Following my previous post, which looked at Genre Networks in five antebellum New York City theatres, I have started working on a graph of actors and actresses through August-December 1839. Using the same theatres (The Park, the Bowery, The National, The National (at Niblo’s), and the Chatham), I built a different graph that attempts to measure what actors appeared with what actresses. I am interested in finding out how connected prominent actresses were to the male stars of the period and vice/versa. The idea for this project began while I was transcribing Odell and noticing that there was a high number of theatrical couples appearing together onstage. I was curious to see if having a husband/wife team onstage was an significant draw, so I recorded how often husband/wife pairs were present in productions compared to those same actors appearing with someone else.
From the data that I have used, the answer is highly variable. What I did find, however, suggests that a look at actor networks in the period can provide insight into the lives of performers that might be more obscure if we only looked at the data on a close basis. This is what I love about digital humanities work: it has the potential to provide a wider perspective that one might not otherwise notice on the day-by-day scale. However, as many people will say, “distant reading” is not an end-product, but a means to indicate directions for further work. In making this graph, I found some great clues that I look forward to researching in detail.
A Note on the History of Actresses
There have been many fine works on the lives of nineteenth century actresses, but, as a group, they remain vastly underrepresented in the historical literature. This is in large part due to the relative importance of male actors at the time. While there were certainly female stars, by and large starring roles tended to be written for men, as the pre-1840s audience was largely male. There are exceptions, but a look over the cast lists of plays from the period shows a distinct emphasis on male leads. If anyone were to study the language of reviews from the time, they would probably discover some pretty awfully sexist discourse. In fact, one of the most surprising things I found in this project was how hard it is to find the first names of many of these women! Actresses are mentioned in the cast lists as “Mrs. So-and-so” or “Miss So-and-such.” Being publicly defined by their husband’s name, it makes them seem even more marginalized. I’ve included a few proper names in the graph, but largely left their surnames.
By the way, there was just ONE example of a production that only featured women: The Single Life by John Baldwin Buckstone (a comedy).
Since Odell tends to focus on the leading roles over the members of a theatre’s company, I decided to work with him for once. I started taking the data I extracted from Odell and noted which actors and actresses appeared together on stage. The data (HERE) looks like this:
and so on…
The “Source” headings are the actors and the “Target” headings are the actresses. The numbers under “Weight” represent the number of times each pair has appeared together in the same production. I did not take length of run into account. Looking at the number of performances each appeared in might have measured popularity, but I am interested in how many plays were produced with each pair. In the productions that I recorded, I looked for the male and female lead, sometimes also including the secondary lead if the role was significant enough (i.e. for Othello, I counted Othello and Iago).
This is an “undirected” network, which means that gephi draws a line between an actor and actress each time they appear together, more or less ignoring who is the “source” and who is the “target.” This does not measure who is the biggest star or who gets better roles, but how connected each individual is in the network of working actors in the season.
Compared to the graph from my previous post, this one is nowhere nearly as connected.
Notice the outlying clusters on the top: these are examples where actors and actresses are fairly isolated to one theatre. Clockwise from the left, the clusters represent the the Park Theatre (operatic performances), the Park Theatre (dramatic performances), and the Bowery Theatre. For example, Mrs. Proctor at the Bowery did not act with anyone who acted at the other theatres. It very well might be that with more data, we would see a higher percentage of connectivity in the graph.
The main cluster shows the two National Theatre companies and the Chatham. These are pretty well-connected, with good historical reason. In the fall of 1839, the National Theatre burned down, so the company moved to a new location, which ultimately failed, and they were picked up by the Chatham as it opened its doors.
The graph’s colors are based on gephi’s modularity partitioning, so nodes that appear closer together share the same colors. Here, I was impressed that the colors largely conform to the different theatres. In the main cluster, the purple is the Chatham, the green is the first National, and the blue is the National after the fire.
The nodes (individual actors) are sized according to their “betweenness centrality.” Instead of measuring the number of productions they were in, which would look at their popularity, this measures their connectiveness by seeing how many shortest paths each actor is on. Whoever is on the most shortest paths between people has the highest betweenness centrality. Basically, a well-connected person will be close to a high number of people. For example, if (A) wants to meet (C) but doesn’t know her, (A) needs to go through someone else, and, all things being equal, they would rather talk to (B), who knows (C), rather than talk to (F), who knows (E), who can get in touch with (D), who knows (C). In that case, (B) would have a higher betweenness centrality than (F) or (A). In this graph, the most connected person is Matilda Flynn, suggestingthat she has shared the stage as a co-star with more different people than anyone else. If you needed to find a go-between to another actor in the cluster, chances are, Flynn would be your person.
By contrast, if I was looking at popularity, one the most-produced actors in the graph is J.R. Scott, but he is represented by a small node because he only appeared with Matilda Flynn (Mrs. Thomas Flynn, in 8 plays) and Amelia Bannister (Mrs. Nathanial Bannister, in 2 plays). Likewise, Edwin Forrest, the biggest American star of the period, only worked with Anna Sefton and Virginia Monier.
On the graph, btw, the thickness of the lines connecting nodes indicates how many times the pair has acted together, so you can tell, for example, that Mr. Bannister worked more with Matilda Flynn than Mrs. Bannister.
I’ve cleaned up the graph a bit, removing nodes with only one connection.
The outlying clusters look fairly straightforward, but they offer a few tantalizing details. For the three clusters on top, I’ve sized the nodes according to degree, meaning the bigger the node, the more productions they were in.
First, the Bowery:
This was a little surprising, as the Bowery is generally thought of as being quite robust in the time period. While this picture should change with more data, the Bowery is interestingly incestuous in the sample. Here, we can see that Thomas Hamblin, the Bowery manager, largely acted with his wife, Mrs. Shaw (well, she wasn’t actually his wife, but that’s the subject of another post…). Elizabeth Shaw was the busiest actress in the theatre, with Mrs. Proctor close behind. The Bowery had a successful season this Fall, so I suppose if it ain’t broke…
The Park Theatre is pretty straightforward, but the two graphs clearly show that the theatre had a separate range of stars for drama and opera:
The Main Cluster
Which brings us to the main cluster:
The most immediate observation I noticed here is that actresses are more connected than their male counterparts. This is a result of the structure of theatres at the time where most theatres had a resident company that shuffled around to accommodate engagements with (usually male) stars. The graph is showing that even though actresses didn’t have as many starring vehicles and were relegated to company roles, they may have had a wider field of experience than the men. Quite possibly, the structure of the theatre marginalized them even as the lived practice provided them with superior social networks. I would love to follow this up with an in-depth look at how actresses related to each other!
With the exception of J.S. Browne, who was constantly working in the sample period, actresses are consistently more central than actors. And, through all this, Matilda Flynn just dominates:
Above, each highlighted node is an actor that Flynn worked with on more than one occasion (note, you don’t see a line connecting her to Browne or Anderson because they only acted together a few time sand gephi doesn’t draw it thick enough).
One can see how central Flynn is to the Chatham, but also to the wider theatrical landscape of the season.
On my initial question of husband/wife teams, some teams were frequent, others not so much. Henry Wallack and his wife, Maria, frequently worked together, but others, such as the Bannisters did not. I did notice that Father/Daughter teams seemed to be more interesting to audiences, with the Vandenhoff’s almost exclusively appearing together.
Now, I’m off to sort through Odell and expand the data that’s being collected…
browse more posts:
What do you do with a footnote to a footnote to history?One of my side-side projects is to look at a window in…
Towards Games as ScholarshipThere was a point in my graduate school career when I stopped…
D3plus and Theatre HistoryAfter reading Anastasia Salter‘s ProfHacker post on D3plus visualizations, I was intrigued…
A Network Perspective on Male/Female Co-Stars in New York City Theatre (1839)Following my previous post, which looked at Genre Networks in five antebellum…