Martina De Moor (Ghent University)
Now 10 years ago, in 1990, a project called the Quantitative database of Belgian Municipalities was started up as a pilot project, at the Department for Contemporary History at Ghent University in Belgium. The project was funded by Ghent University, the Fund for Scientific Research and a few federal services, and was carried out under the guidance of Eric Vanhaute. Sven Vrielinck has been working several years on this project together with several other part-time and short-period researchers. In total about 8 to 9 full-time equivalents worked on the project the past 10 years. I was asked to take care of the further follow-up of the project, a kind of manager.
Due to limited financing it has taken a while to reach the initial goals but we're convinced that the final results are worth to have a closer look at.
In the beginning, the project aimed at digitising the official census figures of the 19th century until 1970. The National Institute for Statistics gathered and processed a treasure of basic statistic materials ever since the foundation of the Belgian state in 1830. The most important series are based on the ten-yearly census takings of the population, agriculture, industry and trade from 1846 onwards. The NIS published all of these censuses. Not only do these periodical statistics bear witness to a high degree of validity and reliability, the data can also mostly be traced back to the municipal level, which was a conditio sine qua non for this research project. The importance of these sources - From a historical point of view but also from a present-day perspective -contrast however sharply with the serious difficulties in consultation of the publications. By digitising the statistics we hoped to improve their accessibility. For the period before 1846, there are no published results available. We had to resort to partial copies preserved in a range of provincial and state archives in Belgium, France and the Netherlands.
Something more about the municipalities in Belgium, before actually talking about the project.
Since their foundation in 1800 we can see 3 waves in the evolution of the number of municipalities in Belgium. Firstly during the French period, and during the Dutch period as well, the government strived at reducing their number. Around the foundation of the Belgian State we see a turning point: Plans for merging municipalities were put aside; and even several hundreds of new municipalities were added. Until the second half of the 20th century, the number of municipalities remained more or less stable. Nevertheless, many boundaries were changed during this period. Most of these changes occurred during the first decades of the 19th century as a result of the introduction of the French revolutionary land registry, called the "cadastre parcellaire". In order to set up the new registry, each boundary had to be defined before the new land system could be introduced.
During the years 1964, 1970 and 1977 the number of municipalities was dramatically reduced. Small communities were absorbed into larger units. The number of municipalities dropped in 20 years from 2663 (1963) to 589 (1983). Due to these changes and to the fact that data and maps are available at the NIS from the 1970s onwards, we have decided to stop the period of research in 1970.
So, our main goal was to collect data on the level of these municipalities, taking their changing boundaries into account. The database that would thus be developed had to be user-friendly and readily accessible by other researchers, allowing them to make long term and diachronic analyses. We thus needed to develop a data system suitable for storage, management and consultation of the statistics. The structure of the official statistics was such that it allowed us to use a simple matrix. In a second phase we decided to make reconstruction maps in order to visualise the statistics. We expected this to be a rather easy tasks that could be done in a short period of time. It turned out differently, however. The data on boundary changes the NIS had, turned however out to be incomplete and often incorrect. So, we had to change plans; more research had to be done on the boundary changing part than initially expected. In fact, so little was known about the actual boundary changes on the municipal level, that a lot of time had to be spend in the archives. Because of this change of plans, the eventual outcome of the project was much more than "just" a database of Belgian Statistics on the municipal level. In fact, we ended up with all the facets of a Geographic Information System.
The database is composed of provincial spreadsheets and maps. Although, the town or municipality is the basic unit of analysis in the database, we chose the province as a level for amalgamation of the statistical data and maps. Creating such a level of amalgamation was interesting because it would make the composition of maps later on much easier. Many macro-studies on 19th-20th century subjects use the province boundaries to demarcate their area of study. Linkage of external data to our maps and statistics would thus become easier. Moreover, files on a provincial level would not be too large and thus easily manageable. Considering the fact, that storage capacities were still limited at the beginning of the 90s, this was not an unimportant reason to keep the data separated in several files.
The spreadsheets were build up out of two types of data. In the first 12 columns, you'll find data that describe the spatial location of the statistical data that follow in the next colums.
In the spreadsheets the first columns refer to the province where the data were collected. In the next columns the administrative districts (arrondissements), the serial number of each geographical unit and the official code of each municipality and district as published by the NIS were entered. Furthermore we had to introduce a cartographic code which links each area in the spreadsheet to a polygon on the digitised map. One can find for example: the indication whether the municipalities carried the legal title of town or city, a variable that indicates the type of record, the year of foundation and/or abolishment of the municipality, the years of boundary corrections, the legal name of the municipalities and earlier and unofficial names in their most frequent variations. As you can see, the spreadsheets have in the end become much more than simple digital copies of the national censuses.
The oldest information used is the census of 1796, when the revolutionary French government decreed the first national census for the entire territory that later would become Belgium. It would have been a too risky and endless operation to go back further in time while holding on to the municipal level. The information on municipal boundaries is too defective and too unreliable to obtain comparable lists on a national or even on a provincial scale. The information in the database ends in 1970. It can be guaranteed that the information between the terminal dates of 1796 and 1970 is comparable; the requirements for a diachronic database have been met. It is now possible to make statistical comparisons over a period of 170 years, unaffected by the many changes in the geographical units. Due to the project, demographic statistics now leap back nearly half a century in time.
As I already said, The NIS has its own database of the statistics that were published after 1970. We are planning to link our database with theirs. Such a linkage would allow future researchers to work on data sets with statistical data of more than 200 years. I presume I do not have to stress here which enormous possibilities for long term diachronic research on all levels varying from the local to the national level, will thus be created.
Basically, a GIS consists out of 2 major components: a database/spreadsheets and maps. So, in a second phase of our project we started composing a large set of maps that can be linked to the database. We started with a base map of 1977, bought from a firm specialised in GIS. This map was used as the master map: on this map the thousands of boundary changes of the municipalities between 1800 and 1977 were put in. For each province we made a collection of maps, on the basis of the principle that whenever boundary changes occurred, a new base map was made and stored. The maps are vector-oriented and map up about 82,000 reference points indicating the boundaries between the towns and villages. For 1977, for example, we're talking about 2600 towns and villages. As projection we used the Belgian Lambert-projection.
For each province we thus have now a collection of maps. Whenever one wants to make a map for several provinces or for the whole of Belgium, one has to make a composition of a selection of files on the bases of a concordance table. The concept is rather simple but very useful. A concordance table allows us to reconstruct easily, maps for any year between 1800 and 1977. When the maps are composed, the statistical data can be linked and several types of maps can be constructed.
When discussing maps, databases and GIS, I should put in a word on SOFTWARE as well. During the project programs such as Plan Perfect, Quattro Pro, Lotus, SPSS, Atlas Mapmaker, Atlas Draw, Atlas Graphics were used. From our point of view now, these programs are outdated. When starting up the project we had to shift with what technology offered at that time. Historians regard a period of 10 years as very short; for computer scientists however, a period of 10 years is equal to eternity. When rereading the texts of the workshop of 1994 held in Firenze, one realises how fast technology has developed during the 1990s. Ten years ago, Geographic Information Systems were still in their infancy and hardly ready for easy use by historians. For example Mapmaker is a kind of premature GIS-program. Considering the state of GIS ten years ago, we probably chose one of the most suitable programs for our goals at that time. It's easy to use -even for computer illiterate historians- and it gives nice results without much foreknowledge. It allowed us to make maps linked to the statistical data, and for us that was precisely we expected the software to do, at that time. At the moment we are converting the data into Arcview in order to make them more easily accessible and exchangeable. Conversion of one GIS-package to another, is probably one of the most difficult problems a lot of long-term projects such as ours are facing. The data collection period takes a long time due to archive research. In the meanwhile one has to keep track of the developments in software. We have been lucky to start at a moment when GIS was in full development, otherwise our results would have probably be lost due to conversion problems. Luckily, we chose for commercial software at that time and we didn't -or couldn't - start programming ourselves. If we would have build our own software package, we probably would have been confronted with enormous problems now. Apart from this, building a well-thought out model has prevented that we're facing more serious problems than conversion. The next challenge for GIS now is to incorporate a new dimension, the time dimension. The historical approach -and certainly the approach of researchers working with quantitative data- is almost by definition a long-term approach. With our concordance table we have tried to avoid this problem. Making the database more user-friendly implies that this concordance table will have to be integrated one way or another into GIS, making querying as easy as possible.
A second important result of the project is 3-volume book that will appear shortly. The book deals with the territorial division of Belgium (1795-1963). It is in fact an administrative-geographic and statistical repertory of the municipalities and supra-communal units (kantons, arrondisements and provinces) of Belgium. In fact this extra "spin-off" of the project was not planned from the beginning. The information in the book is however so vital for any study dealing with geographic units that it we found it more than worthwhile for publishing.
The repertory gives among other things the history of several administrative and judicial units, such as kantons, arrondissements. Several aspects (such as the foundation, composition, changes of boundaries, number of inhabitants, surface) are dealt with, including laws and decrees referring to changes of one of these aspects.
The Repertory also includes an overview of all municipalities which have existed for the period of research. The repertory allows us to locate all of these Belgian municipalities within the larger administrative and judicial units. In the repertory one can find when changes occurred.
It furthermore includes lists of old and modern names of the community SLI DE
a chapter on the laws and decrees concerning the territorial division of Belgium and the figures of all censuses for all municipalities for the period 1796-1961. The book is a repertory and thus hasn’t focussed on answering any fundamental research questions in the field of administrative geography.
It was never our goal to work on the actual meaning of boundary changes within for example the formation a nation,.. We only wanted to offer other researchers a tool which would allow them to work more efficiently. Many other researchers of all fields of historiography and geography have already used the database and especially the maps for their own research.
So far, researchers could however not consult the data themselves. The construction of the database is rather complex and logically historians do not always have the technological know-how to work with GIS. Therefor, we are at the moment trying to make it more easily accessible and user-friendly for historians. The eventual output of this process will most likely be a CD-ROM and access via the internet. At the moment we are also developing a centre for historical cartography and GIS. It is no secret that historians are always in need of a good map for a certain area and a certain period. More and more they want these in a digital format. Though several maps are available in digitised format -for example those of the Quantitative Database- historians can't always find these. The maps are scattered over several institutes and universities and -more importantly- due to intellectual property rights the authors of the maps are not always willing to distribute their data. Consequently, those results end up somewhere in a university cellar without ever being used again. The centre wants to prevent this by bringing together supply and demand of digitised maps via the construction of a meta-database on digitised cartographic and spatial data for the area of Belgium and surrounding areas that were closely related to the Belgian territory, for example Sealand or the North of France. The quantitative database will be the first complete data set in this meta-database. Via the meta-database we hope to accumulate all information on digitised historical maps and related digitised statistics and inform other historians which maps can be found -including their main characteristics such as format, software used, unit of analysis, accuracy, …- and under what conditions they can be used, and so on. This meta-database will probably be modelled on an already existing meta-database made by the Flemish government. Furthermore the centre aims at digitising maps depending on the demand of historians. But right now we're still in the phase of looking for funding.
To summarise briefly the history of the Quantitative database of Belgian municipalities. During the period of 1990-1992 most time was spend on a pilot project for an area around Ghent. The next two years we have extended this area to the whole of (present) Belgium. In the period 1995-2000 we've been working on some specific applications, the repertory and we've made the first steps in the direction of an easily accessible and user friendly system for researchers. The next step might be integration into a database on the European Level.
© Martina De Moor April 2000