Notes on Writing a Novel

Plot.Essential. The Pre-Essential. Plot might seem to be a matter of choice. It is not. The particular plot is something the novelist is driven to. It is what is left after the whittling-away of alternatives. The novelist is confronted, at a moment (or at what appears to be the moment: actually its extension may be indefinite) by the impossibility of saying what is to be said in any other way.

He is forced towards his plot. By what? By the ‘what is to be said.’ What is ‘what is to be said’? A mass of subjective matter that has accumulated—impressions received, feelings about experience, distorted results of ordinary observation, and something else—x. This matter is extra matter. It is superfluous to the non-writing life of the writer. It is luggage left in the hall between two journeys, as opposed to the perpetual furniture of rooms. It is destined to be elsewhere. It cannot move till its destination is known. Plot is the knowing of destination.

Plot is diction. Action of language, language of action.

Plot is story. It is also ‘a story’ in the nursery sense = lie. The novel lies, in saying that something happened that did not. It must, therefore, contain uncontradictable truth, to warrant the original lie.

Story involves action. Action towards an end not to be foreseen (by the reader) but also towards an end which, having been reached, must be seen to have been from the start inevitable.

Action by whom? The Characters (see      characters). Action in view of what, and because of what? The ‘what is to be said.’

What about the idea that the function of action is to express the characters? This is wrong. The characters are there to provide the action. Each character is created, and must only be so created, as to give his or her action (or rather, contributory part in the novel’s action) verisimilitude.

What about the idea that plot should be ingenious, complicated—a display of ingenuity remarkable enough to command attention? If more than such a display, what? Tension, or mystification towards tension, are good for emphasis. For their own sakes, bad.

Plot must further the novel towards its object. What object? The non-poetic statement of a poetic truth.

Have not all poetic truths been already stated? The essence of a poetic truth is that no statement of it can be final.

Plot, story, is in itself un-poetic. At best it can only be not anti-poetic. It cannot claim a single poetic license. It must be reasoned—onward from the moment when its none-otherness, its only-possibleness has become apparent. Novelists must always have one foot, sheer circumstantiality, to stand on, whatever the other foot may be doing. (N.B.—Much to be learnt from story-telling to children. Much to be learnt from the detective story—especially non-irrelevance. (See      relevance.) )

Flaubert’s ‘Il faut interésser.’ Stress on manner of telling: keep in mind, ‘I will a tale unfold.’ Interest of watching silk handkerchief drawn from conjuror’s watch.

Plot must not cease to move forward. (See      advance.) The actual speed of the movement must be even. Apparent variations in speed are good, necessary, but there must be no actual variations in speed. To obtain those apparent variations is part of the illusion-task of the novel. Variations in texture can be made to give the effect of variations in speed. Why are apparent variations in speed necessary? (a) For emphasis. (b) For non-resistance, or ‘give,’ to the nervous time-variations of the reader. Why is actual evenness, non-variation, of speed necessary? For the sake of internal evenness for its own sake. Perfection of evenness = perfection of control. The evenness of the speed should be the evenness inseparable from tautness. The tautness of the taut string is equal (or even) all along and at any part of the string’s length.


Are the characters, then, to be constructed to formula—the formula pre-decided by the plot? Are they to be drawn, cut out, jointed, wired, in order to be manipulated for the plot?

No. There is no question as to whether this would be right or wrong. It would be impossible. One cannot ‘make’ characters, only marionettes. The manipulated movement of the marionette is not the ‘action’ necessary for plot. Characterless action is not action at all, in the plot sense. It is the indivisibility of the act from the actor, and the inevitability of that act on the part of that actor, that gives action verisimilitude. Without that, action is without force or reason. Forceless, reasonless action disrupts plot. The term ‘creation of character’ (or characters) is misleading. Characters pre-exist. They are found. They reveal themselves slowly to the novelist’s perception—as might fellow-travellers seated opposite one in a very dimly-lit railway carriage.

The novelist’s perceptions of his characters take place in the course of the actual writing of the novel. To an extent, the novelist is in the same position as his reader. But his perceptions should be always just in advance.

The ideal way of presenting character is to invite perception.

In what do the characters pre-exist? I should say, in the mass of matter (see      plot) that had accumulated before the inception of the novel.

(N.B.—The unanswerability of the question, from an outsider: ‘Are the characters in your novel invented, or are they from real life?’ Obviously, neither is true. The outsider’s notion of ‘real life’ and the novelist’s are hopelessly apart.)

How, then, is the pre-existing character—with its own inner spring of action, its contrarieties—to be made to play a preassigned role? In relation to character, or characters, once these have been contemplated, plot must at once seem over-rigid, arbitrary.

What about the statement (in relation to PLOT) that ‘each character is created in order, and only in order, that he or she may supply the required action?’ To begin with, strike out ‘created.’ Better, the character is recognized (by the novelist) by the signs he or she gives of unique capacity to act in a certain way, which ‘certain way’ fulfils a need of the plot.

The character is there (in the novel) for the sake of the action he or she is to contribute to the plot. Yes. But also, he or she exists outside the action being contributed to the plot.

Without that existence of the character outside the (necessarily limited) action, the action itself would be invalid.

Action is the simplification (for story purposes) of complexity. For each one act, there are an x number of rejected alternatives. It is the palpable presence of the alternatives that gives action interest. Therefore, in each of the characters, while he or she is acting, the play and pull of alternatives must be felt. It is in being seen to be capable of alternatives that the character becomes, for the reader, valid.

Roughly, the action of a character should be unpredictable before it has been shown, inevitable when it has been shown. In the first half of a novel, the unpredictability should be the more striking. In the second half, the inevitability should be the more striking.

(Most exceptions to this are, however, masterpiece-novels. In War and Peace, L’Education Sentimentale and À La Recherche du Temps Perdu, unpredictability dominates up to the end.)

The character’s prominence in the novel (pre-decided by the plot) decides the character’s range—of alternatives. The novelist must allot (to the point of rationing) psychological space. The ‘hero,’ ‘heroine’ and ‘villain’ (if any) are, by agreement, allowed most range. They are entitled, for the portrayal of their alternatives, to time and space. Placing the characters in receding order to their importance to the plot, the number of their alternatives may seem to diminish. What E. M. Forster has called the ‘flat’ character has no alternatives at all.

The ideal novel is without ‘flat’ characters.

Characters must materialize—i.e., must have a palpable physical reality. They must be not only see-able (visualizable); they must be to be felt. Power to give physical reality is probably a matter of the extent and nature of the novelist’s physical sensibility, or susceptibility. In the main, English novelists are weak in this, as compared to French and Russians. Why?

Hopelessness of categoric ‘description.’ Why? Because this is static. Physical personality belongs to action: cannot be separated from it. Pictures must be in movement. Eyes, hands, stature, etc., must appear, and only appear, in play. Reaction to physical personality is part of action—love, or sexual passages, only more marked application of this general rule.

(Conrad an example of strong, non-sexual use of physical personality.)

The materialization (in the above sense) of the character for the novelist must be instantaneous. It happens. No effort of will—and obviously no effort of intellect—can induce it. The novelist can use a character that has not yet materialized. But the unmaterialized character represents an enemy pocket in an area that has been otherwise cleared. This cannot go on for long. It produces a halt in plot.

When the materialization has happened, the chapters written before it happened will almost certainly have to be recast. From the plot point of view, they will be found invalid.

Also, it is essential that for the reader the materialization of the character should begin early. I say begin, because for the reader it may, without harm, be gradual.

Is it from this failure, or tendency to fail, in materialization that the English novelist depends so much on engaging emotional sympathy for his characters?

Ruling sympathy out, a novel must contain at least one magnetic character. At least one character capable of keying the reader up, as though he (the reader) were in the presence of someone he is in love with. This is not a rule of salesmanship but a pre-essential of interest. The character must do to the reader what he has done to the novelist—magnetize towards himself perceptions, sense-impressions, desires.

The unfortunate case is, where the character has, obviously, acted magnetically upon the author, but fails to do so upon the reader.

There must be combustion. Plot depends for its movement on internal combustion.

Physically, characters are almost always copies, or composite copies. Traits, gestures, etc., are searched for in, and assembled from, the novelist’s memory. Or, a picture, a photograph, or the cinema screen may be drawn on. Nothing physical can be invented. (Invented physique stigmatizes the inferior novel.) Proust (in last volume) speaks of this assemblage of traits. Though much may be lifted from a specific person in ‘real life,’ no person in ‘real life’ could supply everything (physical) necessary for the character in the novel. No such person could have just that exact degree of physical intensity required for the character.

Greatness of characters is the measure of the unconscious greatness of the novelist’s vision. They are ‘true’ in so far as he is occupied with poetic truth. Their degrees in realness show the degrees of his concentration.


—Is a derivative of Plot. Gives actuality to Plot.

Nothing can happen nowhere. The locale of the happening always colours the happening, and often, to a degree, shapes it.

Plot having pre-decided what is to happen, scene, scenes, must be found, so chosen, as to give the happening the desired force.

Scene, being physical, is, like the physical traits of the characters, generally a copy, or a composite copy. It, too, is assembled—out of memories which, in the first place, may have had no rational connection with one another. Again, pictures, photographs, the screen are sources of supply. Also, dreams.

Almost anything drawn from ‘real life’—house, town, room, park, landscape—will almost certainly be found to require some distortion for the purposes of the plot. Remote memories, already distorted by the imagination, are most useful for the purposes of scene. Unfamiliar or once-seen places yield more than do familiar, often-seen places.

Wholly invented scene is as unsatisfactory (thin) as wholly invented physique for a character.

Scene, much more than character, is inside the novelist’s conscious power. More than any other constituent of the novel, it makes him conscious of his power.

This can be dangerous. The weak novelist is always, compensatorily, scene-minded. (Jane Austen’s economy of scene-painting, and her abstentions from it in what might be expected contexts, could in itself be proof of her mastery of the novel.)

Scene is only justified in the novel where it can be shown, or at least felt, to act upon action or character. In fact, where it has dramatic use.

Where not intended for dramatic use, scene is a sheer slower-down. Its staticness is a dead weight. It cannot make part of the plot’s movement by being show in play. (Thunderstorms, the sea, landscape flying past car or railway-carriage windows are not scene but happenings.)

The deadeningness of straight and prolonged ‘description’ is as apparent with regard to scene as it is with regard to character. Scene must be evoked. For its details relevance (See      relevance) is essential. Scene must, like the characters, not fail to materialize. In this it follows the same law—instantaneous for the novelist, gradual for the reader.

In ‘setting a scene’ the novelist directs, or attempts to direct, the reader’s visual imagination. He must allow for the fact that the reader’s memories will not correspond with his own. Or, at least, not at all far along the way.


Must (1) Further Plot. (2) Express Character.

Should not on any account be a vehicle for ideas for their own sake. Ideas only permissible where they provide a key to the character who expresses them.

Dialogue requires more art than does any other constituent of the novel. Art in the celare artem sense. Art in the trickery, self-justifying distortion sense. Why? Because dialogue must appear realistic without being so. Actual realism—the lifting, as it were, of passages from a stenographer’s take-down of a ‘real life’ conversation—would be disruptive. Of what? Of the illusion of the novel. In ‘real life’ everything is diluted; in the novel everything is condensed.

What are the realistic qualities to be imitated (or faked) in novel dialogue?—Spontaneity. Artless or hit-or-miss arrival at words used. Ambiguity (speaker not sure, himself, what he means). Effect of choking (as in engine): more to be said than can come through. Irrelevance. Allusiveness. Erraticness: unpredictable course. Repercussion.

What must novel dialogue, behind mask of these faked realistic qualities, really be and do? It must be pointed, intentional, relevant. It must crystallize situation. It must express character. It must advance plot.

During dialogue, the characters confront one another. The confrontation is in itself an occasion. Each one of these occasions, throughout the novel, is unique. Since the last confrontation, something has changed, advanced. What is being said is the effect of something that has happened; at the same time, what is being said is in itself something happening, which will in turn, leave its effect.

Dialogue is the ideal means of showing what is between the characters. It crystallizes relationships. It should, ideally, so be effective as to make analysis or explanation of the relationships between the characters unnecessary.

Short of a small range of physical acts—a fight, murder, love-making—dialogue is the most vigorous and visible interaction of which characters in a novel are capable. Speech is what the characters do to each other.

Dialogue provides means for the psychological materialization of the characters. It should short-circuit description of mental traits. Every sentence in dialogue should be descriptive of the character who is speaking. Idiom, tempo, and shape of each spoken sentence should be calculated by novelist, towards this descriptive end.

Dialogue is the first case of the novelist’s need for notation from real life. Remarks or turns of phrase indicatory of class, age, degree of intellectual pretension, idées reçues, nature and strength of governing fantasy, sexual temperament, persecution-sense or acumen (fortuitous arrival at general or poetic truth) should be collected. (N.B.—Proust, example of this semi-conscious notation and putting to use of it.)

All the above, from class to acumen, may already have been established, with regard to each character, by a direct statement by the novelist to the reader. It is still, however, the business of dialogue to show these factors, or qualities, in play.

There must be present in dialogue—i.e., in each sentence spoken by each character—either (a) calculation, or (b) involuntary self-revelation.

Every piece of dialogue must be ‘something happening.’ Dialogue may justify its presence by being ‘illustrative’—but the secondary use of it must be watched closely, challenged. Illustrativeness can be stretched too far. Like straight description, it then becomes static, a dead weight—halting the movement of the plot. The ‘amusing’ for its own sake, should above all be censored. So should infatuation with any idiom.

The functional use of dialogue for the plot must be the first thing in the novelist’s mind. Where functional usefulness cannot be established, dialogue must be left out.

What is this functional use? That of a bridge.

Dialogue is the thin bridge which must, from time to time, carry the entire weight of the novel. Two things to be kept in mind—(a) the bridge is there to permit advance, (b) the bridge must be strong enough for the weight.

Failure in any one piece of dialogue is a loss, at once to the continuity and the comprehensibility of the novel.

Characters should, on the whole, be under rather than over articulate. What they intend to say should be more evident, more striking (because of its greater inner importance to the plot) than what they arrive at saying.


The question of angle comes up twice over in the novel.
Angle has two senses—(a) visual, (b) moral.

(a) Visual Angle.—This has been much discussed—particularly I think by Henry James. Where is the camera-eye to be located? (1) In the breast or brow of one of the characters? This is, of course, simplifying and integrating. But it imposes on the novel the limitations of the ‘I’—whether the first person is explicitly used or not. Also, with regard to any matter that the specific character does not (cannot) know, it involves the novelist in long cumbrous passages of cogitation, speculation and guesses. E.g.—of any character other than the specific (or virtual) ‘I’ it must always be ‘he appeared to feel,’ ‘he could be seen to see,’ rather than ‘he felt,’ ‘he saw.’ (2) In the breast or brow of a succession of characters? This is better. It must, if used, involve very careful, considered division of the characters, by the novelist, in the seeing and the seen. Certain characters gain in importance and magnetism by being only seen: this makes them more romantic, fatal-seeming, sinister. In fact, no character in which these qualities are, for the plot, essential should be allowed to enter the seeing class. (3) In the breast or brow of omniscient story-teller (the novelist)? This, though appearing naïve, would appear best. The novelist should retain right of entry, at will, into any of the characters: their memories, sensations and thought-processes should remain his, to requisition for appropriate use. What conditions ‘appropriateness’? The demands of the plot. Even so, the novelist must not lost sight of point made above—the gain in necessary effect, for some characters, of their remaining seen—their remaining closed, apparently, even to the omniscience of the novelist.

The cinema, with its actual camera-work, is interesting study for the novelist. In a good film, the camera’s movement, angle and distance have all worked towards one thing—the fullest possible realization of the director’s idea, the completest possible surrounding of the subject. Any trick is justified if it adds a statement. With both film and novel, plot is the pre-imperative. The novelist’s relation to the novel is that of the director’s relation to the film. The cinema, cinema-going has no doubt built up in novelists a great authoritarianism. This seems to me good.

(b) Moral Angle.—This too often means, pre-assumptions—social, political, sexual, national, aesthetic, and so on. These may all exist, sunk at different depths, in the same novelist. Their existence cannot fail to be palpable; and their nature determines, more than anything else, the sympatheticness or antipatheticness of a given novel to a given circle of readers.

Pre-assumptions are bad. They limit the novel to a given circle of readers. They cause the novel to act immorally on that given circle. (The lady asking the librarian for a ‘nice’ novel to take home is, virtually, asking for a novel whose pre-assumptions will be identical with her own.) Outside the given circle, a novel’s pre-assumptions must invalidate it for all other readers. The increasingly bad smell of most pre-assumptions probably accounts for the growing prestige of the detective story: the detective story works on the single, and universally acceptable, pre-assumption that an act of violence is anti-social, and that the doer, in the name of injured society, must be traced.

Great novelists write without pre-assumption. They write from outside their own nationality, class or sex.

To write thus should be the ambition of any novelist who wishes to state poetic truth.

Does this mean he must have no angle, no moral view-point? No, surely. Without these, he would be (a) incapable of maintaining the conviction necessary for the novel; (b) incapable of lighting the characters, who to be seen at all must necessarily be seen in a moral light.

From what source, then, must the conviction come? and from what morality is to come the light to be cast on the characters?

The conviction must come from certainty of the validity of the truth the novel is to present. The ‘moral light’ has not, actually, a moral source; it is moral (morally powerful) according to the strength of its power of revelation. Revelation of what? The virtuousness or non-virtuousness of the action of the character. What is virtue in action? Truth in action. Truth by what ruling, in relation to what? Truth by the ruling of, and in relation to, the inherent poetic truth that the novel states.

The presence, and action, of the poetic truth is the motive (or motor) morality of the novel.

The direction of the action of the poetic truth provides—in fact, is—the moral angle of the novel. If he remains with that truth in view, the novelist has no option as to his angle.

The action, or continuous line of action, of a character is ‘bad’ in so far as it runs counter to, resists, or attempts to deny, the action of the poetic truth. It is predisposition towards such action that constitutes ‘badness’ in a character.

‘Good’ action, or ‘goodness’ in the character, from pre-disposition towards such action, is movement along with, expressive of and contributory to, the action of the poetic truth.

If the novelist’s moral angle is (a) decided by recognition of the poetic truth, and (b) maintained by the necessity of stating the truth by showing the truth’s action, it will be, as it should be, impersonal. It will be, and (from the ‘interest’ point of view) will be able to stand being, pure of pre-assumptions—national, social, sexual, etc.

(N.B.—‘Humour’ is the weak point in the front against pre-assumptions. Almost all English humour shows social (sometimes, now, backed by political) pre-assumptions. (Extreme cases—that the lower, or employed, classes are quaint or funny—that aristocrats, served by butlers, are absurd. National pre-assumptions show in treatment of foreigners.)


It has been said that plot must advance; that the underlying (or inner) speed of the advance must be even. How is this arrived at?

(1) Obviously, first, by the succession, the succeedingness, of events or happenings. It is to be remembered that everything put on record at all—an image, a word spoken, an interior movement of thought or feeling on the part of a character—is an event or happening. These proceed out of one another, give birth to one another, in a continuity that must be (a) obvious, (b) unbroken.

(2) Every happening cannot be described, stated. The reader must be made to feel that what has not been described or stated has, none the less, happened. How? By the showing of subsequent events or happenings whose source could only have been in what has not actually been stated. Tuesday is Tuesday by virtue of being the day following Monday. The stated Tuesday must be shown as a derivative of the unstated Monday.

(3) For the sake of emphasis, time must be falsified. But the novelist’s consciousness of the subjective, arbitrary and emotional nature of the falsification should be evident to the reader. Against this falsification—in fact, increasing the force of its effect by contrast—a clock should be heard always impassively ticking away at the same speed. The passage of time, and its demarcation, should be a factor in plot. The either concentration or even or uneven spacing-out of events along time is important.

The statement ‘Ten years has passed,’ or the statement ‘It was now the next day’—each of these is an event.

(4) Characters most of all promote, by showing, the advance of the plot. How? By the advances, from act to act, in their action. By their showing (by emotional or physical changes) the effects both of action and of the passage of time. The diminution of the character’s alternatives shows (because it is the work of) advance—by the end of a novel the character’s alternatives, many at the beginning, have been reduced to almost none. In the novel, everything that happens happens either to or because of one of the characters. By the end of the novel, the character has, like the silk worm at work on the cocoon, spun itself out. Completed action is marked by the exhaustion (from one point of view) of the character. Throughout the novel, each character is expending potentiality. This expense of potentiality must be felt.

(5) Scene promotes, or contributes to, advance by its freshness. Generically, it is fresh, striking, from being unlike the scene before. It is the new ‘here and now.’ Once a scene ceases to offer freshness, it is a point-blank enemy to advance. Frequent change of scene not being an imperative of the novel—in fact, many novels by choice, and by wise choice, limiting themselves severely in this matter—how is there to continue to be freshness? By means of ever-differing presentation. Differing because of what? Season of year, time of day, effects of a happening (e.g., with house, rise or fall in family fortunes, an arrival, a departure, a death), beholding character’s mood. At the first presentation, the scene has freshness; afterwards, the freshness must be in the presentation. The same scene can, by means of a series of presentations, each having freshness, be made to ripen, mature, to actually advance. The static properties in scene can be good for advance when so stressed as to show advance by contrast—advance on the part of the characters. Striking ‘unchangingness’ gives useful emphasis to change. Change should not be a factor, at once, in both scene and character; either unchanged character should see, or be seen against, changed scene, or changed character should see, or be seen, against unchanged scene. Two changes obviously cancel each other out, and would cancel each other’s contribution to the advance of plot.


Relevance—the question of it—is the headache of novel-writing.

As has been said, the model for relevance is the well-constructed detective story: nothing is ‘in’ that does not tell. But the detective story is, or would appear to be, simplified by having fact as its kernel. The detective story makes towards concrete truth; the novel makes towards abstract truth.

With the detective story, the question ‘relevant to what?’ can be answered by the intelligence. With the novel, the same question must constantly, and in every context, be referred to the intuition. The intelligence, in a subsequent check over, may detect, but cannot itself put right, blunders, lapses or false starts on the part of the intuition.

In the notes on Plot, Character, Scene and Dialogue, everything has come to turn, by the end, on relevance. It is seen that all other relevances are subsidiary to the relevance of the plot—i.e., the relevance to itself that the plot demands. It is as contributory, in fact relevant, to plot that character, scene and dialogue are examined. To be perfectly contributory, these three must be perfectly relevant. If character, scene or dialogue has been weakened by anything irrelevant to itself, it can only be imperfectly relevant—which must mean, to a degree disruptive—to the plot.

The main hope for character (for each character) is that it should be magnetic—i.e., that it should attract its parts. This living propensity of the character to assemble itself, to integrate itself, to make itself in order to be itself will not, obviously, be resisted by the novelist. The magnetic, or magnetizing, character can be trusted as to what is relevant to itself. The trouble comes when what is relevant to the character is found to be not relevant to the plot. At this point, the novelist must adjudicate. It is possible that the character may be right; it is possible that there may be some flaw in the novelist’s sense of what is relevant to the plot.

Again, the character may, in fact must, decide one half of the question of relevance in dialogue. The character attracts to itself the right, in fact the only possible, idiom, tempo and phraseology for that particular character in speech. In so far as dialogue is illustrative, the character’s, or characters’, pull on it must not be resisted.

But in so far as dialogue must be ‘something happening’—part of action, a means of advancing plot—the other half of the question of dialogue-relevance comes up. Here, the pull from the characters may conflict with the pull from the plot. Here again the novelist must adjudicate. The recasting and recasting of dialogue that is so often necessary is, probably, the search for ideal compromise.

Relevance in scene is more straightforward. Chiefly, the novelist must control his infatuation with his own visual power. No non-contributory image, must be the rule. Contributory to what? To the mood of the ‘now,’ the mood that either projects or reflects action. It is a good main rule that objects—chairs, trees, glasses, mountains, cushions—introduced into the novel should be stage-properties, necessary for ‘business.’ It will be also recalled that the well-set stage shows many objects not actually necessary for ‘business,’ but that these have a right to place by being descriptive—explanatory. In a play, the absence of the narrating voice makes it necessary to establish the class, period and general psychology of the characters by means of objects that can be seen. In the novel, such putting of objects to a descriptive (explanatory) use is excellent—alternative to the narrator’s voice.

In scene then, relevance demands either usefulness for action or else explanatory power in what is shown. There is no doubt that with some writers (Balzac, sometimes Arnold Bennett) categoricalness, in the presentation of scene, is effective. The aim is, usually, to suggest, by multiplication and exactitude of detail, either a scene’s material oppressiveness or its intrinsic authority. But in general, for the purposes of most novelists, the number of objects genuinely necessary for explanation will be found to be very small.

Irrelevance, in any part, is a cloud and a drag on, a weakener of, the novel. It dilutes meaning. Relevance crystallizes meaning.

The novelist’s—any writer’s—object is, to whittle down his meaning to the exactest and finest possible point. What, of course, is fatal is when he does not know what he does mean: he has no point to sharpen.

Much irrelevance is introduced into novels by the writer’s vague hope that at least some of this may turn out to be relevant, after all. A good deal of what might be called provisional writing goes to the first drafts of first chapters of most novels. At a point in the novel’s progress, relevance becomes clearer. The provisional chapters are then recast.

The most striking fault in work by young or beginning novelists, submitted for criticism, is irrelevance—due either to infatuation or indecision. To direct such an author’s attention to the imperative of relevance is certainly the most useful—and possibly the only—help that can be given.

Orion II, 1945.

Copyright 1950 by Elizabeth Bowen. Used by permission of Curtis Brown, Ltd. All rights reserved.