The monsters and the textual critics

Tom Davis

Imre Lakatos, in his brilliant deconstruction of the practice of mathematics [1], suggested that one of the ways that mathematicians cope with exceptions to a theorem is by a strategy that he called 'monster-barring'. You simply label the exception as so bizarre as not to qualify as a true refutation, and then can behave is if it did not exist: as the Lacanians would say, it becomes part of the Other. This strategy, however, as both Lakatos and Lacan knew well, does not work. A problem is a problem, and does not cease to exist by a simple re-labelling. The theory of textual criticism, in my view, is beset by these monsters, in the form of paradoxes, impossibilities, deceits. This paper is an attempt to let the monsters speak, and to see if thereby they can be accommodated. In its first version, this paper was rejected from the Festschrift for which it was written on grounds of flippancy. It became, in fact, itself a monster. It is very tempting to take this as vindication of its contents.

I would like to offer three propositions.
1 Textual criticism is necessary.
2 Textual criticism is impossible.
3 Textual criticism is universal.
In other words, we must go on, we can't go on, we go on. For Becket this seems to have been agonising, and for textual critics too, life sometimes seems a little insoluble; beset, in fact, by monsters of every kind. But for the rest of us, this is how life is lived, from compromise to compromise: something we do constantly, without too much trouble.

In order to demonstrate this I will begin with a classic textual crux, from Othello, which is intended to demonstrate the impossibility of textual criticism. I will then discuss how textual criticism is something we do all the time, and offer a description of this real-life textual criticism. Next I intend to compare and contrast this with textual criticism proper, and show (I hope) how much of the theoretical argument inspired by the latter arises from a process of mystification, that obscures the essential identity of the two processes, specifically with regard to two traditional causes of dispute: the science/art distinction in textual criticism, and the problems around author's intention. I end with a personal reminiscence, a cheerful conclusion, and the answer to a riddle proposed by A.E.Housman.

When I was an undergraduate, in a more heroic age of literary criticism, there seemed to be two views available of the play Othello. One of them, which one could call the Tom Brown view, was associated with the Victorian critic Bradley, and depicted the hero of the play as an essentially decent chap who had been led astray in an inadvertent moment. The Tom Brown view went with the Flashman view of Iago, naturally. The critic F.R.Leavis, on the other hand, who had spent the first world war in a Quaker ambulance unit, was understandably less enthusiastic about generals, and reread Othello as the villain of the play; a superficial individual, the leading astray of whom was no problem at all. This view naturally went with the demotion of Iago to a narrative convenience. One of the arguments in favour of Leavis's view, which I remember finding particularly plausible, centred round Othello's final soliloquy; a damaging blow to the Tom Brown theory was that Othello, having transgressed rather seriously even by the more relaxed ethical conventions of Jacobean tragedy, doesn't appear to be particularly apologetic. In later years, as the result of my metamorphosis from Leavisite to textual bibliographer, I discovered that that really rather depended on which text you read.

For instance, if, like a large proportion of the undergraduate population you happened to consult the New Arden Shakespeare, you would read a text based in a crucial place on the Quarto of 1522 [2], the original of which reads as follows:
When you shall these vnlucky deedes relate,
Speake of them as they are, nothing extenuate,
Nor set downe ought in malice; then must you speake,
Of one that lou'd not wisely, but too well:
Of one not easily iealous, but being wrought,
Perplext in the extreame; of one whose hand,
Like the base Indian, threw a pearle away,
Richer then all his Tribe...
Here Othello is comparing himself to an Indian, who would not understand the value of what he has just thrown away, however precious to others; his act in so doing could therefore not be blamed. The primary signification of 'base' would therefore have to be, not moral, but social: low in the scheme of things. Since what he is talking about is the brutal murder of his own wife, it is hard not to see this, as Leavis would say, as self-justification.

If however you happened to buy the cheap Signet edition, which believes (following Greg) that the Folio is right wherever it is not obviously wrong, then the text you would read would derive from this:
When you shall these vnluckie deeds relate,
Speake of me, as I am. Nothing extenuate,
Nor set downe ought in malice.Then must you speake,
Of one that lou'd not wisely, but too well:
Of one, not easily Iealious, but being wrought,
Perplexed in the extreame: Of one, whose hand
(Like the base Iudean) threw a Pearle away
Richer then all his Tribe
Here, in the Folio [3] version, the situation is quite different. Opinions have varied slightly as to who this Judean might be, but a solid majority feels that Othello is actually to be comparing himself to Judas, the supreme betrayer, whose remorse was so great at the realisation of what he had done, and lost, that he, as Othello is about to do, committed suicide. In that case 'base' now implies a very heavy self-condemnation­p;the heaviest possible, in fact­p;and Leavis's unrepentant Othello vanishes into thin air.

What are we to do with this situation? Or, rather, what are literary critics to do? The first thing that we can observe is how important textual criticism is to literary criticism: how completely and utterly literary critics are at the mercy of the vagaries of textual transmission, and, in the case of Leavis (notoriously) but of many others too, how little they seem to be aware of this. Here we have two famous and seminal interpretations, that affected the reading practice of generations, almost entirely dependent (it is not too strong to say) on variation of one letter in one word; and each critic is completely unconscious of this fact. It is a distinguishing characteristic of the way that literary texts are read that immense importance can suddenly be invested in a single word, letter, or even comma, out of any that compose it unpredictably. Other texts­p;computer programs, for instance, or Bank Statements­p;may similarly have momentous consequences attached to tiny items, but this significance is predictable. With literary texts, as Peckham [4] has pointed out, you can never tell which word may be seized upon and installed as the core of an argument.

The second piece of information that we can derive from our reading of the readings in Othello is that there is absolutely no way of knowing with any degree of certainty at all which of the two readings to choose as the basis of our interpretation of the play. Let us look at what a textual editor might do with this problem.

Essentially there are two ways of doing textual editing, which correspond to what used to be called the Art and the Science of textual criticism. The Science approach in this case will begin, perhaps, with the insights of analytical bibliography, which hopes to be able to tell us which compositor set which bit of text, what the relation is between the Folio and the Quarto (though in this latter instance I suspect the bibliographer would have to use the techniques of the traditional textual critic, collation and recension) and so on. To cut a long story and much debate short, there is no agreement of any significance between textual editors that will help us at all with the Indian/Judean problem: neither of the two witnesses has preponderant authority for any instance of variance. Greg's view as to the supremacy of the Folio, in spite of the respect which his opinion properly attracts, is not widely accepted, but no alternative view seems to have clearly emerged.

Another Science solution would be to offer the heady possibility that we are talking about a turned letter: an /n/ upside down. Unfortunately this is not the case; if it were, the letter would appear to be raised above the base-line, and neither is. Bibliography, like its estranged twin Forensic Science, abounds in such potential solutions, that never seem to actually actualise themselves for the instance at hand, and the literature of both is full of papers that offer beautiful solutions­p;if only one could find a suitable problem for them to solve.

All that Science can tell us is that the problem is probably paleographic in origin. In other words, somewhere behind one or other version there may well have been a bit of handwriting in which it was possible to mistake the /u/ for the /n/ or vice versa. This doesn't help much, except in one important respect: if it is true, it makes it unlikely that Shakespeare wrote both words, one as a revision of the other.

But apart from this there is nothing that we can do except use Art: or, as it is commonly called, literary judgement. We must look at the context of the whole play, and of Shakespeare's plays, and of what we know about literature, to find the answer. In this activity the textual critic and the literary critic are effectively one and the same.

The arguments between those who favoured Art and adherents of the supremacy of the Science approach are now, mercifully, part of the interesting history of mid-twentieth century Bibliography, and there is a general air of comfortable agreement that when Science won't help you, Art will have to do; unreliable and disputatious though it may be. Or, to put it in the normal and more respectful and optimistic terms, the textual editor will, when other methods are not useful, resort to literary judgement; he or she will operate as a literary critic. However, this is to some extent a cover-up, as every textual critic faced with the sheer daily insolubility of textual cruces knows very well.

This, I would suggest, is because there are two essential differences between literary and textual criticism, which are disguised by the reverence attaching to the concept of 'literary judgement'. The first difference is this. The readings produced by textual critics, once produced, are taken as true; the person who produced them, and his or her labours, become immediately invisible. The edited text is what people read, not the lemmata or even, much, the textual introductions; and they expect this text to be correct. I would suspect that Leavis's un[textual]critical approach to Othello, after nearly half a century of sustained propaganda about the importance of textual criticism, is still the norm rather than the exception. However influential the readings of a literary critic, no-one treats them with the same­p;quite extraordinary, when one thinks of it­p;faith which is commonly applied to the labours of textual criticism: these are taken for granted as being the true, correct, and only readings, whereas everyone knows that a literary critical reading is only an interpretation, that can be readily superseded by a more plausible one, and no doubt will be, as fashions change.

Secondly, even if this misapplied faith were not applied, and the reading public, having learned the tedious mysteries of textual critical analysis, knew just how provisional the apparent solidity of the edited text often is­p;even then the textual critic's position would not be much improved: there is another, and much more painful, distinction between textual and literary criticism. Literary critics make an entire profession out of not agreeing with each other: interpretations must be individual, or they are worthless. But textual critical judgements aren't like that: their job is not to be interesting, but (in some sense) to be 'true'. So, not surprisingly, there is definitely no settled literary critical opinion as to the meaning of Othello; how is a textual critic to operate? Telling him or her to behave like a literary critic doesn't really help matters, I would suggest; in fact, the reverse.

In practice, as a reading of the textual notes attached to this particular crux in various editions of Othello makes clear, this literary-critical technique has not produced anything like a single view, or a solution to the problem. For instance (taken more or less at random):
Othello is lamenting his ignorance, stupidity, gullibility and descent to savagery, none of which is applicable to Judas. the kiss Othello gives Desdemona is hardly a sign of betrayal so much as an instinctive sexual response at odds with his mental conviction of her guilt. [5]

Judas did not betray through ignorance of the value of his victim, nor did he throw his 'pearl' away, but received thirty pieces of silver. Nor did Othello betray Desdemona in intention, but only in fact. [6]

F. 'Iudean', which some prefer and take as referring to Judas. But he was the archetype of treachery, of which Othello, an 'honourable murderer', never accuses himself. [7]
And so on. What is remarkable, apart from the air of weighty certainty of these insubstantial suppositions, is the fact that while the textual critics are dutifully behaving like literary critics, they seem to be completely unaware of the rather famous critical debate about this play, which Leavis initiated (for non-readers of Scrutiny) in 1952; as unaware, in fact, of Leavis as he was of them.

Nonetheless what the textual critics are doing is what textual critics do: they pretend to be literary critics: to behave with some confidence as if they knew the answer, and to argue a case for it. This seems natural: after all, both editors and readers of edited texts have been trained to behave in just this way, in undergraduate literary courses. But as I have said there is a clear distinction between the kind of truth content attaching to literary interpretation and that which is appropriate to textual criticism, and to confuse one with the other, and use rhetoric to reinforce it, only serves to conceal what I think should be clear from the Othello crux: the answer is impossible. We cannot know. The degree of knowledge attainable by textual critics, at those points­p;the cruces­p;where their Art or Science is truly tested, is not commensurate to the importance placed by literary critics on its results. In that sense, it seems to me, textual criticism cannot be done.

In spite of all this, it is still true that textual criticism, in the widest sense of the words, is something that we all do; it is a natural and instinctive activity. We all edit the world we perceive, otherwise we could not make sense of it. If the world is a book, then in order to read it we have no choice but to be editors. In a slightly narrower sense, that is with regard to all of the many linguistic texts that we meet in everyday life, both written and spoken, we expect, as part of the normal course of nature, a certain amount of corruption, and make our ad hoc emendations as often as we find it necessary without giving the matter too much thought. In speech, if we mishear, or hear what we take to be a slip of the tongue, we can either silently edit to what we believe is a more acceptable version, or we can ask the speaker if that was what they meant to say. In my daily paper I expect a misprint or two, and emend as necessary; if I can't work out the true reading there is perhaps a minor irritation, but no more; it is not worth spending the time, over breakfast. If I really need to know what the author meant, say in the case of a mail order advertisement for something that I might wish to buy, I can phone, or write, or go ahead and test my conjectural emendation by acting upon it and seeing what happens. Our inevitable response to the inevitable muddle is that we negotiate, compromise, and muddle through; specifically, we try quite hard to negotiate a balance between the amount of effort spent in editorial activity and the value of that results from it. Were it not for this universal balancing my daily newspaper would be deluged with phone calls and letters, and breakfast would stretch into the middle of the afternoon.

Indeed, perfect accuracy is not something we particularly like: it seems soulless and stupid. My word-processor is perfect. It responds with idiotic perfection to every keystroke I make, right or wrong: it doesn't edit the world in any way whatsoever. All of the errors which, since (unlike it) I happen to be human, find their way into the process of typing, are reproduced with complete fidelity. It behaves, in fact, like that impossible ideal of the classical textual critic, the scribe who is an accurate moron. This fidelity, this lack of a textual critical function, is so counter-intuitive that one of the first add-ons developed for the basic word-processing program was a crude textual editor in the form of a spelling checker.

Before I acquired the word processor I would write in my not very legible longhand, and give the result to the Departmental Secretary, who, because she is also a human being, is a textual editor, and would correct my spelling mistakes, using, as humans do, whatever seemed to her to be an appropriate amount of effort to make sense of the bit of the universe at hand­p;in this case, my illegible longhand. And, being human, she would also make mistakes, which I would then edit. Corruption is universal; textual criticism is natural to humans. We don't give the matter much thought.

Perhaps for this reason it is important to think about it. What do we do, when we do textual criticism? In doing this universal human activity we operate, I would suggest, a very simple, but, for its purpose, quite adequate communications model. Something like this. A message originates (by some occult means) in someone's mind. In order to communicate it he or she encodes it in any one or more of a wide range of physical forms (paper, sound waves, electronic impulses) by which means it is transmitted to the recipient. Unfortunately the process of rendering this rather insubstantial object substantial also tends to introduce what information theory people call 'noise', and what textual critics, with their characteristic leaning towards theology, call 'corruption'. The recipient has to decode the immaterial message from the material vehicle, and this process will often involve a certain amount of editing out of this 'noise' so that the message is received in a form as free from it as is thought necessary for adequate comprehension.

Now, this model is useful, but it is only useful for the purposes to which it is normally put: that is, everyday commonsense textual criticism. It is however a powerfully persuasive explanation of what goes on, and therefore has been roped in to account for what goes on in a much more complicated task: the textual criticism of literary artefacts. How, in practice, does this everyday model work?

I will offer as an example the address of a letter sent to me once:
Mr T.R. Davis
Lecturer in Didliography and Paleography
Department of English
University of Birmingham
I would suggest that when we are confronted with a message like this, the following process, or something like it, takes place:
1. Something is wrong

The message doesn't fit with our (lexical and encyclopedic) expectations of what it should be like. In this instance, I noticed immediately, having studied many envelopes addressed to myself, that this was deviant, and my lexical/encyclopedic knowledge told me that (wishful thinking aside) 'didliography' is not the normal term for the study of the history of printing and allied matters.
2. What is the 'true reading'?

Almost at the same time as stage (1) notices the deviance, the answer suggests itself: the 'error' evokes the norm in the act of identifying it as deviant.
3. How did the 'error' come about?

This part of the process is optional, and often feels supernumerary or not worth the trouble. If I conjecture that a typist somewhere must have made a mistake from dictation, on the basis of my knowledge that (a) typists are often dictated to, and (b) the sounds associated with the graphemes /b/ and /d/ are heard as resembling each other in English, then this feels like a different kind of process from (1) and (2). Firstly it is conscious and worked out, while the other steps are intuitive and rather fast. Secondly in this case it is unnecessary, because the range of solutions to the problem is highly limited and easily available: the answer is obvious, and all stage (3) can do is to confirm it, or to satisfy intellectual curiosity.

But if the answer is less obvious, then (3) can sometimes come into its own. Here is another (again genuine) address:
Mr T.R.Davis
Department of English
University of Birmingham
My internal encyclopedia tells me that this message is quite abnormal, since it refuses to allow a connection between the institution I work in and the well-known car hire firm. But stage (2) springs less readily to mind. Why this mysterious acronym? Maybe it's important, and I should know about it? Stage (2) is not helpful, because the immediate linguistic and encyclopedic norms don't suggest a valid reason for the acronym to be there. Therefore I move to stage (3), and, after a reasonably short period of intense thought, the answer springs into mind: knowledge of the process of transmission points out to me that (a) this message, like the last one, may have been derived from dictation, (b) there are two possible spellings of my surname which dictators need to distinguish, and (c) typists often have better things to think about than typing yet another tedious address. Thus stage (3) gives us the answer to stage (2).

But, as readers will have noticed, this rather satisfactory solution to the problem is unusual. Mostly stage (3) is of no help whatsoever, and we are driven back to stages (1) and (2): to trying to fit, into the empty slot in the context vacated by what we have spotted as odd, for whatever reason, some new item that we feel might make more sense.

It is, I hope, quite clear that this process is in important respects identical with that of traditional textual criticism. Stages (1) and (2) are Art, and stage (3) is Science. Stage (1) contains in itself the process of collation, here done in memory against all of the other envelopes addressed to myself that I have seen. Stage (2) is the intuitive reference to whatever norms are appropriate and available (in the case of textual critics, these are usually not only the normal linguistic models, but also specifically literary or theological ones, or sometimes both). And stage (3), of course, is the domain of Bibliography, and Stemmatics, and Paleography, and all the rest of the repertoire that the textual critic uses when the problem can't easily be solved by simple commonsense.

All of these stages, in traditional textual criticism, have however been somewhat disguised, by a process that one can only call mystification. Stage (2) is, as we have seen, normally called 'literary judgement', with consequent confusion; stage (3) is often given the equally mystificatory title 'scientific', a term that has attracted an equivalent, but quite opposed, privileging process in this society. Here are the first two monsters that beset textual criticism; these, I hope, can fairly readily be identified as non-entities.

If we look at the two processes, 'literary judgement' and 'bibliographical analysis', in the light of the three-stage process identified above, we can see that they are in fact rather similar: they both attempt to resolve a perceived anomaly by reference to supposed norms. In the first case what we look at is a range of similar linguistic circumstances, and in the second, we summon up and relate a set constructed normal circumstances surrounding production and dissemination of documents or other texts. Stage (1) feels intuitive because its medium is language, whether literary, theological, or normal speech, and we are able to access and select huge amounts of normative linguistic material of whatever kind very quickly indeed; it is part of usual (and utterly remarkable) language use. But the process is so quick that in applying it to problems of the textual critical kind it feels like a kind of magic flash, wonderfully useful but somewhat occult and therefore unreliable; as indeed it is, because there is no way that we can subject the process to a conscious process of validation. Contrast this with the bibliographically based insights of even a giant like Greg, which are on the whole accessible, and can be checked and subjected to falsification processes. But I only have conscious access to a tiny fraction of all of the contexts that provide the meanings for the words that I myself use, let alone those of other people's meanings. It comes down to a matter of intuition, which is often undistinguishable from blind faith; and this, I suggest, is why textual critics traditionally have such violent arguments.

I can here offer an example, that shows this intuitive process, and its mystification, in a somewhat dramatic light. The passage is from A.E.Housman, [8] who, of all textual critics, managed to elevate rudeness to his colleagues, and indeed (in the following passage) to everyone else, into a high art.
The following stanza of Mr de la Mare's 'Fare well' first met my eyes, thus printed, in a newspaper review.
Oh, when this my dust surrenders
Hand, foot, lip, to dust again,
May these loved and loving faces
Please other men!
May the rustling harvest hedgerow
Still the Traveller's Joy entwine.
And as happy children gather
Posies once mine.
I knew in a moment that Mr de la Mare had not written rustling, and in another moment I had found the true word. But if the book of poems had perished and the verse survived only in the review, who would have believed me rather than the compositor? The bulk of the reading public would have been perfectly content with rustling, nay they would sincerely have preferred it to the epithet which the poet chose. If I had been so ill-advised as to publish my emendation, I should have been told that rustling was exquisitely apt and poetical, because hedgerows do rustle, especially in autumn, when the leaves are dry, and when straws and ears from the passing harvest-wain (to which 'harvest' is so plain an allusion that only a pedant like me could miss it) are hanging caught in the twigs; and I should have been recommended to quit my dusty (or musty) books and make a belated acquaintance with the sights and sounds of the English countryside. And the only possible answer would have been ugh!
Clearly Housman is making the point that he has magical powers; a position that he reinforces by the brilliant manoeuvre of not telling the reader what the answer is. Anticipating in his practice the theoretical work of Stanley Fish by half a century, he leaves the reader gasping at his/her own inability to solve the conundrum: surprised, as it were, by sin.

Or this reader, at any rate; to such an extent that (with profound feelings of inadequacy at ever having pretended to be a textual critic) I went over to the Library and looked up the answer in De La Mare's works. The answer to Housman's riddle is... not very obvious.

Even if we take this anecdote as being fictional, it is still instructive: what the fiction is concerned to instil is precisely mystification of the kind that I have described: as if 'literary judgement' were somehow different from the normal process of language use. At base, it is not. The fictional (or not) hero of Housman's narrative knew a lot of contexts; that's all.

We can now turn to the main target of my analysis: the author

Let us look into this model a little more closely. If someone makes a mistake in talking to you, you might well ask them what they meant. We want (following our simple communications model) to know what the sender intended to encode and transmit, but was prevented by whatever vagary in the system. Intended meaning, in the shape of a pure version of the message, uncontaminated by any corruption, to which the sender (and only the sender) has access and can refer without trouble, is a vital part, perhaps the most vital part, of the commonsense model of textual criticism. It is the truth, the essence, the original pure message, which (we think) absolutely needs to exist in order that we can decide that that which is not it, is false. All of this seems very true; unfortunately, as far as textual criticism is concerned, it is defective.

The problem is, at its root, that the model works reasonably well for speech, but not well for written language; and textual criticism is only done on written language. The model's defect is that it makes no distinction at all between written and spoken language, and since this distinction is lacking, there is confusion, of a quite complicated nature, because our image of written language is influenced by the spoken variety, and of the spoken, on the other hand, by the written. I will explain.

Firstly, most of the messages which we encounter, and which most humans have encountered since language came into existence, are spoken: straightforward, pitched at the comprehension of the hearer, face-to-face, and short. Since they are spoken, there is normally easy access to a speaker, who can inform us of his or her intended meaning. In these cases the model works well; so well, in fact, that we export it instinctively to situations for which it is not appropriate. It is one thing to apply an intention model when someone says 'pass the salt'; quite another if the text at hand is Middlemarch. Nonetheless, it seems that since, however literate we may be, speech comes first, it is taken as the norm of communication, with, in Derrida's now famous formulation, its 'metaphysic of presence'.

However the situation is more complicated than that. Speech may be primary, but we do live in a highly literate culture­p;where, for instance, spelling (irrationally) influences pronunciation, and not the other way round­p;and so our view of face-to-face spoken interaction is that it is something like the kind of conversations that people have in novels, where people do indeed seem to pass messages, in the form of coherent packages of words, to each other, as in our simple communications model. We could call this (following Popper) the bucket theory of communication: the contents of one bucket are poured, unproblematically, into another, as meaning is transferred. This depends, in its turn, on a bucket theory of mind: a receptacle in which things can be contained and preserved, steadily and intact.

But actual conversation, and actual minds, are not at all like that. For one thing, as well as the words spoken there is a great deal of use of gesture, intonation, facial expression, pausing, eye contact, and much much else. What we need this rich repertoire for is the negotiation of meaning. Whether the dialogue is between strangers or lovers, in and around the words spoken is a complex tissue of questions and and answers: am I being rude, respectful, too subservient? Am I invading your space or breaking your rules? Or, alternatively, am I winning? And so on. But the principal question, over, above, and within the variety of social codes, is: do you understand what I mean to say? Now, the important point is that this question is normally negotiated moment-by-moment. Intentions in speech, like messages in speech, are not as written discourses seem to be, somehow wholly intended and delivered en bloc; they are part of the normal to-and-fro of conversation. The mind is not a bucket: it operates in time, not spatially, so that it finds it very difficult (without years of training) to hold large structures at once; normally the mind is wayward and hard to control, with thoughts, intentions, and impulses fleeting in and out of existence, like dust floating in sunlight. Once said, the words and their intentions fade quickly in comparison to that moment, and the words to come are vague, provisional, and highly dependent on incoming information that has not yet arrived. To re-use Saussure's illuminating analogy, conversation is like chess: what tends to matter most is the current position, not how we got there.

Since intentions are so evanescent, I would suggest that only in this highly interactive situation, this plenitude, is the concept of intention at all meaningful. It is a truism of grammatology to say that in written discourse the speaker is absent, in time and in space: this, indeed, was one reason why writing was invented. The other reason was that writing captures and preserves in time the momentary existence of spoken words, which was found to be very useful for the invention of (for instance) novels, among many other items. In speech the speaker is present and the text is (outside the moment of speaking) absent; in writing the author is absent and the text is (seems) present, all of it: an object we hold in our hands. In our model of communication we (naturally) take the best of both worlds: the comforting presence of the author, the comforting solidity of the text, and suppose that we can somehow obtain from this author (who is, perhaps, not just absent, but dead) a coherent, sustained, and solid set of intentions and meanings for all of the words in the whole of this supposedly solid text.

It is of course possible to deconstruct both author and text even further. We can deconstruct the author, for instance, by following Derrida and Lacan, who question the presence and the existence (respectively) of an intending subject even in spoken interaction. So for a (perhaps predominant) group of current literary theorists the preferred model of spoken discourse would be altogether more vertiginous than the discourse-analysis model that I have offered: what is being negotiated, they would suggest, is not only meanings but the illusion of a meaning subject.

There is another useful way to refine our model of what goes on in the act of reading and writing: we can deconstruct the seeming solidity of the text, the presence of the book in our hand as we read. Take Middlemarch. What is it? Of course it is a book, so many pages, so many words. So solid is the presence of this printed object that it is very hard to think of Middlemarch without referring to some aspect of its writtenness, its physical form. But the meaning of Middlemarch only happens when someone is reading, writing, remembering, or anticipating it.

This meaning, therefore, is never anything but the mental construct of a reader or writer. In the course of reading the book exists only as the result of a negotiation, constantly changing as we read on, between selective memory and a pattern of anticipations, together with a small area on which the consciousness is currently focussed. Just like in conversation, meaning is negotiated, but the difference is that the words to come are fixed, and what we negotiate with is the text itself: the words written on the page. It is comforting and customary to feel that someone is present in the text and addressing us, but in fact there is no-one there: the words and the reader are all there is. Similarly it is a normal illusion to suppose that when we read or think or talk or write about an extensive­p;say, a literary­p;text we are in possession of the whole thing, as solid and present as the book we read it in; but this is not true either: like everything else we perceive or know, it is only made up of memory, anticipation, and the fleeting moment of awareness.

In the act of writing too the book exists only as a pattern of anticipation and memory; here, as in conversation, the words to come are negotiable, but again the negotiation is not with a hearer, but the text itself. It is as impossible for a writer as for a reader to hold the whole text constantly in mind, like a deity for whom his or her creation is eternally present; if we discuss Middlemarch then you and I are are comparing our selective and unique recollections of the text, which will be constructed on the basis of various theories that we have about it. And this would be true if we were to discuss it with George Eliot too: even for a short document I cannot possibly recall the complex patterns of impulse and desire, of negotiation and selective memory and anticipation that went through my head as I produced it, and (I assume) neither could she; precisely because we are constructed by a culture that enables me to create written documents, the art of memory and the training of concentration are not much taught, but even if they were, the task would be impossible.

Since this is so, if I am offered someone else's reading of a passage I have written I will probably respond to it as just another reader; in any case, I know well that what the words will mean, however I may have intended them, is what a significant proportion of readers will read them as meaning, and once they are written and promulgated I do not have much control over that. This is of course a truism of current Literary Theory; but it is also true that if I am asked my opinion as to the wording of a passage I have written I will respond principally as just another textual critic, [9] normalising to whatever I expect the passage to be.

For this is what we do, in fact, in our daily textual criticisms. If a passage looks odd for some reason, or if we see two or more texts that are assumed to be versions of the same, but which differ, what we do is, we make the situation make sense by normalising the offending irregularity against expectations derived from the text, and other texts, and from a large number of other sources, just as we do when we read, or for that matter when we perceive the world in any way whatsoever. In the case of reading and textual criticism (and for some people, of course, in perceiving the world) it is comforting to think that we are thereby getting in touch with the intentions of an author, an author who is in charge, in control, and has the whole text in his or her head. This is not so.

I think it may be helpful at this point to offer an piece of personal experience which I have found extremely helpful in thinking about this problem. Some years ago I found that my training in bibliography and paleography was quite applicable to the solution of problems in forensic document analysis, as we call it: the detection of forgery, in other words. I have now been operating as what lawyers call a handwriting expert for some time, and it is my primary research interest. Readers who notice a certain detachment from the rather exquisite agonies of textual bibliographers will now know why.

I found, as I say, my training very useful, and entirely appropriate; in some ways more so than that of my colleagues in this field, who all received their initial training in the 'hard' sciences of Physics and Chemistry. In one important respect, however, I was at a disadvantage. I found that I regarded the mysteries that are given to me every day as personal challenges, and any degree of insolubility in them as a personal affront. They were monsters, to be defeated, with my very own sword. This is not entirely bad: it gave a certain energy to my work, and to my defence of it in Court, but what went with it was a very marked temptation to use rhetoric to make the evidence give up more than it had to offer. This, as is well known, is a notorious characteristic of bibliographers and textual critics, and a fairly harmless part of life's rich comedy. Forensic science is however a more serious business even than cruces in Othello: it puts people in prison, and takes away their money, their livelihood, their reputation­p;or saves them from these fates. It is not to be trifled with. Such trifling had been beaten out of my colleagues the scientists at a very early age: if the evidence won't play ball, then you don't help it along. This is fundamental; it is the scientist's concept of sin.

The distinction crystallised for me in a conversation with a lawyer; we agreed that the essential difference between us was that his job was to make a case, and mine was not; the scientist's role is to let the evidence speak through their knowledge, skill, and theoretical framework; but it can say only what it has to say, not what they want from it. This may be hard for the defendant or (more normally) the prosecution, but that is the way the real, as opposed to the fictional, world happens to be constructed. Scientists run up all the time against the intransigence of nature, who will not easily conform to human expectations and desires; it is part of the job, taken for granted: expected, in fact. Textual critics, who deal in fiction all their working lives, are locked into the messy and rather corrupt obfuscations of ideology, which creates monsters that are not theirs. If the evidence will not deliver what the literary criticism needs, either because it is intransigent or absent, or because the needs are incompatible with the way the world is, then it is not for the textual critic to do anything other than point this out. They must offer a solution, since the job-description (dictated by literary critics) demands one, but they should make absolutely clear the degree of unlikelihood, or impossibility, that that solution represents. Modesty, and a sense of humour, are the answer; not reverent appeals to literary judgement, or a non-scientist's version of science; nor, most of all, to the reification of the text, the author, and his or her intentions. If we can do this, and can simply accept, like any scientist, the intransigence of nature, his or her stubborn refusal to open up secrets to even the most dedicated enquirer, then the monsters, I suggest, vanish away. Textual criticism is then a perfectly possible and satisfactory activity: after all, we do it every day.

What de la Mare seems to have written, incidentally, is 'rusting'.