Digital archivists tend to disagree about the place of paratexts. Whereas Google Books often scans texts at such a low resolution that anything but printed words are difficult to discern, Andrew Stauffer’s Book Traces project and Steven Olsen-Smith and Peter Norberg's The Melville Marginalia Project aims to identify individual copies of nineteenth- and early-twentieth-century books in libraries by highlighting their unique marginalia and inserts. Illustrations, advertisements, marginalia, boards, and decorative initials—the effluvium of the print form—does not digitize easily. Moreover, in terms of library and information science, paratexts resist standard means of categorization. Paratexts are problematic because they offer an exception rather than a type. To scholars, they often seem extraneous or even detrimental to the written texts they accompany. Marginalia, for instance, simultaneously defaces and compliments a text. Advertisements are a distracting and commercial accretion to an artwork. And yet, all paratexts provide necessary context for understanding the complexity and fullness of print history. The question I will address in this paper is how archivists ought broadly to understand paratexts, and how specifically should they treat nineteenth-century illustrations.

Numerous digital archives have taken on the task of scanning, categorizing, and tagging illustrations ( e.g. the William Blake Archive, the Cervantes Project, Cardiff University’s Illustration Archive), and yet the purpose and constraints of this task remain unfixed. In fact, Julia Thomas notes in her recent Nineteenth-Century Illustration and the Digital (2016), that owing to the uniquely important role of context for these paratexts—usually the book or periodical—"the digital might appear an alien environment for historic illustrations." While the role of the digital image archive concerned with illustrations remains unsettled, recent scholars have used the affordances of the digital archive to open up new avenues for curation and exploration. Using as a case study a digital archive that I direct and edit titled Visual Haggard, a NINES indexed and peer reviewed archive that contextualizes and improves access to the illustrations of Victorian novelist H. Rider Haggard (1856 - 1925), I argue that digitizing illustrations must be inclusive.

I will consider the problem of inclusion and exclusion in digital archive curation. As paratexts, illustrations are lumped together with a number of visual objects that initially accompanied fictions. For this reason I explain the necessity of using metadata to differentiate illustration types. The large decorative initials which appear in many nineteenth-century texts, but originated in medieval manuscripts, are less illustrations of the text than embellishments. However, their ideological function is significant and multifold. Similarly, advertisements were often in conversation with serialized fictions—whether thematically or stylistically. In this paper I discuss strategies to enable digital image archivists committed to creating an authentic encounter with the history of print to avoid ignoring or marginalizing these types of unique and difficult paratexts.