I’ve started to add data to NodeXL to create a graph of the first 45 documents of the witch trials. This covers the accusations and examinations of Sarah Good, Sarah Osborne, Tituba, Dorothy Good, Martha Corey, Rachel Clinton, and Rebecca Nurse. It also reaches the naming of Elizabeth Proctor and Sarah Cloyce. Within these 45 documents, there are 91 individuals named with 665 unique relationships connecting these individuals together (over 1100 with all the repeated connections). Here is that graphed out with the labels on the vertices removed for a clearer image of what the graph shows:
The lighter blue dots are for the accusations and depositions of Rachel Clinton. The Dark blue dots show most of the main accusers. The light green represents Rebecca Nurse and those who support her. The dark green dots are people more involved in the first three accusations. The colors are for the groups that the program identifies and puts together in a cluster based on the number of connections between the individuals.
The data from this doesn’t show much about the witch trials because of how the roles of individuals will change. John Proctor is only mentioned once as Elizabeth’s husband, but after I add documents accusing him, he will have more connections formed to change the data shown on the graph.
The way relationships are determined can be subjective. In the reading of the documents, a lot of relationships are clear, but many need to be inferred. For example, the deposition of Elizabeth Hubbard against Sarah Good gives them a relationship based on the accusation. However, Thomas Putnam wrote the deposition so there is a connection between him and Hubbard which I include when adding the data, but this does not mean Thomas Putnam and Sarah Good have a connection since Putnam might only serve as a scribe. Arrest warrants are treated differently. When Edward Putnam files a complaint for the warrant against Sarah Good, his complaint implies a connection since he takes part in having her arrested. Edward Putnam would also be filing a complaint for the afflictions of Abigail Williams, so that connection is included. This makes it so a lot of connections can be inferred with warrants. I do not add John Hathorne and Sarah Good as a connection based on just the warrant. The judges did not include “Elizabeth” on the warrant for Elizabeth Procter because they don’t know her, so it can’t be inferred for any warrant unless the judge is involved in the complaint.
I will also add that the data will not include relationships based on spectral evidence. When the afflicted claim George Burroughs led a mock Sabbath and Rebecca Nurse served him as a deacon, Burroughs and Nurse do not have a connection based on this testimony.
One other discretion I am taking with defining relationships involves the courtroom. If a person plays a notable role at an examination, they have a connection to the others who play a role as well, i.e. Ezekiel Cheever as the scribe has a connection to Nathaniel Ingersoll who owns the building where the examinations take place. This does not include Giles Corey with everyone at the first examinations even though we know he was in attendance based on testimony about Martha Corey’s attempt to stop him from going on March 1st. His connections are made when he takes part, such as testifying against his wife later in the month and even later at his own examination.
As it stands right now, Ann Putnam Jr. has the most direct connections, 45, mostly from being at the examinations as one of the afflicted. This will change drastically as the next several hundred documents are graphed and the documents from before the accusations.
-Bernard Rosenthal, Records of the Salem Witch-hunt, Cambridge: Cambridge University Press, 2009.
-Smith, M., Milic-Frayling, N., Shneiderman, B., Mendes Rodrigues, E., Leskovec, J., Dunne, C., (2010). NodeXL: a free and open network overview, discovery and exploration add-in for Excel 2007/2010, http://nodexl.codeplex.com/ from the Social Media Research Foundation, http://www.smrfoundation.org