Você está na página 1de 23

The British Society for the Philosophy of Science

Ideal Types and Historical Explanation


Author(s): J. W. N. Watkins
Source: The British Journal for the Philosophy of Science, Vol. 3, No. 9 (May, 1952), pp. 22-43
Published by: Oxford University Press on behalf of The British Society for the Philosophy of
Science
Stable URL: http://www.jstor.org/stable/685472 .
Accessed: 17/02/2015 12:10

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

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

Oxford University Press and The British Society for the Philosophy of Science are collaborating with JSTOR to
digitize, preserve and extend access to The British Journal for the Philosophy of Science.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL
EXPLANATION*
J. W. N. WATKINS
I Introduction
IN thispaper1 I shallconsider: first,whatsortof creatures
idealtypes
shouldbe if they areto be usedin the constructionof socialtheories;
andsecondly,whatwe do whenwe try to explainhistoricaleventsby
applyingsuchtheoriesto them.2
I
Originallywrittento meet a problemwhich arosein the courseof Professor
K. R. Popper'sseminarat the LondonSchoolof Economicson 'The Philosophyof
History.' I owe a very greatdebt to ProfessorPopper'steachingsand criticisms.
A muchalteredandexpandedversionwas readto the OxfordSocialStudiesAsso-
ciationon 23rdNovember1951,whereit was subjectedto somepertinentcriticism
fromwhichthe presentversionhas,I hope,benefited.
2 The formalstructure of a predictionis, of course,the sameas that of a full-
fledgedexplanation. In both caseswe have: (a) initialconditions; (b) universal
statements; and (c) deductiveconsequences of (a) plus (b). We explaina given
event (c) by detecting(a) and by postulatingand applying(b); and we predicta
futureevent(c)by inferringit fromsomegiven(a)andpostulated(b). Nevertheless,
I thinkthatin socialscienceexplanationandpredictionshouldbe considered separ-
ately,for two reasons. First,as ProfessorG. C. Hempelhas pointedout in a most
illuminatingdiscussionof this problem(seehis 'The Functionof GeneralLawsin
History'in Readings Analysis,ed. H. FeiglandW. Sellars,New York,
in Philosophical
1949,pp. 462-5) in historywe often have to be content(andin fact arecontent)
with what he calls an explanationsketch,i.e. a somewhatvague and incomplete
indicationof (a) and(b)fromwhich (c)is not strictly deducible. Andif we go back
to a timewhen (a) butnot (c)hasoccurred,thispartialsketchof (a) and (b)will not
allow us to predict(c). Forexample,we may be satisfiedby the explanation that
SmithinsultedJones becauseJones had angeredhim, althoughwe shouldnotbe
prepared to admitthatifJonesangersSmithin thefuture,Smithwill necessarily react
by insultingJones.
Secondly,eventhe socialscientistwho canprovideafull-fledged explanationof a
pasteventwill runinto difficultiesif he triesto predictsimilarevents,becausethey
will occurin a systemwhich is not isolatedfrom the influenceof factorswhichhe
cannotascertain beforehand. TheAstronomer Royalcanprepare a NauticalAlmanac
for 1953becausehe is predictingthe movementsof bodiesin a systemisolatedfrom
extraneousinfluences,but the Chancellorof the Exchequercannot preparean
EconomicAlmanacfor 1953 because,even if he possessedsufficientknowledgeto
explaincompletelythe 1951levelsof prices,production,investment,exports,etc.,his
* Received15. i. 52
22

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

2 HolisticandIndividualistic
IdealTypes
It is only decentto begina discussionof idealtypesby considering
Weber'sviews; but he held two successiveconceptionsof what an
idealtypeshouldbe anddo, without,I think,realisingwhatimportant
differences lay betweenthem.
His earlierversionis set out in an articletranslated underthe title
'" Objectivity"in SocialScienceand SocialPolicy.' At thistime
(1904) Weberbelievedthatthe socialscientistshouldnot tryto imitate
the naturalscientist'sprocedureof systematically subsumingobserva-
tion-statements and low order theoriesundermore comprehensive
laws. The socialscientistshouldfirstdecidefromwhatpointof view
to approachhistory. Having decided,say, to treat its economic
aspect,he shouldthen selectfrom this some uniqueconfigurationof
activitiesand institutions,such as 'the rise of capitalism.'Then he
shouldpin down and describeits components. His final taskis to
drawin thecausallinesbetweenthesecomponents, imputing'concrete
effectsto concretecauses.'2
This programmecould never be carriedout; 'in any actual
economicsystemso manyfactorsareat work simultaneously thatthe
effectof a singlefactorby itselfcanneverbe known,for its tracesare
soon lost sight of.' 3 Andseparate factscannotbe linkedtogetheras
causesandeffectswith no referenceto generallaws. However,I will
not pressthesecriticismsof a methodologicalpositionwhich Weber
tacitlyabandoned later.
To assistthe socialscientistin this task of explainingparticular
eventsby relatingthemto theirparticular antecedents, Weberproposed
his first versionof the ideal type. This was to be constructedby
abstracting the outstandingfeaturesfrom some (moreor less clearly
demarcated) historicalcomplex,andby organising theseintoa coherent
word-picture. The idealityof sucha typelies in its and
simplification
aloofnessfrom detail: it will be freefromthe detailedcomplexityof
predictionsof futurelevels would undoubtedlybe upsetby unforeseeable,world-wide
disturbingfactors,the effectsof any of which might be cumulative.
Hence, the problem of social predictionraisesquestionsnot raisedby the problem
of historicalexplanation ; and this paperis not concernedwith the former.
1 Max Weber, The
Methodology of theSocialSciences,trans.and ed. E. A. Shils and
H. A. Finch, Illinois, I949, ch. 2
2 Op. cit. p. 79
" Walter Eucken, The Foundations of Economics,trans.T. W. Hutchison, London,
1950, p. 39
23

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
the actuality to be analysedwith its aid. As this kind of ideal type
emphasisesthe 'essential ' traits of a situation consideredas a whole,I
callit ' holistic,'in contrastwith the 'individualistic' idealtype described
by Weber in Part I of his posthumous Wirtschaft undGesellschaft.'
In this work he held that the social scientist'sfirst task was to build
up a generally applicabletheoreticalsystem; and for arrivingat this
he proposedthe use of ideal types similarto the models used in deduc-
tive economics. These are constructed,not by withdrawingfrom the
detail of social life, but by formalisingthe resultsof a close analysisof
some of its significant details considered in isolation. The holistic
ideal type was supposedto give a bird'seye view of the broad charac-
teristics of a whole social situation, whereas the individualisticideal
type is constructedby inspecting the situationsof actual individuals,
and by abstractingfrom these : (a) general schemes of personalpre-
ferences; (b) the different kinds of knowledge of his own situation
which the individualmay possess; and (c) varioustypicalrelationships
between individualsand between the individualand his resources. An
individualisticideal type places hypotheticalactorsin some simplified
situation. Its premissesare: the form (but not the specific content)
of the actors' dispositions, the state of their information, and their
relationships. And the deductive consequences of these premisses
demonstrate some principle of social behaviour, e.g. oligopolistic
behaviour. The ideality of this kind of ideal type lies: (i) in the
simplificationof the initialsituationand in its isolationfrom disturbing
factors; (ii) in the abstractand formal, and yet explicit and precise
characterof the actors'schemesof preferencesand statesof information;
and (iii) in the actors'rationalbehaviourin the light of (ii). It is not
claimed that a principleof social behaviourdemonstratedby an indi-
vidualistic ideal type will often have an exact empirical counterpart
(though the principle of perfect competition has been precisely
manifested,for instancein commodity-markets). But economists do
claim that there is a limited number of basic economic principles,and
that any economic phenomenon is a particularconfigurationof some
of these, occurring at a particularplace and time, which can be ex-
plainedby a synthesisof the relevantideal types, and by specifyingthe
content of their formal premisses.2
by A. R. HendersonandTalcottParsonsas TheTheoryof Socialand
1 Translated
Economic introd.by TalcottParsons,London,1947
Organisation,
2 ' This
morphologicalstudyof economichistoryrevealsa limited
numberof pure
formsout of which all economicsystemspastand presentaremadeup.' Eucken,
24

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION
Weber was no Platonist; he proposed both kinds of ideal type as
heuristic aids which, by themselves, tell you nothing about the real
world, but which throw into relief its deviations from themselves.
The individualisticideal type was to assistin the detectionof disturbing
factors, such as habit and tradition, which deflect actual individuals
from a rationalcourse of action--a proposalI shall examine later (see
p. 41). Now I shall examine the assumptionsunderlying Weber's
earlierproposalto use holistic ideal types.
One might improve one's appreciationof the shape of a roughly
circularobject by placing over it an accuratetracingof a circle. This
analogy brings out Weber's conception of the purpose,and mannerof
employing, holisticideal types in threerespects. (i) By comparingan
impure object with an ideal constructionthe deviationsof the former
from the latterare thrown into relief; and Weber did regardthis kind
of ideal type as a ' purelyideal limitingconceptwith which the real
situation... is compared andsurveyedfor the explicationof certainof
its significantcomponents.' x (ii) Both the objectand the construct
areconsideredas a whole. (iii) The analogyinvolveswhatis presup-
posedby the ideaof comparison, namely,a simultaneousawareness of
the characteristics of both things being compared. And in 1904
Weber did assumethat the socialscientistcan placehis knowledge
of a realsituationalongsidehisknowledgeof anidealtypehe hashim-
self constructed,and comparethe two.2 It is the simultaneous
' of the pureprinciples
op. cit.,p. io (myitalics). The ' de-idealisation of economic
theory which occurs when they arecombined into a particular
configuration which
is appliedto an empiricalcounterpart, is exactlyparalleledin the naturalsciences.
Forexample,Galileocombinedthe Lawof Inertia(whichdescribes the motionof a
body not acteduponby anyforce-a conditionwhichcanneverbe realised),andthe
Law of Gravity(which describesthe motion of a body in a vacuumwhich the
experimenter cannotobtain),and the principlesof air resistance,into a theoretical
configurationwhich allowscompletepredictionof the trajectories of e.g. cannon-
balls,if the initialconditionsareknown. 'All universalphysicalconceptsandlaws
... arearrivedat by idealisation.They therebyassumethatsimple... formwhich
makesit possibleto reconstruct any facts,howevercomplicated,by syntheticcom-
binationof theseconceptsand laws, thus makingit possibleto understand them.'
(ErnstMach,quotedby F. Kaufmann,Methodology of theSocialSciences,
New York,
1944,p. 87.) 1Methodology, p. 93
2 Thushe
speaksof' the relationshipbetweenthe logical structureof the con-
ceptualsystem . . . andwhatis immediatelygivenin empiricalreality' (op. cit.,
p. 96). The term 'immediatelygiven' shouldnot, I think,be takentoo seriously.
What this phrasedoesimplyis thatthe socialscientist's knowledgeof idealtype and
corresponding realityareon anequalfooting.
25

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
knowability of the featuresof both which enablesholistic ideal types
to be ' used as conceptualinstrumentsfor comparison with and measure-
mentof reality.'1
At this point an awkwardquestionarises: If the characteristics of a
historical situation have already been charted beforethe ideal type is
brought into play, why bother with ideal types ? They are not
hypotheses2 which guide the social scientistin his searchfor facts, for
they are not supposed to be realistic, or empirical. A holistic ideal
type is not a guessabout reality, but an a prioriword-picture--in other
words, a definition. What Weber's earlierproposalamountsto is that
holistic ideal types should be used as explicit definitions of those
'hundreds of words in the historian'svocabulary [which] are ambig-
uous constructscreatedto meet the unconsciouslyfelt need for adequate
expressionand the meaning of which is only concretely felt but not
clearly thought out.' 3
Thus the holistic ideal type transpiresto be something of a mouse,
a mere demandfor definitions; 4 and I shudderwhen I imagine each of
those 'hundreds of words' being replaced by lengthy verbal defini-
tions, though such defining maybe helpful in particularcircumstances.
For instance,to order and classify a collection of variegatedinstances
it may be necessaryto construct a scale with limiting ideal types at
either end. The survey of the constitutionsof 158 Greek stateswas
probably tidier and more systematic than it would have been if
Aristotle's ' Monarchy-Aristocracy-Polity' and ' Tyranny-Oligarchy-
Democracy' scales,or some equivalent,had not been used.
But such scales are for classifying facts already analysed, not for
analysing raw material; and the real weakness of Weber's earlier
proposallies in the method of historicalanalysiswhich was to accom-
pany the use of holistic ideal types. With individualisticideal types, it
will be remembered,we startwith individuals'dispositions,informa-
tion and relationships,and work outwards to the unintendedconse-
quencesof their interaction(deducinga price-level,for example, from
demand and supply schedules). But with holisticideal types the
analysisis supposed to proceed in the opposite direction. Here, the
historianis supposedto startwith the broad (or ' essential') character-
istics of an entire historicalsituation, and then to descendto an ever
closer definition of its deviationsfrom the ideal type with which it is
1 2
Op. cit. p. 97 Op. cit. p. 90 3 Op. cit. pp. 92-3
4 For a criticism of such demands, see K. R. Popper, The Open Societyand Its
Enemies,London, 1945, vol. 2, ch. ii, sect. ii
26

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

being compared. In principle,this descentfrom overall traits to


detailedingredientsmight continueuntil, at the endof the analysis,
the relevantdispositions,information,andrelationships of the people
concernedhadbeenestablished.
The idea that we can apprehendthe overallcharacteristics of a
socialsituationbeforelearningsomethingof theindividualsituationsof
the actorsin it appearsto be borneout by a statementsuchas, 'The
Britisheconomyin 1850wascompetitive.' Thisstatementapparently
attributesan overallcharacteristic to a demarcated whole,whilesaying
nothing about individuals(justas 'The lake'ssurfacewas calm' says
nothing about water-particles).Now the unintendedmerit of the
holisticidealtypeis thatits useforcesus to recognisethe falsityof this
idea. If, in orderto assessthecompetitiveness of the Britisheconomy
in 185o,we try to establishan idealtype of' perfectcompetition'we
shallat oncefindthatwe canonly defineit in termsof thepreferences,
informationandrelationships of individuals--an assertionwhichcanbe
confirmedby turningto any economicstext-book. In otherwords,
we shallhave established an individualisticidealtype.' But if know-
ledge of the generalcharacteristics of a social situationis always
derivativeknowledge,piecedtogetherfrom what is known of indi-
viduals'situations,thenit is not possiblefor historicalanalysisto pro-
ceed from overallcharacteristics towards individuals'situations. The
formeris logicallyderivativefrom the latter. Weber'searliercon-
ceptionof an idealtype presupposed thatone can detectthe essential
traitsof somehistoric'whole' while remainingaloof from the detail
of personalbehaviour; but this beliefis shownto be falsewhen we
actuallyconstructsucha type. It wasprobablythisexperiencewhich
laterled Weber tacitlyto abandonholisticideal types and the im-
possiblemethodassociated with them,in favourof individualistic ideal
types and the method of reconstructing historicalphenomenawith
their aid.2
1 Similarly, if we try to construct an ideal type for 'feudalism,' say, we shall at
once find ourselves speaking of people's obligations and privileges towards their
superiors,inferiors, the land, and so on.
2 What I call a 'holistic ideal type' roughly correspondsto what Eucken called a
' real type,' a name he used to denote the 'stages,' such as 'city economy,'
'early
capitalism,' 'mature capitalism,'through which, according to the Historical School
of economists, any economic system develops. He also rejectedsuch types in favour
of individualisticideal types (which he simply called ' ideal types ') and he criticised
Weber for confusing the two, but from a somewhat differentviewpoint to my own.
His fascinatingbook, The Foundationsof Economics,contains a sustainedplea for the
27

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
The assertionthat knowledge of social phenomena can only be
derived from knowledge about individualsrequiresone qualification.
Fortherearecertainovertfeatures1 whichcanbe established without
of
knowledge psychologicalfacts, such as the level of prices,or the
death-rate(butnotthe suicide-rate).And if we detectmore or less
regularchangesin suchovert featureswe have somethingeminently
suitablefor analysis. But somepeople,over-impressed by the quasi-
for a
regularityof, example, long-term'wave' in economic life, have
supposed that such a thing possessesa sort of internaldynamic,and
obeysits own laws; and thatwhile it must thereforebe takenas a
datum, many other phenomena(such as burstsof inventiveness,
emigrationmovements,outbreaksof war) canbe explainedas conse-
quencesof it.2
The Israelitesalso imputedtheir fortunesand misfortunesto a
superiorentityimmunefrom theirown activities; but they rightly
calledthis ' God.' All socialphenomenaare, directlyor indirectly,
humancreations. A lump of mattermay exist which no one has
perceived,but not a pricewhichno one hascharged,or a disciplinary
code to whichno one refers,or a tool whichno one would dreamof
using. Fromthis truismI inferthe methodologicalprinciplewhich
underliesthis paper,namely, that the social scientistcan continue
fertilemarriageof abstracttheoryandconcretefact,anda powerfulcriticismof the
HistoricalSchoolforblurringthedistinction betweenthetwo ; whereasI amarguing
againstmethodologicalholism, and for methodologicalindividualism.Our argu-
mentstendto coincidebecause' historicism ' is closelyrelatedto 'holism ' : thebelief
in laws of developmentpresupposes a ' whole' whichundergoesthe development.
For his discussionof realandidealtypes,see pp. 69-70, 173, 300, 326, 331,andes-
peciallypp. 347-9.
1 By 'overt feature'I do not meansomethingwhichcannecessarily be directly
perceived-it maybe a highlytheoretical construct. But whetherit be the priceof
a markedarticlein a shop-window,or the averagelevel of pricesin 1815,an overt
featureis somethingwhich canbe ascertained withoutreferringto people'sdisposi-
tions, etc. See R. Stone, TheRoleof Measurement in Economics, Cambridge,1951,
p. 9.
2 I havewrittenthe abovewith the Russian economistKondratieff in mind. He
assertsthat the view that long waves 'are conditionedby casual,extra-economic
circumstances andevents,suchas (I) changesin technique,(2) warsandrevolutions,
(3) the assimilation of new countriesinto the worldeconomy,and(4) fluctuations in
. ..
gold production reverse[s] the causal connections and the
take[s] consequence to
be thecause.' (N. D. Kondratieff, 'The LongWavesin EconomicLife,'Readings in
Business CycleTheory,BlakistonSeries,London,1950,ch. 2, p. 35.) In otherwords,
thelongwaveis thefundamental datum,in termsof whichallpolitical,technological,
etc., changesmustbe explained.
28

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

searchingfor explanations of a socialphenomenonuntilhe hasreduced


it to psychologicalterms.1 I am not, of course,denyingthatsucha
thing as a long-termprice movementwill partiallydetermineother
events,whichwill be partiallyexplicablein termsof it. I only assert
thatit too is, in principle,explicable,andexplicablein termsof indi-
vidualattitudestowardsthingsandotherpeople.
To sum up my argumentso far: An understanding of a complex
socialsituationis alwaysderivedfroma knowledgeof the dispositions,
beliefs,andrelationships of individuals. Its overt characteristics
may
be establishedempirically,but they areonly explained by beingshown
to be the resultantsof individualactivities.2
All this was recognisedby the later Weber. In The Theoryof
SocialandEconomic Organisation ideal type constructionmeans (not
detectingandabstracting theoverallcharacteristics
of a wholesituation,
andorganisingtheseinto a coherentscheme,but)placinghypothetical,
rationalactorsin some simplifiedsituation,andin deducingthe con-
sequencesof theirinteraction.
Suchintellectualexperimenting maybe fruitfulevenif someof the
premissesare very unrealistic.For instance,the conceptof a static
economy in equilibriumaids the analysisof the changesand dis-
equilibriaof actualeconomies. And grossexaggeration of one factor
may show up an influencewhich would otherwisehave been over-
looked. This is particularlyimportantin social sciencewhere the
influenceof differentfactorscan seldombe accuratelycalculated. If
1 Professor
M. Ginsbergtakesa contraryview : 'These principles[of historical
interpretation]are not necessarilyexclusivelypsychologicalor even teleological:
there may well be sociallaws sui generis. . .' (Aristotelian Society,Supplementary
Volume,XXI, 1947,' Symposium: The Character of a HistoricalExplanation,' p.
77). The only examplehe givesof somethingdetermined by suchlawsis phonetic
change. But I thinkthatthe problemof phoneticchangeis so differentfromthe
typicalproblemsof the socialsciences(war,unemployment, politicalinstability, etc.)
thatthereis a bettercasefor leavingit outsidetheirdomainthanthereis forwidening
theirscopeand methodsto let it in. The studyof thiskind of problemmightbe
classedwithvitalstatistics
andthestudyof contagiousdiseases which,ProfessorHayek
hasargued,shouldbe regardedas ' naturalsciencesof society' ratherthanas ' social
sciences.'(Individualism
andEconomic Order,London,1949,p.
2 An 57-)
explanation maybe in termsof the typicaldispositions of moreor lessanony-
mousindividuals,or in termsof the dispositions of specificindividuals. (Thisis the
basisof my distinctionbetween' explanation in principle' and' explanation in detail.'
Seep. 30.) Thus,you might try to explainan electionresultin termsof how 'the
Lancashire shop-keeper ' and'the non-partyprofessional man' etc., felt ; or, if you
hadan unlikelyamountof knowledge,in termsof the dispositions of eachelector.
29

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
E is the sortof effectproducedby F1,andif F1andE arebothpresent,
the socialscientisttendsto assumethatF, is thecauseof E, whereas
F, mayhavecausedonly a partof E, andan undetectedfactorF, may
havecausedtherestof E. Forexample,the domesticeconomicpolicy
of countryA will be a majorinfluenceon its own economy; but this
may alsobe influencedby the domesticeconomicpolicyof countryB.
In orderto show up this secondaryinfluence,we might assumepro-
visionallythat A exportsall its productionto, and importsall its
consumptionfrom,B, andthendeducethe effecton A of a changeof
policyin B.'
But that would be a preliminaryintellectualexperiment. The
premissesof a finishedidealtypeshouldbe sufficiently
realisticforit to
be applicableto historicalsituations. I now turnto the problemof
application.

3 Historical
Explanation
I shallconsiderthreelevelsof historicalexplanation
: (I)colligation
(whereidealtypesplayno significantrole) ; (II)explanationin prin-
ciple (which is the field par excellencefor ideal types); and (III)
explanation in detail(whereideal are
types mostlyconstructed adhoc,
andrenderedincreasinglyrealisticuntilthey becomeempiricalrecon-
structions).2
(I) Colligation.The term 'colligation' has been revivedby Mr
Walsh3 to denotea procedurewhich is important,not becauseit is
methodologically powerful,but becausemost 'literary' historiansdo
in fact useit when history. It
they write,for example,constitutional
means'explainingan event by tracingits intrinsicrelationsto other
eventsand locatingit in its historicalcontext.'4 Thuswe begin to
understand why a bill was enactedin May 1640 condemningStrafford
to deathwhen we learnof suchmattersas : his autocraticpowerin
Ireland; Parliament's fearof the IrisharmyandPym'sruthlessness as
J. E. Meade.
1 owe thisexampleto Professor
21
understandthat ProfessorF. A. Hayekalso drawsa distinctionbetweenex-
plainingin principleand explainingin detail,but thathe wishesto distinguishan
explanationof why, say,a pricerose,froma quantitative of the amount
explanation
by which it rose; whereasI wish to distinguishbetweenexplanationsin terms
of typical etc.,andexplanations
dispositions, andpersonal
in termsof thecharacteristics
of
idiosyncrasies the principalactors
concerned.
3 SeeW. H. Walsh,An Introduction toPhilosophyof History,London,1951,ch. 3,
53 4Op. cit.p. 59
30

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION
a parliamentary leader; the King'sdependenceon Parliament to pay
indemnitiesto the Scottisharmyin the north; and the angryanti-
royalistmob which beset Westminsterduringthe bill's passage. It
may also be better understoodby being colligatedwith subsequent
events. Thus the Long Parliament'slater treatmentof Laud and
Charlessuggeststhatits treatmentof Strafford was not eccentric,but
part of a campaignagainstextra-parliamentary power.
However,asMrWalshadmits,colligationyieldsonly whathe calls
a ' significantnarrative',whichis morethana chronicle,butlessthan
a full explanation, of the eventscolligated.
(II) Explanationin Principle.The principle of the automatic
governorcanbe demonstrated in a simplemodelwhichshowsthata
fallin sometemperature, voltage,speed,pressure,etc.,belowa certain
level will move a lever which will increasethe supplyof heat, etc.;
and vice versa. Understandingthis, you can explain the constant
temperature of your car'scirculatingwater in principle if you know
thatanautomaticgovernorcontrolsit, althoughyou do not understand
its detailedoperation.'
Analogousexplanations areusedin appliedeconomics. Consider
the bargainingprocess. The principleof thisis demonstrated in the
following ideal type. Two rational agents are postulated. Each
possesses one homogeneous, divisible good, and each knows the
scheduleof thosecombinations of variousportionsof his own andthe
other'sgood whichhe would exchangeindifferently for the whole of
his presentgood. Thesepremissesarehighlyprecise,andalsohighly
formal. Call them ocand/. Fromtheseit is deducedthatonly the
limitswithinwhich a bargainwill be struckaredetermined,andthat
within thoselimits the outcomewill be arbitrary. Call this conse-
quencew. Now considerpost-warAnglo-Argentinian tradenegotia-
tions. Here,we can,I think,detectfactorsA, B, c, d,... Z where:
A andB arethe resources andpoliciesof the tradedelegations, andare
concreteexamplesof o and 3 ; c, d, . . . areminorfactorswhosesmall
influenceson Z may partlycancelout ; andZ is the outcomeof the
negotiations, theactualinstabilityof Anglo-Argentinian traderelations,
whichis a roughempiricalcounterpart of o. The ' , 3, o 'idealtype
explainsin principlethe ' A, B, c, d, . . . Z ' situation.2
In this exampleI have assumedthat only one main economic
1 It is the principleof an inventionratherthanits physicaldetailwhichis usually
described in patents. SeeM. Polanyi,TheLogicofLiberty, London,1951,P. 21.
2 I owe this
exampleto ProfessorLionelRobbins.
31

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
principle,demonstrable in a single ideal type, was at work in the
historicalsituation. But the situationwill usuallybe morecomplex.
Considera wage-bargain.Perhapsthereis a closedshopandlimited
entryinto the tradeunion. The firm,a centrallyplannedorganisa-
tion,buysits rawmaterials, whicharerationed,througha government
agency, and its at
machinery the bestpriceit canget fromoligopolistic
suppliers. By law it mustexporta proportionof its produce,andthe
export marketis highly competitive. Its home pricesarefixed by a
cartelagreement.?The generalsituationis inflationary.
Here, the outcomeof the bargainingprocesswill be shapedby a
numberof economicprinciplesbesidesthatillustratedin the previous
example. Andin orderto understand thewholesituationin principle
it would be necessaryto buildup a complexmodelfromthe relevant
simpleidealtypes. Here,an' (a,f ), (A,), (ar,-r),.. . c ' modelwould
be usedto explainin principlean ' A, B, c, . . . L, M, n,.. . S, T, u,
Z' situation(wherec . .., n . .., u representcomparatively
uninfluentialfactors).
The socialscientist'sexplanations in principlelackthe quantitative
precisionof explanations in mathematical physics. But he mayclaim
that his explanationsare at any rate 'intelligible' and 'satisfying,'
whereasthose of the naturalscientistare not. The most universal
lawswhichthelatterappliesin hisexplanations andpredictionscontain
'
terms(e.g. elementaryquantum of action') whose connotation the
'
laymancannot picture'or ' graspintuitively.' Moreover,the status
of thesemost universallaws is probablyonly temporary: they will
probablycome to be subsumedunderhigherorderlaws.
But the ultimatepremissesof socialsciencearehumandispositions,
i.e. somethingfamiliarand understandable.They 'are so much the
of oureverydayexperiencethattheyhaveonly to be statedto be
,stuff
recognisedas obvious.'2 And while psychologymay try to explain
these dispositions, they do providesocialsciencewith a naturalstop-
1The definitionsof perfectcompetition,oligopoly and monopoly provide,
incidentally, good illustrations individualism.An
of the principleof methodological
entrepreneur faces: (a) perfectcompetitionif thepriceatwhichhe sellsis determined
mayleadto price
for him ; (b)oligopoly,if he canalterhisprice,butif thisalteration
his
changesby competitors which may force him to make undesiredaltera-
further,
tions to his own price; and (c) monopoly,if he canalterhis pricewithoutcausing
undesiredrepercussions.Competition,oligopoly and monopolyare nothingbut
the outcomeof the behaviourof interactingindividualsin certainrelationships.
2Lionel Robbins,TheNatureandSignificance of Economic
Science,London,1935,
P. 79
32

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

ping-placein the searchfor explanationsof overt socialphenomena.


The socialscientistmight claimmore. The naturalscientistcannot,
strictlyspeaking,verifyvalidhypotheses; he canonly refute falseones.-
He can say, ' If H, then E. But not-E. Thereforenot-H.' But
if he says, 'If H, then E. MoreoverE. ThereforeH' he com-
mits the fallacyof affirmingthe consequent.2But a socialscientist
mightclaimthata validsocialtheorycanbe verifiedbecauseboth its
conclusionsandits premissescan be confirmed-' you assentto the
formerbecausetheycorrespond with recognisedsocialfacts; andyou
assentto the latterbecausethey correspondwith your ideasof how
peoplebehave.' An exampleof the beliefthata socialtheorycan be
wholly verifiedby being confirmedat both endsis to be found in
Keynes'GeneralTheory. Therehe asserts'the fundamental psycho-
logical law, uponwhichwe areentitledto dependwith great con-
fidence. . . fromour knowledgeof humannature. . . , thatmenare
disposed,as a rule and on the average,to increasetheirconsumption
as theirincomeincreases,but not by as muchas the increasein their
income'; andviceversa.3He thenshowsthatthe empiricalfactthat
no depressionhasworseneduntil 'no one at all was employed' is a
deductiveconsequenceof this law.4 The theory is thus doubly
confirmed,andthereforeverified: ' it is certain thatexperiencewould
be extremelydifferentfromwhatit is if the law did not hold.'5 No
naturalscientistcould claimso muchfor hislaws. His explanations
are' surprising ' in thesensethathe explainsthefamiliarin termsof the
unconfirmableunfamiliar. So are the explanationsof a Freudian
psychologist. But the socialscientistexplainsthe familiarin termsof
the familiar. The elementof surprisein his explanationslies in the
logicaldemonstration of connectionswhichhadnot beenseenbefore
betweenfactswhichareprima faciediscrete.
But a doublecautionmustbe enteredagainstthe idea of double
confirmation in socialscience: (i) The sameconclusioncan,of course,
be deducedfrom differentsetsof premisses,andwe cannotbe certain
that our set of psychologicalassumptions is the correctset. (ii) Even
if ourpsychologicalassumptions arecorrect,andevenif we dofindthat
1 SeeK. R. Popper,LogikderForschung,
Vienna,1935,passim
2 See F. S. C. Northrop,The
Logicof theSciences andtheHumanities,New York,
1948,pp. 108-9; and e.g. H. W. B. Joseph,An Introductionto Logic,Oxford,1916,
pp. 522-3
3 TheGeneral Interest
andMoney,London,1936,p. 96
Theoryof Employment,
5
4 Op. cit. p. 252 Op. cit. p. 25I (my italics)
c 33

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
their deductive consequencescorrespondto recognisedfacts, we may
neverthelessbe mistakenif we explain these facts as a consequenceof
those psychological factors. This is becausewe can seldom calculate
the relative influence of different psychological factors (see p. 29).1
Thus Keynes' belief that people are disposed to save a smaller pro-
portion of their income if their income diminishes may well be
correct; his demonstrationthatthis generaldispositionwould not allow
depressionsto worsen indefinitely is immaculate; and the fact that
depressionsdo not worsen indefinitelyis undoubted. It is nevertheless
conceivablethat no depressionhas been halted becauseof this disposi-
tion. One may have been halted by an outbreakof war, another by
an upsurgeof confidence,anotherby a public works policy, and so on.
In explaining social phenomena we must not be content with the
detection of one factor which, singly, would have produced,after an
unstatedperiod, an unstatedamount of an effect which may, in any
particularsituation,have been causedmainly by quite differentfactors.
If I am right in supposing that social theories derive sociological
conclusions from psychological premisses,we should expect to find
that major theoretical advancesin social science consist in the per-
ception of some typical feature of our mental make-up which had
previously been disregarded,and in its formulationin a way which is
more deductively fertile and which goes to explain a wider range of
facts, than the psychological generalisationsrelied on hitherto. And
this is preciselywhat we do find. I think that it would be generally
conceded that economicsis the most maturesocialscience,and that the
two most strikingadvancesmadein economics duringthe last century
are: (i) the 'revolution' which occurred in the early 1870's when
Jevons, Mengerand Walrasintroducedthe conceptof marginalutility;
and (ii) the Keynesian'revolution.'
(i) The classicaleconomists saw that the price of a good must be
partly determinedby the demand for it, and that that demand must
reflect the buyers' estimatesof the good's utility-and yet diamonds,
whose utility is low, fetch a far higher price than water, whose utility
is high. So they tried to escapefrom their dilemmaby sayingthat the
price of a good is determinedby the cost of its production,though this
would obviously be untrue of an unwanted good which had been
expensivelyproduced. This difficultydissolvedwith the introduction
of the idea of the utility, not of a whole good, but of its leastimportant,
SI was myself inclined to accept the idea of double confirmation until Professor
Popper pointed out to me the relevanceof this consideration.
34

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION
or 'marginal,'unit. For-and this is the recognitionof a psycho-
logicalcontour-linewhichhadnotbeenclearlymappedbefore-it is in
termsof thatunitthatwe tendto valuea whole good ; andthe more
we have of the samegood, the more its marginalutility diminishes.
Hence, if diamondsbecameabundantand water very scarce,their
subjectivelydeterminedvalueswould be reversed. F. H. Knighthas
given a vivid descriptionof the eleganceandpowerof the conceptof
marginalutility:
To its admirersit comesnear to being the fulfilmentof the
eighteenth-century cravingfora principle whichwoulddo forhuman
conductandsocietywhatNewton'smechanics haddonefor thesolar
system. It introducessimplicityand order,evento theextentofmaking
it possibleto statetheproblems in theformof mathematical functions
dealtwithby themethodsof infinitesimal calculus.'
(ii) The readerwho is unfamiliar with Keynes'contributionto the
theoryof employmentmusttakeits valueon trust,for it is impossible
to describeit briefly. But hereagainwe find thatwhatit restson is
the perceptionand preciseformulationof certainhumandispositions
whichKeynesregardedas' ultimateindependent 2 andfrom
variables,'
whichhe coulddeducesuchdependentvariables(orovertphenomena,
asI havecalledthempreviously)astheamountof employmentandthe
generallevel of prices. At the heartof his GeneralTheoryKeynes
placed' threefundamental psychologicalfactors,namely,the psycho-
logicalpropensityto consume,the psychologicalattitudeto liquidity
andthe psychologicalexpectationof futureyieldfromcapital-assets.' 3

(III)Explanation inDetail. Themarkof anexplanation in principle


is its relianceon typical dispositionsand its disregardof personal
differences.But it is oftenimpossibleto disregardthese,forinstance,
in diplomatichistory. Here,the premissesof a historicalexplanation
must be the specificdispositions,beliefsand relationships of actual
people. This is what I call 'explanation in detail.'
So far,I haveallowedtwo questionsto lie dormant: (i) Whatis
the statusof these dispositions,and whereinlies their explanatory
power? (ii) What assumptionsconcerningpeople'srationalityare
we obliged to make when we explainsomethingin termsof their
dispositionsand beliefs? Thesequestionswere not acuteso long as
1 F. H. Knight,TheEthicsof Competition,
London,1935,P. 158
2
Op. cit. p. 246. Of course,the variablesare only 'independent'from the
socialscientist'spoint of view. The psychologistwould probablyconsiderthem
' 3
dependent'. Op. cit. pp. 246-7
35

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
explanations in principle were being considered. An explanation
requires a general statement as its major premiss; and when we
postulate a typical disposition we assert that all men (with trivial
exceptions and minor deviations, and, perhaps, within a limited
historico-geographicalarea) are prone to behave in a certain kind of
way; and this gives us the generality we require. And we explain
in principleby combining types which are, after all, ideal, and which
may therefore be expected to contain idealised simplificationsof real
life, such as the assumptionof fully rationalbehaviourin the light of
preferencesand beliefs.
But when we turn to explanationsin detail these two questionsdo
become acute. For we are here concerned with the variegated dis-
positionsof actualpeople, and theseappearto lack the generalitywhich
the majorpremissof an explanationneeds. And actualpeople do not
behave altogether rationally, which suggests that we cannot go on
assumingthat they do. I shalldiscussthe first questionunderthe head
of' Personality', and the second under the head of' Rationality and
Purposefulness'.
(i) Personality. A series of occurrencesconstitutesa person'slife,
and a complex and evolving system of dispositions constitutes his
personality.' Dispositions'are not laws, for they mention particular
things or persons. On the other hand they resemble laws in being
" "
partly variable" or open." ' 2 The dispositionswhich comprise
a uniquepersonalityare, so to speak, 'laws ' which apply to only one
man over a limited period of time. It is as if the laws of chemistry
concerning, say, mercury, applied only to a period in the life of one
1I haveadoptedthe terminologyof Professor G. Ryle'sConceptofMind(London,
1949; see especiallych. 5), but not that book's famousdenialthat a man has
'privilegedaccess'to his own mind. Sittingbesidethe driverof a car who turns
whiteandwrenchesthe steering-wheel over, I may perceiveinstantaneouslythathe
fearsan accident,but I do notfeelhisfear. Moreover,the historianis usuallyin the
positionof the policemanwho triesto reconstruct whathappenedfrom skid-marks
and reportsof witnesses; and for him the dualismbetweenuninterpreted overt
behaviour(e.g.Jan Masaryk's fallfroma Praguewindow) and its interpretation in
psychological termsis veryreal.
But the following characteristicremarkssuggestthat ProfessorRyle has now
modifiedhis originalanti-dualism : ' We have... a sortof (graduatedly)privileged
accessto suchthingsas palpitations of the heart,cramps,and creaksin the joints.'
'I haveelsewherearguedfortheideathata ticklejustisa thwartedimpulseto scratch.
... ButI do notthinknow thatthiswill do.' (' Feelings,'ThePhilosophical
Quarterly,
April 1951, I, 198-9)
2
Ryle, op. cit. p. 123
36

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

solitarybottleof mercurywhichhascomeinto existence,matured,and


will dissolve,andwhosetwin, we may confidentlyassume,neverhas
existed,andneverwill.
All this presupposesthatmen do havepersonalities, i.e. thattheir
behaviouris fairlyconsistentovera periodof timeif theirpersonalities
are not subjectedto dissolventshocks. This assumptionof the
quasi-permanence of personalities
corresponds roughly--veryroughly
-to the naturalscientist'sbelief in the permanenceof the natural
order.
The generalisations of psychology fit into this scheme in the
followingways : (a) Someattributea certaindispositionto all men.
The theory of the associationof ideas is an example. (b) Others
attributecertaindispositions to a certaintype of man,e.g. the ' intro-
vert.' (c)Yet othersattemptto describethedynamicsof personality-
development,deriving later dispositionsfrom prior determining
conditionsin the light of psychologicaltheory. (It is this searchfor
the primitivedeterminingconditionswhichleadsbackto the ' forma-
tive years' of early childhood.) An exampleis the theory of the
'incest-complex,' whichassertsthata manwho idealisedhis sisteras a
childwill be proneto hypoxesthesia on marriage.
A dispositionattributedto one man is no weakerthanthe same
dispositionattributedto all menin explainingandpredictingthatone
man'sbehaviour. 'X will acceptoffice' can be deducedfrom the
minorpremiss,'X believesthat if he refusesthe officehe has been
offeredhe will find himselfin the wilderness'in conjunctionwith
either(a) the majorpremiss,'All men seekpower,'or (b) the major
premiss,' X is a power-seeker '; butwhereas(b)maybetrue,(a)is the
sort of statementwhich is likely to be false becausemen are not
uniform,1
Similarly,a detaileddescriptionof one man'schess-playingdis-
positions (his knowledgeof the rules, evaluationsof the different
pieces,and abilityto see a certainnumberof movesahead)together
withhispresentbeliefsabouthisopponent'sintentionsandthepositions
of the pieces,implyhis next move, whichcouldnot be deducedfrom
propositionsabout chess-playersin generalin conjunctionwith a
descriptionof the presentstateof the game.
Thus the idea that the historian'sinterpretativeprinciplesare
simplygeneralisations abouthumannature,into which he musthave
1On law-like dispositionsof very limited generality,see R. Peters, ' Cure, Cause
and Motive,' Analysis,April I950, Io, no. 5, io6
37

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
specialinsight,is inadequate.' His knowledgeof humannaturein
generalhas to be supplemented by a knowledgeof the peculiarper-
sonalitiesof the principalactorsconcernedin the situationhe is trying
to understand, whetherhis problembe X's behaviour,or the chess-
player'snext move, or the rise of Christianity,or the Congressof
Vienna.
The dispositionswhich the historianattributesto a personalityhe
is trying to reconstructresemblescientificlaws in two furtherways.
(a) They arepostulatedhypotheseswhich correspondto nothing
observable,althoughobservablebehaviourcanbe inferredfromthem
in conjunctionwith factual minor premisses. Consequently,in
judging theirvaliditywe want to know, not the mentalprocessby
which the historianarrivedat them, but their degreeof successin
accountingfor what is known of the man'sbehaviour. The hypo-
theticaldispositions postulatedby the historianwho has' sympathetic-
ally identifiedhimself with his hero' may be richerthanthoseof the
historianwho hasnot done so, but it is not this which gives them a
certificateof reliability. ProfessorHempelhas put the mattervery
clearly:
Themethodof empathy is,no doubt,frequently appliedby laymen
andby expertsin history. But it doesnot in itselfconstitute an ex-
planation; it is rather a
essentiallyheuristic
device ; its function is to
suggest certain psychologicalhypotheses which might serve as ex-
planatoryprinciples in thecaseunder consideration.2
And the historian is no moreprecludedfrom reconstructing a strange
andunsympathetic personalitythanis the scientistfromreconstructing
the behaviourof an atomwhich does thingshe would not dreamof
doinghimself."
1 This idea underliesMr Walsh's contributionto the symposium on' The Charac-
ter of a Historical Explanation' (AristotelianSociety, SupplementaryVolumeXXI,
1947.) From it he infers that, since 'men's notions of human nature change from
age to age' we must recognise 'the subjectiveelement which history undoubtedly
contains ' (p. 66). The point is, do historians'notions of, say, Napoleon's personality
change from age to age (not because of the discovery of fresh evidence, etc., but)
arbitrarily?
2
Op. cit. p. 467. Failureto realisethis is, I think, the weaknessof R. G. Colling-
wood's The Ideaof History(ed. T. M. Knox, Oxford, 1946).
3 Failure to recognise this vitiates, I think, some of the argument in Professor
F. A. Hayek's ' Scientism and the Study of Society' (Economica, 1942, 9, 267-291 ;
1943, 1o, 34-63 ; 1944, II, 27-39). There, despite all the work done in abnormal
psychology, he asserts: 'When we speak of mind what we mean is that certain
38

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

(b) The dispositionswhich constitutea personalityalso resemble


scientificlaws in thatthey form a hierarchical system; andthisis of
considerablemethodologicalimportance. It is, of course,essential
that the dispositions which a historianattributesto a historicalfigure
shouldnot be mere ad hoctranslations of known occurrencesinto
dispositionalterms. It is explanationof Brutus'behaviourto say
no
thathe wasdisposedto assassinate Caesar,thoughit wouldbe a ground
for an explanationto saythatBrutuswas disposedto placehisloyalty
to the Stateabovehis loyaltiesto his friends,if independentevidence
were foundto supportthishypothesis. Moreover-andit is herethat
the ideaof a hierarchyof dispositions is important-the historianwho
canexplainsomeaspectof a person'sbehaviourup to a certain timein
termsof certaindisposition,althoughhissubsequent behaviourconflicts
with thisdisposition,mustnot merelysay thatat thattime the earlier
dispositiongave way to another. He should find a higherorder
dispositionwhichhelpsto explainboth earlierand laterlower order
dispositions,and hence the whole rangeof the person'sbehaviour.
For example: supposethatRussianforeignpolicyis controlledby a
consistent,integratedpersonality. Before1939Russiawas disposedto
pursuean anti-fascistforeignpolicy. But in 1939came the Russo-
GermanPact. In orderto explainthisaberration it is not enoughfor
the historianto say thatthe anti-fascist dispositionwas replaced. He
must find a higherorderdisposition(e.g. 'Russianforeignpolicy is
determinedby considerations of nationalexpediency,not by ideo-
logical factors') from which,in conjunction withfactualpremisses, the
in
change policy is derivable. In doing thisit is clear thatthe historian
will not be translatingan occurrence(the signingof the pact) into
dispositionalterms,but derivingboth the occurrenceand the change
in lower orderdispositions froma more permanentand fundamental
disposition.
In conclusionit shouldbe said that the personalityof a man in
society comprisesdispositionsboth of a more privateand tempera-
mental kind, and of a more public and institutionalkind. Only
phenomenacanbe successfully on the analogyof our own mind.... To
interpreted
recognisemindcannotmeananythingbut to recognisesomethingasoperatingin the
sameway as ourown thinking.'Fromthisfalsepremisshe correctlyinfersthe false
conclusionthat' historycannevercarryusbeyondthestagewherewe canunderstand
the workingof the mindsof the actingpeoplebecausethey aresimilarto our own'
(vol. 10,pp.61-2). Onlya war-likehistoriancantacklea GenghizKhanor a Hider!
I hastento add that I owe muchto otherpartsof ProfessorHayek'sargument,
andthatI suspectthatmy papercontainsa good dealof unwittingplagiarismof it.
39

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
certainindividualsare disposedto weep duringthe death-scenein
Othello,but all policemenare disposedto blow theirwhistlesunder
certain circumstancesand any Speakerin the House of Commons is
disposedto disallow parliamentarycriticismof exercisesof the
Prerogative. And these more public and institutionaldispositions,
which may vary very little when one man undertakesanother's role,
can be abstracted fromthe total,variegated andso
fluxof dispositions,
provide the social scientist
with a fairlystablesubject-matter.1
(ii) Rationality andPurposefulness. Beforeaskingwhatassumptions
the historianis obligedto makeaboutthe rationalityof thosewhose
behaviourhe is trying to interpret,we must establisha satisfactory
' definitionin use' of the term'rationalbehaviour.' Weberdefined
it, veryausterely,asthe deliberate andlogicalchoiceof meansto attain
explicit goals, in the light of existingfactualknowledge. This is
unsatisfactory for two reasons. (a) Whiteheadsaidsomewherethat
' civilisationadvancesby extendingthe numberof importantopera-
tionswe canperformwithoutthinkingaboutthem.' Thismorning's
tooth-brushingwas not irrationalbecausedone from habit and not
from deliberations on dentalhygiene. Our pursuitof goalsneednot
be consciousin orderto be rational. (b) Behaviouroften does not
conformto the end-means pattern. I maytell thetruth,or go fishing,
simply from a desire to do so, with no furtherendin mind.2
We escapethese difficultiesby sayingthat a personhas behaved
rationallyif he wouldhavebehavedin the sameway if, with the same
factualinformation,he had seen the full logicalimplicationsof his
behaviour,whetherhe actuallysaw them or not. And if we define
purposefulbehaviouras trying (consciouslyor otherwise)to do or
achievesomethingwanted,it followsthatfullyrationalbehaviouris a
limitingcaseof purposefulbehaviour.
The historianwho triesto interpretovertbehaviourmustassume
that it is purposefulbut not necessarilyfully rational.3 Consider a
crimepassionelcommittedby an enragedhusband. A judge who
assumedthatthe husbandhadbehavedpurposelessly couldreconstruct
the eventin a numberof quitearbitrary ways--perhapscrampcaused
1SeeHayek,op. cit. vol. 9, p. 284.
2 ' We investour in the hopeof gettingdividends.... But the
capitalreluctantly
anglerwould not acceptor understand an withoutthe activities
offerof the pleasures
of angling. It is anglingthat he enjoys,not somethingthat anglingengenders.'
Ryle, op. cit. p. 132. See also H. A. Pritchard,MoralObligation, Oxford, 1949,
pp. 10o-I.
3 See
Robbins, op. cit. ch. 4, sect. 5
40

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION
his fingerto contractroundthe triggerof a gunwhichhappenedto be
pointingat his wife's lover.' But while thejudge mustnot assume
purposelessness he need not assumefull rationality. The husband
would probablyhaveconfinedhimselfto threatsandremonstrances if
he hadpausedto considerthe lessimmediateconsequences of a violent
courseof action.
The assumption of purposefulness is constantlymadeby thosewho
attempt the most intensive of
analysis humanbehaviour,i.e. practising
psycho-analysts. It has often been pointed out that the psycho-
analystis on the sideof rationalityin thathe triesto curehis patients.
More interestingfrom our point of view is his assumptionthat the
behaviourof an uncured patientis thoroughlypurposeful. Supposea
patientforgets wind his watch, and so arriveslate at his father's
to
funeral. Unlikethe layman,the psycho-analyst will not attributethe
stopped watch to accidentalforgetfulness, to a purposelesspsychic
aberration.He will askhis patientwhyhe wanted his watchto stop-
maybe he felt on
guilty having a death-wish fulfilledandso createdan
excusefor avoidingthe funeral. Thiswould certainlybe purposeful
behaviour,andmightevenbe regardedas rationalbehaviourbasedon
misinformation.2
(iii) Conclusion.Havingconsideredthe statusof dispositionsand
theproblemof rationality, we cannow returnto explanations in detail.
Weber advocatedusing individualistic ideal types, which depict
rationalbehaviour,to show up the partialirrationalityof actual
behaviour. But thisis unacceptable.Supposethata historianwishes
to interpreta general'sbehaviourduringa battle. He has recon-
structed,as best he can, both the dispositionswhich constitutethat
aspectof the general'spersonality with whichhe is concerned,andthe
general's information aboutthe militarysituation. Supposethat, in
conjunction, these dictateretreatas the rationalcourseof action,but
thatthe generalis knownto havegiventhe signalto advance. Now
the historian,like the psycho-analyst, will not want to leavepuzzling
overt behaviouruninterpreted;but accordingto Weber he should
simplycall this a deviationfrom the ideallyrationalcourseof action
implied by the premissesof his theoreticalreconstructionof the
situation. But since an irrationalaberrationcan be attributedto
1 As ProfessorPopper once said to me,' Judges have a vested interestin murder ! '
2 The mixture of
rationality and misinformation due to childhood associations
which psycho-analysisbrings to the surface was pointed out to me by Professor
Popper.
41

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
J. W. N. WATKINS
anythingfrom boredomto panic, this procedurewould resultin
thoroughly arbitraryreconstructions.Rather, the historianmust
discoverthe mostsatisfactoryamendmentto the premissesof his ideal
type (constructedmore or lessadhocto depictthemainfeaturesof the
general'spersonality situation)whichwill removethe discrepancy
and
between what it implies and what happened.1Perhapsthere is
independentevidenceto suggest that the generalwas more lion-
heartedthanthe historianhad supposed; or perhapshe had under-
estimatedthe enemy's strength,or, in estimatingthe immediate
consequences of an advance,he had overlookeda more distantun-
desirablerepercussion.When ad hocidealtypesare usedin detailed
they have to be amendedand amendeduntil
historicalexplanations,
they ceasebeingidealconstructsandbecomeempiricalreconstructions.
The historianwho claimsto have interpreteda historicalsituation
shouldbe ableto show : (a)thatthebehaviourof theactorsin it flows
fromtheirpersonalitiesandsituationalbeliefs; and (b)thatsignificant
events which no one intendedare resultantsof the behaviourof
interactingindividuals.
4 Summary
An individual'spersonalityis a systemof unobservable dispositions
which, together with his factual beliefs, determineshis observable
behaviour. Societyis a systemof unobservable relationshipsbetween
individualswhoseinteractionproducescertainmeasurable sociological
phenomena. These, together with observable individual behaviour,
are the only overt facetsof a socialsystem. We can apprehendan
unobservable socialsystemonlyby reconstructing from
it theoretically
what is known of individualdispositions,beliefsand relationships.2
Henceholisticideal types,which would abstractessentialtraitsfrom
a socialwhole3 while ignoring individuals,are impossible: they
1
On satisfactoryand unsatisfactoryamendments of systems of dispositions, see
p. 39.
2
'The social sciences . . . do not deal with " given " wholes but their task is to
constitutethese wholes by constructingmodels from the familiarelements .. .' ' The
whole is never directly perceived but always reconstructedby an effort of our
imagination. Hayek, op. cit. vol. zo, p. 44, and p. 42 n. I.
3 Weber certainlydid not equate' essentialtraits' with overt, measurablepheno-
mena. Indeed, I do not think he would have regardedthe latter even as candidates
for inclusion among the former. By an 'essential trait' he meant what he called a
meaningfulconcept, or historicalclusterof ideas, such as ' Christianity,'' methodism,'
' socialism ' (see Methodology, pp. 94-7). But an uninterpreted,unexplained,observ-
able sociological phenomenon, such as a price-level, is not meaningful in this sense.
42

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions
IDEAL TYPES AND HISTORICAL EXPLANATION

alwaysturninto individualistic idealtypes. Individualisticidealtypes


of explanatorypower are constructedby firstdiscerningthe form of
typical,sociallysignificant,dispositions,and then by demonstrating
how, in varioustypicalsituations,theselead to certainprinciplesof
socialbehaviour.
If sucha principle,or a numberof suchprinciples,is at work in a
historicalsituation,the outcome of that situationcan be explained
anonymously,or in principle,by an applicationto it of the relevant
idealtype, or combinationof idealtypes. If the idiosyncrasies of the
actors concernedsignificantlyinfluencedthe outcome, it must be
explainedin termsof theirpeculiardispositions atndbeliefs. In either
case, the hypothetico-deductive method is used. The hypotheses
consistof postulateddispositions, beliefsand relationships
of (anony-
mous or specific)individuals; andtheirtestliesin the correspondence
or otherwisebetweentheirdeductiveconsequences andwhatis known
of the overtcharacteristicsof the situationbeingreconstructed.How
the historianestablishes of a vanishedsituation
the overtcharacteristics
is anotherstory.

The LondonSchoolof Economicsand PoliticalScience


HoughtonStreet,Aldwych,LondonW C 2

43

This content downloaded from 128.235.251.160 on Tue, 17 Feb 2015 12:10:03 PM


All use subject to JSTOR Terms and Conditions

Você também pode gostar