Builders of digital reading environments must constantly take stock of changes in the information landscape. The new developments in cloud computing and social media over the last decade, to take but two of the most significant changes, have fundamentally altered how we interact with and share data, while the constant surge of new hardware continues to impact where, when, and how we work. But all this change can blind us to what remains the same. As a textual culture we continue to gather, digest, classify, and divide our ideas when we speak, write, and type, a point also explored by Adrian van der Weel earlier in this volume. These kinds of actions, what John Unsworth has famously termed scholarly primitives, are so internalized that they remain largely invisible to us (Unsworth 2000). The designing of digital reading environments, therefore, requires an investment that anticipates future readerly needs through careful study of the long continuum of textual communication. Put another way, looking back helps us to gauge what waits in the distance.

Implementing New Knowledge Environments (INKE) is a major, seven-year, federally funded collaborative research project committed to the design of new digital reading environments. That the designing and prototyping of new digital tools might derive through study of the history of reading, particularly through a close analysis of the intersection of people, technology, and culture, remains central to the project. Since many INKE designs have begun with a survey of related codex exemplars before considering the new affordances available on screen, INKE prototypes attempt to strike a balance between past and future thinking. In the following pages, I will survey a selection of INKE tools, and in doing so attempt to show how INKE-based textual theory unfolds. While I will use different scenarios to illustrate the potential uses of each tool, I will also situate each tool within its own particular textual history. As such, this essay attempts to engage not only with the function of such tools, but the logic they imbue.

Testing affordances

According to the OED, an affordance is "2. Psychol. A property of an object or an aspect of the environment, esp. relating to its potential utility, which can be inferred from visual or other perceptual signals" (OED 2016, emphasis in original). For textual scholars, the word affordance denotes the use-potential of a given medium or substrate. In the long history of the book, one might consider, for example, the properties of stone tablet, papyrus, vellum, and paper. A more granular approach might then look to various designs such as scrolls and codices to consider the actions that these different substrates inspire. Measuring the affordances of digital media is similar. We might begin by considering the substrate we work with or through, for example, the silicon or other material used for computer, tablet, and phone screens; but we must also examine the interfaces we work in and the various tools we rely on to navigate within these environments. To speak of digital affordances, then, is to consider the full potential of the medium, not only what has been achieved, but also what might be realized.

The history of computer-assisted reading tools can be seen as part of a long series of experiments aimed at maximizing the affordances of electronic and digital media. If we do not immediately think to classify such tools within the domain of textual studies, it is because we ignore the acts that such tools allow us to perform, be it searching, indexing, annotating, or collating. The starting point for any discussion of computer-assisted reading is Roberto Busa, the Italian-born Jesuit. In 1949, Busa teamed up with Thomas J. Watson, the founder of International Business Machines Corporation (IBM), to create a tool for searching the works of Aquinas. Thirty years later, the fifty-six printed volumes of the Index Thomisticus were complete. What makes Busa so important in the context of this chapter is that he saw the potential of computer assistance for tracking patterns across a massive corpus and subsequently embraced the affordances of the new media for textual study. Busa's computer-assisted experiments in corpus linguistics should thus be regarded as a turning point in the use of computers for the study and production of texts, one that would continue with the introduction of hypertext in the 1980s, the widespread take-up of the Internet in the 1990s, and the current experiments in social and cloud computing today (Hockey 2004). The following discussion of various INKE tools is thus part of the history just briefly discussed, since it considers the affordances of various in-development tools for different textual acts and the textual traditions that inspire such designs.

One way of assessing the affordances of a new digital tool, particularly one built for textual study, is by considering how it might be used for an established author's works, especially a writer whose corpus has been expertly edited in print. Such an approach is valuable for not only exposing the strengths and weaknesses of the particular tool under review, but also for shedding light on the similarities and differences between print and digital environments, particularly in matters of design and navigation, a point also taken up by Sondheim et al in the next chapter. The Adages produced by the famous humanist scholar Desiderius Erasmus offers a particularly good test case for measuring the affordances of different INKE tools because the textual history of the Adages during Erasmus' lifetime reveals a work of staggering bibliographic complexity: multiple editions, innumerable variants, and a host of other variables all contribute to what is often characterized as a single work. Capturing the flux of the Adages is extremely difficult, and yet ignoring the work's highly iterative nature is nothing short of misrepresentation. How might different digital tools be used to confront this challenge? How might a digital tool be employed to curate variance and versioning in the early editions, and how might the tool be used to alter our reading of modern criticism of those editions? How might searching and sorting through a modern edition of the Adages differ on screen? How might a graphical visualization assist the way we see the iterative nature of humanist editorial work, or the co-occurrence of complex relations in a modern translated edition? Before attempting to answer these questions we must address the work in question, in order to see the challenge before us.

First printed in 1500, Erasmus' original selection of Latin and Greek proverbs, known as the Collectanea, offered readers 818 unnumbered adages with brief notes. Eight years later, Erasmus would work in-house with Venice's most famous printer, Aldus Manutius, to produce a much-expanded folio edition. The Adagiorum Chiliades, or Thousands of Adages, contained 3260 proverbs, this time with detailed philological and historical notes. Subsequent folio editions, especially those printed after 1513 by the famed Basel publisher-printer, Johann Froben, would see the introduction of new proverbs and expanded notes for existing ones. The final edition produced before Erasmus' death in 1536 contained no fewer than 4,151 proverbs (Barker 2001). Any serious textual scholar of Erasmus' Adages faces the daunting task of studying a work under constant revision, one that not only expanded over its thirty-five year gestation, but one that was regularly rearranged to conform to its author's latest vision. Rather than enter this maze, scholars of the early printed editions of the Adages often settle on a single early imprint or turn to the latest modern edition: the desire for a single, authorial text leads to compromise.

NewRadial

But what if users could access different digitized editions of the Adages and all of their component parts? How then might they select, sort, and mix the primary evidence to fit their individual research needs? These kinds of questions inform the design of NewRadial, a tool developed by INKE member Jon Saklofske and assisted by Jake Bruce and Ian Brunton (programmers), and Stephen MacNeil and Amy Robinson (interface design and user testing). By arranging source material page by page in circular "nodes," as in the example of William Blake's Europe: A prophecy shown below, NewRadial's visualization scheme allows a user to see the book in pages (Figure 1). But this is just one of its affordances: users can then select different individual pages or nodes, mix the nodes to form radials, and then annotate the cluster. Users can also collect nodes from different database searches into personal collections that can be curated, annotated, then shared with other users or exported to other tools. The end result is a highly flexible environment tailored to fit different, individual research needs (Figure 2). Returning to Erasmus, how might this tool be used to study the Adages? In particular, how might NewRadial make it easier to enter the textual maze mentioned above.

Figure 1: Screenshot from INKE's NewRadial showing William Blake's Europe: A prophecy (London: 1794)

Figure 2:

One way to appreciate the complexity of the Adages is by following a single, edited proverb through its evolution in the early printed editions. The textual apparatuses in current print editions do this, in part, often citing the first edition in which the adage appeared, its position in that edition, and the changes it underwent in subsequent editions. In volume 31 of the Collected works of Erasmus, for example, the first of several volumes devoted to the Adages, one starts with Amicorium communia omnia (between friends, all things are common) (Erasmus 1982). This adage was numbered first in the 1536 edition, the last of the printed editions to be edited by Erasmus prior to his death. In the Collected works the 1536 edition serves as copy text. In the textual notes at the foot of the page the reader encounters the following statement: "In the Adagiorum collectanea of 1500 this was no 94 ... here it has been entirely rewritten" (Erasmus 1982, 29). This is a good starting point, but for one interested in the details of the rewriting—not only the changes in content, but also the graphics of the change, such as the employment of different typography, the addition and subtraction of notes or the inclusion of new paratexts—it falls short. Simply put, modern print editions of Erasmus' Adages do not allow us to see how the Adages was rewritten and reset over the author's lifetime. The proposed introduction to the Adages in the Collected works will cover the textual history of the work in more detail, but whether it will allow us to see those changes is as yet unknown.

The reasons for this are two-fold. First, the primary role of the Collected works is to provide an authoritative English translation, which it does. Visualizing all the textual variants is not only impractical; it is not the project's principal aim. But is this agenda, in part, a consequence of medium? For example, would the editors have made these same decisions if the primary outlet of the Collected works were digital? Might we be able to see the rewriting referred to differently if the textual notes to the Works were synced with a tool like NewRadial? Imagine, for example, that the three different versions of Erasmus' adage Amicorium communia omnia were available as nodes on the NewRadial canvas, namely pages from the 1510, 1515, 1536 editions. Readers could then compare key editorial moments from the larger textual history of the work or examine changes in mise-en-page over these three editions. They might also take note of (and share) such changes using NewRadial's annotative capabilities (via edge creation and commentary). Readers working closely with the Collected works, either in print or digital form, might also compare the original against the translation. In other words, a NewRadial rendering of Erasmus' Adages could offer a valuable complement to the Collected works by pairing images of primary materials with an acclaimed modern edition of the text.

One argument in favour of NewRadial is that the plasticity of its design mirrors modern critical practice; specifically, the penchant in the humanities for broad historicization and associative thinking. This is especially true in scholarship on Erasmus. While there are examples of highly specialized readings of the Adages, most scholars use the Adages to tackle larger contextual questions, be it humanist editorial practice, early modern political thought, or the history of common-placing (Phillips 1964). For example, Erasmus scholars regularly examine the 1515 edition of the Adages when discussing Erasmus' multiple scholarly endeavours following his arrival in Basel. Hilmer M. Pabel's recent monograph on Erasmus' nine-volume edition of the works of St. Jerome, printed in 1516, is typical in situating that work within a larger scholarly frame (Pabel 2008). Mark Vessey offers a similar approach in his reading of the 1515 edition of Erasmus' Lucubrationes (Vessey 2012).

In 1514, Erasmus teamed-up with the Basel printer Johann Froben, and over the next two years, the partnership would spur the production of a flurry of significant works, including Erasmus' edition of Plutarchan Opuscula (1514), his edition of Seneca's works (1515), the famous double-columned Greek and Latin translation of the New Testament, the Novum Instrumentum (1516), and a much-revised edition of the Adages (1515). Another context for Erasmus' edition of the New Testament was the Complutensian polyglot (1514-1517) which was in production in Spain at the same time. For Pabel, examining Erasmus' edition of Jerome requires study of Erasmus' other works published in ca. 1515. To ignore Erasmus' editions of Plutarch and Seneca, not to mention the Novum Instrumentum and the Adages, would be to ignore a larger intertextuality from which the Jerome edition was conceived.

Indeed, Erasmus himself saw the Adages in this larger editorial context. In countless letters written throughout 1514, Erasmus frames his latest work on the Adages as one among many scholarly endeavours. The weight of such editorial work is especially clear in a letter to Udalricus Zasius dated 14 September 1514:

My volume of Adagia is being so much enriched that it might be thought a new book. Jerome is being got ready for the press with my summaries and notes. The New Testament is being prepared, revised, and expounded in my notes. The Copia is being published with my own revisions, and a volume of Parallels will follow it. Those works of Plutarch which I had translated are now printed. Anneus Seneca is also in hand, corrected with immense labour by myself. Since any one of these works is capable of absorbing an entire man, and no Erasmus either but a man of adamant, you can easily guess how little spare time I have. (Erasmus 1974, 35-36)

Similar references to work-in-progress would continue in Erasmus' subsequent correspondence throughout the remainder of the year, and in each instance, the challenge of working on the Adages figures prominently. When modern critics read the Adages in a larger early modern context, they are essentially repeating a tendency initiated by its author, for Erasmus' letters marks the initial stage in their critical reception.

Jeanne Nuechterlein's Translating nature into Art: Holbein, The Reformation, and Renaissance rhetoric offers just such an example with its repeated references to Erasmus' Adages and the ca. 1515 context just discussed (Nuechterlein 2011). As an artist working in Basel, often in partnership with Johann Froben, Holbein found himself connected to a network that included Erasmus. A famous copy of the 1515 edition of Erasmus' Moriae Encomium (tr. In praise of folly) includes pen and ink drawings by Holbein. Holbein would also produce portraits of Erasmus on several occasions. Many of the ornamented borders and decorative initials used in Erasmus-Froben imprints were also the work of Holbein. In these three instances the Holbein-Erasmus connection is apparent, and, in at least two instances, references to the Adages are made explicit. In one of the pen and ink drawings, Erasmus is shown composing his Adages, and in Holbein's painting for William Warner, Erasmus sits with a closed book that contains an inscription from the Adages along the book's fore-edge. Any close reading of Holbein's art requires examination of these textual borrowings, and therefore the study of Holbein necessarily leads to Erasmus. Nuechterlein looks beyond these explicit connections, however, to suggest that Erasmus' approach to translation and paraphrase served more broadly as a source for Holbein's artistic method. By looking beyond Holbein's famous contemporary artists, such as Albrecht Dürer and Lucas Cranach, to a famous contemporary writer, Nuechterlein associates two domains usually kept separate. One might say that the strength of her argument depends, in part, on her examination of different media usually kept apart (Nuechterlein 2011).

As readers, we evaluate Nuechterlein's argument largely on the basis of her reading of the evidence. While some readers will be more familiar than others with the Erasmus material she cites, even specialists will rarely consult the sources she cites when making their assessments. Indeed, it would be cumbersome to do so for all of the documents referred to. However, if an article or chapter of a book were represented as a node in NewRadial, and linked to the primary and secondary sources it referenced through edge links or child nodes, both the argument and the documentary evidence on which that argument is based would be readily accessible to the reader. In the case just discussed, readers would then engage not only with the scholarly monograph, but with the evidence behind it. In doing so, they would evaluate her argument largely from their reading of the evidence. By using NewRadial to map the networks of relation between primary and secondary sources and to offer a portal to the full text of such sources is of benefit to both writers and readers. For writers, keying the primary documents serves to curate their evidence—both during the writing process and after its completion. For readers, the keying of primary documents provides access to the blueprints behind the author's scholarship and a more granular approach to the author's arguments and methodology. In the above scenarios, NewRadial affords a new approach to the scholarly essay, chapter, or monograph, by making process and product equally accessible. However, to maximize the functionality of this comparative approach for users, both the essay and the evidence must occur in proximate relation, that is, in a single interface that allows one to see the layers in the same space, or at least in conjoined panels. Such a design, as Brent Nelson argues in his chapter in this volume, takes account of the larger environment in which such actions are performed. Constant hyperlinking will only lead to the opening and closing of countless windows, increase the chance of distraction, and ultimately offer little advantage over print.

It is important to emphasize NewRadial's place within the longer history of textual assemblage. The assembling of texts, especially those related by topic, has been central to the history of book production for centuries. One of the most interesting examples of textual assembly, and one that informs NewRadial's logic, is the sammelbände. Sammelbändes, or books comprised of multiple texts, be it manuscripts, imprints, or a combination, involve the bundling of material, often related by subject, bound into one volume. A prime example of the later is the Schoner Sammelbände, a fascinating volume of early sixteenth-century cartographic and astronomical maps, charts, gores (the paper sections that are cut out and placed on a globe), and other fragments currently held at the Library of Congress (Hessler 2013). Seen from our vantage, the sammelbände is valuable primarily because it contains examples of rare printed material; the scarcity of its contents is what generates its modern meaning. From the early modern perspective, however, it was a compilation of up-to-date scientific materials conveniently located in one place. For early modern readers, mixing media was the norm. Given the affordances of digital media, is it not time for a large-scale take-up of remixing again?

While it is typical to piece out the past based on modern divisions of author, work, and date, the material evidence from earlier periods, as just seen, suggests otherwise. The title page shown here (Figure 2) is taken from one of Galileo's earliest imprints, here produced under the pseudonym De Cecco di Ronchitti. The image, however, is misleading for it fails to show the larger context in which the imprint appears. The Galileo imprint is number eight of fifteen in a fascinating sammelbände consisting of astronomical imprints ranging in date from 1520 to 1609. Previously owned by the French aristocrat, Léonor d'Estampes de Valençay (1589-1650), it reflects a near-contemporary assemblage of astronomical texts. Here, Galileo is positioned as one amongst many authors in a customized collection.

Figure 3: The titlepage to Galileo's Dialogo di Ronchitti. Image courtesy of the Thomas Fisher Rare Book Library.

Figure 4: The front board and armorial to the Estampes sammelbände in which the Galileo work appears. Image courtesy of the Thomas Fisher Rare Book Library.

As Jeffrey Todd Knight has recently noted, "in the early handpress era, the printed work was relatively malleable and experimental—a thing to actively shape, expand and resituate as one desired ... Every bound volume was a unique, customized assemblage" (Knight 2013, 4-5). Such assembling continues in today's digital environments. The ability to remix and repurpose in NewRadial adheres to the spirit of the sammelbände, as it provides a textual space to personalize one's reading and research.

Erasmus in the Dynamic Table of Contexts (DTOC)

If NewRadial finds its logic in the past practices of textual assemblage, the Dynamic Table of Contexts looks to the past to develop new strategies in the navigating, searching, and indexing of books for its inspiration. The Dynamic Table of Contexts offers readers new affordances by bridging two of the most familiar features of the codex: the table of contents and the index. The tool adopts the tabular structure of a traditional table as a core part of its interface, but it blends the sectional divisions used for structuring the contents (typically into chapters) with the more finite and granular search terms commonly alphabetized in an index. The end result is a more dynamic reading environment, one that not only encourages the combining of two core practices in textual navigation, but a space where the user customizes that combination (Nelson et al. 2012). The Dynamic Table of Contexts (see Figures 5 and 6) supports both continuous and discontinuous reading, especially of monographs and collections of essays. Users of the tool are able to read digital books from start to finish, search for patterns in the book using the index and or the XML-based tagging functions, or do a combination of both actions at once. One might, for example, begin reading a chapter, and then stop to search for the frequency of place names (for example, Montreal, Quebec, Canada) if one were interested in geography and demographics. A more fine-tuned search using the tag function might focus on the co-occurrence of specific individuals in each of these specified places, and perhaps by season or time of day. Once the information has been located, the reader might then continue reading the remainder of the chapter, leaving the examination of the search results to a later time.

Figure 5: Index and tag searches in the Dynamic Table of Contexts.

Figure 6:

Consider the opening adage from the Collected works of Erasmus discussed earlier, namely Amicorium communia omnia (between friends, all things are common). One can imagine, for example, that after an initial reading of the adage the reader might then choose to focus on references to classical gods, not only as they appear in the adage under review, but also in the ca. 3500 additional adages still unread. In this instance, the reader conducts an indexical search for patterns across a larger Erasmian corpus. Here, close reading gives way to machine reading, at least momentarily; but acts of continuous and discontinuous reading do not have to remain so neatly divided (Stallybrass 2002). The Dynamic Table of Contexts is designed for flexible reading, a space where linear reading and sporadic searching occur interchangeably and, as with much Web navigation, erratically (Hayles 2012).

As noted in an article on the tool, the Dynamic Table of Contexts allows users to add, subtract, and rearrange XML-encoded index items from the table of contents to meet individual criteria catered to a researcher's needs (Nelson et al. 2012). In other words, the reader is able to redesign the book. This process begins with a search, like the one for classical gods. Since the search terms are displayed in various colours, in both the text proper and in the graph used to capture frequency of occurrence, the customization process has already begun through the alteration of graphical codes. But the reader can take this a step further by adding and subtracting from the machine-generated results, or highlighting new items. Taken as a whole, then, the Dynamic Table of Contexts allows scholars a space to both read and research, a space to balance close and distant reading, and perhaps, most importantly, a space that the user helps to shape and design (Moretti 2005).

If the Dynamic Table of Contexts design offers a prediction of future reading habits, it also responds to a centuries-old desire to customize textual artefacts. While for centuries readers have relied heavily on the printed architectures of the book and page (for example, tables, indexes, headers, sidenotes, cross-references, images) to manage and navigate information, they have also used manuscript additions (in the form of insertions and supplements, commonplaces, interleaving etc.), in part, to compensate for the inadequacies of print architecture. As David McKitterick notes, "the composite text has enjoyed an existence that has been exploited—openly or clandestinely, but usually consciously—since the fifteenth century" (McKitterick 2003, 10). The history of manuscript in printed books is about customization. It is to that history I now turn.

The textual features of any printed book, especially those created for searching and sorting, mark but one incarnation in a longer history of experimentation. As Brent Nelson illustrates in his discussion of the page architectures for various Bibles in the previous chapter, most notably the Walton Polyglot (1655-57), many of the codex's most iconographic features, including the table of contents and index, have evolved through a continuous process whereby authors or editors model textual features in an attempt to gauge readerly needs. In their mappings of time, for example, fifteenth-century writers of printed chronicles struggled with how to maximize searching without fragmenting narrative. The use of indexes, both as separate paratexts and as part of the mise-en-page, offered two ways to reach this balance. In the fasciculus temporum, a universal chronicle from Creation to the present, circles of text help to separate "biblical, classical and modern rulers in the flow of historical time," while the work's woodcut illustrations, such as the woodcut of the burning of Troy, offer additional visual cues (see Figures 7 and 8 below) (Rosenberg and Grafton 2010, 31). While one can read the fasciculus sequentially, from start to finish, one can also conduct intermittent searches of the full text through its various indexical page architectures. In the fasciculus navigation is optimized through graphic experimentation.

Figure 7: Opening from Werner Rolevinck, fasciculus temporum (Strasbourg, 1490). Image Courtesy of the Thomas Fisher Rare Book Library.

Figure 8: Hand-coloured woodcut of the Burning of Troy with misprinted caption from Werner Rolevinck, fasciculus temporum (Strasbourg, 1490). Image Courtesy of the Thomas Fisher Rare Book Library.

Or at least this is true in theory. Since reading is largely idiosyncratic, and since the individual research needs of each reader will vary, no book design can anticipate the needs of all readers. In the case of printed texts, therefore, manuscript has often served as a form of supplement to aid the reading experience. The first example shown below (Figure 9) is a rather simple case in point. A reader of Seneca's tragedies has offered a fairly rudimentary table of contents with a list of plays and page references. While the manuscript table of contents provides a point of reference for the larger sectional divisions in the book, thus offering assistance for the future locating of particular plays, it also serves to show what a printed text, by 1634, was expected to include. The customization of this book is inspired as much by cultural norms as by individual expectations.

Figure 9: MS Index and titlepage to Seneca's Tragedies (London, 1634). STC 22220. Image Courtesy of the Thomas Fisher Rare Book Library.

In this second example (Figure 10) taken from the tabula of the fasciculus just discussed, a reader has used the blank margins to add manuscript page references by hand. The added medium serves to address what the printed index fails to provide.

Figure 10: Two pages from alphabetized printed index with manuscript additions from Werner Rolevinck, fasciculus temporum (Strasbourg, 1490). Image Courtesy of the Thomas Fisher Rare Book Library.

The third example (Figure 11), taken from an early copy of Erasmus' Adages, is similar to the manuscript table just shown, as it too illustrates an early modern reader customizing the printed text with manuscript. But here the reader uses different colours of ink to index different adages. One could argue that the three, coloured additions shown here are more than supplements since they offer a nuanced approach to indexing, a personalized system designed to address the inadequacies of the single-coloured printed index. Red, brown, and sepia serve to subdivide the alphabetical listing into categories. Evidence from the period would suggest as much as annotators often used customized icons and different coloured inks as part of their active reading. The thirteenth-century Bishop of Lincoln, Robert Grosseteste, used more than four hundred symbols for marking his text (Sherman 2008, 27, 29) while the famous sixteenth-century Hellenist, Guillaume Budé, used a combination of colour and symbols to mark his notebooks (Blair 2010, 73). In allowing digital readers to customize their indexes, the Dynamic Table of Contexts is similar to the example shown above. Manuscript and digital indexing may require different substrates, but they are part of the same textual tradition.

Figure 11: Index with ms. supplements from Erasmus, Adagiorum Chiliades (Basel, 1526). Image Courtesy of the Centre for Reformation and Renaissance Studies, University of Toronto.

Bubblelines

If NewRadial and the Dynamic Table of Contexts are best understood as tools born out of the history of textual assemblage and indexing respectively, Bubblelines is best seen through the history of illustration and design. As a visualization tool designed for textual study, Bubblelines exemplifies what Stan Ruecker has described as rich prospect browsing, a system that allows the user not only the ability to conduct robust searches, but also to modify the terms for searching, and even parts of the interface that support it (Ruecker 2011). In the spirit of Franco Moretti's understanding of distant reading—a kind of reading that combines with searching to interpret the patterns within a large corpus—Bubblelines searches for patterns across multiple texts (Moretti 2005). The bubbles in Bubblelines offer a graphic representation of the large amount of data recovered through searches made over one or many XML-encoded texts. As with other visualizations, this tool translates a series of complex, semantic relations into visual form. The results generated through machine reading afford a new kind of research, what Stephen Ramsay has described as algorithmic criticism (Ramsay 2011).

The example shown here (Figure 12) details a search across eight Brontë novels encoded in XML. The findings illustrate a search query comparing the co-occurrence of pre-chosen terms, in this case different kinds of characters, particularly those of masters (green), sisters (blue), and friends (pink). The differently-sized bubbles visualize when each of the characters is present in each of the novels, and the frequency of each. The centre of the first bubble marks the first mention and the centre of the last bubble marks the last mention. The diameter is based on mentions of the character name. In visualizing character relations in this way, one begins to identify larger patterns across a corpus of related texts; one also gets a snapshot of an important element of a particular novel's blueprint.

What we see in the previous example could easily be adopted for Erasmus' Adages. Taking our imagined search for classical gods in the Dynamic Table of Contexts as our starting point, Bubblelines could then supply a visualization of those search results. Here one tool would complement another to generate two ways of seeing the same results. But one could do even more. Taking the Brontë example as our template, let us imagine Erasmus' Adages as one text in a string of humanist publications XML-encoded and visualized in Bubblelines. What might a search in Thomas More's Utopia and Erasmus' Praise of folly, as well as works by Colet and Luther, tell us about the use of classical literature in this period? More significantly, how might the use of Bubblelines reorient future research in the field?

Figure 12: Screenshot from INKE's visualization tool Bubblelines.

Having just looked at Bubblelines' potential for the study of Erasmus' Adages, I want now to look at its place within a larger textual history. While visualizations are often discussed by digital humanists in relation to techniques fostered in the sciences and computing, they are rarely discussed in relation to the history of illustration. An important exception is Martyn Jessop's "Digital Visualization as a Scholarly Activity" (2008). Edward Tufte's Beautiful Evidence (2008) similarly approaches visualization with a wide temporal lens.

Bubblelines, like New Radial and the Dynamic Table of Contexts, is also shaped by centuries of experimentation in paper-based media. Understanding the logic of Bubblelines depends on looking beyond the immediate past to former experiments in visualizing textual phenomena. The Bible, and in particular the Book of Revelation, is a particularly good source for comparison.

No biblical text has proven more difficult to interpret than the Book of Revelation. In the sixteenth century, commentary after commentary attempted to unpack its hidden, allegorical meanings, and the result was a most amazing set of experiments (Sharpe 2003). John Napier's A plaine discovery of the whole Revelation of S. John (Figure 13) weaves chronology with narrative and annotation in an amazing display of mise-en-page. In addition to managing a three-column page, and flipping back and forth to the longer notes, readers must also rotate the text to see the current stage in the book, which is denoted by textual references to trumpets, vials, and other objects. The addition of Junius' annotations to the Geneva Bible after 1592 resulted in a similarly complex page, one complete with different icons to help navigate the crowded margins. Reading the Book of Revelation in the early modern period was more of an amplified than a linear experience since the architecture of the page encouraged engagement with layers of commentary before moving to the next book or section.

Figure 13: Typical opening from John Napier's A plaine discovery of the whole Revelation of S. John (London, 1611). Image courtesy of Thomas Fisher Rare Book Library.

Such complex textual experimentation was not limited to words, but can be found in illustrations and visualizations of the book's text. Hugh Broughton's highly polemical rendering of the whore of Babylon interprets Revelation with reference to the Catholic Church. The whore of Babylon (here wearing the papal tiara), and the calf in the centre of the image, identified by the word "Pope," stand in front of Rome's seven hills (see Figure 14). Such polemicized iconography is, in theory, unnecessary given that Broughton makes the same associations in his written text, but the picture helps to visualize a series of relations in a text that is highly figurative and whose interpretation is highly referential. Luther seemed to understand this point too, for while he resisted the inclusion of illustrations in the Bible, he made an exception for Revelation. In the 1520s, he would work with Lucas Cranach to illustrate many of the same polemical associations used by Broughton above (Parshall 1999).

Figure 14: Engraving of the whore of Babylon from Hugh Broughton's A concent of scripture (London, 1590). Image courtesy of Thomas Fisher Rare Book Library.

The second illustration offers a very different visual of the Book of Revelation. Rather than honing in on a single iconic figure, like the Whore of Babylon, in Joseph Mede's rendering, or "key," the entirety of the book is mapped (see Figure 15). The scrolling of the book and the opening of the seven seals occurs, as Peter Stallybrass has shown, through a series of textual media and devices (open books, scrolls, and bookmarks) (Stallybrass 2002; Stallybrass 2009). Most interesting in this context is how the single image attempts to digest hundreds of chapters found over twenty-three books of the Bible in a single visualization.

Figure 15: Foldout illustration from Joseph Mede, The key of the Revelation (London, 1650). Image courtesy of Thomas Fisher Rare Book Library, University of Toronto.

But interpreting the schema's meaning is another matter, and as Kevin Sharpe has shown, readers of Mede interpreted his work in very different ways (Sharpe 2003, 137-139). While Mede may have offered a conservative commentary on Revelation, the work was subjected to radical reinterpretation. As Sarah Hutton notes, "by an irony of history, his scheme, or variants of it, enjoyed huge popularity among those he would have regarded as his theological nemeses" (Hutton 2003, 32). Indeed, Mede died in 1628, just years before the outbreak of the English Civil Wars, and only three years before his work was translated at the bequest of radical sects. Seeing Broughton's polemic read into the Mede schema, especially in the translated editions that appeared in the turbulent decades of the 1640s and 1650s, illustrates that reading is always a subjective act. Such a reminder is relevant to how we assess the functionality of digital tools like Bubblelines, for while the tool may be designed to visualize relationships in novels and other kinds of corpora, these visualizations must be interpreted, and readers will interpret the patterns they present differently.

The designers of current visualizations, like the makers of early woodcuts and engravings, may offer new and exciting ways into material, but answers can only be derived from readers. Builders of digital tools must never lose sight of this fact, for the strength of any tool lies largely in its plasticity, the ability for readers to use the tool in ways that mesh with their research. One of the strengths of INKE's current tools and prototypes is that they are designed with the reader in mind. In fact, the three tools discussed all allow the reader to modify and/or curate the tool and/or the findings they generate. As we have seen, by enabling the reader in this way, these tools invest in the logic of different textual traditions. The annotated index and the sammelbände are both driven by desires to customize, while visual schemas are often designed to amplify text. As we continue to adjust to an ever-evolving electronic space, one where our readings are supplemented by those of the machine, and as we attempt to map the place of the reader in this complex exchange, we must keep one eye on the common denominators that persist no matter what media humans work through. To look at past ways of reading when designing tools for future readers is simply to recognize the small, if important, chapter we are currently writing in the larger history of communication.