The Spatial Analysis of Chinese History and Society

Organizer: Peter Bol, Harvard University

Chair: Jianxiong Ge, Fudan University, China

Discussant: G. William Skinner, University of California, Davis

Geographic information systems (GIS) are giving new importance to geography by providing researchers in social sciences with the practical means of carrying out sophisticated spatial analyses of their data. Online utilities (e.g. GoogleEarth) are making basic GIS procedures available to the general public; new free browsers (e.g. ArcExplorer 4) allow desktop users easily to use spatial datasets. China in Time and Space, based on 1990 data, was the first China GIS. ArcChina has more data but at high cost ($3000).

  The challenge for a GIS that serves the social sciences is to link the available data with the geographic places (points and polygons) to which the data pertain. Since 2000 the China Historical GIS project has been producing a base GIS for Chinese history (222 BCE-1911 CE, to be extended to 1911-2000) by georeferencing all changes in the administrative hierarchy, thus allowing scholars to analyze historical data in spatial contexts. By the end of 2006 CHGIS will be providing coverage for the territory in which about 90% of the population resided. All CHGIS datasets are freely available for unrestricted educational use. The panel deals with fundamental issues in creating a temporal GIS –the changing ontology of space in China's history (Bol and Berman) and using remote sensing and historical sources (Man) – and applies GIS to analyzing urban migration today (Henderson) and changing sex rations (Lavely).

Peter Bol, Harvard University

Fundamental to a temporally sensitive GIS is the division of space into units that correspond to data with spatial referents. For the first millennium of imperial history these were prefectural and county (xian) level units; for the second millennium, particularly in the south, local historical sources also provide data on settlements and locations below the county seat. Early historical geography is point rather than polygon based; i.e. it envisions the administrative system as a hierarchy of points represented by administrative seats rather than as clearly bounded territory, contradicting the contemporary assumption that administrative space should be represented as territorial expanses within sharply defined borders. Conceptualizations of space changed during the Tang-Song transition but the continuing distinction between household and cadastral registration and between natural villages and local control hierarchies ensured a high degree of spatial ambiguity. There is some evidence, however, that the degree of these spatial distinctions varied regionally as well as temporally, being weaker in the North China plain and stronger in the south. CHGIS allows diverse spatial ontologies and thus a more accurate display of data. Future efforts should give greater attention to settlement points and the bureaucratic, economic, kinship, and religious networks that linked these points together and provided pathways along which state, community, and family interests could be negotiated.

Peter K. Bol and Merrick Lex Berman (JOINT PAPER, Bol is first author)

Zhimin Man, Fudan University, China

The China Historical GIS (CHGIS), as a joint research project of the Harvard Yenching Institute and the Center for Historical Geography at Fudan University, aims to build a basic temporal geographical information system of Chinese history beginning in 222 BC. River courses usually serve as one of the basic elements in the division of administrative regions. So it is important to reconstruct the main river courses as they changed over time before plotting administrative boundaries for CHGIS data. The are three types of evidence for past river courses: 1) Remote sensing data. Land satellite imagery can be used to show differences in land use, usually in the form of differences in ground level. High resolution DEM data can show tiny uprisings in the terrain, such as left by riverbeds and embankments. 2) Historical maps. Maps drawn in the past carry some datable geographic information, thus providing a basis for identifying later changes. 3) Documentary records. Historical documents are both the traditional and the main evidence for reconstructing changes in river courses, providing the "real time" first-hand and secondary accounts. The temporal specificity in such evidence is unique and irreplaceable. GIS provides the working platform that allows us to combine these three types of evidence in an unified geographical coordinate system with a common spatial resolution.

Mark G. Henderson, University of California, Davis

Migration to coastal metropolises has been one of the most remarked aspects of China's economic development in the past two decades, but the phenomenon of migration is more widespread and differentiated than commonly acknowledged. This paper presents spatially disaggregated data from the 1990 and 2000 censuses showing the varied composition of the migrant population and migrants' choices of destinations in the North China macroregion. G. William Skinner first proposed that hierarchical systems of cities within macroregions would be the most appropriate units for the analysis of urbanization in late imperial China, and subsequent studies have applied that framework to other times and features of Chinese society. Illustrating how Skinner's Hierarchical Regional Space model is implemented in a geographic information system, this paper offers a new perspective on the spatial contours of contemporary migration and how it is in turn reshaping China's urban landscape.

William R. Lavely, University of Washington

Literature on rising sex ratios in contemporary China shows secular change and regional variability. This study examines sex ratio variation at the local level to ask if variation reflects contemporary or transient phenomena or is instead rooted in persisting ecological and social structures.

The study uses two unique sources. First is the Jurong County Survey of 1933 led by Zhang Xinyi (Shiban Jurong xian renkou nongye zongdiaocha baogao, 1934) is the best of a very few scientific Republican era demographic surveys. It was in essence an agricultural census with tabulations for all 144 administrative villages. Analyzed as a cross section, these data can provide insight into the economic correlates of sex ratios in a county that was at the time characterized by high sex ratios (the county sex ratio at ages 0-4 was 118 in 1933). Second is the township level tabulations of the 2000 Census with data for 23 Jurong townships in 2000 (with the sex ratio age 0-4 stood at 114).

GIS is used to create a spatially-referenced database comparing the behavior across time of the populations residing in areas demarcated by the 23 contemporary townships. The GIS also provides data on terrain (such as elevations) and changes in transport infrastructure and enables the construction of indicators of change in cropping patterns and outputs over the 67-year period. This allows us to test various propositions about the persistence of local demographic behaviors and the reasons for change.