Of the problems arising from the lack of a technical vocabulary of poetics, two demand special attention. The fact, already mentioned, that there is no word for a work of literary art is one that I find particularly baffling. One may invoke the authority of Aristotle for using "poem" in this sense, but usage declares that a poem is a composition in metre, and to speak of Tom Jones as a poem would be an abuse of ordinary language. One may discuss the question whether great works of prose deserve to be called poetry in some more extended sense, but the answer can only be a matter of taste in definitions. The attempt to introduce a value-judgement into a definition of poetry (e.g., "What, after all, do we mean by a poem that is, something worthy of the name of poem?") only adds to the confusion. So of course does the antique snobbery about the superiority of metre which has given "prosy" the meaning of tedious and "prosaic" the meaning of pedestrian. As often as I can, I use "poem" and its relatives by synecdoche, because they are short words; but where synecdoche would be confusing, the reader will have to put up with such cacophonous jargon as "hypothetical verbal structure" and the like.
The other matter concerns the use of the word "symbol," which in this essay means any unit of any literary structure that can be isolated for critical attention. A word, a phrase, or an image used with some kind of special reference (which is what a symbol is usually taken to mean) are all symbols when they are distinguishable elements in critical analysis. Even the letters a writer spells his words with form part of his symbolism in this sense: they would be isolated only in special cases, such as alliteration or dialect spellings, but we are still aware that they symbolize sounds. Criticism as a whole, in terms of this definition, would begin with, and largely consist of, the systematizing of literary symbolism. It follows that other words must be used to classify the different types of symbolism.
For there must be different types: the criticism of literature can hardly be a simple or one-level activity. The more familiar one is  with a great work of literature, the more one's understanding of it grows. Further, one has the feeling of growing in the understanding of the work itself, not in the number of things one can attach to it. The conclusion that a work of literary art contains a variety or sequence of meanings seems inescapable. It has seldom, however, been squarely faced in criticism since the Middle Ages, when a precise scheme of literal, allegorical, moral, and anagogic meanings was taken over from theology and applied to literature. Today there is more of a tendency to consider the problem of literary meaning as subsidiary to the problems of symbolic logic and semantics. In what follows I try to work as independently of the latter subjects as I can, on the ground that the obvious place to start looking for a theory of literary meaning is in literature.
The principle of manifold or "polysemous" meaning, as Dante calls it, is not a theory any more, still less an exploded superstition, but an established fact. The thing that has established it is the simultaneous development of several different schools of modern criticism, each making a distinctive choice of symbols in its analysis. The modern student of critical theory is faced with a body of rhetoricians who speak of texture and frontal assaults, with students of history who deal with traditions and sources, with critics using material from psychology and anthropology, with Aristotelians, Coleridgians, Thomists, Freudians, Jungians, Marxists, with students of myths, rituals, archetypes, metaphors, ambiguities, and significant forms. The student must either admit the principle of polysemous meaning, or choose one of these groups and then try to prove that all the others are less legitimate. The former is the way of scholarship, and leads to the advancement of learning; the latter is the way of pedantry, and gives us a wide choice of goals, the most conspicuous today being fantastical learning, or myth criticism, contentious learning, or historical criticism, and delicate learning, or "new" criticism.
Once we have admitted the principle of polysemous meaning, we can either stop with a purely relative and pluralistic position, or we can go on to consider the possibility that there is a finite number of valid critical methods, and that they can all be contained in a single theory. It does not follow that all meanings can be arranged, as the medieval four-level scheme implies, in a hierarchical sequence, in which the first steps are comparatively elementary and apprehension gets more subtle and rarefied as one goes on. The  term "level" is used here only for convenience, and should not be taken as indicating any belief on my part in a series of degrees of critical initiation. Again, there is a general reservation to be made about the conception of polysemous meaning: the meaning of a literary work forms a part of a larger whole. In the previous essay we saw that meaning or dianoia was one of three elements, the other two being mythos or narrative and ethos or characterization. It is better to think, therefore, not simply of a sequence of meanings, but of a sequence of contexts or relationships in which the whole work of literary art can be placed, each context having its characteristic mythos and ethos as well as its dianoia or meaning. I call these contexts or relationships "phases."
Literal and Descriptive Phases: Symbol as Motif and as Sign
Whenever we read anything, we find our attention moving in two directions at once. One direction is outward or centrifugal, in which we keep going outside our reading, from the individual words to the things they mean, or, in practice, to our memory of the conventional association between them. The other direction is inward or centripetal, in which we try to develop from the words a sense of the larger verbal pattern they make. In both cases we deal with symbols, but when we attach an external meaning to a word we have, in addition to the verbal symbol, the thing represented or symbolized by it. Actually we have a series of such representations: the verbal symbol "cat" is a group of black marks on a page representing a sequence of noises representing an image or memory representing a sense experience representing an animal that says meow. Symbols so understood may here be called signs, verbal units which, conventionally and arbitrarily, stand for and point to things outside the place where they occur. When we are trying to grasp the context of words, however, the word "cat" is an element in a larger body of meaning. It is not primarily a symbol "of" anything, for in this aspect it does not represent, but connects. We can hardly even say that it represents a part of the author's intention in putting it there, for the author's intention ceases to exist as a separate factor as soon as he has finished revising, Verbal elements understood inwardly or centripetaly, as parts of a verbal structure, are, as symbols, simply and literally verbal elements, or units of a verbal structure. (The word "literally" should be kept  in mind.) We may, borrowing a term from music, call such elements motifs.
These two modes of understanding take place simultaneously in all reading. It is impossible to read the word "cat" in a context without some representational flash of the animal so named; it is impossible to see the bare sign "cat" without wondering what context it belongs to. But verbal structures may be classified according to whether the final direction of meaning is outward or inward. In descriptive or assertive writing the final direction is outward. Here the verbal structure is intended to represent things external to it, and it is valued in terms of the accuracy with which it does represent them. Correspondence between phenomenon and verbal sign is truth; lack of it is falsehood; failure to connect is tautology, a purely verbal structure that cannot come out of itself.
In all literary verbal structures the final direction of meaning is inward. In literature the standards of outward meaning are secondary, for literary works do not pretend to describe or assert, and hence are not true, not false, and yet not tautological either, or at least not in the sense in which such a statement as "the good is better than the bad" is tautological. Literary meaning may best be described, perhaps, as hypothetical, and a hypothetical or assumed relation to the external world is part of what is usually meant by the word "imaginative." This word is to be distinguished from "imaginary," which usually refers to an assertive verbal structure that fails to make good its assertions. In literature, questions of fact or truth are subordinated to the primary literary aim of producing a structure of words for its own sake, and the sign-values of symbols are subordinated to their importance as a structure of interconnected motifs. Wherever we have an autonomous verbal structure of this kind, we have literature. Wherever this autonomous structure is lacking, we have language, words used instrumentally to help human consciousness do or understand something else. Literature is a specialized form of language, as language is of communication.
The reason for producing the literary structure is apparently that the inward meaning, the self-contained verbal pattern, is the field of the responses connected with pleasure, beauty, and interest. The contemplation of a detached pattern, whether of words or not, is clearly a major source of the sense of the beautiful, and of the pleasure that accompanies it. The fact that interest is most easily aroused by such a pattern is familiar to every handler of words, from  the poet to the after-dinner speaker who digresses from an assertive harangue to present the self-contained structure of verbal inter relationships known as a joke. It often happens that an originally descriptive piece of writing, such as the histories of Fuller and Gibbon, survives by virtue of its "style," or interesting verbal pattern, after its value as a representation of facts has faded.
The old precept that poetry is designed to delight and instruct sounds like an awkward hendiadys, as we do not usually feel that a poem does two different things to us, but we can understand it when we relate it to these two aspects of symbolism. In literature, what entertains is prior to what instructs, or, as we may say, the reality-principle is subordinate to the pleasure-principle. In assertive verbal structures the priority is reversed. Neither factor can, of course, ever be eliminated from any kind of writing.
One of the most familiar and important features of literature is the absence of a controlling aim of descriptive accuracy. We should, perhaps, like to feel that the writer of a historical drama knew what the historical facts of his theme were, and that he would not alter them without good reason. But that such good reasons may exist in literature is not denied by anyone. They seem to exist only there: the historian selects his facts, but to suggest that he had manipulated them to produce a more symmetrical structure would be grounds for libel. Some other types of verbal structures, such as theology and metaphysics, are declared by some to be centripetal in final meaning, and hence to be tautological ("purely verbal"). I have no opinion on this, except that in literary criticism theology and metaphysics must be treated as assertive, because they are outside literature, and everything that influences literature from without creates a centrifugal movement in it, whether it is directed toward the nature of absolute being or advice on the raising of hops. It is clear, too, that the proportion between the sense of being pleasantly entertained and the sense of being instructed, or awakened to reality, will vary in different forms of literature. The sense of reality is, for instance, far higher in tragedy than in comedy, as in comedy the logic of events normally gives way to the audience's desire for a happy ending.
The apparently unique privilege of ignoring facts has given the poet his traditional reputation as a licensed liar, and explains why so many words denoting literary structure, "fable," "fiction," "myth," and the like, have a secondary sense of imtratfa, like the  Norwegian word digter which is said to mean liar as well as poet. But, as Sir Philip Sidney remarked, "the poet never affirmeth," and therefore does not lie any more than he tells the truth. The poet, like the pure mathematician, depends, not on descriptive truth, but on conformity to his hypothetical postulates. The appearance of a ghost in Hamlet presents the hypothesis "let there be a ghost in Hamlet." It has nothing to do with whether ghosts exist or not, or whether Shakespeare or his audience thought they did. A reader who quarrels with postulates, who dislikes Hamlet because he does not believe that there are ghosts or that people speak in pentameters, clearly has no business in literature. He cannot distinguish fiction from fact, and belongs in the same category as the people who send cheques to radio stations for the relief of suffering heroines in soap operas. We may note here, as the point will be important later, that the accepted postulate, the contract agreed on by the reader before he can start reading, is the same thing as a convention.
The person who cannot be brought to understand literary convention is often said to be "literal-minded." But as "literal" surely ought to have some connection with letters, it seems curious to use the phrase "literal-minded" for imaginative illiterates. The reason for the anomaly is interesting, and important to our argument. Traditionally, the phrase "literal meaning" refers to descriptive meaning that is free from ambiguity. We usually say that the word cat "means literally" a cat when it is an adequate sign for a cat, when it stands in a simple representative relation to the animal that says meow. This sense of the term literal comes down from medieval times, and may be due to the theological origin of critical categories. In theology, the literal meaning of Scripture is usually the historical meaning, its accuracy as a record of facts or truths. Dante says, commenting on the verse in the Psalms, "When Israel went out of Egypt," "considering the letter only, the exodus of the Israelites to Palestine in the time of Moses is what is signified to us (significatur nobis)." The word "signified" shows that the literal meaning here is the simplest kind of descriptive or representational meaning, as it would still be to a Biblical "literalist".
But this conception of literal meaning as simple descriptive meaning will not do at all for literary criticism. An historical event cannot be literally anything but an historical event; a prose narrative describing it cannot be literally anything but a prose narrative.  The literal meaning of Dante's own Commedia is not historical, not at any rate a simple description of what "really happened" to Dante. And if a poem cannot be literally anything but a poem, then the literal basis of meaning in poetry can only be its letters, its inner structure of interlocking motifs. We are always wrong, in the context of criticism, when we say "this poem means literally" and then give a prose paraphrase of it. All paraphrases abstract a secondary or outward meaning. Understanding a poem literally means understanding the whole of it, as a poem, and as it stands. Such understanding begins in a complete surrender of the mind and senses to the impact of the work as a whole, and proceeds through the effort to unite the symbols toward a simultaneous perception of the unity of the structure, (This is a logical sequence of critical elements, the integrity, consonantia, and claritas of Stephen's argument in Joyce's Portrait. I have no idea what the psychological sequence is, or whether there is a sequence - I suppose there would not be in a Gestalt theory.) Literal understanding occupies the same place in criticism that observation, the direct exposure of the mind to nature, has in the scientific method. "Every poem must necessarily be a perfect unity," says Blake: this, as the wording implies, is not a statement of fact about all existing poems, but a statement of the hypothesis which every reader adopts in first trying to comprehend even the most chaotic poem ever written.
Some principle of recurrence seems to be fundamental to all works of art, and this recurrence is usually spoken of as rhythm when it moves along in time, and as pattern when it is spread out in space. Thus we speak of the rhythm of music and the pattern of painting. But a slight increase of sophistication will soon start us talking about the pattern of music and the rhythm of painting. The inference is that all arts possess both a temporal and a spatial aspect, whichever takes the lead when they are presented. The score of a symphony may be studied all at once, as a spread-out pattern: a painting may be studied as the track of an intricate dance of the eye. Works of literature also move in time like music and spread out in images like painting. The word narrative or mythos conveys the sense of movement caught by the ear, and the word meaning or dianoia conveys, or at least preserves, the sense of simultaneity caught by the eye. We listen to the poem as it moves from beginning to end, but as soon as the whole of it is in our minds at once we "see" what it means. More exactly, this  response is not simply to the whole of it, but to a whole in it: we have a vision of meaning or dianoia whenever any simultaneous apprehension is possible.
Now as a poem is literally a poem, it belongs, in its literal context, to the class of things called poems, which in their turn form part of the larger class known as works of art. The poem from this point of view presents a flow of sounds approximating music on one side, and an integrated pattern of imagery approximating the pictorial on the other. Literally, then, a poem's narrative is its rhythm or movement of words. If a dramatist writes a speech in prose, and then rewrites it in blank verse, he has made a strategic rhythmical change, and therefore a change in the literal narrative. Even if he alters "came a day" to "a day came" he has still made a tiny alteration of sequence, and so, literally, of his rhythm and narrative. Similarly, a poem's meaning is literally its pattern or integrity as a verbal structure. Its words cannot be separated and attached to sign-values: all possible sign-values of a word are absorbed into a complexity of verbal relationships.
The word's meaning is therefore, from the centripetal or inward-meaning point of view, variable or ambiguous, to use a term now familiar in criticism, a term which, significantly enough, is pejorative when applied to assertive writing. The word "wit" is said to be employed in Pope's Essay on Criticism in nine different senses. In assertive writing, such a semantic theme with variations could produce nothing but hopeless muddle. In poetry, it indicates the ranges of meanings and contexts that a word may have. The poet does not equate a word with a meaning; he establishes the functions or powers of words. But when we look at the symbols of a poem as verbal signs, the poem appears in a different context altogether, and so do its narrative and meaning. Descriptively, a poem is not primarily a work of art, but primarily a verbal structure or set of representative words, to be classed with other verbal structures like books on gardening. In this context narrative means the relation of the order of words to events resembling the events in "life" outside; meaning means the relation of its pattern to a body of assertive propositions, and the conception of symbolism involved is the one which literature has in common, not with the arts, but with other structures in words.
A considerable amount of abstraction enters at this stage. When we think of a poem's narrative as a description of events, we no  longer think of the narrative as literally embracing every word and letter. We think rather of a sequence of gross events, of the obvious and externally striking elements in the word-order. Similarly, we think of meaning as the kind of discursive meaning that a prose paraphrase of the poem might reproduce. Hence a parallel abstraction comes into the conception of symbolism. On the literal level, where the symbols are motifs, any unit whatever, down to the letters, may be relevant to our understanding, But only large and striking symbols are likely to be treated critically as signs: nouns and verbs, and phrases built up out of important words. Prepositions and conjunctions are almost pure connectives. A dictionary, which is primarily a table of conventional sign-values, can tell us nothing about such words unless we already understand them.
So literature in its descriptive context is a body of hypothetical verbal structures. The latter stand between the verbal structures that describe or arrange actual events, or histories, and those that describe or arrange actual ideas or represent physical objects, like the verbal structures of philosophy and science. The relation of the spatial to the conceptual world is one that we obviously cannot examine here; but from the point of view of literary criticism, descriptive writing and didactic writing, the representation of natural objects and of ideas, are simply two different branches of centrifugal meaning. We may use the word "plot" or "story" for the sequence of gross events, and the connection of story with history is indicated in its etymology. But it is more difficult to use "thought," or even "thought-content," for the representational aspect of pattern, or gross meaning, because "thought" also describes what we are here trying to distinguish it from. Such are the problems of a vocabulary of poetics.
The literal and the descriptive phases of symbolism are, of course, present in every work of literature. But we find (as we shall also find with the other phases) that each phase has a particularly close relationship to a certain kind of literature, and to a certain type of critical procedure as well. Literature deeply influenced by the descriptive aspect of symbolism is likely to tend toward the realistic in its narrative and toward the didactic or descriptive in its meaning. Its prevailing rhythm will be the prose of direct speech, and its main effort will be to give as clear and honest an impression of external reality as is possible with a hypothetical structure. In the documentary naturalism generally associated with such names as  Zola and Dreiser, literature goes about as far as a representation of life, to be judged by its accuracy of description rather than by its integrity as a structure of words, as it could go and still remain literature. Beyond this point, the hypothetical or fictional element in literature would begin to dissolve. The limits of literary expression of this type are, of course, very wide, and nearly all the great empire of realistic poetry, drama, and prose fiction lies well within them. But we notice that the great age of documentary naturalism, the nineteenth century, was also the age of Romantic poetry, which, by concentrating on the process of imaginative creation, indicated a feeling of tension between the hypothetical and the assertive elements in literature.
This tension finally snaps off in the movement generally called symbolisme, a term which we expand here to take in the whole tradition which develops with a broad consistency through Mallarme and Rimbaud to Valery in France, Rilke in Germany, and Pound and Eliot in England. In the theory of symbolisme we have the complement to extreme naturalism, an emphasis on the literal aspect of meaning, and a treatment of literature as centripetal verbal pattern, in which elements of direct or verifiable statement are subordinated to the integrity of that pattern. The conception of "pure" poetry, or evocative verbal structure injured by assertive meaning, was a minor by-product of the same movement. The great strength of symbolisme was that it succeeded in isolating the hypothetical germ of literature, however limited it may have been in its earlier stages by its tendency to equate this isolation with the entire creative process. All its characteristics are solidly based on its conception of poetry as concerned with the centripetal aspect of meaning. Thus the achieving of an acceptable theory of literal meaning in criticism rests on a relatively recent development in literature.
Symbolisme, as expressed for instance in Mallarmé, maintains that the representational answer to the question "what does this mean?" should not be pressed in reading poetry, for the poetic symbol means primarily itself in relation to the poem. The unity of a poem, then, is best apprehended as a unity of mood, a mood being a phase of emotion, and emotion being the ordinary word for the state of mind directed toward the experiencing of pleasure or the contemplating of beauty. And as moods are not long sustained, literature, for symbolisme, is essentially discontinuous,  longer poems being held together only by the use of the grammatical structures more appropriate to descriptive writing. Poetic images do not state or point to anything, but, by pointing to each other, they suggest or evoke the mood which informs the poem. That is, they express or articulate the mood. The emotion is not chaotic or inarticulate: it merely would have remained so if it had not turned into a poem, and when it does so, it is the poem, not something else still behind it. Nevertheless the words suggest and evoke are appropriate, because in symbolisme the word does not echo the thing but other words, and hence the immediate impact symbolisme makes on the reader is that of incantation, a harmony of sounds and the sense of a growing richness of meaning unlimited by denotation.
Some philosophers who assume that all meaning is descriptive meaning tell us that, as a poem does not describe things rationally, it must be a description of an emotion. According to this the literal core of poetry would be a cri de coeur, to use the elegant expression, the direct statement of a nervous organism confronted with some thing that seems to demand an emotional response, like a dog howling at the moon. L'Allegro and Il Penseroso would be respectively, according to this theory, elaborations of "I feel happy" and "I feel pensive." We have found, however, that the real core of poetry is a subtle and elusive verbal pattern that avoids, and does not lead to, such bald statements. We notice too that in the history of literature the riddle, the oracle, the spell, and the kenning are more primitive than a presentation of subjective feelings. The critics who tell us that the basis of poetic expression is irony, or a pattern of words that turns away from obvious (i.e., descriptive) meaning, are much closer to the facts of literary experience, at least on the literal level. The literary structure is ironic because "what it says" is always different in kind or degree from "what it means." In discursive writing what is said tends to approximate, ideally to become identified with, what is meant.
The criticism as well as the creation of literature reflects the distinction between literal and descriptive aspects of symbolism. The type of criticism associated with research and learned journals treats the poem as a verbal document, to be related as fully as possible to the history and the ideas that it reflects. The poem is most valuable to this kind of criticism when it is most explicit and descriptive, and when its core of imaginative hypothesis can be most easily  separated. (Note that I am speaking of a kind of criticism, not of a kind of critic.) What is now called "new criticism," on the other hand, is largely criticism based on the conception of a poem as literally a poem. It studies the symbolism of a poem as an ambiguous structure of interlocking motifs; it sees the poetic pattern of meaning as a self-contained "texture," and it thinks of the external relations of a poem as being with the other arts, to be approached only with the Horatian warning of favete linguis, and not with the historical or the didactic. The word texture, with its overtones of a complicated surface, is a most expressive one for this approach. These two aspects of criticism are often thought of as antithetical, as were, in the previous century, the corresponding groups of writers. They are of course complementary, not antithetical, but still the difference in emphasis between them is important to grasp before we go on to try to resolve the antithesis in a third phase of symbolism.
Formal Phase: Symbol as Image
We have now established a new sense of the term "literal meaning" for literary criticism, and have also assigned to literature, as one of its subordinate aspects of meaning, the ordinary descriptive meaning that works of literature share with all other structures of words. But it seems unsatisfactory to stop with this quizzical antithesis between delight and instruction, ironic withdrawal from reality and explicit connection with it. Surely, it will be said, we have overlooked the essential unity, in works of literature, expressed by the commonest of all critical terms, the word form. For the usual associations of "form" seem to combine these apparently contradictory aspects. On the one hand, form implies what we have called the literal meaning, or unity of structure; on the other, it implies such complementary terms as content and matter, expressive of what it shares with external nature. The poem is not natural in form, but it relates itself naturally to nature, and so, to quote Sidney again, "doth grow in effect a second nature."
Here we reach a more unified conception of narrative and meaning. Aristotle speaks of mimesis praxeos, an imitation of an action, and it appears that he identifies this mimesis praxeos with mythos. Aristotle's greatly abbreviated account here needs some reconstruction. Human action (praxis) is primarily imitated by histories, or verbal structures that describe specific and particular actions.  Mythos is a secondary imitation of an action, which means, not that it is at two removes from reality, but that it describes typical actions, being more philosophical than history. Human thought (theoria) is primarily imitated by discursive writing, which makes specific and particular predications. A dianoia is a secondary imitation of thought, a mimesis logou, concerned with typical thought, with the images, metaphors, diagrams, and verbal ambiguities out of which specific ideas develop. Poetry is thus more historical than philosophy, more involved in images and examples. For it is clear that all verbal structures with meaning are verbal imitations of that elusive psychological and physiological process known as thought, a process stumbling through emotional entanglements, sudden irrational convictions, involuntary gleams of insight, rationalized prejudices, and blocks of panic and inertia, finally to reach a completely incommunicable intuition. Anyone who imagines that philosophy is not a verbal imitation of this process, but the process itself, has clearly not done much thinking.
The form of a poem, that to which every detail relates, is the same whether it is examined as stationary or as moving through the work from beginning to end, just as a musical composition has the same form when we study the score as it has when we listen to the performance. The mythos is the dianoia in movement; the dianoia is the mythos in stasis. One reason why we tend to think of literary symbolism solely in terms of meaning is that we have ordinarily no word for the moving body of imagery in a work of literature. The word form has normally two complementary terms, matter and content, and it perhaps makes some distinction whether we think of form as a shaping principle or as a containing one. As shaping principle, it may be thought of as narrative, organizing temporally what Milton called, in an age of more exact terminology, the "matter" of his song. As containing principle it may be thought of as meaning, holding the poem together in a simultaneous structure.
The literary standards generally called "Classical" or "neo-Classical," which prevailed in Western Europe from the sixteenth to the eighteenth centuries, have the closest affinity with this formal phase. Order and clarity are particularly emphasized: order because of the sense of the importance of grasping a central form, and clarity because of the feeling that this form must not dissolve or withdraw into ambiguity, but must preserve a continuous relationship to the nature which is its own content. It is the attitude characteristic of "humanism" in the historical sense, an attitude marked on the one hand by a devotion to rhetoric and verbal craftsmanship, and on the other by a strong attachment to historical and ethical affairs.
Writers typical of the formal phase - Ben Jonson for instance - are sure that they are in contact with reality and that they follow nature, yet the effect they produce is quite different from the descriptive realism of the nineteenth century, the difference being largely in the conception of imitation involved. In formal imitation, or Aristotelian mimesis, the work of art does not reflect external events and ideas, but exists between the example and the precept. Events and ideas are now aspects of its content, not external fields of observation. Historical fictions are not designed to give insight into a period of history, but are exemplary; they illustrate action, and are ideal in the sense of manifesting the universal form of human action. (The vagaries of language make "exemplary" the adjective for both example and precept.) Shakespeare and Jonson were keenly interested in history, yet their plays seem timeless; Jane Austen did not write historical fiction, yet, because she represents a later and more externalized method of following nature, the picture she gives of Regency society has a specific historical value.
A poem, according to Hamlet, who, though speaking of acting, is following a conventional Renaissance line of poetics, holds the mirror up to nature. We should be careful to notice what this implies: the poem is not itself a mirror. It does not merely reproduce a shadow of nature; it causes nature to be reflected in its containing form. When the formal critic comes to deal with symbols, therefore, the units he isolates are those which show an analogy of proportion between the poem and the nature which it imitates. The symbol in this aspect may best be called the image, We are accustomed to associate the term "nature" primarily with the external physical world, and hence we tend to think of an image as primarily a replica of a natural object. But of course both words are far more inclusive: nature takes in the conceptual or intelligible order as well as the spatial one, and what is usually called an "idea" may be a poetic image also.
One could hardly find a more elementary critical principle than the fact that the events of a literary fiction are not real but hypothetical events. For some reason it has never been consistently  understood that the ideas of literature are not real propositions, but verbal formulas which imitate real propositions. The Essay on Man does not expound a system of metaphysical optimism founded on the chain of being: it uses such a system as a model on which to construct a series of hypothetical statements which are more or less useless as propositions, but inexhaustibly rich and suggestive when read in their proper context as epigrams. As epigrams, as solid, resonant, centripetal verbal structures, they may apply pointedly to millions of human situations which have nothing to do with metaphysical optimism. Wordsworth's pantheism, Dante's Thomism, Lucretius' Epicureanism, all have to be read in the same way, as do Gibbon or Macaulay or Hume when they are read for style instead of subject-matter.
Formal criticism begins with an examination of the imagery of a poem, with a view to bringing out its distinctive pattern. The recurring or most frequently repeated images form the tonality, so to speak, and the modulating, episodic and isolated images relate themselves to this in a hierarchic structure which is the critical analogy to the proportions of the poem itself. Every poem has its peculiar spectroscopic band of imagery, caused by the requirements of its genre, the predilections of its author, and countless other factors. In Macbeth, for instance, the images of blood and of sleeplessness have a thematic importance, as is very natural for a tragedy of murder and remorse. Hence in the line "Making the green one red," the colors are of different thematic intensities. Green is used incidentally and for contrast; red, being closer to the key of the play as a whole, is more like the repetition of a tonic chord in music. The opposite would be true of the contrast between red and green in Marvell's The Garden.
The form of the poem is the same whether it is studied as narrative or as meaning, hence the structure of imagery in Macbeth may be studied as a pattern derived from the text, or as a rhythm of repetition falling on an audience's ear. There is a vague notion that the latter method produces a simpler result, and may therefore be used as a commonsense corrective to the niggling subtleties of textual study. The analogy of music again may be helpful. The average audience at a symphony knows very little about sonata form, and misses practically all the subtleties detected by an analysis of the score; yet those subtleties are really there, and as the audience can hear everything that is being played, it gets them  all as part of a linear experience; the awareness is less conscious, but not less real. The same is true of the response to the imagery of a highly concentrated poetic drama.
The analysis of recurrent imagery is, of course, one of the chief techniques of rhetorical or "new" criticism as well: the difference is that formal criticism, after attaching the imagery to the central form of the poem, renders an aspect of the form into the propositions of discursive writing. Formal criticism, in other words, is commentary, and commentary is the process of translating into explicit or discursive language what is implicit in the poem. Good commentary naturally does not read ideas into the poem; it reads and translates what is there, and the evidence that it is there is offered by the study of the structure of imagery with which it begins. The sense of tact, of the desirability of not pushing a point of interpretation "too far," is derived from the fact that the proportioning of emphasis in criticism should normally bear a rough analogy to the proportioning of emphasis in the poem.
The failure to make, in practice, the most elementary of all distinctions in literature, the distinction between fiction and fact, hypothesis and assertion, imaginative and discursive writing, produces what in criticism has been called the "intentional fallacy," the notion that the poet has a primary intention of conveying meaning to a reader, and that the first duty of a critic is to recapture that intention. The word intention is analogical: it implies a relation between two things, usually a conception and an act. Some related terms show this duality even more clearly: to "aim at" something means that a target and a missile are being brought into alignment. Hence such terms properly belong only to discursive writing, where the correspondence of a verbal pattern with what it describes is of primary importance. But a poet's primary concern is to produce a work of art, and hence his intention can only be expressed by some kind of tautology.
In other words, a poet's intention is centripetally directed. It is directed towards putting words together, not towards aligning words with meanings. If we had the privilege of Gulliver in Glubbdubdrib to call up the ghost of, say, Shakespeare, to ask him what he meant by such and such a passage, we could only get, with maddening iteration, the same answer: "I meant it to form part of the play." One may pursue the centripetal intention as far as genre, as a poet intends to produce, not simply a poem, but a  certain kind of poem. In reading, for instance, Zuleika Dobson as a description of life in Oxford, we should be well advised to allow for ironic intention. One has to assume, as an essential heuristic axiom, that the work as produced constitutes the definitive record of the writer's intention. For many of the flaws which an inexperienced critic thinks he detects, the answer "But it's supposed to be that way" is sufficient. All other statements of intention, however fully documented, are suspect. The poet may change his mind or mood; he may have intended one thing and done another, and then rationalized what he did. (A cartoon in a New Yorker of some years back hit off this last problem beautifully: it depicted a sculptor gazing at a statue he had just made and remarking to a friend: "Yes, the head is too large. When I put it in exhibition I shall call it 'The Woman with the Large Head.'") If intention is still thought to be apparent in the poem itself, the poem is being regarded as incomplete, like a freshman's essay where the reader has continually to speculate about what the author may have had in his mind. If the author has been dead for centuries, such speculation cannot get us very far, however irresistibly it may suggest itself.
What the poet meant to say, then, is, literally, the poem itself; what he meant to say in any given passage is, in its literal meaning, part of the poem. But literal meaning, we have seen, is variable and ambiguous. The reader may be dissatisfied with the ghost of Shakespeare's answer: he may feel that Shakespeare, unlike, say, Mallarme, is a poet he can trust, and that he also meant his passage to be intelligible in itself (i.e., have descriptive or rephrasable meaning). Doubtless he did, but the relationship of the passage to the rest of the play creates myriads of new meanings for it. Just as a vivid sketch of a cat by a good draughtsman may contain in a few crisp lines the entire feline experience of everyone who looks at it, so the powerfully constructed pattern of words that we know as Hamlet may contain an amount of meaning which the vast and constantly growing library of criticism on the play cannot begin to exhaust. Commentary, which translates the implicit into the explicit, can only isolate the aspect of meaning, large or small, which is appropriate or interesting for certain readers to grasp at a certain time. Such translation is an activity with which the poet has very little to do. The relation in bulk between commentary and a sacred book, such as the Bible or the Vedic hymns, is even more  striking, and indicates that when a poetic structure attains a certain degree of concentration or social recognition, the amount of commentary it will carry is infinite. This fact is in itself no more in credible than the fact that a scientist can state a law illustrated by more phenomena than he could ever observe or count, and there is no occasion for wondering, like the yokels in Goldsmith, how one small poet's head can carry the amount of wit, wisdom, instruction, and significance that Shakespeare and Dante have given the world.
Still there is a genuine mystery in art, and a real place for wonder. In Sartor Resartus Carlyle distinguishes extrinsic symbols, like the cross or the national flag, which are without value in themselves but are signs or indicators of something existential, from intrinsic symbols, which include works of art. On this basis we may distinguish two kinds of mystery. (A third kind, the mystery which is a puzzle, a problem to be solved and annihilated, belongs to discursive thought, and has little to do with the arts, except in matters of technique.) The mystery of the unknown or unknowable essence is an extrinsic mystery, which involves art only when art is also made illustrative of something else, as religious art is to the person concerned primarily with worship. But the intrinsic mystery is that which remains a mystery in itself no matter how fully known it is, and hence is not a mystery separated from what is known. The mystery in the greatness of King Lear or Macbeth comes not from concealment but from revelation, not from something unknown or unknowable in the work, but from something unlimited in it.
It could be said, of course, that poetry is the product, not only of a deliberate and voluntary act of consciousness, like discursive writing, but of processes which are subconscious or preconscious or half-conscious or unconscious as well, whatever psychological metaphor one prefers. It takes a great deal of will power to write poetry, but part of that will power must be employed in trying to relax the will, so making a large part of one's writing involuntary. This is no doubt true, and it is also true that poetic technique, like all technique, is a habitual, and therefore an increasingly unconscious, skill But I feel that literary data are in the long run only explicable within criticism, and I am reluctant to explain literary facts by psychological cliches. Still, it seems now almost impossible to avoid the term "creative," with all the biological analogies it suggests, when speaking of the arts. And creation, whether of God,  man, or nature, seems to be an activity whose only intention is to abolish intention, to eliminate final dependence on or relation to something else, to destroy the shadow that falls between itself and its conception.
One wishes that literary criticism had a Samuel Butler to formulate some of the paradoxes involved in this parallel between the work of art and the organism. We can describe objectively what happens when a tulip blooms in spring and a chrysanthemum in autumn, but we cannot describe it from the inside of the plant, except by metaphors derived from human consciousness and ascribed to some agent like God or nature or environment or élan vital, or to the plant itself. It is projected metaphor to say that a flower "knows" when it is time for it to bloom, and of course to say that "nature knows" is merely to import a faded mother-goddess cult into biology. I can well understand that in their own field biologists would find such teleological metaphors both unnecessary and confusing, a fallacy of misplaced concreteness. The same would be true of criticism to the extent that criticism has to deal with imponderables other than consciousness or logically directed will. If one critic says that another has discovered a mass of subtleties in a poet of which that poet was probably quite unconscious, the phrase points up the biological analogy. A snowflake is probably quite unconscious of forming a crystal, but what it does may be worth study even if we are willing to leave its inner mental processes alone.
It is not often realized that all commentary is allegorical interpretation, an attaching of ideas to the structure of poetic imagery. The instant that any critic permits himself to make a genuine comment about a poem (e.g., "In Hamlet Shakespeare appears to be portraying the tragedy of irresolution") he has begun to allegorize. Commentary thus looks at literature as, in its formal phase, a potential allegory of events and ideas. The relation of such commentary to poetry itself is the source of the contrast which was developed by several critics of the Romantic period between "symbolism" and "allegory," symbolism here being used in the sense of thematically significant imagery. The contrast is between a "concrete" approach to symbols which begins with images of actual things and works outward to ideas and propositions, and an "abstract" approach which begins with the idea and then tries to find a concrete image to represent it. This distinction is valid enough  in itself, but it has deposited a large terminal moraine of confusion in modern criticism, largely because the term allegory is very loosely employed for a great variety of literary phenomena.
We have actual allegory when a poet explicitly indicates the relationship of his images to examples and precepts, and so tries to indicate how a commentary on him should proceed. A writer is being allegorical whenever it is clear that he is saying "by this I also (allos) mean that." If this seems to be done continuously, we may say, cautiously, that what he is writing "is" an allegory. In The Faerie Queene, for instance, the narrative systematically refers to historical examples and the meaning to moral precepts, besides doing their own work in the poem. Allegory, then, is a contrapuntal technique, like canonical imitation in music. Dante, Spenser, Tasso, and Bunyan use it throughout: their works are the masses and oratorios of literature. Ariosto, Goethe, Ibsen, Hawthorne write in a freistimmige style in which allegory may be picked up and dropped again at pleasure. But even continuous allegory is still a structure of images, not of disguised ideas, and commentary has to proceed with it exactly as it does with all other literature, trying to see what precepts and examples are suggested by the imagery as a whole.
The commenting critic is often prejudiced against allegory without knowing the real reason, which is that continuous allegory prescribes the direction of his commentary, and so restricts its freedom. Hence he often urges us to read Spenser and Bunyan, for example, for the story alone and let the allegory go, meaning by that that he regards his own type of commentary as more interesting. Or else he will frame a definition of allegory that will exclude the poems he likes. Such a critic is often apt to treat all allegory as though it were naive allegory, or the translation of ideas into images.
Naive allegory is a disguised form of discursive writing, and belongs chiefly to educational literature on an elementary level: schoolroom moralities, devotional exempla, local pageants, and the like. Its basis is the habitual or customary ideas fostered by education and ritual, and its normal form is that of transient spectacle. Under the excitement of a particular occasion familiar ideas suddenly become sense experiences, and vanish with the occasion. The defeat of Sedition and Discord by Sound Government and Encouragement of Trade would be the right sort of  theme for a pageant designed only to entertain a visiting monarch for half an hour. The apparatus of "mass media" and "audiovisual aids" plays a similar allegorical role in contemporary education. Because of this basis in spectacle, naive allegory has its center of gravity in the pictorial arts, and is most successful as art when recognized to be a form of occasional wit, as it is in the political cartoon. The more solemn and permanent naive allegories of official murals and statuary show a marked tendency to date.
At one extreme of commentary, then, there is the naive allegory so anxious to make its own allegorical points that it has no real literary or hypothetical center. When I say that naive allegory "dates," I mean that any allegory which resists a primary analysis of imagery that is, an allegory which is simply discursive writing with an illustrative image or two stuck into it will have to be treated less as literature than as a document in the history of ideas. When the author of II Esdras, for instance, introduces an allegorical vision of an eagle, and then says, "Behold, on the right side there arose one feather, which reigned over all the earth," it is clear that he is not sufficiently interested in his eagle as a poetic image to remain within the normal boundaries of literary expression. The basis of poetic expression is the metaphor, and the basis of naive allegory is the mixed metaphor.
Within the boundaries of literature we find a kind of sliding scale, ranging from the most explicitly allegorical, consistent with being literature at all, at one extreme, to the most elusive, anti-explicit and anti-allegorical at the other. First we meet the continuous allegories, like The Pilgrim's Progress and The Faerie Queene, and then the free-style allegories just mentioned. Next come the poetic structures with a large and insistent doctrinal interest, in which the internal fictions are exempla, like the epics of Milton. Then we have, in the exact center, works in which the structure of imagery, however suggestive, has an implicit relation only to events and ideas, and which includes the bulk of Shakespeare. Below this, poetic imagery begins to recede from example and precept and become increasingly ironic and paradoxical. Here the modern critic begins to feel more at home, the reason being that this type is more consistent with the modern literal view of art, the sense of the poem as withdrawn from explicit statement.
Several types of this ironic and anti-allegorical imagery are familiar. One is the typical symbol of the metaphysical school of the  Baroque period, the "conceit" or deliberately strained union of normally disparate things. The paradoxical techniques of meta physical poetry are based on a sense of the breakdown of the internal relation of art and nature into an external one. Another is the substitute-image of symbolisme, part of a technique for suggesting or evoking things and avoiding the explicit naming of them. Still another is the kind of image described by Mr. Eliot as an objective correlative, the image that sets up an inward focus of emotion in poetry and at the same time substitutes itself for an idea. Still another, closely related to if not identical with the objective correlative, is the heraldic symbol, the central emblematic image which comes most readily to mind when we think of the word "symbol" in modern literature. We think, for example, of Hawthorne's scarlet letter, Melville's white whale, James's golden bowl, or Virginia Woolf s lighthouse. Such an image differs from the image of the formal allegory in that there is no continuous relationship between art and nature. In contrast to the allegorical symbols of Spenser, for instance, the heraldic emblematic image is in a paradoxical and ironic relation to both narrative and meaning. As a unit of meaning, it arrests the narrative; as a unit of narrative, it perplexes the meaning. It combines the qualities of Carlyle's intrinsic symbol with significance in itself, and the extrinsic symbol which points quizzically to something else. It is a technique of symbolism which is based on a strong sense of a lurking antagonism between the literal and the descriptive aspects of symbols, the same antagonism that made Mallarmé and Zola so extreme a contrast in nineteenth-century literature.
Below this we run into still more indirect techniques, such as private association, symbolism intended not to be fully understood, the deliberate spoofing of Dadaism, and kindred signs of another approaching boundary of literary expression. We should try to keep this whole range of possible commentary clearly in mind, so as to correct the perspective both of the medieval and Renaissance critics who assumed that all major poetry should be treated as far as possible as continuous allegory, and of the modern ones who maintain that poetry is essentially anti-allegorical and paradoxical.
What we have now is a conception of literature as a body of hypothetical creations which is not necessarily involved in the worlds of truth and fact, nor necessarily withdrawn from them,  but which may enter into any kind of relationship to them, ranging from the most to the least explicit. We are strongly reminded of the relationship of mathematics to the natural sciences. Mathematics, like literature, proceeds hypothetically and by internal consistency, not descriptively and by outward fidelity to nature. When it is applied to external facts, it is not its truth but its applicability that is being verified. As I seem to have fastened on the cat for my semantic emblem in this essay, I note that this point comes out sharply in the discussion between Yeats and Sturge Moore over the problem of Ruskin's cat, the animal that was picked up and flung out of a window by Ruskin although it was not there. Anyone measuring his mind against an external reality has to fall back on an axiom of faith. The distinction between an empirical fact and an illusion is not a rational distinction, and cannot be logically proved. It is "proved" only by the practical and emotional necessity of assuming the distinction. For the poet, qua poet, this necessity does not exist, and there is no poetic reason why he should either assert or deny the existence of any cat, real or Ruskinian.
The conception of art as having a relation to reality which is neither direct nor negative, but potential, finally resolves the dichotomy between delight and instruction, the style and the message. "Delight" is not readily distinguishable from pleasure, and hence opens the way to that aesthetic hedonism we glanced at in the introduction, the failure to distinguish personal and impersonal aspects of valuation. The traditional theory of catharsis implies that the emotional response to art is not the raising of an actual emotion, but the raising and casting out of actual emotion on a wave of something else. We may call this something else, perhaps, exhilaration or exuberance: the vision of something liberated from experience, the response kindled in the reader by the transmutation of experience into mimesis, of life into art, of routine into play. At the center of liberal education something surely ought to get liberated. The metaphor of creation suggests the parallel image of birth, the emergence of a new-born organism into independent life. The ecstasy of creation and its response produce, on one level of creative effort, the hen's cackle; on another, the quality that the Italian critics called sprezzatura and that Hoby's translation of Castiglione calls "recklessness," the sense of buoyancy or release  that accompanies perfect discipline, when we can no longer know the dancer from the dance.
It is impossible to understand the effect of what Milton called "gorgeous Tragedy" as producing a real emotion of gloom or sorrow. Aeschylus's The Persians and Shakespeare's Macbeth are certainly tragedies, but they are associated respectively with the victory of Salamis and the accession of James I, both occasions of national rejoicing. Some critics carry the theory of real emotion over into Shakespeare himself, and talk about a "tragic period," in which he is supposed to have felt dismal from 1600 to 1608. Most people, if they had just finished writing a play as good as King Lear, would be in a mood of exhilaration, and while we have no right to ascribe this mood to Shakespeare, it is surely the right way to describe our response to the play. On the other hand, it comes as something of a shock to realize that the blinding of Gloucester is primarily entertainment, the more so as the pleasure we get from it obviously has nothing to do with sadism. If any literary work is emotionally "depressing" there is something wrong with either the writing or the reader's response. Art seems to produce a kind of buoyancy which, though often called pleasure, as it is for instance by Wordsworth, is something more inclusive than pleasure. "Exuberance is beauty," said Blake. That seems to me a practically definitive solution, not only of the minor question of what beauty is, but of the far more important problem of what the conceptions of catharsis and ecstasis really mean.
Such exuberance is, of course, as much intellectual as it is emotional: Blake himself was willing to define poetry as "allegory addressed to the intellectual powers." We live in a world of threefold external compulsion: of compulsion on action, or law; of compulsion on thinking, or fact; of compulsion on feeling, which is the characteristic of all pleasure whether it is produced by the Paradiso or by an ice cream soda. But in the world of imagination a fourth power, which contains morality, beauty, and truth but is never subordinated to them, rises free of all their compulsions. The work of imagination presents us with a vision, not of the personal greatness of the poet, but of something impersonal and far greater: the vision of a decisive act of spiritual freedom, the vision of the recreation of man.
Mythical Phase: Symbol as Archetype
In the formal phase the poem belongs neither to the class "art," nor to the class "verbal": it represents its own class. There are thus two aspects to its form. In the first place, it is unique, a techne or artifact, with its own peculiar structure of imagery, to be examined by itself without immediate reference to other things like it. The critic here begins with poems, not with a prior conception or definition of poetry. In the second place, the poem is one of a class of similar forms. Aristotle knows that Oedipus Tyrannus is in one sense not like any other tragedy, but he also knows that it belongs to the class called tragedy. We, who have experienced Shakespeare and Racine, can add the corollary that tragedy is something bigger than a phase of Greek drama. We may also find tragedy in literary works which are not dramas. To understand what tragedy is, therefore, takes us beyond the merely historical into the question of what an aspect of literature as a whole is. With this idea of the external relations of a poem with other poems, two considerations in criticism for the first time become important: convention and genre.
The study of genres is based on analogies in form. It is characteristic of documentary and historical criticism that it cannot deal with such analogies. It can trace influence with great plausibility, whether it exists or not, but confronted with a tragedy of Shakespeare and a tragedy of Sophocles, to be compared solely because they are both tragedies, the historical critic has to confine himself to general reflections about the seriousness of life. Similarly, nothing is more striking in rhetorical criticism than the absence of any consideration of genre: the rhetorical critic analyzes what is in front of him without much regard to whether it is a play, a lyric, or a novel. He may in fact even assert that there are no genres in literature. That is because he is concerned with his structure simply as a work of art, not as an artifact with a possible function. But there are many analogies in literature apart altogether from sources and influences (many of which, of course, are not analogous at all) and noticing such analogies forms a large part of our actual experience of literature, whatever its role so far in criticism.
The central principle of the formal phase, that a poem is an imitation of nature, is, though a perfectly sound one, still a principle which isolates the individual poem. And it is clear that any poem may be examined, not only as an imitation of nature, but as an  imitation of other poems. Virgil discovered, according to Pope, that following nature was ultimately the same thing as following Homer. Once we think of a poem in relation to other poems, as a unit of poetry, we can see that the study of genres has to be founded on the study of convention. The criticism which can deal with such matters will have to be based on that aspect of symbolism which relates poems to one another, and it will choose, as its main field of operations, the symbols that link poems together. Its ultimate object is to consider, not simply a poem as an imitation of nature, but the order of nature as a whole as imitated by a corresponding order of words.
All art is equally conventionalized, but we do not ordinarily notice this fact unless we are unaccustomed to the convention. In our day the conventional element in literature is elaborately disguised by a law of copyright pretending that every work of art is an invention distinctive enough to be patented. Hence the conventionalizing forces of modern literature - the way, for instance, that an editor's policy and the expectation of his readers combine to conventionalize what appears in a magazine - often go unrecognized. Demonstrating the debt of A to B is merely scholarship if A is dead, but a proof of moral delinquency if A is alive. This state of things makes it difficult to appraise a literature which includes Chaucer, much of whose poetry is translated or paraphrased from others; Shakespeare, whose plays sometimes follow their sources almost verbatim; and Milton, who asked for nothing better than to steal as much as possible out of the Bible. It is not only the inexperienced reader who looks for a residual originality in such works. Most of us tend to think of a poet's real achievement as distinct from, or even contrasted with, the achievement present in what he stole, and we are thus apt to concentrate on peripheral rather than on central critical facts. For instance, the central greatness of Paradise Regained, as a poem, is not the greatness of the rhetorical decorations that Milton added to his source, but the greatness of the theme itself, which Milton passes on to the reader from his source. This conception of the great poet's being entrusted with the great theme was elementary enough to Milton, but violates most of the low mimetic prejudices about creation that most of us are educated in.
The underestimating of convention appears to be a result of, may even be a part of, the tendency, marked from Romantic times  on, to think of the individual as ideally prior to his society. The view opposed to this, that the new baby is conditioned by a hereditary and environmental kinship to a society which already exists, has, whatever doctrines may be inferred from it, the initial advantage of being closer to the facts it deals with. The literary consequence of the second view is that the new poem, like the new baby, is born into an already existing order of words, and is typical of the structure of poetry to which it is attached. The new baby is his own society appearing once again as a unit of individuality, and the new poem has a similar relation to its poetic society.
It is hardly possible to accept a critical view which confuses the original with the aboriginal, and imagines that a "creative" poet sits down with a pencil and some blank paper and eventually produces a new poem in a special act of creation ex nihilo. Human beings do not create in that way. Just as a new scientific discovery manifests something that was already latent in the order of nature, and at the same time is logically related to the total structure of the existing science, so the new poem manifests something that was already latent in the order of words. Literature may have life, reality, experience, nature, imaginative truth, social conditions, or what you will for its content; but literature itself is not made out of these things. Poetry can only be made out of other poems; novels out of other novels. Literature shapes itself, and is not shaped externally: the forms of literature can no more exist outside literature than the forms of sonata and fugue and rondo can exist outside music.
All this was much clearer before the assimilation of literature to private enterprise concealed so many of the facts of criticism. When Milton sat down to write a poem about Edward King, he did not ask himself: "What can I find to say about King?" but "How does poetry require that such a subject should be treated?" The notion that convention shows a lack of feeling, and that a poet attains "sincerity" (which usually means articulate emotion) by disregarding it, is opposed to all the facts of literary experience and history. The origin of this notion is, again, the view that poetry is a description of emotion, and that its "literal" meaning is an assertion about the emotions held by the individual poet. But any serious study of literature soon shows that the real difference between the original and the imitative poet is simply that the former is more profoundly imitative. Originality returns to the  origins of literature, as radicalism returns to its roots. The remark of Mr. Eliot that a good poet is more likely to steal than to imitate affords a more balanced view of convention, as it indicates that the poem is specifically involved with other poems, not vaguely with such abstractions as tradition or style. The copyright law, and the mores attached to it, make it difficult for a modern novelist to steal anything except his title from the rest of literature: hence it is often only in such titles as For Whom the Bell Tolls, The Grapes of Wrath, or The Sound and the Fury, that we can clearly see how much impersonal dignity and richness of association an author can gain by the communism of convention.
As with other products of divine activity, the father of a poem is much more difficult to identify than the mother. That the mother is always nature, the realm of the objective considered as a field of communication, no serious criticism can ever deny. But as long as the father of a poem is assumed to be the poet himself, we have once again failed to distinguish literature from discursive verbal structures. The discursive writer writes as an act of conscious will, and that conscious will, along with the symbolic system he employs for it, is set over against the body of things he is describing. But the poet, who writes creatively rather than deliberately, is not the father of his poem; he is at best a midwife, or, more accurately still, the womb of Mother Nature herself: her privates he, so to speak. The fact that revision is possible, that a poet can make changes in a poem not because he likes them better but because they are better, shows clearly that the poet has to give birth to the poem as it passes through his mind. He is responsible for delivering it in as uninjured a state as possible, and if the poem is alive, it is equally anxious to be rid of him, and screams to be cut loose from all the navel-strings and feeding-tubes of his ego.
The true father or shaping spirit of the poem is the form of the poem itself, and this form is a manifestation of the universal spirit of poetry, the "onlie begetter" of Shakespeare's sonnets who was not Shakespeare himself, much less that depressing ghost Mr. W. H., but Shakespeare's subject, the master-mistress of his passion. When a poet speaks of the internal spirit which shapes the poem, he is apt to drop the traditional appeal to female Muses and think of himself as in a feminine, or at least receptive, relation to some god or lord, whether Apollo, Dionysus, Eros, Christ, or (as in Milton) the Holy Spirit. Est deus in nobis, Ovid says: in modern times we may compare Nietzsche's remarks about his inspiration in Ecce Homo.
The problem of convention is the problem of how art can be communicable, for literature is clearly as much a technique of communication as assertive verbal structures are. Poetry, taken as a whole, is no longer simply an aggregate of artifacts imitating nature, but one of the activities of human artifice taken as a whole. If we may use the word "civilization" for this, we may say that our fourth phase looks at poetry as one of the techniques of civilization. It is concerned, therefore, with the social aspect of poetry, with poetry as the focus of a community. The symbol in this phase is the communicable unit, to which I give the name archetype: that is, a typical or recurring image. I mean by an archetype a symbol which connects one poem with another and thereby helps to unify and integrate our literary experience. And as the archetype is the communicable symbol, archetypal criticism is primarily concerned with literature as a social fact and as a mode of communication. By the study of conventions and genres, it attempts to fit poems into the body of poetry as a whole.
The repetition of certain common images of physical nature like the sea or the forest in a large number of poems cannot in itself be called even "coincidence," which is the name we give to a piece of design when we cannot find a use for it. But it does indicate a certain unity in the nature that poetry imitates, and in the communicating activity of which poetry forms part. Because of the larger communicative context of education, it is possible for a story about the sea to be archetypal, to make a profound imaginative impact, on a reader who has never been out of Saskatchewan. And when pastoral images are deliberately employed in Lycidas, for instance, merely because they are conventional, we can see that the convention of the pastoral makes us assimilate these images to other parts of literary experience.
We think first of the pastoral's descent from Theocritus, where the pastoral elegy first appears as a literary adaptation of the ritual of the Adonis lament, and through Theocritus to Virgil and the whole pastoral tradition to The Shepheardes Calender and beyond to Lycidas itself. Then we think of the intricate pastoral symbolism of the Bible and the Christian Church, of Abel and the twenty-third Psalm and Christ the Good Shepherd, of the ecclesiastical over tones of "pastor" and "flock," and of the link between the Classical  and Christian traditions in Virgil's Messianic Eclogue. Then we think of the extensions of pastoral symbolism into Sidney's Arcadia, The Faerie Queene, Shakespeare's forest comedies, and the like; then of the post-Miltonic development of pastoral elegy in Shelley, Arnold, Whitman, and Dylan Thomas; perhaps too of pastoral conventions in painting and music. In short, we can get a whole liberal education simply by picking up one conventional poem and following its archetypes as they stretch out into the rest of literature. An avowedly conventional poem like Lycidas urgently demands the kind of criticism that will absorb it into the study of literature as a whole, and this activity is expected to begin at once, with the first cultivated reader. Here we have a situation in literature more like that of mathematics or science, where the work of genius is assimilated to the whole subject so quickly that one hardly notices the difference between creative and critical activity.
If we do not accept the archetypal or conventional element in the imagery that links one poem with another, it is impossible to get any systematic mental training out of the reading of literature alone. But if we add to our desire to know literature a desire to know how we know it, we shall find that expanding images into conventional archetypes of literature is a process that takes place unconsciously in all our reading. A symbol like the sea or the heath cannot remain within Conrad or Hardy: it is bound to expand over many works into an archetypal symbol of literature as a whole. Moby Dick cannot remain in Melville's novel: he is absorbed into our imaginative experience of leviathans and dragons of the deep from the Old Testament onward. And what is true for the reader is a fortiori true of the poet, who learns very quickly that there is no singing school for his soul except the study of the monuments of its own magnificence.
In each phase of symbolism there is a point at which the critic is compelled to break away from the range of the poet's own knowledge. Thus the historical or documentary critic has sooner or later to call Dante a "medieval" poet, a notion unknown and unintelligible to Dante. In archetypal criticism, the poet's conscious knowledge is considered only so far as the poet may allude to or imitate other poets ("sources") or make a deliberate use of a convention. Beyond that, the poet's control over his poem stops with the poem. Only the archetypal critic can be concerned with its relationship to the rest of literature. But here again we have  to distinguish between explicitly conventionalized literature, such as Lycidas, where the poet himself starts us off by referring to Theocritus, Virgil, Renaissance pastoralists, and the Bible, and literature which conceals or ignores its conventional links. The conception of copyright and the revolutionary nature of the low mimetic view of creation also extends to a general unwillingness on the part of authors of the copyright age to have their imagery studied conventionally, and in dealing with this period, most archetypes have to be established by critical inspection alone.
To give a random example, one very common convention of the nineteenth-century novel is the use of two heroines, one dark and one light. The dark one is as a rule passionate, haughty, plain, foreign or Jewish, and in some way associated with the undesirable or with some kind of forbidden fruit like incest. When the two are involved with the same hero, the plot usually has to get rid of the dark one or make her into a sister if the story is to end happily. Examples include Ivanhoe, The Last of the Mohicans, The Woman in White, Ligeia, Pierre (a tragedy because the hero chooses the dark girl, who is also his sister), The Marble Faun, and countless incidental treatments. A male version forms the symbolic basis of Wuthering Heights. This device is as much a convention as Milton's calling Edward King by a name out of Virgil's Eclogues, but it shows a confused, or, as we say, "unconscious" approach to conventions. Again, when we meet the images of a man, a woman, and a serpent in the ninth book of Paradise Lost, there is no doubt of their conventional links with similar figures in the Book of Genesis. In Hudson's Green Mansions the hero and heroine first meet over a serpent in a quasi-Paradisal setting: here the conventional nature of the imagery is a matter on which the author gives us no help. When a critic meets St. George the Redcross Knight in Spenser, bearing a red cross on a white ground, he has some idea what to do with this figure. When he meets a female in Henry James's The Other House called Rose Armiger with a white dress and a red parasol, he is, in the current slang, clueless. It is clear that a deficiency in contemporary education often complained of, the disappearance of a common cultural ground which makes a modern poet's allusions to the Bible or to Classical mythology fall with less weight than they should, has much to do with the decline in the explicit use of archetypes.
Whitman, as is well known, was a spokesman of an  anti-archetypal view of literature, and urged the Muse to forget the matter of Troy and develop new themes. This is a low mimetic prejudice, and is consequently appropriate enough for Whitman, who is both right and wrong. He is wrong because the matter of Troy will always be, in the foreseeable future, an integral part of the Western cultural heritage, and hence references to Agamemnon in Yeats's Leda or Eliot's Sweeney among the Nightingales have as much cumulative power as ever for the properly instructed reader. But he is of course perfectly right in feeling that the content of poetry is normally an immediate and contemporary environment. He was right, being the kind of poet he was, in making the content of his own When Lilacs Last in the Dooryard Bloomed an elegy on Lincoln and not a conventional Adonis lament. Yet his elegy is, in its form, as conventional as Lycidas, complete with purple flowers thrown on coffins, a great star drooping in the west, imagery of "ever-returning spring" and all the rest of it. Poetry organizes the content of the world as it passes before the poet, but the forms in which that content is organized come out of the structure of poetry itself.
Archetypes are associative clusters, and differ from signs in being complex variables. Within the complex is often a large number of specific learned associations which are communicable because a large number of people in a given culture happen to be familiar with them. When we speak of "symbolism" in ordinary life we usually think of such learned cultural archetypes as the cross or the crown, or of conventional associations, as of white with purity or green with jealousy. As an archetype, green may symbolize hope or vegetable nature or a go sign in traffic or Irish patriotism as easily as jealousy but the word green as a verbal sign always refers to a certain colour. Some archetypes are so deeply rooted in conventional association that they can hardly avoid suggesting that association, as the geometrical figure of the cross inevitably suggests the death of Christ. A completely conventionalized art would be an art in which the archetypes, or communicable units, were essentially a set of esoteric signs. This can happen in the arts - for instance in some of the sacred dances of India - but it has not happened in Western literature yet, and the resistance of modern writers to having their archetypes "spotted," so to speak, is due to a natural anxiety to keep them as versatile as possible, not pinned down exclusively to one interpretation. A poet may be showing an esoteric  tendency if he specifically points out one association, as Yeats does in his footnotes to some of his early poems. There are no necessary associations: there are some exceedingly obvious ones, such as the association of darkness with terror or mystery, but there are no intrinsic or inherent correspondences which must invariably be present. As we shall see later, there is a context in which the phrase "universal symbol" makes sense, but it is not this context. The stream of literature, however, like any other stream, seeks the easiest channels first: the poet who uses the expected associations will communicate more rapidly.
At one extreme of literature we have the pure convention, which a poet uses merely because it has often been used before in the same way. This is most frequent in naive poetry, in the fixed epithets and phrase-tags of medieval romance and ballad, in the invariable plots and character types of naive drama, and, to a lesser degree, in the topoi or rhetorical commonplaces which, like other ideas in literature, are so dull when stated as propositions, and so rich and variegated when they are used as structural principles in literature. At the other extreme we have the pure variable, where there is a deliberate attempt at novelty or unfamiliarity, and consequently a disguising or complicating of archetypes. Such techniques come very close to a distrust of communication itself as a function of literature. However, extremes meet, as Coleridge said, and anti-conventional poetry soon becomes a convention in its turn, to be explored by hardy scholars accustomed to the dreariness of literary bad lands. Between these extreme points conventions vary from the most explicit to the most indirect, along a scale parallel to the scale of allegory and paradox already dealt with. The two scales may be often confused or identified, but translating imagery into examples and precepts is a quite distinct process from following images into other poems.
Near the extreme of pure convention is translation, paraphrase, and the kind of use which Chaucer makes of Boccaccio in Troilus and The Knight's Tale. Next we come to deliberate and explicit convention, such as we have noted in Lycidas. Next comes paradoxical or ironic convention, including parody often a sign that certain vogues in handling conventions are getting worn out. Then comes the attempt to reach originality through turning one's back on explicit convention, an attempt which results in implicit convention of the kind we detected in Whitman. Then comes a  tendency to identify originality with "experimental" writing, based in our day on an analogy with scientific discovery, and which is frequently spoken of as "breaking with convention." And, of course, at every stage of literature, including this last one, there is a great deal of superficial and inorganic convention, producing the kind of writing that most students of literature prefer to keep in the middle distance: run-of-the-mill Elizabethan sonnets and love lyrics, Plautine comedy-formulas, eighteenth-century pastorals, nineteenth- century happy-ending novels, works of followers and disciples and schools and trends generally.
It is clear from all this that archetypes are most easily studied in highly conventionalized literature: that is, for the most part, naive, primitive, and popular literature. In suggesting the possibility of archetypal criticism, then, I am suggesting the possibility of extending the kind of comparative and morphological study now made of folk tales and ballads into the rest of literature. This should be more easily conceivable now that it is no longer fashionable to mark off popular and primitive literature from ordinary literature as sharply as we used to do. Also, we shall find that superficial literature, of the kind just spoken of, is of great value to archetypal criticism simply because it is conventional. If throughout this book I refer to popular fiction as frequently as to the greatest novels and epics, it is for the same reason that a musician attempting to explain the rudimentary facts about counterpoint would be more likely, at least at first, to illustrate from "Three Blind Mice" than from a complex Bach fugue.
Every phase of symbolism has its particular approach to narrative and to meaning. In the literal phase, narrative is a flow of significant sounds, and meaning an ambiguous and complex verbal pattern. In the descriptive phase, narrative is an imitation of real events, and meaning an imitation of actual objects or propositions. In the formal phase, poetry exists between the example and the precept. In the exemplary event there is an element of recurrence; in the precept, or statement about what ought to be, there is a strong element of desire, or what is called "wish-thinking." These elements of recurrence and desire come into the foreground in archetypal criticism, which studies poems as units of poetry as a whole and symbols as units of communication.
From such a point of view, the narrative aspect of literature is  a recurrent act of symbolic communication: in other words a ritual. Narrative is studied by the archetypal critic as ritual or imitation of human action as a whole, and not simply as a mimesis praxeos or imitation of an action. Similarly, in archetypal criticism the significant content is the conflict of desire and reality which has for its basis the work of the dream. Ritual and dream, therefore, are the narrative and significant content respectively of literature in its archetypal aspect. The archetypal analysis of the plot of a novel or play would deal with it in terms of the generic, recurring, or conventional actions which show analogies to rituals: the weddings, funerals, intellectual and social initiations, executions or mock executions, the chasing away of the scapegoat villain, and so on. The archetypal analysis of the meaning or significance of such a work would deal with it in terms of the generic, recurring, or conventional shape indicated by its mood and resolution, whether tragic, comic, ironic, or what not, in which the relationship of desire and experience is expressed.
Recurrence and desire interpenetrate, and are equally important in both ritual and dream. In its archetypal phase, the poem imitates nature, not (as in the formal phase) nature as a structure or system, but nature as a cyclical process. The principle of recurrence in the rhythm of art seems to be derived from the repetitions in nature that make time intelligible to us. Rituals cluster around the cyclical movements of the sun, the moon, the seasons, and human life. Every crucial periodicity of experience: dawn, sunset, the phases of the moon, seed-time and harvest, the equinoxes and the solstices, birth, initiation, marriage, and death, get rituals attached to them. The pull of ritual is toward pure cyclical narrative, which, if there could be such a thing, would be automatic and unconscious repetition. In the middle of all this recurrence, however, is the central recurrent cycle of sleeping and waking life, the daily frustration of the ego, the nightly awakening of a titanic self.
The archetypal critic studies the poem as part of poetry, and poetry as part of the total human imitation of nature that we call civilization. Civilization is not merely an imitation of nature but the process of making a total human form out of nature and it is impelled by the force that we have just called desire. The desire for food and shelter is not content with roots and caves: it produces the human forms of nature that we call farming and architecture. Desire is thus not a simple response to need, for an animal may  need food without planting a garden to get it, nor is it a simple response to want, or desire for something in particular. It is neither limited to nor satisfied by objects, but is the energy that leads human society to develop its own form. Desire in this sense is the social aspect of what we met on the literal level as emotion, an impulse toward expression which would have remained amorphous if the poem had not liberated it by providing the form of its expression. The form of desire, similarly, is liberated and made apparent by civilization. The efficient cause of civilization is work, and poetry in its social aspect has the function of expressing, as a verbal hypothesis, a vision of the goal of work and the forms of desire.
There is however a moral dialectic in desire. The conception of a garden develops the conception "weed" and building a sheepfold makes the wolf a greater enemy. Poetry in its social or archetypal aspect, therefore, not only tries to illustrate the fulfilment of desire, but to define the obstacles to it. Ritual is not only a recurrent act, but an act expressive of a dialectic of desire and repugnance: desire for fertility or victory, repugnance to drought or to enemies. We have rituals of social integration, and we have rituals of expulsion, execution, and punishment. In dream there is a parallel dialectic, as there is both the wish-fulfilment dream and the anxiety or nightmare dream of repugnance. Archetypal criticism, therefore, rests on two organizing rhythms or patterns, one cyclical, the other dialectic.
The union of ritual and dream in a form of verbal communication is myth. This is a sense of the term myth slightly different from that used in the previous essay. But, first, the sense is equally familiar, and the ambiguity not mine but the dictionary's; and, second, there is a real connection between the two senses which will become more apparent as we go on. The myth accounts for, and makes communicable, the ritual and the dream. Ritual, by itself, cannot account for itself: it is pre-logical, pre-verbal, and in a sense pre-human. Its attachment to the calendar seems to link human life to the biological dependence on the natural cycle which plants, and to some extent animals, still have. Everything in nature that we think of as having some analogy with works of art, like the flower or the bird's song, grows out of a synchronization between an organism and the rhythms of its natural environment, especially that of the solar year. With animals some expressions of  synchronization, like the mating dances of birds, could almost be called rituals. Myth is more distinctively human, as the most intelligent partridge cannot tell even the most absurd story explaining why it drums in the mating season. Similarly, the dream, by itself, is a system of cryptic allusions to the dreamer's own life, not fully understood by him, or so far as we know of any real use to him. But in all dreams there is a mythical element which has a power of independent communication, as is obvious, not only in the stock example of Oedipus, but in any collection of folk tales. Myth, therefore, not only gives meaning to ritual and narrative to dream: it is the identification of ritual and dream, in which the former is seen to be the latter in movement. This would not be possible unless there were a common factor to ritual and dream which made one the social expression of the other; the investigation of this common factor we must leave for later treatment. All that we need to say here is that ritual is the archetypal aspect of mythos and dream the archetypal aspect of dianoia.
The same distinction in emphasis that we noted in the first essay between fictional and thematic literature recurs here. Some literary forms, such as drama, remind us with particular vividness of analogies to rituals, for the drama in literature, like the ritual in religion, is primarily a social or ensemble performance. Others, such as romance, suggest analogies to dreams. Ritual analogies are most easily seen, not in the drama of the educated audience and the settled theatre, but in naive or spectacular drama: in the folk play, the puppet show, the pantomime, the farce, the pageant, and their descendants in masque, comic opera, commercial movie, and revue. Dream analogies are best studied in naive romance, which includes the folk tales and fairy tales that are so closely related to dreams of wonderful wishes corning true, and to nightmares of ogres and witches. Naive drama and naive romance, of course, also interpenetrate. What naive drama dramatizes is usually some kind of romance, and the close relation of romance to ritual can be seen in the number of medieval romances that are linked to some part of the calendar, the winter solstice, a May morning, or a saint's eve; or else to some class ritual like the tournament. The fact that the archetype is primarily a communicable symbol largely accounts for the ease with which ballads and folk tales and mimes travel through the world, like so many of their heroes, over all barriers of language and culture. We come back here to the fact that literature  most deeply influenced by the archetypal phase of symbolism impresses us as primitive and popular.
By these words I mean possessing the ability to communicate in time and space respectively. Otherwise they mean much the same thing. Popular art is normally decried as vulgar by the cultivated people of its time; then it loses favor with its original audience as a new generation grows up; then it begins to merge into the softer lighting of "quaint," and cultivated people become interested in it, and finally it begins to take on the archaic dignity of the primitive. This sense of the archaic recurs whenever we find great art using popular forms, as Shakespeare does in his last period, or as the Bible does when it ends in a fairy tale about a damsel in distress, a hero killing dragons, a wicked witch, and a wonderful city glittering with jewels. Archaism is a regular feature of all social uses of archetypes. Soviet Russia is very proud of its production of tractors, but it will be some time before the tractor replaces the sickle on the Soviet flag.
It is at this point that we must notice and avoid the fallacy of a theory of mythological contract. That is, there may be such a thing as a social contract in political theory, if we keep the discussion to observable facts about the present structure of society. But when these facts are attached to a fable about something that happened in a past too remote for any evidence to disturb the fabler's assertions, and we are told that once upon a time men surrendered or delegated or were tricked into surrendering their power, political theory has merely become one of Plato's indoctrinating lies. And because the only evidence for this remote event is its analogy to the present facts, the present facts are being compared with their own shadows. A precisely similar fabling process has taken place in the literary criticism concerned with myth, which has hardly yet emerged from its historical contract stage.
As the archetypal critic is concerned with ritual and dream, it is likely that he would find much of interest in the work done by contemporary anthropology in ritual, and by contemporary psychology in dreams. Specifically, the work done on the ritual basis of naive drama in Frazer's Golden Bough, and the work done on the dream basis of naive romance by Jung and the Jungians, are of most direct value to him. But the three subjects of anthropology, psychology, and literary criticism are not yet clearly separated, and the danger of determinism has to be carefully watched. To the  literary critic, ritual is the content of dramatic action, not the source or origin of it. The Golden Bough is, from the point of view of literary criticism, an essay on the ritual content of naive drama: that is, it reconstructs an archetypal ritual from which the structural and generic principles of drama may be logically, not chronologically, derived. It does not matter two pins to the literary critic whether such a ritual had any historical existence or not. It is very probable that Frazer's hypothetical ritual would have many and striking analogies to actual rituals, and collecting such analogies is part of his argument. But an analogy is not necessarily a source, an influence, a cause, or an embryonic form, much less an identity. The literary relation of ritual to drama, like that of any other aspect of human action to drama, is a relation of content to form only, not one of source to derivation.
The critic, therefore, is concerned only with the ritual or dream patterns which are actually in what he is studying, however they got there. The work of the Classical scholars who have followed Frazer's lead has produced a general theory of the spectacular or ritual content of Greek drama. The Golden Bough purports to be a work of anthropology, but it has had more influence on literary criticism than in its own alleged field, and it may yet prove to be really a work of literary criticism. If the ritual pattern is in the plays and it is fact, not opinion, that one of the main themes of Iphigeneia in Tauris, for example, is human sacrifice the critic need not take sides in the quite separate historical controversy over the ritual origin of Greek drama. Hence ritual, as the content of action, and more particularly of dramatic action, is something continuously latent in the order of words, and is quite independent of direct influence. Even in the nineteenth century, we find that the instant drama becomes primitive and popular, as it does in The Mikado, to repeat an example given before, back comes all Frazer's apparatus, the king's son, the mock sacrifice, the analogy with the festival of the Sacaea, and many other things that Gilbert knew and cared nothing about. It comes back because it is still the best way of holding an audience's attention, and the experienced dramatist knows it.
The prestige of documentary criticism, which deals entirely with sources and historical transmission, has misled some archetypal critics into feeling that all such ritual elements ought to be traced directly, like the lineage of royalty, as far back as a willing  suspension of disbelief will allow. The vast chronological gaps resulting are usually bridged by some theory of race memory, or by some conspiratorial conception of history involving secrets jealously guarded for centuries by esoteric cults or traditions. It is curious that when archetypal critics hang on to a historical framework they almost invariably produce some hypothesis of continuous degeneration from a golden age lost in antiquity. Thus the prelude to Thomas Mann's Joseph series traces back several of our central myths to Atlantis, Atlantis being clearly more useful as an archetypal idea than as a historical one. When archetypal criticism revived in the nineteenth century with a vogue for sun myths, an attempt was made to ridicule it by proving with equal plausibility that Napoleon was a sun myth. The ridicule is effective only against the historical distortion of the method. Archetypally, we turn Napoleon into a sun myth whenever we speak of the rise of his career, the zenith of his fame, or the eclipse of his fortunes.
Social and cultural history, which is anthropology in an extended sense, will always be a part of the context of criticism, and the more clearly the anthropological and the critical treatments of ritual are distinguished, the more beneficial their influence on each other will be. The same is true of the relation of psychology to criticism. The first and most striking unit of poetry larger than the individual poem is the total work of the man who wrote the poem. Biography will always be a part of criticism, and the biographer will naturally be interested in his subject's poetry as a personal document, recording his private dreams, associations, ambitions, and expressed or repressed desires. Studies of such matters form an essential part of criticism. I am not of course speaking of the silly ones, which simply project the author's own erotica, in a rationalized clinical disguise, on his victim, but only of the serious studies which are technically competent both in psychology and in criticism, which are aware how much guesswork is involved and how tentative all the conclusions must be.
Such an approach is easiest, and most rewarding, with what we have called thematic writers of the low mimetic that is, chiefly, the Romantic poets, where the poet's own psychological processes are often part of the theme. With other writers, say a dramatist who is aware from the first word he writes that "They who live to please must please to live," there is danger of making an unreal abstraction of the poet from his literary community. Suppose a  critic finds that a certain pattern is repeated time and again in the plays of Shakespeare. If Shakespeare is unique or anomalous, or even exceptional, in using this pattern, the reason for his use of it may be at least partly psychological. If there were any evidence that he had persisted in using it when it failed to please an audience, the probability of a personal psychological element would be very high. But if we can find the same pattern in half a dozen of his contemporaries, we clearly have to allow for convention. And if we find it in a dozen dramatists of different ages and cultures, we have to allow for genre, for the structural requirements of drama itself. Now as a matter of fact we do find in Shakespeare's comedies that the same devices are used over and over, and it is the business of the literary critic to compare these devices with those of other dramatists, in a morphological study of comic form. Otherwise we shall deprive ourselves of the perfectly legitimate appreciation of the scholarly qualities of Shakespeare, of seeing in the repeated devices of his comedies a kind of Art of Fugue of comedy.
A psychologist examining a poem will tend to see in it what he sees in the dream, a mixture of latent and manifest content. For the literary critic the manifest content of the poem is its form, hence its latent content becomes simply its actual content, its dianoia or theme. And this dianoia on the archetypal level is a dream, a presentation of the conflict of desire and actuality. We seem to be going around in a circle, but not quite. For the critic, a problem appears which does not exist for a purely psychological analysis, the problem of communicable latent content, of intelligible dream, Plato's conception of art as a dream for awakened minds. For the psychologist all dream symbols are private ones, interpreted by the personal life of the dreamer. For the critic there is no such thing as private symbolism, or, if there is, it is his job to make sure that it does not remain so.
This problem is already present in Freud's treatment of Oedipus Tyrannus as a play which owes much of its power to the fact that it dramatizes the Oedipus complex. The dramatic and psychological elements can be linked without any reference to the personal life of Sophocles, of which we know nothing whatever. This emphasis on impersonal content has been developed by Jung and his school, where the communicability of archetypes is accounted for by a  theory of a collective unconscious - an unnecessary hypothesis in literary criticism, so far as I can judge.
What we have found to be true of the writer's intention is also true of the audience's attention. Both are centripetally directed, and implications exist in the response to art as they do in the creation of it, implications of which the audience is not explicitly aware. Discrete conscious awareness can take in only a very few details of the complex of response. This state of things enabled Tennyson, for instance, to be praised for the chastity of his language and read for his powerful erotic sensuousness. It also makes it possible for a contemporary critic to draw on the fullest resources of modern knowledge in explicating a work of art without any real fear of anachronism.
For instance, Le Malade Imaginaire is a play about a man who, in seventeenth-century terms, including no doubt Moliere's own terms, was not really sick but just thought he was. A modern critic might object that life is not so simple: that it is perfectly possible for a malade imaginaire to be a malade veritable, and that what is wrong with Argan is clearly an unwillingness to see his children grow up, an infantile regression which his wife - his second wife, incidentally - shows that she understands completely by coddling him and murmuring such phrases as "pauvre petit fils." Such a critic would find the clue to Argan's whole behavior in his unguarded remark after the scene with the little girl Louison (the erotic nature of which the critic would also notice): "II n'y a plus d'enfants." Now whether this reading is right or wrong, it does not swerve from Moliere's text, yet it tells us nothing about Moliere himself. The play is generically a comedy; it must therefore end happily; Argan must therefore be brought to see some reason; his wife, whose dramatic function it is to keep him within his obsession, must therefore be "exposed" as inimical to him. The plot is a ritual moving toward a scapegoat rejection followed by a marriage, and the theme is a dream-pattern of irrational desire in conflict with reality.
Another essay in this book will be concerned with the details and practice of archetypal criticism: here we are concerned only with its place in the context of criticism as a whole. In its archetypal aspect, art is a part of civilization, and civilization we defined as the process of making a human form out of nature. The shape  of this human form is revealed by civilization itself as it develops: its major components are the city, the garden, the farm, the sheep fold, and the like, as well as human society itself. An archetypal symbol is usually a natural object with a human meaning, and it forms part of the critical view of art as a civilized product, a vision of the goals of human work.
Such a vision is bound to idealize some aspects of civilization and ridicule or ignore others; in other words the social context of art is also the moral context. All artists have to come to terms with their communities: many artists, and many great ones, are content to be the spokesmen of them. But in terms of his moral significance, the poet reflects, and follows at a distance, what his community really achieves through its work. Hence the moral view of the artist is invariably that he ought to assist the work of his society by framing workable hypotheses, imitating human action and thought in such a way as to suggest realizable modes of both. If he does not do this, his hypotheses should at least be clearly labelled as playful or fantastic. Marxism takes more or less this view of art, and thereby repeats the argument reached at the end of the Republic. We are told there, if we follow the argument simply as it stands, that according to justice, or social work properly done, the painter's bed is an external imitation of the craftsman's bed. The artist, therefore, is confined either to reflecting or to escaping from the world that the true worker is realizing.
We have adopted the principle in this essay that the events and ideas of poetry are hypothetical imitations of history and discursive writing respectively, which in their turn are verbal imitations of action and thought. This principle brings us close to the view of poetry as a secondary imitation of reality. We are interpreting mimesis, however, not as a Platonic "recollection" but as an emancipation of externality into image, nature into art. From this point of view the work of art must be its own object: it cannot be ultimately descriptive of something, and can never be ultimately related to any other system of phenomena, standards, values, or final causes. All such external relations form part of the "intentional fallacy." Poetry is a vehicle for morality, truth, and beauty, but the poet does not aim at these things, but only at inner verbal strength. The poet qua poet intends only to write a poem, and as a rule it is not the artist, but the ego in the artist, who turns  away from his proper work to go and chase these other seductive marshlights.
It is an elementary axiom in criticism that morally the lion lies down with the lamb. Bunyan and Rochester, Sade and Jane Austen, The Miller's Tale and The Second Nun's Tale, are all equally elements of a liberal education, and the only moral criterion to be applied to them is that of decorum. Similarly, the moral attitude taken by the poet in his work derives largely from the structure of that work. Thus the fact that Le Malade Imaginaire is a comedy is the only reason for making Argan's wife a hypocrite - she must be got rid of to make the play end happily.
The pursuit of beauty is much more dangerous nonsense than the pursuit of truth or goodness, because it affords a stronger temptation to the ego. Beauty, like truth and goodness, is a quality that may in one sense be predicated of all great art, but the deliberate attempt to beautify can, in itself, only weaken the creative energy. Beauty in art is like happiness in morals: it may accompany the act, but it cannot be the goal of the act, just as one cannot "pursue happiness," but only something else that may give happiness. Aiming at beauty produces, at best, the attractive: the quality of beauty represented by the word loveliness, a quality which depends on a carefully restricted choice of both subject and technique. A religious painter, for instance, can produce this quality only as long as churches keep commissioning Madonnas: if a church asks for a Crucifixion he must paint cruelty and horror instead.
When we speak of the human body as "beautiful," we usually mean the body of someone in good physical condition between eighteen and about thirty, and if Degas, for example, shows us pictures of thick-bottomed matrons squatting in hip baths, we interpret the shock to our propriety as an aesthetic judgement. Whenever the word beauty means loveliness or attractiveness, as it is bound to do whenever it is made the intention of art, it be comes reactionary: it tries to restrict either what the artist may choose for a subject or the method in which he may choose to treat it, and it marshals all the forces of prudery to keep him from expanding his vision beyond an arid and insipid pseudo-classicism. Ruskin spoiled many of his finest critical insights with this fallacy; Tennyson often hampered the vigor of his poetry by it, and in some of the lesser beauticians of the same period we can see clearly what the neurotic compulsion to beautify everything leads to. It  leads to an exaggerated cult of style, a technique of making everything in a work of art, even a drama, sound all alike, and like the author, and like the author at his most impressive. Here again the vanity of the ego has replaced the honest pride of the craftsman.
The formal or third phase of narrative and meaning, although it includes the external relations of literature to events and ideas, nevertheless brings us back ultimately to the aesthetic view of the work of art as an object of contemplation, a techne designed for ornament and pleasure rather than use. This view encourages us to separate aesthetic objects from other kinds of artifacts and to postulate an aesthetic experience different in kind from other experiences. Corresponding to the bibliographical view of literature as the aggregate or pile of all the books and plays and poems that have been written, we find the aesthetic view of criticism as a discrete series of special (sometimes vaguely sacramental) apprehensions. There is no reason for not granting this view of literary experience its own validity; one objects to it only when it excludes other approaches.
The archetypal view of literature shows us literature as a total form and literary experience as a part of the continuum of life, in which one of the poet's functions is to visualize the goals of human work. As soon as we add this approach to the other three, literature becomes an ethical instrument, and we pass beyond Kierkegaard's "Either/Or" dilemma between aesthetic idolatry and ethical freedom, without any temptation to dispose of the arts in the process. Hence the importance, after accepting the validity of this view of literature, of rejecting the external goals of morality, beauty, and truth. The fact that they are external makes them ultimately idolatrous, and so demonic. But if no social, moral, or aesthetic standard is in the long run externally determinative of the value of art, it follows that the archetypal phase, in which art is part of civilization, cannot be the ultimate one. We need still another phase where we can pass from civilization, where poetry is still useful and functional, to culture, where it is disinterested and liberal, and stands on its own feet.
Anagogic Phase: Symbol as Monad
In tracing the different phases of literary symbolism, we have been going up a sequence parallel to that of medieval criticism. We have, it is true, established a different meaning for the word  "literal." It is our second or descriptive level that corresponds to the historical or literal one of the medieval scheme, or at any rate of Dante's version of it. Our third level, the level of commentary and interpretation, is the second or allegorical level of the Middle Ages. Our fourth level, the study of myths, and of poetry as a technique of social communication, is the third medieval level of moral and tropological meaning, concerned at once with the social and the figurative aspect of meaning. The medieval distinction between the allegorical as what one believes (quid credas) and the moral as what one does (quid agas) is also reflected in our conception of the formal phase as aesthetic or speculative and the archetypal phase as social and part of the continuum of work. We have now to see if we can establish a modern parallel to the medieval conception of anagogy or universal meaning.
Again, the reader may have noticed a parallelism gradually shaping up between the five modes of our first essay and the phases of symbolism in this one. Literal meaning, as we expounded it, has much to do with the techniques of thematic irony introduced by symbolisme, and with the view of many of the "new" critics that poetry is primarily (i.e., literally) an ironic structure. Descriptive symbolism, shown at its most uncompromising in the documentary naturalism of the nineteenth century, seems to bear a close connection with the low mimetic, and formal symbolism, most easily studied in Renaissance and neo-Classical writers, with the high mimetic. Archetypal criticism seems to find its center of gravity in the mode of romance, when the interchange of ballads, folk tales, and popular stories was at its easiest. If the parallel holds, then, the last phase of symbolism will still be concerned, as the previous one was, with the mythopoeic aspect of literature, but with myth in its narrower and more technical sense of fictions and themes relating to divine or quasi-divine beings and powers.
We have associated archetypes and myths particularly with primitive and popular literature. In fact we could almost define popular literature, admittedly in a rather circular way, as literature which affords an unobstructed view of archetypes. We can find this quality on every level of literature: in fairy tales and folk tales, in Shakespeare (in most of the comedies), in the Bible (which would still be a popular book if it were not a sacred one), in Bunyan, in Richardson, in Dickens, in Poe, and of course in a vast amount of ephemeral rubbish as well. We began this book by remarking  that we cannot correlate popularity and value. But there is still the danger of reduction, or assuming that literature is essentially primitive and popular. This view had a great vogue in the nineteenth century, and is by no means dead yet, but if we were to adopt it we should cut off a third and most important source of supply for archetypal criticism.
We notice that many learned and recondite writers whose work requires patient study are explicitly mythopoeic writers. Instances include Dante and Spenser, and in the twentieth century embrace nearly all the "difficult" writers in both poetry and prose. Such work, when fictional, is often founded on a basis of naive drama (Faust, Peer Gynt) or naive romance (Hawthorne, Melville: one may compare the sophisticated allegories of Charles Williams and C. S. Lewis in our day, which are largely based on the formulas of the Boy's Own Paper). Learned mythopoeia, as we have it in the last period of Henry James and in James Joyce, for example, may become bewilderingly complex; but the complexities are designed to reveal and not to disguise the myth. We cannot assume that a primitive and popular myth has been swathed like a mummy in elaborate verbiage, which is the assumption that the fallacy of reduction would lead to. The inference seems to be that the learned and the subtle, like the primitive and the popular, tend toward a center of imaginative experience.
Knowing that The Two Gentlemen of Verona is an early Shakespeare comedy and The Winter's Tale a late one, the student would expect the later play to be more subtle and complex; he might not expect it to be more archaic and primitive, more suggestive of ancient myths and rituals. The later play is also more popular, though not popular of course in the sense of giving a lower-middle class audience what it thinks it wants. As a result of expressing the inner forms of drama with increasing force and intensity, Shakespeare arrived in his last period at the bedrock of drama, the romantic spectacle out of which all the more specialized forms of drama, such as tragedy and social comedy, have come, and to which they recurrently return. In the greatest moments of Dante and Shakespeare, in, say The Tempest or the climax of the Purgatorio, we have a feeling of converging significance, the feeling that here we are close to seeing what our whole literary experience has been about, the feeling that we have moved into the still center of the order of words. Criticism as knowledge, the criticism which is  compelled to keep on talking about the subject, recognizes the fact that there is a center of the order of words.
Unless there is such a center, there is nothing to prevent the analogies supplied by convention and genre from being an endless series of free associations, perhaps suggestive, perhaps even tantalizing, but never creating a real structure. The study of archetypes is the study of literary symbols as parts of a whole. If there are such things as archetypes at all, then, we have to take yet another step, and conceive the possibility of a self-contained literary universe. Either archetypal criticism is a will-o'-the-wisp, an endless labyrinth without an outlet, or we have to assume that literature is a total form, and not simply the name given to the aggregate of existing literary works. We spoke before of the mythical view of literature as leading to the conception of an order of nature as a whole being imitated by a corresponding order of words.
If archetypes are communicable symbols, and there is a center of archetypes, we should expect to find, at that center, a group of universal symbols. I do not mean by this phrase that there is any archetypal code book which has been memorized by all human societies without exception. I mean that some symbols are images of things common to all men, and therefore have a communicable power which is potentially unlimited. Such symbols include those of food and drink, of the quest or journey, of light and darkness, and of sexual fulfilment, which would usually take the form of marriage. It is inadvisable to assume that an Adonis or Oedipus myth is universal, or that certain associations, such as the serpent with the phallus, are universal, because when we discover a group of people who know nothing of such matters we must assume that they did know and have forgotten, or do know and won't tell, or are not members of the human race. On the other hand, they may be confidently excluded from the human race if they cannot understand the conception of food, and so any symbolism founded on food is universal in the sense of having an indefinitely extensive scope. That is, there are no limits to its intelligibility.
In the archetypal phase the work of literary art is a myth, and unites the ritual and the dream. By doing so it limits the dream: it makes it plausible and acceptable to a social waking consciousness. Thus as a moral fact in civilization, literature embodies a good deal of the spirit which in the dream itself is called the censor. But the censor stands in the way of the impetus of the dream.  When we look at the dream as a whole, we notice three things about it. First, its limits are not the real, but the conceivable. Second, the limit of the conceivable is the world of fulfilled desire emancipated from all anxieties and frustrations. Third, the universe of the dream is entirely within the mind of the dreamer.
In the anagogic phase, literature imitates the total dream of man, and so imitates the thought of a human mind which is at the circumference and not at the center of its reality. We see here the completion of the imaginative revolution begun when we passed from the descriptive to the formal phase of symbolism. There, the imitation of nature shifted from a reflection of external nature to a formal organization of which nature was the content. But in the formal phase the poem is still contained by nature, and in the archetypal phase the whole of poetry is still contained within the limits of the natural, or plausible. When we pass into anagogy, nature becomes, not the container, but the thing contained, and the archetypal universal symbols, the city, the garden, the quest, the marriage, are no longer the desirable forms that man constructs inside nature, but are themselves the forms of nature. Nature is now inside the mind of an infinite man who builds his cities out of the Milky Way. This is not reality, but it is the conceivable or imaginative limit of desire, which is infinite, eternal, and hence apocalyptic. By an apocalypse I mean primarily the imaginative conception of the whole of nature as the content of an infinite and eternal living body which, if not human, is closer to being human than to being inanimate. "The desire of man being infinite," said Blake, "the possession is infinite and himself infinite." If Blake is thought a prejudiced witness on this point, we may cite Hooker: "That there is somewhat higher than either of these two (sensual and intellectual perfection), no other proof doth need than the very process of man's desire, which being natural should be frustrate, if there were not some farther thing wherein it might rest at the length contented, which in the former it cannot do."
If we turn to ritual, we see there an imitation of nature which has a strong element of what we call magic in it. Magic seems to begin as something of a voluntary effort to recapture a lost rapport with the natural cycle. This sense of a deliberate recapturing of something no longer possessed is a distinctive mark of human ritual. Ritual constructs a calendar and endeavors to imitate the precise and sensitive accuracy of the movements of the heavenly  bodies and the response of vegetation to them. A farmer must harvest his crop at a certain time of the year, but because he must do this anyway, harvesting itself is not precisely a ritual. It is the expression of a will to synchronize human and natural energies at that time which produces the harvest songs, harvest sacrifices, and harvest folk customs that we associate with ritual. But the impetus of the magical element in ritual is clearly toward a universe in which a stupid and indifferent nature is no longer the container of human society, but is contained by that society, and must rain or shine at the pleasure of man. We notice too the tendency of ritual to become not only cyclical but encyclopaedic, as already noted. In its anagogic phase, then, poetry imitates human action as total ritual, and so imitates the action of an omnipotent human society that contains all the powers of nature within itself.
Anagogically, then, poetry unites total ritual, or unlimited social action, with total dream, or unlimited individual thought. Its universe is infinite and boundless hypothesis: it cannot be contained within any actual civilization or set of moral values, for the same reason that no structure of imagery can be restricted to one allegorical interpretation. Here the dianoia of art is no longer a mimesis logou, but the Logos, the shaping word which is both reason and, as Goethe's Faust speculated, praxis or creative act. The ethos of art is no longer a group of characters within a natural setting, but a universal man who is also a divine being, or a divine being conceived in anthropomorphic terms.
The form of literature most deeply influenced by the anagogic phase is the scripture or apocalyptic revelation. The god, whether traditional deity, glorified hero, or apotheosized poet, is the central image that poetry uses in trying to convey the sense of unlimited power in a humanized form. Many of these scriptures are documents of religion as well, and hence are a mixture of the imaginative and the existential. When they lose their existential content they become purely imaginative, as Classical mythology did after the rise of Christianity. They belong in general, of course, to the mythical or theogonic mode. We see the relation to anagogy also in the vast encyclopaedic structure of poetry that seems to be a whole world in itself, that stands in its culture as an inexhaustible storehouse of imaginative suggestion, and seems, like theories of gravitation or relativity in the physical universe, to be applicable to, or have analogous connections with, every part of the literary  universe. Such works are definitive myths, or complete organizations of archetypes. They include what in the previous essay we called analogies of revelation: the epics of Dante and Milton and their counterparts in the other modes.
But the anagogic perspective is not to be confined only to works that seem to take in everything, for the principle of anagogy is not simply that everything is the subject of poetry, but that anything may be the subject of a poem. The sense of the infinitely varied unity of poetry may come, not only explicitly from an apocalyptic epic, but implicitly from any poem. We said that we could get a whole liberal education by picking up one conventional poem, Lycidas for example, and following its archetypes through literature. Thus the center of the literary universe is whatever poem we happen to be reading. One step further, and the poem appears as a microcosm of all literature, an individual manifestation of the total order of words. Anagogically, then, the symbol is a monad, all symbols being united in a single infinite and eternal verbal symbol which is, as dianoia, the Logos, and, as mythos, total creative act. It is this conception which Joyce expresses, in terms of subject-matter, as "epiphany," and Hopkins, in terms of form, as "inscape."
If we look at Lycidas anagogically, for example, we see that the subject of the elegy has been identified with a god who personifies both the sun that falls into the western ocean at night and the vegetable life that dies in the autumn. In the latter aspect Lycidas is the Adonis or Tammuz whose "annual wound," as Milton calls it elsewhere, was the subject of a ritual lament in Mediterranean religion, and has been incorporated in the pastoral elegy since Theocritus, as the title of Shelley's Adonais shows more clearly. As a poet, Lycidas's archetype is Orpheus, who also died young, in much the same role as Adonis, and was flung into the water. As priest, his archetype is Peter, who would have drowned on the "Galilean lake" without the help of Christ. Each aspect of Lycidas poses the question of premature death as it relates to the life of man, of poetry, and of the Church. But all of these aspects are contained within the figure of Christ, the young dying god who is eternally alive, the Word that contains all poetry, the head and body of the Church, the good Shepherd whose pastoral world sees no winter, the Sun of righteousness that never sets, whose power can raise Lycidas, like Peter, out of the waves, as it redeems souls from the lower world, which Orpheus failed to do. Christ does not  enter the poem as a character, but he pervades every line of it so completely that the poem, so to speak, enters him.
Anagogic criticism is usually found in direct connection with religion, and is to be discovered chiefly in the more uninhibited utterances of poets themselves. It comes out in those passages of Eliot's quartets where the words of the poet are placed within the context of the incarnate Word. An even clearer statement is in a letter of Rilke, where he speaks of the function of the poet as revealing a perspective of reality like that of an angel, containing all time and space, who is blind and looking into himself. Rilke's angel is a modification of the more usual god or Christ, and his statement is all the more valuable because it is explicitly not Christian, and illustrates the independence of the anagogic perspective, of the poet's attempt to speak from the circumference instead of from the center of reality, from the acceptance of any specific religion. Similar views are expressed or implied in Valry's conception of a total intelligence which appears more fancifully in his figure of M. Teste; in Yeats's cryptic utterances about the artifice of eternity, and, in The Tower and elsewhere, about man as the creator of all creation as well as of both life and death; in Joyce's non-theological use of the theological term epiphany; in Dylan Thomas's exultant hymns to a universal human body. We may note in passing that the more sharply we distinguish the poetic and the critical functions, the easier it is for us to take seriously what great writers have said about their work.
The anagogic view of criticism thus leads to the conception of literature as existing in its own universe, no longer a commentary on life or reality, but containing life and reality in a system of verbal relationships. From this point of view the critic can no longer think of literature as a tiny palace of art looking out upon an in conceivably gigantic "life." "Life" for him has become the seed-plot of literature, a vast mass of potential literary forms, only a few of which will grow up into the greater world of the literary universe. Similar universes exist for all the arts. "We make to ourselves pictures of facts," says Wittgenstein, but by pictures he means representative illustrations, which are not pictures. Pictures as pictures are themselves facts, and exist only in a pictorial universe. "Tout, au monde," says Mallarmé, "existe pour aboutir a un livre."
So far we have been dealing with symbols as isolated units, but  clearly the unit of relationship between two symbols, corresponding to the phrase in music, is of equal importance. The testimony of critics from Aristotle on seems fairly unanimous that this unit of relationship is the metaphor. And the metaphor, in its radical form, is a statement of identity of the "A is B" type, or rather, putting it into its proper hypothetical form, of the "let X be Y" type (letters altered for euphony). Thus the metaphor turns its back on ordinary descriptive meaning, and presents a structure which literally is ironic and paradoxical. In ordinary descriptive meaning, if A is B then B is A, and all we have really said is that A is itself. In the metaphor two things are identified while each retains its own form. Thus if we say "the hero was a lion" we identify the hero with the lion, while at the same time both the hero and the lion are identified as themselves. A work of literary art owes its unity to this process of identification with, and its variety, clarity, and intensity to identification as.
On the literal level of meaning, metaphor appears in its literal shape, which is simple juxtaposition. Ezra Pound, in explaining this aspect of metaphor, uses the illustrative figure of the Chinese ideogram, which expresses a complex image by throwing a group of elements together without predication. In Pound's famous black board example of such a metaphor, the two-line poem "In a Station of the Metro," the images of the faces in the crowd and the petals on the black bough are juxtaposed with no predicate of any kind connecting them. Predication belongs to assertion and descriptive meaning, not to the literal structure of poetry.
On the descriptive level we have the double perspective of the verbal structure and the phenomena to which it is related. Here meaning is "literal" in the common sense which we explained would not do for criticism, an unambiguous alignment of words and facts. Descriptively, then, all metaphors are similes. When we are writing ordinary discursive prose and use a metaphor, we are not asserting that A is B; we are "really" saying that A is in some respects comparable with B; and similarly when we are extracting the descriptive or paraphrasable meaning of a poem. "The hero was a lion," then, on the descriptive level, is a simile with the word "like" omitted for greater vividness, and to show more clearly that the analogy is only a hypothetical one. In Whitman's poem Out of the Cradle Endlessly Rocking, we find shadows "twining and twisting as if they were alive," and the moon swollen "as if with tears."  As there is no poetic reason why shadows should not be alive or the moon tearful, we may perhaps see in the cautious "as if" the working of a low mimetic discursive prose conscience.
On the formal level, where symbols are images or natural phenomena conceived as matter or content, the metaphor is an analogy of natural proportion. Literally, metaphor is juxtaposition; we say simply "A; B." Descriptively, we say "A is (like) B." But formally we say "A is as B." An analogy of proportion thus requires four terms, of which two have a common factor. Thus "the hero was a lion" means, as a form of expression which has nature for its internal content, that the hero is to human courage as the lion is to animal courage, courage being the factor common to the third and fourth terms.
Archetypally, where the symbol is an associative cluster, the metaphor unites two individual images, each of which is a specific representative of a class or genus. The rose in Dante's Paradiso and the rose in Yeats's early lyrics are identified with different things, but both stand for all roses all poetic roses, of course, not all botanical ones. Archetypal metaphor thus involves the use of what has been called the concrete universal, the individual identified with its class, Wordsworth's "tree of many one." Of course there are no real universals in poetry, only poetic ones. All four of these aspects of metaphor are recognized in Aristotle's discussion of metaphor in the Poetics, though sometimes very briefly and elliptically.
In the anagogic aspect of meaning, the radical form of metaphor, "A is B," comes into its own. Here we are dealing with poetry in its totality, in which the formula "A is B" may be hypothetically applied to anything, for there is no metaphor, not even "black is white," which a reader has any right to quarrel with in advance. The literary universe, therefore, is a universe in which everything is potentially identical with everything else. This does not mean that any two things in it are separate and very similar, like peas in a pod, or in the slangy and erroneous sense of the word in which we speak of identical twins. If twins were really identical they would be the same person. On the other hand, a grown man feels identical with himself at the age of seven, although the two manifestations of this identity, the man and the boy, have very little in common as regards similarity or likeness. In form, matter, personality, time, and space, man and boy are quite unlike. This is the only type of image I can think of that illustrates the process of identifying two  independent forms. All poetry, then, proceeds as though all poetic images were contained within a single universal body. Identity is the opposite of similarity or likeness, and total identity is not uniformity, still less monotony, but a unity of various things.
Finally, identification belongs not only to the structure of poetry, but to the structure of criticism as well, at least of commentary. Interpretation proceeds by metaphor as well as creation, and even more explicitly. When St. Paul interprets the story of Abraham's wives in Genesis, for instance, he says that Hagar "is" Mount Sinai in Arabia. Poetry, said Coleridge, is the identity of knowledge.
The universe of poetry, however, is a literary universe, and not a separate existential universe. Apocalypse means revelation, and when art becomes apocalyptic, it reveals. But it reveals only on its own terms, and in its own forms: it does not describe or represent a separate content of revelation. When poet and critic pass from the archetypal to the anagogic phase, they enter a phase of which only religion, or something as infinite in its range as religion, can possibly form an external goal. The poetic imagination, unless it disciplines itself in the particular way in which the imaginations of Hardy and Housman were disciplined, is apt to get claustrophobia when it is allowed to talk only about human nature and subhuman nature; and poets are happier as servants of religion than of politics, because the transcendental and apocalyptic perspective of religion comes as a tremendous emancipation of the imaginative mind. If men were compelled to make the melancholy choice between atheism and superstition, the scientist, as Bacon pointed out long ago, would be compelled to choose atheism, but the poet would be compelled to choose superstition, for even superstition, by its very confusion of values, gives his imagination more scope than a dogmatic denial of imaginative infinity does. But the loftiest religion, no less than the grossest superstition, comes to the poet, qua poet, only as the spirits came to Yeats, to give him metaphors for poetry.
The study of literature takes us toward seeing poetry as the imitation of infinite social action and infinite human thought, the mind of a man who is all men, the universal creative word which is all words. About this man and word we can, speaking as critics, say only one thing ontologically: we have no reason to suppose either that they exist or that they do not exist. We can call them divine if by divine we mean the unlimited or projected human.  But the critic, qua critic, has nothing to say for or against the affirmations that a religion makes out of these conceptions. If Christianity wishes to identify the infinite Word and Man of the literary universe with the Word of God, the person of Christ, the historical Jesus, the Bible or church dogma, these identifications may be accepted by any poet or critic without injury to his work the acceptance may even clarify and intensify his work, depending on his temperament and situation. But they can never be accepted by poetry as a whole, or by criticism as such. The literary critic, like the historian, is compelled to treat every religion in the same way that religions treat each other, as though it were a human hypothesis, whatever else he may in other contexts believe it to be. The discussion of the universal Word at the opening of the Chhandogya Upanishad (where it is symbolized by the sacred word "Aum") is exactly as relevant and as irrelevant to literary criticism as the discussion at the opening of the Fourth Gospel. Coleridge was right in thinking that the "Logos" was the goal of his work as a critic, but not right in thinking that his poetic Logos would so inevitably be absorbed into Christ as to make literary criticism a kind of natural theology.
The total Logos of criticism by itself can never become an object of faith or an ontological personality. The conception of a total Word is the postulate that there is such a thing as an order of words, and that the criticism which studies it makes, or could make, complete sense. Aristotle's Physics leads to the conception of an unmoved first mover at the circumference of the physical universe. This, in itself, means essentially that physics has a universe. The systematic study of motion would be impossible unless all phenomena of motion could be related to unifying principles, and those in their turn to a total unifying principle of movement which is not itself merely another phenomenon of motion. If theology identifies Aristotle's unmoved mover with a creating God, that is the business of theology; physics as physics will be unaffected by it. Christian critics may see their total Word as an analogy of Christ, as medieval critics did, but as literature itself may be accompanied in culture by any religion, criticism must detach itself accordingly. In short, the study of literature belongs to the "humanities," and the humanities, as their name indicates, can take only the human view of the superhuman.
The close resemblance between the conceptions of anagogic  criticism and those of religion has led many to assume that they can only be related by making one supreme and the other subordinate. Those who choose religion, like Coleridge, will, like him, try to make criticism a natural theology; those who choose culture, like Arnold, will try to reduce religion to objectified cultural myth. But for the purity of each the autonomy of each must be guaranteed. Culture interposes, between the ordinary and the religious life, a total vision of possibilities, and insists on its totality for whatever is excluded from culture by religion or state will get its revenge somehow. Thus culture's essential service to a religion is to destroy intellectual idolatry, the recurrent tendency in religion to replace the object of its worship with its present under standing and forms of approach to that object. Just as no argument in favor of a religious or political doctrine is of any value unless it is an intellectually honest argument, and so guarantees the autonomy of logic, so no religious or political myth is either valuable or valid unless it assumes the autonomy of culture, which may be provisionally defined as the total body of imaginative hypothesis in a society and its tradition. To defend the autonomy of culture in this sense seems to me the social task of the "intellectual" in the modern world: if so, to defend its subordination to a total synthesis of any kind, religious or political, would be the authentic form of the trahison des clercs.
Besides, it is of the essence of imaginative culture that it transcends the limits both of the naturally possible and of the morally acceptable. The argument that there is no room for poets in any human society which is an end in itself remains unanswerable even when the society is the people of God. For religion is also a social institution, and so far as it is one, it imposes limitations on the arts just as a Marxist or Platonic state would do. Christian theology is no less of a revolutionary dialectic, or indissoluble union of theory and social practice. Religions, in spite of their enlarged perspective, cannot as social institutions contain an art of unlimited hypothesis. The arts in their turn cannot help releasing the powerful acids of satire, realism, ribaldry, and fantasy in their attempt to dissolve all the existential concretions that get in their way. The artist often enough has to find that, as God says in Faust, he "muss als Teufel schaffen," which I suppose means rather more than that he has to work like the devil. Between  religion's "this is" and poetry's "but suppose this is," there must always be some kind of tension, until the possible and the actual meet at infinity. Nobody wants a poet in the perfect human state, and, as even the poets tell us, nobody but God himself can tolerate a poltergeist in the City of God.