A (not so) short history of historical network research – part 2

by Cindarella Petz

Let us resume our dive into the history of historical network research! In this (three-part series) post, I first recounted the history of computational thought in historical research, and next I will discuss networks for history, before highlighting some of the cornerstones of network methodologies in historical studies.¹ This is part 2 on the development of network research.

While now we are diving deep into the history of network research, it is time to review the fundamental developments in the field of social network research, which are the basis for integrating network methodology/ies into historical research. This second post builds upon the long history of computational thought and computation-based methods in the historical studies (part 1 of this series), before I tackle networks in historical research (in part 3).

Networks! On the development of (social) network research

In social network research, interactions between (social) entities are modeled through the relational perspective of a network–a semantically annotated graph with a set of links (the so-called edges) among a set of entities (nodes) in order to “interpret the social behavior of the persons involved” as Mitchell (1969) classically defined this.

The ready availability of computers and increasing capabilities of computing since the 1950s did not only influence the rise of quantitative history, as we have seen in part 1 of this series, but also that of network research. The key breakthrough in the formalization of social network research was to describe these social networks and their mechanisms mathematically in the 1950s.

Building upon the mathematical foundations laid by Dénes König’s breakthrough in graph theory in 1936 (which itself was based on Leonhard Euler’s “Königsberger Brückenproblem” of 1736)², the 1950s saw the birth of the mathematical description of substructures of social structures (Cartwright and Harary 1956), insights on the behavior of random graphs and their transferability to social networks (Erdös and Rényi 1959), and the description of the small world phenomena (Milgram 1967; Travers and Milgram 1969), colloquially popularized under the term “Six Degrees of Separation” named after the Broadway Show written by John Guare (1990). These formal innovations coincided with the advancement of the computerization of the universities, and quickened the analysis of quantitative data–the time had been “ripe” for a relational turn, as Raab (2010, p. 32) put this.

The historiography of network research commonly associates the inauguration of network research with the studies on sociometry by Jakob L. Moreno and Hellen H. Jennings in the 1930s. However, the origins of depicting social relationships in networks and matrices are continuously traced back further to, e.g., the sociomatrices by Johannes Delitsch depicting friendships between pupils as early as 1880 (Heidler et al. 2014).

The origins of studying social relationships in networks, however, are closely related to the evolvement of the field of sociology: studying how people form social relationships, groups, and societies. As such, intellectual ancestors of network research include the work by Georg Simmel (1908, who theorised the concept of “social circles” into which individuals were born into or choose to belong), and his student Leopold von Wiese’s theories and studies on relationships (1924/1933) in order to explain the formation of society (Schnegg 2010, p. 21-2).

Modern network research centers emerged in the late 1930s and 40s in

• • New York, such as most famously the study on ‘Lady Bountiful’ by Lundberg and Lawsing (1937) theorising a connection between social status and prosperity, which mechanism would be later known as “homophily” (Schnegg 2010, p. 25),
  • Harvard around structural functionalist Radcliffe-Brown, who devised ethnology as the study of “networks of actually existing relations” (Radcliffe-Brown 1940, p. 2). His student W. Lloyd Warner was affiliated with the famous study on the ‘Hawthorne Works’ and the effect of the same name (1924–32; Roethlisberger & Dickson 1939), and directed/co-authered the “Old City” and “Yankee City” studies on the emergence of cliques in the latter (Warner and Lunt 1941) and the racial biases in the class society of Natchez, Mississippi  (published under the moniker “Deep South”, Davis et al 1941), which is regarded as an exemplary data set of two-mode matrices (Borgatti and Everett 1997; Nitschke 2016, p. 15),
  • as well as in Manchester and London in the 1950s, which ethnologist research objective led to the development of ego-centred network analysis of personal ego networks instead of more broad analysis of societies (Freeman 2004, pp. 103–5; Nitschke 2016, p. 16-8), such as the innovations in data collection and perspectives on networks in the survey of both marriage partners in the studies by Bott (1957) on family structures (Jansen 2006, p. 43).

In the 1960s, a new network research center formed in Harvard around Harrison White of a relational sociology perspective, which led to a renewed interest in network research; as well as by the group surrounding Linton Freeman at UC Irvine California and the group by Edward Lauman and James Coleman at the University of California. Since the late 1970s, network research experienced a successive formalization and upturn in reception under the joined efforts of the “International Network for Social Network Analysis,” founded in 1978. Key methodological innovations included the concept of embeddedness (Granovetter 1985), equivalence (Borgatti and Everett 1989), structural holes (Burt 1992), small world networks (Watts and Strogatz 1998), and power law (Barabási and Albert 1999), to name just a few in the wake of the professionalization of the field of network research since the 1980s.

Further developments of formalized network methodologies and their implementations in case studies have since resulted in an increasingly unmanageable amount of publications since the late 1990s: for example, on exponential random graph modeling (Lusher et al. 2012), temporal network analysis (Freeman 2003), brokerage (Burt 2005), or studies in context-aware network analysis as by my former colleagues Mirko Schönfeld and Jürgen Pfeffer (2020).

Network terminology and methodology are increasingly adopted in other disciplines, such as history. This incorporation of a network research methodology in historical studies is considered a transdisciplinary research paradigm by its practitioners.

In the final part of this series, I will tie together the influence of these inaugurating network methodologies on historical studies, and the creation of the field of historical network research.

---

---

Image Credits: Featured Image: Charles-Joseph Minard: Carte figurative des pertes successives en hommes de l’armée française dans la Campagne de Russie 1812-1813. Régnier & Dourdet 1869. https://gallica.bnf.fr/ark:/12148/btv1b52504201x/f1.item.zoom#.

1. This series draws in large parts on the chapters on historiographical perspectives on the digital humanities and network research of my dissertation (Petz 2022). [↩]
2. Compare also to the “Introduction to historical (social) network analysis” part 1 and part 2 by my colleague Demival Vasques Filho. [↩]

---

---