Charles Hodge (1797-1878) was professor at Princeton Theological Seminary from 1820 until his death in 1878.  With a tenure of more than 40 years, Hodge’s leadership coincided with the height of the American Civil War, 1861-1865. He examined and wrote pivotal works on a Biblical rationale of slavery and slave ownership. His life and work generate much discussion today.

I used metadata from the correspondence series of the Charles Hodge finding aid from the PTS Library Special Collections to create a network visualization to see if new insights into Hodge’s views on slavery might be gleaned from examining relationships in his letters to and from colleagues.

Network visualizations consist of nodes and edges, with nodes representing in this case correspondents or individuals, and edges being the connections or links between the individuals. Networks can be analyzed in two ways: as whole networks focusing on connections of an entire system, and as ego-networks focusing on the connections of one dominant node to the rest of the network.  The Hodge correspondence lent itself to an ego-network, with Hodge as the dominant node.

First, I imported, cleaned, and normalized metadata from the correspondence series in Google Refine. I created columns labeled “Source” for the letter writer, “Target” for the recipient, and “Type” for the direction. In this case all were directed. A fourth column, “Date,” listed the year of correspondence.

I then imported this data as a comma separated value (csv) file into Gephi network visualization software, where I tested several layout types.

Gephi offers 12 layout types on the left side panel in the Overview pane. The Data Laboratory holds the data set uploaded as a csv file. The Preview pane displays the final version of the network visualization.

ForceAtlas 2 layout produces a network of Hodge’s correspondence that is very strong visually with edges exploding from the central node, representing Hodge. Relative edge thickness can be set in preview settings. For this project it is at the default setting of 1.0. Even in the default setting, two edges are significantly thicker than all others, illustrating extensive correspondence from Hodge to one individual and comparable correspondence from another individual to Hodge. ForceAtlas 2 also displays a few thin edges between nodes exclusive of Hodge. Labels can be added under Presets options when in the Preview pane, and the identity of these nodes linked outside Hodge will be revealed. Labels can be made proportional to the size of the node, but some layouts of this dataset, such as ForceAtlas 2, make proportional labels illegible and the display confusing.

I then tried the Radial layout, which is better suited to an ego-network visualization. The Radial layout retains edge thickness that was shown in ForceAtlas 2, but now all nodes radiate from one central Hodge node and are dispersed more evenly. The result is a network that is easier to read and interpret, though it doesn’t convey the force and power of the central node as well. Using the Radial layout, in the graph window I clicked the Heat Map icon, which sets color intensity by distance from the center node and by edge weight or thickness. This produces an even better graphic visualization of the central node’s relationship to the surrounding ones.

I also discovered that unchecking the curved box under the Edges preview settings in the Preview pane not only straightens the edges but also gives them directional arrows, indicating the direction of correspondence between nodes.

Orange and black are Princeton University colors, so I reset the color palette to purple, a more ecclesiastical color.

By filtering the data through a partition count of the date ranges, I created a new graph with only correspondence from 1861, the first year of the Civil War. The new visualization shows most correspondence in the PTS collection was between Hodge and William B. Sprague, a Presbyterian minister in Albany, NY, with Sprague writing to Hodge. To create this visualization, I dragged the Partition (Date) filter down below Queries, selected a date from Partition (Date) Settings, and ran the filter. I then switched to Preview Pan and refreshed the graph, then saved an image of it. To create a full representation of all dates, all of these steps would need to be performed for each date.

These visualizations were created using a small dataset with fewer than 300 entries. Possible future expansion of the visualization could include finding aid metadata from the correspondence files of other individuals who knew Charles Hodge personally or professionally, such as his wife and children, colleagues at Princeton Theological Seminary, and fellow clergy. Additionally, the Charles Hodge Papers at Princeton University is a much larger collection with thirteen boxes of correspondence dating from 1810 to 1884 and includes letters to and from Princeton Seminary colleague Archibald A. Alexander, more letters from Sprague, and some  from classmate John Maclean. Adding data from other related collections would result in a more complex network with clusters outside the ego-network visualization created just from Charles Hodges’ Princeton Seminary correspondence.

This visualization offers an initial glimpse into dynamics between Charles Hodge and his contemporaries. Further research is necessary on Hodge and about other individuals in the visualization to determine whether new insights on Hodge’s stand on slavery can be gleaned from the graph. Next steps would include biographical studies of persons represented by thicker edges or larger nodes, along with addition of more data to this network and creation of related networks.