3. Why the Crossing Costs Something

Craft advice asserts that transitions must orient the reader quickly and does not say why, or how quickly, or what “orient” names. All three can be answered, and from independent directions: from narratology, which describes what a gap is; from discourse psychology, which measures what a gap costs; from the theory of exposition, which explains what a gap can be made to do; and from film theory, which supplies the vocabulary most working writers actually use and which needs correcting before it can be used safely. A historical strand shows that the whole apparatus is younger than the novel.

3.1 Genette on duration: the transition as a managed ellipsis

The frame for everything in this chapter is Gérard Genette’s treatment of duration in Narrative Discourse (1972; trans. Jane E. Lewin, 1980) — the relation between the time an event takes in the story and the space it occupies in the discourse. Chapter 1 (The Purposes of a Scene) §6 sets out the four canonical movements and builds a twelve-mode continuum from them; that material is assumed here and not repeated. What this chapter needs from Genette is narrower and sharper: his account of ellipsis, which is the most neglected of the four categories in craft pedagogy and the most useful.

A transition, in these terms, is an ellipsis whose edges have been finished. The gap is a stretch of story time the discourse declines to narrate; the transition is everything the writer does at its two ends to make the omission legible and survivable. Hence the chapter’s structural definition: a managed ellipsis plus a re-anchoring.

Genette’s three formal types of ellipsis

Genette classifies ellipses along several axes; three of his distinctions are directly operational for a writer, and the first is the one this chapter uses most.

The practical value of the distinction is that an ellipsis a reader detects only retrospectively is a different craft object from a signposted one, and it fails in a different way and is repaired by different means.

• An explicit ellipsis is cheap and safe. Its risk is tedium and mechanism (§8.3): if every gap declares itself in the same idiom, the reader begins to hear a metronome.
• An implicit ellipsis is elegant and risky. It costs no words at the join, but it borrows against the new unit’s opening sentences, which must now carry evidence of elapsed time as well as of place and person. The orphaned cut (§8.1) is an implicit ellipsis that never redeems the loan.
• A hypothetical ellipsis is a structural device rather than a connective one. It is what The Good Soldier and Absalom, Absalom! are made of; it is why a reader of Ford cannot say when anything happened. Used deliberately it produces the sensation that the narration is not in control of its own chronology, which for some books is the effect required. Used accidentally it produces the sensation that the author is not in control of it, which is never the effect required.

Genette adds two further axes that revision can use directly. An ellipsis is definite when its duration is specified (three days later) and indefinite when it is not (some time afterwards, later that year); and it is qualified when the text supplies information about what happened inside it (three days later, after the funeral) and unqualified when it does not. Most workshop advice about transitions is, when translated, advice about these two axes: say how long it was is a plea for definiteness; don’t summarise what we don’t need is a plea against over-qualification.

Frequency: the iterative bridge

Genette’s third category, frequency, concerns how many times an event is narrated relative to how often it occurred, and it supplies the single most under-used transition instrument in contemporary fiction.

• The singulative narrates once what happened once. It is the default, and it is what most transitions in most manuscripts are made of.
• The iterative narrates once what happened many times: Every Thursday he counted the till twice. An iterative sentence covers an indefinite stretch of story time in one grammatical gesture, and it does so while characterising — because a habit is a fact about a person as well as a fact about a calendar.

This matters here because the iterative is a transition technique in disguise. A paragraph of habitual action is a bridge that crosses months, establishes the normal against which the next scene’s departure will register, and costs the reader almost nothing in processing, since habitual grammar suppresses the demand for specific temporal localisation. Its near-disappearance from contemporary manuscripts, noted at Chapter 1 §6.3, is one reason so many drafts reach for the time-stamp: the writer has one bridging tool where they should have two.

Genette’s pseudo-iterative — iterative grammar carrying singulative detail, as in Proust’s Combray evenings — is worth flagging as a transition device too. It is logically incoherent (nobody’s Thursdays are identical down to the remark someone made) and readers accept it without difficulty, which tells us something useful about how loosely temporal bookkeeping is actually maintained during reading.

One more Genettean term, held in reserve

Paralipsis — Genette’s name, under the heading of mood rather than duration, for the omission of information the adopted focalisation ought to have supplied — is the exact mechanism of the failure pattern this chapter calls the false ellipsis (§8.8). A gap that skips a stretch of time is an ellipsis. A gap that skips the event the book is about, while continuing to present itself as a mere interval, is a paralipsis wearing an ellipsis’s clothes. The distinction is worth carrying because the two look identical on the page and are read completely differently once the reader catches on.

3.2 Emmott’s contextual frames: what an anchor actually does

If Genette explains what a gap is, Catherine Emmott explains what the reader has to do about it — and her account is the closest fit between narrative theory and practical craft anywhere in this textbook.

Narrative Comprehension: A Discourse Perspective (1997) proposes that readers maintain contextual frames: mental representations of a specific episode’s context, holding who is present, where they are, and when. Characters are bound into a frame; a frame that a reader is currently using is primed. Emmott’s argument is that these frames, not sentences, are what makes extended narrative processable — a pronoun four pages into a scene is resolved not by searching backwards through the text but by consulting the primed frame, in which one referent is available and the others are not.

Three of her operations describe, with unusual precision, what happens at a transition.

The consequences for practice are direct, and they are the theoretical underwriting of the whole of §6.

First: the anchor is a frame-priming instruction. When a reader crosses white space, the previous frame is not automatically discarded — it is the only one available, so it remains primed until something dislodges it. The first concrete cue on the far side is what tells the reader to build a new frame or recall an old one. Until that cue arrives, the reader is processing new sentences inside the old frame, which is why an unanchored cut does not produce a blank so much as a wrong reading that must later be undone.

Second: frame repair is expensive in a way that frame building is not. A reader who has assembled two sentences into the old context and then discovers they belong to a new one must not only construct the new frame but dismantle a partial and now false representation. This is a better account of the felt experience of a bad transition than “confusion”. The reader is not lost. The reader has confidently gone the wrong way, and the cost is the walk back.

Third: frame recall is much cheaper than frame switching, and writers systematically over-pay for it. When a novel returns to a thread the reader has already inhabited — Elena in the shop, the second Levin chapter, a viewpoint the book has established — the reader does not build the context from scratch. They retrieve it, and a single cue is enough to trigger the retrieval. This is the mechanism behind the recap bridge failure (§8.6): a writer treats a return as though it were an arrival, and re-establishes what the reader could have recovered from one proper noun.

The corollary is a useful revision rule:

Orient in proportion to novelty, not in proportion to elapsed page count. A new frame must be built; an old frame need only be named. The reader’s memory for a scene they have inhabited is far better than most drafts assume, and far worse for details they were merely told.

Fourth: binding is what makes a location an anchor. Emmott distinguishes characters who are bound into a frame from those merely mentioned. A place-noun by itself primes a location; a place-noun with a bound character in it primes a scene. This is why The shop was cold is a weaker anchor than Elena let herself into the shop, even though the second contains no more information about the location. The second binds a person to it, and the frame is now populated.

Emmott’s model also explains a phenomenon craft advice notices without explaining: that a reader can cross an enormous gap effortlessly if the far side is a frame they already hold, and stumble over a small one if it is not. Ten years elapse in To the Lighthouse and the reader arrives in a house they know. A cut of two hours into an unnamed corridor with an unnamed occupant is harder work. The size of the ellipsis and the difficulty of the crossing are only loosely related, which is the first thing that recommends “budget” over “distance” as the governing metaphor.

3.3 The event-indexing model: measuring the cost of a shift

Emmott describes the architecture. A parallel body of work in discourse psychology measures the load.

The event-indexing model (Zwaan, Langston and Graesser, 1995) holds that readers index incoming events along a small number of dimensions and monitor each for continuity with what came before. Rolf Zwaan and Gabriel Radvansky’s synthesis in Psychological Bulletin (1998) gives the canonical list:

The empirical finding that matters for this chapter is simple and robust: a discontinuity on any of these dimensions slows reading measurably at the point of the shift. Readers take longer over a sentence that moves the action to a new place, or to a new time, or to a new protagonist, than over one that continues in the same place, time and protagonist. Zwaan’s work on temporal shifts (1996) found the effect graded rather than binary — an hour later costs more to process than a moment later — which is what the craft intuition about “large” and “small” gaps has always claimed without evidence.

Three refinements deserve emphasis, because they turn a general finding into usable craft.

(1) The costs are largely additive across dimensions. This is the finding the chapter leans on hardest. A shift on two dimensions costs roughly what two shifts cost; a shift on four costs roughly what four cost. Craft advice treats the POV-and-location handoff as the most expensive technique in the repertoire and offers no reason. The reason is here: a handoff to a new viewpoint character, in a new place, at a new time, pursuing a different goal, shifts four of the five dimensions at once, and the reader pays for all four in the same two sentences.

The dimension count is the best single predictor of how much orientation a gap requires. Count the dimensions that change; that is the size of the bill.

This gives a diagnostic instrument, formalised at §7.1 as test 2. It also explains why some very large ellipses are cheap — a jump of five years that keeps the same protagonist in the same house shifts one dimension — and why some very small ones are expensive: a cut of ten minutes to a different character in a different building, with an unrelated goal, shifts four.

(2) Causality and intentionality are the dimensions writers forget. Time, space and protagonist are the ones craft advice names; they are the when, where, whose vantage of this chapter’s working definition. But the two remaining dimensions are the ones that make a transition feel arbitrary rather than merely fast. A cut that preserves time, place and person but breaks the causal chain — the new material has no relation to what preceded it — will read as a non sequitur even though every question of orientation has been answered. This is the failure Chapter 2 (Scene and Sequel) diagnoses as a broken chain, and it is why that chapter and this one must be read together. Orientation is necessary and not sufficient. The reader can know when and where they are and still not know why they are there.

(3) The cost is a cost, not a defect. This is where the chapter must be careful with its own evidence. That readers slow down at a shift does not mean shifts are bad; it means shifts are expensive, and expense buys things. Segmentation research — Zacks and Tversky’s account of event structure (2001), and Speer, Zacks and Reynolds’s finding (2007) that neural activity tracks narrative event boundaries — suggests that readers naturally parse continuous experience into discrete events at these discontinuities, and that boundaries are where memory is consolidated. A narrative with no shifts would not be effortless; it would be unsegmented, and therefore unmemorable. The processing cost at a boundary is partly the cost of filing what came before.

This has a consequence craft literature rarely draws: the transition is where the previous scene gets remembered. A join is not only an entrance but a moment of consolidation, which is one reason a scene that ends well is remembered better than one that merely stops, and one reason the launching line (§5.3) repays attention out of proportion to its length.

A methodological caution, in keeping with the series’ practice. These are findings about reading times and about comprehension under laboratory conditions, mostly with short constructed texts. They describe tendencies in processing, not laws of literary value. That a shift costs three hundred milliseconds tells us nothing about whether Woolf should have made it. Their proper use is to explain why a well-attested craft intuition is well attested, and to convert a vague sense that a transition is “too much at once” into a countable quantity.

3.4 Sternberg: the gap as an expositional opportunity

Meir Sternberg’s Expositional Modes and Temporal Ordering in Fiction (1978) reframes exposition as a distribution problem: every narrative must convey its antecedents, and the only variables are when and in what order. Chapter 9 (The Opening Pages) uses this to argue that front-loading backstory spends an asset. The transition-side consequence is different and is not usually drawn.

Gaps are where exposition is distributed. The moment after a break is one of the few positions in a novel at which a reader will accept summary without resistance, because they have just been told, typographically, that something has been skipped and are actively looking for information about what. A paragraph of exposition dropped into the middle of a scene stalls it; the identical paragraph placed immediately after a scene break reads as the transition doing its job.

This gives the summary bridge (§4.4) a defence it is rarely granted. It is not merely a way of covering uninteresting intervals. It is the natural home of distributed exposition, and a novel that never uses it must find some other way to place its antecedents, usually by making characters tell each other things they both already know.

Sternberg’s distinction between suspense (uncertainty about the future) and curiosity (uncertainty about the past) also bears directly on the choice of technique, and can be tabulated:

The third row is the one drafts get wrong. A great deal of unnecessary mystery in manuscripts is manufactured by leaving intervals unspecified in the hope that vagueness reads as depth. Sternberg’s framework makes the question answerable: an interval should be made mysterious only if the reader’s curiosity about it is an asset the book intends to spend later. If nothing will be paid out, say how long it was and move on.

3.5 The film strand, admitted and then corrected

Almost every working writer’s mental model of the transition is cinematic. The vocabulary gives it away: cut, cut to, match cut, cross-cutting, establishing shot, fade. The borrowing is old, productive and, taken uncritically, a source of specific and identifiable errors. This section admits the debt and then draws the line.

What film theory supplies

David Bordwell and Kristin Thompson’s Film Art codifies the continuity editing system — the set of conventions by which classical narrative cinema makes cuts unobtrusive. Several of its devices have real prose analogues:

Sergei Eisenstein’s account of montage supplies the other half. Where continuity editing hides the join, Eisenstein’s montage is the meaning: two shots collide and the spectator produces a third thing that is in neither. The Kuleshov effect — an identical face read as hunger, grief or desire depending on the shot beside it — is the purest demonstration that juxtaposition is itself a signifying operation. This is the theoretical parent of the match-cut transition (§4.3), and it justifies the claim there that a well-chosen join can carry an argument no sentence in either unit states.

Where the analogy breaks, and what it costs

Three disanalogies matter, and the third is the one that produces the errors.

(1) The film cut is instantaneous; the prose cut is not. A film cut delivers an entire new frame in a twenty-fourth of a second, complete with location, time of day, weather, who is present, their ages, their clothes, their relative positions and the emotional register of the lighting. Prose delivers information serially, one item per phrase, and the reader assembles a frame over several seconds of reading. There is no prose operation that corresponds to a cut, and the shorthand “cut to” quietly imports the assumption that the new frame arrives whole. It does not; it arrives in a queue, in whatever order the writer put it.

(2) Film orientation is involuntary and free; prose orientation is voluntary and paid for. A director who wants the viewer to know it is raining does not spend anything: the rain is in the frame whether or not it is important. A novelist who wants the reader to know it is raining spends a clause, and — per Rabinowitz’s rules of notice, discussed in Chapter 9 §3.2 — that clause is read as a claim that the rain matters. Prose has no free background. This is the deep reason the “establishing shot” transfers so badly: the establishing paragraph is not neutral scene-setting but a promise, and a writer who opens every scene with one has promised that fourteen rooms are significant.

(3) The viewer cannot go back; the reader can, and does. A confusing cut in a film is survived — the viewer waits, and the film resolves it or does not. A confusing cut in a novel produces the backtrack, which is the specific behaviour this chapter’s diagnostics are designed to prevent. The costs of an unclear join are therefore asymmetric between the media: film’s confusion is transient, prose’s confusion is corrective.

The generalised warning. “Cut to” is a good instruction about sequence and a bad instruction about cost. Everything the camera would have given the viewer for nothing must, in prose, be either written or dispensed with. A writer thinking cinematically will reliably under-budget the arrival.

What does transfer

Two things, both of use.

• Cutting late and leaving early. The screenwriting maxim — enter a scene as late as possible and leave before it winds down — is sound in prose for the same reason it is sound on screen: the audience infers the removed material without difficulty, and the removal raises the density of what remains. This is a claim about where the gap goes, and therefore belongs mostly to Chapter 6, but its transition-side consequence is that a late entry demands a faster anchor.
• Juxtaposition as an argument. Eisenstein’s insight survives translation into prose intact, because it is a claim about how readers and viewers alike make meaning out of adjacency. The match-cut (§4.3) is its prose form.

3.6 The historical strand: how novels crossed gaps before white space

Contemporary transition practice looks like a set of natural laws and is in fact a young and local convention. The relevant history has three phases.

Phase one: the gap is spoken. The eighteenth- and nineteenth-century novel bridged its gaps by saying so. The narrator addresses the reader, announces the change of scene, and takes responsibility for it: we must now return to, let us leave him for the present, while these things were passing at X, let us see what became of Y. Fielding gives each Book of Tom Jones an introductory chapter of open authorial commentary, so that the largest joins in the novel are not concealed but occupied. Thackeray, Trollope, Dickens and Balzac all use the explicit narratorial handover as ordinary equipment. The device is not clumsiness; it follows from a narrative situation in which the narrator is a presence with a relationship to the reader, and in which concealing the machinery would be a stranger choice than displaying it.

Bleak House (1853) shows the range within a single novel: the third-person present-tense narrator moves between London institutions with explicit deictic gestures — turning aside from Chancery, in Chapter 2, for what he calls a glimpse of the world of fashion on the same miry afternoon — while the alternation with Esther’s retrospective first person is handled with no bridging language at all, by chapter division alone. §9.8 treats the system in full.

Phase two: the gap is marked. Typographic section division within a chapter is old — printers have used ornaments, rules and asterisks to mark divisions for centuries — but the modern convention, in which a blank line stands alone as a narrative signal with no verbal accompaniment, becomes ordinary equipment in the twentieth century. Nicholas Dames’s The Chapter (2023), the standard modern history of segmentation, is the place to pursue the longer story; its central lesson for this chapter is that the units we take for granted are historically contingent artefacts with their own periods and fashions, and that many of them originated as editorial or scribal conveniences rather than as authorial art.

Phase three: the gap is cut. The unbridged hard cut — white space, then a new scene beginning in medias res with no verbal acknowledgement of the discontinuity — is a twentieth-century norm, and cinema is part of its history, though the modernist prose experiments of the 1910s and 1920s run alongside and sometimes ahead of the film techniques they are supposed to derive from. Hemingway’s practice, Woolf’s, Dos Passos’s and Joyce’s each contribute; by mid-century the unmarked break is unremarkable, and by the 1980s it is the default in Anglo-American commercial and literary fiction alike.

Two inferences follow, and they pull against each other, which is the point.

• The advice in §§4–8 is calibrated to contemporary convention. Nearly every prohibition this chapter issues describes Anglo-American publishing practice since roughly 1960. That is a reason to know it and a reason not to mistake it for a law of narrative.
• The historical practices are not merely tolerated. Fielding’s essayistic bridges are among the best things in Tom Jones; the Victorian narratorial handover is a resource contemporary fiction has largely lost and occasionally rediscovers. A pedagogy that can only say this would be edited out today has explained the market and not the book.

There is also a live counter-current worth noticing, because students often assume the direction of travel is one-way. Explicit narratorial address has returned in a substantial body of recent literary fiction, and the dated section heading — London, March 1943 — has become so standard in historical and multi-thread commercial fiction that it now functions as a genre signal in its own right (§10.6). The hard cut is not the end of a progression. It is one convention among several currently in use.

3.7 The first principle, stated precisely — and qualified

The chapter’s first principle can now be given in full.

The orientation budget. Every gap in a narrative imposes a processing cost on the reader that varies with the number of situational dimensions it changes. A transition is the expenditure of narrative material to meet that cost. The correct expenditure is the least that closes the gap for this gap — which may be a paragraph, a clause, or nothing.

Three components make the principle usable:

1. The bill is set by the gap, not by the writer’s habits. Count the dimensions that shift (§7.1, test 2). Four dimensions require more than one.
2. The payment can be made on either side. A launching line that names the destination (§5.3) discharges part of the bill before the gap; a self-anchoring opening sentence discharges the rest after it. Writers who think of transitions as arrival-side only are working with half the instrument.
3. Over-payment is a fault, not a safety margin. Words spent on orientation the reader had already inferred delay the new unit’s engine and read as throat-clearing (§8.2). The failure is symmetrical with under-payment, and rather harder to see, because over-explained prose is never confusing.

And now the qualification, which is where the more interesting teaching lies.

The budget is a metaphor, and it encodes an assumption the chapter should not pretend is neutral. It assumes that orientation is a cost to be minimised, that reader attention is a scarce resource, and that the well-made transition is the invisible one. Every one of those assumptions is a description of a particular literary culture — broadly, the Anglo-American realist and commercial traditions of the last half-century — rather than of narrative as such.

Consider what the metaphor cannot describe.

• “Time Passes” (§9.1) is an ellipsis of ten years narrated at length. By the logic of the budget it is an absurdity: a transition that costs more than the units it joins. It is also one of the great formal achievements in English fiction, and no account of transitions that rules it out has understood the subject.
• Pedro Páramo (§9.6) declines to signal its shifts at all, between speakers, between times, and between the living and the dead. The reader is not under-oriented by mistake. The novel’s proposition is that these categories do not admit of the distinctions the budget presupposes.
• The Victorian narratorial bridge spends lavishly and buys something the budget cannot price: the presence of a narrator, and a relationship with a reader, which is itself a large part of what those novels are for.

The budget therefore survives as a revision instrument, not as an aesthetic. Its proper use is to make an expenditure visible so that it can be confirmed as a choice — which is the use §7 puts it to, and the use the Coda warns against exceeding.