Working on the Chain Gang: Interpretation in Law and Literature

Stanley Fish

In his essay Law as Interpretation Ronald Dworkin is concerned to characterize legal practice in such a way as to avoid claiming either that in deciding a case judges find the plain meaning of the law “just ‘there'” or, alternatively, that they make up the meaning “wholesale” in accordance with personal preference or whim. It is Dworkin’s thesis that neither of these accounts is adequate because interpretation is something different from both. Dworkin is right, I think, to link his argument about legal practice to an argument about the practice of literary criticism, not only because in both disciplines the central question is, “What is the source of interpretative authority?,” but also because in both disciplines answers to that question typically take the form of the two positions Dworkin rejects. Just as there are those in the legal community who have insisted on construing statutes and decisions “strictly” (that is, by attending only to the words themselves), so there are those in the literary community who have insisted that interpretation is, or should be, constrained by what is “in the text”; and just as the opposing doctrine of legal realism holds that judges’ “readings” are always rationalizations of their political or personal desires, so do proponents of critical subjectivity hold that what a reader sees is merely a reflection of his predispositions and biases. The field is divided, in short, between those who believe that interpretation is grounded in objectivity and those who believe that interpreters are, for all intents and purposes, free. Dworkin moves to outflank both of these positions by characterizing legal and critical practice as “chain enterprises,” enterprises in which interpretation is an extension of an institutional history made up of “innumerable decisions, structures, conventions and practices.” Interpretation so conceived is not purely objective since its results will not “wring assent from a stone” (there is still “room for disagreement”), but neither is it wholly subjective, since the interpreter does not proceed independently of what others in the institution have done or said.

In general, I Find this account of interpretation and its constraints attractive, in part because I find it similar in important ways to the account I have offered under the rubric of “interpretive communities” in Is There a Text in This Class? There are, however, crucial differences between the two accounts, and in the course of explicating those differences I will argue that Dworkin repeatedly falls away from his own best insights into a version of the fallacies (of pure objectivity and pure subjectivity) he so forcefully challenges.

We can begin by focusing on the most extended example in his essay of a “chain enterprise,” the imagined literary example of a novel written not by a single author but by a group of coauthors, each of whom is responsible for a separate chapter. The members of the group draw lots and the

lowest number writes the opening chapter of a novel, which he or she then sends to the next number who adds a chapter, with the understanding that he is adding a chapter to that novel rather than beginning a new one, and then sends the two chapters to the next number, and so on. Now every novelist but the First has the dual responsibilities of interpreting and creating, because each must read all that has gone before in order to establish, in the interpretivist sense, what the novel so far created is. He or she must decide what the characters are “really” like; what motives in fact guide them; what the point or theme of the developing novel is; how far some literary device or Figure, consciously or unconsciously used, contributes to these, and whether it should be extended or refined or trimmed or dropped in order to send the novel further in one direction rather than another.

In its deliberate exaggeration this formulation of a chain enterprise is helpful and illuminating, but it is also mistaken in several important respects. First of all, it assumes that the First person in the chain is in a position different in kind from those who follow him because he is only creating while his fellow authors must both create and interpret. In an earlier draft of the essay Dworkin had suggested that as the chain extends itself the freedom enjoyed by the initiator of the sequence is more and more constrained, until at some point the history against which “late novelists” must work may become so dense “as to admit only one good-faith interpretation”; and indeed that interpretation will not be an interpretation in the usual sense because it will have been demanded by what has already been written. Dworkin has now withdrawn this suggestion (which he had qualified with words like “probably”), but the claim underlying it – the claim that constraints thicken as the chain lengthens – remains as long as the distinction between the First author and all the others is maintained. The idea is that the First author is free because he is not obliged “to read all that has gone before” and therefore doesn’t have to decide what the characters are “really” like and what motives guide them, and so on. But, in fact, the first author has surrendered his freedom (although, as we shall see, surrender is exactly the wrong word) as soon as he commits himself to writing a novel, for he makes his decision under the same constraints that rule the decisions of his collaborators. He must decide, for example, how to begin the novel, but the decision is not “free” because the very notion “beginning a novel” exists only in the context of a set of practices that at once enable and limit the act of beginning. One cannot think of beginning a novel without thinking within, as opposed to thinking “of,” these established practices, and even if one “decides” to “ignore” them or “violate” them or “set them aside,” the actions of ignoring and violating and setting aside will themselves have a shape that is constrained by the preexisting shape of those practices. This does not mean that the decisions of the First author are wholly determined, but that the choices available to him are “novel-writing choices,” choices that depend on a prior understanding of what it means to write a novel, even when he “chooses” to alter that understanding. In short, he is neither free nor constrained (if those words are understood as referring to absolute states), but free and constrained. He is free to begin whatever kind of novel he decides to write, but he is constrained by the finite (although not unchanging) possibilities that are subsumed in the notions “kind of novel” and “beginning a novel.”

Moreover, those who follow him are free and constrained in exactly the same way. When a later novelist decides to “send the novel further in one direction rather than in another,” that decision must follow upon a decision as to what direction has already been taken; and that decision will be an interpretive one in the sense that it will not be determined by the independent and perspicuous shape of the words, but will be the means by which the words are given a shape. Later novelists do not read directly. from the words to a decision about the point or theme of the novel. but from a prior understanding (which may take a number of forms) of the points or themes novels can possibly have to a novelistic construction of the words. Just as the first novelist “creates” within the constraints of “novel practice” in general, so do his successors on the chain interpret him (and each other) within those same constraints. Not only are those constraints controlling, but they are uniformly so; they do not relax or tighten in relation to the position an author happens to occupy on the chain. The last author is as free, within those constraints, to determine what “the characters are really like” as is the first. It is tempting to think that the more information one has (the more history) the more directed will be one’s interpretation; but information only comes in an interpreted form (it does not announce itself) . No matterhow much or how little you have, it cannot be a check against interpretation because even when you first “see” it, interpretation has already done its work. So that rather than altering the conditions of interpretation, the accumulation of chapters merely extends the scope of its operation.

If this seems counterintuitive, imagine the very real possibility of two (or more) “later” novelists who have different views of the direction the novel has taken and are therefore in disagreement as to what would constitute a continuation of “that” novel as opposed to “beginning a new one.” To make the example more specific, let us further imagine that one says to another, “Don’t you see that it’s ironic, a social satire?,” and the second replies, “Not at all, at most it’s a comedy of manners,” while a third chimes in, “You’re both wrong; it’s obviously a perfectly straightforward piece of realism.” If Dworkin’s argument is to hold, that is, if the decisions he talks about are to be constrained in a strong sense by an already-in-place text, it must be possible to settle this disagreement by appealing to that text. But it is precisely because the text appears differently in the light of different assumptions as to what is its mode that there is a disagreement in the first place. Or, to put it another way, “social satire,” “comedy of manners,” and “piece of realism” are not labels applied mechanically to perspicuous instances; rather, they are names for ways of reading, ways which when put into operation make available for picking out the “facts” which those who are proceeding within them can then cite. It is entirely possible that the parties to our imagined dispute might find themselves pointing to the same “stretch of language” (no longer the same, since each would be characterizing it differently and claiming it as a “fact” in support of opposing interpretations. (The history of literary criticism abounds in such scenarios.) Each would then believe, and be able to provide reasons for his belief, that only he is continuing the novel in the direction it has taken so far and that the others are striking out in a new and unauthorized direction.

Again, this does not mean that a later novelist is free to decide anything he likes (or that there is no possibility of adjudicating a disagreement) , but that within the general parameters of novel-reading practice, he is as free as anyone else, which means that he is as constrained as anyone else. He is constrained in that he can only continue in ways that are recognizable novel ways (and the same must be said of the first novelist’s act of “beginning”), and he is free in that no amount of textual accumulation can make his choice of one of those ways inescapable. Although the parameters of novel practice mark the limits of what anyone who is thinking within them can think to do, within those limits they do not direct anyone to do this rather than that. (They are not a “higher” text.) Every decision a later novelist makes will rest on his assessment of the situation as it has developed; but that assessment will itself be an act of interpretation which will in turn rest on an interpreted understanding of the enterprise in general.

This, then, is my first criticism of Dworkin’s example: the distinction it is supposed to illustrate – the distinction between the first and later novelists – will not hold up because everyone in the enterprise is equally constrained. (By “equally” I mean equally with respect to the condition of freedom; I am making no claims about the number or identity of the constraints.) My second criticism is that in his effort to elaborate the distinction Dworkin embraces both of the positions he criticizes. He posits for the first novelist a freedom that is equivalent to the freedom assumed by those who believe that judges (and other interpreters) are bound only by their personal preferences and desires; and he thinks of later novelists as bound by a previous history in a way that would be possible only if the shape and significance of that history were self-evident. Rather than avoiding the Scylla of legal realism (“making it up wholesale”) and the Charybdis of strict constructionism (“finding the law just ‘there”), he commits himself to both. His reason for doing so becomes clear when he extends the example to an analysis of the law:

Deciding hard cases at law is rather like this strange literary exercise. The similarity is most evident when judges consider and decide “common-law” cases; that is, when no statute figures centrally in the legal issue, and the argument turns on which rules or principles of law “underlie” the related decisions of other judges in the past. Each judge is then like a novelist in the chain. He or she must read through what other judges in the past have written not simply to discover what these judges have said, or their state of mind when they said it, but to reach an opinion about what these other judges have collectively done, in the way that each of our novelists formed an opinion about the collective novel so far written. Any judge forced to decide any law suit will find, if he looks in the appropriate books, records of many arguably similar cases decided over decades or even centuries past by many other judges of different styles and judicial and political philosophies, in periods of different orthodoxies of procedure and judicial convention. Each judge must regard himself, in deciding the case before him, as a partner in a complex chain enterprise of which these innumerable decisions, structures, conventions and practices are the history; it is his job to continue that history into the future through what he does. He must interpret what has gone before because he has a responsibility to advance the enterprise in hand rather than strike out in some new direction of his own.

The emphasis on the word “must” alerts us to what is at stake for Dworkin in the notion of a chain enterprise. It is a way of explaining , how judges are kept from striking out in a new direction, much as later novelists are kept by the terms of their original agreement from beginning a new novel. Just as it is the duty of a later novelist to continue the work of his predecessors, so it is the duty of a judge to “advance the enterprise in hand.” Presumably, the judge who is tempted to strike out in “some new direction of his own” will be checked by his awareness of his responsibility to the corporate enterprise; he will then comport himself as a partner in the chain rather than as a free and independent agent.

The force of the account, in other words, depends on the possibility of judges comporting themselves in ways other than the “chain-enterprise” way. But is there in fact any such possibility? What would it mean for a judge to strike out in a new direction? Dworkin doesn’t tell us, but presumably it would mean deciding a case in such a way as to have no relationship to the history of previous decisions. It is hard to imagine what such a decision would be like since any decision, to be recognized as a decision by a judge, would have to be made in recognizably judicial terms. A judge who decided a case on the basis of whether or not the defendant had red hair would not be striking out in a new direction; he would simply not be acting as a judge, because he could give no reasons for his decision that would be seen asreasons by competent members of the legal community. (Even in so extreme a case it would not be accurate to describe the judge as striking out in a new direction; rather he would be continuing the direction of an enterprise – perhaps a bizarre one – other than the judicial.) And conversely, if in deciding a case a judge is able to give such reasons, then the direction he strikes out in will not be new because it will have been implicit in the enterprise as a direction one could conceive of and argue for. This does not mean that his decision will be above criticism, but that it will be criticized, if it is criticized, for having gone in one judicial direction rather than another, neither direction being “new” in a sense that would give substance to Dworkin’s fears.

fears are equally groundless with respect to the other alternative Dworkin imagines, the judge who looks at the chain of previous decisions and decides to see in it “whatever he thinks should have been there.” Here the danger is not so much arbitrary action (striking out in a new direction) as it is the willful imposition of a personal perspective on materials that have their own proper shape. “A judge’s duty,” Dworkin asserts, “is to interpret the legal history he finds and not to invent a better history.” Interpretation that is constrained by the history one finds will be responsible, whereas interpretation informed by the private preferences of the judge will be wayward and subjective. The opposition is one to which Dworkin repeatedly returns in a variety of forms, but in whatever form it is always vulnerable to the same objection: neither the self-declaring or “found” entity nor the dangerously free or “inventing” agent is a possible feature of the enterprise.

First of all, one doesn’t just find a history; rather one views a body of materials with the assumption that it is organized by judicial concerns. It is that assumption which gives a shape to the materials, a shape that can then be described as having been “found.” Moreover, not everyone will find the same shape because not everyone will be proceeding within the same notion of what constitutes a proper judicial concern, either in general or in particular cases. One sees this clearly in Dworkin’s own account of what is involved in legal decisionmaking. A judge, he explains, will look in the “appropriate books” for cases “arguably similar” to the one before him. Notice that the similarity is “arguable,” which means that it must be argued for; similarity is not something one finds, but something one must establish, and when one establishes it one establishes the configurations of the cited cases as well as of the case that is to be decided. Similarity, in short, is not a property of texts (similarities do not announce themselves), but a property conferred by a relational argument in which the statement A is like B is a characterization (one open to challenge) of both A and B. To see a present-day case as similar to a chain of earlier ones is to reconceive that chain by finding it in an applicability that has not always been apparent. Paradoxically, one can be faithful to legal history only by revising it, by redescribing it in such a way as to accommodate and render manageable the issues raised by the present. This is a function of the law’s conservatism, which will not allow a case to remain unrelated to the past, and so assures that the past, in the form of the history of decisions, will be continually rewritten. In fact, it is the duty of a judge to rewrite it (which is to say no more than that it is the duty of a judge to decide), and therefore there can be no simply “found” history in relation to which some other history could be said to be “invented.” All histories are invented in the weak sense that they are not simply, discovered,” but assembled under the pressure of some present urgency; no history is invented in the strong sense that the urgency that led to its assembly was unrelated to any generally acknowledged legal concern.

To put it another way, there could be no such strongly invented history because there could be no such strong inventor, no judge whose characterization of legal history displayed none of the terms, distinctions, and arguments that would identify it (for competent members) as a legal history. Of course, someone who stood apart from the enterprise, someone who was not performing as a judge, might offer such a history (a history, for example, in which the observed patterns were ethnic or geographical), but to accuse such a historian of striking out in a new direction or inventing a better history would be beside the point since whatever he did or didn’t do would have no legal (as opposed to sociological or political) significance. And, conversely, someone who was in fact standing within the enterprise, thinking in enterprise ways, could only put forward a history that was enterprise-specific, and that history could not be an invented one. It is true, of course, that jurists can and do accuse each other of inventing a history, but that is a charge you level at someone who has “found” a history different from yours. It should not be confused with the possibility (or the danger) of ” really” inventing one. The distinction between a “found” history and an “invented” one is finally nothing more than a distinction between a persuasive interpretation and one that has failed to convince. One man’s “found” history will be another man’s invented history, but neither will ever be, because it could not be, either purely found or purely invented.

As one reads Dworkin’s essay, the basic pattern of his mistakes becomes more and more obvious. He repeatedly makes two related and mutually reinforcing assumptions: he assumes that history in the form of a chain of decisions has, at some level, the status of a brute fact; and he assumes that wayward or arbitrary behavior in relation to that fact is an institutional possibility. Together these two assumptions give him his project, the project of explaining how a free and potentially irresponsible agent is held in check by the self-executing constraints of an independent text. Of course, by conceiving his project in this way – that is, by reifying the mind in its freedom and the text in its independence – he commits himself to the very alternatives he sets out to avoid, the alternatives of legal realism on the one hand and positivism on the other. As a result, these alternatives rule his argument, at once determining its form and emerging, again and again, as its content.

An example, early in the essay, involves the possibility of reading an Agatha Christie mystery as a philosophical novel. Such a reading, Dworkin asserts, would be an instance of “changing” the text rather than “explaining” it because the text as it is will not yield to it without obvious strain or distortion. “All but one or two sentences would be irrelevant to the supposed theme, and the organization, style and figures would be appropriate not to a philosophical novel but to an entirely different genre.” The assumption is that sentences, figures, and styles announce their own generic affiliation, and that a reader who would claim them for an inappropriate genre would be imposing his will on nature. It is exactly the same argument by which judges are supposedly constrained by the obvious properties of the history they are to continue, and it falls by the same analysis. First of all, generic identification, like continuity between cases, is not something one finds, but something one establishes, and one establishes it for a reason. Readers don’t just “decide” to recharacterize a text; there has to be some reason why it would occur to someone to treat a work identified as a member of one genre as a possible member of another. There must already be in place ways of thinking that will enable the recharacterization to become a project, and there must be conditions in the institution such that the prosecution of that project seems attractive and potentially rewarding. With respect to the project Dworkin deems impossible, those ways and conditions already exist. It has long been recognized that authors of the first rank – Poe, Dickens, Dostoyevski – have written novels of detection, and the fact that these novels have been treated seriously means that the work of less obviously canonical authors – Wilkie Collins, Conan Doyle, among others – are possible candidates for the same kind of attention. Once this happens, and it has already happened, any novel of detection can, at least provisionally, be considered as a “serious” work without a critic or his audience thinking that he is doing something bizarre or irresponsible; and in , recent years just such consideration has been given to the work of Hammet, Chandler (whom Dworkin mentions), Highsmith, Sayers, Simenon, Freeling, John D. MacDonald, and Ross Macdonald. In addition, the emergence of semiotand structural analysis has meant that it is no longer necessarily a criticism to say of something that it is “formulaic”. a term of description, which under a previous understanding , of literary value would have been invoked in a gesture of dismissal, can now be invoked as a preliminary to a study of “signifying systems.” The result has been the proliferation of serious (not to say somber) formalist readings of works like Fleming’s Goldfinger. Whatever one might think of this phenomenon, it is now a recognized and respectable part of the academic literary scene. At the same time the advocates of “popular culture” have been pressing their claim with a new insistence and a new rigor (prompted in part by the developments I have already mentioned), and a measure of their success is the number of courses in detective fiction now offered in colleges and universities at all levels.

Given these circumstances (and others that could be enumerated), it would be strange if a sociological or anthropological or philosophical interpretation of Agatha Christie had not been put forward (in fact, here we have an embarrassment of riches), and as a longtime reader of her novels it has occurred to me to put one forward myself. I have noticed that Christie’s villains are often presented as persons so quintessentially evil that they have no moral sense whatsoever and can only simulate moral behavior by miming, without understanding, the actions and attitudes of others. It is typical of these villains also to be chameleons, capable almost at will of changing their appearance, and one can see why: since they have no human attachments or concerns, they can clothe themselves in whatever attachment or concern suits their nefarious and often unmotivated ends. (The parallel with Shakespeare’s Iago and Milton’s Satan is obvious.) It would seem, then, that Christie has a theory of evil in relation to personal identity that accounts for (in the sense of generating a description of) many of the characteristics of her novels: their plots, the emphasis on disguise, the tolerance for human weakness even as it is being exposed, etc. Now, were I to extend this general hypothesis about Christie into a reading of one or more of her works, I would not be proceeding as Dworkin’s pronouncement suggests I would. I would not, that is, be changing the novel by riding roughshod over sentences bearing obvious and inescapable meanings; rather, I would be reading those sentences within the assumption that they were related to what I assumed to be Christie’s intention (if not this one, then some other), and as a result they would appear to me in an already related form. Sentences describing the weaknesses of characters other than the villain would be seen as pointing to the paradoxical strength of human fraility; sentences detailing the topography and geography of crucial scenes would be read as symbolic renderings of deeper issues, and so on. This interpretive action, or any other that could be imagined, would not be performed in violation of the facts of the text, but would be an effort to establish those facts. If in the course of that effort I were to dislodge another set of facts, they would be facts that had emerged within the assumption of another intention, and they would therefore be no less interpretive than the facts I was putting in their place. Of course, my efforts might very well fail in that no one else would be persuaded of my reading, but neither success nor failure would prove anything about what the text does or does not allow; it would only attest to the degree to which I had mastered or failed to master the rules of argument and evidence as they are understood (tacitly, to be sure) by members of the professional community.

The point is one that I have made before: it is neither the case that interpretation is constrained by what is obviously and unproblematically “there,” nor the case that interpreters, in the absence of such constraints, are free to read into a text whatever they like (once again Dworkin has put himself in a corner with these unhappy alternatives). Interpreters are constrained by their tacit awareness of what is possible and not possible to do, what is and is not a reasonable thing to say, what will and will not be heard as evidence, in a given enterprise; and it is within those same constraints that they see and bring others to see the shape of the documents to whose interpretation they are committed.

Dworkin’s failure to see this is an instance of a general failure to understand the nature of interpretation. The distinction between explaining a text and changing it can no more be maintained than the others of which it is a version (finding vs. inventing, continuing vs. striking out in a new direction, interpreting vs. creating). To explain a work is to point out something about it that had not been attributed to it before and therefore to change it by challenging other explanations that were once changes in their turn. Explaining and changing cannot be opposed activities (although they can be the names of claims and counterclaims) because they are the same activities. Dworkin opposes them because he thinks that interpretation is itself an activity in need of constraints, but what I have been trying to show is that interpretation is a structure of constraints, a structure which, because it is always and already in place, renders unavailable the independent or uninterpreted text and renders unimaginable the independent and freely interpreting reader. In searching for a way to protect against arbitrary readings (judicial and literary), Dworkin is searching for something he already has and could not possibly be without. He conducts his search by projecting as dangers and fears possibilities that could never be realized and by imagining as discrete concepts entities that are already filled with the concerns of the enterprise they supposedly threaten.

One of those entities is intention. Dworkin spends a great deal of time refuting the view that interpretation in law and in literature must here concern itself with the intentions of the author. He argues, first, that the intention of a novelist or legislator is “complex” and therefore difficult to know. Second, he argues that even if the intention were known, it would be only a piece of “psychological data,” and therefore would be irrelevant to the determination of a meaning that was not psychological but institutional. In short, he argues that to make intention the key to interpretation is to bypass the proper interpretive context – the history of practices and conventions – and substitute for it an interior motion of the mind. This argument would make perfect sense if intentions were, as Dworkin seems to believe them to be, private property and more or less equivalent with individual purpose or even whim. But it is hard to think of intentions formed in the course of judicial or literary activity as “one’s own,” since any intention one could have will have been stipulated in advance by the understanding of what activities are possible to someone working in the enterprise. One could no more come up with a unique intention with respect to the presentation of a character or the marshaling of legal evidence than one could come up with a new way of beginning a novel or continuing a chain of decisions. Simply to do something in the context of a chain enterprise is ipso facto to “have” an enterprise-specific intention, and to read something identified as part of a chain enterprise is ipso facto to be in the act of specifying that same intention. That is to say, the act of reading itself is at once the asking and answering of the question, “What is it that is meant by these words?,” a question asked not in a vacuum, but in the , context of an already-in-place understanding of the various things someone writing a novel or a decision (or anything else) might mean (i.e., intend).

In Dworkin’s analysis, on the other hand, reading is simply the construing of sense and neither depends nor should depend on the identification of intention. He cites as evidence the fact that authors themselves have been kto reinterpret their own works. This, Dworkin asserts, shows that “[a]n author is capable of detaching what he has written from his earlier intentions . . . of treating it as an object in itself.” But in fact this only shows that an author is capable of becoming his own reader and deciding that he meant something other by his words than he had previously thought. Such an author-reader is not ignoring intention, but recharacterizing it; he is not interpreting in a “non-intention-bound style,” but interpreting in a way that leads to a new understanding of his intention. Nor is there anything mysterious about this; it is no more than what we all do when sometime after having produced an utterance (it could be in less than a second) we ask ourselves, “What did I mean by that?” This will seem curious if intentions are thought of as unique psychological events, but if intentions are thought of as forms of possible conventional behavior that are to be conventionally “read,” then one can just as well reread his own intentions as he can reread the intentions of another.

The crucial point is that one cannot read or reread independently of intention, independently, that is, of the assumption that one is dealing with marks or sounds produced by an intentional being, a being situated in some enterprise in relation to which he has a purpose or a point of view. This is not an assumption that one adds to an already construed sense in order to stabilize it, but an assumption without which the construing of sense could not occur. One cannot understand an utterance without at the same time hearing or reading it as the utterance of someone with more or less specific concerns, interests, and desires, someone with an intention. So that when Dworkin talks, as he does, of the attempt to “discover” what a judge or a novelist intended, he treats as discrete operations that are inseparable. He thinks that interpretation is one thing and the assigning of intention is another, and he thinks that because he thinks that to discover intention is to plumb psychological depths unrelated to the meaning of chain-enterprise texts. In fact, to specify the meaning of a chain-enterprise text is exactly equivalent to specifying the intention of its author, an intention which is not private, but a form of conventional behavior made possible by the general structure of the enterprise. This, of course, does not mean that intention anchors interpretation in the sense that it stands outside and guides the process; intention like anything else is an interpretive fact; that is, it must be construed; it is just that it is impossible not to construe it and therefore impossible to oppose it either to the production or the determination of meaning.

The fact that Dworkin does so oppose it is of a piece with everything else in the essay and is one more instance of its basic pattern. Once again he has imagined a free-floating and individualistic threat to interpretation – in this case it is called “intention” – and once again he has moved to protect interpretation by locating its constraints in a free-standing and self-declaring object – in this case “the work itself,” detached from the antecedent designs of its author. And this means that once again he has committed himself in a single stroke to the extremes he set out to avoid, the objectivity of meanings that are “just there” and the subjectivity of meanings that have been “made up” by an unconstrained agent.

I cannot conclude without calling attention to what is perhaps the most curious feature of Dworkin’s essay, the extent to which it contains its own critique. Indeed, a reader sympathetic to Dworkin might well argue that he anticipates everything I have said in the preceding pages. He himself says that “the artist can create nothing without interpreting as he creates, since . . . he must have at least a tacit theory of why what he produces is art,” and he also points out that the facts of legal history do not announce themselves but will vary with the beliefs of particular judges concerning the general function of the law. In another place he admits that the constraint imposed by the words of a text “is not inevitable,” in part because any theory of identity (i.e., any theory of what is the same and what is different, of what constitutes a departure from the same) “will be controversial.” And after arguing that the “constraint of integrity” (the constraint imposed by a work’s coherence with itself) sets limits to interpretation, he acknowledges that there is much disagreement “, about what counts as integration”; he acknowledges in other words that the constraint is itself interpretive.

Even more curious than the fact of these reservations and qualifications is Dworkin’s failure to see how much they undercut his argument. Early in the essay he distinguishes between simple cases in which the words of a statute bear a transparent relationship to the actions they authorize or exclude (his sample statute is “No will shall be valid without three witnesses”), and more difficult cases in which reasonable and knowledgeable men disagree as to whether some action or proposed action is lawful. But immediately after making the distinction he undermines it by saying (in a parenthesis), “I am doubtful that the positivists’ analysis holds even in the simple case of the will; but that is a different matter I shall not argue here.” It is hard to see how this is a different matter, especially since so much in the essay hangs on the distinction. One doesn’t know what form the argument Dworkin decides not to make would take, but it might take the form of pointing out that even in a simple case the ease and immediacy with which one can apply the statute to the facts is the result of the same kind of interpretive work that is more obviously required in the difficult cases. In order for a case to appear readable independently of some interpretive strategy consciously employed, one must already be reading within the assumption of that strategy and employing, without being aware of them, its stipulated (and potentially controversial definitions, terms, modes of inference, etc. This, at any rate, would be the argument I would make, and in making it I would be denying the distinction between hard and easy cases, not as an empirical fact (as something one might experience), but as a fact that reflected a basic difference between cases that are self-settling and cases that can be settled only by referring them to the history of procedures, practices, and conventions. All cases are so referred (not after reading but in the act of reading), and they could not be anything but so referred and still be seen as cases. The point is an important one because Dworkin later says that his account of chain enterprises is offered as an explanation of how we decide “hard cases at law”; that is, his entire paper depends on a distinction that he himself suggests may not hold, and therefore, as we have seen, his entire paper depends on the “positivist analysis” he rejects in the parenthesis.

One can only speculate as to what Dworkin intends by these qualifications, but whether they appear in a parenthesis or in an aside or in the form of quotation marks around a key word, their effect is the same : to place him on both sides of the question at issue and to blur the supposedly hard lines of his argument. As a result, we are left with two ways of reading the essay, neither of which is comforting. If we take the subtext of reservation and disclaimer seriously, it so much weakens what he has to say that he seems finally not to have a position at all; and if we disregard the subtext and grant his thesis its strongest form, he will certainly have a position, but it will be, in every possible way, wrong.