Figure 1. Figure 1: 20 most frequentyl mentioned indivituals of the corpus.
Ancient historians often deal with the matter of scarcity of sources (Rollinger, 2020). Those who deal with the Ptolemaic Egypt ( c. 306-30 BCE) , however, struggle with the abundance of papyrological material. The royal administration of Egypt strongly depended on papyri to control the machinery of tax collection, promoting the production of written evidence and relying on the many scribes who diligently recorded all sorts of information. As a result, there are thousands of surviving papyri that record the daily dealings of the Ptolemaic administration and middle-ranking officers all over Egypt.
The Zenon archive is an exemplarly case of the caches of papyri found in Egypt. It was found by sebbakh diggers at the beginning of the 20th century in Gharabet el-Gerza in Egypt (Vandorpe, 2013). It is the largest archive of papyri from the Ptolemaic period with over 1,800 texts and was likely assembled by a man named Zenon, the business representative and private secretary the minister of finances of the king . During the years 263-229 BCE, Zenon collected registers of his work and exchanged letters on various matters with other prominent individuals, giving us an insight into his interpersonal relations, ranging from the inner circle of the king to petitioners from the Egyptian countryside.
The abundant number of surviving papyri is both a blessing and a curse. As much as the ancient historian is pleased with the perspective of working with an extensive group of sources, this can represent a high cost in terms of time and resources. In this paper, we propose the application of Social Network Analysis (SNA) for the exploration of large corpora of papyri within the context of a case study focused on interpersonal relations as represented in a copus of texts from the Zenon archive.
The presentation will focus on the goal of i dentifying social actors who occupy strategic positions in the network. A combination of manual and automated methods was used to create a database of the material. First, the texts and their metadata were collected from two online papyrological databases (Trismegistos and papyri.info). Then, the social interactions and relationships alluded to in the texts were interpreted and manually added to a local research database which served as the basis for the network analysis. The research database was modelled as a relational database and administered with proprietary software (Microsoft Access) that enabled the easy implementation of a data input graphical user interface for the qualitative analysis of the sources and the manual extraction of the interactions and relationships.
The relational database format allows for exporting specific datasets which were then used for the SNA mainly using Gephi (v. 0.10.1; Bastian et al 2009). In this context , the nodes are defined as all mentioned individuals in the texts and the edges are registered as either a social interaction or a relationship, and these two types are further characterized and grouped into different classes. Within the scope of the research, w e define interaction and relationship as follows: an interaction is any occurrence of interpersonal engagement that involves significant actions guided by societal norms, values, and roles and a relationship encompasses the links or associations between two individuals involving continuous interaction.
Historical network studies often focus on relational ties that can be extracted in bulk from sources, collections, or gazetteers, for example, the exchange of information through networks of correspondence (i.e., senders and addressees), co-attestations or the relationships between geographical origin, actors, and marked events (Ruffini, 2008, 2020; Tambs 2022) . Modeling interpreted interactions between historical actors into relational data for a study focused on the patterns of connections is a relatively innovative way of working with ancient historical sources (Alexander and Danowski, 1990; Rollinger, 2020) . Dealing with social interactions brings about the challenge of modeling nuanced information, often not explicitly given in the sources nor easily extracted with computational methods.
As anticipated, Zenon’s protagonism in the archive along with that of his employer, the finance minister Apollonios (see Figure 1), manifested themselves strongly in the network, causing an overshadowing effect that was tackled by filtering the individuals out of the graph and analyzing the resulting effects. Ultimately, w e focused the analysis on centrality measures (betweenness centrality, degree centrality and eigenvector centrality) that account for an individual’s overall presence in the archive, their overall structural importance in the network, and their influence relative to their connections.
This approach allowed the identification of 16 individuals of interest other than the known protagonists of the archive in a universe of 2,295 total mentioned individuals for further investigation. Within the same approach it is possible to filter the results to specific types of interactions (e.g., conflict, benefit, transaction ), for the identification of individuals within specific socio-political contexts.
Overall, we have concluded that the network analytical approach efficiently identifies key actors in the sources, considering centrality–in its different manifestations–as what defines the importance of an individual in this specific research context (cf. Ahnert et al, 2021). This approach favors specific types of connectedness as a defining factor and identifies sub-corpora of sources for further investigation of well-connected, powerful and centrally positioned social actors. The presented study exemplifies how adopting computational methods can help researchers overcome the challenges of working with large and complex datasets and is part of the author’s doctoral research focused on a social network analytical approach to the Zenon archive (2018-2023).