GaWC Research Bulletin 149

This Research Bulletin has been published in Political Geography, 24 (6), (2005), 703-730.

doi:10.1016/j.polgeo.2005.01.009

Please refer to the published version when quoting the paper.

(Z)

New Political Geographies: Global Civil Society and Global Governance through World City Networks

P.J. Taylor

Abstract

This study is an investigation of network processes associated with globalization to see whether they herald ontological, new, political geographies that challenge the reproduction of modern states through Westphalian processes. Using an interlocking city network model, three world city networks are defined: an inter -state city network with state departments as network-makers, a supra -state city network with UN agencies as network-makers, and a trans -state city network with NGOs as network-makers. These networks are interpreted as representing worldwide spaces of flows for Westphalian, global governance, and new global civil society processes respectively. The inter-state city network is treated as a benchmark against which to compare the two newer global networks. Activity matrices describing the geographies of the network-makers are analysed and principle components interpreted as common global locational strategies. It is found that the three city networks exhibit very different spatialities: the inter-state city network is primarily intra-regional and horizontal; the supra-state network is primarily inter-regional and hierarchical; and the trans-state city network is primarily inter-zonal (straddling the ‘North-South divide'). These spatialities are discussed in terms of what they mean for contemporary political geographies.

 

INTRODUCTION: THE SPATIALITIES OF NEW POLITICAL GEOGRAPHIES

Taking the long perspective of the modern world-system, new political geographies are not unusual. Defining political geography as the spatial distribution of the deployment of political power, the focus of study has been the territorial state and therefore new political geographies have arisen with every geopolitical transition. For instance, following the collapse Soviet power in 1989-91, a new, post-Cold War political geography emerged. There are important differences of interpretation of this new pattern of power deployment but nobody doubts there is a new political geography. But this is not the only form that new political geographies take. Taking an even longer ‘trans-system' perspective, ‘deeper' new political geographies can be defined in terms of the changing spatialities of deploying political power. For instance, the transition from the complex medieval European distribution of power to the ‘Westphalia state system' of territorial sovereign states is such changed political geography. I call the process ‘deeper' because new spatialities imply a changing ontology of political power. It is such new political geographies that are my subject matter here.

The initial reaction to the end of the Cold War was to proclaim a ‘new world order' in which the ‘soft power' of the USA would be deployed to convert erstwhile ‘totalitarian states' into new ‘liberal democratic states'. Note the continued assumption that states are the prime units of power. The violent breakdown of this territorial ‘order' into a ‘new world disorder' in the 1990s re-emphasized the potency of states as defining what is the political on the world stage. But there was a second equally familiar scripting of the 1990s: globalization. The end of the Cold War saw the demise of the ‘second world' and thus problematized the concept of a ‘third world'. In addition, the rise of Pacific Asia in the 1980s meant that the old ‘North Atlantic' core region was being economically challenged for the first time in the history of the modern world-system. Supported by ‘one world' rhetorics, globalization appeared as a ‘natural' progression, a new geographical scale of human activities. As such it implied the demise of the state as the prime unit of spatial organization. In other words, globalization heralds a different world spatiality to replace the Westphalia mosaic. This argument implies creation of new political geographies in the deeper sense, which is what I explore in this paper.

I have been developing my thoughts on this topic for over a decade and this paper builds upon and synthesizes two earlier papers, previous excursions into spatialities. In the first of the earlier papers I introduced the concept of interstateness – the multiplicity of modern states - in an attempt to emphasize the intrinsic relational basis of the inter-state system (Taylor 1995). This was a necessary precursor to understanding how to identify the demise of territorial states and their replacement by a different spatial organization. In the process I identified the conundrum that, in an ever-changing modern world, how can changes in states as adaptions to new circumstances -‘ordinary' change – be distinguished from signs of their dissolution – ‘extraordinary' change? I offered the balance between inter-state processes (operating between states) and trans-state processes (operating across or beyond states) as a solution. The former reproduce the Westphalian mosaic, the latter undermine it, so that a strong relative rise in the latter can be interpreted as more than simple adaption. In this paper I reinterpret this argument as the spatiality-ontology position presented above. Through this rethinking I find three processes, with supra-state processes (operating above states) joining the original inter-state and trans-state processes. In the globalization thesis it is supra-state and trans-state processes that challenge the ‘Westphalian norm' of inter-state processes. Unlike for the earlier paper, I am now in a position to provide some empirical evidence that describes contemporary operations of these three processes.

I first introduced this new empirical work in the second earlier paper I build upon that was concerned with ideas on relations between world cities and territorial states (Taylor 2000). Here I identified the spatiality as explicit: states define a political space of places, cities represent an economic space of flows. Thus, these alternative spaces were originally described within different spheres of social activity, the political versus the economic. But it does not have to be that way. Although most the evidence harnessed to support the thesis of globalization has been economic, there are now sizeable literatures that describe ‘global governance' and ‘global civil society' to set alongside the more numerous writings on the ‘global economy'. In this paper I take the two former concepts seriously and consider their spatiality as city-centred spaces of flows.

The synthesis that is my current position can therefore be stated as follows. I research the supra-state and trans-state processes as world city networks that operate both above and across or beyond states. It is through these networks that states are being undermined and this extends beyond the economic to encompass the political. Treating city networks as process is central to Castells' (1996) seminal argument for the rise of ‘network society'. Although he initially implied that spaces of flows were transcending spaces of places, he now considers these two space-forms to be inter-related (Castells 2000) and I strongly hold to this position for any spatiality. For instance, the world political map that describes the state mosaic is only the outcome of Westphalian processes, and those processes actually operate as spaces of flows. For instance, interstateness is produced and reproduced through the diplomatic circles generated in and through capital cities. Thus, I no longer emphasize the simple spaces of flows versus spaces of places argument that was the basis of my earlier treatment of cities and states (Taylor 2000). In this paper I investigate globalization as a world city network making process in which the balance between inter-state activities and supra-state plus trans-state activities is appreciably tilting away from the former.

The paper is primarily empirical: I present new analyses that inform contemporary debates on the global. This emphasis is a simple necessity because very little is known about inter-city relations within globalization, especially for non-economic processes. Clearly it is not possible to debate global governance and global civil society through city-centred lenses before there is some evidential basis upon which to develop and evaluate ideas. My previous empirical research on world cities has described the organization of the global economy through a world city network defined by the activities of major financial and business service firms (Taylor 2004a). But world cities are much, much more than ‘global service centres'. For instance, most of them are capital cities, clearly implicated in the Westphalian processes of interstateness. But these great cities are also more than political centres. They are important cultural centres and they are the sites where global social practices are emerging. In other words, they can be interpreted as the organizational nodes of global governance and global civil society as well as of global economy (Taylor 2004c). This is the line of argument followed here.

There are five substantive parts to the paper. First, a model is introduced that provides a rigorous way of specifying a city-centred view of the world. This ‘interlocking' model was developed to describe how global service firms have created a world city network for the global economy (Taylor 2001); here I adapt it to other, non-economic, institutions. The key process is city network formation and the most important step is to identify the agents, the city network-makers. Using this model, I replace service firms as city network makers by the departments of states that conduct foreign affairs, United Nations agencies, and non-governmental organizations (NGOs). These three agents produce three very different types of network that are described in turn as an inter -state city network, a supra -state city network, and a trans -state city network. The first is a network of flows between states: it represents the contemporary operation of the Westphalia process through cities. This is employed as a baseline against which alternative political use of contemporary cities are evaluated. The second is a network of flows above states: it represents an interpretation of globalization processes as an increasing geographical scale of operation. The third is a network of flows across and beyond states: it represents an interpretation of globalization processes as transcending states and their boundaries. Each case is described in detail in terms of its global configurations. The findings are then compared in a discussion that highlights how the spatialities of global governance and global civil society diverge from Westphalian spatiality. I end with a discussion on the meaning of the findings.

A CITY-CENTRED VIEW OF THE WORLD: THE INTERLOCKING MODEL

Cities do not create city networks. City governments may and do build, or support the building of, infrastructure networks (airports, smart buildings) but these do not constitute city networks: it is the people, commodities and information that flow through the infrastructures that define the inter-city networks. I am interested in the flows that reproduce institutions, flows that are part and parcel of their everyday business. In their generation of flows, such institutions produce inter-city networks. Previous work has concentrated on advanced producer services firms (e.g. in law and advertising) as the city network-makers (Taylor 2001, 2004a). However, network formation is by no means limited to this economic activity: any organized activity that requires working in and through several cities is part of world network formation. Thus, in what follows I draw directly from the ideas of a world city network produced through service firms (Taylor 2001) but extrapolate the argument to network-making institutions in general. This will allow for subsequent use of non-economic actors within a model that has proven very fruitful for analysing the world city network as the skeletal structure of the global economy (Taylor 2004a). The conjecture is that the model and methods employed for the latter will also describe the skeletal structures of global governance and global civil society.

Model

The network model I use is an interlocking model. This defines an unusual network that operates at three levels. Most networks consist of just the net level that is the outcome of processes operating through the nodal level: the nodes are the agents, the network-makers. Thus in a study of intra-gang social relations, the agents are the individual members of the gang that create the network that is the gang. However, I have already argued that even though cities are the nodes of the network they are not the network-makers. Thus there has to be a third level in the network, the sub-nodal, that consists of the institutions that create the world city network. It is called an interlocking network because the institutions ‘interlock' cities in the everyday conduct of their activities. Hence law partnerships preparing inter-jurisdictional contracts and advertising agencies devising global campaigns interlock cities through the flows of information, knowledge and personnel between project-relevant cities, but so do foreign-policy departments and embassies in their diplomatic negotiations, UN agencies in their fulfilment of missions, and NGOs in their prosecution of projects. All these institutions use cities to do their work and in the process interlock those cities through network-making practice.

In an ideal statistical world researchers would have access to counts of flows within important institutions. But these are never available: in the private sector they are commercial secrets, in the public sector they are state secrets, and elsewhere there is no incentive to collect the information within institutions that have more urgent things to accomplish. Fortunately, there is a relatively easy way around this problem. It is usually possible to find reliable information on the office networks of these institutions. Certainly the cities in which an institution has an office can be found. Beyond this presence/absence data, information is often available on an institution's activities within a city such as the size of an office or its special functions (e.g. headquarters). However, the specific information available for an institution's offices is usually unique to that institution: there are different types and degrees of information across institutions. This problem is overcome by devising simple coding of information, institution by institution, to produce comparable data. This is described in detail for advanced producer service firms in Taylor et al. (2002a) where the coding of cities ranges from 5 for housing a firm's headquarters through to 0 for no office in a city. These are the activity values for a given institution in a given city. The basic assumption is that the larger the activity value the larger the flows of intra-institution information generated. Further, this can be used to estimate flow-levels between cities: two cities with large activity values for the same institutions will have more inter-city flows than two cities sharing few activity values of the same institutions. From these assumptions measures of interlock connectivity can be derived (Taylor 2001, 2004a) and these are reported for the non-economic data used below in separate papers (Taylor 2004b and c). In this paper, I focus upon the configurations of the world city networks.

Analysis

If information to produce activity values is collected from n institutions across m cities, these can be arrayed as an n x m activity matrix where columns are institutions and rows are cities. These data provide the basic input to any interlock network analysis. The matrix is easy to interpret. Looking at it vertically, every column of activity values represents the locational strategy of a given institution across the cities. Looking at it horizontally, every row of activity values indicates the mix of institutions present and their relative strengths in the given city. This is the form of data matrix upon which all subsequent analyses are based.

Activity data matrices can be analysed in many ways. The most straightforward, and one that has proven fruitful with service activity matrices (Taylor 2004a and Taylor et al . 2004), is the basic R-mode principal components analysis with varimax rotation, which produces particularly precise results (Rummel 1970). It is a data reduction technique that searches out the basic dimensions contained in large data sets. The general idea is to reduce the n variables to k components where k is considerably less than n, and yet a large proportion of the original variability in the data is retained. The components are derived from correlations between variables and are combinations of parts of them. Every combination of variables is different and independent, and the components are described in terms of the variables that contribute most to their constitution. In the analyses reported below it is the institutions' locational strategies that are the variables so that the components are interpreted as independent ‘common' locational strategies of a combination of institutions.

Each principal components analysis provides three key pieces of information through which to interpret results: the component loadings on variables (institutions), the component scores on objects (cities), and a measure of the importance of each component.

1. There are sets of loadings showing the levels of correlation between each original variable and each component. Variables with high loadings are the clusters of variables that constitute the ‘composite variable', which is the component. In the analyses below I use a threshold of 0.4 to denote ‘high' loadings. It is clusters of institutions with such high loadings that are used to interpret the components as common locational strategies.
2. There are sets of scores showing how each composite variable (component) is distributed across the m objects in the matrix (cities). Scores are given as standardised values, which means they have a mean of zero and most values lie between minus two and plus two. The pattern of cities with high scores on a component describes the composite locational strategy; this is interpreted as a subnet of cities within the overall network of cities.
3. The importance of a component is given by the proportion of the overall variation in the data (the variance of all the variables) that it incorporates – it is usually referred to as ‘variance accounted for'. The sum of these proportions is the amount of variation retained through reduction from n variables to k components.

A key decision in principal components analysis is the choice of number of components. There is no generally accepted way to make this choice; in the analyses below I have followed an experimental method to find particularly stable components - ones that exist through many choices of number of components (Taylor et al . 2002b).

Data Sets

Operationalizing the interlocking network model for analysis requires construction of activity matrices linking institutions to cities. As previously noted, such data were produced initially for advanced producer service firms (Taylor et al . 2002a). The key goal of that exercise was to produce reasonably robust data relatively quickly from publicly available information. The end result was a 100 firms x 315 cities services activity matrix of activity scores ranging from 5 (headquarter city) to 0 (no presence of firm in city). Such data matrix construction has been replicated successfully for other world city network-makers: media firms (Kratke and Taylor 2004) and architectural firms (Knox and Taylor 2004). I continue this replication here using government foreign affairs departments representing institutions making inter-state city networks, UN agencies making supra-state city networks, and NGOs making trans-state city networks. Details of the three resulting matrices are given in Table 1. Note that two matrices are described for each process. Initial matrices indicate all the data collected, these were abridged to decrease the sparseness of the matrices by eliminating both network-makers and cities with less than five presences. It is this second adapted matrix that is used for analysis.

The activity matrix for foreign affairs (diplomatic) relations was derived from the web-based worldwide Embassy database (www.embassyworld.com). The site was accessed in May 2004. All diplomatic presences were recorded; as well as embassies and consulates, diplomatic missions, tourist bureaus, and trade offices were also included. In this case the coding was quite simple because in diplomatic networks the capital cities are overwhelmingly the senders and receivers of information. Thus for each state foreign affairs office, its capital city location is scored a maximum 5 and all other offices are scored the minimum presence of 1. After culling the matrix to reduce sparseness, the result is a matrix of activity values scoring 0, 1 and 5 for 114 foreign affairs departments across 170 cities (Table 1).

The activity matrix for UN agencies was derived from the UN website (www. un.org). The site was accessed in October 2003. Taking each UN institution separately, all locations where it operated were identified and presences were coded at three levels: headquarter cities were scored 3, 'regional or 'liason' offices were scored 2, and 'field', 'branch', 'area', 'national' office, 'advisory' team locations and 'representatives' were scored 1. With this information there were a lot of places with just one presence (e.g. field offices) and therefore the cull from initial matrix to the matrix for analysis was quite severe resulting in just 34 institutions operating across 92 cities (Table 1).

The activity matrix for NGOs was derived from the 2001-2002 Yearbook of Inter-state Organizations . This source provides a very wide range of institutions and since I am ultimately concerned with processes of global civil society, four categories were chosen for study: Environment, Development (including subcategories), Human Rights and Humanitarian. To ensure the focus was on global-scale activities, only NGOs with offices on at least three continents were selected. 86 were found to fit the criterion and detailed information on their offices was obtained from individual NGO websites. Using the scale of an office's responsibility as criterion, inter-state headquarter offices were coded 4, regional offices 3, state offices 2, and sub-state offices 1. After culling, the matrix for analysis includes the offices of 58 NGOs over 100 cities (Table 1).

The methodology I use relies on large amounts of simple information to provide basic estimates of the uses of cities by multi-city organizations. The data collected originally encompassed over 100,000 pieces of information, and even after matrix sparseness is reduced, there are still over 28,000 pieces of information that are analysed below (Table 1). It is this huge quantity of data that necessitates the use of a data reduction technique such as principal components analysis. Application to the three data matrices has resulted in the 114 state diplomatic locational strategies being reduced to 12 common strategies, the 34 UN agency strategies to 6 common strategies, and the 58 NGO strategies to 10 common strategies (Table 2). These account for just over a third of the variability in the first matrix, and just over one half for the other two matrices. Thus although the data reduction has produced different numbers of subnets of cities, there remains a big amount of variance unaccounted for. This is interpreted as all three matrices incorporating large quantities of variability unique to individual network-makers. Thus, Table 2 suggests that state diplomatic strategies in particular have a surfeit of idiosyncratic patterns. This makes sense because these network-makers are territorial representatives who might be expected have more ‘local' geographical links than UN agencies and global NGOs. This should become apparent in interpretation of the various components where particular spatial patterns are discussed. Through these interpretations three specific dimensions of spatiality are revealed: local regional subnets versus inter-regional subnets, core zone city subnets versus core/non-core subnets, and primate hierarchical subnets versus ‘horizontal' network subnets. Each data analysis tends to emphasize one each of these spatiality dimensions.

WESTPHALIAN POLITICAL PRACTICE THROUGH AN INTER-STATE WORLD CITY CONFIGURATION: DIPLOMATIC SUBNETS OF CITIES

The city network created through state apparatuses is used as a yardstick against which to compare the other two city networks. It can be thought of as the contemporary inter-city expression of Westphalian political organization of sovereign territorial states. Such ‘diplomatic circles' have kept this spatial organization operating through the modern era and certainly continue to reproduce it in our times. But this is the political arrangement being confronted by globalization; if the latter is a source of serious challenges to this political order then there will be different inter-city patterns being created by other political network-makers. Obviously different identification of new political geographies requires a benchmark for comparison, hence this initial analysis of the inter-state city network.

The inter-state world city network privileges one class of cities: capital cities. In world city terms, these are the political ‘command and control centres': the headquarters of the network-making institutions, the foreign affairs arm of state governments. This is recognised loosely in our inter-state relations language when it is said that "Washington is doing this" or "Paris is supporting that". But this is rarely developed as indicating an inter-city space of flows. It might be expected that such research would be found in political geography but this not the case since studies of capital cities are currently unfashionable. Hall (1993) has provided a useful comparative study leading to a classification of capital cities but this does not deal with the network-making done in capital cities. Van der Wusten (2003) does broach inter-city relations through his concern for centrality and decentralization in European politics and I follow his lead here. Using the foreign affairs activity matrix provides another step towards developing this neglected research field.

There is, of course, a research tradition for quantitative analyses of the spatial distribution of embassies; Nierup (1994) provides the most through example in recent political geography. But such studies always conceptualise the relations described as between states. Of course, there are very good reasons for describing the Westphalia network in this manner, but it is not the only way. I have chosen to provide a city-centred description of these relations to facilitate direct comparison with the supra-state and trans-state inter-city networks but this is not just a pragmatic research decision. As emphasized previously, cities are where the work of political network-making goes on which is why modern states have capital cities. But the inter-state world city network encompasses more than capital cities: envoys are sent out to missions in other cities. This is why the foreign affairs activity matrix is not square; in the analysis reported here there are 114 capital cities (senders making the network) and 170 cities (receivers) many of which are not capital cities. Examples include New York and Geneva as important UN centres, and there are many countries where the capital city is not a country's economic centre thus additional missions are established in non-capital cities such as: Frankfurt, Hong Kong, Istanbul, Lagos, Mumbai, Sao Paulo, Sydney and Toronto. There are, of course, several such US cities and Los Angeles, Chicago, San Francisco and others feature in the data. Hence, unlike studying a country x country square matrix, my analysis will pick up this extra inter-city dimension that previous studies have been unable to describe.

The details of the 12 principal components derived from the foreign affairs activity matrix are detailed in the appendix (Table A1) where high loadings and scores are listed. There is one immediate feature that is very noticeable: the subnets of cities include relatively large numbers of members. For instance, for both components I and VI, 24 cities are listed, the highest number for all analyses. This means that the results are quite complex and I do not intend to describe each component in detail. However, they do fall into two distinct types of inter-city subnets: regional subnets are most common with 8, but there are also 4 inter-regional subnets, which are, perhaps, particularly interesting for globalization comparisons.

Regional Subnets of Cities

The big three components (I, II, and II in Table 2) are typical regional subnets linking together cities in Latin America, Europe and South/South East Asia respectively. Capital cities are invariably from these regions, the receiving cities likewise but with important additions of leading political cities notable Washington. The latter feature is, however, most developed for central Asian cities (component XII) where links to leading world cities dominate. Further Asian city subnets are found for Islamic cities, wherein leading East Asian cities are also featured (component IV). There are two regional subnets reflecting the former second world: component IV and component XI, the latter also including a northern Europe element. Africa also has its own subnet (component IX).

These findings are consistent with other studies of geographical patterns of embassies wherein regional links dominate: Nierup (1994), for instance, defines 8 ‘regional arenas'. The extra-regional cities that are found in the results (Table A1) – Washington, London, etc. – are equivalent to Nierup's ‘intruders' (p. 79) reflecting the hierarchy in the network. Thus this part of the analysis shows nothing new except that regionality plus hierarchy has survived the Cold War and continues to be the prime feature of the Westphalia city network under conditions of contemporary globalization.

Inter-regional Subsets of Cities

There are four inter-regional subsets and I begin by focusing on just three for reasons that will soon become apparent. The classic example is component V that largely features capital cities from medium-sized states across the world: the three capital cities as senders – Copenhagen, Canberra and Mexico City – illustrate the geographical spread. A more geographically coherent subnet of cities is shown in component VIII where the EMEA (Europe-Middle East-Africa) combination (found in some organizations of global service firms, see Taylor 2000) is produced. Finally component VIII has a very unusual structure with no important sending capital cities featuring but with a subnet consisting of an array of minor cities. The key point is that the latter is totally dominated by Washington, Tokyo and New York. This appears to be a power-seeking subnet that bypasses the traditional, western European ‘imperial core'.

These three components do seem to be distinctive to this analysis and are suggestive, therefore, of a new extension of political city subnets beyond local regions. Certainly Nierup's (1994) analysis throws up no such inter-regional groupings. Nevertheless, I do not want to over-interpret here since Nierup's quantitative methodology is different from mine and therefore dissimilar findings may reflect techniques as much as actual change. However, there can be no such uncertainty with the final inter-regional subnet of cities.

I can be sure that component X is unique to this study because it features many non-capital cities. It is a fascinating for two reasons. First, the sending capital cities feature three countries that are in some sense politically separate from their local regions – Israel, Costa Rica and Taiwan. Second, the lack of local political links is compensated through this component in links to economically important cities, particularly in the USA. Thus the two leading cities in the subnet are Los Angeles and San Francisco and all other US cities (except Washington and New York) appear in the top 12 cities (Houston, Miami, Chicago, Boston and Atlanta). Although US-dominated, other non-capital cities are featured: Hong Kong, Tel Aviv and Toronto. However, this should not be interpreted as incipient globalization since inter-state relations reinforce rather than undermine the Westphalia process. This is an example of states adapting to changing circumstances, a network manifestation of the rise of the (economic) ‘competition state' (Cerny 2000).

In Summary

So what sort of benchmark do I have for searching out new city subnets? Clearly this first analysis of a world political space of flows shows a configuration including numerous cities, highly fragmented in subnets: the initial evidence of a high level of unaccounted variability (despite using a relatively large number of components -Table 2) has been supplemented by most subnets having only local-regional ranges (Table A1).

GLOBAL GOVERNANCE THROUGH A SUPRA-STATE WORLD CITY NETWORK: UN AGENCY SUBNETS OF CITIES

According to Roseneau (1995: 182) ‘Global governance is the sum of myriad – literally millions – control mechanisms driven by different histories, goals, structures, and processes'. This differs starkly from Westphalian inter-state relations not only in its inherent complexity but also in the nature of the politics. As Roseneau tells it, traditional command and hierarchy are challenged by ‘steering' behaviours in a framework of more horizontal relations, in short through network structures. States and their institutions, notably the UN, are part of this new complexity, but only part, for there are numerous other ‘channels' of political behaviours. Nevertheless it is within the UN family of institutions that overt political power above states is mostly concentrated (Scholte 2000: 148-51). Hence I use these institutions to describe a supra-state world city network.

The creation of the UN family of institutions at the end of World War II was an inter-state process with supra-state implications. It was created by states for states and membership has become the accepted symbol of territorial sovereignty. Within the Westphalia framework, the UN is commonly viewed in evolutionary terms whereby the ‘liberty of states' to wage war in the eighteenth century was, first, curtailed by irregular ‘great power' Congresses in the nineteenth century, and then, was limited to ‘defence' by the permanent organizations of the twentieth century, first the League of Nations and then the United Nations. Thus a product of war and peace, the UN appears as the archetypal inter-state institution. But it was never just that. Although failing to embark down the path to world government that many of its supporters have advocated (e.g. Barnaby 2001), it has moved beyond an inter-state agenda. The most overt early example of this is the 1948 Declaration of Human Rights that directly challenges territorial sovereignty. But it is the practices of the UN across a wide spectrum of activities that has led to the UN having a supra-state presence in the world political space of flows.

Rosenau (1992, 69) has referred to the UN as a potential ‘global service'. This is how I interpret it in this study. Like other global services, the UN family of institutions uses cities to supply the wide-ranging public goods it provides. Whether in the fields of health, food, science, labour, development, finance, communications, human rights or refugees, UN agencies operate through cities to make networks of practice. This political world space of flows is little understood because locational strategy has not been a topic of research in UN studies. There have been proposals for decentralization to reduce concentration of power in New York and Geneva (Chapman 1991) and Gorbachev did suggest geographically rotating meetings of the UN Security Council (Frank 1991), but the only comprehensive proposal of a locational strategy that I have found is Rosenau's (1992, 74) idea for countering ‘system overload' by locating ‘UN missions in every member state'. The latter is, of course, consistent with the UN's inter-state character and would, presumably, approximate the Westphalian space of flows described above. But there are locational strategies in the UN family of institutions even though not generally recognised as such: every agency has presences of different degrees across selected cities. These define a supra-state network of cities, my subject here.

The six components that describe the locational strategies of UN agencies (Table 2) differ greatly in terms of the inter-state subnets of cities in two important ways. First, the number of cities featured in the supra-state subnets is generally much smaller implying more selective locational decisions. Second, the scores for the leading city of each component are very large for the subnets (Table A2). With one exception, all are above 6, whereas for the inter-state subnets none are this high (Table A1). Such high scores show the existence of key articulator cities, cities that completely dominate a subnet of cities (Taylor 2004c). This indicates that the supra-state network of cities has very strong hierarchical tendencies. It implies top-down processes through ‘primate city subnets'. These subnets divide into two types. The majority, 4, are dominated by core zone cities and two are dominated by non-core zone cities.

Subnets of Cities Dominated by Core Zone ‘Primate' Cities

It will come as no surprise that the primate cities of the three large components are Geneva, Washington and New York. But these components are very different in their clusters of agencies. Geneva, previously headquarters of the League of Nations, has inherited some institutions from that era and has continued to be a key city for UN activities. The main agencies whose locational strategies place Geneva in this position are the World Population Fund, the World Health Organization, the Inter-state Telecommunications Union and the UN Children's Fund. There are no necessary functional links between the locational strategies that focus on Geneva, they simply pinpoint this city as the general articulator of UN practice.

All other components have a functional bias in their cluster of agencies. This is very clear for Washington's primacy in component II. Here the focus is simply on financial functions, with a very small subnet of cities including just Tokyo and Paris. The New York primate subset (component III) is more like Washington's primacy rather than Geneva's but with a concentration on development and women's issues. This is perhaps a surprise given that New York is the headquarters of the UN. What this result suggests is that it has been Geneva and not New York that has largely reaped the benefits of expansion beyond the formative political concerns of war and peace. Finally, Vienna is primate of one of the smaller components, its neutral location between the Cold War opponents favouring it in having the atomic UN agencies.

Subsets of Cities Dominated by Non-core Cities

The two small components with non-core cities heading their subnets are much less primate: component IV has Bangkok with a score just over six which makes it closer to the highest leading scores in the inter-state subnets (Table A1) than those just described; component VI's highest score (for Addis Ababa) is below four and thus is actually less than most leading scores in the inter-state subnets (Table A1). Both subnets consist largely of non-core cities from three continents and both reflect locational strategies of development agencies. The small component VI focused on Addis Ababa is the more specialist, derived from the two development agencies of the World Bank, it represents a sort of mirror image of Washington's component II. One further feature that they share is a larger number of cities in each subnet, 10 for Bangkok's and 15 for Addis Ababa's.

What these two subnets suggest is that in spreading their largesse across non-core cities, the hierarchical tendencies of UN practices are partly ameliorated and this is particularly so for the World Bank.

In Summary

It is hard to imagine a more contrasting space of flows to the Westphalian benchmark: the dominance of regional fragmentation has been completely replaced by inter-regional subnets that are typically tightly structured and very hierarchical. This second world political space of flows largely consists of top down processes dominated by just four ‘UN primate cities' located in the core zone.

GLOBAL CIVIL SOCIETY THROUGH A TRANS-STATE WORLD CITY NETWORK: NGO SUBNETS OF CITIES

Although it is commonly agreed that global civil society is a ‘fuzzy concept' (Anheier et al. 2001b: 11; An-Na'im 2002: Chandhoke 2002)) with its ‘organizational infrastructure' still in a ‘state of flux' (Anheier and Themudo 2002: 191), nevertheless Keane's (2001: 23) description provides the essence of the subject: ‘Global civil society is a vast, interconnected, and multilayered social space that comprises many hundreds of self-directing or non-governmental institutions and ways of life'. Through its ‘cross border networks' global civil society is constituted of ‘chains of interactions linking the local, regional and planetary orders' (p. 24), This new social world is constituted by ‘networks, coalitions, partnerships and social movements' but NGOs are commonly taken as the key operators (Anheier and Themudo 2002): in Chandhoke's (2002: 38) words, ‘NGOs play a large-than-life role in global civil society'. So much so that it is often found to be necessary to remind readers that global civil society is actually more than the activities of just NGOs (e.g. Anheier et al. 2001: 4; Chandkoke 2002: 38; Anheier and Themudo 2002: 191). While accepting the latter, NGOs remain the obvious starting point for describing the geography of global civil society because of their explicit mode of operation beyond states. Hence I use these institutions to describe a trans-state world city network.

The growth of NGOs has been exponential since the mid-nineteenth century and they now constitute a formidable number of social agents at all geographical scales. It has recently been reported that just under eighteen thousand NGOs are ‘international' or ‘internationally-orientated' (Glasius, Kaldor and Anheier 2002: 322). As noted previously, the NGOs in this analysis are all worldwide in scope. Each has a specific purpose guiding its actions and the latter generate geographical fields of activity that are my subject here. This activity has involved NGOs using cities as organizational foci and, unlike for UN agencies, a city-centred approach to understanding their operations is emerging (Glasius and Kaldor 2002; Sassen 2002; Taylor 2004c). But these studies do not describe the global configuration of NGO activities.

The principal components analysis of the NGO activity matrix identifies ten common uses of cities (Table 2) and these locational strategies are reported in Table A3. The resulting subnets of cities fall into three categories: strongly hierarchical subnets wherein one primate city totally dominates; moderately hierarchical subnets in which one city stands out but rather less so than above; and subnets without a primate city.

Strongly Primate Subnets

There are three components that generate component scores at the level found in the UN agencies analysis above. As in the latter, the primates are leading ‘first world' cities: London. Washington, and Geneva.

The London subnet (component I) is the most hierarchical as measured by the difference between the first and second highest scores. This is by no means a ‘regional ‘ subnet: other Western European cities do not feature and the five featured cities below London (Table A3) cover all continents and different levels of world city status. Other cities scoring positively on this loading show a preponderance of ‘third world' cities. Not surprisingly, the NGOs with the high loadings on this component have their headquarters in London. Thus this is a city subnet with London at the head of a worldwide, but extremely, hierarchical location strategy.

The other two extremely hierarchical subnets are UN-related. As expected, the Washington subnet (component III) is the product of several Washington-based NGOs. They can be interpreted a common group of NGOs promoting, and mopping up after, the ‘Washington consensus' policy programmes. The geography being articulated is quite specific. Operating also through Brussels and New York, this locational strategy involves capital cities of countries that have made abrupt changes to neo-liberal economic policies. Former-COMECON cities are listed in Table A3, but below the 0.4 threshold other such cities plus Asian and Latin American cities are found. In contrast the Geneva subset (component IV) is generated by a different set of NGOs with the focus upon lobbying UN agencies. The subnet of cities is distinctive with Pacific Asia prominently featured.

Clearly, although the specifics are different, the structure found here is very similar to the top three UN agency locational strategies: replacing New York by London as primate, NGOs have created another trio of very hierarchical city subnets.

Other Primate Subnets

The smallest four components (Table 2) also have primate cities but their dominance over other cities in their subnet is far less than above. A particularly interesting feature is that two of the four subnet primates are non-core zone cities.

Brussels and Ottawa are the two core cities with the moderately hierarchical subnets. The Brussels subnet (component VIII) is very much a core pattern of cities featuring western European, USA and Pacific Asian cities. Eastern Europe is also represented by two cities. Ottawa's subnet derives from the smallest component (X). There is certainly a hint of an Americas bias in this subnet (5 out of 11 cities featured) but African cities also feature (4 out of 11). This locational strategy reflects particular Canadian NGOs and shows how a relatively minor political city from a core country can become a subnet primate city in world city networks.

Nairobi and Manila are the two non-core cities that head hierarchical subnets (components VII and IX). In Nairobi's case it confirms the city's role as a leading NGO city in terms of connectivity (Taylor 2004d). Note that its articulation role is not regional: the city subnet includes more Asian than African cities. The core city link is with Tokyo. Thus this locational office strategy most definitely does not show Nairobi as part of an emerging African city network. Unlike the Nairobi articulation, Manila's subnet does exhibit a locational bias towards its own region (Pacific Asia) plus a strong showing in neighbouring South Asia. The core city link is European, mainly Brussels but also Frankfurt.

The key point from this part of the analysis is that Nairobi and Manila join Bangkok from the previous UN agency analysis as examples of non-core city primates within world city networks.

Non-primate Subnets

There are three city subnets that are without dominance by a single city. They feature many more cities than other subnets: component II with 15 cities, component V with 16 cities, and component VI with 18 cities. Note their similarity to the two city subnets from the previous supra-state analysis headed by Bangkok and Addis Ababa. In general, such results define a different spatial organization: common ‘horizontal' locational strategies defining network fields of activity rather than simple primate hierarchies.

It is particularly interesting that the second most important component in the analysis defines a common horizontal locational strategy. It is a horizontal field of UN associated activity (component II), easily identified as such through the NGOs that load high (Table A3) where UN NGOs feature prominently. It is ironic that UN association generates a horizontal field of activity with no articulator city given the tendency towards city primacy in the previous analysis of UN agencies (Table A2). In this city subnet New York and Geneva are very important but they in no way dominate: for example, they have scores below Cairo. In general it can be noted that this network field of activity transcends the core/non-core division. But it is not worldwide; there is a very strong Afro-Asian locational bias with Latin American cities conspicuous by their absence.

The other two non-primate city subsets are both ‘medium-sized' components. There is a horizontal field of core city activity (component V) that is organised through Canadian and Scandinavian cities (6) plus other European capitals (4) and two important Pacific Asian cities. Only four non-core cities feature, one for each major region: sub-Saharan Africa, Latin America, North Africa/Middle East, and South Asia. The other non-primate city net shows an inter-regional locational bias: there are eight Western European cities and four West African cities in this horizontal field of activity (component VI). This is the only office locational strategy that appears to link two specific regions in a core/non-core interaction. However it is not limited to just these two regions: there are two cities each from North America, South America, Pacific Asia in this city subnet.

The common feature of the horizontal city subnets is that they imply a core/non-core interaction. But this geographical interaction takes three different forms: a basic non-core field of activity plus UN centres (component II); a basic core field of activity with selected non-core centres (component V); and a targeted inter-regional field of activity across the core/non-core divide (component VI).

In Summary

NGOs define two basic types of locational strategy: primate and horizontal. Core zone cities dominate the primacy findings but there are two non-core cities that emerge as moderate primates. In addition the horizontal city subnets feature other non-core cities as important centres.

COMPARING SPATIALITIES IN WORLD POLITICAL SPACES OF FLOWS

The last three sections delineated city-centric political geographies; spatial distributions of power operating through very different institutions. I have termed the resulting networks inter-, supra-, and trans-state structures on the basis of what is known about how state departments, UN agencies and NGOs conduct their business. This interpretation is obviously also informed by globalization theses that invariably posit new spaces of activity. But now that findings from equivalent analyses of the three inter-city networks are available, it is possible to take the argument further. In this section, I compare the city-centric political geographies in terms of the basic dimensions of spatiality that are revealed.

Four basic facets of inter-city spatialities are summarised in Table 3. I will describe the spatiality of the benchmark network, Westphalian inter-state process, to introduce these four facets. First, the overall structure describes the degree of configuration in the network. This is indicated by the number of components required to describe the network, and by how much of the total variance they account for. The inter-city network turns out to be poorly configurated showing weak commonality in a highly fragmented structure. This means that the agents of network formation, the state departments, do not operate through major common locational policies. Second, the regional structure shows a definite bias towards sub-regional city nets. This indicates that the fragmentation takes the form of largely intra-regional locational strategies. Third, the zonal structure is a relatively simple separation between core and non-core zones. Thus the locational decision making adheres to the basic global ‘North-South divide'. Fourth, the network structure shows a weak level of primacy suggesting a relatively ‘horizontal' collection of city subnets. In this case the diplomatic activity produces a spread of power across the network. Thus the Westphalian inter-state process described by this analysis is very weakly patterned as a global structure.

Before I continue using Table 3 for comparisons, I need to comment on the last finding and to generalise its limitations to all findings. In a time of US political dominance with so much popular talk of an ‘American empire', it may seem perverse to describe the current Westphalian process as weakly primate. Two points need to be made clear. First, the measures employed to describe Westphalian process are derived from diplomatic activity and not military deployment and action. It is in the latter where exceptional US superiority lies, and analysis focussing on the military would inevitable find a most hierarchical order. I argue that the more subtle diplomatic circles are most instructive for understanding basic operation of the Westphalian process. In this case Washington actually appears more than any other city (as receiver in 7 of the 12 subnets) and records the highest level of primacy in the analysis. But the fact remains that my findings must be interpreted as partial, I hope they are reasonable descriptions but they are certainly not complete ones. Second, this limitation applies to both the other city networks. Supra-state activities extend beyond UN agencies and trans-state activities are not the sole work of NGOs. Thus all three of the city networks described here are only specific representations of their respective processes, albeit generally recognised as main representations, as argued in the introduction. With this important caveat in mind, I can now continue to make some modest but interesting comparisons between processes.

Although the supra-state analysis uses less components for its description and accounts for much more variation than the Westphalia case, I still only term its commonality moderate. However, the components imply a less fragmented picture in that there are some functional specialisms among the subnets producing a compartmentization of structure. It is quite remarkable how inter-regional this spatiality is: every subnet includes cities in three continents even though two subnets consist of only three cities. But this does not translate into an inter-zonal structure: the two three-city subnets are just core-centred. Thus I denote this network as only moderately inter-zonal. However, its distinctiveness is the strong primacy within its subnets. The supra-state process through cities appears to be very hierarchical. Note that this is not a global govern ment process creating a national-urban hierarchy writ large: the hierarchy is multiple, in subnets that are trans-continental. This is because, although the network makers are all UN agencies, this is only a ‘family' of institutions, without singular direction: the agencies range from largely autonomous to wholly independent in their locational activities. This makes the primacy finding particularly interesting: the global governance implied by this analysis is a basic process of top-down hierarchical activity through functional compartments across the world.

In terms of overall structure the trans-state network appears to be somewhat between the previous two patterns: it has a commonality the same as for the supra –state network but the number of subnets is very close to the inter-state network. It also appears to be ‘in-between' in terms of hierarchical structure with some very primate subnets although this feature is not so pervasive as in the supra-state network. But in the other two facets of spatiality, the trans-state network is distinctive. The remarkable finding is in the simple geography of the subnets: they are ALL both inter-regional and inter-zonal. Thus it appears that it is this incipient global civil society being pioneered by NGOs that has the strong global structure in my analysis. However, given the primacy in these networks wherein the articulator cities are invariably in the core-zone, this global civil society's power structure is clearly top down rather than bottom-up.

So what do these comparative findings tell us about new globalization processes in contrast to long-standing Westphalian processes? I draw four fundamental conclusions.

1. Supra-state and inter-state processes are indeed creating more globally structured political geographies than is apparent with inter-state processes.
2. While encompassing some similarities, supra-state and inter-state processes are creating distinctively different political geographies.
3. Especially supra-state processes, but also trans-state processes, are clearly creating a more hierarchical global structure than has been apparent with Westphalian inter-state processes.
4. Especially trans-state processes, but also supra-state processes, are definitely creating political structures that transcend the ‘North-South divide'.

And, I suggest, these conclusions define new ‘deep' political geographies, spatialities of power not subject to rapid change like the inter-state alliances that create new spatial dispositions of power in geopolitical transitions.

DISCUSSION: BACK TO PROCESS

All three analyses have been cross-sectional, indicating global patterns at a given time. This I have measured outcomes of processes; in this final discussion I interrogate the processes further by asking what do these deep new political geographies mean? Certainly it is to be expected that altering spatialities cannot be an isolated facet of fundamental social change. This brings the discussion into the realms of Wallerstein's (1974) argument on the demise of the modern world-system. Certainly the latter will be premised on new spatialities but I do not wish to pursue this argument for this final discussion. I will narrow the scope to strictly socio-spatial features of change.

The first point to note is that my simple equating of network making agents to types of network and spheres of activity is not as clear cut as it has been presented. This point has been conceded previously but it must be re-examined for broader discussion of the results. In particular, the role of NGOs exemplify this blurring of spheres. Following the literature of global civil society (especially the new Yearbooks ), I have used NGOs to represent this sphere of activity. But NGOs also feature in discussions of global governance. For instance, Roseneau's (1995: 182) global governance involves ‘steering' behaviours in a multilayered framework of relations in which NGOs are key contributors to a global informal politics. In terms of practice, therefore, it is the combination of these formal (state-based) and informal institutions as ‘a set of interlocking but separate bodies' that ‘produce the system of global governance' (Halliday 2000: 334). This makes my equating of NGO activity as the representative process in global civil society building problematic. Civil global society looks very much like the informal half of Roseneau's global governance: both are composed of diffused networks that transcend sovereign territories with NGOs featuring prominently.

This is grist to the mill for Chandhoke's (2002: 35) ‘three sector fallacy', the common assumption that politics, economy and society are separate and distinct realms of action. Extrapolated to the global level, this fallacy specifies a global civil society as the ‘third sector' ‘which can not only be distinguished from but which is an alternative to both the state-centric international order and the networks of global markets' (p. 36). Chandhoke shows that global civil society to be inherently ‘contaminated' by both the other two sectors (through massive state and private fundings of NGOs, and through NGOs working within the UN). Her emphasis on the limits of global civil society tempers claims for its radical transformative role; she interprets it as being firmly embedded in a particular post-Cold war conjuncture dominated by rich countries. Certainly my findings on city primacy and hierarchy in both supra-state and trans-state fit Chandhoke's position better than Roseneau's challenge to hierarchy through more diffuse ‘horizontal' channels of power. It should not be forgotten that undermining the state at a global scale also means eroding the most important twentieth century radical geography outcome: decolonisation, that late, and perhaps last, major political geography product of the Westphalian process.

However, I do not interpret my findings as implying that the three spheres as organizing frames for our understanding of contemporary globalization should be abandoned at this time. My research conjecture that the interlocking model could be transferred from economic to non-economic agents of city network formation has been proven reasonable through producing new results that counter assumptions (often masquerading as definitions) common in the literature. Thus Roseneau's suggestion that global governance is through more horizontal power structures is disputed by my analyses. Westphalian interstateness, with its diffusion of diplomatic circles to all states, is actually a relatively ‘horizontal' benchmark that neither my global governance nor global civil society results are able to match. Thus the next step for this research does not involve dissolving spheres of activity but being more mindful of inter-sphere relations. For example, returning to the role of NGOs, their position at the heart of both global governance and global civil society arguments has to be explained. For this it is necessary to go beyond process outcomes and rethink the processes themselves. One perspective is to argue that NGOs constitute the key integrating institutions between global governance and global civil society (Taylor 1999: 92-3). Their role is analogous to political parties in intra-state politics. Neither part of government nor national civil society, parties have operated as conduits between political elites and society. From the time when governments relied on parties for their election and therefore power, policies were proposed for society by parties to government, and also propagated to society by parties for government. NGOs have now found themselves in a similar ambiguous position, most noticeably in their links to UN activities (Otto 1996), and links to states through the distribution and organization of the development aid programmes from ‘North to South' through cities across the world (Townsend 1999; Townsend et al . 2002). This is a basic source of both the additional primacy and the inter-zonality to be found in subnets not subject to the Westphalian process.

ACKNOWLEDGEMENTS

The work that produced the interlocking city network model was supported by a research grant from ESRC. Data collection was carried out through the Metropolitan Institute at Virginia Tech. Specifically, the state departments' activity matrix was created by Kevin Matthews, the UN agency's activity matrix by Clare Blake, and the NGO's activity matrix by Troy Gravitt.

REFERENCES

Anheier, H. (2002). ‘Measuring global civil society', Introducing global civil society'. In H. Anheier, M. Glasius, and M. Kaldor (eds) Global Civil Society 2001 . Oxford: Oxford University Press.

Anheier, H., Glasius, M. and Kaldor, M. (eds) (2001a). Global Civil Society 2001 . Oxford: Oxford University Press.

Anheier, H., Glasius, M. and Kaldor, M. (eds) (2001b). ‘Introducing global civil society'. In H. Anheier, M. Glasius, and M. Kaldor (eds) Global Civil Society 2001 . Oxford: Oxford University Press.

Anheier, H. and Themudo, N. (2002) ‘Organizational forms of global civil society: implications of going global'. in M. Glasius, M. Kaldor, and H. Anheier (eds) Global Civil Society 2002 . Oxford: Oxford University Press

An-Na'im, A. (2002). ‘Religion and global civil society: inherent incompatibility or synergy and interdependence?'. In M. Glasius, M. Kaldor, and H. Anheier (eds) Global Civil Society 2002 . Oxford: Oxford University Press.

Barnaby, F. (1991). The CAMDUN conferences. In F. Barnaby (ed.) Building a More Democratic United Nations . London: Frank Cass.

Castells, M. (1996). The Rise of Network Society . Oxford: Blackwell.

Castells, M. (1999). Grassrooting the space of flows. Urban Geography , 20, 294-302.

Chandhoke, N. (2002). ‘The limits of global civil society'. In M. Glasius, M. Kaldor, and H. Anheier (eds) Global Civil Society 2002 . Oxford: Oxford University Press.

Chapman, D. (1991). A method for the direct election of a popular second assembly of the UN. In F. Barnaby (ed.) Building a More Democratic United Nations . London: Frank Cass.

Frank, A. G. (1991). Gorbechev's United Nations initiatives: steps in the right direction. In F. Barnaby (ed.) Building a More Democratic United Nations . London: Frank Cass.

Glasius, M. and Kaldor, M. (2002). ‘The state of global civil society before and after September 11 th ' . In M. Glasius, M. Kaldor, and H. Anheier (eds) Global Civil Society 2002 . Oxford: Oxford University Press.

Glasius, M., Kaldor, M. and Anheier, H. (eds) (2002). Global Civil Society 2002 . Oxford: Oxford University Press.

Halliday, F. (2000) ‘Global governance: prospects and problems'. Citizenship Studies 4, 330-48.

Keane, J. (2002) ‘Global civil society?', Introducing global civil society'. In H. Anheier, M. Glasius, and M. Kaldor (eds) Global Civil Society 2001 . Oxford: Oxford University Press.

Knox, P. L. and Taylor, P. J. (2004). Globalization of architectural practice. Journal of Architectural Education ,

Kratke, S. and Taylor, P. J. (2004). A world geography of global media cities. European Planning Studies ,

Nierop, T. (1994). Systems and Regions in Global Politics . Chichester, UK: John Wiley.

Otto, D. (1996) ‘NGOs in the UN system: the emerging role of international civil society'. Human Rights Quarterly, 18, 107-41.

Roseneau, J. N. (1992). The United Nations in a Turbulent World . Boulder, CO: Lynne Rienner.

Roseneau, J. N. (1995) ‘Governance and democracy in a globalizing world'. In D. Archibugi, D. Held and M. Kohler (eds) Re-imagining Political Community: Studies in Cosmopolitan Democracy . Cambridge: Polity.

Rummel, R. J. (1970). Applied Factor Analysis . Evanston: Northwestern University Press.

Sassen, S. (2002). ‘Global cities and diasporic networks: microsites in global civil society'. In M. Glasius, M. Kaldor, and H. Anheier (eds) Global Civil Society 2002 . Oxford: Oxford University Press.

Scholte, J. A. (2000). Globalization: a Critical Introduction . London: Macmillan

Taylor, P.J (1995). Beyond containers: internationality, interstateness, interterritoriality. Progress in Human Geography , 19,1-15.

Taylor, P. J. (1999). Modernities: a Geohistorical Introduction . Cambridge: Polity.

Taylor, P. J. (2000). World cities and territorial states under conditions of contemporary globalization. Political Geography , 19, 5-32

Taylor, P. J. (2001). ‘Specification of the world city network'. Geographical Analysis , 33, 181-94

Taylor, P. J., Catalano, G., and Walker, D. R. F. (2002a). ‘Measurement of the world city network'. Urban Studies, 39, 2367-76

Taylor, P. J., Catalano, G., Walker, D. R. F. (2002b). ‘Exploratory analysis of world city network'. Urban Studies , 19, 2377- 94.

Taylor, P. J. (2004a). World City Network: a Global Urban Analysis . London: Routledge.

Taylor, P.J. (2004b). The new geography of global civil society: NGOs in the world city network. GaWC Research Bulletin 144.

Taylor, P. J. (2004c). Leading world cities: empirical evidence of urban nodes in multiple networks. GaWC Research Bulletin 146.

Taylor, P. J., Catalano, G., and Walker, D. R. F. (2004). ‘Multiple globalisations: regional, hierarchical and sectoral articulations of global business services through world cities'. The Services Industries Journal ,

Townsend, J. (1999). ‘Are non-governmental organizations working in development a transnational community?'. Journal of International Development ,11, 613-23

Townsend, J., Porter, G. and Mawdsley, E. (2002). ‘The role of the transnational community of NGOs: governance or poverty reduction?' Journal of International Development , 14, 829-39.

Van der Wusten, H. (2003) The distribution of political centricity in the European states system. GaWC Research Bulletin 91.

Table 1: Details of data matrices

Table 2: Principal components of the three data matrices

Table 3: Spatialities of political world city networks

Appendix: Detailed Principal Components Results

Table A1: The inter-state city network: state departments as networkers

Table A1 (cont.)

Table A2: The supra-state city network: UN agencies as networkers

Table A2 The supra-state city network: UN agencies as networkers