6. Anchor and Drift

6.1 The anchor’s job

Working mechanism. An anchor is the first concrete cue after a gap that answers when, where or whose vantage this is. A drift is a deliberate delay of the anchor, holding the reader in productive uncertainty for a beat before orienting them — a controlled effect, not an oversight, and distinguished from an oversight by the fact that the anchor still arrives.

§3.2 supplies the account of what an anchor does, and it is restated here in operational form because it is more specific than the craft formulation.

When a reader crosses a gap, the contextual frame they were using is not cleared. It remains primed, because nothing has replaced it. Everything they read on the far side is therefore assimilated to the old frame until a cue arrives that forces a frame switch or a frame recall. The anchor is that cue. Its function is not to inform the reader — a reader who has read nothing yet has no information deficit to fill — but to stop them reading in the wrong place.

Three consequences follow, and each corrects a common misunderstanding.

(1) An anchor is not a description. It is a single referent, correctly chosen, that triggers a retrieval or licenses a construction. A paragraph of scene-setting is not a better anchor than a proper name; it is a slower one, and during the delay the reader is still operating in the abandoned frame.

(2) The cost of a late anchor is not confusion but repair. This is the point §3.2 draws from Emmott, repeated because it changes what a writer is looking for in revision. A reader who lacks an anchor does not stop; they continue confidently in the wrong frame, and the correction, when it comes, requires them to dismantle a partial representation they had already committed to. The felt experience is not “I am lost” but “I have to go back.” In revision, the symptom to hunt is not incomprehension but backtracking.

(3) The strength of an anchor depends on how many dimensions it resolves. A proper name resolves the protagonist dimension and, in a novel with an established cast, often implies place and time as well. A weather description resolves nothing except mood. The taxonomy at §6.3 is ranked implicitly by this.

6.2 Timing: where the anchor should fall

The craft convention is “within the first sentence or two”, and it is approximately right for approximately the right reason. The reason is not politeness but the repair cost curve: the further a reader has read in the wrong frame, the more of a false representation they have built, and the cost of correcting it rises faster than linearly with the number of sentences committed.

A serviceable set of thresholds, offered as calibration rather than as law:

Two refinements.

The thresholds tighten as the dimension count rises. A one-dimension shift can survive a late anchor, because the reader’s default assumption — same place, same person, a little later — will usually be right, and no repair is needed. A four-dimension shift cannot, because every default assumption will be wrong.

They also tighten with the reader’s fatigue and the book’s pace. A reader forty pages into a fast commercial novel has been trained to expect immediate orientation and will experience a twenty-word delay as a fault. A reader of Woolf has been trained otherwise. Convention is part of what an anchor is measured against, which is one more reason the chapter’s numbers are calibrations and not thresholds of correctness.

6.3 A taxonomy of anchor types

Twelve options, listed roughly in descending order of resolving power. The purpose of the list is substitution: a join that is not working is very often the right technique anchored by the wrong type.

Notes on the types most often forgotten

The resumed object (4) is the chapter’s proposed sixth technique in its anchoring aspect (§4.6), and it is the type most systematically under-used in drafts, because it requires planting and drafts plant late. A writer who has particularised three or four objects early has a set of free anchors for the rest of the book.

The continued action (5) is the anchor that produces the least perceptible seam of any in the list, and it is entirely dependent on the launching line. It cannot be added at the arrival; it must be built at the departure. This is why §5.3 puts the launching line first.

The register shift (11) is the only anchor that works without naming anything at all, and it is therefore the only one available to a novel that wants to withhold names, places and times — which is to say, the only anchor available to a large part of modernist fiction. It is slow: a reader needs two or three sentences of a voice to recognise it. That slowness is why register-anchored transitions look like drift, and why the distinction at §6.5 matters so much for reading Faulkner and Rulfo.

The paratextual anchor (12) is free and is therefore over-chosen. A dated heading resolves time at no narrative cost, but it also announces that the book is the sort of book that has dated headings, and in contemporary publishing that is a genre claim (§10.6). It also lets a writer avoid learning to anchor in prose, which is a poor bargain over a career.

6.4 How to use the taxonomy: substitution testing

As with every taxonomy in this series, the purpose is not classification but substitution. Naming the anchor type a transition currently uses is a preliminary; the value is in rewriting it with a different one.

Procedure.

1. Take a join that is not working. Identify the anchor: the first cue that resolves when, where or who. Record its type and its word position.
2. Choose two other types at maximum distance from the current one. If the join currently anchors on a time-marker (3), try a resumed object (4) and a continued action (5). If it anchors on a proper name (1), try a sensory cue (6) and a register shift (11).
3. Rewrite the first two sentences three times — once per anchor type — without changing anything else.
4. Do not judge until all three exist. The first version has the advantage of familiarity and will always seem right while it is the only one.

What the results usually show.

The last row is the most frequently useful diagnosis in this chapter. A high proportion of transitions that resist repair at the arrival are being asked to do work that a two-word change at the departure would have made unnecessary.

6.5 Drift: the anchor delayed on purpose

Drift is the deliberate withholding of the anchor. It is a real and important technique, and it is also the alibi under which most orphaned cuts are defended in workshops, so the criterion separating them has to be stated exactly.

The criterion. In drift, the anchor still arrives — late, and on purpose. In an orphaned cut, it arrives at a point determined by the writer’s failure to notice its absence, or does not arrive at all.

The criterion is not about when the anchor comes. It is about whether its lateness is doing something. Three conditions distinguish the two in practice, and a draft that meets all three is drifting; a draft that meets none is orphaned.

1. The uncertainty is the character’s. Drift works when the reader’s disorientation reproduces the viewpoint character’s own — waking in an unfamiliar room, coming round after an injury, arriving somewhere unannounced, being in the grip of dread or grief that has detached them from place and time. Drift that reproduces nobody’s experience is merely absent orientation.
2. The uncertainty is bounded. The reader can tell what they do not know. This is the same discrimination Chapter 9 §8.6 makes about withheld information: a bounded gap generates curiosity, an unbounded one generates noise. A drift in which the reader knows they do not yet know where they are is productive; one in which they cannot tell whether they have missed something is not.
3. Something else is being supplied meanwhile. Drift is only tolerable if the sentences during the delay are rich in some other dimension — a voice, a body, a sensation, a thought. Delay plus thin prose is simply thin prose.

A scale of drift

The fourth row is where the chapter’s model begins to strain, and it should be admitted rather than smoothed over. In The Sound and the Fury’s first section and throughout Pedro Páramo, orientation is not delayed; it is redistributed, so that the reader assembles a frame from cues scattered across the whole text rather than receiving one at each join. The orientation budget does not describe this. What it can say is that these books pay in a different currency — they offer, in place of orientation, a texture of such specificity that the reader stays without knowing where they are. That is a real bargain, and it is available only to writers who can supply the texture.

A note on the workshop defence

The commonest exchange in a workshop about a confusing scene break runs: I meant it to be disorienting. The response the chapter recommends is not disagreement but a request for the schedule. When does the anchor arrive, and what is the reader doing until it does? A writer who can answer both questions is drifting. A writer who cannot has discovered the intention after the fact, which is a perfectly respectable way to find a technique but should not be confused with having used one.

6.6 How gaps are signalled typographically

The typographic mark at a gap is not neutral punctuation. It is a claim about the size and kind of the discontinuity, and readers interpret it with considerable consistency. A writer who chooses the wrong mark has misinformed the reader before the prose begins.

Four practical observations.

(1) A mark that overstates the gap is misinformation. A white-space break inserted between two consecutive beats of the same conversation tells the reader that something has changed when nothing has, and they will spend the next sentences looking for the change. This is a small fault repeated many times, and it is one of the commonest causes of a manuscript that feels choppy without any individual passage being at fault.

(2) A mark that understates the gap forces the prose to compensate. Crossing three years on a blank line is legitimate, and it obliges the first sentence on the far side to be unambiguous about the interval, because the mark has under-claimed.

(3) The blank line is unreliable in production, and this is a craft problem, not only a design one. A white-space break that falls at the foot of a page disappears: the reader turns the page and sees an ordinary paragraph beginning. Publishers use the dinkus, an ornament, or a raised initial to prevent this, and a writer submitting a manuscript should mark section breaks with a visible character rather than an empty line for the same reason. In reflowable digital text, where page boundaries are not fixed, the problem is unavoidable without a visible mark. A transition that depends entirely on a blank line to signal its gap has a single point of failure in typesetting, and the prose should be written so that the far side still anchors if the mark is lost.

(4) The mark and the technique are independent. Any of the six techniques can appear under any of the signals. Students frequently assume that a chapter break licenses a bigger ellipsis than a white-space break; convention weakly supports this, but the association is a habit rather than a rule, and Chapter 6 shows how variably it has been observed.

6.7 Historical drift, and a caution against universalising the present

§3.6 sets out the broad history. Three narrower points belong here, because they bear directly on how the anchor and the typographic mark are used.

The unmarked white-space break is a young convention with a specific origin. It requires cheap paper, a compositorial tradition willing to leave blank lines, and — decisively — a reading culture that accepts unbridged discontinuity. All three are twentieth-century conditions. A novelist of 1840 could not have used the modern hard cut even if they had wanted to, because the reader’s contract with the narrator did not include silent jumps.

The anchor conventions are genre-specific and currently diverging. Contemporary practice is not converging on a single norm; it is splitting. The table is this chapter’s estimate from reading, not a survey; the Core chapter gives named cases (The Girl on the Train for the paratextual handoff, One Day for the dated metronome).

And the caution. Every prohibition in §8 describes the middle two rows. A student who internalises them as laws will misread the first row as crude and the last two as broken. The properly critical position is the one this series maintains throughout: learn the current conventions well enough to break them deliberately, and read historically well enough to know that they are conventions.