Craft discussion of endings is even more prone to terminological drift than discussion of beginnings, because the everyday words are all in use for several things at once. “The ending” names, variously: the last sentence, the last chapter, the last movement, the resolution of the plot, and the reader’s feeling on closing the book. These are five different objects with five different failure modes, and a revision note that does not say which is meant is unusable. The chapter therefore begins by separating them.
2.1 The unit under study
Working definition. The ending is the final unit of a novel, distinguished from every other unit by two properties: it has no subsequent scene to correct, complicate or pay off what it delivers; and it is the material through which the reader retroactively re-reads everything that came before it. It comprises the climax and whatever aftermath the book requires in order to land.
Four elements of that definition carry weight.
- “Final unit.” Not the final sentence. The ending is a stretch of narrative, and in most novels it is longer than writers assume — commonly the last chapter or two, sometimes the last movement of the book, occasionally (as in Anna Karenina) an entire part.
- “No subsequent scene.” This is the constitutive property, and it is what makes the ending a kind of unit rather than merely a position. It is also why an epilogue does not escape the definition: an epilogue is not something after the ending, it is part of it.
- “Retroactively re-reads.” The ending is defined partly by an effect it has on other text. No other unit in this textbook is.
- “Climax and aftermath.” The composite structure is the source of most proportion errors, and §2.3.1 separates the two parts.
Note what the definition does not mention: resolution, catharsis, satisfaction, or completion. Those are things endings often do, and §4 anatomises them. They are not what an ending is. An ending that resolves nothing is still an ending, and it is still subject to every constraint in this chapter.
2.2 How long is an ending?
There is no principled answer, and any number offered is a convention. Four practical delimitations circulate, and — as with the opening (Chapter 9 §2.2) — they answer different questions.
| Delimitation | Extent | Answers the question |
|---|---|---|
| The last sentence or paragraph | 1–150 words | What is the reader left holding? (the final image, §4.5) |
| The last scene | ~800–3,000 words | Does the ending land as a scene, with its own value shift? |
| The last chapter or movement | ~3,000–12,000 words | Is the aftermath proportionate, and are the domains delivered? (§4, §6.6) |
| From the climax to the last word | Highly variable; 5–20% of the book | Is the structural shape of the close right? |
This chapter treats the third delimitation as primary — roughly the last chapter or movement — because that is the span across which the domains of §4 are actually discharged, and because it is the unit a revision decision usually operates on. Where a point applies specifically to the last line, or to the whole post-climactic stretch, the text says so.
A rule of thumb worth carrying, and the inverse of Chapter 9’s:
The opening’s contracts must be opened within two pages and honoured within the first chapter. The ending’s accounts must be closed within the last chapter and felt within the last page.
The asymmetry in that formulation is deliberate. An opening can afford to work slowly because the reader has agreed to keep reading. An ending cannot, because the reader is leaving.
2.3 The two confusions
Two distinctions do more work in this chapter than any others, and both are routinely collapsed in craft discussion.
2.3.1 Climax is not ending
The climax is the last major confrontation or disaster — the peak of conflict, in the sense Chapter 2 (Scene and Sequel) gives the term, the final disaster beat of the final goal–conflict–disaster sequence. The ending is the climax plus whatever aftermath the book needs.
The distinction matters because a book that stops at its climax and a book that lets its climax breathe into an aftermath are making two different structural choices, and choosing wrongly is one of the most common and most damaging errors in the repertoire (§8.2, §8.3). Aristotle already had the vocabulary: the desis, the tying of the knot, runs to the turning point; the lusis, the unravelling, runs from the turning point to the end. The lusis is not an appendix to the play. It is half the structure.
| Climax | Aftermath | |
|---|---|---|
| Function | Discharges the accumulated conflict | Establishes what the discharge means |
| Chapter 1’s terms | The largest value shift in the book | The reader’s opportunity to register that the value has shifted |
| Chapter 2’s terms | The final disaster | The final sequel — reaction, dilemma, decision — for the whole novel |
| Typical domains served | Plot, catharsis | Character, theme, the final image |
| Failure if omitted | The book has no engine | The book has no meaning; §8.2 |
| Failure if overlong | Exhaustion; diminishing returns on violence | Momentum death; §8.3 |
The most useful reframing here, because it turns a question of proportion into a structure the reader already knows, is to recognise the aftermath as the novel’s own sequel in Chapter 2’s technical sense. Every scene in that model is followed by a sequel in which the viewpoint character reacts, faces a dilemma and takes a decision; the aftermath of a novel is the same structure applied to the book. That is why an aftermath consisting only of reaction feels indulgent (a character grieving for thirty pages), and why an aftermath containing a decision — however small — feels complete. Nadia putting on the coat in §9.10 is a decision, not a reaction, and that is the whole reason it can end a novel.
2.3.2 Resolution is not closure
The second confusion is more consequential and much less often named.
Resolution is a property of the plot: the book’s dramatic questions are answered, or are legibly and deliberately left open. It is, in principle, checkable — one can list the questions and mark them.
Closure is a property of the reader: the felt sense that the emotional and thematic arc is complete, that the book has arrived rather than stopped. It is not checkable, only reportable.
These come apart in both directions, and all four combinations occur in published fiction. The 2×2 is the most useful diagnostic object in the chapter’s first half.
| Closure felt | Closure absent | |
|---|---|---|
| Resolved | The classical settlement. Jane Eyre; Great Expectations (revised); Emma. Every question answered and the answering feels like arrival | The hollow tidy ending. Every question answered and the reader feels nothing. This is the only one of the four that is a defect by default (§8.5) |
| Unresolved | The open close that closes. The Handmaid’s Tale; Blindness; Beloved; A Passage to India. Real ambiguity, complete feeling | The deliberate refusal. The Castle (unfinished, and read as though not); parts of the nouveau roman. Legitimate, difficult, and much rarer than drafts imagine |
Three inferences follow, and they are the operative content of the distinction.
- Resolution is neither necessary nor sufficient for closure. A book can answer every question and feel unfinished; it can leave the central question open and feel absolutely done. Writers who diagnose a flat ending as a resolution problem and respond by answering more questions usually make it worse.
- Only one of the four quadrants is a default failure. Resolved-without-closure is the hollow ending, and it is a failure because the reader can see the work being done and cannot feel it. The other three are choices.
- The mechanisms that produce closure are largely formal, not informational. Herrnstein Smith’s whole book is about this: closure is produced by the arrangement of the last lines, not by the quantity of information they carry (§3.2). This is why a novel can close on a sentence that answers nothing.
The clearest single illustration is A Passage to India (1924). Its final page answers nothing: Aziz and Fielding cannot be friends, and the earth and the sky say, in the novel’s closing words, “No, not yet” and “No, not there”. No plot question is resolved by that. The reader nonetheless experiences complete closure, because the refusal is delivered in the book’s own governing terms — the failure of connection, which is the novel’s subject — and because the sentence is formally terminal: a negation, in the mouths of the landscape, with nothing possible after it. Resolution declines and closure arrives, in the same clause.
2.4 Adjacent terms
| Term | Meaning | Relation to the ending |
|---|---|---|
| Climax | The peak of conflict; the last major confrontation or disaster | A part of the ending; §2.3.1 |
| Denouement (dénouement) | Literally “untying”; the unravelling of the complications after the climax | The French inheritance of Aristotle’s lusis; roughly synonymous with aftermath, with a stronger implication that facts are being disentangled |
| Desis / lusis | Aristotle’s tying and unravelling | The two-part structure underlying §2.3.1 |
| Resolution | The answering of the plot’s dramatic questions | A domain of the ending (§4.1), not the ending |
| Closure | The felt sense of completion | The effect an ending may or may not produce; §2.3.2 |
| Termination | Simply stopping | Herrnstein Smith’s contrast term; the thing an ending must not merely be (§3.2, §7.1 test 7) |
| Epilogue | A separated final unit, usually displaced in time, voice or vantage | The structural inverse of the prologue (Chapter 10); a strategy within the ending, not an alternative to it (§6.4, no. 7) |
| Coda | A short final movement in a different key, after the main structure is complete | Musical borrowing; often synonymous with a brief epilogue, but implies tonal rather than temporal displacement |
| Finale | A last movement gathering the whole cast or theme | Eliot’s own term for Middlemarch’s last section (§9.2) |
| Aftermath | Everything between climax and last word | This chapter’s preferred neutral term, because it presupposes no shape |
| Anticlimax | A close deliberately smaller than the expectation it follows | A strategy (§6.4, no. 11), not a defect, when chosen |
| Last line | The final sentence, considered as a citable unit | The most privileged position in the book after the first line (§3.6) |
| Apparatus | Appendices, notes, indexes and other paratext following the narrative | The vehicle of several formal-closure effects (§4.6); Genette’s territory |
| Series hook | A deliberate non-closure creating an appetite for a further volume (this edition keeps sequel for Chapter 2’s sense) | A commercial strategy with a structural cost (§8.11) |
Two of these deserve a note, because they are used loosely even in scholarship.
Denouement and aftermath are not quite synonyms. Denouement carries the implication that something is being untied — that facts previously entangled are being separated, mysteries explained, identities revealed. That is a nineteenth-century plot’s characteristic aftermath. A modern novel’s aftermath frequently unties nothing and consists entirely of consequence. Using denouement for both smuggles in an assumption about what endings are for.
Epilogue and coda are frequently conflated and should not be. An epilogue is displaced along an axis — time, voice, vantage — in the same way Chapter 10 (The Prologue) describes a prologue as displaced. A coda is not displaced at all; it is continuous, and merely changes key. The distinction matters in revision because the two have different costs: an epilogue must justify its exteriority, as a prologue must (Chapter 10 §1.2), while a coda need only justify its length.
2.5 What this chapter excludes
For clarity of scope, the following lie outside it:
- Openings — see Chapter 9 (The Opening Pages), the formal inverse, whose §4.6 formal contract this chapter’s §4.6 closes.
- Prologues, and the epilogue as a separated unit — Chapter 10 (The Prologue) treats separation as such, and its §8.9 and §9.8 deal directly with the epilogue that reframes. This chapter treats the epilogue only as one closural strategy among sixteen (§6.4).
- Chapter endings — see Chapter 6 (The Chapter Break), which treats the temporary stopping place. The relation is exact: a chapter break offers the reader a place to stop; the ending offers the place they must.
- The structure of the climactic scene itself — see Chapter 1 (The Purposes of a Scene) and Chapter 2 (Scene and Sequel). This chapter takes the climax as given and asks what must follow it.
- Subplot convergence — see Chapter 14 (The Subplot), whose account of convergence and of the attention tax supplies this chapter’s proportion test (§7.1, test 6).
- Series endings as a commercial problem — touched in §10.3; Chapter 17 (Series and Genre) treats the series instalment, and the economics of multi-volume publication are outside scope.
- The unfinished novel — The Mystery of Edwin Drood, The Castle, The Man Without Qualities. These raise real questions about whether closure can be attributed posthumously, and The Castle appears twice below (§2.3.2, §6.4) as a limit case, but a text that has no ending cannot be analysed as having a bad one.