10. Endings Across Forms

The model above was built for the novel. It transfers to adjacent forms with adjustments that need stating, because students routinely import novel-ending habits into forms that punish them — and because two of the forms below invert the model’s economics entirely.

10.1 The short story

The story’s ending is the most technically demanding in prose, and the reason is arithmetical. A novel’s ending can rely on four hundred pages of deposit; a story’s ending can rely on fifteen. The ratio of accumulated context to closural demand is the worst in the form, which is why so much of the twentieth century’s thinking about endings was done by story writers.

Four differences.

The dominant modern convention deserves naming, because students absorb it without noticing. The epiphanic close — a moment of unspoken recognition, usually rendered as an image, arriving in the last paragraph — descends from Joyce’s Dubliners and became, through the American story of the mid-twentieth century, the default. It is now in the position §3.7 describes for the ambiguous novel ending: no longer a discovery, and no longer self-justifying. A story that closes on a character looking out of a window at snow is not concluding; it is signalling that stories conclude this way.

The corrective is the same as for novels: the strategy must be chosen against alternatives (§6.5), not inherited.

10.2 The novella

The novella’s ending problem is the mirror of its opening problem (Chapter 9 §10.2). It has a novel’s tonal amplitude and a story’s word count, which means it can build a novel’s worth of pressure and has a story’s room in which to discharge it.

Two consequences.

• The novella is the natural home of the arrested ending (§6.4, no. 12) and of death as a structural terminus (§8.8). Both discharge maximum pressure in minimum space.
• Proportion is unforgiving. A three-page aftermath is under one per cent of a novel and five per cent of a novella. The §6.6 table’s percentages should be halved for the form.

The Death of Ivan Ilyich remains the model, and the reason is that its structure solves the proportion problem in advance: by announcing the death first and then retracing the life, Tolstoy converts the entire novella into an aftermath, so that the ending has nothing left to do but arrive.

10.3 Series, serials and the volume ending

The series inverts several of this chapter’s economics, and a writer working in one should know which.

The last row is the significant one and it produces a rule that many series writers discover too late:

In a series, the retroactive lens is not fired until the final volume. Every earlier volume’s ending is provisional, and a later book can retroactively ruin an ending that worked. This is the structural reason that series conclusions disappoint so reliably: they must close the accumulated accounts of every previous volume, at a length calibrated for one.

The serial — the work published in parts and read as a continuous whole — is a different case, and Chapter 6 (The Chapter Break) covers its instalment endings. What concerns this chapter is the pressure serialisation exerts on the final ending, described at §3.7: an audience that has waited eighteen months requires a conclusion legible as one, at speed, which historically produced the wind-up chapter. Contemporary episodic television is under the same pressure and produces the same result, with the same complaints.

10.4 The screen comparison

Three things transfer from screen endings to prose, and three do not.

Transfers:

• The proportion problem is identical, and screen has a vocabulary for it: the “false ending”, the sequence of scenes each of which could have closed the film. The complaint made about the endings of long fantasy films is the pattern of §8.3.
• Terminal modification works the same way. A film that has cut fast and cuts slowly at the last, or a series that drops its title sequence for a finale, is doing what a novel does when it alters a chapter shape.
• The retroactive reframe transfers completely, and screen exploits it more systematically than prose, because a visual medium can replay earlier images literally.

Does not transfer:

• Screen has involuntary closural signals that prose lacks. Credits roll; music resolves; the image fades. A film cannot fail test 7 (termination) in the way a novel can, because the medium supplies the signal regardless of the storytelling. Prose has no fade to black, which is why the formal domain (§4.6) has to be constructed rather than inherited.
• Duration is fixed and visible. A viewer with twelve minutes left knows the ending is coming; a reader with fifty pages left knows the same, and both facts shape expectation — but the film’s audience cannot skim, cannot stop, and cannot check. The reader’s awareness of the remaining page-block is a closural instrument no screen work possesses, and it is why an ending that arrives twenty pages before the physical end of a book is disquieting in a way no film can reproduce.
• Serial television’s economics reward non-closure in a way publishing does not, because renewal is decided after transmission. The result is a form structurally biased towards §8.11, and craft advice imported from it should be treated accordingly.

10.5 Translated fiction

The series’ standing distinction applies with particular force at endings: structural claims survive translation; texture claims do not. Three specific observations.

1. Last lines are the most translation-sensitive sentences in a book. Everything §4.5 says about grammatical closure — terminal stress, unqualified assertion, the last word — is a claim about a specific language’s prosody and syntax, and none of it survives. A translator who preserves the sense of a final sentence may lose the closural force entirely, and one who preserves the force may have to alter the sense. Compare the endings of The Stranger across Gilbert, Ward and Smith: the structural fact (Meursault arrives at an assent) is invariant; whether the assent sounds serene, defiant or flat differs measurably between them, and so does the reader’s final impression of the novel.

2. Word order determines what the last word can be. Languages with different default orders cannot place the same element last without distortion. A German or Latin sentence can end on a verb the English must move; a Japanese sentence’s final particle has no English equivalent. Any analysis that rests on “the last word of the novel is X” is, in translation, an analysis of the translator’s decision.

3. Apparatus is added and removed in translation. Afterwords, translator’s notes, glossaries and scholarly introductions accrete around translated novels, and they sit where a formal-closure device would sit (§4.6). A student analysing whether a novel ends with apparatus must check the original’s contents page. Chapter 10 §10.6 makes the same point about front matter, and the two cautions are the same caution.

The practical instruction for a seminar: when analysing a translated ending, state which claims are structural and which are textural, and — where the point is textural — name the translator in the sentence.

10.6 Interactive and multiple-ending forms

A brief note, because the form falsifies one of the chapter’s assumptions and it is worth seeing which.

Video games, hypertext fiction and branching narratives frequently offer several endings, selected by the reader’s choices. This is not the same as §6.4’s double ending (no. 15), which exhibits two endings to a reader who receives both; here the reader receives one and knows others exist.

The consequence is that the retroactive lens fires differently. A player who reaches an ending knows it is one of several, which means the ending cannot claim that this is what the story was — only what it became. Kermode’s concord fiction (§3.1) is weakened at the point where it does its work: a tock that could have been a different tock organises the interval less firmly.

Two observations follow, and both are useful for prose writers.

• Multiple endings tend to produce weaker closure and stronger authorship-feeling. The reader’s awareness of the branch converts the ending from a discovery about the material into a record of their own choices.
• The forms that solve this do so by making the multiplicity thematic — by being about choice, determinism or replay. This is the same solution §6.4 no. 15 requires of the double ending in prose, and the same solution §8.9 requires of ambiguity: a structural device that draws attention to itself must be about something.

10.7 Concordance: concept to work

A study aid. Each entry names the work whose ending most clearly displays the concept.