This project has manually transcribed all available census records and each individual entry in the city directories and tabulated this information in a spreadsheet, and these individuals’ first and last names and occupations in the census records have been matched with the corresponding records listed in city directories, in order to obtain address information and append it to the wealth of data present in the census. Adjustments were made throughout the matching process to ensure high quality matches even in the absence of perfect links. When occupations could be linked up and names in the census also matched phonetically to names in the city directory, these records were assumed to be properly linked.
In the spirit of staying true to historical realities, this work also gives extra attention to capturing the details at each location. Existing literature has suggested a method of aggregating wealth across each same-household family into one “economic unit” for the purposes of counting wealth. This approach is especially reasonable because wealth occasionally was divided up among family members, and so looking only at the heads of household ignores the comprehensive financial wealth and real estate holdings of a family. Likewise, this research also combines the wealth of all family members and tabulates it as a single variable attached to the family record, in order to avoid the problem of undercounting family wealth that may have been spread among several members. Those who are not heads of household were identified in the database as additional rows associated with each head of household record, so that they could be studied further, although each family is represented on the map only once for these particular quantitative ratio variables. The full dataset was then exported as a CSV file for use within ArcGIS Desktop.
There are additional reasons that families, rather than individuals, were chosen as the data points. Primarily, this method helps to mitigate the issue of children and dependent family skewing distributions. While some individuals did live alone, the typical household size in the mid-nineteenth century was greater than 6 people, usually members of the same family living together. It would not be logical to analyze clusters and differences across space using individuals as the data points because most individuals were dependents who were not choosing their locations. This potential problem could be considered a major flaw with any simple analysis of population counts in general, but the larger problem of misrepresenting housing choice would be even more exaggerated at the finest level of detail.
In order to reconstruct each city’s street grid and orientation properly, this research synthesized all available accurate historical base maps for Washington, Omaha and Nashville and georeferenced these maps in order to register them to contemporary GIS systems. A street centerline shapefile subsequently was digitized from the georeferenced base map images. Using city directories’ descriptions of house numbering systems (usually the “Street Directory” section) as an initial point of reference, individual line segments in the centerline shapefile were coded manually with starting and ending address numbers for each side of the street, which is intended to correspond to the historical address numbering system.
All records containing address information are geocoded using this customized address locator service, generating a point shapefile containing every applicable record. However, some exceptions were necessary due to incomplete data. Omaha and Nashville revealed very common use of nearest intersections or even a generic location “between” two streets. Details in the georeferenced historical map also helped to provide clues about the location of these features. All other points are offset from the centerline file, placed at their most probable locations along the block. Where it was possible to match records precisely to an associated structure, the points were edited and relocated manually using street directories.