The household novel that runs through this book has reached the end of a chapter, and the writer has to decide where the end falls. The chapter opened in Elena’s truck outside her brother’s house. Dov has given her a new key to the family shop and withheld the reason their mother changed its locks; Elena has thought of telephoning their mother and decided instead to drive to the shop. After a scene break she is at the register drawer that sticks on the K, and the draft ends the chapter two sentences later:
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 next chapter opens in Dov’s thread, where he is arranging to sell the shop. A workshop reader calls the ending gripping, and in one sense she is right: it is hard to stop reading there. But the ending is chosen for a property it has on its own, and a chapter ending has very few properties on its own.
What has gone wrong
Consider what a reader who does stop here carries with her. She has an anomaly, someone in a drawer that should have been untouched, and an unease attached to no person, since the only figure in the scene is Elena, who knows no more than she does. That is a real charge, and the next page spends it. The chapter after the break cuts to Dov, and a reader who has just been asked who has been in this drawer? and is then shown the brother at his business will answer the question herself, from the cut. The cliffhanger is answered by adjacency: it becomes a delayed caption to the following chapter rather than a source of suspense. And the ending would be the third open close in a few pages: the doorway closed on a withheld reason, and Elena’s sequel ended at a scene break on a decision not yet carried out, in a book whose register is domestic and quiet.
The writer cannot easily see any of this. She composes the last line of a chapter already knowing the first line of the next; for her the white space is formatting, and she will never close the book there and sleep with the last sentence working on her. The reader will, at some breaks, and the sentence she closes the book on is the last thing the novel says to her for a night or a week. This is the segmentation problem: the chapter break is the one structural decision in a novel whose consequences are felt outside the book, in the interval when the reader is not reading, and the writer, who is never on the far side of her own breaks, has no direct access to that interval. Chapter endings are therefore found in revision, by reading the last paragraph of a chapter and then deliberately not reading on.
The claim
The chapter argues that every chapter break is an offered exit, a point at which the reader is licensed to stop, and that the writer sets its price.
A chapter break is an offered exit, and its last sentence sets the price of taking it. The price is relative, so whether a given break should be expensive or free depends on the breaks around it, and a chapter ending is a decision about a position in a sequence, not a property of the chapter.
The claim does not say that the price should always be high. A book in which every exit costs the most has no way to signal that one stop is different from the others, and the failure patterns describe what happens to it. It does not say that the reader is a customer to be manipulated: the mechanism is old and neutral, and the same device sold newspapers for The Count of Monte Cristo and made readers of Bleak House carry a crossing-sweeper’s death with them for a month. Nor does it say that chapter breaks are chiefly about suspense. Most chapter endings in most good novels are not cliffhangers and would be worse as cliffhangers; the belief that they should be descends from a commercial format, which How It Works describes. The household chapter is revised in the worked analyses, and the revision moves the break one paragraph later instead of rewriting the sentence.
See also: The Transition (Chapter 5), for the hard cut into the drawer; The Dialogue Exchange (Chapter 3), for the doorway; the Reference introduction, with the claim and its limits set out at length.