A Work in Progress
I want to reconstruct the historical racial demography of New Haven with the longer-term goal of connecting it with patterns in Medical Institution of Yale College student housing and the location of institutions like the hospital. (The Medical Institution has since become the Yale School of Medicine.) Today, housing in New Haven is significantly segregated by race but that history is often told from the perspective of redlining in the twentieth century.[1] Redlining, however, reflected existing patterns of racial segregation that had emerged in the nineteenth century, so reconstructing the city’s nineteenth-century spatial demography will reveal when and how these patterns developed before the twentieth century.[2]
Reconstructing early nineteenth-century spatial demographics is challenging. Modern spatial demographics are relatively easy to lookup and map. Today, federal U.S. census data is divided into census blocks which are designed to be as fine grained as possible without revealing individual households’ statistics. Census blocks may not contain any population at all, but an average of 51 of them make up a census block group (in 2010) which includes 600 to 3,000 people. However, census block data was only published starting with the 1940 census, and the Census Office only began supplying data for areas under 50,000 people from the 1870 census.[3] In the period I’m interested in, roughly 1810-1860, census data was only tabulated at the county, and sometimes the city or town, level. Though the census schedules showing individual household data (and all free individuals from 1850) are now publicly available, these records are not geographically located in any more detail, meaning no street or address information is included.
So, how can we reconstruct nineteenth-century demographic patterns within cities and counties?
City directories (precursors to the telephone books of the twentieth century) very helpfully locate people spatially. They include people’s addresses, sometimes both home and work ones, and they often include personal information such as people’s professions and even how they were racialized. The drawback to these sources is that they are much less comprehensive than censuses. Nevertheless, we can use city directories to locate a significant number of people and compare who is included with the overall census numbers to get a sense of who is missing.
For this project, I selected the New Haven directory from 1841 because it was one of the earliest directories for the town, was in the middle of the period I’m interested in, and was only one year removed from a federal decennial census that could be used for comparison.[4] The 1841 directory includes people’s names and, depending on the individual listing, professions, addresses (both home and work), as well as labelling some people “col’d.” Assuming “col’d” means “colored” and represents historically assumed racialized categories, mapping the addresses of the people labeled with it will reveal patterns of residential racial segregation.
Part of a page from the 1841 directory. Note John Duke and Miss Elizabeth Dunn listed as “col’d” and the variety of address formats (those preceded by ‘h’ are home addresses).
However, mapping the addresses in the directory is not straightforward. The directory lists addresses in a variety of ways, including a street number, at cross streets, on a street between two others, on a street (with no other details), at a particular building, etc. Also, a quick comparison between modern and historical maps reveals that street layouts, names, and numbering schemes have changed over time. So to map people in the directory, historical streets and addresses must be reconstructed.
This is the 1868 Beers map of New Haven georeferenced to modern road data. You can see with the exception of the freeways and their onramps that the modern road grid significantly matches the 1868 grid.
To create the historical map, I combined data from the modern road network, contemporary maps, and descriptions of streets from the directory. Road networks tend to be static outside of major changes because moving a road involves moving everything around it, so modern road data was used as a basis for the historical road network. The road network was reconstructed in ArcGIS Pro, by removing, adding, renaming, and altering modern GIS data according to historical maps, including a rough map included in the 1841 directory.[5] The directory also helpfully listed streets as having a start and end point, indicating where numerical addresses began and which direction they increased in.
The 1841 directory helpfully includes where streets stop and start, implying the direction of their numbering. However, how the street numbers increment is not specified.
However, these street descriptions do not include the number range per block, which varies by street and block. To locate addresses given as numbers on a street, I mapped the numerical addresses of buildings listed as on or near another identifiable location, such as an intersection. This creates a range of address numbers for each block, and numbers of buildings not listed in relation to a landmark can then been interpolated between the locatable ones. To approximate street numbers for roads without any locatable addresses, the numbering is assumed to start with number “1” and continue through the highest address number listed in the directory on that street.
I used a physical map to explore how streets were labeled and numbered in the 1841 directory before encoding them in GIS to make an address locator.
Having constructed this mapping infrastructure, the next step will be to actually map the directory data and compare it, where possible, to the 1840 census and to records related to the Medical Institution of Yale College. I plan to make a heat map of New Haven’s “col’d” population, indicating where (if anywhere) people of color were concentrated, and then overlay medical student housing locations to check for any correlation. I am particularly interested to see if the Medical Institution’s few Black students roomed with white students or in potentially more segregated areas. I also plan to map New Haven’s enslaved population based on the 1840 federal census to again look for connections with with medical student housing.
This process is the basis for my Race and Space in Nineteenth-Century New Haven project, conducted as part of the larger Yale School of Medicine and Slavery project.
[1] For instance, “CT Data Story: Housing Segregation in Greater New Haven | DataHaven,” accessed November 29, 2022, https://www.ctdatahaven.org/reports/ct-data-story-housing-segregation-greater-new-haven.
[2] Robert Austin Warner has reconstructed where African Americans lived in nineteenth century New Haven, see New Haven Negroes, a Social History (1940; New York: Arno Press, 1969), 29-31.
[3] US Census Bureau, “Census Blocks and Block Groups,” in Geographic Areas Reference Manual, 1994, chap. 11, https://www2.census.gov/geo/pdfs/reference/GARM/Ch11GARM.pdf; US Census Bureau, “What Are Census Blocks?,” Census.gov, accessed November 29, 2022, https://www.census.gov/newsroom/blogs/random-samplings/2011/07/what-are-census-blocks.html; US Census Bureau, “Glossary,” Census.gov, accessed November 29, 2022, https://www.census.gov/programs-surveys/geography/about/glossary.html.
[4] James M. Patten, Patten’s New Haven Directory, for the Years 1841-2 (New Haven: James M. Patten, 1841), http://archive.org/details/1841NewHavenCTDirectory.
[5] Original road data from Arc: https://services.arcgis.com/P3ePLMYs2RVChkJx/arcgis/rest/services/TIGER_Roads_2021_view/FeatureServer. Historical road data from maps held by Yale University Libraries: F. W. Beers, “Plan of the City of New Haven : New Haven, Conn. – Yale University Library,” 1868, https://collections.library.yale.edu/catalog/15675098; D. W. Buckingham, “Map of the City of New Haven. – Yale University Library,” 1847, https://collections.library.yale.edu/catalog/15830186; “Plan of Fair Haven, New Haven Co., Conn. – Yale University Library,” 1855, https://collections.library.yale.edu/catalog/15675059.
