9. Worked Analyses

Nine close readings and a revision case. The analytical apparatus is the same throughout: identify the proposition and the payoff (§6.3), name the ending type (§4) and the placement strategy (§6.4), and ask what the break cost the reader.

The readings are weighted towards the serial tradition for the reason given at §3.3 — it is where the evidence is. The last four are dissidents: novels that use segmentation against its ordinary purposes, and which between them mark the limits of everything the chapter has argued.

9.1 Dickens, Bleak House (1852–53) — the number as the unit of design

Bleak House appeared in twenty monthly numbers, nineteen of them at a shilling and the last a double, between March 1852 and September 1853. Sixty-seven chapters were distributed across them, most numbers carrying three, some four, the last a larger group. Each number ran to thirty-two pages with two illustrations by Hablot Knight Browne. Dickens’s working notes — the “mems” — survive for the novel and are reproduced in the Clarendon edition and in Harry Stone’s Dickens’ Working Notes for His Novels (1987); John Butt and Kathleen Tillotson’s Dickens at Work (1957) remains the standard analysis of what they show.

What the notes demonstrate. Three things, all of them relevant to a novelist who will never publish serially.

1. Dickens planned in blocks, not in chapters. The design unit was the number: a topic list, a set of characters to be brought forward, and a decision about where in the block each effect would fall. The strongest material was reserved for the last chapter of the number, because that was the chapter that had to survive a month.
2. Withholding was scheduled explicitly. The notes contain deferrals — questions posed to himself and answered not yet — with the withheld item carried forward to a later number. Sternberg’s asset-management model of exposition (Chapter 5 §3.4) is visible here as an actual working practice.
3. Two narrators produced two ending grammars. The present-tense panoramic narrator closes chapters on systems, weather and moral address; Esther’s retrospective first person closes on personal discovery. The alternation gives the novel two different kinds of chapter ending, and readers learn to distinguish them long before they could describe the difference.

Two endings worth close attention.

Chapter 47, “Jo’s Will”, ends with the crossing-sweeper’s death and the narrator’s abandonment of the scene for direct address: Dead, your Majesty. Dead, my lords and gentlemen… And dying thus around us every day. The proposition of the chapter is a boy’s death; the payoff refuses to be about the boy. The reader is put outside the fiction, in their own country, holding an accusation. As an ending type it is an echo close whose echo is political (§4.4), and its placement is strategy 3 — just past the disaster, with the reaction handed to the reader rather than performed by a character.

Chapter 59 closes Esther’s pursuit of her mother at the burial-ground gate: And it was my mother, cold and dead. Seven monosyllables after a chapter of accelerating motion. This is a pivot in the strict sense — nothing is left open, but the whole preceding search is retrospectively reframed — and it is a demonstration of §5.2’s point about contraction. The chapter’s rhythm has been long and pursuing; the final clause is short and arrived. Nothing in the sentence is doing work except its brevity and its placement.

The transferable lesson. Dickens’s practice inverts the modern default. A contemporary novelist plans chapters and lets the larger shape emerge; Dickens planned the larger shape and let the chapters fall inside it. The result is a novel in which roughly one ending in three is structurally weighted and the rest are allowed to be ordinary — which is, arithmetically, the opposite of a manuscript in which every chapter ends on a hook. Reserving importance is a technique, and the serial format enforced it.

9.2 Dickens, Great Expectations (1860–61) — the weekly slot and the labelled Stage

Great Expectations was written under a different and much tighter constraint. It appeared weekly in All the Year Round from December 1860 to August 1861, in thirty-six instalments of one or two chapters each, and it was begun at short notice to rescue the magazine’s circulation. Fifty-nine chapters at a mean of roughly three thousand words is a very different rhythm from Bleak House’s five and a half thousand, and the difference is visible on every page: the novel is faster, tighter, less digressive, and has almost no room for the panoramic set-piece.

What the weekly slot did to the endings. Three consequences.

• Endings arrive twice as often, so they cannot all be large. Dickens’s solution is a much higher proportion of question closes and echo closes than in the monthly novels, with the heavy artillery reserved for the Stage boundaries.
• The chapters are internally shapelier. A weekly part must be satisfying on its own, which forces each chapter towards a single completed movement — the convention that would later become the twentieth-century norm.
• The largest boundaries had to be protected. Which brings us to the labelled Stage.

The Stage endings. The novel is divided into three Stages of Pip’s Expectations, and Dickens marks the first two with a sentence in the text: This is the end of the first stage of Pip’s expectations. The device repays attention, because it is a solution to a problem created by the format. In a weekly serial the reader receives a chapter ending every seven days; there is no typographic way to make one of those endings feel categorically larger, because the reader is holding a magazine and not a book. So Dickens says it. Genette’s intertitle function — the paratextual announcement of a structural division — is performed inside the narrative because the paratext is unavailable.

The first Stage’s close (Chapter 19) is dissected at §5.2. The second Stage closes on Pip’s realisation of what Magwitch’s money has made of him: the disaster has already occurred inside the chapter, and the ending is given over to its meaning rather than to its consequences. Placement strategy 3, again — past the disaster, into the beginning of reaction, and out.

A note for Chapter 16. The two endings of Great Expectations — the original and the revised — are the definitive case study in the language and belong to Chapter 16 (The Ending). The observation that belongs here is a structural one: a novel that has twice labelled its own internal boundaries has taught its reader to expect a marked conclusion, and the disagreement about the endings is partly a disagreement about which of them supplies the marking the book has trained the reader to want.

9.3 Dumas, The Count of Monte Cristo (1844–46) — the feuilleton and the daily gap

Le Comte de Monte-Cristo appeared in instalments in the Journal des Débats between August 1844 and January 1846. The roman-feuilleton — the serial printed at the foot of the daily newspaper — had been established in the previous decade and had already produced the era’s great popular successes; the format’s economics were direct, since circulation depended on the reader buying tomorrow’s paper.

The formal consequence. A hundred and seventeen chapters, short by the standards of the English monthly novel, ending with extreme regularity on turns: a name half-disclosed, a document produced, a question asked and unanswered, a visitor announced. Read consecutively in a modern edition — the standard English version is Robin Buss’s (Penguin, 1996) — the effect is exhausting, and readers routinely report that the novel is gripping and monotonous at the same time. Both reactions are correct, and the explanation is arithmetical rather than aesthetic.

A device deployed once a day for eighteen months, encountered instead forty times in an evening, has been administered at roughly a hundred times its designed dose.

What survives the change of format, and what does not. This is the reading’s real value, because it separates two things craft discussion usually confuses.

The structural claims survive; the tonal ones do not. This is the same distinction the series maintains about translation, applied to format rather than to language, and it is the reason this chapter treats delivery format as a first-order variable rather than as publishing history.

The teaching use. Set students the exercise of reading twenty consecutive chapters of Monte Cristo in one sitting and then twenty over twenty days. The difference in reported experience is large, reproducible, and impossible to obtain from any argument.

9.4 Eliot, Middlemarch (1871–72) — Books, mottoes, and a designed segmentation

Middlemarch was issued in eight Books, published at roughly two-month intervals across 1871 and 1872 in five-shilling paper-covered parts — a format devised for it rather than inherited, and different from the standard twenty-number shilling serial. The Books are titled: Miss Brooke; Old and Young; Waiting for Death; Three Love Problems; The Dead Hand; The Widow and the Wife; Two Temptations; Sunset and Sunrise. A Prelude opens the novel and a Finale closes it. Eighty-six chapters sit inside this frame, each headed by a motto.

Three levels of segmentation, doing three different jobs.

The Book titles repay attention as a set. Waiting for Death, The Dead Hand, Two Temptations — each names a condition rather than an event, which tells the reader that this is a novel in which situations are more important than incidents. Compare a set of Book titles naming events and the difference in promise is immediate. A Part title is the most compressed statement of intent available to a novelist, and Eliot’s are a masterclass in using them to declare a book’s subject rather than its plot.

The mottoes. Many are Eliot’s own, composed in a borrowed idiom and unattributed; others come from Shakespeare, Chaucer, Dante, Blake, Bunyan and elsewhere. Their characteristic operation is counterpoint (§8.11): a proposition is stated in general terms, and the chapter then supplies an instance that qualifies, complicates or embarrasses it. The reader is placed in the position of judging the fit, which is the activity the novel is training them in.

The ending types. Eliot’s chapters typically close on narratorial reflection — a lift out of the scene into a general observation. This is the soft landing in its most fully developed form, and it produces a specific reading experience: the reader leaves each chapter holding an idea rather than a question. The cost, which should be stated plainly, is that Middlemarch is not compulsive. The benefit is that it is one of the small number of long novels that readers finish and then re-read, and the two facts are related. A book that ends its chapters with thought rather than with threat is recruiting a different kind of attention, and it lasts longer.

Chapter 20’s closing movement — the passage about the roar on the other side of silence, and the wadding of stupidity that lets us live — is the clearest single instance of the mode. The chapter has narrated a young woman’s disillusion in Rome; it closes by generalising the disillusion into a proposition about the limits of human sympathy. Nothing is pending. The reader is released, holding an idea that will make the next four hundred pages legible.

9.5 Thackeray, Vanity Fair (1847–48) — the title as narratorial commentary

Vanity Fair appeared in twenty monthly numbers between January 1847 and July 1848, illustrated by Thackeray himself, and subtitled A Novel without a Hero. Its sixty-seven chapters are titled, and the titles are a sustained performance.

What Thackeray’s intertitles do. Genette’s category (§3.4) is realised here in its most active form. The titles are not summaries but interventions: they adopt a voice, take a position, and frequently mock the material they introduce.

The consequence for the reading experience is that every chapter is framed by a voice before it begins. The narrator of Vanity Fair is present in the table of contents. This is a formal decision with a thematic point: a novel that treats society as a puppet-show requires a showman, and the intertitles are where the showman stands when he is not narrating.

The Waterloo chapter. Chapter 32 is the novel’s best-known ending and one of the great uses of placement strategy 12 — the chapter-length silence. Thackeray declines to narrate the battle. The chapter stays in Brussels with the civilians, the rumours, the flight of Jos Sedley, and the guns audible in the distance. Its final sentence brings darkness down on the field and the city, and reports that Amelia was praying for George, who was lying on his face, dead, with a bullet through his heart.

Read as a chapter ending it does four things at once:

• Pivot — the withheld fact arrives last.
• Recasting close — Amelia’s prayer, narrated in the same sentence, is retrospectively made futile, and so is much of the chapter.
• Echo close — the final image is a body, given no commentary.
• Ellipsis — the battle itself falls into the white space after the full stop, and the novel never goes back for it.

Why the omission survives. Because it is foregrounded. Twenty pages of listening to guns have made the battle’s absence conspicuous, so the reader cannot mistake omission for oversight. This is the condition §6.4 places on strategy 12, and Vanity Fair is where it is met most completely.

9.6 Cervantes, Don Quixote (1605, 1615) — headings that summarise and mock

Don Quixote is the founding demonstration that a chapter’s apparatus can be part of its comedy. The headings promise, in the manner of the chivalric romances Cervantes is parodying, to relate what the chapter contains — and then relate it inaccurately, or grudgingly, or with editorial complaint. (English versions cited: John Rutherford, 2000; Edith Grossman, 2003. Structural claims below survive translation; the tone of individual headings does not, and students comparing versions will find the headings differ considerably in register.)

Three operations the headings perform.

1. They discharge suspense in advance, as a capitulum was designed to (§3.1) — and thereby announce that this is not a book whose pleasures are suspenseful.
2. They characterise a narrator who is not the author: an editor working from sources, with opinions about them.
3. They become, in Part II, self-referential. The later headings comment on their own function, on what the reader will see, and on the difficulty of naming a chapter’s contents — a joke that only works because the convention was still live.

The interrupted duel. The most consequential chapter break in the history of the novel occurs between Part I chapters 8 and 9. The battle with the Basque is suspended with the swords raised; the chapter ends. Chapter 9 opens not with the resumption but with the narrator explaining that his source broke off at this point, and describing how he came upon the continuation — an Arabic manuscript by Cide Hamete Benengeli, found by chance in the Alcaná of Toledo and translated for him.

The break is therefore doing three incompatible things simultaneously, which is why it is inexhaustible:

• It is a cliffhanger of the crudest kind, on an interrupted physical action.
• It is a false cliffhanger (§8.8), since the danger is dispelled at once and not by events.
• It is a recasting close of maximum reach: the resumption reveals that the entire narrative is a translation of a found document by a possibly unreliable Moorish historian, which changes the status of everything read so far and everything to come.

The teaching point. Cervantes converts the cheapest available ending into the novel’s deepest formal move, and the conversion works because the deflation is more interesting than the resolution would have been. That is the precise condition under which the false cliffhanger becomes legitimate, and it is a high bar.

9.7 Melville, Moby-Dick (1851) — segmentation as method

A hundred and thirty-five chapters and an epilogue, all titled, ranging from a few lines to several thousand words. No novel in English uses chapter division so variously, and the variety is the method rather than a symptom of it.

Four distinct segmentation strategies operate in the same book.

Cetology as the argument in miniature. Chapter 32 classifies whales by the format of books — Folio, Octavo, Duodecimo. A novel much concerned with whether the world can be systematised organises its central taxonomy by bibliographic size, in a chapter that is itself a unit of a book. The joke is precise and it is about this chapter’s subject: the categories we impose on the world are the categories of the technology we use to record it. Dames’s argument that the chapter is an inherited technology of segmentation has no better literary illustration.

What this does to the reader’s stopping behaviour. The uneven lengths mean the reader can never predict how long a chapter will take, and therefore cannot use the chapter as a unit of time budgeting (§3.2). Moby-Dick refuses to tell you whether you have nine minutes’ reading ahead or forty. Combined with the titles — which promise a topic rather than an event — the effect is a book in which the reader stops when they are tired rather than when the book releases them. Some readers experience this as freedom and some as difficulty, and both responses are responses to the same formal decision.

Against the proportion principle. By §5’s account, Moby-Dick’s chapter-length variance ought to read as noise. It does not, because the variance is correlated with kind: short chapters are dramatic or lyric, long ones are expository, and the reader learns the correlation by chapter twenty. Variance becomes rhythm when it is predictable in type even when it is unpredictable in length. That is a refinement §5 could not have arrived at without this novel, and it should be added to the principle.

9.8 Heller, Catch-22 (1961) — chapters that defeat sequence

Forty-two chapters, most of them named for a character: “The Texan”, “Doc Daneeka”, “Clevinger”, “Major Major Major Major”, “The Chaplain”, “Aarfy”, “Snowden”, “Yossarian”. A handful name places or things instead. Very few name events.

What the naming scheme does. It converts the novel’s segmentation from a temporal to an associative system. A chapter titled with a date or an event tells the reader when they are; a chapter titled with a person tells them whose orbit they are in. Since the novel’s chronology is deliberately scrambled — events are recounted out of order, and the same episode is returned to repeatedly with more of it disclosed each time — a sequence-based segmentation would have been fighting the material. The character-naming scheme instead reinforces it: the book is organised as a set of portraits, and the portraits can be arranged in any order.

The consequence for the ending types. Heller’s chapter endings are predominantly question closes and recasting closes, which is what an associative structure requires. Nothing much can be left physically pending, because the next chapter may be six months earlier; what can be left pending is a meaning. The recurring return to Snowden’s death — glimpsed, alluded to, partially told, and finally given in full in the chapter that bears his name — is the clearest large-scale case of curiosity rather than suspense as a novel’s engine (Sternberg; §6.2). The reader is not waiting to find out what happens. They are waiting to find out what already happened, and the chapters can therefore be as discontinuous as the satire requires.

The limit case for §6.3. Does a Catch-22 chapter have a proposition? Sometimes barely. Several chapters propose only here is a person, and pay off with and here is what is absurd about him. That is a proposition and a payoff, but of a kind the model was not built for, and it is worth conceding: the proposition–payoff mechanism describes narrative chapters well and portrait chapters only approximately. Where a novel’s unit is a person rather than a movement, the test to apply is not “what question did this raise?” but “what did I learn to see differently?”

9.9 Sterne, Tristram Shandy (1759–67) — the chapter as a joke about chapters

The last of the dissidents, and the one that puts the whole apparatus of this chapter in question, which is why it is placed here rather than earlier.

What Sterne does to the unit. Chapters run from a sentence to many pages. A chapter is announced as having been torn out, and the volume’s pagination skips accordingly. Two chapters of the ninth volume appear blank and are supplied later, out of place. The author’s preface arrives well into the book rather than before it. There is a chapter devoted to the theory of chapters, arguing for their necessity as a relief to the mind — an argument delivered in a work whose chapters relieve nothing.

The serious content of the joke. Three propositions, each of which this textbook has to answer.

1. A life cannot be divided into chapters. Tristram’s attempt to narrate his own existence fails because every event has antecedents and every antecedent has antecedents; the chapter, which requires a beginning and an end, is the wrong instrument for the job. This is an argument about the fit between narrative segmentation and lived experience, and it is not obviously wrong.
2. The chapter’s boundaries are the author’s convenience, not the story’s. Sterne makes this visible by placing them absurdly. Once seen, it cannot be unseen, and it is the same claim Dames makes historically (§3.1) two and a half centuries later.
3. The reader’s expectation of segmentation is a habit that can be exploited. Every one of Sterne’s formal outrages depends on the reader having internalised the convention it violates.

The answer this chapter gives. Not a refutation — Sterne is right — but a distinction. That the chapter is a convention rather than a natural joint does not make its placement arbitrary in practice, any more than the arbitrariness of the metre makes measurement useless. What follows from Sterne is not that placement does not matter but that it matters as a rhetorical decision rather than as a discovery, which is the position §16 argues for.

9.10 A revision case: where to end the chapter at the drawer

The nine readings above are of published fiction. What follows applies the diagnostic protocol to a draft, using the household that runs through this series: a small shop, a father dead in March, his ledger, three adult children, and a mother who changed the locks and will not say why.

The material, and where it was planted. The register drawer that sticks on the K and comes free on the third try is planted in Chapter 10 (The Prologue). The changed locks are planted in Chapter 3 (The Dialogue Exchange), in the doorway exchange where Dov gives Elena the key and withholds the reason. Elena’s decision to drive to the shop rather than telephone their mother is the closing beat of the sequel worked in Chapter 2 (Scene and Sequel). The hard cut into the drawer, and the smell of fresh ink, are Chapter 5 (The Transition)’s worked example. The ledger is first named in Chapter 4 (The Descriptive Detail), in a clause about the coat the father wore behind the counter over it; the father’s careful slanting hand with its crossed sevens, his entry of 9 March broken off halfway along its third line, and the change of handwriting beneath it are this chapter’s material.

The sibling at the drawer is Elena — the investigator, the one who wants to know why the locks were changed, the one who has had a key since she was twelve.

Two cuts are available for the same material.

Option A

The drawer came free on the third try, same as always, and the smell that met her wasn’t dust. It was ink, fresh ink, the sweetish smell of a biro, wrong for a till nobody had opened since March.

— chapter ends here —

Option B

The drawer came free on the third try, same as always, and the smell that met her wasn’t dust. It was ink, fresh ink, the sweetish smell of a biro, wrong for a till nobody had opened since March.

The ledger was in the drawer, where it had always lived. She lifted it out and set it on the counter, the leather faintly warm, as though someone had been leaning an arm on it an hour ago and not four months. It fell open near the back. Her father’s careful hand ran down the page, the same slant, the same crossed sevens, to an entry dated the ninth of March that stopped halfway along its third line. Under it someone had written the date again, and a count, in a looser hand that was not his, and the looser hand went on over the page. She did not read any of it. She sat down on the stool because her legs asked her to, turned back to the first page, eleven years back, and started from the beginning.

— chapter ends here —

The protocol applied

The verdict, and why it depends on the neighbours

Neither option is correct in the abstract, and a student who concludes that Option B is “better writing” has missed the point of the exercise. The right cut is determined by what the chapters on either side do, and in this book they are known.

What comes before: the doorway exchange (Chapter 3’s material), which closes on a withheld reason; then, in this chapter, Elena’s sequel in the truck, ending at a scene break on a decision she carries out immediately (Chapter 2), then the hard cut into the shop (Chapter 5). The reader has just been given a scene that ended on an unexecuted intention and a transition that opened on a disturbance. Two costly exits have been offered in quick succession.

What comes after: Dov’s thread — the sale, the deadline, the papers signed without telling his sisters (Chapter 14’s material).

This settles it, and the reasoning runs as follows.

1. Option A’s cliffhanger would be answered by adjacency. If the chapter ends on someone has been in this drawer and the next chapter opens in Dov’s thread, the placement itself supplies the answer. The reader does not need to be told; they will infer it from the cut. A cliffhanger whose next chapter names the culprit by juxtaposition is not a cliffhanger; it is a delayed caption.
2. Option B converts the same information into dramatic irony. With the changed hand already on the page, the cut to Dov is no longer a reveal but a view of the person the reader now suspects, going about his business. Everything he does in that chapter is read twice. This is a straight application of Sternberg’s distinction: A runs on suspense, which the following chapter deflates; B runs on curiosity, which the following chapter feeds.
3. The variety test is against A. Three consecutive unresolved exits is the beginning of cliffhanger fatigue (§8.2) in a novel whose register is domestic and whose whole effect depends on quietness. A book of this kind can spend an expensive exit perhaps five times, by this case’s estimate.
4. The re-entry cost is against A. Ending mid-discovery obliges the following chapter either to resume the scene — which forfeits the handoff to Dov and the whole subplot architecture — or to abandon it, which leaves the reader impatient through a chapter they should be reading closely.
5. The payoff map is for B. The series’ ending requires the changed handwriting to have been planted early and lived with for a long time, so that when all three siblings finally read the ledger the reader has been holding the fact for two hundred pages. Option A defers the plant by a chapter and reduces the time it has to work.

Verdict: Option B, on the grounds of what precedes and follows it — not on the grounds that a partial resolution is intrinsically better than a full withholding.

The condition under which A becomes right. If this chapter were the last chapter of Part One, A is the better cut, and by some distance. A Part boundary can carry an expensive exit that an ordinary chapter break cannot, because the reader is being told to expect a change and will accept a long-held question across it; and the next Part could then open in Dov’s thread without the juxtaposition reading as an answer, because a Part boundary licenses discontinuity. The same two sentences are a mistake at one structural position and the best available choice at another. That is the whole content of this chapter, in one comparison.

The substitution test, run once more

§6.5 requires the break to be tried a beat earlier and a beat later. Both are instructive.

• A beat earlier, ending on the leather faintly warm: an echo close. The chapter would end on a physical sensation with an unstated implication — someone’s arm, an hour ago — and no information at all. It is the most atmospheric of the three and the least useful, because the proposition would be raised twice and paid off never.
• A beat later, ending after Elena has read the first of the changed entries: the pivot becomes a disclosure, and the chapter acquires a second proposition it cannot pay off. This is the chapter that is really two (§8.7).

Three positions, three legitimate chapters, one right answer given the neighbours. The exercise is not about finding the good sentence; it is about discovering that the sentence was never the variable.

What Option B has not solved

Four things, and a workshop should say them.

• It resolves a question the novel might have wanted open longer. Once the reader knows a second hand continued the ledger, the field of suspects is small and domestic, and a good deal of the book’s remaining mystery is transferred from who to why. That is the right trade for this novel, whose subject is withholding rather than crime — but it is a trade, and a version of the book that wanted to sustain uncertainty about Dov for another hundred pages should choose A and accept the costs listed above.
• The last sentence is doing a great deal at once. She sat down on the stool because her legs asked her to, turned back to the first page, eleven years back, and started from the beginning carries a physical reaction, a characterisation, a decision and a structural handoff. It is close to the density ceiling that Chapter 1 (The Purposes of a Scene) §5 warns about, and read aloud it is slightly crowded.
• The restraint is convenient. She did not read any of it asks the reader to accept an investigator’s restraint at the moment she has most reason to look. It is motivated, since Elena means to know the ledger in order, from its first page; but it also protects the ending, in which the illness entries and the list near the end of the father’s pages must reach all three siblings at once (canon C10). The book does not return to Elena’s reading until the ledger has left her hands, which Chapter 14 (The Subplot) arranges. A sceptical reader may still see the author’s hand in her patience.
• The chapter’s proposition is inherited rather than raised. Elena arrives at the shop with a question given to her by the previous chapter. That is legitimate — propositions frequently carry across breaks — but it means this chapter cannot be assessed in isolation, and it is therefore a poor model for a stand-alone exercise. A chapter ending is not a portable skill. It is a decision about a position in a particular sequence.