Computers and the Humanities 36: 27–48, 2002.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.

27

Digital Facsimiles: Reading the
William Blake Archive

∗

JOSEPH VISCOMI

University of Noth Carolina, Department of English, Greenlaw Hall, CB #3520, NC 27599-3520,
Chapel Hill, USA
E-mail: [e-mail]

Tyger, Tyger, burning bright,
In the forest of the night:
What immortal hand or eye,
Could frame thy fearful symmetry?

(1–4)

1

1. The Case for Facsimile Reproduction

The lines of the epigraph are from the most anthologized poem in English
literature.

2

They are, of course, the first four lines from William Blake’s “The

Tyger,” one of twenty-six poems in his Songs of Experience, first printed in 1794
with twenty-one poems and combined at that time with his Songs of Innocence
of 1789 to create his best known illuminated book, Songs of Innocence and of
Experience. These bibliographical details, though, only hint at the complexities
of studying and editing Blake’s poetry, complexities revealed more quickly and
clearly by a simple comparison between the lines as they appear above and the
form in which they were originally read (Figure 1).

Immediately, we see that Blake’s text is calligraphic, illustrated, and finished

in watercolors, features which textual scholars have argued theoretically – and, I
believe, we sense intuitively – as contributing to the meaning of the whole.

3

Indeed,

we recognize the typographic translation as grossly distorting the original artifact,
a hand-colored impression printed from a copper plate executed in “illuminated
printing” (i.e., “relief etching”), a technique Blake invented in 1788. Instead of the
needles, burins, and other metal tools of the graphic artist, Blake worked on copper
plates with pens, small brushes, and an ink impervious to acid (probably asphaltum
in turpentine mixed with a little lampblack); he wrote text backward, illustrated it,
and etched the untouched metal below the surface to leave the integrated design
in printable relief. Working on metal with the tools of poet and painter enabled
Blake to create a multi-media space, a “site” where poetry, painting, and print-

28

JOSEPH VISCOMI

Figure 1. “The Tyger,” Songs of Innocence and of Experience copy C.

making came together in ways both original and characteristic of Romanticism’s
fascination with autographic gesture, with spontaneity, intimacy, and organicism.

Blake used illuminated printing as a mode of production rather than repro-

duction, combining text and illustration on the plate for the first time rather than
reproducing a pre-existent page design. The etched design, however, is continually
recreated when printed, for it is printed in different inks and colored differently
at different periods. The result is a “printed manuscript,” an oxymoron coined by
Robert Essick to describe Blake’s ability to produce repeatable yet unique works.
For example, “The Tyger” above (Figure 1) is from an early copy of Songs desig-
nated as copy “C”; its Experience impressions were printed in 1794 in yellow ochre
ink on both sides of fine wove paper (which when bound with other leaves created
facing pages) and finished in light watercolor washes. This version looks and feels
quite different from the impression in late copy Z (Figure 2), which was printed in
1826 in orange-red ink on one side of the leaf, elaborately colored, strengthened in
pen and ink, and given frame lines like a miniature painting.

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

29

Figure 2. “The Tyger,” Songs of Innocence and of Experience copy Z.

Now, add to this basic comparison numerous other impressions produced from

the same plate at different times and in different production styles, each looking
slightly to dramatically different from the other, each with the possibility of textual
variants, and nearly all occupying a different place in each copy of the book, and
you will begin to glimpse the bibliographical and textual difficulties confronting
the editor – and student – of Blake.

How editors resolve these difficulties affects directly how Blake is known and,

of course, what Blake we come to know. Typographic transcriptions, which abstract
texts from the artifacts in which they are versioned and embodied, made good
economic but poor editorial sense. They made possible inexpensive editions of
Blake’s poetry and his inclusion in anthologies – and thus classrooms – but, as we
see, at the expense of Blake’s intentions. Such transcriptions are “reader’s texts”
when Blake’s idiosyncratic punctuation is corrected, with stops determined by
modern rules of syntax or grammar. For example, Keynes edits the lines as:

30

JOSEPH VISCOMI

Tyger! Tyger! burning bright
In the forest of the night,
What immortal hand or eye
Could frame thy fearful symmetry?

(1–4)

4

Erdman, on the other hand, attempts to reflect Blake’s markings as closely as type
allows (many of Blake’s marks have no typographic equivalent), choosing marks
by consensus, by comparing numerous printings of the text to see which mark –
e.g. comma or period – is most often present:

Tyger Tyger, burning bright,
In the forest of the night;
What immortal hand or eye,
Could frame thy fearful symmetry?

(1–4)

5

The resulting composite text may claim to have excavated the text as executed on
the copper plates, but it no more corresponds to an actual printed work than the
user friendly reader’s text.

Relative to type, unscaled monochromatic reproductions of colored illuminated

pages move readers closer to Blake’s vision, though they fail to capture size, color,
and texture of the original. At the other end of the reproductive continuum are
facsimiles, and the finest of these, like the hand-colored collotypes produced by
the William Blake Trust between 1952 and 1978, represent the originals more
successfully, but as expensive limited editions they are themselves not readily
available to students. And even they fail to provide the kinds of detail necessary
for good art historical and editorial analyses. To discern, for example, whether
a mark was etched on the copper or added or changed afterwards in printing or
coloring the impression would still require close first-hand scrutiny of the original
works, which are housed in international collections at widely separated locations.
Moreover, facsimiles and reproductions both present an edited Blake, edited in the
sense of works selected for reproduction and in the way images are reproduced.
Indeed, the public’s exposure to Blake – and this includes many advanced students
and not a few scholars – has been narrowly restricted to a small number of items
that have been too frequently reproduced.

6

For example, Songs copy Z (Figure 2)

is one of only two copies (both late) commercially reproduced in color, whereas
the less flashy early copy C (Figure 1) has never been reproduced. Reproductions
and facsimiles also edit Blake by reproducing only the image and not the full
sheet, which means bibliographical information, such as paper’s size and the plate’s
registration to the paper, goes unrecorded and Blake – or Mrs. Blake – appears to
be a neater printer than he/she was and the pages visually more uniform than they
are.

Typographic editions and reproductions of only about 20% of Blake’s illumin-

ated canon (40 or so of the 175 copies of the 19 illuminated books Blake produced

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

31

between 1788 and 1827, when he died), reproduced sometimes well, sometimes
execrably, but in no coherent historical order and insufficient detail to sustain
scholarly and editorial research – this was the state of Blake studies when in 1993
the editors of the Blake Archive began to conceive of reproducing Blake digitally.
Morris Eaves, Robert Essick, and I had just finished editing nine illuminated works
for the Blake Trust and, while pleased with the scholarly apparatus we developed,
we were frustrated by the relatively small number of reproductions allowed. With
the economic restraints of the codex form fresh in mind, we visited the Insti-
tute for Advanced Technology in the Humanities at the University of Virginia,
where we began to envision a critical hypertext of approximately 3000 images,
2/3rds drawn from the illuminated works and the remainder from Blake’s paint-
ings, drawings, prints, and manuscripts, with all texts and images deeply encoded
in SGML. We would represent the illuminated canon by exemplary copies from
each printing of each illuminated book, as well as copies from the same printing
session with important variants in coloring, motifs, arrangements, etc., along with
related material, such as drawings, proofs, and sketches, so that the production
history of each book would be recorded.

7

About half of the books selected for

inclusion had never been reproduced before – including six of our eight copies of
Songs. Our typographic transcriptions of texts would be, in the terms of textual
criticism, as “diplomatic” as the medium allows. That is, in line with the archival
dimension of our project, our texts are conservative transpositions of the original
into conventional type fonts, retaining not only Blake’s capitalization, punctuation
(within the limitations of typography), and spelling, but also (for the first time
in a complete edition) his page layout. Unlike printed editions of Blake, which,
as noted, have typically chosen among the textual features of various copies to
produce a single printed text, the texts in the Archive are specific to individual
plates; each transcription is of a particular plate in a particular copy and no
other.

8

Once archived digitally, structured and tagged (indexed for retrieval in SGML,

adapted to the purpose), annotated with detailed descriptions, and orchestrated
with a powerful search engine (in this case DynaWeb software), the images in
the Archive could be examined like ordinary color reproductions. But they could
also be searched alongside the texts, enlarged, computer enhanced, juxtaposed in
numerous combinations, and otherwise manipulated to investigate features (such
as the etched basis of the designs and texts) that have heretofore been imper-
ceptible without close examination of the original works. Even scholars who are
able to globetrot from collection to collection end up relying heavily upon their
inadequate memories, notes, photocopies, and photographs to compensate for the
distances in time and space between collections. Seeing the original prints, paint-
ings, manuscripts, and typographical works is good in itself; but seeing them in
fine, trustworthy reproductions, in context and in relation to one another is the
scholarly ideal. Difficulty of access to originals and reliance on inadequate repro-
ductions has handicapped and distorted even the best efforts. Again, the result has

32

JOSEPH VISCOMI

all too frequently been distortions of the record, misconstructions, and the waste of
considerable scholarly labor.

In the broadest terms, the Blake Archive is a contemporary response to the

needs of a dispersed and various audience of readers and viewers and to the
corresponding needs of the collections where Blake’s original works are currently
held. Both the audience and the collections, institutions, and curators on which it
depends share a strong interest in the accessibility and the preservation of Blake’s
works. The Blake Archive attempts to serve both sets of needs at once by providing
free access to its Web site, where access to Blake’s works is possible to a degree
heretofore impossible. But we have designed the site primarily with scholars in
mind. For them we believe that the Archive will soon become not merely useful
but indispensable – as a handy reference, a point of departure, or a site of sustained
research – because of the fidelity of our reproductions. Reproductions can never be
perfect, and our images are not intended to be “archival” in the sense sometimes
intended – virtual copies that might stand in for originals after a fire. But we recog-
nize that, if we are going to contribute as we claim to the preservation of fragile
originals that are easily damaged by handling, we must supply reproductions that
scholars can depend upon in their research. Hence our benchmarks produce images
accurate enough to be studied at a level heretofore impossible without access to the
originals. As we shall see, in side-by-side comparisons, images in the Archive are
more faithful to the originals in scale, color, and texture than the best photomechan-
ical (printed) images in all but the most extraordinary instances, and provide more
detail than even unmagnified originals.

2. Digital Facsimiles in the Blake Archive

The Archive’s digital images are scanned from three types of color transparencies:
35mm slides, 4

× 5-inch, and 8 × 10-inch transparencies. While both of the larger

size transparencies include color bars and gray scales to ensure color fidelity, the 4
× 5 inch transparencies are our preferred and most often used reproductive source.
They are much less expensive to produce and can be scanned directly instead
of through glass (see below). For the sake of consistency, if not also quality of
image, we prefer film to digital images from contributing collections, where equip-
ment and protocols for digital capture vary. As the editor responsible for color
correcting the digital images, I have verified most of our transparencies for color
accuracy against the original artifact, and in all cases they have been verified by the
photographer employed by the owner institution. Indeed, for three of the largest
collections in the Archive, together consisting of over 1400 images, we brought in
our own photographer and I was the assistant on the shoot. The transparencies were
developed at the end of each day and compared the next morning, before that day’s
shoot, to the originals on a color-corrected light box. We used Kodak Kodachrome
Professional EPP and EPY film, which captures the soft gradations of watercolors

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

33

Figure 3. The Book of Urizen copy G, plate 4, full sheet.

and paper tone, and is the film type most of our contributing institutions use as
well, and we used cool strobe light to illuminate the originals.

The size of our originals range from 2.5

× 2 inches for the smallest illuminated

prints to over 30

× 20 inches for the largest watercolors and paintings. These

measurements, though, refer to the size of the image and not leaf or canvas. For
example, plate 4 of The Book of Urizen copy G is approximately 6

× 4 inches, but

was printed on a sheet 11.5

× 9 inches (Figure 3).

Rather than photographing the full sheet, we cropped to the image, leaving just

enough room on the 4

× 5 inch film for the color bars and gray scale (Figure 4).

34

JOSEPH VISCOMI

Figure 4. The Book of Urizen copy G, plate 4, uncorrected transparency with color bars and
gray scale.

Admittedly, cropping discards bibliographical information, such, as in this case,
the plate’s poor registration to the sheet, and thus fails to represent the artifact
itself. Consequently, like printed reproductions and facsimiles, our digital images
(Figure 5) also hide poor registrations, but our reasons for cropping are more
technical and pedagogical than economic and aesthetic.

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

35

Figure 5. The Book of Urizen copy G, plate 4, corrected in Archive’s Object – View Page.

Recording the image on the film as large as possible reduces the scaling ratio

between reproductive source and original, which ensures greater accuracy in the
digital image (see below); cropping out the image’s margins significantly reduces
file size, and, for most images, enables the image to be shown within the Archive’s
interface on monitors 17" or larger without scrolling. For us, cropping at this
initial stage of photography is determined by the current material constraints of
the electronic medium – by file sizes, monitor size, screen resolutions, and band
width.

Until May of 1998, we scanned our transparencies at IATH on a Microtek

Scanmaker III, a flatbed scanner with a transparent media adapter. Since then, we
have scanned them at UNC on a Microtek Scanmaker V, which, instead of the

36

JOSEPH VISCOMI

adapter built into the lid, uses a separate drawer to hold 4

× 5-inch film inside

the scanner’s body, where it is scanned using Microtek’s EDIT (Emulsion Direct
Imaging Technology) system. Scanning transparencies directly and not through
glass creates a sharper scan (one less layer to absorb light), eliminates Newton rings
(interference patterns resulting from the contact of film and glass), and reduces
dust. Slides (which have only rarely been used in the Archive) are scanned on
either a Nikon LS-3510AF Slide Scanner or a Microtek 35t Plus Scanner.

9

The baseline standard for all scanned images to date is 24-bit color and a resol-

ution of 300 dots per inch (dpi). Depending on the size of the original, the scanned
image is scaled against the source dimensions of the original artifact at 1:1, 1:2/3,
or 1:1/2. We have found that these settings produce excellent enlargements and
suitably sized inline images. Nonetheless, we have begun to scan at 600-dpi now
that our hardware (the Macintosh G3 is equipped with 384 MB of RAM and 9
GB of internal storage and another 5 GB external) can better accommodate the
larger file sizes that result, and we have also changed the type of image file we
color-correct. Initially, we saved our raw 300-dpi scans as TIFFs and as JPEGs,
storing the TIFFs (Tagged Image File Format) on 8mm magnetic Exabyte tape –
data which we have since migrated to CD-ROM (in Mac/ISO 9660 hybrid format).
Now, by scanning at 600-dpi, we capture more information, including paper tone
and texture, but the resulting files are quite large – artifacts a mere 8.5

× 6.5 inches

produce files in excess of 100 MB. Consequently, we store these raw scans on
CD-ROM as (LZW-compressed) TIFFs, which cuts the file size in half. From the
original 600-dpi TIFF image, we also derive a 300-dpi (compressed) TIFF, which
significantly reduces the file size with no noticeable loss of information. Even so,
like its parent image, the 300-dpi TIFF is saved on CD-ROM and stored outside
the Archive proper.

The color-corrected images that we display to our users in the online Archive

are all served using the JPEG (Joint Photographic Experts Group, ISO/IEC 10918)
format. A 300-dpi JPEG is derived as the final step of the color-correction process
and sent by electronic file transfer from UNC to the Archive’s server at Virginia,
where an assistant logs its arrival on our internal tracking sheets (this 300-dpi JPEG
becomes the “enlargement” we offer of each image and which on most systems
takes a few seconds longer to load for viewing). The assistant then derives a100-
dpi JPEG (which will become the inline image, i.e., the image in our Object View
Page) from the color-corrected 300-dpi JPEG using ImageMagick, software that
enables the batch processing of image files from the UNIX command line. Users
are thus presented with an inline image at 100-dpi, which is fine enough for most
purposes and requires graphics files of modest size that facilitate downloading and
movement from image to image. And they have the option to view an enlargement
at 300-dpi for the study of details.

Scanned images less than 9.5

× 7 inches – which includes all the illumin-

ated works – are scaled at 1:1 so the inline image will display at true-size on a
monitor with a 100-dpi screen-resolution. Its enlargement will be approximately

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

37

three times the original. Images between 9.5

× 7 and 16 × 12 inches are scaled

at 1:1 to produce enlargements three times the original, but the inline images are
resized manually so that they can be viewed in their entirety without scrolling;
users wishing to see it at true size can increase its size manually till it matches
the measurements of the original, which are given under the image. Images larger
than 16

× 12 inches are scaled at 1:2/3 to produce enlargements two times the

size of the original, and images larger than 24

× 30 inches are scaled at 1:1/2 to

produce enlargements one and a half times the original, and the inline images in
both categories are resized manually. Though only two times or less the originals,
on screen these enlargements provide enormous amounts of detail.

As part of the scanning process for each image, a project assistant completes a

form known as an Image Production (IP) record. The IP records contain detailed
technical data about the creation of the digital file for each image, including owner
institution, film stock, type, generation, and date made, measurements of original
image, reproductive source, and output, the scaling ratio, file size, and scanner
hardware and software. These records, retained in hard copy at the project’s office,
become part of the Image Information record that is inserted into each image as
metadata. Each and every image in the Archive also contains textual metadata
comprising its Image Information record (see below).

The 300-dpi (compressed) TIFF images generated by the scans are each indi-

vidually color-corrected against the original transparency or slide while it is on a
color-corrected light box (the same one used in the photo shoots sponsored by the
Archive). As noted, the fidelity of the transparencies can be trusted, having them-
selves been verified for color accuracy against the original artifacts. We correct
the image using Adobe Photoshop and calibrated, hooded Radius PressView 17SR
and 21SR monitors. We now color-correct the TIFF instead of JPEG because the
lossless compression does not discard image data each time changes are made
to the file in Photoshop. This 300-dpi (compressed) color-corrected TIFF is then
saved on CD-ROM. The color-correction process, which takes upwards of thirty
minutes – and sometimes as long as several hours – for each of the Archive’s
3000 images is necessary to bring the color channels of the digital image into
alignment with the hues and color tones of the original artifact. This is a key step
in establishing the scholarly integrity of the Archive because, although we cannot
control the color settings on an individual user’s monitor, the color-correction
process ensures that each image will match the original artifact when displayed
under optimal conditions (which we specify to users).

Images need to be “color corrected” because the very process by which they

are digitized results in a loss of color quality. Color correcting compensates for
this loss by changing an image’s shadows, midtones, and /or highlights, hue, and
saturation so that the colors of the final output, whether print or screen, conforms
to the original – or original idea of the graphic artist. Color correcting images is a
form of editing, and as such is work requiring decisions best made by an editor and
not scanning assistants. Our aim is accuracy, to make reproductions appear as close

38

JOSEPH VISCOMI

as possible to what they reproduce, but the means of doing so may seem overly
subjective, of requiring too much human intervention. To assume that machines –
camera and scanner – yield more objective results of and by themselves, however,
is mistaken. At the very least, the color cast created by the scanning process
(true of even high-end drum scanners used by fine-art printers) must be removed
and the image sharpened with “unsharp mask.”

10

Most likely it will need more

work, though only a handful of Photoshop’s numerous features, such as curves,
levels, unsharp mask, color balance, and hue and saturation, will be employed.
But one also needs to master the temptation to “improve” the original aesthetically
– eliminate foxing and other blemishes, for example, or increase or decrease the
saturation of certain colors – “restore” it – bring a two hundred year old faded
wash drawing to its presumed original pristine condition. While absolute accuracy
is not possible – not even for Photoshop – the electronic medium combined with
the skills and good eyes of an editor can create trustworthy reproductions.

But good eyes and skill in Photoshop will be for naught if the quality of the

reproductive source is compromised. As noted, we prefer to use 4

× 5 inch trans-

parencies instead of 35mm slides, in part because the former comes with color bars
and gray scales, has a smaller scaling ratio, and is what photographers working for
museums and libraries routinely provide for color reproductions. Slides, on the
other hand, while technically transparencies, are not meant for publication. Indeed,
all the institutions we deal with prohibit their reproduction and make them for
study, or, more exactly, reference purposes only. A similar situation between high
fidelity and reference exist in digital reproduction. Many images of art works on the
Web today are from slides scanned at 72-dpi; these can give a sense of composition
or design, but cannot be used for scholarly and editorial work, no more than Dover
reproductions can (see below). Images from slides, even when scanned at very high
resolutions – which results in files of enormous size – cannot match the quality
of images scanned from transparencies because they have a far lesser amount of
photonic data.

A comparison between a transparency and a slide, both scanned at 300-dpi,

scaled at 1:1 to the original, and shown at 100%, the size of our enlargements, is
most revealing. In a bit of “fearful symmetry,” the title plate to Blake’s The Book
of Urizen copy C is here reconstructed with a left side from a transparency and
a right from a slide (Figure 6). The transparency captures the texture of the paint
and subtle gradations, missing in the slide. The two versions of Blake’s America a
Prophecy copy O plate 13 are even more telling, because here transparency (left)
and slide (right) are uncorrected raw scans. The transparency comes in sharper and
with more information, requiring less manipulation than the slide (Figure 7).

The difference between the two remains as pronounced after they have been

color corrected and sharpened (Figure 8), though both are far superior to the middle
image, which is from a slide scanned at 72-dpi but enlarged 400% to provide detail
analogous to the two 300-dpi scans. Unlike them, however, it badly pixelates at this
size and distorts its information.

11

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

39

Figure 6. The First Book of Urizen copy C, title plate, reconstructed from transparency (left)
and slide (right) scanned at 300-dpi, detail.

Figure 7. America a Prophecy copy O plate 13, uncorrected transparency (left) and slide
(right) scanned at 300-dpi, detail.

40

JOSEPH VISCOMI

Figure 8. America a Prophecy copy O plate 13, uncorrected transparency (left) and slide
(right) scanned at 300-dpi, with image from slide scanned at 72-dpi (middle), detail.

Trying to coax detail out of a 72-dpi scan is like using a magnifying glass

to examine a photomechanical reproduction, which, in place of pixels, yields the
interfering dots of the photographic screen (Figure 9).

Unless one examined originals with slide loops (small but powerful magnifying

glasses for 35mm slides placed directly on the artifact and thus unacceptable to
curators and librarians) or conventional magnifying glasses, one had no choice but
to resort to reproductions and facsimiles for detail. The original Blake Trust facsim-
iles, produced through a combination of collotype and hand-painting through
stencils, have no screen, but, when magnified, do reveal visual “boundaries”
between colors (the result of using stencils). Indeed, the Archive’s images have
achieved greater color fidelity in some notable instances – for example, several
plates in the Blake Trust’s facsimile of The Book of Urizen, copy G (left) have a
reddish hue not found in the originals. In this respect, the images in the Archive
(right) are truer to the tonality of the originals (Figure 10).

Our 300-dpi (JPEG) enlargement yields superb detail for close inspection of

printing and coloring, enabling one to study paint layers, differentiate what was
printed from plates and what added to impressions, examine emendations in manu-
scripts (Figure 11) and the tiny hieroglyphic like figures and decorations between
lines of text (Figure 12).

They provide the raw data for editing, enabling one to compare multiple

versions of texts and take on those most puzzling of all marks, the ones simultan-
eously part of the text and design. For example, in plate 4 (lines 13–16) of Songs

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

41

Figure 9. The Book of Urizen copy G, plate 5, photographic screen of Dover Publication
reproduction (left) and Archive digital image (right), detail.

Figure 10. The Book of Urizen copy G, plate 5, collotype and stencilled coloring of Blake
Trust facsimile (left) and Archive digital image (right), detail.

42

JOSEPH VISCOMI

Figure 11. Island in the Moon manuscript, page XX, detail.

Figure 12. The Marriage of Heaven and Hell copy D, plate 11, hieroglyphic-like figures
between lines 13 and 15, detail.

of Innocence and of Experience (Figure 13), is the mark after “write” punctuation
(comma or fat period) or tendril – or both?

After “sight” and “reed,” Erdman records a colon and a period, though the latter

is more oblong than the former. In copy F (top), Blake colored the tendril yellow
but left the tip green in the shape of a period, forming a colon after the fact. A happy
accident? More challenging are designs like that of plate 13 of Songs (Figure 14)
which force one to ask if figures or parts of them can function as stops, whether
non-linguistic marks belong to the linguistic code.

Enlargements of good digital images can yield more information than the

originals, as can enhancements, which can reveal such subtleties as a black-inked
signature faded into a black wash, or, when creatively “deformed,” as demonstrated
by Jerome McGann, yield hidden structural information.

12

Often as important as

enlargement is comparison; in the Archive one can, with almost all the illuminated
books, immediately check an image in one copy against other copies of that book,

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

43

Figure 13. Songs of Innocence and of Experience copies C (bottom) and F (top), plate 4, lines
13–16, detail.

Figure 14. Songs of Innocence and of Experience copy C, plate 13, lines 5–8, detail.

which has never been possible with the printed record and rarely in any one real
collection.

Reproducing images more accurately than print, images not in print, and

multiple copies of images, makes the Archive the first place to stop when studying
Blake. The Archive’s scholarly integrity, however, also requires that its images be
true size, because scale can be a significant aspect of the experience and meaning
of an object. We grant structural priority to the physical object by accounting
archivally and editorially for the original size of Blake’s works, whether plates,
paintings, drawings, manuscripts, or printed pages. We have done that in two
ways, by displaying the actual size of every object directly beneath it, and by
providing ImageSizer, a sophisticated image manipulation tool, available from
every Object-View Page. A new Java applet developed at IATH with the Blake
project in mind, ImageSizer can resize images on the fly at the size of originals on
monitors whatever their resolution. It is this feature in the Archive that makes our
images digital “facsimiles.”

ImageSizer’s principle function for the Archive is to allow users to view Blake’s

work on their computer screens at its actual physical dimensions. Users may invoke

44

JOSEPH VISCOMI

Figure 15. ImageSizer’s calibration applet setting a “cookie” informing the ImageSizer of
user’s own unique screen-resolution.

the ImageSizer’s calibration applet to set a “cookie” informing the ImageSizer of
their own unique screen-resolution (Figure 15).

Users hold an actual ruler up against the one on the screen and use the slider bar

to align the measurements of both. Screen resolution thus determined, the applet
automatically resizes the images (which have already been scaled to display at true
size on a 100-dpi screen) to appear at true size on the user’s monitor. Based on
this data (recorded in the cookie), all subsequently viewed images will be resized
on the fly so as to appear at their true size on the user’s screen. If a user returns
to the Archive at some later date from the same machine (and the resolution has
not been changed), the data stored by the cookie will remain intact, and there will
be no need to recalibrate. Users may also set the ImageSizer’s calibration applet
to deliver images sized at consistent proportions other than true size, for example,
at twice normal size (for the study of details). In addition, the ImageSizer allows
users to enlarge or reduce the image within its on-screen display area, and to view
the textual metadata comprising the Image Information record embedded in each
digital image file (Figure 16).

The Image Information record combines the technical data collected during the

scanning process from the Image Production record with additional bibliographic

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

45

Figure 16. Textual metadata comprising the Image Information record embedded in each
digital image file.

documentation of the image, as well as information pertaining to provenance,
present location, and the contact information for the owning institution. These
textual records are, at the most literal level, a part of the Archive’s image files.
Image files are typically considered to be nothing but information about the images
themselves (the composition of their pixelated bitmaps, essentially); but in practice,
an image file can be the container for several different kinds of information. The
Blake Archive takes advantage of this by slotting its Image Information records
into that portion of the image file reserved for textual metadata. Because the textual
content of the Image Information record now becomes a part of the image file
itself in such an intimate way, this has the great advantage of allowing the record
to travel with the image, even if it is downloaded and detached from the Archive’s
infrastructure. The Image Information record may be viewed using either the “Info”
button located on the control panel of the Archive’s ImageSizer applet or with the
Text Display feature of standard software such as Adobe Photoshop or X-View.

Also from the Object View Page, a user can evoke Inote, the Archive’s other

Java applet. Inote is an image-annotation tool which permits us to append textual
notes (“annotations”) to selected regions (or “details”) of a particular image; these
annotations are generated directly from the SGML-encoded illustration descrip-
tions prepared by the editors. Inote functions most powerfully when used in

46

JOSEPH VISCOMI

Figure 17. Inote used in conjunction with the Archive’s image searching capabilities, where it
can open an image found by the search engine, zoomed to the quadrant of the image containing
the object(s) of the search query, with the relevant textual annotation displayed in a separate
window.

conjunction with the Archive’s image searching capabilities, where it can open an
image found by the search engine, zoomed to the quadrant of the image containing
the object(s) of the search query, with the relevant textual annotation displayed in
a separate window (Figure 17). From there, Inote allows the user to enlarge the
image for further study and/or to access additional annotations located in other
regions of the image. Inote may also be invoked directly from any of the Archive’s
Object View pages, allowing users to “browse” the annotations created for a given
image.

3. Conclusion

Blake’s illuminated works are widely dispersed and more and more severely
restricted as a result of their value, rarity, and fragility. Studying them, at both
the introductory and advanced levels, requires accurate reproductions of as many
copies of each book as is possible. One might wish to argue that translating Blake’s
analogue hand into the discreet digits of types for the purpose of publication merely
recreates the relation between manuscript (fair copy) and letterpress, and hence

DIGITAL FACSIMILES: READING THE WILLIAM BLAKE ARCHIVE

47

appropriate for texts from the period. Such an argument, though, would not only
ignore Blake’s intentions, but it would also dismiss as meaningless the historical,
technical, and aesthetic information provided by the poem’s form. Though not
reproducible in the digital technology of his day, and unevenly so by the dots
per inch that followed, Blake’s pictorial poetry can be accurately reproduced in
today’s digital technology when imaging protocols such as those outlined here
are followed. For most scholarly purposes, the resulting reproductions will more
than suffice, or, at the very least, adequately prepare one for examining originals,
those not reproduced in the Archive and those for which bibliographic information
necessarily requires handling the artifact.

The Archive’s exceptionally high standards of site construction, digital repro-

duction, and electronic editing have made possible reproductions that are more
accurate in color, detail, and scale than the finest commercially published repro-
ductions and facsimiles, and texts that are more faithful to Blake’s own than
any collected edition has provided. We have applied equally high standards in
supplying a wealth of contextual information, which includes full and accurate
bibliographical details and meticulous descriptions of the content of each image.
Finally, users of the Archive can attain a new degree of access to these works
through the combination of powerful text-searching and (for the first time in any
medium) advanced image-searching tools that are made possible by the editors’
detailed image descriptions and innovative software.

We are hoping, of course, that the Archive, once extended to encompass the full

range of Blake’s work, will ultimately set a new standard of accessibility to a vast
collection of visual and textual materials that are central to an adequate grasp of
the art and literature of the eighteenth and nineteenth centuries. But we have also
come to see the Blake project as a pacesetting instance of a fundamental shift in
the ideas of “archive,” “catalogue,” and “edition” as both processes and products.
Though “edition” and “archive” are the terms we have fallen back on, in fact we
have envisioned a unique resource unlike any other currently available – a hybrid
all-in-one edition, catalogue, database, and set of scholarly tools capable of taking
full advantage of the opportunities offered by new information technology.

Notes

∗

I am grateful to my co-editors on the Archive, Morris Eaves and Robert Essick, and to our technical

editor, Matthew Kirschenbaum, for their contributions to our About the Archive documentation pages
<http://www.iath.virginia.edu/blake/public/about/>, a few paragraphs from which have been reused
in this article.

1

Grant, John E. and Mary Lynn, Johnson, eds., Blake’s Poetry and Designs. Norton Critical Edition,

W.W. Norton, New York, 1979, pp. 49–50.

2

According to Professor Paul Zall, who examined all editions of Granger’s Index to Poetry, Blake’s

“Tyger” was the most anthologized poem.

3

See, for example, Jerome McGann’s Black Riders, Princeton University Press, Princeton, N.J.,

1993 and Textual Condition, Princeton University Press, Princeton, N.J., 1991, and Joseph Viscomi’s
Blake and the Idea of the Book, Princeton University Press, Princeton, N.J., 1993.

48

JOSEPH VISCOMI

4

Keynes, Geoffrey, ed. Blake: Complete Writings with Variant Readings. 1966. Repr. with

corrections, Oxford University Press, Oxford, 1979, pp. 214.

5

Erdman, David V., ed. With Commentary by Harold Bloom. The Complete Poetry and Prose of

William Blake. 1965. Rev. ed., University of California Press, Berkeley, 1988, pp. 24–25.

6

Rule of thumb: reproductions get reproduced. Scholars will reproduce images they know through

reproductions, and, for economic reasons, prefer to acquire negatives or transparencies on file than
to incur new photography costs. The Archive addresses this problem by greatly enlarging the pool
of reproductions, clearly identifying the image’s owner institution or collection, and informing users
how to acquire permission to reproduce.

7

Only in the past six years, largely as a result of the Archive’s editors’ extensive research, has the

history of the production of the illuminated books been correctly understood. That in turn has made
possible for the first time a sound scholarly archive, including numerous copies of the illuminated
books that have been neglected because their place in the history of production was not understood.

8

We must recognize, though, that they are, however accurate, necessarily approximations, simply

because any transcription of Blake’s irregular etched texts into the uniformities of conventional print
is at best a translation. In printed editions, differences are relegated to the editorial apparatus. In the
Blake Archive, users can easily compare the texts of different etched copies side by side.

9

Current versions of Microtek’s ScanWizard software (3.1.2) are used with the Microtek scanners;

the Nikon runs version 4.5.1 of its supporting plug-in. The Microtek III flatbed and the Nikon slide
scanner are attached to a Macintosh Power PC; the Microtek Scanmaker V is attached to a Macintosh
G3; the Microtek 35t Plus is attached to a Macintosh G3. All three Macs run OS 8.1.

10

Unsharp mask is a film compositing technique used to sharpen edges on an image. It corrects

blurring in the original source and compensates for blurring that occurs in sampling and printing
processes. It is recommended whether output will be print or online, though the sharpening effect
is more pronounced in online image (on-screen) than in high-resolution output (print). The unsharp
mask filter locates every two adjacent pixels with a difference in brightness values that you specify
and then increases the pixels’ contrast by the amount you specify. You also specify the number of
surrounding pixels to which the sharpening effect is applied.

11

The image (America plate 13) reproduced is 23.7

× 17.2 cm, input size (the image on the trans-

parency) is 10.2

× 7.4 cm, and scaling is 231%, producing an output of 23.7 × 17.2 cm and a 22.7

MB TIFF and a 1.8 MB JPEG. The slide, also scanned at 300dpi, but with an input size of only 2.7
× 2.0 cm, requires a scaling of 860% to make an output of 23.7 × 17.3 cm, a 23.2 MB TIFF and a
1.5 MB JPEG. The slide scanned at 72 dpi, same scaling, produces a 1.3 MB TIFF and a 323k JPEG.

12

For signatures, see Viscomi, Joseph, “A Breach in a City, the Morning After the Battle: Lost

or Found?” Blake/An Illustrated Quarterly (Fall, 1994), pp. 50–53. McGann’s demonstration of
“deformation” was presented as part of the session chaired by Matthew Kirschenbaum, “Redefining
Our Notions of What (Digital) Images Really Are,” Association for Computers in the Human-
ities/Association for Literary and Linguistic Computing, joint international conference, University
of Virginia, Charlottesville, Va., 6/10/99.