View all Blog posts

Posted by: Hilde de Weerdt 6 years, 8 months ago


Two frameworks for understanding spatial control and political integration in Chinese history

Hilde De Weerdt

            In this post I review the principal criticisms of the Skinnerian approach to Chinese empire and space and compare the Skinnerian model with Social Network Analysis (SNA) models, asking to what extent SNA can address criticisms of the former. This effort fits into a larger project to re-examine explanations for the maintenance of empire in the historical sociology of empires by adding political communication and political literacy to the more common variables of economic and military power, and political and ideological hegemony. Within the context of the Skinnerian model, it can also be understood as an attempt to move away from the binary of marketing networks and administrative hierarchies in the analysis of the structure of Chinese society to relational models that set out to test the relevance of different categories of relationships (social, economic, spatial, cultural, or political) over time.


This post is based on part of a presentation delivered at the panel "Empire and Space" at The Annual Meeting of the Association for Asian Studies, Philadelphia, March 26, 2010. The panel was dedicated to G. William Skinner and also aimed to provide alternative models to conceive of Chinese space. I thank my fellow panelists Fei Siyen, Karen Wigen, and Mark Edward Lewis, and the many members of the audience who raised questions afterwards.

Regional systems theory

            Analyses of the relationship between space and empire in imperial China are to a large extent still defined by the work of William Skinner so much so that terms such as the macroregion have become part of our standard vocabulary. As suggested by the handful of panels dedicated to his legacy at the 2010 Association for Asian Studies meeting, Skinner has remained an authoritative interpreter of the spatial structure and dynamics of socio-economic and cultural practices in imperial China.

            The macroregion has become representative of the repertoire of concepts and methods applied in Skinner’s work. It entered the language of the Chinese humanities and social sciences as a synecdoche for Skinnerian analysis in the 1980s.[1] The macroregion was the result of Skinner’s creative adaptation and combination of, in the first instance in the 1960s, central place theory, and, in the second instance in the 1970s, regional systems theory. From an examination of the spatial pattern of marketing systems in Sichuan he moved on to a regional model of the Chinese territories as a whole. Skinner’s answer to the question of the extent to which Chinese society was integrated in imperial times changed significantly after his exposure to regional systems theory. In “Marketing and Social Structure in Rural China” (1964-1965) he assumed that the nesting of marketing structures in ever more encompassing marketing systems culminates in an empire-wide network that parallels and reinforces the hierarchy of administrative jurisdictions:

"It is the joint participation of standard markets in two or three intermediate marketing systems, of intermediate markets in two or three central marketing systems, and so on, which articulates and unites the little local economies centered on each market town into, first, regional economic structures and eventually into a single society-wide economy. Thus, marketing had a significance for societal integration in traditional China which at once paralleled and surpassed-which both reinforced and complemented-that of administration."[2]

By the 1970s, in “Mobility Strategies in Late Imperial China: A Regional Systems Analysis” (1976) and his contributions in The City in Late Imperial China (1977), Skinner noted that physiographic features such as rivers and topography constrained the hierarchy of nested economic systems. The constraints appeared particularly significant at the regional level: “The magnitude of interregional transactions was insufficient to link these regional systems of economic central places into a single integrated urban system.”[3] In the latter work in particular Skinner proposed that Chinese socio-economic history (and by extension all Chinese history) can be best approached through the ups and downs of the physiographic macroregions. He sketched the history of urbanization in the macroregions through the last two millennia and argued against the use of other spatial units such as provinces or the empire in historical analysis because he considered those mere administrative units which blurred the economic systemic boundaries of the physiographic regions.

            Because the concept, and the powerful map demarcating the physiographic regions of China within which socio-economic as well as religious and cultural behaviors were to be understood, became paradigmatic, its origins, its theoretical assumptions, its methodological choices, and its application received little sustained scrutiny. Carolyn Cartier wrote a multi-faceted critique raising fundamental questions about the theoretical assumptions of the macroregional model. First, even though not an advocate of environmental determinism as earlier geographers had been, Skinner’s criteria for the definition of macroregions assume the primacy of economic and physiographic factors.[4] He sees economic central functions as basic and other religious, political, social, and cultural functions as dependent on them.[5] Research on the social and cultural construction of places in recent decades underscores the inadequacy of such an understanding of the relationship between economics and culture. Faure and Siu have shown that the spatial dimensions of human behavior in southern China cannot be adequately captured within the confines of the geo-economic logic of transportation costs fundamental to central place theory.[6] Its inability to explain historical change (such as the processes that form and transform regions, cores, and peripheries) and to include human agency along structural factors in the formation of places are other aspects of the model that have attracted critical attention. Crucially, the insistence on the discreteness of the macroregions leaves the question of how empire-wide territorial coherence could be maintained unanswered.

            Others have noted that Skinner’s model neglects or underestimates inter- and supraregional economic exchange; the same could be said about the modeling of politics. Despite the valuable insight that socio-economic and cultural behaviors operate according to logics that both differ (at lower levels of human settlement) and overlap/interact (at the higher levels) with bureaucratic/administrative structures and dynamics (his analysis of the bureaucratic labels affixed to administrative jurisdictions is particularly revealing in this regard), political behavior receives only limited attention in Skinner’s analysis of the spatial organization of Chinese empires. Within Skinner’s model, political behavior is limited to either administrative/ bureaucratic behavior or non-bureaucratic local leadership in welfare operations. The former fits into a strict hierarchical model of discrete administrative organization where each lower-level unit is made to fit within one and only one higher-level unit and so effectively controlled by the center. The latter is in his analysis part of the marketing structure and so brings elites of different types together in the teashops of the standard and intermediate level marketing towns. The central market towns that were absorbed within the jurisdictional structure also become the scene for the intermingling and joint leadership of local government and local elites. Missing within this picture is any consideration of communication and exchange among elites resident in standard, intermediate, or central market towns that were not adjacent within the hexagonal structures of the central place model of marketing towns.


Network analysis

            Social Network Analysis (SNA) can address some of the theoretical and methodological problems in Skinnerian central-place analysis, but it has also been subject to some of the same criticisms.

            The term network itself deserves some further scrutiny as its meaning has changed significantly in the course of the last decades and it continues to mean very different things in academic and non-academic discourse. Skinner, for example, used the term “networks” to describe marketing structures: “Marketing structures, unlike administrative structures, take the form of interlocking networks.”[7] “Network” in Skinner’s metaphorical use of the term refers to the non-discrete nature of marketing structures particularly at the lower, sub-regional, level. He thus differentiates the marketing networks at the lower level from the hierarchy of non-overlapping administrative jurisdictions. Even though the boundaries of the network are similarly fluid in current network analytical definitions, a crucial difference between the former and the latter use of the term concerns the role of different factors and attributes in the formation of networks. 

            Whereas economic functions are primary in Skinner’s model, SNA is agnostic about the weighting of the different categories of relations that shape networks and individual or collective attributes such as physical address, occupation, or class membership. In the words of one advocate and critic, SNA’s point of departure is the anti-categorical imperative.[8] It is thus less likely to be subject to criticisms of economic and environmental determinism. It is in part because of the anti-categorical imperative that SNA is typically imagined as a set of methods, as a heuristic rather than as a body of theory.[9] Indeed, as a set of methods to examine individual and collective behaviors through actors’ involvement in structured social relations SNA tests may also lead to the conclusion that network effects do not obtain and so allows for its own irrelevance.

            Despite its representation as a body of methods, it has been rightly pointed out that, as any methodology, SNA is based on a set of theoretical assumptions.[10] In historical social network analyses the influence of theoretical assumptions may vary depending on the uses to which network analytical methods are put; generally speaking, the more transformations applied to historical data, the more influential theoretical assumptions become.[11] Irrespective of such differences, SNA is a relational sociology and thus conceives of social relationships as a determinant of human behavior; it privileges interaction, i.e., social relationships, the measurement of social relations, and the analysis of relationship patterns over attribute variables or socio-cultural norms in the explanation of individual and group practices. A network is conceived as a social space. Just as individuals or communities occupy locations in geographic space where their proximity may have implications for socio-economic and other practices, they can be shown to be socially proximate or distant in network diagrams with similar implications for the flow of goods, ideas, diseases, information, etc.[12] The structural features of the network can, just as the diagrams of marketing structures or the macroregional map, suggest the potential for as well as the constraints on exchange amongst individuals and groups, but in the case of SNA the nature of the enabling and constraining factors remains to be determined by the investigator.

            Such a conceptualization of social space is like central place theory a construct that emerged from mid- and late twentieth century European and American university campuses (most notably the University of Manchester and Harvard) and differs radically from indigenous conceptualizations of social space and human relatedness as embodied in the nested hierarchies of body, family, region, and the realm.[13] Nevertheless, even though its genealogy shows the signs of its western provenance and use in modern western contexts, SNA is not tied to these contexts but, as a set of techniques, broadly applicable to various objects and contexts. Unlike central place theory and the Skinnerian model, SNA does not adhere to a well-defined hierarchy and as noted above allows for its own inapplicability. The network of relationships should be seen as one environment of action amongst others (such as broader cultural environments).[14]

            Network analysis is furthermore based on the idea that the elements that constitute the network (the nodes) can be defined and redefined flexibly; analyses can be conducted at both the individual and group levels and range from individual subjects to countries and groups of countries. Some historical network analysts have applied network methods to concepts or discursive elements at a remove from concrete social relations.[15] Networks can thus be flexibly reconfigured to explore questions from different angles. Scalability is built into SNA and networks can be understood both in their entirety and at the level of ever-smaller constituent parts. In their comparison of SNA to other types of social structuralism Emirbayer and Godwin therefore conclude that it is both more general and more concrete because the larger structures (whether groups or institutions) can be disaggregated and made concrete in the lower-level components of actors and relations.

            In the context of spatial analysis, SNA is compatible with alternative models of mapping human socio-economic and other activity geographically which focus on “interconnected spatial scales of activity.” “Such attention to scale can help explain the actual workings of social and economic processes through a spatial hierarchy, moving from local to regional, national, and international levels of concern.”[16] Such ideas hold more promise for explaining the maintenance of China's territorial coherence.

            SNA can thus overcome some of the problems raised with regard to Skinnerian central place and macroregional theory, but it also suffers from some of the same limitations. Ahistoricity is a potential danger as networks are often based on static data; this is, however, a methodological problem that can and has been overcome through the temporal sequencing of networks allowing for the variation of both ties and attributes within a given network. Network analysts are, moreover, interested in processes as well as patterns and have been tracing the formation, development, as well as the disintegration of networks. Whether network analysis is in itself sufficient to explain such processes is, however, a more challenging question.

            Arguably the most powerful type of criticism that has been raised of SNA in historical research is the lack of network analysis’ ability to explain the historical reasons for the formation and transformation of the mapped networks. Emirbayer’s and Goodwin’s critical review of SNA’s contributions to relational sociology and history highlighted the persistent neglect of broader cultural and political explanations for the historical dynamics of networks as well as the role of human agency in the use of cultural and political ideational resources. This critique has borne result and recent work such as Paul McClean’s study of strategic interaction in Renaissance Florence demonstrates how the analysis of cultural and political discourses can be combined with network analysis to produce a history of patronage and networking itself.[17]

            The conceptual repertoire, the quantitative methods, and the increasingly diverse array of statistical and visualization software that have appeared since the 1970s have turned SNA into a powerful toolbox for the investigation of the social space within which human histories are written in ways that are both reminiscent of [18] but also transcend the limitations of the economic and spatial configurations of the Chinese empire in Skinner’s work. Nevertheless, network graphs, like macroregional maps, remain texts for examination based on data and algorithms that remain subject to re-evaluation. The historian Claire Lemercier notes that network graphs may result in a fallacy of objectivity, but ultimately concludes: “Ce n’est pas non plus un nouvel objectivisme qui permettrait de cartographier la structure sociale et de prévoir tout comportement individuel. Elle peut présenter le lien à la fois comme une resource et comme une contrainte, envisager l’individu et ses entourages ou bien les fractures et hiérarchies d’une structure plus large.”[19]


The interaction between quantitative and qualitative approaches 

            The interaction between history and network theory need not bring back the specter of positivist and cliometric social history. Even though historians and sociologists have on the whole different aims, with the former focused on the particular and the explanation of particular historical changes and the latter on generalizable patterns,[20] the tensions engendered by an encounter between the two can yield productive results. Cultural historians calling for a critical hermeneutics of texts as well as the recovery of historical discourses can benefit from the work of sociologists working on the construction of social discourses in present-day and historical contexts. Paul McLean's already-mentioned work on strategic networking was based on the analysis of thousands of letters exchanged during the Italian Renaissance. Its analysis of changing discourses of honor shows how the translation of texts into an analysis of historical discourses is not as immediate as cultural historians make it appear to be and can benefit from quantitative analyses. John Mohr's articles on the historical meaning of poverty and the construction and impact of social welfare classification systems in the poverty directories of Progressive Era New York are further examples of the significant contributions sociologists are making to cultural history.[21] Bill Sewell similarly underscores the compatibility of interpretive and quantitative methods in Logics of History: "Quantitative reasoning works because semiotic practices of all sorts (precisely because they are generated by more or less stable codes or paradigms) produce regularities in human action." And, "this assumption that quantitative and interpretive methods belong to incompatible epistemic universes is both theoretically troubling and practically doubtful."[22]



            Skinner proposed, justifiably so, that Chinese history needs to be disaggregated into a small set of constituent regions based on physiographic criteria. Within this model (mostly designed for the analysis of marketing and economic exchange) political exchanges take place within the administrative hierarchy imposed by the state (top-down flows of information) or are limited to the reach of local marketing networks—where local elites can gather in teashops. I propose that a network approach allows us to tackle the question of the coherence and continuity of empire better. By overlaying the administrative hierarchy with horizontal communication networks across vast distances and by interpreting cultural and political discourses within the context of social relationships we should be able to better understand the role of elites in the long-term maintenance of empire in the second millennium of Chinese history.


[1] Cartier, “Origins and Evolution of a Geographical Idea: The Macroregion in China.”

[2] Skinner, “Marketing and Social Structure in Rural China,” 31.

[3] Skinner, “Mobility Strategies in Late Imperial China,” 330.

[4] Skinner, The City in Late Imperial China, 281-82.

[5] Skinner, The City in Late Imperial China, 276.

[6] Faure and Siu, Down to Earth: The Territorial Bond in South China.

[7] Skinner, “Marketing and Social Structure in Rural China,” 31.

[8] Emirbayer and Goodwin, “Network Analysis, Culture, and the Problem of Agency,” 1414-15.

[9] See, for example, Scott, Social Network Analysis, 37.

[10] Emirbayer and Goodwin, “Network Analysis, Culture, and the Problem of Agency”; Emirbayer, “Manifesto for a Relational Sociology.”

[11] This is especially the case if historical data are used for modeling their impact on other social phenomena. Gould, “Uses of Network Tools in Comparative Historical Research,” 250-51.

[12] Robins and Pattison, “Interdependencies and Social Processes: Generalized Dependence Structures,”211-12.

[13] Lewis, The Construction of Space in Early China.

[14] Emirbayer and Goodwin, “Network Analysis, Culture, and the Problem of Agency,” 1443.

[15] An excellent illustration of the historical relevance of this approach is Mohr, “Soldiers, Mothers, Tramps and Others: Discourse Roles in the 1907 New York City Charity Directory.” See also McLean, The Art of the Network, chs. 3-5. For a critical review of the development of such approaches, see Mohr and Franzosi, “New Directions in Formalization and Historical Analysis,” and Mohr, “Network Methods, Historical Texts, and the Formal Analysis of Institutional Systems”; Gould, “Uses of Network Tools in Comparative Historical Research,” 242, 259-63.  

[16] Cartier, “Origins and Evolution of a Geographical Idea: The Macroregion in China,”124.

[17] McLean, The Art of the Network.

[18] SNA also maps core/periphery structures, centrality, density and integration but such measurements apply to positions and ties in social space which may include but are not limited to geographic space.

[19] Lemercier, “Analyse de réseaux et histoire,” 111.

[20] On the different approaches towards history in history and in sociology, see, for example, Skocpol and Somers, "The Uses of Comparative History in Macrosocial Inquiry."

[21] McLean, The Art of the Network; Mohr, "Soldiers, Mothers, Tramps and Others," and Mohr and Duquenne, “The Duality of Culture and Practice:  Poverty Relief in New York City, 1888-1917.”

[22] Sewell, Logics of History, 350, 370; also 78, 346-72. Sewell takes a more critical stance towards mechanistic and positivist types of social analysis and that includes in his view network analysis. Sewell's critique of network analysis appears to be based on more mathematically oriented work and does not appear to be informed by recent trends in the broader field of social network analysis. Ibid., 60.



Cartier, Carolyn. "Origins and Evolution of a Geographical Idea: the Macroregion in China." Modern China 28, no. 1 (2002): 79-142.

Emirbayer, Mustafa. "Manifesto for a Relational Sociology." The American Journal of Sociology 103, no. 2 (1997): 281-317.

———, and Jeff Goodwin. "Network Analysis, Culture, and the Problem of Agency." The American Journal of Sociology 99, no. 6 (1994): 1411-54.

Faure, David, and Helen F Siu, eds. Down to Earth: The Territorial Bond in South China. Stanford, Calif.: Stanford University Press, 1995.

Gould, Roger V. "Uses of Network Tools in Comparative Historical Research." In Comparative Historical Analysis in the Social Sciences, edited by James Mahoney and Dietrich Rueschemeyer, 241-69. Cambridge: Cambridge University Press, 2003.

Lemercier, Claire. "Analyse de réseaux et histoire." Revue d’histoire moderne et contemporaine 52, no. 2 (2005): 88-112.

Lewis, Mark Edward. Writing and Authority in Early China, Suny Series in Chinese Philosophy and Culture. Albany: State University of New York Press, 1999.

McLean, Paul Douglas. The Art of the Network: Strategic Interaction and Patronage in Renaissance Florence, Politics, History, and Culture. Durham [N.C.]: Duke University Press, 2007.

Mohr, John W. "Soldiers, Mothers, Tramps and Others: Discourse Roles in the 1907 New York City Charity Directory." Poetics 22 (1994): 327-57.

———, and Vincent Duquenne. "The Duality of Culture and Practice: Poverty Relief in New York City, 1888-1917." Theory and Society 26 (1997): 305-56.

———. “Network Methods, Historical Texts, and the Formal Analysis of Institutional Systems.” Paper presented at "The 162th Annual Meeting of American Historical Association," Chicago, US, January, 5-8, 2012.

Robins, G.L., and P. Pattison. "Interdependencies and Social Processes: Generalized Dependence Structures." In Models and Methods in Social Network Analysis, edited by Peter J Carrington, John Scott and Stanley Wasserman, 192-214. Cambridge: Cambridge University Press, 2005.

Scott, John. Social Network Analysis. A Handbook. Newberry Park, CA: Sage, 2000.

Sewell, William H. Logics of History. Social Theory and Social Transformation. Chicago: The University of Chicago Press, 2005.

Skinner, G. William. "Mobility Strategies in Late Imperial China : A Regional Systems Analysis." In Regional Analysis, edited by Carol A. Smith, 327-64. New York: Academic Press, 1976.

———. The City in Late Imperial China. Stanford: Stanford University Press, 1977.

———. Marketing and Social Structure in Rural China. Ann Arbor, Mich.: Association for Asian Studies, 2001.

Skocpol, Theda, and Margaret Somers. "The Uses of Comparative History in Macrosocial Inquiry." Comparative Studies in Society and History 22, no. 2 (1980): 174-97.


  • There are currently no comments

New Comment

required (not published)

Recent blog posts

International Medieval Congress 2015 by mchu, July 30, 2015, 3:11 p.m.

Team members Hilde De Weerdt, Chu Mingkin and Julius Morche contributed to the panel “Historical Knowledge Networks in Global Perspective” more

MARKUS update and new tools by hweerdt, March 12, 2015, 6:38 a.m.

The MARKUS tagging and reading platform has gone through a major update. New features are more

Away day for the "State and society network" at LIAS by mchu, Dec. 5, 2014, 12:40 p.m.

Team members Hilde De Weerdt, Julius Morche and Chu Ming-kin participated in the Away Day of the “state and society more

See all blog posts

Recent Tweets