Visible and Invisible Books: Hermetic Images in N-Dimensional Space

Jerome McGann

To see a world in a grain of sand
And a heaven in a wild flower,
Hold infinity in the palm of your hand
And eternity in an hour.
William Blake, "Auguries of Innocence"

Or if it indeed be so, that this other Space is really Thoughtland, then take me to that blessed region where I in Thought shall see the insides of all solid things. . . . In that blessed region of Four Dimensions, shall we linger on the threshold of the Fifth, and not enter therein? Ah, no! Let us rather resolve that our ambition shall soar with our corporal ascent. (E. A. Abbott, Flatland. A Romance of Many Dimensions)

All the news organs have picked up the story: "After five centuries of virtually uncontested sway, the Book seems to be facing a serious threat to its power. Informed sources report a large computerized force continues its sweep through traditional centers of bookish institutional control. Resistance has been fierce in certain quarters, and vast areas remain wholly under Book authority. Spokesmen from both sides describe the situation as volatile. According to militia leader Sven Birkerts. .

That kind of report shapes much of the public discussion about the relation of books and an array of new computer-based tools generically named "Hypertext" and "Hypermedia". This report will be different, coming instead from what Dante Gabriel Rossetti called "an inner standing point".<1> I am writing this in the Alderman Library at the University of Virginia, from an office on the third floor, where The Institute for Advanced Technology in the Humanities (IATH) is located. I reach for an apt Rossetti quotation as a matter of course since in this office, under the sponsorship of IATH and with its technical resources, I and a group of English Literature graduate students have spent the past six years developing The Complete Writings and Pictures of Dante Gabriel Rossetti. A Hypermedia Research Archive.<2>

IATH occupies something under 2000 square feet. Anyone who knows libraries knows the premium they put on their floor space. They guard it like a certain dog at the gate of hell. Yet this large area was cleared by the library and given to IATH for its use in 1993. "It has been reliably reported that major centers of Book power throughout the country have been voluntarily joining forces with the Electronic invaders and. . . ."

Those reports are true, but only factually true, like Kenneth Starr's report to Congress. Under a mask of objectivity that kind of reporting generates a different kind of "hyper"media. In matters of some moment it helps to have an inner standing point rather than an agenda. The inner standing point gives you access to the complexities.

Nor is there any doubt that we are, at this millennial moment, passing through the first stages of a major shift in how we think about and manage texts, images, and the vehicles that carry them into our attention. From a literary person's point of view, the relevance of these changes can appear purely marginal: for whatever happens in the future, whatever new electronic poetry or fiction gets produced, the literature we inherit (to this date) is and will always be bookish.

Which is true -- although that truth underscores what is crucial in all these events from the scholar's point of view: we no longer have to use books to analyze and study other books or texts. That simple fact carries immense, even catastrophic, significance.

Trying to think clearly in this kind of volatile situation is not easy. In fact, after working for most of this decade to implement The Rossetti Archive, an online hypermedia tool for studying all the writings and art works of Dante Gabriel Rossetti, I am beginning to see some simple but fundamental truths about books, digital tools, and what we might think about or expect of them. These simplicities are what I want to talk about today. Ultimately I hope to persuade you that setting these two forms of thought and expression into a mutually critical relation -- encouraging each to interrogate and explore the other -- is probably the most fruitful thing we could do right now.

Let me start, then, with a fundamental misconception: that a digital work is prima facie more complex and more powerful than a book. Well, it isn't so, they are just tools designed to manage knowledge and information at different scalar levels. Our worlds are differently constituted by spoons on one hand and by steamshovels on the other. Nor is one of these instruments "better" or more powerful. They do different things. Right now and in the foreseeable future, books do a number of things much better than computers. There is no comparison, for example, between the complexity and richness of paper-based fictional works, on one hand, and their digital counterparts -- hypermedia fiction -- on the other. Nor does the difference simply measure a difference of writing skill -- Italo Calvino, say, versus Stuart Moulthrop. The history of the book medium and the development of fictional conventions within that medium have evolved an extraordinarily nuanced and flexible set of tools for the imagination. The truth is that the "hyper"media powers of the book, in this area of expression if not prima facie, far outstrip the available resources of digital instruments.

But the latter, even in this moment of their earliest history, confidently declare and establish their authority in other areas of knowledge processing and communication. A like situation emerged in the 15th century with the invention of movable type. The printed book quickly supplanted the manuscript as a primary vehicle for storing, retrieving, and transmitting information. On the other hand, even the finest early printed books -- and there are many such -- lack the expressive and intellectual resources available to works produced in the manuscript tradition. That clear deficiency, needless to say, did not hinder the development of printed works -- on the contrary, it inspired and promoted that development.

Today we stand in a similar set of circumstances. My special interests as an educator, a writer, and a scholar have brought me to engage the authority of our new digital tools. I undertook the development of The Rossetti Archive in 1992 as an experimental effort to exploit the special powers of digital technology -- specifically, to try to design a model for a critical edition that would overcome certain of the key limitations of critical editions organized in book form. Scholars need tools that can efficiently manage large bodies of related literary and artistic objects. This is exactly what the traditional critical edition does. But it's clear, prima facie, that digital tools can execute many of the tasks of scholarly editing much better, much more thoroughly and much more precisely, than books can.

For instance, in certain important respects even works of imagination will and should be treated as we might treat ordinary "material objects" like (say) screwdrivers or business records. The corpus of Rossetti's visual and textual works is very large and its interrelations are very complex. Simply building a scholarly space that facilitates accessing these material objects for study and analysis, including complex kinds of comparative study and analysis, is a very useful thing to do. Digital space is in this case a much richer and more flexible space than bibliographical space. So in 1992 we set out to explore and exploit that space by building The Rossetti Archive, and after seven years' work we aren't unhappy with the results. No paper-based book or set of books could have done what The Rossetti Archive offers to scholars. The book medium is physically incapable of the kinds of storing, integrating, and accessing operations we had held out as a basic scholarly demand for the Archive.

But artistic works are also very different from screwdrivers and business records, and our deepest interest in them focuses on those special qualities and characteristics. So we ask of our new electronic tools: can you also help us to see and understand (in a Blakean sense) works of imagination, can you "advance our learning" of such works? And we ask ourselves: how might we manipulate these new tools to meet these special desires and requirements?

Of course, simply to ease access to literary and artistic work enhances our abilities to understand. These results are apparent in such splendid constructions as The Perseus Project or any of the electronic "Archives" being developed, for instance, at University of Virginia's Institute for Advanced Technology in the Humanities - electronic tools centered in the works of Blake, Dante, Dickinson, Rossetti, Whitman.<3> But we want to go beyond what these instruments are designed to do - out farther and in deeper. Can we do this? Are digital resources useful here?

This question brings to mind Edward Tufte's work.<4> Tufte is interested in the ways inventive people have used paper instruments to organize and elucidate various kinds of information. His studies inevitably lead us to an important set of important meta-questions: what is a page, what is a book, what are their parts and how do they function?

One of the most interesting and unforeseen consequences of building The Rossetti Archive has been our encounter with those kinds of questions and problems. Initially it was clear to me how digital tools gave us great practical advantages over paper-based critical editions. The new engines could handle, in full and unabbreviated forms, vast amounts of data -- far more than any book or reasonable set of books. They could also handle different kinds and forms of material data -- not just textual, but visual and audial as well. These capacities made it possible to edit critically certain works that could not be adequately handled in a paper medium: the works of Rossetti, of course, but also those of Burns, of Blake, of Dickinson. The digital tools also exposed the critical deficiencies of the paper-based medium as such. Any kind of performative work -- dramatic works, for example, and pre-eminently Shakespeare's dramas -- gets more or less radically occluded when forced into a bookish representation. These differentials led me to see what I regarded at the time as an important general insight into books, computers, and their scholarly relation. I got a lot of satisfaction out of writing, in 1993, the following dicta:

When we use books to study books, or hard copy texts to analyze other hard copy texts, the scale of the tools seriously limits the possible results. In studying the physical world, for example, it makes a great difference if the level of the analysis is experiential (direct) or mathematical (abstract). In a similar way, electronic tools in literary studies don't simply provide a new point of view on the materials, they lift one's general level of attention to a higher order.<5>
While I wouldn't dissent from those sentences today, when I wrote them I was certainly unaware of much that they implied. My own levels of attention would be considerably raised as we undertook to implement the logical design of The Rossetti Archive between 1993 and 1997.

Some of these matters I have already written about -- for example, the dysfunction that arises when one tries to use standard markup forms, SGML and all its derivatives, to elucidate the functional structures of imaginative works.<6> The recursive patterns that constitute an essential -- probably the essential -- feature of poetry and imaginative works in general cannot be marked, least of all captured, by SGML and its offspring. At first we engaged this dysfunction as a set of practical problems for building, or modifying our original plans for, The Rossetti Archive. But this deep asymmetry between our primary bibliographical data and our digital tools forced us to realize that we would not get very far with our practical problems if we didn't begin to think more rigorously about a pair of difficult elementary questions -- questions that we had hitherto treated far too casually, as if they did not involve problems, for us at any rate, at all.

Here are the questions. First, what is a literary work, what are it parts, how do they function? We assumed we knew how to answer such questions but our attempts to translate our bibliographical materials into coded instructions showed us that we didn't. (The principle here is simple and known to every teacher: if you can't explain what you know to someone else so that they also understand, you don't really know what you think you know.) Second, what constitutes a critical representation of a literary work and how does such a representation function? With one notable exception, every critical method and theory known to me assumes that the measure of critical adequacy is the degree of equivalence that can be produced between the object of critical attention and the critical representation of that object. Behind this view lies the ancient idea that representation is a mirroring relation. As we kept building The Rossetti Archive the flaw in that traditional understanding became more and more clear. A hypermedia work by choice and definition, the Archive therefore obligated us to integrate in a critical way both textual and visual materials. Our efforts were continually frustrated, however, because while digital texts lie open to automated search and analysis, digital images do not. Consequently, our critical mirror never adequately reflected the reality we knew was there. Indeed, so far as the mirror was concerned much of that reality might as well have been a large population of vampires.

I can only give the most schematic indication here of how we are addressing those questions and problems now -- and I say "are addressing" because I am acutely aware of how little we understand about these matters. I will take up the second set of questions first and use that discussion to open up the issues involved in the first -- and logically primary -- set of questions.

For the past few years I and some colleagues have been experimenting with ways of manipulating the texts and images of imaginative works in order to enhance our perception and understanding of how they function. The initial phases of this work have been sketched in two papers: my essay "Imagining What You Don't Know: The Theoretical Goals of the Rossetti Archive", which I wrote in 1997; and a collaborative piece called "Deformance and Interpretation", which I co-authored with the poet Lisa Samuels in 1997-1998.<7> Both of these essays argue, in different ways, that "adequacy" in any critical representation cannot be measured by a scale of equivalence. A true critical representation does not accurately (so to speak) mirror its object, it consciously (so to speak) deforms its object. The critical act therefore involves no more (and no less) than a certain perspective on the object, its acuity of perception being a function of its selfconscious understanding of its own powers and limitations. It stands in a dialectical relation to its object, which must always be a transcendental object so far as any act of critical perception is concerned. This transcendental condition is a necessity because the object perpetually shifts and mutates under the influence of its perceivers. The critical act is a kind of conversation being carried on in the midst of many like and impinging conversations, all of which might at any point be joined by or merge into any of the others.

Works of art recreate -- they "stage" -- a world of primary human intercourse and conversation. As with their reciprocating critical reflections, they manipulate their perceptual fields to generate certain dominant rhetorics or surface patterns that will organize and complicate our understandings. An important critical maneuver, then, involves dislocating or "deforming" those dominant patterns so as to open doors of perception toward new opportunities and points of view. A dominant self-representation of Paradise Lost is to "justify the ways of God to Man". That famous dislocater of texts, William Blake, accepted the literality of Milton's text but utterly deformed its meanings, as it were: to Blake, the words "justify", "God", and "Man" signify in ways that Milton could hardly have imagined. The all but complete inversion that Blake's interpretive moves bring to Milton highlights one of the most important features of imaginative works: that they are incommensurate with themselves at all points.

The Blake/Milton relation highlights the general relation that critical deformations bear to aesthetic incommensurability. Blake knew very well that he had deformed the great Puritan, who was also his master spirit. Blake's works are what he called Buildings of Los(s), consciously written under the rubric "I must create my own system or be enslaved by another man's". That famous declaration draws on a peculiar Blake lexicon, however, where the word "create" and its cognates are synonymous with the word "error". This is why Blake will speak of "Error or Creation" and go on to assert that "Error is Created Truth is Eternal". His brief epic Milton is a deformed reading of Paradise Lost and Paradise Regained. Its acuity -- that is to say, its power to elucidate Milton's work -- is a direct function of the "errors" that it deliberately creates in relation to that work.

Critical deformations can be usefully undertaken either randomly or according to a set of pre-arranged protocols. I have found, for example, that when certain of the standard filter protocols in Adobe Photoshop are applied to paintings - D. G. Rossetti's paintings, for instance - interesting structural features get exposed to view. Using the edging protocol to make arbitrary transformations of a number of Rossetti's pictures revealed, for example, that many of the pictures, and almost all of his famous portraits of women, are dominated by patterns of interlocking vortices and spirals. He plays numerous variations on these patterns, which are evidently the result of conscious purpose. This key structural feature of Rossetti's pictorial work has not been previously noticed or commented upon. It is a feature that leaps into prominence when these random deformations are passed through the pictures.

We now believe that a useful set of image-editing operations could be established that would have two important critical functions: first, to expose characteristic formal features of pictorial works; and second, to release perception from the spell of precisely those kinds of characteristic formal patterns, and open a perception of different arrangements and patterns. For the truth about works of art - textual, pictorial, auditory - is that they are, in Tufte's word, "multivariate".

There is an interesting moral to the story I've just told about critical reading as a deformance procedure. Although I've been familiar with the idea since at least the mid-1960s, when I first read Galvano della Volpe and when my lifelong interest in Blake's work began, I did not come to realize its claim to generality until I encountered the recalcitrance of digital images. Unlike language objects, once a visual object - a painting or drawing or photograph - is digitally reconstituted, it resists any further moves to mirror or translate it. Playing and doodling with digital images in Adobe Photoshop one day - it happened casually and with no deliberate goal in mind - I suddenly saw that the resistance of the image was in fact a critical opportunity, and not an impasse at all.

That realization brought additional unexpected consequences for the way we were conceiving The Rossetti Archive's digital texts and the problems we were having in marking them for automated computational analysis. We knew from the outset of the project that digital images stood apart from the computational resources of the new technology and we came quickly to realize how difficult it would be, except in the most elementary ways, to integrate automated text analysis to the information contained in digitized images. But it was dismaying to discover how much of Rossetti's poetry -- how much of his strictly textual work -- escaped our powers to represent it critically. Although I've already touched on the reasons for these computational deficiencies, some elaboration here will help to clarify how our difficulties were forcing us to rethink in fundamental ways the "nature" of poetical and imaginative works.

Our failures with implementing some of the goals of The Rossetti Archive were bringing a series of paradoxical clarities not only about our digital tools, but even more about the works those tools were trying to reconstitute. We realized that we were making inadequate assumptions about such works, and that we were using tools designed through those assumptions. That realization turned us back to reconsider the logical and ontological status of the original works. I am convinced none of us will get very far with our new digital tools unless we first undertake a thorough reconsideration of this kind.

First of all, a little history. The discipline of Humanities Computing developed in the field of linguistics, where scholars realized that computers would be extremely useful for carrying out automated pattern searches across large bodies of linguistic data. As a consequence, the textual corpus, even if it was in fact a poetical corpus, was framed for computational purposes as if it were informational or expository. Consequently, the tools that emerged to mark electronic texts for search and analysis also assumed that their object would be the exposure of the informational content and expository structure of the text.

The problem is that poetical works, insofar as they are poetical, are not expository or informational. Because works of imagination are built as complex nets of repetition and variation, they are rich in what informational models of textuality label "noise". No poem can exist without systems of "overlapping structure", and the more developed the poetical text, the more complex are those systems of recursion. So it is that in a poetic field no unit can be assumed to be self-identical. The logic of the poem is only frameable in some kind of paradoxical articulation such as: "a equals a if and only if a does not equal a".

Let me illustrate the truth of that formulation with a couple of very traditional interpretive examples. I'll begin with a famous sonnet by Gerard Manley Hopkins that illustrates in a dramatic way how textual objects of this kind are not self-identical.

As Kingfishers Catch Fire
1:   As kingfishers catch fire, dragonflies dráw fláme; 
2:   As tumbled over rim in roundy wells 
3:   Stones ring; like each tucked string tells, each hung bell's 
4:   Bow swung finds tongue to fling out broad its name; 
5:   Each mortal thing does one thing and the same: 
6:   Deals out that being indoors each one dwells; 
7:   Selves??goes itself; myself it speaks and spells, 
8:   Crying Whát I do is me: for that I came.
9:   Í say móre: the just man justices; 
10:   Kéeps gráce: thát keeps all his goings graces; 
11:   Acts in God's eye what in God's eye he is?? 
12:   Chríst??for Christ plays in ten thousand places, 
13:   Lovely in limbs, and lovely in eyes not his 
14:   To the Father through the features of men's faces.

The first statement in this text offers a paradigm of its duplicities. The word "As" here operates simultaneously in a formal and in a temporal sense (so here it means both "Just as" or "In just the way that" and also "While" or "At the same time as"). The repetition of the word in line 2 underscores its variational possibilities because the poem's second statement introduces an altogether new grammar. Then comes what at first might be taken for a synonym of "As", the word "like" which introduces the sonnet's third syntactic unit (running from the third word of line three through line four). This unit of syntax appears to have the same general form as the sonnet's opening unit, but when we press it more closely we watch it shapeshift into a new and unexpected grammar. Once again the move comes through duplicitous word usage. The word "like" here functions simultaneously as a conjunction (a synonym for "as"), as an adverb (meaning "alike"), and as a noun (in the sense of "kind", as in the word "mankind").

There's nothing unusual about this passage from Hopkins. Poets do this kind of thing all the time, it is the very essence of poetical textuality. I choose the passage not exactly randomly, however, but because its complexities are so apparent and so dramatic. In four lines an amazing kind of textual metastasis has unfolded, nor have I even come close to an adequate exegesis of what is happening here. The phrase "catch fire", for example, normally suggests - as our dictionaries tell us - a passive eventuality, but in this case a feedback loop causes another textual metamorphosis, so that the word "catch" turns active, as if this kingfisher were catching fire as it hunts and catches fish. This transformation occurs because the phrase is affected retroactively, as it were, by the syntactic rhymes that immediately follow the phrase in the next two lines ("kingfishers catch fire", "dragonflies draw flame", "stones ring").

Imaginative textual objects regularly work through these kinds of transformations, feedback loops, and complex repetitions. All are forms or types of what we call "rhymes", that staple poetic device illustrating the algorithm I set out above: "a==a if and only if a=/=a".

The nonhierarchical character of these transformations and rhymings emerges very clearly in the sestet of this sonnet. Look carefully at lines 12-14. The word "plays", probably the pivotal word in the poem, involves a most cunning kind of textual wit. It conceals a pun whose "other meaning", so to speak, is "prays". Why is this so? Because the word is syntactically linked to a predicate complement that only comes to us in the final line, in the phrase "To the Father". The text of the poem generates the literal phrase "plays...To the Father". The oddness of that phrase doesn't reach us until we have transacted the hiatus of line 13, however, when we suddenly realize that the text has been (mis)leading us to reconstitute the phrase into something more linguistically apt. No one reading such a phrase in the poem's plain context of religious usages can fail to hear the absent but secretly prepared alternative phrase: "prays. . .To the Father". This is simultaneously a playful and a prayerful text.

But the text has not finished with its games of self-generation and self-transformation. For the play/pray wordgame regenerates itself yet again in a kind of conceptual meta-text: the word "prays" means as well "praise". The poem as a whole is a kind of playful prayer of praise "for" Christ and "To" the Father, the word "Christ" being here the text's key figure of individuation, or what Hopkins called "selving".

In all this commentary I've tried to keep my remarks free from any kind of thematic or ideational/ideological references. Everything I've talked about has to do with Hopkins's text as a functioning sign system, a structure of signifiers and signifieds. I've done this not because I think "meaning" in a referential sense isn't a crucial part of every textual field, but because I want to demonstrate how full of meaningful activities these fields are even when their referentialities are held in abeyance. Look again at line 12 of the sonnet and think about how it prepares us to register the word game that only gets fully exposed in line 14. In line 12 Hopkins has made a text that our mouths will find difficult to transact: "Christ -- for Christ plays". The problem comes as we try to negotiate a passage from those 3 r's to the l in "plays". Our mouths would find it easier to read "prays" here rather than "plays", we have to make some physical and mental effort to ensure that we get the given phonetic sequence right. The effort is a perceptual signal that our bodies will not let our minds forget when we come to line 14. And we are prepared for this exercise with r's and l's because the sonnet in fact opens its textual field in line 1 with a major deployment of just those phonetic signs.

What is this kind of text, really? First of all, it is both -- and simultaneously -- a perceptual and a conceptual event. Informational texts seek to minimize their perceptual features in the belief that texts calling attention to their vehicular forms interfere with the transmission of their ideas. The textuality of poetry reminds us of the intimate part that phonetics play in the signifying operations of language. It also reminds us of a second important feature of text: that while it may deploy ordered, even hierarchical, structures of ideas, its object (as it were) is to play with and within such structures and not be consumed by them. Are there such things as pure, non-languaged "ideas"? Perhaps. However that may be, when ideas function textually, they commit themselves to fields of perception as well as systems of conception. So in the case of this sonnet we will want to see that while Hopkins's Scotist ideas play throughout the text and even comprise its argument, the sonnet is not comprehended in those ideas or reducible to a Scotist description or exposition. No textual event -- not the Scotist word "Selves", not even the word "Christ" -- is ever self-identical or self-transparent. Most especially is this true for imaginative texts -- where alone we will see an effort to exploit the full resources of textuality.

Let me point out one other feature of this text, a moment of its physical visibility that we may hardly recognize as a visible thing. The wordplay realized in line14 (Plays/Prays. . .to the Father") would fail in its remarkable effect were it not for the hiatus in lines 12-13, a hiatus that is constructed as a visible space and a temporal rhythm. I leave for another occasion any discussion of that temporal rhythm and its perceptual character because I want to concentrate here on the visible forms being deployed.

We tend not to notice an elementary fact about printed or scripted texts: that they are constituted from a complex series of marked and unmarked spaces. The most noticeable are the larger regular units -- the lines, the paragraphs, or (in verse) the stanzas, as well as the spaces between them. Every one of these spatial units, as well as all the others on a page or in a book, offer themselves as opportunities for non-lexical expression. For a helpful comparison think of the cartoon strip with its sequence of frames separated by gutters. The force of cartoon narrative is always a function of the energy generated in those gutters, where the work's inexplicit but crucial relations are solicited in the reader's imagination. Ballad poems regularly treat their stanzas in exactly the same way, and all good writers learn to exploit the spatial fields of their texts. A procedural gap organizes the continuous play of differences between the physical lines of a poetic form and the grammatical order playing in the form. The divisions in long poems and prose fictions create opportunities for building relational nets across the framed areas of the text.

It is highly significant that readers of books move from recto to verso, that their field of awareness continually shifts from page to "opening" (i.e., the space made by a facing verso/recto), and that the size of the book -- length, breath and thickness -- help to determine our reader's perceptions at every point. Texts are not laid out flat on plane pages, and if I were to open the subjects of typefaces or calligraphic forms, of ink, of paper, and of the various ways marks can be scripted or printed, the multivariate manifold of the book would be easily recognized. Entering those subjects shows why a fine press book is not just another pretty face -- at least not the ones that have given thought to themselves. When William Morris re-issued The House of Life, his friend Rossetti's masterwork, as a Kelmscott Press book, the point was to help readers perceive the sonnets more thoroughly than they might in the trade editions. The Kelmscott edition radically alters the spatio-temporal field of the sonnet sequence. It is nothing less than what we would now call a new "reading" of the sequence.<8>

But even these examples can be misleading if they suggest that bibliographical space is a matter of solid geometry. To help dispel that possible illusion I offer the example of a seventeenth-century poem titled "To the Post Boy". This example comes to shift our angle of focus, so to speak, and to expose networks of dispersed visibilities.

To the Post Boy
1:   Son of A whore God dam you can you tell 
2:   A Peerless Peer the Readyest way to Hell? 
3:   Ive out swilld Baccus sworn of my own make 
4:   Oaths wod fright furies and make Pluto quake. 
5:   Ive swived more whores more ways than Sodoms walls 
6:   Ere knew or the College of Romes Cardinalls. 
7:   Witness Heroick scars, look here nere go 
8:   Sear cloaths and ulcers from the top to toe. 
9:   Frighted at my own mischeifes I have fled 
10:   And bravely left my lifes defender dead. 
11:   Broke houses to break chastity and died 
12:   That floor with murder which my lust denyed. 
13:   Pox on it why do I speak of these poor things? 
14:   I have blasphemed my god and libelld Kings; 
15.   The readyest way to Hell come quick?? 
Boy						 nere stirr 
16:   The readyest way my Lords by Rochester.

This work illustrates another mode of textual instability operating at a trans-linguistic level. The issue gets focused as a problem of attribution: we aren't sure who authored this work, and the uncertainty affects every aspect of the poem's textuality.<9> Most of the primary textual witnesses - late 17th and early 18th century manuscripts and printed texts - assign the poem to Rochester, hence seeing it as an astonishing piece of self-directed satire perhaps designed to frustrate and undermine his enemies and their literary devices. The dialogue-poem would be coming to show that Rochester could write satire, even against himself, that none of his antagonists could match.

But certain early witnesses, as well as some later scholars, don't read the poem as Rochester's but as the work of one of his enemies.

The issue, on current evidence, is in fact undecidable, although scholarly opinion today inclines toward favoring Rochester's authorship. (Not very long ago opinion went the other way.)

The poem therefore gets framed in three optional ways: as Rochester's work, as the work of someone else satirizing Rochester, and as a kind of duck-rabbit lying open to either and both readings simultaneously. Those frames, we want to remember, are part of the textuality of the work and they are deeply imbedded. But they run through the text in visibilities that extend far beyond what we might register as the work's plane or solid geometries. Indeed, they only appear as bibliographical and manuscript data scattered in disparate and disjunct materials - documents now housed separately in many libraries (the British Library, the Victoria and Albert Museum, Ohio State University library, the Osterreichische Nationalbibliotek, Vienna, and the Bodleian). In those documents and their complex interfaces we trace out that crucial and fundamental feature of every text: its transmission history - which is to say, we trace out the remains of those earliest readers who half perceived and half created this text.

Every document, every moment in every document, conceals (or reveals) an indeterminate set of interfaces that open into alternate spaces and relations.

Traditional criticism will engage this kind of Radiant Textuality more as a problem of context than a problem of text, nor is there any reason to fault that way of seeing the matter. But as the word itself suggests, "context" is a cognate of text, and not in any abstract Barthesian sense. We construct the poem's context, for example, by searching out the meanings marked in the physical witnesses that bring the poem to us. We read those witnesses with scrupulous attention, that is to say, we make our detailed way through the looking glass of the book and thence to the endless reaches of the Library of Babel where every text is catalogued and multiply cross-referenced. In making this journey we are driven far out into the deep space, as we say these days, occupied by our orbiting texts. There objects pivot about many different points and poles. The objects themselves shapeshift continually and the pivots move, drift, shiver, and even dissolve away.

"Ah, a cosmological metaphor," you will tell me, "for thinking about books and texts."

But in that metaphor, I ask you, what is the figure and what is the ground? The metaphor itself has a bibliographical history that might be traced and described. Which came first, as it were, the metaphor or the book? After Derrida, it's harder than ever to say. But it's not hard to say that what we register as the phenomenal world has been a bibliographical function for more than two millenia at least.<10>

Not hard to say, perhaps, but still difficult to realize or know: because our models for knowing have been shaped in scientific models cast in informational and expository forms. Those forms do not normally cultivate self-reflection, however deeply they may reflect upon matters they set apart from themselves to observe and interrogate; and least of all do they practice self-reflection on their medium of exchange.<11> But that kind of reflection is precisely what happens in imaginative work, where the medium is always the message, whatever else may be the subjects of the work.

Content in poeisis therefore tends to involve more broadly "semiotic" rather than narrowly "linguistic" materials. The perceptual features of text are as apt for expressive purposes as the semantic, syntactic, and rhetorical features -- at least so far as the poets and readers who make such texts are concerned. Every feature represents a determinate field of textual action, and while any one field might (or might not) individually (abstractly) be organized in a hierarchical form, the recursive interplay of the fields appears topological rather than hierarchic. The organization is more like a mobile with a shifting set of poles and hinge points carrying a variety of objects, many of an "opposite and discordant" character, as Coleridge might say.

Which bring me back Edward Tufte and the opening sentence of his influential book Envisioning Information (1990):

Even though we navigate daily through a perceptual world of three spatial dimensions and reason occasionally about higher dimensional arenas with mathematical ease, the world portrayed on our information displays is caught up in the two-dimensionality of the endless flatlands of paper and video screen" (12).

So acute and arresting is Tufte's appreciation of textual graphics that we tend to pass over a crucial piece of misinformation that his work has envisioned. Despite what he says, we do not "navigate daily through a perceptual world of three spatial dimensions", although it is true that we often think we do and even represent ourselves as doing so. Nor are we doomed, when we transact our books and our monitors, to "the two-dimensionality of the endless flatlands of paper and video screen". Even our daily movements are "multivariate"and n-dimensional, and when we imagine ourselves passing through a world of three dimensions we are merely surrendering to a certain type of perceptual filter. It is a filter regularly exposed and repudiated by an imagination like William Blake's, as my epigraph suggests. Every page, even a blank page, even a page of Dan Quale's prose, is n-dimensional. The issue is, how clearly has that n-dimensional space of the page -- its "multivariate" character -- been marked and released?

To see that truth about paperspace seems to me especially useful in an age fascinated to distraction by the hyperrepresentational power of digital technology. We want to remember that books possess exactly the same powers, and we want to remember not simply to indulge a farewell nostalgia at the twilight of the book. One of the great tasks lying ahead is the critical and editorial reconstitution of our inherited cultural archive in digital forms. We need to learn to do this because we don't as yet know how. Furthermore, we scholars need to learn because it is going to be done, if not by us, then by others. We are the natural heirs to this task because it is we who know most about books.

When we study the world of books with computers we have much to learn from our subjects. In crucial ways, for instance, a desk strewn with a scholar's materials is far more efficient as a workspace -- far more hypertextual -- than the most powerful workstation, screen-bound, you can buy. Or consider this: if these new machines can deliver stunning images to our view, the only images they understand are their own electronic constructions. Original objects -- visual, audial -- remain deeply mysterious to a computer. If a computer serves up, say, a facsimile of Rossetti's painting The Blessed Damozel, its most effective means for understanding that image -- for analyzing it -- are through sets of so-called metadata, that is, logical descriptions introduced into the electronic structure in textual form. Even when (some would say "if") that limitation gets transcended, logical ordering through metadata will never not be a part of computerized scholarship of literary works. The objects of study demand it -- just as the physical sciences, for all their use of mathematical models, cannot do without empirical investigations.

And there are more serious problems. Scholars are interested in books and texts as they are works of "literature" and imagination, but those who design computerized tools sometimes seriously misunderstand their primary materials. So far as I can see, nearly all the leading design models for the scholarly treatment of imaginative works operate from a naive distinction between a text's "form" and "content". So in a recent essay the brilliant computer-text theorist Steven DeRose writes that "A book is 'the same' if reprinted from quarto to octavo and from Garamond 24 to Times 12 in all but a few senses".<12> Aldus and the fifteenth-century humanist printers knew better. Those "few senses" are never non-trivial, and in many cases -- a list is too easy to develop -- they carry the most profound kinds of "content".

DeRose's view is now commonplace among those who are making decisions about how to design scholarly tools for the computerized study of literary works. Poems for example are inherently non-hierarchical structures that promote attention to varying and overlapping sets of textual designs, both linguistic and bibliographical. But the computerized structures being imagined for studying these complex forms approach them as if they were expository, as if their "information" were indexable, as if the works were not made from zeugmas and puns, metaphors and intertexts, as if the textual structure were composed of self-identical elements. Some textual information in poems is indexable, but nearly everything most salient about them is polyvalent. So far as imaginative works are concerned the equation remains: a equals a if and only if a does not equal a.

Not to despair, however. Like the appearance of the codex nearly two-thousand years ago, like the advent of printing in the fifteenth-century, the computer comes bearing great promise to literary scholars.

"But will we be assimilated? Is resistance futile?" There are no aliens here, no struggle between books and computers. From now on scholarship will have both, willy-nilly. The question is -- the choice is -- whether those with an intimate appreciation of literary works will become actively involved in designing new sets tools for studying them.


  • 1. This paper is a revised version of a report I gave under the rubric "What is Text?" at the 1999 joint ACH/ALLC annual convention on humanities computing held at U. of Virginia.Back
  • 2. The Rossetti Archive is scheduled for public online release in July 2000. Back
  • 3. For the Perseus Project see:; the homepage of the Institute for Advanced Technology in the Humanities is:
  • 4. See especially his Envisioning Information (Graphics Press: Chesire, CT, 1990). Back
  • 5. "The Rationale of HyperText", in Electronic Text. Investigations in Method and Theory, ed. Kathryn Sutherland (Clarendon Press: Oxford, 1997), 20 (online version at http://~jjm2f/rationale.html).Back
  • 6. See for example "Hideous Progeny, Rough Beasts. Editing as a Theoretical Pursuit," in TEXT (U. Of Michigan Press: Ann Arbor, 1998), 1-16 (online version at href://~jjm2f/chum.html). Back
  • 7. The latter was printed in New Literary History (Winter 1999), 25-56; the other is an online essay accessible at: http://~jjm2f/deform.html. Back
  • 8. I've given attention to this kind of elementary bibliographical "expressivity" for two reasons. First, it gives a simple but arresting reminder of the determinate materiality of every language form. (An example for an oral event of language would not be difficult to construct). Second, although the matter is beyond the scope of this essay, I believe that an analysis of these kinds of visible features of text holds promise for exploiting computerized resources for the interpretation of imaginative works. A more detailed treatment of this matter is given in a paper (unpublished) being written in tandem with this one. The latter lays out the theoretical basis for a set of experimental operations presently being carried out by several people at the Institute for Advanced Technology in the Humanities. Back
  • 9. For a discussion of this poem and its attribution problems see David M. Vieth, Attribution in Restoration Poetry. A Study of Rochester's Poems of 1680. (Yale UP: New Haven, 1963), 199-203. Back
  • 10. Because our knowledge of phenomena is a textual/bibliographical function.Back
  • 11. The twentieth-century has produced a number of notable exceptions among scientists from Einstein and Godel to G. Spencer Brown and Roger Penrose. Brown's Laws of Form (1969) is particularly important, not least because of Brown's self-conscious use of textspace to develop his argument. Back
  • 12. See DeRose's essay "Structured Information. Navigation, Access, and Control" ( Back.