Reconstructing the geographical framework of the 1901 census of Canada.

he 1901 census of Canada, both in its original published form and in the sample created by the CanaT dian Families Project (CFP), offers very rich potential for geographical analysis of Canadian families. The reason is quite straightforward. When one examines documents of the 1901 census, the centrality of geographical space seems to permeate all of the surviving manuscript records and each of its printed volumes. Whether enumerating individuals or publishing results, census officials paid particular attention to geographical units. Their goal was to gather and disseminate population and economic information for provinces, cities, and census districts, as well as for smaller geographical areas such as towns, villages, and census subdistricts. In fact, the precise locational reference of any dwelling (its street address or legal property description), and therefore of any family occupying the dwelling, was carefully recorded in 1901 for the first time in Canadian census taking. Although census officials did devise a geographical framework for the census, it is no easy task to reconstruct the spatial parameters of this historical document. More important, the published volumes of the census, unlike later and even some earlier censuses, offer no cartographic evidence of the boundaries of the statistical units used to administer and report census results. Other than the locational references written on manuscript census schedules, little tangible evidence remains with which to retrace the actual boundaries and areas referenced and canvassed by census enumerators. Without a geographic& framework, it would be impossible to analyze the data sampled by the CFP when the purpose is to interpret the spatial processes, patterns, and structures associated with family life in Canada’s cities and rural areas. To meet this goal, methods have been devised to reconstruct the boundaries of the various types of statistical units that census officials used when gathering and publishing census information. These units include census districts, subdistricts, polling subdivisions, and urban places. Using areal units of varying size will, of course, influence the geographical interpretation of family life. To illustrate this fundamental fact, we have mapped, at different geographical scales, the patterns of average family size that characterize the social space of Montreal and its rural hinterland. With a population of 267,730 in 1901, Montreal was Canada’s largest city. This total does not include the population of twelve contiguous suburban municipalities that were incorporated as villages, towns, or cities and housed nearly 85,000 people. We make no attempt to explain the effect of scale differences on the revealed map patterns, other than to reaffirm the assertion of Fernand Braudel (1984, 21) that “geographical space as a source of explanation affects all historical realities, all spatially-defined phenomena.”