Sigma Xi, The Scientific Research Society

Architecture and Evolution Author(s): Robert Mark Reviewed work(s): Source: American Scientist, Vol. 84, No. 4 (JULY-AUGUST 1996), pp. 383-389 Published by: Sigma Xi, The Scientific Research Society Stable URL: http://www.jstor.org/stable/29775710 . Accessed: 01/05/2012 16:19

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

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

Sigma Xi, The Scientific Research Society is collaborating with JSTOR to digitize, preserve and extend access to American Scientist.

http://www.jstor.org Architecture and Evolution

In over thedebate adaptation,which view is best served by themetaphors of the "spandrels" of SanMarco and thebosses ofKing's College?

Robert Mark

their criticism of the evo? dral de San in support (Basilica Marco) Venice, primary documents and from archaeo? of has come to To lutionary precept adaptationism, pervade evolutionary dis? logical examination, have greatly en? Gould and course. For in texton biologists Stephen Jay example, his 1992 hanced our understanding of architec? Richard Lewontin in 1979 devised an evolution Williams George categorical? tural innovation. They have also helped rooted in historical architec? states "New structures arise in analogy ly evolu? iUuminate themethods used by presci ture. a In paper that soon became fa? tion in one of two ultimate ways, as re? entific builders to achieve remarkable mous, asserted that in a dundancies or Gould success in struc? they designing spandrels/' himself erecting large-scale it is absurd to decora? recounted how a building, accept colleague asserted "We tures. Although primary geometric tion to the interioras the deter have all applied been spandrelized." This mind? forms?particularly those that could be minant of the form of its structural ele? not set, however, is shared by every simply laid out using common instru? ments. Gould and Lewontin commentator on ments as a a By analogy, evolutionary theory. such straight edge and criticized about evolution The came under firedur? in arguments analogy heavy compass?were extensively used that immediate the emphasize biological ing past year from philosopher conceptual design, theywere modified and little attention to other Daniel C. as a as utility pay Dennett, part of general needed to assure structural stability. attributes of form. for ex? assessment of They decry, Gould's beliefs. Gould, he Realizing that confidence in a new the various "untestable notes in his ample, specula? book Darwins Dangerous building's ability to resist collapse could tions based on of? "needs tohave a term never secondary utility" Idea, for the (pre? have been gained merely from fered to the stunted front explain legs sumably many) biological features that applying simple geometric rules, I have of a itmakes farmore are not Tyrannosaurus; adaptations." Yet, according to sought to identify other sources of the sense, to its ab? there are they contend, accept Dennett, problems with the early builders' structural knowledge. normal form as "the reduced product analogous architectural elements chosen Chief among these is the fact that of functional homo Gould and as as conventionally by Lewontin, well with building technology itself follows an in ancestors/' In other Gould's logues words, biological speculations. evolutionary pattern. Practical experi? so much faith in I leave it to others to the evo? ence adaptationsim places pursue with earlier buildings once fur? natural as an selection optimizing agent lutionary side of the argument; the nished much of the information about thatan "is broken into function of the organism unitary historical architectural building performance that today comes 'traits' and an for each is can adaptive story elements, however, be clarified. By from engineering modeling. In effect,an In some a proposed separately." cases, considering the structuralmechanics of earlier building served as a "model" for viable based on ancient and a new explanation adaptation buildings by following the design. The technological ele? cannot be devised and should not historical be, path of development that led gance of many early structural solu? Gould and Lewontin wrote. to the of me adoption architectural ele? tions led also to discern thatduring Gould and Lewontin's principal ments in the buildings cited by Gould construction themaster builder used a based on the so-called deco? and we can metaphor, Lewontin, analyze the suit? technique that, although available to rated of Saint Mark's Cathe of the spandrels ability architectural metaphors today's designers, is rarely employed, used. they because of the usual separation of the modern design office from the construc? Robert Mark, civil and ar? professor of engineering The Evolution of Building Technology tion site (and sometimes, too, because of chitecture at Princeton University, has pioneered A of the struc? confidence inmodern the thorough understanding misplaced analy? application ofmodern engineering modeling tural mechanics of ancient buildings has sis methods). During construction the for the of historical Several study buildings. of become in recent builder made detailed observations of these studies, which have new to ar? possible only decades, brought focus with chitectural as well as to the of undesirable history the conservation of development experimental any behavior?particularly historic and numerical ca? in the fabric. tak? buildings, have been reported on previous? modeling techniques cracking building Steps inAmerican of his? en to these then ly Scientist (Winter 1968, Septem? pable reliably analyzing complex remedy shortcomings ber-October 1978 and torical architectural forms. to March-April 1987). Ad? These tech? led refinements in the design. dress: School ofArchitecture, Princeton niques, particularly when applied in the None of the results of these new ar? 08544-5264. University, Princeton, NJ context of information available from chitectural studies has informed any of

1996 July-August 383 Figure 1. of Saint Mark's Cathedral inVenice, painted by Canaletto in the early 18th century, have become the unlikely subjects of a continuing debate concerning the role of adaptation in evolutionary theory. Evolutionists Stephen Jay Gould and Richard Lewontin proposed in 1979 an architec? tural metaphor to elucidate their view that not all aspects of form have immediate biological utility. They cited the decoration of the "spandrels"?the areas as triangular between Saint Mark's arches and the domes?as well the embellished bosses in the vaults of King's College Chapel in Cambridge, as nose architectural by-products, analogous to the furnishing support to spectacles. Critics, most notably philosopher Daniel Dennett, have taken Gould and Lewontin to task for both the architectural elements chosen and their biological speculations. Meanwhile, experimental and numerical modeling techniques have recently made it possible to structurally analyze the complex elements of historical architecture. The author applies such knowledge to the function of spandrels, bosses and similar architectural elements, judging theirmetaphoric suitability to the evolutionary debate.

the arguments for or against the evolu? as part of a more general discussion of ing more on architectural matters, tionarymetaphor. Considering how re? evolution, one could have expected at opens himself up to stronger criticism cently these resultswere published, it is least a passing reference to the evolu? on this score. But before I go further into not surprising thatGould and Lewontin tionary aspect of architectural design. all of this,we should examine the archi? made no reference to them?although, Dennett, writing much later and focus tectural issues raised by both sides.

384 American Scientist, Volume 84 The Birth of aMetaphor The design is so elaborate, harmo? Cambridge, for example, the spaces we Gould and Lewontin begin theirpaper nious, and purposeful that are contain bosses alternately embell? rose with the observation that: tempted to view it as the starting ished with the Tudor and a point of any analysis, as the cause in portcullis [seeFigure 2]. In sense, this ar? an The great of St.Mark's Cathe? some sense of the surrounding design represents "adaptation," dral inVenice [see Figure 1] presents chitecture. But thiswould invert the but the architectural constraint is as a in its mosaic design a detailed analysis. The system begins with the clearly primary. The spaces arise neces? iconography expressing the main? architectural constraint: the necessary by-product of fan vaulting; use a stays of Christian faith.Three circles sary four spandrels and their taper? their appropriate is secondary a a tried to that of figures radiate out from central ing triangular form. They provide effect.Anyone who argue the structure exists because the alter? image of Christ: angels, disciples space inwhich themosaicists worked; rose so and virtues. Each circle is divided they set the quadripartite symmetry ation of and portcullis makes sense in a would into quadrants, even though the of the dome above. much Tudor chapel same that dome itself is radially symmetrical Such architectural constraints be inviting the ridicule un? on Dr. in structure. Each quadrant meets abound and we find them easy to Voltaire heaped Panglos: we cannot one of four spandrels in the arches derstand because do not impose "Things be other than they are.... below the dome. Spandrels?the ta? our biological biases upon them. Everything ismade for thebest a noses were car? pering triangular spaces formed by Every fanvaulted ceilingmust have purpose. Our made to Yet bi? the intersection of two rounded series of open spaces along themid ry spectacles...." evolutionary to ex? arches at right angles?are necessary line of the , where the sides of ologists, in their tendency focus on to architectural by-products ofmount? the fan intersectbetween the pillars. clusively immediate adaptation are to ar? ing a dome on rounded arches. Each Since the spaces must exist, they local conditions, do tend ignore a spandrel contains design admir? often used for ingenious ornamental chitectural constraints and perform an inversion ably fitted into its tapering space.... effect. In King's College Chapel in just such of explanation.

Figure 2. Ceiling ofKing's College Chapel inCambridge, England, is formedby fanvaulting, a structureof ribs radiatingfrom piers along a are at Gould and Lewontin that the the building's periphery. As result, opening created the perimeters of the conoid tops. suggested a ornate bosses within these spaces are a secondary design element. In recent book, Dennett argued that the metaphor is inappropriate, because it cannot be determined whether the bosses are secondary or primary?perhaps fan vaulting was chosen to make possible the no In these to be filled for beautiful bosses. He suggests, in any event, that they have function other than visual. fact, though, spaces need structural surety, to prevent the spread of fire, to improve acoustics and to exclude birds in the roof from entering the chapel. (Photograph courtesy of E. C. Robison.)

1996 July-August 385 Figure 3], all of them about equal in structural soundness and ease of

building.... Here [too] there is a terminologi? cal confusion that seriously impedes discussion.... Strictly speaking, the tapering, roughly spherical surface [in between the dome and the four round arches] is called a pendentive, a not spandrel. Strictly speaking, spandrels are what remains of a wall once an you punch arch through it [see Figure 4].... Speaking more loosely, spandrels are places-to-be dealt-with, and in that looser sense ... another variety of spandrel would be a squinch [see Figure 5], Now, why does all this matter? Because, when Gould and Lewontin say that spandrels are "necessary ar? chitectural by-products," what they say is false if they are using "span? drel" in the narrow sense (synony? 3. Domes mounted on arches create in the corners, between Figure roughly triangular spaces mous with "pendentive") and true the upper sides of the arches and the base of the dome. Gould and Lewontin called these only ifwe understand the term in whereas Dennett identified them as The former spaces spandrels, (correctly) pendentives. the loose, all-inclusive sense. But in saw these as for decoration left the choice of whereas the spaces opportunities by structure, that sense of the are was term, spandrels latter suggested that the structure chosen tomake possible the spaces for the display of mosaics. In the are a not that fact, pendentives (left), which but the visual surfaces in front of heavy design problems, features are either be surcharge, necessary structural elements. Brackets (right), suggested by Dennett as sub? might designed (adapta? stitutes for are as as or not. in the loose pendentives, not "equivalent" because of they lack such surcharge well tions) Spandrels sense are nec? support from any other massive structure. indeed "geometrically essary" in one regard: ifyou place a Dennett's Controversion by-products ofmounting a dome on dome over four arches, you have Dennett ac? rounded arches." In what sense nec? begins his rebuttal by what you might call an obligatory de? now use as knowledging the wide spread essary? Itmight appear at first if sign opportunity: You have to put were no of the term "spandrels" by evolution? there alternatives to smooth something there to hold up the on ists, and then he goes to appraise tapering triangular surfaces in be? dome?some shape or other, you de? the architectural side of themetaphor: tween the dome and the four round? cide which. But ifwe interpret span? are ed arches, but there in fact indef? drels as obligatory places for one we or are hard? The spandrels of San Marco, initelymany ways that these spaces adaptation another, they are told, "are necessary architectural could be filled with masonry [see ly a challenge to adaptationism. But is there nevertheless some other way inwhich spandrels in the narrow sense?pendentives?truly are nonoptional features of San Mar? co? That iswhat Gould and Lewon? tin seem to be asserting.... Not only were the pendentives just one ^^^Rb^^^Kt^ ^J^^^^Si^r^ ^^^^^T among many imaginable options; theywere just one among the readi? ly available options. [According to Richard Krautheimer] squinches had '^^^^ been a well-known solution to the ^^^^^^^^ problem of a dome over arches in since about the seventh century. What the actual design of San Marco?that is, pendentives?has going for it are mainly two things. 4. on as First, it is a mini? Figure Arches based half-circles, such these supporting the Segovia Aqueduct (first cen? (approximately) are not mum tury A.D.), optimal structural forms. The surcharge placed in the triangular spaces, or "energy-surface" (what you spandrels, prevents the arch from deflecting laterally at its haunches. Pendentives, with their would get ifyou stretched soap film serve same heavy surcharge behind, the function in San Mark's. (Photograph by the author.) in a wire model of the corner), and

386 American Scientist, Volume 84 hence it is close to a rrurrimumsur? monly used for transition to a square face area.... Second, the smooth sur? planned space below. face is ideal for themounting ofmo? Conventional architectural defini? saic images?and that is why the tions, unfortunately, rarely touch on Basilica of San Marco was built: to technical function. The description of a provide a showcase formosaic im? spandrel indeed seems to indicate only ages. The conclusion is inescapable: geometric necessity. When the span? the spandrels of San Marco aren't drel is applied with semicircular ma? spandrels even inGould's extended sonry arches, however, it also takes on a sense. They are adaptations chosen crucial structural function. An arch from a set of equipossible alterna? can span relatively long distances tives for largely aesthetic reasons.... (compared with a simple beam) be? Their primary function [was] to pro? cause its rise generates horizontal as vide a showcase for [Christian] sym? well as vertical forceswithin the arch ... bols. The Byzantine style ofmosa? that confine the individual stones (or ic decoration had provoked the voussoirs) comprising it.Yet arches may admiration of powerful Venetians prove unstable and collapse unless who wanted to create a local exam? their form an approaches optimal 5. Dennett Otto the that induces resultant internal Figure Squinches, suggested, ple. Demus, great authority shape could also be substituted for ... pendentives, on San Marco mosaics, shows forces to act the axis of the arch: a along making them one "among many imaginable that themosaics are the raison d'etre for an arch of uniform catenary depth options." In fact, squinches?shown here as of and hence of of its own dead San Marco, many supporting weight. (For arches placed across square corners to form a an sufficient its architectural details.... most practical applications, catenary octagon?are structurally only Gould and Lewontin's other exam? is closely enough approximated by a for relatively small structures. Pendentives became the element of choice in the sixth ple fromarchitecture was also ill cho? parabola, or for shallow arches, even a.D. to to sen, as it turns out, since we simply by a circular arc.) Semicircular, or "Ro? century provide support large domes over arches springing from four piers. don't know whether the King's Col? man," arches, used well into theMid? Since pre-scientific builders relied heavily on lege bosses alternating rose and dle Ages because they are relatively proven approaches in their designs, it is are the raison d'etre of the to out and construct, are not portcullis simple lay unlikely that the designer of St. Mark's fan or vice versa.... The an arch form.Hence re? vaulting pur? optimal they would have considered other systems. pose of thebosses isprobably entirely quire heavy, solid fill (surcharge) to to provide focal points forornament. prevent outward displacement at the Unlike pendentives, needed for the Did the bosses have to be there any? haunches of the arch that could lead to support of large domes, squinches are way? No. From an engineering point failure. Such a surcharge, as illustrated appropriate only for relatively small of view, there could have been neat in Figure 4, is ubiquitous in all but dec? scale domes?a salient characteristic of round holes there, "lanterns" letting orative Roman arches and constitutes Byzantine church domes after the sev? indaylight fromabove if itwasn't for what is normally referred to in the ar? enth century?where outward-acting the [timber roof above the fanvaults]. chitectural literature as a spandrel. forces are also relatively slight. Both Maybe fan vaulting was chosen by The conventional definition of the the conventional definition of the the builders so that the ceiling could pendentive also indicates only a geo? squinch and Richard Krautheimer's carry Tudor symbols! metric transition. Yet, as I have demon? comment about its almost boundless strated from structural modeling of use fail to account forbuilding scale? These architectural issues can best be Imperial Roman and Byzantine monu? although I have no doubt that the late a Krautheimer dealt with firstby augmenting the or? ments, the pendentive is actually Professor would have or use. dinary definitions of the elements to in? translation of the surcharge, the so agreed with this limitation in as clude their technological functions, and called "step rings" typically found As to building history, San Marco, then by reviewing the course of archi? above large Roman domed buildings themain church ofVenice and the pala? tectural development that led to their (including the Pantheon constructed in tine chapel of the doge (the city's chief employment in San Marco and in the Rome between 118 and 128 A.D.). The magistrate), was constructed between chapel of King's College. step rings are, in turn, a translation of 1063 and about 1095 A.D. and was con? the planar arch spandrel to a three-di? sciously based on the five-domed Greek Saint Mark's, Venice mensional dome. The solid-fill sur? cross sixm

1996 July-August 387 ca, San Marco. Although the central suiting from the desire to increase the This description of the type of vault? me? dome of San Marco, just short of 13 size of cathedral clerestories in order to ing that came into prorninence inmon? ters in diameter, ismuch smaller than display more glass. These explanations umental English halls after themiddle the 31-meter-diameter dome of Hagia neglect the fact thatHigh Gothic archi? of the fourteenth century applies equal? was a Sophia, it is considerably larger than tecture largely phenomenon of lywell to the central-pier-supported ra? typical Byzantine domes constructed af? extraordinary urban growth where dial vaulting of polygonal English era. ter the Justinian Without benefit of mercantile and ecclesiastical centers cathedral chapter houses thatdate from detailed stress analysis, it cannot be de? vied with one another over height of after themiddle of the 13th century (see as as finitively argued that squinches?or, building, Chicago and New York Figure 7). Indeed, the dating suggests even an were suggested by Dennett, array of to do seven centuries later. Indeed, that the concept of fan vaulting derived brackets?might suffice for supporting to fully understand the raison d'etre of from the chapter houses. Yet the adop? one ac? the San Marco central dome. But bear anymonumental building, must tion of radially ribbed vaulting from a inmind that the of structural count the itwas central to a or process for all functions de? plan square- rectangular era was design in theprescientific large? signed tomeet, including the icono planned bay leaves an opening created on ly based lessons learned from ob? graphic and symbolic associations of its by theperimeter of the conoid tops that serving the behavior of earlier build? forms and the cultural contexts of itspa? requires closure (see Figure 2)?to pro? a ings. Hence I strongly doubt that trons and users, as well as available vide horizontal restraint at the top of or an sixth- 11th-centurybuilder would technology. I.M. Pei's design of the East the vault ribs, to help prevent the have dared consider any system other Wing addition to theNational Gallery spread of fire from the timber roof than pendentives to support a dome of inWashington was farmore directed to above the hall, to enhance the hall's San Marco's scale set over four the art piers. single purpose of displaying acoustic ambience and, perhaps not at ar? Concerning themerit ofDennett's thanwas themain church ofVenice, but all least in importance, to provide a bar? for the of the all visitors to the are made rier the avian com? gument primacy (indeed building against ever-present aware truly splendid) mosaics of San Marcos, acutely that itsdesign was affect? munity in the roof. This ismost simply one a must realize that all specialists tend ed by host of other issue as well. effected by the placement over the see own as to their discipline being of opening of a large, flat stone plate. The a greatest import to large undertaking. King's College Chapel, Cambridge boss is actually only the ornament pro? The marked increase in the height of Fan Vaulting: A ceiling or roof structure truding from that plate. Gothic cathedrals at the beginning of formed of conoid surfaces of rotation In late Gothic architecture, particu? the 13th century, for example, has been about the axes of supporting piers larly in England and in Germany, the as explained by stained-glass experts re from an array of uniform ribs. classic quadripartite ribbed vault com? posed of diagonal and cross ribs was frequently supplanted by vaults dis? playing farmore intricate rib patterns. In these later vaults, the closely spaced ribs effectively bear the weight of the vaulting, whereas the thin stone web? bing (classically composed of voussoirs that function in three dimensions much like the load-bearing voussoirs a of two-dimensional arch) merely fills - interstices between the ribs. The clus? <-;.;.';'??- tering of ribs also provides a strong decorative motif described as having "an atmosphere of heavy luxuri? vault re? ance"?hardly any surface mains available for applied decoration. The 13-1/2-meter-span fan vaulting of King's College Chapel at Cambridge dates frombetween 1466 and 1515. De? spite possible cost savings from the use of standardized, radial ribs, the build? ing account gives the cost of construct? ing the fan vault as two-thirds more than that of a conventional quadripar? titevault of equal span that could have displayed considerably more applied decoration. Rather than providing an appropriate "ceiling [to] carry the Tu? dor it is farmore that Figure 6. Sketches of comparative sections taken through the early second-century Pantheon symbols," likely a fan was selected for the col? in Rome (left) and along diagonal of the sixth-century Hagia Sophia in Constantinople vaulting hall in order to an (right) reveal the vital stiffening surcharges around the exterior of the Pantheon dome and lege adopt up-to behind the pendentive in Hagia Sophia, a design element consciously used in St. Mark's. date, high architectural style.

388 American Scientist, Volume 84 a as at was in houses in the Figure 7. Radial ribbed vaulting around central pier, shown here Salisbury Cathedral, employed English chapter to a came into in second half of the 13th century. Fan vaulting, the translation of this vaulting pattern rectangular-planned space, prominence construction was itwas for such as the English Halls a century later. And although the cost of such high, probably adopted buildings Chapel of the of Kings College because itwas considered up-to-date high architectural style. (Photograph courtesy author.)

The Score of surface decoration that can be altered Gould, S. J., and R. C. Lewontin. 1979. The Metaphor of San Marco and the Gould and Lewontin's of at will?"You have to spandrels Panglossian misapplication put something A of the the term for there tohold the dome?some paradigm: critique adaptationist spandrel pendentive per? up shape programme. Proceedings of theRoyal Society a wider of or decide the B205581-598. haps implies latitude design other, you which"?ignores of London, choice than intended for theiranal? or in some cases even of they years, centuries, Gould, S. J. 1993. Fulfilling the spandrels of the to world and mind. In ogy. But Dennett's critique of the archi? construction experience that led their Understanding Scientific ed. Seltzer. Madison: of tectural basis of the goes even inhistoric bu?dings. Prose, J University analogy incorporation Wisconsin Press. furtherastray because he slights the tech? W. 1980. Fan A nical rationale of the architectural ele? Acknowledgments Leedy, C, Jr. Vaulting: Study of Form, and Santa Moni? ments in He isnot alone in this. Mark's recent research on ancient and me? Technology, Meaning. question. ca, Calif.: Arts & Architecture Press. David and I dieval structurehas been supported by the My colleague Billington McDonald, W. L. 1992. Roman experimental have out that the is National Science Foundation, theNational In pointed problem design and the great church. Hagia Sophia Foundation theHumanities and theAn? widespread both in historical analysis for from theAge of Justinian to thePresent, ed. R. drewW. Mellon Foundation.He is indebted Mark and A. S. Cakmak. New York: Cam? and in current architectural theory: "Fol? also to author The bridge University Press. lowing art history,contemporary writing Jonathan Weiner, of R. 1982. in Gothic Structure. on architecture tends to focus on formal Beak of the Finch (1995),for bringing the Mark, Experiments Cambridge: The MIT Press. where visual ideas dominate the ongoing spandreldebate tohis attention. analysis and D. P. 1989. Structural discussion of the and of Mark, R, Billington. origin meaning and the of new form. Tech? is touched References imperative origin style.Technology rarely upon, and Culture, vol. 30(2):300. O. 1984. The Mosaics San Marco in nology and stmcture, under? Demus, of although generally Venice. of Press. Mark, R. 1990. Light, Wind, and Structure. Cam? stood as is seen as a le? Chicago: University Chicago necessary, hardly bridge: The MIT Press. Dennett, D. C. 1995. Darwin's Dangerous Idea: of form, even for ed. Architectural to gitimate giver large Evolutionand theMeaning ofLife. New York: Mark, R, 1993. Technology Up To Dennett Revolution. MIT Press. scale building." his credit, Simon & Schuster. theScientific Cambridge: does not focus on But his treatment N. 1974. An Outline Archi? style. Krautheimer, R. 1979. Early Christian Byzantine Pevsnar, of European 7th ed. Harmondsworth: of crucial structural elements as a kind Architecture. Harmondsworth: Penguin. tecture, Penguin.

1996 July-August 389