For once, historians deploying network analysis were able to skip the exposition. The vocabulary and theory of nodes, edges, and graphs required no introduction. Instead, it became possible to focus on the challenges of bringing together the domains of mathematical networks and source criticism. While the conference covered a period from the Middle Ages to the 21st century and discussed disparate networks, a conversation emerged across the debates that tied the varying topics together. This conversation explored issues of archival silences, the completeness of networks, and the creation of new perspectives on source materials revealed through abstracting relations between historical actors, places, or even objects, into network graphs.
The conference opened with a case study by DELFI I. NIETO-ISABEL (London) on the 14th-century Beguines in southern France. Drawing mainly on documents from inquisitorial courts, Nieto-Isabel traced the network of Beguines and actors associated with their community. Despite being most probably incomplete, the model helps to shed new light on how their prosecution spread through the region, and subsequently how new narratives of sainthood and the end of times formed. Afterwards, GRACE DI MÉO (Oxford) discussed leadership change in the all-female Forty Elephants gang in interwar London. This change coincided with a drop in the connectivity of the network, seemingly indicating a decreasing degree of co-offending within the gang. Di Méo warned against taking the data at face value as at the time the gang also started deploying new tactics, which aimed to avoid having crimes attributed to them. Subsequently, MARTIN NICASTRO (Pavia) expanded the bounds of network analyses by including geographical data and field research to map the venues of the music scene in Milan around 1960. Nicastro found that most performances took place within a 2 km radius of the city centre. At the same time, there was only some overlap between musicians performing different genres, e.g. classic and pop music, while others, like opera musicians, stayed mostly to themselves.
The second panel revolved around networks of literature production and books. MACIEJ MARYL (Warsaw) used bibliometric data to quantify the attention Polish authors received through publications written about them, comparing the results with their career landmarks. Maryl's findings contrasted with some of the expected narratives about career trajectories. While, for example, an author might win a Nobel Prize, the attention of monographs and articles does not necessarily focus on such events, but on the popularity of a religious poet. EMMA WILSON (Dallas) explored an ‘imagined’ network of authors and printing locations in the early modern era. The network centred around Paris and Venice, which deployed tactics like the perpetual reprinting of popular Cicero texts (dubbed the “Cicero hack” in subsequent discussions). Furthermore, it became apparent that authors with significant numbers of publications, printing in different cities, e.g. Philipp Melanchthon, were able to gain significant network centrality. FABIAN DOMBROWSKI (Brunswick) presented experiments conducted by himself and SEBASTIAN KLAES (Brunswick) at the library of the Georg Eckert Institute, using catalogue data of digital collections to reconstruct networks of textbook production. He recommended allowing research questions to determine digital methods rather than vice versa. The experiments aim to give a more accurate account of the network of textbook production involving a two-step process: using Named Entity Recognition (NER) to process title pages and prefaces to expand the catalogue data and link the entities to authority files enriching the data with additional information.
CAITLIN BURGE (Luxembourg) provided insights into how Thomas Cromwell utilized his position as secretary to Cardinal Wolsey to establish his own network of influence. Building on Granovetter's idea of triadic closure (two actors will most likely come into contact if they independently have a strong relationship with a third actor), Burge traced the network of letter exchange between Cromwell, Wolsey, and the court society. Cromwell regularly closed triads, meaning that, rather than merely acting as a mouthpiece for Wolsey, his contacts with court members became direct contacts of his own. RACHAEL HASLAM (York) showed that even with disconnected networks, there is potential to create new perspectives on a phenomenon, in this case: medieval letters of recommendation. The letters reveal a set of consciously created ego-networks, centered around specific individuals, which were used to create and advance careers. Haslam focused on the networks of high-profile actors like Anselm of Canterbury, Bernard of Clairvaux, and Thomas Becket. MELITA THOMAS (London) worked on filling gaps in the social networks around Mary I. Not many of Mary’s letters survived, so Thomas had to consider other hints in the source material to be indications of relations. Using accounts of titles and gifts awarded in Mary's household in conjunction with letter mentions, Thomas was able to reconstruct parts of the network to consider the role of soft power in Mary’s inner circle.
The keynote of RUTH AHNERT (London) addressed the methods deployed in the research for her 2023 book Tudor Networks of Power.1 Central to the development of these methods is the notion that the data itself does not provide historians with answers, especially not when considering only a single metric in a network. Instead, it is necessary to layer metrics, to make the data human-readable through techniques such as visualisation, and to maintain a strong connection with the source material. Ahnert also explained the process of data cleaning, which is one of the project’s biggest outputs since the data is available to interact with online.
CAROLINE FISH (West Lafayette) emphasised the problem of the embajadoras. While early modern letters of Spanish ambassadors survived, much of the source material on their wives has been discarded by the archives. This survivability makes it difficult to place these women in the diplomatic networks of their time. Fish chose a case study on Catalina de Moncada, an embajadora who, even with a lack of material, shows up as the second most central figure in her husband’s network (as inferred from other letters mentioning her writing). Fish showed that even more relations can be added, as certain rhetorical motifs in letters to Catalina's husband imply her personal connections. This shows Catalina de Moncada to be a diplomatic actor at least on par with her spouse.2 KONSTANTIN FERIHUMER (Vienna) examined the process of “Aryanization” of Jewish businesses in Vienna in 1938/39, the effects of which are still present in modern-day Austria. Though Ferihumer admitted that the data collected is incomplete, he was able to reconstruct a pattern of behaviour by the perpetrators in the watch and jewellery industry: low and mid-level fascists organised the takeover and re-distribution by themselves, while the Nazi party and state institutions channelled this process.3
An intellectual network of Polish expatriates was introduced by ALEKSANDRA KAYE (Berlin), who also presented for MALTE VOGL (Berlin), RAPHAEL SCHLATTMANN (Berlin), JASCHA SCHMITZ (Berlin), LEA WEISS (Berlin) and LAURA VON WELCZECK (Berlin). Using the handbooks and bibliographies of Stanisław Zieliński about important Polish publications and people abroad, Kaye proposed a “multilayered historical network”. The concept, developed by Martin Grandjean, allows for a combined analysis of different levels of connections between the actors, for example through working or publishing together. SERGI LOZANO (Barcelona) and MARC BADIA-MIRÓ (Barcelona) presented a way of moving beyond the usual interpretations of the 19th-century Cobden-Chevalier trade treaties, as many of those interpretations focus on international trade volume, to focusing instead on the level of connectivity between states. Indeed, the global network itself existed before this network of bilateral treaties but grew denser through them. Additionally, the positions marked through metrics like degree and betweenness centrality changed and helped some states to move from the periphery to a more central position. An entirely different application of network analysis was implemented by GABOR TOTH (Los Angeles). Toth tracked the flow of topics in the Shoah Foundation's testimonies of Holocaust survivors to consider the differing roles of solidarity in male and female survivors. A network emerged that represents all potential paths through a narrative and the data showed that male and female survivor narratives do differ, as female narratives choose paths through the topic graph that cross more nodes representing topics of solidarity. However, Toth warned that this might not be an expression of the gendered qualities of actors, but an effect of mechanisms of control in the camps.
Researching antisemitism within artist communities of the 30s and 40s, KONSTANTIN SCHISCHKA (Vienna) tackled the problem of lost archives. Schischka substituted this lack of source material by relying on exhibition catalogues. This helped to reconstruct a fluctuating network, where only one percent of the central actors stayed in a stable position. The picture might be misleading, as Schischka noted, since the catalogues did not help to collect data about female artists and persecuted groups. The conclusion therefore stressed the need for a hermeneutic approach alongside network analysis. The next paper was given by JESPER VERHOEF (Rotterdam), whose work on the network of Dutch transgender websites analysed a larger queer web sphere. This turned out to be only partly integrated into the online network of the wider community, reflecting the “marriage of convenience” between the trans*- and the gay-lesbian activists in the Netherlands. Verhoef showed that both groups work together to be recognized by society but commit to different and conflicting theories of gender. MEGAN PALMER (Nottingham) concentrated on the Transylvanian city of Brașov, a community shaped by German, Hungarian, and Jewish families. For actors to gain influence, it was necessary for them not to work against the network formed between these groups, but to partake in it. Only when nationalism started to grow in the region in the 19th century did the network drift apart into more distinct groups.
To conclude, the conversation at Our Interlocked Universe was not so much driven by discussions of methodological approaches, but by shared theoretical concepts and their implications. There were three main lines of discussion. Firstly, it was agreed that in most cases it is not possible to produce complete networks, as the historical record is never complete. Only some gaps in the networks can be filled, as demonstrated by Fish and Thomas. Additionally, many contributors made the point that while the completeness of their graphs was unachievable, reliable interpretations can be offered by reaching what Ferihumer called the “saturation” of data. This leads to the second key topic of discussion, that the network metrics alone provide no evidence. Networks are valuable as they help historians to navigate large data sets and are a useful abstraction which can be leveraged to create new ways of observing connections. The overall conclusion was that working with network data is an intermediate step and must always be related back to understandings of historical phenomena. The last thread of the conversation concerns the biases of selection practices at the archives. Those practices shape the data even before it becomes digital data, let alone before that data is transformed into network visualisations. In some cases, this means that a network analysis reflects more about the archive than it does about the actual topic of interest. As an addition to the much-discussed data literacy, it might be necessary to start reinforcing collection literacy, to ensure that (digital) historians reflect on the influence of selection processes that determine early on the limits of possible data sets.4
Even though the debate was in-depth and established the conclusions above, there is still a lot left to explore. What, for example, can serve as general criteria for the sufficient saturation of data? Likewise, as is always the case following a conference of specialists, the debate must be continued in the wider community. These objectives must be next on the agenda if the valuable method of network analysis is to be established further as a trusted method and theory of historical research.
Conference overview:
Panel One: Centrality Analysis
Delfi I. Nieto-Isabel (London): Connected Afterlives. The Centrality of Martyrs and Relics in a Dissident Network at the End of Times.
Grace Di Méo (Oxford): Women, Organised Crime and Criminal Networks in Twentieth-Century England.
Martin Nicastro (Pavia): Urban Space and Performance Circuits. Analyzing Historical Live Music Through Network Analysis and GIS.
Panel Two: Multipartite Networks
Maciej Maryl (Warsaw): Network Analysis of Career Trajectories in Polish Literature after 1989.
Emma Wilson (Dallas): The Logic of Early Modern International Relations. Digital Network Analysis of Imagined Intellectual Communities, 1500–1700.
Fabian Dombrowski (Brunswick) / Sebastian Klaes (Brunswick): Hidden on Page One. Reconstructing the Networks of Textbook Production with Named Entity Recognition (NER).
Panel Three: Network Evolution
Caitlin Burge (Luxembourg): “Some of the glory stuck to his hands”. Thomas Cromwell, Thomas Wolsey, and Triadic Closure.
Rachael Haslam (York): Networks of Recommendation and the Shaping of Central Medieval Religious Careers.
Melita Thomas (London): The Power of Networks and the Networks of Power. The Social and Political Networks of Mary I of England (1516–1558).
Keynote
Ruth Ahnert (London): Tudor Networks of Power.
Panel Four: Hidden Structures
Caroline Fish (West Lafayette): A Network of Challenges. The Historical Quandaries of Social Network Analysis and Early Modern Diplomatic History.
Laura del Alisal (London): Emma Goldman’s Network of Revolutionary Women. Exploring Historical Social Network Analysis Visualisation as a Method for Digital Public History
Konstantin Ferihumer (Wien): ‘Aryanisation’ Networks. The Viennese Watch and Jewellery Industry 1938-1939.
Panel Five: Modelling and Frameworks
Aleksandra Kaye (Berlin) / Malte Vogl (Berlin) / Raphael Schlattmann (Berlin) / Jascha Schmitz (Berlin) / Lea Weiß (Berlin) / Laura von Welczeck (Berlin): Networks of Knowledge. Exploring Socio-Epistemic Structures in Stanisław Zieliński’s Historical Compilations, 1830–1935.
Sergi Lozano (Barcelona) / Marc Badia-Miró (Barcelona): Beyond the ‘Cobden-Chevalier network’. Effects of Preferential Trade Agreements during the First Globalisation from a relational perspective.
Gabor Toth (Los Angeles): Voices of Solidarity from Auschwitz. Simulation of Survivor Narratives by Women and Men as Directed Graphs.
Panel Six: Communities
Konstantin Schischka (Wien): From Art to Ideology. Tracing German Nationalist and Anti-Semitic Networks in the Vienna Secession, 1898–1945.
Jesper Verhoef (Rotterdam): A Community within a Community. Transgender Websites within Queer Hyperlink Networks, 2009–2022.
Megan Palmer (Nottingham): Social networks and formal institutions. using Social Network Analysis to investigate the relationship between informal identity-based groups and formal local power structures in the Habsburg Dual Monarchy (1867–1918).
Notes:
1: Ruth Ahnert / Sebastian E. Ahnert, Tudor Networks of Power, Oxford 2023; see, for the visualisations id. / Kim Albrecht, Tudo Networks, URL: <https://tudornetworks.net/> (23.06.2024).
2: The paper was awarded a highly deserved ECR prize at the conference.
3: The presentation of LAURA DEL ALISAL (London) about Emma Goldman scheduled for this panel sadly could not take place.
4 See, for example the edited volume Anke Hertling / Peter Carrier (Ed.), Collecting Educational Media. Making, Storing and Accessing Knowledge, New York / Oxford 2022.