University of Virginia Library


Page 1

The Textual Criticism of Visual
and Aural Works


Textual criticism—the study of the relationships among
variant texts of works—has primarily been associated,
throughout its long history extending back to antiquity, with
verbal works as transmitted on tangible objects such as
parchment and paper. But all works, whether constructed of words or
not, have had histories that—if fully told—would reveal stages of growth
and change, reflecting not only their creators' intentions but also the ef-
fects of their passage to the public and through time. All works, in other
words, have textual histories. Whether or not one chooses in every case
to use the word "text" to refer to the arrangement of elements that make
up a work is irrelevant; the point is that the issues and problems dealt
with in the textual criticism of verbal works have their counterparts in the
study of all other works.

One reason that this point has not been as widely acknowledged as
it ought to have been is perhaps the fact that the varying media used
in different arts affect the nature of the accompanying scholarship. The
textual criticism of verbal works, for example, often leads to the produc-
tion of scholarly editions, which embody insights derived from the study
of textual history. Because the medium of verbal works—language—is
fundamentally intangible, a work can be represented by a text in a newly
produced physical object (the one conveying the scholarly edition) with-
out making any alterations to the historic artifacts that had transmitted
earlier texts of the work to the present. But a work in a tangible medium,
like a painting or a sculpture, cannot as freely be accorded scholarly edi-
tions, since any alteration deemed appropriate by the editor would per-
manently alter the artifact that uniquely is the site of the intended work
and thus would deprive the future of some of the evidence that had been
available to the editor.

Regardless of whether textual scholarship leads to editions or to es-
says, the medium employed in each art determines the nature of the
evidence available for reconstructing textual history. (I am not speaking
of the quantity of evidence, which can vary irrespective of the medium.)


Page 2
The evolution of the text of a painting, for example, may be attested to
by sketches, which are analogous in some respects to the drafts of a verbal
work. But, being made of a physical material that is applied to a physi-
cal surface, a painting may preserve earlier stages of its text that exist on
the same surface beneath the layer of paint that is now on top, and these
stages can be revealed, in varying degrees, by modern technology. Some
works in intangible media, such as literature and music, can be transmit-
ted either by direct imitation or by tangible aides-memoirs. When a textual
tradition is entirely of the intangible kind, knowledge of a work's history
is dependent on memory; when texts are passed along in physical objects,
there is direct access to certain moments of the past (the moments when
the documents were prepared), but the texts thus reported are not neces-
sarily more trustworthy than those carried forward in human memory.
The objects, however, inevitably carry traces of their own manufacture;
and when those clues are uncovered through analysis, they can help ex-
plain how the text came to be constituted as it is. This procedure is the
counterpart to scrutinizing a canvas for underlying layers of paint: the
examination of objects can be as crucial to reconstructing textual histories
of works in intangible media as it is for those in tangible media. Textual
critics must of course assess whatever evidence is available to them, but
the process must take into account the fact that some evidence may be
transmitted by media different from the medium of the work itself.

Furthermore, the textual history of a work proceeds beyond the point
at which the creator or creators of the work die or cease to make changes
in it. The discipline of textual criticism has traditionally focused on the
evolution of texts only up to that point, though the evidence often has
to come, by default, from later documents. But the alterations that texts
undergo as they are disseminated to the public, both during and after
their creators' lifetimes, are part of the full story of the life of any work
(as textual critics have increasingly come to recognize). Such alterations
take different forms for works in different media. Since works in tangible
media exist in their originally created form only at the geographical loca-
tions where they are placed at any given time, they are often reproduced
in order to allow people to experience them, at least to some extent,
without traveling to those locations. Such reproductions cannot dupli-
cate precisely the physical material of the original, and they frequently
use different media, as when an oil painting on canvas is reproduced by
a photographic process on paper (inevitably with differences in coloring
and perhaps with different dimensions as well). Many more people may
be familiar with an image in its reproduced forms than have ever seen it
in the original, and those textually different forms are therefore relevant
to the historical study of a work's reputation and influence. In the case
of verbal works transmitted on paper, their physical forms were not, in


Page 3
the large majority of instances, regarded by their creators as parts of the
works. But because the physical features of books reflect contemporary
culture and frequently have an effect on readers, those features become
textual elements for the study of cultural change and reception history.

None of these points is surprising: they follow naturally from the rec-
ognition that some works (such as paintings, sculptures, and buildings) are
basically made of tangible materials and that others (such as literature,
music, dance, and film) are basically intangible in nature. In practice,
works may consist of mixed media—not merely in the sense that differ-
ent ingredients associated with a single art form are brought together
(as when watercolor and crayon are used in the same work) but, more
significantly, in the sense that elements of different arts (including those
with a different ontological status) are made parts of a single work (as
when words and music, or words and visual effects, are joined). The tex-
tual study of any work, mixed or not, can proceed more thoughtfully and
logically if it is conducted with an awareness of the relationships among
the media employed by all arts and the conditions set by each medium in
regard to textual change.

In what follows, I shall look at a variety of arts and comment on some
of the similarities and distinctions among them from the point of view of
textual criticism. Although I outlined these considerations in A Rationale
of Textual Criticism
(1989), I wish now to expand the discussion in the hope of clarifying some of the issues that are often raised, sometimes confus-
ingly, in popular discourse on the arts. (My examples come from the New
York Times, to which all parenthetical dates below refer.) I hope that these
comments, which are intended to be no more than suggestive, can begin
to show how a framework might be formed for thinking about the textual
histories of human creations in all media.

Work in Intangible Media

1. Verbal Work

Works constructed of verbal languages are a useful starting point be-
cause they have been so extensively discussed in the literature of textual
criticism that for many people they constitute a natural frame of refer-
ence for textual questions in general. Yet (or perhaps not surprisingly) the
medium of language is the most mysterious of all media. It would not be
so if it were unquestionably oral, with tangible texts being only the re-
cords (like musical scores) by which the vocal work is recreated. Not only
is it possible, however, to have languages without oral components but
also—and more importantly for present purposes—it is common for au-
thors to expect their works to be read silently, without the spoken sounds
of the words being formed imaginatively in the mind. Of course, some


Page 4
writers (and not only poets) do wish the sounds to be heard internally;
but even in those cases, or most of them, the authors feel that the works
have been experienced in their intended medium whether read silently or
aloud. Whatever may be said about the origins of language, in practice it
is a medium that can exist apart from speech. But its written presentation
on a physical surface is not its primary form either: most writings, from
the point of view of their creators, are not visual works, for the works still
exist when presented in different handwriting or typefaces on different
papers. However difficult it is to define the medium of language, one thing
is therefore clear: the medium is intangible. Like all works in intangible
media, verbal works require performance of some kind (oral or internal)
for their realization; and because performances have duration, such works
consist of sequential texts, intended by their creators to be experienced
from a beginning to an end.

One of the problems that these conditions pose for textual criticism is
the likelihood of confusing the work with its means of transmission. When
verbal works are communicated on paper, readers (and even, at times,
the class of readers called textual critics) may think of the works as what
they see rather than as what they silently perform. They may automati-
cally correct a typographical error, for example, without realizing that
other places in the text may also be erroneous (from the point of view of
the author's or the publisher's intention). What textual critics and other
readers have available to them in documentary texts are various attempts,
from various past moments, to provide instructions for recreating a work.
The fact that words and punctuation on paper are perceived by the eye
does not in itself make the text a work of visual art, for the artistry of the
physical design is not normally a part of the intended verbal work. (Punc-
tuation is no more visual from this point of view than the words; it, like
them, is an element of linguistic meaning, transferrable into pauses and
intonations in recited texts.) Authors do at times, however, choose to use
visual effects, creating mixed-media works that employ both a tangible
medium (since the pictorial images and spatial arrangements exist in ink
on paper) and an intangible one (language). The relative proportions of
each medium can vary: at one end, for example, are primarily verbal
works in which some of the text is shaped (like George Herbert's bird-
wings and Lewis Carroll's mouse-tail, or indeed stanza indentations and
spacing), and at the other are primarily visual works in which language is
simply a significant element (like concrete poetry).

Textual criticism, which elucidates these conditions, controls the op-
tions for editors: in the first of these situations, new editions (with critical,
reset texts) are possible so long as the shaping is maintained; in the sec-
ond, where the work is totally dependent on physical presentation, only a
facsimile reprinting could be considered (and not even that if the quality


Page 5
of the paper were a part of the work and one were concerned with pre-
serving authorial intention). There is no doubt that the physical presen-
tation of a verbal work conveys information about the cultural milieu in
which the document was produced and can have an influence on readers,
regardless of whether any visual effects were intended by the author. But
many readers have always been interested in authorially intended texts
and make an effort to exclude documentary features that they suspect
were not so intended; thus authorial intention is a factor in the history
of reading. Textual criticism properly studies all these interchanges over
time, but any editions that result will inevitably follow different guidelines
according to whether the focus is on authorial intention or social-cultural

The intended texts of a verbal work and the varying physical texts
aimed at conveying that work are all worthy of study, but understanding
that "the texts of works" and "the texts of documents" are not synony-
mous phrases is central to clear thinking about the texts of works made
of language. This point applies equally to verbal works transmitted in
oral form (as in nonliterate societies, but not only in them), for each oral
rendition is a "document" and not necessarily the work. When the texts
of such performances are written down (as they often are at some point,
if only by anthropologists or other historians), a tangible documentary
record becomes available for analysis by textual critics, supplementing
any ongoing oral tradition.

2. Music

The points that can be made about the nature of verbal works are
directly applicable to thinking about music, even though the medium of
music, sound, is less enigmatic than the ontology of language. What is
immediately comparable about the two is that in each case the notation
present on paper provides instruction for some kind of performance and is
not in itself—except in unusual instances—part of the work of art. Com-
posers do sometimes create scores that are meant to have visual appeal,
but the result is not necessarily a mixed-media work, unless the composer
intended that anyone listening to a performance of the music should also
be looking at the score. Nevertheless, such scores limit the freedom of edi-
tors to alter the design of the notation, in much the same way that shaped
verbal texts do. Another similarity between musical scores and verbal
texts on paper is that each can be the basis for a silent performance as
well as a rendition out loud. Many people are adept at "hearing" music
by reading scores, though when they do so, they are not experiencing the
work in the medium in which its creator intended it to be experienced
(unlike the silent readers of verbal texts). No two performances, silent or
aloud, can ever be precisely identical, for the most that a score (like a


Page 6
tangible verbal text) can do is to provide a framework that encompasses,
and delimits, a variety of performances.

The composer Richard Wernick once said, "Conductors and compos-
ers are privileged in that our training has given us the capacity to hear a
piece of music as we see it on the written page. But, unfortunately, there
is an intermediary between the music and the public." This comment was
criticized as "silly" by a writer for the New York Times (John Rockwell,
on 17 December 1989), but one understands what Wernick was getting at:
every performance may depart from what the composer intended. To me,
the problem with Wernick's statement is not the word "unfortunately"
(which bothered Rockwell) but the idea that the musical work is present
on paper and that those trained to read scores can experience "the mu-
sic" directly. Will every trained reader of a score of Wernick's "hear" the
music exactly as Wernick does? Will even Wernick "hear" it the same way
every time he looks at it? In other words, the reader of a score is an "in-
termediary" just as much as a person performing aloud. It is in the nature
of works in intangible media that, however determinate the instructions
for their recreation may be, the texts of the works themselves are always
indeterminate. Textual critics of music have available to them not only
written and printed texts but also evidence of specific performances, both
in anecdotal accounts and in sound recordings. All this material is essen-
tial for following the textual history of a piece of music over time, but only
a part of it is relevant to the study of authorial intention. Editors of music
have to realize, along with editors of verbal works, that no single text can
accommodate all aspects of textual history simultaneously.

And many of those who use or discuss editions of music need (like
readers of verbal editions) to learn more fully what the function of a
scholarly edition is. When Donal Henahan commented (20 March 1983)
on the first volume ( Rigoletto ) of the University of Chicago critical edition
of Verdi, he noted that not every "discovery or interpretation" in the edi-
tion will be "honored" by every performer. "Interpreters," he said, "will
always want to slip the bonds. The eternal struggle between purity and
practicality is not merely inevitable but one of the dynamic forces that
keep the musical world going around." This is a peculiar way to talk about
a scholarly critical edition, especially since Henahan recognizes that the
text is a product of critical judgment throughout and that the editorial
annotation calls attention to many unresolved cruxes. To speak of "slip-
ping the bonds" when one does not "honor" the text, which suggests that
one is being a renegade, fails to acknowledge a primary aim of a scholarly
edition: to give readers evidence for reconsidering the text. The poles of
"purity" and "practicality" (if they are poles) have no meaning in relation
to texts of music, or other works in intangible media. Accepting any text


Page 7
(scholarly or not) unquestioningly is not being "pure" but is simply being
negligent; and departing from any text as a result of critical evaluation
is not being "practical" but is simply being responsible—it shows an ac-
ceptance, that is, of the responsibility that follows from understanding the
status of texts as fallible instructions for the recreation of works. Whether
one's goal is a stage of authorial intention (as in the Verdi edition) or a
stage of a work's later development, the makeup of the tangible text is a
matter of judgment, as is the range of performance that seems justified
by it.

Distinguishing between authorial intention and authorial expectation,
which I have often found useful in discussing textual questions relating to
verbal works, can also help to clarify textual dilemmas posed by music.
Thus the amount of interpretive variation and outright embellishment
engaged in by performers is a matter of convention that has differed from
one period to another. Composers of the past naturally expected their
work to be performed in whatever manner was conventional at the time,
but their own textual intentions may have been different. (There is, for
example, reason to believe that Verdi came to disapprove of some of the
embellishments regularly added by singers in his time.) To the extent that
evidence is available, textual criticism and scholarly editions should delin-
eate intended texts as well as those that were actually heard by listeners.

Furthermore, the concept of expectation helps to approach the issues
argued about in the "early-music" debates of recent decades. Composers
clearly expected their works to be performed on the instruments available
in their own time, and they probably thought of the sounds produced by
those instruments when they read music silently. But does it follow, as
"early-music" adherents claim, that one is not being faithful to authorial
intention if one listens to music played on instruments developed later
than the composers' lifetimes? Not necessarily, if one regards composers'
expectations about instruments as a separate matter from their intentions
to have certain notes and interpretive instructions in their scores. The
differing sounds of instruments made at different times are (like varia-
tions in tempo and volume) just one of the variable aspects of individual
performances. As the pianist Malcolm Bilson has pointed out (14 June
1998), instruments have always "played a role in determining interpreta-
tion," and performing a work on instruments of different periods is one
"avenue" for exploring what the music has to offer.

The use of modern instruments for earlier music is not analogous to
the modernizing of verbal texts, an issue that has generated equally ve-
hement debate. Such modernizing normally refers to spelling and punc-
tuation, and the equivalent in music would be the modernizing of the
notation in scores; neither practice is defensible in scholarly editing since


Page 8
it deprives readers of textual elements that may be (and often are) relevant
to interpretation. For verbal works, the equivalent of modern instruments
would be modern pronunciation, which also is a feature of performance
encompassed within the framework that is provided by the set of printed
or handwritten instructions (a point not contradicted by the occasional
presence of rhymes or puns that require earlier pronunciation). The aural
texts of performances (including the nature of the sounds produced by
varieties of the "same" instrument) are thus, for the most part, materials
for social textual criticism (dealing with public texts and their reception),
whereas the study of authorial intention is primarily focused on visible
(notated) texts—at least when there is a written tradition that can be con-
nected with the creator(s).

But of course some music, like some verbal work, does not have a
written tradition. Folk music of the past may have been transmitted by
imitation over a long period of time before being written down, and at
present there is music that has not yet been written down (as there was at
any given moment of the past). Performances of pieces that originated in
the past but that are not yet reported in written form or in sound record-
ings (or for which there is an independent living tradition) have a different
status for textual criticism from performances in the present by the person
or persons who are creating or improvising the music as they perform it.
Performances of the first kind provide evidence for recreating intended
texts, whereas those of the second kind constitute the intended works
themselves. Thus jazz, often created in performances of this second type,
is analogous to such verbal (or partially verbal) works as those produced
by performance artists and story-tellers.

The comparison between works of music and those of verbal language
brings to mind another connection: verbal texts are often combined with
musical ones (as in songs and opera) to form mixed-media works. The fact
that in vocal music some sounds are produced by human voices is not
what makes such works mixed, for the voices are simply musical instru-
ments; rather, the mixed-media status of these works is produced by the
presence of words that have meaning apart from the music. (Opera is of
course mixed in a further way because of its use of the physical art of stage
sets and costumes.) Textual criticism of these verbal texts can be pursued
as one would any other verbal text; and, in the case of those written as
separate works (with no thought of their being set to music), variants from
the original texts cannot be considered errors if they were the intended
readings of the creators of the musical-verbal works. The many similari-
ties between literature and music underlie the fact that in musicology as
in literary scholarship there is a tradition of textual study; and many of
the scholars currently engaged in editing music are well versed in the
theoretical writings that editors of literature have produced.


Page 9

3. Dance

A work of dance may of course be transmitted by direct instruction
from one choreographer or dancer to another. This kind of transmission,
like the oral transmission of verbal works, presents at any moment the lat-
est stage (or one of the latest stages) in a continuously evolving tradition.
This social product may be judged aesthetically satisfying and is certainly
of importance for cultural history. But another legitimate interest—in the
intentions of the creator(s) of the work—may in many instances be more
accurately conveyed by a physical document, even if the earliest surviving
one does not go back to the time of the creator(s), since the text of any
given document is relatively stable (barring unusual physical deteriora-
tion). In any case, the physical recording of dance texts, like that of music
texts, is obviously desirable. But there is a major difference in the histori-
cal development of such recording in the two arts, for symbolic dance
notation on paper (as opposed to written verbal accounts or printed step
tablature) goes back only to 1700 (the Beauchamp-Feuillet system), and
no one notation system even today has such widespread acceptance as
the now conventional music notation does. It is instructive to look at the
illustrations in Mary Ann Malkin's 2003 Dancing by the Book, which give a
sense of the variety of systems proposed in the eighteenth century.

At present, Labanotation, a system introduced in the 1920s, may be
regarded as the dominant form, having been championed by the Dance
Notation Bureau, which possesses over seven hundred Labanotated scores
(see Erika Kinetz's article, 7 November 2005). The Benesh system, how-
ever, has been favored by the Royal Ballet in London. Although stan-
dardization is helpful, any comprehensible notation is important for the
preservation of individual works. Nijinsky, for example, who was much
interested in the problem, created his own notation and on more than
one occasion wrote out a score for L'Après-midi d'un faune (as reported by
Jennifer Dunning, 9 December 1989). Textual critics of dance may in
some cases have variant texts on paper at their disposal; and any dance
score—like any piece of musical or verbal notation—may contain erro-
neous readings, identifiable both through an intimate knowledge of what
is stylistically plausible for the choreographer in question and through
recourse to such external sources as reviews and other contemporary

If the search for appropriate notation has been a significant strand
in dance history, another—not surprisingly—has been experimenta-
tion with technology that can record movement. For many decades now,
dances have routinely been filmed or videotaped, often from multiple
angles. More recently, digital photography has been used, and there has
been research on software (such as LabanDancer) to convert scores into
animation. Computer programs are used by some choreographers as an


Page 10
aid to composition; and stages of their work, if saved, can give textual
critics access to drafts. Motion-capture technology can create a digital
record of a performance in a sequence of shifting lines and shapes, which
can in turn be given vestigial human form. Trisha Brown's how long is an
example of a dance piece that includes a video projection of a digital re-
cord of the movements being made simultaneously by the human dancers.
(It was discussed by Matthew Mirapaul on 13 April 2005.) A film record
of a performance of this piece would thus also preserve a second kind
of record of the movements involved. There have been choreographers
who regard pictorial representations of dance as perhaps more valuable
than scores. Martha Graham's admiration for certain still photographs of
dancers is well known; and Murray Louis recently expressed satisfaction
with the preservation of his legacy through videotapes of his repertory
(reported by Anna Kisselgoff on 17 December 2005).

One can understand how still and motion photography can be crucial
for the study of an individual dancer's technique. And of course the tex-
tual criticism of dance must take into account every kind of documenta-
tion that exists for any given work. But the idea that a visual record of a
dance might be superior to a score, or even render it unnecessary, simply
because the dance is a visual form of art, betrays a misunderstanding of
the relation of work and performance in the performing arts. Each visual
record preserves only a single performance, which is an interpretation of
a set of instructions; every performance is a postulation of the work, and
the instructions accommodate them all, except when an interpretation
goes beyond what can be regarded as implicit in the instructions. The
instructions themselves may be consciously altered over time, as differ-
ent generations adapt the work to their own sensibilities. But for anyone
interested in authorial intention, even a film of a performance directed or
performed by the choreographer has its deficiencies, since it still reflects
only one occasion and since (even with multiple cameras) it cannot show
everything that could be observed by viewers of the live performance.
The situation is precisely the same as the one that exists in music, where
no one performance can be a substitute for the score. Muriel Topaz, one
of the directors of the Dance Notation Bureau, made the point well at
a 1962 conference: a dance, she said, like works in any of the other per-
forming arts, has "a substance, a compositional integrity that transcends
the initial [and, one should add, any other] performance" (quoted in Jack
Anderson's obituary of her, 1 May 2003). That transcendent integrity is
distilled in the score or scores, which encompass various interpretations
reflected in performance.

Topaz's comment was concerned with the originating choreographer's
intention, but it could be applied to the revising vision of later choreogra-
phers who alter a piece. Pictorial evidence is of course basic for studying


Page 11
the history of performance and thus of reception. But for the history of
intended texts, as they changed over time because of shifting intentions
on the part of the creator of a piece as well as of others, scores are of
crucial importance, when they exist (though pictorial evidence should in
any case be scrutinized critically for clues to such intentions not available
elsewhere). Scores may contain music notation and words along with the
dance notation if the work includes music and/or words (sung or spoken),
as most works of dance do. The music and/or words may have their own
independent textual history, but once they are joined with dance, the cho-
reographer's intention for them takes precedence over earlier intentions,
and what might have been regarded as a textual error in the music or
the words under other circumstances would not be erroneous if intended
by the choreographer. Nevertheless, a dance score may contain errors in
any of its conjoined elements, and a knowledge of the textual history of
each of those elements gives an editor the basis for making emendations.
What may seem to be an error in the musical text may be shown to be
an error indeed by analyzing it in conjunction with the dance notation
at the same point. The situation is in principle the same as that faced by
textual critics of all mixed-media works, whether or not the elements in
them have separate histories.

4. Film, Video, and Digital Art

Confusion about the distinction between work and document in the
field of film is epitomized by Andrew Pollack's article (16 March 1998)
on the issues raised by digital film restoration. He points out that in an
opening scene of Gone with the Wind there is a gap between a flagpole
and the building to which it is supposed to be attached (showing that the
scene was not a natural shot but was produced by special effects, ineptly
handled). This error can be corrected digitally by scanning the frames
involved and then copying and inserting certain pixels. But Pollack raises
a question: "Should the flaw be fixed or retained as an intrinsic part of the
original masterpiece?" The trouble with his question is that the "master-
piece"—that is, the work—does not necessarily consist of everything that
is on the film The levitating flagpole is no more a part of Gone with the
(the work) than the typographical error "fastidions" (for "fastidious")
in the original American edition of Moby-Dick is a part of that work. Both
are obvious flaws in the documents that make possible the transmission of
those two works, but they are not parts of the works as intended by an
one at any time. A printed edition of a verbal work, consisting of multiple
supposedly (but not quite) identical copies, is analogous to the copies of
a film made for distribution to theaters: in both cases, the multiplicity of
copies allows for the simultaneous experiencing of the work at different
locations, even though there is no guarantee in either case that the cop-


Page 12
ies, or the performances based on them (the silent or oral reading in the
one case, the projection in the other), are the same. This point proves (if
proof is needed) that cinematic works employ an intangible medium: the
work is not a roll of celluloid but rather the sequence of moving images
(now usually supplemented by sound) that it enables us to recreate. But we
can never be sure whether the basis for this recreation is accurate—and
whether the details of the projection lead to what was intended at any
previous time.

Intention is of course not the only possible aim, but people often seem
to think that a choice has to be made between preserving the document
and restoring authorial (or some other) intention. Pollack quotes and
paraphrases Michael Friend (director of the film archive at the Academy
of Motion Picture Arts and Sciences) as follows: "The goal should be to
'restore the original achievement,' not the presumed original intention
of the film maker." The statement would be clearer if Friend had said
"document" instead of "achievement"; but the main point to be made
about his comment is that there is no need to choose between the goals.
It is always important for historical study to have the documents that
were the basis for public showings in the past; but critically emended texts
enable a fuller appreciation of the filmmaker's artistry. Another archivist
opposed to editorial alteration also confuses the discussion. Pollack quotes
Richard R. May (vice president of film preservation for Warner Brothers)
as saying, "To change something just because we have the technology to
do it, I am, and I think other archivists are, against it." Obviously the
existence of technology is not in itself an appropriate reason for making
a change; but when technology makes possible the correction of an error,
it is a valuable tool for the study of artistic intentions.

The matter of technical knowledge is worth pausing over for a mo-
ment. The fact that digital technology did not exist when Gone with the
was made does not mean that using it to correct the flagpole error
is an act of modernizing—for the correct flagpole image was achievable
by special-effects technology in 1939. For the same reason there can be no
objection to using modern technology to add an animal's missing foot in
one frame of Fantasia. It is wrong to object to this kind of emendation
by regarding it as "akin to efforts to fix the poor sense of perspective in pre-
Renaissance paintings" (a view reported by Pollack). Restoring what was
intended at a past time is not at all the same as imposing a more recent
visual vocabulary on a work from the past.

There have been, however, some attempts at film alteration that are
indeed akin to revising pre-Renaissance perspective. The two best-known
ones are probably stretch-printing and colorizing. The former attempts
to smooth out the jerky, stylized movements in some silent films (such as
Chaplin's) by repeating every second or third frame (see Walter Kerr's


Page 13
criticism of this practice, 1 July 1984). Colorizing obviously tries to make
black-and-white films look as if they had been shot in color (the subject of
many complaints in the Times, such as those by Vincent Canby on 2 and
30 November 1986). Neither of these techniques truly achieves its aim;
but even if they were successful, they would be unjustifiable to anyone
interested in authorial intention, for both involve changes in the visual
language used by the filmmaker. Whether that language was employed by
necessity or by choice (whether, for example, black-and-white photogra-
phy was used in a film of 1920 or one of 1970), it was what the filmmaker
was working with to express a vision. Those who defend stretch-printing
and colorization without admitting that new works are thereby created
put themselves in a position just as indefensible as those who think that
the spelling and punctuation of a verbal work can be modernized without
affecting meaning.

In other words, a distinction must be made between, on the one hand,
altering the cinematic vocabulary of a film (which is an act of adapta-
tion, not a historical undertaking) and, on the other, correcting errors,
reincorporating unauthorized deletions, or restoring degraded images (all
of which aim to recover authorial intention). The latter category is il-
lustrated by two "critical editions" of classic films that were unveiled in
New York within a few days of each other in early 1989. On 30 January,
a restored Gone with the Wind brought back the intended colors: a new
negative had been made in a two-year process that involved combining
the three original negatives (each for a different dye), thus resulting in new
prints that were faithful to the original intention, unlike the degraded and
unstable film stocks that had been used for showings in previous decades.
On 8 February, a reconstructed Lawrence of Arabia restored twenty minutes
of footage that had been deleted without David Lean's permission; it also
corrected such errors as the reversed printing of the second reel (which
had caused Lawrence's wristwatch to switch from one wrist to the other).
(See the articles by Max Alexander and Janet Maslin on 29 January 1989.)
These two "editions" show how the properly researched alteration of a
received film text can produce a version that is closer to the intention of
a director at a particular past time, just as the scholarly emendation of
documentary verbal texts, when undertaken with a historical goal like the
restoration of authorial intention, can take readers back to certain past
moments more faithfully than can the documents that physically survive
from the past. In both cases, given the relation of documents to works in
these arts, the historic documents need not be damaged in the process of
creating new texts.

The fact that Lean did not approve of the cuts made in Lawrence of
points to a major strand of film history that must be investigated
by textual critics. Just as authors of verbal works have often felt that their


Page 14
artistic goals were compromised by the alterations demanded by publish-
ers, many film directors have been angered but powerless in the face of
deletions required by studios. The changes in each case are of interest for
what they reveal about the cultural climate of the times that produced
them, but textual critics will also wish to understand the stages of textual
evolution that preceded publication or public release. Whether the evi-
dence that leads to such knowledge can support the production of new
texts varies from case to case, but in general one may say that the editor of
a verbal work may more often be able to recreate the text of a deleted pas-
sage than the editor of a cinematic work can, for even the survival of the
script of lost footage does not in itself make possible the reassembly of all
that would be required to reshoot it. Rarely can one expect to duplicate
even the limited remaking that happened during the re-editing of Lawrence
of Arabia:
the original actors were asked to re-record lost portions of the
soundtrack. That such an event took place underscores the mixed-media
nature of cinema and video, allowing for the separate reconsideration and
alteration of the soundtrack and the images.

Orson Welles once said that the only film he was "allowed" to com-
plete as he wished was Citizen Kane; his intentions for the others can be
reclaimed only in varying degrees. One can write about his aims for The
Magnificent Ambersons
, but without the lost footage one cannot produce a
new text. For Touch of Evil, however, there is not only a surviving preview
print, which includes some later deleted material (fifteen minutes' worth),
but also a memorandum by Welles detailing what he found unsatisfac-
tory in the film as released. With those materials available, Walter Murch
was able to prepare a new "edition" (for 1998 release, forty years after
the original) that incorporates digital repairs, restoration of cut scenes,
deletion of scenes added by the studio, a different structuring, and the
replacement of Henry Mancini's music (the removal of which allowed the
recovery of a suppressed layer of sound effects under it). (Murch's account
of how his version follows Welles's memo "scrupulously" appeared on
6 September 1998.)

Despite the expense of such restoration, there are other notable ex-
amples, such as the 1993 re-editing of A Streetcar Named Desire, similarly
made possible not only by the survival of excised footage but also by
external sources, including Elia Kazan's autobiography. (See Bernard
Weinraub's report, 16 September 1993.) This kind of careful recreation
of directorially intended film texts is to be encouraged, and the advent of
videocassette distribution of films for home viewing provided additional
occasions for undertaking re-editing. A film like Angel Heart, which Alan
Parker was forced to cut in order to receive an "R" rather than an "X"
rating, was released in 1987 as a video with the uncut text. The trend
toward providing home viewers with texts not available to theater audi-


Page 15
ences has expanded with the now seemingly established custom of in-
cluding both commentary and excised footage on the digital video disks
(DVDs) of commercial films. These disks have in effect become what are
called documentary editions in the field of verbal-text editing: they aim
to present an unaltered reproduction of a received text, accompanied by
annotation and related documentary texts.

I should add immediately, however, that the main text is "unaltered"
in the sense that its constituent elements have not been changed; but the
transfer from film to DVD, with the result that the work is viewed on
a small (or relatively small) rather than a large screen, may in itself in
many cases be a departure from authorial intention. This point does not
of course apply to works made for video and computer-monitor viewing
in the first place. The textual criticism of early television works is compli-
cated by the fact that before the late 1940s there was no adequate means
for preserving them and that the kinescope technology that followed was
not very satisfactory. Early digital art poses a somewhat different problem
for historical study: whereas there was no difficulty copying it, the
programs and equipment for retrieving it sometimes became obsolete and
unavailable. But despite the technological differences between celluloid
film, videotape, and digital disks, the theoretical framework for textual
criticism is the same. What is preserved in each visual document is not
simply (as with recordings of music and films of dance) the record of an
individual performance; rather it is (when displayed) a version of the work
itself—that is, the version of it represented by the particular copy of the
physical film being used. Nevertheless, the uncertainties attaching to the
constitution of all works in intangible media remain present, since one
may still question the content of the physical document and the manner of showing it.

Mention of television and digital art requires a brief word about
broadcasting—that is, the transmission (by radio waves, telephone, cable,
and the like) of a visual or aural work from a central source to any number
of locations simultaneously. Broadcasting is both a means of reproduction
(as when a movie made for theaters is telecast or a concert is radiocast)
and a means of original dissemination (as when a made-for-television
documentary is released on a television network or when a work of digi-
tal art is mounted on the artist's website). Broadcasting enables a work
to be experienced in multiple places at the same time (given appropriate
receiving equipment), without the need to transport any documents (like
scripts, scores, or films) to those places. And it does not always require au-
diences to be available at given tunes, since the internet and on-demand
television channels allow viewers to choose their own times for looking at
particular works. Although there are obviously significant differences be-
tween broadcasting and other means of bringing visual and aural works to


Page 16
the public, the chief difference affecting textual criticism is the increased
difficulty of ascertaining the variant texts that have been experienced.
Individual television stations, for example, may alter the programs they
receive from their network headquarters (especially by making deletions),
and each viewer's television set or computer monitor may display the
material somewhat differently. Yet so little of this information may be
recoverable as to hamper severely the writing of comprehensive textual
histories of works that have been broadcast.

5. Drama and Performance Art

It is not surprising that textual critics of literature have written ex-
tensively about drama, since plays are often read in printed form
and since dramatic works—especially those by Shakespeare and his
contemporaries—have figured heavily in the development of English-
language textual criticism. Whereas earlier textual critics and editors of
Renaissance drama focused on authorial intention (treating plays as they
did nondramatic literature), some recent scholars have been more inter-
ested in the performance texts of plays, arguing that drama is essentially
a collaborative art. Although it is true that drama, like all performing
arts, cannot normally be experienced in its intended form without col-
laborative effort, it does not follow that playwrights' intentions are neces-
sarily superseded by what emerges from the production process. Many
playwrights, even when they participate in that so-called "development,"
are not fully sympathetic with the resulting evolution of their plays. Ter-
rence McNally once said, "I worry that in the process of developing my
new play I lose it" (7 December 1986). And Andrew Bergman, noting
that movies "get filtered through other sensibilities," wanted to create a
play that would, he said, "be my voice purely, the way a novel is." He
rejected the idea of writing a screenplay: "I didn't want to have to open it
up and put some action in it where I don't want to" (reported by Mervyn
Rothstein, 12 April 1986). Bergman may have been unrealistic in imagin-
ing the amount of control he would have in the theater, but he made his
point of view very clear.

A celebrated example of a dispute over the alteration of a playwright's
intentions is the lawsuit brought by Samuel Beckett's American publishers
and lawyers against the American Repertory Theater, whose director,
Joanne Akalaitis, wished to set Endgame in a subway station rather than
an empty room. Partly in response to this event, New York University
in March 1985 held a panel discussion entitled "Authors versus Direc-
tors: Who Has the Right of Interpretation?" Opinions were expressed on
both sides (described by Samuel G. Freedman, 14 March 1985), but not
much could have been expected from a discussion responding to such
an unenlightened question. Thus John Guare's comment—"Theater is


Page 17
a collaborative art. But the playwright can choose who he or she wants
to collaborate with"—was not helpful, being limited to performances of
works by living playwrights (or during the life of a copyright) and over-
simplifying the issues even for those cases. "Interpretation" by performers
(whether or not so instructed by directors or authors) is a natural part of
the recreation of works in intangible media. The key point for discussion
should have been how to draw the line between interpretation that is in
keeping with the author's intention and interpretation that goes beyond
it and creates a different work. If the goal of a performance is compliance
with authorial intention, the range of permissible interpretation is usually
determined by the author-derived script, which in general (making allow-
ances for errors and variants) is the primary set of authorial guidelines,
subsuming certain kinds of interpretation. Thus the pacing of the action,
the exact brightness and tone of the lighting, and the actors' manner of
speaking can in many instances vary without violating the guidelines. But
a change that contravenes explicitly stated directives (such as moving the
setting of Beckett's play from a bare room to a subway station) is an act
of "interpretation" not in line with authorial intention.

If authorial intention is not a concern, directors are of course free to
make such changes or more drastic ones (at least for works out of copy-
right). And authorial intention need not be, and certainly is not, always
the aim: works in intangible media can be expected to undergo all man-
ner of changes in their performed texts over the years. Alterations not in
line with authorial intention may seem aesthetic improvements to some
people, and there is no reason in principle to disapprove of such changes.
What may legitimately be deplored, however, is confused thinking about
what does, or does not, conform with authorial intention and thus about
whether or not a given performance is within the matrix of possibilities
embodied in an authorial text.

The range of those possibilities varies in inverse proportion to the
explicitness of the stage directions and set descriptions. Some playwrights
compose extremely detailed paragraphs of such commentary, which in
published form make the plays read almost as novels but also provide sub-
stantial insight into how the playwright visualized stage presentation. The
use of furniture and other physical objects in staging is one indication of
the mixed-media nature of drama, for such settings resemble installation
art; but in drama, where physical details are usually ancillary, some of
those details can be altered without violating the guidelines of the script,
whereas in an installation any change of a detail (except one aimed at
correcting an error) would compromise the artist's intended text. (These
points obviously apply to stage settings in opera as well.) Controversy
has recently arisen as to whether directors should be granted copyrights
for their staging: playwrights believe that such copyrights would limit


Page 18
their control over their plays (the issue is commented on by Jesse Green,
29 January 2006). But since any given staging would normally not be the
only one consistent with the authorially intended script (and might well be
inconsistent with it), any copyrighted staging could not logically be tied to
the copyright of the play, and permission to perform the play would not
entail the use of that particular staging.

Generally speaking, then, the consequential alteration of settings is an
example of a directorial change that would depart from authorial inten-
tion; but the exact pronunciation of the words would not—except when
a particular dialect is specifically indicated—be a matter controlled by
the concept of an authorially intended text. The 2004 Globe Theatre
production of Romeo and Juliet in what was thought to be Elizabethan
pronunciation was naturally an enlightening experiment in historical re-
construction, just as there is a valid historical interest in hearing a piece
of music played on instruments from the composer's time; but in neither
case are those precise sounds dictated by authorial intention. Similarly,
a filmed record of a dramatic performance, even one in which the play-
wright was involved, does not supersede (though it may well supplement)
a printed text, any more than a film of a dance work can replace choreo-
graphic notation. In a cinematic work, of course, the angles of the shots
selected for inclusion, as well as the pacing of the alternation between
close-ups and more distant views, constitute the work; but for dance and
drama, a film is only one limited record of one performance—and the
performance itself cannot demonstrate all the possibilities inherent in the
score or script.

Performance art presents a somewhat different situation in that nor-
mally the performer and the creator of the spoken language are the same
person. (Lecturing is clearly one category of performance art.) There
may or may not be a script or outline, but in any case whatever the per-
former does and says reflects authorial intent, except in those instances
where a mistake is made or where the performer is reluctantly conform-
ing to the advice of someone else. Deciding when these situations obtain
is one of the judgments that a textual critic must make. Similarly, when
a performer-author adapts the language and action to a specific local
situation, the textual critic must consider whether the adaptation is a
departure from the essential work or whether the work encompasses all
such variants that occur in individual performances. (The textual criti-
cism of performance art, in other words, entails much the same process
of thought as that applicable to jazz.) Filmed records of performance
works are in one sense only documentation, since the work itself is the
live performance. Yet these records have a different status from the filmed
documentation of a conventional play, for the variants preserved in them
(and perhaps nowhere else) may be parts of the intended work as much


Page 19
as if they had been written into a script. From the point of view of autho-
rial intention, the work may comprise all (or many) of its variant perfor-
mances; this observation, however, could be made about a conventional
play only by thinking of the work as a social product rather than as a
reflection of individual authorial intention. Sport, which involves impro-
visation within clearly defined frameworks and plans, may be regarded as
a genre—predominantly collaborative—of performance art; the textual
history of a sporting event takes into account not only the various docu-
mentary records of what happened but also the variations that occur,
as a result of editorial decision or the equipment employed, when those
records in visual media are shown on different occasions.

Work in Tangible Media

6. Painting, Drawing, and Calligraphy

One might at first believe that critical editions of paintings (as opposed
to writings about the textual histories of paintings) would be extremely
rare, since editorial emendations would alter the unique art objects, forc-
ing viewers in each instance to look at the emended text, and it alone.
Such results, one might imagine, would not often be permitted by own-
ers of paintings. (An editor could of course make alterations on a photo-
graphic or other reproduction of a painting rather than on the original,
or alternatively could make a photographic record of the appearance of
the original before beginning editorial work, but either way one or more
of the versions would simply be reproductions, not the work expressed
in the materials intended by the painter.) To some extent, this surmised
rarity of scholarly editions of paintings is correct, for even when permis-
sion to edit is given, there can be only one edition at any one time: each
successive edition would obliterate the one that went before (except in
the form of a photographic or digital record). The freedom of editors to
create new editions is inevitably restricted in the case of works in tangible
media—certainly as compared with the theoretically unlimited freedom
editors have to edit works in intangible media, where documentary evi-
dence need not be destroyed in the process.

In another sense, however, there have been many more scholarly
editions of paintings than might have been expected, for cleanings and
"restorations" now routinely take place in museums. The technology for
analyzing the underlying layers and chemical makeup of paintings (in-
cluding infrared reflectography, Raman microspectrometry, and digital
imaging) and the skill of the restorers who remove substances from, and
add them to, paintings have become so sophisticated that museum cura-
tors feel justified in allowing the editing of paintings to take place. Such
editions often have a more dramatic effect on the text than occurs in edi-


Page 20
tions of works in intangible media: a cleaning, after all, affects the entire
text, every square inch of it. Despite the impressive expertise that now
generally underlies these operations, there are grounds for raising disqui-
eting questions about them.

In the first place, the goal of both cleaning and other kinds of restora-
tion is normally final authorial intention—to bring the text of a painting
back to what it was when the painter considered it finished. But this is not
the only moment in the history of a painting that is of interest, and some
would argue that it is not the moment of the greatest interest. However
one feels about this matter, the fact remains that every painting, simply
because it is a physical object, inevitably undergoes alterations over the
years resulting from the atmospheric conditions under which it has been
kept, and sometimes from accidental damage as well. The present state of
each painting is a summation of all that has gone before and is the base on
which its future evolution will rest. Unless the painting is recent, it surely
will not look as it did when it left the painter's hands; but it probably has
not looked that way for a considerable time, and many viewers will have
responded to intermediate states. Any old painting that has been the sub-
ject of commentary over the years will have presented a different appear-
ance to different writers. Since modern technology allows the recovery
of a great deal of the evidence that lies beneath the top layer of dirt and
paint, textual critics can write accounts of the evolution of a painting and
can offer illustrative reproductions of various stages in its history without
taking any intrusive action that affects the physical object itself. Some
people would therefore say that the editing of paintings should not take
place at all (except perhaps to stop physical deterioration), leaving the
body of evidence embedded in the object to remain untouched for future
textual critics to examine, perhaps using even more advanced technology,
and in any case perhaps arriving at different conclusions.

This position values the physical evidence of the object and the con-
stantly changing appearance of the work over the text of the painting as
intended (at some point) by its creator. But the interest in artistic intention
is great enough to cause many people to feel that the recovery of intended
forms of paintings is worth the price of losing some of the accumulated
evidence reflecting the history of the object. In instances where this view
is dominant, there is still room for disagreement about the precise goal
and about whether that goal has been achieved. An artist's intentions may
change, not only during the process of original painting but also at a later
time (after the painting was considered "finished"), and the layers of paint
reflecting these changing intentions may be recoverable. Only one stage
can be selected; and even if agreement were reached as to which is the
most appropriate, there would probably still be disagreements as to how
much needs to be removed from the present surface in order to reach


Page 21
the level that represents the desired stage. Was a given layer of paint, for
example, applied by the original painter, perhaps overpainting a previ-
ous "final" layer, or was it added by a later painter in order to make the
painting conform better to a later taste? Can one tell the difference be-
tween atmospheric contamination and the residue of the artist's method
of obtaining a particular color or finish?

These uncertainties are analogous to those attaching to critical edi-
tions in every field, since any text that emerges from acts of judgment,
however learned they may be, can be questioned by equally informed
persons who would have made different judgments. Uncertainty, in other
words, is unavoidable and has not prevented restoration projects from
being undertaken for major works. One of the best known examples is
the Vatican's long and painstaking program (beginning in 1980) of clean-
ing Michelangelo's frescoes in the Sistine Chapel. The idea of removing the
candle soot and other pollutants that have darkened these paintings
may sound admirable in the abstract; but only the naïve could imagine
that the process would not be problematic, and there have in fact been
serious criticisms made by James Beck, Tuti Scialoja, and others. They
have raised points that relate both to the long-term effects of the clean-
ing and to the determination of artistic intention. They say, for example,
that the surface dirt and wax may have served as a preservative and that
removing them may expose the paints to potentially destructive modern
pollutants. There is also the possibility that Michelangelo himself used
candle soot and animal fat as darkening agents to create shadowy effects.
Furthermore, arguments have been made that removing paint added af-
ter the plaster was dry, as if it were overpainting by another hand, may at
some points destroy Michelangelo's final touches, since he may sometimes
have regarded the paint he placed on the wet plaster as a first draft. Thus
the restorer's every act of alteration is an act of judgment; but unlike the
rejection of a reading in a literary text, it cannot be reversed. (Two of the
many instances of the Times coverage of the controversy over the Sistine
restoration are articles by Douglas C. McGill on 6 November 1986 and
by Mary Davis Suro on 4 January 1987.)

If these problems illustrate some of the issues inevitably raised by criti-
cal editing, another kind of problem—not inevitable—has also come up
in connection with part of the Sistine work: lack of coherence in formulat-
ing the goal of the restoration. Over the years, various forms of drapery
were painted on forty nude bodies in the Last Judgment; but only the sev-
enteen coverings added after 1750 were removed in the restoration. In
defense of this retention of some of the coverings, Kathleen Brandt has
said that the drapery was added "at the request of the patron, namely
the papacy," and that it constitutes "a chart of notions of decorum over
time." (See John Tagliabue's article, 9 April 1994.) But this reasoning


Page 22
would logically lead to preserving all forty of the additions. Even if one
were to find a better justification for the significance of the 1750 state of
the work, there would still be the inconsistency between the handling of
this fresco and the others (in which the artist's final intention is the aim).
One cannot—in a painted work or a printed text—simultaneously re-
flect authorial intention and the accretions that emerged from later social

The current reputations of artists have a great deal to do with which
accretions to paintings get preserved. No one would be likely to raise
great objection to the removal of any of the added draperies in the Last
, since they are by lesser painters than Michelangelo, nor would
anyone wish Michelangelo's work to be removed in order to reveal the
frescoes of Perugino that lie beneath. When the National Gallery in Lon
don undertook to restore Bellini's The Feast of the Gods, completed in 1514,
it decided to concentrate on the painting's appearance in 1529, since
Titian made major additions to it at that time, and the idea of removing
work by Titian could not be contemplated. Preserving the 1529 form of
the work can be justified not simply on the grounds of Titian's stature but
because it also keeps the portions of Dosso Dossi's intermediate revisions
that were not covered over by Titian and thus reveals the social attitude
toward painting texts in the sixteenth century, when an owner of a paint-
ing was less concerned with maintaining its integrity as the work of a
particular artist than with causing it to evolve in conformity with current
taste. But of course the work that illustrates this important historical point
cannot be regarded as a work by Bellini. The National Gallery's 1990 ex-
hibition of the restored Feast admirably showed how modern technology
allows some knowledge of a painting's textual history by displaying X-ray,
infrared, and ultraviolet photographs of the underlying images along-
side the presumed 1529 state of the oil. (On this exhibition, see Michael
Kimmelman's 21 January 1990 piece.)

One of the Louvre's great possessions, Veronese's Marriage at Cana,
can serve as an epitome of the concerns and occurrences attendant on
cleaning and restoration. The recent cleaning (1990–92) removed yellowed
varnish, revealing bright colors not seen by viewers and commentators
for a very long time. The restorers, using information obtained by X-ray
and chemical tests, removed paint that they considered had been added
by a different painter, thereby arousing a storm of protest from a group
of artists. The most controversial decision was to change the color of the
coat worn by a major figure in the foreground: the red layer was taken
away, exposing the green one underneath. Critics argued that only the
lower part of the red coat showed brushwork uncharacteristic of Veronese
and that this area had been subjected to earlier repair; they also noted
that the coat is red in the earliest known copy, made in 1607 (nineteen


Page 23
years after the artist's death). If the red was indeed Veronese's, part of his
intended text is now lost, whereas evidence of his earlier extensive altera-
tions, moving and inserting figures, is safe by being on a lower level and
has been made visible by X-ray pictures. The X-ray investigation also
made clear some of the vicissitudes endured by the physical object, for
it has many repaired nail-holes, and it retains the signs of having been
unstitched horizontally (to form two sections for ease of hanging) and
then—fifty years later— of having been stitched back. As if to continue
the punishment, an accident occurred when the canvas was being rehung
in June 1992, resulting in several gashes, three of them about three feet
long. An unexpected job of restoration was then required, in which the
cut threads were individually glued together and retouched. (See Marlise
Simons, 17 November 1992.) The object therefore tells the story of acci-
dental and intentional mutilations, of the artist's changing intentions, and
of several restorations. If the story is particularly dramatic in this case, the
combination of elements that make it up is not at all uncommon as the
underpinning of the texts of paintings that we now see.

We are by now accustomed to significant textual revelations being
brought about by modern technology applied to paintings. Even in the
absence of surviving preliminary studies, we are beginning to have for
some paintings the kind of evidence of revisions that literary scholars
have long had for verbal works in the form of rough-draft holographs.
(The same technology is of course being used in the literary field to un-
cover words invisible or illegible to the naked eye.) Fortunately, museum
exhibitions are also increasingly paying attention to the textual criticism
of paintings, as in the Bellini and Veronese instances—or in the 2004
exhibition at the Art Institute of Chicago, showing that under the surface
of Seurat's A Sunday on La Grande Jatte—1884 is evidence not only of his
extensive revisions of figures and colors but also of his brush work, which
originally consisted of conventional strokes, not pointillist dots. (See the
report by Holland Cotter, 20 August 2004.)

More discoveries will come, relating not only to oil paintings but also
to works in watercolor, tempera, charcoal, and other media capable of
covering up earlier drafts. Revisions in some media, like ink or pencil,
may be more likely to involve erasure than cover-up, but such evidence
as there is will be uncovered in those cases as well. Textual criticism of
calligraphic art must also take into account the linguistic text formed by
the characters, just as if the work were a concrete poem, which in some
ways it is. Regardless of the amount of textual evidence available, all
visual works that consist of the actual surfaces on which the artist made
marks are ontologically the same and thus pose the same problems for
textual criticism and scholarly editing.


Page 24

Since these works are so often presented in frames, it is perhaps worth
adding that the same observations apply to frames as to the visual ele-
ments in the presentation of tangible verbal texts: they very often have
an effect on audience response, but they are not always intended as parts
of the works. Frames, even if designed or selected by the creators of the
works they enclose, may not be regarded by those creators as textual ele-
ments in the works, though they must be taken into account in analyzing
the history of responses to the works. And many of them are deserving of
study as works of art in their own right.

7. Sculpture, Craft, and Installation Art

The extent to which some of the evidence for the textual history of
a piece of sculpture can be found in the object itself varies according to
the material out of which the work is made, just as it does in the case of
two-dimensional works (that is, essentially two-dimensional ones, despite
the three-dimensional surface of paint or collage). Sculptures produced
by cutting material away, like those in marble or wood, do not readily
provide evidence of previous drafts. Any cuts that the sculptor regarded
as mistakes are obliterated by new cuts or were made to fit into a revised
plan for the area involved. Analysis by a textual critic may suggest likely
spots where such alterations occurred, but no editorial emendation would
follow because the artist's final intentions are reflected in those altera-
tions. (It is as if a writer, having made a slip of the pen, incorporated the
accidental reading into the final form of the sentence—an analogy that
can have an exact sculptural parallel, when a letter-cutter makes a mis-
take in carving an inscription in stone or wood.) And no one would be
likely to propose recreating a hypothetical earlier intention at the price
of destroying the final one—or would be allowed to undertake the task,
if proposed, in any form other than a replica. On the other hand, sculp-
tures produced by casting or by a process of building up or adding on, as
with some works in metal, glass, or other materials (where separate units
are affixed to each other), may on occasion offer more internal evidence
of revision, but rarely on the scale of what technology has uncovered
beneath the surfaces of paintings. External evidence, as in sketches or
models, may of course exist, as it may for works of any genre; and such
evidence, when it exists at all, is more likely to be the only evidence than
is the case with painting.

The issues faced by editors of sculpture were discussed in a front-page
article in the New York Times on 15 July 2003, written by Alan Riding to
explain the international controversy aroused by plans to clean Michel-
angelo's David. Riding asked, "Should the marble colossus be restored to
its original perfection or simply cleaned of grime? Or should it learn to
live with the inevitable streaks and blotches of venerable old age?" These


Page 25
two questions, though not carefully phrased, allude to the two basic ap-
proaches available to all editors: the critical approach, in which altera-
tions are made with the goal of recreating a previous version of a work
(as intended at some past moment but not fully realized in any surviving
artifact); and the documentary, in which no changes are made to the
inherited state of a work, thus preserving the cumulative effects of its his-
tory. The former, as set out by Riding, subsumes two activities, restoring
and cleaning. But it is important to understand that cleaning is a form of
restoring, for its goal is the restoration of a work's earlier appearance. And
Riding's use of the word "simply" in connection with cleaning implies,
incorrectly, that it is necessarily less intrusive or less a matter of judgment
than the kind of restoring that involves additions. In cleaning marble, for
example, a decision has to be made not only about how much to take off
but also between a dry method (using brushes and rubbing) and a wet one
(using poultices), each of which has a different effect on the pores of the
stone and produces a different look.

Even though cleaning certainly involves critical judgment, there is un-
deniably a widespread perception that a restoration consisting of cleaning
alone is more conservative (that is, safer) than one that also includes such
actions as filling cracks and nicks or supplying a replacement element (like
the marble and plaster substitute toe that was added to the David in 1991).
The argument that a less intrusive procedure (whatever it may be) is more
justifiable can perhaps be supported on practical grounds for works in
tangible media, though it can never be supported logically. It is reminis-
cent of the claim made with pride by some literary editors that they have
tried to keep their alterations to a minimum. But in a critical edition one
must make whatever changes are necessary to bring about the desired
goal, and there is no virtue in doing less than that out of respect for the
misguided notion that a small number of changes is desirable in itself. But
such editions do not affect the documentary evidence, whereas in the case
of sculpture and other works in physical media there are good practical
grounds for protecting such evidence at the price of editorial illogic.

In any case there would be no possibility of restoring the authori-
ally intended texts of many sculptures, particularly those kept outdoors
and made of materials that can be irreversibly affected by exposure to
weather. When Riding asked whether the David should be "restored to
its original perfection," he knew that he was not describing a realizable
goal, for later in the article he quoted a restorer who pointed out that
"there is not one millimeter of its original surface left"—partly as a re-
sult of weather damage, partly because of an 1843 cleaning that used
hydrochloric acid. The application of a new surface, though a theoretical
editorial possibility, would surely not be permitted, and it probably should
not be permitted, because one could persuasively argue that in this situ-


Page 26
ation the editorial emendation is too conjectural to force on all viewers
of the work (unlike what happens with editorial conjectures in literature,
which leave readers free to experience the work without them). Even in
cases where the surface is not severely damaged, bringing a work back to
"its original perfection" would always be conjectural, as critical editing
necessarily is, and Riding's statement—like any expression of the goal
of critical restoration—should include wording that acknowledges the
contingencies inherent in such an aim. One must also remember, when
speaking of sculpture, that weather-produced alterations may sometimes
actually produce authorially intended texts, for sculptors may take into
account the changes (as in the colors of metals) that their materials will
undergo. Although there is always reason to be interested in textual shifts
over time, even when not intended by the creators of works, here is a situ-
ation in which some later changes—brought about by the course of time
and not by a scholarly editor—can be accepted as representing authorial

The attempt to restore the intended texts of some modern sculptures
poses an additional difficulty in that these sculptures may contain a wider
range of fragile materials than earlier sculptures do. When such elements
show deterioration or damage, conservators hope to find replacement
parts; but identifying the source of the materials and trying to learn whether
they are still available are not easy tasks. For this reason Carol Mancusi-
Ungaro, now the director of conservation at the Whitney Museum, began
in the mid-1980s a video oral-history project in which artists (painters as
well as sculptors) are asked to give advice on the conservation of their own
works (see Randy Kennedy's account, 29 June 2006). These conversations
are valuable documents in ways that go beyond their original purpose,
for artists' comments on the specific materials they used naturally lead to
broader reflections on what they meant to convey in their works. But the
textual usefulness of these interviews is not as straightforward as some
might imagine. Any specific identifications of materials and their sources,
when an artist's memory is accurate, can indeed be helpful in restoring a
text; but one must remember that any recommendations as to what ought
to be done reflect the artist's thinking at the time of the conversation and
may, if carried out, produce a new version of the work. The result of an
interview, then, may be to provide one more stage of authorial intention
for the conservator-editor (who can select only one) to consider.

There are of course many other three-dimensional objects, the prod-
ucts of human creativity just as sculpture is, that are not usually called
"sculpture." Conventionally these objects have been regarded as examples
of "craft" or "design" or "decorative art" rather than of "fine art," but
this distinction does not have an ontological basis, or even one founded
on the genres of works involved. Fortunately art museums, and not simply


Page 27
museums with anthropological interests, have increasingly recognized the
importance of "crafts," and specialized craft museums have also been es-
tablished. For present purposes, it is not necessary to worry about these
divisions, for the same points can be made about the texts of such objects
as about the texts of those usually called sculptures. This point, though
not stated in terms of "texts," was at the heart of a 1990 conference held
at the American Craft Museum in New York (and reported by Roberta
Smith on 22 January 1990), where George A. Kubler reiterated (in the
words of an earlier book of his) that "the idea of art can be expanded to
embrace the whole range of man-made things." Because many of those
things usually regarded as craft are utilitarian objects, the study of their
original states can be particularly difficult, owing to the wear that comes
with use and to the fact that people rarely hesitate to alter their utensils,
furniture, and clothing when alterations will improve (in their view) the
functioning of the objects.

One may justifiably be interested in the authorially intended texts of
such objects (whether created by anonymous artisans or famous silver-
smiths, cabinetmakers, and couturiers) as well as the texts that evolved
through use and the passage of time. The textual stages that are tradi-
tionally preferred vary according to the medium: curators of silver wish
to remove tarnish (as grime is lifted from paintings), whereas specialists
in old furniture prize the patina it acquires over time. These are simply
matters of convention, for every stage in the life-story of every object is
of historical and aesthetic interest. Even so, there is a particular appeal in
the current state of any utilitarian object, since it bears the traces of the
object's shared life with human beings. For this reason Brian Murphy, in
his 2005 book on Persian rugs, The Root of Wild Madder, takes the view that
wear and what might be called imperfections (such as awkward repairs)
are not objectionable. As Kubler eloquently put it at the craft conference
(making the case in fact for the social approach to all art): "to be in use
among the young who transmute and re-enact the work of the dead is the
best of all eternities." None of this negates an interest in the earlier (or
earliest) states of objects; but it does suggest that the destruction of later
evidence in an effort to bring back an earlier state may often be harder to
justify for "crafts" than for "arts" in the minds of many people.

Stained-glass art, often classified as a craft, can afford unusual op-
portunities for uncovering post-production textual histories, especially
in such complicated structures as windows made up of many individual
pieces of glass held together by lead strips. The sophisticated research
now being done is illustrated by Drew Anderson's detailed reports on the
Gothic windows at the Cloisters in Manhattan. After comparing current
and earlier photographs of a window digitally and making rubbings of
each piece of glass, Anderson describes the deterioration, amateur repairs


Page 28
(such as the insertion of any piece of glass that came to hand and the
enlarging of the lead to hold it in), and previous conservation efforts that
have occurred at each spot in the window. He is thus reading the windows
in much the same way that analytical bibliographers read the physical
evidence in books—a point that is particularly apt since medieval picto-
rial windows were meant to function in part as books, conveying biblical
stories to those who could not read. Textual critics must always try to
extract the story that the physical details tell, as preparation for evaluat-
ing the text of the work. "Each window tells its own story," Anderson
says. "And the more you work on them, the more you find out" (quoted
by Carol Vogel, 17 June 2006). This knowledge helps one to appreciate
the work, whether or not any editorial alterations seem advisable; and in
the case of stained-glass windows, attempting to recreate their original
condition would often be incompatible with preservation, since cleaning
can alter the colors. The situation once again illustrates how, with works
in tangible media, practical considerations may limit the editorial activity
that follows from a detailed knowledge of textual history.

Installation art, another form of sculpture, poses a few additional tex-
tual problems. Because installations are likely to consist of a variety of
objects and because the spatial relationships among them are crucial to
the work, the task of moving an installation (if it is not a site-specific one)
to a different location is especially challenging. Although it is possible to
recreate an installation precisely, so long as all the same objects remain
available, it is likely that there will be small textual differences (at least)
in each reinstallation. If the artist supervises reinstallations, any differ-
ences may constitute authorially intended versions, unless they result from
concessions made in deference to a particular space, in which case they
are analogous to adaptations for a special audience. Exact reinstallations
require exact measurements of the original installation; photographs are
obviously not adequate, and any sketches or notes by the artist do not
necessarily predict or record what the artist actually did in setting up
the original installation. Furthermore, installations are in many instances
mixed-media works par excellence, combining sculptural items, paintings,
photographs, videos or film loops, music or other sound, verbal texts in
visible form, and so on. Textual criticism and scholarly editing of such
works therefore involve the issues characteristic of each of the media in-
dividually, as well as the analysis of the texts that have resulted from the

8. Architecture, Interior Design, and Gardening

As a class, works of architecture undergo significant alterations at
the hands of their owners more often than any other works of art. The
changes go far beyond the repairs necessitated by aging and weathering;


Page 29
they often involve the construction of new rooms and wings, the recon-
figuration of existing rooms, and the attachment of decorative elements
and coverings to the exterior, perhaps totally transforming its appear-
ance. Because buildings are meant to be inhabited, they are like large-
scale utensils, which preserve the traces of daily use as well as conscious
alteration aimed at increasing utility. The architect's original intentions
are less important to many (probably most) owners than creating spaces
that they deem more convenient or attractive. The present text of every
house or office building is interesting (like the present state of all works
of art in physical media) as the momentary culmination of its interaction
with human beings. The textual history of a structure is often partly vis-
ible without any prodding beneath the surface, but a great many details
are always hidden from sight. Some of them can be learned in the course
of ordinary repairs, but uncovering others may require the destruction of
part of what is presently visible.

The decision as to whether such destruction is justifiable—whether
what is recovered is more desirable than what is lost—involves essentially
the same issues that arise with any other work of physical art, despite
the fact that buildings have a utilitarian function. There are increasing
numbers of owners of old houses who are eager not simply to preserve
what they possess but to restore it to what it was at a previous time, re-
gardless of whether the result will be more, or less, comfortable. Thus the
scholarly editing of buildings is not an uncommon activity, encouraged
by the historic preservation movement. Preservation at the most basic
level is of course the prevention of destruction, but it inevitably leads to
a second stage: since buildings require maintenance for survival, repairs
(textual emendations) must be made, and if they are to be made responsi-
bly, they must be directed toward a stated goal. As with all critical editing,
one must first decide what point in the textual history of the work is to
be the focus—the latest, the earliest, or some intermediate one. And after
the historical moment is settled on, one must determine what evidence
there is for recreating it—whether, indeed, there is enough to make the
attempt possible.

Among the factors that often affect the answer to the first of these
questions is the presence of alterations and additions made by a cele-
brated architect. The 1980s restoration of the main immigration building
on Ellis Island, for example, did not have as its goal the original 1900
state but rather that of 1918, when the Guastavino tile vaults were added.
(See Paul Goldberger's discussion, 14 August 1990.) Christopher Wren's
great addition to Hampton Court Palace in the 1690s destroyed a Tudor
courtyard, but no one would be likely to advocate returning the palace to
the form in which Henry VIII experienced it, since Wren's work would be
lost in the process. So when a serious fire in 1986 gutted the south wing


Page 30
of Wren's addition, any course of action other than restoration seemed
unthinkable. Surviving fragments (crystal, fabric, wood carving, and the
like) were pieced together with newly fabricated matching material. The
resulting mixture is not unlike what exists in the older parts of the palace
(or any other old building that has undergone repair). As Simon Thurley,
a curator at Hampton Court, said, "half of the 'Tudor' brick walls are
not really Tudor, and some are 19th-century" (quoted by Sherry Marker,
28 March 1993). The line between restoration and replica is not a distinct
one, for any emendation that involves new material can only produce an
approximation of the intended text, since the medium is physical and the
new material is a different physical object.

Determining the past time to be aimed for in a restoration is occa-
sionally taken out of the owner's hands if the building is located within a
designated historic district that has rules governing such matters. Gener-
ally the regulations insist on the preservation of the mix that existed at
the time the rules went into effect (with some alterations allowed by ad
permission); the focus, in other words, is on the social text—of each
house, as it has evolved, and of the neighborhood as a whole. The prob-
lems that can arise are epitomized by a case that occurred in the 1980s in
East Hampton, Long Island. The owners of an eighteenth-century house
in the historic center of the village removed a Victorian porch that had
been added, because they preferred the house to look as it had in 1790
and not to be an amalgam of the styles of two periods. They had unwit-
tingly failed to apply for a certificate of appropriateness, but it would not
have been granted in any case, for the Design Review Board argued that
the Victorian "Boardinghouse Era" was a distinctive part of the village's
history, and it ruled that the porch had to be built back.

This decision seems questionable—since the materials of which the
porch was constructed had been destroyed and since the porchless version
of the house does represent another period of East Hampton history—but
I am concerned less with the rightness of the decision than with what this
episode illustrates about the editing of buildings. Judgment is naturally
involved, as it is in all editing, and judgment is often affected by fashion,
by evaluations embedded in the cultural milieu. Thus Robert Hefner, a
preservationist employed by East Hampton, noted, "Thirty years ago a
restoration architect would say, 'Rip this porch off.' The philosophy was
to go back to the original form. Now preservation philosophy is more
refined and objective, taking into account historical periods" (quoted by
Michael Winerip, 20 July 1990). Although one might question "refined"
and "objective," there is no doubt about the main point: shifting scholarly
and intellectual predispositions influence editorial decisions. Preserving
the mixture of period styles that are the natural byproduct of a function-


Page 31
ing community may not be "objective," but it is certainly cautious, since
evidence of past living is not intentionally destroyed. One cannot help
but think of the contrast between East Hampton's firm (perhaps obstinate)
insistence on preserving the historical record and Colonial Williamsburg's
equally inflexible desire to erase the nineteenth century.

The question of what evidence there is for a restoration, when res-
toration (to whatever moment) is settled upon, involves recognizing that
surviving plans or other external documents cannot be accepted at face
value. A plan may of course indicate what an architect intended at the
time when it was drawn, but the building may have been built in a dif-
ferent way—because the builder did not follow the plan, or the architect
had second thoughts (possibly recorded on a now-missing plan), or the
client asked for alterations. Evidence of the inadequacy of plans turns up
all the time: to cite only one instance, when restoration of the ship Moni-
was being considered, it was found, as William J. Broad states, that
its "remains are often quite different from plans and period drawings"
(30 July 2002). Complex structures like buildings and ships, which in-
volve assembly from disparate parts, frequently contain within themselves
many indications of their own history; and sophisticated techniques are
now available to restoration architects for uncovering such evidence, as
when they determine the sequence and dates of multiple layers of paint.

When intrusive procedures (involving cutting, stripping, or even par-
tial demolition) are deemed necessary—whether the occasion is resto-
ration or remodeling—one is brought face to face with the actions of
human beings in the past with an immediacy not often matched by tex-
tual investigations into other arts. Whenever Paul Eisemann, a New York
carpenter, cuts into a wall, "he views it [according to John Freeman Gill]
as an opportunity to practice a kind of workingman's archaeology." Eise-
mann says that "buildings will talk to you if you listen"; for example, when
"you can see the original mortar on brickwork, sometimes you can tell
what group of immigrants did the work" (1 January 2006). As one undoes
past work—pulling out an old nail, for instance—one learns something
about the technique and attitude of the person who did the work in the
first place. When Verlyn Klinkenborg was taking apart the oldest room
of his house, he thought of the original carpenter and "how solidly he
did his job. He stinted nothing when it came to lumber and nails and,
especially, screws." Behind the walls "there is another house and another
set of lives." Klinkenborg, intrigued by the decisions of previous owners,
recognizes that future owners will equally marvel at his own decisions.
"The trouble," he says, "is that you can see only what remains—not what
has been erased. When I'm finished with my work, an era in the life of
this house will have vanished" (25 November 2005). This is the essence of


Page 32
textual alterations in physical media, but at least in the case of buildings
there still often remains enough evidence for detailed textual histories to
be recovered.

The wall decoration, furnishings, and arrangement of objects within
a house, or the landscaping and gardens that surround a house, may
or may not have been created by the same person who designed the
house. From the point of view of authorial intention, the extent to which
a house, its interior decoration, and its gardens can be approached as
a single work will vary with the circumstances; but from the social or
audience-response point of view, they are always inextricable, since they
are experienced together, and the ensemble affects one's response to each
element. One object that landscape architecture and interior design share
with other architecture is that of creating spaces to be lived in—they are
all "living rooms"—and thus gardens and room decorations, like houses,
often undergo textual alterations to accommodate their owners' changing
needs and tastes. Interior design, textually considered, involves the same
questions as installation art, for both create arrangements of objects to
be experienced within defined spaces—though interior design is more
frequently altered, since owners of houses usually regard it as a utilitarian
craft rather than a "fine" art.

The textual criticism of designed landscape must take into account
the primary fact that distinguishes it from the textual study of buildings
and room decoration: the predominant elements in such texts are living
plants that change continuously. The human-made inanimate objects in
gardens—like flagstones, sundials, and statuary—are no different textu-
ally from sculpture and craft. But whereas sculptors and architects of
buildings often take into account the effects that weather will have on the
materials they use, they do not have to think about variations that are
perceptible on a daily basis. The plan for a garden, whether on paper or
in the mind, provides a framework that subsumes the constantly changing
details of the stages of growth of the individual plants as well as the altera-
tions produced by necessary maintenance. Such changes are part of what
is intended (by the original designer or a later one who modifies the earlier
plan); the elements that are intended to be relatively more stable include
color, height, and massing, as well as the specific varieties of plants. Thus
the only editorial emendations that are at odds with what was intended
are those that change elements of this kind and their relationships. (That
the textual history of gardens has been flourishing is suggested by the
intense interest in the work of Gertrude Jekyll and Beatrix Farrand, by
programs of the National Trust in England and the Smithsonian Institu-
tion in America, and by the existence of the Thomas Jefferson Center for
Historic Plants at Monticello and Mac Griswold's Sylvester Manor Proj-
ect on Shelter Island.) Because plants are living and changing, they bring


Page 33
an element of the performing arts to the tangible-medium art of garden-
ing; for that reason, the relation of written plans or designs to gardens is
like that of scripts and scores to drama and music.

9. Printmaking, Photography, and Book Design

The ability of audiences in different locations to experience a work
simultaneously is primarily associated with works in intangible media,
where multiple copies of notation for recreating the works can be widely
distributed. But some works using tangible media can exist in a number of
exemplars if the artist chooses to create an intermediate object that pro-
vides the means for mechanically producing multiple copies. Sculptures
in liquefiable materials like metal or glass, for example, are sometimes
cast in molds; and when the process and results are supervised and ap-
proved by the sculptor, each resulting piece constitutes the work from the
point of view of authorial intention. The most voluminous category of
visual art intended to exist in multiple exemplars consists of those works
on paper that are usually called "prints." Before photography (to which
we shall turn in a moment), the intermediate objects used to "print" the
paper with ink could take various forms, offering three classes of printing
surface: a relief (protruding) surface, as on a woodblock, where the area
not to print is cut away; an intaglio (or sunken) surface, as on a metal plate
prepared for engraving or etching, where grooves are cut to hold the ink
that will be transferred to the paper; and a planographic (level) surface,
as on a lithographic stone, where the greasy lines of a crayon drawing will
hold the ink. These differing processes naturally affect the character of
the prints that result, and an understanding of them is therefore relevant
to textual criticism as well as to art criticism, for textual criticism (in any
field) can never be divorced from the effort to understand and appreciate
the works it addresses. But the differences among these production pro-
cesses need not be pursued here since the same textual issues are raised
by all of them.

The fundamental textual question to be asked about prints is how
each of several exemplars can equally be "the work." One answer, from
the point of view of authorial intention, is that if the artist approves these
exemplars and regards them as "the same," each one by definition is the
work. (The artist who makes the image sometimes does the work of pre-
paring the intermediate object and printing from it, though often one or
both of those operations are performed by others, subject to the artist's
approval.) But of course the exemplars cannot possibly be the same in ev-
ery minute detail, for no two physical objects ever are. Even without hu-
man intervention—that is, intentional alteration—prints may vary as a
result of inking differences, for example, or the deterioration of the block
or plate as a result of wear and the passage of time. The artist may reject


Page 34
and destroy certain copies, but even those that are saved and approved
are bound to have slight textual differences. What "the same" means
here can only be that the differences are so small as not to be regarded as
meaningful by the artist. Nevertheless, they may be noticeable to viewers,
whose responses may be affected. Alternatively, perhaps the artist does
not regard all the copies as essentially the same but is willing to consider
the variations as falling within the intended conception of a given work.
Either way, the idea that the work exists in each exemplar is shattered:
one ideally needs to see every copy in order to experience the full range of
textual nuance present in the "work," now taken to comprise the totality
of all the exemplars. (This point reminds one of the comparable necessity,
in studying the texts of verbal and musical works transmitted in printed
editions, of examining multiple copies.)

However impractical the goal of seeing every copy may be, it is worth
pursuing because the differences one will locate may go beyond the small
(but not necessarily insignificant) variants created by the printing process.
They may also include alterations made by the artist (or at the artist's
direction), reflecting early trials or changed intentions. Although a block,
plate, or stone may not carry clear evidence of the alterations made to
it, the impressions taken from it at various times do serve as a record
of the states (or some of them) that it has gone through. This point was
illustrated by a 2004 exhibition at the Frick Collection called "The Un-
finished Print," which gave viewers the opportunity of comparing prints
made from the same plates at different times (see Roberta Smith's ac-
count, 4 June 2004). One of the most dramatic examples consisted of the
first and eighth states of Félix Bracquemond's etched portrait of Edmond
de Goncourt, which are different enough that they could be regarded as
separate works. Although the title of the exhibition implied that states
prior to the last are "unfinished" (as some in the show clearly were),
there is no reason that more than one finished state cannot exist, each
representing the artist's final intention at a particular time. In the case
of Blake's illuminated books, Joseph Viscomi has brilliantly shown—in
his 1993 Blake and the Idea of the Book—that the impressions made at one
time, though differing in small ways, share certain characteristics that
link them together and distinguish them as a group from the impressions
made at another time. Thus a comprehensive examination of the impres-
sions taken from a given plate puts one in a position to judge (and it is
always a matter of judgment) which textual differences can be subsumed
under a single version and which create another version (or even perhaps
a distinct work).

The full textual history of an image created for printing includes forms
not intended by the artist, and they often exist in quantity, for a block
or plate can be used for printing (and be altered) by persons not associ-


Page 35
ated with the artist; and even if the block or print does not survive, there
has been the possibility, for the past century and a half, of reproducing
the image by various photographic means. An important distinction for
textual criticism is the one conventionally made between a "print" and
a "reproduction": the former is the direct product of the block, plate, or
stone on which an image was created, whereas the latter is at least another
generation removed, being a copy (photographic, xerographic, digital,
and the like) either of a print or of another reproduction. (The term "re-
productive print" is sometimes used to signify a print made from an object
that did not involve the collaboration of the original creator of the image,
as when a copyist engraves a plate after a painting, without the painter's
oversight; but of course such a print is still a print, not a reproduction, so
long as it is made directly from the plate.) Whether or not one is focusing
on authorial intention, this distinction is of intense textual significance,
since it refers to differences that profoundly affect viewers' responses.
A print pulled from a relief or intaglio surface, for example, has three-
dimensional attributes, whereas a reproduction of such a print normally
lacks them. Even a reproduction of a lithographic print shows differences,
if less dramatic. Yet the number of people who have experienced repro-
ductions of famous print-images is far greater than those who have seen
the prints themselves; and any study of the history of these works has to
take into account not only the prints, with all their variations, but all the
reproductions as well.

The art of photography raises the same considerations, since the pho-
tographic print is produced from an intermediate object, the negative
(reflecting choices made with the camera and the developing), and since
the handling of the printing process can lead to variant texts among the
finished prints. (Digital photography and printing employ different tech-
nology from traditional photography, but the textual upshot is that varia-
tions, including gross manipulation of the images, can be produced more
easily.) Furthermore, the distinction between prints and reproductions is
just as applicable (even though photography is the most common process
used for making reproductions), since a print is produced directly from
the artist's negative, whereas a reproduction is derived from a photograph
of one of the prints (or from another reproduction). A good example of
a photographer's changing intentions, as seen in variant prints from the
same negative, is offered by Ansel Adams. Late in his life, he reinter-
preted many of his earlier pictures by printing from the old negatives in
a way that replaced "elegance with melodrama," in the words of John
Szarkowski's 2001 exhibition catalogue, Ansel Adams at 100 (quoted in
Sarah Boxer's review, 1 September 2001).

For Adams, each act of printing could be seen in terms of the perform-
ing arts: "the negative," he said in 1943 (and often repeated the idea),


Page 36
"can … be compared to a musical score. It's ready for its performance—
the print. … it can be performed so as to recreate the original visual-
ized intention." But it can also be performed, as his own later printing
showed, in a way that reflects a different intention. Although a musical
work can similarly be performed in diverse ways, Adams's comparison of
photography and music cannot usefully be carried further and indeed is
seriously misleading. Because musical works use an intangible medium,
they must be recreated (performed) each time they are experienced (that
is, in their intended "live" form, not as a recorded reproduction). But
photographs, which use a tangible medium and are stationary, do not
need to be printed afresh whenever one wishes to experience them, since
prints made earlier are physical objects that can be handed down through
time. What "performance" means in the two cases is very different: in
music it is an integral part of the work, whereas in photography it is part
of the work's prehistory. And these points remain the same whether one
is talking about authorial intention or the intentions of others.

Another art form involving multiple exemplars—each of which is a
"print" of an image transferred from an intermediate object—is the de-
sign of printed books, magazines, and newspapers (and of their constituent
parts like advertisements). As with the kinds of prints just discussed, which
can preserve evidence of changes in the underlying plates and negatives
(and other such objects), books and periodicals from a single edition (a
single act of typographical layout and design) vary among themselves as
a result of changes, both intentional and accidental, in the type-formes
or plates used—changes that may occur during one printing session or
between such sessions. The conventional terminology used by bibliogra-
phers is different from that used by art historians, and both have their
limitations. In the book field, an "impression" (or "printing") comprises
all the units produced in a single continuous session (of however many
hours or days are required to complete the desired number of copies),
and a second distinct session, separated in time from the first, produces
another impression. In the study of art prints and posters, on the other
hand, each copy is called an impression—a usage that makes more literal
sense, since each one does result from a separate act of impressing (or at
least printing on) the paper. If the usage in art thus accords better with
the fact that each copy is (if only slightly) different, the bibliographical
usage more clearly accommodates another reality, that a group of copies
made in one limited period of time may share characteristics not present
in groups of copies made in other periods.

But the terminology, despite its awkwardness, should not prevent tex-
tual critics of any of these arts from dealing with variations. The existence
of differences among copies of any given edition of a printed book is
widely understood by editors of verbal texts, for whom such variations are


Page 37
part of the textual history of the works they are investigating. But every
edition can of course be viewed as a work of visual art, in which the art-
istry of type designers, papermakers, and (sometimes) binders is deployed
in a particular way by a book designer; and verbal variation is only one
of the many sources of difference among the exemplars of any such work.
(I am speaking here only of conventional books; artists also create quasi-
book objects that are best regarded as works of sculpture.) Even when
design drafts, pieces of type, type-formes, proofs, and other preliminary
materials survive, the primary evidence for the textual history of a piece
of book design is found in the finished objects themselves, since their
production over time allows them collectively to preserve a record of the
variations that occurred (or some of them). Approaching books as visual
art rather than as vehicles for transmitting verbal language brings us full
circle from where we began.

These notes are intended to illustrate a way of thinking, not to be com-
prehensive. Nor are the sections into which I have divided them meant to
be self-contained: certain issues are dealt with more fully under one head-
ing than under another, but many of those issues are equally applicable
to all the discussions. This interdependence reflects the fact that all the
arts are related and that thinking about the textual criticism of one art
can clarify the thinking about others, including those I have not touched
on, such as the olfactory and gustatory arts. After all, texts of every kind
of human creation are unstable (like the natural objects studied by scien-
tists), and it is this basic condition that textual critics in all fields are track-
ing. They are historians of metamorphosis, chronicling the changes that
have in turn affected the responses to human works at different times.

Textual critics' recognition of the inescapability of impermanence is
not at odds, however, with the urge to produce scholarly editions, which
are attempts to help an audience to encounter various past moments
in the history of a work, rather than merely to read about them. How
textual criticism and scholarly editing are carried out is contingent not
only on the surviving evidence but on the distinctive characteristics of
the different media in which works can be created. Yet the primary issues
are identical, and that is why textual study in any field can benefit from
being conducted with a knowledge of the questions that have arisen, and
the answers that have been offered, in other fields. Approaching every
human creation with an understanding of its textual history, seen against
the panorama of all other textual histories, helps us to appreciate the
humanity movingly embedded in each version of a work and to enjoy the
hard-won accomplishment represented there.


Page 38