Laurence Perrine,
"The Nature of Proof in the Interpretation of Poetry

To what extent can teachers demand "correct" interpretations of poetry? Are there no incorrect readings? Professor Perrine, Southern Methodist University, presents well-illustrated answers to these crucial questions.

That a poem may have varying interpretations is a critical commonplace. That all interpretations of a poem are equally valid is a critical heresy, but one which perennially makes its reappearance in the classroom. “Why can't a poem mean anything that a reader sees in it?” asks the student. “Why isn't one person's interpretation of a poem as good as anyone else's?” According to his theory the poem is like an ink blot in a Rorschach personality test. There are no correct or incorrect readings: there are only readings which differ more or less widely from a statistical norm.

This theory is one that poets themselves have sometimes seemed to lend support to. T.S. Eliot, in response to conjectures about the meanings of his poems, has replied, "If it suits you that way, then that is all right with me." Yeats once wrote to a friend: "I shall not trouble to make the meaning clear--a clear vivid story of a strange sort is enough. The meaning may be different with everyone." But one is not really quarreling with Eliot or Yeats in challenging this point of view. Eliot, as a critic dealing with the poetry of others, has been constantly concerned with determining precise meanings. No poet, however, likes to be caught in the predicament of having to explain his own poems. He cannot say, "What I really meant was . . ." without admitting failure, or without saying something different (and usually much less) than what his poem said. And in doing so, he gives this diminished reading the stamp of his own authority. "A writer," E. A. Robinson once told an interviewer, "should not be his own interpreter." It is significant that Yeats was quite willing to write, for an anthology, a comment on one of his poems so long as the comment did not appear over his own name. "If an author interprets a poem of his own," he explained to the editor, "he limits its suggestibility." The poet is eager to be understood. But where as the comments of a critic may raise the curtain on a reader's understanding of a poem, the poet's own comments drop the curtain. We must therefore not mistake the defensive gestures of a poet like Yeats or Eliot for a declaration of his critical theory.

In this paper, accordingly, I wish not to advance any new propositions, but only to re-assert the accepted principal that for any given poem there are correct and incorrect readings, and to illustrate the process which the correctness of a reading may be proved or disproved. For logical proof, though not experimental proof, is at least as possible in the interpretation of poetry as it is, say in a court of law.

The criteria used for judging any interpretation of a poem are two: (1) A correct interpretation, if the poem is a successful one, must be able to account satisfactorily for any detail of the poem. If it is contradicted by any detail it is wrong. Of several interpretations, the best is that which most fully explain the details of the poem without itself being contradicted by any detail. (2) If more that one interpretation satisfactorily accounts for all the details of the poem, the best is that which is most economical, i.e. which relies on the fewest assumptions not grounded in the poem itself. Thomas Huxley illustrates this principle of judgment in a different area in one of his essays. If, he says, on coming downstairs in the morning we find our silverware missing, the window open, the mark of a dirty hand on the window frame, and the impress of a hobnailed boot on the gravel outside, we logically conclude that the silverware has been stolen by a human thief. It is possible, of course, that the silverware was taken by a monkey and that a man with dirty hands and hobnailed boots looked in the window afterwards; but this explanation is far less probable, for, though it too accounts for all the facts, it rests on too many additional assumptions. It is, as we would say, too "far-fetched."

These two criteria, I ask you to notice, are not different from those we bring to the judgment of a new scientific hypothesis. Of such we ask (I) that it satisfactorily account for as many as possible of the known facts without being contradicted by any fact, (2) that it be the simplest or most economical of alternative ways of accounting for these facts.

Problems in Interpretation

Let me illustrate by presenting two problems in interpretation. The first is an untitled poem by Emily Dickinson:

Where ships of purple gently toss
On seas of daffodil,
Fantastic sailors mingle,
And then-the wharf is still

The second consists of a pair of poems, one by Walt Whitman, the other by Herman Melville. The poem by Whitman appeared in his volume of Civil War poems, Drum-Taps. Melville, who was Whitman's almost exact contemporary, also published a book of war poems (Battle-Pieces), though the following poem did not appear in it. I ask you with the Dickinson poem merely to decide what it is about; with the Whitman and Melville poems, to determine the principal difference between them. (See page 395.)

Several years ago I presented the Emily Dickinson poem to a number of students and colleagues and discovered that not one of them interpreted the poem as I did. Almost universally they read the poem as being descriptive of a scene in a garden or meadow. A consensus of their interpretations runs as follows:

Tall purple flowers (iris?) stand above the daffodils and are tossed in the breeze. Bees and butterflies ("fantastic sailors") mingle with the flowers. The wind stops, and then the garden is still,

Beside this let me place the interpretation which I hope to prove the correct one:

The poem is a description of a sunset. The "ships of purple" are clouds. The "seas of daffodil" are skies colored golden by the setting sun. The "fantastic sailors" are the old-fashioned seamen dressed in gorgeous garments of many colors brought from exotic lands. The sun sinks and the wharf (the earth where the sun set-the scene of this colorful activity is still.

How do we demonstrate the “sunset” reading to be correct and the “garden” reading to be incorrect? By some such argument as this:

“Ships of purple” is ore apt metaphor for clouds than for flowers, both as to size and to motion (we often speak of clouds as “sailing”). “Daffodill” would normally be in the plural if it referred to flowers rather than to colors: why would not the poet say “On a sea of daffodils”? “Mingle” fits better the intertwining colors of the sunset than it does the behavior of bees, which mingle with flowers perhaps but not, except in the hive, with each other (and the flowers here are “seas”). The “garden” reading provides no literal meaning for the “wharf.” The “garden” reading, to explain why the wharf becomes “still”, demands the additional assumption that the wind stops (why should it? and would the bees and butterflies stop their activity if it did?); the disappearance of the sun, in contrast, is inevitable and implicit in the sunset image. Finally, the luxuriance of imagination manifested in the poem is the more natural consequence of looking at clouds and sunset sky than at flowers. We look at clouds and see all sorts of things-ships, castles, animals, landscape-but it takes some straining to conjure up a scene such as this one from a garden. The "garden" reading is therefore incorrect because it fails to account for some details in the poem (the wharf), because it is contradicted by some details (the singular use of “daffodil”), because it explains some details less satisfactorily than the “sunset” reading (“ships of purple,” “mingle”), and finally because it rests on assumptions not grounded in the poem itself (the wind stops). The “sunset” reading explains all these details satisfactorily.



With its cloud of skirmishers in advance,
With now the sound of a single shot snapping like a whip, and now
an irregular volley.
The swarming ranks press on and on, the dense brigades press on,
Glittering dimly, toiling under the sun-the dust-cover'd men,
In columns rise and fall to the undulations of the ground,
With artillery interspers'd-the wheels rumble, the horses sweat,
As the army corps advances.


With banners furled, and clarions mute,
An army passes in the night;
And beaming spears and helms salute
The dark with bright.

In silence deep the legions stream,
With open ranks, in order true;
Over boundless plains they stream and gleam
No chief in view!

Afar, in twinkling distance lost,
(So legends tell) he
lonely wends
And back through all that shining host
His mandate sends.

Ordinarily we have only the internal evidence of the poem itself on which to rest an interpretation. In this instance, as I discovered some time after the incident related, there is external proof also of the "sunset" reading. The poem was first published in 1891 under the title "Sunset." Though this title was editorially supplied by T. W. Higginson after Emily's death, its correctness is established by two other poems in which the poet uses substantially the same imagery (yellow and purple, sea and ships). One poem itself contains the word "sunset"; the other was entitled "Sunset" by the poet in a letter to a friend.

The Whitman-Melville problem I presented more recently as a theme assignment to an Honor section in freshman English. Again I received not a single correct solution. I should confess at the outset, however, that I am guilty of having planted a false clue. The false clue lies in the information that Melville wrote a book of poems about the Civil War-perfectly true, of course, but totally irrelevant. This poem is not about the Civil War, as is manifest from "spears and helms"-items not stocked by Civil War quartermasters. More important, this poem is not about war at all. The main difference between this poem and Whitman's is that Whitman's is literal, Melville's metaphorical. Whitman's is about an army corps on the march, Melville's is about the stars.

My freshmen immediately identified subject matter when I wrote the Melville poem n the board and circled words: "beaming," "bright," "gleam," "twinkling," "shining." The five words together form a constellation whose reference, once the pattern is recognized, is almost immediately clear. That "twinkling" modifies "distance" and that "shining" modifies "host" provides additional confirmation. The phrase "host of heaven" is used extensively for stars in the Bible.

From this starting point the proof proceeds with logical rigor: (1) The close repetition of "beaming," "bright," "gleam," "twinkling," and "shining" immediately suggests stars. (2) The setting is night. (3) The poem emphasizes the silence of the procession, which moves "in silence deep" and "with clarions mute." (No actual army, of course, no matter how secret its movements, is ever quite so stealthy. In Whitman's poem "the wheels rumble," as indeed wheels do.) (4) The poem also emphasizes the idea of infinite space: the army marches "over boundless plains"; its leader is "Afar, in twinkling distance lost." (5) The army marches "With open ranks, in order true"-a formation more star-like than military. No actual legions ever "stream" in perfect order; but the stars keep an eternally fixed but open relation to each other. (6) Finally, no commander of this army is in view -a situation especially unusual in an army proceeding in perfect order. Indeed, the "army" interpretation cannot explain this detail without assumptions grounded outside the poem.

The real difficulty of interpreting the Melville poem comes, of course, at this point, for the Melville poem is not simple descriptive, as Whiteman's is, but philosophical. As I read it, the poem poses the question of the existence of God. No God is observable in the heavens (which are silent), yet the start follow an “order true,” and legends (e.g. the Bible) tell us that God orders them. These stories, however, are indeed “legends,” i.e., they are of doubtful authenticity; and even if they be true, the God they speak of is “Afar, in twinkling distance lost,” not in daily confrontation of man or nature. One hundred years earlier a poet writing on this theme would have declared without hesitation that "The heavens declare the glory of God, the firmament showeth his handiwork"; Melville ends his poem with a question or a doubt. In the nineteenth century the argument from design had been shattered.

If a poem, then, does have a determinable meaning-if, in the interpretation of poetry, we can't say that “anything goes”-why does the opposite theory so often arise? It is because of some false analogy drawn with music or abstract art? Perhaps. But, first of all it arises because within limits there is truth in it. A poem-in fact, any pattern of words-defines an area of meaning, no more. Any interpretation is acceptable which lies within that area. The word “horse” may justifiably call up in a reader's mind the image of a black, a roan, or a white horse; a stallion, a mare, or a gelding; even a wooden sawhorse, a human “work horse,” or a female “clothes horse.” But as soon as the word is combined with another, say “roan,” the area of meaning is drastically reduced. It can still be stallion, mare, or gelding; but it cannot be a white or black horse, a sawhorse, “work horse,” or “clothes horse.” Further expansions of the context limit the meanings still further. But even without context the word cannot mean cow.

In poetry, context may function to expand meaning as well as to limit it. Words in poetry thus have richer meanings than in prose-they may exhibit purposeful ambiguities-but the meanings are still confined to a certain area. With a poem like Whitman's that area is fairly narrowly circumscribed. The reader may legitimately see a Northern or Southern army (if he knows nothing of Whitman's life); in fact, if the poem is removed from its context in DrumTaps he may legitimately see a Revolutionary War army, but in no case may he interpret the poem as being about stars.

The Problem of Symbols

The areas of greater meaning are created by symbolical poems. “A symbol,” writes John Ciardi, “is like a rock dropped into a pool; it sends out ripples in all directions, and the ripples are in motion. Who can say where the last ripple disappears?” True, But even a symbol does not have unlimited meaning. The pool in which the rock is dropped has borders. A symbol in literature is made up of words which, by the way they are used, have acquired a sometimes tremendously increased area of meaning. To switch from Ciardi's figure, we may envision such a symbol as a powerful beam of light flashed out into the darkness by a searchlight from a point on earth. The cone of light is the area of meaning. Its point is precise and easily located. But its base fades out into the atmosphere. Its meanings are therefore almost infinite. But they are not unlimited. They must be found, at whatever distance from the apex, within the circumference of the cone.

By the very nature of the case the process of proof or demonstration with symbolic literature is more difficult than with nonsymbolic, just as complex logical problems are more difficult than the simple ones by which logicians demonstrate their principles. Scholars will continue to debate the meanings of the “white whale” in Moby-Dick for years to come. Their argument, however, has meaning. Some interpretations do make more sense than others. More than one meaning may be valid, but not just any meaning can be. The white whale is not an ink blot, not even a white ink blot.

Let me illustrate with a poem by William Blake:


O Rose, thou art sick!
The invisible worm
That flies in the night,
In the howling storm,

Has found out thy bed
Of crimson joy,
And his dark secret love
Does thy life destroy.

The essential difference between a metaphor and a literary symbol is that a metaphor means something else than what it is, a literary symbol, means something more than what it is. In the word of Robert Penn Warren, a symbol "partakes of the reality which it renders intelligible"; in the words of E. E. Stoll, a symbol "means what it says and another thing besides." if we use I. A. Richards' terms vehicle and tenor for the two things equated by a metaphor, we must say that with a symbol the vehicle is part of the tenor. The vehicle in this case is not like one of those long trucks we see on the highways carrying automobiles from manufacturer to dealer; it is more like a new automobile filled with presents at Christmas time in which the automobile is part of the gift. Melville's "Night-March” is not really about an army at all; Blake's poem is about a rose and a cankerworm.

But Blake's poem is so richly organized that the rose and the worm refuse to remain a rose and a worm. The phrase “dark secret love” is too strong to be confined to the feeding of the worm on the rose; “thy bed of crimson joy” suggests much more than the rose bed which it literally denotes. The powerful connotation of these phrases, added to those of “sick,” “invisible,” “night,” and “howling storm,” and combined with the capitalization of “Rose” and the personification of the flower, force the reader to seek for additional meanings. Almost immediately the Rose suggest a maiden and the worm her secret lover; but these meanings in turn suggest still broader meanings as the cone of light broadens. The poem has been read by different readers as referring to the destruction of joyous physical love by jealousy, deceit, concealment, or the possessive instinct; of innocence by experience; of humanity by Satan; of imagination and joy by analytic reason; of life by death. Some of these meanings are suggested entirely by the poem itself, some by a knowledge also of Blake's other writings.

It is not my purpose here to make a detailed examination of these interpretations in the light of my two criteria. My belief is that a case can be made for all of them; that the symbols allow them all; that we are not forced to choose between the two interpretations of the Dickinson poem or the one by Melville. But if the rose can mean love, innocence, humanity, imagination, and life; and if the worm can mean the flesh, jealousy, deceit, concealment, possessiveness, experience, Satan, rationalism, death (and more), can the two symbols therefore mean just anything? The answer is No. The rose must always represent something beautiful or desirable or good. The worm must always be some kind of corrupting agent of youth. Both symbols define an area of meaning and a viable interpretation must fall within that area. Blake's poem is not about the elimination of social injustice enlightened society; it is not about the eradication of sin by God; it is not about the triumph of freedom over tyranny. Any correct interpretation must satisfactorily explain the details of the poem without being contradicted by any detail; the best interpretations will reply on the fewest assumptions not grounded in the poem itself.

A rose is a rose is a rose, and is more than a rose. But a rose is not an ink blot. Nor is a poem.

from The English Journal (September 1962; pp.393-98)