Interrogating fiction’s civic effects

In order to answer these questions, Civic Fictions will focus on the historical period generally viewed as foundational for the development of the modern novel, and for debates on the function of the novel,² the long 18th century and especially the Age of Revolution (c. 1760–1830), by carrying out a series of longitudinal, prosopographic studies of readers and their books during this period. Drawing on the mass of bibliographic and biographic data that has been harvested, curated and made available over the past decades through multiple digitisation projects, it seeks to develop new methodologies enabling historians to use large-scale data on the circulation of books among identifiable, individual readers as an effective reader-reception source. Thus, the project also hopes to build new digital infrastructure for future inquiries by aggregating different types of source material, leveraging digital tools and, eventually, the potential of Linked Open Data (LOD) to make data from different sources and national contexts interoperable.

Civic Fictions starts out by operationalising a number of frequently invoked effects that fiction supposedly has on readers. Modern-day literary theorists and historians have identified several mechanisms that might have historically enabled fiction to foster citizenship. Thus, historians like Lynn Hunt, supported by philosophers including Martha Nussbaum and Richard Rorty, have hypothesised that fiction can elicit the empathy and perspective-taking essential to putting oneself in another’s position and hence recognising others’ humanity and rights (Hunt 2007; Nussbaum 1995; Rorty 1993). Benedict Anderson influentially theorised that fiction helps individuals imagine new societal formations or ‘imagined communities’, aiding both processes of nation-building and – more problematically – of imperialistic expansion (Anderson 1983; Bell 2021). And historians and philosophers have argued that phenomena like subscription libraries, and other forms of reading-related sociability in coffeehouses and salons, decisively helped constitute civil society, according to Jürgen Habermas’s influential thesis that the literary public sphere was the precursor of the ‘public sphere’ tout court (Habermas 1989).

But arguing against the empathy theory of fiction, other theorists have postulated instead that fiction achieves its effects not through empathy, but on the contrary through a process that might be described as ‘empathic unsettlement’. Psychoanalytically-inspired theorists have in particular argued that fiction can aid individuals working through loss and trauma, through mechanisms such as displacement or transference (Caruth 1996; LaCapra 2001). Literary trauma studies often adopt a multi- or trans-generational focus, exploring how fiction can link past and present in sometimes unexpected ways, shaping memories and (re)writing history in the process (Hirsch 2012; Rothberg 2009). Such memory work, acting out and working through trauma may thereby support collective, political community formation, and underwrite original new forms of citizenship.

Book historians like Elisabeth Eisenstein, finally, have studied how specific books, as ‘agents of change’, may have historically incited readers towards political action, catalysing sweeping societal transformations. In this context, the 18th-century Age of Revolution has attracted particular attention, with commentators and critics of the French Revolution, from its inception to the present day, repeatedly asking: ‘Do books cause revolution?’ (Eisenstein 1980; see also Febvre and Martin 2010; on the books and revolution thesis, see Burrows 2015 and Mornet 1933). The most recent iteration of this perennial historical question, the so-called Darnton thesis, according to which fictional genres like utopian science fiction and philosophical pornography ultimately tipped readers towards revolt, has been especially productive, generating a rich body of scholarship of its own (Darnton 1996).³ More recently still, adopting an explicitly global, even postcolonial perspective, and paying attention to the ‘reverberations of revolution’ over time, other historians have posited that 18th-century print culture played a central role in issuing a ‘call to liberty’ that resonated with individuals across the globe, from bourgeois housewives to enslaved Africans (Polasky 2015; see also Amann and Boyden 2021).

Accounts of fiction’s effects on readers, however, raise several issues. First, literary scholars and historians – not surprisingly, given their own professional affinities – generally perceive these effects to be beneficial. Fiction’s more sinister potential – inclining readers toward hate rather than empathy, totalitarianism rather than rights-based regimes – remain understudied.⁴ Yet from Plato to today’s book-banning activists, there have also been more critical perspectives on the place of literature in the polis.⁵ Might fiction’s effects, as Plato controversially suggested, actually be morally harmful, justifying his proposal to ban poets from the ideal republic? Just as problematically, modern-day accounts of the effects of fiction on readers operate at a high level of theory, or are based on evidence that is anecdotal or fragmentary at best. Attempts to validate the empathy–altruism hypothesis empirically remain inconclusive. Psychologists note that ‘the relationship between empathy and prosocial behaviour is neither direct nor inevitable’ (Eisenberg and Strayer 1987, p.11). Readers may empathise with fictional characters while immersed in reading, yet fail to subsequently transform that empathy into substantive action, political or otherwise. Conversely, readers may resist or reject meanings overtly conveyed by books, thwarting fiction’s citizenship-building potential (Chartier 1997; Fetterley 1978; Rose 2018).

Given these issues, one approach to the problem of measuring fiction’s effects on readers has been to collect and analyse readers’ own accounts of their reading experience. This has resulted in a number of ambitious undertakings, most prominent among them the UK-based RED (Reading Experience Database, 1450–1945) and its national counterparts elsewhere, which seeks to document every extant ‘recorded engagement with a written or printed text – beyond the mere fact of possession […] as much information as possible about what people read, where and when they read it and what they thought of it’. The RED project has to date produced some 30 000 records, and has inspired much follow-up research. At the same time, however, it has also drawn attention to one of the major problems in book history: the lack of comprehensive, qualitative reception data on how individuals actually responded to fiction in specific contexts.

Focusing on the 18th century, an alternative favoured by some historians has been to adopt instead a micro-historical approach. This has resulted in a number of important studies of ego-documents left by individual readers during the course of the long 18th century, documenting their reactions to specific works of fiction (Blaak 2009; Brewer 1996; Darnton 1984; Towsey 2019). Yet in self-reported accounts of reading, readers often shape their narratives to describe responses that fit into overarching cultural discourses. Readers may selectively tailor their accounts, leaving out books read but judged of little prestige, and they may provide socially desirable accounts when recording their own reading experience, reflecting short-lived reactions rather than long-term transformations.

The unreliability of self-reported accounts of reading reactions, coupled with the extreme paucity of historical, direct reading-reaction sources, has led some reader-reception theorists to propose that the solution may lie elsewhere. Suzanne Keen, in her insightful study Empathy and the Novel, has thus pointed to ‘the need for longitudinal studies of readers’ lives’ in order to establish relations between reading and (political) action (2007, p.xiii). Creating such longitudinal studies is, however, a massive undertaking, and presents daunting logistical issues if working with modern-day readers, and seeking to document life-long reading careers spanning 50-year periods or more. This is one reason why turning to the past may provide a way out of this reader-reception scholarly dilemma.

Leveraging digital tools to understand novels’ effects on readers

Indeed, longitudinal studies of readers’ lives are precisely what the historical archive is potentially able to provide, as decades-long work in historical bibliometric databases across the world is beginning to bear fruit. The 18th century has attracted some of the most exciting digital humanities research to date – a fact to which the launching of the present journal amply testifies – and some of the most comprehensive collections of book-history sources available. Extensive, curated and searchable data on books that circulated during the Age of Revolution, roughly understood as the period 1760–1830, linking books to owners and readers, is available not only in databases such as the UK’s Arts and Humanities Research Council-funded Libraries, Reading Communities and Cultural Formation in the Eighteenth-Century Atlantic database (henceforth Reading Communities database), currently under construction,⁶ and Western Sydney University’s French Book Trade in Eighteenth-Century Europe (FBTEE) database.⁷ The University of Helsinki’s Eighteenth-Century Collections Online (ECCO) metadata reuse project allows historians to reuse in their own databases the metadata created by Gale for its commercially marketed ECCO database of 180 000 titles published in the 18th-century British Isles (Tolonen, Mäkelä and Lahti 2022). And other, older databases in addition are on the cusp of being made available in new digital formats, such as the Mapping the Eighteenth-Century French Novel database, also at Western Sydney University, that will integrate the still-unsurpassed MMF-2 bibliographic database of the French novel, originally prepared by Angus Martin and his late collaborators, Richard Frautschi and Vivienne Mylne, in the 1970s, and updated ever since (original print publication: Martin, Mylne and Frautschi 1977).

More interesting still for the purposes of establishing the effects of fiction on readers, many of these databases also include more or less extensive data on readers, making new economies of scale feasible. The Reading Communities database, currently under construction at the University of Liverpool under the direction of Mark Towsey, is collecting in a single open-access database contextualised bibliometric data on book holdings, membership and book usage at subscription libraries in Britain and North America in the period 1731–1801, together with other lending-record databases addressing the period 1750–1830. My own European Research Council-funded MEDIATE database of 18th-century private-library catalogues (Measuring Enlightenment: Disseminating Ideas, Authors, and Texts in Europe, 1665–1830) gathers together data on over half a million individual books, each of which was once in possession of an identifiable owner.⁸ Most of these 18th-century owners have left multiple traces in the historical record, while a few are exceptionally well-documented individuals due to their ecclesiastic, cultural or government functions. Catalogues’ descriptions of their books may include details – such as the presence of interfoliation or annotations by the owner, sometimes even reproducing these annotations verbatim – indicating how these books were used. And the totality of these half-million books actually represent what Franco Moretti has described as a comprehensive ‘literary system’,⁹ or, as I define it on the MEDIATE project website, ‘a set of relations between higher- and lower-prestige texts, geographic regions and languages, and between authors closer and farther away from centres of cultural authority’. Understanding a library as a system provides context for understanding the significance of the reception of any individual title by a specific reader, and allows researchers to harness quantitative, ‘library profile’ analyses to understand how any individual work might have been understood by its readers.¹⁰ Within the framework of Itamar Even-Zohar’s polysystem theory, a systemic approach also sheds new light on the ways in which literary relations between different geographic regions and languages may have affected long-term processes such as canon formation and the global impact of a movement such as the Enlightenment.¹¹

These bibliometric databases, together with other digital tools, large-scale bibliographic databases and digitisation projects, have produced repositories containing literally millions of pieces of (meta)data on 18th-century persons and books, including rich data on book owners and readers, and presenting new possibilities for historians. There are now enough resources, technical expertise and analytic tools to aggregate and compare the source material available, using data that historians were unable to access as little as a decade ago. Economies of scale, for the first time, allow us to counteract the qualitative issues of comparison across different, often incomplete datasets by harnessing the power of algorithmic processes, using Semantic Web technologies to interlink data across datasets. The University of Liverpool’s Reading Communities database, and Western Sydney University’s FBTEE databases are currently in the process of being integrated into a common database framework, using the Heurist open-source online database builder tool. Similarly, dynamic data-linking workflows have already been established between the MEDIATE database and the early-modern bibliographic open-access thesaurus managed by the Consortium of European Research Libraries (CERL). The CERL Thesaurus aggregates bibliographic data created by over 300 research libraries in Europe and thousands of librarians and researchers worldwide, and currently contains 1.3 million records documenting forms of imprint places, imprint names, personal names and corporate names found in material printed before the middle of the 19th century. The CERL Thesaurus is updated and enriched with new data on a daily basis by a distributed network of hundreds of librarians across Europe and beyond. Thus, creating new authority files for Persons (book owners) and other entities directly in the CERL Thesaurus allows us both to use them in the MEDIATE database, through this dynamic record linkage, and allows the broader community of researchers consulting the CERL Thesaurus to benefit directly from the research work carried out in individual projects.

The Civic Fictions project proposes to take the data-linking work already underway one step further, by combining large-scale data on the circulation of books with biographical data on owners and readers, in order to develop a new model to understand the effects that books may have had on 18th-century readers over the course of their lives. Ever since intellectual historians started experimenting with digital, web-based quantitative tools in the 1990s, their transformative potential to not only expand, but fundamentally alter the datasets we work with has been evident (Burrows and Roe 2020; Tonra 2021; Moretti 2005). So what happens, this project asks, when the dataset we work with is no longer hundreds or even thousands of books, but millions, in a more accurate reflection of the ever-expanding European book market, which has been estimated, in the second half of the 18th century, at some 628 million individual copies for new books alone (Buringh and van Zanden 2009)?

The individual postdoctoral and PhD subprojects within Civic Fictions will feed into the project’s digital backbone, which will be constituted by an ecosystem of major historical bibliometric databases. These include both the above-mentioned MEDIATE and FBTEE databases, and other historical bibliometric databases currently under construction. In addition, this ecosystem will be supplemented with significant additional datasets to be harvested by the postdoctoral and PhD researchers – primarily taken from the 18th-century Luchtmans bookseller-publisher archives, from the Dutch KB–National Library lending records, and archives on French Revolutionary library confiscations, on which more below. Thus, the Civic Fictions project will create tools to query all this data simultaneously through a common interface, ultimately allowing historians to address a wide range of questions. Of most direct relevance to the project, these include a set of complex, interrelated questions about the historical relationship between reading fiction and citizenship:

• Can processes related to empathy, trauma and community building be linked to specific works of fiction circulating between 1760 and 1830, as evidenced by the societal interventions of readers identified through ownership and borrowing patterns?

• How do these works of fiction relate to other macro-patterns and clusterings of (non-fictional) books in the larger corpus of books circulating in this period, i.e. to the broader literary polysystem, as well as to the 18th-century publishing field?

• What is the historical readership of these works of fiction, both in terms of individual readers and larger categories (socio-economic, professional, gender, etc.) of readers, and how did these constituencies develop and change over time?

• What methodologies can historians use to infer readers’ reactions to books from ownership and borrowing patterns, and how can these methodologies complement and enrich other sources, such as self-reported reader reactions in ego-documents?

A glimpse of what the results of such an enquiry might reveal is provided by previous, suggestive case studies of individual readers, which, in the absence of quantitative studies made possible by modern digital instruments, is the only kind of analysis currently possible. Thus, historians studying the reading habits of brutal slaveowner Thomas Thistlewood have noted with surprise how he transcribed entire passages from the fourth chapter of Rousseau’s Social Contract, ‘On slavery’, but, seemingly compartmentalising his own different roles, subsequently neglected to apply Rousseau’s critique to his own actions (see Kates 2022). In this case, Thistlewood’s reading apparently produced no direct effect on his own actions. Might this have been different if instead of reading works of political theory, he had read novels seeking to elicit empathy for the plight of their enslaved characters? And what about slaveowners as a category? 18th-century debates on slavery and abolition provide an ideal case to use quantitative data such as will be aggregated in the Civic Fictions ecosystem to test the validity of theories of fiction-induced empathy (or in 18th-century terms, ‘sympathy’) as a force for societal change. While historians such as Hunt have suggested that narrative (fictional) texts that expressly sought to elicit sympathy for the enslaved underwrote modern human-rights thinking, contributing decisively to abolitionism, others have instead viewed fictional representations of enslaved suffering as a form of pornographic voyeurism, or have even argued that abolitionist fiction was used to promote imperialism (Absillis 2022; Wood 2002). Despite recent re-evaluations, the complex 18th-century reception of these texts remains poorly understood (Huisman 2015; Moore 2019). Was fiction (rather than non-fiction) most effective in moving readers, which texts had the greatest impact, who were their readers, and how were they read? Civic Fictions will attempt to leverage ‘big data’, or extremely large bibliographic corpora, to quantitatively study these kinds of questions.

Operationalising 18th-century reading

Civic Fictions builds, first, on the existing MEDIATE database, which used distant-reading techniques to reveal the (changing) values assigned to specific books during the long 18th century, by cross-referencing biographic data on readers with data on books circulating in the period 1760–1830. It will mobilise literary theory to develop new understandings of the emerging patterns, combining quantitative, big-data methods with qualitative, close-reading techniques to empirically test longstanding ideas about the civic effects of fiction. Thus, the project addresses two major challenges in book history, namely the lack of comprehensive reader-reception data like ego-documents and commonplace books on the one hand, and the difficulty of access at scale of phenomena like marginalia on the other, by developing methods to use large-scale data on the circulation of books among identifiable, individual readers as a viable reader-reception source. The project links book ownership (private-library catalogues, bookseller’s archives), borrowing practices (library lending-records) and other sources to individual actors and larger (socio-economic, professional, gender, political) groupings. Coupling data on ownership to longitudinal studies of book owners’ lives and societal interventions, including analysis of the social networks in which they participated, it seeks to reveal macro-patterns allowing historians to infer how works of fiction might have historically moved readers. In selecting a methodology to address this question, Civic Fictions starts out from four hypotheses.

First, the project posits that the concept of ‘reading’ is notoriously slippery, and requires critical unpacking. At what point can one be said to have ‘read’ a work: when one has read half? three-quarters? And if fiction shapes society, what about books rarely read cover to cover, that nonetheless massively impacted readers (the Bible – although arguably not a work a fiction – immediately comes to mind)? Books influence even those who never read a single sentence in them, as conversation and discussion acquaint ‘readers’ with content in ways sometimes as profound as direct reading. These considerations call for an expansive definition of reading that understands it as any form of (mediated) engagement with written text, including its physical manifestations (books as material objects). It follows, then, that book ownership and borrowing should not be considered a proxy for ‘real’ reading – as they are, for example, in the RED – but as one of many forms it can take.

Second, forms of access to books are multiple. 18th-century readers accessed books through reading aloud in the home, through informal lending networks, and through multiple varieties of sociability, in salons, coffeehouses and beyond, often perusing print material without buying it (Raven, Small and Tadmor 1996; Staves 2007; Towsey and Roberts 2018; Williams 2017). Thus, the Civic Fictions digital infrastructure will also need to effectively model the 18th-century ‘bookscape’, that is, the multiple points of entry through which individuals accessed books. Heeding calls for a richer understanding of ‘the landscape of book ownership’ in the 17th century, Civic Fictions will model what James Raven has described as ‘cultural topographies’, or the networks of people, places and practices constituting 18th-century book culture (Pearson 2012; Raven 2014; see also Knight, White and Sauer 2019 and for the 18th-century Dutch Republic, van Goinga 1999). Individuals could access books – for ‘reading’ or other purposes – in multiple ways, in settings running the gamut from the intensely private and personal to the highly public and institutionalised (Figure 1):

Figure 1

How did 18th- and early 19th-century individuals access books?

No single source, it is clear, could do justice to the complexity of any individual’s interactions and access to print. Civic Fictions posits, then, that to produce an integrated account of how books moved individuals in various ways, historians must gather evidence from multiple sources, including publisher’s archives, probate inventories, printed (private) library catalogues, library lending records, and ego-documents, none of which is sufficient on its own to provide an accurate picture of the complete communication circuit, or the full range of 18th-century individuals’ interactions with books.¹² But in order to make this possible, existing data first needs to be standardised, aggregated, enriched, and in some cases completed with additional, complementary data, creatively curating and extrapolating from the patchwork of evidence available.

Civic Fictions further posits that all reading and engagement with books has a public dimension. Reading is socially situated, and how individuals access books matters. Whether a book came from a family library or was merely heard about in conversation, and whether reading took place in private or in public, are key variables influencing how that book might affect its reader. Reading mediated by cultural and family gatekeepers was necessarily a collaborative, social experience. Such collective interactions with books complicate private/public dichotomies, compelling historians to redefine Habermas’s bourgeois public sphere in ways that attend to gendered or socially circumscribed forms of access to its institutions (Goodman 1992). Because all fiction reading has this public dimension, the Civic Fictions project will take the contexts of access (where, when, how, who else) as an essential variable in modelling how works of fiction might move their readers.

Finally, the effect of any book is not fixed, but changes over time, depending on the context of reading, and the historical accumulation of inherited readings. Interpretations exist in a dynamic ecosystem, as no book is read in isolation, but rather interacts with others, and is subject to historical layerings of meaning. Thus, the value and effect of specific books could change – sometimes radically – over time, depending on myriad contextual factors. (Jensen 2011; Towsey 2019). What did it mean, for example, that, as shown by the private-library auction catalogues gathered in the MEDIATE database, neither Rousseau’s Contrat social nor anti-monarchical political pamphlets were topping bestseller lists in the 1790s, but rather, it was Ovid’s Metamorphoses that was reported in the highest percentage of auction catalogues (Montoya 2021)? Did this mean that amid unprecedented political turmoil, individuals turned to this classic depiction of a mutable world to cope with events unfurling around them? In that case, what did individuals value in Ovid at this particular moment in time? Was it the physical book itself, possibly a remnant of a family heritage or shared culture slipping beyond reach, or was it the text that held most appeal? Were individuals in the 1790s ‘using’ Ovid in the same way as they did in the 1780s, or the 1690s, and how might these usages differ from, or build upon previous ones? Or alternatively, might the massive presence of Ovid in auction catalogues simply point to the fact that, as the Revolution radicalised and aristocrats prepared to flee, they flooded the market with books hastily gathered from family bookshelves? In other words, how might this finding be skewed by source bias, and how might books sold at auction differ from books reported in other sources? Did individuals access different books from a family library than from an institutional library, or from a coffeehouse, and how did all these books fit together into any individual’s reading culture? Only a richly contextualised understanding of the presence of this book in an individual’s library, that adequately models the full 18th-century bookscape, can help address these questions. In order to account for these multiple factors, and multiple temporalities in the reception of fiction, the Civic Fictions data ecosystem will therefore need to adopt a diachronic, multi-generational perspective, that will allow users to play with multiple variables in viewing data, including temporal ones.

New book history corpora to be digitised

Civic Fictions therefore seeks to create new digital infrastructure capable of providing a ‘thick’, multi-layered description of individuals’ interactions with books by aggregating data harvested from different types of source material, and developing sophisticated faceted search functionality to query it. Leveraging existing digital tools and making data from different sources and national contexts interoperable, it will integrate curated datasets from several major, existing 18th-century historical bibliometric databases, with long-term data sustainability ensured by collaboration with CERL. In doing so, it also aims to lay the groundwork for the creation of accessible, open-access data infrastructure for future projects in intellectual history – which may well seek to address different questions altogether. To this end, data will be prepared for future conversion into LOD by linking it to standardised authority files such as VIAF and CERL’s Thesaurus, and creating new authority files where necessary, with input from major data-linking projects such as Canada’s LINCS (Linked Infrastructure for Networked Cultural Scholarship) throughout the project duration.

The project primarily draws on four types of archival source material relating books to individuals: private-library auction catalogues, publisher-bookseller archives, government records on library confiscations, and library lending records. Three of these sources have already been (partially) sourced, transcribed and converted into enriched and curated bibliographic datasets, representing decades of accumulated research time investment. These sources have been selected to encompass different types of material, different actors, geographic contexts and time periods.

The MEDIATE database

The backbone of the project is provided by the existing, open-access MEDIATE database, which focuses on private-library sales and auction catalogues. During the course of the 18th century, an estimated 10 000+ sale – most often, auction – catalogues were produced of privately-owned libraries, primarily in the Netherlands, British Isles, Denmark and France. Today, some ten to twenty per cent of these are still extant, scattered in libraries around the world (Blom, Jagersma and Reboul 2020; for an overview of this source material, see also Jagersma et al. 2023). These library catalogues contain a wealth of bibliographic information on books owned by named individuals, and provide prime source material to link specific books to identifiable readers and their wider social and family networks. During the MEDIATE project, a database was created that brings together both full-text, searchable transcriptions of 600 auction catalogues of private libraries sold in the Dutch Republic, France, the British Isles and Italy between 1665 and 1830, and data extracted, standardised and further enriched from these catalogues. The MEDIATE database currently contains over half a million individual, subject-categorised book records, and enriched data on owners, dynamically linked to standardised authority files for VIAF Works, Persons (authors, translators, editors, annotators, etc.) and CERL Thesaurus Publishers and Places. Data-cleaning and enrichment is ongoing, and will continue throughout the Civic Fictions project, including the addition of data from several dozen new catalogues to the existing corpus. Most importantly, the MEDIATE database provides a tried and tested data model and workflows that can serve as a basis for further modelling work in the larger Civic Fictions data ecosystem.

Publisher-bookseller archives

Supplementing the private-library auction catalogues already digitised in the MEDIATE database, the second major source that Civic Fictions will harvest are publisher-bookseller archives. Thousands of publisher-booksellers were active in 18th-century Europe, yet no more than a dozen of their business archives have been preserved. The two most complete extant publishers’ archives are those of the modest Swiss publisher-bookseller Société Typographique de Neuchâtel, active during the period 1769–1794, and the Leiden-based Luchtmans firm, active from 1697 to 1848. Twenty times as large as the STN, and with an archive running to eleven metres in length, Luchtmans was one of the ten biggest publisher-booksellers in 18th-century Europe. In total, the Luchtmans archive holds an estimated 1.23 million transactions, not including stock-takes found in booksellers’ catalogues.

Of these two archives, the smallest, that of the Société Typographique de Neuchâtel, has been entirely digitised and rendered into searchable book and client records in what is one of the most comprehensive 18th-century historical bibliometric databases to date, the FBTEE database at Western Sydney University. This database houses 70 000 records on individual books sold between 1769 and 1794, and has piloted a unique, granular subject classification system that is ready for import into other bibliometric databases, giving researchers a head start in the complex subject categorisation work ahead (Burrows 2018).¹³ As a follow-up to the FBTEE database project, in 2017 its creator Simon Burrows, together with colleagues at the University of Amsterdam, spearheaded the production of digital scans of the entire Luchtmans archive, which was subsequently made digitally available in open access through Amsterdam University Library. This means that the Luchtmans archive, too, is now ripe for conversion into structured, searchable data in a modern bibliometric database. Because of the massive scale of data in this archive, however, historians have hitherto barely touched it, as a comprehensive approach is not feasible even within the scope of a multi-year project.¹⁴ Civic Fictions will take a first bite out of the massive Luchtmans archive, by selectively harvesting data on the private (international) client transactions recorded in it, using focused research questions to keep the task manageable.

Focusing specifically on the period 1760–1830, a postdoctoral subproject will explore community formation among the private clients of the Luchtmans firm, in a diachronic, comparative perspective. The Luchtmans private client accounts list 285 500 transactions (sales of a title involving Luchtmans and another identifiable private client on a given date) in the Dutch Republic, across an estimated 5000 to 6000 private accounts, and several thousand transactions in the separate ‘Foreign booksellers accounts’ from 1741 to 1788. This subproject will ask how individual book buyers, as evidenced by their buying patterns, negotiated the interplay between individual and collective identities, and how buyers’ networks shaped the emerging bourgeois public sphere and new kinds of community-building. It will do so by gathering all data on Luchtmans private clients in a database, collating further data on the (other) clients they represent, and places they resided, and identifying and enriching client details (names, professions, places of residence, family members, direct professional colleagues, life events) with data from other sources. Preliminary samplings suggest that an estimated eighty-five per cent of Dutch clients will be relatively easy to identify, drawing on the mass of person data available in resources like ECARTICO,¹⁵ so-called Naamwysers or 18th-century Who’s Who books for professional groups, the KNAW–Huygens Institute’s Biografisch Woordenboek van Nederland¹⁶ and Biografisch Portaal van Nederland.¹⁷ Identifying clustering that might point to community formation and broader societal engagements, digital analytic tools such as network analysis will be used to answer these prosopographic research questions. Initial groundwork has shown that behind every individual client, there emerge (family) networks of readers, interpretive communities and patterns of book acquisition. A key early finding is the importance of gatekeepers in mediating access to books, particularly for less entitled groups such as female buyers (Rozendaal forthcoming). Tracing changes in book acquisition patterns over a long period, and viewing these within a multi-generational framework, it should thus become possible to offset different, overlapping networks – large households, extended kinship networks, professional and religious communities – each of which developed distinctive sets of interactions with works bought at Luchtmans.

French Revolutionary library confiscation records

The second new book history source that Civic Fictions will harvest are the French government records recording Revolutionary confiscations and restitutions of Parisian aristocratic libraries between 1792 and 1803. Not only does this material provide a unique archival dataset to study relations between books and readers, it also provides historians with a textbook example of a population that suffered unprecedented traumatic upheaval, hence potentially allowing literary theorists to test hypotheses about fiction’s role in working through trauma. The French Revolution witnessed successive waves of property seizures, which also included the libraries of aristocrats and clerics who had left France, as émigrés or as the result of forcible deportations (1792), and libraries of political suspects and condemned individuals (1793). Restitutions followed as early as 1794, as under the Convention thermidorienne émigrés, transported criminals and especially their widows and heirs could claim restitution of seized books. Inventories drawn up by claimants offer precious insights into the emotional, sometimes specifically memorial attachments that books held for their owners. Widows regularly invoked deceased husbands and books supposedly used by their children, strongly gendering the genre of the demande de restitution.

But despite the extraordinary richness of the data they contain, the French Revolutionary library confiscation archives – primarily the T and F17 series in the National Archives, and the Restitutions des livres series in the Bibliothèque de l’Arsenal, in Paris – have been almost entirely neglected by historians. There has been no comprehensive study of this material, and only a few studies focusing on specific provinces, confiscations of ecclesiastic libraries, and the role of confiscations in creating national libraries and a national bibliography (Mellot 2019; Varry 2005). The Parisian records, surprisingly, have never been the object of a systematic inventory. While precise totals are lacking, an indication of the extent of the material is the fact that by 1797, Paris depots held some 450 000 volumes, many of them confiscated from private individuals. The reading culture of émigrés, similarly, has only incidentally been addressed, despite the insights it furnishes on how this particularly well-documented population of individuals worked through collective trauma (Desplats 1987; Fossier 1989; Reboul 2019).

However, recent digitisation efforts mean that much of the Arsenal material is now available as open-access scans – again, ripe for conversion into structured, searchable data in a modern bibliometric database. Significantly for the Civic Fictions data ecosystem, some of this material pertains to families and individuals also present in the MEDIATE dataset, further underlining the desirability of comprehensive record linkage. Similarly, almost all of the titles (VIAF Works) in the confiscation archives are already recorded in the MEDIATE database, hence facilitating digital record linkage. Harvesting this corpus, the Civic Fictions project will create a database of the material, standardising Person and Work records, and linking data from other sources – auction catalogues, archival material elucidating individuals’ family networks – in order to identify patterns in confiscations and restitution claims. Building on the standardisation and identification work, a number of significant case studies will be selected that allow for a longitudinal, multi-generational approach, so as to provide the groundwork for an analysis of how books generally, and works of fiction specifically, might have functioned as a nexus of (competing) values for specific owners or their heirs: as objects of traumatic displacement, as prosthetic memories of a vanished world, or even as building-blocks for community formation and new civic identities.¹⁸

Public and commercial library lending records

Finally, an extraordinarily rich source for book history is the extant lending registers of various 18th-century commercial and public libraries. Library borrowing records have been most fully preserved and inventoried for the English-speaking world, and are currently being harvested in the Reading Communities database. But lending registers are also extant for other parts of the world. In the Netherlands, the most comprehensive source is the KB–National Library lending records in The Hague, that cover the period from 1799, when Stadholder Willem V’s abandoned library was nationalised, all the way to 1903 (van Delft 2000; van Delft 2017; for a more recent analysis see also Huijsmans 2022). The period until 1830, the main focus of the Civic Fictions project, totals some 25 000 borrowing records (date, description of book, surname of lender, expected and real date of return). Since the library was open only to members of the political elite, in active government service, the registers provide ideal material to study the relation between books, citizenship and the ideological positionings of these especially well-documented individuals (Elias and Schölvinck 1991; Oddens 2012). Most strikingly, given the government functions of the borrowers, over a third of the borrowings fell into the category of belles-lettres: principally novels, but also classical authors and language manuals (Huijsmans 2022). This raises questions about relations between fiction-reading and citizenship. Was this library used primarily for recreational ends, complementing other forms of reading and books accessed elsewhere? Were books used exclusively by registered borrowers, or can it be inferred that behind each named borrower, there were other (perhaps female) household members and extended family networks also accessing them? How did borrowers negotiate private and public identities in their choices of fiction, and how might books borrowed contribute to expanding, or alternatively reinforcing borrowers’ political outlooks, in an 18th-century variant of modern-day ‘bubble formation’?

To answer these questions, Civic Fictions will create a database of the material, again standardising Person and Work records, and identifying works of fiction that appear most frequently in the registers, patterns linking specific individuals to specific titles, subjects, and possibly even specific policies those individual may have helped shape. This process will be facilitated by the existence of a first, pilot database developed under the supervision of Paul Hoftijzer in 2002, and recording the lending transactions in the KB–National Library registers from 17 September 1799 to 16 September 1801. Digital scans have now been produced of the full registers, which are highly structured in their layout, potentially allowing for the use of handwriting transcription software such as Transkribus to speed up the transcription process. Since uniquely, almost all books described in the registers are still in the KB–National Library collection, we will also be able to inspect the physical copies of these books for the presence of reader annotations or other reader-response evidence. In short, by adopting a prosopographic, comparative approach, and cross-referencing data with additional data from other sources on The Hague’s bookscape during this period (e.g. de Kruif 1999; Lubbers 2014), including the Luchtmans archives in the neighbouring city of Leiden, a rich analysis will become possible both of long-running patterns and salient discontinuities in this particular group of 18th-century readers’ historical interactions with books, and especially works of fiction.

Database and interface tool development

By building public-facing infrastructure to study 18th-century source material, bringing together a distributed community of research projects and institutions, Civic Fictions aims to cement existing collaborations between individual research projects, between universities and the GLAM (galleries, libraries, archives and museums) sector in creating common open-access scholarly infrastructure. Together, the databases that will be aggregated in this historical bibliometric data ecosystem already contain data on over one million individual books circulating in the period 1760–1830. This will allow historians to move between macro and micro levels, distant-reading approaches, and individual case-studies, mobilising big-data analytic tools to aggregate and link different types of source material.

The data model underlying the Civic Fictions data ecosystem will be designed, first, to be compatible with the existing MEDIATE, FBTEE and Reading Communities databases, and should be able to accommodate granular, longitudinal Person data, and capture bibliographic data in a manner that can describe incomplete, uncertain and approximate data. Reflecting the structure of the existing MEDIATE and FBTEE databases, it will draw on the FRBR (Functional Requirements for Bibliographic Records) hierarchy to describe books at four distinct levels: Work, Expression, Manifestation, Item (Tillett 2004). Additionally, data will be aligned with standardised authority files (VIAF, CERL) in preparation for comprehensive LOD linking further down the road. Where functional, individual databases may be created by the postdoctoral researchers and PhDs themselves in the Heurist Open Source online database builder and CMS publisher tool, which enables non-technically trained scholars to directly create their own individual, tailored databases. Significantly, Heurist is already being used by the FBTEE and Reading Communities databases, as well as by larger institutional data creator-aggregators such as the Bibliothèque nationale de France, allowing individual researchers to work collaboratively with colleagues studying similar datasets. Further enhancing collaborative workflows, Civic Fictions project members will be able to work directly in the CERL Thesaurus to create new Person authority files, dynamically linked to the relevant records in the other databases in the ecosystem, for all book owners documented in the project.

Second, complementing this data ecosystem, the project will create a common interface so that users can simultaneously query all these datasets. By allowing users to query multiple kinds of source material at once, new links and unforeseen patterns will become visible, mobilising the accumulated expertise of generations of historians, and rendering data from different (national) sources fit for large-scale comparison. Allowing researchers eventually to use Semantic Web technologies to interlink research data across datasets, this will represent a major step forward in overcoming the current silo effect of databases whose disparate formats, organisation and interfaces make it difficult for historians to effectively connect and understand the data and underlying historical processes they document. The common interface, in short, should ideally enable historians, at the click of a mouse, to access literally millions of pieces of bibliographic data on the circulation and reception of books in Europe during the 18th century.

Within this common interface, finally, a suite of analytic tools will allow users to conduct time-series analyses, carry out complex (statistical) queries, as well as produce dynamic visualisations (for example, a time-lapse map of the location of all individuals who owned or borrowed a specific title like Jean-Jacques Rousseau’s Nouvelle Héloïse between 1761 and 1830). A particular desideratum will be functionality similar to Amazon’s sixteen different recommender systems, using machine learning to match books and readers according to different kinds of similarity measures. The interface and suite of analytic tools should therefore represent a real improvement for historical scholarship on any of several already existing, individual databases.

Conclusion

The Civic Fictions data ecosystem seeks to create new bibliographic tools for the international community of scholars working on the 18th century. Moving well beyond 18th-century studies alone, the project aims to aid and deepen historians’ understanding of the process and impact of reading, both in the historic past and the present, linking it not only to (political) community formation, but also to origins and processes of societal change. In the long term, the findings may potentially also be of interest to literary theorists and public advocates of the arts and humanities, as they will provide empirical evidence to validate or disprove some of their most cherished hypotheses. Finally, the project will be significant in exploiting, bringing together and further analysing data from multiple existing digital book-history resources, and systematically preparing it for conversion into LOD formats. Impact will be achieved through a public-facing data interface and collaboration with organisations like CERL to disseminate findings to a wider audience, promoting use of the interface tool among audiences interested in fiction and its effects. In short, Civic Fictions would ultimately like to provide a blueprint, model and virtual laboratory demonstrating the intellectual benefits of linking bibliographic data, well beyond the field of 18th-century studies and book history, to address questions that have long been fundamental to the humanities.