Escolar Documentos
Profissional Documentos
Cultura Documentos
CENTURY
Series Editor: Andrew Stauffer
NINETEENTH-
CENTURY
ILLUSTRATION
AND THE DIGITAL
Studies in Word
and Image
Julia Thomas
The Digital Nineteenth Century
Series Editor
Andrew Stauffer
University of Virginia
Charlottesville
Virginia, USA
The Palgrave Pivot series publishes short-form monographs on topics at
the intersection of nineteenth-century studies and the digital humani-
ties. Partnering with the NINES Center (Networked Infrastructure
for Nineteenth-Century Electronic Scholarship) at the University of
Virginia, this series will be retrospective and prospective, involving not
only explications of digital projects and theoretical considerations of
methods, results, rhetorics, and audiences, but also projections that chart
a course for future work. The series will also include free-standing titles
for scholars throughout the world not tied to a specific digital project,
but rather synoptic studies of a particular method, approach, or thematic
in digital nineteenth-century studies. The series aims to provide a grow-
ing archival record of the digital nineteenth century across the years.
Nineteenth-Century
Illustration
and the Digital
Studies in Word and Image
Julia Thomas
Cardiff University
Cardiff, Wales, UK
Cover illustration: Pattern adapted from an Indian cotton print produced in the 19th century
vii
viii Acknowledgements
1 Frontispiece 1
2 (In)visibility 15
3 Searchability 33
4 Crowdsourcing 65
5 Tailpiece 95
Bibliography
105
Index
115
ix
List of Figures
xi
xii List of Figures
Frontispiece
The fundamental question that the book addresses is: how does its
migration to a digital environment challenge and change our under-
standing of nineteenth-century illustration, both in terms of its cultural
importance and the way that it signifies? A key component of this signifi-
cation is the dialogue between word and image that illustration puts into
play. W.J.T. Mitchell has claimed that ‘all media are mixed media’, there
is no purely visual, or, for that matter, verbal genre.5 But illustration is
more of a mixed medium than most. To turn to a related theoretical
discourse, illustration is the ‘intermedial’ genre that Irina O. Rajewsky
defines as ‘media combination’, where at least two distinct media com-
bine: ‘These two media or medial forms of articulation are each present
in their own materiality and contribute to the constitution and significa-
tion of the entire product in their own specific way’.6 It is the ‘media
combination’ of word and picture that distinguishes illustration, I would
argue, from Rajewsky’s other example of intermedial relations: ‘medial
transposition’, where a media product is transformed into another
medium (a book, for example, is turned into a film).7 I want to make this
distinction clear from the outset because illustration has often been (mis)
understood in precisely these terms, with the ‘original’ text regarded
as the creative inspiration for and source of the image into which it is
transposed. Illustration here is little more than a pale reflection of the
originary words. However, this is to misconstrue how nineteenth-cen-
tury illustration works. The text is not always the primary, or even prior,
partner in the illustrative relation, as Landon’s picture–poems written
to accompany images indicate. Moreover, an illustration is not simply a
‘transposition’ or ‘transformation’ of the words, but stands in a nuanced
relation of complementarity and conflict, sameness and difference.
Notions of what an illustration is and how it makes its meanings have
been central to my research for some years, but these notions were thrown
into relief during the development of the Database of Mid-Victorian
Illustration (DMVI) (http://www.dmvi.org.uk) and The Illustration
Archive (http://illustrationarchive.cardiff.ac.uk). Both of these resources,
funded by the Arts and Humanities Research Council in the UK, make
nineteenth-century illustrations available online, but their content and
methodologies are very different. Work on DMVI began in 2004 at a time
when major digitisation projects were underway (in December of that
year, Google announced its ‘Library Project’).8 DMVI was funded as part
of a ‘Resource Enhancement’ programme, the resource that it enhanced
being a little-known collection of illustrations from periodicals of the
4 J. THOMAS
1860s and 1870s housed in the School of Art Gallery and Museum in
Aberystwyth University. In fact, so little known was this collection that
it had languished in a disused lift shaft for over 30 years.9 The project
involved travelling from Cardiff to Aberystwyth (not always easy, as any-
one who has attempted the journey by train will know) and setting up a
flatbed scanner in the reading room of the gallery. We could almost have
been Google. The project was not only concerned with the digitisation of
these illustrations, though. A major part of the remit was to explore ways
of making the content of the illustrations searchable so that the images
could be retrieved in terms of what they depicted. After considering sev-
eral image classification systems, we decided to create a bespoke keyword-
ing structure and to mark up each illustration according to this schema. It
was this aspect of the project that turned out to be the most challenging,
even more than the commute from Cardiff to Aberystwyth.
One advantage of our frequent train journeys was the time that it gave
us to think through and discuss ideas. A founding principle of DMVI
was agreed upon while the team was laden with scanning equipment and
waiting for a train to Oxford to see the Forrest Reid collection of illus-
trations in the Ashmolean Museum: the decision to include illustrations
published in 1862, a year that was chosen because it was particularly rich
in terms of illustrated books and periodicals. There were pragmatic rea-
sons for this decision: this was a 3-year project and our ambitions had
to be realised within this fixed timeframe. But there were also theoreti-
cal considerations: we wanted the database to provide a snapshot of the
range of literary illustrations published in a period that was, arguably,
one of the most significant in the history of British book illustration, and
this would not have been possible if we had focused on, say, the work of
a specific artist, illustrator or author. The choice of 1862 (which encom-
passed illustrated novels that began or ended their serialisation in this
year) meant that we could include more familiar illustrated works, like
Frederic Leighton’s illustrations for George Eliot’s Romola and John
Everett Millais’ illustrations for Anthony Trollope’s Orley Farm, along-
side the less familiar images that accompanied poems and short stories in
the popular Victorian periodicals, London Society, Once a Week and Good
Words. As it stands today, the database contains 868 illustrations from lit-
erary works, all of which are described bibliographically and iconographi-
cally (in terms of what the images depict), enabling the user to search
across multiple fields.
1 FRONTISPIECE 5
The Illustration Archive had a different genesis. This was the output of
a Big Data project that aimed to make searchable over a million illustra-
tions from books in the British Library’s collection. The decade between
the development of DMVI and The Illustration Archive was one that
marked a defining shift from what Bethany Nowviskie describes as the
‘boutique digitization’ undertaken by humanities scholars to the mass
digitisation projects carried out by large corporations.10 The images in
The Illustration Archive had already been digitised by Microsoft; so when
we embarked on this project in 2014 there was no setting up camp with
a scanner in the reading room of the British Library. As we soon discov-
ered, though, big data presents its own transportation issues. For a start,
there is too much of it to transfer electronically. The data was copied on
to five external hard drives and these drives travelled by courier from the
British Library down the M4 to Cardiff. When they arrived, we had very
little idea of the illustrations they contained. The books in which the
images appeared covered ‘history’, ‘geography’, ‘literature’ and ‘philoso-
phy’; but these categories were established in the 1850s when they were
much broader. We were dealing, then, not only with considerably more
illustrations than in DMVI, but also different types and taxonomies of
illustrations (the focus of DMVI was on illustrations that accompanied
‘literary’ texts), although the main objective of the two projects was the
same: to create a searchable archive of illustrations that would be freely
accessible online. Because of the number of illustrations in the British
Library dataset, it was impossible to keyword all of them in-house. So,
the project developed the infrastructure for crowdsourced tagging while
exploring the potential for computer recognition of image content. The
resource that we created (a web-accessible front-end to a database of
illustrations, bibliographic metadata and descriptive tags) is currently the
largest searchable online resource dedicated to illustrations.
While these projects provide the foundation for the argument in this
book, the discussion moves beyond their specific mechanisms and out-
comes to develop a way of conceptualising the relation between the digi-
tal and illustration studies. Since I first made the case for the recognition
of illustration studies as a discipline ‘with its own methodologies, vocab-
ularies and critical strategies’, the agenda for this discipline has been set
out by a number of critics.11 Paul Goldman suggests that illustration
studies should involve a re-evaluation of the significance of illustration
across academic institutions and the development of courses devoted to
the subject.12 The images that might form the basis of illustration studies
6 J. THOMAS
are vast in type and number. Illustration crosses different periods and
genres and the potential modes of analyses that such studies embrace are
equally varied, from a close analysis of the illustrations that accompany
a particular work to an examination of illustration as a cultural practice
and its place in the history of the book. However, the critical emphasis
of illustration studies is a specific one, which, as David Skilton argues,
can be distinguished ‘from the more limited objectives of bibliographi-
cal description and aesthetic appreciation’.13 Put simply, the objective of
illustration studies is to analyse how illustrations signify, how they make
their meanings and how these meanings are embedded in the historical
moment of their production and reception.
This objective is not as straightforward as it may initially seem because
the study of illustration needs to begin by advocating the relevance of
illustration and countering the all-too-common misconceptions to be
found in scholarly work, the sorts of misconceptions that view illustra-
tion as a straightforward example of Rajewsky’s ‘medial transposition’, or
worse. Some of the typical fallacies about illustration are as follows: that
illustrations are purely decorative or ornamental; that illustrations are
supplements or ‘added extras’ and can, therefore, be extracted from the
text without loss; that illustrations have a subordinate relation to the text;
that the text exists prior to the illustration; that the text is the primary
means of signification; that the illustration replicates or reflects textual
meanings. References to ‘mere illustration’ are scattered throughout crit-
ical articles and books, so much so that they are simply taken for granted.
It is unsurprising, then, that these errors have found their way into the
field of digital humanities. Martyn Jessop, for example, uses this familiar
notion of illustration as a means of establishing his idea of ‘visualization’:
The term ‘Illustration’ implies an image which serves only to support writ-
ten language; thus the main carrier of information is the associated text not
the image. Many consider that an image can only be a true visualization
when it is the primary carrier of information, not a supplement to a piece
of text. I shall settle on a distinction in which an illustration is intended
merely to support a rhetorical device (usually textual) whereas a visualiza-
tion is intended either to be the primary rhetorical device or serve as an
alternative but parallel (rather than subordinate) rhetorical device.14
advances in illustration studies that have taken place in tandem with digi-
tal initiatives such as DMVI. These resources, Jung asserts, ‘have con-
tributed significantly to shedding light on, and to the understanding of
the importance of, corpora and illustration techniques that did not fea-
ture in traditional accounts of literature and book history’.20 Likewise,
Christina Ionescu points to the increased availability of image databases
and digitised editions as a key component in the growth of studies in the
eighteenth-century book illustrations.21
As Ionescu recognises, the digital resource allows a greater number of
historic illustrations to be seen than ever before. But it does more than
this. The digital archive is not simply a transparent vehicle through which
illustrations are made accessible, but a critical and analytic space in which
they are continually re-viewed. This re-viewing, moreover, takes place in
the creation of the digital resource as well as in the final product. It is
important to make this point because digital humanities has sometimes
been regarded, even among exponents, as a largely ‘instrumentalist’ field
in which the traditional interpretative questions of the ‘humanities’ (that
is, of literary scholarship) are forsaken in favour of the project-based
creation of digital tools and archives. A recent article in the Los Angeles
Review of Books makes this point: ‘What Digital Humanities is not about,
despite its explicit claims, is the use of digital or quantitative methodolo-
gies to answer research questions in the humanities. It is, instead, about
the promotion of project-based learning and lab-based research over
reading and writing’.22 It is true that a humanities research agenda has
not always been evident in digital projects that ‘apply’ innovative com-
putational tools, sometimes arbitrarily, to research areas. It is also true
that applications of computational tools, even to big data, can be narrow
in their conceptual approaches and research implications. But this is not
always the case. An alternative view sees the humanities and the digital
as mutually reinforcing principles, with digital humanities initiatives at
their strongest when the computational infrastructure and the humani-
ties research agenda are interdependent.23 Indeed, this book argues that
there is not a clear line, never mind a binary opposition, between the
development of the computational infrastructure and the research ques-
tions: these research questions are articulated as much in the creation of
the digital resource as in what the computational ‘tools’ make possible.
Questions of what an ‘illustration’ is, how it signifies, and even what
it means to say that a picture ‘illustrates’ a text, are fundamental, then,
to the development of digital illustration archives. The discussion that
1 FRONTISPIECE 9
word and image, and this juxtaposition takes place in spite of Friedrich
Kittler’s assertion that ‘computers … are not designed for image-process-
ing at all’.27 Computers might not be designed for image processing, but
perhaps it is this very fact that has spurred on attempts to bring word
and image together in productive, and explicitly ‘illustrative’, ways. Take,
for example, the creation of Mosaic, the web browser that for the first
time enabled the integrated display of word and image and led to the
popularity of the World Wide Web, a software development that bears
more than a passing resemblance to wood engraving, which for the first
time allowed pictures and letterpress to be printed simultaneously on the
same page and led to the burgeoning of illustration.28
The emergence of interface studies and its focus on the configura-
tion of the GUI (graphical user interface) and WIMP interface (win-
dows, icons, menus and pointers) may offer opportunities for exploring
the illustrative dialectic that is ever present on our screens.29 Johanna
Drucker has denaturalised the workings of the interface by stressing
its graphic, and therefore visual, nature: ‘We tend to think of graphi-
cal interface as the screen display, a portal into the online world with
menu bars, buttons, and icons to manipulate. As a result, we ignore its
graphicality, its constructedness, the very features that support its opera-
tions and make it work’.30 But the interface is not ‘purely’ graphical, if,
indeed, there is ever such a thing. These graphical features have a textual
component. Interfaces are a space where word and image are in constant
dialogue. This can be seen in our everyday encounters with the digital
where we come across pictorial capital letters, like Apple’s ‘A’ for its ‘App
Store’, which is made up of a pencil, paintbrush and ruler, and Google’s
ever-changing logo that makes pictures out of, around and between the
letters. Website menus consist of thumbnail images with text under-
neath, while websites themselves are combinations of words and images.
In social media, there are illustrated tweets, tagged images and, in the
case of platforms like Tumblr, pictures with captions. The blogging con-
vention of heading a post with an illustration comes straight out of the
Victorian illustrated press. Even with other developments—audio and
moving components, the tactile experience of touch screens—the appear-
ance of the digital, what we see, is a combination of text and image,
an Illustrated User Interface. This is not just a matter of appearances.
It is in this combination of word and image that the interface generates
meanings, meanings that are bound up in cultural values and ideologies.
What does it signify that an icon for an app store is composed of the
1 FRONTISPIECE 11
tools of writing, painting and geometry, or that a blog post about lib-
erty is preceded with a picture of an American flag? An understanding of
the significance and workings of historical illustrations might help us to
understand and interpret this illustrated digital environment.
In 1999, Christian Vandendorpe made the astute prediction that hyper-
media would lead to a dismantling of the hierarchy that privileges text
over image and give rise to more interactions between the two forms:
Notes
1. Eduardo Urbina, ‘Visual Knowledge: Textual Iconography of the
Quixote’, Don Quixote Illustrated: Textual Images and Visual Readings/
Iconografía del Quijote, ed. Eduardo Urbina and Jesús G. Maestro,
Biblioteca Cervantes 2 (Pontevedra: Mirabel Editorial, 2005), pp. 15–38.
Available online at http://cervantes.tamu.edu/V2/CPI/iconography/
publ.html.
2. Jerome McGann, Radiant Textuality: Literature after the World Wide Web
(New York and Basingstoke: Palgrave, 2001), pp. 63–67.
3. See, for example, Andrew Stauffer’s introduction and the accompany-
ing forum, ‘Evidence and Interpretation in the Digital Age: Searching
Engines, Reading Machines’, Victorian Studies, 54:1 (2011): 63–68.
4. McGann, Radiant Textuality, p. 173.
1 FRONTISPIECE 13
Victorian Studies. There has also been a noticeable rise in the number
of special journal issues devoted to illustration, such as the ‘Illustration
and Gender’ issue of Nineteenth-Century Gender Studies (Summer 2015)
and the ‘Picturing the Eighteenth-Century Novel’ issue of the Journal for
Eighteenth-Century Studies (December 2016).
20. Sandro Jung (ed.), British Literature and Print Culture (Cambridge: D.S.
Brewer, 2013), pp. 1–2.
21. Christina Ionescu (ed.), Book Illustration in the Long Eighteenth Century:
Reconfiguring the Visual Periphery of the Text (Cambridge: Cambridge
Scholars, 2011), pp. 1–2.
22. Daniel Allington, Sarah Brouillette and David Golumbia, ‘Neoliberal
Tools (and Archives): A Political History of Digital Humanities’, Los
Angeles Review of Books, 1 May 2016.
23. For a discussion of how recent work in digital humanities is turning away
from a preoccupation with process and instrumentalism to an engage-
ment with the theoretical, see Anthony Mandal, ‘Digital Humanities’,
The Year’s Work in Critical and Cultural Theory, ed. Neil Badmington
and David Tucker, vol. 24 (Oxford: Oxford University Press, 2016), pp.
317–342.
24. Jay David Bolter, Writing Space: The Computer, Hypertext, and the History
of Writing (New Jersey and London: Lawrence Erlbaum Associates,
1991), p. 72.
25. See, for example, Lev Manovich, who traces the intersections between cin-
ema and new media in The Language of New Media (Cambridge, MA:
MIT Press, 2001).
26. Mary E. Hocks and Michelle R. Kendrick, ‘Introduction: Eloquent
Images’, Eloquent Images: Word and Image in the Age of New Media, ed.
Mary E. Hocks and Michelle R. Kendrick (Cambridge, MA: MIT Press,
2003), pp. 1–16, p. 1.
27. Friedrich Kittler, Optical Media: Berlin Lectures 1999, trans. Anthony
Enns (Cambridge: Polity, 2010), p. 226.
28. I am grateful to Michael Goodman for drawing this to my attention.
29. For an intervention in studies of the interface, see Alexander R. Galloway’s
political analysis in Alexander R. Galloway, The Interface Effect (Cambridge:
Polity, 2012).
30. Johanna Drucker, Graphesis: Visual Forms of Knowledge Production
(Cambridge, MA: metaLABprojects and Harvard University Press,
2014), p. 138.
31. Christian Vandendorpe, From Papyrus to Hypertext: Toward the Universal
Digital Library, trans. Phyllis Aronoff and Howard Scott (1999; Urbana:
University of Illinois Press, 2009), p. 96.
CHAPTER 2
(In)visibility
t Fig. 2.1 Titlepage for Sir Gilbert Edward Campbell, Bart., Mysteries of the
Unseen; or, Supernatural Stories of English Life (London: Ward, Lock and Co.,
1889). Available on The Illustration Archive: http://illustrationarchive.cardiff.
ac.uk
of the most enigmatic forms of representation, and the one that seems
most resistant to analysis. Part of the reason for this murkiness is that
illustration has conventionally fallen between generic categories, seen
neither as text nor as image. Writing in 1928, Forrest Reid, one of the
first collectors of mid-Victorian illustration, recognised the problematic
status of illustration: it ‘has come to be regarded’, he states, ‘as a dubious
mixture of art and something that is not art’.15 In some ways, it is also
the very ubiquity of illustration in the nineteenth century, with which
Reid, as a collector, had to contend, that has made its analysis so pro-
hibitive. Gleeson White, another early collector, described his task as a
‘magnificent’ one in which the piles of manuscripts at his side ‘prove the
impracticability of the enterprise’.16 Illustrations are simultaneously invis-
ible and omnipresent: forgotten today, yet everywhere in nineteenth-
century culture, to the extent that there are too many to examine with
any degree of thoroughness. Gérard Genette acknowledges this in his
account of the paratext, where he deals only briefly with illustration
because the field is too large. Illustration, in his words, is an ‘immense
continent’.17 One of the main difficulties with illustration, however, is
that it is not a continent at all: it does not occupy a distinct, continuous
space. On the contrary, illustration is dangerously incontinent: an unruly
and hybrid form that refuses to be fixed.
Critics have tried to find their way around this ‘immense continent’
using the foundational map of the archive. Reid and White were com-
pelled to create their own archives by cutting the illustrations from mid-
nineteenth-century publications and using these archives as the bedrock
for their research.18 White designs an artist-centred catalogue, moving
diligently through the images that appeared in illustrated weekly papers,
some illustrated books and the key periodicals of the 1860s, including
Once a Week, the Cornhill Magazine, Good Words and London Society,
his guiding principle being to detail the work of ‘every artist of the first
rank’.19 Reid takes a similar approach, creating a ‘survey’ of the illustra-
tions of 58 British artists working in the 1860s, which is arranged by
artist and school, including ‘The Pre-Raphaelite Group’ and ‘The Idyllic
School’.20
As the examples of these early collectors suggest, the content and
organising principles of the physical archive go hand in hand with the
scholarship on illustration. The same can be said of illustration studies,
which, out of necessity, have been governed by the images that are made
available and easily identifiable in the physical archive. It is unsurprising,
2 (IN)VISIBILITY 21
or, rather, ‘photographs’. As the homepage states: ‘The key goal of The
Commons is to share hidden treasures from the world’s public photog-
raphy archives. Please help make the photographs you enjoy more dis-
coverable by adding tags and leaving comments. Your contributions and
knowledge make these photos even richer.’25 On a day that I happen to
be browsing, the ‘photographs’ that are displayed on the homepage are
the book illustrations from the British Library’s dataset. Some are steel
engravings, some are etchings, others are wood engravings, and they are
all, of course, digital images; but none are ‘photographs’.
I am not simply being pedantic. The dilution of generic difference on
Flickr might not affect the average oil painting, but it does have reper-
cussions for historical illustrations, which are already marginalised. While
a general hosting platform like Flickr sidesteps the specificity of illustra-
tion, specialist digital illustration resources have directly to confront and
negotiate the issue of what constitutes the visual difference of an illus-
tration. In a sense, there is no more to illustration than meets the eye.
Illustrations are inescapably visual, a visuality that stands out in the nine-
teenth-century book where the pictures are often printed on thicker, yel-
low paper, with tissue interleaves separating them from the text, and in
landscape format, which means that the book has to be turned around
to view them. Indeed, Leah Price’s contention that ‘The Victorians
cathected the text in proportion as they disowned the book’ goes some
way towards explaining the deep-rooted suspicion of illustration that
emerged alongside its proliferation because the presence of illustrations
inevitably draws attention to the book as a book.26 There are other visual
constituents that define illustrations in their printed forms: the features
of the method by which they are reproduced (the scratched lines of an
etching, the cross hatching of an engraving, the porous texture of a lith-
ograph); and the size of these images (their conformity to the dimen-
sions of a page, a plate, a woodblock, although blocks were often bolted
together to make a larger image).
But, while there might be a specific ‘look’ to an illustration that dis-
tinguishes it from, say, a painting, its visual identity is never fixed or
static. On the contrary, an illustration has multiple visual incarnations:
it is the artist’s design (which is lost when the image is drawn or pinned
onto the woodblock), the engraved woodblock or etched/engraved
plate (which could go through several modifications and corrections),
the proof engraving on India paper, and the stereotype or electrotype
made from the block. Even in its printed form, the visual specificity of
24 J. THOMAS
Notes
1. Richard Maxwell argues that there are three main factors that explain the
burgeoning of illustration in these years: experimentation and innovation
in book layout and typography, which led to mass-market publishing; the
rise of the notion of British literature as an institution; and a subliminal
resistance to the presence of images in books that actually spurred on
invention. Richard Maxwell (ed.), introduction, The Victorian Illustrated
Book (Virginia: University of Virginia Press, 2002), pp. xxi–xxx.
2. [Catherine Gore], ‘The New Art of Printing. By a Designing Devil’,
Blackwood’s Edinburgh Magazine, 55:339 (January 1844): 45–49, p. 47.
3. The illustrations are not present in the edition from which I am quoting.
Anthony Trollope, Orley Farm, ed. David Skilton (1861–1862; Oxford:
Oxford University Press, 1985), 2 vols., vol. ii, p. 230.
4. William Makepeace Thackeray, Vanity Fair. A Novel Without a Hero
(London: Bradbury and Evans, 1848), p. 56.
5. For a discussion of the exclusion of the illustrations from the Penguin
Classics edition of Vanity Fair (2001), see Peter L. Shillingsburg,
‘Practical Editions of Literary Texts’, Variants 4: The Book as Artefact:
Text and Border, ed. Anne Mette Hansen, Roger Lüdeke, Wolfgang
Streit, Cristina Urchuaguía and Peter Shillingsburg (Amsterdam and New
York: Rodopi, 2005), pp. 29–55, pp. 36–40.
6. Paul Goldman, ‘Defining Illustration Studies’, Reading Victorian
Illustration, 1855–1875: Spoils of the Lumber Room, ed. Paul Goldman
and Simon Cooke (Aldershot: Ashgate, 2012), pp. 13–32, pp. 17–18.
30 J. THOMAS
Searchability
Fig. 3.1 John Everett Millais, illustration for Harriet Martineau, The Anglers of
the Dove, Once a Week 7 (16 August 1862) p. 197. Engraved by Joseph Swain.
Available on the Database of Mid-Victorian Illustration: www.dmvi.org.uk
situates these images within their historical context, but also has implica-
tions for the relation between word and image that illustration puts into
play. This relation is one that defines illustration, generates its meanings,
and determines how it is viewed, yet it has received scant attention from
critics. Lorraine Janzen Kooistra, one of the few scholars to examine the
interaction between word and image in the context of illustration of the
1890s, remarks that ‘Although most studies of illustrated books pay lip
service to the significant interactions of image and text, few critics have
gone further to investigate just how these interactions work, or to theo-
rize visual/verbal relationships’.1 Kooistra proposes a notion of ‘bitextual-
ity’ that accounts for the dialogue between word and image through an
engagement with the sexualised discourse in which illustration is described.
As Kooistra’s definition implies, the relation between word and image
in illustration is bound up in power relations. I have also discussed this
idea elsewhere, arguing that the interaction between text and picture in
illustration is frequently articulated in hegemonic terms, with the picture
positioned as subordinate to an authoritative text.2 A digital environment
is not free from these hierarchies. On the contrary, they are embedded in
the act of digitisation itself. When Microsoft set out to scan the books in
the British Library that form the basis of The Illustration Archive, it was
the words that it wanted to capture. The illustrations just happened to be
there. The identification and isolation of the illustrations in the digitised
and OCRed (Optical Character Recognition) pages were enabled not so
much because of their visual difference but because of their negative rela-
tion to the text: they were recognised as images because they were not
words. The only metadata recorded for the illustrations were the dimen-
sions of the page scan in pixels and the page number on which the image
appears as counted from the front cover (not conforming to the actual
pagination of the books).
Within this signifying system of OCRs and scans, it is the text that is
the primary signifier: it is the sought after product, the positive term by
which the illustration is defined. In this sense, a digital environment seems
to support conventional ideas of illustration that position the words as
superior to the images. As Kooistra puts it, ‘Since the image has no indi-
vidual significance, but only gains meaning in relation to the written word,
illustration can be ignored or omitted without loss’.3 But the ‘loss’ of the
illustration is never complete. The text may be privileged in a digital con-
text, but it is a text that is always already invaded by the visual. After all,
the very act of digitising the pages turns the words into images. Moreover,
36 J. THOMAS
the dataset are all visually similar: the birds are positioned in side pro-
file, dominate the picture space, and are the central focus of the image.
They conform exactly to the Google ‘idea’ of a bird. While it is likely
that this tool would retrieve an illustration from an ornithological work,
it is unlikely that it would find a picture of a dead pheasant flung over
someone’s shoulder or doves flying into a dovecote, both of which fea-
ture in one of Millais’ illustrations to Trollope’s Framley Parsonage.12
Likewise, although the Instance Search tool is effective at searching for
more or less identical images, what is much more difficult computation-
ally is finding ‘sort of similar but not the same’ image content, a problem
that is exacerbated in the dataset of illustrations used in The Illustration
Archive because of the scale and variety of the images.13 Even recent
developments in computer vision such as ‘deep learning’, which attempts
to mimic the brain by using machine-learning algorithms to model high-
level abstractions in data, are heavily dependent on using similar types
and formats of images for analysis.14
A bird is not always a bird, at least to a computer. As we discovered
in our, admittedly more modest, experiments with image similarity algo-
rithms, the features of birds can resemble anything, from decorative
coats of arms to Victorian gentlemen’s facial hair. The Kestrel and the
Beard Problem, as I have come to think of it, suggests that the possibili-
ties of computer vision are tantalisingly close, but, at the moment, tech-
niques for making image content searchable need to be supplemented by
other manual processes in order to be effective across a range of images.
Mario Klingemann, an artist and practitioner of ‘computational craft’,
has adopted just such a mixed approach in his clustering of thousands
of illustrations from the British Library’s dataset. Klingemann’s work is
on display in The Illustration Archive, where typing almost any search
term is likely to return one of his collections, which include illustrations
of skulls, prams, shoes, maps (almost 40,000 of them), women with
books, artists, painters, nudity and pictorial letters (arranged by letter).
The brilliance of Klingemann’s identification of the content of the illus-
trations lies not in a reliance on a specific method, but on a combination
of approaches. He employs multiple data clustering algorithms alongside
a manual clustering of image training sets organised using human vision.
The most prevalent method for making image content searchable,
though, does not so much involve ‘vision’ as text. It is the spectral
presence of words that haunt these various techniques, from the ‘Bag
of Words’ model, and the textual metadata that underpins the pictures
3 SEARCHABILITY 39
• Take the British Library’s designated title for a book: The Sports of
the Genii. [Etchings from drawings by Miss Susan Macdonald. With
verses by A. Home]
• Split this title into words, separated by spaces
• Look at each word in turn, checking for the occurrence of the
word-part: ‘draw’
• Count how many words into the title this word is
• Join the words before and after this word into two strings: The
Sports of the Genii. [Etchings from … by Miss Susan Macdonald. With
verses by A. Home]
• If the second string, which comes after the word-part ‘draw’, con-
tains the word ‘by’, continue
• Find the ‘by’ word in the second string (here it is the first word, but
this is not always the case)
• If the second string contains the word ‘author’, make a leap of faith
that it is the author who also designed the drawings and identify the
author’s name from the book’s metadata
• ‘Clean’ the data, removing any parentheses: Miss Susan Macdonald.
With verses by A. Home
• Take the start of the string up until the first word that is longer than
four characters (and so likely to be a name rather than a title) and
that ends in a full stop: ‘Miss Susan Macdonald’
• Remove any punctuation that may have been dragged along
• And this is the Eureka moment: the illustrator is ‘Miss Susan
Macdonald’
Don Quixote that is used to link the illustrations to the text, enabling
a search for illustrations such as ‘DQ knighted at inn by innkeeper’.
The use of this taxonomy to retrieve illustrations gives a fascinating insight
into the history of these images, indicating what scenes are most often
illustrated and what scenes are not (there are 250 of this scene alone
whereas there are no illustrations of ‘Night encounter with Maritornes’.
The taxonomy can also indicate at what historical point the illustration of
a particular scene becomes popular, how the images evolve, change and
differ from each other, and instances of their reuse and modification in
various editions (the presence of numerous illustrations in this archive that
are mirror images of each other suggests that these illustrations were cop-
ied from earlier woodblocks and reprinted). This search mechanism is an
effective way of connecting the illustrations to the text of Don Quixote,
but such an extraordinarily detailed taxonomy of episodes can only be
created for a single or small number of texts; it could not be scaled up
to dozens or hundreds of texts. Another limitation lies in the fact that,
because the taxonomy is based on episodes, adventures and scenes
from Don Quixote, it does not specify image content other than what is
included in the description of the episode. We can infer that there is prob-
ably an inn and an innkeeper in the illustration of ‘DQ knighted at inn
by innkeeper’, but the presence of other people, objects and animals is
detailed in a supplementary keyword list.
Although the taxonomy used in the Cervantes Project allows the
text and the illustrations to be brought together, it stands indepen-
dently from the text, as another layer of metadata. But there is always,
of course, the text ‘proper’. If the words of the text coincide with the
content of the illustrations, then the issue of making the images search-
able could be resolved. In the case of The Illustration Archive, all 68,000
volumes could potentially be mined to search for relevant images. This
idea that the text ‘maps’ onto the content of illustrations was fundamen-
tal to a recent project by Kalev Leetaru that used the accompanying text
to make searchable millions of illustrations from books in the Internet
Archive. The software that Leetaru developed extracted the caption from
each image and the paragraphs before and after the illustration in the
book and made this information searchable along with the bibliographic
metadata. When users type in search terms, the system interrogates these
texts in order to find relevant images.21
The searchability of the illustrations in Leetaru’s project, however,
relies on assumptions about how illustrations signify that are highly
3 SEARCHABILITY 47
Fig. 3.3 Detail from Myles Birket Foster, illustration for Robert Bloomfield,
‘The Farmer’s Boy’, Favourite English Poems (London: Sampson Low, 1862)
p. 174. Engraved by Edmund Evans. Available on the Database of Mid-Victorian
Illustration: www.dmvi.org.uk
50 J. THOMAS
Birket Foster, along with this incongruous detail, suggests that this could
be an illustration that has previously been used to illustrate a different
text. Certainly, Birket Foster was well known for mass-producing hun-
dreds of stock ‘rural’ illustrations that could potentially illustrate any
number of texts that happened to be set in the countryside. The dis-
crepancy between turnip and hay would have repercussions in a digi-
tal resource where the text is being used to search for images: if a user
searches for pictures containing ‘turnips’, the image returned via the text
would be wrong; and if the user searches for illustrations showing ‘hay’,
the picture would not be retrieved because the word ‘hay’ does not
appear in the text. In a similar, though more ethereal, example from ‘The
Rime of the Ancient Mariner’ in Favourite English Poems, a supernatural
figure appears in the illustration when the mariner shoots the albatross.
This ghost-like being would remain forever intangible if the user is reli-
ant on the words of the text to retrieve the images.26
These illustrations, which are by no means extraordinary, undermine
the notion of equivalence between word and image, an equivalence that
is necessary if the surrounding words are used to search for and retrieve
the content of pictures. Even in those cases where the text and image
do coincide in some way, the relevant text is not necessarily located in
proximity to the illustration. Although the actions and events in the text
might be physically and thematically ‘close’ to the actions and events
depicted in an illustration, the textual description of the setting for those
actions might be given pages earlier. ‘Peregrine’s Eloquence’, an illus-
tration from Trollope’s Orley Farm, appears facing page 289 in the first
book edition of the novel while the textual passage it illustrates is a short
section at the end of page 295 that spills over onto page 296 (this is the
only textual moment where the protagonist stands with his back against
the fire, the position in which he is pictured in the illustration). However,
the caption of the illustration, ‘Peregrine’s Eloquence’ relates to an ear-
lier moment on page 295, and the description of the room in which
the scene takes place is described a couple of pages before that, on page
293.27 In this example, the caption would be of little help in retrieving
the content of the illustration, unless the user was searching for pictures
of ‘Peregrine’ (in which case, numerous illustrations of falcons would be
returned, too!). ‘Peregrine’s Eloquence’ demonstrates that, in the case of
literary illustrations, the subject matter of an image can be taken from
multiple parts of the text, and these parts do not always surround the pic-
ture. In periodical publications and monthly instalments (Orley Farm was
3 SEARCHABILITY 51
first published in monthly shilling parts), it was common for the illustra-
tions to appear at the front of the instalment, and in one such example
from London Society, a frontispiece illustration is separated from the poem
it illustrates by 65 pages.28 In cases like this, the caption of the illustra-
tion often points the reader in the direction of the text (here it states
‘see p. 162’), a device that presents its own problems if the caption is
employed as a simple descriptor of the image that can ‘find’ iconographic
content.
What all these examples suggest is that the relation between word
and image in illustration is a highly complex one. In terms of making
the content of illustrations searchable online, the textual information
surrounding the picture—whether this is the bibliographic metadata,
the caption or the words—is unreliable. The best option, perhaps, is to
let the illustrations speak for themselves. This, in theory at least, is the
approach adopted in what has become the most prevalent method for
making the content of images searchable online: keywording or tagging,
a process in which the images are looked at and described in words, and
users are able to find relevant pictures from this textual markup. Tagging
has been employed in a variety of digital contexts, from individual files
such as images and video to websites and collections of connected
resources. Search engines like Google rely heavily on tags, which are
often automatically retrieved from formalised data structures. As Mayer-
Schönberger and Cukier put it, ‘Tagging has emerged as the de facto
standard for content classification on the Internet’.29
In the case of illustration resources, tagging also has implications for
the relationship between word and image that defines illustration. As a
mechanism for describing the visual content of a picture in words, tag-
ging introduces another layer of text that comes to characterise illustra-
tion. In this sense, the marking up of image content could itself be called
an ‘illustrative’ act. Indeed, the activity of tagging illustrations, of add
ing words to pictures, is reminiscent of the ways in which nineteenth-
century texts were often ‘written to’ images.30 Traces of this prac-
tice, which was relatively commonplace in the case of albums and gift
books, can be found in the titles of books: Birket Foster’s Pictures of
English Landscape … with pictures in words by Tom Taylor (1862); Death’s
Doings; consisting of numerous original compositions, in prose and verse,
the … contributions of various writers; principally intended as illustra-
tions of 24 plates designed and etched by R. Dagley (1826). These titles
position the prose and verse compositions as ‘pictures’ or ‘illustrations’,
52 J. THOMAS
includes thousands of prints and book illustrations from the early mod-
ern period that have been marked up by hand and arranged in a hierar-
chical structure by topic. Although the terms of British Printed Images
are different to DMVI’s, reflecting the historical differences between the
two datasets, DMVI uses a similar tree-like structure that allows users
to drill down and across categories. The keywording system developed
on DMVI has since been adapted for use in The Yellow Nineties Online,
Illustrating Scott and Nineteenth-Century Serials Edition (NCSE)
(http://www.ncse.ac.uk).35
Keywording an illustration is not as straightforward or transparent as
it might seem, however. One problem that emerged during the develop-
ment of the keywording system for DMVI was the extent to which we
could, or should, look into the text to mark up the content of the illus-
tration. In some cases, this recourse to the text was difficult for practical
reasons (the illustrations from Aberystwyth and the Ashmolean museum
had already been extracted from the periodicals), but there was also a
critical issue at stake in using the text to describe the image. The activ-
ity of keywording those illustrations that had been divorced from the
text and then going back to the texts in order to shape and modify the
tags emphasised the complexities of the interaction between word and
image. To use Roland Barthes’ term, the words can ‘anchor’ the illus-
tration, steering an interpretation of the image or suggesting new and
different meanings that the image does not have independently.36 For
example, the text might specify a familial relationship between figures
that is not obvious from the picture alone: figures marked up as ‘boy’
and ‘girl’ become ‘brother’ and ‘sister’. Equally, the text can clarify what
is happening in an illustration or can fill in the details rendered by a few
scratched lines. When the text is read alongside the picture, an indistinct
snowy landscape turns into ‘Canada’. This textual ‘anchoring’ also means
that illustrations often marginalise or exclude details that are promi-
nent in the words (an illustration does not need to visually demarcate or
define a feature that is a textual given). In this sense, illustrations interact
with texts as much in what they do not show as what they do.
This is not a one-way ‘anchorage’, though: the illustration also
‘anchors’ the meanings of the words and determines how they are read.
A simple ‘drawing room’ in the text becomes furnished in the illustra-
tion, and these pictorial details generate their own meanings (about class,
gender and domesticity). An illustration can define what someone looks
like: the iconic figure of Sherlock Holmes, deerstalker cap and all, was
54 J. THOMAS
Du moment qu’un type est fixé par le crayon, il perd ce caractère de géné-
ralité, cette concordance avec mille objets connus qui font dire au lect-
eur: “J’ai vu cela” ou “Cela doit être”. Une femme dessinée ressemble à
une femme, voilà tout. L’idée est dès lors fermée, complète, et toutes les
phrases sont inutiles, tandis qu’une femme écrite fait rêver à mille femmes.
Donc, ceci étant une question d’esthétique, je refuse formellement toute
espèce d’illustration.38
tag perched on a man’s bowler hat, a ‘duck’ tag impeded the progress of
a duck as it glided across a lake, and a young child could be forgiven for
looking rather red in the face as a ‘People>>Children’ tag attached itself
to the side of her head. The relationship between text and image here
was presented as an obvious and straightforward one. ‘All we ask’, Your
Paintings encouraged its potential taggers, ‘is that you look carefully at
the art and tell us what you see’.40
But ‘telling what you see’ is not a simple task, and this was exposed
in the tagging demonstration pages of the site that included sugges-
tions on how the images had been tagged ‘by the system’. ‘There is no
100% right answer to this’, the user was informed. ‘However, in this case
we would have chosen the following tags …’. Ironically, it is the very
requirement for this statement and the inclusion of the ‘recommended’
tags that reveals the fissure between words and images, the fact that they
are not the same, even though the act of tagging depends on the illusion
of their equivalence. As the ‘system’ and user-generated keywords appear
on the screen beside the image, this fissure becomes even more apparent.
Tags never fully succeed in labelling and identifying aspects of the pic-
ture. The luggage is always already lost.
The gap between word and image is as much the result of the visual
specificity of the image as of the instability of words. Gotthold Lessing
famously argued that images are spatial rather than temporal: they
exist as stationary objects in space rather than objects that progress in
time.41 A text has a discreet beginning and end according to the logic
of the sentence; images, on the other hand, have no such logic and this
becomes only too apparent in the ‘blind’ keywording with no prompts
from the system that is encouraged by Flickr and on other social media
sites. When it comes to keywording a picture, where does the tagger
even begin? The tagging structures on sites like Your Paintings impose
a temporality, a textual logic, on the image in the form of questions
about the image that the tagger is asked to answer. But even within the
constraints of these questions, there is no natural or obvious order in
which the features of an image should be tagged. Nor is there a predict-
able order in which these features are actually seen. This unpredictability
was the focus of First Impressions, a crowdsourced tagging project run
by Indianapolis Museum of Art, where users were asked to click the first
thing that caught their eye in a series of artworks. The results showed
some variation in what the visitors saw ‘first’, especially in those composi-
tions where single elements were not placed in the middle foreground.
56 J. THOMAS
Commenting on the results, the museum blog states that, ‘Through this
experiment it seems that individuals are viewing art however they want
to view art, which, is exactly correct’.42 Or not correct. What this experi-
ment indicates is that images are not viewed in the same way by all spec-
tators and the order in which things are seen is not self-evident, and this
is assuming that the activities of seeing and clicking/tagging are as simul-
taneous and as instantaneous as is implied here.
Some aspects of an image might not be seen at all. Keywording raises
the question of what is visible and invisible in an image. Paradoxically,
while digital image processing privileges the visuality of illustration (as
argued in Chap. 2), this is precisely the aspect of the image that tends to
be marginalised in keywording systems. The colour illustrations in The
Illustration Archive, for example, are rarely tagged as ‘colour’. Then,
there is the associated issue of what should or should not be marked up
in a picture. Are the keywords ‘man’ or ‘woman’ sufficient, or should
features and clothing also be described (a test tagger for The Illustration
Archive tagged ‘nose’, ‘mouth’ and ‘eyes’ for every portrait, a level of
detail that may seem excessive, but is by no means incorrect). And what
of the relationship between objects? This relationship cannot be articu-
lated in single keywords: the tags ‘woman’, ‘letter’, ‘cat’ and ‘sleep-
ing’ do not express how these features interact in an image. The Yellow
Nineties Online attempts to counteract this limitation by including a
prose description of each image alongside keywords. This combination of
approaches means that the illustrations are searchable using both sets of
markup (the keywords and the descriptions), with the prose markup hav-
ing the advantage, as The Yellow Nineties Online asserts, of describing the
image for the visually impaired.43
When employed as the sole mechanism for describing and searching
the content of images, however, prose descriptions can be as problematic
as keywords. The more ‘spontaneous’ nature of representing an image
in sentences rather than adhering to the rigid categories of a keywording
schema means that aspects of an image might be left out and the image
would not be retrieved if the end user were searching for the excluded
term.44 Prose descriptions, like keywords, give only a partial and frag-
mentary account of the visual impact of the picture. The illusion of
the equivalence of word and image on which both keywords and prose
descriptions depend is only ever an illusion: words do not seamlessly map
onto pictures. Indeed, in many ways, the activity of tagging is not actu-
ally one of ‘mapping’ at all, but of interpretation.
3 SEARCHABILITY 57
were there: an allusion from Exodus to the pillar of light that helps the
Israelites, which is depicted in an illustration by Myles Birket Foster47; a
toy ark in a picture by Arthur Boyd Houghton for Good Words (Fig. 3.4).
The keywords here are not merely adjuncts to the image, but consti-
tute its meanings, influencing how the illustration is viewed. The map-
ping of the Iconclass codes onto DMVI’s keywording system meant that
‘interpretations’ of the same image could be seen side by side, and, by
analysing these interpretations together, their cultural meanings became
apparent. The marking up of an image is always a culturally inflected
activity, and this takes on some political significance when decisions
have to be made about how illustrations should be keyworded. How,
for example, should a tagger describe an illustration of a fictional fig-
ure who is frequently described in the text in terms that would now be
regarded as racist? Should the keyworder use the racist term in fidelity
to the text, or omit it on the basis that it occurs in the text alone (even
though the illustration might include the stereotypical traits associated
with this signifier)?48 Such issues point towards the complex dynamics of
word and image in illustration, suggesting the ways in which keywording
exposes the values of previous historical moments as well as the ideologi-
cal assumptions of our own.
The methods for making the content of illustrations searchable online,
whether computer vision, the use of textual metadata or the activity of
keywording, are not simply ‘tools’ that are applied to the illustrations
from the outside. Rather, they are bound up in the relation between
word and image that defines illustration. The development of digital
illustration resources involves a negotiation of how illustration signifies,
raising the question of the extent to which an illustration can be said to
‘reflect’ or re-produce the text. Illustrations are distinctly visual entities:
they have their own modes of reproduction, their own tropes, allusions,
and traditions, but, like the digital methods that promise to make them
searchable, their visuality is far from unmediated. In illustrations, words
are never completely out of the picture.
Notes
1. Lorraine Janzen Kooistra, The Artist as Critic: Bitextuality in Fin-de-
Siècle Illustrated Books (Aldershot: Scolar, 1995), p. 3.
2. Julia Thomas, Pictorial Victorians: The Inscription of Values in Word and
Image (Athens, OH: Ohio University Press, 2004), p. 15, passim.
60 J. THOMAS
21. For an account of this project, see Kalev Leetaru, ‘500 Years of Book
Images’, http://kalevleetaru.com.
22. Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution
that Will Transform How We Live, Work and Think (London: John
Murray, 2013), p. 13.
23. This project ran in 2008 and was funded by the Cardiff Undergraduate
Research Opportunities Programme. The team consisted of two final year
English Literature undergraduate students, Marianne Fisher and Simon
Eckstein.
24. For a discussion of the complexities of these illustrations, see Joan
Stevens, ‘Thackeray’s Pictorial Capitals’, Costerus: Essays in English and
American Language and Literature, 2 (1974): 113–140.
25. Robert Bloomfield, ‘The Farmer’s Boy’, Favourite English Poems
(London: Sampson Low, 1862), pp. 116–193, p. 175.
26. Edward Henry Wehnert, illustration for Samuel Taylor Coleridge, ‘The
Rime of the Ancient Mariner’, Favourite English Poems, p. 223. Available
on the Database of Mid-Victorian Wood-Engraved Illustration. www.
dmvi.org.uk.
27. These page references are for the first book edition of Anthony Trollope’s
Orley Farm, illustrated by John Everett Millais and engraved by the
Dalziel Brothers (London: Chapman and Hall, 1861–1862).
28. This is John Dawson Watson’s illustration, ‘A Summer’s Eve in a Country
Lane’ (p. 97), which illustrates a poem by T. H. (pp. 162–163) in London
Society, 2 (August 1862).
29. Mayer-Schönberger and Cukier, Big Data, p. 43.
30. In the case of their fine art books, the Dalziel brothers would commis-
sion a set of pictures from an artist and hire poets to write a page of
verse to accompany each plate. According to Kooistra, this led to ‘a
new kind of ekphrastic poetry for the gift-book market.’ See Lorraine
Janzen Kooistra, Poetry, Pictures, and Popular Publishing: The Illustrated
Gift Book and Victorian Visual Culture 1855–1875 (Athens, OH: Ohio
University Press, 2011), pp. 102–103.
31. Murtha Baca notes the ‘art information community’s very real need for
practical guidelines on how to lead end users to relevant images of art
and architecture online.’ Murtha Baca (ed.), Introduction to Art Image
Access: Issues, Tools, Standards, Strategies (Los Angeles: Getty Research
Institute, 2002), p. v.
32. For more details on the process of keywording DMVI and the issues it
raised, see Julia Thomas, ‘Getting the Picture: Word and Image in the
Digital Archive’, European Journal of English Studies, 11: 2 (August
2007): 193–206.
3 SEARCHABILITY 63
44. This is sometimes the case in the Rossetti Archive, which depends on prose
descriptions to make the content of images searchable. For a discussion of
these issues, see Thomas, ‘Getting the Picture’, p. 196.
45. For more information on this project, see Etienne Posthumus, ‘DMVI
Iconclass Mapping’, Cardiff Book History blogpost, 22 February 2011,
https://cardiffbookhistory.wordpress.com/tag/arkyves-project/.
46. John Everett Millais, illustration for Anon. ‘Margaret Wilson’, Once a
Week 7 (5 July 1862), p. 42. Available online at www.dmvi.org.uk.
47. Myles Birket Foster, illustration for Thomas Campbell, ‘The Pleasures of
Hope’, Favourite English Poems, p. 280. Available online at www.dmvi.
org.uk.
48. This particular example came out of an email correspondence with Kate
Holterhoff, the editor of Visual Haggard, http://www.visualhaggard.
org, an archive of illustrations of Rider Haggard’s novels. As Holterhoff
points out, there may be an ‘ethics of access’ in making some of these
racially discomfiting illustrations accessible. Digital accessibility raises
the question of ‘whether digitization recapitulates an image’s nine-
teenth-century ideology, or whether the online image has become some-
thing altogether new.’ Kate Holterhoff, ‘Ethics and the Digital Archive:
The Case for Visualizing H. Rider Haggard’, Journal of Visual Culture
Online, 21 October 2013. Available online at http://blogs.tandf.co.uk/
jvc/2013/10/21/ethics-and-the-digital-archive-the-case-for-visualizing-
h-rider-haggard/.
CHAPTER 4
Crowdsourcing
this activity is articulated and described. The aim of this chapter, then, is to
move away from discussions of the history, logistics and benefits of crowd-
sourcing5 and instead to identify this activity as an inherently problematic
one. The discussion focuses on three intersecting areas, which, together,
reveal what is at stake in the new access to images offered by crowdsourced
image tagging: the construction of the tagger and tagging ‘community’ in
a crowdsourcing environment; the infrastructure of digital resources and
attempts to control the ‘messiness’ of taxonomies; and the definitions of
‘illustration’ that emerge from the development of crowdsourcing tasks.
The politics of crowdsourced image tagging is apparent in one of the
earliest and most successful image tagging initiatives: the mounting of
a corpus of images on Flickr by the Library of Congress and the launch
of ‘The Commons’, a designated area of Flickr where cultural heritage
institutions share photographs that are tagged by visitors to the site.6
Central to this project was the strategic objective to increase awareness
of the collection and ‘to expand outreach’, an objective that seems to
have taken priority over making the images searchable, which is why the
tags are ‘free’, that is, without what prompts requisite in crowdsourc-
ing projects that aim for increased image searchability.7 For Mia Ridge,
crowdsourcing is ‘a powerful platform for audience engagement with
museums, offering truly deep and valuable connection with cultural her-
itage through online collaboration around shared goals or resources’.8
Moreover, the audience that is being engaged here is not just the typical
demographic, with a concerted effort being made to reach those who
are not conventionally interested or invested in art. In the case of Your
Paintings, another crowdsourced image tagging project, one-fifth of tag-
gers reported that they had never, or very rarely, visited art galleries.9
There are, of course, any number of ‘external’ reasons why users
become engaged in such resources, and the collaborative and pleasurable
aspects of image tagging might be some of them. But I want to suggest
that at least part of the success (or otherwise) of crowdsourced image
tagging lies in the textual strategies of the resources themselves: the
ways in which they address, construct and enlist the users, or how they
‘play the crowd’. It is noticeable, for instance, that the activity of tag-
ging images is often articulated as way of giving taggers (an illusion of)
‘ownership’ of the works of art: these are ‘Your paintings’. This strategy
has historical precedents in the Victorian period, where numerous meth-
ods were employed to try and interest the working classes in the arts,
from the offer of cheap rail fares to places of cultural significance, to the
4 CROWDSOURCING 69
opening of galleries at times that fitted in with the working week. Art
was regarded as an ennobling force, able to imbue the working classes
with the values of the middle classes. A century on and John Berger was
lamenting the fact that the history of art was too elitist and appealing to
the public to reclaim it.10 While such ideas seem far removed from the
space of the digital archive, the terms of the debate are strikingly similar.
Crowdsourced image tagging plays a key role in outreach programmes
precisely because it is driven by the assumption that anyone—and every-
one—can tag a picture.11 Its attempts to address and engage the crowd
take place in the context of a concomitant attempt to break down barri-
ers, whether these barriers are intellectual, geographic, social, political or
cultural. Crowdsourced image tagging is defined as a democratic act.
Where crowdsourcing differs from Victorian attempts to engage the
public, however, is in its move away from the notion that cultural institu-
tions can benefit the individual to the idea that the individual can ben-
efit the institution. It is not what your gallery can do for you, but what
you can do for your gallery. As Ridge argues, ‘GLAM projects provide
an opportunity for altruistic acts’.12 This ‘opportunity’ is intensified in
cultural heritage crowdsourcing projects, which, according to Ridge, ‘are
well positioned for appeals to altruism’.13 In fact, several studies have
indicated that an emphasis on altruistic motivations in crowdsourcing
initiatives increases levels of participation.14 This debate has been taken
up by Trevor Owens, who raises the status of the ‘crowd’ by defining
them as ‘amateurs’ in the sense that they are part of the history of ‘con-
tributors to the public good’.15 Owens advocates a model of crowd-
sourcing that empowers a community of viewers, and where tagging is a
valued activity: ‘We are in a position to let the users of these collections
leave a mark on the collections. Instead of browsing through a collec-
tion they literally become the authors of our historical record’.16 What
Owens’ comments reveal is that the role of visitors to digital archives has
radically changed with the onset of crowdsourcing. Crowdsourcing pro-
vides an opportunity for viewers to become active partners rather than
passive recipients in their exchange with the image, to ‘write’ the picture
as they tag it, and this activity compromises the seemingly fixed distinc-
tion between the ‘expert’ and the ‘amateur’.
There is a fundamental sense, though, in which the appeal to altru-
ism that has come to characterise crowdsourcing projects is embedded
in the textual interfaces of these sites. Johanna Drucker has drawn atten-
tion to the fact that ‘An interface is a space in which a subject, not a user,
70 J. THOMAS
projects are ‘intimately linked with notions of community, and the sense
of community felt by participants’.23 Certainly, crowdsourced image tag-
ging projects often take their lead from the so-called ‘Citizen Science’
initiatives, which construct their anonymous users as nameable and
acknowledged ‘citizens’. Such a ‘community’ is organised and known,
and in a literal sense, too, since user profiles are frequently stored. Even
as they label pictures, taggers are themselves labelled.
The forging of a ‘community’ with a specific identity was central to
Brooklyn Museum’s Tag! You’re It! game that launched in 2008 with the
objective of keywording art objects in the collection. Taggers here were
described collectively as a ‘posse’, and suitably diverse members of this
posse demonstrated the tagging process in a short video set to music in
which they held up handwritten keywords. The altruistic appeal of this
project was inseparable from the creation of a community, objectives that
were built into the tagging game, where users played against each other:
‘Help us out by playing tag with other posse members. Your tags will
help everyone find objects in our collections’. At the end of the game,
users were told the names of the posse members with whom their tags
matched. Tag! You’re It! was as much about interacting with other posse
members as interacting with the art objects, even though this interac-
tion was illusory since the game was not conducted in real time and the
tags of other participants had already been collected and entered into the
system. The attempt to personalise the posse of taggers was accompanied
by an attempt to personify the museum, to give it a face, or faces. There
were ‘thank you’ videos in which museum professionals addressed the
taggers and told them how well they had done. A prolific tagger could
‘collect’ 12 such videos.
Tag! You’re It! addressed the tagger directly as both a member of
the posse and a ‘subject’. As such, it attempted to dismantle the barrier
between the tagger and the museum. The 12 videos were so appeal-
ing and collectible because of their informality: the tagger encountered
museum workers at the photocopier, on a bench eating lunch, or sat
at a paper-strewn desk. The implication was that these were real peo-
ple, just like the tagger, and that these curators truly valued the tagger’s
input. This informality was mirrored by the informality of the language
of the interface. There were no detailed instructions on how to tag or
any ‘Frequently Asked Questions’, but there was constant encourage-
ment throughout the tagging process: ‘Blast! You just passed’, ‘Work
it! You just passed’, ‘Chirp! You just passed’. The voice of the interface,
72 J. THOMAS
moreover, was a ventriloquised one that also spoke for the tagger: ‘What
else ya got?’, ‘Nah, skip this one’. This linguistic register emulates the
language of the street, of youth culture or an idea of youth culture. It
suggests an attempt to render art more ‘real’ and relevant, less ‘grown
up’. This language was in opposition to the rest of the website, which
adopted the more formal discourse typical of museums: ‘The Museum’s
collections were initially developed, in the early decades of the twenti-
eth century, by such outstanding curators as Stewart Culin, Herbert
Spinden, and William Henry Goodyear, with the generous sup-
port of collectors and donors from Brooklyn and around the country.
Continuing to build upon their pioneering work, the Brooklyn Museum
has amassed one of the largest and most diverse collections in the
United States’.24 It is hard to imagine Stewart Culin, Herbert Spinden
and William Henry Goodyear, the great curators of the past, giving the
thumbs up to prolific taggers.
In July 2014, Tag! You’re It! was decommissioned. A blogpost
by Shelley Bernstein, the Vice Director of Digital Engagement and
Technology at Brooklyn Museum, explained what had gone wrong.25
The problem seems not to have been the number of tags generated (over
230,000), but the fact that this was, indeed, an intimate ‘posse’: the tag-
ging community consisted of only 1100 taggers in the 6 years that the
game was running and the most consistently engaged taggers turned out
to be members of Brooklyn Museum staff. As Bernstein writes, ‘If our
institutional mission centers around community with the aim to engage
a broad audience, are the Posse and our tagging games doing that effec-
tively?’26 The answer, according to Bernstein, was a resounding ‘no’.
Tag! You’re It! was retired on the basis of the limited size and breadth
of the tagging community and the ineffectiveness of the gaming struc-
ture at engaging a wider user group (Bernstein admitted that more than
half of the tags were actually generated directly via the object page rather
than through the game).
Where Tag! You’re It! failed, Your Paintings seems to have succeeded,
having accumulated over six million tags inputted by 12,000 taggers
before the site was rebranded as Art UK in 2016. There was no gam-
ing incentive here, although there were different coloured paintbrushes
that indicated the number of paintings tagged by the user. In terms of
the motivational techniques that the site used to increase and sustain
participation, there was the familiar use of the language of altruism,
but here it came with an added dimension that may help to explain the
4 CROWDSOURCING 73
project’s impressive results. In Your Paintings, the focus was on the own-
ership and national identity of the user: these are ‘your paintings’, and
the ‘you’ to whom these images ‘belong’ was addressed specifically as a
British citizen. The appeal to altruism, then, was couched in nationalist
terms: ‘help tag the nation’s art!’ the site implored its visitors.27 Visitors
to Your Paintings were constructed, and actively engaged, as British
Citizens, who would be even better citizens if they tagged the images on
display. Paradoxically, however, at the very moment that Your Paintings
addressed its ‘British’ users, it also revealed the ambiguous notion of the
‘crowd’ or ‘community’ that underlies such tagging projects: whether it
is a posse or a nation, the ‘crowd’ is always exclusive as well as inclusive.
And just as the taggers are characterised and labelled by the digital
environment with which they interact, so are the tags themselves. The
steve.museum, a collaboration of museums and galleries that was estab-
lished to examine the possibilities of user-generated descriptions of art,
acknowledged that ‘the anarchy of emergent folksonomy seemed a cause
for concern’.28 Of particular ‘concern’ was the apparent gap between
user-generated tags and the descriptions of the images provided by
experts. It was estimated that up to 90% of the terms users contributed
were not present in the museum documentation for the art object.29 The
language employed in such accounts of crowdsourced tagging suggests
that the problematics of this activity extend to the terms in which the
tags themselves are described. Tags are never merely tags. They acquire
a political momentum: they can be good or bad, fixed or unruly, ‘eve-
ryday’ or expert, valuable or valueless, anarchic or democratic.30 The
Powerhouse Museum in Sydney, Australia, advocates the use of crowd-
sourcing to mark up its collections (users can delete as well as add key-
words) but the resultant tags are defined as informal and potentially
unreliable, their difference from the more formal tags stressed in their
visual incarnation as a word cloud.31 Such a tactic has parallels in archi-
tectural restoration projects that make visible the new additions to build-
ings so that they can never be mistaken for the original. Powerhouse
even provides a ‘warning’ to visitors: ‘Sometimes museums describe
objects in language that is highly specialist and user added keywords
are useful in bridging the “semantic gap” between the language of the
museum and that of the user…. Please note that this is an experimental
way of navigating the collection and that the “accuracy” of these key-
words is not verified by the Museum. We recommend their use primarily
as exploration and navigation tools’.32
74 J. THOMAS
this painting relate to any event? What subjects do you see in this paint-
ing? What type of painting is this? Very little is left to chance in terms of
how the tagger responds to these questions. The keywords entered for
the ‘things and ideas’ question were connected to definitions from the
Oxford English Dictionary and the tagger selects the most appropriate
term. Likewise, the subject categories for the painting (including ‘Home
and family’, ‘Ideas and Emotions’, ‘Religion and belief’) appeared as a
drill-down list, and the options for the ‘type’ of painting were also pre-
sented to the tagger in list form.
According to Dunn and Hedges, the editorial control in crowdsourc-
ing projects is ‘reassuring’ to users: ‘There also needs to be some ele-
ment of peer review/quality control to reassure the community that
their efforts create academically viable outputs’.45 Such control, how-
ever, exposes an implicit uneasiness about unruly tags embedded in
these systems, a suspicion that, instead of calling a spade a spade, taggers
might call it a shovel, or they might upset the entire system by labelling
it a hoe. The activity of tagging is reined in, so much so that, to some
extent, the resultant descriptions of the images derive as much from the
structures and methodologies of the tagging systems as from the tags
themselves. The architecture of crowdsourced image tagging quite liter-
ally provides its own added value.
In The Illustration Archive, targeted questions about the illustrations
function as cues to the user, adding a formal structure to the tags that,
in turn, optimises the searchability of the images. The questions asked of
the tagger are, therefore, crucial because they determine how the digital
archive can be interrogated and what information can be gleaned about
the illustrations. Initial attempts to adapt the tagging questions used in
Your Paintings for The Illustration Archive led to an awareness of the
very differences between these artistic genres. While a relatively open
question like ‘What things or ideas can you see in this painting?’ could
usefully be adapted for illustration, other questions are not so relevant.
The question ‘Can you name any people in this painting?’ (the examples
given on the site were those heavyweights of British history, Winston
Churchill, Florence Nightingale and Horatio Nelson) suggests the prom-
inence of portraiture; while the question, ‘Does this painting relate to an
event?’ (the examples given were the Battle of Trafalgar, the First World
War and the Coronation of Queen Elizabeth II) refers to the tradition of
history painting. While illustrations do, of course, include portraits and
represent historic events, these are not distinct genres that they are in
78 J. THOMAS
painting. Even an open question like ‘What subjects do you see in this
painting?’ raises questions about how illustration signifies. Sara Shatford
Layne offers useful advice to art cataloguers on how the ‘subjects’ of
works of art might be classified. As she acknowledges, subject access is
‘One of the most important means of enabling users to locate art images’
but it is also ‘a complex and sometimes messy process’.46 Identifying the
subject of a picture, it seems, can be as ‘messy’ in formal classifications
as it is in user-generated folksonomies. This is exacerbated in the case of
illustration because the ‘subject’ or content of the picture is not neces-
sarily contained within the frame of the image as it is in painting, but is
generated in the interaction between the image and the text.
Illustration, then, needs to be defined in its own terms rather than
by the definitions of fine art. One of the most straightforward tagging
questions in terms of painting, ‘What type of painting is this?’, is actu-
ally one of the most problematic when it is asked of illustration. Your
Paintings offered its taggers suggestions on how to answer this question:
a painting might be ‘abstract’, a ‘landscape’, ‘portrait’, ‘seascape’ or ‘still
life’. Again, however, while an illustration may appear ‘abstract’ (like the
images in some twentieth-century artists’ books, for instance), this is not
the same thing as an ‘abstract’ painting; neither is an illustrated landscape
the same as a painted one. And there are ‘types’ of image peculiar to
illustration that do not have equivalents in painting: illustrated advertise-
ments, vignettes, decorations, pictorial capitals, titlepages, frontispieces
and tailpieces.
Our objective in asking taggers in The Illustration Archive to spec-
ify the ‘type’ of illustration was not to formulate a definitive, classifica-
tory list of illustrations, but, rather, to make some sense of the diverse
illustrations contained in the British Library dataset and to allow the
end user to search the Archive for broader categories of illustration as
well as particular iconographic features or concepts, to be able to find
a ‘scientific’ or ‘photographic’ illustration alongside those illustrations
that depict ‘cats’ or ‘bridges’. But just as there is no standard formula
for bibliographically describing an illustration, neither is there an estab-
lished typology of illustrations. The very notion of a ‘type’ of illustra-
tion is fragmented and multifaceted. ‘Type’ itself has meanings that cross
the boundary between the visual and textual: a type in its earliest deriva-
tion is a picture of something, a representation, an image, but the word
also resonates with letters and printers’ type. Likewise, there is a slippage
between the typology of an illustration and its iconographic content. The
4 CROWDSOURCING 79
borders and pictorial capital letters. Our list of types also needed to take
into account the demands and protocols of crowdsourced image tagging,
which were often at odds with the ‘expert’ curatorial tagging assumed
by subject indexes. For example, we took the decision to visually define
our illustrative types using images from the Archive because our tagging
‘community’ was a potentially diverse one that might include anyone
from school-going children to practising illustrators. Each ‘type’ of illus-
tration is displayed on the site in the form of a thumbnail illustration as
this screenshot shows (Fig. 4.2):
Although many of the ‘types’ of illustration that are displayed in
this screenshot drill down to other ‘types’ (choosing ‘a decoration’,
for example, leads to ‘decorative border’, ‘decorative letter’, ‘coat of
arms’ and ‘decorative motif’), the main categories needed to be concise
enough to be displayed on a single screen, which meant that we were
highly selective in what we included as ‘types’. A key part of this selec-
tion process lay in determining if the aspect of an illustration that might
be called a ‘type’ was likely to be tagged in subsequent questions (the
marking up of an aspect of an image in The Illustration Archive only
requires inputting once for the image to be retrieved using this term).
The second question we asked the tagger, ‘Are there any other things or
ideas in this illustration?’, mops up the majority of iconographic informa-
tion about the image, meaning that we could remove certain categories
from our list of types, such as ‘animal illustration’ (if the tagger input-
ted ‘dog’ in the second question, the hypernyms for this term would
automatically be added by WordNet).50 Another determinant in what
was included and excluded from our list of illustrative types was the rela-
tive ease with which these types could be identified by the tagger. This
applied particularly to types of reproductive technique, which are diffi-
cult to identify, especially in digitised versions of images where the qual-
ity is not always consistent. We included ‘photography’ on the basis that
it is easier to recognise than most of the other processes, and the system
allows details of reproductive techniques or, indeed, any other additional
information to be inputted by the tagger in a free text box at the end of
the tagging task. The titles of the books, which would be interrogated in
a general search, also often include information about the reproductive
technique, especially if it was regarded as a marketable feature.51
The question ‘What type of illustration is this?’ involves a confron-
tation with ideas of ‘illustration’ that are determined by the develop-
mental processes that have gone into articulating this tagging question
82 J. THOMAS
Fig. 4.2 Screenshot from the ‘Tag Illustrations’ page, The Illustration Archive:
http://illustrationarchive.cardiff.ac.uk
4 CROWDSOURCING 83
in the first place. To put it simply, there is a sense in which the defini-
tions of ‘illustration’ are created by the digital archive itself, and these
definitions are multiple and unstable, transformed in the click of an
illustrated button. This comes to the fore in Science Gossip (https://
www.sciencegossip.org), a Zooniverse tagging project that crowdsources
information about the illustrations in nineteenth-century science peri-
odicals, including Hardwicke’s Science-Gossip, the journal that gives
the project its name. Part of the AHRC-funded project, ‘Constructing
Scientific Communities: Citizen Science in the nineteenth and twenty-
first Centuries’, Science Gossip makes parallels between the current infor-
mation revolution and the nineteenth-century revolution in print and
journal publishing when the number of science periodicals jumped from
around 100 titles in 1800 to 10,000 by the end of the century.52 This
crowdsourcing project has two main objectives: it is intended to sup-
plement metadata on the illustrations by adding information about who
created the images, their subject matter and any particular species they
portray; and it sets out to make an explicit connection between nine-
teenth-century citizen scientists who contributed to these periodicals
and the twenty-first century citizen scientists doing the tagging. As the
site states, ‘This is the first Zooniverse project where citizen scientists are
both the researchers and the subject of the research’.53
As an investigation into the communication of science in the Victorian
period and today, the project is extremely valuable, but its negotiation
of illustration is more problematic. This is exposed in the seemingly
transparent question that Science Gossip begins by asking its taggers, ‘Are
there any illustrations on this page?’ Like The Illustration Archive, this
site is forced to confront the difficulty of defining ‘illustration’, a dif-
ficulty that is compounded in the absence of an established vocabulary
around illustration. The ‘types’ of scientific illustration given as exam-
ples here are ‘drawing/painting/diagram, chart/table, photograph and
map’; but what is most revealing are those illustrations that are excluded
from this definition. The pages of these science journals are full of illus-
trated advertisements, decorated titlepages and pictorial capital letters,
but these are not defined as illustrations and, as such, are not identified
as ‘illustrations’ by the taggers. As Science Gossip openly declares, ‘we’re
only looking for certain types of illustrations’. The identification of only
‘certain types of illustrations’ is the source of some anxiety for its ‘citizen
scientists’, though. In one discussion thread, opening with the subject
line, ‘When is an illustration an illustration?’, the tagger writes, ‘On this
84 J. THOMAS
But can ‘decorative letters’ be disregarded on the basis that they ‘have
no relation to the text?’ The very presence of these images on the page
means that they are, inevitably, bound up in their interaction with the
words, even if this interaction is one of disjunction. The example given
on the site, a vignette of shells, crabs and seaweed that appears on a
page about astronomy, is a case in point. The image might not depict
anything explicitly ‘astronomical’, but this very deviation signifies.
4 CROWDSOURCING 85
Even leaving aside their relation to the text, such images make mean-
ings in other, distinctly visual, ways that call into question their defini-
tion as ‘purely decorative’. Decorations and ‘ornaments’ actually have a
key place in the history of the book, constituting some of the earliest
examples of graphic design. This significance has been attested in another
digital project, Fleuron (https://fleuron.lib.cam.ac.uk/), which contains
thousands of images of printers’ ornaments extracted from the pages of
eighteenth-century books. A century on from these fleurons and con-
temporaneous with the journals in Science Gossip, the role and aesthetic
importance of decorative and ‘ornamental’ illustrations was being revived
and redefined. Owen Jones’s classic book on the subject, The Grammar
of Ornament, was published in 1856 and stands alongside the ‘ornamen-
tal’ philosophy of the Kelmscott Press. Walter Crane uses the signifier
‘ornament’ in preference to ‘illustrate’ because of its associations with
the ideals of fine art.57 James Mussell has made a convincing case for the
recognition, in both theoretical and digital terms, of all the ‘visual codes’
that exist on the pages of nineteenth-century newspapers and periodi-
cals.58 Lorraine Janzen Koositra has similarly attested to the important
place of ornamentation in the periodical press in this period. Referring to
the Scottish periodical, The Evergreen, she comments on ‘the structural
patterns that emerge from the cumulative effect of the designs across
the periodical’s pages’.59 It is easy to envisage that the ornaments on the
pages of Hardwicke’s Science-Gossip function in the same way, but the
refusal to acknowledge their status as illustrations means that they could
be lost in the very digital space that promises to ‘unlock’ so many ‘illus-
trative treasures’.60
What is also lost in the definition of these ‘decorations’ as ‘not-
illustrations’ is the opportunity to explore the practice of recycling and
modifying illustrations to which the editor of Science Gossip refers in the
comments about the reuse of ‘decorative letters’. Indeed, the very pres-
ence of such reused images in Hardwicke’s Science-Gossip attests to the
ubiquity of a practice that crossed the boundaries of publications. The
editor, Charles Knight, for instance, often reused illustrations across his
versions of Shakespeare’s Works (which included a Stratford edition,
Pictorial edition, Cabinet edition, National edition and Imperial edition).
Even the illustrations in a seemingly stable form like the Victorian gift
book are frequently re-mixed. As Kooistra notes, ‘publishers increasingly
took advantage of the gift book’s modular form and recycled poems and
pictures from old stock in different combinations, tricked out in new
86 J. THOMAS
Don’t you think that it would be judicious to take the French artists’ and
gravers’ names from the pictures in Illustrated Travels? They take away the
idea of originality as your readers may easily know them to be French, and
there is no reason why they should know it. I adopted the plan myself with
some of your cuts which I purchased from Mr Turner in New York for my
magazine there. The least touch of the chisel will do it without damaging
the picture in any way.
Notes
1. For a sense of the range of these projects, see Mia Ridge (ed.),
Crowdsourcing our Cultural Heritage (Aldershot: Ashgate, 2014).
2. For an outline of this project, see Tim Causer and Valerie Wallace, ‘Building
A Volunteer Community: Results and Findings from Transcribe Bentham’,
DHQ: Digital Humanities Quarterly 6:2 (2012). Available online at
http://www.digitalhumanities.org/dhq/vol/6/2/000125/000125.
html. The crowdsourcing aspect of Transcribe Bentham is discussed in Tim
Causer and Melissa Terras, ‘“Many Hands Make Light Work. Many Hands
Together Make Merry Work”: Transcribe Bentham and Crowdsourcing
Manuscript Collections’, Crowdsourcing our Cultural Heritage, ed. Mia
Ridge (Aldershot: Ashgate, 2014), pp. 57–88.
3. See Luis von Ahn and Laura Dabbish, ‘Labeling images with a com-
puter game’, Proceedings of the SIGCHI conference on Human Factors
in Computing Systems (Vienna, Austria: ACM, 2004), pp. 319–326.
Available online at http://ael.gatech.edu/cs6452f13/files/2013/08/
labeling-images.pdf.
4. A recent study has identified three main aims of crowdsourcing: to explore
new forms of public engagement, to enrich institutional resources,
and to build novel resources. Laura Carletti, Derek McAuley, Dominic
4 CROWDSOURCING 89
Tailpiece
Keywords Nineteenth century · Illustration · Digital · Intertextuality
This tailpiece is not the end of the tale. So far, this book has focused
on the issues of making illustrations visible and searchable online, iden-
tifying the development of the digital resource as a process that engages
with notions of what an ‘illustration’ is and how it makes its meanings. I
want to shift the emphasis in these concluding pages to outline the con-
ceptual paradigms opened up when these new modes of visibility and
searchability are operational.
Digital archives offer a new accessibility for historic illustrations, mak-
ing available images housed in disparate publications and collections and
allowing them to be retrieved and viewed in diverse ways. Such features
are of benefit to researchers well beyond those working in the immedi-
ate field of illustration studies. Indeed, the identification of the ‘academic
they describe, where images can be moved around from box to box and
viewed in the context of other images, anticipates the modes of display
enabled by the digital, which emphasises and exposes affillustration. In
The Illustration Archive, the illustrations can be seen in a ‘list’ or ‘gal-
lery’ format (with or without metadata), viewed alongside illustrations
from the same book, viewed with ‘similar’ images, or in the context of
user-generated exhibitions. In DMVI, we added a ‘lightbox’ function
when we enhanced the site so that users can make their own collections
of illustrations. Digital resources can even emulate the practice of extra-
illustration. The Victorian Illustrated Shakespeare Archive brings together
illustrations from four nineteenth-century editions of Shakespeare and, at
the click of a button, reconfigures them in new ways, generating numer-
ous extra-illustrated versions.
Recognition of affillustration is not simply the result of being able
to view the images together, however. The research potential of digital
illustration resources lies in a combination of visibility and searchability,
of making illustrations available and allowing them to be interrogated
across bibliographic and iconographic fields. These features vary from
archive to archive, but, as a general rule, a bibliographic search mines
the metadata associated with the illustration, allowing users to find illus-
trations by publication title, author, publisher, date, illustrator and (in
the case of DMVI) engraver. This bibliographic search option offers up
illustrations to different modes of analysis. The diverse dataset in The
Illustration Archive, for example, makes it possible to analyse illustra-
tions by date range or decade, to look at illustrations of specific works,
or to gauge what texts were illustrated when. Such bibliographic searches
of illustrations lend themselves to multiple critical approaches, from
a ‘close’ viewing that analyses a single or small group of images to a
‘distant’ viewing, or what Matthew L. Jockers calls ‘a bird’s-eye view’,
which looks for overarching patterns.10
Whereas the bibliographic metadata consists of, and supplements,
information found in library catalogues, the iconographic details of illus-
trations are not so easily identifiable from catalogues, metadata or shelves
full of illustrated books. The ability to search the content of illustra-
tions and to identify the interrelations between images in terms of what
they depict is a direct result of digital intervention, having emerged
from the methodologies of digital illustration resources. In the absence
of a sophisticated method of Content-Based Image Retrieval, the icon-
ographic markup of illustrations is generated through the intermediary
100 J. THOMAS
the basement storage of the British Library and collected between 1850
and 1975. In terms of digitisation, the only selection criteria concerned
the size of the books, which were between 18.5 and 30.5 cm in height
and no thicker than 10 cm because of the capabilities of the scanners.
The sheer variety of this dataset means that the material is characterised
more by its diversity than uniformity and it is the very randomness or
‘unpredictability’ of the illustrations that has proved most fruitful. While
the 68,000 volumes limit the maximum number of authors to 68,000
(give or take multi-authored volumes), in any practical sense, the names
of the authors are ‘unpredictable’. This is also the case with the date
range: the collection is known as ‘The Nineteenth-Century Printed
Books Dataset’, but there are a significant number of eighteenth-century
books and we have identified books published as early as 1511 and as late
as 1946. It goes without saying that the iconographic features of these
illustrations are unquantifiable and unknown. As I have suggested in pre-
vious chapters, the unpredictability of the illustrations in The Illustration
Archive presents numerous challenges in terms of the development of the
archive, but there is a sense in which it also liberates the research poten-
tial of the digital resource, making it possible for the first time to identify
and explore the interplay between illustrations in ways that might com-
plement but are not limited to bibliographic imperatives.
With nineteenth-century illustrations now in the sight, if not in the
hands, of the user, it is time to draw a line under this tailpiece. This book
has focused on nineteenth-century material and archives, but its aim has
been to develop conceptual models for digital illustration studies that
might inform future practice more broadly, both in terms of how we
conceive of illustration and the relation between word and image, and
in how digital resources might be constructed to take account of and
expose the multiplicity and plurality of illustration. As I have argued,
the interaction between the digital and illustration is a dynamic one in
which the meanings of illustration are negotiated and played out. A fun-
damental aspect of these meanings is the visual dialogue between illus-
trations that I have called affillustration, which situates these images as
part of a network that shapes and is shaped by culture. I want to con-
clude by making the case that digital archives are uniquely positioned to
reveal these affiliative relations and to enable new and nuanced analyses
of how illustrations signify. The potential for this analysis of illustration,
102 J. THOMAS
moreover, lies as much in the development of the digital archive and the
modes of visibility and searchability that it institutes as in the tools and
features made available in the ‘live’ resource. The digital, then, is never
just a means of bringing illustration to light. Illustration might be allu-
sive, as the affiliative relations between these images suggest, but it is also
elusive, an ‘immense continent’ that is still largely unexplored. A digital
environment illuminates the way—it allows illustrations to be seen and
traversed; but it also gestures to hidden territories—to those areas that
continue to keep us in the dark (Fig. 5.1).
Fig. 5.1 ‘The End’, tailpiece from Edward Lamplough, Yorkshire Battles (Hull:
William Andrews and Co., 1891). Available on The Illustration Archive: http://
illustrationarchive.cardiff.ac.uk
5 TAILPIECE 103
Notes
1. Andrew Piper, Book Was There: Reading in Electronic Times (Chicago and
London: University of Chicago Press, 2012), pp. 55–56.
2. Piper, Book Was There, pp. 55, 56.
3. Stephen C. Behrendt, ‘The Functions of Illustration—Intentional
and Unintentional’, Imagination on a Long Rein. English Literature
Illustrated, ed. Joachim Möller (Marburg: Jonas Verlag, 1988),
pp. 29–44, p. 29.
4. Roland Barthes, ‘The Death of the Author’, Image Music Text, ed. and
trans. Stephen Heath (London: Fontana, 1977), pp. 142–148, p. 146.
5. See James Mussell, The Nineteenth-Century Press in the Digital Age
(Basingstoke: Palgrave Macmillan, 2012), pp. 73–74.
6. David Skilton, ‘The Centrality of Literary Illustration in Victorian Visual
Culture: The Example of Millais and Trollope from 1860 to 1864’,
Journal of Illustration Studies, 1 (December, 2007). Available online at
http://jois.uia.no/articles.php?article=30.
7. For a history of extra-illustration, see Erin C. Blake and Stuart Sillars,
Extending the Book: The Art of Extra-Illustration (Washington D.C.:
Folger Shakespeare Library, 2010).
8. Gleeson White, English Illustration, ‘The Sixties’: 1855–1870 (1897; Bath:
Kingsmead Reprints, 1970), p. 7.
9. Forrest Reid, Illustrators of the Eighteen Sixties (1928; Toronto: Dover,
1975), p. 11.
10. Jockers is referring to a way of analyzing literature that he calls ‘macro-
analysis’ and that draws on Franco Moretti’s concept of ‘distant’ read-
ing. The ‘scanning’ or ‘looking across’ of illustration that I propose,
although embedded in viewing rather than reading, is equally indebted
to the possibilities opened up by large-scale datasets. Matthew L. Jockers,
Macroanalysis: Digital Methods and Literary History (Urbana: University
of Illinois Press, 2013), p. 19. See also Franco Moretti, Distant Reading
(London and New York: Verso, 2013).
Bibliography
Ahn, Luis von and Laura Dabbish, ‘Labeling images with a computer game’,
Proceedings of the SIGCHI conference on Human Factors in Computing Systems
(Vienna, Austria: ACM, 2004) pp. 319–326. Available online at http://ael.
gatech.edu/cs6452f13/files/2013/08/labeling-images.pdf.
Allington, Daniel, Sarah Brouillette and David Golumbia, ‘Neoliberal Tools (and
Archives): A Political History of Digital Humanities’, Los Angeles Review of
Books, 1 May 2016.
Baca, Murtha (ed.), Introduction to Art Image Access: Issues, Tools, Standards,
Strategies (Los Angeles: Getty Research Institute, 2002).
Barthes, Roland, ‘The Death of the Author’, Image Music Text, ed. and trans.
Stephen Heath (London: Fontana, 1977) pp. 142–148.
Barthes, Roland, ‘Rhetoric of the Image’, Image, Music, Text, ed. and trans.
Stephen Heath (London: Fontana, 1977) pp. 32–51.
Beegan, Gerry, The Mass Image: A Social History of Photomechanical
Reproduction in Victorian London (Basingstoke: Palgrave Macmillan, 2008).
Behrendt, Stephen C., ‘The Functions of Illustration—Intentional and
Unintentional’, Imagination on a Long Rein. English Literature Illustrated,
ed. Joachim Möller (Marburg: Jonas Verlag, 1988) pp. 29–44.
Bell, Peter, Joseph Schlecht and Björn Ommer, ‘Nonverbal Communication in
Medieval Illustrations Revisited by Computer Vision and Art History’, Visual
Resources 29:1–2 (2013): 26–37.
Berger, John, Ways of Seeing (London: Penguin and BBC, 1972).
Bernstein, Shelley, ‘Clear Choices in Tagging’, Brooklyn Museum blogpost,
22 July 2014, https://www.brooklynmuseum.org/community/blogo-
sphere/2014/07/22/clear-choices-in-tagging/.
Blake, Erin C. and Stuart Sillars, Extending the Book: The Art of Extra-Illustration
(Washington D.C.: Folger Shakespeare Library, 2010).
Bloomfield, Robert, ‘The Farmer’s Boy’, Favourite English Poems (London:
Sampson Low, 1862) pp. 116–193.
Bolter, J. David, and Richard Grusin, Remediation: Understanding New Media
(Cambridge, MA: MIT Press, 1999).
Bolter, J. David, Writing Space: The Computer, Hypertext, and the History of
Writing (New Jersey and London: Lawrence Erlbaum Associates, 1991).
‘Book-Prints’, Chambers’s Journal of Popular Literature, Science and Arts, 30
August 1862, pp. 135–137.
Bradley, John and Stephen Pigney, ‘Towards a Bibliographic Model of
Illustrations in the Early Modern Illustrated Book’, paper delivered at the
Digital Humanities conference, University of Hamburg, 2012. Available
online at https://lecture2go.uni-hamburg.de/konferenzen/-/k/13923.
Burdick, Anne, Johanna Drucker, Peter Lunenfeld, Todd Presner, Jeffrey
Schnapp, Digital_Humanities (Cambridge, MA: MIT Press, 2012).
Carletti, Laura, Derek McAuley, Dominic Price, Gabriella Giannachi and Steve
Benford, ‘Digital Humanities and Crowdsourcing: An Exploration’, paper
delivered at MW 2013: Museums and the Web conference, 2013. Available
online at http://mw2013.museumsandtheweb.com/paper/digital-humanities-
and-crowdsourcing-an-exploration-4/.
Causer, Tim and Melissa Terras, ‘“Many Hands Make Light Work. Many Hands
Together Make Merry Work”: Transcribe Bentham and Crowdsourcing
Manuscript Collections’, Crowdsourcing our Cultural Heritage, ed. Mia Ridge
(Aldershot: Ashgate, 2014) pp. 57–88.
Causer, Tim and Valerie Wallace, ‘Building A Volunteer Community: Results
and Findings from Transcribe Bentham’, DHQ: Digital Humanities Quarterly
6:2 (2012). Available online at http://digitalhumanities.org:8081/dhq/
vol/6/2/000125/000125.html.
Chapman, Alison, ‘Virtual Victorian Poetry’, Virtual Victorians: Networks,
Connections, Technologies, ed. Veronica Alfano and Andrew Stauffer (New
York and Basingstoke: Palgrave Macmillan, 2015) pp. 145–166.
Chung, J. S., R. Arandjelović, G. Bergel, A. Franklin and A. Zisserman,
‘Re-presentations of Art Collections’, paper presented for the Workshop on
Computer Vision for Art Analysis, ECCV, 2014. Available online at http://
www.robots.ox.ac.uk/~vgg/publications/2014/Chung14/.
Cooke, Simon, ‘A Bitter After-Taste: The Illustrated Gift Book of the 1860s’,
Reading Victorian Illustration, 1855–1875: Spoils of the Lumber Room, ed.
Paul Goldman and Simon Cooke (Aldershot: Ashgate, 2012) pp. 53–78.
Crane, Walter, Of the Decorative Illustration of Books Old and New (1896;
London: Bracken, 1984).
Bibliography 107
Ruskin, John, The Cestus of Alglaia, The Complete Works of John Ruskin, ed.
E. T. Cook and Alexander Wedderburn (London: George Allen, 1903–1912)
vol. XIX, pp. 139–140.
Sabharwal, Arjun, Digital Curation in the Digital Humanities: Preserving and
Promoting Archival and Special Collections (Waltham, MA: Chandos, 2015).
Saussure, Ferdinand de, Course in General Linguistics, ed. Charles Bally and
Albert Sechehaye, trans. Wade Baskin (New York: McGraw-Hill, 1959).
Seed, Patricia, ‘A Map is Not a Picture: How the Digital World Threatens the
Validity of Printed Maps’, Between Humanities and the Digital, ed. Patrik
Svensson and David Theo Goldberg (Cambridge, MA and London: MIT
Press, 2015) pp. 401–409.
Shillingsburg, Peter L., ‘Practical Editions of Literary Texts’,Variants 4: The
Book as Artefact: Text and Border, ed. Anne MetteHansen, Roger Lüdeke,
Wolfgang Streit, Cristina Urchuaguía and Peter Shillingsburg (Amsterdam
and New York: Rodopi, 2005) pp. 29–55.
Skilton, David, ‘The Centrality of Literary Illustration in Victorian Visual
Culture: The Example of Millais and Trollope from 1860 to 1864’, Journal
of Illustration Studies 1 (December 2007). Available online at http://jois.uia.
no/articles.php?article=30.
Skilton, David, ‘Illustration Studies and the Infinite Archive’, Journal of
Illustration Studies 2 (December 2013). Available online at http://jois.uia.
no/articles.php?article=50.
Springer, Michelle, Beth Dulabahn, Phil Michel, Barbara Natanson, David Reser,
David Woodward and Helena Zinkham, For the Common Good: The Library
of Congress Flickr Pilot Project, report (30 October 2008). Available online at
http://www.loc.gov/rr/print/flickr_report_final.pdf.
Stauffer, Andrew, ‘Evidence and Interpretation in the Digital Age: Searching
Engines, Reading Machines’, Victorian Studies 54:1 (2011): 63–68.
Stevens, Joan, ‘Thackeray’s Pictorial Capitals’, Costerus: Essays in English and
American Language and Literature 2 (1974): 113–140.
Terras, Melissa, ‘Crowdsourcing in the Digital Humanities’, A New Companion
to Digital Humanities, ed. Susan Schreibman, Ray Siemens and John
Unsworth (Chichester: Wiley-Blackwell, 2016) pp. 420–438.
Terras, Melissa, ‘Cultural Heritage Information: Artefacts and Digitization
Technologies’, Cultural Heritage Information: Access and Management, ed.
Ian Ruthven and G. G. Chowdhury (London: Facet, 2015) pp. 63–88.
Terras, Melissa, ‘Reuse of Digitised Content (1)’, blogpost, 6 October 2014,
http://melissaterras.blogspot.co.uk/2014/10/reuse-of-digitised-content-
1-so-you.html.
Thackeray, William Makepeace, Vanity Fair. A Novel Without a Hero (London:
Bradbury and Evans, 1848).
112 Bibliography
Thomas, Julia, ‘Getting the Picture: Word and Image in the Digital Archive’,
European Journal of English Studies 11:2 (August 2007): 193–206.
Thomas, Julia, Pictorial Victorians: The Inscription of Values in Word and Image
(Athens, OH: Ohio University Press, 2004).
Thomas, Julia, ‘Reflections on Illustration: The Database of Mid-Victorian Wood-
Engraved Illustration’, Journal of Illustration Studies 1 (December 2007).
Available online at http://jois.uia.no/articles.php?article=37.
Trant, Jennifer, David Bearman and Susan Chun, ‘The Eye of the Beholder:
Steve.Museum and Social Tagging of Museum Collections’, proceedings of
ICHIM, 24–26 October 2007, Toronto, Canada. Available online at http://
www.archimuse.com/ichim07/papers/trant/trant.html.
Toronto, Jennifer, ‘Tagging, Folksonomy and Art Museums: Early Experiments
and Ongoing Research’, Journal of Digital Information 10:1 (2009).
Trollope, Anthony, Orley Farm, illustrated by John Everett Millais and engraved
by the Dalziel Brothers (London: Chapman and Hall, 1861–1862).
Trollope, Anthony, Orley Farm, ed. David Skilton (1861–1862; Oxford: Oxford
University Press, 1985).
Urbina, Eduardo, ‘Visual Knowledge: Textual Iconography of the Quixote’,
Don QuixoteIllustrated: Textual Images and Visual Readings/Iconografía del
Quijote, ed. Eduardo Urbina and Jesús G. Maestro, Biblioteca Cervantes 2
(Pontevedra: Mirabel Editorial, 2005) pp. 15–38.
Vandendorpe, Christian, From Papyrus to Hypertext: Toward the Universal Digital
Library, trans. Phyllis Aronoff and Howard Scott (1999; Urbana: University
of Illinois Press, 2009).
Vaughan, William, ‘Facsimile Versus White Line: An Anglo-German Disparity’,
Reading Victorian Illustration, 1855–1875: Spoils of the Lumber Room,
ed. Paul Goldman and Simon Cooke (Aldershot: Ashgate, 2012) pp. 33–52.
Wallace, Doireann, ‘Words as Keys to the Image Bank’, Revisualizing Visual
Culture, ed. Chris Bailey and Hazel Gardiner (Aldershot: Ashgate, 2010)
pp. 83–96.
Wan, Ji, Dayong Wang, Steven C. H. Hoi, Pengcheng Wu, Jianke Zhu,
Yongdong Zhang and Jintao Li, ‘Deep Learning for Content-Based Image
Retrieval: A Comprehensive Study’, 2014. Available online at https://pdfs.
semanticscholar.org/391b/86cf16c2702dcc4beee55a6dd6d3bd7cf27b.pdf.
White, Gleeson, English Illustration, ‘The Sixties’: 1855–1870 (1897; Bath:
Kingsmead Reprints, 1970).
‘Zooniverse releases Science Gossip’, Biodiversity Heritage Library blogpost,
4 March 2015, http://blog.biodiversitylibrary.org/2015/03/zooniverse-
releases-science-gossip.html.
Bibliography 113
Web Resources
Arkyves http://arkyves.org.
Art and Architecture Thesaurus Online www.getty.edu/research/tools/vocabu-
laries/aat.
ARTigo www.artigo.org.
British Museum Collection Online http://www.britishmuseum.org/research/
collection_online/search.aspx.
British Printed Images to 1700 www.bpi1700.org.uk, dir. Michael Hunter.
Broadside Ballads Online http://ballads.bodleian.ox.ac.uk, dir. Alexandra
Franklin.
Brooklyn Museum www.brooklynmuseum.org.
Cervantes Project http://cervantes.tamu.edu, dir. Eduardo Urbina.
Citizen Archivist www.archives.gov/citizen-archivist.
Connected Histories www.connectedhistories.org.
Database of Mid-Victorian Illustration www.dmvi.cardiff.ac.uk, dir. Julia Thomas.
Database of Mid-Victorian Illustration (enhanced site) www.dmvi.org.uk, dir.
Julia Thomas.
Fleuron: A Database of Eighteenth-Century Printers’ Ornaments https://fleuron.
lib.cam.ac.uk/, dir. Hazel Wilkinson.
Flickr: The Commons www.flickr.com/commons.
HathiTrust Digital Library www.hathitrust.org.
Iconclass www.iconclass.nl.
Illustration Archive http://illustrationarchive.cardiff.ac.uk, dir. Julia Thomas.
Illustrating Scott http://illustratingscott.lib.ed.ac.uk, dir. Peter Garside.
Internet Archive https://archive.org.
Library of Congress Subject Headings http://id.loc.gov/authorities/subjects.
html.
Marie Duval Archive http://www.marieduval.org, dir. Simon Grennan.
MatchEngine https://services.tineye.com/MatchEngine.
Metadata Games www.metadatagames.org.
NINES www.nines.org.
Nineteenth-Century Serials Edition www.ncse.ac.uk, dir. Laurel Brake.
Philadelphia Museum of Art www.philamuseum.org.
Powerhouse Museum https://maas.museum/powerhouse-museum.
Rossetti Archive www.rossettiarchive.org, dir. Jerome McGann.
Science Gossip www.sciencegossip.org, dir. Sally Shuttleworth.
Transcribe Bentham www.transcribe-bentham.da.ulcc.ac.uk, dir. Philip Schofield.
Ukiyo-e http://ukiyo-e.org, dir. John Resig.
Victorian Illustrated Shakespeare Archive www.shakespeareillustration.org, dir.
Michael John Goodman.
114 Bibliography
A B
Aberystwyth University, School of Art Bag of Words, 36. See also Content-
Gallery and Museum, 4 Based Image Retrieval (CBIR)
Adobe illustrator, 60 Baltussen, Lotte Belice, 89
Adobe Photoshop, 25, 26, 28 Barthes, Roland, 53
Advertisements, 78, 80, 83, 88 Bearman, David, 90
Affillustration, 12, 97–101 Beegan, Gerry, 31
Afsar, Kamil, 89 Behrendt, Stephen C., 97
Ahn, Luis von, 88 Bell, Peter, 61
Alexander, Mrs., A Life Interest, 40 Benford, Steve, 89
Allington, Daniel, 14 Bennett, Charles, 67
Althusser, Louis, 70 Beowulf, 80
Apple, 10 Bergel, Giles, 60
Aquatint, 37 Berger, John, 69
Arandjelović, Relja, 60 Bernstein, Shelley, 72, 74
Arkyves, 57 Bible, 17, 57
Aroyo, Lora, 89 Bibliographic metadata, 5, 21, 39, 40,
Art and Architecture Thesaurus Online 42, 43, 46, 51, 99
(AAT), 79, 80 Big Data, 5, 8, 47, 48
Art UK, 54. See also Your Paintings Blake, Erin C., 103
ARTigo, 67 Bloomfield, Robert‚ ‘The Farmer’s
Arts and Humanities Research Council Boy’, 49
(AHRC), 3, 83 Bolter, Jay David, 9, 29
Ashmolean Museum, Oxford, 4, 53 Bradley, John, 61
F Greg, Andrew, 89
Facebook, 74 Grennan, Simon, 39. See also Marie
Falstaff, 79. See also Shakespeare, Duval Archive
William Grusin, Richard, 29
Family Paper, 86
Favourite English Poems, 48–50
Fenn, George Manville, 86 H
First Impressions, 55 Hardwicke’s Science-Gossip, 83, 85. See
Flanagan, Mary, 91 also Science Gossip
Flaubert, Gustave, 7, 54 Hardy, Thomas, 21
Fleuron, 79, 85 Harpring, Patricia, 22
Flickr, 22, 23, 47, 55, 67, 70, 74, 75 Harvey, Ian, 61
Fogg, Martha, 63 HathiTrust Digital Library, 42
Folksonomies, 74, 78 Hayles, Katherine N., 27
Foster, Myles Birket, 49, 51, 59 Hedges, Mark, 70
Franklin, Andrew, 60 Heilmann, Ann, 93
Frick Photoarchive, 37 Historical Texts, 39
Frontispieces, 78, 80 Hockney, David, 22
Fyfe, J. Hamilton, ‘About Toys’, 58 Hocks, Mary E., 9
Hofstede, Geert, 70
Hoi, Steven C.H., 61
G Holterhoff, Kate, 21. See also Visual
Galloway, Alexander R., 14 Haggard
Genette, Gérard, 20 Home, Anne, The Sports of the Genii,
Getty Research Institute, 79 41. See also Macdonald, Susan
Giannachi, Gabriella, 89 Hood’s Own: or, Laughter from Year to
Gift books, 21, 51 Year, 22
Gligorov, Riste, 89 Houghton, Arthur Boyd, 59
Goldman, Paul, 5, 18, 25 Hypermedia, 11
Golumbia, David, 14
Good Words, 4, 20
Goodman, Michael, 28. See also I
Victorian Illustrated Shakespeare IBM, 70
Archive Iconclass, 57, 59
Goodyear, William Henry, 72 Idyllic School, 20
Google, 3, 4, 10, 37, 39, 51, 67 Illuminated Magazine, 98
Google Image Labeler, 66 Illustrated London News Historical
Gore, Catherine, 29 Archive, 39
Gould, Emilie West, 70 Illustrating Scott, 21, 53, 100
Graham, Margaret E., 60 Illustration Archive, 2, 3, 5, 8, 21, 22,
Granger, James, Biographical History 25, 28, 35, 37, 38, 40, 42, 43,
of England, 98
118 Index
46, 56, 67, 75–78, 81, 83, 88, Layne, Sara Shatford, 78
96, 97, 99–101 Leech, John, 98
Illustration Studies, 5–9, 12, 20, 100, Leetaru, Kalev, 46, 47
101 Leighton, Frederic, 4, 49
ImageNet, 39 Lessing, Gotthold, 55
Indianapolis Museum of Art. See First Li, Jia, 60
Impressions Li, Jintao, 61
Instagram, 74 Library of Congress Subject Headings
Interface, 9, 10, 69–71 (LCSH), 80
Intermediality, 13 Library of Congress, 68. See also
Internet Archive, 42, 46 Flickr; Library of Congress Subject
Intertextuality, 97 Headings (LCSH)
Ionescu, Christina, 8 Limonard, Sander, 89
Lithography, 37, 79
Lloyd, Nicky, 61
J London Society, 4, 20, 51
Jackson, John, 32 Los Angeles Review of Books, 8
James, Henry, 7, 24 Lunenfeld, Peter, 63
Jessop, Martyn, 6, 7
Jockers, Matthew L., 99
Jones, Owen, 85 M
Joshi, Dhiraj, 60 Macdonald, Susan, The Sports of the
Jung, Sandro, 7 Genii, 41
Mackay, Charles, 91
Maidment, Brian, 31
K Mandal, Anthony, 14
Kaufman, Geoff, 91 Manovich, Lev, 27, 28, 70
Kelmscott Press, 85 Manzo, Christina, 91
Kendrick, Michelle R., 9 Maps, 38–40, 46, 88
Keywording, 4. See also Tagging Marcus, Aaron, 70
Kim, Sung-Do, 30 Marie Duval Archive, 39
Kingston, W. H. G., 86 Martineau, Harriet, The Anglers of the
Kirschenbaum, Matthew, 37 Dove, 34
Kittler, Friedrich, 10, 19 Maxwell, Richard, 29
Klingemann, Mario, 38 Mayer-Schönberger, Viktor, 48, 51, 75
Knight, Charles, 85 McAuley, Derek, 88
Kooistra, Lorraine Janzen, 34. See also McGann, Jerome, 2. See also Rossetti
Yellow Nineties Online Archive
McPherson, Tara, 21
Mendelsund, Peter, 54
L Metadata Games, 67
Landon, Letitia Elizabeth, 2 Meyrick, Robert, 13
Index 119
Schnapp, Jeffrey, 13 U
Science Gossip, 83–85, 87, 88 Ukiyo-e.org, 37
Scientific illustration, 80, 83 Urbina, Eduardo, 2. See also Cervantes
Seed, Patricia, 93 Project
Shakespeare, William, 80
Shillingsburg, Peter L., 29
Sillars, Stuart, 103 V
Skilton, David, 6, 98 Vandendorpe, Christian, 11
Spinden, Herbert, 72 Vaughan, William, 28, 42
Springer, Michelle, 89 Vervaart, Just, 89
Stauffer, Andrew, 12, 92 Victorian Illustrated Shakespeare
Steve.museum, 73 Archive, 26, 28, 99, 100
Stevens, Bethan, 32 Vignettes, 24, 78
Stevens, Joan, 62 Viscomi, Joseph, 2. See also William
Swain, Joseph, 34 Blake Archive
Visual Geometry Group, University of
Oxford, 37
T Visual Haggard, 21
Tag! You’re It!, 71, 72, 74 Visualisation, 27
Tagging, 5. See also Keywording,
Crowdsourced image tagging
Tenniel, John, 15, 21 W
Tennyson, Alfred Lord, 21 Waal, Henri van de. See Iconclass, 57
Terras, Melissa, 22, 25 Waite, Julian, 39. See also Marie Duval
Text Encoding Initiative (TEI), 52 Archive
Thackeray, William Makepeace; Vanity Wallace, Doireann, 75
Fair, 17 Wallace, Valerie, 88
The Times Digital Archive, 39 Wan, Ji, 61
Thomas, Julia, 3. See also Database Wang, Dayong, 61
of Mid-Victorian Illustration; Wang, James Z, 60
Illustration Archive Watson, John Dawson, 62
Titlepages, 78, 80, 83, 88 Wehnert, Edward Henry, 62
Transcribe Bentham, 65 Welsh Newspapers Online, 39
Trant, Jennifer, 90 White, Gleeson, 20, 24, 99
Trollope, Anthony; Framley Parsonage; Wilde, Oscar, 21
Orley Farm, 4, 38, 50, 98 William Blake Archive, 2, 21, 26, 27,
Tumblr, 10 52, 100
Twitter, 74 Wilson, Margaret, 57
Index 121