By Halima Haruna.
Dr. Julia Flanders, Dr. Nick Beauchamp, and Dr. Lawrence Evalyn provided mentorship and feedback on this project.
African American women applied for pensions between 1861 and 1934 when their Civil War veteran husbands died. These women, often the sole breadwinners in their families, flexed their rights as citizens for the first time with the support of what Brandi C. Brimmer calls a “grassroots pension network” of social ties and associational life in their neighborhoods in Claiming Union Widowhood (2020). The grassroots pension network included neighbors, other veterans, acquaintances, family relations, and religious leaders who testified on behalf of widows in letters, affidavits, and depositions. The network also included black professional men who served as claims agents alongside white attorneys in South Carolina and Washington, D.C.
This project-prototype collects data from digitally available pension claims (via the National Archives) from widows of veterans of the 2nd and 33rd United States Colored Troops (U.S.C.T.) regiments. The project aims to create networks that uncover trends and patterns in the communities that supported the widows as they navigated the application process. The project’s findings contribute to our understanding of the lives of African Americans post-Civil War and their engagement with their communities and governmental bodies. Methodologically, the project proposes new definitions of relationships previously limited by the archive and an argument for using network visualization and analysis in historical research.
Historical Background
At the start of the 19th century, the institution of slavery was already crumbling. Abolition grew in political popularity due to the efforts of black and white abolitionists—the newly formed Republican Party in 1854 codified values like abolition and free market trade. The Republican Party and its candidate, Abraham Lincoln, entered office in 1860. His election, among other factors like debates over federal enforcement of the Fugitive Slave Law in 1850, led to Southern Secession from the United States Union and the Civil War (1861 – 1865) (Blackett 2018, 38).
The Lincoln administration initially claimed that the issue of slavery was unrelated to the war. In July 1861, Congress signed a policy that mandated army commanders to close their bases to self-emancipated people, but the law quickly unraveled. Still more refugees arrived at the Union base camps. Some men who arrived at the refugee camps around Union army bases in the South were threatened with being turned away if they did not serve in the Union army. Others who served with the Confederate Army abandoned the Confederate Army for the Union Army. Up to 210,000 African Americans served in the Union Army from all over the country, over one-fifth of the nation’s adult black male population under the age of 45.
The Act of July 14, 1862, set up the general law pension system for war veterans of the Civil War and their dependents. African American widows were initially excluded from this system. Protests by African American widows at Fort Pillow, Tennessee, in 1864 prompted lawmakers to address this injustice. The supplemental Act of July 1864 accepted eyewitness testimony to support claims for pensions, which allowed African-American widows to prove their marriages to veterans. These efforts were necessary because marriages between African Americans were not legally recognized before emancipation. Furthermore, in June 1866, African Americans’ emancipated status was recognized by eliminating references to their formerly enslaved status.
However, the law of 1882 used sexual morality as a basis for determining eligibility for a pension, undermining the rules introduced in 1864 and 1879 (Brimmer 2020, 12). The federal government, via the Department of the Interior and Veterans’ Affairs, employed investigators known as Special Examiners to determine the veracity of widows’ claims. The service law of 1890 allowed a veteran who had honorably served 90 days in the military to apply for a pension if he could find a physician to affirm his unfitness for manual labor. While the service law was ostensibly passed in good faith, it was not in practice. Many veterans were refused their claims to disability pensions because bureau officials did not believe them (Logue & Blanck 2008, 390-1).
Making a Claim
Making a claim was often an arduous event that required claims agents to take on responsibilities of benevolent societies, providing shelter and financial assistance to disabled veterans and poor widows before their quarterly stipends arrived. This was because African American women faced intense scrutiny from the federal government as they applied for pensions. Many African American women and men took steps to obtain marriage licenses as soon as possible. However, African American women and men also continued to embrace fluid domestic relationships, a leftover from the antebellum period during which long-term monogamous cohabitation and marriage were difficult to maintain. Women who remarried after their husbands died while receiving a pension would have their payments terminated. Oftentimes, veterans separated or abandoned their wives during the Civil War and remarried. These situations led to contesting widows’ pension claims, which Special Examiners investigated. Investigations could take any number of years. These cases, as well as those with missing documents, resulted in pension files with hundreds of pages, and that took 2 years on average to be approved. Due to the combined factors of most widows’ old age and the length of time before approval, many pension payments would only be approved after the widow or dependent had passed away.
Pension claims were essential for the widows applying for them. Most African American women were unable to acquire regular wages or property and were providing for their families or depending on their adult children. Some widows also depended on mutual aid associations and benevolent societies for basic amenities. Despite the obstacles to securing a pension, African American women asserted their right to citizens’ welfare.
Datasets and Data Dictionary
The primary source for the data is from the Department of the Interior, Bureau of Pensions (1849 – 1930) and Veterans’ Administration (07/21/1930 – 03/15/1989), Case Files of Approved Pension Applications of Widows and Other Dependents of the Army and Navy Who Served Mainly in the Civil War and the War With Spain, Records of the Department of Veterans Affairs. There are 417 pension claims with “U.S. Colored Troops” in the title according to the search I performed on the National Archives interface, and I have data on 172 of them so far, about 40% of the filed claims. Each pension claim comprises a mix of forms, letters, and depositions. Each claim includes a Civil War widow’s pension claim, her late veteran husband’s pension claim (most veterans filed for pensions before they passed), and the pension or reimbursement claims of the widow’s family members when the widow passes, if they file for such a claim.
The typology of the claims informed decision-making about the data. The forms used by widows and attorneys made certain data points visible and repeatable in multiple pension claims. Some categories, like dates, were identifiable across pension claims due to the bureaucratic form of the pension claim. The key documents I focused on for data collection were the approval forms, pension declaration forms, and the depositions. The approval forms are written and signed by the state. The approval forms record key dates like the date of the veteran’s death, the first date the widow’s pension claim was filed, and the approval date. The pension declaration forms are written by or on behalf of the widow, as she declares her application for a pension. The declaration records the names of minor children if the widow had any, and the location where her husband was enrolled, among other information. The deposition is written by testifiers (family relations, other veterans, etc.). The relations in these depositions are categorized and defined independently in this project.
Data Collection Process and Methodology
I manually collected data from digitized petitions in a simple spreadsheet. I used key strategies including close reading, skimming, organizing the master document, building an evolving data dictionary, keeping detailed process notes, and scoping the type of data collected through advisory meetings with instructors. This process was supplemented by reading recent scholarship on U.S. Civil War veterans and their widows. Based on a literature review conducted for an Independent Study course on U.S. history, I outlined my principal objectives in data collection, restricting the data to the bureaucratic process and relationships in the pension claims.
To assess the feasibility of the data collected, I paused data collection to prototype. I used OpenRefine to “clean up” a small sample of data (10 rows in CSV format), that is, to make the data machine-readable. I used common transforms to trim whitespace and to replace tabs in names with underscores for better machine readability. I used these directions to change the date format in the dataset from a MM/DD/YYYY numeric format to a DD month YYYY format. During a prototyping session with Lawrence Evalyn, I explored the data using RAWGraphs, a web-based platform that creates quick graphs. The prototyping process revealed gaps in the data collected for network analysis based on the pension claims. Dr. Julia Flanders suggested including geographic locations in the data. I also introduced controlled vocabularies into the data spreadsheet using the Google Sheets data validation tool.
Decisions
The project data’s uniqueness is a function of the lack of access to tools for analyzing handwritten documents. Contrary to projects using machine-generated corpora, to generate any data, I had to read, make decisions about, and transcribe the primary source. These decisions included choices about what data to collect, e.g., which individuals are most relevant to a network, and how to collect it in order to fulfill the project’s objectives.
The pension files varied widely in content and format. Few documents were standardized across pension claims. This is due to the wide geographic and temporal range covered in the sources: the data is collected from sources dated from 1864 to 1934, and the data collected is from pensions filed in 23 different states in the Northeastern, Southeastern, Midwestern, and Southwestern regions. This temporal and geographic variance resulted in differences in claim and approval form formats and variance in the content recorded in the sources. Despite the variance, dates were more consistently recorded in the petition files and were easier to recognise in writing than letters due to the more standard visual form of numbers than that of letters in handwriting. Other data points, like the locations of widows when filing, attorney names, and attorney offices’ locations, were also consistently recorded.
The project required a process or logic for defining the relationships present in the archive. Based on secondary scholarship on African American kinship during and after the Civil War, I defined the “Family Relation” category of relationship between widow and testifier using fictive kinship logic. People became fictive kin because they worked or lived together during the antebellum period. For example, Sylvia Jones’s pension claim includes a deposition from Maria Bryan saying that she had “known Sylvia since she was a little girl,” so I categorised their relationship as Family Relation. Relations like “Neighbor” and “Acquaintance” were determined by the definitions given in the pension claims. The data dictionary has the complete list of category definitions.
Certain categories, such as “Soldier Alias,” seemed to be consistently recorded in the primary sources. However, after prototyping, the category was not pertinent to the project objectives. Another discontinued category was “Free/Formerly Enslaved,” recorded as a boolean value. This data point was discontinued because the later petitions did not include this data, especially after the 1866 act that fully acknowledged emancipation for all. I started collecting “Attorney location” and “Initial Rejection” categories following conversations with Dr. Flanders. I also started collecting data on “Children” as a boolean category. Ultimately, data on the widows and the testifiers stayed constant.
Network Analysis
At the start of the project, the objective was to visualize and analyze the “grassroots pension network” described by Brandi Brimmer in Claiming Union Widowhood. After discussion with Dr. Julia Flanders, it was determined that the project’s goal required the use of network analysis software. Early trials were conducted in INSH 7910, NULab Workshop with Dr. Julia Flanders.
Cytoscape
Cytoscape is a desktop package that visualizes and analyzes networks using nodes and edges. Cytoscape offers network visual customization after analysis. It also has a robust user manual. For the first trial, I used data from 59 petitions filed between 1871 and 1924 by widows of soldiers in the 2nd regiment of the United States Colored Troops. Regiments proved to be a useful category for scoping the large dataset.
An edge list is required to visualize a network on Cytoscape. The node list identifies each entity in the network. The node list for the first trial included 59 widows, their late husbands, two persons recorded as a “Witness” to a widow’s affidavit or as a Testifier making affidavits as evidence in a widow’s claim, and an attorney working with each widow. Each node was identified by the name recorded in the files. Node attributes or roles were assigned based on this categorization: “Widow,” “Veteran,” “Civilian,” and “Attorney.” The gender of each person was recorded as a second node attribute using a “M” or “F” boolean. Gender was assumed based on cultural understanding of the genders assigned to specific names. The attribute of Gender and assuming its binary nature and its relation to the individual was a complex matter. Discussions with Dr. Flanders led me to the conclusion that first names are usually used to signal gender based on the context of the documentary record.
The edge list sets up a directed network with the primary node as the widow and the secondary node as the veteran, civilian, or attorney. The edge relationship was defined by the widow’s relationship to the secondary node: “widowOf,” “clientOf,” and “wasTestifiedForBy.” Once the edge and node lists had been created from the parent dataset, the data was ingested into Cytoscape.
The first trial network revealed the connected nature of the pension claims of widows in the 2nd regiment, supplementing Brimmer’s thesis of the “grassroots pension network.” Using Cytoscape’s network analysis tools, measures such as betweenness centrality and degree could be calculated. In this particular regiment, widows had a higher betweenness centrality and degree value (betweenness and degree both refer to measures that calculate the significance and strength of a node’s connections to other nodes in a network) than attorneys. The widows’ higher measures of betweenness centrality and degree suggest that their actions in community-making were more critical than those of attorneys who had the larger legal and financial freedom to act in pension claims-making.
Gephi
After the first attempt at network visualization and analysis, I discussed the project with Dr. Nick Beauchamp, who suggested the Gephi platform, a network visualization tool. To better visualize and analyse the data collected, it proved imperative to explore Gephi in addition to Cytoscape. This exploration would also support my academic goal to become familiar with network analysis broadly and develop my capacity with the tools in the field.
After collecting additional data, I decided to approach the next phase of network analysis by scoping the data by regiments. Scoping the networks by regiment also provides consensus on the location of the widows, increasing the likelihood of the presence of a network. 38 of the 57 widows in the 2nd regiment network were located in Virginia at the time they filed a pension claim. 21 of the 42 widows in the 33rd regiment network were located in South Carolina, and 13 of the 42 widows were in Georgia at the time they filed a pension claim. The 2nd regiment network, the 33rd regiment network, and the interactive versions of the 2nd and 33rd regiment networks are directed networks with edges that indicate relationships from the widows to all other nodes. The first network (the 2nd regiment) uses the same data as the Cytoscape network, and the second network (the 33rd regiment) uses newly collected data.
To prepare the data for both networks, I used this converter to convert CSV files to GEXF files and used Gephi to produce the graph visualization. When that failed, I assigned each node an ID number in the node list and used facets in OpenRefine to rename nodes in the edge list with the corresponding ID number. I used this tutorial to get started on Gephi. I used the Force Atlas layout and changed Repulsion Strength to 750 and Attraction to about 0.1. I used the Partition (discrete mapping instead of continuous mapping) module to configure node color by Degree. The networks also featured weighted edges. Edges from attorneys are weighted at 1, edges from testifiers are weighted at 3, and edges from late husbands are weighted at 5. Nodes with the highest degree value (the number of edges directed to or from a node) are colored in darker green, with the color continuously changing from dark to light based on degree value.
Early Findings
Gephi’s data laboratory supports calculations for degree, stress, and bridging centrality values in networks. I used these measures to closely read and analyse the nodes to ascertain which nodes and edges (relationships) make up “invisible” modules in the network.
In the 2nd regiment, Daniel Langley, a Civil War veteran, testified in the claims of four widows with husbands who served in Company B on the 2nd regiment in Norfolk County, Virginia. He was a corporal in the Union Army and became a shoemaker after the war. In Betsy Whitehurst’s pension claim, he confirmed Isaiah Whitehurst’s visual disability in a deposition. Daniel Langley was part of a larger network. Betsy Whitehurst’s husband, Isaiah Whitehurst, testified for Sarah Bright and Lucinda Ackiss. Lucinda Ackiss’s husband, George Ackiss, testified for Betsy Whitehurst. All three widows lived in Norfolk County, Virginia, and were clients of Simon Lyon, a Washington D.C. attorney.
In the 33rd regiment network, the attorney, R.F. Greaves or Renty Franklin Greaves, has the highest degree value. Further exploration into Renty Franklin Greaves reveals that he was a formerly enslaved person who served in the Union Army, and after the Civil War, rose to prominence as a leader of the African American community in Beaufort, South Carolina.
Three Civil War veterans were shown to be testifiers in other widows’ pensions claims: John Floyd, Lazarus Williams, and Samuel Marks. John Floyd was married to Maria Floyd and testified in Mintie Williams’ claim in 1892, and before he died in 1905 in Camden County, Georgia. Lazarus Williams was Mintie Williams’s husband until his passing in 1893, and he testified in John Floyd’s pension claim in 1891, which Maria Floyd continued to pursue after her husband’s passing. The Floyds and the Williamses were a group of families in Camden County, Georgia, that supported each other to receive pensions. This shows that within counties, there were relationships that resulted in or were caused by pension claims, and these relationships were part of the social support needed to apply for pensions after the Civil War.
In summary, network analysis allows for an additional reading of large datasets, essentially revealing the clusters of relationships that took place in the counties. The 2nd and 33rd regiment network and analysis revealed that attorneys typically had the highest in-degree value, i.e., fewer attorneys were operating in the pension claims space than other individuals, and these individual attorneys had more connections. The networks also reveal the differences in the importance of locality by geography. The 2nd regiment features a majority of widows in Virginia and Washington D.C., as opposed to the 33rd regiment, which features a majority of widows, and testifiers and attorneys with high degree values located in South Carolina.
Interactive Networks
To meet the project objectives of publication to support further research into the pension claims, it was necessary to make the networks public and interactive. This blog post by Volodymyr Miz, a machine learning specialist, provided information on how to host Gephi networks on GitHub Pages sites. The deployment required the installation of the SigmaJS exporter plugin, a JavaScript library aimed at visualizing graphs of thousands of nodes and edges on the web, on Gephi. The plugin was developed by InteractiveVis at the University of Oxford. Once the file was exported and hosted on a separate GitHub repository, the GitHub Pages site with the network could be deployed. The two networks can be viewed on the Web at the links to the Network of the 2nd Regiment and the Network of the 33rd Regiment. The networks are open for exploration and analysis by users based on the node attributes provided. The interactive networks could be useful as a research or pedagogical tool to explore networks, notice patterns in, and draw new conclusions about African American communities at the turn of the twentieth century.
Conclusion
The Pension Bureau and its lobbyists were a significant influence in the lives of African American families. African American communities supported individual families and widows in their claims to federal pensions. The intertwined domains of the taxing red tape, the communities, and the pensions themselves are symbols of the changes taking place in African American life in the late 19th century, including increased state oversight, a turn to self-help, and the development of private finance. This project zeroes in on the processes that African American widows participated in to legitimize effective citizenship and acquire their share of the country’s welfare, and the communities that upheld them.
Following feedback from peers and mentors, development efforts will focus on improving the interactive network interface through edits to the JavaScript library that supports the visualizations. Development will also focus on the analysis and interpretation of the networks and the publication of relevant figures and findings. Close analysis of connections between pension claims will support the thesis of the “grassroots pension network,” research into the individuals who contributed to the pension networks, and conclusions on the larger trends among the connections within networks. In the future, the project could collect data on veterans’ disability claims and identify patterns in the claims and the bureaucratic process relating to them.
References
Blackett, R. J. M. The Captive’s Quest for Freedom: Fugitive Slaves, the 1850 Fugitive Slave Law, and the Politics of Slavery. Slavery since Emancipation. Cambridge: Cambridge University Press, 2018. https://doi.org/10.1017/9781108275439.
Brimmer, Brandi C. Claiming Union Widowhood: Race, Respectability, and Poverty in the Post-Emancipation South. Durham, UNITED STATES: Duke University Press, 2020.
———. “Tracing the Intimate Histories of Poor Black Women in the Late-Nineteenth-Century US South.” The Journal of the Civil War Era 12, no. 4 (December 2022): 528-553,596.
Department of the Interior. Bureau of Pensions. (1849 – 1930) and Veterans’ Administration. (07/21/1930 – 03/15/1989). Approved Pension File for Betsy Whitehurst, Widow of Private Isaiah Whitehurst (Alias Isaiah Whitehouse), Company B, 2nd U.S. Colored Troops Infantry Regiment (WC-430553). Records of the Department of Veterans Affairs. Accessed May 9, 2025. https://catalog.archives.gov/id/469788754?objectPage=95.
Foner, Eric. Reconstruction: America’s Unfinished Revolution, 1863-1877. 1st ed. The New American Nation Series. New York: Harper & Row, 1988.
Hahn, Steven. A Nation under Our Feet: Black Political Struggles in the Rural South from Slavery to the Great Migration. Harvard University Press, 2003. https://doi.org/10.2307/j.ctv1bzfp2g.
Logue, Larry M., and Peter David Blanck. “‘Benefit of the Doubt’: African-American Civil War Veterans and Pensions.” Journal of Interdisciplinary History 38, no. 3 (2008): 377–99.
National Archives. “In Freedom’s Shadow,” August 15, 2016. https://www.archives.gov/publications/prologue/2010/fall/greaves.html.
Pinheiro, Holly A. “Gendered Rhetoric and Black Civil War Military Recruiting.” American Nineteenth Century History 20, no. 3 (September 2, 2019): 273–91. https://doi.org/10.1080/14664658.2019.1696974.
Rothbard, Murray. “Beginning the Welfare State: Civil War Veterans’ Pensions.” Quarterly Journal of Austrian Economics 22, no. 1 (June 17, 2019): 68–81. https://doi.org/10.35297/qjae.010019.
Veterans Administration. (7/21/1930 – 3/15/1989) and Department of the Interior. Bureau of Pensions. 1849-1930. Approved Pension File for Maria Floyd, Widow of Private John Floyd, Company A, 33rd U.S. Colored Troops Infantry Regiment (WC-612242). Records of the Department of Veterans Affairs. Accessed May 9, 2025. https://catalog.archives.gov/id/313300116?objectPage=4.
