Introduction: historical network analysis and social groups in the Enlightenment

Dan Edelstein and Chloe Summers Edmondson

While many periods of history are popularly known by their ‘great men’, the Enlightenment stands out for the prominence of its ‘great groups’.¹ Already in the eighteenth century, the leading French figures of the Enlightenment were identified in the plural: they were les philosophes, as the title of Charles Palissot’s 1760 satirical play dubbed them. They banded together in salons, the most notorious gathering being the coterie holbachique; they co-wrote major works, from the Encyclopédie (the product of a ‘société de gens de lettres’, as its title page announced) to the Histoire des deux Indes; they fell in love with each other’s mistresses (the marquis de Saint-Lambert managed to break both Voltaire’s and Rousseau’s hearts). And they had famous falling-outs, usually involving Rousseau.

These writers made up the ‘little flock’ that Peter Gay memorialised in his monumental study.² Since then, historians have shed light on some of the other groups that operated around them. Members of provincial and foreign academies, salon-goers, foreign writers, antiquarians, Freemasons, booksellers, pamphleteers: these are just some of the networks that extended beyond, and on occasion intersected with, the closely knit band of the philosophes.³ The particularly social nature of the Enlightenment was not limited to France: in Scotland, universities created an academic network for professors to develop and debate Enlightenment ideas; in Milan, the irreverent Academy of Fists (Accademia dei pugni) provided the forum where Cesare Beccaria met Pietro Verri, and was introduced to Enlightenment thought; and in the Dutch Republic, Masonic lodges served a similar function.⁴ While sociability was of course central to the intellectual activity of every historical period, what is particularly notable about the Enlightenment is how these different networks were central to their participants’ identity. One could not take part in the Enlightenment on one’s own.

Where historians have widened the parameters for participation in the Enlightenment, technological innovations have also made it easier to identify a greater number of these participants. Comprehensive lists of academy members can now be found on dedicated websites; little-known individuals can be identified thanks to a quick search on the Internet; and the digitisation of correspondences has been a double boon, first in terms of accessibility and searchability of letter collections, but also for metadata production. Indeed, one of the side benefits of such digitisation projects is the creation of spreadsheets filled with names, dates, places and relationships. For instance, it was thanks to metadata produced by the Electronic Enlightenment project at the University of Oxford that the Mapping the Republic of Letters project at Stanford University released its first data visualisations.⁵

While geography offers an effective way to analyse historical data, it typically obfuscates other dimensions, particularly those dealing with individuals. We lose sight of people when we look at points on a map, and, more specifically still, we lose sight of groups. But correspondence data (along with data from other sources, including academies, lodges and salons) can also be used to explore social relations. Points on a map can become nodes in a network, as cartographic space gives way to conceptual space. It is at this intersection of Enlightenment historiography, data capture and social network analysis that the chapters collected in this volume all fall. They take advantage of new data sources, configurations and modes of analysis to deepen our understanding of how Enlightenment sociability worked, who it included and what it meant for participants.

Social network analysis in history

While the sources and methods employed today to study social networks were not available until recently, scholars have long explored the place of networks and social groups in the Enlightenment. In the late 1960s, a team of French researchers around Henri-Jean Martin (including François Furet and Roger Chartier) explored the networks that connected authors, censors and printers to booksellers and collectors.⁶ A complex network of paper producers, printers, publishers, booksellers and authors was also the focus of Robert Darnton’s study of the Encyclopédie’s quarto edition (1775-1800).⁷ Other types of networks attracted equal attention: Alan Charles Kors examined the social make-up and connections of the coterie holbachique;⁸ Daniel Roche explored the constitution of provincial academies;⁹ Frank Kafker analysed the social relations of the encyclopédistes;¹⁰ and the salons, which had been a source of fascination since the nineteenth century (Sainte-Beuve dedicated many studies to them), were the subject of monographs by Dena Goodman and Antoine Lilti, among others.¹¹ Correspondence networks were not overlooked either, with Lorraine Daston, Anne Goldgar and Laurence Brockliss calling attention to the different types of networks that existed in this period, and highlighting important formal and rhetorical differences between the older, erudite Republic of Letters and the newer, worldly connections of the Enlightenment.¹²

If scholars have long zeroed in on social groups as a key to understanding Enlightenment practices, it is in no small part because their importance had already been recognised in the eighteenth century. To be sure, réseau was not yet used in the contemporary sense of a network. But one can find plenty of cognate concepts. If society was redefined, during the Enlightenment, as ‘the ontological frame of our existence’, it still also retained the more voluntaristic sense of an ‘union de plusieurs personnes pour quelque objet qui les rassemble’.¹³ All were members of society, but only some were members of the Royal Society. The philosophes similarly appropriated the notion of a parti, particularly in their political skirmishes at the French academy.¹⁴ Here, too, we find a cognate concept to a network: both refer to a delineated group, with some control over membership access, which shares a set of goals.

So what is different about exploring Enlightenment sociability from the perspective of networks? One might be tempted to answer ‘data’, but that would be to forget the many earlier studies that also relied on substantial amounts of data. Indeed, the Annales school, in which many of the above-cited French historians were trained, pioneered the use of statistical methods in history. Spreadsheets only became available in the 1980s, but these earlier historians still crunched large data sets and produced sophisticated data visualisations.¹⁵

A more noticeable difference is the adoption of social network analysis (SNA) by historians, a methodology developed for the social sciences that lies at the intersection of social theory, mathematics and statistics, and has proved fertile for interdisciplinary studies over the past few decades.¹⁶ Albert-László Barabási calls it ‘the next scientific revolution, the science of networks’.¹⁷ At the most fundamental level, a social network is a system of actors (nodes) and the ties between them (edges). The actor is a social entity that can range from an individual in a group to a department in a corporation. The other key components are the ‘relational ties’ between actors. These ties can be of very different natures and are sometimes referred to as ‘structural variables’. Variables allow for the filtering of the larger network into subgroups of actors.¹⁸ An SNA study will include not only information about the individual actors, but also data about the relationships between them.

The main attraction of SNA is that it allows for the study of the network as a whole, not just the ties between two or three actors.¹⁹ Networks are systems: changes to one part of the structure can affect all other parts. As a system, the network ‘provides opportunities or constraints on individual actions’.²⁰ By disseminating information and resources, social networks create shared interests, identities, norms and values, providing pathways for communication of all sorts: knowledge and ideas, as well as gossip and rumours.²¹

SNA can be applied to almost any type of network. For instance, many of the networks examined in this volume are ‘ego-centred networks’ (i.e. networks structured around a focal person and the set of actors with ties to that individual).²² In all cases, the primary objective of SNA is the same: to transform the network into a mathematical matrix so that observers can quantify the dimensions of the whole, as well as of all its parts. Common calculations include the centrality of a node (how important it is in the system), its degree (the number of edges attached to it), its eccentricity (the distance between a given node and the other nodes in a network); it is also possible to determine which nodes are bridges (the nodes critical to the connectedness of the network). Within networks, there are often a few actors with an extremely high degree (a high number of ties) that constitute hubs, as well as clusters of closely knit actors within the broader network.²³ All these measures can be determined quantitatively, though this reliance on calculation may pose problems for historical data.

Social scientists often stress the important work that goes into curating the data before it can be analysed with SNA methods. The researcher must establish what or who are the nodes in the object of study and how they are connected (i.e. the types of relational ties). Then comes the important task of defining the network. How do you determine membership? Does it have to do with frequency of communication or interaction, and how is that frequency established?²⁴ As Mark Newman notes, conceiving of objects as a network can produce new and fruitful insights in many fields. But the network also simplifies the system by reducing it to an abstract pattern of connections, and a lot of information will accordingly be ignored. Not all tools are suited to all networks, and the researcher must tailor the SNA methods employed to what is appropriate to the object of study.²⁵

Indeed, the issues that historians face when applying SNA mainly reflect differences in the quality of historic sources. Social scientists also need to curate the data before it can be analysed with SNA methods. Yet, as Claire Lemercier points out, a major problem for historians is the inability to compare networks in a formal way, since sources tend to be particularly patchy.²⁶ In such cases, the researcher must tailor the SNA methods to fit their object of study.²⁷ This approach is accordingly best viewed as a valuable complementary tool or technique for historians, especially when used in conjunction with qualitative analyses.²⁸ Lemercier argues that the greatest value of SNA for the historian is that it provides measurements and schemas based on something other than intuition.²⁹ Thinking in network terms and adapting SNA concepts can sometimes be more valuable than the calculations themselves.³⁰

Historians have already begun to demonstrate how these concepts can be deployed effectively to a vast range of topics, particularly in early-modern studies. In volumes such as Communities in early modern England: networks, place, rhetoric, edited by Alexandra Shepard and Phil Withington, scholars examine manuscript networks, the networks of female medical practitioners, networks of Catholic dissent and social networks in Restoration London.³¹ In Poetry and the police: communication networks in eighteenth-century Paris, Robert Darnton endeavours the difficult task of reconstructing a communication network concerning the dissemination of satirical poems. Through police records of the investigations, Darnton is able to reconstruct a communication network that takes into account both written communication and the traces of oral communication (spoken rhymes and songs) found in the records, revealing processes by which information circulated throughout the streets of Paris in early-modern times.³² William Warner has rearticulated the American Revolution as an ‘event in the history of communication’ through an analysis of its (epistolary) communication networks. Warner examines how networks shaped the unfolding of revolutionary events, and argues that innovations in communication and the networks that formed as a result afforded the conditions of possibility for the American Whigs to network their way to power for a full-blown revolution of colonial America.³³ More recently still, Ruth Ahnert and Sebastian Ahnert have reconstructed the underground Protestant community living under the reign of Mary I in England, using the mathematical and computational tools of SNA to analyse correspondence data, uncovering features of the organisation of this networked community.³⁴

These are just a few of many historical studies and projects that have emerged in recent years, which adapt network analysis to historical research. Not only do these applications of SNA make valuable contributions to historical research, but, as Lemercier also proposes, historians are uniquely situated to provide insights into social network analysis through their deeper understanding of social, historical and political context, and of the historicity of different kinds of relational ties and of agency in a social environment.³⁵

When your data isn’t edgy

But the adoption of SNA methods by historians at times demands particular inventiveness and a thorough reimagination. The extent to which one must be methodologically creative largely depends on the kinds of data at one’s disposal. Most of the data we can gather from correspondence networks, for instance, comes in the form of ‘ego-networks’: we know whom Voltaire wrote to, and who wrote to him. But we rarely have much information about connections between the correspondents themselves. On some occasions, we can superimpose multiple ego-networks and discover more relationships; and we can always fill in relationships between correspondents with data from other sources. But our data sets will invariably remain grossly incomplete.

This problem is particularly acute for correspondence data, but it is a common issue with many types of historical data. Attendance records for Enlightenment salons are very patchy as well, and it is exceptional when we can establish that two individuals met each other in a salon (i.e. attended at the same time). Data about academy memberships tends to be more complete, but again we can often only infer relationships between members.

These problems may sound like just an inconvenience, as though it simply means that one needs to do more work cleaning up and enriching the data to ready it for SNA. But they are in fact more insurmountable than that. One could spend ten years in the archives digging up relational data about Voltaire’s correspondents, and still be nowhere close to recreating all the existing edges. In most cases, the data simply isn’t there. The historical record will never offer the same level of accuracy as, say, a Twitter feed (or even a novel or a play, for that matter).³⁶

From a methodological perspective, the difficulties we confront with these historical data sets can be reduced to one major challenge: we can often ascertain with great confidence that two individuals knew each other, but it can be much more difficult, if not impossible, to determine whether two individuals did not. This is a considerable hurdle if you wish to adopt SNA methods to analyse your data. Typically, SNA metrics require precise and accurate data about edges, both existing and absent. Most of the calculations used in SNA involve a contrast between potential relationships and existing ones. Network density, for instance, is calculated by comparing the number of existing relationships between nodes with potential ones; a similar ratio is what determines the clustering coefficient of a node or network (see Figure 1).

In our case, however, we do not know whether a potential relationship between two nodes was actually non-existent (i.e. that person A had no connection with person B), or simply that this relationship was not recorded anywhere. For this reason, standard SNA methods – say, calculating degree distribution, clustering coefficients, network density or average path length – can be meaningless when applied on such a network. Worse yet: they would produce results that seemed perfectly valid, but that grossly distorted historical reality.

Studying historical networks, then, is not simply a matter of importing existing sociological methods. We need to seek out another approach. If our data isn’t very ‘edgy’, we do have more reliable information about the nodes themselves. Here as well, we are surely missing some: in the case of Voltaire, for instance, it has been estimated that we have lost about half of his total correspondence (and there are nearly 19,000 extant letters).³⁷ But the total number of missing correspondents is probably much lower, given that most of the loss is from Voltaire’s inbox (i.e. the letters he received). So, even though we are missing a large number of letters, our data about his correspondents is more reliable. Even if we lose 90 per cent of the letters exchanged between Voltaire and a correspondent, so long as at least one letter survives, the person will not be lost.

Using node data to analyse social networks may sound like a strange idea: what does someone’s birth year, for instance, tell us about their connections? In fact, prosopographical data contains many social indicators; and when it is enriched and multiplied, it can shed light on social connections.

We realised the potential of prosopography for studying networks when we received a data dump from the Electronic Enlightenment project. In addition to letter metadata, the EE researchers had collected basic information about correspondents: date and place of birth and death, nationality and occupation. This last dimension, in particular, struck us as promising: knowing that someone was a ‘playwright’, and lived in a given place during a certain period of time, was a valuable piece of social knowledge. In all likelihood, this person would have belonged to a number of social networks around, say, theatres and literary authors. If you combine that knowledge with data about salon attendance, academic membership, social status and bibliography, a fairly detailed picture begins to emerge about the person’s milieu and connections. It can also be valuable, however, to synthesise the many details that metadata provides into broader categories that can help bring the bigger picture into focus.

Some of the authors included in this volume pioneered this approach in a recent article on ‘The French Enlightenment network’, inventing a prosopographical schema (which we called ‘Procope’) to organise data in a controlled vocabulary.³⁸ This schema was designed to capture the social, knowledge, professional and religious affiliations, or ‘networks’ as we called them, characteristic of individuals in early-modern Europe. These groupings have a certain heuristic benefit as loose categories that reflect actor categories used at the time, such as gens du monde or gens de lettres. In our ‘Procope’ schema, those individuals would be classified, respectively, as ‘Aristocracy’ (or ‘Elite’) and ‘Letters’, or, in some cases, both. It goes without saying that these classifications are done according to the metadata available. Individuals sorted into ‘knowledge networks’, for instance, are almost exclusively published authors, or otherwise known to have been very involved in the world of letters or sciences, while ‘Aristocracy’ includes individuals with noble titles. In ‘Procope’, the schema was designed to be hierarchical, to account for specificities while at the same time being as inclusive as possible. Therefore, a well-known individual who published on theological matters would be placed in the knowledge network ‘Letters_Religious’, while an individual for whom little information is available might simply be categorised in ‘Letters’. Individuals are also placed into as many ‘networks’ as possible to recognise the multiple components of their identities and engagements within society. The marquis de Condorcet, for instance, would be categorised as ‘Letters_Literary’, ‘Political Economy’, ‘Sciences_Mathematical’, ‘Aristocracy’ and ‘Government’.³⁹

The point of these classifications is not to make absolute judgements about individuals’ identities, even less to make a claim, for instance, about what the sciences were in the eighteenth century, but simply to provide a methodical system for sorting people into general categories. Employing a schema is helpful, even essential, in data analysis because, if the individuals were only tagged according to their specific biographical information – as a duke or a prince, a poet or a novelist, a priest or a cardinal – it would be impossible to identify the larger patterns within the data set, study the subnetworks of ‘Aristocracy’, ‘Sciences’ and ‘Clergy’, and trace the overlaps between them. Such classifications make visible macro patterns that provide rich and nuanced insight into the social, epistolary and academic networks of early-modern Europe. In the chapters that follow, many of the authors in this volume employ prosopographical schemas in their studies, either applying ‘Procope’ to a new object of enquiry or generating their own schemas along similar lines to suit their data and research questions.

Adapting a classification system tailored to early-modern data is precisely the kind of work that we must do when adapting SNA methodologies to historical data. The results we obtain by focusing on nodes, rather than edges, are of course of a very different sort from what SNA metrics provide. But the questions we hope to answer as intellectual, cultural and literary historians are also of a different sort. In ‘The French Enlightenment network’, for instance, we focused on the remarkably high number of aristocrats and State officials among the correspondents of Voltaire, Rousseau, D’Alembert and others. And we also noted the extensive overlap between two social groups known at the time as gens de lettres and gens du monde. Our approach was still based in quantification, but we counted and measured different values.

There is of course no reason why historians and sociologists should study the same questions. It is precisely because scholars of different stripes are interested in different aspects of human society and culture that we have multiple disciplines. In adapting SNA for historical data, however, we were led to question some of its sociological premises. In particular, we came to ask ourselves whether SNA had a sufficiently robust theory of the social, and whether a core SNA concept, ‘small world theory’, made sense from a more humanistic perspective. It is to these questions that we turn in conclusion.

Social groups and identity

Group identity is a critical element for understanding how social networks are structured, how individuals interact between groups, and why we seek out membership in groups in the first place. To be sure, for some kinds of SNA, group identity is of no concern. Epidemiologists, for instance, do not necessarily need to factor social memberships in to study the spread of the disease. Diseases may spread more rapidly among the members of a given group, but they also spread quickly and seamlessly from one group to another, given the opportunity. Production networks, too, need not take group identity into account: iPhone owners in New York are part of an economic network which extends to Foxconn workers in China, whether or not they identify as such. But for other kinds of networks, group identity will determine what spreads and what remains within the group. This is because information circulates differently from disease. Voltaire might indirectly catch a cold from a Savoyard merchant, who had passed the virus on to a farmer in Ferney; but it is highly unlikely that the Savoyard merchant would indirectly catch Voltaire’s deism (see Figure 2).

This distinction has unfortunately been overlooked by many social network theorists. Possibly this oversight reflects the fact that some of the earliest work on social networks involved information transmission. In Stanley Milgram’s famous experiment, inhabitants of Omaha, Nebraska and Wichita, Kansas, were asked to send letters back to Boston, Massachusetts, forwarding them to whomever they might know along the way.⁴⁰ The point of the experiment was to demonstrate that we are much more closely connected to other people in society than we imagine (hence the name Milgram gave to his study: ‘small world’). But there was another element of Milgram’s experiment that received less attention: only 28 per cent of the letters came back.⁴¹ When Duncan Watts and his group repeated Milgram’s experiment using email, on a much larger scale (60,000 participants, global targets), the completion rate was even lower: a mere 1.6 per cent of chains were completed.⁴²

Because these and other studies have reinforced the finding that there are roughly ‘six degrees of separation’ between any two people on the globe, the fact that so many chains go uncompleted is rarely viewed as problematic.⁴³ But it points to another feature of social networks that is equally crucial: namely, the filtering function that determines what kinds of information we transmit. While this filtering function is normally operative in every social encounter between individuals, it is particularly acute with respect to information sharing between groups (see Figure 3). You do not forward your cousin’s chain letter to your work listserv; and you might not reveal your political views to certain family members. There are of course instances when information spreads across social groups like a disease; then we give this process a biological name (‘going viral’). But the viral transmission of information is obviously the exception rather than the norm.⁴⁴ Most information does not circulate broadly, but stays within the confines of a social group. Gossip is a good example: its social currency is only recognised by those who are familiar with the protagonists. Gossip helps to solidify the ties of a social network, but in so doing also establishes its boundaries.⁴⁵

As every teenager knows, social groups are just as much about exclusion as about inclusion. The filtering function between groups thus serves a dual purpose: we regulate our own flow of information so that it is relevant (and appropriate) to the audience in question;⁴⁶ but we also withhold information from others in order to preserve (and enhance) its value. To ignore the boundaries between social groups, then, is to ignore a fundamental feature of social life: as Groucho Marx famously joked, all we really want is to join the groups that will not let us in. It is the existence of these boundaries that makes social networks genuinely social: indeed, ‘a phenomenon may be collective without being social.’⁴⁷

Rather than embrace small-world theory, a more humanistic approach to social network analysis could centre on a ‘many-worlds’ theory, in which individuals who belong to different social groups may only be separated by a few degrees, but can still be ‘worlds apart’, socially speaking. This approach is particularly helpful for studying networks in early-modern times, when societies were even more hierarchic. Social groups at this time also tended to be more sharply delimited: individuals who belonged to these networks recognised them as defined collections of people.⁴⁸ Belonging to one of these networks was an act of choice: it meant something. This meaning was in fact twofold: as an individual, participating in a network brought certain benefits, in the form of social capital;⁴⁹ but as a group, the network also had socially recognisable meaning. In a word, it had an identity.⁵⁰

But where does group identity come from? Bruno Latour has argued that ‘to delineate a group, no matter if it has to be created from scratch or simply refreshed, you have to have spokespersons which “speak for” the group existence.’⁵¹ In the case of the Enlightenment, this network function (of ‘speaking for the group existence’) may help us better understand the role of the philosophes. Conceiving of the philosophes as spokespersons who established the social identity of the network can indeed help clear up some confusions. The philosophes were not the ideological masterminds of any specific ‘agenda’, nor the ringleaders of an activist party. What they did was literally ‘speak for’ the Enlightenment by corresponding with the aristocrats, military officers, civil servants and others who wanted to gain admission into this group. They also functioned as the default gatekeepers for the group: receive a response from Voltaire, or an invitation to Lespinasse’s salon, and you were in. But you could also be kicked out, as when Voltaire excommunicated Charles Palissot for mocking the Parisian philosophes.⁵²

Chapter overview

This volume presents a series of case studies that draw upon newly available data sets to explore the social networks of the Enlightenment period, ranging geographically from Russia and Sweden to Italy and England, with particular focus on France. These studies engage with methodological approaches drawn from SNA, but also pioneer new historically driven methods for thinking about networks in early-modern societies that combine quantitative and qualitative methods. Many studies draw on metadata from digitised correspondence, academy membership, data extracted from texts and self-generated data sets from correspondence, memoirs and biographies. Authors thus not only examine networks of diverse natures and compositions, but also use the term ‘network’ in very different ways. While for some the network in question concerns concrete instances of communication as in a correspondence network, others understand the network as the members of large institutional organisations. For others still the network is a group of people who interacted socially on a regular basis, or a group of acquaintances and contacts built over a lifetime of travel. Many use the term ‘network’ to think about early-modern social identity within the broader network of study in order to capture the distinctions between aristocrats or government officials for instance, while others still consider fields of study or intellectual schools as networks of knowledge. As such, the volume groups the chapters in three parts: correspondence networks, social networks and knowledge networks.

Part I: correspondence networks

The case studies in the opening section of this volume focus largely on the ways in which individuals leveraged correspondence networks to increase their own celebrity. Nicholas Cronk asks us to consider what it means to analyse and interpret a correspondence network. This chapter addresses methodological issues surrounding the study of correspondence networks, taking Voltaire’s vast epistolary corpus of over 18,000 letters as a case study. Reconstructing this network poses several challenges, as many letters are missing or destroyed, some of those known to us are forgeries, and many still are literary performances. Cronk proposes a typology to distinguish between different types of letters: personal letters, public letters and semi-public ones, where there is an expectation they may end up in the public sphere. Cronk concludes with some reflections on how to further studies of Voltaire’s correspondence network through new digital methods, text analysis, social network analysis and, in particular, ways to reconstruct the holes in the network.

Kelsey Rubin-Detlev and Andrew Kahn examine Catherine the Great’s correspondence using the basic concepts of SNA, and reveal certain defining features of her epistolary network. This analysis of the subnetworks that composed her larger ego-network (administrative networks, diplomatic networks, courtly networks, the network of European heads of State, as well as the parental, amorous and cultural networks) is complemented by a more in-depth look at a specific epistolary episode. Catherine’s matchmaking for her son Paul serves as an illustrative example of how correspondence played a key role in consolidating the empress’ power at home and abroad. Rubin-Detlev and Kahn demonstrate how Catherine the Great influenced Europe’s power-brokers, managed her own and Russia’s image abroad, and promoted her celebrity as an enlightened ruler, all through epistolary networking.

Cheryl Smeall examines how Francesco Algarotti’s published and unpublished correspondence with over 145 individuals allowed the Venetian to establish his networks and make a name for himself across Europe. In what can be described as a reverse Grand Tour, Algarotti extended the in-person networking from his extensive travels, where he gained access to intellectual circles across Europe, through active correspondence, building scientific, literary, artistic and political networks. Smeall shows how this was important not only for his international celebrity, but also for the success of his Il Newtonianismo per le dame. Smeall demonstrates how Algarotti combined his extensive travelling with correspondence to activate and maintain networks everywhere he went, expanding his international renown in tandem with his networks.

In a similar vein, Pierre-Yves Beaurepaire provides a rich window into the life story of the Huguenot pastor Jacques Pérard through an analysis of his unpublished correspondence. Beaurepaire highlights how Pérard played a crucial role in the development of the French-language press in Prussia and in the Baltic area. An analysis of his epistolary strategies reveals how Pérard used correspondence for the purpose of intensive cultural and social networking, to further his position in the European French-language book and periodical trade. Beaurepaire demonstrates that, through communication covering a broad expanse of territory, Pérard was able to gain and maintain a foothold in the Republic of Letters that extended throughout Europe.

Part II: social networks

The next section of the volume moves from correspondence data to social networks that are reconstructed from the far less tangible data of who interacted with whom in milieux of sociability. These authors reconstruct social networks from data extracted from memoirs, biographies, correspondence and police records, employing the broadest notion of the concept of a ‘network’ – including in the network anyone the individual of interest is documented as having met at least once.

Chloe Summers Edmondson addresses the question of how to distinguish the ‘philosophical’ from the mondain salon in eighteenth-century France through a case study of Julie de Lespinasse’s salon. Often referred to as the ‘muse’ of the Encyclopédie, Lespinasse earned her claim to posterity through the prominent salon she hosted in Paris from 1764 to 1776. While considering Lespinasse’s correspondence and the testimonies of her contemporaries, the primary objective of this chapter is to analyse the demographics of the broader network of Lespinasse’s salon attendees, with particular attention to its intellectual composition. Edmondson makes a case for an ‘enlightened’ breed of salon, demonstrating how several aspects of Lespinasse’s network coalesce to substantiate the intellectual dimension of the salon: from its integration with published authors, the academies and correspondents of major Enlightenment figures, to its similarities and overlap with the baron d’Holbach’s salon.

Charlotta Wolff uncovers the intersections between diplomacy and the Enlightenment through her analysis of the Parisian networks of Swedish ambassador Gustav Philip Creutz between 1766 and 1783. This quantitative and functional analysis of Creutz’s networks in Paris based on data from French police records reveals that his network was composed of aristocrats, diplomats, elite women, philosophes, writers, artists, patrons and collectors. His network of sociability was thus, as Wolff argues, mondain, political and philosophical all at the same time. This case study ultimately is suggestive of the cultural significance of the Enlightenment as a movement at this time – being a part of the enlightened networks mattered, even for a diplomat. Creutz’s network highlights the role of diplomats and practices of sociability for the transmission of ideas in Enlightenment Europe.

Maria Teodora Comsa traces the creation of Giacomo Casanova’s personal network of influence in Paris circa 1750 using his memoir Histoire de ma vie as a data source. Comsa examines how Casanova developed his network from a small theatrical coterie to connect with some of the most prominent members of the Parisian elite. Comsa reveals that the theatre network was more powerful and better connected than one might expect, serving in fact as Casanova’s portal into high society. Manipulating three networks – the theatre network, the intellectual network and the French aristocratic network – Casanova pursued a conscious and methodical plan in his social networking, going to the theatre to make important social connections, and deploying his talents of wit and social performance in forging his network.

Part III: knowledge networks

In the last section of this volume, contributors investigate ‘knowledge networks’ in two case studies of very different natures, and employing two distinct methods. Melanie Conroy analyses the complex role of the academies in the Republic of Letters using a prosopographical and quantitative approach. In this study, there is a particular focus on the relationship between academic membership and geography, as well as the presence in the academies of the sciences, the encyclopédistes and correspondents of major Enlightenment figures. Conroy demonstrates that the French academies formed a single network, based in Paris, which extended throughout France and across Europe, and suggests that this academic network was fundamental to the development of the Enlightenment precisely because it drew on the power of State sponsorship.

Finally, Mark Algee-Hewitt examines the knowledge networks present within a text, Johnson’s Dictionary, and the system of interpretation created through its network of citations. Algee-Hewitt reassembles and makes visible the embedded knowledge system of Johnson’s Dictionary by computational methods. What emerges is an understanding of Johnson’s Dictionary as an implicit work of literary criticism at a foundational moment for the formation of the English canon. By exploring the groupings and metrics of this hidden network, Algee-Hewitt sheds light on the relationship between definition and knowledge domain in the eighteenth century, as it offers a new history of Enlightenment textual authority that remained embedded in British lexicographic practices for the following two centuries.