GaWC Research Bulletin 276

GaWC logo
  Gateways into GaWC

This Research Bulletin was delivered as the 10th GaWC Annual Lecture at Loughborough University in January 2008.


The Study of Cities: Historical and Structural Approaches

P. Saey*


Structural approaches relate to the logic of a system, while historical approaches relate to chains of causes and effects. Structural approaches make use of knowledge of the outcomes in the analysis of past events, which creates the pitfall of historical necessity. There are two strategies to avoid this pitfall, i.e. (i) investigating periods in which the outcome is still undetermined and (ii) combining the historical and structural approaches. This paper clarifies the implications of the second strategy through a critical analysis of the replacement of Christallerian ideas about central places and Taylorian ideas about global cities. With regard to Christallerian ideas, three authors are dealt with: Bobek's definition of urban settlements as nodes in a network of flows, Christaller's definition of a spatial structure, and Barton's definition of agency. Each of the three authors rejected the historical approach of explaining the location and development of urban settlements based on site and situation, albeit that Barton had a different concept of centrality than Bobek and Christaller. There are indeed two concepts of centrality, one referring to city-ness and one referring to town-ness. The structural interpretation of Christaller's approach provides the opportunity to define the logic of a system in terms of Elias's concept of rules of figuration. The analytical connection between Christallerian and Taylorian ideas is made through a discussion of Hall's paper Christaller for a global age: it is argued that it runs counter to both Christaller's own theory as well as Taylor's conception of a world city network. Taylor conceives of cities as transhistorical entities. The legitimacy of this concept is examined through a confrontation between three authors who defend a city-based view of history with inter-city relations as the formative force: Pirenne who presents a classical historical story, Jacobs who transubstantiates, and Taylor who (arguably) abstracts. This confrontation serves as a stepping stone for explaining how the structural and historical approaches interact: a structural approach determines the unit of analysis. The importance of this thesis is illustrated by the example of the archipelago of cities 1250-1350 investigated by Abu-Lughod.


In this paper I make the proposal to combine the historical approach to the study of cities with the structural approach. This may sound like a triviality, but as the following sections aim to show, this is most certainly not the case.

A structural approach relates to the logic of systems. A historical approach, in contrast, relates to (i) chains of causes and human purposes and motives, and their intended or unintended results and the possible generalisations or general historical trends that can be deduced from the investigation of these chains; and to (ii) the actual formation, development and disintegration of systems. The historical approach includes the examination of idiographic causal chains as well as Janet Abu-Lughod's analysis of the world system in the period 1250-1350.

This unargued definition relieves me of the arguably impossible task of providing a survey of the several kinds of structural and historical approaches to the study of cities. Cities have an internal structure, they belong to networks, they are undoubtedly important to economic development, and they are an essential part of civilizations. However, the way in which cities are central to economic development and the meaning of being an essential part of civilizations is far from unambiguous. There is a great variety of structural approaches, reaching from pure structuralism without a subject, over all sorts of functionalism to dualist structurationism and Giddens's dual structuration theory. The set of historical approaches is no less varied. And there is also the structural historical approach of, for example, the Annales School. It is therefore simply impossible to deal with all these kinds of investigations even in a superficial way.


Defining both approaches in this way raises what Abu-Lughod calls “the real methodological problem of historiography”: the part that knowledge of outcomes plays in the analysis of past events. “The usual approach”, she claims in her book on the progenitors of the modern world-system, “is to examine ex post factor the outcome – that is, the economic and political hegemony of the West in modern times – and then to reason backward, to rationalize why this supremacy had to be.”1 Since this conveys the impression that the rise of the West was an inherent historical necessity, she wants to avoid this approach by investigating a period in which the outcome was still undetermined.

Undoubtedly the investigation of a period in which the outcome is still undetermined is a perfectly legitimate strategy to avoid the pitfall of historical necessity. However, it is not the only possible strategy. The other strategy is to combine the historical with the structural approach. In this strategy the knowledge of the outcomes is not put aside, but used precisely in the way Abu-Lughod deplores. This implies that a structural approach does not entail the idea of historical necessity. Authors who oppose structural approaches for this very reason seem to commit at least one of the three following errors.

The first error is the error of confusing the notions of existential necessity and necessary existence. An example will clarify the difference between both. Profit making is essential to capitalism, capital cannot exist without profit making, and therefore when capitalism exists, profit making also exists of necessity. However, that does not mean that the existence of capitalism is necessary; it is merely a contingent fact.

The second error is the error of overlooking the fact that most explanatory research, including historical research, involves precisely reasoning backwards from outcomes. This is known as abduction or retroduction, and many philosophers of science take the view that abduction rather than induction or deduction, is the logic of science.

The third error is the mistake of equating structural causality with efficient causality. To put it in Aristotelian terms: the structural approach should reveal formal causes, the historical approach should reveal efficient and final causes. In present-day language: structure is causally effective; it constrains and enables, and it is about the why behind path dependency and equifinality. Agency, in turn, is causally efficient, whereby human agency is intentionally causally efficient.

My argumentation proceeds through a critical analysis of the replacement of Christallerian ideas about central places by Taylorian ideas about global cities. It is based on a continental background: the rejection of the so-called morphogenetic approach in the pre-positivist geography of the first half of the twentieth century, and the narrowing down of the field of application of central place theory. Taylor, on the contrary, has developed his ideas on the basis of a rejection of state centric and afterwards territorial thinking.


My starting point is the following quote from Bobek: “Um städtische Siedlungen hervorzurufen, ist die Zusammenhang, die Konzentration, die Brechung dieser Verkehrsfäden an gewissen Punkten nötig” (“The gathering, the concentration, the breaking of these threads of traffic in certain points is necessary to bring about urban settlements”.)3 In present-day language: “Urban settlements come into being as nodes in networks of flows of traffic.” In this article on the foundations of urban geography, Bobek reacted against what was then called the morphogenetic approach in urban geography. This approach consisted of four parts:

  • the description of the site or topographic location of the urban settlement,
  • the description of the situation or geographic location of the urban settlement,
  • the historical development of the urban settlement insofar as it is connected to the site and situation
  • the urban settlement as part of the broader landscape and as a landscape in itself.

Bobek argued that this approach only dealt with superficial characteristics of urban settlements. If geography wanted to say more interesting things about these settlements, it should focus on their function “als lebendigen Wirtschaftskörpers innerhalb des Wirtschafstgetriebes der Landschaft” (“as living economic bodies within the economic machinery of the landscape”)4. Later in this article, Bobek extended the notion of traffic to political and cultural communication.

The idea that underlies Bobek's functional line of thought can be summarized as follows. The division of labour within a reasonably advanced economy, that is, an economy that has advanced beyond self-sufficient households, leads to exchange and traffic flows. Initially (or in principle), these linkages interconnect all places involved. However, after a certain time or in reality), the resulting costs of transaction and transport are substantially reduced by creating nodes, i.e. urban settlements. These nodes or urban settlements arise or are built intentionally at favourable locations. In Bobek's line of reasoning a ‘favourable location' relates to the power to attract trade activities: “In dieser Ansichziehung aller früher ungeordneten Verkehrsfäden, welche sich so wie durch die Wirkung eines Magnetpols zu einem radialen Netze um die städtische Siedlung anordnen, scheint uns das geographisches Fassbare und Wesentliche an städtischer Wirtschaft zu liegen. Dem Namen Stadt können nur solche Orte verdienen, welche eine derartige beherrschende Stellung im wirtschaftlichen, kulturellen und politischen Verkehr eines Gebietes besitzen” (“It appears to us that the geographically explicable and essential aspects of the urban economy are to be found in this attraction of all formerly unordered threads of traffic that are organized into a radial network around the urban settlement as if the latter functions as a magnetic pole. Only those places that hold such a controlling position in economic, cultural and political communications deserve the name of city or town”.)5

Bobek's network of flows lacked a precise spatial structure and lacked agency. It was Christaller who provided the network with a precise structure (the famous hierarchical hexagonal structure), and it was Barton who injected agency into the network.

Christaller was aware of the article of Bobek. Bobek's view, he wrote in a footnote, largely coincides with mine. As a consequence, his theoretical approach equally starts with a reaction against the historical approach of explaining the distribution and development of urban settlements in terms of site and situation. However, Christaller's dismay over this approach takes a somewhat different angle. Christaller emphasized that a favourable site and situation were neither a sufficient nor a necessary condition for the location and development of urban settlements. There are, indeed, countless favourable topographical sites that have never hosted an urban settlement. And there are equally countless urban settlements located in favourable geographical situations that did not develop into large cities.

Christaller's alternative to the historical approach of site and situation is set out in his doctoral dissertation of 1933, which dealt with the distribution of the central places in Southern Germany. Christaller comes to the conclusion that most central places in Southern Grmany appear to be arranged according to the marketing principle. For this reason, Christaller chose to dub this principle the primary law of distribution. Frequent deviations from this arrangement were explained by two other principles, the traffic principle and the principle of separation respectively, whereby the latter is also often referred to as the administrative principle or principle of authority. And finally, there are also a number of tertiary deviations of a regional or local nature.

According to Christaller, the marketing principle represents the highest rationality, because it is the arrangement which ensures that the entire population is supplied with all central functions from the smallest possible number of central places. However, to optimize administration or traffic, or to consider special regional or local circumstances, it may be necessary to deviate from this arrangement. Deviations may therefore still benefit the public interest. However, the realization of an arrangement that benefits the public interest is met by effective resistance from the promotion of private interests and from irrational behaviour. In sum: there is a basic mechanism at work which in principle leads to a specific hexagonal distribution of central places, but which is in practice hampered because of interference from a series of secondary and tertiary mechanisms. When, where and how all these mechanisms interfere with the basic mechanism cannot be theorized, this has to be detected through appropriate empirical research.

From a purely methodological point of view, Christaller certainly succeeded in achieving his objective to do better than the historical or morphogenetic approach of site and situation. His theory meets the requirements of a scientific explanation. Science tries to explain facts, and a fact is scientifically explained when it can be established that it was possible at time T1 and place P1, but impossible at time T2 and place P1, and at time T1 and place P2. This definition of scientific explanation conveys the idea that a fact could not have happened in another way than it actually did, without being inevitable. It must be emphasized that an until then hardly applied method in geography lies at the origin of this methodological success of Christaller: the method of abstraction. To arrive at his theory, Christaller made abstraction of a number of possible urban functions such as military, manufacturing and harbour functions. As a result, his theory is not a theory of urban settlements, but rather a theory of of central places.

In a surprisingly rarely cited article on The creation of centrality, published in the Annals of the Association of American Geographers of 1978, Barton reacted against what she termed the received theory of central places. She rejected the neoclassical framework in which the theory was formulated, and proposed an alternative entrepreneurial/exchange framework based on classical economics. By using this framework she attempted to solve the problem of the creation of centrality. According to Barton, centrality arises from the activities of the entrepreneur, which can be a merchant, a trader, a retailer, a wholesaler, a banker. The entrepreneur mediates exchange, he makes it possible. In her case study of colonial New England, she arrived at the conclusion that three levels of centrality were discernable. At the lowest level, agricultural towns were occasionally able to supply surplus products. These are distributed through merchants either for the urban population or for export purposes. The degree of centrality achieved under these circumstances is both small and unstable. At the next level, places may achieve centrality by providing services specifically oriented to consumption by the mercantile community. Inn and taverns are examples of these types of functions. Other towns were promoted by the merchants for the provision of naval stores. Finally, towns in which the entrepreneurial sector is concentrated provide an excess of goods and services created exclusively by these agents. Functions related to the maintenance of the mercantile community, such as those geared to shipping, are typical for such towns. Furthermore, because the entrepreneur is the mediator between consumer and producer, access to information is a basic requirement of his activity. As a result newspapers, bookstores, and printers are concentrated in these towns of highest centrality. The final outcome is a two-layered landscape consisting of a set of agricultural towns with an unstable centrality status, and a system of central places powered by the entrepreneur.

As pointed out earlier, Bobek's network of flows lacked both a precise spatial structure and agency. Christaller provided the network with a precise structure (the famous hierarchical hexagonal structure), while Barton later injected agency into the network. But we should be very careful in simply combining the insights provided by these three authors.

Christaller does not deal with the creation of centrality, but with factors that turn a settlement into a central place of a certain importance. He thereby lists a number of establishments of which the presence in the settlement indicates its importance as central place of a certain level: institutions of administration, of commerce and finance, of transportation and communication, of cultural and religious importance, of importance in health and sanitation, of social importance, and so on. It may seem obvious to identify the initiators, founders, managers, directors, operators of these establishments as agents that create centrality. However, the hexagonal spatial structure is deduced from the distances that consumers are willing to travel to obtain the central functions. This implies that the theory considers central functions from the standpoint that they serve households. In a way, this is indeed the case. After all, schools provide pupils with education, while banks provide depositors with a safe-deposit box and a savings account. However, schools have a function in the reproduction of labour power. Banks enhance or accelerate the circulation of capital through consumer and producer financing. In other words, Christaller passes over the crucial role that central functions play in the (re)production of society. Barton, in contrast, starts from the role of the entrepreneur in the economic system. Settlements derive their centrality from the presence of the activities of these entrepreneurs, which are mainly business services, and from the presence of services that these entrepreneurs need.

This means that we are confronted here with two different kinds of centrality: centrality created by central functions that serve households (including consumer services, education, administration, and so on), and centrality created by business services. Following Taylor, we can conceive of the first kind of centrality as ‘town-ness', and of its creation as a process of town formation. The second kind of centrality can be envisaged as ‘city-ness', whereby its creation relates to a process of city-formation6.

This implies that Christaller's solely applies to the first kind of centrality, i.e. to town-ness. In this case it is (or was) possible to specify a mechanism that – if working without interference – would lead to a hierarchical hexagonal system of central places serving hinterlands, whereby the size of the hinterland depends on the nature of the central functions provided. The possibility of specifying a spatial structure is due to the fact that members of households are either not able or willing to travel beyond a certain distance to obtain the required central function. In the case of the second kind of centrality (the city-ness) it is impossible to specify a spatial structure in the same way. The distribution of business services depends on the motives of the entrepreneurs to locate their enterprise in a certain settlement. Distances to actual and potential harbours, centres of production, centres of consumption, mineral deposits, and so on enter the fray.

The morphogenetic approach is most certainly an example of a historical approach. Is the Christallerian approach a structural approach? A structural approach relates to the logic of the system. What is the system? At first sight, the answer may seem straightforward. The system is the network of central places and their hinterlands. But if this is the system, what is its logic?

We have seen that a distinction should be made between the spatial system on the one hand, and the system of agents whose behaviour results in that spatial system on the other hand. The system of agents constitutes a figuration, a term that is borrowed from the sociologist Elias7. Elias rejected the commonplace view on the relation between an individual and society. The commonplace view on this relation assumes that each individual person perceives him- or herself as being surrounded by institutions such as the household, the school, the town, the class, the state, and so on, which together make up society. Elias replaces this common sense view by the image of many people depending on each other and who, on the basis of their mutual dependencies, form figurations or structures of interdependence. These figurations are governed by rules of figuration, which are essentially rules that the people that are making up the figuration (i.e. the agents) have to observe if the figuration is to be maintained. Rules of figuration are generalized procedures of action. They formulate actions that, under specified conditions, must, should, or can usefully, legitimately, meaningfully, advisedly, and so on, be carried out. This makes it possible to give the following definition of both the logic of a system as well as its structure: the ‘logic of a system' is the functioning of the system according to the rules of figuration as long as the agents perfectly observe these rules. The term ‘structure of the system' refers either to the body of rules or to the relationships, distributions, balances which are the result of the application of the rules.

In the case of the Christallerian system of central places, the agents are the suppliers of central functions (services to households) and the consumers of central functions (the members of the households living in the hinterland). When these suppliers and consumers of central functions act according to certain rules of figuration, the result will be the spatial system with its typical arrangement of central places (which can be called the spatial structure of the system). The rules concerned are that suppliers locate either as closely as possible to each other or as far away as possible from each other, while consumers patronize the nearest central place. But these are, of course, not the only rules of figuration. The other rules pertain to those rules that need to be observed if the figuration of the agents themselves is to be maintained. And because the central functions are needed to reproduce society, we are referring here to the entire body of rules without which the society concerned would disintegrate.


It is said that we have entered a new era, the age of globalization. This has prompted Hall to redraw Christaller's urban hierarchy8. He proposes to drop the three lowest levels, because these settlements have ceased to perform a significant role as central places (at least in Europe). Furthermore, he suggests to add two new levels above the L-centres (the highest level in Southern Germany), i.e. ‘global cities' and ‘sub-global cities'. He identifies both categories by referring to the alpha global cities and beta/gamma global cities identified in a GaWC-classification of global cities. In my opinion, Hall's redrawing of Christaller's hierarchy is in contradiction with Christaller's theory, and it also runs counter Taylor 's conception of the world city network.

Hall does not mention that Christaller himself discerned two levels above the L-centres: RT-centres and R-centres. Examples of RT-centres are Hamburg, Cologne/Düsseldorf/Essen, Munich, Bordeaux, Lyon/Marseille, Milan, and Naples. Examples of R-centres are Berlin, Paris and Rome. Christaller called these R-centres Weltstädte, world cities. There can be little doubt that an exhaustive list of the R- en RT-centres would show a great similarity with Hall's list of global and sub-global cities. However, this is beside the point. This is because the subject matter of Christaller's concept of world cities is quite different from the content of the GaWC-concept of global cities. Christaller identifies Paris as a world city because it provides services to the households of a hinterland larger than that of RT- or L-centres. Theoretically, the size of the hinterland of an R-centre is about 300,000 km², while the size of the hinterland of a RT-centre is about 100,000 km² and that of an L-centre about 32,000 km². The GaWC-group, on the other hand, identifies Paris as an alpha global city because it provides advanced business services to firms in the whole world and shows a high degree of connectivity with other global cities. This interpretation of the GaWC-classification is based on Taylor 's conception of the world city network and its measurement by means of relational data9.

The nodes of the world city network are thus central places in a different way than in Christaller's central place theory. They do not serve hinterlands, but a hinterworld. The term hinterworld refers to the geographical nature of the global distribution of service connections which lay behind a city's global city formation. The city's functional reach is considered to be global, and the key to the spatial organization involved is the degree of connectivity, the level to which a city participates in advanced business service relationships between actors with a global radius of action.

The world city network considers connectivity rather than hierarchy to be the areal functional principle of the world economy. Yet it remains a network of central places. As a consequence, questions similar to those in central place theory arise. These are questions about the relation between central functions and economic development. How are central functions related to growth poles in every possible sense of the word, including the sense of poles of development and their spatial organization? In what way do the possible contributions of central functions to growth pole or development mechanisms depend on local, regional, central state and interstate intervention in the fields of law, general conditions of production involving infrastructure and public utilities, reproduction and regulation of the labour power, and protection of the level of capital accumulation?

In the case of the world city network the first of these questions has received a tentative answer. Taylor equates Sassen's concept of global cities with Jacobs's perception of dynamic cities10. As a result, business services occupy the key position in a development or growth pole mechanism. A dynamic city develops as the product of a virtuous circle of import replacing and export multipliers. The driving force behind this process is ‘new work', which occurs when firms in the city start producing – through innovative imitation – things that were formerly imported. This extraverted growth pole concept makes the development of a city dependent on the development of a network of cities, which is understandably more useful to Taylor 's theory construction in which business services perform the Jacobsian new work, than the more introverted classical growth pole theories. But neither the spatial organization of these mechanisms nor the role of state and interstate intervention is thereby clarified, although Jacobs does once again provide insightful ideas about the economic forces of expansion and the related processes which ripple out across the space economy far beyond the city-regions11. Nevertheless the key question remains: to what degree does the key position of business services entail a coordinating or even organizing role of these services in capital accumulation? The answer to this question requires a thorough investigation into the relationships between global cities and commodity chains.

Within the framework of a confrontation of structural and historical approaches, the question of the key position of business services in capital accumulation raises an even more fundamental question of the legitimacy of transhistorical notions. The question of the key position of business services is indeed linked to the borrowing by Taylor of Jacobsian transhistorical ideas. It makes him conceive of cities as transhistorical entities. Is this a legitimate conception? Or is it a variant of transubstantiation?


Transubstantiation is the change of the substance of bread and wine into the body and blood of Christ during the Eucharist. According to the teaching of some Christian Churches, including the Roman Catholic Church, during the Eucharist bread and wine are transformed into Jesus Christ's body and blood. In the process they take a higher, more sublime form. In a review of the development of geographical thought, Kesteloot & Kesteloot have used this term to characterize one of the procedures that geographers have applied to neutralize the importance of social classes for their research. Three procedures can be discerned:

  • abstraction: the existence of social classes is simply ignored, and Man therefore appears in the shape of Economic Man;
  • banalization: social classes have no particular role in the explanation of geographical facts;
  • transubstantiation: the importance of social classes pales before the fact that the cultural differentiation of humankind is the continuation of biological evolution, and Man therefore appears in the shape of Wagnerian Ecological Man. Wagner was a rival of Darwin, who developed his own theory of evolution, emphasizing the selective role of physical environment. Basically transubstantiation consists in applying the principles of natural history to human history.12

Of course, the question whether or not Taylor 's conception of cities as transhistorical entities is a variant of transubstantiation does not entail the suggestion that Taylor falls back on this Wagnerian Ecological Man of environmentalists such as Ratzel. Rather, the question refers to a Jacobsian Innovative City Man.

Two citations from the English translation of Pirenne's Les Villes du Moyen Age provide a good starting point to develop the argumentation13. Taylor places Pirenne at the beginning of the lineage of contemporary students of inter-city relations. The reason is that Pirenne was the first important scholar to create a city-based view of European history with inter-city relations as its formative force14.

The first citation: “History is obliged to recognize that, however brilliant it seems in other respects, the cycle of Charlemagne, considered from an economic viewpoint, is a cycle of regression … The ninth century is the golden age of what we have called the closed domestic economy and which we might call, with more exactitude, the economy of no markets … The closed demesnial organization, which made its appearance at the beginning of the ninth century, was … an abnormal phenomenon … [T]he period which opened with the Carolingian era knew cities neither in the social sense, nor in the economic sense, nor in the legal sense of the word.”15

The second citation: “[T]he ‘abbey-merchants' were … not free agents, but employees exclusively in the service of their masters. It is not apparent that any of them ever carried on business on his own account.”16

Taylor refers to the first passage in his outline of the lineage of research on inter-city relations: Charlemagne represents regression, not revival. The second citation raises the problem of transubstantiation.

According to Pirenne, the necessary and sufficient conditions of progress are the development of commerce by a strong independent merchant class and the creation of cities with a strong middle class. These are in fact two sides of the same coin. Pirenne deals with European history. Jacobs deals with the history of humankind17. In her view, the necessary and sufficient condition of progress is the development of a network of trading cities. In other words: human history is a history of economic parentage, of dynamic cities within city networks, of disintegration of some sub-networks and the decline of the cities involved, of which the role is taken over by other cities creating new networks.

In this view, the development of the forces of production (what Marxists think to be the driving force of human development) is:

  • not alternately promoted and hindered by the relations of production (as Marxists say);
  • not caused by the development of Man's creative capacities through his exploitation of the physical environment (as, for instance, environmentalist such as Ratzel say);
  • not caused by applying best practices of allocation (as is implied in the concept of Economic Man);

but rather caused by the victory of the innovators in a permanent economic conflict between innovators and conservatives. This is what the concept of Jacobsian Innovative City Man essentially refers to.

Note the difference between Pirenne and Jacobs. Pirenne brings up cities with a strong middle class and independent merchants as opposed to abbey-merchants. According to Jacobs, however, merchants do not have to be independent. However, we know that in societies based on the tributary mode of production, mercantile activity cannot be granted too much room for manoeuvring. Commerce may develop, but when it grows too strong, it will be curtailed in a number of ways. Jacobs needs to transcend the differences in the status of merchants in order to be able her story of economic parentage. As can be deduced from her book The nature of economies, she does this by transubstantiating: “The theme running throughout this exposition – indeed, the basic premise on which the book is constructed – is that human beings exist wholly within nature as part of natural order in every respect … In describing natural processes and selecting examples to illustrate them, I have jewed to information from the fields of biology, evolutionary theory, ecology, geology, meteorology, and other natural sciences as the information is cuurently understood and interpreted by practitioners in these sciences.”18. Nonetheless, the use of transhistorical concepts can still be justified in spite of this.

The reason for this is that transhistorical concepts may also be the result of abstraction. In one sense, abstraction is one of the procedures to neutralize the importance of social classes. The existence of social classes is ignored, and Man takes the form of Economic Man. But in another sense it is the basis of the methodological success of Christaller. This is because there are two types of abstraction. The first type consists of creating a fiction; the second type consists of focusing on particular features of the study object. When a researcher takes the latter approach, he/she still deals with real things. For instance, when Christaller makes abstraction of military and industrial functions in order to be able to focus on central functions, he still deals with reality. A town does supply central functions. Economic Man, on the contrary, is a fiction. One does not neglect some characteristics of real people to keep some other features as focal point. Rather, one invents a being that simply cannot be found in reality.

Let us take the example of Karl Marx. In order to define the capitalist mode of production, Marx abstracts from those qualities that belong to our society as part of other systems. These other systems are manifold, and are defined at other spatiotemporal levels: e.g. human society as a whole (which takes in the whole history of our species), class society (which takes in the entire period of class history), modern capitalist society (which only takes the most recent stage of capitalism), and the unique society that exists at this time and at this place (which only takes in what is here and now)19. In this definition of the capitalist mode of production by means of the method of abstraction, Marx did not follow the Aristotelian scheme of genus/species/individual. He did not try to define the genus in a first phase in order to be able to continue with the deduction of the species. Rather, he starts from the present and draws in the past in order to be able to better understand the present on the basis of the actual differences between present and past. Marx thereby looks for the things that make the capitalist mode of production essentially different from the feudal mode of production, and comes to the conclusion that, say, productive labour is geared to exchange value rather than use value, and that ground rent plays a subordinate role rather than a dominant role. A general theory about productive labour or about ground rent cannot be derived from this. However, what can be derived is the finding that, within the capitalist mode of production, labour geared to exchange value and subordination of ground rent are internally related to each other.20 This means that relations rather ‘things' are the fundamental materials of reality. Human beings do not exist as such, they only exist at a certain spatiotemporal level of abstraction, as slaves and masters, serfs and lords, wage labourers and capitalists, and so on. At another spatiotemporal level they exist as direct producers and appropriators of surplus. At still another spatiotemporal level they exist as parents and children, and so on. It is exactly this type of abstraction that may lead to transhistorical concepts that are not hampered by the problems associated with transsubstantiation.

Does Taylor use this method of abstraction, in spite of his references to Jacobs? Taylor's chapter in Van Christaller tot Wallerstein21, and especially his chapter in Cities in globalization may lead to such an interpretation. In a first move, he deals with processes of city and town formation operating at the spatiotemporal level of the entire period since the first urban revolution 9.000 years ago. In a second move, he deals with processes operating at the spatiotemporal level of modern times. As a consequence, the concept of cities as transhistorical entities is a legitimate concept – it makes sense22 Whether it constitutes a valid concept is, of course, another matter. This depends on the explanatory power of the concepts of moral syndromes, borrowed from Jacobs's Systems of survival23. If these syndromes, the commercial and the guardian syndromes, can be shown to be rules of figuration, their use is a sign of the structural approach. Pirenne, on the other hand, applies the historical approach, because he takes a same causes/same effects approach24.


There are two strategies to avoid the pitfall of historical necessity. The first strategy consists of investigating a period in which the outcome was still undetermined, and has been adopted by Abu-Lughod. The second strategy consists of comibing of historical and structural approaches. However, it goes without saying that such a combination is much more than a simple juxtaposition. The nature of this combination needs to be explained.

Let us begin by returning to the world city network. Taylor calls it an interlocking network because it consists of three levels. Normally networks are specified at two levels. The first level defines the scope of the relations, and is the level of the network itself. The second level is the nodal level that defines the agents whose relations constitute the network. The world city network is different, because it has three levels:

  • firstly, the world-economy as the network level were services are dispensed;
  • secondly, the cities as the nodal level, constituting the knowledge constellations for production of services; and
  • thirdly, the sub-nodal level, the advanced producer service firms that create and provide the services.

Such a triple configuration is called an interlocking network.

The cities at the nodal level and the firms at the sub-nodal level are necessary to the formation and reproduction of the network. Cities as local networks of political and economic institutions and business service firms exporting services beyond the local market are two key agencies. Other key agencies are the multiplicity of supervisory institutions that oversee the practice of individuals and firms within particular service sectors and the nation-states. Or, more accurately: the state apparatus, especially that involving economic policy, and general national culture of conducting business and its relation to national society. This implies that, taken together, service firms, city governments, service-sector institutions and nation-states are the key responsibles in shaping the world city network. Two causal nexuses and four identity assignments constitute their mutual interaction. The causal nexuses are the interactions between firms and cities on the one hand, and between the sectors and nation-states on the other hand. They are essentially two-way connections that are necessary for both agencies to reproduce themselves. Firms and cities need each other, just as sectors and nation-states need each other. Identity assignments are one-way relations in which one agency moulds the character of another agency. This is the case for sectors and firms, states and cities, sectors and cities, states and firms.25

This constellation of four agencies, two causal nexuses and four identity assignments can be interpreted as a complex of internally related elements, defined at the appropriate level of spatiotemporal abstraction. This level is the first level of the interlocking network, the level of the network itself: the present-day world-economy, Wallerstein's modern world-system in the ongoing phase of intensified globalization. It is a system consisting of agents with their specific logic, the functioning of the network according to the rules of the causal nexuses, and identity assignments when the four agents act completely according to these rules. This implies that the structural approach determines the unit of analysis. The structural approach determines the causal chains that have to be investigated by the historical approach.

The bearing of this conclusion can be illustrated by its implications for the study of the so-called world system in the period 1250-1350 by Abu-Lughod. Since Abu-Lughod focuses on cities rather than countries, her study takes the form of a study of a network of cities. However, because of the low degree of integration, Abu-Lughod chooses to call it an archipelago of towns, or an archipelago of cities in the terminology of Taylor. Furthermore, this is not a study of a ‘real' world-system, but rather a study of a system in formation. The system-in-formation is Eurasiatic, and consists of eight interlinked subsystems. The latter can be grouped into three larger circuits – the western European, the Middle Eastern, and the Far Eastern. In contrast with the modern world-system, the archipelago as a whole does not exhibit an overarching hierarchy; there is no encompassing core-semiperiphery-periphery structure, while a number of coexisting core powers exist instead of one hegemonic power. The subsystems were more self-sufficient than they are today and therefore less vitally dependent upon one another for common survival. The transformation into an integrated world-system, in turn, has not been realised because the subsystems followed their own development path to the degree that the trade network failed to become an integrative force, after which the system ultimately fragmented.26 Translated in structural terms: the creation of a world-system failed because the systems on the verge of being integrated ultimately followed their own logic. They did so to such a degree that the trade network was not able to develop a logic of its own. This implies that in the end there was no system to speak of, and it therefore makes no sense to speak of subsystems. From a structural viewpoint, this implies that the archipelago of cities cannot be a unit of analysis, and it would therefore make no sense to try to investigate the relation between cities and chains of commodities (at least not in the sense that this is possible with regard to the world city network).

This is not to say that the choice of the archipelago of cities in the thirteenth century as study object is either a wrong or an untenable choice. On the contrary, it is a perfectly legitimate choice within the framework of developing a methodological strategy to avoid the pitfall of historical necessity. However, it is completely misleading to use terms such as ‘world system' in the sense that Wallerstein uses these kinds of terms27. Abu-Lughod herself notes the following about the term ‘world system': “The term ‘world system', as it is currently used, has unfortunately been conflated with the particular hierarchical structure of organization that developed from the sixteenth century onward. This makes debates over world systems less than fruitful. It is important to remember that a system is simply ‘a whole composed of parts in orderly arrangement according to some scheme' (Oxford Dictionary)28. This justification of her use of the term ‘world-system' is, in my opinion, utterly unsatisfactory. The issue is not whether a world-system must be similar to the modern world-system. Rather, the issue is whetherthere is such a thing as complexes of internal relations between agents exhibiting a logic of their own. To speak of a world-system is only justifiable when it is has become certain that its logic will carry on because the factors that can prevent it no longer operate or are made ineffective. In the case of the modern world-system this means the sixteenth century, which – as Wallerstein argues – witnessed the failure of the empire builders. Of course, it is only through a retrospective analysis that the dating of these turning points in history can be achieved. As a consequence, you do need the knowledge of the outcomes.



* Pieter Saey, Department of Geography, Ghent University.

1. Abu-Lughod J.L., Before European Hegemony. The World System A.D. 1250-1350, New York/Oxford, 1989, p. 12.

2. Bobek H., Grundfragen der Stadtgeographie, Geographischer Anzeiger, 28 (7), 1927, pp. 213-224; Christaller W., Die Zentralen Orte in Süddeutschland, Jena (translated by C.W.Baskin, Central Places in Southern Germany, Prentice-Hall, 1966); Barton B., The creation of centrality, Annals of the Association of American Geographers, 68, 1978, pp. 34-44.

3. Bobek H., op.cit., p. 216.

4. Idem, p. 214.

5. Idem, p. 216.

6. Taylor P., Cities within spaces of flows, in: P.J. Taylor, B. Derudder, P. Saey and F. Witlox (eds), Cities in Globalization. Practices, Policies and Theories, London/New York, pp. 287-297.

7. Elias N., Wat is sociologie? Utrecht/Antwerpen, 1976.

8. Hall P., Christaller for a global age: redrawing the urban hierarchy, in: A. Mayr, M. Meurer and J. Vogt (eds), Stadt und Region: Dynamik von Lebenswelten, Leipzig, 2002, pp. 110-128.

9. Taylor P.J., World City Network. A Global Urban Analysis, London/New York, 2004.

10. Jacobs J., The Economy of Cities, New York, 1969; Sassen S., The Global City, Princeton, 2001.

11. Jacobs J., Cities and the Wealth of Nations. Principles of Economic Life, New York, 1984.

12. Kesteloot C. and Saey P., La géographie classique et la neutralisation du role des classes sociales dans l'explication des faits géographiques, L'Espace Géograpique, 1986, pp. 222-230.

13. Pirenne H., Medieval Cities. Their Origins and the Revival of Trade, Princeton, 1980 (translated from the French 1925 by F.D. Halsey).

14. Taylor P.J., Prologue. A lineage for contemporary inter-city studies, in: Taylor et al. (note 6), pp. 1-12.

15. Pirenne, op.cit., p. 40, 46, 47, 75.

16. Idem, p. 109.

17. Jacobs, op.cit. (note 10).

18. Jacobs J., The Nature of Economies, New York, 2000, pp. ix-x.

19. Ollman B., Dance of the Dialectic. Steps in Marx's Method, Urbana/Chicago, 2003, p. 162.

20. Van Dijk J., Westers marxisme als sociale wetenschap. Object, methode en praktijk van een onderzoekstraditie, Nijmegen, 1984, p. 52.

21. Taylor P., Cities and states: an elemental reinterpretation of their orgins and relations, in: Van Nuffel N. (red.), Van Christaller tot Wallerstein. Liber Amicorum Prof. Dr. Pieter Saey, 2007, pp. 119-130.

22. Taylor P., op.cit. (note 6).

23. Jacobs J., Systems of Survival. A Dialogue on the Moral Foundations of Commerce and Politics, New York, 1992.

24. Pirenne, op.cit., p. 54.

25. Taylor, op.cit. (note 9), 56-61.

26. Abu-Lughod, op.cit., pp. 32-38.

27. Wallerstein I., The Modern World-System I. Capitalist Agriculture and the Origins of the European World-Economy in the Sixteenth Century, New York/San Francisco/London, 1974; Wallerstein I., Historical Capitalism with Capitalist Civilization, London/New York, 1995; Wallerstein I., World-Systems Analysis. An Introduction, Durham/London, 2004.

28. Idem, p. 38.


Edited and posted on the web on 3rd July 2008