Background and Methods

This project uses Maria Weston Chapman’s correspondence, which has been digitized by the Boston Public Library as part of their ongoing project to digitize their entire Antislavery Collection on the Internet Archive to create two sets of visualizations: an interactive map and a network analysis.

The correspondence collection consists of approximately 1500 items, which range from full letters with envelopes and enclosures to partial drafts of unaddressed correspondence. The metadata for the letters (sender, recipient, title, description, year, publisher, and link) were scraped from the Digital Public Library of America API using a  javascript scraper.

All 1500 items were then checked, cleaned and the sender’s and recipient’s addresses were entered into the database by hand.¹A great deal of thanks goes to my undergraduate research assistant, Derek Smith, who assisted with the data entry. The publisher line had some but not all of the sender’s addresses and not all the letters contained explicit addresses. Using other letters written around the same time as those without addresses and biographies of the various individuals, many of the unknown addresses were filled in with approximate locations.  Those locations were then run through GPS Visualizer using Bing Maps API (a couple of test addresses discovered that Bing gave the best results for historical addresses) to obtain coordinates for each of the points.

Network Analysis

The sender and recipient data and the addresses were then run through Gephi to produce network graphs and statistical analysis of the network.²I followed W. Caleb McDaniel’s tutorial for the initial directions on how to use Gephi for network analysis. McDaniel uses all of the letters in the BPL Antislavery Collection for his analysis, although significantly more documents have been digitized since he wrote this tutorial. W. Caleb McDaniel, “Mining the BPL Anti-Slavery Collection on the Internet Archive.” Accessed May 18, 2015. http://wcm1.web.rice.edu/mining-bpl-antislavery.html After a lot of tweaking (and reloading the data into the software as it crashed or I messed some thing up and had to start over), I got the following graphs:

This first graph is a standard network chart. It shows Maria Weston Chapman at the center of a very complex network with mostly single connections to the 283 people in the network. There are a handful a serious standouts, most notably Richard Davis Webb, who wrote or received 115 letters; Mary Anne Estlin, with whom she exchanged 99 letters, and her sisters, who wrote and received 182 letters, Deborah Weston being the most prolific individual in the network outside Chapman herself.  The average number of exchanges within the network was just over 5.  Elizabeth Pease Nichol, Samuel May and Edmond Quincy were also statistically significant correspondents in the network. William Lloyd Garrison was a secondary node within the network outside of the five Weston sisters, although the vast majority of his letters to and from Chapman were written after 1870.

A directional graph gives a slightly different picture, although before analyzing it, one must consider the way in which the archival  collection was gathered. The Antislavery Collection was mostly gathered in the 1880s and 90s as institutional histories of various reform movements and biographies of the leadership were written by the children and grandchildren of early nineteenth century reformers. Many of the Weston sisters’ correspondence was marked as private, or burn after reading, and although some of the letters survived, many others were clearly destroyed.³Lee Chambers, The Weston Sisters. (Chapel Hill: University of North Carolina Press, 2014), 181-83.  The selection of which letters to keep and which to dispose of was heavily influenced by the authors and editors of the institutional histories and many of these letters were shared with other people writing histories of other movements, including Elizabeth Cady Stanton and Susan B. Anthony who were writing the first volume of The History of Woman Suffrage around the same time. The vast majority of outgoing correspondence from Chapman was in the form of drafts, many of which had no dateline or address line, and therefore have only estimated locations based on the content or other letters dated around the same time.

As a result of the methods of collection and retention, these letters tend to be written to Chapman, except for the draft copies which often had no clear recipient and are more heavily skewed towards letters from men and about the National Antislavery Bazaar, which Chapman and her sisters ran from 1835 to 1858.  However, two people are notable recipients of letters from Chapman, namely William Lloyd Garrison and Elizabeth Pease Nichol who both received far more letters than they sent to Chapman. As more of the Garrison papers are digitized, we may find that Garrison’s status changes.

Mapping

Originally I intended to use PostGIS to generate the maps for this project, but the software had a very steep learning curve and time was of the essence (I was presenting it at Women’s History and the Digital World at Bryn Mawr). With a lot of help from Tommy Tavenner, we built a browser-based Javascript mapping tool that reads the data from a .csv file.  The software, which is still under development, aggregates and disaggregates the points as you zoom in and out of each region (the colored dots with numbers show the total number of points aggregated, not the number of letters).

The results were similar to those found in the network analysis: transatlantic connections were significant. Boston had significantly more connections to Europe than it did to the American midwest or even to Philadelphia, both major centers of antislavery activities. Chapman’s travels in Europe are also detailed, she spent several years in Paris and England and continued writing to family and colleagues back on Boston throughout that time.

Mapping the letters revealed a number of problems with mapping correspondence networks. Most notably, there were dozens of letters addressed from Boston to Boston, which were impossible to map with any meaning. Not quite half the letters had “Boston” as one or both of the addresses, which makes a closer analysis of networks within the Boston antislavery community very difficult. As I work on this project further, I will explore ways in which we can extract meaningful visualizations of correspondence data without exact addresses.