pattern language nikos

Architectural Research Quarterlyhttp://journals.cambridge.org/ARQ

Additional services for Architectural Research Quarterly:

Email alerts: Click hereSubscriptions: Click hereCommercial reprints: Click hereTerms of use : Click here

The structure of pattern languages

Nikos A. Salingaros

Architectural Research Quarterly / Volume 4 / Issue 02 / June 2000, pp 149 - 162DOI: 10.1017/S1359135500002591, Published online: 19 August 2008

Link to this article: http://journals.cambridge.org/abstract_S1359135500002591

How to cite this article:Nikos A. Salingaros (2000). The structure of pattern languages. Architectural Research Quarterly, 4, pp 149-162doi:10.1017/S1359135500002591

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/ARQ, IP address: 128.243.112.64 on 26 Aug 2015

http://journals.cambridge.org Downloaded: 26 Aug 2015 IP address: 128.243.112.64

^ u fii! \ \ //

NikosA. Salingaros

The structure ofpattern languages

Author's addressDivision of MathematicsUniversity of Texas at San AntonioSan AntonioTX 78249USA

Pattern languages help us tackle the complexity of a variety of systemsrangingfrom computer software, to buildings and cities. Each'pattern'represents a rule governing one working piece of a complex system,and the application of pattern languages can be done systematically.Design that wishes to connect to human beings needs the informationcontained in a pattern language. This paper describes how to validateexisting pattern languages, how to develop them, and how theyevolve. The connective geometry of urban interfaces is derived fromthe architectural patterns of Christopher Alexander.

We observe the world around us and learn itsstructure by abstracting cause and effect, and bydocumenting recurring solutions obtained underdifferent conditions. Such empirical rules,representing regularities of behaviour, are called'patterns'. Visual patterns are the simplestexpression of the pattern concept (Salingaros, 1999).Many patterns are hard-wired into our mind: weinherit actions and reactions that guarantee oursurvival. Other patterns have to be learned, and forman artificial extension of the human mind. Theability to observe patterns gives us the humanadvantage of both adapting to, and changing ourenvironment. Of course, the complexity enveloping apattern in each specific setting has to be partiallycleared so as to get at its basic mechanism.

The language of a group of patterns forms thegroundwork for any discipline. Learned patternlanguages - not intrinsic to the human mind - werecarefully preserved in the past. Many patterns ofhuman relations are codified into religions, myths,and literary epics. A collective intelligence developsfrom pooling discoveries accumulated overgenerations. This process is entirely general. Thesciences rely on mathematics for the ability toorganize data and explain phenomena by means ofregularities, or logical patterns (Steen, 1988).Breakthroughs occur when patterns in one area linkto patterns in other areas.

This paper discusses the language that linkspatterns. A pattern language contains usefulconnective information that helps both to validatethe patterns, and to apply them. We are going to cast

the structure of a pattern language in terms of theproperties of pattern combinations. Such anapproach reveals the ordering of patterns in space,time, and human dimensions. I will assume readershave a minimal familiarity with the architecturalpatterns of Christopher Alexander as published in APattern Language (Alexander, Ishikawa et al., 1977).1Although introduced into architecture more than 20years ago, their true significance has beenappreciated by only a few practitioners. Patterns area powerful tool for controlling complex processes,but because of misunderstandings, they have notplayed a wide role in architectural design. Instead,patterns have found unexpected success in computerscience.

The audience for this paper is anyone interested inconnecting their designs to human beings. We willshow that this cannot be done withoutincorporating patterns. After describing in generalterms what patterns are, and the ways they cancombine, I will discuss the relationship betweenpatterns and science. Graph theory visuallyillustrates some key aspects of pattern languages:how patterns combine to form higher-level patternscontaining new information; how linked patternsexist on different levels; how to find patterns in anew language; and how a pattern language isvalidated through its connective structureindependently of each individual pattern's validity. Amajor concern is how a pattern language is damagedthrough the imposition of arbitrary stylistic rulesand anti-patterns, which are often mistaken forpatterns. All too often, people have tried to change a

theory arq V0I4 no 2 2000 149


150 arq V0I4 no 2 2000 theory

1 Small parking lots(#103). One of the 253design 'patterns' thatrecur in architecturea Each of theAlexandrine patternsopens with aphotographsummarizing thepattern subject. Noneis identified: this oneillustrates the Fellows'(faculty) car park at St.Catherine's College.Cambridgeb The illustrationencapsulating the textsection entitled'Vastparking lots wreck theland for the people'c The illustrationconcluding thediscussion sectiond The prescription andthe pattern

The illustrations inthese Figures and Figs.2-14 and 21 are from APATTERN LANGUAGE byChristopher Alexander,copyright-1977 byChristopher Alexander.Used by permission ofOxford UniversityPress, Inc. (Availablefrom Oxford UniversityPress, 1171 pp.Hardcover, GBP40 or$40. ISBN 019 5019199www.oup.com)

Make parking lots small, serving no more than five toseven cars, each lot surrounded by garden walls, hedges,fences, slopes, and trees, so that from outside the cars arealmost invisible. Space these small lots so that they are atleast 100 feet apart.

five to seven cars

id

society by changing its architectural patternlanguage. An application to the geometry of urbaninterfaces is given from the patterns approach.

What is a pattern?In A Pattern Language, Alexander and his colleaguespublished 253 solutions or design 'patterns' thatrecur in architecture, such as the need for SMALLPARKING LOTS (#103) [Fig. 1a], or SIX-FOOT BALCONY -the minimum depth that makes it useful - (#167)[Fig. 2] (Alexander, Ishikawa et al., 1977). They arguedthat built designs violating the derived patterns werenoticeably less successful than those that followedthem. The Alexandrine format fixing a patternconsists of a statement summarizing the philosophyabout a specific topic (i.e., for SMALL PARKING LOTS):'Vast parking lots wreck the land for people' [Fig. ib[.

They follow the pattern statement by anexplanation that supports the pattern: statisticaldata; a scientific analysis; discovering thesimultaneous occurrence of this pattern in totallydifferent cultures; psychological, structural, orcultural reasons; and so on. For example, thediscussion following the above pattern includes:'... the fabric of society is threatened by the mereexistence of cars, if areas for parked cars take upmore than 9 or 10% of the land in a community....tiny parking lots are far better for the environmentthan the large ones, even when their total areas arethe same ... Large parking lots, suited for the cars,have all the wrong properties for people' [Fig. ic].

A pattern ends with some sort of prescription inpractical terms, to help incorporate the pattern intoan actual design. For example: 'Make parking lotssmall, serving no more than five to seven cars, eachlot surrounded by garden walls, hedges, fences,slopes, and trees, so that from outside the cars arealmost invisible ... [Fig. id]'.

Many criticisms of Alexander's Pattern Languageare valid to some degree - that it reflects thephilosophy of the 1960s, that it is too radical and noteasily incorporated into contemporary design andplanning, that it ignores almost all of what isconsidered important architecture in the twentiethcentury - but these are trivial compared with theimportant message it offers. This paper will attemptto show that any design that ignores patterns cannever hope to connect to human beings.

Combining Alexandrine patternsYou can combine design patterns in an infinitenumber of ways. However, the connective rules - i.e.,the language - were only briefly sketched out. Toobtain an understanding of the relationshipbetween patterns, you have to go back to Alexander'searlier work (Alexander, 1964; Alexander, 1965).Other than Chapter 16 of The Timeless Way of Building(Alexander, 1979), Alexander himself has not dwelton the synthesis between patterns. Any perceivedweakness of patterns could he in individual patterns,but it is more likely the result of not understandingtheir combinatorial language. Although designpatterns written in Alexandrine form allude to theirconnectivity to other patterns (in the prelude and

Nicos A. Salingaros The structure of pattern languages



2 Six Foot Balcony(#167). Another ofthe patterns - seealso 21a Encapsulatingillustrationb The pattern. Theaccompanyingprescriptive noterecommends that, ifpossible, part of thebalcony should berecessed into thebuilding so that it isnot cantilevered outand separated fromthe building by asimple line and thatit should be partiallyenclosed

six feet deep

2b

postscript), it is difficult to visualize those without aconnective map. Even architects who use patternstend to be unaware of how patterns link to eachother, so the resulting design frequently lacks large-scale coherence.

In an entirely unanticipated development, thePattern Language format has found a basicapplication in computer programming. Anyprogramming solution that reappears in separateinstances maybe identified as a 'pattern', and besubsequently re-used as a unit. Patterns are nowrecognized as a powerful theoretical framework inwhich to assemble complex computer programs(Coplien and Schmidt, 1995; Gabriel, 1996; Gamma,Helm et al., 1995). The proponents of softwarepatterns believe that patterns can help to solve awide range of practical problems that wouldotherwise be too cumbersome or time-consuming.

To give readers a better sense of what is meant bypatterns connecting to each other, we list someexamples of coupling. One pattern contains or generalizes another

smaller-scale pattern. Two patterns are complementary and one needs

the other for completeness. Two patterns solve different problems that overlap

and coexist on the same level. Two patterns solve the same problem in

alternative, equally valid ways. Distinct patterns share a similar structure, thus

implying a higher-level connection.With connective rules, two different aspects of apattern come into play. On one hand, a pattern'sinternal components will determine its inclusioninto a larger pattern. On the other hand, it is theinterface that determines overlap, or connection on

the same level. Two patterns on the same level mayeither compete, loosely coexist, or necessarilycomplement each other.

One criticism of Alexandrine patterns arises fromtheir clash with existing economic practice andconstruction process. The Pattern Language extendsfrom the scale of surface detail, to the scale of a largecity, and covers Alexander's ideas on how to bestimplement a more human built environment(Alexander, Ishikawa et al., 1977). Some of the urbanpatterns flatly contradict land speculation and theerection of mega towers, while the building patternsmake obvious the need for more structural qualitythan today's contractors are used to providing. Bothof these points threaten a profit source in theconstruction industry. While it is not yet clear how toreconcile those differences, Alexander's critics find inthis an excuse to dismiss all of the Pattern Languageas impractical and unrealistic (Dovey, 1990). That isvery short-sighted.

A more serious concern comes from practitionerswho attempt to apply Alexandrine patterns to shapethe built environment. The Pattern Language is not,and was never claimed to be a design method and itis always a struggle to integrate patterns into anactual design project. Architects, however,desperately need a self-contained design method,and, not finding it in Alexander's theories, will adoptwhatever design method is currently in fashion. Thetools that Alexander is proposing are therebybypassed, appearing useful only in retrospectiveanalysis, which also explains the Pattern Language'srelative lack of impact. Design is tremendously hardwork, and I would like to help show how to utilizepatterns in practice.

A set of connected patterns provides a framework

The structure of pattern languages Nikos A. Salingaros



upon which any design can be anchored. Thepatterns do not determine the design. By imposingconstraints, they eliminate a large number ofpossibilities while still allowing an infinite numberof possible designs. The narrowing of possibilities is,after all, an essential part of a practical designmethod. In this case, the remaining choices areprecisely those that connect to human beings eithervisually, emotionally, functionally, or by facilitatingtheir interactions and activities. People havefundamental physical and emotional needs thatshould be satisfied by the built environment, thoughmost of them are neglected nowadays. Architecturaldesign that accommodates - or, better still, enhances- a framework of Alexandrine patterns will be felt asmore 'natural' than one which doesn't.

The connective geometry of urban interfacesIn a living city, boundaries define and connectdifferent regions, and encourage many h u m a nprocesses that make the city successful. Whetherthese functions take place is largely a consequence ofthe geometry of the u rban boundaries: it has to bebo th crinkly and permeable. (In mathematicalterms, it is accurate to call such a line a 'fractal', sinceit is nei ther continuous, nor perfectly smooth.) Theneeded information for this already exists in severalAlexandrine patterns, which combine to give adefinite u rban geometry very different from thatfound in contemporary cities.

In practice, it is very cumbersome to work from acomplete catalogue of discovered patterns to create aproduct. A simplified connective list can drasticallyimprove the utility of any pat tern language. Aprocedure for generating such a map is based on theconceptual 'chunking' of information (Miller, 1956).The goal is to cluster patterns into groups of aboutfive or fewer on each level of scale. Suppose oneneeds to design something using available patterns;pick those tha t are most relevant to the problem athand, then choose not more than about a dozenrelated pat terns from an existing patterns catalogue.Identify a vertical dimension (e.g. time, space, orgroup size) appropriate to the process that generatesthe end product, and study how the generativeprocess develops as one moves u p the levels of scale.

Once you assemble a group of patterns from apatterns catalogue, you can go back and developothers for related processes, which will includepatterns left out in the initial round. Pattern groupsfor different results should be separate, and notconfuse each other's clarity. In the case of urbaninterfaces, several patterns are directly relevant. Ihave listed t h e m here, numbered as in A PatternLanguage (Alexander, Ishikawa et al., 1977).

13. SUBCULTURE BOUNDARY15. NEIGHBOURHOOD BOUNDARY42. INDUSTRIAL RIBBON53. MAIN GATEWAYS108. CONNECTED BUILDINGS119. ARCADES121. PATH SHAPE122. BUILDING FRONTS

124. ACTIVITY POCKETS160. BUILDING EDGE165. OPENING TO THE STREET166. GALLERY SURROUND

These dozen patterns serve as an empiricalfoundation for a geometry of urban interfaces.

Reversing the order of the patternsAlexander numbered the patterns according todecreasing size, yet I will reverse the order in theabove list for our discussion. GALLERY SURROUND[Fig. 3] proposes that people should be able to walkthrough a connecting zone such as a balcony to feelconnected to the outside world. OPENING TO THESTREET [Fig. 4] is the corollary: people on a pavementshould feel connected to functions inside a building,made possible by direct openings. BUILDING EDGE[Fig. 5] should be such as to encourage life, creatingpedestrian nodes and the necessarily crinkly,crenellated geometry that they require. ACTIVTrYPOCKETS [Fig. 6] reveal that any public space issuccessful only if its edge contains andaccommodates successful pedestrian nodes.

galleriesadjacent indoor rooms

balconies

open, without glass

i

) M '- straddling the pathpedestrianpftttl

crenelation

depth along the edge

shelter

Nicos A.Salingaros The structure of pattern languages



BUILDING FRONTS [Fig. 7] define the life at the builtedge of a street, while uniform set-backs 'almostalways destroy the value of the open areas betweenthe buildings'. PATH SHAPE [Fig. 8] requirespedestrian nodes along a path, and these will deformany straight edges into a more fractal form. ARCADES[Fig. 9] connect the inside of buildings with the worldoutside via an intermediate partially-enclosed space;without them, the transition is too abrupt.

CONNECTED BUILDINGS [Fig. 10] create both a

boundary and a path along it, which is destroyed byhaving intermediate space between the buildings.MAIN GATEWAYS [Fig. 11] give significance - bydefining access - to what would otherwise be auseless space between buildings. INDUSTRIALRIBBON [Fig. 12] functions as one possible way tocreate a wide boundary for separating regionscontaining other types of buildings. Finally, the twopatterns NEIGHBOURHOOD BOUNDARY [Fig. 13] andSUBCULTURE BOUNDARY [Fig. 14] stress the necessity

paths

.pocketsof activity connections

gateway

no setbacks

slight angles

boundary

ribbons 200 to500 wide

bulge in the middle

road

narrow ends

continuous arcades

1 i 111 i 11 nTflYf meeting places

3-14 a dozen patternswhich serve as anempirical foundationfora geometry ofurban interfaces

3 Gallery Surround(#166) forms aconnecting zoneproviding a feeling ofconnection to theoutside world

4 Opening to the Street(#165) provides aconnection betweenthe pavements andthe functions insidebuildings

5 Building Edge (#160)should encourage life,creating space forpedestrian nodes

6 Activity Pockets(#124). Public spaceonty works if its edgescontain successfulpedestrian nodes

7 Building Fronts (#122)define the life at thebuilt edge of thestreet

8 Path Shape (#121)requires pedestriannodes along a path,deforming straightedges

9 Arcades (#119)provide a transitionbetween buildinginteriors and theworld outside

10 Connected Buildings(#108) create both a

boundary and a pathalong it

11 Main Gateways (#53)give significance tootherwise uselessspace

12 Industrial Ribbon{#42) forms a way ofcreating a wideboundary forseparating groups ofother types ofbuildings

13 NeighbourhoodBoundary (#15)stresses thenecessity forcontainment in aliving city

14 Subculture Boundary(#13) is similar

gateways

restricted access

200 "feet of land

man-madeboundaries

meeting places

' natural boundaries




of containment in a living city, and show how onezone can destroy an adjoining zone if theappropriate boundaries are absent. Together, theabove patterns combine to create the picture of aliving city that depends in large part on itsconvoluted, permeable interfaces. The informationgathered by Alexander and his colleagues in puttingtogether the Pattern Language offers a conception ofthe urban fabric as a highly connected structure,whose subdivisions are defined by complexboundaries.

Some critics may wish to dismiss the first group ofpatterns as relevant only to a pedestrian city, whichin their estimation, no longer exists. Quite theopposite is true. The discussion of this paper makesit clear that, since human beings are anatomicallygeared for walking as their principal mode oftransport, these patterns are timeless and relevanteven if their domain is restricted in today's car-dominated urban landscape. They still applywherever we walk, whether it be in parking lots,along storefronts, suburban pavements, or indoorshopping malls. Decades of suppression by patternsfor the automobile network has erased mostpedestrian patterns (Newman and Kenworthy, 1999).Whenever there is an architectural opportunity,however, these patterns re-emerge spontaneously tocreate a living interface.

Validation of the patternsAlexander presents the Pattern Language as apractical tool, and orders the patterns in roughlydecreasing size. That is the correct ordering whenone is using them for design, since decisions on thelargest scale have to be made first. Nevertheless, thatpresupposes that the patterns are understood to betrue in a fundamental sense. The problem is thatmainstream architecture never entirely acceptedAlexandrine patterns; it was the more sensitive andspiritual fringe movements that did. To validate theabove patterns, they have to be read in the oppositeorder: small to large. The human mind can combinethe smaller patterns into groups; the larger patternsutilize these groupings and also generate newproperties that are not present in the componentpatterns. The mind is capable of validating thepatterns subconsciously when we read the patternsin an evolving (small-to-large) order.

Even now, more than 20 years after its publication,the fundamental significance of the PatternLanguage is hardly appreciated. Many people stillthink of it as a catalogue of personal preferences,which is a total misconception (Dovey, 1990). Eventhose who realize that each pattern is establishedeither though empirical observation, or by scientificreasoning, often fail to see its inevitability. Irecommend, though, that you photocopy therelevant patterns from A Pattern Language (Alexander,Ishikawa et al., 1977), and staple them together in thereversed order. Reading them without thedistractions of all other patterns helps to connectthem in the reader's mind, and the naturalprogression small to large reveals the connectionsbetween successively larger scales. Doing this leads to

the conclusion that the type of urban boundarydescribed is not simply our suggestion, but isnecessary for a living city.

Quite separate from the internal validation offeredby their ability to combine, what demonstrates thepatterns' inevitability is their connection tofundamental patterns of human behaviour andmovement. Many human functions and interactionsare facilitated by the proposed urban geometry, andwe could graphically link behavioural patterns tothese architectural patterns directly. In mostinstances, this connection is revealed as an intuitionthat the patterns for urban boundaries 'feel right'.Alexander based much of the validation for thePattern Language on this intuitive assessment(Chapter 15 ofThe Timeless Way of Building [Alexander,1979]). which was dismissed as unscientific. But agraphic and theoretical basis underlies this.

The smaller the scale on which a pattern acts, themore immediately it connects to human beings.Architectural patterns on the human range of scaleslcm-im create a visceral response because we canexperience them with most of our senses. Largerpatterns that cannot be touched or felt requiresynthesis and recognition; they become moreintellectual. People who have not experienced themin person (in some region of the world where theystill exist) can rarely imagine their emotionalimpact. This is the reason why the sequence small-to-large works in a validation process: it brings in thestrongest personal connection at the beginning, andsuccessive patterns build upon an intuitivelyaccepted base.

Patterns and scienceIn the remainder of this paper, I will discuss patternsin very general terms, with the intention ofdemonstrating their inevitability. A pattern is adiscovered solution that has been tested for sometime, and under varying conditions. Forarchitectural and urban patterns, the time-framecan be several millennia. A pattern is not usuallyinvented, so creativity is subordinated here toscientific inquiry and observation. Although you canfind novel ways to combine and relate patterns,creativity is reserved for the products arising from anapplication of the pattern language, not the process.Since patterns are derived empirically fromobservations, they differ from scientific theory,which derives solutions starting from firstprinciples. Nevertheless, discovered patterns providea phenomenological foundation out of whichscientific theories can grow. Once established, thosetheories explain why some patterns work.

Sometimes, a pattern may arise as an informedconjecture. It has to survive the intense criticism andscrutiny that are part of the scientific method ofvalidation. Although patterns are pre-scientific, theyare in fact much broader than science. A pattern maybe the intersection of separate scientificmechanisms. Many patterns do not yet have ascientific explanation; for others that do, theexplanations may be bulky and convolutedcompared with the simplicity of the pattern itself.



theory arq V0I4 no2 2000 155

Medicine, pharmacology, and psychology are basedat least partially on pattern languages, while theirphenomenological foundation is slowly beingreplaced by a biological/chemical basis.Morphological and scaling rules that apply broadlyacross many different disciplines (West and Deering,1995) are patterns that are useful independently ofthe particular mechanisms that generate theobserved phenomena.

Unfortunately, architecture as a disciplinecurrently has no means of validating anarchitectural pattern, so the basic mechanism forpattern formation doesn't exist. Architects who arenot also trained in the scientific method will notdistinguish between a design method or procedurethat gives successful results and one that fails; thevalidation process that should follow any proposedsolution does not form part of architecturaleducation (Stringer, 1975). The reasons why somebuildings fail - in the sense of being unpleasant anddifficult to use - are never seriously examined.Consequently, design mistakes tend to be repeatedindefinitely.

A philosophical reversal presents an even moreserious impediment to the use of architecturalpatterns. Architecture has changed in this centuryfrom being a trade serving humanity withcomfortable and useful structures, to an art thatserves primarily as a vehicle for self-expression forthe architect. In the current architectural paradigm,the emotional and physical comfort of the user areof only minor importance. Architects resist using thePattern Language because they erroneously believe ithinders artistic freedom. Declaring that they wish toexpress their creativity freely, they nevertheless forcethemselves to work within irrelevant stylisticconstraints. Contemporary architecture has becomeself-referential, validated only by how well itconforms to some currently accepted style, and notby any objective external or scientific criteria(Stringer, 1975).

The nature of a pattern languageIn practice, pattern languages arise from two verydifferent needs: (a) as a way of understanding, andpossibly controlling, a complex system; (b) asnecessary design tools with which to buildsomething that is functionally and structurallycoherent. To visualize patterns and theirinterconnections, we use a graph representation.Patterns may be identified with nodes in a graph,and the graph is connected by edges of differentlengths [Fig. 15]. A pattern is an encapsulation offorces; a general solution to a problem. The'language' combines the nodes into anorganizational framework. A loose collection ofpatterns is not a system, because it lacks connections.

The rules by which the patterns (nodes) connectare just as important as the patterns themselves.Words without connection rules cannot make up alanguage. A coherent combination of patterns willform a new, higher-level pattern that possessesadditional properties [Fig. 16]. Not only does eachoriginal pattern work in combination as well as it

did individually, but the whole containsorganizational information that is not present inany of its constituent patterns. A higher-level patterncannot be predicted from the lower-level patternsalone. Sticking patterns together without properordering will not provide an overall coherence. Eachcomponent might work individually, but the wholedoes not work, precisely because it is not a whole.

A pattern language is more than just a patternscatalogue. Individual patterns are easier to describethan their language, yet a catalogue is only adictionary. It does not give a script; it has no rules forflow, internal connections, or ordered substructures.A patterns catalogue lacks the essential validationthat comes from recognizing the combinatorialproperties in the language. Some patterns willrequire other complementary patterns forcompleteness, and the allowed combinations are

15 Individual patternsgroup to form sixhigher-level patternshaving additionalproperties

16 Further connectionsorganize thepatterns in Figure 1into a pattern onthe next higher

level. Newproperties ofthe wholecorrespond tonew symmetries

The structure of pattern languages Nikos A.Salingaros



usually infinite. A language tells you which of themcan be combined, and in what manner, in order tocreate a higher-level pattern. Drawing an analogywith biological systems, the system works because ofthe connections between subsystems (Passioura,1979)-

Hierarchical connections across scalesEvery complex system has a hierarchical structure;i.e., different processes are occurring on differentscales or levels. Connections exist both on the samelevels, and across level (Mesarovic, Macko et al., 1970).The same is true for a pattern language. The'language' generates a connective network by whichthe ordering of nodes on one level creates nodes at ahigher level. This process goes on all the way up, andall the way down in levels [Fig. 17]. The cohesiveframework provided by the language enables theupward transition to all the higher levels. We canbetter understand a language if it has organizationat different levels, because each level is shielded fromthe complexity in all the other levels.

A pattern language does not have a strictlymodular rule structure - as would be the case if thelanguage were defined by only a few basic units - but

adds new rules as the scales grow. Higher levels in asystem are dependent on all lower levels, but notvice-versa (Passioura, 1979). Even thoughdisconnected lower-level patterns can work withoutnecessarily forming a higher-level pattern, such asystem is not cohesive, because it exists on only onelevel. Each level in a complex hierarchical system issupported by the properties of the next-lower level.The combination of patterns acting on a smallerlevel of scale acquires new and unexpectedproperties not present in the constituent patterns,and these are expressed in a higher-level pattern [Fig.18]. Patterns on higher levels are therefore necessarybecause they incorporate new information.

Many failures in describing a complex system aredue to not allowing for enough levels. A gap betweenlevels disconnects the pattern language, since thepatterns on different levels are then too far apart tobe related [Fig. 19]. We tend to fall into this trapbecause of non-hierarchical thinking. Some urbanpatterns work on the scale of 100m and containarchitectural patterns that work on the scale of 1 m,but what about the patterns on all the intermediatescales? An even more serious problem is thewidespread association of importance with size in

17 Hierarchicalconnections showhow patterns onhigher levels dependon those on lowerlevels

18 Patterns on one levelcombine to helpdefine a new patternon a higher level

19 Two groups ofpatterns are too farapart in scale toconnect effectively




our culture. Working within that mindset, it is veryeasy to concentrate only on the large-scale patterns(or anti-patterns), and ignore those on lower levels.That makes it impossible to validate patternsthrough their vertical connections, which areillustrated in Figures 17 and 18.

One of the principal methods of validating apattern language is that every pattern be connectedvertically to patterns on both higher and lowerlevels. Damage to a pattern language can beunderstood visually, by crossing out any singlepattern in Figure 17. This will remove thecoordination of all the linked patterns below it;moreover, if a vertical relation is one of inclusion,then obviously those patterns below are alsoeliminated. In addition, all linked patterns above thecrossed pattern are automatically eliminated.Therefore, removing one pattern withoutunderstanding its connections damages a significantportion of the pattern language because it alsoremoves at least one vertical chain of patterns.

It is necessary to address a misunderstanding thatidentifies any multi-level structure with an invertedtree-like hierarchical ordering. In a tree, everythingis ordered from a single node above, and nodes onthe same level do not link directly. Although someauthors use this terminology, that is not what ismeant here. Figure 17 shows that the hierarchy wepropose for pattern languages is not an inverted tree,because it has multiple tops and horizontalconnections; i.e., several times more connectionsthan a tree has. A hierarchical inverted tree structureis too restrictive, since all communication has to passthrough higher-level nodes. Inverted tree-likehierarchies are associated with systems that exerttop-down control (Alexander, 1965).

Finding patterns for new disciplinesA new discipline needs to abstract its patterns as theyappear. It is building its own foundation and logicalskeleton, upon which future growth can besupported. Knowing its basic patterns early on willspeed up the language's development, and guide it inthe right direction. You may obtain insight into anew field lacking a pattern language by studyingpatterns from established disciplines. A universalhigh-level structure is inherent in all patternlanguages. The solution space, which is distinct fromthe parameter space, is rarely one-dimensional,which means that knowing what doesn't workcannot give what works simply by doing theopposite. There may be an infinity of differentopposites. One needs to exhaust the solution spaceby identifying many neighbouring anti-patternsbefore zeroing in on the pattern itself.

Here we need to warn against the destructivetendency in our times of judging patternsprematurely using strict criteria such as efficiency,cost reduction, and streamlining. It is not that theseare inappropriate criteria, but rather that they tendto ignore the linkage between patterns. In otherwords, patterns in a pattern language depend oneach other in a complex manner, and a hasty cullingof what are erroneously deemed 'superfluous'

patterns may damage the cohesion of the language.Many fundamental patterns have been discarded inthe false interest of economy, without realizing thatthey are essential to a system's coherence and overallperformance. The long-term consequences of thisare negative, and significant. You may attempt tostreamline a process after its complexity is wellunderstood, but not before. Promising new patterns,and time-honoured old ones, have been ruthlesslyscrapped by short-sighted thinking, borne out of thebelief that complex systems have to conform to somesort of 'minimalist design'. This comes from asuperficial understanding of how a system works.

The most elegant complex systems are nearly (butnot perfectly) ordered. Having to accommodatepatterns on the smaller and intermediate scales -indeed, actually growing out of them - the larger-scale patterns cannot be perfect in the sense of beingpure or too simple. Good design avoids unnecessarycomplication. It is balanced between arising out ofloosely organized small-scale patterns, which couldlead to somewhat random forms or processes, andpatterns which might pay too much attention to thelarge scale. Going too far in either extreme damagesthe coherence (and therefore the efficiency) of thesystem.

The general ideas offered here prove useful inextending urban patterns to the electronic city. Thenotion of an 'intelligent environment' defines theurban connectivity of the new millennium. On top ofthe existing path structure governed by Alexandrinepatterns (Salingaros, 1998), we need to develop rulesfor electronic connectivity (Droege, 1997; Grahamand Marvin, 1996). To define a coherent, workingurban fabric, the pattern language of electronicconnections (which is only now being developed)must tie in seamlessly to the language for physicalconnections. Already, some authors misleadinglydeclare that the city is made redundant by electronicconnectivity. Such opinions ignore new observedpatterns, which correlate electronic nodes tophysical nodes in the pedestrian urban fabric. Thetwo pattern languages will most likely complementand reinforce each other.

Consistency and connectivityOf the two criteria: (a) internal consistency, and (b)external connectivity, the second is by far the moreimportant. A system's complexity - the extent ofwhich may not be known for some time, if ever - canprevent a new pattern language from having asmooth internal structure. It is essential, however,that any pattern language link to existing languagesat its boundaries [Fig. 20]. For example, a buildingthat is internally inconsistent would be unusable.Once a building has achieved a minimum degree ofinternal consistency, however, external connectivitywith other patterns becomes more important. Thepoint is to avoid the isolation of pathologicalsystems, which then survive because they are notsubject to interactive checks and balances.

It is possible to define a set of anti-patterns that'clean up' complexity by imposing rigid, one-dimensional ideas. Such a language could itself be



158 arq vol 4 no 2 2000 theory

perfectly consistent internally, hut it cannot coexistwith other pattern languages that respectcomplexity. The best example comes fromgovernment. Fascism and totalitarianism clean upthe messiness of human society, but clash with ourmost deeply-held patterns of human values. In thesame way, any organizational pattern language thatattempts to create a positive work environment willnecessarily connect with and provide a transition toAlexander's architectural pattern language, whichdetermines built form on all levels of scale(Alexander, Ishikawa et al., 1977).

The architectural pattern SIX-FOOT BALCONY[Fig. 2] helps to illustrate connectivity (Alexander,Ishikawa etal., 1977). Many social patterns of familylife, such as sitting around a table; eating a meal;children playing with toys on the floor; growingplants in large pots; outdoor cooking on a charcoalgrill; and so on, can occur on a balcony only if it is atleast 6ft (2m) deep [Fig. 21]. When a balcony is madetoo narrow so as to follow some arbitrary designcanon or simply to be cheap (which satisfiesinternally consistent criteria), it fails to connect tothe above social patterns. Connection here meansaccommodation and inclusion among patternsbelonging to two different languages. Mathematicalisolation, as in Figure 20, guarantees the physicalisolation of the balcony from potential users.

We don't appreciate how completely architecturalpatterns connect to social patterns; the former makeup a significant part of the traditional culture in anysociety. Losing them irreparably damages the way asociety functions, because architectural patternshelp to define all the higher-level social patterns[Fig. 22]. Especially among the rural poor, tradition isthe only way of safeguarding their culture. Traditionembodies solutions evolved over countlessgenerations, so design patterns are connected withand have become part of a way of life. This point hasbeen stressed by Alexander (Alexander, 1979), and isvery eloquently argued by Hassan Fathy (Fathy, 1973)(pp. 24-27). Sensitive architects pay attention so thattheir designs accommodate and nurture socialpatterns.

Sometimes, a pattern might have an unwantedsecondary characteristic; the same way an inheritedtrait in an organism may be essential for survival, buthave a mildly negative side-effect. The same pattern isexpressed as two different features. Attempting toremove the secondary, unwanted feature (forexample, getting rid of every architectural elementor social pattern that 'spoils' an overall perfectsymmetry) without realizing what it connects to candestroy the entire language. By condemningsecondary features of human patterns because theyare not consistent with arbitrary ideas of style, orbecause of some anti-social aversion, architects havesucceeded in eliminating traditional patternlanguages around the world.

Stylistic rules and the replication of virusesDuring a time of crisis, or in the desire to be totallyinnovative, established disciplines sometimeswillingly replace their pattern languages by stylistic

rules. Those are entirely arbitrary, however, comingeither from fashion or dogma (someone in authoritypronounces a rule that is never questioned), or theyrefer to a very specific situation that does not applybroadly. Stylistic rules are incompatible withcomplex patterns such as the one shown in Figure22. The mechanism by which stylistic rules propagatebears essential similarities to the replication ofviruses. A stylistic rule is usually given as a template,and proponents are required to replicate it in theenvironment. Its success is measured not by how wellit serves any human activity, but rather by how manycopies are produced.

Stylistic rules frequently have no connection tohuman needs: they are just images with a superficialsymbolic content. While some are benign, many arepathological. An information code for built form -

/ V^ Six feet

Six feet deep.

NicosA. Salingaros The structure of pattern languages



for example, 'flat, smooth, continuous walls at streetlevel' - enters the mind of a designer either throughteaching, or from seeing built examples. Otherwiseintelligent people are easily seduced by simplisticideas in a design method, which is easy to applybecause it eliminates or suppresses naturalcomplexity. That individual then becomes an agentfor replicating the virus. Every time this code isreplicated, it destroys human connections in thatregion of the city; the result is obvious because thisparticular virus undoes all the patterns forconnective urban interfaces discussed previously.

By contrast, a pattern is not dictated or forced, butarises out of use, and is accepted on its benefits. Itfacilitates human life and interactions, and has tocontinually stand up to tests of its efficacy in thisrespect. An essential difference is that, because of its

I

20 The enclosed patterncandidates areinternally consistentbut fundamentallyflawed, because theyfail to connect toexternal patterns

21 Six-Foot Balconypattern shows howmany social patternsof family life canoccur within such aspace

22 Architecturalpatterns that pairwith social patterns(solid) furthercombine to create asocio-architecturalpattern on a higherlevel

underlying forces, no architectural pattern can berepresented as a simple visual image. A pattern solvesa complex problem; it is not a template to bemindlessly copied. It is far easier to reproduce avisual template than to solve a fundamental designproblem, however, because the former requires noreasoned thought; only intuitive matching. Theintellect does not need to work, and the designer canwithdraw from the responsibility of making difficultdecisions about the complex interactions betweenbuilt forms and human activities. Partly as a result ofthis shift, architectural design is now heavilyoriented towards visual templates defined by designstyle.

Many stylistic rules are anti-patterns: they areneither accidental, nor the simple preferences of anindividual. They intentionally do the opposite ofsome traditional pattern for the sake of novelty. Bymasquerading as 'new' patterns, they misuse apattern language's natural process of repair todestroy it. Patterns work via cooperation to build upcomplex wholes that coexist and compete in somedynamic balance. By contrast, stylistic rules tend tobe rigid and unaccommodating. Their replication inmany cases fixes the geometry of built form so as toexclude human patterns. Any single stylistic rule iscapable of suppressing an entire chain of linkedpatterns on many different scales [Fig. 17]. Adestructive stylistic rule, like a virus, is aninformational code that dissolves the complexity ofliving systems.

Today's architects are trained to use a limitedvocabulary of simple forms, materials, and surfaces.Their possible combinations are insufficient to evenapproach the structure of a language. This replacesan accumulated literature of patterns correspondingto words, sentences, paragraphs, chapters, and booksthat encapsulates meaning from human experienceand life. Few people realize the enormousconsequences on society of adopting a particulardesign vocabulary. Decisions concerningarchitectural style affect the surrounding culture;contrary to what is widely proclaimed, one person'svisions are not restricted to a building as a single artwork. A single visual template can eventually destroya culture just as effectively as a deadly virus.

Evolution and repair of pattern languagesValidated patterns are more or less permanent, yetthere exists a process of repair and replacement. Nowand then, we may play devil's advocate and ignore oldsolutions to see new, innovative ones in an olddiscipline. A new pattern is superior if it increasesthe connectivity with the majority of establishedpatterns compared with the old pattern it isreplacing. It could have a broader context, orsupersede several older patterns, thus tightening thelanguage. This is a process whose goal is tostrengthen an existing pattern language by repairand evolution, to preserve accumulated wisdom bykeeping it relevant to changing needs.

Much less frequently, a paradigm shift occurs tomake an entire pattern language irrelevant: e.g.,horse-drawn vehicles are replaced by automobiles.



160 arq V0I4 r>0 2 2000 theory

That does not invalidate the pattern languageshowing how to create the former; it just makes thatend product less desirable. While the technology andmaterials changed, however, many patterns weresaved almost intact in going from carriages to cars. Ingeneral, the adoption of innovation is greatlyfacilitated by minimizing the perception of change;and consequently the number of patterns that needto be replaced. It is wasteful to throw out a repositoryof patterns, some of which may have beenestablished over millennia.

The introduction of a new pattern language neednot displace an older one entirely. Coexistence ofcompeting or complementary patterns is oftendesirable and even necessary, especially if the newpatterns occupy different positions in the hierarchy(by acting on different scales). If properly connected,they will lead to a richer and more stable complexsystem. Patterns for the automobile transportationnetwork were falsely believed to be threatened bypatterns for pedestrian and mass-transit networks.On the basis of this misunderstanding, urbanplanners and car manufacturers simply suppressedthe latter (Newman and Kenworthy, 1999). Nowadays,we are beginning to understand that a balancedcoexistence of all three languages - describingpedestrian, automobile, and mass-transit movement,respectively - is a necessary prerequisite for acomprehensive transportation system (Salingaros,1998).

A few patterns might work equally well ondifferent levels, though most patterns' contextestablishes their place in a particular scale of thepattern language. Some patterns can be moved up ordown vertically within a language. Such a propertyleads to economy in a pattern language through self-similar scaling, which means that one scale looks thesame as another scale when magnified. A patternlanguage that develops coherence over time may alsodevelop a degree of self-similar scaling as a result ofthe connections across levels. As the ensemble ofpatterns evolves a cooperative structure, driven bythe alignment of patterns (or anti-patterns) ondifferent levels, it creates unexpected similarities.Thus, each level of a coherent structure expresses aproperty that is characteristic of the whole.

The importance of detailA language requires patterns on as many levels as ittakes to connect to natural processes. Every level isimportant by itself. In any complex system, detail ispart of the lower scales in a hierarchy. If these areunconnected, or missing, then the system is notcoherent, and cannot work (Mesarovic, Macko et al.,1970). Neglecting a pattern because it is on a lowerlevel handicaps the entire structure. It is not alwaysobvious what the lowest level of a system is uponwhich all the higher levels depend. Detail that is partof a scaling hierarchy will be connected to all higherlevels of complexity, and is not just 'added on'.Physical forms have structural features on differentscales as a result of internal and external forces. Fromthe microscopic to the macroscopic through allintermediate scales, different levels of scale cooperate.

In the design of buildings, there are several scales -corresponding to the human range of scales, 1cm to1m - that are difficult to justify purely on structuralgrounds. Yet, to define a connected hierarchy ofscales, those scales have to be present in the structure(Salingaros, 2000). Therefore, either the designshould allow the emergence of structure andsubdivisions on those scales, or substructure has tobe intentionally generated on those scales. This needcreates traditional 'ornament' and all the patternsthat generate it (Alexander, Ishikawa et al., 1977;Salingaros, 1999). The appropriate ornament isessential for a large form to be coherent (Salingaros,2000). An analysis of structural coherence arisingfrom a linked hierarchy of scales reveals thenecessity for ornament, though nowadays,ornament is discordant because it is unrelatedto the larger form.

Detail is a separate question. The smallestperceivable detail at arm's length goes down to0.25mm, which relates to a visual system such as atextile or a computer display. While such detail isavailable in richly-textured materials, it is usuallythe scales between texture and ornament(imm-icm) that are missing from contemporarybuildings. Our minimalist design tradition removesthe intermediate and smaller scales from builtform. After half a century of training in this idiom,we tend to forget that the best-loved architecture(Modernist included) works especially well on thesescales. People need to connect to structure on everyscale.

ConclusionPattern languages encapsulate human experience,and help us cope with complexity in ourenvironment. They apply to everything fromcomputer programs, to buildings, to organizations,to cities. A civilization's pattern languages are oftensynonymous with its technical and cultural heritage.New spheres of human endeavour develop their ownpattern language, which must link to existingpattern languages in related fields. Individualpatterns are validated empirically over time. Thelanguage itself will be on the right track if it evolves aconnective structure that incorporates scaling andhierarchy. Architecture and urban design in thetwentieth century rely on a set of stylistic rules thatfail to connect to patterns of human life. People havebeen taught by schools, critics, television, andmagazines to prefer abstract visual forms, and toignore the fact that environments generated by suchtemplates cannot accommodate their ownbehavioural patterns. An example of this wastraced to a fundamental misunderstanding abouturban geometry. It is believed that the removalof urban interfaces would help to create thecontemporary city, but it has seriously damagedit instead.

This paper argued that patterns provide anecessary foundation for any design solution toconnect with human beings. Contradicting themdisconnects the built form from people. Thisconclusion has profound consequences for



theory arq V0I4 no2 2000 161

architectural practice. It drastically shifts theposition of pattern languages in contemporaryarchitecture. From the peripheral position at thefringes they have occupied for more than twodecades, theyjump to a central point of architecturalrelevance. Pattern languages were revealed as the'taproot' of all architecture, from which design

draws its life by virtue of satisfying human needs.This is true even if one disagrees with one or more ofAlexander's patterns. Our results imply that designstyles which cut themselves off from this source oflife are condemned to remain forever sterile. Thosethat intentionally do so have to admit from now onthat this is indeed their aim.

Note1. A Pattern Language by Christopher

Alexander, Sara Ishikawa andMurray Silverstein with MaxJacobson, Ingrid Fiksdahl-King andShlomo Angel is published byOxford University Press andavailable in a 1171 page hardbackedition for 40 or $64. ISBN 019 5019199. See also the 'insight' section atthe end of this issue.

An illustrated appraisal entitled'Christopher Alexander and PatternLanguage' by Mark Gelernter waspublished in The Architects'Journal, 5January 1983, pp. 17-21.

ReferencesAlexander, C. (1964). Notes on the

Synthesis of Form, Harvard UniversityPress, Cambridge, Massachusetts.

Alexander, C. (1965). 'A City is Not aTree' in Architectural Forum, vol. 122,April, No. 1, pp. 58-61 and No. 2, pp.58-62. [Reprinted in: Design AfterModernism, John Thackara, ed.,Thames & Hudson, London, 1988,pp. 67-84].

Alexander, C. (1979). The Timeless Way ofBuilding, Oxford University Press,New York.

Alexander, C, Ishikawa, S., Silverstein,M., Jacobson, M., Fiksdahl-King, I.and Angel, S. (1977). A PatternLanguage, Oxford University Press,New York.

Coplien, J. O. and Schmidt, D., Ed.(1995). Pattern Languages of ProgramDesign, Addison-Wesley, Reading,Massachusetts.

Dovey, K. (1990). 'The Pattern Languageand its Enemies' in Design Studies,vol. 11, pp. 3-9.

Droege, P., Ed. (1997). Intelligent

Environments, Elsevier, Amsterdam.Fathy, H. (1973). Architecture for the Poor,

University of Chicago Press,Chicago.

Gabriel, R. (1996). Patterns of Software,Oxford University Press, New York.

Gamma, E., Helm, R., Johnson, R. andVlissides, J. (1995). Design Patterns,Addison-Wesley, Reading,Massachusetts.

Graham, S. and Marvin, S. (1996).Telecommunications and the City,Routledge, London.

Mesarovic, M. D., Macko, D. andTakahara, Y. (1970). Theory ofHierarchical Multilevel Systems,Academic Press, New York.

Miller, G. A. (1956). 'The MagicalNumber Seven Plus or Minus Two:Some Limits on Our Capacity forProcessing Information' in 27tePsychological Review, vol. 63,pp. 81-97.

Newman, P. and Kenworthy, J. (1999).Sustainability and Cities, Island Press,Washington D.C.

Passioura, J. B. (1979). 'Accountability,Philosophy, and Plant Physiology' inSearch (Australianjournal of Science),vol. 10, no. 10, pp. 347-350.

Salingaros, N. A. (1998). 'Theory of theUrban Web' in Journal of UrbanDesign, vol. 3, pp. 53-71. [Earlierversion published electronically byResource for Urban DesignInformation in 1997]

Salingaros, Nikos (1999). 'Architecture,Patterns, and Mathematics' in NexusNetworkjournal, vol. 1 (Casalini Libri,Florence, Italy) to appear in 1999.Electronic version available fromhttp://www.math.utsa.edu/sphere/

salingar/ArchMath.htmlSalingaros, N. A. (2000). 'Hierarchical

Cooperation in Architecture, andthe Mathematical Necessity forOrnament' in Journal of Architecturaland Planning Research, vol. 17, pp. [toappear]

Steen, L. A. (1988). 'The Science ofPatterns' in Science, vol. 240, pp.611-616.

Stringer, P. (1975). 'The Myths ofArchitectural Creativity' inArchitectural Design, vol. 45,pp. 634-635.

West, B. J. and Deering, B. (1995). TheLure of Modern Science, WorldScientific, Singapore.

Illustrationsarq gratefully acknowledges

Christopher Alexander and OxfordUniversity Press for consent toreproduce photographs, diagramsand text from A Pattern Language.

The author: 15-20 and 22.

AcknowledgementsThe author's research is supported inpart by a grant from the Alfred P.Sloan Foundation. I am grateful to G.Arbon, P. L. Briggs, J. O. Coplien, C. L.Jeffery, R.Johnson, J. Tidwell, M.Waddington, and S. Woo for helpfulcomments.

BiographyNikos A. Salingaros is professor ofmathematics at the University ofTexas at San Antonio. He is workingwith Christopher Alexander todevelop a new theory of architecturebased on scientific principles, aproject supported by the Alfred P.Sloan Foundation.



After the CityLars LerupAn architect's view of the newmetropolitan consciousness and thesuburban metropolis as the futurefrontier.216 pp., 52 illus. $24.95

The CIAMDiscourseon Urbanism,1928-1960Eric MumfordThe first complete history of a movementwhose mission was to revolutionizearchitecture and create an agenda formodern urbanism.408 pp., 107 illus. $45

The Unknown CityContesting Architectureand Social Spaceedited by lain Borden, Joe Ken,and Jane Rendell with Alicia PivaroEssays on architecture as narrative andurban space as experience and the newgeographies they create.600 pp., 150 illus. $50

TheProvisional CityLos Angeles Stories ofArchitecture and UrbanismDana CuffHow the modern city is formed throughan endless cycle of upheavals,demolition, and debate.400 pp., 188 illus. $40

The ArchitecturalDrawings ofAntonio daSangallo theYounger andHis CircleVolume II: Churches, Villas,the Pantheon, Tombs,and Ancient InscriptionsChristoph L Frommel, general editorNicholas Adams, editorThe second volume of a trilogy containingthe complete drawings of the famoussixteenth-century Italian workshop.512 pp., 509 illus. $95

To order call 800-356-0343(US & Canada) or 617-625-8569.Prices subject to change without notice.

Perspecta 31The Yale Architectural Journal"Reading Structures"edited by Carolyn Ann Fougand Sharon L JoyceExamines the relationship betweenarchitectural design and structuralengineering through the processes andproducts of both disciplines.160 pp., $20.00 paper

nnw in

Eccentric SpacesRobert Harbison"[MJakes me want to rush out in everydirection at once and reexamine all I haveever seen. You can hardly ask a book to domore than that." New York Times192 pp. $15 paper

The Springboardin the PondAn Intimate Historyof the Swimming PoolThomas A. P. van Leeuwen"A bracing intellectual plunge."Los Angeles Times330 pp., 244 illus., 26 color $35 paper

Architecture Theorysince 1968edited by K. Michael Hays"This collection insistently raises importantquestions and helps us elucidate problemsthat might not have otherwise occurred to us.' Architecture824 pp., 70 illus. $30 paper

Architectureand ModernityA CritiqueHilde Heynen"[A] very helpful synthetic overview of theprincipal positions in critical theory'sarguments over modernity." South Carolina Review280 pp., 103 illus. $22 paper

Le Corbusier,the Noble SavageToward an Archaeologyof ModernismAdolf Max Vogt"Scholarly and witty." Building Design381 pp., 255 illus. $25 paper

http://mitpress.mit.edu

The MIT Press2000/4

pattern language nikos

Documents