The United Nations Educational, Scientific, and Cutural Organization

Internationalorganizations as teachersof norms: the United Nations Educational,Scientific, and Cutural Organizationand sciencepolicy MarthaFinnemore

The structureof states is continuallyevolving. Since theirestablishment in Europe some fivehundred years ago and particularlysince World War I, states have grown in terms of both the varietyof tasks they performand the organizationalapparatuses with which they perform these tasks. The researchoutlined below investigatesthe causes underlyingthis process of state change in the case of one recentlyadopted set of statebureaucracies, those designedto coordinatescientific research. In the last fiftyyears science policymakingorganizations have sprungup in virtuallyall developedcountries and in most developingones. Most explanationsfor the appearance of these new pieces of state machineryfound in the political science or economics literaturesdescribe this development as demand-driven,that is, some domestic groupperceives a problemto whicha sciencepolicy bureaucracy is thesolution. Social groups such as producers of science (e.g., scientists)or consumers of science (e.g., technology-intensivebusinesses) may come to perceive that state coordinationand directionof a growingscience establishmentare in theirinterest. State officialsmay come to perceivethat the intimaterelation- ship between science and securitymakes controlof science in the national interest.Depending on the perspectiveadopted, one would predict different configurationsof science bureaucraciesserving different interests, but in all cases, the impetusfor creatingthose organizationswould be a demand by state or societal actors that the governmentshould direct and control science. This studyquantitatively tests these demand-drivenhypotheses by compar- ing a varietyof indicatorsof stateconditions that have been arguedto prompt demand with the timingof adoption of science policy bureaucracies.The

I am gratefulto Laura Helvey, Peter Katzenstein,Steve Krasner,Forrest Maltzman, Rose McDermott,John Meyer, John Odell, FranciscoRamirez, Nina Tannenwald,Kurt Weyland, and twoanonymous reviewers for helpful comments on earlierdrafts.

IntemationalOrganization 47, 4, Autumn1993, pp. 565-597 ?3 1994 byThe 10 Foundationand the MassachusettsInstitute of Technology 566 InternationalOrganization resultsprovide little support for any of the demand-drivenhypotheses. Consequently,an alternativeexplanation is investigated.Early in thediffusion of thisbureaucratic innovation, several international organizations took up sciencepolicy as a cause and promotedit amongmember states. The article tracesthe process whereby one of these international organizations, the United Nations Educational,Scientific, and CulturalOrganization (UNESCO), "taught"states the value and utility of science policy organizations. I arguethat the creation of this teaching mission, whereby UNESCO would supplythe organizational innovation to states,was a reflectionof a newnorm elaboratedwithin the international community.' This norm held that coordinationand directionof science are necessarytasks of the modern state and that a sciencepolicy bureaucracy having certain well-specified characteristics was the appropriatemeans to fulfillthose tasks. States createdscience bureaucracies,with UNESCO's help,to complywith the new normabout states'responsibility for science.Thus, the organizationalinnovation was suppliedto states from outside, from an internationalorganization, rather than beingthe productof anycharacteristics internal to or inherentin the state itself. The articlemakes contributionsto three differenttheoretical debates ongoingin the field.First, the findingsoutlined above lend supportto constructivistor reflective theoretical approaches that treat states as social entities,shaped in part by internationalsocial action.State policiesand structuresin thiscase are influencedby changing intersubjective understand- ingsabout the appropriate role of the modern state. However,in most cases the causes of those changed understandings lienot at thenational level but at thesystemic level: it is an internationalorganization' thatpersuades states to adoptthese changes. Thus, a secondcontribution of this articleis to demonstratethe role of internationalorganizations as principals,rather than agents, in internationalpolitics. Finally,the article raises questions about the nature and roleof epistemic communities.While many of the UNESCO officialsinvolved in this reorganizationof international science had scientificcredentials, their reasons for acting had moreto do withtheir status as internationalbureaucrats than with their professionalsocialization or principled beliefs about science. This suggests that the "epistemic" aspect of groups may not always be their most important featureand thatcaution is warrantedin ascribingcausal status to specialized knowledgewhen explaining political behavior.

1. Forpurposes of this article a "norm"is definedas a rulelikeprescription which is both clearly perceptibleto a communityof actorsand whichmakes behavioral claims upon thoseactors. Althoughcomments of McElroywere influential in formulatingthis definition, McElroy's own definitiondiffers significantly from mine; see RobertMcElroy, Morality and AmericanForeign Plicy: TheRole ofMoral Nonns in InternationalAffairs (Princeton, N.J.: Princeton University Press, 1992). UNESCO 567

The developmentof science policy

The relationshipbetween states and science by no means begins with the establishmentof formalstate science policy bureaucracies.2National academies and royalsocieties of science,many of whichenjoyed some amountof state sponsorshipand whose memberswere in frequentcontact with govern- mentofficials, date back to the seventeenthcentury. Similarly, state-sponsored universitiesoften housed scientistsand theiractivities. However, state sponsorshipof the sciencesin thisearly period was understoodto be analogousto state sponsorshipof the arts; greatnessand accomplishmentin arts and sciences reflectedstate power ratherthan being a means to achieve power. Further, patronageof thiskind usuallyentailed minimaldirection and control.Acad- emies and universitiesmay (or maynot) have benefitedfrom state funding, but theywere not part of the state apparatus and wereleft free to pursuetheir work witha minimumof stateinterference. The modern concept of science policy differson both these issues. It understandsscience as a means to nationalpower and consequentlyseeks to bringscience activity under the control of the state. Most oftenthis has entailed the creationof a new piece of state apparatusdedicated explicitlyto thistask. The firsteffort to do thiswas made bythe Britishin 1915when they established the Departmentof Scientificand IndustrialResearch to wean Britishscience and industryfrom continental, especially German, innovations, expertise, and technicalequipment during World War 1.3 A few Commonwealthmembers mimickedthe Britishlead and establishedsimilar organizations, but it was not untilafter World War II thatscience policy bureaucracies became widespread. Before 1955 onlya handfulof countries(fourteen) had such entities;by 1975 eighty-ninecountries did. This researchseeks to explainhow and whythe state interestin and use of sciencechanged in thisway.

2. The historyof states'changing attitudes toward science obviously is muchmore complex than theoverview presented here. For moreon thissubject, see JosephBen David, "The ScientificRole: The Conditionsof Its Establishmentin Europe," Minerva4 (Autumn1965), pp. 15-54; A. Hunter Dupre, Sciencein theFederal Govemment:A Historyof Policies and Activitiesto 1940 (Cambridge, Mass.: Harvard UniversityPress, 1957); Philip Gummett,Scientists in Whitehall(Manchester, England: ManchesterUniversity Press, 1980); Ros Herman, The European ScientificCommunity (Harlow, England: LongmanPress, 1986); Eric Hutchinson,"Scientists as an InferiorClass: The EarlyYears of the DSIR," Minerva8 (July1970), pp. 396-411; Daniel Kevels, ThePhysicists: The Historyof a ScientificCommunity in ModemAmerica (New York: AlfredA. Knopf,1978); Frank Pfetsch,"Scientific Organization and Science Policy in Imperial Germany,1871-1914: The Foundingof the Imperial Instituteof Physicsand Technology,"Minerva 8 (October 1970), pp. 557-80; JarlathRoyane, Science in Govemment(London: Edward Arnold, 1984); Ian Varcoe, "Scientists,Government, and OrganizedResearch in Great Britain,1914-1916: The EarlyHistory of the DSIR," Minerva8 (April 1970), pp. 192-216; and RobertWuthnow, "The World Economy and the Institutionalizationof Science in SeventeenthCentury Europe," in AlbertBergesen, ed., Studiesof the Modem World-System(New York: Academic Press,1980), pp. 57-76. 3. Peter Alter,The ReluctantPatron: Science and theState in Britain,1850-1920 (Oxford:Berg, 1987), pp. 201ff. 568 InternationalOrganization

For purposesof thisstudy I definescience policybureaucracies as organsof the state that have as their primarymission the tasks of coordinating, organizing,and planningscientific and technologicalactivities at a national level. I exclude frommy definitionthe followingtypes of organizations:(1) nonstateorganizations (such as scientists'professional societies); (2) organizations dealingwith only one branchof science (such as the National Weather Service or medical and health organizations);(3) educational organizations whose primarymission is to trainscientific and technicalpersonnel rather than coordinateactivities broadly; and (4) researchorganizations whose primary missionis to conductresearch rather than to make policy.This definitionis based on UNESCO's definitionused in compilingits world directoriesof national science policymakingbodies and so ensures that the UNESCO activitieschronicled below and myanalysis concern the same phenomena.4

Demand-drivenexplanations for science policyorganizations

Most explanationsfor the creationof new state bureaucraciestrace the cause to some changein materialconditions that reconfigures the interestsof actors withinthe state. Functionalistsmight regard such an objectivechange to be sufficientas well as necessaryfor the new bureaucracyto appear. Othersless sanguineabout the efficacyof politicalsystems in meetingall needs or fulfilling all functionswould regardchange in materialconditions only as a necessary conditionand look to the process by which demands are voiced and, once voiced, are realized for sufficientconditions. Even in this latter set of explanations,however, some material change must prompt the demand- makingprocess. Thus in most explanationsthere is some prerequisitecondition associated withthe creation of new state bureaucracies.5 Three kindsof prerequisites have been argued to be relevant.The firstare what I call issue-specificconditions. Here, it is the situationin the issue-area particularlyrelevant to the new

4. The firstof these directoriesappeared duringthe 1960s. See United Nations Educational, Scientific,and Cultural Organization(UNESCO), WorldDirectory of National Science Policy- makingBodies, 3 vols. (Paris: UNESCO, 1966-68). Volume 1 coveredEurope and NorthAmerica; volume2, Asia and Oceania; and volume 3, Latin America.A second directorywas publishedin 1984. See UNESCO, WorldDirectory of National Science Policy-makingBodies, Science Policy Studies and Documents Series, vol. 59 (UNESCO: Paris, 1984). A second edition of this 1984 directorywas publishedin 1990. See UNESCO, WorldDirectory of National Science Policy-making Bodies,Science PolicyStudies and DocumentsSeries, vol. 71 (Paris: UNESCO, 1990). I have made severalrefinements to the UNESCO definitions.For furtherexplanation, see the appendix. 5. I have borrowedthe term"prerequisite" from Collier and Messick'sanalysis of the spreadof social securityacross states. See David Collier and Richard Messick, "Prerequisitesversus Diffusion:Testing AlternativeExplanations of Social SecurityAdoption," AmericanPolitical ScienceReview 69 (December 1975), pp. 1299-315. UNESCO 569 organizationthat prompts its creation. Applied to science,this argument links the creation of a state science policymakingapparatus to the growthand strengthof the domesticscience community. An argumentof just thistype has been made by David Dickson to explainthe originsof sciencepolicymaking in the United States.6 In Dickson's view the growthof the domestic science establishmentprompted the creationof a statescience policy apparatus in two ways. On the one hand state actors saw a science policybureaucracy as an opportunityto directand controlthis new activity.On theother hand scientists saw such an organizationas a potentialconduit for state aid and coordination. This thesiswould predictadoption of science policyorganizations to be highly correlated with domestic levels of science activity,for example, with the numberof scientistsin the countryand the amountof researchand development(R&D) spending. The nexttwo typesof conditionsapply to consumersrather than producers of science. Developmentor modernizationlevels are argued to promptthe creation of science policy entities throughthe actions of the economic consumersof science,particularly industry. The idea here is that as a state's economydevelops, it will become more technology-intensiveand so require morescientific support. Economic actorstherefore put pressureon the stateto organize and supplythis support; a new science policy organizationis the result.In manymixed economies, these actorsmay be state economicactors; what is importantfor this analysisis that the purpose of demand-makingis economic.According to thisthesis, indicators of economicdevelopment, such as per capita grossdomestic product (GDP), should predictthe creationof a sciencepolicy organization. Securityconditions are argued to promptthe creation of science policy bureaucraciesthrough the actions of militaryconsumers of science. In the modernera ofwarfare, scientific prowess has been clearlylinked to technological and hence to militarysuccess. Thus statesperceiving threats to theirpower and/orsecurity will be pushed to findnew and more effectivetechnologies to meet those threats.Militaries in these states will demand that the state organize and support the scientificestablishment for reasons of national defense. The timingof sciencebureaucracy creation in Britain(during World War I) and in the United States (immediatelyfollowing World War II) has led a numberof scholarsto drawcausal connectionsbetween security concerns and sciencepolicy. Sanford Lakoff, Jean-Jacques Salomon, and HarveySapolsky all point to these wars as well as to another perceived securitythreat-the launchingof Sputnik-as the catalystsfor government interest in harnessing scienceto achievenational objectives in the UnitedStates and Europe. Having organized science to meet securitythreats duringwartime with apparent

6. David Dickson. TheNew Politics of Science (New York: Pantheon,1984), pp. 25ff. 570 InternationalOrganization success, these wartimeinstitutions were then redeployedby states to meet peacetimeobjectives.7 Robert Gilpinmakes a more detailed and broadersecurity argument based on his investigationsof French science.8He argues that France's creationof science policy organizationswas the direct result of a perceived threat to French influenceand independence froma preponderanceof U.S. power immediatelyfollowing World War II. At one level,this threat was understood militarilyand led the Frenchto use theirscience communityto upgradetheir defenseestablishment, notably to establisha separatenuclear strike force. But threatsto influenceand securityin the Frenchview were not limitedto the militarysphere. The French were also concerned about loss of economic dominance.U.S. economicstrength following World War II was viewedwith trepidation,and direct U.S. investmentin France was viewed as a formof imperialismby a foreignpower. During that time the French spoke of a "technologygap" that theymust bridge by harnessingFrench science in the service of French industryto protect French economic independence and integrity.9 Securityunderstood in this sweepingway, as any threatto influenceand independence,could operate in so manyarenas thatdeveloping tidy objective indicatorsto test for its presence is probably impossible.10The narrower argumentsabout securitythreats understood in a militarycontext are some- what easier to uncover.If armed conflictor the threatof armed conflictis critical,indicators of perceivedmilitary threat, such as defensespending as a percentageof GNP, should be correlatedwith the creationof science policy organizations.States perceivingmilitary threats should be among the firstto adopt science policy;conversely, relatively secure states should be clustered amongthe late adopters.

Testingthe demand-driven explanations

Each of these explanationsposits a material conditionthat then sparks a demand forthe state to adopt new tasks and to create new bureaucraciesto carryout those tasks. While it would be impracticalto investigatethe actual

7. See the followingchapters in Ina Spiegel-Rosingand Derek de Solla Price, eds., Science, Technologyand Society:A Cross-disciplinaryPerspective (London: Sage, 1977): Jean-Jacques Salomon, "Science Policy Studies and the Developmentof Science Policy,"pp. 43-70; Sanford Lakoff,"Scientists, Technologists, and Political Power," pp. 355-92; and Harvey Sapolsky, "Science, Technology,and MilitaryPolicy," pp. 443-72. 8. Robert Gilpin,France in theAge of theScientific State (Princeton,N.J.: Princeton University Press,1968). 9. Ibid. 10. For example,in the Frenchcase, threatsto influenceand independenceextended to cultural mattersand led France to pursue a numberof foreignpolicy initiatives aimed at preservingand extendingFrench language and culturein otherstates. UNESCO 571 demand-makingprocess over a largenumber of countrieshaving very different politicalsystems, it is quite simpleto checkfor the existenceof conditionssaid to be prerequisiteto those demands. The test reportedhere compiled and analyzed quantitativeindicators of domestic conditionsthat mightprompt creationof a science policybureaucracy in a sample of forty-fourcountries chosen to be globallyrepresentative in termsof both geographyand developmentlevels. As suggestedin the foregoingdiscussion, these were percentageof GDP spenton R&D; proportionof scientistsand engineersin the population; per capita GDP; and percentageof gross national product(GNP) spent on defense. A complete descriptionof the indicatorsused and the method of compilingthem can be foundin the appendix. Testinga global sampleof statesraises issues about comparabilityamong the unitsof analysis,particularly comparability of developingand industrialized states. Cameroon and the United States, for example,are so differenton so many measures that one may question whetherthe units of analysis are sufficientlyalike to make comparisonappropriate and meaningful. In thiscase, comparabilityof the unitsis ensuredby the researchquestions being asked. The hypothesesbeing testedconcern the behaviorof states as a politicaland organizationalform: What promptsstates to adopt new tasksand constructnew apparatuses to carryout those tasks? The hypothesesdo not carrywith them caveats about degrees of stateness,state capacity,or other potentiallylimiting characteristics. Instead theymake argumentsabout the behaviorof states qua states. Cameroon and the United States may be very differentstates, but theyare both states nonetheless.In fact the articlewill suggestthat what is going on in this case is a redefinitionof the state as a politicaland organizationalform; that is, a redefinitionof what is necessaryand appropriatebehavior for a state. Figures1-4 showthe distributionof values foreach of the indicatorsof state conditions at the time science policy bureaucracies were created in the countriesstudied. A quick look at figuresreveals that none of the patterns corresponds to the expected patterns described above. If any of these conditionswere both necessary and sufficient,there would be a largecluster of adoptionson the histogramat thatnecessary and sufficientvalue. Instead,the adoptionsappear to occur at a verywide range of values for all fourof the variables. No singlevalue of any variable appears likelyas a necessaryand sufficientcondition for adoption. In fact,countries adopted these science bureaucraciesat wildlydifferent levels of each of these domesticconditions. Some elaborationfrom the raw data willmake the extremelywide rangeof variation in values even clearer: (1) Countriescreated these bureaucracieswhen theyhad as few as nine scientistsemployed in R&D (e.g., Congo) or as manyas halfa million(e.g., the United Statesand the SovietUnion). (2) R&D spendingas a percentageof GDP rangedfrom 0.01 percentat the timeof adoption(Bangladesh) to 1.5 percent(France). (3) Per capita GDP in constantU.S. dollarsranged from a low of $118/year 572 InternationalOrganization

24_

22-X

18 2 - 16 4 4-

0.0-0.3 0.3-0.6 0.6-0.9 0.9-1.2 1.2-1.5 1.5-1.8 1.8-2.1 2.1-2.4 2.4+

Percentageof grossdomestic product spent on R&D FIGURE 1. Researchand development(R&D) spendingat thetime of science policyadoption

(Pakistan) to a highof more than $9,000/year(Denmark) at the time these bureaucracieswere created. (4) Defense spendingas a percentageof GNP ranged from0.7 percent (Mexico, Sri Lanka) to more than 10 percentat the timeof adoption (France, Iraq, Jordan,and the SovietUnion). The rangeof variationon the defensevariable is more than a factorof ten; the rangeof variation on all of the othervariables is a factorof one hundredor more. Ranges of variation this large do not readily suggest any causal connectionbetween sufficientstate conditionsand the adoption of science bureaucracies. Similarly,Figures 1-4 provide little support for the necessarycondition hypothesis,that is, that there is some minimumthreshold value of these variablesthat triggers demand for the bureaucracy.If such a value existed,we shouldsee veryfew (or no) adoptionsat the low end of thevalue rangefor one or moreof thesevariables; all values would be spread across the upper end of the rangeat or above thenecessary condition level. Again, the far-flungdistribution of values revealed in these figuresand elaboratedin the textabove does not readilysupport this proposition. Rather than clusteringat the upper end of the value ranges,there seems to be a concentrationof values at the low end, particularlyon the science variables. This is clearlynot a bunchingthat would supportthe existenceof a necessary UNESCO 573

z~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

8 - II 4-_18-

scec12 poic adoptio

wereqal unIkeycniae o cec ueurc.

11 The Cog crae t osi ainld aRceceSiniiu n16 hni reore hain olnInesinit nae nR Djb n hnsedn nR Dwsol 0.1 pecn ofGPI esrdi ..dlas D e aiawsol 23ta er n trigermiitr spndnemad aconeIo fr aSciencetsan ny20policyebrs ecn eauracy,tha fG P.Cmro threshtoldmsnrae t fieNtoaeso de la Rehrh- ScEniiu etTcnqei195we itrpre emlygoly ihy asGUtoscetit havin reerhjbI explanatoryann pnigol poer.nesi 1 eceto t D nrsac.Prcpt D was 33 fo veoprytittlef hpplainahieot N ndfne siencfalct Ihtya,adtecutysetony23pretoadlargeonubrosml,po,tcnlgclyushiiaedad militarilyiunthreatneditountrsinescraedteeburcigneauhcraeies incopthed19byan lrendmeFigureso1i4.sGatemalaof forhexamle reatediscae ConstejoiNacionars de bneastrgongnesaCientoficasny necnscasy-uinot1966fwhenit-reordtedn haviumng. onl threvngat,setsomemnlym1.07tprcenthofd GNPe ondhefese. Caramleroo ansppsd thCong wrgeredequlyulklmand for iaea plc science bureaucracy."httrehl ut es o

11. ThetCog cargeatumed its Coseill Natonal deltecherogchl Scintifhiqueina1963whndi reportehavingonlyrenined scintrists engaged inhR&D jbsreucandihenspeningone R&Dws only 190s.1pecet oftDPs MeasuredaicontUs. dollrs GDP plsericapitawasa ponly$25tat yhearand militar spningurs14accuntedmoroly, 2.04 perntofmp.lCmeo created its Offieo Nationald dnestiaRchrche SCientifiqueet Technique in 1965 when it reportedemploing onlyegt forenscientistsresarc inepoe jobs,an spendin 0.16 only ofitpercent rsac.PrcptGDP on D whrats,$3o htyar nspent 2.37honr only percent of GNP onit defense.Cmro n h og 574 InternationalOrganization

24 22

t16-

) I _4-

av12- E0- z 8-

t 6 8-- | 10|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~------

0-750 750-1,500 1,500-3,0003,0(X4,500 4,500-6,000 6,000-7,5$ 7,50-9,O00 9.000+

GDP per capita in constant$US (1980)

FIGURE 3. Grossdomestic product (GDP) per capitaat thetime of science policyadoption

At the same time,the histogramsdo showthat some countriescreate science bureaucraciesat reasonablyhigh levels of all the indicatorvariables. Signifi- cantly,the first instances of thisscience bureaucracy creation occur amongthis group,suggesting that demand-driven explanations may fit some of the earliest adopters of science policy. Britain, the firstadopter, clearly created its Departmentof Science and IndustrialResearch in 1915 forsecurity reasons to counter German advances in chemicals and machinerythat were directly supportingthe Germanwar effort.12The establishmentof theNational Science Foundation in the United States in 1950 was explicitlyrelated to concerns about militaryand industrialcompetitiveness and was stronglyinfluenced by thecreation of the atombomb.13 French science policymaking, as chronicledby

12. Alter,The Reluctant Patron. See also Roy McLeod and E. Kay Andrews,"The Originsof the D.S.I.R.: Reflectionson Ideas and Men, 1915-1916,"Public Administration, vol. 48, no. 1, 1970,pp. 23-48; and Ian Varcoe, "Scientists,Government, and Organized Research in Great Britain 1914-1916,"pp. 192-216. The United Kingdomis not includedin the quantitativeanalysis above because sciencedata forthat country for 1915 are unavailable. 13. Dickson,The New Politics of Science.See also J.Merton England, A Patronfor Pure Science: The National Science Foundation's FormativeYears, 1945-1957 (Washington,D.C.: National Science Foundation,1982); N. Dupree, Sciencein theFederal Government; Bruce Smith,American Science Policy Since WorldWar II (Washington,D.C.: BrookingsInstitution, 1990); and U.S. Congress,House Committeeon Science and Technology,Task Force on Science Policy,A History of SciencePolicy in the UnitedStates, 1940-1985, Science PolicyStudy Background Report, no. 1, 99thCongress, 2d sess., 1986,serial R. UNESCO 575

24 22

18-

m 14

0 , 12- 10-

6 4_

0.0-1.0 1.0-2.0 2.0-3.0 3.0-4.0 4.0-5.0 5.0-6.0 6.0-7.0 7.0-8.0 8.0+

PercentageGNP spenton defense

FIGURE 4. Defensespending as a percentageof gross nationalproduct (GNP) at thetime of science policy adoption

Gilpin, seems also to have been promptedby securityand competitiveness concerns,albeit of a more general nature since the French were concerned about a general loss of influencein the world,especially vis-a-vis the United States.14 But how do we explainthe creationof science policyorganizations in more than a hundred other states covering the extremes of science capacity, developmentlevels, and militarysituations in the subsequenttwenty years? Countriesas dissimilaras Bulgaria,Czechoslovakia, El Salvador,the Federal Republic of Germany,Indonesia, Italy, Lebanon, Mali, Pakistan,and Sweden all createdtheir first science policy bureaucracy during the peak adoptionyear of 1962.15 It willbe arguedbelow thatthese twophenomena, the apparentresponsive- ness to stateconditions in a fewearly-adopting states followed by a patternof adoptionunrelated to stateconditions, can be reconciledin the followingway: science policybureaucracies appeared as an innovationin the international systemin responseto clear domesticdemands in a fewprominent developed countries.The innovationwas thenpicked up and popularizedby an international organization,UNESCO, for reasons of its own and spread by that

14. Gilpin,France in theAge ofthe Scientific State. 15. "Peak adoptionyear" in thiscase meansthe singleyear in whichthe largestnumber of states createdthese science policy bureaucracies. 576 InternationalOrganization

organizationto other states in which the conditionsthat ordinarilywould promptsuch demandsdid not exist.

A supply-drivenexplanation

Since demand-sideexplanations for the creation of science policy organizations appear to be on weak groundfor most states, the nextlogical question is, what if these bureaucraticinnovations are not demanded inside the state but supplied fromoutside? In fact,most (roughly70 percent) of these science policyorganizations were created between 1955 and 1975.16Beginning in the early 1950s two internationalorganizations, UNESCO and the Organization for Economic Cooperation and Development (OECD), began activelyto promotethis science policyinnovation among theirmember states.17 In this section I describe the promotionactivities of one of these organizations, UNESCO, and present evidence that its activitieswere the impetus for widespreadadoption of science policy.Such an explanationallows us to make sense of the data presentedearlier since it is consistentwith both (1) the low-endclustering on the histograms,indicating that many countries adopted these bureaucraciesat verylow levels of the indicatorvariables, and (2) the timingof adoption across states,specifically the rapid increase immediately afterthe international organizations begin to promotethe science policy innovation. Specifically,I will argue that from its inception UNESCO as an organization has had to address two constituencies-thestates comprising its membership and the professionalexperts in its substantiveareas of concern.As will be describedbelow, the relationshipbetween these twoconstituencies within the organizationhas changedover time. These shiftsin turnhave been responsible forchanges in UNESCO's programs,particularly the rise of science policy as an area of concern.18

Origins of UNESCO's interestin science As originallyconceived, UNESCO was to be named the United Nations Educational and CulturalOrganization. Science was understoodto be partof culture.The notionthat science was qualitativelydifferent from other aspects

16. By extendingthe period by fiveyears to include the years 1976-80, the percentageof adoptingstates rises to 84.4. 17. For more on the science policypromotion activities of the OECD see Martha Finnemore, "Science, the State,and InternationalSociety," Ph.D. diss.,Stanford University, 1991, chap. 3. 18. The followingaccount of UNESCO activitiesis based on researchdone at the libraryand archivesof UNESCO's Paris headquarters.While these contain a wide varietyof documents authoredby nationalgovernments, it mustbe acknowledgedthat carrying out the researchat the internationalorganization's headquarters does run the riskof bias in favorof an international organization-drivenexplanation at the expense of a nationalone. A researchdesign in whichthe science policyarchives were consultedat a varietyof national capitals in countriesof different regions and developmentlevels and facing differentsecurity situations would be superior. However,such a designwas notfeasible in thiscase. UNESCO 577 of culture and thereforemerited special recognitionin the organization's missionand titlehad to be foughtfor by scientistsand science promotersin governmentduring the preliminarymeetings and negotiationsthat took place in 1942-45. The fact that they succeeded in gettinga piece of the new organizationto servetheir own interestswas due in part to the organizational strengthand coherenceof the internationalscience communityand in partto their abilityto influenceworld affairs,as had been demonstratedat Hiro- shima.19The combinationof these two convinced the president of the conferencecharged with establishing the new organizationto supportaddition of theword "Scientific"to the titlewith the followingremarks: "In these days, whenwe are all wondering,perhaps apprehensively, what the scientistswill do to us next,it is importantthat they should be linkedclosely with the humanities and shouldfeel that they have a responsibilityto mankindfor the result of their labours."20

UNESCO's early science programs and organizational structure Givingscience a titularrole in the new interstateorganization was a way of recognizingthe importanceof science to the state but did not, in those early years,entail state directionor controlof scientificresearch. UNESCO's early science programswere designed to serve science and scientistsrather than states.They aimed to increasethe world sum of scientificknowledge and access to thatknowledge without regard to nationalboundaries. The notionimplicit in sciencepolicy, that science is a nationalresource to be developedby individual states, is very differentfrom UNESCO's original notions of science as a transnationalenterprise. Indeed, the principalrationale for internationalizing science and forbringing it underthe auspices of the United Nationsin thefirst place was to free it fromthe meddlingof self-interested(and presumably self-aggrandizing)states. Apart fromthe dangersof statesexploiting scientific discoveries for military gain,state interferencein science had longbeen understoodto stiflescientific progress.Science was believed to proceed most efficientlyand productively whenleft to scientists.Certainly this was the attitudeof the League of Nations' InternationalInstitute for Intellectual Cooperation in the interwarperiod, and itcontinued to be theattitude of most scientists' professional organizations and of individualscientists active in internationalaffairs.21

19. Details on the lobbyingefforts of scientistsfor special recognitionin the embryonic UNESCO can be foundin JamesSewell, UNESCO and WorldPolitics (Princeton, N.J.: Princeton UniversityPress, 1975). 20. Conferencefor the Establishmentof UNESCO, "Opening Addressby the Presidentof the Conference,the Rt. Honorable Ellen Wilkinson,MP," Conferencefor the Establishmentof UNESCO, London,1-16 November1945 (Paris: UNESCO, 1946), p. 24. 21. JulianHuxley, the firstexecutive director of UNESCO, and Joseph Needham, the first directorof UNESCO's Natural Sciences Department,were instrumentalin the foundingof that organizationand wrote extensivelyon theirviews of science as a transnationalactivity. See, for example,Julian Huxley, Unesco: Its Purposeand Its Philosophy(Washington, D.C.: Public Affairs 578 InternationalOrganization

Earlystatements of UNESCO's purposewith regard to sciencereflect this view.At thefirst session of theGeneral Conference in November1946, the Subcommissionof theGeneral Conference for Natural Sciences summed up theaims of UNESCO's Departmentof Natural Sciences in thefield of science as follows: 1. to establisha world-widenetwork of field science cooperation offices; 2. to supportthe scientific unions, add totheir number and assist them in theirwork;22 3. to organizeand operate an internationalclearing house for scientific information; 4. to supportthe work of the United Nations and its specialized services; 5. to informthe general public in all countriesof the international implica- tionsof scientificdiscoveries; 6. to createnew forms of international scientific cooperation (international observatoriesand laboratories, etc.).23 Sciencepolicy and promotingscience capabilities of memberstates were not evenmentioned. EarlyUNESCO scienceprograms conformed very much to theseaims.24 Sciencecooperation field offices were set up inCairo, Nanking, New Delhi, and Rio de Janeiroto facilitate the movement of scientists and information around theglobe.25 Financial support was extended to nongovernmental organizations, particularlythe InternationalCouncil of ScientificUnions, as a meansof promotingcooperation among scientists, and more of these unionswere established.26Research institutes were established, such as theInstitute of the HyleanAmazon, the Instituteof the Arid Zone, and an International ComputationCenter, to bringscientists from different countries together to

Press, 1947). Sir HenryDale, who persuaded the establishingconference to accept science as a distinctpart of the embryonicUNESCO and who had been part of the League of Nations' InternationalInstitute for Intellectual Cooperation, held similarviews. See Sewell, UNESCO and WorldPolitics. 22. "Scientificunions" are scientists'professional organizations, such as the International AstronomicalUnion and the InternationalGeodesy and Geophysical Union. Their umbrella organizationis the InternationalCouncil of ScientificUnions. 23. This summaryis fromMarcel Florkin,"Ten Years of Science at UNESCO," Impact of Scienceon Society7 (September1956), pp. 123-24. 24. See, forexample, "Activities of Unesco in the NaturalSciences During 1948," 14 February 1949,Natural Sciences (NS)/67, UNESCO Archives,Paris. 25. The Rio de Janeirooffice was movedto Montevideoin 1949,and in 1951 the Nankingoffice was relocated to Djakarta in the wake of the Chinese revolution.In creatingthese fieldoffices JosephNeedham (head of UNESCO's Natural Sciences Department)was realizingthe Interna- tionalScience CooperationService he had proposed duringthe war. See JosephNeedham, "An InternationalScience Cooperation Service,"Nature 154 (25 November 1944), pp. 657-59. For originalplans forthe fieldoffices, see "UNESCO Science CooperationOffices," 12 June1947 Nat Sci/28/1947,UNESCO Archives,Paris. For a briefhistory of earlyfield office program see Florkin, "Ten Years of Science at UNESCO." 26. Specifically,UNESCO helped foundthe Union of InternationalEngineering Associations and the Council of InternationalOrganizations of Medical Sciences. See Florkin,"Ten Years of Science at UNESCO." UNESCO 579 workon problemsof mutualinterest. All of these activitiesprovided services directlyto scientistsrather than states. The earlyorganizational structure of UNESCO also reflectedthis understanding of science and culture as transnationaland often nongovernmental activities.While the General Conferencewas composed of equal member states,UNESCO's executiveboard was to be composedof eighteen individuals, elected by General Conference delegates for their distinction in the organization'ssubstantive fields. Board memberswere to serveon behalfof the conferenceas a whole and not as representativesof theirrespective governments.27

Change The principle of nongovernmentalismenshrined in the compositionof UNESCO's executiveboard soon came under attack.While board members were to be elected as individuals,the originalUNESCO constitutionprovided that no state could supplymore than one of its nationals to the board. In practice this focused pressure on board memberswhen their governments wanted to pursue particularpolicies throughthe organization.(James Sewell citesone highAmerican official's description of the U.S. governmentbringing itsboard memberback to Washingtonto "briefthe hell out of [him]to tryto get [him] to see thingsthe State Departmentway."28) Accounts of UNESCO conferencedelegates from the period describethe decline of participationby scientists,scholars, educators, and writersand the increased presence of "governmenttechnicians" who viewed themselvesas governmentspokesper- sons.29 In 1954 UNESCO membersvoted to amend the constitutionand reorganize the executiveboard into a body of twenty-twogovernmental representatives. The shifttoward a governmentalizedsecretariat was justifiedon several grounds.The reason cited most oftenwas a financialone; since states were footingthe bill for UNESCO's operations,the organization should serve states. In the wordsof one officialwho leftthe organizationabout thattime, the shift was "the price for financialsupport." A cynical French delegate, Roger Seydoux,predicted that withthis change "the Finance Ministries... would become mastersof UNESCO and itsprogramme."30 However,participants also understoodthe shiftto be in keepingwith a larger shiftin the internationalclimate frompostwar Kantian transnationalismto

27. Note that this was a deliberate shiftfrom the way in which the Conferenceof Allied Ministersof Education's executivebureau and UNESCO's preparatorycommission executive committeewere constituted.Both of thesewere composedof nationalrepresentatives. See Sewell, UNESCO and WorldPolitics. 28. As quoted in Sewell, UNESCO and WorldPolitics, p. 169. 29. Ibid., pp. 168-69. 30. Both quotationsare fromibid., p. 169. 580 InternationalOrganization cold war Hobbesian nationalism.The originalnongovernmental structure was a reflectionof 1940sbeliefs (or at least hopes) thatideas were a unifyingforce in the world. Education, science, and culturecould weave a web thatwould draw a dividedworld of nation-statestogether. If "wars begin in the mindsof men," then the UNESCO solutionfollows logically, namely, that "it is in the mindsof men that the defences of peace mustbe constructed."'31Harry Truman voiced similarbeliefs in his appeal to the 1945 United Nations founding conferencein San Franciscoto "set up an effectiveagency for consistent and thoroughinterchange of thoughtand ideas, forthere lies the road to a better and moretolerant understanding among nations and amongpeoples."32 By the mid-1950sthis view had been eclipsed.At best,ideas were irrelevant to the intensepower struggle raging in theworld; at worst,ideas wereviewed as divisiveand dangerous,in whichcase statescould not affordto leave themto a collection of unaccountable individuals. As realpolitik replaced Kantian liberalism,states were reintegratedas majorplayers in determiningUNESCO policies. Not surprisingly,the UnitedStates, then in thegrip of anticommunistfervor, was thestandard-bearer for this new attitude. UNESCO came to be viewedas a politicalinstrument in the cold war; itspurpose was to be a "Marshall Plan for ideas," whichwould block the expansionof "intolerant"communism. Failure to convertUNESCO wholeheartedlyto its own foreignpolicy agenda led U.S. officialsto erect organizationalbarriers to containand controlits influence. Chief amongthese was the InternationalOrganizations Employment Loyalty Board, establishedin early1953 to overseethe employment of U.S. representa- tivesto all United Nations agencies. Such screeningensured that American executiveboard memberswould be sympatheticto theirgovernment's views. This screeningclearly flew in theface of the nongovernmentalprinciples under whichUNESCO was formed. While the Americans succeeded in replacing intellectualsof dubious leaningswith patriots and loyalists,they failed in theirultimate goal of making the organizationa tool of U.S. foreignpolicy. Just as theirgovernmentalist reformswere enacted in 1954,the SovietUnion and severalof its Easternbloc allies finallyagreed to join UNESCO. Shortlythereafter came the influxof newlyindependent developing states with an agenda of theirown, effectively blockingcontrol by the UnitedStates or byany other great power. The shiftfrom nongovernmentalism to governmentalismrepresented a shift in the balance of power among UNESCO's two constituencies.As the internationalclimate changed and optimismabout the utilityof transnational

31. Constitutionof the United Nations Educational, Scientific,and CulturalOrganization as reprintedin William Preston,Edward S. Herman, and HerbertI. Schiller,Hope and Folly. The UnitedStates and UNESCO, 1945-1985(Minneapolis: University of MinnesotaPress, 1989), p. 315. The originalphrase, "war beginsin the mindsof men," was coined by ClementAtlee. See ibid., p.33. 32. Trumanis quoted in Preston,Herman, and Schiller,Hope and Folly,p. 33. UNESCO 581 activitiesdeclined, the champions of those activities-scientists,scholars, artists,and educators-lost ground.Their rhetoric about building bridges to all humankindbecame less and less appealingto UNESCO's memberstates, each of whom increasinglyviewed the world as hostile and much of the rest of humanitywith suspicion.Thus, duringthe cold war states reassertedthem- selvesas UNESCO's chiefconstituents, and UNESCO officialsreshaped their programsto accommodatethem.

Effectsof organizational change on science programs UNESCO's science activitiessoon reflectedthe shiftin worldviewand the organization'srediscovered constituency of states.While internationalscien- tificprojects begun previouslyhad continued,UNESCO also became con- cernedwith promoting science at the nationallevel. UNESCO began to focus its attentionon helping states to organize, direct, and expand their own domesticscience establishments. Its preferredmethod of doingthis was to help statesto create a new stateagency to take care of thesetasks. As a firststep in thisdirection, UNESCO conducteda surveyof the national researchcouncils of memberstates in 1953.33The statedpurpose was twofold. The firstpurpose was to collectreference material for anyone asking advice on how to set up a research council. The second purpose was to provide backgroundmaterial for the establishmentwithin UNESCO of an advisory committee(the InternationalAdvisory Committee on ScientificResearch) whose role would be to provide assistance to states seeking science policy advice.34 At thisstage, UNESCO stillviewed its role in science policypromotion as a relativelypassive one; it waited for states to ask for advice and assistance. Followingthe 1954 reforms,the organization became moreactivist and science policyactivities grew by leaps and bounds. In 1955, partlyas an outgrowthof the 1953 survey,UNESCO convened a meeting of directorsof national researchcenters in Milan at whichthirty countries were represented.35 The first agenda itemfor this meeting was discussionof "the role of nationalplans for the developmentof scientificresearch."36 At the conferenceUNESCO staff membersoutlined the virtuesof nationallydirected science activity,discussed differentmodels forsuch direction,and emphasizedthe role UNESCO could

33. The originalsurvey analysis is containedin UNESCO Archives,document NS/107. Survey resultswere also publishedas "Reportsand Documents:Survey of NationalResearch Councilsfor Pure and Applied Science in the Member States of UNESCO," Impact of Scienceon Society4 (Winter1953), pp. 231-55. 34. "Reportsand Documents,"p. 231. See also PierreAuger, "UNESCO and the Development of Research in the Field of NaturalSciences," UNESCO Chronicle1 (July1955), p. 5. 35. See UNESCO/NS/124, UNESCO Archives,Paris, forthe finalreport of thismeeting. See also Auger,"UNESCO and the Developmentof Research in the Field of NaturalSciences," p. 5. 36. UNESCO Chronicle1 (July1955), p. 26. 582 InternationalOrganization play in providingassistance to governmentswishing to pursue the outlined objectives.37 By the late 1950s UNESCO had begun activelyto assistcountries in setting up science policyorganizations: in 1957 when the Belgian governmentasked forhelp in settingup its National Science PolicyCouncil, UNESCO sent the chiefof its Science PolicyDivision to directthose activities.38UNESCO also providedassistance to theLebanese governmentin creatinga NationalCouncil forScientific Research.39 These activitiesreceived a boost in 1960when Pierre Auger, then acting as a special consultantto the United Nations as a whole,40issued a report recommendingthat national scientific policy be one of the "foremostpreoccu- pationsof governments": States shouldmake it theirbusiness to ensure[that] interaction between the encouragementof scientificresearch, on the one hand,and economic and social progress,on the other,operates smoothly to the advantageof both.It is, at the same time,the dutyof organizationsin the UnitedNations familyto assistStates in thismatter.4' Auger's detailed report,requested and approved by the larger United Nations,validated the science policyactivities in whichUNESCO had been engagedover the last fiveyears and provideda basis forformalizing what had been an ad hoc collectionof activities.42Beginning in 1960,UNESCO General Conferenceresolutions included instructions that the Director-General"col- lect, analyse and disseminateinformation concerning the organizationof scientificresearch in MemberStates and the policiesof MemberStates in this respect."43By 1963 the General Conferenceresolutions were more explicit. The Director-Generalwas authorizedto assistmember states "in theestablish- ment or improvementof science policyplanning and researchorganization,

37. Several of the presentedpapers were later publishedin the journal Impactof Science on Society.Most relevantis Werner Moller's "National Research Councils and Science Policy," Impact of Science on Society6 (September 1955), pp. 155-68. Moller was a memberof the Departmentof NaturalSciences at UNESCO. 38. The Belgian National Science PolicyCouncil was establishedin 1959. UNESCO Archives, NS/(Research OrganizationUnit)ROU/100, UNESCO Archives,Paris. 39. See NS/ROU/Lebanon (LEB) 1-23; and NS/ROU/100, UNESCO Archives,Paris. 40. Auger had been the second head of the Natural Sciences Department after Joseph Needham and had recentlyretired from the Secretariat. 41. PierreAuger, Current Trends in ScientificResearch (Paris: UNESCO, 1961), p. 220. 42. This reportis routinelycited as the basis forUNESCO's science policyactivities. See, for example, Y. de Hemptinne,"UNESCO's Role in the Organizationof ScientificResearch," UNESCO Chronicle9 (July1963), p. 245; and Alexi Matveyevopening speech to the meetingof the Coordinatorsof Science PolicyStudies, Karlovy Vary, Czechoslovakia, June 1966. Matveyevwas UNESCO's AssistantDirector-General for Science. His speech is reprintedin Principlesand Problemsof NationalScience Policies, Science PolicyStudies and Documents Series,vol. 5 (Paris: UNESCO, 1967),p. 12. Also see, "Surveyof UNESCO's activitiesand achievementswith regard to sciencepolicy," NS/ROU/100, UNESCO archives,Paris, p. 3. 43. UNESCO, General Conference,11th sess., 1960, Resolutions,2.1131 (Paris: UNESCO, 1960). UNESCO 583 throughsending advisorymissions, conducting scientific and technological potentialsurveys, with particular regard to human resourcesand budgets,or organizingtraining seminars and, to thisend, to participatein theiractivities in the field."44From thispoint on the goal of spreadingand improvingscience policy organizationswas firmlyentrenched in UNESCO's officialscience program.

The new norm What is interestingabout UNESCO's programin termsof the questions raisedin thisarticle is thatthe languageused is prescriptivenot evaluative, and in this sense it is normative.UNESCO officialssimply declared science policymakingto be necessaryand good; therewas no seriousattempt to prove thatthat was so: "Statesshould make it theirbusiness" to coordinateand direct science,45or, "The developmentof science policyshould be the responsibility of an organizationat the highestlevel of governmentin the country."46Also "the Science PolicyProgramme of UNESCO is formulatedon the basis of the principlethat the planningof sciencepolicy is indispensable"for the coordination and promotionof scientificresearch.47 These assertionsare not coupled withany evidence that such bureaucraticentities enhance sciencecapabilities. This is surprisinggiven that until only a fewyears earlier, conventional wisdom had held exactlythe opposite-that governmentinvolvement stifled scientific creativity. In addition the language is universal;it promotesthese bureaucraciesas good for all states and at all levels of scientificcapability. This ignores an obvious potential strategyfor many countries,particularly less developed countries(LDCs)-free-riding. Science, with its imperativesto disseminate resultswidely and immediately,has manyproperties of a collectivegood. The economicadvantages of being a followerrather than a leader in technological innovationhave been widelydiscussed in othercontexts. Thus, froma functionalstandpoint, it is not obviouswhy all statessuddenly "needed" a sciencepolicy bureaucracy at thatparticular point in time.In fact, it appears thatthese eventswere not related to functionalneed in any strict sense. Rather,they cOnstitute a redefinitionof the normsand expectationsof state roles withregard to science. Initiallyscientists sought to harness state resourcesto furthertheir own scientificprojects by claiminga piece of an interstateorganization. To do so theyhad to proclaimscience an appropriate

44. UNESCO, General Conference,13th sess., 1964,Resolutions, 2.112(d) (Paris: UNESCO, 1964), p. 32. 45. PierreAuger, Current Trends in ScientificResearch, p. 220, emphasisadded. 46. UNESCO, Principlesand Problemsof SciencePolicy, Science PolicyStudies and Documents veries,vol. 5 (Paris: UNESCO, 1966),p. 87. 47. "The Proposed Science Policy Programmeof UNESCO for 1967-68," NS/ROU/117, UNESCO Archives,Paris, p. 1, emphasisadded. 584 InternationalOrganization concernof governments. The debateover the S in UNESCO and Wilkinson's commentsin prompting itreveal the scientists' success in this. However, when scientistsand the other epistemic communities lost control of UNESCO to the memberstates, the situation did not simply revert to thestatus quo ante.48The normthat science was nowan appropriateconcern of statesremained firmly entrenched,but the relationship between science and stateswas redefinedto reflectthe newworld climate and UNESCO's new dominantconstituency. Ratherthan states collectively within an internationalorganization promoting and directingscience as a transnationalenterprise, Natural Sciences Depart- mentofficials in UNESCO now arguedthat states individually should take responsibilityfor promoting and directing science within their own borders. By proclaimingscience policymaking tobe an appropriateand necessary function ofstates qua statesand by offering themselves as a sourceof knowledge about thisnew function, UNESCO scienceofficials successfully redefined their role in a waythat was neither irrelevant nor dangerous to their new clients. The interestin andsuccess of UNESCO's effortsis notunrelated to thefact that duringthis period large numbersof new stateswere beingcreated, virtuallyall ofwhich were LDCs. Whileit was developedcountries such as the UnitedStates and the UnitedKingdom that initially had pushedfor the reorganizationof UNESCO's executiveboard to favorstates for cold war- relatedreasons in the 1950s,UNESCO's reorientationtoward states fit well withthe climate of national self-determination in the 1960s. Among the large and growingmembership of newlyindependent LDCs, thenotion that states shouldand could promote and direct science, with all itseconomic and military applications,was popular. At the1963 United Nations Conference on Science and Technologyfor the Benefit of Less DevelopedAreas, the agenda item on organizingand planningscientific and technologicalpolicies was reportedto yield"one of the mostfruitful discussions in the wholeconference."49 In additionto laudingscience policy as an activity,conference delegates stressed theimportance of building up inthe developing countries indigenous programs of research.They argued that "just as no countrycould develop economically on importedgoods, so nonecould developintellectually on importedideas alone."50For thesestates, science as a transnationalactivity spelled continued dependence.Science had to be a nationalpursuit to be normativelycompatible

48. The term"epistemic community" refers to "a communityof expertssharing a beliefin a commonset of cause-and-effect relationships as well as commonvalues to which policies governing theserelationships will be applied."See PeterHaas, "Do RegimesMatter? Epistemic Communi- ties and MediterraneanPollution Control," Intemational Organization 43 (Summer1989), pp. 377-403;the quotationis drawnfrom footnote 20, p. 384. Haas providesa moreextensive discussionof this definition on page3 ofPeter Haas, "Introduction:Epistemic Communities and InternationalPolicy Coordination," International Organization 46 (Winter 1992), pp. 1-35. 49. UnitedNations, Secretariat, Economic and Social Council,36th sess.,Report to the Secretary-Generalon theResults of theUnited Nations Conference on Scienceand Technologyfor the Benefitof Less DevelopedAreas, 1963, E/3772, annexes, agenda item15. 50. Ibid.,section 181, p. 24. UNESCO 585 withthe nationalism of the newly created state and provide it with the means of resistingencroachments from outside.51 This nationalor statistconception is distinctfrom the understandingof sciencethat prevailed in UNESCO's earlyyears. States were now understood to be theprimary purveyors of development and progress.Thus it was states, notscientists, who could best bring the fruits of science and technology to their citizens.Scientific knowledge could be translatedinto increased wealth and securityor improvedstandards of living only if it was harnessedby states and integratedinto their larger economic and militaryestablishments. Scientific capacityor "scientificpotential" was viewedas a nationalresource, not as a branchof some larger collectively held international resource. The redefinitionof scienceas a state concerndid not necessarilyhave negativeimplications for the independenceor productivityof the scientific community.In factthe establishmentof nationalscience bureaucracies very oftenhad theeffect of giving scientists more power at thenational level and accessto moreresources. The argumenthere is notthat states would now run and controlscientists and science but that sciencewould be organized nationally,for national benefit, rather than internationally, foroverall benefit.

Spreading science policy Teachingstates to fulfilltheir new role quickly became UNESCO's principal sciencemission; by 1960 a specialResearch Organization Unit of the Natural SciencesDepartment had been established to dealwith these tasks.52 Efforts to establishand expandscience policy organizations were undertaken on several fronts.First, following instructions from the executive board, UNESCO field officesorganized a seriesof meetings to promote the idea ofscience policy and disseminateinformation about establishing the necessary policy machinery.53 At thetime these meetings took place (1959-60) in Latin America, the Middle East, and SoutheastAsia veryfew of the attendingcountries had science bureaucracies.Only three of the elevencountries at the Latin American regionalmeeting did; onlyone of theMiddle Eastern countries at theCairo meetinghad sucha body.54In all threecases, these regional conferences were

51. This interpretationis compatiblewith the conclusionsof Stephen D. Krasnerin Structural Conflict:The ThirdWorld Against Global Liberalism(Berkeley: Universityof CaliforniaPress, 1985). 52. The Research OrganizationUnit was subsequentlyrenamed the Science PolicyDivision. For more on the earlyactivities of the Research OrganizationUnit, see de Hemptinne,"UNESCO's Role in the Organizationof ScientificResearch," pp. 244-48. 53. "UNESCO Science Cooperation Offices,"UNESCO Chronicle7 (December 1961), pp. 433-435. 54. See Guy B. Grestford,"The Developmentof Science in South-eastAsia," Nature186 (11 June 1960), pp. 859-60; and B. A. Houssay, "Organization of ScientificResearch in Latin America,"Nature 188 (31 December 1960), pp. 1157-58. The three Latin American countries havingscience bureaucracieswere Argentina,Brazil, and Mexico. Resolutionsand declarations fromthis seminarare found in NS/ROU/36, UNESCO Archives,Paris. On the Middle East 586 InternationalOrganization

onlythe first of what became a seriesof meetingson sciencepolicy, for once all statesin the regionhad created the policymachinery, talks continued on how thismachinery could be refinedand improved.55 In addition, UNESCO undertooka series of studies on science policy issues.56Many of these worksexamined the science policyestablishments of individualstates and were designed to provide ideas and models to others seekingto establishand improvescience policymakingin theirown countries. Others examinedgeneral issues of science policymaking.Studies oftenwere coupled withmeetings of governmentscience officialsfrom member states.57 Governmentalparticipation in producingthe recommendationsof the studies ensuredthat these recommendations were reachingthe desiredaudience. Perhapsmost interesting, UNESCO officialswould, if requested, come intoa countryand provide on-site consultingservices about how a science policy programmight be established.By early 1966 UNESCO had science policy promotionprograms of this kind either completedor under way in fifteen countriesplus programsto revise existingstate science policybureaucracies alongpreferred UNESCO linesin severalothers.58 The preferredUNESCO form for a science bureaucracyhad two key features.First, the entity making policy about sciencecould notalso do science; that is, it could not also be a research organization.The new science organizationcould not objectivelyassess national researchpriorities when it also had a vested interestin certainlines of researchbeing done in its own laboratories.(The science policybody had to be liberatedfrom any possible conflictsof interest.)Second, the sciencepolicy body was to have access to the highestlevels of government.It was eitherto be a ministerial-levelbody or be located close to the seat of power,for example in the President'soffice. It was thoughtthat it should not be subservientto any otherministry (for example, education or planning),since such an arrangementwould seriouslylimit the organization'sindependence and preventthe nationwidecoordination across all aspectsof sciencethat was required. conference,see UNESCO, Structuraland OperationSchemes of National SciencePolicy, Science Policy Studies and Documents Series, vol. 6 (Paris: UNESCO, 1967); and "Science Planning, Developmentand Co-operationin theCountries of theMiddle East and NorthAfrica," Nature 189 (4 February1961), pp. 362-63. 55. Results of some of the later meetingswere subsequentlypublished as part of the Science PolicyStudies and DocumentsSeries. 56. These were publishedin the (ongoing)book seriesUNESCO, SciencePolicy and Documents Series (UNESCO: Paris). The studies to which I referhere were published duringthe period 1965-90. 57. This was truefor two of thevolumes in particular.See UNESCO, Principlesand Problemsof NationalScience Policies, Science Policyand Documents Series,vol. 5, and UNESCO, Structural and OperationalSchemes of NationalScience Policy, Science Policyand Documents Series,vol. 6 (UNESCO: Paris,1966 and 1967,respectively). 58. Science policy establishmentmissions were complete or under way in Algeria, Congo (Leopoldville), Ethiopia, Guinea, Iran, Iraq, Kenya, Lebanon, Madagascar, Morocco, Senegal, Sudan, Tanzania, Venezuela, and Zambia. Science policymodification or reorganizationprograms were undertakenin Indonesia, Nepal, the Philippines,and the United Arab Republic, among others. UNESCO 587

UNESCO consultants'insistence on these two featuresguided its science policypromotion efforts in memberstates. The organization'scommitment to this particularform of the bureaucracyshows up consistentlyin its on-site "teaching"activities in memberstates. The followingexamples illustrate the ways in which UNESCO succeeded in persuadingstates to set up science bureaucracies and, more specifically,science bureaucracies of the kind UNESCO preferred.

Examples ofUNESCO's promotionof science policy Lebanon One of the firstplaces UNESCO officialsundertook this kind of science policyconsulting was in Lebanon. Because itwas one of thefirst, the Lebanese case became a prototypefor UNESCO consultantson latermissions. A more detailed examinationof the Lebanese case reveals the extentof UNESCO's influenceon the constructionof a science bureaucracythere. UNESCO officialsdid notjust sit on the sidelinesand make suggestions;the head of the UNESCO NaturalSciences Departmentactually drafted the enablinglegislation for the new bureaucracy,while other membersof the secretariatstaff lobbied relevant Lebanese politicians to get it passed. In so doing they squelcheda conflictingLebanese proposalfor the new bureaucracywhich they consideredinappropriate and inadequate. The startingpoint for UNESCO's involvementin Lebanon was the Decem- ber 1960 regionalconference on science planningorganized by the UNESCO Middle Eastern field officein Cairo. At that conferencefield officestaff memberspresented reports on the organizationof sciencein variouscountries of the region.The reporton science and technologyin Lebanon revealedthat Lebanese researchlacked any practicalorientation and that coordinationof researchwas almostnonexistent.59 However, the reportdid not have precisely the intendedeffect. Rather than prompting the Lebanese governmentto begin organizingand coordinatingscientific research, as the Cairo conferencehad recommended,it promptedthe Lebanese ForeignAffairs Ministry to request UNESCO's help in settingup a scientificresearch center, to be part of the Universityof Lebanon, which could carry out (rather than coordinate) scientificresearch in Lebanon in an efficientand effectiveway.60 This requestwas channeledto Yvan de Hemptinne,then scientific secretary to the directorof the Natural Sciences Department at UNESCO. After

59. Unless otherwiseindicated, citations of lettersand memorandabelow are fromUNESCO SecretariatRegistry files, UNESCO Archives,Paris. Where documents were assignedfile numbers, these are so noted. For a descriptionof the reporton Lebanese science and technology,see memorandumby Yvan de Hemptinne,scientific secretary to the directorof the Departmentof NaturalSciences, to Malcolm S. Adiseshiah,UNESCO AssistantDirector-General, May 1961,NS memo50.085. 60. This proposalis describedin a letterfrom Lebanese DirectorGeneral of National Education Fouad Sawaya to UNESCO AssistantDirector-General Adiseshiah, 23 May 1961. 588 InternationalOrganization reviewingthe request,de Hemptinneresponded that a countrylike Lebanon did not need more laboratories or research centers; instead it needed coordinationof its existingresearch efforts.He proposed that ratherthan sendingan expertto set up a researchcenter, UNESCO shouldsend an expert to set up a coordinativescience policybody in Lebanon.61Negotiations over whichproject UNESCO would support,the centerto carryout researchor the science policybody to coordinateresearch, were eventuallyresolved when the Director-Generalof UNESCO, Rene Maheu, intervenedwith the Presidentof Lebanon.62The new body was to be a national researchcouncil thatwould organizeresearch and make policyabout sciencerather than a researchcenter thatwould do research. On Maheu's instructions,de Hemptinnespent several weeks duringthe summerof 1961 in Lebanon draftingenabling legislation for this council.63 Key featuresof de Hemptinne'sproposal were: (1) thatcoordination of researchin all scientificdisciplines were to be centralizedunder the council and (2) that the councilwas under no circumstancesto operate any typeof laboratoryor researchfacility itself. The nextstage of UNESCO's workinvolved sending a second science policyexpert, Charles Boschloos,to Lebanon to workwith the Lebanese to refineand revisethe proposed legislationdrafted by de Hempt- inne.By thistime the Lebanese had formeda nationalscientific commission of theirown whose purpose was to workwith UNESCO specialistsin designing the new council. In December, Boschloos and the Lebanese commission circulatedtheir revised draft of the enablinglegislation. It differedin several ways from de Hemptinne's draft,notably by weakeningboth of what de Hemptinnehad consideredto be the keyprovisions of the legislation-thatall scientificdisciplines were to be broughtunder council jurisdiction and thatno directinvolvement in researchactivities would be permitted. De Hemptinneand other secretariatmembers were furious.64Boschloos's contractwas terminatedand elaboratenegotiations were undertakenwith the Lebanese National ScientificCommission to push the legislationback onto what secretariatmembers considered to be the righttrack. UNESCO submit-

61. Memorandum from de Hemptinne to Adiseshiah, May 1961, NS memo 50.085. De Hemptinnealso proposed an elaborate three-tieredstructure for this coordinative bureaucracy. These proposalswere greatlysimplified under pressure from F. Karam at the Bureau of Member States (BMS) and fromthe Director-Generalof UNESCO himself.See memorandumfrom F. Karam,BMS, to Jos6Correa, director of BMS, 5 May 1961,BMS 80/memo100; and memorandum fromUNESCO Director-GeneralRene Maheu to M. V. Kovda, directorof UNESCO Department of NaturalSciences, n.d. (possiblyJune or July1961). 62. Memorandumfrom Maheu to Kovda, n.d. (possiblyJune or July1961). The factthat the Director-Generalof UNESCO and the President of Lebanon both were involved in these negotiationsindicates the importanceattached to themby both parties. Maheu, in fact,goes on to say in the above-citedmemo: "For manyreasons, I attachthe utmostimportance to thisproject which,in myview, has greatvalue as an example";translation mine. 63. This proposed legislationcomprises NS/ROU/7, 8 February1962, UNESCO Archives, Paris. 64. See memorandumfrom F. Karam, BMS, to A. K. Kinany,chief, Unit of Arabic-speaking Countries,BMS, 15 December 1961; and letterfrom de Hemptinneto T.O.P. Lilliefelt,permanent resident,Technical Assistance Bureau, Beirut,20 December 1961,NS 801/226(40). UNESCO 589 ted formalcomments to the commissionon the revised draft,arguing for changes back to the original proposals on these importantissues.65 After several weeks of negotiation,de Hemptinne was permittedto compile a synthesisof the two existing drafts for considerationby the Lebanese Parliament.66In it de Hemptinneincluded strong statements about bothof his chiefconcerns while making concessions to the commission'sdraft on issues of less consequence. The synthesisproposal was submittedto the Lebanese Parliament in February1962. Now the Parliamenthad two alternativesto consider: the Lebanese scientificcommission's proposal and de Hemptinne'srevised proposal. To promotehis alternative,de Hemptinnetraveled to Beirutthat spring to answer questions and speak with membersof Parliamentabout the new council proposals. He feared that the commissionwould produce some new counterproposaland derail his efforts.To offsetthis possibility he enlistedthe help of the National ScientificCommission's new president,Joseph Najjar, to keep himinformed of anynew developments.67In the end, the lobbyingefforts by Maheu, de Hemptinne,F. Karam, Malcolm Adiseshiah, and others at UNESCO paid off.On 28 August 1962 the Lebanese Parliamentapproved withoutdiscussion de Hemptinne'ssynthesis proposal for a National Scientific Research Council.68 UNESCO's activitiesin Lebanon did not stop with the creation of the council.Following passage of the enablinglegislation, UNESCO immediately plungedinto the taskof helpingthe Lebanese set up the new bureaucracyand ensuringthat that bureaucracy headed in the desireddirection. Before the end of 1962,UNESCO was recruitingtwo "expertsin the organizationof scientific research"to go to Lebanon and draftoperating regulations, budgets, and an organizationalchart for the new ScientificResearch Council.69UNESCO also

65. These commentsare containedin "Commentairesde l'UNESCO suirl'avant-projet de loi portantcr6ation d'un 'Conseil National de la Recherche Scientifique'redige en novembre1961 pare la CommissionScientifique National du Liban," (Commentsby UNESCO on the proposed legislationconcerning the creationof a "National Council for ScientificResearch" draftedin November1961 by the National Scientific Commission of Lebanon), 8 February1962, NS/ROU/9, UNESCO Archives,Paris. 66. "Avant-projetde loi portantcreation d'un conseil nationalde la recherchescientifique au Liban: Synthesedes avant projets de loi etablis par M. Y. de Hemptinne,Chef du Groupe d'organisationde la recherchescientifique de l'UNESCO et par la CommissionScientifique Nationale du Liban" (Proposed legislationconcerning the creationof a National Council for ScientificResearch for Lebanon: synthesisof proposed legislation drawn up by M. Y. de Hemptinne,group leader of UNESCO's Research Organization Unit, and by the National ScientificCommission of Lebanon), 8 February1962, NS/ROU/10, UNESCO Archives,Paris. 67. Letterfrom de Hemptinneto JosephNajjar, National ScientificCommission President, 20 February 1962. De Hemptinne was then head of the new Research Organization Unit at UNESCO's NaturalSciences Department. 68. Letter fromChafic Moharram,technical counselor to the Presidentof Lebanon, to de Hemptinne,3 October 1962. De Hemptinne'senabling legislationspecified that the council's budgetwas notto be less than 1 percentof the state'sbudget. 69. See, for example,B. K. Blount,"Report to the National Research Council of Lebanon," compiled 10 March-7 April 1964, Lebanon file,Secretariat Registry Files, UNESCO Archives, Paris. Blountwas deputydirector of the BritishDepartment of Scientificand IndustrialResearch and was a temporaryconsultant to UNESCO. 590 InternationalOrganization conducted externalreviews of Lebanese science policymakingat frequent intervalsover the nextdecade and offeredsuggestions for improvements.70

East Africa ObviouslyUNESCO's experiencesin promotingscience policy bodies among itsmembers differed from country to country.Records from one ofUNESCO's largesubsequent science policy campaigns in East Africaduring 1967 and 1968 providessome insightinto the rangeof experienceencountered by UNESCO consultants. First,UNESCO consultantsdid not alwaysspoon-feed the structureof the new sciencepolicy bureaucracy to countriesas theydid in Lebanon. Sometimes the originaldraft of enablinglegislation for the new bureaucracycame from some groupwithin the country,often a Ministryof Education or a Ministryof Planning,and was then sent either to UNESCO's headquarters or to its regionaloffice for comments and suggestions.71However as discussedearlier, UNESCO officialshad some firmnotions about what these science policy bureaucraciesshould look like and did not let opportunitiesto impose their views escape. Most often the draftswere returnednot just with extensive commentsbut also with a visitby a UNESCO expertwho would meet with relevantlocal officialsabout what UNESCO perceivedas shortcomingsof the country'splan and UNESCO's proposedremedies. For example,in 1966 theEthiopian government sent a draftof their order on the establishmentof a nationalresearch council to UNESCO's regionaloffice, which forwardedit along with suggested revisionsto the Science Policy Divisionstaff at UNESCO headquartersin Paris. Headquartersthen sent one of theirscience policyexperts out to Addis Ababa to attendmeetings of the draftingcommittee for the researchcouncil order and to provideinformation regardingcertain aspects of the proposed council's potential activities.72A similarcourse of eventstook place in Tanzania.73 Zambian officialson the other hand were makinggood progresstoward

70. See, for example, P. Piganiol, "Organisation de la politique scientifiqueau Liban" (Organization of science policy in Lebanon), 1967-68; and M. Steyaert,"Liban: politique scientifiquenational et organisationdes recherches oceanographiques" (Lebanon: national science policyand the organizationof oceanographicresearch), 1968, Lebanon file,Secretariat RegistryFiles, UNESCO Archives,Paris. 71. It should be noted that even where enabling legislation originatedin the countries themselves,UNESCO still provided some of the impetus for creatingthe new bureaucracy. Virtuallyall locallydrafted enabling documents cite UNESCO regionalscience policy conferences (forexample, the 1964Lagos conferenceamong African countries) as promptinglocal activity,and mostfollow conference recommendations to a largeextent. 72. Memorandumfrom I. C. Koupalov-Yaropolkto A. Matveyev,13 April 1967,science policy memo 541, Secretariat RegistryFiles, UNESCO Archives,Paris. Koupalov-Yaropolkwas a UNESCO policyconsultant and Matveyev,UNESCO AssistantDirector-General. In particular, see confidentialannex I, "Ethiopia." 73. Ibid., annexV, "Tanzania." UNESCO 591 creatinga nationalscience policyboard withoutUNESCO interventionwhen UNESCO staffofficials discovered their activity. UNESCO consultantsimme- diatelyinserted themselves into the process,offering advice and suggestions withoutany direct appeal fromthe Zambian government.74 The obstaclesencountered by UNESCO officialsin settingup these science bureaucracies also varied fromcountry to country.In Ethiopia, UNESCO experts complained that creation of the research council was "not very popular" and that those workingon the project "seem to be interestedin safeguardingthose rightsand privilegesof their institutionsthat mightbe delegated to the N.R.C. [National Research Council]. Hence theytry to re- duce the would-bepowers of the N.R.C."75 In Sudan, UNESCO officialshad trouble findingenough qualified scientiststo drafta proposal for the new science policy body, let alone staffit once it was created.76In Tanzania, UNESCO officialscomplained that a generalapathy about the entireproject prevailed.77In Kenya consultantscomplained of attemptsto subordinatethe sciencepolicy body to theMinistry of Economic Planning rather than making it partof the President'soffice and givingit directaccess to the highestlevels of government.78 One featureof all of UNESCO consultants'experience was the lack of familiaritywith the notionof a science policybureaucracy in memberstates, even in the highestgovernment and science circles,and the "necessity"for UNESCO officialsto spread the word. For example,in Ethiopia UNESCO consultantswere astonishedto findthat they were the firstpeople to present the idea of a nationalresearch council to the deans of the medical,engineering and buildingcolleges at Haile Selassie I University,despite the factthat the vice-chairmanof the committee drafting the enabling legislation for the council was the dean of the Facultyof Sciences at theuniversity.79 Despite thesedifficulties, all of the above-mentionedEast Africanstates had installedscience policybureaucracies of a type in keepingwith UNESCO's guidelinesby 1970-within three years of the UNESCO consultants'initial visits.80

74. Ibid., annexVIII, "Zambia." 75. Ibid.,confidential annex I, "Ethiopia." 76. Ibid., annexVII, "Sudan." 77. UNESCO consultantKoupalov-Yaropolk described the situationas follows:"The draft Constitutionof the National Research Council has been lyingfor some 14 monthsin theMinistry of Agriculture.[This] indicatesthat there are few people reallyinterested in the establishmentof N.R.C. or thatthey do nothave influenceenough to push thismatter forward." See ibid.,annex V, "Tanzania." 78. See discussionof twokey features of UNESCO's preferredform of a science policymaking body,above. 79. "Ethiopia," confidentialannex I to sciencepolicy memo 541. 80. Kenya is the only exception,since it did not create its own national science policy bureaucracyuntil 1977. The Kenyanrationale for not creatingsuch a bureaucracyearlier was that thenation could derivethe necessarybenefits from an existingEast Africanregional science policy bureaucracy. 592 InternationalOrganization

The relationshipbetween demand and supply

When initiallyanalyzing the data on creationof sciencepolicy organizations, I notedthat for the earliestinnovators in thisarea such as the United States and theUnited Kingdom, demand-driven explanations may be sufficient.UNESCO did not afterall inventscience policy.Rather, it picked up the notion from successfuland powerfulstates and popularizedit. Thus,while the firstscience policyorganizations may have been created in responseto domesticdemand, subsequentadoptions were stronglyinfluenced by systemic norms promoted by UNESCO. This kind of systemicsupply operation differsfrom simple imitationor mimeticexplanations-the more common explanation for diffusion patterns in the organization theory literature-on two counts. First, mimesis is an unmediatedprocess; it locates the impetusfor imitative actions in the imitator. In thiscase mimesiswould claim that each state,looking out at the worldof states,decided it wanted a science bureaucracylike the ones createdby these fewprominent trendsetters. As the foregoinganalysis makes clear, thisis not whathappened. Knowledgeof these innovationsand assertionsof thevalue of theseinnovations were suppliedto statesby a thirdparty, UNESCO. Second,mimesis is a process,not a cause. It saysnothing about why countries would choose to imitateone particularinnovation and not others. I have arguedthat in thiscase the drivingforce behind adoption of thisinnovation is normative.Positive evidence for the effectsof norms is always difficultto provide,which is one reason such explanationsare oftentreated as residuals. Here I cite rhetoricalshifts in discourseabout the organizationof scienceboth withinUNESCO and amongmember states that coincide with the behaviors described. This positive evidence, coupled with the failure of alternative explanations,provides a strongcase forthe role of norms.

Conclusionsand implications

The analysisoffered here is suggestivein severalways. First and mostobviously, it shows how forcesexternal to states can shape choices about internalstate structure.This in itselfis noteworthysince, as was discussed earlier, the literatureon statestructure has littleto say about international-levelcauses of statestructural change.

81. In theiranalysis of the spread of municipalreforms across U.S. cities at the turnof the century,Tolbert and Zucker provide a similar explanation of a diffusionprocess. Their interpretationis as follows:"As an increasingnumber of organizations adopt a programor policy,it becomes progressivelyinstitutionalized, or widelyunderstood to be a necessarycomponent of rational organizationalstructure. The legitimacyof the procedures themselvesserves as the impetusfor the lateradopters." See Pamela Tolbertand LynneZucker, "Institutional Sources of Change in the Formal Structureof Organizations:The Diffusionof Civil Service Reform, 1880-1935,"AdministrativeScience Quarterly 28 (March 1983), p. 35. UNESCO 593

In addition,the UNESCO storyreveals a relationshipbetween the international systemand states that is not easily accommodatedwithin traditional state-centricneorealist analysis. In neorealism,the force exercised by the internationalsystem on states is constraint;the systemis passive or at best reactive.The systemprevents states frompursuing certain policies that they maywant to pursue,but the identificationand definitionof preferredpolicies come from states. In this case, the system-levelactors were proactive; identificationand definitionof the policy options were supplied by an international-levelactor, UNESCO. Interestin international-levelsources of state policyhas been growingin recentyears. The literatureson epistemiccommunities, ideas, and transnational relationsall deal in one way or anotherwith international sources of statepolicy in whichsystem-level actors contribute to statepolicy debates in a positiverather than merely a negativeand constrainingway. Surprisingly little attentionhas been paid to internationalorganizations in this research;the analysishere suggeststhat international organizations have an importantrole to play. Third, the analysispresented here suggeststhat states are more socially responsiveentities than is recognizedby traditionalinternational relations theory.State policies and structuresare influencedby intersubjectivesystemic factors,specifically by normspromulgated within the internationalsystem. In thiscase, stateswere socialized by internationalorganizations and an international communityof experts-in thiscase scientists-to accept the promotion and directionof science as a necessaryand appropriaterole.82 Before 1955 most states had no perceptionthat a science policybureaucracy was in their interest;it was not part of theirutility function. Actions by UNESCO and examplesof a fewprominent developed states persuaded statesthat making science policywas an appropriateand necessarytask of states,regardless of objectivescience, developmental,or securityconditions. Thus, the empirical anomalyidentified at thebeginning of this research-that stateshave coordinative science bureaucracies regardlessof whetherthey have any science to coordinate-is the result of a behavioral norm (that states should direct science) makingsimilar claims on dissimilarstate actors. This findingis morecompatible with constructivist or "reflective"theoretical approachesthan it is withthe more conventionalapproaches to international relations.83The fact that states adopt policies not as an outgrowthof their individualcharacteristics or conditionsbut in responseto sociallyconstructed normsand understandingsheld bythe wider international community demon-

82. For an alternativeperspective on socializationof states in which hegemonsrather than internationalorganizations are the socializingforce, see G. John Ikenberryand Charles A. Kupchan,"Socialization and HegemonicPower," Intemational Organization 44 (Summer1990), pp. 283-315. 83. Robert Keohane, "International Institutions:Two Approaches," IntemationalStudies Quarterly32 (December 1988),pp. 379-96. 594 InternationalOrganization

stratesan embeddednessof statesin an internationalsocial systemthat conventionalapproaches ignore. In concreteterms, this research suggests at leastthree avenues for further research.First, it indicates a largelyunexplored role for international organizations,namely, their role as an arenain whichnorms and convergentexpecta- tionsabout internationalbehavior are developed.The regimesliterature correlatesnorms and convergentexpectations with regimes that often have affiliatedinternational organizations, but it understandsthe normsand convergentexpectations to producethe international organizations.84 It does not focuson the reverseprocess documented here in whichinternational organizations,once established,produce their own new convergent expecta- tionsand norms.Thus, the regimes literature has continuedto treatinterna- tionalorganizations as agentsonly; it also has had littleto say aboutthese organizationsas principalsand the active role played by these organizations in promotingnew norms. Second,this research builds on thegrowing literature on epistemiccommu- nitiesby showing how international communities of experts may use internationalorganizations as a base fromwhich to wield influence. However, this case does raise questionsfor epistemic communities scholars. While most of the peopleinvolved in UNESCO teachingactivities had scientific credentials, their motivationsfor acting derived from their status as internationalbureaucrats ratherthan from professional norms in thescience community or principled beliefsabout science. In fact,these scientists were challenging existing norms in thescience community, norms about state-science relations. The case heresuggests that specialized knowledge may not provide compel- lingprinciples for political action; those principles came from understandings about appropriatepolitical organization that were developedin UNESCO. Furtherwork on theconditions under which "epistemic" characteristics are determinativeand theways these interact with political and organizational variablesmight strengthen and clarify the epistemic communities framework. Third,this research underscores the now widely recognized need for more theoreticalwork to addressthe increasingly well-documented feedback effects ofsocial structures such as norms,shared expectations, and even international organizationson actorssuch as states.85Neorealist theory takes preferences as givenand understandsthem to driveinternational interaction. To theextent

84. Stephen D. Krasner,ed., IntemnationalRegimes (Ithaca, N.Y.: Cornell UniversityPress, 1983). 85. Krasner,in the conclusionof IntemationalRegimes, raised thisissue as did Keohane in his morerecent address to the InternationalStudies Association. Implications of the structure-actor relationshiphave been exploredby Ashley and by Kratchowiland Ruggie, among others.See Krasner,International Regimes; Keohane, "InternationalInstitutions: Two Approaches"; Richard Ashley,"The Povertyof Neorealism,"in Robert Keohane, ed., Neorealismand Its Critics (New York: Columbia UniversityPress, 1986), pp. 255-300; and FriedrichKratchowil and JohnGerard Ruggie, "InternationalOrganization: A State of the Art on an Art of the State," Intenational Organization40 (Autumn1986), pp. 753-75. UNESCO 595

that preferencesare shown to be the productrather than the producerof internationalinteraction, a new theoreticalapparatus to guide research in theseareas is needed.

Appendix:data and sources Science policy organizations Definition.UNESCO has definedscience policyorganizations as organiza- tionswhose centralpolicy making function [is] ... nationallevel. . . planning,organization, or co-ordinationofscientific and technological activities. Organizations such as Minis- triesor Departments of Science and Technology, National Research Councils, and Academiesof Science, as wellas otherbodies with similar overall responsibilities, havethus been included in the new UNESCO directory;bodies whose responsibili- tiesare limited to specificsectors of the economy or particular fields of science and technologyhave, on thecontrary, not been included.86 Two ambiguitiesarose in coding. The firstconcerns generalized state planningagencies whose responsibilityis to plan all aspectsof the economy.If these plans include science, do theyqualify as science policymakingbodies? The 1984 directoryis silent on this point,but the earlier 1960s directories specificallyexclude entities with such general responsibilities.87 I have done the same in requiringthat science policy organizations have scienceas theircentral concern. The second ambiguityconcerns the statusof nationalacademies of science. For theoreticalreasons made clear in the article,I am interestedonly in state organizations.However, not all academies are partof the state apparatus.(In the United States,for example, the NationalAcademy of Sciences is a private professionalsociety.) However, in many countries academies enjoy some amountof state support;in the formerSoviet Union and Soviet-stylestates, academies are constitutedin sucha wayas to make themdifficult to distinguish fromthe state apparatus.In such cases, the activepolicymaking and advising role playedby academies mightvery well be consideredthe firststate science policymakingorganization. To determinewhether or not academies shouldbe countedas state science policybureaucracies, I deferredto the UNESCO Science PolicyStudies and DocumentsSeries. These documentswere authoredby officials of the countries under study.If they presented their academy as their firstscience policy organization,as Cuba does, then it was coded as such. If theytreated the academyas a forerunnerof the "real" science policymakingapparatus, as the formerSoviet Union did,then it was notcoded as a sciencepolicy organization.

86. UNESCO, WorldDirectory of National Science Policy-making Bodies, Science PolicyStudies and DocumentsSeries, vol. 59 (Paris: UNESCO, 1984); p. viii.See also thedefinitions in theearlier directories,listed in thetextual notes above. 87. See UNESCO, WorldDirectory of National Science Policy-makingBodies, 3 vols. (Paris: UNESCO, 1966-68). 596 InternationalOrganization

Date of creation of initial science policy organizations. Data on creation dateswere obtained from UNESCO's SciencePolicy Studies and Documents Series.Most itemsin thatseries are analysesof sciencepolicy activity in a regionor UNESCO membercountry and usuallyinclude a briefhistory of sciencepolicy activity in membercountries. Often countries have formeda seriesof science policymaking organizations as differentgovernments reorga- nizedtheir bureaucracies. Ambiguity about which of thesemight be thefirst sciencepolicymaking organization was resolved by deferring to thenationals of the countryor regionin questionwho had authoredthese studies. This is clearlypreferable to codingthat would have allowed UNESCO designationof thefirst organization that qualifies as a sciencepolicy body.

Science data Data on thenumber of scientistsand engineersinvolved in R&D and the amountof spending on R&D as a percentageof GDP inthe year science policy organizationswere createdwere obtainedfrom the UNESCO Statistical Yearbook.88Not all countriescollect science data in all years. Data collectionby developingcountries in particular is sporadic.Where figures were not available forthe year of creation, the figure for the closest year available was used. In two cases (Chileand Tanzania), R&D spendingfigures were unavailable; thus the analysisof R&D as a percentageof GDP has been done fora sampleof forty-tworather than forty-four countries. The yearbookscontain extensive definitions of "scientist,""engineer," and "researchand development" used in compiling their data. For purposes here it is worthnoting that all ofthese figures, including R&D spending,are forboth publicand private sectors.

Development data GDP percapita in constant 1980 U.S. dollarswas used as a roughmeasure of development.These data werenot available in constantdollars of thesame base yearfor the relatively large span of yearsunder study here; they were calculatedbased on data publishedby the InternationalMonetary Fund (IMF).89Where necessary, conversions from one base yearof U.S. dollarsto 1980dollars was madeusing producer price indexes found in theStatistical Abstractof the United States.90Populations figureswere taken from the UNESCO StatisticalYearbooks. For countries not members of the IMF, figures forper capita GNP weresubstituted.91

88. UNESCO, StatisticalYearbook (Paris: UNESCO, variousyears). 89. InternationalMonetary Fund (IMF), IntemationalFinancial StatisticsYearbook (City of publication:IMF, variousyears). 90. U.S. Departmentof Commerce,Bureau ofthe Census,StatisticalAbstract ofthe United States (Washington,D.C.: U.S. GovernmentPrinting Office, 1992). 91. U.S. Arms Control and Disarmament Agency, World Expendituresand Anns Trade, 1963-1973 (Washington,D.C.: U.S. GovernmentPrinting Office, 1975). UNESCO 597

Securitydata Defensespending as a percentageof GNP in the yearof sciencepolicy creationwas used as a measureof perceived security threat. These figures were obtainedwhen available for the necessary years.92 In caseswhere science policy organizationswere created prior to 1963(when the U.S. ArmsControl Agency begancollecting these data), the figures were obtained from national statistical abstracts. Defensespending is onlya roughmeasure of perceived security threat since thereare a hostof domestic reasons why states may spend on defensehaving to do withmaintaining stability of governments. However, since these distortions generallyincrease rather than decrease defense spending, they should make us suspiciousof false-positive findings rather than false-negative ones. Thus, if the data revealeda correlationbetween high defense spending and creationof sciencepolicy institutions, we wouldwant to lookmore closely at ourdefense figures.The fact that even with these distortionsstates create science institutionsat consistently low levels of defense spending supports rather than underminesmy hypotheses.

92. U.S. ArmsControl and Disarmament Agency, World Military Expenditures and Arms Trade, 1963-1973(Washington, D.C.: U.S. GovernmentPrinting Office, 1975).