| In the last decade, historical GIS has exhibited its advantages in dealing with enormous historical data, which was out of reach for traditional historians, even with computers. Major historical GIS projects like the Chinese Historical Geographic Information System (Ge Jianxiong葛剑雄& Peter Bol, 2001), the Great Britain Historical GIS (Gregory, Bennett, Gilham& Southall, 2002), and the USA National Historical GIS (Master & Noble, 2005) provide robust infrastructures that can serve multiple research purposes. These projects have made great efforts to exhibit the spatial dynamics of administrative boundaries of various spatial units combined with demographic and other socio-economic data. Another trend of historical GIS is represented by projects that focus on more specific territories, especially cities, like Virtual Shanghai (http://virtualshanghai.net), Virtual Jamestown (http://www.virtualjamestown.org) or eWilliamsburg (http://research.history.org/ewilliamsburg), etc. These platforms apply GIS to urban history to provide new visions of history and forms of visualization of historical data. Mapping and spatial analysis, including interactive web-mapping, are a major component of these projects.
While the practice of historical GIS depends on various types of data from archival sources, historical maps occupy a preeminent position for geo-referencing and vectorisation of historical data. Maps provide a “symbolized image of geographic reality, representing selected features or characteristics resulting from the creative effort of its author’s execution of choices, and are designed for use when spatial relationships are of primary relevance” (Daniel Dorling & David Fairbairn, 1997). As a particular type of mapping, cadastral maps constitute as especially valuable source, both for their degree of reliability and the wealth of data associated to such documents.
A cadastre is the basic instrument for the registration of landed property. Cadastral records include basic data such as the name of owners, location, area, value, tax base, etc. Research on cadastres has attracted more and more interest among historians in recent years. On the one hand, as an effective means of government monitoring and control of land, the cadastre was regarded as the foundational institution of municipal administration. Many historians have approached the issue of cadastre as a major key to understand local history. Kain, R. J.P. & Baigent, E. (1992) have argued: “we view cadastral maps as instruments for effecting state policies with respect to landed property and for exerting political and economic control over land.”
On the other hand, a cadastre constitutes a very useful additional source for historical research. Owing to the important role the cadastre played in local administration, the literature on cadastres, even those established in antiquity, offers solid references on the sources and history of cadastres. More recently, Peter Ekamper (2010) applied the methodology of Geographic Information Systems (GIS) to link cadastral map data, population census data and population register data for the Dutch city of Leeuwarden in the mid-nineteenth century.
This paper takes the cadastre of French Concession in Shanghai as a case study. The French Concession was a specific area, like its British-dominated counterpart, the International Settlement. Both were established on a territory that remained under Chinese sovereignty, even while foreigners were entrusted with the power to rule these areas where more than 90% of the population was Chinese. For this study and reconstruction of the cadastre, aside from the technical difficulties, the major challenge was to collect continuous and complete data. On the one hand, not all the cadastral maps seem to have been preserved well up to the present. On the other hand, access to such documents remains difficult in the local archives. The cadastral map of 1931 and 1941, and its ledgers for 1932, 1934 constitute the documentary basis for this paper. The period of 1931- 1941 represents a special phase when the Service du Cadastre was reorganized and when the land of the French Concession was entirely surveyed again.
This paper was inspired by Peter Ekamper’s (2010) work on the combined use of historical cadastres and population census data in the city of Leeuwarden in the mid-nineteenth century. In this paper, we examine the structure of landed property in the French Concession in relation with issues of ownership, land value, and population. We argue that the use of a data-enriched historical cadastre offers a unique vantage point to understand the spatial and economic dynamics at work in the French Concession.
2. Historical Background
Shanghai was officially opened to foreign trade in 1843 under the Treaty of Nanking, and the International Settlement (1843-1943) and the French Concession (1849-1943) were built one after another. Under the various treaties signed thereafter, the Chinese government and the foreign representatives adopted the Land Regulations, a text that governed the conditions under which foreigners could acquire land in Shanghai. Since China retained it sovereignty over the leased land, foreigners could not purchase land, they could only rent it “in perpetuity”. These dispositions created a special institution, that of the “Title Deed” (daoqi 道契) by which foreigners paid a “rent” to the Chinese authorities every time they “acquired” a piece of land. On the other hand, the municipal organizations established in the foreign settlements instituted a land tax to support their operations. The title deeds and its plot survey were registered with the Chinese authorities in order to embody the sovereignty China retained on land ownership.
Therefore, there coexisted two parallel cadastral maps: one was the China-sponsored private plot map which was affixed to the title deed to identify the extent and ownership of land; the other was the Settlement-sponsored cadastral map which recorded a number of properties in a large-scale block map. At the beginning, there was no strict land survey or cadastre office in the French Concession in the 1848-1889 period, even if cadastral maps of French Concession were created by employees under the management of the Public Works Department since 1863.
In order to improve the precision of the land deeds and to meet the needs of land transfers, a Chinese Land Office (Huizhangju 会丈局) was established in 1889. The Land office took the lead in land survey in the French Concession, yet with Chinese traditional and inaccurate techniques that influenced seriously the precision of the cadastral map in the period of 1889-1917 (SMA, U38-1-1052) because the data on land area in the cadastral ledger came from the Chinese survey.
With the municipal and economic development and spatial expansion of the Settlement, the power of the settlement authorities to rule over land within the French Concession became stronger and a great deal of the legal rights initially ascribed to the Chinese authorities was weakened or captured by the foreign municipal authorities. Figure 1 shows the spatial expansion of the French Concession from 1849 to 1943. The creation of the Service du Cadastre in 1917 embodied this tendency. Before 1917, the area of private land in the cadastral register mainly came from measure of private land by the Chinese Land office, while after 1917 the work of measuring and mapping private land came under the Service du Cadastre. It introduced a better professional survey organization and modern survey techniques. Moreover, owing to a reorganization of the Service du Cadastre in 1932, the work of land survey became stronger and the precision of cadastral maps was further improved.
3. The nature and content of the cadastre
The data for this study was drawn from the cadastre of French Concession in the period of 1931-1941, and related data retrieved from various official archival sources. The cadastre was composed of two parts: the cadastral map and the cadastral ledgers. The cadastral map included the maps of 1931 and 1941, and each cadastral map was composed of one general map and a succession of detailed section maps indicating the boundary and the number of each land plot, block number, name of the owner, road names, dotted lines for planned enlargement or new constructions, waterways, etc. The cadastral ledgers included the years 1932, 1934. Each cadastral ledger consisted of one table of “Titres” by nationality, which constituted an index of all title deeds with corresponding cadastral number, and numerically-ordered tables with the names of the owner, the number of the title deed (Titre consulaire), the number of the cadastral lot (Lot cadastral), area, land value per mu (mow) and block number.
Generally speaking, the degree of precision of the map was relatively high. First, the cadastral maps were created by trained and experienced surveyors of the Service du Cadastre. After the creation of the Service du Cadastre in 1917, the work of measuring and mapping the land in the French Concession was carried out over five years (1916-1921). The technicians employed the technique of triangular surveying and polygonation, using precise apparatus such as theodolite and chain etc. Second, on 9th April 1936, the chief of the Service du Cadastre wrote a letter to the municipal engineer that described the two different types of calculation used before and after 1917. The method of calculation before 1917 was provided by the Chinese land Office (SMA:U38-1-1052 p. 9):
The terrain was divided by many quadrangles and triangles, we obtained the area of a quadrangle by multiplying the half summation of two opposed sides by the half summation of other two sides, and the area of a triangle equaled to the half of multiplying the two sides of right-angled.
The result so obtained was naturally inaccurate. After 1917, the terrain was divided into triangles and all the area was calculated by triangle, so it was more precise than before. Third, after the reorganization of the Service du Cadastre in 1932 all the land in the French Concession was remeasured, with all the changes of combination or division presented on the cadastral maps of the different periods.
As to the precision of the cadastral ledgers that relied on the data from the title deeds, it was also very high. First, after the creation of the Service du Cadastre in 1917, the work of measuring and mapping the private land parcels was carried out in a very thorough way. In his report “Rapport sur la précision des opérations topographiques du cadastre” (SMA:U38-1-105 p. 33-81) the municipal engineer stated that:
“Land under title deeds was drawn by our most experienced operators using a Zeiss or Wild theodolite and chain ... ... to establish one or more title deed plans, we need to draw the entire block that contains them. The errors are those encountered in the general survey of the Concession, yet with the difference that controls are more effective here. Since the plans for several title deeds must fit perfectly, the operator performs his survey so as to have several overlap; if the report onto paper does not fit, it is enough for him to immediately return to the field to remeasure the questionable measures. ”
Yet the limitations of the cadastre should also be noted for two aspects: the precision was relative because errors were generated in the process of land survey and map drawing. The reasons should not all be ascribed to technique, but to the special circumstances and topography of Shanghai. As noted by the municipal engineer (SMA U38-1-1052 p. 33-81) more than one third of the land was covered with creeks that flooded low-lying terrain in times of heavy rain, not an unusual condition in Shanghai. This resulted in increased difficulties for measuring land. Another issue was the coexistence of land under Chinese land deed (fangdan 方单), land under the title deeds from the various foreign consulates, land without registration, etc, which was difficult to manage. Moreover, the property of the land covered with creeks and public paths added an element of complexity because not all the creeks were privately-owned possession. Some were possessed collectively by several persons, while others were public. So were the numerous Chinese paths. These particular conditions resulted in some confusion about parcel boundaries. Over time, owing to the increasing function of the Service du Cadastre, these questions were dealt with step by step in the 1932-1943 period, in part because most of the creeks and paths were filled in and turned into new roads. This process made it easier to determine parcel boundaries.
4. Geo-referencing and vectorisation of the cadastral maps
When we got the original cadastral maps, the first task was to produce raster maps by scanning the printed maps, then georeference the raster maps into a map coordinate system, in order to be combined or integrated with other geo-referenced maps. After digitizing and georectifying the original maps, the main task was the conversion of the geographic symbols into vector objects like points, lines and polygons. For the cadastral map, most of the spatial data in the parcel could be expressed by polygons, while lines were used to represent the roads and rivers. Given the degree of detail included in cadastral maps, the vectorization of spatial data was a highly time-consuming process. We used ArcGIS to draw and store the different types of geographic data, such as land lot, roads, rivers, etc. as separate layers.
The workflow for the successive operations is illustrated in Figure 2. Figure 2 shows the digitizing of the original cadastral maps of 1931, and illustrates the different types of objects, such as parcels (polygon), roads (line), river (line) stored in different layers in ArcGIS. In addition, of course, the data from the original registers was imported into a database that mirrored the original fields of the register, with the individual land lot as the basic piece of data. The content of the database was injected into the attribute tables of the map layers in ArcGIS.
5. Cadastres of the French Concession: a preliminary analysis
Once we had transformed the cadastral maps of 1931 and 1941, we could overlay the two layers in the same map coordinate system and make some preliminary analysis. Generally speaking, the existing historical studies concerning the real estate ownership focused mainly the general developing history of the real estate, were concerned with the institutions of the real estate, such as various land regulations, the land deed (Chen Zhengshu陈正书,1996) and related institutes or companies(Chen Yanlin 陈炎林1933), or linked with the influence of the real estate, such as the urbanization, the landscaping, or social variance (Guo Qizheng, 2004). These general historical studies help to pave the way for a basic understanding of the cadastre. However, most previous studies about the real estate have been qualitative analysis, which only describes the phenomena recorded in the historical literature and other secondary sources. And this method cannot do proper justice to the study of the various spatial components of the real estate, which has omitted to consider by the historians but was key to understand the history of social spatial process evolving over long and different periods of modern shanghai.
First, we could illustrate the changes of individual parcels in the period from 1931 to 1941. Owing to frequent land sales, the area and figure of the lots changed considerably between 1931 and 1941. The two basic processes were the merging and the division of land lots. In modern Shanghai, the process of merging occurred more frequently in the earlier period of urbanization, while division took place in the period after the completion of urbanization. In the period studied in this paper, urbanization had been accomplished and the division of lots was more frequent.
The spatial changes of land lots that intervened between 1931 and 1941 are exhibited in Figure 3. Firstly, it is apparent that the distribution of the lots that were divided or subdivided was linked to the process of urban expansion of the French Concession: the division of lots was more pronounced in the 1914 extension of the French Concession than the old French Concession before 1914 because it was a new urban development where land was exploited after 1910. Second, the division of lots was mainly located along the roads, especially the newly-built roads or the roads planned to be built. On the one hand, a number of lots were divided into two or more parts by road building itself; on the other hand, road building accelerated changes in land use and raised land value. As a result, landowners found it beneficial to divide or subdivide their lot for sale after the building of new roads. One can also see, of course, that a few parcels were particularly large and were inevitably prone to being split into small properties under urbanization.
Second, land value is considered by researchers as an important medium with which to assess the extent to which land is exploited and the forms of land use. Land values in turn can also exert a great influence on changes in land use. The general historical studies on the land values of modern Shanghai focuses on the land price (Zhanghui,1935;Ma Xueqiang,1999) and its influence exerting to the urbanization(Wang Shaoqing, 2009) or other social questions. These studies want to show the basic temporal changes in various periods, while the spatial changes have been omitted. The land value of each parcel recorded in the cadastral ledgers of the French Concession in 1932 provides rich data for the reconstruction of the spatial dynamics of land value.
Figure 4 shows the spatial distribution of land value in 1932. Land value followed a general descending curve from east to west. The curve reflected the increased land value that resulted from the more intense urbanization and commercialization of the land located in the inner core and close to the Bund. The same pattern was replicated in the International Settlement. Secondly, land value was connected with infrastructures, just as Figure 2 has shown. Land values were higher along the major commercial thoroughfares. A clear indication that urban infrastructures could be regarded as one of the most important factors that drove the increase of land value.
Third, the land owner was another factor that links with the development of urbanization. In the French Concession, the exploitation of land was determined by the foreign land owners. Because the municipal administration of the French Concession was mainly composed of foreign land owners and it was easier for them to influence municipal policies on land, they could induce the government to build the basic facilities and services in order to increase the value of their land. We established in a previous study (Mou Zhenyu, 2010) that the place where foreign landowners concentrated was most likely to be exploited and benefit from municipal infrastructures. Therefore, it is helpful to understand the process of urbanization in applying GIS to recover the nationalities of landowners.
The data on real estate in the cadastral ledgers of the French Concession consisted of not only the landowner’s name and lot number, but its corresponding number of title deed. Since all foreigners who purchased land from Chinese had to register with their consulate and get the title deed issued by the Consulate, each title deed was given a unique number in the cadastral map. For example, if the title deed was issued by the British Consulate, it carried the mention B.C. for “British Consulate” lot. Therefore, the nationality of the landowners can be known by the title deed.
The distribution of land property by nationality is shown in Figure 5. By a simple query in ArcGIS, it is possible to map out the different nationalities in different colors. From Figure 5 we can note that several nationalities coexisted in the real estate market of the French Concession, although the most important landowners were French, British, and American. The spatial distribution of French real estate was distributed most broadly and was more extensive throughout the entire territory. Obviously too, the French had a total monopoly on land in their settlement in the early phase, before the 1900 extension. Not a single lot was acquired by other nationals. British landowners favored lots in the 1900 extension, in the inner core with good infrastructures, along commercial thoroughfares, in particular along avenue Edward VII. American real estate was quite widely distributed, but it concentrated mainly in the suburban area, especially near Route Ghisi, Avenue Petain etc. We shall examine whether this was related to certain types of real estate (apartment buildings, clubs, etc.). Chinese real estate was also widely distributed and located mainly in the suburban area. From the small size of Chinese real estate properties, it appears that for the most part it reflected places where once old villages had stood.
The single thematic maps in the previous sections were produced with the basic functions of GIS for spatial analysis. In order to examine the characteristics of the spatial distribution of real estate, we applied multiply spatial data analysis. Overlay operations are the basic methods of the multiply spatial data analysis of GIS. These operations combine several raster maps in the same coordinates and produce new information that was not present in the single layers. In overlay operations, we can establish new spatial elements on the basis of multiple input maps.
The population data of the French Concession in 1935 and 1937 held by the French Concession included the block numbers, the number of Europeans and Chinese residents in each block, and the amount of commercial shops of each block. Figure 6 shows first that there was a positive correlation between the density of population and commercial shops, especially for Chinese residents. In other words, where there was more population, there were more commercial shops. Second, land value had a positive correlation with population and commercial activity. Land values were higher where there were more commercial shops and population. Third, the real estate owned by foreigners from various nationalities had a correlation with the spatial distribution of European residents, but the correlation was not absolute. For example, there were few foreigners living east of Route Père Robert where foreigners owned a great number of properties. Owing to the high density of the population of Chinese residents and the numerous commercial shops, land value there was higher in the eastern part of the settlement.
These preliminary results provide further evidence to the acknowledged views expressed in recent historical studies that suburbanization foreigners in modern shanghai from 1930 to 1940. Guo Qizheng (2004) studied the social production process and the transformation of alleyway houses in the suburban areas of Shanghai, which was driven by the authority of the French Concession and by foreigners. From F.6, we can see that the inner core where land value was high was mostly inhabited by Chinese residents, while foreigners resided mostly outside of the core of city where land value was not as high, in the central section of the settlement, and some of them in the suburban area. Second, the trend of suburbanization was also testified by the active changes (division or merging) of lots located mainly on the edge of the city core and in the suburban areas. The changes of lots embodied to some extent the activities of the land. It could be concluded from the spatial distribution of commercial shops that there was a long period for suburbanization because there were still far less commercial shops in the suburban area than in the inner core of the settlement by the early 1940s.
6. Summary and discussion
Historical GIS provides a new methodology for historical research to investigate and exhibit the spatial dynamics of various topics like population density, land value, etc in urban history. For historians, the difficulty of historical GIS is not so much using the methodology of ArcGIS than how to convert the historical sources into digital data and how to find the exact location in today’s geographic coordinates. Compared with other types of historical maps, cadastral maps are comparatively exact, especially after the 19th century due to the use of modern techniques of land survey and modern cartographic methods. Thus cadastral maps can become a useful source in historical GIS. Researches of cadastral maps focus on the geovisualization of specific topics recorded in cadastre registers so as to exhibit the spatial change of real estate, land lots or land value. On the other hand, it is necessary to use multiply spatial analysis with additional socio-economic data to make the cadastral map tell a richer story.
In this paper, we used the cadastral maps of the French Concession as an example of historical GIS. We established the particular spatial distribution of land lots among the various nationalities and exhibited the characteristics of the spread of land value as well as the spatial changes of individual lots. Finally, we explored the spatial connections between land value, ownership, population density, nationality, and density of commercial shops. This example highlights the need for accurate and complete historical data and how this plays a decisive role in the process of the historical reconstruction in historical GIS.
There remains much work to be done on the cadastral maps because there exists a large number of original land deeds with a full record of land transfers since the first acquisition in correspondence with cadastral lots. One of the next steps will be to combine the data from land deeds and to reconstruct for every lot the timeline of its change over time, a highly time-consuming process. Yet this will allow us to explore more fully the social history of land use in the urban space of modern Shanghai.
I would like to thank Prof. Christian Henriot for his constructive and helpful comments and patient correction of this article. We gratefully acknowledge the contribution of the Virtual Shanghai Project that made all its GIS data, maps and other sources available to us to achieve the present study during our visitorship at the Lyon Institute of East Asian Studies (Institut d’Asie Orientale) in 2010–2012.
We gratefully acknowledge the contribution of the Virtual Shanghai Project
that made all its GIS data, maps, and other sources available to us to achieve the present study during our visitorship at the Lyon Institute of East Asian Studies (Institut d’Asie Orientale) in 2010-2011.
Land value data for the 1932 is missing for the no. 1-1000 parcels.
Different from other region in Shanghai, all the Title Deeds of Old French Concession before 1900 were registered in French Consulate, so F.4 shows all the land belong to French, in fact, there existed small part of foreign properties of other nationalities and Chinese properties under the French Title Deed.