Biol Philos (2009) 24:215–230 DOI 10.1007/s10539-008-9141-y
San Marco and evolutionary biology Alasdair I. Houston
Received: 15 October 2008 / Accepted: 15 October 2008 / Published online: 6 January 2009 Springer Science+Business Media B.V. 2008
Abstract Gould and Lewontin use San Marco, Venice, to criticise the adaptationist program in biology. Following their lead, the architectural term ‘‘spandrel’’ is now widely used in biology to denote a feature that is a necessary byproduct of other aspects of the organism. I review the debate over San Marco and argue that the spandrels are not necessary in the sense originally used by Gould and Lewontin. I conclude that almost all the claims that Gould makes about San Marco are wrong and that it is reasonable to view the architectural spandrel as an adaptation. The spandrels example has not provided a good illustration of why adaptive explanations should be avoided. In fact, it can be used as an example of how adaptive explanations can be dismissed even when there is evidence in their favour. I also discuss the use of the concept of a spandrel in biology. Keywords Adaptation Constraint Mosaic Necessity San Marco Spandrel Symmetry
The spandrels of San Marco Gould and Lewontin (1979) introduced the concept of a spandrel as a necessary byproduct of an overall design. Their paradigmatic case is San Marco (St Mark’s), Venice, whose spandrels were taken to be a necessary consequence of mounting domes on rounded arches. The discussion of spandrels forms part of a criticism by Gould and Lewontin of ‘‘adaptive’’ explanations in evolutionary biology. Gould and Lewontin (1979) is a very influential paper. Both the biological argument (e.g. Mayr 1983; Stephens and Krebs 1986; Williams 1992) and its style of presentation (Selzer 1993) have been evaluated. Spandrels are discussed in a wide range of contexts, A. I. Houston (&) School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, UK e-mail:
[email protected]
123
216
A. I. Houston
including the complexity of biological networks (Sole and Valverde 2006), biochemistry (Landweber and Pokrovskaya 1999; Weiss et al. 2002), locomotion (Aerts et al. 2003), language (Botha 2001; Pinker 1994; Pinker and Bloom 1990), music (Jackendoff and Lerdahl 2006; Peretz 2006), psychology (Hampton 2004; Schacter and Dodson 2001; Van Leeuwen 2007), technological change (Turnbull 2000) and aesthetics (De Sousa 2004) and are mentioned in many books (e.g. Conway Morris 1998; Dupre 2001; Gazzaniga 1994; Pinker 1994; Willis 1995; Wilson 2002). In this paper I evaluate various claims made by Gould and Lewontin (1979) and Gould (1997, 2002) about the structure and decoration of San Marco. Before turning to the biological issues, there are three related points that I will address. 1. 2. 3.
The necessity of certain aspects of the architecture of San Marco. The possibility that the structure was chosen so as to display mosaics. The view that the structure influenced the design of the mosaics.
Architectural necessity Gould and Lewontin (1979) criticise what they call the adaptationist programme or Panglossian paradigm, i.e. the approach in which every aspect of an organism’s morphology and behaviour is explained in terms of the contribution that it makes to the organism’s ability to survive and reproduce. The paper begins with a description of the central dome of San Marco, Venice. The dome rests on four round arches, with a structure known as a spandrel or pendentive occurring where each arch meets a neighbouring arch (see Fig. 1b, c). The dome and the spandrels are decorated with mosaics. Gould and Lewontin (1979) suggest that the harmonious way in which the decoration is displayed might tempt an adaptationist to see the spandrels as being chosen in order to display the mosaics. They argue that this would be a completely incorrect view. According to Gould and Lewontin, the spandrels are a necessary byproduct of mounting a dome on round arches. Spandrels were forced on the builders because they built a dome on round arches, and the spandrels were then used as a surface that could be decorated. Gould sees this architectural example as making two related points that are important for evolutionary biology. 1.
2.
A feature may originate from geometric constraints. Just as the spandrels are a necessary consequence of the design of the dome, so various biological features are the result of architectural constraints (Gould and Lewontin 1979, p. 595). A feature’s current utility may not indicate the reason for its original appearance. Gould (2002, p. 1233) defines an adaptation as a feature that has been shaped by natural selection for its current use. A spandrel is a feature that may be useful but is not an adaptation.
Before discussing necessity, I will define the architectural terms that are used in this paper. I follow the terminology of Mark (1996, p. 387).
123
San Marco and evolutionary biology
Spandrel
Pendentive Squinch
217
A triangular wall area over the haunches of an arch, or in multi-story buildings, the area of wall between the vertical supports and upper and lower windows (see Fig. 1a). A triangular segment of vaulting at the base of a dome used to make the transition from a round- to a square-planned space (see Fig. 1b, c). An arch, or corbeling, at the base of a dome or … vault used to make the transition to a square-planned space below (see Fig. 1d, e).
Gould and Lewontin (1979, p. 581) introduce spandrels as follows: ‘‘Spandrels—the tapering triangular spaces formed by the intersection of two rounded arches at right angles (figure 1)—are necessary architectural byproducts of mounting a dome on rounded arches.’’ The same point is also made at the end of the paper (p. 595): ‘‘Spandrels must exist once a blueprint specifies that a dome shall rest on rounded arches.’’ I will use the term ‘‘GL spandrel’’ for a spandrel in the sense of a feature that is a geometrically necessary byproduct. Types of necessity It is important to distinguish between various ways in which a feature can be said to be necessary. Broadly speaking, there seem to be three relevant senses: geometric, functional and cultural. (a) Geometric necessity. Following Dennett (1995), I refer to a feature as a geometric necessity if its existence in a building follows from the geometric implications of the design specification of the building. Gould sometimes calls this ‘‘structural necessity’’ (e.g. Gould 2002, p. 1250). (b) Functional necessity. Functional necessity is to do with whether a feature is required in order for a structure such as a building of a given design to function adequately. In the case of a building, an obvious requirement is that it does not collapse. It will often be difficult to decide if a given feature is a functional necessity. The stability of a building depends on not just its design but also on the materials used, the environmental conditions and the period of time over which performance is evaluated. Although it may be hard to decide if a feature is a functional necessity, it will often be easy to decide if a claim about necessity is based on function rather than geometry. If the claim is based on the performance of the building and depends on its size or the materials used in its construction, then the claim is concerned with functional necessity. (c) Cultural necessity. We might say that builders are culturally constrained if they are limited in the number of construction methods that they know. If only one method is known, then we could say that it is adopted because of cultural necessity. In moving from an architectural to a biological context we can draw up the following approximate correspondences:
123
218
A. I. Houston
a
b
d
c
e
Fig. 1 a A spandrel; b, c a dome on pendentives; d, e a dome on squinches. (a based on Houston (1997).)
123
San Marco and evolutionary biology
Geometric necessity
219
Physical or geometrical constraints
Functional necessity
Strong selection on good design
Cultural necessity
Genetic constraints on possible forms
Are pendentives necessary? ‘‘Spandrels’’ in the above quotes from Gould and Lewontin (1979) are what we are now calling pendentives, so what Gould and Lewontin (1979) claim is that if a dome is mounted on rounded arches, pendentives are inevitable byproducts, i.e. pendentives are GL spandrels. Note that the quotations make no mention of the size of the building, the materials to be used in its construction, or the environment in which it must stand. This justifies my belief that we are concerned here with geometric necessity rather than functional necessity. Dennett (1995), Lauder (1996) and Houston (1997) give counterexamples to show that pendentives are not necessary. Gould (1997) does not refute the counterexamples. Instead he justifies his position by claiming that Mark (1996) views pendentives as necessary. As I now show, this misrepresents Mark. Gould (1997, p. 10752) says Robert Mark, the only participant in this debate with requisite architectural expertise, affirms our central point (while also noting the direct utility of pendentives in buttressing). He writes of our San Marco example (ref. 3, p. 386): ‘‘Domes mounted on arches create roughly triangular spaces in the corners, between the upper sides of the arches and the base of the dome.’’ Mark (1996) goes on to say that ‘‘pendentives … are necessary structural elements.’’ He does, however, consider the possibility of a dome without pendentives in buildings of smaller scale. Here is the full quotation (caption to Mark’s Fig. 3. The figure that Mark refers to as ‘‘left’’ is similar to Fig. 2a of this paper, the figure that Mark refers to as ‘‘right’’ is similar to Fig. 2b). Figure 3. Domes mounted on arches create roughly triangular spaces in the corners, between the upper sides of the arches and the base of the dome. Gould and Lewontin called these spaces spandrels, whereas Dennett (correctly) identified them as pendentives. The former saw these spaces as opportunities for decoration left by the choice of the structure, whereas the latter suggested that the structure was chosen to make possible the spaces for the display of mosaics. In fact, the pendentives (left), which are but the visual surfaces in front of a heavy surcharge, are necessary structural elements. Brackets (right), suggested by Dennett as substitutes for pendentives, are not ‘‘equivalent’’ because of [sic] they lack such surcharge as well as support from any other massive structure. Although the sentence quoted by Gould might be taken as support for the geometric necessity of pendentives, it can be seen from the above quotation that Mark is discussing an alternative to pendentives. (The quotation also shows that
123
220
A. I. Houston
a
b
Fig. 2 A pendentive (a) and an alternative using brackets (b). Based on Dennett (1995) and Mark (1996)
Mark supported Dennett in the question of terminology, something that is not obvious from Gould (1997).) If Mark thinks that pendentives are geometrically necessary, how is it possible for him to compare pendentives with alternatives such as brackets or squinches? In fact, Mark distinguishes between the geometry of a pendentive and the solid material behind the surface that prevents the dome from being pushed outwards. It is this surcharge rather than the shape of the surface that Mark regards as necessary. A structure that looked like a pendentive but lacked the surcharge would not provide adequate support. Mark (1996, p. 387) says: ‘‘Unlike pendentives, needed for the support of large domes, squinches are appropriate only for relatively small-scale domes.’’ Note that the need is based on the performance of the structure, i.e. it is functional necessity. Mark does not say that a building the size of San Marco had to be built using pendentives; he says that builders of the time would not have dared consider any alternative to pendentives. Dennett (1995, p. 274) says that pendentives were ‘‘chosen from a set of equipossible alternatives largely for aesthetic reasons.’’ Mark does not dispute the possibility of alternatives, but he does not regard them as equivalent because they differ in stability. In other words, Mark challenges Dennett’s suggestion that variation in dome support is neutral in terms of structural soundness and argues that pendentives are functionally necessary. In his response to Dennett (1995), Rose (1997) defends the claim of Gould and Lewontin (1979) that pendentives are inevitable, but does so on the grounds that they are required for functional reasons. Rose states that there are several possibilities including pendentives and that pendentives are advantageous in terms
123
San Marco and evolutionary biology
221
a
b
c
Fig. 3 Domes without pendentives or squinches. a exterior view. b and c possible interior views. b based on Fig. 498 of Creswell (1979), c based on Fig. 499 of Creswell (1979)
of the stability of the structure. Far from refuting Dennett, Rose is supporting him against Gould and Lewontin. Creswell (1979) discusses the evolution of the pendentive. He gives examples of domes without pendentives or squinches (see Fig. 3). Far from being geometrically necessary, the pendentive was an invention, see for example Crouch and Johnson (2001, p. 345), Lassus (1966, p. 72), Nuttegns (1983, pp. 118–119), see also Lauder (1996). The claim by Gould (1997, p. 10751) that pendentives are ‘‘not explicitly designed’’ is untenable. Logic and architectural history show that they are designed. After mentioning Dennett’s argument and Mark’s response, Segerstra˚le (2000, p. 117) summarises the architectural debate by saying that spandrels are still ‘‘a necessity, not a choice, for historically constrained structural reasons.’’ Although the builders of San Marco were influenced by a previous building (e.g. Howard 2002; Mark 1996), it is not clear that they were historically constrained in that they knew no alternatives to the pendentive.
Was the structure chosen to display the mosaics? Gould and Lewontin (1979) claim that it would be a mistake to regard the pendentives of San Marco as being chosen in order to display the mosaics. Dennett (1995) challenges this. To decide who is correct, we need to know whether the plans for the form and decoration of San Marco started with the structure based on
123
222
A. I. Houston
pendentives or with the desire for mosaics. Gould (1997, pp. 10752–10753) puts the issue as follows: Did the architects decide to mount a dome on arches, thus engendering the spandrels as a necessary consequence, or had the designers devised such a good plan for decorating spandrels that they persuaded the architects to provide the four pendentives so that they could execute their evangelical design? (see also Gould (2002, p. 1255)) Gould (1997, p. 10753) describes two methods that he claims can be brought to bear on this question of ‘‘causal sequence,’’ i.e. the sequence of intentions. First (and evidently superior for relying on raw observation rather than inference), we might obtain evidence for an actual historical order and therefore be able to know which feature arose first as a primary adaptation and which subsequently as a coopted byproduct. Second (and more generally applicable in relying on available data of present cases, but necessarily inferential), we may tabulate the ‘‘comparative anatomy’’ of current examples in a cladistic context and try to determine a historical order from the distribution. The same two methods are advocated by Gould (2002, p. 1256). Gould does not provide any justification that these methods are valid. I now look at how Gould (1997) applies each of these methods to San Marco and then make a general assessment of both of them. The historical approach In the case of San Marco, Gould (1997, p. 10753) states that: ‘‘The domes of San Marco, and all accompanying structures, were built three centuries before the mosaicists placed their designs of such excellent fit into the pendentives.’’ The reference given is Demus (1984). This claim is repeated by Gould (2002, p. 1256): Direct historical data. We know that the spandrels were not built to house the evangelists because San Marco stood and operated in its appointed role as a church for at least three centuries before the mosaicists applied their astonishing work to a series of constrained and previously unornamented spaces (see Demus’s classic four volume monograph (1984) on the history, architecture and iconography of San Marco). Does this tell against Dennett’s claim that the pendentives were chosen in order to display mosaics? The first point to note is that the mosaics could only be put in place once the dome and pendentives were constructed, so regardless of the intentions of the builders, the sequence of construction has to be dome then mosaics. The crucial question concerns intention: was the structure chosen because it was intended to decorate it with mosaics? I presume that the gap of three centuries between finishing the structure and decorating it that Gould mentions is taken to be
123
San Marco and evolutionary biology
223
evidence against the view that the mosaics were planned first. There are two problems with this argument. 1.
2.
Medieval building programmes often lasted for hundreds of years (e.g. Kraus 1979). So although delay might suggest that the mosaics were not planned first, by itself is does not rule out the possibility; we need information on the intentions of the builders. I have not been able to find anything explicit on this in Demus (1984, 1988), but Howard (2002, p. 27) states that mosaics were always intended. The gap that Gould refers to does not exist. Demus (1984, p. 2; 1988, p. 3) states that there are various dates for the consecration of San Marco, ranging from 1084–1085 to 1117. If Gould’s assertion is correct, this would mean that the mosaics of the dome date from about 1400. According to Demus (1988, p. 62) ‘‘The mosaics of the central dome and its pendentives date from the great renewal of the twelfth century following the catastrophe whose traces are visible in the east dome.’’
Comparative analysis of domes Gould says that Dennett’s claim about the mosaics can be investigated by a comparative analysis of buildings with domes. For reasons that are not given, this analysis is confined to Western buildings (Gould 1997, p. 10753, 2002, p. 1256): Thousands of Western buildings feature domes atop rounded arches—and every single one of them generates tapering triangular spaces at the intersections. Gould uses this statement to support the claim that pendentives are necessary. I find this strange. One does not establish a geometric necessity by citing confirming examples. Furthermore, if pendentives are necessary, why is attention confined to Western buildings? Are pendentives not necessary in Eastern buildings? In any case, the statement is not correct—domes on squinches (Fig. 1d, e) can be found in Western buildings, including Sta Fosca, Torcello in the Venetian lagoon (Lassus 1966; Nuttegns 1983). Gould’s two approaches, the historic method and the comparative method, are meant to be shedding light on what Gould (1997, p. 10753) calls the causal sequence. What needs to be established is whether the pendentives were chosen for their own sake or in order to display the mosaics; see Gould (1997, pp. 10752– 10753; 2002, p. 1255). We are concerned with the truth of the statement: The builders of San Marco chose pendentives in order to display mosaics. (Dennett (1995, p. 274) is explicitly concerned with San Marco.) The analysis presented by Gould (1997, 2002) claims that a building’s causal sequence can be determined from the historical sequence, which can be observed directly (Method 1) or inferred from a comparative analysis (Method 2). Gould provides no support for this claim. In his analysis of San Marco, Gould takes the historical sequence to be the sequence of construction, from which it might be assumed that it is equivalent to ontogeny in the case of organisms, but in the context of snails Gould (1997, p. 10753) says ‘‘We
123
224
A. I. Houston
cannot use the first method of actual historical sequence to resolve this question because we do not know whether the first coiled snails brooded their eggs in an umbilical chamber.’’ This leaves me unsure as to what Gould takes the biological historical sequence to be. In addition, the biological analogy of architectural intention has not been specified. I have pointed out that the sequence of construction at San Marco is subject to its own constraints and need not reflect the causal sequence. Such constraints on the order of construction are likely to be common. Given all these points, why should Gould’s comparative analysis of buildings shed light on the causal sequence of a particular building?
Did the structure influence the mosaics? To illustrate his view that secondary features can have an important impact on the overall design of an object, Gould (1997, p. 10754, 2002, p. 1254) claims that the number of pendentives (four) accompanying each of San Marco’s domes is reflected, in four out of five cases, in the (four-part) symmetry of the mosaics of the domes themselves. He also claims that in these four cases, the four-part symmetry of the dome design is related geometrically and iconographically to the pendentives. In contrast to Gould’s claim that four out of five domes have four-part symmetry, I find that at best two out of five domes have four-part symmetry (details in Table 1). Contrary to the claim of Gould and Lewontin (1979) and Gould (1997, 2002) the central dome is not one of them. There is no convincing evidence of iconographic links between particular pendentive mosaics and neighbouring mosaics of the dome. Structure and mosaics Gould’s discussion of the symmetry of the domes is part of an argument in favour of secondary features (‘‘spandrels’’) exerting an important influence. Even if there had been consistent four-part symmetry in the dome mosaics, it is not clear that it should be attributed to the pendentives rather than the decision to use four arches. The number four has special significance (four evangelists, four horsemen, four last things, four elements, four humours, four cardinal directions, four seasons, etc.) and so any four-part symmetry of a dome may reflect this basic theme rather than the fact that there are four pendentives. A possible predominance of four-part symmetry over structure is illustrated by the dome of La Martorana in Palermo. This dome is mounted on four squinches such that the structure below the dome forms in plan an octagon, but the dome decoration of four archangels that surround Christ has fourpart symmetry ((Demus 1950) plate 46 and p 79). In a comparison of Byzantine churches, Ousterhout (1992) argues for the influence of mosaics on structure. He says (p 60) that compared to Hosios Loukas and Daphni ‘‘Nea Moni offers an innovative alternative for the placement of mosaic.’’ ‘‘Whatever its ultimate source, the building seems to have been designed specifically for mosaic decoration. The proposed changes in its design and its bold, new formulation are best seen as a direct response to aesthetic concerns, to the important mosaic program with its imperial overtones.’’ For further discussion see
123
San Marco and evolutionary biology
225
Table 1
The dome mosaics of San Marco
Dome
Main Circle
Pendentives
4-part?
Bilateral?
Icon
Themes, basic symmetry and iconographic links (‘‘Icon’’) between pendentive mosaics and neighbouring mosaics of the dome Central
Virgin, 2 angels, 12 apostles
Evangalists
No
Yes
East
Virgin & 13 prophets
Symbols of Evangalists
No
No
?
West
12 apostles
angels
Yes
Yes
No
North
St John & 5 scenes from his life
Doctors of the Church
No
Yes
No
South
4 male saints
Female saints
Yes
Yes
Weak
Dome
Dome mosaic
Weak
Dome mosaic ? pendentives
Symmetry groups Central
m
m (if all Evangalists equivalent)
East
1
1
West
4 mm
2 mm
North
m
m
South
4 mm
m (If all female saints equivalent)
The table uses the terminology of Stevens (1980): 1, asymmetric; m, bilateral symmetry; 2 mm, 2 rotations & 2 reflections; 4 mm, 4 rotations & 4 reflections Notes: If each figure in a mosaic is represented exactly, there will typically be no symmetry (i.e. only the identity transformation is possible). To avoid this, I sometimes take similar figures to be identical (cf. the remarks on relative equality in Shubnikov and Koptsik (1974)). For example, in the central dome, the figure of the virgin, together with the figures of the angels and the apostles are all of about the same size and so can be treated as identical. It is also necessary to assess the symmetry of the design of the dome itself, as well as that of the dome plus the pendentives. This is because including the pendentives automatically restricts the design to at most four-part symmetry
Ousterhout (1999, pp. 97–98). Regardless of the origin of pendentives (or squinches) this argument indicates that details of form might be influenced by the desire to apply mosaics.
Adaptation and constraint So far I have concentrated on architectural issues. I now consider San Marco in a broader context. The pendentive as an adaptation Gould stresses the importance of distinguishing between the historical reasons for a feature and the feature’s current utility. Gould and Vrba (1982) suggest that an adaptation should be defined as a character that natural selection has shaped for its current use. An obvious extension of this definition to artefacts would be as follows. A feature of an artefact is an adaptation if its origin and form can be traced to selection for its current role, cf. Williams (1966, p. 6). The pendentive satisfies this
123
226
A. I. Houston
definition. Creswell (1979) mentions the long period of unsatisfactory domes that preceded the invention of the pendentive. Furthermore, the pendentive was invented to fulfil its current role of improving the stability of a dome. Gould (1997, p. 10751) rejects the claim that pendentives are adaptations. He suggests that people take a pendentive to be an adaptation rather than a GL spandrel because it does something useful. This is not my argument. My view is that pendentives are not GL spandrels because they are not geometrically necessary. The extra facts that they do something useful and were introduced to do this thing clinch the case for taking them to be architectural adaptations (cf. Gould and Vrba (1982, p. 4)—‘‘designed specifically for the task it performs.’’) Gould and Vrba (1982) introduce the term exaptation to refer to a character that does something useful but whose origin cannot be traced to its current function. They give two cases: 1. 2.
A character, previously shaped by natural selection for a particular function (an adaptation), is coopted for a new use. A character whose origin cannot be ascribed to the direct action of natural selection (a nonaptation), is coopted for current use.
GL spandrels form a subset of case (2) above. Unlike a GL spandrel an exaptation does not have to meet the requirement of being ‘‘necessarily and predictably shaped in a certain way, … arising as an inevitable side consequence …’’ (Gould 1997), so not all exaptations are GL spandrels. Gould provides evidence that certain biological features are exaptations, but for a feature to be a spandrel it must have a necessary form that arises as an inevitable side-effect. Gould typically does not provide the extra evidence about necessary form. Some of the biological examples Gould and Lewontin (1979) suggest that Seilacher (1972) takes the divaricate pattern of molluscs to be a spandrel, and state (p 596) ‘‘The divaricate pattern is a fundamental architectural constraint.’’ I do not find this claim convincing. There does not seem to be any reason to view the pattern as a necessary feature. As Seilacher (1972, 1973) points out, it is not found in all species. Seilacher views the pattern to be an example of ‘‘fabricational noise’’—something that is an incidental consequence of the way in which an object is made. He gives the seam on modern South American plastic water jars as an example. These jars are made in the same shape as traditional earthenware jars, but the new method of construction produces a seam. Gould (2002) gives a similar example based on using a mould to make glass bottles. In these cases, there is nothing necessary about the shape of the object that requires it to have a seam. The seam is a contingent side-effect of a particular construction process. A test for whether a feature is geometrically necessary is to ask whether the object can be imagined without the feature. It is clear that we can have jars and bottles without seams. Similarly, we can have shells without divaricate patterns. These patterns are not a ‘‘fundamental architectural constraint’’ and hence are not spandrels.
123
San Marco and evolutionary biology
227
Gould (1997) gives some other possible examples of spandrels. One of them concerns the genitalia of female spotted hyenas. The suggestion is that selection for females to be larger than males has resulted in high levels of testosterone in females, with ‘‘masculinized’’ female genitalia being a by-product. As in the case of shell patterns, there is no justification here for regarding the feature as a geometrically inevitable consequence. Gould (1997) discusses the extent to which various aspects of human psychology are adaptations shaped by evolution or are side-effects of changes in the brain over evolutionary time. Once again he goes from a plausible case for a side-effect to an unjustified claim for a spandrel. ‘‘Spandrel’’ is used in two different senses in biology. Gould says that spandrels ‘‘arise as necessary byproducts of another decision in design.’’ They are ‘‘inevitable architectural byproducts,’’ or ‘‘ineluctable architectural byproducts’’ (Gould 1997, p. 10754). Williams (1992, p. 78) adopts a less restrictive definition. He says ‘‘A spandrel, in the sense of Gould and Lewontin (1979), is a structure arising as an incidental consequence of some evolutionary change.’’ Under Gould’s definition, the nature of the spandrel can be deduced from the design of the primary object (‘‘a predictable form that arises as a side consequence’’ Gould (1997, p. 10751)). Furthermore, Gould’s definition puts limits on the extent to which a spandrel can be modified. Unless the design specification responsible for the spandrel changes, the spandrel cannot be modified in a way that would remove the necessary features. These properties do not hold under the definition of Williams. Like Williams, Buss et al. (1998) and Hampton (2004) use a definition of a spandrel that makes no mention of the fact that a spandrel has an inevitable form. By removing any mention of architectural constraint and necessity, such definitions make ‘‘spandrel’’ a synonym of ‘‘side effect’’ or ‘‘by-product.’’ As a result, ‘‘spandrel’’ becomes a redundant term. Although the example of San Marco was chosen to clarify the biological argument, it has probably had the opposite effect (a ‘‘revenge effect’’—a special sort of side-effect discussed by Tenner (1996)). Even after Mark (1996) gave his evaluation, Lauder (1996) and Houston (1997) came to one conclusion and Gould (1997) came to another. This shows how hard it can be to reach an agreement about whether a feature is an adaptation. Part of the problem may be the apparent simplicity of the pendentive. Pinker and Bloom (1990) say that no-one would suggest that the mosaic designs arose as a side-effect of the structure or as a result of random processes. But the case of the pendentive is not so clear-cut. It looks simple enough to be a side-effect. But both structural analysis (Mainstone 1999; Mark 1996) and history (Creswell 1979; Hamilton 1933; Lassus 1966) tell a different story. Early attempts to put a dome on arches did not involve pendentives. They are not a side-effect but a specific invention that improved the structural stability of the dome. Gould’s general reason for using an architectural example is that it avoids the automatic tendency to look for an adaptive account for biological features. It turns out, however, that an ‘‘adaptive’’ account may be appropriate for the understanding of architecture in general (Mark 1982, 1990) and pendentives in particular (Mark 1996). In addition, the use of an architectural example may have caused problems.
123
228
A. I. Houston
Gould’s architectural spandrels are based on geometric constraints, whereas many of Gould’s biological examples involve constraints that are not geometrical.
Conclusions A dome can be mounted on four arches without the use of pendentives. This means that pendentives are not a geometric necessity. I have presented evidence that the pendentive is an invention rather than a necessary side-effect and that it is reasonable to view the pendentive as an adaptation. The spandrels of San Marco do not provide a good illustration of why adaptive explanations should be avoided. In fact, they can be used as an example of how an adaptive explanation can be dismissed even when there is evidence in its favour. The comparative analysis offered by Gould (1997, 2002) is flawed. What Gould calls the causal sequence is concerned with the intentions of the people who designed San Marco. Although Gould (1997, p. 10753) regards the sequence of construction of a particular building as providing ‘‘observation rather than inference’’ about the causal sequence, the two sequences are logically distinct. Gould claims that a comparative analysis of buildings can be used to shed light on the intentions of the builders of San Marco. It is hard to see why this claim should be accepted. Gould (1997, 2002) makes some claims about the symmetry of the mosaics in the domes of San Marco. I have evaluated these claims in the light of Demus (1984, 1988) and have found that they do not hold. Some people are prepared to use the term ‘‘spandrel’’ as if it meant no more than ‘‘side-effect’’ or ‘‘by-product’’ despite the fact that this is not how Gould defines it. Although this terminology is understandable in that many of the biological examples that Gould calls spandrels do not satisfy his definition, I suggest that this use of ‘‘spandrel’’ can create confusion and should be abandoned. We could retain ‘‘spandrel’’ in the sense of a GL spandrel, but it seems perverse to use a term based on a feature that is an architectural adaptation to characterize features that are not biological adaptations. I think that evolutionary biology would be better off without the term. Acknowledgments I thank Lutz Fromhage, Robert Mark, Andrew Phillips and Sean Rands for helpful comments on previous versions of the ms and Tim Colborn for producing the figures. This work was supported by a Leverhulme Research Fellowship.
References Aerts P, Van Damme R, D’Aout K, Van Hooydonck B (2003) Bipedalism in lizards: whole-body modelling reveals a possible spandrel. Philos Trans R Soc Lond B Biol Sci 358:1525–1533. doi: 10.1098/rstb.2003.1342 Botha RP (2001) How much of language, if any, came about in the same sort of way as the brooding chamber in snails? Lang Commun 21:225–243. doi:10.1016/S0271-5309(01)00002-7 Buss DM, Haselton MG, Shackelford TK, Bleske AL, Wakefield JC (1998) Adaptations, exaptations, and spandrels. Am Psychol 53:533–548. doi:10.1037/0003-066X.53.5.533
123
San Marco and evolutionary biology
229
Conway Morris S (1998) The crucible of creation. Oxford University Press, Oxford Creswell KAC (1979) Early muslim architecture. Hacker Art Books, New York Crouch DP, Johnson JG (2001) Traditions in architecture. Oxford University Press, New York Demus O (1950) The mosaics of Norman Sicily. Routledge & Kegan Paul, London Demus O (1984) The mosaics of San Marco in Venice. University of Chicago Press, Chicago Demus O (1988) The mosaic decoration of San Marco, Venice. University of Chicago Press, Chicago Dennett DC (1995) Darwin’s dangerous idea. Simon & Schuster, New York De Sousa R (2004) Is art an adaptation? Prospects for an evolutionary perspective on beauty. J Aesth Art Crit 62:109–118. doi:10.1111/j.1540-594X.2004.00144.x Dupre J (2001) Human nature and the limits of science. Oxford University Press, Oxford Gazzaniga MS (1994) Nature’s mind. Penguin, Hammondsworth Gould SJ (1997) The exaptive excellence of spandrels as a term and prototype. Proc Natl Acad Sci USA 94:10750–10755. doi:10.1073/pnas.94.20.10750 Gould SJ (2002) The structure of evolutionary theory. Harvard University Press, Cambridge Gould SJ, Lewontin RC (1979) The spandrels of San Marco and the panglossian paradigm: a critique of the adaptationist programme. Proc R Soc Lond B Biol Sci 205:581–598 Gould SJ, Vrba ES (1982) Exaptation—a missing term in the science of form. Paleobiology 8:4–15 Hamilton JA (1933) Byzantine architecture and decoration. Batsford, London Hampton SJ (2004) Domain mismatches, scruffy engineering, exaptations and spandrels. Theory Psychol 14:147–166. doi:10.1177/0959354304042014 Houston AI (1997) Are the spandrels of San Marco really panglossian pendentives? Trends Ecol Evol 12:125. doi:10.1016/S0169-5347(96)20112-0 Howard D (2002) The architecture of Venice. Yale University Press, New Haven Jackendoff R, Lerdahl F (2006) The capacity for music: what is it, and what’s special about it? Cognition 100:33–72. doi:10.1016/j.cognition.2005.11.005 Kraus H (1979) Gold was the mortar: the economics of cathedral building. Routledge & Kegan Paul, Boston Landweber LF, Pokrovskaya ID (1999) Emergence of a dual-catalytic RNA with metal-specific cleavage and ligase activities: the spandrels of RNA evolution. Proc Natl Acad Sci USA 96:173–178. doi: 10.1073/pnas.96.1.173 Lassus J (1966) The early christian and Byzantine world: landmarks of the world’s art. Paul Hamlyn, London Lauder GV (1996) The argument from design. In: Rose MR, Lauder GV (eds) Adaptation. Academic Press, New York, pp 55–91 Mainstone RJ (1999) Structure in architecture: history, design and innovation: variorum collected studies series, v. CS659. Ashgate, Aldershot Mark R (1982) Experiments in gothic structure. MIT, Cambridge Mark R (1990) Light, wind, and structure. MIT, Cambridge Mark R (1996) Architecture and evolution. Am Sci 84:383–389 Mayr E (1983) How to carry out the adaptationist program. Am Nat 121:324–334. doi:10.1086/284064 Nuttegns P (1983) The story of architecture. Phaidon, Oxford Ousterhout R (1992) Originality in Byzantine architecture ? the Chois Katholikon - the Case of NeaMoni. J Soc Archit Hist 51:48–60. doi:10.2307/990640 Ousterhout R (1999) Master builders of Byzantium. Princeton University Press, Princeton Peretz I (2006) The nature of music from a biological perspective. Cognition 100:1–32. doi: 10.1016/j.cognition.2005.11.004 Pinker S (1994) The language instinct. Penguin, London Pinker S, Bloom P (1990) Natural language and natural selection. Behav Brain Sci 13:707–784 Rose S (1997) Lifelines. Allen Lane, London Schacter DL, Dodson CS (2001) Misattribution, false recognition and the sins of memory. Philos Trans R Soc Lond B Biol Sci 356:1385–1393. doi:10.1098/rstb.2001.0938 Segerstra˚le U (2000) Defenders of the truth. Oxford University Press, Oxford Seilacher A (1972) Divaricate patterns in pelecypod shells. Lethaia 5:325–343. doi: 10.1111/j.1502-3931.1972.tb00862.x Seilacher A (1973) Fabricational noise in adaptive morphology. Syst Zool 22:451–465. doi: 10.2307/2412952 Selzer J (1993) Understanding scientific prose. University of Wisconsin Press, Madison Shubnikov AV, Koptsik VA (1974) Symmetry in science and art. Plenum, New York
123
230
A. I. Houston
Sole RV, Valverde S (2006) Are network motifs the spandrels of cellular complexity? Trends Ecol Evol 21:419–422. doi:10.1016/j.tree.2006.05.013 Stephens DW, Krebs JR (1986) Foraging theory. Princeton University Press, Princeton Stevens PS (1980) Handbook of regular patterns. MIT, Cambridge Tenner E (1996) Why things bite back. Forth Estate, London Turnbull D (2000) Gothic tales of spandrels, hooks and monsters: complexity and association in the explanation of technological change. In: Ziman J (ed) Technological innovation as an evolutionary process. Cambridge University Press, Cambridge Van Leeuwen DSN (2007) The spandrels of self-deception: prospects for a biological theory of a mental phenomenon. Philos Psychol 20:329–348. doi:10.1080/09515080701197148 Weiss MA, Nakagawa SH, Jia WH, Xu B, Hua QX, Chu YC, Wang RY et al (2002) Protein structure and the spandrels of San Marco: insulin’s receptor-binding surface is buttressed by an invariant leucine essential for its stability. Biochemistry 41:809–819. doi:10.1021/bi011839? Williams GC (1966) Adaptation and natural selection: a critique of some current evolutionary thought. Princeton University Press, Princeton Williams GC (1992) Natural selection: domains, levels, and challenges. Oxford University Press, Oxford Willis D (1995) The sand dollar and the slide rule. Adison-Wesley, Reading Wilson DS (2002) Darwin’s cathedral. University of Chicago, Chicago
123