Mapping The Republic of Letters
Social scientists started using quantitative methods in the 1960s. These efforts contributed to the field of cliometrics, or the “new economic history”, which utilizes computational methods to interpret a large amount of quantitative data to interpret and gain new insights into economic history. Digital history represents a new trend in historical studies to analyze a vast amount of digitized and born digital material, which was formerly only available at the location of physical archives. Historical scholars often needed to travel to these archives, and painstakingly extract information from the immense quantity of obscure sources that drive historical scholarship beyond the more common, widely available literature. Digitization, online archives, and particularly searching and sorting capabilities, have facilitated this scholarship to a large degree. However, digital techniques, which have provided new methodological and research opportunities, are complementary to traditional close readings of the sources (Edelstein et al., 2017).
In a large, multidisciplinary, multisite application to historical research, Big Data methods were used to create a research system to study the early modern Republic of Letters. The Republic of Letters, or Respublica literaria, was an intellectual community and correspondence network of writers, scholars, artists, politicians, philosophers, and other members of the intelligentsia during the 1600s and 1700s, coinciding with the “Age of Enlightenment”. The network was primarily based in Europe and the Americas. The Republic of Letters was so named (by the participants themselves, not by historians) because the main mode of correspondence was the exchange of letters, which allowed the communication network to cover great geographical distances. It was therefore a large “traffic analysis” project to explore geospatial and temporal relationships (Edelstein et al., 2017).
The Republic of letters has been widely studied by generations of scholars, mostly by close reading and analysis of the content of the individual letters. However, with the availability and accessibility of digitized sources, online archives, search and sort mechanisms, database technology, computational and visualization tools, and particularly metadata, new questions can be posed and answered. The project Mapping the Republic of Letters (henceforth called Mapping or the Mapping project), an early large scale digital history project, began in 2008, and involved Stanford University scholar Dan Edelstein, Professor of French and History; Paula Findlen, Professor of Italian History; Giovanna Ceserani (Classics and History); Caroline Winterer, Professor and Director of the Stanford Humanities Center; and digital research architect Nicole Coleman, among many others. Although the project was primarily based at Stanford University, it also included researchers, software developers, and students from the University of Oklahoma and Oxford University, as well as members of the Politecnico di Milano in Italy for developing the visualization system. The Mapping project is therefore an example of an international, multisite, and multidisciplinary large-scale project in the digital humanities. Over 55,000 letters and documents, sent and received by approximately 6,400 different individuals, comprise the Big Data that the Mapping project was conceived to analyze. A major challenge was discovering patterns in large sets of documents that were traditionally analyzed through close reading (Chang et al., 2009).
The objective of the project was to visualize information in databases and archives related to the Republic of Letters, and, specifically, to generate maps and networks that allow scholars to study the geographic range, temporal characteristics, and quantity of correspondences of various participants. Questions such as changes in recipients and their geographic locations of letters for a specific individual over time, or how widespread an individual’s letters were both temporally and spatially, can be investigated with intuitive visualizations. Consequently, metadata from the digitized sources and generated from previous digital history projects are the main resources for the project because the metadata contains information about the sender, recipient, dates, and addresses. It was on the basis of the metadata that the mappings were performed. The actual content of the letters is not part of the metadata and was not the main focus of the work. However, a very important point is that the visualizations based on the metadata provided new, interesting, and unexpected findings that prompted new lines of inquiry and research questions, which are subsequently studied by going back to the content of the letters. In this way, the Mapping project is analogous to a “distant reading” to discover new and unexpected patterns, trends, and anomalies, followed by a “close reading” of the letters to explain what was found in the visualizations (Edelstein et al., 2017).
From a computational standpoint, in the early stages of the project, around 2009, several innovative computational tools were developed. The main visualization tool was “RPLViz”, running in Adobe Flash player. Because of the complexity and richness of the information being analyzed, multiple views and animations were provided to users. The “connections view” provided the interactive network visualization with nodes representing cities and other geographic locations and edges representing connections between correspondents. Colour-coded information represented by hue and opacity (translucency), as well as the line thickness of the edges, was provided for interpretability. The “volume view” employed circles of different diameters to represent the volume of incoming and outgoing correspondence for each location. The “flow view” complemented the connections and volume views by indicating the directions of the correspondence. This view was animated to reduce clutter and to more accurately represent flow. Opacity was employed as a visual cue to indicate the volume of a particular flow. The comparison view presents correspondences between two individuals selected by the user. These views, taken together, provided new exploration capabilities that were found to be very useful for historians. This component of the project also emphasized the need for a close collaboration between programmers, computational specialists, and domain experts – in this case, historians. Humanities scholars were an integral part of the development team, assisting in the design of the system, and also providing computational specialists with insights into what historians’ needs are, and what tools and techniques would be useful. The collaboration also elucidated the challenges involved in incomplete data and missing attributes, and how this missing information can be specified and integrated into visualizations. Another challenge that was highlighted (especially to the computer scientists) is that in digital humanities projects (and Big Data in general), data are not only incomplete or missing, but they are highly heterogenous and non-numerical. However, meeting these challenges led to the development of new computational paradigms, especially in visualization, and spurred the creation of innovative analysis tools that specifically benefit historical scholarship. The visualizations enabled by the Mapping project also led historians back to the sources in the archives to explain anomalies in the visualizations, such as the disappearance of nodes over time (Chang et al., 2009).
More recent work on the Mapping project includes development of the open access “Palladio” software suite, specifically designed at Stanford University for visualizing complex historical data. Palladio is a web-based visualization system that directly grew from the Mapping project, but is used in other projects as well. It allows users to upload, analyze, and visualize their data in a browser, although the Palladio system does not store any data uploaded by users. An example of an interactive network visualization created with Palladio, with associated charts, and tables, illustrates Voltaire’s correspondence network.
Other innovative tools were created during the early stages of the development of the Mapping project. The “Corrispondenza” tool performed the mapping functions, and charts and statistics on the number of letters plotted and those letters with insufficient information that could not be plotted. The “Inquiry” tool enabled users to perform queries directly from a map or timeline, enhancing the standard searching procedure of entering search terms into a text box followed by viewing the results returned by the database. The “Ink” tool allows users generate statistics on the quantities of letters through an interactive map. Bibliographical information on the senders and recipients of letters is sorted with the “Fineo” tool.
New tools have been developed and new technologies have been incorporated as the project evolved over time. The open Gephi graph visualization platform was used for a project on John Locke. The Mapping project is a member of the Gephi consortium that developed this tool.
Several case studies emerged from this large-scale project, including “British Architects on the Grand Tour in Eighteenth-Century Italy”, “The Correspondence Network of Benjamin Franklin”, and “ ‘John Locke likes this’: An Ego-Network Analysis of John Locke’s Letters”. A list of the projects and publications can be found on the website (Edelstein et al., 2017). Experiments on metadata in the letters of English philosopher John Locke (1632 – 1704), German Jesuit and polymath Athanasius Kircher (1602 – 1682), American writer, scientist, inventor, and diplomat Benjamin Franklin (1706 – 1790), and French writer and philosopher Voltaire (François-Marie Arouet) (1694 – 1778) were performed.
The project led to some interesting and valuable historical insights. For example, one of the most surprising discoveries was that Voltaire, one of the most well-known individuals in the Republic of Letters, and usually considered to be cosmopolitan, was more focused on France than originally suspected, at least in terms of his letter correspondences. Although there are “high-profile” cases of international correspondence, such as with Catherine the Great in St. Petersburg, and despite his interest in Isaac Newton’s theories, analysis of the visualizations demonstrate that approximately 70% of Voltaire’s correspondents were French, although there were “tails” (less dense groupings of connections) to England, Russia, and Switzerland. Another interesting research result from the Mapping project concerns Benjamin Franklin. As Franklin was based in Philadelphia, which at the time was small compared to European intellectual centers, he did not have the same opportunities for international (but still continental) connections as did his European Republic of Letters colleagues. Before his journey to London, all of Franklin’s correspondence was American. During his stay abroad, all his correspondence was from England. European correspondence continued at a much reduced level (25% of his letters originated overseas) when Franklin returned to America, thereby demonstrating a return to mostly local correspondence. The visualizations are also useful for studying “degrees of separation” among members of the Republic of Letters. For instance, although Voltaire and Franklin never corresponded directly with each other, the individuals with whom both did correspond could be determined (Edelstein et al., 2017).
Although these projects have a substantial geospatial component, exact geographical locations are not of prime importance, and therefore geographic information systems (GIS) were not incorporated into the visualizations. Additionally, abstract maps were preferred over historical maps. Instead, locations were presented as “cultural zones”, defined in relationship to each other, and which do not have strict geographical boundaries. For instance, the French and English cultural zones are centered around Paris and London, respectively. In general, cultural zones were located at city locations or at clusters of cities (Edelstein et al., 2017).
Finally, a major endeavor of the Mapping project on the Grand Tour of Italy is described. The Grand Tour was a custom wherein young male members of the European upper classes would undertake a journey throughout Europe. It was considered to be part of an upper-class education. The Grand Tour was predominant in the period from the 17th century to the mid-19th century. In the context of the Republic of Letters, the Grand Tour of Italy was undertaken by thousands of Europeans during the 18th century, and was a major factor in the emergence of modernity, spurring new concepts of politics, the arts, leisure, and culture (see Grand Tour for more information). One research project created a searchable database and visualization system to document the lives and adventures of travelers on the Grand Tour. Over five thousand entries in A Dictionary of British and Irish Travellers in Italy, 1701-1800, (edited by John Ingamells, Brinsley Ford Archive, New Haven: Yale University Press, 1997) were processed for this work. More information is found on the Grand Tour website of the Mapping project.
The Grand Tour Project yielded insights into the emergence of new connections as people met in Italian cities, including Rome, Naples, Florence, and Venice. Some of these connections were temporary consequences of traveling, while others were longer-term. The richness and heterogeneity of the metadata provides opportunities to pose and answer new research questions. For instance, it facilitated researching possible connections by tracing the edges of the networks (Edelstein et al., 2017).
The underlying philosophy of the Mapping project was essentially humanistic, complementing and enriching traditional historical scholarship. Many possibilities for this scholarship are opened by the vast scale of Big Data, including rich interactive visualizations and heuristics. As is the case with a lot of humanities data, the methods employed assume some level of incompleteness, uncertainty, and ambiguity in the data. However, computational techniques help scholars discover new patterns and explore new lines of inquiry. However, it is emphasized that the interesting and innovative results emerge from interpretive, non-computational investigations. The goal is to gain insights into and to make sense of visualizations, such as maps and graphs, that necessarily represent ambiguous and imprecise data. In summary, close reading is still mandatory for interpreting the results of computational analysis and visualizations (Edelstein et al., 2017).
In the Mapping project, different sources of metadata were used and aggregated, allowing the historical networks to be studied from multiple viewpoints, and enable studying degrees of separation. The project also suggests many opportunities for future research and development. One is an encyclopedia of metadata. A rich archive can potentially be built that contains and integrates multiple databases, along with associated metadata. In this way, all metadata on scholarly communication can be aggregated. The interactions and connections of each individual could be integrated into a collection of data spanning hundreds of years. Bibliographic information can be added, along with other metadata, such as professional or scholarly society memberships, education, social activities, among others, resulting in a complex, multidimensional archive from which rich, innovative scholarly investigations can emerge. However, further enrichment of tools and techniques, along with the development of new ones, are necessary for these goals to be even partially realized. It is the humanists themselves who will develop future tools to analyze rich metadata and to draw meaning from them (Edelstein et al., 2017).