When museums and archives moved online their roles shifted, and nowhere is this so apparent as with visual objects. Thanks to the tremendous efforts of digital archivists, primary texts–written, aural, and visual–are no longer cordoned off behind marble walls that admit only willing patrons. Once made available online and in open access form, encoded tags, keywords, and annotations exist as part of the entire Internet, and not merely within the context of one single database. Typing a database’s metadata keywords or phrases into a search engine like Google draws-in individuals previously unaware of an archive and its subject matter. Search engine results create continuity between various websites; they are dependent on algorithms, and for that reason the most useful and discoverable metadata must be that which most fully accords with the logic of the search engine. Yet facilitating discovery poses numerous challenges. While the judicious implementation of tags is necessary to the work of the digital archivist, achieving a complete, universal, or unbiased set of metadata tags is obviously impossible. No object can ever be completely searchable. Relatedly, some searches return false connections (through synonyms or trending searches). However, using descriptive metadata that serves the same purpose as annotations might illuminate basic content such as the intent of the original creator and information about the reading experience of historical audiences. Annotations should anticipate the likely needs of scholarly and lay readers while forecasting the possible interests of audiences in the future. Today it is rare for object tags to achieve even a portion of these worthwhile qualities, and possibly for good reason.

Descriptive metadata makes primary materials available immediately, unexpectedly, and intimately. This accessibility affects more than just scholars. Discovering historical documents and artifacts can lead to creative interpretations and remixes in popular culture. In fact, cyberpunk novelist William Gibson states: “Our culture no longer bothers to use words like appropriation or borrowing… The record, not the remix, is the anomaly today. The remix is the very nature of the digital”  [Gibson 2005, ¶10].[2] Placing digitized archival materials online expands and re-contextualizes their significance, and visual texts are particularly liable to remixing. Consider the Internet meme created using the 1793 self-portrait of French painter Joseph Ducreaux (Portrait de l'artiste sous les traits d'un moqueur, Louvre Museum). These witty, entertaining, and topical remixed memes represent the best of online culture. Joseph Ducreaux’s impish expression and pointing fingers have inspired meme creators to make comics combining contemporary sentiments with archaic phrasing. One anonymous author, for example, attached the phrase “Gentlemen, I inquire who hath released the hounds?,” which a little thinking reveals to be an overwrought parody of the Baha Men song “Who Let the Dogs Out?” (2000). More and more educators hold up the meme form, and this subset specifically, as exemplary for expanding vocabulary, encouraging critical thinking, and introducing the public to art history. Unfortunately, facilitating access to primary materials also has the potential to perpetuate, and even proliferate, stereotypes about historically oppressed groups. As the numerous Black Sambo caricatures of Barack Obama available online suggests, racist primary materials also have the ability to resurface as inspiration for contemporary hate speech. Access to historical images is a reality, and in many respects a positive one, but how these texts are discovered and contextualized by way of metadata tags and database structure remains unsettled.

The object collection database for the Pitt Rivers Museum, an institution housing over half a million anthropological and ethnological artifacts, provides a useful example of metadata’s role in archival practice. In addition to more concrete information about each artifact’s dimensions and material, this archive sorts entries using Keywords (amulet, hook, quiver) and Classifications (agriculture, clothing, writing). Generally absent though is interpretative information such as, say, an amulet’s possible significance or the likely ceremonial function of a piece of clothing. This is no great surprise considering that, as Jeremy Coote, a curator and joint head of collections at the Pitt Rivers Museum explains, “the databases were initially constructed for internal use. It was then decided to make them available online for the benefit of researchers and the wider public”  [Coote 2016]. The repercussions of openly releasing museum records online are numerous. Scholars may now conduct research in the Pitt Rivers Museum object collections virtually at anytime and from any place in the world. While access democratizes the collection, it has also divorced these artifacts from curatorial supervision. Experts may not require guidance by way of heavy editorializing to glean needed information from the archive, but the absence of historical and cultural context may be disadvantageous to non-academics tasked with interpreting these objects from metadata created to assist scholars. The Pitt Rivers Museum object collections interface provides the public with access to identifiers like the accession number, geographical region, cultural group, and field collector, but without interpretive and contextual information it remains most appropriate for academics, archivists, and anthropologists. Although the site offers curated blog posts and education guides, a lack of editorial commentary within the structure of the database itself can make visiting the Pitt Rivers object collections database rather disconcerting for non-specialists, and possibly traumatic to visitors encountering its offensive content.

What interests me about the Pitt Rivers Museum’s digital archive is its historicity. This database highlights the museum’s role as not only an active and growing anthropological collection, but also an institution bounded by nineteenth-century traditions and ideological assumptions. More so than other institutions, the Pitt Rivers Museum contains, and was in previous decades responsible for generating, ideologically contentious and sometimes hateful content. For this reason, the collection provides access to not only images and metadata pertaining to the artifacts of ancient and diverse peoples, but also letters, manuscripts, and photographs documenting the work of Pitt Rivers and his colleagues in the field. In order to ensure that users are made aware of the status of these records the database includes the statement: “It is important to see how objects were perceived in the past, therefore we preserve all the information ever recorded. Some of the historic wording may now appear outdated and even offensive”  [Pitt Rivers Museum 2015]. Curators at Pitt Rivers consider this statement to be so important that the popup windows accompanying every object in the database paraphrase this last sentence. Mr. Coote explains the necessity of repeating this disclaimer for ensuring “that such an explanation was present on any single record, which might of course be printed out”  [Coote 2016]. Both historical integrity and transparency about the information’s status is laudable but limited. Disclaimers resemble trigger warnings because they notify visitors of offensive content. Yet disclaimers also serve to disassociate the mission and ideology of an institution from the content it hosts and disseminates. In fact, the Pitt Rivers Museum implicitly supports the notion that publishing information differs from upholding it. While I have no interest in tackling the philosophical and ethical question of whether some documents ought to be made public ever, an issue about which Wikileaks continues to grapple [Zifcak 2012], the disclaimer paradigm is passive and typically neglects the opportunity to engage with visitors, scaffold user experience, and critique the past.

I pause to consider a particularly upsetting example taken from the Pitt Rivers Museum manuscript collection, albeit a transcription rather than an image, and then consider opportunities to critique this traumatic content. This archive records several sexist, racist, and xenophobic letters written by Patrick Michael Byrne (1856-1932) to Anglo-Australian anthropologist Baldwin Spencer (1860-1929) on the topics of science and the “Native Question.” Although Byrne was the officer-in-charge for the Overland Telegraph Line, and not a scientist, he took a keen interest in evolutionary biology and ethnography [Mulvaney 2000, 12]. In a letter dated 24 May 1895, Byrne mocks Alexander Thomas Magarey (1849-1906) and Thomas Worsnop (1821-1898), members of the Royal Geographical Society of Australasia, for what he interprets as their sympathy towards the religion and art of native Australians. Byrne dismisses the notion that these aboriginals have a true spiritual life, joking that a religion as profitable as Theosophy might be invented with the assistance of an aboriginal “conjuror,” because “if a third rate female trickster can deceive an Australian Judge, one should be able to educate a nigger to gull the ordinary Public in a very short space of time”  [Byrne 1895]. Encountering and contextualizing attitudes like those articulated by Byrne and his contemporaries is an unfortunate reality for historians. In past decades, feminists, race theorists, postcolonial critics, cultural studies scholars, and historians of science have all worked to mediate these threads of nineteenth-century bigotry, but I wonder about the consequences of making Byrne’s words publicly available on the Internet. Is it the responsibility of digital archivists to curate and annotate the hateful objects they release into the online public sphere, or should these statements be made outside of the archive in peer-reviewed journals, edited collections, or academic blogs?

Few would argue that Byrne’s letter should be censored or removed entirely, myself included, but scholars must think carefully about the context, accessibility, and descriptive tags that accompany hateful digitized artifacts. Instead of remaining silenced in a box stored physically on site, objects in digital archives have the potential to speak again through the public’s often-unpredictable keyword searches. All voices and ideas rendered searchable online are immortal. Like zombies, search engines ensure that proper keyword entry revives all lexical metadata. But what information might accompany, if not every, at least all problematic texts? Perhaps portions of archives containing obviously offensive content might incorporate specific editorial statements rather than general disclaimers (although it is worth noting that the manuscript collections portion of the Pitt Rivers Museum’s website that contains Byrne’s letters does not include the statement about “outdated” and “offensive” content that accompanies the object collections database). The Pitt Rivers Museum’s treatment of its collections suggests that, in a sense, digitizing hate-promoting primary documents revives these voices, so, in the case of Byrne, archivists might consider including links to an explanation of the historical relationship between bigotry and the discipline of anthropology, the British Empire, Social Darwinism, and/or Spencer and Byrne’s biographies. Archivists might also direct readers to something written by the community of Australian peoples Byrne disparages. Finally, metadata labeling this material as “racist” might be included as a TEI tag so that researchers looking to study this type of material can more easily discover and offer scholarly critique of Byrne’s letter. In short, rather than a static digital record of hate and prejudice, however historically accurate, the Pitt Rivers archivists might use Spencer’s words as an opportunity to talk back to the past.

I admit the grounds of my discussion about the Pitt Rivers Museum’s digital archive are slippery. In the first place, ideas equally if not more offensive to those expressed in Byrne’s letter can be found in the books digitized and made searchable through optical character recognition (OCR) rendering in digital libraries including Project Gutenberg, Google Books, HathiTrust, and Internet Archive. Second, it takes an enormous amount of time and funding to add annotations to a museum or archive’s vast collections–resources that might be better spent on preservation and maintenance initiatives. To the first concern I counter that annotation and curation should be what sets digital archives and museums apart from digital libraries. In the second, it seems to me that because financial barriers vary on a case-to-case basis, institutions might prioritize curation for cases of extreme or obviously hateful content.

Henry Louis Gates Jr., editor of Harvard University’s Image of the Black in Western Art Archive and a self-described collector of what he terms “racist memorabilia,” suggests that there are compelling reasons for exhibiting racist artifacts, with “the most important function of displaying and collecting this stuff [being] a didactic one: critique”  [Gates 2013, ¶5]. In Gates’s words, “we need to study these images in order to deflect the harm that they continue to inflict upon African Americans, at the deepest levels of the American unconscious”  [Gates 2013, ¶20]. Because racism persists in contemporary culture, material artifacts of this hate must be preserved, displayed, and studied. But what does it mean to use visual artifacts as a means of critique in digital archives? In this section I survey examples of online collections devoted to racist American artifacts, as well as strategies for their inclusion in hypertext formats.

The Jim Crow Museum of Racist Memorabilia at Ferris State University has a similar mission to that expressed by Gates of using, “objects of intolerance to teach tolerance and promote social justice”  [Pilgrim 2016]. Highlighting racist American ephemera promotes change by underscoring the trauma these images continue to cause. The legacy of racism in the United States is far from resolved, so it remains necessary for humanists to highlight past inequalities in order to promote social justice in the present. Because this museum’s mission includes advocating “the scholarly examination of historical and contemporary expressions of racism,” founder and curator David Pilgrim has made context and curatorship an extremely important component of this institution [Pilgrim 2016]. It is for this reason that, unlike many other museums, explanatory statements and labels are lengthy and plentiful. The visibility of offensive images must not be suppressed or forgotten, but it should not exist without curation. Images of hate speak too loudly and powerfully to go unmediated by scholars and activists.

Contrasting with the Pitt Rivers Museum’s large database of minimally tagged objects, the website for the much smaller (9,000 objects) Jim Crow Museum accompanies its “objects and collections related to racial segregation, anti-black caricatures, civil rights, and African American achievement” with heavy annotations through informational videos, essays, and lesson plans [Pilgrim 2016]. In fact, online there is no database record. Instead, Pilgrim embeds a sampling of objects within categorical hypertext essays focused on caricature types: Brute, Picaninny, Tom, Nigger, Saphire, Jezebel, Mammy, Coon, the Tragic Mulatto, Golliwog and Nat. To view images of these caricatures visitors must first access an essay contextualizing and historicizing the stereotype. The Mammy essay, for example, includes five images of individual objects, unaccompanied by title or creation date in addition to a sixth panoramic photograph of the physical display case devoted to this caricature type. There is also a link to a slideshow showcasing 75 Mammy stereotype images. Instead of including objects in a labeled and sorted database, the website contextualizes a sampling of artifacts in a way that explains their social, political, and historical significance.

It is important to note that none of the images available on the Jim Crow Museum website are tagged with metadata enabling visitors to search for them using criteria other than their stereotype. For instance, if I wanted to see all examples of ceramics or advertisements, I would have to sort through all the images discretely. Pilgrim explains that the construction of a public database containing more complete records for the museum is underway [Pilgrim 2017]. While this will provide an interface to permit searching, the records for the entire collection will never be made publicly available online. In fact, database access may even be restricted to members of the Ferris State University community. The editorial decision to include educational context about the legacy of racism first, and access to images and examples as a means of supporting this message second, exemplifies the value of heavy editing for supporting the cause of social justice.

Like many archivists tasked with cataloguing offensive subject matter, Pilgrim is wary of enabling racists rather than defeating them. He is well aware that some persons may use the Jim Crow Museum website as a distributor of hateful images, rather than an educational tool. If the website’s caricatures appear remixed on tee shirts and posters then they perpetuate the historical injustices they were meant to combat. In fact, unlike many other museums that highlight unique and rare objects from the collection to attract visitors, the Jim Crow Museum makes it a policy to keep its rare collections offline [Pilgrim 2017]. All the images accessible on the site can be found elsewhere on the Internet without trouble. Pilgrim deliberately keeps these records of hate outside the grasp of well- and ill-intentioned visitors alike because an inclusive database is unnecessary for accomplishing the primary aims of this institution. Because the mission of this museum is public education and challenging racism in contemporary culture, a complete database of Mammy objects is superfluous to the work of demonstrating this caricature’s role as “a figment of the white imagination, a nostalgic yearning for a reality that never had been”  [Pilgrim 2000, ¶9]. While the absence of a public database may hinder some researchers, it achieves the museum’s objectives for the public admirably.

The structure of Pilgrim’s Jim Crow Museum suggests that the key to social justice is the curation of digitized archival materials through thoughtful and well-researched annotations. Several other archives and museum websites devoted to hateful visual objects adopt strategies similar to these. Websites for the Center for Holocaust and Genocide Studies (CHGS) at the University of Minnesota as well as the HarpWeek Cultural History feature titled “Toward Racial Equality: Harper’s Weekly Reports on Black America, 1857-1874” both contain relatively few objects (less than 100) and use disclaimers and critique to contextualize historical objects. Hateful images of African Americans, Native Americans, Jewish peoples, Roma and Sinti (Gypsies) appear in CHGS’s “Visualizing Otherness” series, and disclaimers accompany the first two categories. In the second instance, for example, editors include a note stating:

Many of the images shown here are offensive to Native-Americans. However, they are shown because they represent means within popular culture by which the sense of inferiority was imposed on Native-Americans during the period after the defeat of Native peoples on the Great Plains and the establishment of the ‘Reservation’ system by the Federal Bureau of Indian Affairs.  [CHGS 2017, ¶1]

The images accompanying this disclaimer include toy sets, cigar bands, and historic photographs in the first set, while contemporary Cleveland Indians, Kansas City Chiefs, and Atlanta Braves merchandise dominates the second. Minimal descriptive tags accompany images in both sets, and there are no means in place to search these images: their position in an archive devoted to “Otherness” subsumes competing significance or research criteria (artist, medium, dimensions, etcetera).

HarpWeek, an archive indexing the illustrated American periodical Harper’s Weekly (1857-1916), accompanies “Toward Racial Equality,” with an editorial “Warning” beginning: “Website visitors should be warned that several of the words, descriptions, and images from Harper’s Weekly are considered racially offensive by today’s standards”  [CHGS 2017, ¶1]. Harpweek assumes that any contemporary visitor, and not just those members of the oppressed group (as is the case with the CHGS disclaimer), may find this content traumatic. Like all of the archives containing hateful content that my project considers, editors contend that the need to digitize offensive materials outweighs the implied argument to veil these histories. In fact, my research uncovered no counter to this claim among academics excepting Gates’s assertion that some persons he knows: “think the whole lot [of racist memorabilia] should be assembled into one gigantic bonfire, incinerated, and the ashes buried in an impenetrable vault, or strewn over the broadest reach of the deepest ocean never to be displayed again”  [Gates 2013, ¶2]. Digital archivists and museum curators seem to argue with a single voice that collecting and critiquing offensive objects is necessary for generating positive change; what differs is how this content is framed. Harpweek’s warning continues: “The materials are presented in order to give a true historical picture of the leading 19th-century newspaper’s view of black Americans. We at HarpWeek hope this site will serve as a valuable resource which provides an important perspective on the multifaceted history of black Americans, generates a deeper understanding and respect for the subject, and sparks further interest in its study and discussion”  [CHGS 2017, ¶1]. In point of fact, this site includes not only this general disclaimer, but also a statement written by Randall Kennedy, a law professor at Harvard University, about the necessity of using, and not censoring, the term “nigger” in historical archives. While these warnings absolve the site of disseminating racist ideas and materials, it is the editorial critique–the timelines, introductory essay, and notes accompanying scanned illustrations–that permit “Toward Racial Equality” to create a space for activism and justice.

Because editors at Harpweek are aware of the hurt racist historical objects continue to inflict, however good the archivist’s intentions, many of the collected Harper’s Weekly illustrations published during the 1850s, 60s, and 70s include contextual information beyond the titles and publication dates. While text from Harper’s Weekly accompanies some of the illustrations, which may inform readers of the image’s content and the artist’s identity, a few go so far as to include editorial notes. For instance, “The Louisiana Murders–Gathering The Dead And Wounded” illustration shows two white men on horseback observing numerous African Americans, several of whom are apparently injured or dead. Because the solemn scene does not include text, a note explains that the illustration shows the 13 April 1873 Colfax Massacre, which claimed the lives of 70 black men during southern Reconstruction. This note helps visitors to make sense of an illustration depicting a terrible but largely forgotten incident in United States history. This content adds much needed historical background to an archive concerned with sharing how subscribers to Harper’s Weekly perceived African Americans, which factors into the website’s rhetorical framing journey “toward racial equality.” However, beyond the general categories of “Slavery,” “Civil War,” “Reconstruction,” and “Culture & Society”–sections that accord to the Timelines portion of the site–without access to the paid HarpWeek database there is no way to search them based on criteria such as content, artist, or dimensions within the feature “Toward Racial Equality.”

There can be no doubt that the websites for the educational and social justice promoting Jim Crow Museum, the CHGS, and HarpWeek’s “Toward Racial Equality” use disclaimers, notes, and essays critiquing stereotypes to treat offensive and traumatic images in a sensitive and critically engaged manner. Accountability is a primary concern for these archives, which their editors admirably and unceasingly aim to promote. These projects all use a form of heavy annotation to give voice to historically oppressed groups, but in order to control the narrative these sites also forego the database’s searchable structure. While it is possible to search within any of these websites using Google grammar (“site:syntax”), these hypertext sites adopt the strategy of sorting images by editorial commentary rather than descriptive metadata tags. Visitors are discouraged from searching for specific archived images because on the whole the metadata attached to objects within these three sites is generally limited to image titles and dates. Usually excluded from these digital records is specific information about, say, the artist, dimensions, material, style, or a description of content and subject matter (tree, water, war, love). In “Toward Racial Equality,” for example, it would be difficult to identify all illustrations by a specific artist quickly or conclusively. Displaying objects representing hateful ideas beside thoughtful criticism at the expense of searchability is a calculated decision, but I wonder if annotations and critique of hateful historical content might coexist in the resources and metadata of a database.

Can the structure of a database simultaneously serve the goals of public education, social justice, and academic research? The call to amass and critique artifacts documenting past inequities made by Gates and Pilgrim, among others, is opposed to current digital humanities and archival practice in three significant ways. First, adding digital tags addressing identity politics relating to race, class, gender, and religion, beyond the most blatant examples of bigotry, moves the archive away from objective and mimetic documentation of the past and into the realm of subjective editorializing. What seems biased and hateful to some audiences may appear disinterested and inoffensive to others. Second, because all historical objects reflect in some degree the ideology of the era in which they were created, labeling these ideas using metadata tags might be construed as heavy-handed and unnecessary. For example, if, as Tara McPherson explains in “Why Are the Digital Humanities So White?”, “race, particularly in the United States, is central to… shaping how we see and know as well as the technologies that underwrite or cement both vision and knowledge”  [McPherson 2012, 143], then because all historic images produced in the West that depict non-white persons are ideologically complicit in the discourse of race it is needless to tag images using the Library of Congress subject headings of “race” or even “racism.” Ostensibly in the West this argument extends to gender, class, and sexuality among other social and identity categories. Third, archivists and curators are understandably hesitant to preserve and disseminate materials that might cause offense. Although many museums possess racist objects, these are often buried in backrooms and basements, and away from all visitors and donors. The institutional marginalization of trauma-causing objects persists online. In fact, humanist archivists have been a driving force behind digitizing many non-canonical and often hateful historical objects and texts because “[w]orks widely held as cultural treasures are more likely to be digitized than acutely problematic material”  [Gailey 2011, 137]. Humanists motivated to integrate politicized and identity-driven information about historical collections have arguably done as much as those in the library and museum science field to think through the question of social justice and the digital archive. Although texts out of the library sciences like Museum Frictions [Karp et al. 2006] and Exhibiting Cultures [Karp and Lavine 1991] have contributed greatly to developing ideas about the “museum of conscience,” the difficult task of creating digital archives concerned with offensive objects is truly interdisciplinary. To advance the cause of social justice, literary and cultural studies and digital humanities scholars have much to learn from digital preservationists and library and information scientists, and vice versa.

Because metadata has such significant consequences on scholars, digital archivists must take the issue of metadata creation seriously–particularly when tagging traumatic material in databases. Scaffolding the viewing experience for both popular and academic audiences requires integrating the disclaimers, contextual essays, and lesson plans for sites like the Jim Crow Museum of Racist Memorabilia with database searchability. As I mentioned in the introduction of this article, one project that foregrounds this concern in innovative ways is Gailey’s Race and Children’s Literature of the Gilded Age (RCLGA). Gailey and her collaborators at Washington University in St. Louis facilitate searching by adding TEI tags. In the case of the Reconstruction era Uncle Remus children’s books by Joel Chandler Harris (1848-1908), the team essentially translated the stereotypically exaggerated vernacular of the African American characters into regularized English. This extra, or “heavy,” inscription makes the database more accessible to humans and computers alike. For instance, regularizing Harris’s phrase “ole Brer Wolf want ter eat de little Rabs” allows visitors to find this passage by searching for the word “rabbit” [Gailey 2011, 135]. Strategies for enabling searches including “heavy editing” through TEI or database tags like those used by the Pitt Rivers Museum supports traditional academic researches while opening up new avenues for discovery. Comprehensive search criterion not only facilitates precise historical investigations, it also streamlines the types of distant reading and data mining that digital humanists use to create sophisticated visualizations and models.

Gailey and her team’s technique for making the content of the database more accessible stops short of actually critiquing Harris, whose fictions made apologies for slavery and were responsible for creating many stereotypes about African Americans that persist into the present day. The RCLGA does not use TEI to point out or respond to the racist aspects of the archive. Gailey argues that “heavy editing” by way of a dual transcription–recording Harris’s words and their translation to regularized English–allows the RCLGA to maintain editorial objectivity because it “makes no claim about the rightness or wrongness of the readings, only how standardized their spellings are”  [Gailey 2011, 136]. This concern with objectivity aligns the project with the Pitt Rivers Museum more so than the Jim Crow Museum, the CHGS, and HarpWeek. While social justice is not a central concern for the archive, education certainly is. Gailey explains that: “Because of the controversial nature of these works and the variety of audiences the edition may draw, the editors plan to provide multiple methods of access to the editions. One track would be geared toward elementary schoolers and their teachers, another toward high schoolers, and one, of course, for scholarly readers, each with its own level and kind of editorial intervention”  [WUSTL 2017]. In this way the structure of the RCLGA database facilitates educators at the primary and secondary level, rather than assuming that only researchers will view these texts without support. Gailey and her team acknowledge the offensiveness of the material and have designed the project to scaffold the experience of multiple types of visitors. This strategy supports audiences from academic settings well by providing the context necessary for instructors to use the archive, hopefully with the intention of advancing the message of social justice in their classrooms. How the illustrations archived in the site will factor into this project remains to be seen. Although Gailey does not address the question of heavy editing as it relates to the illustrations archived in the RCLGA, I extend the “heavy editing” approach to the project of making offensive archived images both searchable and contextualized for educators and researchers alike.

As the director and editor of Visual Haggard, a digital archive dealing with the type of images historically complicit in Western systems of oppression, “heavy” descriptive metadata tags provide an obvious avenue for adding contextual and categorical information useful for all audiences and allied to the cause of social justice. Visual Haggard is intended to centralize and improve access to the illustrations of popular Victorian novelist and imperialist H. Rider Haggard (1856-1925). The archive contains over 1,300 images and receives approximately 1,500 page views every month. Visual Haggard has recently passed peer review with NINES, the Networked Infrastructure for Nineteenth-Century Electronic Scholarship. Tagging every illustration to describe its subject matter has been one of the most significant initiatives of Visual Haggard. Inspired particularly by the long-running William Blake Archive, which boasts the most complete and effectively tagged images of any academic database, the metadata accompanying each image on Visual Haggard moves beyond the objective categories of artist, novel, edition, page number, date of publication, publisher, publication city, and image source (specific libraries, athenaeums, or private collections). With the assistance of undergraduates at the Georgia Institute of Technology enrolled in my English 1102 course, “Visual Culture, Digital Archives and H. Rider Haggard,” I am in the process of tagging all objects visible in these illustrations (window, horse, hat, tree), the setting (continent, country, house, forest), the names of all depicted characters, the posture and expression of characters, the number of human figures (e.g. 1 figure, 3 figures, crowd), and the style of art (Pre-Raphaelite, Symbolist), to name a few categories. But in addition to these descriptive tags, the database will also include searchable tags for critical themes relating to identity and social justice.

The need to tag and contextualize images dealing with race and racism in the Visual Haggard database has become increasingly apparent in recent years. As an avid imperialist best known for his time spent in South Africa, Haggard’s views towards race were complex and influential. While the novelist became deeply enamored with the language, history, and culture of the Zulu people, he also expressed stereotypically bigoted disdain towards what he termed Kafirs (also spelled Kaffirs; a racist term for black Africans in South Africa), Hottentots (an offensive term for the Khoikhoi peoples), and half-castes (a derogatory phrase suggesting the inferiority of mixed-race people which formed a part of anti-miscegenationist rhetoric). The heroic and neoclassical appearance of the Zulus depicted by Haggard’s illustrators seems to reflect the novelist’s admiration for these “Romans of Africa”  [Pocock 1993, 21]. However, scenes showing white men aiming their firearms at unarmed native servants, or slaughtering scores of brown men of African, South American, or Middle Eastern descent serve as terrible cultural records of imperialism’s violence and racism in the West. The artworks Visual Haggard collects are graphic reminders of the ideological work that adventure fictions like King Solomon’s Mines (1885) and She (1887) accomplished. The pictures that accompanied Haggard’s popular stories advanced the idea of Empire among British and American audiences, and the archive’s metadata must reflect this.

Visitors can learn about this novelist’s general attitudes and prejudices regarding race, imperialism, and gender from Visual Haggard’s editor’s statement and biography of Haggard. However, encountering these illustrations and deciphering their import often occurs on an image-by-image basis as each illustration archived in the database possesses a unique take on these themes. For this reason, Visual Haggard uses the Disqus commenting application for individual images. Towards the bottom of every illustration’s hypertext page, expert and lay visitors to the archive may contribute commentary and glosses. These individualized annotations democratize the process of creating and disseminating contextual information, however flawed and subjective, which is especially helpful for annotating offensive objects. For instance, I recently added a Disqus comment to Visual Haggard that historicizes, defines, and critiques the racist slur “Kafir” that appears on the illustration for Haggard’s The Witch’s Head (1884) titled “A shapely Kafir girl.” Although I assessed this offensive image created by artist Charles Kerr (1858-1907) in an article for the Journal of Victorian Culture Online [Holterhoff 2013], integrating critique into the interface of the archive educates and promotes social justice for a broader audience (those who might stumble upon the archive rather than access it deliberately). Non-academics form an unlikely audience for periodicals like the JVC, so contextual information about offensive content should be immediately accessible for the archive’s users. While Disqus comments offer one means for reaching a general audience, critical audio guides for select illustrations created by Georgia Tech undergraduates and supported by Knight Lab’s SoundCite JS are also in the works. In addition to these strategies for adding contextual information to the site, I use metadata to identify and group critically significant images.

How to tag offensive because racist or sexist images has been a contentious issue for Visual Haggard since the archive’s early days. Historically, I have tagged images of Africans using the descriptive words “race” and “African,” but recently I have begun to add the more critically weighty term “racism” in cases where illustrators depict black characters as victims of violence or exploitation. Adding this tag–which can be searched for using the site’s search bar–makes the bigotry of Haggard and his illustrators more visible. Tagged examples of “racism,” like other critical tags concerning identity and oppression such as “sexism,” will likely be of the greatest interest to scholars. However, these tags may also serve to educate curious nonacademic visitors while promoting justice through identifying and contextualizing histories of subjugation. Of course, the value and usefulness of these terms is subjective and not without problems. To begin, Visual Haggard only permits plain string wildcard searching, so although searching for “gender” or “racism” works well, typing “race” into the search bar will also turn up “embrace,” “disgrace,” and “Horace” (a protagonist of Haggard’s She). More theoretically, this archive’s critical tags will certainly shift to reflect a variety of voices as interpretations of well-studied images change and less-studied images proliferate. But, however imperfect, critical tags provide a worthwhile layer of editorial critique.

Ambiguity makes Visual Haggard a valuable tool in the undergraduate classroom for provoking discussion about ideology and the cultural history of the “Scramble for Africa,” as students are eager to weigh-in on appropriate labels for these texts and artworks. However, this same resistance to a singular or even limited set of interpretations makes tagging the illustrations archived on the site challenging. In the case of the imminently problematic image “A shapely Kafir girl,” I not only use the tags “race” and “racism,” but also “gender” and “sexism” because, in addition to using a racial slur, this illustration depicts two white men leering at an objectified African woman’s body. Here, the archivist is alleviated from the burden of interpretation, as Haggard’s own words point to, and likely guided the design for, this image’s racist and sexist undertones. Yet in the remainder of the archive the issue of descriptive metadata tags is seldom so clear-cut. I elect to use the term “race” far more often than “racism” because Haggard and his illustrators’ complex ideas about Africans, and especially the numerous heroic if stereotyped Zulu warrior characters, make determining the appropriateness of this label difficult. By applying the paradigm of “heavy editing” to illustrations through Disqus comments, metadata tags, and eventually audio guides, I have begun to make arguments and scaffold user experiences in a manner that explicates critical content in these illustrations to enhance the site’s ability to educate and promote justice.

It is worth noting that Visual Haggard does not use racial slurs as tags. While the Jim Crow Museum uses offensive historical language for the purpose of educating visitors, because the role of racism and sexism in Haggard’s illustrations is more subtle and ideological I adopt accepted cultural studies and Library of Congress keywords as database tags. Although “kafir” does appear as a searchable tag owing to Kerr’s image caption, I have avoided other opportunities in which this term–a frequent one in Haggard’s lexicon–might appear. Consider Russell Flint’s illustration from King Solomon’s Mines of Allan Quatermain and the African character Jim titled “I saw Neville’s wagon move off. Presently Jim came back running” [Flint 1907]. At an earlier point in the text Haggard labels Jim “a Kafir hunter” explicitly [Haggard 2007, 15]. Therefore, readers are meant to interpret Jim’s actions in terms of this defining characteristic whenever he features in the plot. However, I have not elected to do so because once I add “kafir” to the list of descriptive tags attached to this image I will have perpetuated and in a sense authored this term afresh. Because Flint never uses the word “kafir,” and his depiction of Jim is not violent or otherwise obviously “racist,” I instead tag this image with “African” and “race,” and I leave the task of interpreting Jim's representation in text and image to scholars. Illustrations are never mimetic, so this archive's metadata should concern the ideological content of images and their accompanying captions alone, rather than the entirety of Haggard’s text. While Visual Haggard must always be simultaneously over- and under-determined by “heavy editing,” editor’s statements, and Disqus comments, the richness of this digital archive’s metadata and Disqus comment creation process–an accretive project which will remain perpetually incomplete and unsettled–aspires to facilitate literary, cultural studies, digital humanities, and art history scholarship at the same time the archive looks to hold itself accountable to the historically disenfranchised groups traumatized by the objects this site archives.

In this essay I have wrestled with the issue of how digital archivists might best use metadata and annotations to sort and contextualize offensive–and particularly racist objects–in a manner that educates the public, supports social justice, and facilitates scholarly research. I suggest that archivists and curators do well to move away from relying wholly upon disclaimers in order to be held accountable to and act as allies for oppressed groups and persons. Museums, libraries and archives that enhance the world’s cultural heritage by making its collections available online often wrestle with questions relating to digitizing and making traumatic materials accessible. Yet as the metadata of digital archives increasingly forms a significant portion of data mining and digital humanities projects, curators must evaluate shifts in the value and use of the digitized collections they control.

For individuals working to create and sustain digital archives, the concerns articulated in my article are practical, imminent, and must be weighed thoughtfully. Although the process of adding substantial contextual metadata to archives can be time consuming and expensive, I have argued that by prioritizing problematic materials and viewing metadata as a process rather than a conclusive endpoint, heavy editing can allow for a more sensitive approach to digitizing and making historical objects accessible. I take seriously Tara McPherson’s warning: “Today, we risk adding the digital humanities to our proliferating disciplinary menus without any meaningful and substantial engagement with fields such as gender studies or critical race theory”  [McPherson 2012, 150]. It is my hope that this paper advances scholarly conversation about the relationship of digital archivists to traumatic materials at the same time it suggests some best practices for allowing metadata to contribute to the cause of justice.