones of strongpro-
grammeconfrontations.The issue is less, "whatnon-rationalelementsdeterminedbelief?"
than,"how is it thatwe came to call these proceduresthe reasonableones?"Thatthis is my
own preferenceis evident from the use of styles of reasoningin Hacking(1982).
The tenets of the strong programmeare, in my opinion, a minor aspect of a broader
movement, the "social constructionof scientific facts" school. BrunoLatourcalls it con-
structionism;philosopherswill recognize it as close kin to Nelson Goodman's"skeptical,
analytic, constructionalistorientation"(Goodman, 1978, p.1). Latour'sideas are elabo-
ratedin his (1987), but here I attendmore to the LaboratoryLife of Latourand Woolgar
(1978). It is one of several recent books with "construction"in its subtitle,not to mention
Pickering's movement of the word "constructing"to the title of ConstructingQuarks
(1984). The great and long forgottenpioneer of this genre was of course Ludwig Fleck
(1935), writing about the constructionof the Wassermantest, and also, incidentally,writ-
ing about a style of reasoning,or Denkstil.
Constructionhas nothing special to do with experiment,as witnessed by anothercon-
struction-subtitlebook, Donald MacKenzie's (1981) on British statistics. It has, never-
theless, a strongoverlay of scepticism aboutexperiment. Facts settled by experimentare
(it is urged) facts only after they have been settled. Hence the idea of experimentcon-

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
150

strainingtheory by revealing facts is made to collapse. R.B. Braithwaite(1953) asked in

1946, "Is science invention or discovery?"His answerwas classic: "Manproposes a sys-
tem of hypotheses. Naturedisposes of its truthor falsity" (pp. 367-8). The second
clause has been constantlywhittled away at this past half century. The social construc-
tionists make out thatpeople, not nature,dispose of the truthor falsity of the proposed
hypotheses. The issue has been well joined. Robinson v. Gilbert& Mulkay,already
cited, is an instance of the debate. More importantare the mattersat issue between
Galison and Pickering. Their debate is, in my opinion, one good model (among others)
of how to carryon. On the one handthey can drawon instrumentationand experimenta-
tion that they know almost as well as the agents who did the work in the first place (weak
neutralcurrents,the bubblechamber). On the other,they turnthis into an epistemologi-
cal and metaphysicalparable,telling the way it is in a certainkind of science, in general.
There is, incidentally,a problemfor those who want to follow theirexample. Do we
have to learn about spinachchloroplastsand bubblechambers,and thousandsof cunning
instruments?The brief but wrong reply is: philosophersof theoryhave to learn evolution-
ary biology or quantummechanics, so what's new? The truthis that there are more kinds
of instrumentseven thanthere are subatomicparticlesor species. So many experiments,
so many details! The consequence is one you would not at first thinkof: the philosopher
of experimentmust be a betterwriter,a betterartist,have more highly developed literary
skills, thanthe philosopherof theory.
As an outsidersees it from a distance,Pickeringsays thatthereare very few constraints
on how experimentalwork will proceed,whereasGalison sees thatthereare a greatmany.
The outsiderwill notice Pickeringattendingto how the goals of a researchprogrammeare
negotiatedbetween the players,so thatsuccess is defined by the consequenceof social
interactions(Pickering,forthcoming). The same outsiderwill crassly say thatGalison must
have some truthon his side, becausemost experimentswon't work:that'sthe unnegotiable
constraint. But Galisonhas somethingsubtlerin mind, based on the way in which instru-
mentaltechnique,traditionand availabilitydetermineswhatit is possible to do in an exper-
imentalsituation,largelyindependentlyof the theoreticalstructuresin the background.

Pickering, on the other hand,in currentwork, increasinglymakes plain that he does

not think that anythinggoes. Like all of us, he proceeds by delving deeper than the bland
labels "experiment,""observation,"of old. He has three levels: theory,phenomenology
of the apparatus,and the materialinstrumentationand objects being investigated. He
calls these three plastic resources. In the course of an experimentalinvestigationyou
may change your account of how the apparatusworks, your accountof the world that you
are tryingto find out about,or may modify your instruments. The final product-in the
unusualevent thatthere is a final product-is a mouldingof the three of these together.

Philosopherswill see that Pickeringis extendingDuhem's thesis about auxiliary

hypotheses. Duhem's thesis has long been decked out in Quinery and hence been largely
irrelevantto real science. It is an importantachievementto returnit to its properstation.
I combine it first with Cartwright(1983, ch. 6) on phenomenologicallaws and approxi-
mation;secondly, with Ackermann(1985), a much more abstractand less example-ori-
ented vision of a dialectic involving data, instrumentsand theory. Cartwrighthas a good
deal to teach about how phenomenologicallaws of the apparatusare modified,
Ackermannabout how data are on the one hand materialobjects given regardlessof theo-
ry, and on the other hand used as signs andregularlyreinterpreted.
Duhem's thesis is widely regardedas an indeterminacythesis-and so it is, when
restrictedto theoriesand hypotheses. In his simplestand astronomicalexample, he thought
aboutchangingour theoryof the stars,andour changingthe theoryof the telescope;he did
not thinkaboutchangingthe way we make telescopes,or abouthow the dialectic between
our theories(auxiliaryhypotheses)abouttelescopes andour telescopes, each of which we

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
 151

modify to try to keep themin some kind of harmonywith each other. I readPickering's
extension of Duhem's thesis as implyingmore determinacythanindeterminacy.I do not
mean thatthe world pre-ordainswhat shall be our theoryof the stars,our theoryof our
apparatus,and our apparatus.I do thinkthatonly a few such combinationspersist,and that
the plasticitynoted by Pickeringturnsinto a sortof glue thatkeeps so much of our science
stable. PatrickHeelan and I will discuss thatlaterthis year. (Heelan 1988, Hacking 1988).
Now I shall turnto realism. Latourand Woolgardescribeda discovery that won a
Nobel prize. The first readersof the book read it in the anti-rationalitymode: the research
reacheda successful conclusion not because the competinginvestigatorsproducedcom-
pelling evidence, but because they negotiatedwith the largercommunityof endocrinolo-
gists and compelled acceptanceof theiranalysis and synthesisof a particulartripeptide.
That is indeed a sceptical theme of theirbook, and admirersof the scientific achievement
naturallyinsist that therewere far more constraintson the laboratories-constraints
imposed by reason and nature-than LatourandWoolgarwere willing to countenance.
Yet the subtitleaboutconstructingfacts makesplainthattheiriconoclasmis directedelse-
where. Theirbook is the most powerfulworkof scientificanti-realismto have emergedin the
pastdecade. It is entirelydifferentfrominstrumentalist anti-realism.VanFraassen'scon-
structiveempiricismtakesfor grantedthata given theoryeitheris, or is not, empiricallyade-
quateto the phenomena.He is an admirerof science. He has no scepticismaboutphenome-
na, and neverconsiderswhetherfacts areconstructedbeforetheoriescan be adequateto them.
Conversely,LatourandWoolgarhave in principleno anti-realistinstinctsaboutunobservable
(theoretical)entities. They claim only thatthey don'texist untilthey areconstructed.It is
partof the rhetoricof science, they say,to eraseall memoryof the construction,so thatwe
speakof discoveringphenomenaandof discoveringthe ("unobservable") structureof a
tripeptide.Once again,I urgecomparisonwith Nelson Goodman. To distinguishthis radical
positionfrom the merelyverbalanti-realistscience-admiringperspectiveof people like van
Fraassen,we shoulduse Goodman'sself-appellation,andcall Latouran irrealist.
As an admirerof science I restrainmy enthusiasmfor this kind of irrealism. Hacking
(1988b) sketches the developmentof TRH-the substancethatis the topic of Latour's
book-over the past ten years. It furnishesadditionalconsiderationsin favourof Latour's
story. My purpose,however,was not merely to welcome the book into the fold of more
conservativephilosophy. It was ratherto show thatthe "constructionist"story can be
retold in an entirelynon-constructionistway, so long as you do not thinkthatthereis one
uniquedescriptionof the real world thatis the ideal endproductof inquiry. You can under-
standthe negotiationsso highlightedby Latourand Woolgaras negotiationsaiming at set-
tling one possible description,at agreementon one set of criteriaforjudging the specifics
of some endocrinologicalexperiments. The description,says the conservativethinker,was
always trueof the world, and not made true. It excludes otherpossible descriptions,more
on groundsof incommensurabilitythaninconsistency. Thereis no uniquelyright descrip-
tion, but thatis just a pleasantmeta-factaboutthe world and its describers. To say this is
not to become subjective. It is to become pluralisticin one's meta-physics. Most descrip-
tions won't wash (denialof subjectivity). Thereis no reasonto thinkthatonly one will
(pluralism). The world is so complex thatwe cannotcompose the one truecomplete story
aboutit. There is no one exhaustivetrue story:the idea does not make sense, as P.F.
Strawson(1959, p. 128) wisely noted long ago, speakingof Leibniz.
Is there then nothing at issue between Latour'sirrealismand my wishy-washy plural-
istic realism? On the contrary. I thinkthat there is some truthin the notion of natural
kinds (althoughI don't think there are canonical, uniquely natural,kinds). The classifica-
tions of the human sciences are differentfrom naturalkinds; I call them humankinds.
Latourdoubts thatthere is any importantdifference. (Hacking 1988a) Here we are again,
discussing an old question, the identityor disparitybetween the naturaland humansci-
ences. But we are doing it on largely new and chiefly experimentalground.

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
152
This thoughtnaturallyleads on to the traditionalquestionof demarcationof science
from pseudoscience. Thatwas a significantpracticalproblemfor late 1920's Viennaand
Berlin. It is commonly said thatthe whole issue is dead. Yet, as Steve Fuller(1988, ch. 7)
urges,otherproblemsof demarcationsurfaceat present. We can witness some in an ongo-
ing spat in which RichardRorty(1988, p 54) attacksBernardWilliams(1985, p.139). But
those two men debatesolely at the level of theoryand"realism",runningaroundthe usual
squirrelcage of theory-obsessedphilosophy. They barelyacknowledgeinteractionwith the
world, and when they do so, do it is in termsof thatjaded hack,"informationandcontrol."
Thatis a phrasethatconceals the complexitiesof the ways in which we interactwith the
world, and implies thatit is the master,theory,thatdoes the informingand the controlling.
In my opinion one of the fundamentalways in which laboratoryscience and ethics
separateis that the formerengages in what I have called the creationof phenomena:the
purifying and maintainingof phenomenathatdo not exist in a pure state anywherein the
universe. (Hacking 1983, ch. 12). It is those phenomenato which theory answers.
Phenomenamay be maintainedin the laboratory,broughtback when interesting,or trans-
formed into off-the-shelf transportabletechnology. This notion of purifying,creating
and regularizingphenomena(and hence the world thatwe inhabit)certainlyinvolves
thinkingand theorizingabout the materialworld, but it also involves interactionwith the
world, and, in an unmetaphoricalsense of the words, remakingit.
I have most often used examples of creationtaken from physics, so it is importantto
insist that this is by no means necessary. As I wrote this paper,the two most publicized
reportsin Nature(13 September1988) and Science (22 September1988) were from
immunology. Researchershave successfully implantedmajorpartsof the human
immune system into mice, giving them a small working model thatcan be used in testing
drugs and vaccines, and, more importantly,enablingthem to investigate this mysterious
thing we call an immune system withoutdoing experimentsof humansubjects. Both
groups used a special strainof mice that have a genetic defect: they lack any immune sys-
tem. So they are shortlived,and indeed persistonly because experimentersmaintainthe
race. Using differenttechniquestwo differentgroupshave implantedhumanimmune
systems into such mice. This phenomenon,of the humanimmune system living in genet-
ically defective mice, is new to the universe. It is one of many kinds of example of what
I call the creationof phenomena.
I do not quarrelimmediatelywith the constructionistwho says thatthe "result",of a
humanimmunesystem living in defective mice, is countedas resultonly in consequence
of a lot of negotiations. Nor do I discountthe obvious truththatthe widespreadreporting
of these results is due to AIDS. I say only thatwhatevermixtureof experimentalingenu-
ity, the proclivityof nature,and the negotiationof competitiveresearchgroupsbroughtthe
phenomenainto being, a phenomenonwas broughtinto being thatdid not exist outside the
laboratory.We did not find it lying about,hiddenlike some lost island shroudedin mist.
The world with which we interactdoes not fix just what discoveries or inventions
shall be made, nor does our past knowledge determinewhat we shall count as a discov-
ery. It is not preordained which instrumentswe shall devise, nor which phenomenawe
shall be able to stabilize or purify. A science can develop among many possible paths,
bringinginto being differentphenomena. Phenomenathat we create on one possible his-
torical path might not be createdon anotherhistoricalpath, say because we have invented
neitherthe instrumentsnor the theoryof the instrumentsthroughwhich we could recog-
nize or control them. Moreoverany paththat we do follow has its own momentum.
Experimentaltechniquesand instruments,when they producewhat are taken to be stable
results, themselves suggest furthersteps to take by analogy. Had we not startedout on a
path, we would not have createdlateron what we do in fact create. Thus this vision of
experimentalscience is far more open thanthatof the philosopherof theory,who typical-
ly imagines that we are aiming at the one truthabout a subjectmatter.

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
 153
I speak of possibilities in the future. Many are open. But our knowledge also closes
off possibilities. A pictureis this: we set up benchmarksas we choose our paths through
possible things to establish. Ourprocedurescome to fix things-and these may be proce-
dures as direct as the invention and perfectionof modes of instrumentation,which then
define questions for us and ranges of possible answers.
Instrumentationleads me to anothertopic, one that used to occur at each of our bien-
nial meetings as a special symposium:the unity of science. I believe in the disunity of
science, largely for the reasons PatrickSuppes (1979) urged at our meeting a decade ago.
But science does have a unity quite differentfrom that of the GUTs. It is kept togetherby
a motley of disunified unifiers. Most importantis one we now ignore, namely mathemat-
ics-Galileo's language of the authorof nature,the differentialand integralcalculus, the
Langrangiansand Hamiltoniansof the nineteenthcentury. We have some rathernew uni-
fiers, such as endemic statisticaltechniques,and brandnew ones, such as fast computa-
tion that transformsthe articulationof theory,the processing of data, and the simulation
of synthetics (be they molecules or metals). There is anothergreat unifier centralto my
presenttheme: instrumentsand apparatus.
Throughoutmost of the twentiethcentury,regimesandpracticesof experimentationand
instrumentation have been a more powerfulsourceof unitythangrandunifiedtheories.
Instrumentswere speedily transferredfromone disciplineto another,not accordingto theo-
reticalprinciplesbut in orderto interfacewith andparticipatein the creationof phenomena.
An extremeversionof this 'instrumentalist'thesis:it is not high level theorythathas stopped
the innumerablebranchesof science fromflying off in all directionsand becomingdifferent
cultures. It is ratherthe pervasivenessof a widely sharedfamily of devices. Nuclearmag-
netic resonancespectrometers,once on the frontierof experimentalphysics, are now the
pedestrianstuff of biochemicalassays. Not becausethe biochemistthinksthatthe physics
used to make the device is right,butbecausethe instrumentis there,the one to use.
The unity of science is workedby a motley of diverseunifiers. Theory-orientedphilos-
ophy gave us a pictureof natureas not only some hegemonicunified totalitybut also as
passive and inert. It made us thinkthatwe discover her properties,we reveal her secrets. I
saw a strikingimage of this the otherday. 1905 has been called the annus mirabilis,the
year in which Einsteinput forthnot only the special theoryof relativity,but also the photon
accountof light, and the full understandingof Brownianmotion in statisticalmechanics. In
Novemberof thatyear the king of Portugal,a patronof science and literature,made a state
visit to Paris. He was presentedwith a statueby Brassaisover a metrehigh:La naturese
decouvrentdevantla science. A handsomeyoung woman, Science, is lifting a cape high
over her head,undressingherself to reveal her exposed bosom and sightly body.
The sculpturemeans many things. Natureis something to be discovered, revealed, a
sexual object who can be got to undressbefore the inquisitive male, who will then domi-
nate and possess her. She is passive. All thatmust be done is to take off her clothes. The
given is the woman, nature,who gives herself, exposes her one truth. And the science
thatpossesses her is theoreticalscience, which discovers this passive being.
That is an image entirelyat odds with the experimentalattitude. But our experimental
predecessorswere equally forthrightin theirsexual imageryof science and nature. We
had a notableexperimentalphilosophyof science long before the presentone that is
emerging. It was the philosophy of Bacon, or at any rate thatnamed Baconian. It inaugu-
ratedthe aggressive master-slavepictureof the male master,who did not so much uncover
as interferewith MistressNature. This, as writersin the women's movementhave amply
documented,has been incorporatedinto our accountsof how natureherself works, with a
conscious attemptto find controllingforces, triggeringmechanisms,targetsand the like, a
pictureof naturein the image of macho militarism(Merchant1980, Keller 1985).

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
154
With what images will a new mode of philosophizingaboutexperimentreflect its
underlyingand implicit attitudes?Self-consciousness aboutthe sexual imagery of the old
philosophies, both theoreticaland Baconian, has emergedat just the same time as the new
thinkingaboutexperimentationitself. I hope that the image of the futurewill be one of
the experimentercollaboratingwith natureratherthanmasteringit.
This tendency will assuredlybe augmentedby the fact that the most imaginativesci-
ences of the presentdecade are biological and astrophysical,one a life science and one
scarcely a laboratoryscience at all. It may be thatthe self conceptionof the experimental
life sciences will displace the role model set by physics. Physics has long lived by the
following pictureof an experiment:thereis a target,some apparatusused to interfere
with the target,and a detectorused to determinewhat is the effect of the interference.
Those are the words and idea of James ClerkMaxwell. If one resists the militaryover-
tones of targets,recall Rutherford'simageryof splittingthe first atom;he comparedhis
alphaparticles to shells from a 19 inch gun, at thattime the noblest achievementof the
Royal Navy. The more biology liberatesitself from a desire to emulate physics, the more
the physicists' conception of theirwork may in turnbe modified.
That would involve a change in our idea of how we relate to nature. The Baconian
image of the man-scientistinterferingwith the woman-naturewas projectedon to nature
itself. The overridingtheme was one of centralcausal structuresthatdominatedevery-
thing that happened. One can speculatethatif an image of experimenteras 'biological'
collaboratorwere to take hold, we would come to thinkof the autonomousand indepen-
dent activities of naturein a differentlight.
Science has from time to time servedas a model for all culture,most notablyin the
Enlightenment,but also duringthe nineteenthcenturywhen the confrontationbetweenreli-
gion and science so troubledreflectiveEuropeans.At presentit does not seem a model, but
thatis because science itself is so ill understood.The commonimage of science remainsa
modified versionof Enlightenmentscience. The humanitieshave clung to the
Enlightenmentimage of science as a grandunifyingintellectualadventure,one thatstrives
to find the ultimatetheoryof everything. Ourcivilizationnow values accommodation,
variety,choice. It denies foundationsbut yearnsfor a stabilitythatensurescoexistence of a
multitudeof interests. It wants tolerationandrespect,not unifiedhegemony. As the human
sciences have become more and more diversified,as writingandcomposingand dancing
and designinghave become more varied,science had been cast as a stereotypedmonolith.
But science has become as multifloriateas the humanities. It has become a domainin
which therecan be stabilitywithoutfoundations,sharingwithoutcommensurability.It is a
domainthatfavoursrealismaboutthe materialworld,with a maximumof varietybut a
minimumof subjectivism. It has become a domainin which therecan be coherentaction
within a thoroughlydisunifiedworldpicture. I have been speakingfor a modest scientific
humanism,which we might call an experimentalhumanism. It will be rooted,perhaps,
more in the life sciences, thanin physics, thatold bastionof theoryand of unity.

References

Achinstein,P. and Hannaway, O. (eds). (1985), Observation,Experimentand

Hypothesisin ModernPhysical Science. Cambridge,Mass.: M.I.T. Press.
Ackermann,R. (1985), Data, Instruments,and Theory:A Dialectical Approach
to UnderstandingScience. Princeton:PrincetonUniversity Press.

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
 155

Baird,D. (1988), "Five Theses on InstrumentalRealism",PSA 1988, 1: 165-173.

Batens, D. and van Bendegem,J. P. (eds.). (1988), Theoryand Experiment:
RecentInsights and New Perspectiveson theirRelation. Dordrecht:Reidel.

Braithwaite,R.B. (1953), Scientific Explanation(1946 TamerLectures).

Cambridge:CambridgeUniversity Press.
Cartwright,N. (1983), How the Laws of Physics Lie. Oxford:ClarendonPress.
Crombie,A.C. (1981), "PhilosophicalPresuppositionsand Shifting
Interpretationsof Galileo", in Probabalistic Thinking,Thermodynamics
and the Interactionof the History and Philosophy of Science, J. Hintikka
et. al. (eds.). Dordrecht:Reidel.

Fleck, L. (1935), Entstehungund Entwicklungeiner wissenschaftlicher

Tatsache. Einfuehrungin die Lehre vom Denkstil und Denkcollectiv.
Basel: Schwabe. F. Bradley and T. Trenn(trans.)(1979), Genesis and
Development of a Scientific Fact. Chicago: Chicago University Press.
Franklin,A. (1986), The Neglect of Experiment. Cambridge:Cambridge
University Press.
Fuller,S. (1988), Social Epistemology.Bloomington,Ind: IndianaUniversityPress.
Gilbert, G.N. and Mulkay,M. (1984), "Experimentsare the Key: Participants'
Histories and Historians'Histories of Science", Isis 75: 105-25.

Gooding, D., Schaffer,S., and Pinch, T. (eds.). (1989), The Uses of Experiment:
Studies of Experimentationin the Natural Sciences. Cambridge:
CambridgeUniversity Press.
Goodman, N. (1978), Waysof Worldmaking.Indianapolis:Hackett.

Hacking, I. (1982), "Language,Truthand Reason",in Rationalityand

Relativism, M. Hollis and S. Lukes (eds.). Oxford, Blackwell: pp. 48-
66. Enlargedin "Styles of Scientific Reasoning",in Post-Analytic
Philosophy, J. Rajchmannand C. West (eds.) New York:Columbia
University Press, 1985, pp. 146-165.
?______ (1983), Representingand Intervening. Cambridge:Cambridge
University Press.
_ _ __ __ . (1988a), "The Sociology of Knowledge about Child Abuse",Nous
22: 53-63, followed by comment by B. Latour,64-7.

?____ __ . (1988b), "The ParticipantIrrealistat Largein the Laboratory",

British Journalfor the Philosophy of Science, 39: 277-294.

?____ __ . (1988c), "On the Stabilityof LaboratoryScience", The Journal of

Philosophy, 85:507-14.
Heelan, P. (1988), "AfterExperiment:Realism and Research",The Journal of
Philosophy, 85: 515-524.

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions
156
Keller, E. F. (1985), Reflections on Genderand Science. New Haven: Yale
University Press.
Latour,B. andWoolgar,S. (1979), LaboratoryLife: The Social Constructionof
ScientificFacts, Sage, Beverly Hills andLondon. Secondunalterededition
with additionalprefaceand the word"Social"strickenfrom the subtitleon
the groundsthatit is redundant,Princeton:PrincetonUniversityPress, 1986.

Latour,B. (1987), Science in Action. HarvardUniversityPress, Cambridge,Mass.

MacKenzie, D. (1981), Statistics in Britain 1865-1930: The Social Construction
of Scientific Knowledge. Edinburgh:EdinburghUniversityPress.
McMullin, E. (1988a), "The Shapingof Scientific Rationality:Constructionand
Constraint",in McMullin 1988b, pp. 1-48.
_ _ _ _ _ _ __ (ed.) (1988b), Constructionand Constraint:The Shaping of
Scientific Rationality. Notre Dame, Ind.:University of Notre Dame Press.
Merchant,C. (1980), The Death of Nature. San Francisco:Harperand Row.

Pickering, A. (1984), ConstructingQuarks:A Sociological History of Particle

Physics. Edinburgh:EdinburghUniversity Press.
__ __ __ _ . (forthcoming), "Big Science as a Form of Life:, in The
Restructuringof the Physical Sciences in Europeand the United States,
1945-1960, M. de Mariaand M. Grilli (eds.). WorldScientific
Publishing, Singapore. de Mariaand Grilli.
Radder,H. (1988), The Material Realizationof Science. Assen/Maastrict:van
Gorcum.

Ravetz, J.R. (1971), ScientificKnowledgeand its Social Problems. Oxford:

ClarendonPress.

Robinson, J.D. (1986), "AppreciatingKey Experiments",BritishJournalfor the

History of Science 19: 51-56.
Rorty,R. (1988), "Is Science a NaturalKind?"in E. McMullin (1988b). 49-74.
Shapin, S. and SchafferS. (1986), Leviathanand the Air Pump: Hobbes, Boyle
and the ExperimentalLife. Princeton:PrincetonUniversity Press.

Stump,D. (1988), "The Role of Skill in Experimentation",PSA 1988, 1: 302-311.

Suppes, P. (1979), "The Pluralityof Science", PSA 1978, 2: 3-16.
Trenn,T. (1977), The Self-SpittingAtom:A Studyof the Rutherford-Soddy
Collaboration. Taylorand Francis:London.

Williams, B. (1985), Ethics and TheLimits of Philosophy. Cambridge,

Mass.:HarvardUniversity Press.

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 18:20:31 PM

All use subject to JSTOR Terms and Conditions