Space is the machine

Bill Hillier
SinceThe social logic of spacewaspublishedin1984,
BillHillierandhiscolleaguesatUniversityCollegeLondonhave
beenconductingresearchonhowspacefeaturesintheformand
functioningofbuildingsandcities.Akeyoutcomeistheconceptof
spatialconfgurationmeaningrelationswhichtakeaccountofother
relationsinacomplex.Newtechniqueshavebeendevelopedand
appliedtoawiderangeofarchitecturalandurbanproblems.Theaim
ofthisbookistoassemblesomeofthisworkandshowhowitleads
thewaytoanewtypeoftheoryofarchitecture:ananalytictheoryin
whichunderstandinganddesignadvancetogether.Thesuccessof
confgurationalideasinbringingtolightthespatiallogicofbuildings
andcitiessuggeststhatitmightbepossibletoextendtheseideasto
otherareasofthehumanscienceswhereproblemsofconfguration
andpatternarecritical.
Space is the machine
Bill Hillier
A confgurational theory
of architecture
Hardback and paperback
editions frst published in the
United Kingdom in 1996 and 1999,
respectively, by the Press Syndicate
of the University of Cambridge.
This electronic edition published
in 2007 by:
Space Syntax
4 Huguenot Place, Heneage Street
London E1 5LN
United Kingdom
www.spacesyntax.com
Copyright Bill Hillier 2004, 2007
ISBN 978-0-9556224-0-3
Layout and design
by Christian Altmann
Set in Haas Unica
Ahouseisamachineforlivingin
Le Corbusier (1923)

ButIthoughtthatallthatfunctionalstuff
hadbeenrefuted.Buildingsarent
machines. Student

Youhaventunderstood.Thebuildingisntthe
machine.Spaceisthemachine.Nick Dalton,
Computer Programmer at University College
London (1994)
Contents Spaceisthemachine|BillHillier SpaceSyntax
Prefacetothee-edition v
Acknowlegdements xii
Introduction 1
Part one Theoretical preliminaries
Chapterone Whatarchitectureaddstobuilding 10
Chaptertwo Theneedforananalytictheoryofarchitecture 39
Chapterthree Non-discursivetechnique 65
Part two Non-discursive regularities
Chapterfour Citiesasmovmenteconomies 111
Chapterfve Canarchitecturecausesocialmalaise? 138
Chaptersix Timeasanaspectofspace 171
Chapterseven Visiblecolleges 190
Part three The laws of the feld
Chaptereight Isarchitectureanarscombinatoria? 216
Chapternine Thefundamentalcity 262
Part four Theoretical syntheses
Chapterten Spaceisthemachine 288
Chaptereleven Thereasoningart 314
Index 344
Contents
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
Space is the Machine wasfrstpublishedin1996byCambridgeUniversityPress.
ThebookbuiltonthetheoryofsocietyandspacesetoutinThe Social Logic of
Space (CambridgeUniversityPress1984),tooutlineaconfgurationaltheoryof
architectureandurbanism.Unfortunately,althoughThe Social Logic of Space is
stillinprintafter23years,whentheinitialprint-runofSpace is the Machine was
exhausted,thenumberofcolourplatesforbadtheuseofthecheapreprinting
technologythatwouldhavemadeasuccessionofreprintseconomicallyviable.
So,althoughthebookwassellingwellatthetime,itfelloutofprint.Asdemand
forthebookhascontinued,forseveralyearscopiesofthebookhaveeitherbeen
impossibletofndorprohibitivelyexpensive.
IamnowimmenselypleasedthatSpaceSyntaxLimited,withsupportfrom
UniversityCollegeLondon(UCL),havedecidedtorectifythissituationbycreating
anewe-editionofthebookandmakingitavailableforfreeontheweb.Iam
particularlygratefultoTimStonorfortheinitialdecisiontofundtheproject,toTim,
ChrisStutzandShinichiIidafororganizingandmanagingtheprojectandacting
aseffectiveeditorsofthenewedition,andtoLauraVaughanandSuzanneTonkin
ofUCLfortheirencouragementthroughouttheprocess.Thanksandappreciation
arealsoduetoChristianAltmannforthenewdesignofthepublication;toRodrigo
Moraforpreparingelectronicimagesfromtheoriginalartworks;toMarcoGandini,
JosephLaycock,SachaTan,andSaussanKhalilforproofreading;toMollyHallfor
creatinganewindex;andtoChristianBerosforimagemanipulationandcreationof
thewebdistributionpagesforthee-edition.
LookingbackonSpace is the Machine,asonThe Social Logic of Space,
Ifndmyselfpleasantlysurprisedthatthefoundationssetoutthereforthespace
syntaxapproachtohumanspatialphenomenastillseemrobust.Atthesametime,
thedevelopmentsinthesubjectsince1996havebeensubstantial,notleastthrough
theinaugurationofthebi-annualspacesyntaxsymposiain1997(originallythe
brainchildofMarkDavidMajor).Thesehavecreatedaresourceofseveralhundred
papersondevelopingthetheory,methodologyandapplicationsofthespacesyntax
approachandnowconstituteoneofitsmostimportantresources.Tomepersonally,
itismostgratifyingthatasetofideascreatedbyasmallgroupofpeopleworking
togetheratUCLinthenineteenseventieshasnowfoweredintoalargeand
coherentbodyofworkbelongingtoaworld-wideresearchcommunity.
Attheriskofbeingunfairtoothers,however,itdoesseemtomethatcertain
contributionstothetheoryandmethodofspacesyntaxhavebeensosignifcant
astodeservereviewinthisprefacetowhatisnowaneleven-year-oldtext.For
example,onthetheoreticalfoundationsofspacesyntax,thethreepaperspublished
byJohnPeponisandhiscolleaguesoftheGeorgiaInstituteofTechnologyinAtlanta
inEnvironment and Planning B in1997and1998(Peponisetal1997,Peponisetal
1998a,b)onthegeometricalfoundationsseemsofpermanentsignifcance,asdo
thetwopapersofMikeBattyofCASAatUCL(Batty2004a,b)onthegraphtheoretic
Preface to the e-edition
v
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
foundations.Iwouldalsohopethatmyownattemptstoshowthattheeffectson
ambientspaceoftheplacingandshapingofphysicalobjectsaresystematicand
canbemathematicallyexpressedwillprovesimilarlyrobust.Theimportanceofthese
effectsbothfortheunderstandingofurbanform(Hillier2002),andhumanspatial
cognition(Hillier2007)will,Ihope,leadtoamoreunifedunderstandingofthelink
betweenthesetworealms.
Onthemethodologicalside,therehasbeenaremarkablefourishingof
newsyntacticmethodsfrommanysourcesandlocations.FromUCL,themost
signifcantofthesehavebeenthesyntacticisingofvisibilitygraphanalysisby
AlasdairTurnerinhisDepthmapsoftware(Turner&Penn1999,Turneretal2001)
andthedevelopmentofsegmentbasedaxialanalysiswithangular,metricand
topologicalweightings,initiallythroughthepioneeringworkofShinichiIidaand
hisSegmensoftwarewithsubsequentimplementationinDepthmap.Itwasthese
morecomplexanddisaggregatedformsoflineanalysisthatallowedustoshow
notonlythathumanmovementwasspatiallyguidedbygeometricalandtopological
ratherthanmetricfactorsbutalsotoclarifywhyapowerfulimpactofspace
structureonmovementwastobemathematicallyexpected(Hillier&Iida2005).
OtherkeymethodologicaldevelopmentsincludethepioneeringworkofDaltonon
angularanalysis(Dalton2001),nowavailableintheWebMapandWebMapatHome
software;theworkofFigueiredoandAmorimoftheUniversityofPernambucoin
BraziloncontinuitylinesintheMindwalksoftware(Figueiredo&Amorim2005),
whichextendlinesbydiscountingangularchangesbelowacertainthreshold;and
theSpatialistsoftwaredevelopmentbyPeponisandhiscolleaguesinconnection
withthethreepapersreferredtoabove.Othersignifcantsoftwaredevelopments
focusonlinkingspacetootherurbanfactorssuchaslandusepatternsand
densities,notablythePlaceSyntaxsoftwarefromMarcusandhiscolleaguesatthe
RoyalCollegeofTechnologyinStockholm,SequencesoftwaredevelopedbyStegen
atARSISinBrussels,andtheConfeegosoftwarepioneeredbyStutz,Gil,Friedrich
andKlaasmeyerforSpaceSyntaxLimited.
Inthemoresubstantiveareasoftheory,myownresearchhasexplored
theinter-relationsofspace,movementatdifferentscalesandlandusepatterns,
anditcannowarguablybeseentobepointinginthedirectionofadesign-level
(meaningpreciseenoughfortheideastobeusableindesign)theoryofcities
asself-organisingsystems.Thetheoryisintwoparts:ontheonehand,atheory
ofhowthespatialformofcitiesisshapedbyspatiallawslinkingtheemergence
ofcharacteristicallyurbanspacepatternstocognitiveaswellastosocialand
economicfactors;ontheother,atheoryofhowtheemergentpatternsofspace
shapemovement,andthroughthisshapelandusepatterns,leadingthrough
feedbackandmultipliereffects,tothegenericformofthecityasaforeground
networkoflinkedcentresatallscalessetintoabackgroundnetworkoflargely
residentialspace.CriticaltotheemergenceofthistheorywasthepaperCentrality
Preface to the e-edition
vi
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
asaprocess(Hillier1999)whichshowedhowlocalprocesseswithanessentially
metricnaturecombinedwiththelargerscalegeometricandtopologicalproperties
ofthespatialnetworktocreatetheprocessesbywhichcentresandsub-centres
emergeinthenetworkthroughthelogicofthenetworkitself-thougheachisof
coursealsoaffectedbyitsrelationtoothers.
Takentogetherthesedevelopmentsinspacesyntaxsuggestthatitoffers
apowerfulcomplementtotraditionalmethodsformodellingcities,notleasttransport
modellingmethods.Thesehaveconceptualfoundationsquitedifferentfrom
syntacticmodelsandseektoexplaindifferentthings,buttheycouldbebrought
intoasymbioticrelationwithsyntacticmodelstothebeneftofboth.Akey
researchpriorityintheimmediatefuturewillbetoexploretheirinter-relations.
Infact,followingthepioneeringworkofPennontheconfgurationalanalysis
ofvehicularmovement(Pennetall1998)workbyChiaradia,Rafordandothers
inSpaceSyntaxLimitedhasalreadysuggestedthatconfguationalfactorscan
contributeinsightsintootherkindsofmovementnetworks,includingcycles,
buses,andovergroundandundergroundrailnetworks.
Oneaspectofthedeepeningrelationbetweenspacesyntaxandthewider
spatialresearchcommunityhasbeenthedebateastohowfarspacesyntaxs
basictenets,suchastherepresentationofcitiesaslinenetworksandthesetting
asideofEuclideanmetricfactorsatthelargerspatialscaleinfavouroftopological
and/orgeometricones,aretheoreticallyvalidandmethodologicallyviable.Fromthe
syntacticpoint,certainpointsofcriticism,suchasthataxialmapsaresubjective
andmeasuresshouldbemetricised,seemtohavebeenanswered.Turneret
al(2005)haveshowedthatleastlinegraphs(allowingrandomselectionamong
syntacticallyequivalentlines)arerigorouslydefnedandindeedareobjectsofgreat
theoreticalinterestinthemselves,asisshownbyrecentworksuggestingthey
havefractalproperties(Carvalho&Penn2004).Likewisethecriticismthatsyntax
disregardsmetricinformationhasbeenansweredbyshowingclearlythatinterms
offunctionalitythisisascaleissue.Asshownin(Hillier1999)referredtoabove,at
asuffcientlylocalisedscalespaceworksinametricway,perhapsrefectingthe
scaleuptowhichpeoplecanmakereasonablyaccuratejudgementaboutdistance
incomplexspaces,soanaccountofthemetricpropertiesofspaceisnecessary
toafunctionallysensitiveandpredictiveanalysisofspaceatthislevel.Butatthe
non-locallevel,itseemsthatthefunctionalityofspacerefectspeoplesuseof
ageometricalpictureofthenetworkconnectivityratherthanametricpicturein
navigatingtheurbangrid,andatthisscaleintroducingmetricweightingintothe
measuresispositivelymisleading(Hillieretal2007).
Thestudyofspacewithinbuildingsusingspacesyntaxmethodshas
alsomuchadvancedsince1996,notleastofcoursethroughthepublicationof
JulienneHansonsDecoding Homes and Houses (1999),thethirdofthesyntax
booksfromCambridgeUniversityPress.AlsonotablehasbeentheworkofPenn
Preface to the e-edition
vii
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
andhiscolleaguesonspatialformandfunctionincomplexbuildings,inparticular
theinfuentialworkonspatialdesignandinnovationinworkenvironments.
Althoughnotstrictlywithinthesyntaxcontext,thehighlyoriginalworkofSteadman
(Steadman1998,2001)ontheenumerationofbuiltformsthroughaclarifcationof
geometric,constructionalandenvironmentalconstraintsbothanswersquestions
aboutenumerabilityraisedinSpace is the Machine,andoffersaplatformforanew
approachtospatialenumerabilitywhichcouldandshouldbetakenupwithinthe
syntaxcommunity.
Againstthebackgroundofthesetheoreticalandmethodological
developments,andcross-disciplinaryexchanges,spacesyntaxresearchisnow
becomingmuchmoreinterdisciplinary.FollowingaspecialissueofEnvironment
and Behaviourin2003editedbyRuthConroyDaltonandCraigZimringbringing
togetherpapersonspacesyntaxandcognitionfromthe2001AtlantaSymposium,
the2006conferenceonSpatialCognitionattheUniversityofBremenorganised
awell-attendedalldayworkshoponspacesyntax.Thelinkbetweenspace
syntaxandcognitivestudiesisnowbecomingawell-establishedbranchof
syntaxresearch.AtthesametimethepioneeringworkofLauraVaughanandher
colleaguesistakingsyntaxinthedirectionofagreaterengagementwithsocial
studies,andaspecialissueofProgress in Planningwillshortlyappearontheuse
ofspacesyntaxinthestudyofspaceasadimensionofsocialsegregationand
exclusion(Vaughan(ed.)2007).
Overall,spacesyntaxisbecomingafourishingparadigmforspatialstudies,
increasinglywellintegratedwithotherapproachesandincreasinglyexpandingits
scopeandscaleofinvestigation.Buttherealtestoftheoryandmethodisapplication
intherealworldofprojectsanddevelopment.HerethecontributionofSpaceSyntax
Limitedcannotbeoverestimated.Sinceitsfoundationasanactivecompanyoffering
spatialdesignandspatialplanningconsultancyundertheleadershipofTimStonor,it
hastestedthetheoryandtechnologyonawiderangeofprojects,manyofthemhigh
profle.TherearenowasignifcantnumberofprojectsinwhichSpaceSyntaxhas
exertedakeyspatialdesigninfuence,includingofcourseintheredesignofTrafalgar
Square(withNormanFoster)andNottinghamsOldMarketSquare(withGustafson
Porter),arguablythetwomostfamoussquaresintheUK,bothnowfunctioningina
newandhighlysuccessfulwayfollowingtheirrespectivere-designs.Otherupand
runningprojectsincludetheBrindleyPlacedevelopmentinBirmingham,Exchange
SquareandFleetPlaceinLondon,andofcoursetheMillenniumBridge,where
SpaceSyntaxshowednotonlyhowwellthebridgewouldbeusedbutalsohow
strongandbenefcialitslongtermeffectswouldbeontheareasonbothsidesofthe
river.Equallyinterestingtospacesyntaxarecaseswhereaspectsofspacesyntax
advicewasnotfollowed,sinceineachcaseproblemshaveappearedthatwere
clearlyforeseenbysyntaxatthedesignstage.
Preface to the e-edition
viii
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
Carefullyandresponsiblyused,itisclearthatsyntaxworksasadesignandplanning
tool.Oneconsequenceofitssuccessinrelativelysmall-scaledesignandplanning
problemsisthatsyntaxisnowincreasinglybeingusedasthefoundationforthe
space-basedmaster-planningofwholepartsofcitiesorevenofwholecities,andso
ineffectasanewwayofmodellingcities.Itisincreasinglywellunderstoodthata
syntacticmodelofacityhastwogreatadvantagesasacomplementtoanorthodox
model.First,asyntacticmodelallowsthedesignerorplannertoworkacrossall
urbanscalesusingthesamemodel,sothatoneformofanalysiswillidentifythe
largescalemovementnetworksanditslanduseeffects,whileanotherwillsimilarly
identifymicro-scalefeaturesandlandusepotentialsofthelocalurbangrid.Second,
exactlythesamemodelthatisusedinresearchmodetoinvestigateandunderstand
howthecityisworkingnowcanbeusedindesignandplanningmodetosimulate
thelikelyeffectsofdifferentdesignandplanningstrategiesandschemes,allowing
therapidexplorationofthelongtermconsequencesofdifferentstrategies.
SpaceSyntaxLimitedalsoconstitutesanexperimentinhowtherelations
betweenauniversityandaspin-outcompanycanbeorganised.AlthoughSpace
SyntaxLimitedcarriesoutitsownresearch,itmaintainsaverycloserelationto
theuniversityresearchdepartment,feedingproblemsintoitandtestingnewideas
andnewtechnologies.Collaborationisbothatthestrategicresearchlevel,butalso
reachesdowntothelevelofindividualprojectswherenecessary.Theexperience
ofaworkingcollaborationbetweentheuniversityandthecompanyhasconvinced
usallthatinthisfeldeventhemostbasicresearchcannotbeseparatedfromthe
demandsandquestionsraisedbypractice.Manytheoreticaldevelopmentshave
beensparkedbyquestionsraisedbyprojects,andatthesametimeprojectshave
providedasuperbearlytestinggroundforturningresearchideasintoworkable
andproventechnologies.ThefactthatitisSpaceSyntaxLimitedwhichisnow
re-publishingoneofthebasictheoreticaltextsofspacesyntaxisanemblemofthe
closenesswithwhichtheoryandpractice,andtheuniversityandthecommercial
world,havedevelopedcollaborativelyoverthepastdecade.
bh
June6th2007
Preface to the e-edition
ix
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
References
Batty,M.(2004)ANewTheoryofSpaceSyntax,WorkingPaper75,
CentreforAdvancedSpatialAnalysis,UCL,London:availablefromWWW
athttp://www.casa.ucl.ac.uk/working_papers/paper75.pdf

Batty,M.(2004)DistanceinSpaceSyntax,WorkingPaper80,CentreforAdvanced
SpatialAnalysis,UCL,London:availablefromWWWat
http://www.casa.ucl.ac.uk/working_papers/paper80.pdf

Carvalho,R.,Penn,A.(2004)Scalinganduniversalityinthemicro-structureof
urbanspace.PhysicaA332539-547.
Dalton,N.(2001)FractionalconfgurationanalysisandasolutiontotheManhattan
Problem.Proceedingsofthe3
rd
InternationalSpaceSyntaxSymposium,Georgia
InstituteofTechnology,Atlanta,GA,7-11May2001.

Figueiredo,L.,Amorim,L.(2005)Continuitylinesintheaxialsystem.Proceedings
ofthe5
th
InternationalSpaceSyntaxSymposium,TechnischeUniversiteitDelft,the
Netherlands,13-17June2005.

HansonJ(1999)Decoding Homes and HousesCambridgeUniversityPress,

Cambridge.

Hillier,B.(1999)Centralityasaprocess:accountingforattractioninequalitiesin
deformedgrids.Urban Design International 4107-127.

Hillier,B.(2002)Atheoryofthecityasobject.Urban Design International 7153-

179.

Hillier,B.,Iida,S.(2005)Networkandpsychologicaleffectsinurbanmovement.
InCohn,A.G.,Mark,D.M.(eds)Spatial Information Theory: COSIT 2005,Lecture
NotesinComputerSciencenumber3693,475-490,Springer-Verlag,Berlin.

Hillier,B.(2007)Studyingcitiestolearnaboutminds:howgeometricintuitions
shapeurbanspaceandmakeitwork.Environment and Planning B: Planning and
Design (forthcoming).

Hillier,B.,Turner,A.,Yang,T.,Park,H.T.(2007)Metricandtopo-geometric
propertiesofurbanstreetnetworks:someconvergences,divergencesandnew
resultsProceedingsofthe6
th
InternationalSpaceSyntaxSymposium,ITU,Istanbul,
Turkey,12-15June2007.
Preface to the e-edition
x
Preface to the e-edition Spaceisthemachine|BillHillier SpaceSyntax
Preface to the e-edition
xi
Penn,A.,Hillier,B.,Banister,D.,Xu,J.(1998)Confgurationalmodellingofurban
movementnetworksEnvironment and Planning B: Planning and Design 2559-84.

Penn,A.andDesyllas,J.andVaughan,L.(1999)Thespaceofinnovation:
interactionandcommunicationintheworkenvironmentEnvironment and Planning
B: Planning and Design,26(2)193-218

Peponis,J.,Wineman,J.,Rashid,M.,Kim,S.H.,Bafna,S.(1997)Onthedescription
ofshapeandspatialconfgurationinsidebuildings:convexpartitionsandtheirlocal
properties.Environment and Planning B: Planning and Design 24(5)761-781.
Peponis,J.,Wineman,J.,Bafna,S.,Rashid,M.,Kim,S.H.(1998)Onthegeneration
oflinearrepresentationsofspatialconfguration.EnvironmentandPlanningB:
PlanningandDesign25(4)559-576.
Peponis,J.,Wineman,J.,Rashid,M.,Bafna,S.,Kim,S.H.(1998)Describingplan
confgurationaccordingtothecovisibilityofsurfaces.Environment and Planning B:
Planning and Design 25(5)693-708.
Steadman,J.P.(2001)Everybuiltformhasanumber.Proceedingsofthe3
rd
InternationalSpaceSyntaxSymposium,GeorgiaInstituteofTechnology,Atlanta,
GA,7-11May2001.
Steadman,J.P.(1998)Sketchforanarchetypalbuilding.Environment and Planning
B: Planning and Design, 25th Anniversary Issue,92-105.
Turner,A.,Penn,A.(1999)Makingisovistssyntactic:isovistintegrationanalysis.
Proceedingsofthe2
nd
InternationalSpaceSyntaxSymposium,Universidadede
Braslia,Brasilia,Brazil,29March2April1999.
Turner,A.,Doxa,M.,OSullivan,D.,Penn,A.(2001)Fromisoviststovisibilitygraphs:
amethodologyfortheanalysisofarchitecturalspace.Environment and Planning B:
Planning and Design 28(1)103-121.
Turner,A.,Penn,A.,Hillier,B.(2005)Analgorithmicdefnitionoftheaxialmap.
Environment and Planning B: Planning and Design 32(3)425-444.
Vaughan,L.(ed.)(2007)ProgressinPlanning(TheSpatialSyntaxofUrban
Segregation)67(4)(inpress).
Acknowlegdements Spaceisthemachine|BillHillier SpaceSyntax
Acknowledgementsandthanksareduefrsttothemanyfriendsand
colleagueswhohave,overtheyears,madeanenormouscontributiontothe
ideasandresearchsetoutinthisbook,mostnotablyDr.JulienneHanson,Alan
PennandDr.JohnPeponis,eachofwhosecontributionshasbeentoolargeand
diversetoacknowledgeindetail;toNickSheepDaltonforthetitleofthebook,
andalsoforthebrilliantsoftwareonwhichmuchoftheresearchisfounded,the
outwardandvisiblesignofwhichisinthePlatesinthebook;toMarkDavidMajor
formastermindingthegradualandpainfulevolutionofthetextandillustrations;
toMyrto-Gabriella(Petunia)Exacoustouforreading,criticisingandhelpingme
substantiallyimprovethefnaldraft;toProfessorPatOSullivan,Headofthe
Bartlett,forasixmonthssabbaticalin1992,whenIsaidIwouldfnishbuthe
knewIwouldnt;totheEngineeringandPhysicalSciencesResearchCouncil
forcontinuedresearchfunding;tothemanycontributorstotheresearch,and
especiallyTimStonor,KayvanKarimi,BeatrizdeCampos,XuJianming,Gordon
Brown,JohnMiller,TadGrajewski,LenaTsoskounoglou,LauraVaughan,Martine
deMaesseneer,GuidoStegenandChangHuaYoo(whodrewthefrstversion
ofthemapofLondon);totheMScanddoctoralstudentsoftheBartlettSchool
ofGraduateStudieswhocontinuetogivesomuchintellectualbuzztothe
department;toProfessorsPhilipSteadman,TomMarkusandMikeBattyfor
sustainedintellectualsupportovertheyears;toStuartLiptonandhisteamat
StanhopePropertiesPlcforprovidinguswithsomanyopportunitiestoapplyour
researchonrealdevelopmentanddesignprojects,andtolearnsomuchfrom
them,andalsotoGordonGrahamoftheLondonRegenerationConsortium,the
SouthBankEmployersGroup,ChesterfeldPropertiesPlc,OveArupandPartners,
andPeterPalumbo;tothemanypublicbodieswhohaveinvitedustocontribute
totheirworkthroughappliedresearch,includingNationalHealthServiceEstates,
BritishRailways,BritishAirways,Powergen,theDepartmentofEducationand
Science,TechnicalAidforNottinghamCommunities,theLondonBoroughsof
CroydonandCamden,andtheTateGallery;tothemanyarchitecturalpractices
whohaveinvitedustoworkwiththemontheirprojects,butmostespecially
SirNormanFosterandPartners,theRichardRogersPartnership,TerryFarrell
andCompany,Skidmore,OwingsandMerrill,NicholasGrimshawandPartners,
BennettsAssociates,SWArchitects,andAvantiArchitects;toProfessorSheila
HillierandMarthaHillierfortoleratinganobsessivelap-topperinthehousefor
longerthanwasreasonable;toKate,CharlotteandBenHillierforcontinuingto
bethegoodfriendsandsupportersofapreoccupiedandinconsideratefather;to
thethiefwhotookallcopiesofthedraftofthefrstfourchaptersfrommyhome
whenstealingmycomputer,thussavingmefromprematurepublication;toRose
Shawe-Taylor,KarlHowe,EmmaSmith,SusanBeerandJosieDixonofCambridge
UniversityPress;andfnallytoUCLforcontinuingtobethemosttolerantand
supportiveofUniversities.
Acknowlegdements
xii
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction

In1984,inThe Social Logic of Space,writtenincollaborationwithJulienne

HansonandpublishedbyCambridgeUniversityPress,Isetoutanewtheory
ofspaceasanaspectofsociallife.Sincethenthetheoryhasdevelopedintoan
extensiveresearchprogrammeintothespatialnatureandfunctioningofbuildings
andcities,intocomputersoftwarelinkingspacesyntaxanalytictoolswith
graphicalrepresentationandoutputforresearchersanddesigners,andintoan
expandingrangeofapplicationsinarchitecturalandurbandesign.Duringthistime,
alargenumberofarticles,reportsandfeatureshaveappeared,theseshavebeen
writteninmanyuniversitiesusingthetheoryandmethodsofspacesyntax,and
researchhasbeeninitiatedinmanypartsoftheworldintoareasasdiverseas
theanalysisofarchaeologicalremainsandthedesignofhospitals.
Duringthistime,manytheoreticaladvanceshavealsobeenmade,
ofteninsymbiosiswiththedevelopmentofnewtechniquesforthecomputer
representationandanalysisofspace.Onekeyoutcomeoftheseadvancesis
thattheconceptofconfgurationhasmovedtocentrestage.Confguration
means,putsimply,relationstakingintoaccountotherrelations.Thetechniques
ofconfgurationalanalysis-ofwhichthevariousspacesyntaxtechniquesare
exemplars-thathavebeenbuiltfromthisideahavemadeitpossibletobring
theelusivepatternaspectofthingsinarchitectureandurbandesignintothe
lightofday,andtogivequantitativeexpressiontotheage-oldideathatitis
howthingsareputtogetherthatmatters.
Thishasinturnledtoacleararticulationofaphilosophyofdesign.
Architecturalandurbandesign,bothintheirformalandspatialaspects,are
seenasfundamentallyconfgurationalinthatthewaythepartsareputtogether
toformthewholeismoreimportantthananyofthepartstakeninisolation.The
confgurationaltechniquesdevelopedforresearchcan,infact,justaseasilybe
turnedroundandusedtosupportexperimentationandsimulationindesign.In
linkingtheoreticalresearchtodesigninthisway,wearefollowingahistorical
traditioninarchitecturaltheorywhichhasbothattemptedtosubjectthepattern
aspectofthingsinarchitecturetorationalanalysis,andtotesttheseanalyses
byembodyingtheminrealdesigns.Thedifferencenowisonlythatthe
adventofcomputersallowsustobringamuchgreatdegreeofrigour
andtestingtotheoreticalideas.
Theaimofthisbookistobringtogethersomeoftheserecent
developmentsinapplyingconfgurationalanalysistoissuesofarchitecturaland
urbantheoryintoasinglevolume.Thesurprisingsuccessofconfgurationalideas
incapturingtheinnerlogicofatleastsomeaspectsoftheformandfunctioning
ofbuiltenvironments,suggeststhatitmightinduecoursebeusefultoextend
theseideastootherareaswheresimilarproblemsofdescribingandquantifying
confgurationseemtobecentral,includingsomeaspectsofcognitivepsychology,
butalsoperhapssociologyitself.Atpresentweareencouragedbythecurrent
interestintheseideasacrossarangeofdisciplinesand,justasthelastdecade
hasbeendevotedtothedevelopmentandtestingoftechniquesofconfgurational

Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
2
analysiswithinarchitectureandurbandesign,sowehopethatthecoming
decadewillseecollaborationsamongstdisciplineswhereconfgurationisidentifed
asasignifcantproblem,andwheresomedevelopmentoftheconfgurational
methodologycouldconceivablyplayausefulrole.
Theimmediatecontextofthebookisthechangingtheoreticaldebatewithin
andaroundarchitecture.Lookingback,itiseasytoseethatinspiteoftheattention
paidtotheoryinarchitectureinthetwentiethcentury,andinspiteofthegreat
infuencethattheorieshavehadonourbuiltenvironment,architecturaltheoriesin
thelastdecadeshaveingeneralsufferedfromtwodebilitatingweaknesses.First,
mosthavebeenstronglynormative,andweaklyanalytic,inthattheyhavebeentoo
muchconcernedtotelldesignershowbuildingsandenvironmentsshouldbe,and
toolittleconcernedwithhowtheyactuallyare.Asaresult,theoriesofarchitecture
haveinfuencedourbuiltenvironmentenormously,sometimesforgood,sometimes
forill,buttheyhavedonelittletoadvanceourunderstandingofarchitecture.
Second,therehasbeenanexplosionofthehistorictendencytoform
architecturaltheoriesoutofideasandconceptsborrowedfromotherdisciplines.
Asaresult,architecturaldiscoursehasbeendominatedbyaseriesofborrowings,
frstfromengineeringandbiology,thenfrompsychologyandthesocialsciences,
thenfromlinguisticsandsemiology,andmostrecentlyofallfromliterarytheory.
Eachofthesehashadthemeritthatitallowedarchitecturetobecomepartofwider
intellectualdebate.Buttherehasbeenaprice,inthatverylittleattentionhasbeen
giventotheinternaldevelopmentofarchitectureasadiscipline.Throughthisturning
away,architecturehasincreasinglyignoredthelessonswaitingtobelearnedfrom
theintensivestudyofexperimentaltwentieth-centuryarchitecture,andacquired
whatnowamountstoahiddenhistoryinwhichkeyaspectsofrecentarchitectural
realityhavebeensuppressedasthoughtheyweretoopainfultotalkabout.
Theaimofthisbookistobegintheprocessofremedyingthisbiastowards
overlynormativetheoriesbasedonconceptborrowingfromotherdisciplines,by
initiatingthesearchforagenuinelyanalyticandinternaltheoryofarchitecture,
thatis,onebasedonthedirectstudyofbuildingsandbuiltenvironments,and
guidedbyconceptsformedoutofthenecessitiesofthisstudy.Theguidingbelief
isthatwhatweneedattheendofthetwentiethcenturyisabetteranddeeper
understandingofthephenomenonofarchitectureandhowitaffectspeopleslives,
andhowthisrelatestoinnovativepossibilityinarchitecture,andthecentralroleof
thearchitecturalimagination.
Thisbookisthereforeconcernedwithwhatbuildingsandcitiesarelike,
whytheyareastheyare,howtheywork,howtheycomeaboutthroughdesign,
andhowtheymightbedifferent.Thewordtheoryisusednotinthecommon
architecturalsenseofseekingsomesetofruleswhich,iffollowed,willguarantee
architecturalsuccess,butinthephilosophicalandscientifcsensethattheories
aretheabstractionsthroughwhichweunderstandtheworld.Anarchitectural
theory,asweseeit,shoulddeepenourgraspofarchitecturalphenomena,and
onlysubsequentlyandwithgreatmodesty,suggestpossibleprinciplesonwhich
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
3
tobasespeculationandinnovationindesign.Suchatheoryisanalyticbeforeitis
normative.Itsprimaryroleistoenquireintothepuzzlethatweseeandexperience
architecture,butwedonotunderstandwhatweseeandexperience.However
stronglywemayfeelthatarchitecturemaybewrongorright,werarelyunderstand
thearchitecturalgroundsonwhichsuchjudgmentsaremade.Thisbooktherefore
seeksanunderstandingofthetheoreticalcontentofarchitecture.
Thebookisinfourparts.Thefrst,Theoretical Preliminaries,dealswith
themostbasicofallquestionswhicharchitecturaltheorytriestoanswer:whatis
architecture,andwhataretheories,thattheycanbeneededinarchitecture?Inthe
frstchapter,Whatarchitectureaddstobuilding,thekeyconceptsofthebookare
setoutonthewaytoadefnitionofarchitecture.Theargumentisthatinaddition
tofunctioningasbodilyprotection,buildingsoperatesociallyintwoways:they
constitutethesocialorganisationofeverydaylifeasthespatialconfgurationsof
spaceinwhichweliveandmove,andrepresentsocialorganisationasphysical
confgurationsofformsandelementsthatwesee.Bothsocialdimensionsof
buildingarethereforeconfgurationalinnature,anditisthehabitofthehumanmind
tohandleconfgurationunconsciouslyandintuitively,inmuchthesamewayaswe
handlethegrammaticalandsemanticstructuresofalanguageintuitively.Ourminds
areveryeffectiveinhandlingconfgurationinthisway,butbecausewedoworkthis
way,wefnditverydiffculttoanalyseandtalkrationallyabouttheconfgurational
aspectsofthings.Confgurationisingeneralnon-discursive,meaningthatwedo
notknowhowtotalkaboutitanddonotingeneraltalkaboutitevenwhenweare
mostactivelyusingit.Invernacularbuildings,theconfgurational,ornon-discursive,
aspectsofspaceandformarehandledexactlylikethegrammaroflanguage,
thatis,asanimplicationofthemanipulationofthesurfaceelements,orwords
andgroupsofwordsinthelanguagecase,buildingelementsandgeometrical
coordinationsinbuilding.Inthevernaculartheactofbuildingreproducescultural
givenspatialandformalpatterns.Thisiswhyitseldomseemswrong.Architecture,
incontrast,isthetakingintoconscious,refectivethoughtofthesenon-discursive
andconfgurationalaspectsofspaceandform,leadingtotheexerciseofchoice
withinawidefeldofpossibility,ratherthanthereduplicationofthepatternsspecifc
toaculture.Architectureis,inessence,theapplicationofspeculativeandabstract
thoughttothenon-discursiveaspectsofbuilding,andbecauseitisso,itisalsoits
applicationtothesocialandculturalcontentsofbuilding.
Chapter2,Theneedforananalytictheoryofarchitecture,thentakesthis
argumentintoarchitecturaltheory.Architecturaltheoriesareessentiallyattemptsto
subjectthenon-discursiveaspectsofspaceandformtorationalanalysis,andto
establishprinciplestoguidedesigninthefeldofchoice,principleswhicharenow
neededasculturalguidanceisnolongerautomaticasitisinavernaculartradition.
Architecturaltheoriesarebothanalyticinthattheyalwaysdependonconjectures
aboutwhathumanbeingsarelike,buttheyarealsonormative,andsayhowthe
worldshouldberathermorestronglythantheysayhowitis.Thismeansthat
architecturecanbeinnovativeandexperimentalthroughtheagencyoftheories,but
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
4
itcanalsobewrong.Becausetheoriescanbewrong,architectsneedtobeableto
evaluatehowgoodtheirtheoriesareinpractice,sincetherepetitionoftheoretical
error-asinmuchofthemodernisthousingprogramme-willinevitablyleadtothe
curtailmentofarchitecturalfreedom.Theconsequenceofthisistheneedforatruly
analytictheoryofarchitecture,thatis,onewhichpermitstheinvestigationofthe
non-discursivewithoutbiastowardsoneorotherspecifcnon-discursivestyle.
Chapter3,Non-discursivetechnique,outlinestheprimerequirement
forpermittingarchitectstobeginthistheoreticallearning:theneedforneutral
techniquesforthedescriptionandanalysisofthenon-discursiveaspectsofspace
andform,thatis,techniquesthatarenotsimplyexpressionsofpartisanshipfor
aparticulartypeofconfguration,asmostarchitecturaltheorieshavebeeninthe
past.Thechapternotesacriticaldifferencebetweenregularitiesandtheories.
Regularitiesarerepeatedphenomena,eitherintheformofapparenttypingor
apparentconsistenciesinthetimeorderinwhicheventsoccur.Regularitiesare
patternsinsurfacephenomena.Theoriesareattemptstomodeltheunderlying
processesthatproduceregularities.Everysciencetheorisesonthebasisofits
regularities.Socialsciencestendtobeweaknotbecausetheylacktheoriesbut
becausetheylackregularitieswhichtheoriescanseektoexplainandwhich
thereforeoffertheprimetestoftheories.Thefrsttaskinthequestforananalytic
theoryofarchitectureisthereforetoseekregularities.Thefrstpurposeofnon-
discursivetechniqueistopursuethistask.
PartIIofthebook,Non-discursive Regularities,thensetsoutanumber
ofstudiesinwhichregularitiesintherelationbetweenspatialconfgurationand
theobservedfunctioningofbuiltenvironmentshavebeenestablishedusingnon-
discursivetechniquesofanalysistocontrolthearchitecturalvariables.
Chapter4,Citiesasmovementeconomiesreportsafundamentalresearchfnding:
thatmovementintheurbangridis,otherthingsbeingequal,generatedbythe
confgurationofthegriditself.Thisfndingallowscompletelynewinsightsintothe
structureofurbangrids,andthewaythesestructuresrelatetourbanfunctioning.
Therelationbetweengridandmovementinfactunderliesmanyotheraspectsof
urbanform:thedistributionoflanduses,suchasretailandresidence,thespatial
patterningofcrime,theevolutionofdifferentdensitiesandeventhepart-whole
structureofcities.Theinfuenceofthefundamentalgrid-movementrelationisso
pervasivethatcitiesareconceptualisedinthechapterasmovementeconomies,
inwhichthestructuringofmovementbythegridleads,throughmultipliereffects,to
densepatternsofmixeduseencounterthatcharacterisethespatiallysuccessfulcity.
Chapter5,Canarchitecturecausesocialmalaise?thendiscusses
howthiscangowrong.Focussingonspecifcstudiesofhousingestatesusing
confgurationalanalysiscoupledtointensiveobservationaswellassocialdatait
isshownhowtheoverlycomplexandpoorlystructuredinternalspaceofmany
housingestates,includinglow-riseestates,leadstoimpoverishmentofthevirtual
communitythatis,thesystemofnaturalco-presenceandco-awarenesscreated
byspatialdesignandrealisedthroughmovement-andthisinturnleadstoanti-
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
5
socialusesofspace,whicharethefrststageindeclinetowardsthesinkestate.
Becausetheroleofspaceinthisprocessistocreateadisorderlyandunsafe
patternofspaceuse,andthisisthenperceivedandexperienced,itispossibleto
conceptualisehowarchitectureworksalongsidesocialprocessestocreatesocial
decline.Inasense,thecreationofdisorderlyspaceusethroughmaladroitspace
designcreatesthefrstsymptomsofdecline,evenbeforeanyrealdeclinehas
occurred.Inasensethen,itisargued,wefndthatthesymptomshelptobring
aboutthedisease.
Chapter6,Timeasanaspectofspacethenconsidersanother
fundamentaldifferencebetweenurbanforms:thatbetweencitieswhichservethe
needsofproduction,distributionandtrade,andthosewhichservetheneedsof
socialreproduction,thatisofgovernment,majorsocialinstitutionsandbureaucracies.
Aseriesofstrangetownsareexamined,anditisshownhowintheirspatial
properties,theyareinmanysensestheoppositetothenormaltownsconsidered
inChapter5.Thedetailedspatialmechanismsofthesetownsareexamined,and
agenotypeproposed.Anexplanationisthensuggestedastowhycitiesofsocial
reproductiontendtoconstructthesedistinctivetypesofspatialpatterns.
Chapter7,VisibleColleges,thenturnstotheinteriorsofbuildings.It
beginsbysettingoutageneraltheoryofspaceinbuildings,takingintoaccountthe
resultsofsettlementanalysis,andthenhighlightsaseriesofstudiesofbuildings.
Akeydistinctionismadebetweenlongandshortmodels,thatis,betweencases
wherespaceisstronglygovernedbyrules,andthereforeactstoconservegiven
socialstatusesandrelationshipsandcaseswherespaceactstogeneraterelations
overandabovethosegivenbythesocialsituation.Theconceptoflongandshort
modelspermitssocialrelationsandspatialconfgurationtobeconceptualisedin
ananalogousway.Aritualisalongmodelsocialevent,sinceallthathappens
isgovernedbyrules,andaritualtypicallygeneratesaprecisesystemofspatial
relationshipsandmovementsthroughtime,thatis,aspatiallongmodel.Apartyis
ashortmodelevent,sinceitsobjectistogeneratenewrelationshipsbyshuffing
theminspace,andthismeansthatrulesmustbeminimisedbyusingaspatial
shortmodel.Inalongmodelsituationspaceisadaptedtosupporttherules,and
behaviouralrulesmustalsosupportit.Inashortmodelsituation,spaceevolvesto
structure,andoftentomaximise,encounterdensity.
PartIIIofthebook,The Laws of the Field,thenusesthesenoted
regularitiestoreconsiderthemostfundamentalquestionofallinarchitectural
theory:howisthevastfeldofpossiblespatialcomplexesconstrainedtocreate
thosethatareactuallyfoundasbuildings?First,inChapterEight,Isarchitecture
anarscombinatoria?,ageneraltheoryofpartitioningisproposed,inwhichitis
shownthatlocalphysicalchangesinaspatialsystemalwayshavemoreorless
globalconfgurationaleffects.Itisthelawsgoverningthispassageformlocal
physicalmovestoglobalspatialeffectsthatarethespatiallawsthatunderlie
building.Theselocal-to-globalspatiallawsarelinkedtotheevolutionofreal
buildingsthroughwhatwillbecalledgenericfunction,bywhichismeantthe
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
6
spatialimplicationsofthemostfundamentalaspectsofhumanuseofspace,that
is,thefactofoccupationandthefactofmovement.Atthisgenericlevel,function
imposesrestraintsonwhatisspatiallyviable,andthisisresponsibleforwhatall
buildingshaveincommonasspatialdesigns.Genericfunctionisthefrstflter
betweenthefeldofpossibilityandarchitecturalactuality.Thesecondflteristhen
theculturalorprogrammaticrequirementofthattypeofbuilding.Thethirdflteris
theidiosyncrasiesofstructureandexpressionthatthendistinguishthatbuilding
fromallothers.Thepassagefromthepossibletotherealpassesthroughthese
threeflters,andwithoutanunderstandingofeachwecannotdeciphertheform-
functionrelation.Mostofall,withoutaknowledgeofgenericfunctionanditsspatial
implicationswecannotunderstandthatwhatallbuildingshaveincommonintheir
spatialstructuresisalreadyprofoundlyinfuencedbyhumanfunctioninginspace.
InChapter9,Thefundamentalcity,thetheoryofgenericfunctionandthe
threefltersisappliedtocitiestoshowhowmuchofthegrowthofsettlements
isgovernedbythesebasiclaws.Anewcomputermodellingtechniqueofall
lineanalysis,whichbeginsbyconceptualisingvacantspaceasaninfnitely
densematrixoflines,containingallpossiblestructures,isusedtoshowhowthe
observableregularitiesinurbanformsfromthemostlocaltothemostglobalcanbe
seentobeproductsofthesameunderlyingprocesses.Afundamentalsettlement
processisproposed,ofwhichparticularculturaltypesareparameterisations.Finally,
itisshownhowthefundamentalsettlementprocessisessentiallyrealisedthrough
asmallnumberofspatialideaswhichhaveanessentiallygeometricalnature.
PartIVofthebook,Theoretical Syntheses,thenbeginstodrawtogether
someofthequestionsraisedinPartI,theregularitiesshowninPartIIandthe
lawsproposedinPartIII,tosuggesthowthetwocentralproblemsinarchitectural
theory,namelytheform-functionproblemandtheform-meaningproblem,canbe
reconceptualised.Chapter10,Spaceisthemachine,reviewstheform-function
theoryinarchitectureandattemptstoestablishapathologyofitsformulation:how
itcametobesetupinsuchawaythatitcouldnotbesolved.Itthenproposeshow
theconfgurationparadigmpermitsareformulation,throughwhichwecannotonly
makesenseoftherelationbetweenformandfunctioninbuildings,butalsowecan
makesenseofhowandwhybuildings,inapowerfulsensearesocialobjectsand
infactplayapowerfulroleintherealisationandsustainingofhumansociety.
Finally,inChapter11,Thereasoningart,thenotionofconfgurationisapplied
tothestudyofwhatarchitectsdo,thatis,design.Previousmodelsofthedesign
processarereviewed,anditisshownthatwithoutknowledgeofconfguration
andtheconceptofthenon-discursive,wecannotunderstandtheinternalities
ofthedesignprocess.Anewknowledge-basedmodelofdesignisproposed,
withconfgurationatitscentre.Itisarguedfromthisthatbecausedesignisa
confgurationalprocess,andbecauseitisthecharacteristicofconfgurationthat
localchangesmakeglobaldifferences,designisnecessarilyatopdownprocess.
Thisdoesnotmeanthatitcannotbeanalysed,orsupportedbyresearch.Itshows
howeverthatonlyconfgurationallybiasedknowledgecanreallysupportthedesign
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
7
process,andthis,essentially,istheoreticalknowledge.Itfollowsfromthisthat
attemptstosupportdesignersbybuildingmethodsandsystemsforbottomup
constructionofdesignsmusteventuallyfailasexplanatorysystems.Theycan
servetocreatespecifcarchitecturalidentities,butnottoadvancegeneral
architecturalunderstanding.
Inpursuingananalyticratherthananormativetheoryofarchitecture,
thebookmightbethoughtbysometohavepretensionstomaketheartof
architectureintoascience.Thisisnotwhatisintended.Oneeffectofabetter
scientifcunderstandingofarchitectureistoshowthatalthougharchitectureas
aphenomenoniscapableofconsiderablescientifcunderstanding,thisdoesnot
meanthatasapracticearchitectureisnotanart.Onthecontrary,itshowsquite
clearlywhyitisanartandwhatthenatureandlimitsofthatartare.Architectureis
anartbecause,althoughinkeyrespectsitsformscanbeanalysedandunderstood
byscientifcmeans,itsformscanonlybeprescribedbyscientifcmeansina
veryrestrictedsense.Architectureislawgovernedbutitisnotdeterminate.What
isgovernedbythelawsisnottheformofindividualbuildingsbutthefeldof
possibilitywithinwhichthechoiceofformismade.Thismeansthattheimpact
oftheselawsonthepassagefromproblemstatementtosolutionisnotdirectbut
indirect.Itliesdeepinthespatialandphysicalformsofbuildings,intheir
genotypes,nottheirphenotypes.
Architectureisthereforenotpartart,andpartscience,inthesensethatit
hasbothtechnicalandaestheticaspects,butisbothartandscienceinthesense
thatitrequiresboththeprocessesofabstractionbywhichweknowscienceand
theprocessesofconcretionbywhichweknowart.Thearchitectasscientistand
astheoristseekstoestablishthelawsofthespatialandformalmaterialswithwhich
thearchitectasartistthencomposes.Thegreaterscientifccontentofarchitecture
overartissimplyafunctionofthefargreatercomplexityoftherawmaterialsof
spaceandform,andtheirfargreaterreverberationsforotheraspectsoflife,thanany
materialsthatanartistuses.Itisthefactthatthearchitectdesignswiththespatial
stuffoflivingthatbuildsthescienceofarchitectureintotheartofarchitecture.
Itmayseemcurioustoarguethatthequestforascientifcunderstanding
ofarchitecturedoesnotleadtotheconclusionthatarchitectureisascience,but
neverthelessitisthecase.Inthelastanalysis,architecturaltheoryisamatter
ofunderstandingarchitectureasasystemofpossibilities,andhowtheseare
restrictedbylawswhichlinkthissystemofpossibilitiestothespatialpotentialities
ofhumanlife.Atthislevel,andperhapsonlyatthislevel,architectureisanalogous
tolanguage.Languageisoftennavelyconceptualisedasasetofwordsand
meanings,setoutinadictionary,andsyntacticrulesbywhichtheymaybe
combinedintomeaningfulsentences,setoutingrammars.Thisisnotwhat
languageis,andthelawsthatgovernlanguagearenotofthiskind.Thiscanbe
seenfromthesimplefactthatifwetakethewordsofthedictionaryandcombine
themingrammaticallycorrectsentences,virtuallyallareutterlymeaninglessand
donotcountaslegitimatesentences.Thestructuresoflanguagearethelaws
Introduction

Spaceisthemachine|BillHillier SpaceSyntax
Introduction
8
whichrestrictthecombinatorialpossibilitiesofwords,andthroughtheserestrictions
constructthesayableandthemeaningful.Thelawsoflanguagedonotthereforetell
uswhattosay,butprescribethestructureandlimitsofthesayable.Itiswithinthese
limitsthatweuselanguageastheprimemeanstoourindividualityandcreativity.
Inthissensearchitecturedoesresemblelanguage.Thelawsofthefeld
ofarchitecturedonottelldesignerswhattodo.Byrestrictingandstructuringthe
feldofcombinatorialpossibility,theyprescribethelimitswithinwhicharchitecture
ispossible.Aswithlanguage,whatisleftfromthisrestrictivestructuringisrich
beyondimagination.Evenso,withouttheselawsbuildingswouldnotbehuman
products,anymorethanmeaninglessbutsyntacticallycorrectconcatenationsof
wordsarehumansentences.
Thecaseforatheoreticalunderstandingofarchitecturethenrests
eventuallynotonaspirationtophilosophicalorscientifcstatus,butonthenature
ofarchitectureitself.Thefoundationalpropositionofthebookisthatarchitecture
isaninherentlytheoreticalsubject.Theveryactofbuildingraisesissuesaboutthe
relationsoftheformofthematerialworldandthewayinwhichweliveinitwhich
(asanyarchaeologistknowswhohastriedtopuzzleoutaculturefrommaterial
remains)areunavoidablybothphilosophicalandscientifc.Architectureisthe
mosteveryday,themostenveloping,thelargestandthemostculturallydetermined
humanartefact.Theactofbuildingimpliesthetransmissionofculturalconventions
answeringthesequestionsthroughcustomandhabit.Architectureistheirrendering
explicit,andtheirtransmutationintoarealmofinnovationand,atitsbest,ofart.Ina
sense,architectureisabstractthoughtappliedtobuilding,eventhereforeinasense
theoryappliedtobuilding.Thisiswhy,intheend,architecturemusthave
analytictheories.
Part one

Spaceisthemachine|BillHillier SpaceSyntax
Part one
Theoretical preliminaries
Defning architecture
Whatisarchitecture?Onethingisclear:ifthewordistoserveausefulpurposewe
mustbeabletodistinguisharchitecturefrombuilding.Sincebuildingisthemore
basicterm,itfollowsthatwemustsayinwhatsensearchitectureismorethan
building.Theessenceofourdefnitionmustsaywhatarchitectureaddstobuilding.
Thecommonestadditivetheoryisthatarchitectureaddsarttobuilding.In
thisanalysis,buildingisanessentiallypracticalandfunctionalactivityontowhich
architecturesuperimposesanartisticpreoccupationwhich,whilerespectingthe
practicalandfunctional,isrestrictedbyneither.Theextremeversionofthisview
isthatarchitectureistheadditiontobuildingofthepracticallyuselessand
functionallyunnecessary.
1
Themorecommonisthatbuildersmakebuildings
whilearchitectsaddstyle.
Fromthepointofviewoffndingwhatpeoplereallymeanwhentheysay
architecture,thereareseriousproblemswiththeseviews.Themostobviousisthat
itdefnesarchitectureintermsofwhatisnormallythoughtofasitsdegeneration,
thatis,thatarchitectureisnomorethantheadditionofasurfaceappearanceto
building.Evenifwetaketheviewthatthisiswhatarchitecturehasbecome,itis
surelyunacceptableasadefnitionofwhatitshouldbe.Architectsbelieve,and
clientsonthewholebuy,theideathatarchitectureisawayofbeingconcerned
withthewholebuilding,andameansofengagingthedeepestaspectsofwhat
abuildingis.Ifarchitectureisdefnedasanadd-onwhichignoresthemain
substanceofbuilding,thenarchitecturewouldbeanadditiontobuilding,butwould
notbemorethanbuilding.Onthecontrary,itwouldbeconsiderablyless.Ifwe
accusearchitectureofbeingnomorethanthis,weimplythatarchitectureoughtto
bemuchmore.Wearethereforebacktothebeginninginourpursuitofadefnition.
Anequallydiffcultproblemwiththisviewisthatitisveryhardtofnd
examplesofbuildingwithapurelypracticalandfunctionalaim.Whereverwefnd
building,wetendtofndapreoccupationwithstyleandexpression,however
modest.Someofthemoststrikinginstancesofthishavecomefromourgrowing
awarenessofbuildingbytechnologicallysimplesocieties,wherewedonotfnd
thatsimplicityoftechniqueisassociatedwithsimplicityofculturalintentorthe
eliminationofthepreoccupationwithstyle.Onthecontrary,wefndthatthroughthe
idiosyncrasiesofstyle,buildingandsettlementformbecomesoneoftheprimary
thoughmostpuzzlingandvariableexpressionsofculture.
2
Thetermthat
expressesthisdiscovery.architecturewithoutarchitectsconfrmstheexistenceof
architectureassomethingoverandabovebuilding,eventhoughatthesametimeit
affrmstheabsenceofarchitects.
3
Itistheawarenessoftheculturalrichnessofeverydaybuildingthatlead
RogerScruton,inhisThe Aesthetics of Architecturetotrytosolvethedefnition
problemforarchitecturebyarguingthatsinceallbuildingsharesapreoccupation
withtheaestheticandthemeaningful,allbuildingshouldbeseenasarchitecture.
4

Scrutonseekstoreintegratearchitecturewiththewholeofbuilding.Inhisview,all
thatweeverfndinarchitectureisfound,atleastinembryonicform,intheeveryday
The visual impression, the
image produced by differences
of light and colour, is primary in
our perception of a building. We
empirically reinterpret this image
into a conception of corporeality,
and this defnes the form of the
space withinOnce we have
reinterpreted the optical image into
a conception of space enclosed
by mass, we read its purpose from
its spatial form. We thus graspits
content, its meaning.
Paul Frankl
What architecture adds to building
Chapter one
Theoretical preliminaries

Spaceisthemachine|BillHillier SpaceSyntax
0
What architecture adds to building

Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
vernacularinwhichmostofusparticipatethroughoureverydaylives.Thus:Even
whenarchitectshaveadefniteaestheticpurpose,itmaynotbemorethanthe
desirethattheirworkshouldlookrightinjustthewaythattablesandchairs,the
layofplacesatatable,thefoldsinanapkin,anarrangementofbooks,maylook
righttoacasualobserver.Thisleadshimtoadefnition:Architectureisprimarily
avernacularart:itexistsfrstandforemostasaprocessofarrangementinwhich
everynormalman[sic]mayparticipate.
5
Thediffcultywiththisdefnitionisthatitleadstoexactlythewrongkindof
distinctionbetween,forexample,thecarefulformalandspatialrulesthatgoverned
theEnglishsuburbanhouseasbuiltendlesslyandrepetitiouslybetweenthewarsby
speculativebuilders,andtheworksof,say,PalladioorLeCorbusier.Theworkofboth
ofthesearchitectsischaracterisedbyradicalinnovationinexactlythoseareasof
formalandspatialorganisationwhereaccordingtoScrutonsdefnition,thereshould
beapreoccupationwithculturalcontinuityandreduplication.Itwouldseemtofollow
thatScrutonsdefnitionofarchitecturewouldcoverthefamiliarEnglishspecbuilders
vernacularmoreeasilythanitwouldtheworksofmajorarchitecturalinnovators.
Whileitmaybereasonable,then,toprefertheEnglishinter-warvernacular
totheworksofPalladioandLeCorbusier,itdoesnotseemlikelythatadefnition
oftheordinaryuseofthewordarchitectureliesinthisdirection.Onthecontrary,
Scrutonsdefnitionseemstoleadusexactlythewrongway.Architectureseems
tobeexactlynotthispreoccupationwithculturalcontinuity,butapreference
forinnovation.FarfromusingthisasabasisforadefnitionthenScrutons
preoccupationwiththevernacularseemstoaccomplishtheopposite.Ittellsus
morehowtodistinguisheverydaybuildingfromthemoreambitiousaspirations
ofwhatwecallarchitecture.
Is architecture a thing or an activity?
Inwhatdirectionshouldwelookthenforadefnitionofarchitectureasmore
thanbuilding?Refectingonthecommonmeaningsoftheword,wefndlittlehelp
andmorediffculties.Thewordarchitectureseemstomeanbothathingandan
activity.Ontheonehanditseemstoimplybuildingswithcertainarchitectural
attributesimposedonthem.Ontheother,itseemstodescribewhatarchitectsdo,
acertainwayofgoingabouttheprocessofmakingbuildings.Thisdoublemeaning
raisesseriousproblemsforadefnitionofarchitecture.Ifarchitecturemeansboth
attributesofthingsandattributesofactivities,thenwhichreallyisarchitecture?
Thedefnitionsurelycannotencompassboth.Propertiesofthingsseemtoexist
regardlessoftheactivitythatcreatesthem,andactivitiesarewhattheyare
regardlessoftheirproduct.Isarchitecture,then,essentiallyathingoranactivity?
Itmust,itseems,beoneortheother.
However,whenwetryeachdefnitioninisolationwequicklyruninto
paradoxes.Letusexperimentfrstwiththeideathatarchitectureisessentiallya
thing;thatis,certainattributesfoundinsome,butnotall,buildings.Ifthatiswhat
architectureessentiallyis,thenitwouldfollowthatacopyofabuildingwhich
What architecture adds to building
2
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
possessesthearchitecturalattributeswillalsobearchitecture,toexactlythe
samedegreeandinthesamewayastheoriginalbuilding.Butwebaulkatthis
idea.Copiesofarchitecturalbuildingsseemnotthemselvestobearchitecture,but
whatwehavenamedthemas,thatis,copiesofarchitecture.Certainlywewould
notnormallyexpecttowinanarchitecturalprizewithadeliberatecopy.Onthe
contrary,wewouldexpecttobedisqualifed,oratleastridiculed.
Whatthenismissinginthecopy?Bydefnition,itcannotbepropertiesofthe
buildingsincetheseareidenticalinbothcases.Thedisqualifyingfactormustliein
theactofcopying.Theactofcopyingsomehowmakesabuildingwitharchitectural
attributesnolonger,initself,architecture.Thismeansthatwhatismissinginthecopy
isnottodowiththebuildingbuttodowiththeprocessthatcreatedthebuilding.
Copyingisthereforeinsomecrucialsensenotarchitectural.Evenifwestartfromthe
propositionthatarchitectureisattributesofbuilding,andthereforeinsomesense,in
theobject,theproblemofthecopyshowsthatafterallarchitectureimpliesacertain
kindofactivity,onewhichismissingintheactofcopying.
Whatthenismissingintheactofcopying?Itcanonlybethatwhich
copyingdenies,thatis,theintentiontocreate,ratherthansimplytoreproduce,
architecture.Withoutthisintention,itseems,abuildingcannotbearchitecture.So
letuscallthisthecreativeintentionandtrytomakeitthefocusofadefnition
ofarchitecture.Wemayexperimentwiththeideaasbefore.Thistime,letthere
beanambitiousbuttalentlessarchitectwhointendsashardaspossibletomake
architecture.Istheproductofthisintentionautomaticallyarchitecture?Whetherit
isornotdependsonwhetheritispossibletoapprovetheintentionasarchitectural
butdisqualifytheresult.Infactthisisaverycommonformforarchitectural
judgmentstotake.Theproductsofaspiringarchitectsareoftenjudgedbytheir
peerstohavefailedinexactlythisway.Ajurymaylegitimatelysay:Weunderstand
yourintentionbutdonotthinkyouhavesucceeded.Howaresuchjudgments
made?Clearlythereisonlyoneanswer:byreferencetotheobjectiveattributes
oftheproposedbuildingsthatourwould-bearchitecthasdesigned.
Itseemsthenthenormaluseofwordsandcommonpracticehasledusin
acircle.Creativeintentionfailsasadefnitionofarchitecturebyreferencetopositive
attributesofthings,justaspositiveattributesofthingspreviouslyfailedbyreference
tointentions.Yetarchitectureseemsatthesametimetomeanboth.Itseemsit
canonlybethattheideaofarchitectureisatonceathingandanactivity,certain
attributesofbuildingsandacertainwayofarrivingatthem.Productandprocess
arenot,itseems,independent.Injudgingarchitecturewenoteboththeattributes
ofthethingandtheintellectualprocessbywhichthethingisarrivedat.
Thismayseematfrstsightratherodd.Itviolatesthecommonconception
thatattributesofthingsareindependentoftheprocessesthatputthemthere.But
itdoesrefecthowpeopletalkaboutarchitecture.Architecturaltalk,whetherbylay
peopleorbycritics,typicallymixescommentonproductwithcommentonprocess.
Forexample,wehear:Thisisaningenioussolutiontotheproblemof,orThisis
acleverdetail,orThisspatialorganisationisboldlyconceived,Ilikethewaythe
What architecture adds to building
3
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
architecthas,andsoon.Eachoftheseisatoneandthesametimeacomment
ontheobjectiveattributesofthebuildingandacommentonthecreativeintellectual
processthatgaverisetoit.Inspiteoftheunlikelihoodofproductandprocess
somehowbeinginterdependentintheideaofarchitecture,thisdoesseemtobe
exactlythecase.Indescribingourexperienceofarchitecturewedescribenot
onlytheattributesofthings,butalsotheintellectualprocessesofwhichthething
isamanifestation.Onlywiththesimultaneouspresenceofbothdowe
acknowledgearchitecture.
Thereis,itseems,someinconsistencybetweenournormalwayofreasoning
aboutthingsandthewaywetalk,reasonablyandreasoningly,aboutarchitecture.
Wemightevensaythattheideaofarchitectureexhibitssomeconfusionbetween
subjectsandobjects,sincethejudgmentthatabuildingisarchitectureseemsat
oneandthesametimetodependontheattributesoftheobjectivethingandon
attributesofthesubjectiveprocessthatgivesrisetothething.Itmightbereasonable
toexpect,then,thatfurtheranalysiswouldshowthatthisstrangenessintheideaof
architecturewaspathologicalandthat,withamorecarefuldefnition,productand
process,andobjectandsubject,couldandshouldbeseparated.
Infact,wewillfndthecontrary.Asweproceedwithourexplorationof
whatarchitectureisandwhatitaddstobuildingwewillfndthattheinseparability
ofproductsandprocessesandofsubjectandobjectsistheessenceofwhat
architectureis.Itisourintellectualexpectationsthatitshouldbeotherwisewhich
areatfault.Architectureisatonceproductandprocess,atonceattributeofthings
andattributeofactivity,sothatweactuallysee,orthinkwesee,bothwhenwesee
andnamearchitecture.
Howdoesthisapparentinterdependenceofproductandprocessthen
ariseasarchitecturefromtheactofmakingabuilding?Tounderstandthiswe
mustfrstknowwhatbuilding,theallegedlylesseractivity,is,andwemust
understanditbothasproductandasprocess.Onlythiswillallowustoseewhat
isdistinctiveaboutarchitecture,andhowthisdistinctivenessinvolvesbothproduct
andprocess.Toallowthistobecomefullyclear,theargumentthatfollowswillbe
takenintwostages.frstwewilllookatbuildingasaproduct,inordertoaskwhat
itisaboutthebuildingasproductthatarchitecturetakesholdofandadds
somethingto.Thenwewilllookatbuildingasaprocess,inordertoaskhow
theprocessofarchitecture,asaddingsomethingtobuilding,isdifferent.
So what is a building?
Thequestionwhatisabuilding?tendstoprovoketwokindsofsimplifcation.
Thefrstisthatbecausebuildingsarepurposefulobjectswecansaywhatthey
arebysayingwhattheirpurposeis.Thesecondisthattheremustbesomesimple
primordialpurposewhichwastheoriginalreasonforbuildingsandtherefore
constitutesakindofcontinuingessenceofbuilding.Thefrstsimplifcationisa
logicalerror,thesecondahistoricalone.Bothfndtheircommonest,butnotonly,
expressioninsuchideasasthatbuildingsareessentiallyshelter.

What architecture adds to building
4
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Bothsimplifcationsarisebecausepurposesareseentobeanteriortoobjectsand
thereforeinsomesenseexplanatoryofthem.Butlogically,functionaldefnitions
areabsurd.Indefningbuildingintermsofafunction,ratherthananobject,no
distinctionismadebetweenbuildingsasobjectsandotherentitieswhichalso
canordoprovidethatfunction,asforexampletrees,tents,cavesandparasols
alsoprovideshelter.Functionaldefnitionsarealsodishonest.Onewhodefnesa
buildingasashelterhasapictureofabuildinginmind,butonewhichisimplicit
ratherthanexplicit,sothattheimprecisionofthedefnitionisneverrevealedtothe
defner.Ifwesayabuildingisashelterwementallyseeabuildingandconceive
ofitfunctioningasashelter,sothatthefunctionseemstoexplaintheobject.
Functionaldefnitionsonlyappeartoworkbecausetheyconcealanimplicitideaof
theobject.Thispreventstheimprecisionofthedefnitionfrombeingapparenttothe
defner.Evenifthefunctionwerethoughttobeuniquetotheobject,thedefnition
ofanobjectthroughitsfunctionwouldneverbesatisfactorysincewecouldnever
besureeitherthatthisfunctionisnecessarilyuniquetothisobject,orthatthisis
theonlyessentialfunctionofthisobject.
Historicallyinfactalltheevidenceisthatneitheristhecase.Ifweconsider
thephenomenonofbuildingevenintheearliestandsimplestsocieties,oneof
themoststrikingthingsthatwefndisthatbuildingsarenormallymultifunctional:
theyprovideshelterfromtheelements,theyprovidesomekindofspatial
schemefororderingsocialrelationsandactivities,theyprovideaframework
forthearrangementofobjects,theyprovideadiversityofinternalandexternal
opportunitiesforaestheticandculturalexpression,andsoon.Ontheevidence
wehave,itisdiffculttofndhistoricaloranthropologicalgroundsforbelieving
thatbuildingsarenotintheirverynaturemultifunctional.
Noristhereanyreasonwhyweshouldexpectthemtobe.Inspiteofthe
persistenceoftheabsurdbeliefthathumankindlivedincavesuntilneolithictimes
(beginningabout1012,000yearsago),andthenusedthecaveasthemodelforthe
building,
6
thereisevidencethathumanbeingshavecreatedrecognisablebuildings
foraverylongtime,perhapsaslongasatleastthreehundredthousandyears.
7
We
donotknowhowtheantiquityofbuildingcompareswiththatoflanguage,butitis
clearthattheevolutionaryhistoryofeachisverylong,andthatconjecturalhistorical
ontologiesareequallyirrelevanttobothintryingtounderstandthecomplexnature
ofeitherassocialandculturalphenomenon.Thespeculationthatbuildingsare
somehowexplainedbybeingdefnedasshelters,becauseweimaginethatthere
musthavebeenatimewhenthiswasallthatbuildingwas,isaboutasuseful
inunderstandingthesocialandculturalcomplexitiesofbuildingastheideathat
languagebeganwithpointingandgruntingistotheoriesofthestructureand
functioningoflanguage.
Butitisnotonlytimethathasgivenbuildingstheirvarietyofcultural
expression.Thenatureofthebuildingasanobjectitselfhascomplexitieswhich
inthemselvesnaturallytendtomultifunctionalityanddiversityofculturalexpression.
Itisonlybyunderstandingthecomplexnatureofthebuildingasobjectthatwe
What architecture adds to building
5
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
canbegintounderstanditsnaturaltendencytomultifunctionality.Atthemost
elementarylevel,abuildingisaconstructionofphysicalelementsormaterialsinto
amoreorlessstableform,asaresultofwhichaspaceiscreatedwhichisdistinct
fromtheambientspace.Attheveryleastthen,abuildingisbothaphysicalanda
spatialtransformationofthesituationthatexistedbeforethebuildingwasbuilt.Each
aspectofthistransformation,thephysicalandthespatial,alreadyhas,asweshall
see,asocialvalue,andprovidesopportunityforthefurtherelaborationofthisvalue,
inthatthephysicalformofthebuildingmaybegivenfurtherculturalsignifcanceby
theshapinganddecorationofelements,andthespatialformmaybemademore
complex,byconceptualorphysicaldistinctions,toprovideaspatialpatterningof
activitiesandrelationships.
However,eveninthemostprimitive,unelaboratedstate,theeffectofthis
elementarytransformationofmaterialandspaceonhumanbeingsthatis,its
functionaleffectiscomplex.Part,butonlypart,ofthiscomplexityisthefunctional
effectthatthesheltertheoristshavenoted,namelythephysicaleffectthatbodies
areprotectedfromambientelementsthatintheabsenceofthebuildingmightbe
experiencedashostile.Theseelementsincludeinclementweatherconditions,
hostilespeciesorunwelcomeconspecifcs.Whenwesaythatabuildingisa
shelter,wemeanthatitisakindofprotectionforthebody.Tobeaprotective
shelterabuildingmustcreateaprotectedspacethroughastableconstruction.
Whatisprotectiveisthephysicalformofthebuilding.Whatisprotectedisthe
space.Buildingshaveabodilyfunction,broadandnon-specifc,butclassifable
asbodily,asaresultofwhichthebuildinghasspaceabletocontainbodies,and
certainphysicalpropertiesthroughwhichbodiesareprotected.
However,eventhesimplestbodilyactofmakingashelterismorecomplex
thanmightappearatfrstsight.Toencloseaspacebyaconstructioncreates
notonlyaphysicaldistinctiononthesurfaceoftheearth,butalsoalogical,or
categoricdistinction.Weacknowledgethisthroughtermslikeinsideandoutside.
Thesearerelationalnotionswithanessentiallylogicalnature,notsimplephysical
facts.Theyariseasakindoflogicalemergencefromthemoreelementaryphysical
factofmakingaboundary.Therelationalityoftheselogicalemergentscanbe
demonstratedbysimplypointingtotheinterdependenceofinsideandoutside.
Oneimpliestheother,andwecannotcreateaspaceinsidewithoutalsomaking
aspaceoutside.Logicalitycanbedemonstratedbydirectanalogy.Thephysical
processofdrawingaboundaryisanalogoustonamingacategory,sincewhenwe
dosowealsobyimplicationnameallthatisnotthatcategory,thatis,weimply
thecomplementofthatcategory,inthesamesensethatwhenwenamethespace
insidewealsoimplyallthespacethatisoutside.Inthatsensethespaceoutside
isthecomplementofthespaceinside.Logiciansconfrmthisanalogybydrawing
Venndiagrams,thatrepresentconceptsasallthatfallswithinthespaceofacircle,
anexactlyanalogouslogicalgesturetothecreationofaboundaryinrealspace.
AsRussellhaspointedout,
8
relations,especiallyspatialrelations,arevery
puzzlingentities.Theyseemtoexistobjectively,inthesense(tousetheexample
What architecture adds to building
6
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
givenbyRussell)thatEdinburghistothenorthofLondon,butwecannotpoint
directlytotherelationinthewaythatwecantootherentitieswhichseemtoreally
exist.Wemustaccept,Russellargues,thattherelation,likethetermsitrelates,
isnotdependentonthought,butbelongstotheindependentworldwhichthought
apprehends,butdoesnotcreate.Wemustthenaccept,hecontinues,thatarelation
isneitherinspacenorintime,neithermaterialnormental,yetitissomething.
Theobjectivityofrelations,andofthemorecomplexrelationalschemes
wecallconfgurations,willbeacontinuingthemeinthisbook.However,even
atthesimplestlevelofthecreationofaboundarybythesimplestactofbuilding,
mattersareyetmorecomplex.Thelogicaldistinctionsmadebydrawingboundaries
arealsosociologicaldistinctions,inthatthedistinctionbetweeninsideand
outsideismadebyasocialbeing,whosepowertomakethisdistinctionbecomes
recognisednotonlyinthephysicalmakingoftheboundaryandthecreationofthe
protectedspacebutalsointhelogicalconsequencesthatarisefromthatdistinction.
Thisisbestexpressedasaright.Thedrawingofaboundaryestablishesnotonlya
physicalseparateness,butalsothesocialseparatenessofadomaintheprotected
spaceidentifedwithanindividualorcollectivity,whichcreatesandclaimsspecial
rightsinthatdomain.Thelogicaldistinctionandthesociologicaldistinctioninthat
senseemergefromtheactofmakingasheltereveniftheyarenotintended.The
primaryactofbuilding,wemightsay,isalreadycomplexinthatminds,andeven
socialrelations,areengagedbybodilytransformations.
Asisthecasewiththelogicalcomplexity,thesociologicalcomplexity
impliedbytheboundaryisinitsverynaturerelational.Indeed,itisthelogicof
therelationalcomplexthatgivesrisetothesociologicaldistinctionsthroughwhich
buildingfrstbeginstorefectandinterveneinsocialrelations.Itisthisessential
relationalityofformandofspacewhichisappropriatedintheprocessesbywhich
buildingsaretransformedfrombodilyobjectstosocialandculturalobjects.The
fundamentalrelationalcomplexofformandspacecreatedbytheactofmakingthe
simplestbuiltobjectistheseedofallfuturerelationalpropertiesofspacesthrough
whichbuildingsbecomefullysocialobjects.
Abuildingthenbecomessociallysignifcantoverandaboveitsbodily
functionsintwoways:frstbyelaboratingspacesintosociallyworkablepatterns
togenerateandconstrainsomesociallysanctionedandthereforenormative
patternofencounterandavoidance;andsecondbyelaboratingphysicalforms
andsurfacesintopatternsthroughwhichculturallyoraestheticallysanctioned
identitiesareexpressed.Thefundamentaldualityofformandspacethatwenoted
inthemostelementaryformsofthebuildingthuscontinuesintoitscomplexforms.
Bytheelaborationofspace,asocialdomainisconstitutedasalivedmilieu.By
theelaborationofformasocialdomainisrepresentedassignifcantidentitiesand
encounters.Inbothsenses,buildingscreatemorecomplexpatternsfromthebasic
bodilystuffofformandspace.Itisthroughthesepatternsthatbuildingsacquire
theirpotentialatoncetoconstituteandrepresentandthusintimetoappearas
theveryfoundationofoursocialandculturalexistence.
What architecture adds to building
7
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Wemaysummarisewhatwehavesaidaboutthenatureofbuildingsas
objectsinadiagramwhichwewillusefromnowonasakindoffundamental
diagramofthebuildingasobject,(seefg.1.1).Theessenceofthediagramisthata
buildingevenatthemostbasiclevelembodiestwodualities,onebetweenphysical
formandspatialformandtheotherbetweenbodilyfunctionandsocio-cultural
function.Thelinkbetweenthetwoisthatthesocio-culturalfunctionarisesfromthe
waysinwhichformsandspacesareelaboratedintopatterns,or,aswewillindue
coursedescribethem,intoconfgurations.Wemustnowlookmorecarefullyatwhat
wemeanbytheelaborationofformandspaceintoconfguration,sincethiswill
bethekeytoourargumentnotonlyaboutthenatureofbuildings,butalso,indue
course,tohowarchitecturearisesfrombuilding.
Letusbeginwithasimpleandfamiliarcaseoftheelaborationofthe
physicalformofthebuilding:thedoriccolumn.Whenwelookatadoriccolumn,
weseeaplinth,apedestal,ashaft,acapital,andsoon,thatis,weseea
construction.Theelementsrestoneupontheother,andtheirrelationtoeachother
takesadvantageofanddependsonthenaturallawofgravity.Butthisisnotallthat
wesee.Therelationsoftheelementsofacolumngovernedbythelawofgravity
wouldholdregardlessofthedoricnessoftheelements.If,forexample,wewereto
replacethedoriccapitalwithanioniancapital,theeffectontheconstructionwould
benegligible,buttheeffectonthedoricnessoftheensemblewouldbedevastating.
Sowhatisdoricness?Clearlyitisnotatypeofconstruction,sincewemay
substitutenon-doricelementsintheensemblewithoutconstructionalpenalty.We
mustacknowledgethatdoricnessisnottheninitselfasetofphysicalrelations,
althoughitdependsonthem.Doricnessisaschemeinwhichelementswith
certainkindsofelaborationareaboveandbelowothersinacertainrelational
sequencewhichemergesfromconstructionbutisnotgivenbyconstruction.On
thecontrary,thenotionofaboveandbelowaswefndthemindoricnessseem
tobelogicalemergentsfromtheactofconstructioninexactlythesamesense
thatinsideandoutsidewerelogicalemergentsfromthephysicalconstruction
ofaboundary.Doricnessisthenalogicalconstruction,onebuiltonthebackofa
physicalconstructionbutalogicalconstructionnonetheless.Throughthelogical

Figure 1.1
What architecture adds to building
8
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
doricnessoftheensemble,wemaysaythatwemovefromthesimplevisualityof
thephysicallyinterdependentsystem,toentertherealmoftheintelligible.Doricness
isaconfgurationofpropertiesthatweunderstand,overandabovewhatwe
seeasphysicalinterdependencies,aformofrelationalelaborationtosomething
whichexistsinphysicalform,butwhichthroughthiselaborationstandsclearofits
physicality.Thisprocessofmovingfromthevisibletotheintelligibleis,wewillsee
induecourse,verybasictoourexperiencebothofbuildingandofarchitecture,
and,evenmoreso,tothedifferencebetweenoneandtheother.
Spatialpatternsinbuildingsalsoariseaselaborationsonprimitivelogical
emergentsfromthephysicalactofbuilding.Aswithdoricness,theydependonbut
cannotbeexplainedbynaturallaw(asmanyhavetriedtodobyappealtobiological
imperativessuchasterritoriality).Theoriginsofrelationalschemesofspacelie
somewherebetweentheorderingcapacitiesofthemindandthespatialordering
inherentinthewaysinwhichsocialrelationshipsarerealisedinspace.Withspace,
aswithform,wethereforefndasplitinbuildingbetweenabodilynature,albeit
witharudimentaryrelationalnature,andamoreelaboratedconfgurationalnature
whichrelatestomindsandsocialexperienceratherthantobodiesandindividual
experience.Thepassagefromthesimplespacetoaconfgurationofspaceisalso
thepassagefromthevisibletotheintelligible.
Spaceis,however,amoreinherentlydiffculttopicthanphysicalform,for
tworeasons.First,spaceisvacancyratherthanthing,soevenitsbodilynatureis
notobvious,andcannotbetakenforgrantedinthewaythatwethinkwecantake
objectsforgranted.(SeeChapter10forafurtherdiscussionofthisassumption.)
Second,relatedspaces,almostbydefnition,cannotbeseenallatonce,butrequire
movementfromonetoothertoexperiencethewhole.Thisistosaythatrelationality
inspaceisrarelyaccessibletousasasingleexperience.Wemusttherefore
digressforamomenttotalkaboutspaceasaphenomenon,andhowwecan
overcomethediffcultiesthatexistintalkingaboutit.Wewilltakethisintwostages.
First,wewilltalkabouttheproblemofhowfarspacecanbeseenasanobjective,
independentthing-in-itself.Wemustdothisbecausethereisgreatconfusionabout
thestatusofspaceandhowfaritcanberegardedasanindependententityrather
thansimplyasaby-productof,say,thearrangementofphysicalthings.Second,we
willtalkaboutspaceasconfguration,sinceitisasconfgurationthatithasitsmost
powerfulandindependenteffectsonthewaybuildingsandbuiltenvironmentsare
formedandhowtheyfunctionfortheirpurposes.

What architecture adds to building

9
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
About space
Itisfarfromobviousthatspaceis,insomeimportantsense,anobjectiveproperty
ofbuildings,describableindependentlyofthebuildingasaphysicalthing.Most
ofourcommonnotionsofspacedonotdealwithspaceasanentityinitselfbut
tieitinsomewaytoentitiesthatarenotspace.Forexample,evenamongstthose
withainterestinthefeld,theideaofspacewillusuallybetranscribedastheuse
ofspace,theperceptionofspace,theproductionofspaceorasconceptsof
space.Inallthesecommonexpressions,theideaofspaceisgivensignifcance
bylinkingitdirectlytohumanbehaviourorintentionality.Commonspatialconcepts
fromthesocialsciencessuchaspersonalspaceandhumanterritorialityalsotie
spacetothehumanagent,anddonotacknowledgeitsexistenceindependentlyof
thehumanagent.Inarchitecture,whereconceptsofspacearesometimesunlinked
fromdirecthumanagency,throughnotionssuchasspatialhierarchyandspatial
scalewestillfndthatspaceisrarelydescribedinafullyindependentway.The
conceptofspatialenclosureforexample,whichdescribesspacebyreferenceto
thephysicalformsthatdefneitratherthanasathinginitself,isthecommonest
architecturalwayofdescribingspace.
Alltheseconceptsconfrmthediffcultyofconceptualisingspaceasa
thinginitself.Onoccasion,thisdiffcultyfndsanextremeexpression.Forexample,
RogerScrutonbelievesthattheideaofspaceisacategorymistakemadeby
pretentiousarchitects,whohavefailedtounderstandthatspaceisnotathingin
itself,butmerelytheobversesideofthephysicalobject,thevacancyleftoverby
thebuilding.ForScruton,itisself-evidentthatspaceinafeldandinacathedral
arethesamethingexceptinsofarastheinteriorbuiltsurfacesofthecathedral
makeitappearthattheinteriorspacehasdistinctivepropertiesofitsown.Alltalk
aboutspaceiserror,heargues,becauseitcanbereducedtotalkaboutbuildings
asphysicalthings.
9
Infact,evenatapracticallevel,thisisabizarreview.Spaceis,quitesimply,
whatweuseinbuildings.Itisalsowhatwesell.Nodeveloperofferstorentwalls.
Wallsmakethespace,andcostthemoney,butspaceistherentablecommodity.Why
thenisScrutonembarrassedbytheconceptofspace?LetmesuggestthatScruton
ismakinganeducatederror,onethathewouldnothavemadeifhehadnotbeenso
deeplyimbuedwiththewesternphilosophicaltraditioninwhichhehasearnedhis
livingandtowhich,incidentally,hehaswrittenanoutstandingintroduction.
10
Thedominantviewofspaceinwesternculturehasbeenonewemight
looselycalltheGalilean-Cartesian.Thisviewarisesfromaschemeofreasoning
frstsetoutinfullclaritybyDescartes.
11
Theprimarypropertiesofphysicalobjects
are,heargued,theirextension,thatis,theirmeasurablepropertieslikelength,
breadthandwidth.Becauseextensioncanbequantifedbymeasuringdevices
whichdonotdependonhumanagency,extensionscanbeseenastheindubitably
objectivepropertiesofthings,unlikesecondarypropertieslikegreenornice
whichseemtodependinsomewayoninteractionwithobservers.
Nowifextensionistheprimarypropertyofobjects,thenitisashortstep
What architecture adds to building
20
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
toseeitalsoastheprimarypropertyofthespacewithinwhichobjectssit.As
Descartessays:Afterexaminationweshallfndthatthereisnothingremainingin
theideaofbodyexceptingthatitisextendedinlength,breadthanddepth;andthis
iscomprisedinourideaofspace,notonlyofthatwhichisfullofbody,butalsothat
whichiscalledavacuum.
12
Inotherwords,whenwetaketheobjectawayfrom
itsspaceitsextensionisstillpresentasanattributeofspace.Spaceistherefore
generalisedextension,orextensionwithoutobjects.Descartesagain:Inspace
weattributetoextensionagenericunity,sothatafterhavingremovedfromacertain
spacethebodywhichoccupiedit,wedonotsupposewehavealsoremovedthe
extensionofthatspace.
13
Followingthisreasoning,spacecomestobeseenasthegeneralabstract
frameworkofextensionagainstwhichthepropertiesofobjectsaredefned,a
metricbackgroundtothematerialobjectsthatoccupyspace.Thisviewofspace
seemstomostofusquitenatural,nomorethananextrapolationofcommonsense.
Unfortunately,onceweseespaceinthisway,wearedoomednottounderstand
howitplaysaroleinhumanaffairs.Culturallyandsocially,spaceisneversimply
theinertbackgroundofourmaterialexistence.Itisakeyaspectofhowsocieties
andculturesareconstitutedintherealworld,and,throughthisconstitution,
structuredforusasobjectiverealities.Spaceismorethananeutralframeworkfor
socialandculturalforms.Itisbuiltintothoseveryforms.Humanbehaviourdoesnot
simplyhappeninspace.Ithasitsownspatialforms.Encountering,congregating,
avoiding,interacting,dwelling,teaching,eating,conferringarenotjustactivitiesthat
happeninspace.Inthemselvestheyconstitutespatialpatterns.
Itisbecausethisissothatspatialorganisationthroughbuildingsandbuilt
environmentsbecomesoneoftheprinciplewaysinwhichcultureismadereal
forusinthematerialworld,anditisbecausethisissothatbuildingscan,and
normallydo,carrysocialideaswithintheirspatialforms.Tosaythisdoesnotimply
determinismbetweenspacetosociety,simplythatspaceisalwayslikelytobe
structuredinthespatialimageofasocialprocessofsomekind.Thequestionis:
howexactlydoesthishappen,andwhatarethesestructureslike?
Space as confguration
Onethingisclear.Encountering,congregating,avoiding,interacting,dwelling,
conferringarenotattributesofindividuals,butpatterns,orconfgurations,formed
bygroupsorcollectionsofpeople.Theydependonanengineeredpatternofco-
presence,andindeedco-absence.Veryfewofthepurposesforwhichwebuild
buildingsandenvironmentsarenotpeopleconfgurationsinthissense.Weshould
thereforeinprincipleexpectthattherelationbetweenpeopleandspace,ifthereis
one,willbefoundattheleveloftheconfgurationofspaceratherthantheindividual
space.Thisisconfrmedbycommonsense.Individualspacesplacelittlelimiton
humanactivity,exceptforthoseofsizeandperhapsshape.Inmostreasonable
spaces,mosthumanactivitiescanbecarriedout.Buttherelationbetweenspaceand
socialexistencedoesnotlieattheleveloftheindividualspace,orindividualactivity.
What architecture adds to building
2
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure1.2
0
1
2
3
4
5
6
7
8
depth
0
1
2
3
4
depth
c.
b.
a.
d.
Figure 1.2
What architecture adds to building
22
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Itliesintherelationsbetweenconfgurationsofpeopleandconfgurationsofspace.
Totakethefrststepstowardsunderstandinghowthishappens,we
mustunderstandhow,inprinciple,aconfgurationofspacecanbeinfuenced
by,orinfuence,aconfgurationofpeople.Letusthereforeconsidersomesimple
hypotheticalexamples.Thetwonotionalcourtyardbuildingsoffgure1.2aandb
showinthefrstcolumninblack,inthenormalway,thepatternofphysicalelements
ofthebuildings.Thecorrespondingfguresinthesecondcolumnthenshowinblack
thecorrespondingpatternofspatialelements.Thebasicphysicalstructuresand
celldivisionsofthetwobuildingsarethesame,andeachhasthesamepatternof
adjacenciesbetweencellsandthesamenumberofinternalandexternalopenings.
Allthatdiffersisthelocationofcellentrances.Butthisisenoughtoensurethatfrom
thepointofviewofhowacollectionofindividualscouldusethespace,thespatial
patterns,orconfgurations,areaboutasdifferentastheycouldbe.Thepatternof
permeabilitycreatedbythedispositionofentrancesisthecriticalthing.Seenthisway,
onelayoutisanearperfectsinglesequence,withaminimalbranchattheend.The
otherisbranchedeverywhereaboutthestrongcentralspaces.
Nowthepatternofpermeabilitywouldmakerelativelylittledifferencetothe
buildingstructurallyorclimatically,thatis,tothebodilyaspectofbuildings,especially
ifweassumesimilarpatternsofexternalfenestration,andinsertwindowswherever
theotherhadentrancesontothecourtyard.Butitwouldmakeadramaticdifference
tohowthelayoutwouldworkas,say,adomesticinterior.Forexample,itisvery
diffcultformorethanonepersontouseasinglesequenceofspaces.Itoffers
littleinthewayofcommunityorprivacy,butmuchinthewayofpotentialintrusion.
Thebranchedpattern,ontheotherhand,offersadefnitesetofpotentialrelations
betweencommunityandprivacy,andmanymoreresourcesagainstintrusion.
Thesedifferencesareinherentinthespacepatterns,andwouldapplyto
wholeclassesofhumanactivitypatterns.Inthemselvesthespatiallayoutsoffera
rangeoflimitationsandpotentialities.Theysuggestthepossibilitythatarchitectural
spacemightbesubjecttolimitinglaws,notofadeterministickind,butsuchasto
setmorphologicalboundswithinwhichtherelationsbetweenformandfunctionin
buildingsareworkedout.
WewillseefromChapter3onwardsthatitisbyexpressingthese
patternpropertiesinanumericalwaythatwecanfndclearrelationsbetween
spacepatternsandhowcollectionsofpeopleusethem.However,beforewe
embarkonnumbers,thereisavisuallyusefulwayofcapturingsomeofthekey
differencesbetweenthetwospatialpatterns.Thisisadevicewecallajustifed
graph,orj-graph.Inthisweimaginethatweareinaspacewhichwecalltheroot
orbaseofthegraph,andrepresentthisasacirclewithacrossinscribed.Then,
representingspacesascircles,andrelationsofaccessaslinesconnectingthem,
wealignimmediatelyabovetherootallspaceswhicharedirectlyconnectedtothe
root,anddrawintheconnections.Thesearethespacesatdepthonefromthe
root.Thenanequaldistanceabovethedepthonerowwealignthespacesthat
connectdirectlytofrstrowspaces,formingthelineofdepthtwospaces,and
What architecture adds to building
23
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
connectthesetothedepthonespaces,andsoon.Sometimeswewillhaveto
drawratherlongandcircuitouslinestolinkspacesatdifferentlevels,butthisdoes
notmatter.Itisthefactofconnectionthatmatters.Thelawsofgraphsguarantee
thatifthelayoutisallatonelevelthenwecanmakealltherequiredconnections
bydrawinglinesconnectingthespaceswithoutcrossingotherlines.
14
Theresultingj-graphisapictureofthedepthofallspacesinapattern
fromaparticularpointinit.Thethirdcolumninfgure1.2aandbshowsj-graphs
forthecorrespondingspatialstructures,drawnusingtheexteriorspaceasroot.We
canimmediatelyseethatthefrstisadeeptreeform,andthesecondashallow
treeform.Bytreewemeanthatthereisonelinklessthanthenumberofcells
linked,andthattherearethereforenoringsofcirculationinthegraph.Alltrees,
eventwoasdifferentasinthetwointhefgures,sharethecharacteristicthatthere
isonlyoneroutefromeachspacetoeachotherspaceapropertythatishighly
relevanttohowbuildinglayoutsfunction.However,whereringsarefound,the
justifedgraphmakesthemasclearasthedepthproperties,showingthemina
verysimpleandclearwayaswhattheyare,thatis,alternativeroutechoicesfrom
onepartofthepatterntoanother.Theseriesoffguresinfgure1.2cshows
ahypotheticalcase,basedonthesamebasicbuildingasthepreviousfgures.
Wedonothavetojustifythegraphusingtheoutsidespaceasroot.This
isonlyonewaythoughasingularlyusefulwayoflookingatabuilding.We
canofcoursejustifythegraphfromanyspacewithinit,andthiswilltelluswhat
layoutislikefromthepointofviewofthatspace,takingintoaccountbothdepth
andringproperties.Whenwedothiswediscoverafactaboutthespatiallayouts
ofbuildingsandsettlementsthatissofundamentalthatitisprobablyinitselfthe
keytomostaspectsofhumanspatialorganisation.Thisisthesimplefactthata
patternofspacenotonlylooksdifferentbutactuallyisdifferentwhenjustifedfrom
thepointofviewofitsdifferentconstituentelements.Thethreenotionalj-graphs
showninfgure1.2dappearverydifferentfromeachother,butallthreeareinfact
thesamegraphjustifedfromthepointofviewofdifferentconstituentspaces.The
depthandringpropertiescouldhardlyappearmoredifferentiftheyweredifferent
confgurations.Itisthroughthecreationanddistributionofsuchdifferencesthat
spacebecomessuchapowerfulrawmaterialforthetransmissionofculturethrough
buildingsandsettlementforms,andalsoapotentmeansofarchitecturaldiscovery
andcreation.Letusseehow.
Formally defning confguration
Firstweneedtobringalittlemoreformalityintothedefnitionofconfguration.
Likethewordpattern(whichwedonotusebecauseitimpliesmoreregularity
thanwewillfndinmostspatialarrangements),confgurationseemstobea
conceptaddressedtothewholeofacomplexratherthantoitsparts.Intuitively,
itseemstomeanasetofrelationshipsamongthingsallofwhichinterdependin
anoverallstructureofsomekind.Thereisawayofformalisingthisideathatis
assimpleasitisnecessary.Ifwedefnespatialrelationsasexistingwhenthere
isanytypeoflinksayadjacencyorpermeabilitybetweentwospaces,then
What architecture adds to building
24
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
confgurationexistswhenrelationsbetweentwospacesarechangedaccording
tohowwerelateoneorother,orboth,toatleastoneotherspace.
Thisratheroddsoundingdefnitioncanbeexplainedthroughasimple
graphicexample.Figure1.3ashowsacelldividedbyapartitionintotwo,sub-cella
andsub-cellb,withadoorcreatingarelationofpermeabilitybetweenthetwo.Itis
clearthattherelationisformallysymmetricalinthesensethatcellaistocellbas
bistoa.Thesamewouldbetrueoftwocellswhichwereadjacentandthereforein
therelationofneighbourtoeachother.Ifaisbsneighbour,thenbmustalsobeas
neighbour.Thissymmetry,whichfollowsthealgebraicratherthanthegeometrical
defnition,isclearlyanobjectivepropertyoftherelationofaandbanddoesnot
dependonhowwechoosetoseetherelation.
Nowconsiderfgures1.3bandcinwhichwehaveaddedrelationstoa
thirdspace,c(whichisinfacttheoutsidespace),butinadifferentwaysothat
in1.3bbothaandbaredirectlypermeabletoc,whereasin1.3c,onlyaisdirectly
permeabletoc.Thismeansthatin1.3cwemustpassthroughatogettobfrom
c,whereasin1.3bwecangoeitherway.In1.3ctherefore,aandbaredifferent
withrespecttoc.Wemustpassthroughatogettobfromc,butwedonotneed
topassthroughbtogettoafromc.Withrespecttoc,therelationhasbecome
asymmetrical.Inotherwords,therelationbetweenaandbhasbeenredefned
bytherelationeachhastoathirdspace.Thisisaconfgurationaldifference.
Confgurationisasetofinterdependentrelationsinwhicheachisdetermined
byitsrelationtoalltheothers.
Wecanshowsuchconfgurationaldifferencesratherneatly,andclarifytheir
nature,byusingthej-graph,asinfgure1.3dande,correspondingto1.3band1.3c
respectively.Comparedto1.3a,spacesbandcin1.3ehaveacquireddepthwith
a b
c
Figure 1.3
c
a b
a
b
c
a
b
c
a b
c. b. a.
e. d.
2 2
2
3
2
3
Figure 1.3
What architecture adds to building
25
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
respecttoeachother,inthattheirrelationisnowindirectandonlyexistsbyvirtue
ofa.Thenumbersadjacenttoeachspaceinthej-graphindexthisbyshowingthe
totaldepthofeachspacefromtheothertwo.Incontrast,1.3dhasacquiredaring
thatlinksallthreespaces,meaningthateachhasachoiceofroutetoeachofthe
others.Thegraphof1.3disidenticalwhenseenfromeachofitsspaces,whilein
1.3e,bandcareidentical,butaisdifferent.
Society in the form of the object
Nowletususethisconceptofconfguration,anditskeyspatialdimensionsof
depthandrings,totrytodetectthepresenceofculturalandsocialideasinthe
spatialformsofbuildings.Figures1.4a,bandcshow,ontheleft,theground-
foorplansofthreeFrenchhouses,andtotheirimmediateright,theirj-graphs
drawninitiallyfromtheoutside,treatingitasasinglespace,thentotherightagain
threefurtherj-graphsjustifedfromthreedifferentinternalspaces.
15
Lookingat
thej-graphsdrawnfromtheoutside,wecanseethatinspiteofthegeometrical
differencesinthehousestherearestrongsimilaritiesintheconfgurations.Thiscan
beseenmosteasilybyconcentratingonthespacemarkedsc,orsallecommune,
whichisthemaineverydaylivingspace,inwhichcookingalsooccursandeveryday
visitorsarereceived.Ineachcase,wecanseethatthesallecommuneliesonall
non-trivialrings(atrivialringisonewhichlinksthesamepairofspacestwice),links
directlytoanexteriorspacethatis,itisatdepthoneinthecomplexandacts
asalinkbetweenthelivingspacesandvariousspacesassociatedwithdomestic
workcarriedoutbywomen.
Thesallecommunealsohasamorefundamentalproperty,onewhich
arisesfromitsrelationtothespatialconfgurationofthehouseasawhole.Ifwe
countthenumberofspaceswemustpassthroughtogofromthesallecommune
toallotherspaces,wefndthatitcomestoatotalwhichislessthanforanyother
spacethatis,ithaslessdepththananyotherspaceinthecomplex.Thegeneral
formofthismeasure
16
iscalledintegration,andcanbeappliedtoanyspaceinany
confguration:thelessdepthfromthecomplexasawhole,themoreintegratingthe
space,andviceversa.Thismeansthateveryspaceinthethreecomplexescanbe
assignedanintegrationvalue.
17
Nowoncewehavedonethiswecanaskquestionsabouthowthedifferent
functionsinthehousearespatialised,thatis,howtheyareembeddedintheoverall
spatialconfguration.Whenwedothis,wefndthatitisverycommonthatdifferent
functionsarespatialisedindifferentways,andthatthiscanoftenbeexpressed
clearlythroughintegrationanalysis.InthethreeFrenchhouses,forexample,we
fndthatthereisacertainorderofintegrationamongthespaceswheredifferent
functionsarecarriedout,alwayswiththesallecommuneasthemostintegrated,as
canbeseeninthej-graphsbesideeachplan.Ifallthefunctionsofthethreehouses
aresetoutinorderoftheintegrationvaluesofthespacesinwhichtheyoccur,
beginningwiththemostintegratedspace,wecanreadthis,fromlefttoright,as:
thesallecommuneismoreintegrated(i.e.haslessdepthtoallotherspaces)than
What architecture adds to building
26
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 1.4
What architecture adds to building
27
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
thecorridor,whichismoreintegratedthantheexterior,andsoon.Totheextentthat
therearecommonalitiesinthesequenceofinequalities,thenwecansaythatthere
isacommonpatterntothewayinwhichdifferentfunctionsarespatialisedinthe
house.Wecallsuchcommonpatternsinequalitygenotypes,becausetheyrefer
nottothesurfaceappearancesofformsbuttodeepstructuresunderlyingspatial
confgurationsandtheirrelationtolivingpatterns.
18
Theseresultsfowfromananalysisofspace-to-spacepermeability.Butwhat
abouttherelationofvisibility,whichpassesthroughspaces?Thethreerowsof
fguresontherightinfgure1.4(lowerpanel)showallthespacethatcanbeseen
withthedoorsopenfromadiamond-shapedspacewithineachsallecommuneand
oneotherspace,drawnbyjoiningthecentrepointsofeachwallofaroom,andthus
coveringhalfofthespaceintheroom.Theideaofthediamondshapeisthatspace
use(inmostwesterncultures)isnormallyconcentratedwithinthisdiamondshape,
thecornerscommonlybeingreservedforobjects.Thediagramsshowthatineach
casethesallecommunehasafarmorepowerfulvisualfeldthanthesalle.Inother
words,thespatialandfunctionaldifferencesbetweenspacesthatwefndthroughthe
analysisofpermeabilityinthehousesalsoappearintheanalysisofvisibility.These
visibilitydifferencescanalsoformthebasisforquantitativeandstatisticalanalysis.
Thistypeofmethodallowsustoretrievefromhouseplansconfgurational
propertiesthatrelatedirectlytothesocialandculturalfunctioningofthehouse.
Inotherwords,throughspatialconfgurationculturallydeterminedpatternsare
embeddedinthematerialandspatialobjectivityofbuildings.Bytheanalysisof
spacesandfunctionsintermsoftheirconfgurationalrelationswithinthehouse,
andthesearchforcommonpatternsacrosssamples,wecanseehowbuildings
cantransmitcommonculturaltendenciesthroughspatialform.Wemustnowask
howandwhythisisthecase,andwhatfollowsfromit?
The non-discursivity of confguration: ideas we think of and ideas we think with
Theanswerwilltakeustothecentreofourargument:thenon-discursivityof
confguration.Non-discursivitymeansthatwedonotknowhowtotalkaboutit.The
diffcultyoftalkingaboutspatialorformalconfgurationsinarchitecturehasalways
seemedaratherperipheralproblemtoarchitecturaltheory.Isuggestitisthecentral
problem,andpartofamuchmoregeneralprobleminhumanaffairs.
Letusbegintoexploretheintuitiveaspectsoftheideaofconfguration
alittlefurther.Considerthefourgroupsofelementsinfgure1.5.Eachgroupisa
differentsetofthings,butplacedinmoreorlessthesameoverallconfguration.
Thehumanmindhasnodiffcultyinseeingthattheconfgurationsarethesame,
inspiteofthedifferencesintheconstituentthings,andthisshowsthatweeasily
recogniseaconfguration,evenwherewehavenowayofgivingitanameandthus
assigningittoacategoryalthoughwemighttrytodosobymakinganalogies
withconfgurationsforwhichnamesarealreadyathand,suchasL-shaped,or
star-shaped.However,thefactthatourmindsrecognisedconfgurationsasbeing
thesameevenwhenthereisnonameathandtolinkthemshowsthatourability
What architecture adds to building
28
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
torecogniseandunderstandconfgurationispriortotheassignmentofnames.
Confgurationseemsinfacttobewhatthehumanmindisgoodatintuitively,
butbadatanalytically.Weeasilyrecogniseconfgurationwithoutconsciousthought,
andjustaseasilyuseconfgurationsineverydaylifewithoutthinkingofthem,but
wedonotknowwhatitiswerecogniseandwearenotconsciousofwhatitiswe
useandhowweuseit.Wehavenolanguagefordescribingconfgurations,thatis,
wehavenomeansofsayingwhatitisweknow.Thisproblemisparticularlysalient
inbuildingsandarchitecture,becausebothhavetheeffectofimposingspatial
andformalconfgurationontheworldinwhichwelive.Buttheproblemisnot
confnedtoarchitecture.Onthecontraryitappearstobepresenttosomedegree
inmostculturalandsocialbehaviours.Inusinglanguage,forexample,weare
awareofwordsandbelievethatinspeakingandhearingwearehandlingwords.
However,languageonlyworksbecauseweareabletousetheconfgurational
aspectsoflanguage,thatis,thesyntacticandsemanticruleswhichgovernhow
wordsaretobeassembledintomeaningfulcomplexes,inawaywhichmakestheir
operationautomaticandunconscious.Inlanguagewecanthereforedistinguish
ideaswethinkof,thatis,thewordsandwhattheyrepresent,andideaswethink
with,thatis,syntacticandsemanticruleswhichgovernhowwedeploywords
tocreatemeaning.Thewordswethinkofseemtouslikethings,andareatthe
levelofconsciousthought.Thehiddenstructureswethinkwithhavethenature
ofconfgurationalrules,inthattheytellushowthingsaretobeassembled,and
workbelowthelevelofconsciousness.Thisunconsciousconfgurationalityseems
toprevailinallareaswhereweuserulesystemstobehaveinwayswhichare
recognisableassocial.Behaviourattable,ortheplayingofgames,appeartous
asspatio-temporalevents,buttheyaregivenorderandpurposebytheunderlying
Figure 1.5
What architecture adds to building
29
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
confgurationalideas-to-think-withthroughwhichtheseeventsaregenerated.We
acknowledgetheimportanceofthisunseenconfgurationalitylabellingitasaform
ofknowledge.Wetalkaboutknowinghowtobehave,orknowingalanguage.
Wecancallthiskindofknowledgesocialknowledge,andnotethatits
purposeistocreate,orderandmakeintelligiblethespatio-temporaleventsthrough
whichwerecognisethepresenceofcultureineverydaylife.Wemustofcourse
takecaretodistinguishsocialknowledgefromformsofknowledgewhichwelearn
inschoolsanduniversitieswhosepurposeistounderstandtheworldratherthan
tobehaveinit,andwhichwemightthereforecallanalytic,orscientifcknowledge.
Initself(thoughnotnecessarilyinitsconsequences)analyticknowledgeleavesthe
worldasitis,sinceitspurposeistounderstand.Analyticknowledgeisknowledge
wherewelearntheabstractprinciplesthroughwhichspatio-temporalphenomena
arerelatedwemightsaytheconfgurationalityconsciously.Weareaware
oftheprinciplesbothwhenweacquireandwhenweusetheknowledge.Asa
result,throughtheintermediaryoftheabstract,wegrasptheconcrete.Insocial
knowledge,incontrast,knowledgeofabstractconfgurationalityisacquiredthrough
theprocessofcreatingandexperiencingspatio-temporalevents.Socialknowledge
workspreciselybecausetheabstractprinciplesthroughwhichspatio-temporal
phenomenaarebroughttogetherintomeaningfulpatternsareburiedbeneath
habitsofdoing,andneverneedbebroughttoconsciousattention.
19
Inspiteofthesefunctionaldifferences,socialknowledgeandanalytic
knowledgearemadeupofthesameelements:ontheonehand,thereis
knowledgeofspatio-temporalphenomena,ontheother,thereareabstract
confgurationalstructuresthatlinkthemtogether.Butwhereasinsocialknowledge
theabstractideasareheldsteadyasideastothinkwithinordertocreatespatio-
temporaleventsintherealworld,sothattheabstractideasbecomethenormative
basesofbehaviour,inscientifcknowledge,anattemptismadetoholdspatio-
temporalphenomenasteadyinordertobringtheabstractstructuresthrough
whichweinterpretthemtothesurfaceinordertoexaminethemcriticallyand,
ifnecessary,toreconstitutethem.
Thiscanbeusefullyclarifedbyadiagram,seefgure1.6.Thedifference
betweenthetwoformsofknowledgeliesessentiallyinthedegreetowhichabstract
ideasareatthelevelofconsciousthoughtandthereforeatrisk.Thewholepurpose
ofscienceistoputtheabstractideaswethinkwithinmakingsenseofspatio-
temporaleventsatrisk.Insocialknowledge,thewholepurposeoftheknowledge
wouldbeputatriskbybringingthemtoconsciousthoughtsincetheirfunction
istobeusednormativelytocreatesociety.However,itisclearlyapossibilitythat
theabstractstructuresofsocialknowledgecould,aswithscience,themselves
becometheobjectofconsciousthought.This,inanutshell,istheprogrammeof
structuralism.Theessenceofthestructuralistmethodistoask:canwebuilda
modeloftheabstractprinciplesofasystem(e.g.language)thatgeneratesall
andonlythespatio-temporaleventsthatcanlegitimatelyhappen?Suchamodel
wouldbeatheoryofthesystem.Itwould,forexample,explainourintuitivesense
What architecture adds to building
30
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
thatsomestringsofwordsaremeaningfulsentencesandothersmostarenot.
Structuralismisratherliketakingtheoutputofacomputerasthephenomenatobe
explained,andtryingtofndoutwhatprogrammecouldgenerateallandonlythese
phenomena.Structuralismisanenquiryintotheunconsciousconfgurationalbases
ofsocialknowledge,thatis,itisaninquiryintothenon-discursivedimensionsof
socialandculturalbehaviour.
Building as the transmission of culture through artefacts
Thespatialandformalpatternsthatarecreatedthroughbuildingsandsettlements
areclassicinstancesoftheproblemofnon-discursivity,bothinthesenseofthe
confgurationalnatureofideaswethinkwithincreatingandusingspace,andinthe
senseoftheroletheseplayinsocialknowledge.Ashasalreadybeenindicated,
oneofthemostpervasiveexamplesofthisisthedwelling.Domesticspacevaries
inthedegreetowhichitissubjecttosocialknowledge,butitisnotuncommon
forittobepatternedaccordingtocodesofconsiderableintricacywhichgovern
whatspacesthereare,howtheyarelabelled,howboundedtheyare,howthey
areconnectedandsequenced,whichactivitiesgotogetherinthemandwhichare
separated,whatindividualsorcategoriesofpersonshavewhatkindsofrightsin
them,howtheyaredecorated,whatkindsofobjectsshouldbedisplayedinthem
andhow,andsoon.Thesepatternsvaryfromoneculturalgrouptoanother,but
invariablywehandledomesticspacepatternswithoutthinkingofthemandeven
withoutbeingawareofthemuntiltheyarechallenged.Ingeneral,weonlybecome
awareofthedegreeofpatterninginourownculturewhenweencounteranother
formofpatterninginanotherculture.
Butdomesticspaceisonlythemostintensiveandcomplexinstanceofa
moregeneralisedphenomenon.Buildingsandsettlementsofallkinds,andatall
levels,aresignifcantlyunderpinnedbyconfgurationalnon-discursivity.Itisthrough
thisthatbuildingsandindeedbuiltenvironmentsofallkindsbecomepartof
whatMargaretMeadcalledthetransmissionofculturethroughartefacts.
20
This
transmissionoccurslargelythroughtheconfgurationalaspectsofspace
andforminthoseenvironments.Forexample,wethinkconsciouslyofbuildings
asphysicalorspatialobjectsandwethinkoftheirpartsasphysicalorspatial
parts,likecolumnsorrooms.Butwethinkofbuildingsaswholeentitiesthrough
theunconsciousintermediaryofconfguration,inthatwhenwethinkofaparticular


abstractprinciples spacial-temporalevents

socialknowledge codes,rules speech,socialbehaviour,

ideastothink spaces,ideastothink

analyticknowledge theories,hypotheses facts,phenomena
paradigms
Figure 1.6
What architecture adds to building
3
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
kindofbuilding,weareconsciousnotonlyofanimageofanobject,butatthe
sametimeofthecomplexofspatialrelationsthatsuchabuildingentails.Asspace
andalsoasmeaningfulformsbuildingsareconfgurational,andbecause
theyareconfgurationaltheirmostimportantsocialandculturalpropertiesare
non-discursive.Itisthroughnon-discursivitythatthesocialnatureofbuildingsis
transmitted,becauseitisthroughconfgurationthattherawmaterialsofspaceand
formaregivensocialmeaning.Thesocialstuffofbuildings,wemaysay,isthe
confgurationalstuff,bothinthesensethatbuildingsareconfgurationsofspace
designedtoorderinspaceatleastsomeaspectsofsocialrelationships,andin
thesensethatitisthroughthecreationofsomekindofconfgurationintheform
ofthebuildingthatsomethinglikeaculturalmeaningistransmitted.
Building as process
Howthencanthishelpusmakethedistinctionbetweenarchitectureandbuilding?
Wenoteofcoursethatwenowbeginnotfromthenotionthatbuildingspriorto
architectureareonlypracticalandfunctionalobjects,butfromthepropositionthat
priortoarchitecturebuildingsarealreadycomplexinstancesofthetransmissionof
culturethroughartefacts.Thisdoesnotmeanofcoursethatbuildingsofthesame
typeandculturewillbeidenticalwitheachother.Onthecontrary,itiscommonfor
vernaculararchitecturestoexhibitprodigiousvarietyatthelevelofindividualcases,
somuchsothatthegroundsforbelievingthatthecasesconstituteinstancesofa
commonvernacularstyle,eitherinformorspace,canbequitehardtopindown.
Thecrucialstepinarrivingatourdefnitionofarchitectureistounderstand
frsthowthevernacularbuildersucceedsinmakingabuildingasacomplex
relationalstructurethroughwhichcultureistransmitted,whileatthesametime
creatingwhatwilloftenbeauniqueindividualbuilding.Wedonothavetolook
farfortheanswer.Thiscombinationofcommonstructureandsurfacevarietyis
exactlywhatwefndwheresocialknowledgeisinoperationintheforminwhich
wehavejustdescribedit:complexconfgurationalideasatthenon-discursivelevel
guidethewaysinwhichwehandlespatio-temporalthingsatthesurfacelevel,and
asaresultconfgurationalideasarerealisedintherealworld.Inbuildingterms,
themanipulationofthespatialandformalelementswhichmakeupthebuilding
will,ifcarriedoutwithinthescopeofnon-discursiveconfgurationalideastothink
with,whichgovernkeyaspectsoftheirformalandspatialarrangement,leadto
exactlythecombinationofunderlyingcommonstructureandsurfacevarietythat
characterisesvernaculararchitecturesingeneral.
Tounderstandhowthishappensinparticularcases,wecandrawonthe
remarkableworkofHenryGlassie.
21
GlassieproposesthatweadaptfromNoam
Chomskysstudiesoflanguageaconceptwhichhecallsarchitecturalcompetence.
Architecturalcompetenceisasetoftechnological,geometricalandmanipulative
skillsrelatingformtouse,whichconstituteanaccountnotofhowahouseismade,
butofhowahouseisthoughtsetoutlikeaprogrammeaschemeanalogoustoa
grammar,thatwillconsistofanoutlineofrulesetsinterruptedbyprosyexegesis.
What architecture adds to building
32
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Theanalogywithlanguageisapposite.Itsuggeststhattherulesetsthevernacular
designerusesaretacitandtakenforgrantedinthesamewayastherulesetsthat
governtheuseoflanguage.Theyareideasthedesignerthinkswithratherthanof.
Theythereforehaveacertaindegreeofabstractionfromthematerialrealitythey
helptocreate.Theyspecifynotthespecifcbutthegeneric,sothatthevernacular
designermayusetherulesasthebasisofacertainrestrainedcreativityin
interpretingtherulesinnovelways.
NowtheimplicationofGlassiesideaisthatarchitecturalcompetence
providesasetofnormativerulesabouthowbuildingshouldbedone,sothata
vernacularbuildingreproducesaknownandsociallyacceptedpattern.Thehouse
builtbyabuildersharingthecultureofacommunitycomesoutrightbecause
itdrawsonthenormativerulesthatdefnethearchitecturalcompetenceofthe
community.Inthiswaybuildingsbecomeanaturalpartofthetransmissionof
culturebyartefacts.Throughdistinctivewaysofbuilding,aspectsofthesocial
knowledgedistinctiveofacommunityarereproduced.Thusthephysicalactof
building,throughasystemofwelldefnedinstrumentalities,becomesthemeans
bywhichthenon-discursivepatternswecallculturearetransmittedintoand
throughthematerialandspatialformsofbuildings.Thenon-discursiveaspectsof
buildingaretransmittedexactlyaswewouldexpectthemtobe:asunconscious
patternimplicationsofthemanipulationofthings.
So what is architecture?
Tounderstandbuilding,then,wemustunderstanditbothasaproductandasa
process.Havingdonethis,wecanreturntoouroriginalquestion:whatisitthat
architectureaddstobuilding?Byunpackingtheculturalandcognitivecomplexity
ofbuilding,itwillturnoutthatweareatlastinsightofananswer.Whatever
architectureis,itmustinsomesensegobeyondtheprocessbywhichtheculturally
sanctionednon-discursivitiesareembeddedinthespatialandphysicalformsof
buildings.Inwhatsense,then,isitpossibletogobeyondsuchaprocess?
Theanswerisnowvirtuallyimpliedintheformofthequestion.Architecture
beginswhentheconfgurationalaspectsofformandspace,throughwhichbuildings
becomeculturalandsocialobjects,aretreatednotasunconsciousrulestobe
followed,butareraisedtothelevelofconscious,comparativethought,andinthis
waymadepartoftheobjectofcreativeattention.Architecturecomesintoexistence,
wemaysay,asaresultofakindofintellectualprise de conscience:webuild,but
notasculturalautomata,reproducingthespatialandphysicalformsofourculture,
butasconscioushumanbeingscriticallyawareoftheculturalrelativityofbuiltforms
andspatialforms.Webuild,thatis,awareofintellectualchoice,andwetherefore
buildwithreason,givingreasonsforthesechoices.Whereasinthevernacularthe
non-discursiveaspectsofarchitecturearenormativeandhandledautonomically,in
architecturethesecontentsbecometheobjectofrefectiveandcreativethought.The
designerisineffectaconfgurationalthinker.Theobjectofarchitecturalattention
ispreciselytheconfgurationalideastothinkwiththatinthevernaculargovern
What architecture adds to building
33
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
confgurationaloutcomes.Thisdoesnotmeanthatthedesignerdoesnotthink
ofobjects.Itmeansthatatthesametimethedesignerthinksofconfguration.
Theessenceofarchitectureliesthereforeinbuildingnotbyreferenceto
culturallyboundcompetences,andthewayinwhichtheyguidethenon-discursive
contentsofbuildingsthroughprogrammesofsocialknowledgespecifctoone
culture,butbyreferencetoawould-beuniversalisticcompetencearrivedatthrough
thegeneralcomparativestudyofformsaimedatprincipleratherthancultural
idiosyncrasy,and,throughthis,atinnovationratherthanculturalreduplication.Itis
whenweseeinthenon-discursivecontentsofbuildingsevidenceofthisconcern
fortheabstractcomparabilityofformsandfunctionsthatbuildingistranscended
andarchitectureisnamed.Thisiswhythenotionofarchitectureseemstocontain
withinitselfaspectsofboththeproductwhichiscreatedandoftheintellectual
processthroughwhichthiscreationoccurs.
Architectureexists,wemightsay,wherewenoteasapropertyofthings
evidencenotonlyofacertainkindofsystematicintenttoborrowanexcellent
phraseproposedbyacolleagueinreviewingthearchaeologicalrecordforthe
beginningofarchitecture
22
inthedomainofnon-discursivity,butofsomething
liketheoreticalintentinthatdomain.Inakeysensearchitecturetranscendsbuilding
inthesamewaythatsciencetranscendsthepracticalcraftsofmakinganddoing.
Itintroducesintothecreationofbuildingsanabstractconcernforarchitectural
possibilitythroughtheprincipledunderstandingofformandfunction.Theinnovative
imperativeinarchitectureisthereforeinthenatureofthesubject.Weshouldno
morecriticisearchitectsfortheirpenchanttowardsinnovationthanweshould
scientists.Inbothcasesitfollowsfromthesociallegitimationswhichgiveeach
itsnameandidentity.Botharchitectureandscienceusethegroundoftheoretical
understandingtomovefrompastsolutionstofuturepossibility,thelatterinthe
directionofnewtheoreticalconstructs,theformerinthedirectionofnewrealities.
Thejudgmentwemakethatabuildingisarchitectureariseswhentheevidenceof
systematicintentisevidenceofintellectualchoiceanddecisionexercisedinafeld
ofknowledgeofpossibilitythatgoesbeyondcultureintoprinciple.Inthissense,
architectureisaformofpracticerecognisableinitsproduct.Thejudgmentwemake
thatabuildingisarchitecturecomeswhenweseeevidenceinthebuildingbothof
systematicintentwhichrequirestheabstractandcomparativemanipulationofform
withinthegeneralrealmofarchitecturalpossibility,andthatthisexplorationandthis
exerciseofintellectualchoicehasbeensuccessfullyaccomplished.
Architectureisthusbothathingandanactivity.Intheformofthething
wedetectevidenceofasystematicintentofthearchitecturalkind.Fromthebuilt
evidencewecanjudgeboththatabuildingisintendedtobearchitectureand,ifwe
aresoinclined,thatitisarchitecture.Weseenowwhythedefnitionofarchitecture
issodiffcult.Becauseitisthetakingholdofthenon-discursivecontentsofbuilding
byabstract,universalisticthought,itisatonceanintentionalmentalactanda
propertyweseeinthings.Itisbecauseweseeinthingstheobjectivisedrecord
ofsuchthoughtthatwenametheresultarchitecture.
What architecture adds to building
34
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Itisclearfromthisanalysisthatarchitecturedoesnotdependonarchitects,
butcanexistwithinthecontextofwhatwewouldnormallycallthevernacular.To
theextentthatthevernacularshowsevidenceofrefectivethoughtandinnovation
atthelevelofthegenotype,thenthatisevidenceofthekindofthoughtwecall
architecturalwithinthevernacular.Thisdoesnotmeanthattheinnovativeproduction
ofbuildingswhicharephenotypicallyindividualwithinavernacularshouldbethought
ofasarchitecture.Suchphenotypicalvarietyisnormalastheproductofculturally
constrainednon-discursivecodes.Itisonlywhentheinnovation,andthereforethe
refectivethought,changesthecodethatunderliestheproductionofphenotypes
thatwedetectthepresenceofabstractandcomparativeandthereforearchitectural
thoughtwithintheconfnesofvernaculartradition.Itisthereforeperhapsattimes
ofthegreatestchangethatwebecomeawareofthistypeofthoughtinvernacular
traditions,thatis,whenanewvernaculariscomingintoexistence.Thisiswhy
thedemarcationbetweenthevernacularandarchitectureconstantlyshifts.The
reproductionofexistingforms,vernacularorotherwise,isnotarchitecturebecause
thatrequiresnoexerciseofabstractcomparativethought,buttheexploitationof
vernacularformsinthecreationofnewformscanbearchitecture.
Architectureexiststhentothedegreethatthereisgenotypicalinvention
inthenon-discursive,thatis,inventionwiththerulesthatgovernthevariabilitythat
ispossiblewithinastyle.Thepreconditionforsuchinventionisanawarenessof
possibilitieswhicharenotcontainedincontemporaryculturalknowingbutwhich
areatthesametimewithinthelawsofwhatisarchitecturallypossible.Architecture
ischaracterisedthereforebyapreoccupationwithnon-discursivemeansratherthan
non-discursiveends.Thisisnottheoutcomeofaperverserefusaltounderstand
theculturalnatureofbuilding,butatakingholdofthisveryfactasapotentialityto
exploretheinterfacebetweenhumanlifeanditsspatialandphysicalmilieu.Inthe
actofarchitecturalcreation,theconfgurationalpotentialitiesofspaceandform
aretherawmaterialswithwhichthecreatorworks.
Likeanycreativeartist,therefore,thearchitectmustseektolearn,through
intellectualinquiry,thelimitsandpotentialitiesoftheserawmaterials.Intheabsence
ofsuchinquiry,therearemanifestandimmediatedangers.Inthevernacularthe
patternofformandthepatternofspacewhichgivethebuildingitssocialcharacter
arerecreatedthroughthemanipulationandassemblingofobjects.Wecansay
thenthattheform,thespatialpatternandthefunctionalpatterntheform-function
relation,inshortareknowninadvanceandneedonlyberecreated.Because
architectureofitsnatureunlinksthepatternaspectsofthebuildingfromtheir
dependenceonsocialknowledge,theseaspectsofthebuildingandaboveall
theirrelationtosocialoutcomesbecomeuncertain.
Inarchitecturethen,becausethesecrucialrelationsbetweennon-discursive
formsandoutcomesarenotknowninadvance,architecturehastorecreateina
new,moregeneralisedform,theknowledgeconditionsthatprevailinthevernacular.
Becausearchitectureisacreativeact,theremustbesomethingintheplaceof
thesocialknowledgestructureasideastothinkwith.Sincearchitectureisbased
What architecture adds to building
35
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
onthegeneralcomparabilityofpossibleforms,thisknowledgecannotsimply
encompassparticularcases.Itmustencompasstherangeofpossiblecases
andifpossiblecasesingeneral.Thereisonlyonetermforsuchknowledge.Itis
theoreticalknowledge.Wewillseeinthenextchapterthatallarchitecturaltheories
areattemptstosupplyprincipledknowledgeofthenon-discursive,thatis,torender
thenon-discursivediscursiveinawaythatmakesitaccessibletoreason.Inthe
absenceofsuchknowledge,architecturecanbe,asthetwentiethcenturyhas
seen,adangerousart.
Thepassagefrombuildingtoarchitectureissummarisedinfgure1.7.The
implicationofthisisthat,althoughweknowthedifferencebetweenarchitecture
andbuilding,thereisnohardandfastlinetobedrawn.Eithercanbecomethe
otheratanymoment.Takingabroaderviewwhichencompassesboth,wecan
saythatintheevolutionofbuildingwenotetwowaysinwhichthingsaredone:in
obediencetoatradition,orinpursuitofinnovation.Buildingcontainsarchitecture
tothedegreethatthereisnon-discursiveinvention,andarchitecturebecomes
buildingtothedegreethatthereisnot.Vernacularinnovationisthereforeincluded
withinarchitecture,butthereduplicationofvernacularformsisnot.Architectureis
thereforenotsimplywhatisdonebuthowitisdone.
Thebringingofthenon-discursive,confgurationdimensionofbuilt
formfromculturalreproductiontorefectiveawarenessandabstractexploration
ofpossibilityisatonceapassagefromthenormativetotheanalyticandfrom
theculture-boundtotheuniversal,thelattermeaningthatallpossibilitiesare
openratherthansimplythepermutationsandphenotypicalinnovationsthatare
sanctionedbythevernacular.Thepassageisalsoonewhichtransformstheidea
ofknowledgefromculturalprincipletotheoreticalabstraction.
Inastrongsense,then,architecturerequirestheory.Ifitdoesnothave
theoreticalknowledge,thenitwillcontinuetodependonsocialknowledge.Worse,
thereiseverypossibilitythatarchitecturecancometobebasedonsocialknowledge
masqueradingastheoreticalknowledge,whichwillbeallthemoredangerous
becausearchitectureoperatesintherealmsofthenon-discursivethroughwhich
societyistransmittedthroughbuilding.
23
Architectureisthereforepermanently
enjoinedtotheoreticaldebate.Itisinitsnaturethatitshouldbeso.Inthatitisthe
applicationofrefectiveabstractthoughttothenon-discursivedimensionsofbuilding,
andinthatitisthroughthesedimensionsthatoursocialandculturalnaturesare
inevitablyengaged,architectureistheoryappliedtobuilding.Inthenextchapterwe
willthereforeconsiderwhatwemeanbytheoryinarchitecture.
What architecture adds to building
36
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Notes
J.Ruskin:Seven Lamps of Architecture,London1849,chap.1.
Theliteratureonvernaculararchitectureascultureisnowextensive,andgrowing
rapidly.AmongtheseminaltextsofferingwidecoverageareRudovskysArchitecture
Without Architects,1964;PaulOliversShelter and Society,Barrie&Rockliff,The
CressetPress,1969anditsfollow-upShelter in Africa,Barrie&Jenkins,London,
1971;AmosRapoportsHouse Form and Culture,PrenticeHall,1969;Labelle
Prussinsclassicreviewofthecontrastingvernacularswithinaregion,Architecture
in Northern Ghana,UniversityofCaliforniaPress,1969;SusanDenyersAfrican
TraditionalArchitecture,Heineman,1978;andKajAndersensAfricanTraditional
Architecture,OxfordUniversityPress,1977;inadditiontoearlieranthropological
classicssuchasC.DaryllFordesHabitat, Economy and Society,Methuen,1934.
Studiesofspecifcculturesarenowtoonumeroustomention,asarethemuch
large-numberoftextswhichhavenowdealtwiththearchitectureofparticular
culturesandregions,butwhicharenotyetavailableinEnglish.Amongrecent
studiesofthevernacular,themostimportanttomymindandbyfarthemost
infuentialinthistexthasbeentheworkofHenryGlassie,andinparticularhis
Folk Housing in Middle Virginia,UniversityofTennesseePress,1975,references
towhich,explicitandimplicit,recurthroughoutthistext.
Thesamehasoftenbeensaidofindustrialarchitecture.J.M.Richards,for
example,inhisAn Introduction to Modern Architecture,Penguin,1940,describes
Figure 1.7
1
2
3
What architecture adds to building
37
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
ThomasTelfordsStKatharinesDockasTypicalofthesimplebutnobleengineers
architectureofhistime.
RogerScruton,The Aesthetics of Architecture,.Methuen,1977.
Ibid.,p.16.
ForarecentrestatementofthisbeliefseeS.Gardiner,The Evolution of the House,
Paladin,1976.
Seeforexampleprehistoricalsectionsofthemostrecent(nineteenth)editionof
SirBanisterfetchersA History of Architecture (editedbyProfessorJohnMusgrove)
writtenbymycolleagueDrJulienneHanson.Itisacommentonarchitecturalhistory
thatitisonlyveryrecentlythatthetrueantiquityofbuildinghasbeenrefectedin
thehistoriesofworldarchitecture.SomeofDrHansonssourcesareinthemselves
remarkabletextswhichifbetterknownwouldentirelychangepopularconceptionof
thehistorynotonlyofbuildingbutalsoofhumansociety.Thekeytextsaregivenin
DrHansonsbibliography,butIwouldsuggesttheremarkableR.G.Klein,Ice Age
Hunters of the Ukraine,ChicagoandLondon,1973asagoodstartingpoint.
B.Russell,The Problems of Philosophy,HomeUniversityLibrary,1912,Oxford
UniversityPresspaperback,1959;Chapter9Theworldofuniversals.
R.A.Scruton,The Aesthetics of Architecture,p.43etseq.
R.A.Scruton,A Short History of Modern Philosophy: from Descartes to Wittgenstein,
ARKPaperbacks,1984.
R.Descartes,The Principles of Philosophy,Part2,PrincipleXinThe Philosophical
Works of Descartes,CambridgeUniversityPress,vol.1,p.259.
Descartes,PrincipleXI,p.259.
Descartes,PrincipleX,p.259.
Graphswhichhavethispropertyarecalledplanargraphs.Anyspatiallayouton
onelevel,consideredasagraphofthepermeabilityrelations,isboundtobeplanar.
TheseexamplesaretakenfromastudyofseventeenhousesinNormandycarried
outfortheCentreNationaledeRechercheScientifque,andpublishedasIdeas
areinthingsinEnvironment and Planning B, Planning and Design 1987,vol.14,pp.
36385.Thisarticlethenformedoneofthebasicsourcesforamuchmoreextended
treatmentinJ.Cuisenier,La Maison Rustique: logique social et composition
architecturale,PressesUniversitairesdeFrance,1991.
Thenormalisationformulafortakingtheeffectofthenumberelementsinthe
graphoutofthetotaldepthcalculationfromanelementis2(md1)/k2,wheremd
isthemeandepthofotherelementsfromtherootelement,andkisthenumber
ofelements.ThereisadiscussionofthismeasureinP.Steadman,Architectural
Morphology,Pion,1983,p.217.ThemeasurewasfrstpublishedinHillieretal.,
SpaceSyntax:anewurbanperspectiveintheArchitects Journal,no48,vol.178,
30.11.83.Thereisanextensivediscussionofitstheoreticalfoundationsandwhyitis
soimportantinspaceinHillierandHanson,The Social Logic of Space,Cambridge
UniversityPress,1984.Themeasuretheoreticallyeliminatestheeffectofthe
numbersofelementsinthesystem.However,inarchitecturalandurbanrealitythere
isanadditionalproblem:bothbuildingsandsettlements,forpracticalandempirical
4
5
6
7
8
9
10
11
12
13
14
15
16
What architecture adds to building
38
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
reasons(aswillbefullydiscussedinChapter9)tendtobecomerelativelyless
deepastheygrow.Asecond,empiricalnormalisationformulaisthereforerequired
totakeaccountofthis.SuchaformulaissetoutinThe Social Logic of Space,
whichhasprovedrobustinuse,buthasbeenextensivelydiscussed,forexample
inJ.Teklenberg,H.Timmermans&A.vanWagenberg,Spacesyntax:standardised
integrationmeasuresandsomesimulations,Environment & Planning B: Planning &
Design,vol.20,1993,pp.34757.SeealsoM.Kruger,Onnodeandaxialgridmaps:
distancemeasuresandrelatedtopics,paperfortheEuropeanConferenceonthe
RepresentationandManagementofUrbanChange,Cambridge,September1989,
UnitforArchitecturalStudies,UniversityCollegeLondon.
Thereisafurthermeasurecalleddifferencefactor,whichexpresseshowstrong
thesedifferencesare,setoutinIdeasareinthings,citedinnote15above.
Itshouldbenotedthattheargumentinthepaperfromwhichtheseexamplesare
taken,Ideasareinthingsisagreatdealmorecomplexthanthatpresentedhere
toillustratethetechnique.Infact,itwasproposedthattwofundamentaltypological
tendencieswouldbeidentifedwithinthesample,whichweremoretodowith
differ-encesintherelationsofthesexesthananythingelse.Anewversionofthis
paperwillbepublishedinJ.Hanson,The Social Logic of Houses,forthcomingfrom
CambridgeUniversityPress.
Theseissuesaredealtwithatgreaterlengthandforaslightlydifferentpurposein
Chapter7.
MargaretMead,Continuities in Cultural Evolution,YaleUniversityPress,1964,
Chapter5.
Forexample,HenryGlassie,Folk Housing in Middle Virginia.
J.Hanson,writtenfortheintendedEncyclopaedia of Architecture,McGraw-Hill,New
York,butnotyetpublished.
Wewillseeinlaterchapters,andparticularlyinChapters6and11,exactlyhowthis
canoccurandwhatitsconsequencesare.
17
18
19
20
21
22
23
The need for an analytic theory
of architecture
Chapter two
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Do architects need theories?
Inthepreviouschapter,architecturewasdefnedasthetakingintorefectivethought
ofthenon-discursive,orconfgurational,aspectsofspaceandforminbuildings.
Invernaculartraditions,theseaspectsaregovernedbythetakenforgrantedideas
to thinkwithofaculture.Inarchitecture,ideas to think withbecomeideas to think
of.Spatialandformalconfgurationinbuildingsceasestobeamatterofcultural
reproductionandbecomesamatterofspeculativeandimaginativeenquiry.
Itfollowsfromthisdefnitionthatarchitectureisanaspiration,notagiven.To
bringtoconsciousthoughttheprinciplesthatunderliethespatialandformalpatterns
thattransmitculturethroughbuildings,andtoformulatepossiblealternativesthat
workasthough they were culturesincearchitecturemustbeanadditiontoculture
notsimplyaremovalofitisanintellectualaswellasacreativetask.Itrequiresnot
onlytheconceptualisationofpatternandconfguration invacuo,butalsocomparative
knowledgeandrefectivethought.Thisiswhyarchitectureisarefectiveaswellas
animaginativeproject,onewhichseekstoreplaceoratleasttoaddtothesocial
knowledgecontentofbuildingwithanenquiryintoprincipleandpossibility.
Architecturaltheoryistheultimateaimofthisrefection.Anarchitectural
theoryisanattempttorenderoneorotherofthenon-discursivedimensions
ofarchitecturediscursive,bydescribinginconcepts,wordsornumberswhat
theconfgurationalaspectsofformorspaceinbuildingsarelike,andhowthey
contributetothepurposesofbuilding.Inasense,theorybeginsatthemoment
architecturebegins,thatis,whenspatialandformalconfgurationinbuildings,and
theirexperientialandfunctionalimplications,arenolongergiventhroughatradition
ofsocialknowledgetransmittedthroughtheactofbuildingitself.Assoonas
buildingmovesfreefromthesafeconfnesofculturalprogramming,somethinglike
atheoryofarchitectureisneededtosupportthecreativeactbyproposingamore
generalunderstandingofthespatialandformalorganisationofbuildingsthanis
availablewithinthelimitsofasingleculture.
Thisisnottosaythatcreativearchitecturedependsontheory.Itdoes
not.Butinthatarchitectureistheapplicationofspeculativeabstractthoughttothe
materialworldinwhichwelive,therefectiveaspectsofarchitecturalenquirylead
totheformulationifnotoftheorythenatleastoftheory-likeideas.Theneedfor
theorybecomesgreaterasarchitectureadvances.Theoryismostrequiredwhen
architecturebecomestrulyitself,thatis,whenitbecomesthefreeexplorationof
formalandspatialpossibilityinthesatisfactionofthehumanneedforbuildings.
However,thefactthattheoryisaninevitableaspectofarchitecturedoes
notmeanthatalltheorieswillhaveapositiveeffectonarchitecture.Onthecontrary,
thedependenceofarchitectureontheoreticalideascreatesanewtypeofrisk:that
theoriesmaybewrong,maybedisastrouslywrong.Themuchdiscussedfailure
ofmodernisminarchitectureisseenasatleastthefailureofatheorythemost
ambitiousandcomprehensiveeverproposedandevenbysomeasthefailureof
theveryideaofatheoryofarchitecture.
39
The need for an analytic theory
of architecture
40
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Asaresult,inthelatetwentiethcenturyanumberofnewquestionsareposed
abouttheoriesofarchitecturewhicharealsoquestionsaboutarchitectureitself.
Doesarchitecturereallyneedtheories,oraretheyjustapretentiousadjunctto
anessentiallypracticalactivity?Ifarchitecturedoesneedtheories,thenwhatare
theylike?Aretheylikescientifctheories?Oraretheyaspecialkindoftheory
adaptedforarchitecturalpurposes?Ifarchitecturaltheoriescanbewrong,and
haveapparentlyadverseconsequences,thencantheyalsoberight?Howcanwe
setaboutmakingarchitecturaltheoriesbetter?Andmostdiffcultofall:howcan
architectureasacreativeartbereconciledtothedisciplinesoftheory?Arethe
twonotopposedtoeachother,inthatbettertheoriesmustleadinevitablytothe
eliminationofarchitecturalfreedom.
Theanswerproposedinthischapteristhatonceweacceptthattheobject
ofarchitecturaltheoryisthenon-discursivethatis,theconfgurationalcontent
ofspaceandforminbuildingsandbuiltenvironments,thentheoriescanonlybe
developedbylearningtostudybuildingsandbuiltenvironmentsasnon-discursive
objects.Tohaveatheoryofnon-discursivityinarchitectureingeneralwemustfrst
buildacorpusofknowledgeaboutthenon-discursivecontentsofarchitectureas
aphenomenon.Thisofcourserunscountertomostcurrenteffortsinarchitectural
theory,whichseektobuildtheoryeitherthroughtheborrowingofconceptsfrom
otherfelds,orthroughintrospectionandspeculation.
However,theproductofthefrst-handstudyofnon-discursivityinbuildings
andbuiltenvironmentswillleadtoanewkindoftheory:ananalytictheoryof
architecture,thatis,onewhichseekstounderstandarchitectureasaphenomenon,
beforeitseekstoguidethedesigner.Ananalytictheoryofarchitectureis,itwillbe
argued,thenecessarycorollaryofarchitecturalautonomy.Withouttheprotection
ofananalytictheory,architectureisinevitablysubjecttomoreandmoreexternally
imposedrestrictionsthatsubstitutesocialideologyforarchitecturalcreativity.
Analytictheoryisnecessaryinordertoretaintheautonomyofcreativeinnovation
onwhichtheadvanceofarchitecturedepends.
Are architectural theories just precepts for builders?
Beforewecanembarkonthetaskofbuildingananalytictheoryofarchitecture,
however,wemustfrstexploretheideaoftheoryinarchitecturealittletoprevent
ourenquirybeingobscuredbysomeofthemorecommonmisconceptions.
Architecturaltheoriesdotakeaverydistinctiveform,butallisnotasitseemsat
frstsight,anditisimportantthatwedonotallowappearancestodisguisetheir
truenatureandpurposes.
Wemayusefullybeginbyexaminingtheviewsofawell-knowncriticof
architecturaltheories.Inhis1977polemicagainstarchitecturalmodernismandits
intellectualfashions,The Aesthetics of Architecture
1
RogerScrutonisdismissiveof
theveryideaofatheoryofarchitecture:Architecturaltheory,hesaysinafootnote,
isusuallythegestureofapracticalman,unusedtowords.Elsewherehegoes
further.Thereisnotandcannotbeatheoryofarchitecture.Whathasbeencalled
The need for an analytic theory
of architecture
4
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
architecturaltheoryaremerelypreceptswhichguidethebuilder.Whilesuch
preceptscanbeusefulcanons,theycanneveramounttoarealtheory,becausethey
cannotbeuniversal,anditisonlywiththeclaimtouniversalitythattheoryarises.
2
Atfrstsight,Scrutonseemstoberight.Forthemostpartmodernismis
oneofthefewexceptionsweassociatetheoriesofarchitecturewithindividual
architects.WhenwethinkofPalladiosorLeCorbusierstheoryofarchitecture
wetakeittomeansomethingliketheintellectualgroundofastyle,thegeneric
principlesunderlyinganapproachtodesign.Itseemsselfevidentthatnosuch
principlescouldeverbeuniversal.Theideaevenleadstoparadox.Auniversal
formulaforarchitecturewould,iffollowed,renderarchitecturethesameand
unchanging,andthereforeultimatelydull.
Butdoestheoryinarchitecturereallyonlymeanaformulaforarchitectural
success?Ascientistwouldfndthisastrangeuseofthewordtheory.Forascientist
atheoryisarationalconstructintendedtocapturethelawfulnessofhowtheworld
is,notasetofguidelinesastohowitshould be.Scientifctheorieshelpusacton
theworld,butonlybecausetheyhavefrstdescribedtheworldindependentlyofany
viewofhowitshouldbe.Theessenceofscienceisthatitstheoriesareanalytic,not
normativeinintent.Theydescribehowtheworldis,notprescribehowitoughttobe.
Mightwethensuggestthatthisisexactlythedifferencebetween
architecturalandscientifctheories,namelythatscientifctheoriesareanalytic,
andaboutunderstandinghowthingsare,whereasarchitecturaltheoriesare
normative,andabouttellinguswhattodo?Thereseemstobesometruthinthis.It
isreasonabletosaythatarchitectureisabouthowtheworldshouldberatherthan
howitis,andthatitstheoriesshouldthereforetendtoexpressaspirationsratherthan
realities.Infact,oncloserexamination,itturnsoutthatthisisnotandcanneverbe
thecase.Admittedly,architecturaltheoriesarenormallypresentedinnormativeform,
butatadeeperleveltheyarenolessanalyticthanscientifctheories.
Takeforexample,twotheorieswhichareaboutasfarapartastheycould
beinfocusandcontent,Albertistheoryofproportion,
3
andOscarNewmanstheory
ofdefensiblespace.
4
Botharepresentedaspreceptsforsuccessfuldesign,inthat
bothauthorsbooksareaimedprimarilyatguidingthearchitecturalpractitionerin
design,ratherthanexplainingthenatureofarchitecturalexperienceasexperienced,
asScrutonsbookis.Butifwereadthetextscarefully,wefndthatthisisnotall
theyare.Ineachcase,thenormativecontentoftheworkrestsonclear,ifbroad,
analyticfoundations.AlbertistheoryofproportionrestsonthePythagoreannotionof
mathematicalforminnature,
5
andthecoincidenceitassertsbetweentheprinciples
ofnaturalformandthepowersofthemind,asevidencedbytherelationshipbetween
oursenseofharmonyinmusicandthesimplenumericalratiosonwhichthose
harmoniesarebased.Ifarchitecturefollowsthemathematicalprinciplesfoundin
nature,Albertiargues,thenitcannothelpreproducingtheintelligibilityandharmony
thatwefndinnaturalforms.Similarly,Newmansdefensiblespacetheoryrestson
thetheoryofhumanterritoriality,bywhichgenetictendenciesincertainspeciesto
defendterritoryagainstothersofthespecies,aregeneralisedtohumanbeings,both
The need for an analytic theory
of architecture
42
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
asindividualsandmistakenlyinmyviewasgroups.If,Newmanargues,architects
designspaceinconformitywithterritorialprinciples,thenitwillbefollowing
biologicaldrivesbuiltintousbynature.
6
Itisnotablethatinbothofthesetheories,theprinciplesfordesignaresaid
tobebasedonprinciplestobefoundinnature.Inaverystrongsense,then,inboth
casesthenormativecontentofthetheorydependsontheanalytic.Onrefection,
itmustbesotosomedegreeinallcases.Anytheoryabouthowweshouldact
toproduceacertainoutcomeintheworldmustlogicallydependonsomeprior
conceptionofhowtheworldisandhowitwillrespondtoourmanipulations.Careful
examinationwillshowthatthisisalwaysthecasewitharchitecturaltheories.We
invariablyfndthatthepreceptsaboutwhatdesignersshoulddoaresetinaprior
frameworkwhichdescribeshowtheworldis.Sometimesthisframeworkisexplicitly
setout,andrestsonaspecifcscientifcorquasi-scientifcfoundation,asinthe
twocaseswehaveinstanced.Sometimesitismuchmoreimplicit,refectingno
morethanacurrentlyfashionablewayoflookingattheworld,asforexample
manyrecenttheorieshaverestedonthefashionableassumptionthateverything
isalanguagesothatdesignerscanandshoulddesignfollowingtheprinciplesof
linguistictheoriesinmakingtheirbuildingsmeaningful.
Althoughpresentednormatively,then,architecturaltheoriesmusthavea
greatdealofanalyticcontent,whetherthisisexplicitorimplicit.Inpointoffact,
facedwithanarchitecturaltheory,ourfrstreactionwouldusuallybetotreatit
exactlyaswewouldascientifctheory.Offeredageneralpropositiononwhichto
basearchitecturalpreceptsfordesignsayapropositionaboutthepsychological
impactofacertainproportionalsystemsorthebehaviouraleffectsofacertain
kindofspatialorganisationourfrstreactionwouldbetoquestionthegeneral
proposition,oratleasttosubjectittotestbyareviewofcases.Weusuallyfnd
quitequicklythatwould-begeneralpropositionsrunfoulofcasesknowntous,
whichwetheninstanceascounter-examplestothetheory.Inotherwords,wetreat
anarchitecturaltheoryverymuchinthesamewayaswewouldtreatascientifc
theory:thatis,wetreatitasananalytictheorybytryingtofndcounter-examples
whichwouldrefuteitsgenerality.Evenwhenitsurvives,wewouldbeinclinedto
treatitwithcontinuingscepticismasatbestaprovisionalgeneralisation,which
wecanmakeuseofuntilabetteronecomesalong.
Itisamistake,then,totreatarchitecturaltheoriessimplyasnormative
precepts,asScrutondoes.Architecturaltheoriesarenotandcannotbesimply
normative,butareatleastanalytic-normativecomplexes,inwhichthenormativeis
constructedonthebasisoftheanalytic.Itfollowsthatproperlytheoreticalcontent
ofarchitecturaltheoriesisspecifedbytheanalytic.Iftheanalytictheoryiswrong,
thenthelikelihoodisthatthebuildingwillnotrealiseitsintention.Architectural
theories,wemightsay,areabouthowtheworldshouldbe,butonlyinthelight
ofhowitisbelievedtobe.
The need for an analytic theory
of architecture
43
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Theories in design
Whyshouldarchitecturaltheoriestakethisdistinctiveformofcombining
propositionsabouthowtheworldshouldbewithpropositionsabouthowitis
believedtobe?Theansweristobefoundinthenatureofwhatarchitectsdo,
thatis,design.Throughitsnatureasanactivity,designraisesissuestowhich
architecturaltheoristsproposesolutionsintheformofanalytic-normative
complexesoftheoreticalideas.Tounderstandwhythisisso,wemust
understandalittleaboutdesign.
Designisofcourseonlyapartoftheprotractedprocessesbywhich
buildingscomeintoexistence.Thebuildingprocessinvolvesformulatinganeed
forapossiblebuilding,conceptualisingwhatitmightbelike,initiatingaprocessof
resourcing,negotiatingandorganising,creatingsomekindofrepresentation,orseries
ofrepresentationsofincreasingrefnement,ofwhatthebuildingwillbelike,then
constructing,ftting,operationalising,andfnallyoccupyingthecompletedbuilding.
Vernacularbuildingisofcoursealesscomplexprocess.Butifthecircumstances
existinwhichdesignisafunction,thenthecorollaryisthatthismorecomplex
buildingprocess,orsomethingapproximatingtoit,alsoexists.Designdoesnot
existasafunctionindependentofthislargerprocess.Onthecontrary,itimpliesit.
Howthendowedefnedesignwithinthisprocess?First,wenotethatitis
onlyattheendoftheprocessthattheobjectoftheprocessanoccupiedbuilding
exists.Formostofitsduration,theprocessisorganisedaroundasurrogatefor
thebuildingintheformofanabstractideaorrepresentationwhichcontinually
changesitsform.Itbeginsasanideaforthebuilding,thenbecomesanideaofthe
building,thenamoreformalisedconcept,thenaseriesofmoreandmorerefned
representations,thenasetofinstructionsandfnallyabuilding.Forthemostpart,
thecomplexprocessofbuildingtakesplacearoundthisshifting,clarifying,
graduallymaterialisingidea.
Theprocessofseeking,fxing,andrepresentingarealisableconcept
ofabuildingfromanideaforabuildingisdesign.Designiswhatarchitectsdo,
thoughitisnotalltheydo,andnotonlyarchitectsdoit.Butitisdesignthatkeeps
whatarchitectsdowhetherornotitisarchitectsthatdoitfxedintheprocess
ofcreatingbuildings.Therehastobeacontroloftheprocessofsearchingout,
conceptualising,andrepresentingthesurrogatebuildingthroughtheprocess.Let
uscallthisthedesignfunction,sothatwecanseethatitisindependentofwho
actuallycarriesitout.
Thedesignfunctionexistswithinthebuildingprocessforonefundamental
reason:becauseatallstagesoftheprocessthoughwithdifferingdegreesof
accuracythepropertiesandperformanceofthebuildingasitwillbewhenbuilt
mustbeforeseeninadvance,thatis,theymustbeknowablefromthesurrogate.
Withoutthisforesight,thecommitmentsofresourcesnecessaryateachstageofthe
processcannotbemadewithconfdence.Thedesignfunctionisessentiallyamatter
ofstage-managingaconstantlychangingrepresentationofwhatwilleventuallybea
building,sothatatallstagesoftheprocessthereisinviewaproposalforanobject
The need for an analytic theory
of architecture
44
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
thatdoesnotyetexist,andwhichisprobablyuniquesinceifitwereacopythere
wouldbenoneedfordesignbutwhosetechnical,spatial,functionalandaesthetic
propertiesifandwhenbuiltare,asfaraspossible,predictableinadvance.
Thedesignfunctioninthebuildingprocessthereforeinvolvesonthe
onehandsearchingoutandcreatingarepresentationofapossiblesolutionfor
thedesignprobleminhand,andontheotherthepredictionoftheperformance
ofthebuildingwhenbuiltfromtherepresentation.Theactivitiesthatmakeup
thedesignprocessrefectthisduality.Designessentiallyisacyclicprocessof
generatingpossibledesignproposals,thenselectingandrefningthembytesting
themagainsttheobjectivesthebuildingmustsatisfytobebeautiful,tobecheap,
tobeostentatious,torepresentanidea,torepayinvestment,tofunctionforan
organisationbyprovidingadequateandwell-orderedaccommodation,andsoon.
7

Thesetwobasicaspectstothedesignprocesscanbecalledthecreativephases
andthepredictivephases.Inthecreativephasestheobjectistocreatepossible
designproposals.Inthepredictivephases,theobjectistoforeseehowproposals
willworktosatisfytheobjectives.
Onceweunderstandthecreative-predictivenatureofthedesignprocess,
thenitiseasytoseehowthenormativeandanalyticaspectsoftheoriescan
usefullycontributetotheprocess.Theoriescanbeused,andoftenareused,
tacitlyorexplicitly,intwoquitedistinctmodesinthedesignprocess:asaidsto
thecreativeprocessofarrivingatadesign;andasaidstotheanalyticprocessof
predictinghowaparticulardesignwillworkandbeexperienced.Oftenofcourse
thesetwoaspectswillbeconfatedinaundifferentiatedthoughtprocess.The
normativeaspectsofatheorytellsthedesignerwheretosearchforcandidate
solutionsinthecreativephases,theanalyticaspectshowthesolutionwillwork.For
example,ifyouareaPalladian,theninthecreativephasesofdesignyousearchfor
aformalandspatialsolutionwithPalladianpropertiesacertainrangeofenvelope
geometries,certainsymmetriesofplanandfaade,certainkindsofdetailing,and
soonconfdentthatifyouproceedinaPalladianmannerthenyoucanpredict
aPalladianoutcome.IfyouareaNewmanite,thenyousearchforformaland
spatialsolutionswithacertainlayeringofspatialhierarchies,certainpossibilities
ofsurveillance,theavoidanceofcertainformalthemesandsoon,againconfdent
thatbyproceedingthiswayasafeenvironmentwillresult.Theorythusstructures
thesearchforapossibledesigninasolutionspacethatmightotherwisebeboth
vastandunstructured,anditdoessoinawaythatgivesthedesignerconfdence
whichmayofcoursebequitemisplacedthatthenatureandpropertiesofthe
eventualbuildingcanbeknownfromthetheory.
Theuseoftheoryisofcourseonlyonewayofstructuringthedesign
process.Infactfewdesignersclaimtocreatedesignsfromtheory,andmany
wouldgooutoftheirwaytodenyit.Butthisdoesnotmeanthattheydonotdesign
undertheinfuenceoftheory.Muchuseoftheoreticalideasinarchitectureistacit
ratherthanexplicit.Thisisnotduetomalignintentonthepartofdesigners,but
muchmoretodowiththeneedfortheoryindesign,howeverlittlethisisrecognised.
The need for an analytic theory
of architecture
45
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Consider,forexample,theproblemofprediction.Havingcreatedacandidatedesign,
thedesignernowhasthetaskofforeseeinghowtheunknownnon-discursivities
offormandspacethatwillbecreatedbythedesignwillworkandbeexperienced
whenbuilt.Logicallythereareonlytwopossiblebasesforsuchprediction:known
precedentandtheoreticalprinciple.Predictionbyprecedentmeanspredictionby
referencetoknowncasesthatalreadyexist.Predictionbyprinciplemeansprediction
byreferencetothegeneralityofknowncases.Bothareessentiallyclaimsbasedon
experience,buttheformerisspecifc,andthelattergeneral.
Predictionfromprecedentraisestwoproblems.Theideaofarchitecture
includestheideathatthebuildingtobecreatedwillnotsimplybeacopyofone
whichexists.Thismeansthatprecedentcannotbeusedlockstockandbarrel
forthewholebuilding.Precedentcanthereforeonlybeusedpiecemealfor
aspectsorpartsofthebuilding.Sinceformallyandspatiallybuildingsarecomplex
confgurations,andnotsimplyassemblagesofparts,itcanneverbeclearthat
thenewembeddingofaprecedentattributeorpartwillnotworkdifferentlyinthe
contextofthenewwhole.Theuseofprecedentindesignisnecessary,since
itbringsinconcreteevidenceinsupportofprediction,butitisneversuffcient,
becauseeachnewsynthesisrecontextualiseseachaspectofprecedent.The
useofprecedentthereforenecessarilyinvolvesinterpretation.
Thepressureondesignerstoworkatleastinpartfromknowledgeof
theoreticalprincipleisthereforeintense.Theapparentadvantagetothearchitectof
workingwithinaparticulartheorybecomesthesolutiontothepredictionproblem
appearsalreadytobecontainedwithinthetheory.Thenormativetheoretical
conceptsthatguidethegenerationofacandidatedesignalsotaketheformof
analyticconceptswhichindicatethatifthedesignerfollowsthepreceptsofthe
theory,thenitistobeexpectedthatthedesignwillworkinthewaythearchitect
intends.Theanalyticfoundationsofthenormativetheoryreturnatthepredictive
stagestoappeartoguaranteearchitecturalsuccess.Thisiswhyarchitectural
theoriestaketheformofnormative-analyticcomplexes.Theyfulflthetwoprimary
needsofthedesignprocesswithasinglesetofpropositions.
However,itisclearthattheseadvantageswillonlyexisttotheextentthat
thetheorysanalyticfoundationsarenotillusory.Iftheydonotofferarealistic
pictureofhowtheworldworks,thenitislikelythatthedesignerspredictionswill
referonlytoanillusoryreality.Apoorlyfoundedanalytictheorywillnotinhibitthe
designerinthecreativephasesofdesign,butitwouldleadhimorhertolookinthe
wrongplace.Itwouldalsomeanthatthedesignerspredictionswouldbeunlikelyto
besupportedbyeventswhenthebuildingisbuilt.Thisiswhybadtheoriesareso
dangerousinarchitecture.
8
Theymakedesignappeartobemucheasier,whileat
thesametimemakingitmuchlesslikelytobesuccessful.This,inthelastanalysis,
iswhyarchitectsneedanalyticallywellfoundedtheories.
However,thisisnotthesameastosaythatarchitectssimplyneedscientifc
theoriestoguidethemindesign.Thedualuseoftheoryinarchitecturebothto
generatedesignsandtopredicttheirperformancepermitsustointroduceavery
The need for an analytic theory
of architecture
46
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
importantcomparison:betweentheoriesinartandtheoriesinscience,andtoargue
thatarchitectureneedstheoriesbothinthesensethatthewordisusedinartandin
thesensethatthewordisusedinscience.
Theoriesinartarenotanalytic-normativecomplexesofthekindwetypically
fndinarchitecture.Theyareprimarilyaboutsupportingthecreativeprocess,
thatis,theyareinessenceaboutpossibility.Theoriesinartexpandtherealmof
thepossible,bydefninganewwaytoartorevenbydefninganewformofart.
Thereneedinprinciplebenoconstraintsonwhattypeoftheoriesareused.The
roleofatheoryinartisnottoclaimauniversalart,ortosetuponeformofart
assuperiortoanother,buttoopenuponemorepossiblekindofart.Theoryinart
isthenessentiallygenerative.Itdoesnothavetotakemuchaccountoffunctional
orexperientialconsequences.Itusesabstractthoughtonlytogeneratenew
possibilitiesinartthathadnotbeenseenbefore.
Ifarchitectureweresimplyanart,itwouldneedtheoriesonlyinthesense
thatpaintersorsculptorshavetheories:thatis,asspeculativeextensionsofthe
realmoftheartisticallypossible.Itisclearthatarchitectureasarthasandneeds
thiskindoftheory.Butthisisnotallithasandneeds.Thedifferencebetween
architectureandartisthatwhenanartistworks,heorsheworksdirectlywiththe
materialthatwilleventuallyformtheartobjectthestone,thepaintandsoon.
Whattheartistmakesistheworkofart.Architectureisdifferent.Anarchitectdoes
notworkonabuilding,butarepresentationofabuildingwecalladesign.Adesign
isnotsimplyapictureofabuilding,butapictureofapotentialobjectandofa
potentialsocialobjectthatis,anobjectthatistobeexperienced,understoodand
usedbypeople.Adesignisthereforenotonlyapredictionofanobject,ratherthan
anobjectitself,but,howeverfunctionallynon-specifcitclaimstobe,aprediction
ofpeopleinrelationtobuilding.Thisiswhereanalytictheoriesareneeded,and
analytictheoriesareanalogoustoscientifctheories.Theoriesinsciencearesets
ofgeneral,abstractideasthroughwhichweunderstandandinterpretthematerial
phenomenatheworldofferstoourexperience.Theydealwithhowtheworldis,not
howitmightbe.Becausearchitectureiscreativeitrequirestheoriesofpossibilityin
thesensethattheyexistinart.Butbecausearchitectureisalsopredictive,itneeds
analytictheoriesofactualityaswellastheoriesofpossibility.
Itisthisdoublenaturethatmakesarchitecturaltheoriesunique.Theyrequire
atoncetohavethegenerativepoweroftheoriesinartandatthesametimethe
analyticpoweroftheoriesinscience.Thefrstdealswiththeworldasitmightbe,
thesecondwiththeworldasitis.Thequestionthenis:howmaytherebetheoriesof
architecturewhichareatoncecreativeandanalytic.Oneaspectoftheanswerturns
outtobesimple:goodanalytictheoriesarealreadylikelytobealsogoodtheoriesof
possibility.Theentireusefulnessofscientifctheoriesintheirapplicationsinscience
andtechnologyisinfactfoundedonthesimplebutunobviousfact:thatanalytic
theoriesdonotsimplydescribetheworldasitis,butalsodescribethelimitsofhow
itcanbe.Scientifctheoriesarearrivedatthroughtheexaminationoftheworldas
itis.Butitisexactlythetheoreticalunderstandingoftheworldasitisthatopensup
The need for an analytic theory
of architecture
47
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
wholerealmsofnewpossibilitythatdonotyetexist.
Itisthisfundamentallinkbetweenactualityandpossibilitythatopenstheway
toananalytictheoryofarchitecture.Butbeforeweexploreit,wemustfrstlookalittle
morecarefullyatarchitecturaltheoriestoseehowtheyarestructured,andwhy,and
howtheymighteventuallymoveinthedirectionofbecomingmoreanalytic.
The problem of architectural theory
Themostcommonproblemwitharchitecturaltheoriesisthattheyhavetoooften
beenstronglynormativeandweaklyanalytic,thatis,ithasbeentooeasytouse
themtogeneratedesigns,buttheyaretooweakinpredictingwhatthesedesigns
willbelikewhenbuilt.Thetheoriesofmodernismwere,forexample,quiteeasyto
followingeneratingdesignstosatisfynormativelystatedobjectives.Theproblem
wasthatthearchitecturalmeansproposedwerenotthemeansrequiredtoachieve
thoseobjectives.Thetheorieswereweaklyanalytic.Theydidnotdealwiththe
worldasitactuallyis.Thenormativedominatedtheanalytic.
Exactlyhownormativelystrongbutanalyticallyweakarchitecturaltheories
areheldinplacecanbeseenbytakingonemorestepindisaggregatingwhat
architecturaltheoriesarelikeandhowtheywork.Forexample,lookingalittle
morecloselyatourtwoexemplarsofarchitecturaltheoriestheAlbertianand
theNewmanitewefndbothhavetwoquitedistinctcomponents:oneinthe
realmofbroadintention,tellingarchitectswhattheyshouldaimtoachievethrough
architecture,andoneintherealmofwhatwemightcallarchitecturaltechnique,telling
architectshowtorealisethatintention.Albertistheory,forexample,tellsarchitects
thatinordertodesignbuildingsthatpeoplewillexperienceasharmonious,they
shouldaimtorefectintheirbuildingsthemathematicalorderfoundinnature.He
thengoesontoofferamethodforcalculatingproportionstoserveasatechnique
forrealisingthisaiminarchitecturalterms.
9
Newmantellsarchitectstheyshouldaim
todesignspacesbeyondthedwellingsothatinhabitantsmayidentifywiththemand
controlthem,thenspecifeshierarchicaltechniquesofspaceorganisationinorderto
realisethis.Wemightcallthesethebroadandnarrowpropositionsaboutarchitecture
containedinatypicalarchitecturaltheory.Thebroadproposition,orintention,sets
agoalwhilethenarrowproposition,orarchitecturaltechnique,proposesawayof
designingthroughwhichtheintendedeffectwillberealised.
Onedifferencebetweenthebroadandnarrowpropositionsliesinwhatthey
engage.Thebroadpropositionengagesaworldofideaswhichmaybeverylarge
initsscopeandmaycontainmuchthatispoorlydefnedandlittleunderstood.The
narrowproposition,ontheotherhand,engagestherealitiesofarchitecturaldesign
andexperience.Ifingeneraltheoriesareabstractpropositionswhichengagethe
realworldofexperience,thenthebroadandnarrowpropositionsofarchitectural
theoriesoccupyoppositeendsofthespectrumcoveredbytheories.Thebroad
propositionsareintherealmofphilosophicalabstraction,wherethetheoryengages
thevastworldofideasandpresuppositions,implicitandexplicit,whicheventually
restsnowherebutintheevolutionofhumanminds.Thenarrowpropositionsarein
The need for an analytic theory
of architecture
48
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
therealmofdirectexperienceoftheworldwheretheoriesengagetheminutiaeof
everydayexperience.
Broadpropositionandnarrowpropositionalsodifferintheirintended
universality.Broadpropositionsareintendedtobeuniversalisticinthatthey
attempttosaythingsaboutarchitecturewhichareheldtobegenerallytrue,and
tosayitinsuchabroadwayastoallowittobetrueinquitedifferentarchitectural
circumstances.Butitisclearthatweshouldnotregardthenarrowpropositionsas
universalistic.
10
Forthemostpartthenarrowpropositionsareofferedaspossible
techniquesforrealisinganabstractlystatedaim,nottheonlysuchtechniques.
Onrefection,againthismustbeso.Thenarrowpropositionsofanarchitectural
theoryaretechniquesforbridgingbetweentheabstractandtheconcrete.Onlyan
abstractioncanbegeneral.Weshouldnotmistakeatechniqueforrealising
ananalyticabstractionfortheabstractionitself.
Nowconsiderthesebroadandnarrowpropositionsinrelationtowhatis
requiredoftheoryinthetwophasesofdesign,thatis,inthefrstphase,ideas
aboutpossibleformsand,inthesecondphase,ideasabouttherelationsbetween
formsandperformanceoutcomes.Bothofthetheorieswehavebeenconsidering
appeartosupplybothneeds.Ideasofpossibleformsarecontainedinthenarrow
propositions,thatis,theconstructivetechniquesthroughwhichthetheoristadvises
thedesignertogoaboutdesigntoensuresuccess.InthecaseofAlbertistheory,
thismeansthesystemsofworkedoutproportionswhichguidethedesignerin
settingupthebuildingasaphysicalform.InNewmanscase,thismeansthe
diagramsofspatialhierarchywhichthedesignercanfollowinsettingupthe
spatialdesign.Ideasoftherelationbetweenformandfunctionaloutcomearethen
expressedatthemorephilosophicallevelofthebroadpropositions.InAlbertis
case,thismeansthebroadpropositions,basedontheanalogywithmusic,about
thehumanexperienceofvisualharmony.
11
InNewmanscase,itmeansthebroad
propositionsabouthumanterritorialityanditsspatialimplications.
12
Inotherwords,
inbothcases,itisthehighlyspecifcnarrowpropositionswhichguidethecreative
processofdesign,andtheverygeneralisedbroadpropositionswhichguidethe
designerinpredictingfunctionaleffectfromformalconfguration.
Nowtheproblemwithmostarchitecturaltheoriesisthatthisisexactly
theoppositeofwhatisrequiredforarchitecturewhichiscreativelyinnovative
andfunctionallysuccessful.Inthegenerativephaseofdesign,whatisneeded
ifarchitecturalcreativityistobemaximisedisideasaboutformalandspatial
confgurationwhichareasunspecifcaspossibleaboutspecifcsolutions,in
ordertoleavethesolutionspaceasopenaspossibletocreativeinvention.Inthe
predictivephases,whatisneededisprecisionaboutspecifcformssincewhatis
atissueisthepredictionofthefunctionaloutcomeofthisorthatrealdesign.In
thegenerativephases,wherewhatisrequiredareabstractorgenotypicalideas
whichopenuprealmsofpossibilityjustastheoriesdoinart,architecturaltheories
ofthistypeofferarathernarrowrangeofsolutiontypeswhichareessentiallyno
morethanasetofabstractexemplarstofollowparticularsystemsofnumerical
The need for an analytic theory
of architecture
49
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
proportionsinonecase,particulardiagramsofhierarchicalspatialrelationsinthe
other.Thenwheninthepredictivephasesofdesignthedesignerneedsamuch
greaterdegreeofanalyticprecisioninordertoforeseehowthisorthatinnovative
formwillworkfunctionallyorexperientially,allthatthetheoriesofferisthevague
analyticgeneralisationsofthebroadpropositions.
Inotherwords,architecturaltheoriesofthistypeareover-specifcwhere
theyshouldbepermissiveandvaguewheretheyshouldbeprecise.Thedesigner
isgivenconcretemodelstofollowwhenheorsheneedsconstructivecreative
ideastosearchthesolutionspace,andvacuousabstractionswhenheorsheought
tobegiventechniquestopredicttheperformanceofparticulardesigns.Thisis,
inanutshell,theproblemwithmostarchitecturaltheories,andthisishow,inreal
design,thenormativeaspectsoftheorycometodominatetheanalytic.Whatis
neededaretheorieswiththereverseproperties,thatis,theoriesthatareasnon-
specifcaspossibletoparticularsolutionsinthegenerativephasesofdesignin
ordertoleavethesolutionfeldaslargeanddenseaspossible,andasspecifcand
rigorousaspossibleinthepredictivephasesinordertobeabletodealpredictively
withunknownformswheretheneedforeffectivepredictionisgreatest.The
implicationofthisisthatweneedafullyfedgedanalytictheorywhichwouldoffer
abstractunderstandingratherthanspecifcmodelsinthecreativephasesofdesign,
andphenotypicalprecisionratherthanvaguegeneralisationsatthetestingstages.
What exactly, are theories?
Howshouldwegoaboutsettingupsuchatheory?Thefrststepmustbetomake
sureweunderstandexactlywhatananalytictheoryis.Thisturnsouttobenotas
easyaslookingthewordupinadictionary.Fewwordsareinfactmoreambiguous
intheiroriginsthantheory.InitsancientGreekorigins,theverbtheoreeinmeans
tobeaspectator,andtheproductsofthisspeculativeactivity,theoremata,were,
notsurprisingly,speculations.ForBacontheoriesweresimplyerrors,thereceived
systemsofphilosophyanddoctrine,tobereplacedinduecoursebysomething
altogetherbetter.
13
Thismeaningisstillrefectedineverydayuse.Incommonusage,
theoriesarespeculations,oflesserstatusthanfacts,atbestatemporaryfxuntil
thefactsareknown.Afctionaldetectivewithaprematuretheoryaboutacase
willalmostcertainlybeshowntobewrong.Theexpressiononlyatheoryclearly
expectstheorynottobeeventuallysupportedbyfacts,buttobereplacedbyfacts.
Inthesesenses,theoriesembodyirremediableuncertainties,andappeartoconstitute
aformofthoughtwhoseobjectistoreplaceitselfwitha-theoretical,andtherefore
secure,knowledge.Incompletecontrast,inmodernsciencethewordtheorytoday
standsforthedeepestlevelofunderstandingofphenomena.Successfultheoriesin
areaswherenonehadpreviouslyprevailed,likeevolutiontheory,arethemostepoch
makingofintellectualevents.Confictsbetweenrivaltheoriesof,say,theoriginsofthe
universeorthenatureofmatter,conductedontheobscurebattlefeldsofmacroand
microphenomena,areamongtheepicsofthelatetwentieth-centurythought.
Sowhatthenistheory,thatitcanbesubjecttosucharangeof
interpretationsandambiguities?Thesourceofthisambiguityliesofcoursenot
The need for an analytic theory
of architecture
50
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
inthevagariesofetymologicalhistorybutinthenatureoftheoriesthemselves.
Theoriesarefoundintherealmofspeculativethought,becausetheyareatroot,
speculations.Theyarenotinthemselves,forexample,statementsaboutobservable
phenomena,norevenstatementsabouttheregularitiesthataretobefoundin
observablephenomena.Theyarepropositionsabouthypotheticalprocesseswhich
mightberesponsiblefortheregularitiesweseeinphenomena.Assuchtheyhave
anecessarilyabstractnature,andarepurelyconceptualentities.Youcannotseea
theory,onlyitsconsequences,soyoucannotverifyatheory,onlyphenomenathat
areconsistentwithit.Whenwetestatheorywedonotsimplylookatthetheoryto
seeifallthepartsareinworkingorderandproperlyrelated,thoughwedoalsodo
this.Wecheckthetheorybyseeinghowfarthephenomenaavailableinthereal
worldareconsistentwiththetheory,andpreferablywithnoother.Tocheckatheory,
ineffect,welookawayfromthetheory.Theoriesareinthemselvesunobservable
andunexperiencable,andthisiswhyintheendeventhebestandthemostdurable
remaininsomesensespeculative.
Butevenwhenweaccepttheabstractandspeculativenatureoftheories,
wehavenotyetexhaustedtheapparentindeterminacyoftheidea.Nosetof
conceptswhichbecomepartofatheorycanexistinisolation.Onthecontrary,
conceptscanonlyexistaspartofconceptualschemesthroughwhichweinterpret
ourexperienceoftheworldandturninformationintoknowledge.Noconceptorset
ofconceptscanexistinavacuum.Eachmustbeembeddedinabroaderrangeof
propositionsorassumptionsaboutwhattheworldislikeandhowitworks.These
broaderframeworkshavebeenknownasparadigmssinceThomasKuhnfrstdrew
attentiontotheirexistence.
14
Withallthisindeterminacyinwhatwemeanbytheory,howisitthat
theycanbesoimportantandsouseful.Toanswerthiswemustunderstandthe
circumstancesinwhichtheoriesariseandwhatpurposestheyserve.Theorisation
beginswhenwenoteacertaintypeofphenomenonandmakeacertaintypeof
presupposition.Thephenomenonwenoteisthatofsurfaceregularityintheworld
asweexperienceit.Thepresuppositionwemakeisthatsurface regularityimplies
underlying invarianceintheprocessesthatgiverisetothephenomenawesee.
Thefrstofthesethenotingofregularitiestheorisationshareswith
language.Thefactthatlanguagehaswordsforclassesofthingsratherthan
simplyforindividualthingsassumesthatweknowthedifferencebetweenorder
andchaos,thatis,thatwecandiscernintheobjectiveworldstructuralstabilities
15
whicharesuffcientlywelldefnedandrepetitioustosupporttheassignmentof
names.Thesenamesare,asphilosophershaveendlesslynoted,abstracttermsfor
classesintheguiseofnamesforthings,withtheconsequencethatevensucha
simpleapparentlyconcreteactofpointingatathingandnamingitdependsonthe
priorexistencenotonlyoftheabstractuniversalconstitutedbythatclassname,but
alsooftheschemeofsuchabstractionsofwhichthatparticularabstractionforms
apart.Theseschemes,aswehaveknownsincedeSaussure,
16
differfromone
languagetoanothersothatwearecompelledtoacknowledgethatnamesarenot
The need for an analytic theory
of architecture
5
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
neutral,simplehandlesonthings,butconceptualinstrumentsbywhichwecreate
anorganisedpictureoftheworld.Namescreateunderstanding,anditisagainstthe
backgroundoftheorganisedpictureoftheworldalreadygiventousbylanguage
andculturethattheorisationbegins.
Theorybeginsinthesameplaceaslanguagewherewenote,inthefuxof
experience,regularities,butaddsafurtherpresupposition:thatsinceregularityis
unlikelytobetheproductofchance,theremustbesomekindofordernotonlyin
theregularphenomenathatweobservebutalsointheprocessesthatgiveriseto
thephenomena.Whyweshouldmakethispresuppositionisnotclear.Butitseems
plausiblethatjustaslanguageseemsintimatelyboundupwithhowwecognisethe
worldsotheorisationisboundupwithhowweactintheworld.When,forexample,
westrikestonestomakesparksandthenfre,thesequenceofeventsfromoneto
theotherisnotinscribedonthesurfaceofthingsbutimpliessomeinteriorprocess
whichissetinmotionbyouractions.Justastheworldrespondstoouractions
onitbyproducingregularities,sowepresupposethattheexistenceofregularities
whichdonotresultfromouractionsmustbetheresultofinvariantprocesses
analogoustoouractions.Ifthenlanguagearisesfromourbeingintheworldand
needingtoknowitsobjectivepersistences,sotheorisationseemstoarisefromour
actingintheworldandontheworldandneedingtoknowtheinteriorprocessesby
whichoutcomereliablyfollowsfromaction.
Wethusseethatregularitiesarethestartingpointoftheory,buttheyarenot
thetheoryitself.Regularitiesinitiatetheprocessoftheorisationsinceweinferfrom
theexistenceofregularitiesthattheremustbesomeinvariantstructureinwhatever
processitisthatproducesthesesurfaceregularities.Theoriesareconcernedwith
thenatureofthatprocess,morepreciselytheyareattemptstomodeltheinvariant
structureofprocesseswhicharethoughttoexistfortheretobesurfaceregularities.
Atheory,then,isnotalistofregularities.Regularitiesarewhattheoryseeksto
explain,butarenotinthemselvestheory.Theyinitiatethesearchfortheorybutare
notandcannotbeitsendpoint.Atheorywhichseekstoexplainregularitiesisan
entityofanaltogetherdifferentkindfromalistofregularities.
Moreover,althoughtheorisationmovesonfromlanguagebyseekingto
identifythehiddenprocessesthatgiverisetosurfaceregularities,itdoesnotbegin
inaconceptualorlinguisticvoid.Itbeginsintheonlyplaceitcan,intheevolution
ofthoughtandlanguage,andtheirrelationtothespace-timephenomenathat
weexperiencewithouttrying.Becausethoughtandlanguagealreadygiveusa
pictureoftheworldwhich,atsomelevelatleast,seemstorefectitsorderand
thereforetoexplainit,wearecompelledtoacknowledgethatwhenwebeginthe
processoftheorisationwearealreadyinpossessionofaviewoftheworldwhich
inmanywaysisverylikeatheory,inthatitmakestheworldseemamoreorless
coherentandorganisedplace.Thedifferenceisthatthetheory-likeunderstanding
weacquirefromcultureandlanguagerefectsnotaninteriororderwhichgivesrise
tothesurfaceregularitiesbutanorderinthosesurfaceregularitiesthemselves.
Whenforexamplelanguagetellsusthatthesunrises,itrefectstheregularities
The need for an analytic theory
of architecture
52
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
thatwenoteonthesurfaceofthings,notahiddenprocesswhichgivesriseto
thissurfaceregularity.Wemightusefullythenthinkofsucheverydayconstructions
astheoryintheweaksense.Analytic,orscientifctheories,aretheoriesinthe
strongsense.Theyaimatagreatertruthbecausetheyseeknottobringorderto
surfaceregularitiesbuttoshowhowthosesurfaceregularitiesarisefrominvariant
necessitiesburieddeepinthenatureofthings.
Formally defning simple regularities
Becausesurfaceregularitiesaretheobjectoftheory,thefrststepintheorisation
istoformalisetheidea.Infact,thereisabeautifullysimplewaytoextracttheidea
ofregularityfromphenomenaandrepresentitaspureregularity,independentof
theoverallqualitativenatureofthings.Theideaisthatoftranslatingtheproperties
ofobjectsintheworldasweseetheminrealspaceintoanabstractspacewhich
allowsustobequiteclearaboutwhatthesepropertiesare.Thisisdonebythe
familiartechniqueofreplacingthespacewithinwhichtheobjectexistswithan
abstractco-ordinatesysteminwhichtheaxesrepresentthosepropertiesofthe
objectthatseemtobeofinterestasregularities.Thusoneco-ordinatemight
representtheheightoftheobject,anotherthelengthandanotherthebreadth.
Wemaythenrepresentanyobjectwhichhasthesepropertiesasasinglepoint
inthepropertyspace.
Oncewecanrepresentthepropertiesofanobjectasapointina
propertyspaceratherthanasthatsetofactualpropertiesinrealspace,wecan
easilyrepresentexactlywhatwemeanbyaregularityasfarastheseproperties
areconcerned.Forexample,totheextentthatthingsarecomparabletoeachother
inmorethatonepropertyinthepropertyspace,thepointsrepresentingthemin
thepropertyspacewillclusterinaparticularregionofthespace.Clustersinthe
propertyspacegiveaformalmeaningtotheideaofatypeorclassofthings,inso
farasthosepropertiesareconcerned.Ifthingswhenrepresentedaspointsinthe
propertyspacearerandomlydistributedthroughoutthespace,thatis,ifthereareno
clusters,thenwewouldsayeitherthattherewerenotypes,butonlyindividuals,or
thatwehadselectedthewrongpropertiesforanalysis.Ifontheotherhandwesee
clusters,weinferthatthingstendtofallintotypes,bywhichwemeanthatvariation
ononepropertytendstobeassociatedwithvariationonatleastoneother,or
perhapsmanyothers.Thisisshowngraphicallyinthetoptwodiagramsoffgure
2.1.Wemayequallyusethepropertyspacetoformalisetheideathattheregularity
weseeliesnotinapparentclassesortypesofthingsbutsequencesofstates
ofthings.Inthiscaseweask:whenanentitychangesononedimension,does
itchangeinanyother?Ititdoes,thentheregularitywillshowitselfasaregular
patterninthedistributionofentitiesinthepropertyspace.Thisisshowninthe
bottomtwodiagramsinfgure2.1.Whenweseesuchapattern,wewouldinferthat
someprocessifnotofcauseandeffectthenatleastofregularco-variationwasin
operation,sinceeachtimeonevariablewaschangedachangeinanothervariable
regularlyappeared.
The need for an analytic theory
of architecture
53
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 2.1
extensiOn
OrthOgOnal
tO surlace
extensiOn
parallel tO
surlace
We lirst create a cOOrdinate systen in which vertical axis represents
increasing size OrthOgOnal tO the surlace and the hOrizOntal axis increasing
size parallel tO the surlace. We then draw in the space shapes cOnlOrning
tO the cOOrdinates. n the bOttOn lelt cOrner we start with an initial square
shape. This becOnes extended vertically as we nOve upwards, laterally as
we nOve hOrizOntally and in bOth dinensiOns il we nOve in bOth directiOns.
Eecause the prOperties Ol lateral and vertical expansiOn are represented
inthe axes, then the pOints in the prOperty space represent the changes in
spape in the lelt ligure nO less than the actual shapes thenselves. ther
prOperties nOt in the cOOrdinate systen wOuld Ol cOurse be Onitted. The
pOints represent Only the prOperties selected lOr representatiOn in the
prOperty space. The idea Ol types Or classes Ol shapes can thus be
shOwnby pOint clusters.
/ssOciated changes can alsO be represented as pOints in the prOperty space.
The pOsitiOn Ol pOints indicates that a change in One dinensiOn is always
assOciated with a change in the Other. This representatiOn is alsO knOwn as
a 'scatergran'. The degree tO which pOints lOnr a line lrOn bOttOn lelt tO
tOp right can be indexes by a 'cOrrelatiOn cOellicient', a value between 0
and ! which indicates hOw strOngly One change inplies the Other.
n the abOve case, the shapes drawn in the cOOrdinate space shOw that
an increase On One dinensiOn is always related tO an increase in the Other.
This type Ol regular assOciatiOn between changes is the surlace phenOnenOn
we assOciate with 'cause and ellect'One change seens tO bring abOiut anOther.
Eut what is shOwn in the ligure is nOt 'cause andellect', nerely the regular
assOciatiOn Ol changes. TO 'explain' these regularities we wOuld need tO shOw
why they are necessary.
extensiOn
OrthOgOnal
tO surlace
tall, thin
shapes
large
square
shapes
snall
square
shapes
lOw, lat
shapes
extensiOn
parallel tO
surlace
extensiOn
OrthOgOnal
tO surlace
extensiOn
parallel tO
surlace
extensiOn
OrthOgOnal
tO surlace
extensiOn
parallel tO
surlace
Figure 2.1
The need for an analytic theory
of architecture
54
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Wemightthenreasonablysaythatquestionsabouttypes,thatis,aboutsimilarities
anddifferences,arequestionsoftheform:doentitiesclusterinparticularregionsof
thepropertyspace?;whilequestionsaboutcauseandeffectareoftheform:when
entitiesmoveinonedimensionofthepropertyspacedotheymoveinanother?
17

Bothofthesedescribetheapparentregularitiesofsurfacephenomena,thatis,the
appearanceoftypesandtheappearanceofcauseandeffect,inanabstractway.The
propertyspaceisameansofcontrollingtheattributesthataretobeaccountedfor
inthepatternofsimilaritiesanddifferences.Wheretherealobjectispresent,allits
propertiesaremanifest.Inthepropertyspace,onlyselectedpropertiesarepresent.
Ofcourse,everythingdependsonourselectingtherightpropertiesfortheproperty
spaceinthefrstplace.Forthisreasonwecanneverbesurefromtheabsenceofa
regularitythatnoregularitiesarepresentinthesephenomena.
Butevenifwegothroughalongprocessofexperimentingwithdifferent
propertiesuntilweeventuallyfndtheclustersorcovariationsthatindicatethe
presenceofregularities,itwillalwaysstillbethesurfacephenomenathatare
representedregardlessofthedegreeofabstraction.Wearestillseeingthesurface
ofthings,thatis,apparentregularitiesofthingsaspresentedtoourexperience.
Wearenotseeingthetheorythatpurportstoaccountforthoseregularities,thatis,
wearenotseeingthemodelofthestructuresoftheprocesswhichmightaccount
fortheseregularities.Whatwearedoingisrecordingphenomenainsuchaway
astobeabletoseeclearlywhatwemeanbyregularities,bytranslatingproperties
intothedimensionsofacoordinatespaceandlocatingobjectsaspointswithin
thisspacesothatonlytheregularpropertiesarerepresentedinwhatwesee.This
bothseemstobeandisafundamentalwaymaybethefundamentalwayof
rigorouslyrecordingsimilaritiesanddifferences,andconstantassociationsbetween
things,withinanobjectiveandindependentframework.
Themeaningofthewordtheorycanthenbemadeprecise.Aswehavesaid,
justasthe a priori givenforthenotingofregularitiesisthatweknowthedifference
betweenorderandrandomness,the a priori givenfortakingthisintotheorisationis
thatregularityonthesurfaceimpliessomesystemicprocessbelowthesurface,such
thatthestructureofthatsystemisinsomesenseinvariant.Atheoryisanattemptto
modeltheseinvariantsinasystemofinterdependentconcepts.Atheoryisamodel
becauseitdealswiththewayinwhichthingsmustbeinterrelatedinordertoproduce
thesurfacephenomena,andabstractbecauseitrepresentsthesystembysome
meansotherthanthatofthesystemitself.Atheoryisamodel,butnotinthesense
thataphysicalmodelisamodel,thatis,asmallcopyofthethingitself,butinthe
contrarysenseofamodeltakingasabstractaformaspossible,uncommittedtoany
particularkindofrepresentationorembodiment.Initspurestform,atheoryisakind
ofabstractmachine,sinceitisanattempttocreateanabstractrepresentationofthe
workingofprocesseswhichgiverisetowhatwesee.
Nowtheenormouspoweroftheoriesarisesfromoneveryspecifcproperty
ofsuchabstractmachines,apropertywehavealreadytouchedupon.Because
theoriesareabstractworkingmodelsofprocesseswhichgiverisetotheactual,
The need for an analytic theory
of architecture
55
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
theyalsogiveabasisforconjecturingaboutthepossible.Theoriesineffectallow
ustogobeyondtheaccumulatedexperienceofrealityandconjecturepossible
statesofrealitythatarecompatiblewiththemodel.Itisthislinkbetweentheactual
andthepossiblethatmakestheoriessousefulforprediction.Toapplyatheoryis
essentiallytoposethequestion:iswhatisproposedapossiblecase?
Itistoolimitingthentocalltheoriesexplanationsofhowtheworldis.
Atheorydefnestheinvariantsthatunderliemanydifferentstatesofreality.Itisin
principleunlikelythatallpossiblestatesofaparticularsetofphenomenaalready
existorarealreadyknown.Itislikelythenthatthetheorywillalsopredictpossible
statesthatdonotexistbutcouldaccordingtothemodel.Itisthispropertyabove
allothersthatimpartstotheoryitsimmensepowerasatoolofthoughtandasan
agentofhumancreativity,andalsoitspracticalusefulness.However,itisclearthat
thesevirtueswillariseonlytothedegreethatthetheorycapturesinvariantsthat
reallyareoutthere.Buthowcanthisbe?Howcananabstractioncapturewhat
isreallyoutthere.Totakethisnextstep,wemustknowalittlemoreabouthow
theoriesareputtogether,howtheywork,andwhattheyaremadeof.
What are theories made of?
Thefrstthingwemustnoteisthattheoriesaremadeofconcepts,usuallyinthe
formofasystemofinterdependentconceptswithtwoformsofexpression:words,
andformalexpression,usuallymathematical.Sinceeverydaylifeandlanguageis
alsorunonconceptswemustknowthedifferencebetweenascientifcconcept
andanunscientifcone.Whatthenisthedifference?Wecandonobetterthanto
discusstheconceptsonwhichbothlanguageandscienceseemtobefounded,
thatis,thedifferencebetweenorderandrandomness.
Orderandrandomnessarebothconceptswhichhaveapowerfulintuitive
meaning.Bothareverybroadindeedintheirapplication,somuchsothatitisvery
hardtopindownwhatthetwotermsmeanwithanyrealclarity.Bothterms,and
evenmorethewaytheyarerelated,expresscomplexintuitionsaboutthewaythe
worldis.Eachtermcanbeusedinawiderangeofsituations,andthemeaning
onlybecomesclearenoughtofeelunderstoodinthespokenorwrittencontext.
Thisiscommonenough.Theintuitiveconceptsthatpervadeandgivesenseto
ourlanguageshavethisrichnessandimprecision,sotheycanbeusedinagreat
varietyofsituations,andindeeditisonlyinthecontextinwhichaconceptisused
thatitsmeaningbecomesunambiguous.
Inscience,itisexactlythisrichnessandimprecisionthatisrestricted.
Scientifcconcepts,althoughexpressedinlanguage,aremuchnarrowerintheir
potentialapplicationthannormallinguisticconcepts.Buttheyarealsomore
systemic,inthattheycompressandexpressmoreinterrelationshipsbetween
concepts.Theyexpressmoreconnectionbetweenthings,butatthecostofa
narrowingoftherangeofapplication.Theconceptofentropyisagoodexample
ofthisbecauseitrelatesbothorderandchaosinasystemicway,andindoingso
restrictstherangeofapplicationofthenewsyntheticconcepttothosesituations
The need for an analytic theory
of architecture
56
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
wherepreciselythesesystemicrelationshold.Thedegreeofentropyinasystem
describesthatsystemspositioninacontinuumfromordertochaos.Likemany
scientifcconceptsofgreatprofundityandgeneralityitcanbeexplainedsimply,
thoughnotthroughwordsbutthroughasimplemodel.
18
Imaginetwojars,aand
b,withacontaining100ballsnumbered1100,andbempty,andsomesystemfor
selectingarandomnumberbetween1and100say,apointeronaspindlewhich
canbespunsothatitlandswithequallikelihoodonthenumberssetoutina
circle.Spinthepointerandwhenthepointrestsonanumber,fndtheballwiththat
numberandtransferitfromwhicheverjaritisintotheotherone.Thenrepeatthis
operationasmanytimesasnecessary.Whathappens?Intuitivelyandcorrectly
wesaythattheprocesswillsettledowntoabouthalftheballsineachjar.Why?
Theanswertellsuswhatentropyisandhowitcanbemeasured.Thefrsttimethe
pointerselectsanumber,theprobabilitythattheballselectedwillgofromatobis
1,thatis,itiscertain,becausealltheballsareina.Thesecondtime,thereisone
chancein100thatthesingleballinbwillreturntoa,but99chancesoutof100,that
is,aprobabilityof.99,thatanotherballwillgofromatob.Thenexttimethereis
onechancein50thatoneoftheballsinbwillreturntoa,but98chancesoutof100
thatanotherballwillgofromatob.Clearly,astheprocessgoeson,thechancesof
ballsgoingbackfrombtoagraduallyincreaseandthechancesofballsgoingfrom
atobdiminishcorrespondingly.
Whenabouthalftheballsareineachjar,theprobabilitiesareaboutequal,
sothesystemtendstosettledowntosmallvariationsaboutthisstate.Toseewhy
thishappensletusdefneamicrostateofthesystemasaparticulardistributionof
individualballsinjarsandamacrostateasaparticularnumberofballsineachjar.
Thereare,clearly,only200possiblemicrostatesofthesystemforthemacrostatein
whichoneballisinonejarandtherestareintheother,thatis,oneforeachofthe
hundredballsineachjar.Forthemacrostatewithtwoballsinonejarand98inthe
otherthereareallpossiblecombinationoftwoballsforeachjar,thatis,200200.
Forthemacrostatewiththreeinoneand97intheotherthereareallcombinations
ofthreeballs.Inotherwords,thenumberofmicrostatesforthemacrostateis
maximisedwhenthelargestpossiblenumberareintheleastfulljarthatis,when
halftheballsareineachbecausebeyondthatpointtherewillbefewerballsin
theotherjarandeverythinghappensinreverse.
Thisiswhythesystemtendstothehalfandhalfstate.Therearefarmore
microstatescorrespondingtothehalfandhalf(ornearhalfandhalf)macrostates
thanforthoseinwhichafewballsareinonejarandmanyintheother.Inother
wordsallthesystemdoesistotendtoitsmostprobablestate.Thisisalsothe
defnitionofthestateofmaximumentropy.Entropyismaximalinasystemwhen
thesystemisinoneofthemacrostatesforwhichtherearethelargestnumberof
microstates.Anexampleofthis,iswheretwogasesareeachrandomlydistributed
inacontainer,withoutregionswhereoneorothergaspredominates.Therearefar
moremicrostateswithrandomdistributionthanmicrostateswithconcentrationsof
oneorothergasinacertainregion.Ourmodelofjarsandballsisthenastatistical
The need for an analytic theory
of architecture
57
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
representationofthemixingoftwogasesinaclosedcompartmentorforthe
gradualheatdeathoftheuniverseastheuniversetendsfromitscurrentimprobable
statetoitsmostprobablestate,thatis,oneinwhichheatismoreorlessevenly
dispersedthroughouttheuniverse.
19
Inotherwords,entropyrelatesthenotionsoforderandchaosintoasingle
concept,butatthesametimegivesitamuchmorepreciseandlimitedreference
totheworld.However,italsodoessomethingelseofnolessimportance.Itpermits
theconcepttobecapturedinaformalmathematicalexpressionaswellasthrough
words.Itisthroughthisformalexpressionthatthelinkbetweentheconceptand
theobservableworldismade.Thistwo-wayemancipationofconcepts,onthe
onehandreorganisingconceptsintomoreprecisesystemsofinterdependence
andontheotherrelatingthemtotherealworldbyassociatingthemwithformal
expressionsistheessenceofwhattheoriesare.
Theoriesarethereforemadeoftwothings:wordsandformalexpressions.
Butbothrepresentconcepts.Atheoryisasystemofconceptswithonetypeof
expression,theverbal,whichlinkstheconceptsbackintoourunderstanding,
necessarilywithsomeimprecision;andanother,mathematicalformwhichlinksthe
conceptsforwardintophenomena,necessarilywithgreatexactness.Theoriesthus
linkourunderstandingtotheworld,connectedtoourunderstandingbylinguistic
conceptsandconnectedtophenomenabyformalexpressionscorresponding
totheconcepts.
Thistwo-wayrelationusinglanguageandformalismtolinkconceptsto
ourunderstandingontheonehandandtotherealworldontheotheristheheart
ofwhattheoriesare.Wemayclarifyallthesecomplexrelationsinadiagram,see
fgure2.2.Thisfgureshowsnotonlyhowtheoriesintervenebetweenlanguageand
theworld,butalsohowsciencerelatestophilosophy,whichoverlapswithscience
inpartofthisoverallscheme.Theoverallformofthediagramsetstheevolution
oflanguageandideasontheleftandthephenomenaofspace-timeontheright.
Theoriesareinthecentre,defnedasarelationbetweenasystemofconceptsand
asystemofformalexpressionswhichlookstwoways:throughtheconceptsitlooks
Figure 2.2
$/-!)./&0(),/3/0(9
THEEVOLUTION
OFLANGUAGE
ANDIDEAS
PARADIGMS
THELANGUAGESIDE
WEAKRELATION STRONGRELATION
4(%/2)%3 THEPHENOMENASIDE
THECONCEPT
SYSTEM
THEFORMAL
SYSTEM
REGULARITIESIN
SPACETIME
PHENOMENA
UNORDERED
SPACETIME
PHENOMENA
$/-!)./&3#)%.#%
Figure 2.2
The need for an analytic theory
of architecture
58
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
backfrstintothebroaderconceptualschemeswecallparadigms,thenintothe
evolvingstructureoflanguageandideaswhicharebothaninevitablecontextand
aninevitableconstraintontheorisation;andthroughtheformalismitlooksforward
towardstheregularitiesinspace-timephenomenawhichtheoriesseektoaccount
for,andthenonwardsintothegeneralforegroundofspace-timephenomenawhich
donotformpartoftheregularitiesbutwhichmayatanystagearbitrarilyengagethe
theorybyofferingphenomenawhichareinconsistentwiththeabstractmachinefor
generatingphenomenaproposedbythetheory.
Theearliestancestorsofwhatwewouldrecogniseasscientifctheories,
suchasthoseofthePythagoreanswhoaresaidtohavefrstnotedtherelation
betweennumericalratiosandformsoccurringinnature,areprobablybestseenas
paradigmsratherthanasfullyfedgedtheories,althoughintheirpreoccupationwith
therelationbetweenspace-timeregularitiesandformalexpressiontheycertainly
prefguretheoriesinthemodernsense.
20
Pythagoreanism(asweearliernotedas
infuencingAlberti)isageneralisationofasingleconceptwhichgeneratedawayof
lookingattheworldonthebasisofafewresults.Thisislegitimatelyaprecursorto
theorybutnotinitselfwhatweoughttobecallingatheory.Howevertheattraction
ofsuchover-generalisationremains,asisseenintheprevalenceofvariantson
Pythagoreanisminthemysticalsubstitutesfortheorywhichhavecontinuedto
occupythefringesofarchitecturalthoughtthroughoutthetwentiethcentury.
21
Theoriesinthescientifcsenseareonestepinfrombothparadigmson
theonesideandregularitiesontheotherinthattheyarecomposedofconcepts
whicharefocusedandrelatedtoeachothertoformasystem,withpreciserelations
betweeneachconceptandformaltechniquesorexpressionswhichareusedto
checkhowfartheregularitiesimpliedbythesystemofconceptsaredetectable
inspacetimephenomena.Scientifctheoriesthusrequirethreerelationstobe
particularlystrong:therelationsamongconceptswhichformtheconceptualsystem;
therelationsbetweenconceptsandformaltechniquesofmeasurement;andthe
relationsbetweentheseformaltechniquesandspace-timephenomena.Intermsof
thediagram,wemaysaythenthatscienceneedstobestrongfromtheconcept
systeminthedirectionofphenomena.
Scienceis,andmustexpecttobe,weakerintheotherdirection,thatis,
inthepassagebackthroughparadigmsintothemoregeneralevolutionofideas.
Thistendstobegroundoccupiedbyphilosophy.Philosophyoverlapswithscience
inbeinginterestedintheories,andrelatingthembacktobroaderfamiliesof
concepts
22
rightthroughtothosethatprevailineverydaylifeandsocialpractices,
23

butdoesnotnormallypreoccupyitselfwiththerigoroustestingoftheoriesagainst
realspace-timephenomena.Scienceandphilosophyarerivalsintherealmof
theory,butonlybecausetheirpreoccupationsreachoutfromtheoryincontrary
directionswiththeeffectthatbetweenthemscienceandphilosophycoverthe
groundthatneedstobeoccupiedbytheoreticalthought.However,itisbecause
sciencemovesfromconceptstophenomenathatitstheorieseventuallycometo
haveapuzzlingstatus,becausetheintuitivesensethattheyexplainthingscomes
The need for an analytic theory
of architecture
59
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
fromtherelationbetweentheconceptsthatmakeupthetheoryandthesensewe
havefromeverydaylanguagethatourideasexplaintheworld.Scientifctheories
areinthissensepsychologicallystrongestwheretheyareinfactweakest,thatis,
wheretheconceptsthatformthetheoryrelatebackintothebroaderconceptual
systemswhichinformeverydaylife.
24
Towards an analytic theory of architecture
Giventhesedefnitions,howthencantherebeananalytictheoryofarchitecture?
frst,letusbecompletelyclearaboutonething.Iftherearenoobjectiveregularities
intherealworldofarchitecturalformandspace,linkingtheconfgurationalaspects
offormandspacewithbehaviouralandexperientialoutcomes,thenthereareno
groundswhatsoeverforseekingtobuildananalytictheory.Theneedforandthe
possibilityofananalytictheorybothstandorfallwiththeexistenceofsuchnon-
discursiveregularities.
Thismeansthattobuildananalytictheory,non-discursiveregularitiesmust
frstbeinvestigatedand,iftheyexist,broughttolight.Howcanthisbedone?We
mayfrstrecallthatanarchitecturaltheoryisanattempttorenderoneorotherof
thenon-discursiveaspectsofarchitecturediscursive,bydescribingnon-discursivity
inconcepts,wordsandnumbers.Wemaysaythatanarchitecturaltheoryseeksto
createanon-discursivetechnique,thatis,atechniqueforhandlingthosematters
ofpatternandconfgurationofformandspacethatwefndithardtotalkabout.
Inresearchtermswecouldsaythatanarchitecturaltheory,atleastinthenarrow
aspectsthroughwhichitdescribesandprescribesdesigndecisions,isanattempt
tocontrolthearchitecturalvariable.
Now,aswehaveseen,architecturaltheoriesinthepasthavetendedto
bestronglynormativeandweaklyanalytic,becausethenon-discursivetechniques
proposedareonlyabletodescribecertainkindsofconfguration.Thisiswhyin
applicationtheyarepartisanforthatkindofconfguration.Forexample,ifanon-
discursivetechniquedescribessystemsofproportionintermsofnumericalor
geometricratios,itisunlikelytobeabletodealwithconfgurationswhichlack
suchproportionality.Itwillonlydescribethosecaseswheretheseproportionshold.
Inanyattempttoapplysuchpartisantechniquesgenerally,theyaremorelikely
thereforetoactasdistortingmirrorsthanadiscoveryofnewregularities.Likewise,
ifournon-discursivetechniqueisasystemofdiagramsexpressingspatialhierarchy,
itisunlikelythatthosetechniquescanbeusefullyappliedtothevastrangeofcases
wheresuchclearhierarchisationisnotfound.Itfollowsagainthatsuchatechnique
willbeuselessforinvestigatingspatialpatternsingeneral.
Wecansaythenthatanon-discursivetechniquewhichispartisanfor
usuallybecauseitisaproductofapreferenceforoneparticularkindofnon-
discursivity,willnotbeusableasananalytictool,andcannotthereforebeusedfor
thediscoveryofnon-discursiveregularities.Thisdefciency,however,doespoint
usinthedirectionofwhatisneeded.Tobringtolightnon-discursiveregularities,
weneednon-discursivetechniquesforthedescriptionofeitherspatialpatterns
The need for an analytic theory
of architecture
60
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
orformalpatterns(orconceivablyboth)whichareuncommittedtoanyparticular
typeofspatialorformalconfgurationorpattern,andwhicharecapableofgeneral
applicationtodescribeallpossibletypesofpattern.Forexample,itoughttobeable
tohandlespatialpatternsorbuiltformpatternswhichlackgeometricregularityas
wellasthosewhichhaveit.Unlessthiscanbedonewithrigourtherethereislittle
hopethattheoreticalpropositionsinarchitecturecaneverbeanalyticinthesense
thatwerequirethemtobe.
Thenextchapterofthisbookwillintroducesuchasetofnon-discursive
techniquesfortheanalysisofconfguration,frstdevelopedinspatialformas
spacesyntax,butnowbeingbroadenedtocoverotheraspectsofconfguration.
Thesetechniqueshavebeenusedoverseveralyearsfortwoprinciplepurposes,
frsttodiscoverhowfaritwaspossibletobringtolightandsubjecttorigorous
comparativeanalysestheconfgurationalaspectsofspaceandforminbuilding
throughwhichcultureistransmitted,andsecond,throughthesecomparativestudies
todevelopacorpusofmaterialwhichwouldpermitthegradualdevelopmentofa
generaltheoryofarchitecturalpossibility.Theremainderofthisbookisessentially
anaccountoftheprogressthathassofarbeenmadeinthisproject.
Aswewillsee,whatwediscoverthroughapplyingthesetechniquestothe
analysisofspatialandformalpatternsinarchitecture,wherevertheyarefoundand
whatevertheirembodimentineitherbuildingsorurbansystems,areinvariantsin
patternswhichlienotonthesurfaceofthingsbutwhichareburiedinthenatureof
confgurationsthemselves.Theseinvariantswecanthinkofasdeepstructuresor
genotypes.Eachculturalmanifestationthroughbuilding,whetherasabuildingtype
foraparticularpurpose,oraparticulararchitecturalethosorimprintingofculture
onbuilding,doessothroughsuchgenotypes.Forexample,seenassystemsof
organisedspace,itturnsoutthattownsandcitieshavedeepstructureswhichvary
withculture.Likewise,seenasorganisedspaces,buildingsfordifferentfunction
purposesalsohavedeepstructuresorgenotypes.Thesegenotypesareorembody
culturalortypologicalinvariants.Thesearenotofcoursegenerallaws.Theyare
atbestthecoveringlawsofcultures.Therearethegenotypicalinvariantsbywhich
eachsocietyandeachfunctioninsocietyseekstoexpressitselfthrougharchitecture.
However,aswebuildourcorpusofgenotypeswegraduallybegintosee
thatthereisanotherlevelofinvariance:therearegenotypesofthegenotypes.
Belowthelevelofculturalvariationinarchitecturethereexistinvariantsacross
culturesandtypes.Thesegenotypesofgenotypesarenotthecoveringlawsof
culturesbuttheinvariantlawsthatbindhumankindingeneraltoitsartifcialmaterial
world.Theyaretheabstractrawmaterialoutofwhichallconfgurationalpossibility
inspaceandforminthebuiltworldareconstructed.Itisatthislevelofinvariance
andonlyatthislevelthatwecanbuildagenuineanalytictheory.These
possibilitieswillbedealtwithinChapters8and9.
The need for an analytic theory
of architecture
6
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Architecture as art and as science
Ifthistheoreticalprojectiseventuallytosucceedanditisbeyondthescopeof
anysinglebooktodomorethantakeafewfalteringstepstowardssuchatheory
thenitisclearthatsuchatheorywouldliberateratherthanconstraindesign.At
root,theneedforarchitecturaltheoryarisesfromtheneedtoformulateprinciples
fromtheexperienceofhavingbuilttoinformandguideusonhowwemightbuild.
Thisdynamicbetweentheactualandpossibleistheessenceofarchitectural
theorising.Architecturaltheoryarisesfromthefactthatarchitectscanneither
forgetthearchitecturaltradition,norrepeatit.Inarchitecture,theoryisnotsimply
ameanstofxapictureoftheworldinacertainform.Itisalsothemeansby
whichformisdestabilisedandanewfutureisconceived.Architectureprogresses
byincorporatingitsrefectiononthepastintoanabstractframeofpossibility.This
frameistheory.Withoutit,historicalthoughtissterile,andcanonlyleadtoimitation
ofthepast.Throughtheintermediaryoftheory,refectiononthepastbecomes
possiblefuture.Historyconstrains,buttheoryliberates,andthemoregeneralthe
theory,thegreatertheliberation.
Doesthismeanthenthatthelinebetweenarchitectureasscienceand
architectureasartneedstoberedrawnclosertoscience?Idonotbelieveso.We
cancallonthebeautifulideasofErnstCassirerontherelationbetweenartand
science.
25
Languageandscience,hewrites,arethetwomainprocessesbywhich
weascertainanddetermineourconceptsoftheexternalworld.Wemustclassify
oursenseperceptionsandbringthemundergeneralnotionsandgeneralrules
inordertogivethemanobjectivemeaning.Suchclassifcationistheresultofa
persistentefforttowardssimplifcation.Theworkofartinlikemannerimpliessuch
anactofcondensationandconcentrationButinthetwocasesthereisadifference
ofstress.Languageandscienceareabbreviationsofreality;artisanintensifcation
ofreality.Languageandsciencedependononeandthesameprocessof
abstraction;artmaybedescribedasacontinuousprocessofconcretionart
doesnotadmitofconceptualsimplifcationanddeductivegeneralisation.Itdoes
notinquireintothequalitiesorcausesofthings;itgivestheintuitionoftheform
ofthingsTheartistisjustasmuchthediscovereroftheformsofnatureasthe
scientististhediscovereroffactsornaturallaws.
Thoseofuswhobelievethatscienceisonthewholeagoodthing,accept
thatscienceisinonesenseanimpoverishmentthoughinothersanenhancement
ofourexperienceoftheworldinthatitcannotcopewiththedensityofsituational
experience.Ithastobeso.Itisnotinthenatureofsciencetoseektoexplainthe
richnessofparticularrealities,sincetheseare,aswholes,invariablysodiverseas
tobebeyondtheusefulgraspoftheoreticalsimplifcations.Whatscienceisabout
isthedimensionsofstructureandorderthatunderliecomplexity.Heretheabstract
simplifcationsofsciencecanbethemostpowerfulsourceofgreaterinsight.Every
momentofourexperienceisdenseand,assuch,unanalysableasacomplete
experience.Butthisdoesnotmeantosaythatsomeofitsconstituentdimensions
arenotanalysable,andthatdeeperinsightmaynotbegainedfromsuchanalysis.
The need for an analytic theory
of architecture
62
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Thisdistinctioniscrucialtoourunderstandingofarchitecture.Thatarchitectural
realitiesaredenseand,aswholes,unanalysabledoesnotmeantosaythattherole
ofspatialconfguration(forexample)inarchitecturalrealitiescannotbeanalysed
andevengeneralised.Theideathatscienceistoberejectedbecauseitdoesnot
giveanaccountoftherichnessofexperienceisapersistentbutelementaryerror.
Sciencegivesusquiteadifferentkindofexperienceofreality,onethatispartial
andanalyticratherthanwholeandintuitive.Assuchitisinitselfthatitisvaluable.
Itneedstobeacceptedorrejectedonitsownterms,notintermsofitsfailureto
belikelifeorlikeart.
Itisinanycaseclearthatthedependenceofarchitectureontheories,
covertorexplicit,doesnotdiminishitsparticipationinCassirersdefnitionofart.
Thisistruebothinthesensethatarchitectureis,likeart,acontinuousprocess
ofconcretion,andalsointhesensethat,likeart,itsaspectsareinnumerable.
Buttherearealsodifferences.Thethingwhoseaspectsareinnumerableisnota
representationbutareality,andaveryspecialkindofreality,onethroughwhichour
formsofsocialbeingaretransformedandputatrisk.Thepervasiveinvolvement
oftheoryinarchitecture,andthefactthatarchitecturescontinuousconcretion
involvesoursocialexistence,defnesthepeculiarstatusandnatureofsystematic
intentofthearchitecturalkind:architectureistheoreticalconcretion.Architectsare
enjoinedbothtocreatethenew,sincethatisthenatureoftheirtask,butalsoto
renderthetheoriesthattietheircreationtooursocialexistencebetterandclearer.
Itisthisthatmakesarchitecturedistinctandunique.Itisasimpossibletoreduce
architecturetotheoryasitistoeliminatetheoryfromit.
Architectureisthusbothartandsciencenotinthatithasbothtechnicaland
aestheticaspectsbutinthatitrequiresboththeprocessesofabstractionbywhich
weknowscienceandtheprocessesofconcretionbywhichweknowart.The
diffcultyandthegloryofarchitecturelieintherealisationofboth:inthecreationof
atheoreticalrealmthroughbuilding,andinthecreationofanexperiencedreality
whoseaspectsareinnumerable.Thisisthediffcultyofarchitectureandthisis
whyweacclaimit.

Notes
RogerScruton,The Aesthetics of Architecture,Methuen,1977.
Ibid.,p.4.
L.B.Alberti,De Re Aedifcatoria,1486;translationreferredto:
Rykwertetal.(1988), MITPress,1991.
O.Newman,Defensible Space;ArchitecturalPress,1972.
Alberti,Chapter9.
Newman,pp.39.
Howthishappensasacognitiveprocessisthe
subjectofChapter11:Thereasoningart.
Chapter11includesacasestudyinthedangersofbadtheory.
Alberti,forexample,Book9.
1
2
3
4
5
6
7
8
9
The need for an analytic theory
of architecture
63
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Scrutonsfundamentalerroristoconfusethesetwoaspects,andineffectto
believethatthenarrowpropositionsofarchitecturaltheoryareintendedto
beuniversalistic.SeeScruton,The Aesthetics of Architecture,p.4.
Alberti,Book9.
Newman,pp.39.
F.Bacon,The New Organon (1620),BobbsMerrill,1960,
AphorismsBook1,Aphorismcxv,p.105.
T.Kuhn,The Structure of Scientifc Revolutions,UniversityofChicagoPress,1962.
TouseReneThomsadmirableexpressionforwhatweobserveseeStructural
Stability and Morphogenesis,Benjamin,NewYork,1975originallyinFrench,1972,
asStabilite Structurelle et Morphogenese.Seeforexamplep.320.
F.DeSaussure,(originallyinFrench1915)versionusedCourse in General
Linguistics,McGrawHill,1966,translatedbyC.BallyandA.Sechahayewith
A.Riedlinger-seeforexamplepp.10312.
Theseexamplesofcoursedealwithlinearvariation,
butthebasicargumentsalsoapplytonon-linearvariation.
Thismodel,theEhrenfestgame,istakenfromM.KacandS.Ulam,
Mathematics and Logic,PelicanBooks,1971,p.168.OriginallyPraeger,1968.
ForafurtherdiscussionseeH.Reichenbach,The Direction of Time,
UniversityofCaliforniaPress,1971,particularlyChapter4.
SeeK.PopperK,Conjectures and Refutations,Routledge
andKeganPaul,1963,Chapter5:Backtothepresocratics.
SeeforexampleM.GhykaM,Geometrical Composition
and Design,Tiranti,London,1956.
ForexampleintheworkofAlexanderKoyre,e.g.Metaphysics and Measurement,
ChapmanandHall,1968(originallyinFrench)andNewtonian Studies,Chapman
andHall,1965orGeorgesCanguilheme.g.La Connaissance de la Vie,Librairie
PhilosophiqueJ.Vrin,Paris,1971.
AspioneeredintheworkofMichelFoucault.
Inthepast,thishasledtoaquiterapidpermeationbynewscientifcconcepts
oftheconceptualschemesofeverydaylife,bringingchangesinconsciousness
whichmayseementirelyprogressive,asforexamplewiththetheoriesofNewton
orDarwin.Itmayindeedbethelossofthisillusorystrengththathasboughtabout
muchofthealienationfromscienceinthelatetwentiethcentury.Assciencehas
progressedfartherintomicroandmacrophenomenaanddiscoveredpatterns
whichareutterlyremotefromeverydayintuitiontheconceptsthatmakeup
scientifctheoriesbecomesostrangethattheycannotevenbeformulatedsoasto
interfaceeffectivelywiththeestablishedconceptualsystemoflinguisticnormality.
Thishashappenedwithquantumtheory.Butwhathashappenedwithquantum
theoryconfrmsourmodelassetoutinthediagram:scienceintervenesthrough
formalismsbetweenconceptsandphenomena.Itisnopartofitsfunctionorits
moralitythattheseconceptsshouldfallwithinthelightedcircleofintuition(touse
HermanWeylsadmirablephraseseehisPhilosophy of Mathematics and Natural
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
The need for an analytic theory
of architecture
64
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Science,Atheneum,NewYork,1963,p.66)andsobetranslatableintotheavailable
conceptsofeverydaylifeandlanguage.Thereisnogreaterarrogancethanthat
weshouldexpectthemtobe,exceptperhapsthebeliefthattheworlditselfinits
deepestoperationsshouldconformitselftotheapparatusofourintuitions.
ErnstCassirer,An Essay on Man,YaleUniversityPress,1944.
Editionused:BantamMatrix,1970.Chapter9,Onart,pp.15288.

25
Nondiscursive technique
Chapter three
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Object artefacts and abstract artefacts
Oneofthedurableintellectualachievementsofthetwentiethcenturyhasbeento
initiatethescientifcstudyofhumanartefacts.Atfrstsight,suchastudymight
seemparadoxical.Mostartefactsarephysicalobjectsthatadaptnaturallawsto
humanpurposes.Tomakeanobjectforapurposesurelypresupposesthatwe
understandit.Buttwenty-fveyearsago,HerbertSimon,inhisThe Sciences of the
Artifcial,showedthatthiswasfarfromthewholestory.
1
Eveniftheobjectswe
makearenotpuzzlinginthemselves,theyaresowhenseeninthecontextofthe
ramifyingeffectsoftheirdispersionthroughoutoursocio-technicalecosystem.He
wasthinking,amongstotherthings,ofcomputers.Itwouldbeasenlightening,he
argued,tohaveanaturalhistoryofcomputersinourincreasinglyartifcialworld,as
ofanynaturalphenomenon.Empiricalsciencesofartefactswerethereforenotonly
apossibility,butanecessity.
Butobjectartefactsareonlythelesseraspectofthepuzzleoftheartifcial.
Therealsoexistsaclassofartefactswhicharenolessdramaticintheirimpacton
humanlife,butwhicharealsopuzzlinginthemselvespreciselybecausetheyarenot
objects,but,onthecontrary,seemtotakeaprimarilyabstractform.Languageisthe
paradigmcase.Languageseemstoexistinanobjectivesense,sinceitliesoutside
individualsandbelongstoacommunity.Butwecannotfndlanguageinanyregion
ofspace-time.Languageseemsreal,butitlackslocation.Itthusseemsbothreal
andabstractatthesametime.Otherartefactswhichsharesomeoftheattributesof
language,suchascultures,socialinstitutions,andeven,somewouldargue,society
itself,allseemtoraisethiscentralpuzzleofbeing,itseems,abstractartefacts.
Itcannotofcoursebesaidthatabstractartefactsarenotmanifestedin
space-time.Theyappearintheformoflinguisticacts,socialbehaviours,cultural
practices,andsoon.Butthesespace-timeappearancesarenottheartefactitself,
onlyitsmomentaryandfragmentaryrealisations.Weapprehendspeech,asde
Saussurewouldsay,butnotlanguage.
2
Inthesameway,weseesocialbehaviours,
butweneverseesocialinstitutions,andweseeculturaleventsbutweneversee
cultures.Yetinallthesecases,thespace-timeeventsthatwewitnessseemtobe
governedintheirformbytheabstract,unrealisableartefactsthatwegiveanameto.
Thematerialworldprovidesthemilieuwithinwhichtheabstractartefactisrealised,
buttheserealisationsaredispersedandincomplete.Theexistenceoflanguages,
socialinstitutionsandculturescanbeinferredfromspace-timeeventsbutnot
seeninthem.
Inspiteofthisstrangemodeofexistence,abstractartefactsseemtobethe
stuffofwhichsocietyismade.Wecannotconceivewhatasocietywouldbelikeif
deprivedofitslanguages,itscharacteristicsocialbehaviours,itsculturalformsand
itsinstitutions.Itisnotclearthatanythingwouldbeleftwhichwecouldreasonably
callsociety.Wemayconjecture,perhaps,thatabstractartefactsarethewaythey
arepreciselybecausetheirpurposeistogenerateandgoverndispersedevents,
andthroughthistoconvertadispersedcollectivityofspeakers,behavioursorsocial
actorsintosomesemblanceofasystem.Themultipositionalityofthespace-time
Environments are invisible.
Their...ground rules...evade easy
perception. Marshall McLuhan
65
Nondiscursive technique
66
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
realisationofabstractartefactsseemstobeanessentialpartofhowtheywork.
However,tosaythisistorestatetheproblem,nottosolveit.Infact,in
spiteoftheirapparentoddity,abstractartefactsposemanyofthepuzzleswhich
scienceseekstoexplainfornaturalsystems.Forexample,theyseemableboth
toreproducethemselvesovertime,andalsotoundergomorphogenesis,though
whetherthisisbyaconstantorsuddenprocessisentirelyobscure.Ifabstract
artefactshavesuchproperties,thenitwouldseemtofollowthattheymusttherefore
havesomekindofinternalprinciplesorlawswhichgiverisetostabilityandchange,
asdonaturalsystems.
3
Yetwhatevertheselawsarelike,theymustalsopass
throughthehumanmind,sinceitisonlythroughhumanmentalactivitythattheself
reproductionandmorphogenesisofthesesystemsoccurs.Itseemsinconceivable,
therefore,thatthelawswhichgoverntheformsofabstractartefactsaresimilarto,
orevencommensurablewith,thelawsthatgovernnaturalsystems.Atthesame
time,suchlawsmustbepartofnature,sincetheycannotbeotherwise.Theymust
refectsomepotentialitieswithinnature.
Inviewofalltheseapparentparadoxes,itwasthegreatmeritofLvi-
Straussandotherpioneersofthestudyofabstractartefactstohavebothidentifed
thekeyinsightnecessaryfortheirstudy,andtohavepointedtoapossible
methodologyforresearch.
4
Theinsightwastohaveseenthedependenceofthe
concreteontheabstractinsystemslikelanguageandculture,asclearlyasPlato
oncenoteditforthenaturalworld.
5
Now,asthen,thisfundamentalinsightprovides
thestartingpointandinitialstanceforthesettingupofsciences.Themethodology
wasthat,aswithnaturalsystems,wewouldexpecttofndcluestothenatureof
theseorganisinglawsbystudyingtheregularitiesthatabstractartefactsgenerate
inspace-time,thatis,inspeech,behaviour,culturalpracticesandinstitutional
forms.Accordingly,themovementcalledstructuralismaimedtoassignabstract
formalmodelswiththestructureandvarietymanifestedinthespace-timeoutput
ofsuchsystems-observedspeech,socialbehaviour,organisationaldynamics
andsoon-andthroughthistoaccountnotonlyfortheinternalsystemnessof
suchphenomena,butalsotoshowhowthehumanmindwascapableofholding
andcreativelytransformingsuchpowerfullystructuredinformation.Inthissense,
structuralismwasnomoreorlessthanorthodoxsciencerewrittenforthestudy
ofabstractartefacts.
6
Thisresearchstrategyrefectsthefundamentalfactthatabstract
artefactsmanifestthemselvestousintwoways:throughthespace-timeevents
theygenerate;andthroughtheconfgurationalpatternswhichseemtosupport
themandwhichenableusbothtogenerateandinterpretthem.Thesetwoways
inwhichweexperienceabstractartefactsareboundtogetherbythefactthatin
usingconfgurationalstructurestogeneratespace-timeeventswealsoproject
theseconfgurationalstructuresintospace-timeandindoingsohelptotransmit
themintothefuture.Thisdoubletakebetweentheconsciousmanipulationof
space-timeeventsandthetransmissionofconfgurationalstructureisthedefning
characteristicoftheabstractartefactandthereasonitisabletobethestuffof
Nondiscursive technique
67
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
society.Bydeployingobjectsandcreatingspace-timeeventswenecessarily
transmitstructures,andthroughthemtheabstractartefactswhichholdsociety
togetherasacommunicativesystem.Theobjectofstructuralismistocapture
thedynamicsoftheseprocesses.
Formalmethodswerethereforecriticaltostructuralism.However,as
Heisenbergonceremarked:Ourscientifcworkinphysicsconsistsinasking
questionsaboutnatureinthelanguagethatwepossessandtryingtogetan
answerfromexperimentbythemeansthatareatourdisposal.
7
Thisissurely
trueofallscientifcenquiry.Unfortunately,itseemstopointdirectlytothefailure
ofstructuralismtodeliveronitspromises.Examiningthespace-timeregularities
ofthephenomenageneratedbyabstractartefacts,wecannotfailtonoteone
overwhelmingconsistency;thattheyseemtobegovernedbypatternlawsof
somekind.Thewordsthatmakeupspeechandthebehavioursthatseemsocial
areallmanifestedinspace-timeassequencesordispositionsofapparentelements
whoseinterdependenciesseemtobemultiplex,andirreducibletosimplerulesof
combination.Forexample,tosay,asChomskydid,
8
thatsentences,whichappear
tobesequencesofwords,cannotbegeneratedbyaleft-rightgrammar,isa
confgurationalproposition.Somedegreeofsyncreticco-presenceofmanyrelations
isinvolvedwhosenaturecannotbereducedtoanadditivelistofpairwiserelations.
Thisistosaythatthelawsgoverningabstractartefactsseemtobeconfgurational
insomethinglikethesensewehavedefneditinthepreviouschapters.
Itisinthisrespectthatstructuralismseemstohavelackedmethodology.Its
formaltechniquesdidnottrytodrivestraighttotheproblemofconfguration,but
confnedthemselvestothemoreelementaryaspectsoflogicandsettheory,those
branchesofmathematics,thatis,thatsoughttoaxiomatisethethinkingprocesses
ofminds,ratherthantomodelrealworldcomplexity.
9
Consequently,justasthe
languagesavailableforPlatoinhistimewereinadequateforhisvisionofnature,
10

sothetoolspickedupinthemid-twentiethcenturybystructuralismweretoofrailfor
thevisionofartifcialphenomenathathadinitiatedtheirsearch.Thephenomenathat
structuralistanalysissoughttoexplainwereinthemainconfgurational,buttheformal
techniquesthroughwhichinvestigatorssoughttodemonstratethisrarelywere.
Built environments as artefacts
Thepurposesofthisdigressionintoabstractartefactsaretwofold:frst,todraw
attentiontocertainpropertiesofbuiltenvironmentsthatmightotherwisebemissed;
second,topointtocertainadvantagesofthebuiltenvironmentinprovidinga
platformfortakingontheproblemofconfgurationinanewway.First,however,
wemustunderstandtheverypeculiarstatusofbuiltenvironmentsasartefacts.
Builtenvironmentsappeartousascollectionsofobjectartefacts,that
is,ofbuildings,andassuchsubjecttoordinaryphysicallaws,anddeservingof
Simonianenquiry.Butthatisnotallthattheyare.AswenotedinChapter1,in
termsofspatialandformalorganisation,builtenvironmentsarealsoconfgurational
entities,whoseformsarenotgivenbynaturallaws.Ifwewishtoconsiderbuilt
environmentsasorganisedsystems,thentheirprimarynatureisconfgurational,
Nondiscursive technique
68
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
principallybecauseitisthroughspatialconfgurationthatthesocialpurposes
forwhichthebuiltenvironmentiscreatedareexpressed.Thecollectionsof
objectartefactsinspace-timethatwesee,arethenameansthroughwhich
sociallymeaningfulconfgurationalentitiesarerealised.Inotherwords,inspite
ofappearances,builtenvironmentspossessakeypropertyofabstractartefacts.
Itsobjectsaremoredurablethan,say,thespokenwordsofalanguage,orthe
rule-infuencedindividualbehavioursthatmakeupasocialevent,buttheyareof
thesamekind.Theyarespace-timemanifestationsofconfgurationalideaswhich
alsohaveanabstractform.Thebuiltenvironmentisonlythemostdurableofthe
space-timemanifestationsofthehumanpredilectionforconfguration.Thishasan
epistemologicalconsequence.Weshouldnotexpectthebuiltenvironmentmerely
tobethematerialbackdroptoindividualandsocialbehaviour,asitisoftentakento
be.Itisasocialbehaviour,justastheuseoflanguageisasocialbehaviourandnot
justameanstosocialbehaviour.Wecannotthereforeregardthebuiltenvironment
asmerelyaninertthing,andseektounderstanditwithoutunderstandingthe
sociallogicofitsgeneration.
Butjustaswecannottreatabuiltenvironmentasathing,wecanno
moretreatitasthoughitwerenomorethanalanguage.Thebuiltenvironmentis,
apartfromsocietyitself,thelargestandmostcomplexartefactthathumanbeings
make.Itscomplexityanditsscaleemergetogether,because,likesociety,abuilt
environmentisnotsomuchathingasaprocessofspatio-temporalaggregation
subjecttocontinualchangeandcarriedoutbyinnumerableagenciesovera
longperiodoftime.Althoughtheseprocessesofaggregationmaybelocally
characterisedbythesamekindofautonomicrulefollowingaswefndforindividual
actsofbuilding,thereareothernolessfundamentalattributesthatmakethebuilt
environmentaspecialcase.
Themostobvious,andthemostimportant,isthatthespatio-temporal
outputsofbuiltenvironmentprocessesarenotephemerallikethoseoflanguage
orsocialbehaviour.Theyarelong-lasting,andtheyaggregatebyoccupying
aparticularregionofspaceforalongtime.Thismeansthatoverandabove
thinkingofbuiltenvironmentsastheproductsofabstractrulesystems,wemust
alsorecognisethattheyhaveanaggregativedynamicwhichistosomeextent
independentoftheserulesystems,although,aswewillsee,itisrarelyquiteoutof
theircontrol.Theseaggregativeprocesseshavequitedistinctiveproperties.Spatio-
temporaladditionstoasystemusuallyoccurlocally,butthedynamicsofthesystem
tendtoworkatthemoreglobalaggregativelevels.
11
Complexityarisesinpartfrom
therecursiveapplication,inincreasinglycomplexaggregations,ofruleswhichmay
initiallybesimple,butthemselvesmaybetransformedbytheevolvingcontextin
whichtheyareapplied.Alocallydrivenaggregativeprocessoftenproducesa
globalstatewhichisnotunderstood
12
butwhichneedstobeunderstoodinorder
forthelocallydrivenprocesstobeeffective.Thisistheessentialnatureofthe
largeaggregatesofbuildingswhichformmostbuiltenvironments.

Nondiscursive technique
69
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Thiscomplex,processualaetiologyisthemainreasonwhybuiltenvironmentshave
provedsoresistanttoorthodoxattemptstomodeltheirstructuremathematically.
Buildingsandcitiesarenotcrystallineobjects,unfoldingundertheinfuence
onlyoflawsofgrowth.Theelementaryspatialgestuariesofhumankindandits
culturesmayconstructlocalelementalconfgurations,butthesethenoperateas
localorderingswithingrowthprocessesandactasconstraintsonthenatural
evolutionofglobalpatterns.Architecturaland,evenmoreso,urbanformsoccur
attheinterfacebetweennaturalprocessesandhumaninterventions.Human
actionsrestrictandstructurethenaturalgrowthprocesses,sothattheycannotbe
understoodwithoutinsightintobothindividually,andintotherelationsbetweenthe
two.Theinterventionofthemindintheevolvingcomplexitymustbeunderstood,
butsomustitslimitations.
Thebuiltenvironmentmaythenbethemostobviousofobjects,andtheone
thatformsourfamiliarmilieu,butatthesametimeitsinnerlogicandstructureisas
inaccessibletousasanythinginnature.However,ithasonegreatadvantageasan
objectofstudy.Itsveryscale,manifestnessandslowrateofchangeofferitupas
theparadigmcaseforconfgurationalinvestigation.Theessenceoftheproblemis
tocapturethelocal-to-globaldynamicsofarchitecturalandurbansystems,thatis,to
showhowtheelementarygenerators,whichexpressthehumanabilitytocognise
andstructureanimmediatespatialreality,unfoldintotheramifedcomplexitiesof
large-scalesystems.
Inthis,methodologicaldiffcultiesarecentral.Theaimofamethodmust
betocapturethelocalorelementalordering,theemergenceofglobalcomplexity,
andhowbothrelatetothehumanmind.Foranyofthese,themanifestproblemof
confgurationmustbetackledheadon,andmustbeapproachedfrstandforemost
asanempiricalproblem.Ifthespace-timeproductsofabstractartefactsareheld
togetherbyconfguration,thenconfgurationcanbefoundbyexaminingthem.
Thecorpusofconfgurationsthatcanbebuiltthroughthestudyofrealcases
mustbesomeindicatorofwherewemightseekfortheconfgurationalinvariants
ofbuiltenvironmentprocesses.Forthistask,theveryscale,relativestabilityand
availabilityofbuiltenvironmentsmakethemtheidealvehicleforanenquiry.Allwe
needaretechniquesthatpermittheextractionofconfgurationfromitsspace-time
embodiments-thatis,non-discursivetechnique.
Simplicity as a means to complexity
Theconfgurationalformalismsproposedhereasthebasisfornon-discursive
techniqueareinsomewaysmuchsimplerthanothersproposedforthesimilar
classesofphenomenaoverthelasttwentyyears.
13
Yettheyhaveprovedthemost
powerfulindetectingformalandfunctionalregularitiesinrealsystems.Thereare
probablythreereasonsforthis.First,thequantitativemethodsproposedaredirected
straightattheproblemofconfguration,thatis,theproblemofunderstanding
thesimultaneouseffectsofawholecomplexofentitiesoneachotherthrough
theirpatternofrelationships.Lackofattentiontothiscentralproblemistheprime
Nondiscursive technique
70
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
reasonwhypastformalismoftenseemedtooffermathematicalsophistication
outofproportiontotheempiricalresultsachieved.Withconfgurationalanalysis
itistheotherwayround.Exceedinglysimplequantitativetechniqueshaveled
toadisproportionatesuccessinfndingsignifcantformalandform-functional
regularities.Confguration,asdefnedbelow,seemstobeatleastoneofthe
thingsthatarchitecturalandurbanpatternsareabout.
Second,inconfgurationalanalysis,asmuchtheoreticalattentionhasbeen
giventotherepresentationofthespatialorformalsystemthatistobeanalysed
astothemethodofquantifcation.Aswewillsee,thisquitenormallygivesriseto
awholefamilyofrepresentationsofthesamespatialsystem,eachonerelevant
tosomeaspectofitsfunctioning.Itisalsonormaltocombinerepresentations,
literallybylayingonerepresentationontopoftheotherandtreatingtheresulting
connectionsasrealconnectionsinthesystem.Throughthis,wefndthatpairs
oreventriplesofrepresentationstakentogetheryieldformallyorfunctionally
informativeresults.Intermsofresearchstrategy,thismeanstryingtorepresent
spaceintermsofthetypeoffunctioninwhichweareinterested.Forexample,
simplelinestructuresdrawnthroughspaces,temporarilydiscountingother
properties,haveprovedsuffcient(aswewillseeinthenextchapter)to
accountformanyaspectsofmovementwithinbuildingsandurbanareas.
Third,andsynthesisingtheprevioustwo,muchattentionhasbeengiven
tothegraphicrepresentationoftheresultsofmathematicalanalysis,sothatthe
formalstructuresidentifedinspatialorformalcomplexescanbeintuitivelyseen
andunderstoodwithouttheintermediaryofmathematicalformalism.Thismeans
thatmuchcanbeunderstoodbythosewhosetemperamentsleadthemtoprefera
graphicalratherthanamathematicalunderstanding.Byrepresentingmathematical
resultsgraphically,alevelofcommunicationispossiblethatpermitslargenumbers
ofpeopletobeinterestedandknowledgeablewhowouldotherwisefallatthe
frstfenceofmathematicalanalysis.Inparalleltothisgraphicalrepresentationof
results,usuallydrawnbycomputer,thereisaparallelemphasisintheinitialstages
ofinvestigationtothedrawingofspatialorformalideasbyinvestigatorsandby
studentsasaconstantadjunctto,andcheckon,formalanalysis.
Noapologyisthenofferedforthesimplicityofsomeofthenotions
presentedhere.Othershavediscussedsomeofthesepropertiesbuthavenotbeen
mindedtoexploretheirfullempiricalortheoreticalrelevance,orhowtheymight
befttedintotheoverallform-functionpicture.Perhapsonereasonforresearchers
tomisskeyrelationswhilegoingclose,hasbeenwhatwewouldseeasan
overarchingandinsomewaysprematureconcernwithdesignattheexpenseof
theempiricalinvestigationofbuildings.ThespacesyntaxresearchatUCLhas
beendrivenbyaremarkofLionelMarchs:Theonlythingyoucanapplyisagood
theory.
14
Anotherpossiblereasonwhyformalexplorationhasmissedtheoretical
insighthasbeenthefrequentlackofacloseenoughrelationbetweenmathematical
andempiricalaspectsoftheproblemsposedbyrealbuildingsandcities.In
contrast,thetechniquesofspatialrepresentationandquantifcationproposedhere
Nondiscursive technique
7
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
areessentiallysurvivorsofanintensiveprogrammeofempiricalinvestigationspread
overthebestpartoftwodecadesinwhichformalquestionshavebeenexplored
inparalleltotheempiricalpuzzlesposedbyarchitecturalandurbanrealities.
WehavealreadydiscussedtheideaofconfgurationatsomelengthinChapter
1.Nowweneedtodefneitformally,andtoshowsomeofitspowertosay
simplethingsaboutspaceandform.Itshouldbenotedthatwhatfollowsisnota
methodologicalcookbook,butatheoreticalexplorationoftheideaofconfguration.
Atthisstage,theexamplesgivenareillustrationsofideas,notworkedexamplesof
analysis.Casestudieswillcomeinensuingchapters.Therelationofthischapterto
thosethatfollowisthatofaquarry,whichfuturechaptersreturntotopickupone
ofthepossibilitiessetouthere,andrefneitforthepurposesofthatchapter.This
chaptershowsthebasesandconnectionofthewholefamilyofmethods.
Defning confguration
Letusbeginbydefningexactlywhatwemeanbyconfguration,usingan
exampledirectlyanalogoustofgure1.3inChapter1,buttakingaslightlydifferent
form.WemayrecallthatinChapter1,asimplerelationwasdefnedasarelation
-say,adjacencyorpermeability-betweenanypairofelementsinacomplex.A
confgurationalrelationwasthendefnedasarelationinsofarasitisaffected
bythesimultaneousco-presenceofatleastathirdelement,andpossiblyall
otherelements,inacomplex.Infgure3.1i,forexample,aandbaretwocubes
standingonasurface.In3.1ii,thecubesarebroughttogetherfullfacewiseto
makeaconjointobject.Therelationofaandbissymmetricalinthatabeing
Figure 3.1
a b a
b
a b a
b
i ii iii
iiii
a b a b
a
b
c c 2
3 3
2
2 2
3
2
3
v vii
c
td = 8 td = 6 td = 8
22
18
16
16
18
22
28
28 23
17
15
15
17
21
27
23
24
18
14
14
16
20
26
20
25
19
15
13
15
19
25
19
td = 168 td = 158 td = 152 td = 150
Figure 3.2
a b
a b
c
vi
Figure 3.1
Nondiscursive technique
72
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
the(contiguous)neighbourofbimpliesthatbisthe(contiguous)neighbourofa.
Onecouldequallysay,thoughwithlessobviousness,thatin3.1iaandbwere
non-contiguousneighbours,andwerethereforesymmetricalinthissense.Either
way,therelationofthetworemainssymmetrical,andinfactthisisimplicitinthe
neighbourrelation.In3.1iii,theconjointobjectformedbyaandbin3.1iiistaken
andrestedononeofitsends,withoutchangingtherelationofatob.Butbnow
appearstobeabovea,andtherelationofbeingabove,unlikethatofbeingthe
neighbourofisnotsymmetricalbutasymmetrical:bbeingaboveaimpliesthat
aisnotaboveb.
Howhasthishappened?Thetemptationistosaythatrelationslikeabove
andbelowdependonanexogenousframeofreference,likeeastandwest,or
upanddown.Infact,whathashappenedcanbesaidmoresimply,asshownin
3.1iiii.Thesurfaceonwhichthecubesstand-say,thesurfaceoftheearth-was
notreferredtoindescribingtherelationbetweenaandbin3.1iandii.Itshould
havebeen,hadwewantedtoforeseetheeffectsofstandingtheconjointobjecton
itsend.Letuscallitc.In3.1ii,therelationofbothaandb,takenseparately,tothe
thirdobject,c,isalsosymmetrical,asistheirrelationtoeachother.So,incidentally,
istherelationoftheconjointobjectformedbyaandbtothethirdobject.Theseare
allsimplerelations.Butwecanalsosaysomethingmorecomplex:thatin3.1ii,a
andbaresymmetricalwithrespecttoc,aswellaswithrespecttoeachother.This
isaconfgurationalstatement,sinceitdescribesasimplespatialrelationinterms
ofatleastathird.Whathappensin3.1iiiisnowclear.Althoughaandbremain
symmetricalwithrespecttoeachother,theyarenolongersymmetricalwithrespect
toc.Onthecontrary,theyareasymmetricalwithrespecttoc.Thedifference
between3.1iiandiiiisthenaconfgurationaldifference.Therelationofaandbto
eachotherischangedifweaddthewithrespecttoclausewhichembedsthetwo
cubesinalargercomplexwhichincludesc.
Thesituationisclarifedbythejustifedgraphs(orj-graphs:graphsin
whichnodesarealignedabovearootaccordingtotheirdepthfromtherootsee
Figure 3.1
a b a
b
a b a
b
i ii iii
iiii
a b a b
a
b
c c 2
3 3
2
2 2
3
2
3
v vii
c
td = 8 td = 6 td = 8
22
18
16
16
18
22
28
28 23
17
15
15
17
21
27
23
24
18
14
14
16
20
26
20
25
19
15
13
15
19
25
19
td = 168 td = 158 td = 152 td = 150
Figure 3.2
a b
a b
c
vi
Figure 3.2
Nondiscursive technique
73
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Chapter1)oftheconfgurationsshownin3.1v,viandvii.Ineach,thebottomnode
istheearth,andisinscribedwithacrosstoindicatethatitistheroot.In3.1v,a
andbareeachindependentlyconnectedasneighbourstotheearth.In3.1vi,the
relationofneighbourbetweenaandbisadded.In3.1vii,therelationbetweenb
andc,theearth,isbrokencreatingatwodeeprelationbetweenbandc.Onemay
notethatthisset-upalreadyexistsin3.1vbetweenthetwonon-contiguouscubes
withrespecttotheearth.Inthissense,3.1viirecreatesagraphwhichalreadyexists
inv.Thisisalsoshowninthenumbersattachedtoeachofthenodesofthegraph,
whichindicatethesumofdepthfromthatnodetotheothernodesinthesystem.
Thetotaldepthof3.1vandviiistherefore8,whilethatofviis6.Wemightsay,
then,thatthedistributionsoftotaldepthsandtheiroverallsumdescribeatleast
someconfgurationalcharacteristicsofthesecompositeobjects.
Nowletusexplorethissimpletechniquealittlefurtherbyexaminingfgure
3.2,aseriesofsimplefgurescomposedofsquarecellsjoinedtogetherthroughtheir
faces(butnottheircorners)withtotaldepthsforeachcelltoallothersinscribedin
eachcell,andthesumsofthesetotaldepthsforeachfgurebelowthefgure.The
fguresareallcomposedofsevenidenticallyrelatedcells,plusaneighthwhichis
joinedtotheoriginalblockofseveninitiallyatthetopendintheleftmostfgure,then
progressivelymorecentrallyfromlefttoright.Therearetwoprincipaleffectsfrom
changingthepositionofthissingleelement.First,thetotaldepthvaluesandtheir
distributionsallchange.Second,thesumsoftotaldepthforeachfgurechange,
reducingfromlefttorightastheeighthelementmovestoamorecentrallocation.
Theeffects,however,arequitecomplex.Thisisnotofcoursesurprising,butit
illustratestwokeyprinciplesofconfgurationalanalysis.First,changingoneelement
inaconfgurationcanchangetheconfgurationalpropertiesofmanyothers,and
perhapsallothersinacomplex.Second,theoverallcharacteristicsofacomplexcan
bechangedbychangingasingleelement,thatis,changesdonotsomehowcancel
outtheirrelationstodifferentelementsandleavetheoverallpropertiesinvariant.On
thecontrary,virtuallyanychangetoelementsthatisnotsimplyasymmetricalchange,
willaltertheoverallpropertiesoftheconfguration.Wewillseeinduecoursethat
confgurationalchangesofthiskind,evensmallones,playavitalroleintheform
andfunctioningofbuildingsandbuiltenvironments.
Shapes as confgurations
Anotherwayofsayingthis,isthatdifferentarrangementsofthesamenumbers
ofelementswillhavedifferentconfgurationalproperties.Forexample,fgure3.3
isasetofrearrangementsofthesameeightsquarecellsthatweconsideredin
fgure3.2,againwithtotaldepthsinscribedineachcell,butalsowithanumber
ofothersimpleproperties,includingthetotaldepth,setoutclosetothefgure:td
istotaldepth,dbaristheaverageforeachcell,sdisthestandarddeviation, dfis
thedifferencefactorindicatingthedegreeofdifferencebetweentheminimum,
maximumandmeandepthineachcomplex(Hillieretal.1987a),andt/tisthe
numberofdifferentdepthvaluesoverthenumberofcells.
Nondiscursive technique
74
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Intreatingshapesasconfgurationsinthissense,thatis,ascomposites
madeupofstandardisedelements,weareineffecttreatingashapeasagraph,
thatis,asapurelyrelationalcomplexofsomekindinwhichwetemporarilyignore
otherattributesoftheelementsandtheirrelations.Itisclearthatsuchdescriptions
areverymuchlessthanafulldescriptionoftheshape.Formanyshapeproperties,
andformanyofthepurposesforwhichwemightseektounderstandshape,a
confgurationaldescriptionofthiskindwouldbequiteinadequateorinappropriate.
Butthereisonesenseinwhichtheconfgurationalstructureoftheshapeisa
uniquelypowerfulproperty,andgivesinsightsintopropertiesofspatialandformal
shapeswhichareincreasinglymanifestingthemselvesasthemostfundamental,
especiallyinstudiesofarchitecturalandurbanobjects.Thispropertyisthatgraphs
ofshapesandspatiallayoutsaresignifcantlydifferentwhenseenfromdifferent
pointsofviewwithinthegraph.Thiscanbedemonstratedvisuallybyusingthe
j-graph.Bydrawingj-graphsfromallnodesinashape,then,wecanpicturesome
quitedeeppropertiesofshapes.
Forexample,ahighlyinterestingpropertyofshapesisthenumberof
differentj-graphstheyhave,andhowstrongthedifferencesare.Forexample,
fgure3.4showsalldifferentj-graphsforaselectionoftheshapesinfgure3.3.
Thenumbervariesfrom3to6.Thereasonforthisisthatifwefndthatthej-graphs
fromtwonodesareidentical,thenthismeansthatfromthesetwopointsofview,
theshapehasastructuralidentity,whichweintuitivelycallsymmetry.Thisiswhyin
theshapesinfgure3.4,thesmallerthenumberofdifferentj-graphsasaproportion
ofthetotalnumberofj-graphs(thatis,thenumberofelementsinthegraph)then
themoretheshapesappearregularbecausetherearemoresymmetriesinthe
shape.Thisistheratiogivenast/t(typesovertotal)infgure3.3.Thisaspectofthe
structureofthegraphthusseemstorefectoursensethatshapescanberegular
orirregulartodifferentdegrees.
Thisanalogycanbemademoreprecise.Infact,thesymmetryproperties
ofshapescanbeexactlytranslatedasconfgurationalproperties.Mathematically,
symmetryisdefnedintermsofinvarianceundertransformation.Intheirbook
Fearful Symmetry,IanStewartandMartinGolubitskyillustratesthiswithsingular
clarity.Toamathematiciantheyargue,anobjectpossessessymmetryifitretains
itsformaftersometransformation.
15
Theyillustratethiswithadiagramshowing
thesymmetriesofthesquare,asinfgure3.5,inwhichatypicalpointintheplane
ismappedintoeightdifferentimagesbytheeightrigidmotionsthatleavethe
squareinvariant.Thinkingofsymmetriesintermsofpointsinashapeisuseful
confgurationally,sincewemayimmediatelyaskwhatwillbethecharacteristicsofj-
graphsdrawnfromeachofthepoints.Itisimmediatelyclearthatthej-graphsdrawn
fromeachofStewartspointswillbeidentical,andthatthiswouldalsobethecase
foranyothercomparablesetofpointswhichStewarthadselected.Itisalsoclear
thatonceapointhasbeenselectedtherewillonlybesevenotherpointsinthe
shapefromwhichj-graphswillbeidentical.Theprincipleisinfactverysimple:in
ashape,everysymmetrywillcreateexactlyonepointfromwhichthej-graphis
Nondiscursive technique
75
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
td=168
d=21
sd=4.58
i=.667
df=.937
t/t=.5
td=128
d=16
sd=0
i=.429
df=1
t/t=.125
td=148
d=18.5
sd=3.577
i=.524
df=.959
t/t=.375
td=142
d=17.75
sd=3.73
i=.512
df=.913
t/t=.5
td=128
d=16
sd=3.46
i=.429
df=.908
t/t=.25
td=144
d=18
sd=4
i=.524
df=.908
t/t=.75
td=142
d=17.75
sd=3.6
i=.512
df=.935
t/t=.75
td=152
d=19
sd=
i=.571
df=
t/t=.75
td=120
d=15
sd=3
i=.381
df=.949
t/t=.375
td=118
d=14.75
sd=2.72
i=.369
df=.940
t/t=.5
td=124
d=15.5
sd=3.12
i=.405
df=.923
t/t=.625
td=108
d=13.5
sd=.194
i=.310
df=.956
t/t=.5
28 22 18 16 16 18 22 28 28 21 15 13 13 17 23
21
19
16 16 16
28
19 13 13 15 19 25
12 16 22
16 12 12
18 14 14
16 14 14
16 12 12
14 12 14
14 12 14
20 14 14 16 20 26
16 14 14 16
22
22 22
22
19
19
10 14 20 14 20
16
14
20
18
14
24
22 16
14 16
16
22
18
24
20
20
15 13 15
13 11 13
19 19
16 16 16
16 16 16
Figure 3.3
_
_
_
_
_
_
_
_
_
_
_
_
Figure 3.3
Nondiscursive technique
76
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
isomorphic.Ineffect,j-graphisomorphismisatestforsymmetry.Thej-graphallows
ustolookatsymmetryasaninternalproperty,incontrasttothemoreexternalview
presupposedbytheinvarianceundermotiondefnition.Inasense,theinvariance
undermotionexistsbecausetherearedifferentpointswithintheshapefromwhich
theshapeisidentical.Wemightsaythatinashapewithsymmetrytherearepoints
withintheshapewithidentityofpositionalinformationinrelationtotheobjectasa
whole,andthisisdemonstratedbyj-graphisomorphism.
Universal distances
Thedistributionsofdepthsthatareshownthroughthej-graphs,andwhich
underliebotharchitecturalandgeometricaleffects-areinfactthemost
fundamentalideainquantifyingtheconfgurationpropertiesofspatialorformal
complexes.Theideafrstmadeitsappearanceintheliteratureofappliedgraph
theoryin1959whenHararyappliedittosociometryunderthenameofstatus.
StatusisdefnedbyBuckleyandHarary
16
thus:Thestatuss(v)ofanodevin
Figure 3.4
20 14 10 14 20
16
14
20
15 13 15
13 11 13
19 19
16 14 14 16
22
22
22
22
15 13 13 17 23
21
21
19
20 16 14 10
19 15 13 11
23 21 17 13 15 19
Figure 3.5. The eight images of a point under the symmetries of the square. The
original point is in the shaded triangle: small squares indicate transformations used.
22 16 14
Figure 3.4
Figure 3.5
Figure 3.4
20 14 10 14 20
16
14
20
15 13 15
13 11 13
19 19
16 14 14 16
22
22
22
22
15 13 13 17 23
21
21
19
20 16 14 10
19 15 13 11
23 21 17 13 15 19
Figure 3.5. The eight images of a point under the symmetries of the square. The
original point is in the shaded triangle: small squares indicate transformations used.
22 16 14
Nondiscursive technique
77
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
G(agraph)isthesumofdistancesfromvtoeachothernodeinG,distance
meaningthefewestnumberofnodesinterveningbetweenonenodeandanother.
Theproblemwithstatusdefnedinthiswayastotaldepthisthatthevaluewill
beverysubstantiallyaffectedbythenumberofnodesinthegraph.Accordingly,as
discussedinChapter1,anormalisationformulawasproposedinThe Social Logic
of Space
17
whicheliminatesthebiasduetothenumberofnodesinthegraph.
Withthisnormalisation,numericalvaluescanbeassignedexpressingtotaldepth
independentlyofthesizeofthesystem.Thisnormalisationformulawasdiscussed
andclarifedbySteadmaninArchitectural Morphology
18
Wewillcallthese
normalisedvaluesi-values,toexpresstheideaofthedegreeofintegration
ofanelementinacomplex,whichwebelievethesevaluesexpress.
Theneedforthenormalisationformulaandtheintuitionoftheformitmight
takeinfactcamefromusingthejustifedrepresentationofthegraph,orj-graph.
Simplyasaconsistentlyusedrepresentation,thej-graphmakesthestructureof
graphs,andmoreimportantlythedifferencesintheirstructures,extraordinarily
clear.However,byrepresentingtheminastandardformat,italsomakesclearthe
needforcomparativenumericalanalysisandhowitmightbedone.Forexample,
itisimmediatelyclearwhatgraphwillbemaximallyandwhatminimallydeep.It
isasimplematterfromtheretofndthenormalisation.Thefactthatnoonefound
thisusefulexpressionbefore,whenitopensupwholenewvistasfortheempirical
analysisandcomparisonofforms,ispresumablybecausenoonesaweitherits
necessityorpossibility.
However,althoughthei-valueformulaallowsthetheoreticaleliminationof
theeffectsofthesizeofthesystem,itdoesnotdealwiththefactthat,empirically,
architecturalandurbanspatialcomplexesuseonlyasmallproportionofthose
theoreticallypossible,andthisproportionshrinksasthesizeofthesystemgrows.
TheseeffectsarediscussedinfullinChapter9,andinfactbecomethebasisof
afulltheoryofurbanspatialform.Asecond,empiricalnormalisationformulawas
thereforeintroducedtocopewiththisempiricalfact.
19
Thesecondformulaisan
empiricalapproximationwithsometheoreticaljustifcation(thatitapproximates
anormaldistributionofdepthvaluesfromanynodeinagraph)andassuchit
lackselegance.Howeveritsrobustnesshasbeendemonstratedinlargenumbers
ofempiricalstudiesovertheyears,duringwhichtimenoneedhasarisentocall
itintoquestion.
20
Nodoubt,asstudiesadvance,itwillbepossibletoeliminate
thissecondnormalisationformulaandreplaceitwithanexpressionwithmore
theoreticalelegance.Inthemeantime,integrationwillrefertotheoutcomesofboth
normalisations,unlesstotaldepth(status,withnonormalisation)ori-value(status
withthefrst,theoreticalnormalisationforsize)arespecifed.Allthesetermsare
differentwaysofreferringtothesamequantity.
Whyhasthisquantityprovedsofundamentalintheempiricalstudyof
spatialandformalconfgurations?Itispossiblethatitssimplicityconcealsavery
fundamentaltheoreticalproperty:thatitisessentiallyageneralisationoftheideaof
distance.Ourcommonconceptofdistanceisthatofaspecifcnumberofmetric
Nondiscursive technique
78
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 3.6
Figure 3.7
12000
14000
16000
18000
20000
22000
24000
26000
16000 18000 20000 22000 24000 26000 28000 30000 32000 34000
Manhattan 32 x 32
m
e
t
r
i
c
2
3
x
3
2
y = .797x - 311.417, R-squared: .998
10 cm
10.0499 cm
10.198 cm
10.4403 cm
10.7703 cm
11.1804 cm
11.6619 cm
12.2066 cm
12.8063 cm
13.4537 cm
14.1422 cm
Figure 3.6
0
200
400
600
800
1000
1200
1400
1600
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Manhattan distance
m
e
t
r
i
c
d
i
s
t
a
n
c
e
y = .683x + 83.787, R-squared: .952
-100
0
100
200
300
400
500
600
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Manhattan distance
M
a
n
h
a
t
t
a
n
-
m
e
t
r
i
c
y = .317x - 83.787, R-squared: .81
3500
4000
4500
5000
5500
6000
6500
7000
7500
16000 18000 20000 22000 24000 26000 28000 30000 32000 34000
Manhattan 32 x 32
M
a
n
h
a
t
t
a
n
-
m
e
t
r
i
c
3
2
x
3
2
y = .203x + 311.417, R-squared: .972
Figure 3.7
a.
c.
b.
d.
12000
14000
16000
18000
20000
22000
24000
26000
16000 18000 20000 22000 24000 26000 28000 30000 32000 34000
Manhattan 32 x 32
m
e
t
r
i
c
2
3
x
3
2
y = .797x - 311.417, R-squared: .998
10 cm
10.0499 cm
10.198 cm
10.4403 cm
10.7703 cm
11.1804 cm
11.6619 cm
12.2066 cm
12.8063 cm
13.4537 cm
14.1422 cm
Figure 3.6
0
200
400
600
800
1000
1200
1400
1600
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Manhattan distance
m
e
t
r
i
c
d
i
s
t
a
n
c
e
y = .683x + 83.787, R-squared: .952
-100
0
100
200
300
400
500
600
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Manhattan distance
M
a
n
h
a
t
t
a
n
-
m
e
t
r
i
c
y = .317x - 83.787, R-squared: .81
3500
4000
4500
5000
5500
6000
6500
7000
7500
16000 18000 20000 22000 24000 26000 28000 30000 32000 34000
Manhattan 32 x 32
M
a
n
h
a
t
t
a
n
-
m
e
t
r
i
c
3
2
x
3
2
y = .203x + 311.417, R-squared: .972
Figure 3.7
a.
c.
b.
d.
Nondiscursive technique
79
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
unitsbetweenonepointandanotherwithinsomesystemofspatialreference.We
cancallthisaspecifcdistance.Totaldepthsumsallspecifcdistancesfromanode
toallothers.Wemaythereforethinkofitasauniversaldistancefromthatnode.If
specifcdistanceisaboutthemetricpropertiesofshapesandcomplexes,universal
distancesseemtobethekeytoconfgurationalproperties.Universaldistance
seemstobeageneralisationoftheideaofdepththatpermitsconfgurationto
becomethecentralfocusofanalysis.
Itmaybeobjectedthatsuchaconceptofuniversaldistancehasonlybeen
madepossiblethroughanunacceptablesimplifcationoftheideaofashapeto
thatofagraph,ratherthananinfnitesetofpoints.Thisisadiffculty,butitseems
thatitmightnotbeasgreatasitmightatfrstappear.Ifweconsiderasquare
shapemadeupofsquarecells,andthereforerepresentableasagraph,asinfgure
3.6,andmeasuredistancesfromandtothecentroidofeachcell,itisclearthat
graphdistanceswillapproximatemetricdistancesonlywhentheyareorthogonally
related.Onthediagonal,metricdistanceswillbeeithershorterorlongerthangraph
distances,dependingonwhetherornotweconnectthegraphdiagonallyacross
cellcorners,oronlyallowjoinsthroughthefaces.Ifcornerlinksarenotallowed,
thengraphdistanceswillben+m(orManhattandistances,byanalogywiththe
Manhattangrid)wheremisthehorizontaldistanceandntheverticaldistance,while
themetric(orasthecrowfies)distancewillbethesquarerootofmsquared+n
squared.Thiswillbemaximalbetweenoppositetopandbottomcorners.Ifdiagonal
linkstoadjacentnodesareallowed,thenthedistancebetweenoppositetopand
bottomcornerswillbemorn,whicheveristhegreater,whichequallymisrepresents
themetricdistance.Ifweplotgraphdistanceagainstmetricspecifcdistancesin
suchasystemwewillfndthatnotonlyarethedifferencessubstantial,butalso
thattheyvaryindifferentpartsofthesystem.Inotherwords,graphandmetric
specifcdistancesarenotlinearlyrelated,sowecannotuseoneasaproxyforthe
other.Figure3.7aisaplotofmetricspecifcdistanceagainstgraph(Manhattan)
specifcdistancefor1000randomlyselectedpairofpointsina100100square
cellarrangementofthetypeshowninthepreviousfgure,andfgure3.7bplots
thedifferencebetweenmetricandgraphspecifcdistanceontheverticalaxisfor
increasinggraphdistanceonthehorizontalaxis.
However,ifwesubstituteuniversalforspecifcdistances,andcarryout
thesameanalysis,thisproblemissignifcantlydiminished.Figure3.7cshows
graph(Manhattan)againstmetricuniversaldistancesforallnodesina3232
(i.e.1024cells)squarecellcomplex,andfgure3.7dplotsgraphdistanceagainst
thedifferencebetweenmetricandgraphdistances.Althoughthevaluesarestill
exactlyasdifferentoverall,theyarenowmoreorlesslinearlyrelated,sothatit
ismuchmorereasonabletouseoneasaproxyfortheother.Thisfortunatefact
permitsafarmorefexibleuseofgraphbasedmeasureofconfgurationthan
wouldotherwisebethecase.Aswewillsee,suchmattersasshapeandscale,
areaanddistancecanallbebrought,asapproximationsatleast,withinthescope
oftheconfgurationalmethod.Allwillbeinsomesensetheoutcomeofseeinga
Nondiscursive technique
80
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
complexofrelatedelementsasasetofj-graphs.Thej-graphineffectredefnesthe
elementofacomplexintermsofitsrelationtoallotherelementsinthecomplex.
Summingthepropertiesofj-graphstoexpresspropertiesofthewholecomplex
meanssummingthedifferentpointsofviewfromwhichthecomplexcanbeseen
internally.Theeventualjustifcationofthisformalismisthatarchitecturalandurban
systemsareexactlythiskindofcomplex.Theyareglobalsystemswhosestructure,
functioningandgrowthdynamicsaremanufacturedoutoftheinnumerabledifferent
pointsofviewfromwhichtheycanbeseen.
Regular shapes as confgurations
Nowletustaketheideaalittlefurther,andclosertoeverydayexperience.Itis
clearthatanyshapecanberepresentedasaregularlyconstructedmeshofcellular
elements,ortessellation,providedwecanscalethemeshasfnelyasweneed.
Thiscanthenbetreatedasagraph,andthusexpressedasapatternofuniversal
graphdistances.Bydescribingsimpleeverydayshapesinthisway,itturnsoutthat
wecancaptureimportantaspectsofhowtheyftintoeverydaylivingpatterns.
Suppose,forexample,wecreatean(approximately)circulartessellationof
arbitrarilysmallsquarecells,asinfgure3.8a.Wemaycalculatethemeandepthof
eachcellfromallothers,andexpresstheresultsinadistributionofdotdensitiesfor
thesquareelementsinwhichthehigherdensities,ordarkercolours,standforgreater
integration-thatis,lessdepthgradedthroughtolightestcoloursfortheleast
a. b.
d.
e.
c.
Figure 3.8
Nondiscursive technique
8
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
integration,orgreatestdepth.Itisclearthatthecentrehasthehighestintegration,
andthatintegrationreducesevenlyinconcentricringsaroundthecentre.Inaperfect
circle,alledgelocationswillhaveanidenticaldegreeofintegration.
Ifwethenconsiderthesquaretessellationinfgure3.8cwefndthatthepatternof
integrationnotonlyrunsfromcentretoedge,butalsofromthecentreoftheedge
tothecorners.Thesquareformisthusmorecomplexthanthecircularformina
simple,butcriticalway.Wemaysaythatinthesquareform,thecentralintegration
effectoccurstwice:onceintheglobalstructurefromcentretoedge,andoncemore
locallyoneachsideoftheform.Wecanalsoeasilycalculatethatthesquareform
islessintegrated-thatis,hasgreateraverageuniversaldistancespertessellation
element-thanthecircularform.
Asweelongatethesquareintoarectangle,asinfgure3.8d,theoverall
formisevenlessintegrated,andthepropertiesfrstfoundinthesquarebecome
moreexaggerated.Theglobalstructureoftheformisnowagroupofintegrated
centralsquares,whichincludessomeonorneartheperipheryoftheobject,
withthetwoendssubstantiallylessintegratedthanotherparts.Eachsidehas
acentraldistributionofintegration,butoneinwhichthelongsideshavemuch
greaterdifferentiationthantheshortsides,andcorrespondincreasinglytotheglobal
structureofthetessellationasweelongateit.Inthelimitingrectangulartessellation,
thesinglesequenceofsquares,thenthelocalandtheglobalstructuresareall
identical,asinfgure3.8e.
Wemaysummarisethisbysayingthatwhilealltheseformsareglobally
structuredfromcentretoedge,inthecircularformthelocalorlateralstructureis
uniform,inthesquareformthelateralstructureismaximallydifferentfromtheglobal
structure,whileintherectangularformthelocallateralstructuretendstobecome
theglobalstructureasweelongateit,untilthelimitingformofthesinglesequence
isreachedwhenthetwostructuresbecomeidentical.Thecorrespondencebetween
thesestructuresofshapesandthewaysinwhichshapeisexploitedforsocial
purposesineverydaylifeisintriguing.Forexample,onsquarediningtablesthe
centresideismoreadvantageousthancornerlocations,becauseitisamore
integratedlocation.Similarly,theEnglishprimeministersitsinthecentreofthe
longsideofabroadrectangulartable,maximisingthisadvantageinintegration.
Incontrast,wherestatusratherthaninteractionistheissue,caricaturedukesand
duchessessitatoppositeendsofalongtable,maximisingproxemicsegregation
butalsosurveillance,whilestudentsandmonksclassicallysitonthesidesofa
longthinrefectorytablewithnooneattheends,thusmakingallbutlocalised
conversationsdiffcult.Thepoliticsoflandholdingknightswithaperipateticking
sittingataroundtableareequallymanifest,asaretheendlesspoliticaldebates
overtheshapesofconferencetablesandparliamentchambers.Thewaysin
whichshapesareexploitedandusedallfollowthepatternofintegrationinsome
way,thoughwithoppositetendenciesdependingonwhetherinteractivestatusor
symbolicstatusismorecritical.
Nondiscursive technique
82
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Plans as shaped space
Nowletusconsiderthemorecomplexcaseofthehouseplan.Inthesequence
ofplansinfgure3.9iisaslightlysimplifedversionoftheplanofoneofthefarm
housesinruralFrancethatwereconsideredinChapter1.Thesallecommuneis
theeverydayspacewherecooking,eatingandthereceptionofeverydayvisitors
takeplace.Thegrandesalleisaspaceformoreformalreceptionofguests.The
workspacestotherightareadairy,washingroomandstorage,allassociatedwith
thefemaleroleinthehouse,thebureauistheoffceoftheprincipalmaleoccupant,
andthesalleisanindeterminatespace,perhapsfunctionallyassociatedwiththe
bureau.Whatdoesitmeantoanalysethisplanasashape?
Aplanis,frst,ashape,whichcanberepresentedasatessellation,see
3.9ii.Forconvenienceandspeedofanalysisweusearatherlargeelement,and
treatthresholdsassingleelements.Thisleadstosomeunrealisminwallthickness,
butthisdoesnotaffecttheanalysis.Thetessellationmaybeanalysedintoapattern
ofuniversaldistances.Sincethisrefectsthedistributionofcentralityintheshape,
inthiselongatedplantheleastuniversaldistances-showndarkest-arefoundin
thefrontcorridorbetweenthelargespacemid-right-thesallecommune-andthe
mainentrancemid-leftasin3.9iii.
Themetricdistributionofuniversaldistancesrepresentsthedegreetowhich
physicaleffortmustbemadetomovefromonepartoftheshapetoanother.Ifwe
comparetheplanshapetoasquareshapewiththesamenumberofelements
wehaveasimpleindexoftheoverallmetricintegrationoftheshape.Inthiscase,
themeanuniversaldistanceofcellsintheshapeis10.3whereasforanequivalent
squareitwouldbe4.9.Dividingtheformerintothelatter,wefndthatourshape
has2.1timestheuniversaldistanceofanequivalentlysizedsquare,indicating
thatabouttwiceasmucheffortmustbemadetomovearoundthisplanasinan
equivalentsquare.Wemaythinkofthereciprocalofthisnumberasindexingthe
degreetowhichashapegetstowardsbeingasquare.Inthiscasethevalueis
.462.Thedegreeanddistributionofuniversaldistancesthusindexessomethinglike
thephysicaleconomyoftheshape,thehumancounterparttowhichistheamount
ofphysiologicaleffortneededtoovercomeuniversaldistances.Wemayperhaps
thinkofthiswayoflookingattheplanasitsbodilyorphysiologicalstructure.It
representstheinertiaaparticularshapeofferstothehumanbodyoccupyingit.
However,aswesawinChapter1,theplanisalsoanarrangementofconvex
elements,thatis,rooms,corridors,halls,andsoon.Wecanrepresentitassuch,
again,byusingsingleelementthresholds,asin3.9iiii.Againweanalysethisfor
itspatternofintegration,thistimetreatingtheconvexelementsaselements,and
thereforeignoringactualdistancesandsizes,giving3.9v.Nowofcourse,aswas
showninChapter1,thestrongestintegratoristhesallecommune.Thoughthe
colourcodingmakesitlookthesameasthecorridor,theintegrationvalueofthe
space(.197,usingthei-valueformula)isalittlestronger(thatis,hasloweruniversal
distance)thanthecorridor(.205).Thismeansthatintermsofconvexasopposedto
metricorganisation,thefocusofintegrationhasbeendisplacedfromthegeometric
Nondiscursive technique
83
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 3.9
xi.
gr and
salle
gr and
salle
salle
commune
main
entr ance
cor r idor vestibule
work
spaces
salle bur eau
i.
ii .
iii.
iiii.
v.
vi.
vii.
viii.
ix.
x.
xii.
xiii.
xiiii.
xv.
Nondiscursive technique
84
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
centreintooneofthefunctionspaces.Thedistributionsurvivesifweaddfour
linearstripsaroundtheplantorepresenttheoutsideworld(sincetherelationtothe
outsideisoftenacriticalaspectofdomesticspaceorganisation),andreanalysefor
integration(3.9viandvii).Theoffsetsallecommunespaceisstillstrongerthanthe
centralcorridorelement.
Wenowoverlaytheconvexelementsonthetessellationshape,connecting
eachtoallthetessellationsquaresthatlieimmediatelyunderit,andre-analyse
thetwolayersasasinglesystem,sothateachconvexelementisaffectedbythe
numberoftessellationelementsitisdirectlyconnectedto,andeachtessellation
elementisaffectedbythelinksmadetoothertessellationelementsthroughthe
patternofconvexelements.Notsurprisingly,wefndthateachlayerhasaffected
thedistributionofuniversaldistancesintheother.Figures3.9viiiandixshoweach
layerofthetwo-layersystemseparately.3.9viii,theconvexlayerofthetwo-level
analysis,showsthatcomparedto3.9v,thelargespaceontheleft,thebestroom,
hasbecomerelativelymoreintegratedthantheworkspacesontherightandthe
offce.Thisisaneffectofscale.Thefactthatthemuchlargerconvexareaofthe
bestroomoverlaysfarmoretessellationsquaresthanthesmallworkroomshas
theeffectofdrawingintegrationtowardsthebestroomindirectproportiontoits
metricscale,andconverselyforthesmallrooms.Ineffect,theconvexlayerofthe
two-levelsystemshowshowthepatternofintegrationoftheconvexelementsis
affectedbytheirarea,asmeasuredbythenumberofuniformtessellationelements
eachoverlays.Thiseffectisclarifedinfgure3.9ixthetessellationlayerofthe
two-layersystem.Comparingthistofgure3.9iii,weseethattheoverlayingofthe
largerconvexelementonthetessellationsquareswithinthebestroomhasthe
effectofmakingthemmoreintegratedandmoreuniform.Theseresultsshowthat
metricscale,shape,andspatialconfgurationcanallbeexpressedinthecommon
languageofuniversaldistances,orintegration,inlayeredspatialrepresentation
consideredasunifedsystems.
Wemaytakethisalittlefurther.Anotherpotentiallayerintheplanis
thesystemoflinesofsightlinkingtheconvexelementstogetherthroughthe
doorways,assumingforthispurposethattheyareopen.Wecanrepresentthis
layerbydrawingaxialstripscorrespondingtolinesofsightasinfgure3.9xand
analyseitspatternofintegration,fgure3.9xi.Wefndthatthefrontaxispassing
throughthesallecommune,thesalleandthecorridorisnowthemostintegrating
elementbutthemainentrancefront-backlinemid-leftandthesallecommune
front-backlinemid-rightarealmostasstrong.
Wemaythensuperimposethelinearelementsontheconvexelementsand
reanalysetheseasasingletwo-levelsysteminwhichthelineelementsarealldirectly
connectedtotheconvexelementsthatlieimmediatelyunderthem.Theeffectofthis
simultaneousanalysisofthetwolayerswillbetoshowhowintegrationisshared
betweenconvexandlinearelements.Wefndthatthefrontcorridorisstillstrongest,
followedbythefront-backlinethroughthesallecommune,followedcloselybyboth
Nondiscursive technique
85
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
thefrontbacklinethroughthemainentranceandbytheconvexspaceofthesalle
communeitself.Theseresultscanbeshownbykeepingthelineandconvexsystem
together,asinfgure3.9xii,butalsobyshowingthemseparatelyforgreaterclarity.
Finallywecanassembleallthreelayersintoasinglesysteminwhich
bothconvexandlineelementsaredirectlyconnectedtoallthetessellation
elementsthatlieimmediatelyunderthem.Wethenanalyseandprintoutthethree
layersseparately,frstthetessellationlayer,fgure3.9xiii,thentheconvexlayer,
fgure3.9xiiii,andfnallythelinelayer,fgure3.9xv.Thefnalpatternemergingfrom
thethree-layeranalysisisthatthefrontaxislinkingthroughallthefrontspaceis
thestrongestintegrator,followedbythesallecommune,thegrandesalle,thelineto
thebackthroughthesallecommuneandthemainentrancelineandthesecondary
entranceline.
ComparedtothepurelyconvexanalysisoutlinedinChapter1,then,a
numberofnewsubtletieshavebeenadded.Forexample,ithasbecomeclearthat
thepotentiallineofsightlinkingroomsthroughthecorridoratthefrontofthehouse
isamorecriticalelementthanappearedintheearlieranalysis,andineffectimparts
tothehouseafront-backorganisationthathadnotemergedfromtheearlieranalysis.
Also,wecanseethattherelationbetweenwhatwemightcalltheenergyeconomy
ofthehouseplan,thatis,theamountofeffortneededtogofromonelocationto
anotherasshowninthemetrictessellation,andthehigher-levelorganisationis
quitesubtle.Ineffect,convexspaceintegrationforthemajorspacesisdisplaced
fromthemetriccentreofgravity,andthedegreeofdisplacementistosomeextent
compensatedbysize.Thusthegrandesalleismoredisplacedthanthesalle
commune,butcompensatesforthisgreaterdisplacementbyitsgreatersize.
Multi-layeredanalysissuggeststhenthatweshouldnotseeasystemofspace
asonething.Aspatiallayoutisashapewhichcontainsmanyconfgurational
potentials,eachofwhichseemstorelatetoadifferentaspectoffunction.These
potentialsmaybetreatedasindependentsystemsofspacebychoosingtoanalyse
thelayoutonthebasisofoneparticularrepresentationratherthananother,orthey
maybetreatedinselectivecombinations,orevenaltogether.Italldependsonwhat
wearetryingtofndout.
Faades as confgurations
Ifthedistributionofthevariouslayersofintegrationinashaperelatestotheways
inwhichweuseshapes,thenanintriguingpossibilitymightbethatitcouldalso
beimplicatedinhowweunderstandshapes.Forexample,buildingfaadesseen
asshapesseemcapableofbeingunderstoodascommunicatorsofinformationin
somesense.Couldconfgurationbeinvolvedinthistypeofapparentcommunication?
Considerinaveryelementarywayhowwerecogniseobjects.Thetoprow
offgure3.10showsthreefgureswhichareconstructedbyarrangingthirtysquare
elementsindifferentways.Recognisingthesefguresseemstohappenintwostages.
Inthefrststage,weidentifyadistinctshape,differentfromothers.Inthesecondwe
assignthatshapetoacategorybygivingitaname.Infgure3.10aandb,wesee
Nondiscursive technique
86
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
twoshapes.Weeasilyrecognisethedifferencebetweenthetwoshapes,thatis,we
readilymakeapureconfgurationaldistinctionbetweenthetwoobjects.Butwehave
nocategorytowhichwecanassigneitherobject.Theprocessofobjectrecognition
isthereforeendedatthefrststage.Infgure3.10cwealsoseeashape,butthistime
weconjectureacategory:theshapelookslikeanover-regularisedhumanoid,sowe
conjectureitismeanttobeeitherarobot,acaricaturehuman,orperhapsatoy.
Ofcourse,thefguredoesnotreallybearmuchresemblancetoahumanbeing
orhumanoid.Theevidenceonwhichourcategoryconjectureisbasedis,tosay
theleast,fimsy.Howeverthenatureoftheevidenceisinteresting.Itseemsto
beconfgurational.Figures3.10a,bandcarenomorethanoutlinesproducedby
rearranging30squarecellsintodifferentconfgurations.Wehave,itseems,aclear
abilitytodistinguishpureshapesorconfgurationfromeachother,priortoany
intuitionofthecategoryofthingtowhichtheconfgurationmightbelong.
Wecancallthefrstthesyntacticstageofobjectrecognition,andthe
secondthesemanticstage.Thesecondstagehasbeenextensivelydealtwith
byphilosophersandothers,butwhataboutthefrst,syntacticstage,onlynow
beinginvestigatedbycognitivepsychologists?
21
Whatdoesitmeantorecognise
aconfguration?Oneapproachtothisistoreversethequestionandaskwhat
propertiesconfgurationshavethatmightallowthemtoberecognised.Suppose,for
example,weanalysetheconfgurationsasdistributionsoftotaldepthvaluesasin
thesecondrowoffgure3.10.
Thisgivesusseveralkindsofusefulinformationabouttheconfguration.
First,thereisthedistributionofintegrationineachform,asshownbythedark-to-
lightpattern.Thiscanbethoughtofasastructurewithintheshape.Second,there
aretheintegrationcharacteristicsoftheformasawhole,asindexedbythemean
depth(md)valuesandtheirstandarddeviation(sd)asshownbeneatheachform.
Forcomparison,themeandepthandstandarddeviationforasixbyfverectangle
(thatis,aregularformwiththesamenumberofelementsandapproximatinga
squareascloselyaspossible)isalsonoted.Weseethat3.10cismoreintegrated
than3.10a,whichismoreintegratedthan3.10b,andthatallarelessintegratedthan
thesixbyfverectangle.Standarddeviationsfollowasimilarpattern.Thesedepth
valuesseemtocorrespondtocertainintuitionswehaveabouttheforms,asdo
thestandarddeviations,whichshowsthat3.10bhasgreatervariationinthemean
depthsofindividualelementsthan3.10a,whichhasmorethan3.10c,andallhave
morethanthesixbyfverectangle.
However,thereisanotherintuitionwhichisnotexpressedinthesemeasures.
Itisobviousthat3.10cismoresymmetricthaneither3.10aor3.10b,sinceithasthe
propertyofbilateralsymmetry,oneofthecommonestandmosteasilyrecognisable
typesofsymmetryfoundinartefactsorinnature.However,whilefgures3.10aand
3.10bbothlackformalsymmetries,theydonotseemtobeentirelyequivalentfrom
thispointofview.Insomesense,fgure3.10aseemstobeclosertosymmetric
organisationthan3.10b.Thereisapossiblequantifcationforthisproperty.Toexplain
Nondiscursive technique
87
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 3.10
md 4.073 sd .926 md4.609 sd 1.259
2
3
4
5
6
7
8
9
Observations
6
x
5
r
e
c
t
a
n
g
l
e
6x5rectangle
Fig1a
Fig1c Fig1b
L ine C har t f or columns: X
1
X
4
a:si= 21/30 = .7; b:si = 26/30 = .867; c: si =11/30= .367
6x5 rectanglesi = 9/30 = .3
md 5.604 sd 1.389
6x5 rectangle: md 3.554 sd .543
a. b. c.
Nondiscursive technique
88
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
it,wemustconsiderthewholeideaofsymmetryfromaconfgurationalpointofview.
Wehavealreadyseenthatpuresymmetriesinshapescouldbeinterpretedas
confgurationalproperties,namelyj-graphisomorphisms.Fromanarchitectural
pointofview,itisveryusefultoformulatepropertiesofsymmetryinthisway,
since,unlikethenormalinvarianceundermotiondefnitionsofsymmetry,itopens
thewaytoweakerdefnitionsofsymmetry,andpermitsanaccountofintuitively
importantarchitecturalpropertieswhichapproachsymmetrybutcannotbeso
formallydefned.Forexample,wecanspecifyidentityofpositionalinformationwith
respectnottothewholeobjectbuttoaregionwithintheobject,thatis,localrather
thanglobalj-graphisomorphism,anddiscusstherelationbetweenlocalandglobal
j-graphisomorphism.Buildingsarefulloflocalsymmetriestheformofawindow,
orofaparticularmasswithinacomplexwhichsometimesareandsometimesare
notrefectedinaglobalsymmetry.Therelationbetweenlocalandglobalsymmetry
seemsanaturalwaytoexpressthis.
Mostsignifcantly,wecanspecifysimilarity,ratherthanidentity,ofpositional
information,anddosoinapreciseway.Forexample,j-graphisomorphismmeans
thatj-graphssharenotonlythesamenumberofelementsandthesametotaldepth,
butalsothesamenumberofelementsateachlevelofthej-graphandthesame
connectionsbetweenelements.Onewayofweakeningthispropertywouldbeto
maintainallpropertiesexcepttherequirementthattheconnectionsbeidentical.
Anotherwouldbetovarythenumberateachlevel(fromwhichitfollowsthat
connectionswouldbedifferent)buttomaintainthetotaldepththesame.
22
Thesecondoftheseseemsparticularlyinteresting,sinceitoffersapossible
formalisationofthepropertyofbalancedasymmetryoftendiscussedinthe
literatureintheformalpropertiesofarchitecture.
23
Forexample,infgure3.11we
loadasimplelinearshapewithtwosetsoffourbytwocells,onehorizontal,the
othervertical,buteachjoinedtoexactlytwocellsinthebasicform.Althoughthe
twoendshapescreatedaredifferent,andinthemselveshavedifferentdistributions
oftotaldepthvalues(ori-values),allthevaluesinthebottomtworowsarepaired
inthateachcellhasexactlyoneothercellwhichissymmetricallylocatedandhas
thesamei-value.Thisi-valueequalityseemstogivearatherprecisemeaningtothe
ideaofbalancedasymmetry.
Wemayapplythisanalysistothethreeshapesshowninfgure3.10.
Thethirdrowshowseachshapewithcellswithequali-valuesmarkedwiththe
samenumber,fromthemosttotheleastintegrating.Weseethat3.10ahasfar
moreequali-valuesthan3.10b.Also,in3.10atheequalvaluesreachwellinto
theintegrationcoreoftheshape,whereasin3.10btheyaredistinctlyperipheral.
Bothoftheseproperties,aswellasthedegreeofintegration,canberepresented
throughasimplestatisticaldevice:thelinechartshowninthefnalrowof3.10.
Hereeachshapeisrepresentedbyaseriesofi-values,plottedfrommosttoleast
integrated(shownasleasttomostdepth),togetherwithaseriesrepresentingthe
sixbyfverectangle(shownascircles)toprovideabaselineforcomparison:3.10a
Nondiscursive technique
89
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
isrepresentedasdiamonds,3.10bastriangles,and3.10cassquares.Evidently,
theoveralldegreeofintegrationisindexedbythelocationoftheseriesonthe
verticalaxis.Thustherectangleisthemostintegrated,3.10cnext,then3.10aand
fnally3.10b.Also,theshapesdivergeastheymovefromintegratedtosegregated
elements,sothatthemostintegratedelementsineachshapearemuchcloser
togetherthantheleast.Thelinechartsalsoshowthedegreeofbalanced
asymmetryintheshapebyaligningelementswiththesamei-valuenexttoeach
othertoformahorizontalline.Theratioofthetotalnumberofelementstothe
numberofelementsthatformpartofsuchlineswillindexthedegreeofbalanced
asymmetryintheshape.Thesimplestindexisthenumberofi-valuesoverthe
numberofelements.Identicali-valueswillincludeboththoseresultingfromperfect
symmetryasshownbyisomorphicj-graphs,andthosethatonlysharethesame
totaldepth.Thissummaryfguremaythenbethoughtofasabroadsymmetry
index.Sivaluesfor3.10a,bandcarebelowthelinechart.
Integrationanalysisofshapes,then,permitsustoretrievesomeuseful
descriptionsofshapepropertiesinaconsistentway,thoughwithoutanypretence
thatthisisafullaccountofthoseproperties.Oneareawherethisapproachis
useful,however,isinconsideringbuildingsasshapes.Thekeypointhereisthat
buildingsarenotpureshapes,inthegeometricsenseoffree-standingformsin
auniformcontext,butorientedshapes,inthesensethattheyareorientedto
andawayfromthegroundonwhichtheystand.Ifwetakethissimplefactinto
accountinanalysingbuildingfaadesasshapesthenweeasilyfndsomevery
suggestiveresults.Thiscanbedemonstratedbysimplystandingshapesonaline,
whichwewillcalltheearthline.Thethreefguresoffgure3.12arethesquare
andrectangularformsshownearlierwithearthlinesadded.Inthecaseofthe
rectangularform,theearthlineisaddedtwice,oncetocreateashapehorizontally
alignedtotheearthandoncetocreateashapeverticallyaligned.
Thefrsteffectthatmustbenotedisthatinthecaseofthesquare,adding
theearthlinehastheeffectofreducingtheoriginaleightsymmetriesofthesquare
toasimplebilateralsymmetry.Thiscanbeseenvisuallyifwecomparethe
Figure 3.11
Nondiscursive technique
90
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
shadingpatternsofthesquarewithanearthlinetotheoriginalsquareform.The
concentricpatternisstillquitemarked,butnowanadditionalbilaterallysymmetric
patternisdetectable.Thiseffectresults,ofcourse,fromtheearthline,asitwere,
drawingintegrationdowntowardsitself.Thisconfrmsintuition.Itisclearthatwe
donotregardasquareashavingthesymmetriesofafree-standinggeometrical
square.Weseeitasaformanchoredtotheearthandhavingleft-rightsymmetry,
butnottop-bottomsymmetry.Indeedthelanguageinwhichwedescribetheform
-topandbottom,leftandright,showswhichrelationsweseeassymmetricaland
whichasymmetrical.
Thebilateraleffectoftheearthlineisfarmoremarkedinasquareformthan
inanelongatedform,whetherweelongatetheformhorizontallyorvertically.Inthe
verticalform,theeffectoftheearthlineistomakeintegrationrunfromthebottomof
theformtosegregationatthetop.Thisobliteratesanysenseofabilateralsymmetric
effectintheshadingpattern,andsubstitutesadifferentiationfrombottomtotop.
Addinganearthlinetoahorizontallyelongatedform,weagainfndthebilateraleffect
isbarelynoticeableintheshadingpattern,andinsteadthereisatendencytoform
broadlayersintheform,butwithmuchweakerdifferentiationfrombottomtotop.
Intermsofintegrationandsymmetryindexthedifferencesbetweenthe
verticalandhorizontalformsarealsostriking.Theverticalform,becauseofthe
greaterdistanceofmostelementsfromtheearthlineandthefactthatfarfewer
connectdirectlytoit,isalmostassegregatedastheelongatedformwithoutthe
earthline.Inthehorizontalform,however,mostelementsarenowclosertothe
earthline,withmanyactuallytouchingit,andtheeffectisthattheshapehasnow
becomemuchmoreintegratedthanthesquareform,theoppositeofthecase
withouttheearthline.
.178 .477 .124 29/65=.446 .095 20/65=.308
Figure 3.12
Nondiscursive technique
9
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Whenweconsiderthesymmetryindextheeffectsarenolessstriking.Whereas
intheoriginalshapes,thesquareformhadmoresymmetrythantheelongated
form,theadditionoftheearthlinehasoppositeeffectsontheverticalandhorizontal
forms.Theverticalformhaslesssymmetrythanthesquareform,becausefewer
elementsareonthesamelevel,whilethehorizontalformhassubstantiallymore,for
thecontraryreason.Again,thereisacommon-sensereasonfortheseeffects.The
additionofanearthlinetoaverticalformconvertsapatternofintegrationthatinthe
originalformwentfromcentretoedgetoonethatalsonowgoesfromtheearthline
which,asitwere,nowanchorstheformupwardsthroughtheform,from
moreintegrationatthebottom,closesttotheearthline,toleastatthetop,farthest
fromtheearthline.Theverticalformineffectnowrunsverticallyfromintegration
tosegregation.Inthehorizontalform,ontheotherhand,insofaraselementsare
horizontallyrelated,theywilltendtobecomemoresimilartoeachother,byvirtue
oftheirclosenesstotheearthline.Thiscorrespondstotheintuitionthatthemore
shapesarealignedalongasurface,themoreequaltheybecome.Incontrast,the
verticaldimensionstressesdifference,inthattherelationsofaboveandbeloware
asymmetrical.Horizontality,wemaysay,equalisesandintegrates,whileverticality
segregatesanddifferentiates.
Theanalysisoffaadesaslayersisalsosuggestive.Forexample,ifwe
takeasimplifedrepresentationofaclassicalfaade,wecanrepresentitfrstasa
shape,thatis,asametrictessellation,then,bydrawingthedominantelementsinthe
faade,asapatternofconvexelements.Byanalysingeachseparately,asinfgure
3.13aandb,weseethattheshape,asrepresentedbythetessellationshowsa
centralisedpatternofintegrationfocussedabove,andrunningdowninto,thecentral
column,givingthedistributionastronglyverticalemphasis.Incontrast,theconvex
analysisfocussesintegrationonthefrieze,creatingahorizontalemphasis.Onemight
conjecturethatinlookingatafaadeweseeashape,andourviewofthatshapeis
thenmodifedbythelarger-scaleorganisationofelementsimposedonthatshape.
Thesecentralisedverticalandlinearhorizontalstructureswhichare
revealedbytheanalysisare,takenseparately,amongthecommonest-perhapsthe
commonest-formalthemeswhichbuildersanddesignershavecreatedinwhole
classesofbuildingfaadesacrossmanycultures.Thefactthatanalysisdiscovers
thesestructuresseems,atleast,aremarkableconfrmationofintuition.Theanalysis
perhapssuggeststhatonereasonwhytheclassicalfaadehasoften,fromLaugier
onwards
24
beenarguedtoconstituteafundamentalmodeoffaadeorganisation,
isexactlybecausethroughitsshapeandconvexorganisationitbothexpressesand
createsatensionbetweenthetwomostfundamentalmodesoffaadeorganisation.
Ifthiswerethecase,thenitwouldsuggestthatwhatthehumanmindreadswhen
itlooksattheformofabuildingis,oratleastincludes,thepatternofintegrationat
morethanonelevel,andtheinterrelationsbetweenthelevels.
Nondiscursive technique
92
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Urban space as layers: the problem of intelligibility
Whateverthecasewithfaades,oneareawheresubstantialempiricalresearch
hasestablishedtheneedtoconsiderlayersofconfgurationalpotential,andtheir
inter-relations,isurbanspace.Consider,forexample,thetwohypotheticalurban
layoutsinfgure3.14aandb.Thetwolayoutsarecomposedofthesameblocks
orislandsofbuildings.Inthefrstcase,theyarearrangedinawaywhichhas
acertaindegreeofirregularity,butlooksmoreorlessurban,inthatthepattern
ofspacecreatedbythearrangementoftheblocksandthisisallthaturban
spaceessentiallyisseemstohavetherightkindsofspacesintherightkindsof
relations,andasaresultappearsintelligibleasanurbansystem.Inthesecond
layout,alltheblocksarethesamebuteachhasbeenmovedslightlywiththe
effectthatthesystemofspaceseemsmuchlessurban,andmuchlesseasily
intelligible.Itisclearthatanyusefulanalysisofurbanspacemusteithercapture
theseintuitionsorshowwhytheyareillusory.Itwillturnoutthattheyarenot
illusoryatall,andthattheyarisefromwell-defnedrelationsamongstthedifferent
spatialpotentialsthatmakeupthelayout.
25
Inonesense,bothlayoutsrepresentthecommonesttypeofurbanspace
structure.Wecancallitthedeformedgrid,becausewhilemadeupofoutward
facingislandsofbuildingseachsurroundedonallsidesbycontinuousspacein
themannerofaregulargrid,thestructureofthatspaceisdeformedintwoways:
itislinearly,oraxiallydeformed,inthatlinesofsightandaccessdonotcontinue
rightthroughthegridfromonesidetotheother,astheywouldinaperfectlyregular
grid,butcontinuallystrikethesurfacesofthebuildingblocksandchangedirection
asaresult;secondlyitisconvexlydeformedinthattwo-dimensionalspaces
continuouslyvaryintheirdimensionsandshape,makingapatternofwiderand
narrowerspaces.Thevisibilityfeldatanypointinthespaceforsomeonemoving
a. b.
Figure 3.13
Nondiscursive technique
93
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
inthegridwillbemadeupofbothkindsofelement.Wherevertheobserveris,
therewillalwaysbealocalconvexelementofsomekind,inwhicheverypointis
visiblefromeveryotherpoint,plustheshapemadebyalllinesofsightandaccess
passingthroughthepoint.Theeasiestwaytodescribethedifferencesbetween
thetwolayoutsintuitivelyistosaythatamovingobserverineitherlayoutwould
experiencecontinuouschangesinthevisibilityfeld,butthatthekindsofvisibility
feldexperiencedinthefrstarequitedifferenttothoseinthesecond.Theapparent
differencesinintelligibilityinthetwolayoutswillturnouttoberelatedtothese
formaldifferencesinthesuccessionofvisibilityfelds.
Wecanbuildupananalysisofthetwolayoutsbyinvestigatingthesedifferent
potentials.First,wewillconsidertheoverlappingconvexelementsthataredefned
bythesurfaceofthisblock.
26
Hereconvexelementsaredefnedbyreferencetothe
surfaceofeachblock,eachofwhichdefnesitsmaximalconvexfeld.Thesefelds
willinevitablyoverlap,andwheretheydo,theareaofoverlapwillitselfformasmaller
convexelementfromwhichbothoverlappingconvexspaceswillbefullyvisible,
thatis,willbeconvex,althoughthesespacesarenotconvextoeachother.The
samewillbetruewhenfurtheroverlappingspacesareadded.Certainsmallspaces
willindeedbeconvextoasubstantialnumberofconvexspacesbecauseallthose
spacesoverlapinthatarea.Suchareaswillasaresulthavelargevisibilityfelds,
whereasareaswherethereisnooverlapwilltendtohavemuchsmallervisibility
felds.Overlappingconvexelementsarevirtuallyimpossibletointuit,becausethe
overlappingissodiffculttorepresent.Computeranalysisisthereforerequired.
Letuslookfrstatthepatternofoverlappingconvexspacesgeneratedin
ourtwolayouts.Figures3.14candd,aretheresultoftheanalysisoftheopen-space
structureofthetwolayouts.Thecomputerhasfrstdrawnalltheoverlappingconvex
elementsdefnedbythefacesofeachblockandthencarriedoutanintegration
analysisofthepattern,withintegrationtosegregationshownfromdark-to-light,as
before.Inthefrsturbanlayout,thedarkestspacesoftheresultingintegrationcore
(theshapemadebythedarkestareas)crosseachotherintheinformalmarket
square,anddarkspaceslinkthemarketsquaretowardstheedgeofthetown.In
thesecond,thereisnolongerastrongfocusofintegrationlinkingasquaretothe
edgesofthesystemand,ineffect,theintegrationcorehasbecomediffused.Infact,
themostintegratingspacesarenowfoundattheedge,andnolongergettotheheart
ofthesystem.Onaverage,thelayoutasawholeismuchlessintegratedthanthe
frst,thatis,ithasmuchgreatertotaldepthfromallspacestoallothers.
Inotherwords,themarginalrearrangementoftheurbanblocksfromthe
frsttothesecondlayoutresultinginaspatialstructurewhichisquitedifferentboth
inthedistributionandinthedegreeofintegration.Intuitively,wemightsuspectthat
theedge-to-centreintegrationcorestructureofthefrstlayouthasmuchtodowith
theoverallsenseofurbanintelligibility,anditslossinthesecondlayout.Intelligibility
isachallengingpropertyinanurbansystem.Sincebydefnitionurbanspaceat
groundlevelcannotbeseenandexperiencedallatonce,butrequirestheobserver
tomovearoundthesystembuildingupapictureofitpiecebypiece,wemight
Nondiscursive technique
94
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
suspectthatintelligibilityhassomethingtodowiththewayinwhichapictureof
thewholeurbansystemcanbebuiltupfromitsparts,andmorespecifcally,from
movingaroundfromoneparttoanother.
Thereisinfactasimpleandpowerfulwayinwhichwecanrepresent
exactlythisproperty.Itisillustratedinthetwoscattergramsinfgures3.14eand
f,correspondingtothetwolayouts.Eachpointinthescatterrepresentsoneof
theoverlappingconvexspacesinthefgureabove.Thelocationofthepoint
ontheverticalaxisisgivenbythenumberofotherconvexspacesthatspace
overlapswith,thatis,theconnectivityofthespacewithotherspaces,andonthe
horizontalaxisbytheintegrationvalueofthespace,thatis,itsdepthfromall
others.Nowconnectivityisclearlyapropertythatcanbeseenfromeachspace,
inthatwhereveroneisinthespaceonecanseehowmanyneighbouringspaces
itconnectsto.Integration,ontheotherhand,cannotbeseenfromaspace,since
itsumsupthedepthofthatspacefromallothers,mostofwhichcannotbeseen
fromthatspace.Thepropertyofintelligibilityinadeformedgridmeansthedegree
towhichwhatwecanseefromthespacesthatmakeupthesystem-thatis,how
manyotherspacesareconnectedto-isagoodguidetowhatwecannotsee,that
is,theintegrationofeachspaceintothesystemasawhole.Anintelligiblesystem
isoneinwhichwell-connectedspacesalsotendtobewell-integratedspaces.An
unintelligiblesystemisonewherewell-connectedspacesarenotwellintegrated,
sothatwhatwecanseeoftheirconnectionsmisleadsusaboutthestatusofthat
spaceinthesystemasawhole.
Wecanreadthedegreeofintelligibilitybylookingattheshapeofthe
scatter.Ifthepoints(representingthespaces)formastraightlinerisingat45per
centfrombottomlefttotopright,thenitwouldmeanthateverytimeaspacewasa
littlemoreconnected,thenitwouldalsobecomealittlemoreintegrated-thatisto
say,therewouldbeaperfectcorrelationbetweenwhatyoucanseeandwhatyou
cantsee.Thesystemwouldthenbeperfectlyintelligible.Infgure3.14e,thepoints
donotformaperfectline,buttheydoformatightscatteraroundtheregression
line,whichisevidenceofastrongdegreeofcorrelation,andthereforegood
intelligibility.Infgure3.14fwefndthatthepointshavebecomediffusedwellaway
fromanyline,andnolongerformatightftabouttheregressionline.Thismeans
thatconnectivityisnolongeragoodguidetointegrationandthereforeaswemove
aroundthesystemwewillgetverypoorinformationaboutthelayoutasawhole
fromwhatweseelocally.Thisagreesremarkablywellwithourintuitionofwhatit
wouldbeliketomovearoundthislabyrinthianlayout.
27
Nowletusexplorethetwolayoutsinmoredetail.Infgure3.14gandh,we
haveselectedapointinthesquareintheanalysisofthefrstlayout,anddrawn
alltheoverlappingconvexelementsthatincludethispoint.Thescatterthenselects
thesespacesinthescattergrambymakingthemcolouredandlarger.Wecansee
thatthespacesthatoverlapatthispointareamongthebestconnectedandmost
integratedinthelayoutandthatthepointsalsoformareasonablelinearscatterin
themselves,meaningthatforthesespacesmorevisibleconnectivitymeansmore
Nondiscursive technique
95
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 3.14a Figure 3.14b
Figure 3.14c Figure 3.14d
Figure 3.14e Figure 3.14f
Nondiscursive technique
96
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Figure 3.14l Figure 3.14m
Figure 3.14j Figure 3.14k
Figure 3.14g Figure 3.14h
Nondiscursive technique
97
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
integration.Boththeshapemadebythesetofspaces,reachingoutfromthesquare
inseveraldirectionstowardstheedgeofthesystem,andthescattergramproperties
confrmthatthispointinthesquarespacehasahighstrategicvalueinthelayout
asawhole.Ifwetrytodothesameforpointsinthesecondlayout,asinfgure3.14j
andk,wefndthatthepointsareburiedinthescatterandhavenospecialstrategic
value.Byexperimentallyclickingonaseriesofpoints,andcheckingboththevisual
feldsandthescattergrams,onecanestablishthattherearenocomparablestrategic
pointsfromwhichaseriesofkeyspatialelementsinthelayoutcanbeseen.
Wemayalsoexperimentwiththeeffectsofchangestothelayout.Suppose,
forexample,wedecidethatthecurrentmarketsquare,althoughstrategically
placed,istoosmallandthatitshouldthereforebemovedelsewhereinorderto
enlargeit.Infgure3.14landm,theoldmarketsquarehasbeenbuiltoveranda
new,largersquarehasbeencreatedtowardsthetopleftofthelayout.Thelayout
hasbeenanalysedandtheconvexelementsoverlappinginthenewsquarepicked
out.Inspiteofitssize,thenewsquarehaspoorintegration,anditsoverlapping
Figure 3.14q Figure 3.14r
Figure 3.14n Figure 3.14p
Nondiscursive technique
98
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
spacesoccupyapoorpositioninthescatter.Themostintegratedspacesremain
thosepointingintotheoldmarketsquare.Inotherwords,thespatialconfguration
asawholecontinuestopointtotheoldsquare.Animportantconclusionfrom
this,amplyconfrmedbytheexaminationofrealtownplans,isthatasquareis
morethanalocalelement.Howitisembeddedintheconfgurationasawhole
isequally,ifnotmore,important.Ifweweretoseektoexploitthisbyexpanding
theoldmarketsquarebyremovingadjacentblocks,wewouldfndthesquare
becomesmuchmoredominant,andthatthelargestspacewithinthesquare(i.e.as
opposedtothoseenteringandleavingwhicharenormallymoredominant)isnow
itselfthesecondmostintegratedspace.Inotherwords,wewouldbegintoshiftthe
emphasisofintegrationfromlinearelementstotheopenspaceitself.Again,this
woulddistorttheessentialnatureoflayout.Thesize,location,andembeddingof
majoropenspacesareallformallyconfrmedasaspectsofwhatweintuitivelyread
astheurbannatureofthelayout.
Convexelementsarenot,ofcourse,themostglobalspatialelementsina
layout,anddonotexhaustallrelationshipsofvisibilityandpermeability.Theselimits
arefoundbylookingnotattwo-dimensionalconvexelements,butatone-dimensional
lineelements.Inadeformedgrid,theelementsmostspatiallyextendedlinearlywill
bethesetofstraightlinesthataretangenttotheverticesofblocksofbuildings.
Relationsbetweenpairsoftheseverticesineffectdefnethelimitsofvisibilityfrom
pointswithinthesystem.Thiscanbeexploredthroughaxialoralllineanalysis,and
infgure3.14n-rwherethecomputerhasfoundandcarriedoutanintegrationanalysis
ofallthelineelementstangentialtoblockvertices.Wefndthattheintelligibilityof
thesystemseenaxiallyisbetterthanseenconvexly,becauselinesaremoreglobal
spatialelementsthanconvexelements,inthattheyexplorethefulllimitsofvisibility
andpermeabilitywithinthelayout.Linesthereforemaketherelationbetweenthe
localspatialelementandtheglobalpatternofspacelookasgoodaspossible.The
differencesbetweenthetwolayoutsthatwefoundthroughtheoverlappingconvex
analysisarehowevermoreorlessreproducedintheall-lineanalysis.Thisagreement
betweenthetwokindsofanalysisisitselfasignifcantpropertyofthelayouts.
Fromthepointofviewofhowlayoutswork,bothtypesofanalysisare
important.Movement,forexample,canbepredictedfromastrippeddownversion
oftheaxialanalysisinwhichonlythelongestandfewestlinesneededtocover
thewholesystemformthelinematrix.Similarly,manyaspectsofstaticurban
behaviours,especiallytheinformaluseofopenspaces,exploitthetwo-dimensional
visibilityfeldpropertiesofspace,withthehighestlevelsofusenormallyadjacent
tothemoststrategicspaces.
Designing with confgurational models
Becausethesetechniquesallowustodealgraphicallywiththenumericalproperties
ofspatiallayouts,wecanalsousethemcreativelyindesign,bringinginmuchnew
knowledgeaboutspaceandfunctionaswedoso.Forexample,extensiveresearch
hasshown
28
thatpatternsofmovementinurbanareasarestronglypredictedbythe
Nondiscursive technique
99
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
distributionofintegrationinasimplelinerepresentationofthestreetgrid.Byusing
confgurationalanalysistechniquesinsimulationmode,wecanexploitboththis
knowledge,andthepotentialforconfgurationalanalysistogiveinsightintopossible
urbanpatternsthatwillnotbeatallcleartointuition.Thispotentialhasnowbeen
exploitedinalargenumberofurbandesignprojects,ofteninvolvingthemodellingof
wholecitiesinordertosimulatetheeffectsofnewdesigns.
29
Todemonstratetheessentialsofthetechnique,asimplifedhypothetical
modelwillsuffce.Thetopleftfgureoffgure3.15isananalysedaxialmap
(thelongestandfewestlinesthatcoverthestreetgrid)ofasmallareaarounda
hypotheticalredevelopmentsite,withintegrationfromdarktolightasbefore,with,
toitsright,thescattergramofitsintelligibility,showingaweaklyintelligiblesystem.
Wecanexperimentbyasking,whatwouldhappenif,forexample,weimposed
aregulargridonthesitewithouttakingtoomuchaccountofthesurrounding
structure,asthesecond-rowfgureandscatter.Weseethatinspiteofthe
geometricregularity,ourlackofconcernfortheglobalpatternhasleftuswitha
ratheruniformlysegregatedspacepatternwithinthesite,withtoopoorarelationto
thesurroundingareas.Asaconsequence,weseefromthescatterthattheareaas
awholehasbecomeevenmoreunintelligible.
Supposewethengotheotherway,andtrytodesignthesitebyextending
stronglines,andlinkingthemtoothers,asinthethirdrowfgureandscatter.The
resultisanintegratingsite,andgoodintelligibility.Thespatialstructureinthesite
alsohasagoodrangeofintegratedandsegregatedspaceincloseproximityto
eachother.Aswewillseeinlaterchaptersthisisanimportanturbanproperty
(seeChapters4and5.)Thisisasimpleexample,butitshowstheabilityof
confgurationalanalysisnotonlytoaidthedesignersintuitioninthinkingabout
patterns,andinparticularintryingtounderstandthepatternconsequencesof
individualdesignmoves,butalsoitsabilitytopermitthedesignertothinkmore
effectivelyabouttherelationofnewandexistingpatterns,andingeneralaboutthe
relationofpartsandwholesincities.
Wemayagainillustratethisbyasimplifedsimulation.Plate1istheaxial
mapofahypotheticalurbansystemwithwell-defnedsub-areas.Researchhas
shownthatthecriticalthingabouturbansub-areasishowtheirinternalstructures
relatetothelarger-scalesysteminwhichtheyareembedded.Thebestwayto
bringthisoutistoanalysethesystemforitsintegrationattwolevels.Firstwedo
ordinaryintegration,whichcountshowdeeporshalloweachlineinisfromevery
otherline.Secondwecounthowdeeporshalloweachlineinisfromalllinesupto
threestepsaway.Thelatterwecallradius-3integration,sinceitlooksateachline
uptoaradiusof3.Theformerwecancallradius-nintegration.Radius-3integration
presentsalocalisedpictureofintegration,andwecanthereforethinkofitalsoas
localintegration,whileradius-nintegrationpresentsapictureofintegrationatthe
largestscale,andwecanthereforecallitglobalintegration.
Wewillseeinduecoursethatlocalintegrationinurbansystemsisthebest
predictorofsmaller-scalemovement-thatusuallymeanspedestrianmovement
Nondiscursive technique
00
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax Figure 3.15
a
b
c
Figure 3.15
Nondiscursive technique
0
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
becausepedestriantripstendtobeshorterandreadthegridinarelativelylocalised
way-whileglobalintegrationisthebestpredictoroflarger-scalemovement,including
somevehicularmovement,becausepeopleonlongertripswilltendtoreadthegridin
amoreglobalisedway.Inhistoricalcities,aswillbeshown,therelationshipbetween
thesetwolevelsofintegrationhasbeenacriticaldeterminantofthepart-whole
structureofcities,becauseitgovernsthedegreeofnaturalinterfacetherewould
naturallybebetweenmorelocal,andthereforemoreinternalmovement,andmore
globalandthereforemorein-outmovementandthroughmovement.
Someofthedifferenteffectsonthisrelationshipthatdifferenttypesoflocal
areadesignwillhavecanbeshownbyhighlightingtheareasinscattergramsof
thewholesystemandexaminingthescatteroflocalagainstglobalintegration.The
areashowninthebottomrow,forexample,isaclassicallystructuredareafora
Europeancity,withstronglinesinalldirectionsfromedgetocentre,withaless
integratedstructureoflinesrelatedbothtothisinternalcoreandtotheoutside.
Thisensuresthatthosemovingintheareawillbeconsciousofboththelocaland
globalscalesofspaceastheymovearound,andtherewillbeagoodinterface
betweenlocalandglobalmovement.Thescatterformedbythesub-areasisshown
totheright.Thepointsoftheareaformagoodlinearscatter,showingthatlocal
integrationisagoodpredictorofglobalintegration,andcrosstheregressionline
forurbanareaasawholeatasteeperangle,showingthatthereisastronger
degreeoflocalintegrationforthedegreeofglobalintegration.Alineonthecore
ofthewholesettlementwill,incontrast,lieatthetopendofthemainregression
line.Thisshowshowsubtlyurbanareascreateasenseoflocalstructurewithout
losingtouchwiththelarger-scalestructureofthesystem.(SeeChapter4foran
examinationofrealcases).
Theareashownimmediatelyabove,inthesecondfrombottomrow,is
typicalofthelayoutswetendtofndinhousingestates,withfewconnectionsto
theedgeandlittlerelationbetweentheedgetocentrestructureandtheinternal
structureofthelayout.Thistypeoflayoutisinvariablyshownasaseriesoflayers
intheredpointscatterwithvirtuallynocorrelationbetweenlocalandglobal
integration.Suchlayoutsinvariablyfreezeallournaturalmovementandbecome
structurallysegregatedlumpsintheurbanfabric.
30
Theareasinthetoptworows
showothervariationsonlocalareastructure,oneproducingeffectsrathersimilar
tothoseintheexperimentalgridinthedesignexperimentoffgure3.16,whilethe
otherisarandomscatteroflines,showingthatinspiteoftheapparentinformality
ofmuchgoodurbandesign,randomlinessimplydonotworkexceptbychance.
Future urban models: intelligent analogues of cities
Inadditiontotheirroleindesign,confgurationalmodelsarenowbeingdeveloped
asabasisforresearchingintothemultidimensionaldynamicsofcities.Consider,
forexample,oneofthebroadestandleasttractableofissuesfacingthebuilt
environmentindustry:thatoftheeconomic,socialandenvironmentalsustainability
ofcities.Eventomonitoreffectivelyandcomparecitiesonsustainabilitycriteria,
Nondiscursive technique
02
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
whatevertheymightturnouttobe,wemustbringdataonthephysicaland
environmentalperformanceofcitiestogetherwithdataontheireconomicand
socialperformance,andtorelatebothtosomekindofdescriptionofthecity.For
example,energyconsumptionandpollutionproductiondepend,amongother
factors,onsettlementpatterns.Shouldsettlementsbedenseorsparse,nucleated
ordispersed,monocentricorpolycentric,oramixofalltypes?Forresearchtogive
ananswer,measurementdataonenvironmentalperformance,anddataonthe
implicationsofdifferentbehaviouralassumptions(forexampleaboutthedistribution
ofworkandhome)andknock-oneffectssuchastheeconomic,socialandcultural
consequencesofspatialaggregationanddisaggregationpolicies,mustberelated
todescriptionsofthephysicalandspatialformofcitieswhichrefecttherangeof
variationfoundintherealworld.
Toworktowardsatheoreticalmodelofhowthismightbedone,wemay
beginwiththepurelyconfgurationalmodelswehavepresented,andshowhow
otherkeyspatialattributessuchasmetricdistance,area,density,plotratios,
shape,politicalboundaries,andsooncanbeexpressedwithintheconfgurational
modelbyusingtheideaofintegratinglayeredrepresentationsofspaceintoa
singlesystem.Forthepurposesofillustrationwewillagainusenotional,simplifed
examples.First,werepresentastreetnetworkasaseriesoflinesorstrips,and
analysetheirpatternofintegration,asinfgure3.16aandb.Inthisanalysis,no
accounthasyetbeentakenofmetricdistance.However,insomecircumstancesat
least,thisseemslikelytobeanimportantvariable.Wecansupplythisbyselecting
anarbitrarymodule-sayaten-metresquareandlinkingmodulesintothepattern
ofthegridandanalysingthisasatessellationshape,asinfgure3.16c.Onits
own,thisisnotofgreatinterest,sinceitinevitablyrefectsthepatternofmetric
centralityinthegrid,asinfgure3.16d,butifwesuperimposethelinenetwork
ontothemetricmodularsystemandanalysethetwolayersasasinglesystem,
thentheeffectistoweighteachlinewithanumberofmodulesdirectlyrelatedto
itslength.Theoutcomeofthislengthweightedintegrationanalysisisshownat
bothlevelsofthecombinedanalysis:intermsofthemodularunitsinfgure3.16e,
andintermsofthelinesuperstructureofstripsinfgure3.16f.Thestriplevelis
muchthesameaspreviously,butthemodularelementsshowaninteresting-and
verylifelike-localisedstructureinwhichgreaterintegrationisconcentratedatthe
streetintersections,withlessintegratedmodulesinthecentresoflinksawayfrom
theintersections.Thisimmediatelyenablesustocaptureanewandfunctionally
signifcantaspectofspaceorganisationinarepresentation.
Therelationshipbetweenmetricareaandconfgurationcanbedealtwithin
ananalogouswaybyunderlayingconvexelementswithatwo-dimensionalmodular
layer,asinfgure3.17a-f.Inacweseehowasimplesysteminwhichfourconvex
spacesofequalsizeandshapeandtheconnectionsbetweenthemarerepresented
asalayerofmodularelementswithfourconvexelementsandfourstripsforthe
connectionsuperimposed.Thetwo-layersystemisthenanalysed.Whetherwelook
attheresultwiththeconvexlayeruppermostorthemodularlayer,theresultswill
Nondiscursive technique
03
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
a. b.
c. d.
e. f.
Figure 3.16
Nondiscursive technique
04
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
beasymmetricaldistributionofintegrationdominatedbythestrips.Infgure3.17
dfwegivetheconvexelementsdifferentareasandunderlaymodularelements
accordingly,sothateachisnowweightedbythenumberofmodularelementsit
overlays.Analysisseparatelythentogethershowsthatintegrationisdrawninto
theconvexelementsaccordingtotheirarea.Notehoweverthattheintegrationof
thetwosmallerconvexareas(onthetop)areinthewrongorder.Thisisbecause
theoneontheleftisclosertothelargest-scaleconvexarea(bottomleft)andthis
affectsitsownintegrationwithrespecttotherestofthesystem.Thustheresults
showacombinationofconfgurationaleffectsandmetricareaeffects.Fromthiswe
canseethatifwemakealargeandsmallsquareconfgurationallyequivalentinan
urbansystemthenthelargesquarewillintegratemore.Metricarea,itturnsoutis
likedistance,apropertycapableofexpressionasanaspectofconfguration.
Wemaysimulatetheeffectofplotratiosanddensitiesbyequallysimplemeans.For
example,ifwewishtoattachabuildingwithagivennumberoffoorstoastreet
network,allweneedtodoisattachaconvexspacethesizeofthegroundareaof
thebuildingtotheappropriatepositioninthestreetsystem,thenoverlayonthata
convexelementforeachfoor,makingsurethateachelementabovethegroundis
detachedfromthestreetandonlyconnectedthroughthegroundlayerasitwould
beinreallife.Thiswillnotappearvisuallyasathree-dimensionalstructure,butit
willexactlyrepresenttheadditionofabovegroundfoorspacetotheurbansystem.
Wemaynowbuildamodelofanurbansysteminthefollowingway.
First,wedividethecityupintoanarbitrarynumberofareasandrepresentthem
asnon-contiguouspolygons.Thesemaybeassmalloraslargeasweneed,
accordingtothelevelofresolutionrequiredbytheresearchquestion.Thepolygons
maybebasedonpoliticalboundaries,likewards,administrativeboundarieslike
enumerationdistricts,segmentsdefnedbyanarbitrarilyfnegrid,ortheymay
bedefnedbyobjectivemorphologicalpropertiesofthebuiltenvironment.These
polygonsrepresentingareasarethefundamentalunitsofanalysisforthetechnique.
Figure3.18ashowsourimaginarysimplifedcaseinwhichthestreetnetwork
ofthecity(orpart-city)issuperimposedonthepatchworkofpolygonssothateach
polygonislinkedintotheurbansystembyallthestreetsorpart-streetsthatpass
throughitoralongsideit.Thistwo-levelspatialsystemisanalysedconfgurationally
tofndthepatternofintegrationinthewholesystem.Evidently,thestreetpattern
willtendtodominatetheareapolygonssimplybecausethestreetsareconnectors.
However,thestreetsystemcanthenbepeeledoffthepolygons,asinfgure3.18b,
leavingapatternofpolygonswiththeirspatialcharacteristicsinrelationtothecity
areaaroundthem,andtothecitysystemasawhole,recordedasasetofnumbers.
Thisbasicprocessoflinkingareastogetherbythestreetnetworkinasingle
confgurationalmodelisthebasisofwhatwecallanintelligenturbananalogue
model.Oncethisisestablished,wecanthencomplicatethemodelinalltheways
wehavedescribedpreviously.Forexample,wecanunderlaythestreetnetwork
withmetricmodulessothattheanalysisofthestreetsystemtakesdistancesinto
account.Wecanunderlaythepolygonswithmetricmodulessothatthemetricarea
Nondiscursive technique
05
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
f.
d. e.
c. b. a.
Figure 3.17
Nondiscursive technique
06
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
ofapolygonistakenintoaccount.Wecanalso,ifwewish,superimposelayerson
thepolygonsrepresentingoffthegroundfoorspace.
Thereisalsoaneasywayoffurtherdisaggregatinganymodelfromthe
levelofresolutionoriginallyselected.Eachoftheoriginalareapolygonscanbe
itselfsubdividedintomuchsmallerpolygonsandanalysedasbefore.Thismore
localisedanalysiswillgiveamuchricheranddenserpictureofthedetailed
characteristicsofthearea.Thesemaythenbefedintoalarger-scalemodelas
moredetailedenvironmentaldescriptors.Thereisnoreasoninfactwhybothlevels
ofthemodelshouldnotbeanalysedasasinglesystem.Theprincipalbarrierwould
becomputingtime.Inourexperienceaddinganewleveloffnestructuretoan
existingmodelleavesthelarger-scalepicturemoreorlessintactprovidedthatthe
disaggregationisdoneuniformlyandisnotconfnedtoparticularregions.
Attheotherendofthescale,wemayalsoderivenewmeasuresofthemost
macro-propertiesofthecitysystem,suchasshape,andshapeloadedwithdifferent
densitiesindifferentregions.Thiscanbedonebysimplylinkingtheareapolygons
togetherandanalysingthedistributionofintegrationinthesystemwithoutthe
superimposedstreetsystem.Shapewillbeindexedbythedegreeanddistribution
ofintegration,andcanbeshownbothbydirectgraphicalrepresentationofthecity
system,orbystatisticalrepresentationssuchasfrequencydistributions,orsimply
bynumbers.Theeffectsofweightingshapesbyloadingdifferentregionswith
higherdensitiescanbeexploredbysimplyoverlayingthespacesrepresentingthe
additionaldensitiesontotherelevantpolygonsofthecontiguouspolygonsystem,
thenproceedingasbefore.Byvaryingthepatternanddensityofcentreswecan
exploretheireffectsontotaldistancetravelled,otherthingsbeingequal,indifferent
kindsofthree-dimensionalurbansystem.Theeffectsofothernearbysettlements
canalsobeinvestigatedbysimplyaddingthemasextensionstothemodel.
Thenumericaldataresultingfromtheanalysisoftheurbansystemcanthen
beusedinanumberofways.First,mostobviously,theparametricdescriptorsforthe
polygonsresultingfromanalysis,refectingastheydothepositionandconfguration
ofeachfniteelementinthecitysystemasawhole,thenbecometheframefor
otherkindsofdatawhichcanbeassignedasdescriptorstothepolygons.Thiscan
bedonewithanyfunctionalvariablethatcanbenumericallyindexedforthatarea
suchaspopulationdensities,pollutionlevels,traffcmovement,pedestrianmovement,
unemploymentrates,crimerates,counciltaxbanding,andsoon.Becausespatial
andotherdescriptorsarenowallinnumericalform,simplestatisticalanalysescan
begintorevealpatterns.Second,thedistributionofanypropertymayberepresented
graphicallyintheurbansystemasavisualdistributionofthatpropertyinthecity
system.Thismeans,inpractice,thatallthevisualisingandcartographicalpotentials
thathavebeendevelopedinthepastfewyearsthroughgeographicinformation
systemscanbeinterfacedwith,andpotentiallybroughtwithinthescopeof,an
analyticmodelwithprovenabilitytolinkmorphologicalandfunctionalpropertiesof
builtenvironmentsystems,hopefullyinamorepredictiveway.
Layeredmodelsarethefutureofconfgurationalmodellingofspace.
Nondiscursive technique
07
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
b.
a.
Figure 3.18
Nondiscursive technique
08
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Thesenewtechniquesarisefromtheresultsofresearchoverseveralyearsin
whichvarioustypesofconfgurationalmodellinghavebeenusedfrsttoidentify
non-discursiveregularitiesinthewaysinwhicharchitecturalandurbansystems
areputtogetherspatiallyandidentifythegenotypesofspatialform;secondto
correlatethesenon-discursiveregularitieswithaspectsofhowhumanbeingscan
beobservedtofunctioninspace;andthird,tobegintobuildfromtheseregularities
apictureofhighergeneralityofhowspatialsystemsingeneralareputtogetherand
functioninresponsetothedemandsthathumanbeingsandtheircollectivitiesmake
ofthem.Inthenextchapterweintroducethemostfundamentalofallcorrelates
withspatialconfguration:humanmovement.

Notes
H.SimonH,The Sciences of the Artifcial,MIT,1969.
F.DeSaussureF,Course in General Linguistics,McGrawHill,1966translatedbyC.
BallyandA.SechahayewithA.RiedlingerSeepp.915(originallyinFrench1915).
IthasofcoursebecomefashionabletofollowthelaterWittgensteinsPhilosophical
Investigations (BasilBlackwell1953;Editionused1968)anddenyanysystemic
propertiestosuchthingsaslanguages,andseeinthemonlyshiftingcontingencies.
Forexample:Insteadofproducingsomethingcommontoallthatwecalllanguage,
Iamsayingthatthesephenomenahavenoonethingincommonwhichmakesus
usethesamewordforall,buttheyareallrelatedtoeachotherinmanydifferent
ways.Anditisbecauseofthisrelationship,ortheserelationships,thatwecallthem
alllanguage.Wittgenstein,para65.Or:Languageisalabyrinthofpaths.You
approachfromonesideandknowyourwayabout;youapproachthesameplace
fromanothersideandnolongerknowyourwayaboutWittgenstein,para203.The
useoftheurbananalogyisinteresting.Aswewillseeinlaterchapters,thisisthe
onetypeofartefactwhereitcanbeshownquiteclearlythatWittgensteinwaswrong.
ThecleareststatementisstillprobablytheOverturetoClaudeLvi-StrausssThe
Raw and the Cooked,JonathonCape,London1970,originallyinFrenchasThe Cru
et le Cuit,Plon,1964.
Plato,The Republic,forexampleVI,50911,pp.7447inPlato,The Collected
Dialogues,eds.E.HamiltonandH.CairnsH,PrincetonUniversityPress,Bollingen
Series,1961.Seealsoed.F.M.Cornford,The Republic of Plato,OxfordUniversity
Press,1941,pp.21621.
Fortheclearestformulation,seeR.Thom,StructuralismandBiology,ined.C.H.
Waddington,Theoretical Biology 4,EdinburghUniversityPress,1972,pp.6882.
W.Heisenberg,Physics and Philosophy GeorgeAllen&Unwin,1959p.57.
N.Chomsky,Syntactic Structures,Mouton,TheHague,1957.
Thereareimportantexceptionstothis,forexampleLvi-Strausssattempt,in
collaborationwithAndreWeil,tomodelcertainmarriagesystemsasAbeliangroups.
SeeLvi-Strauss,The Elementary Structures of Kinship,Eyre&Spottiswoode,1969,
pp.2219.OriginallyinFrenchasLesStructuresElementaire de la Parente,
Mouton,1949.
1
2
3
4
5
6
7
8
9
Nondiscursive technique
09
Theoretical preliminaries

Spaceisthemachine|BillHillier

SpaceSyntax
Forexample,hisingeniousattempttomodeltheelementarypropertiesofmatter
throughthefveregularsolidsintheTimaeus.SeePlato,Timaeus 33etseq.p.1165
inThe Collected Dialogues (seenote5above)
ThisprocessisthesubjectofChapter9.
Asdescribed,forexample,inChapter2ofThe Social Logic of Space.
Foralucidsummary,seeP.Steadman,Architectural Morphology,Pion,1983.
L.March,Inconversation.
I.StewartandM.Golubitsky,Fearful Symmetry,Penguin,1993,p229.
F.BuckleyandF.Harary,Distance in Graphs,AddisonWesley,1990,p.42.
B.HillierandJ.Hanson,The Social Logic of Space,
CambridgeUniversityPress,1984,p108.Seealsonote16inChapter1.
Steadman,p.217.
Hillier&Hanson,pp.10913.
However,seethereferencesinnote16ofChapter1.
Forexample,I.Biederman,Higherlevelvision,ineds.D.Oshersonetal.,
Visual Cognition and Action,MITPress,1990.
Foradiscussionofsomeofthesevariationsfromthepointofview
ofgraphtheoryseeBuckleyandHarary,Distance in Graphs,pp.17985.
Forexample,P.Tabor,Fearfulsymmetry,Architectural Review,May1982.
AbbeMarc-AntoineLaugier,Essai sur larchitecture,Paris1755.
SeeHillier&Hanson,The Logic of Space,p90.
Itshouldbenotedattheoutsetthattheseoverlappingconvexelementsareunlike
theconvexelementsdescribedinThe Social Logic of Space,whichwerenot
allowedtooverlap.SeeHillier&Hanson,pp.978.
Itisexactlythispropertythatlabyrinthsexploit.Ateverypointthespaceyou
seegivesnoinformationormisleadinginformationaboutthestructureofthe
labyrinthasawhole.Ingeneralthoughnotinvariablyagoodurbanformdoes
exactlytheopposite.
SeeChapter4.AlsoB.Hillieretal.,Naturalmovement:orconfguration
andattractioninurbanpedestrianmovement,Environment & Planning B, Planning &
Design,vol.20,1993.
As,forexample,inthecaseofthenewShanghaiCentralBusinessDistrictonwhich
wecollaboratedwithSirRichardRogersandPartners,ortheoriginalplanforthe
KingsCrossRailwaysLands,LondonwithSirNormanFosterandPartners.Seefor
exampleB.Hillier,SpecifcallyArchitecturalTheory,HarvardArchitecturalReview,
vol.9,1993.AlsopublishedasB.Hillier,Specifcallyarchitecturalknowledge,
Nordic Journal of Architectural Research,2,1993.
TheproblemsgeneratedbythistypeoflayoutareexaminedindetailinChapter5.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30