Exploring the Mount Vernon Enslaved Community Using Network Analysis

 

In 1799 George Washington wrote the most comprehensive slave list of enslaved people on his Mount Vernon estate. Three hundred and seventeen people are recorded on that document, living across the five farms that comprised Mount Vernon and beyond. Researching documentary evidence and data contained within the Mount Vernon Slavery Database, however, uncovers even more people and the relationships between them that formed before 1799.[1] The Lives Bound Together: Slavery at George Washington’s Mount Vernon exhibit, which opens today, explores the lives and experiences of this enslaved community on the estate. This interactive visualization (available soon) is part of that exhibit and examines the kinship relationships among the enslaved community at Mount Vernon using network analysis. It represents the structure of the overall enslaved community, including all known individuals who lived on the estate or established kinship ties on the estate from 1754 to 1799. The network represents a total of 517 people, most of whom were enslaved. How are these 517 enslaved people connected through kinship and how are those kinship relations connected across Mount Vernon?

Compiling the Data:

Expanding the social network from 317 people to 517 people involved multiple steps. The basis for that expansion is rooted in two documents: the 1786 slave list for Mount Vernon and the 1799 slave list, both written by George Washington. [2] He recorded each enslaved person who lived and worked at Mount Vernon like a line on a ledger, listing the farm location, owner, work assignment, gender, and age. Washington also recorded children by their mothers and included spousal relationships in the 1799 slave list. Working from these two documents, it is possible to parse out initial family relationships and structures.

Each person and each relationship in this data represents more than just the connection of two historical documents. The ability to compile this small dataset would not have been possible without the tremendous effort of those who worked on the Mount Vernon Slavery Database, which features over 30,000 data points just on the enslaved community.

For example, the dataset for this network includes children who perished between censuses, rendering their presence in the historical record almost non-existent. Without the collective work of Mount Vernon staff, volunteers, and interns working on the database project and digging through historical documents, these enslaved children would remain hidden in the documentary record. These children can be uncovered through meticulously cross-referencing Mount Vernon census data with other documents, such as farm work reports. Washington required his overseers to turn in weekly farm reports that documented the work performed on each of the five farms. These farm reports also listed people who were sick or otherwise unable to work, including pregnant enslaved women in childbed. Usually the mother can be matched with the child on the 1786 or 1799 slave list based on approximate age and birth year. If the child does not appear on a census, he or she most likely passed away before those slave lists were recorded. Although those small children did not survive, they were still a part of the overall kinship network at Mount Vernon.

The resulting dataset includes spouses, children, and siblings, resulting in a total of 1,061 kinship connections. This relationship data only represents those three familial relationships, so the extended kinship connections of grandparents, grandchildren, aunts, uncles, etc. are implicitly, but no less importantly, included. It should be noted that the Mount Vernon Slavery Database uses alpha-identifiers to distinguish people with the same name. For example, there are over 20 Bettys in the database. The capital letter following each first name functions only as a distinguisher and does not represent the last name of the individual.

Relationships table

Relationships table

Building the Visualization:

This visualization was created using network analysis, which plots the connections between entities, such as people, places, and ideas. Each node, or icon, in the network represents one entity, and each edge, or line connecting the nodes, indicates relationships.[3] In this case, nodes represent individuals and edges represent kinship relationships, such as spouses, children, and siblings. The people documented as living in the 1799 slave list are color-coded by farm, with yellow representing people who lived outside of Mount Vernon. Gray nodes symbolize people who were deceased, sold, or had run away by 1799. Each person is plotted according to the last location they lived in the documentary record. The names in uppercase represent the people featured in the Lives Bound Together exhibit.

Network graphs can be directed or undirected and even connect different types of nodes together. A directed network graph depicts one-directional relationships. This particular visualization is an undirected network graph, because the relationships between people can go in both directions. For example, the relationship between a mother and daughter is both parental and filial- they both share that kinship connection.

This social network was built in R using the visNetwork package. R is a programming language used for statistical analysis and visualizing data. The package visNetwork in R uses the vis.js javascript library to create interactive visualizations on the web (an htmlwidget). The spatial orientation of the networked family groupings does not reflect geographic location and their proximity does not suggest relative space or relationships. Given the complexity of these connections in time and space, the Fruchterman-Reingold layout algorithm offers the clearest representation of these family groups within the network. The layout only models the relationships between people and is not meant to represent their location relative to each other in the plane. Multiple factors had to be considered for the interactive network visualization, including the ability to zoom in and out on individuals within the network, color-coding nodes and edges, changing node shape, filtering individuals by name and location, and providing tool tips. The visNetwork package in R was the best fit for this functionality.[4]

Uses of the Visualization:

  • The visualization reveals the variety of family groupings among the enslaved community, such as two-parent groups, mother-child groups, spousal family groups, and multi-generational family groups. This variety connotes agency in the formation of kinship ties and speaks to the different experiences in family life of each group on the estate. Most of these family groups included at least one person who lived and worked on another farm. Only a handful of families all lived and worked on the same farm in 1799. The visualization also reveals the size of family groups. For example, approximately one third of the enslaved people featured in the Lives Bound Together exhibit were a part of the large, multi-generational family group that connected almost every farm and even some outside locations (the group including Frank Lee and William Lee). In this case, relationships between spouses connect all of the smaller family groups into the large family.
  • Visualizing this data in this manner emphasizes the importance of scale. It allows the user to view the whole structure of kinship relations on the estate, as well as the family groupings and individuals that comprise it. The network shows the number of people who formed kinship relationships from outside the estate, at least those visible in the documentary record. These outside locations include neighboring farms and even cities, such as Georgetown.
  • The visualization reveals the ways in which the database can be used to make connections between documents and people otherwise obscured in the historical record. In another sense of scale, this network graph represents two hundred more individuals than are recorded on the 1799 slave list and countless other relationships, a few of which include perished children and spousal connections buried within historical letters.
  • Lastly, and perhaps most importantly, this network visualization generates further research questions about the enslaved community at Mount Vernon. Its main goal is to provide an access point into this history and the larger picture of slavery in eighteenth-century Virginia. Using this project in conjunction with the Mount Vernon Slavery Database contextualizes the individuals and family groups within this kinship network and hopefully prompts questions concerning the formation of social relationships, daily lives of enslaved people, and broader connections to the surrounding Chesapeake area and beyond.

Limits of the Visualization:

  • First, this visualization only includes familial relationships and only those that were visible in the historical record. George Washington recorded the two main documents that these relationships are based upon, and those kinship connections do not necessarily contain the same meanings for the enslaved community that Washington ascribed to them or that we ascribe to them today. The family groups that are represented in the network, such as two parent families, or a connection of three families, are not necessarily the kinship groups that the enslaved people at Mount Vernon experienced or created for themselves. Extended kinship groups could have existed between mothers on the farms, or other social relationships could have contained more meaning to an enslaved person than a blood or kinship relationship. Along this point, fathers within the network are based on spousal connections only and by no means represent definitive biological connections.
  • Second, using one network to represent the whole structure of the enslaved community does not account for change over time. Since this visualization is a snapshot of 1799, it does not allow the user to see when relationships first appear in the documentary record, when people were born and/or died, or the changes between certain years. Generational relationships can be inferred from the visualization, but because it does not account for change over time, those relationships are not as easily accessible.
  • Lastly, the visualization does not easily show the overall connections between farms and their relative sizes. It is difficult to discern whether the ties between farms were largely between spouses, siblings, or parents and children. The number of people and family groups on each farm are also not easily discernable from this network. These limitations should be taken into consideration when exploring the visualization.

Moving Forward:

This project is an introductory exploration of kinship relations within the Mount Vernon enslaved community and can be expanded in many ways. For example, developing an algorithm that would identify family structures and provide statistical analysis would further refine the number of size and family groups. A time series would show how the community changed over time and could include factors such as Washington’s buying, selling, and renting of enslaved people, in addition to births and deaths. Layers could be added to the visualization that represent work assignments or owners. Lastly, the individuals with the network could be plotted by relative farm location to show overall kinship connections across space.

Overall, this project shows how digital methods can connect and uncover people and documents otherwise hidden in the historical record. The visualization is also a useful way to explore historical data and generate further research questions about the enslaved community at Mount Vernon.

Acknowledgments:

The data compiled and used for this visualization was pulled from the Mount Vernon Slavery Database (Microsoft Access version) on April 8, 2016. This project would not have been possible without the collective knowledge and work of Molly Kerr, Dr. Esther White, Dr. Eleanor Breen, Mary Thompson, and all of the volunteers and interns who worked on the Mount Vernon Slavery Database. They put forth tremendous efforts to uncover information about these individuals over several years. Thank you also to Jannelle Legg, Amanda Regan, Dr. Lincoln Mullen, and other faculty and students in the History department at George Mason University and the Roy Rosenzweig Center for History and New Media for their invaluable advice and support.

Notes:

[1] This online version of the database represents only a curated sample made for the Lives Bound Together exhibit. The entire project is currently in Microsoft Access at Mount Vernon. If you are interested in exploring this database further, please contact the Fred W. Smith National Library for the Study of George Washington to make a research appointment.

[2] George Washington, “List of Negroes in Diary Entry for February 18, 1786,” February 18, 1786, The Papers of George Washington Digital Edition, ed. Theodore J. Crackel, Diaries (11 March 1748–13 December 1799), Volume 4 (1 September 1784–30 June 1786). Charlottesville: University of Virginia Press, Rotunda, 2008.

[3] To learn more about network analysis, consult these works: Easley, David and Jon M. Kleinberg. Networks, Crowds, and Markets: Reasoning About a Highly Connected World. Cambridge: Cambridge University Press, 2010; Newman, Mark. Networks: An Introduction. Oxford: Oxford University Press, 2010; Weingart, Scott B. “Demystifying Networks, Parts I & II.” Journal of Digital Humanities 1 (2011).

[4] The first iteration of this project used the igraph package in R to produce static network graphs that compared the structure of the enslaved community over time.