7.0 How to use the protocol
The ten tests below are for revision, not composition. Applied while drafting they will stop a writer at every white space, which is the one place in a first draft where momentum should be preserved at all costs. Applied to a completed draft they are the fastest triage instrument in this chapter.
Five procedural notes, the first two specific to this subject.
- Read the join cold. Every test in this section assumes a specific reading procedure: cover the preceding unit, and read the transition and the first two hundred words after it as though the book began there. This is the only way to simulate the reader’s position, and it is the whole methodological trick of the section. A writer who reads a join in sequence will never find its faults, because they arrive at it already oriented.
- Run them in order. Tests 1–4 are structural, and their failures invalidate the later ones. There is no point tuning an echo at a join whose anchor never arrives.
- Record the answer, not the verdict. The output of a test is a sentence describing what the transition does — the anchor is a place-noun at word thirty-four — not a pass or a fail.
- A failed test is a question. Every test here is failed by at least one novel discussed in §9. The test’s job is to make a choice visible so that it can be confirmed as a choice.
- Sample rather than exhaust. A three-hundred-page novel has too many joins to test individually. Run the transition inventory (§5.5) on the whole manuscript, then run the ten tests on the dozen joins the inventory flags.
7.1 The ten tests
Test 1 — The anchor test
Reading the far side cold, at what word position can I answer when, where and whose vantage this is?
Count the words. Record the position and the anchor type (§6.3). Then ask the follow-up, which is the part of the test that does the work: is any delay deliberate, and if so, what is it doing? Refer to §6.5’s three conditions — is the uncertainty the character’s, is it bounded, is something else being supplied meanwhile? A writer who can answer all three is drifting. A writer who cannot has an orphaned cut (§8.1).
Practical form: give the join to a reader who has not read the preceding scene, and ask them to mark the point at which they knew where they were. Their mark will usually be later than yours.
Test 2 — The dimension-count test
How many of the five situational dimensions change across this gap?
List them: time, space, protagonist, causality, intentionality (§3.3). Count. The count is the size of the bill, and it should govern everything else about the join.
| Dimensions changed | Expected treatment |
|---|---|
| 0–1 | Hard cut; no anchor beyond a resumed referent needed |
| 2 | Hard cut with a clear anchor in the first clause, or a light stamp |
| 3 | An explicit anchor resolving at least two dimensions, immediately |
| 4–5 | Either a paratextual anchor, or a compound transition, or — better — a structural rethink: does the gap need to be this expensive? |
The test’s chief value is that it catches expensive joins that feel cheap because the prose is fluent. A four-dimension shift executed in beautiful sentences is still a four-dimension shift.
Test 3 — The launch test
What does the last sentence before the gap do for the reader on the far side?
Read the launching line alone. Does it name a destination, an intention, a person or an object that the arrival can pick up? If it does, the arrival is pre-paid and can begin inside its own action. If it does not — if the outgoing unit simply stops — the arrival must carry the whole bill, and the join should be reconsidered from the departure rather than the arrival.
This is the test most likely to solve a transition that has resisted three rewrites at the arrival, and it is the one most often skipped, because writers think of transitions as an arrival problem (§5.3).
Test 4 — The technique-fit test
Which of the six techniques is this, and was it chosen or defaulted to?
Name the technique (§4). Then ask what the gap actually required: does the interval contain anything (summary bridge)? does the reader need the clock reset (time-stamp)? is there an object available (bridging object)? do the two units comment on one another (match-cut)? is the destination already announced (hard cut)?
If the technique is a time-stamp and the answer to “does the interval matter” is no and the answer to “is the destination announced” is yes, the stamp is a default and can go.
Test 5 — The momentum test
How many words after the gap does the new unit’s own business begin?
Mark the first sentence that belongs to the new scene rather than to the transition. Count. In contemporary practice the figure is usually under forty words; over a hundred, the transition has become a preamble and should be examined for throat-clearing (§8.2).
The refinement that makes the test useful: ask what the reader could have inferred. Words spent telling a reader something they would have assumed anyway are the specific waste this test looks for, and they are invisible in sequence because they are never wrong.
Test 6 — The gap-content test
What happened inside the ellipsis, and does the reader need it?
Write, in one sentence, everything that occurs in the story during the gap. Then classify:
| Contents of the gap | Correct treatment |
|---|---|
| Nothing of consequence | Hard cut or time-stamp; do not summarise |
| Ordinary events with a cumulative effect | Summary bridge, probably iterative |
| One necessary fact | Qualified ellipsis: state it in the stamp (three days later, after the solicitor’s letter) |
| Something the reader will later need to have understood | Stop. Either dramatise it, or make the omission a deliberate paralipsis with a purpose (§8.8) |
The fourth row is the false ellipsis, and this test is the only one in the protocol that catches it, because a false ellipsis produces no local symptom at all. The join reads perfectly. The damage appears sixty pages later, when a consequence arrives without its cause.
Test 7 — The redundancy test
Is the reader oriented more than once?
Look for a time-stamp followed by a summary of the same interval; a heading naming a viewpoint followed by a paragraph establishing that viewpoint; a place named in the heading and again in the first line. Each is a double anchor (§8.7). The rule: the second element must resolve a dimension the first did not.
Also look for the recap bridge (§8.6): material the reader has already read, restated because the writer fears the reader has forgotten it. Per §3.2, a returning reader performs frame recall, not frame construction, and a single cue is enough to trigger it.
Test 8 — The simultaneity test
Does the reader know whether this unit happens after the previous one, or alongside it?
Prose has no tense for simultaneity between narrated threads. Nothing in the grammar of She parked outside the shop tells a reader whether this is after the doorway scene or during it, and readers default — strongly — to sequence. A book that means “meanwhile” and does not say so has silently rearranged its own chronology in the reader’s model.
Practical form: for every handoff, write “after” or “during” in the margin. Then check whether the text says so. Where it does not, the reader will assume “after”, and the writer must decide whether that assumption is acceptable.
Test 9 — The echo test
Could this transition be doing a second job — and if it already is, will a first-time reader notice?
Two halves. The first is generative: is there an object, an image, a phrase or an action available at this join that would carry the crossing and mean something? The second is corrective: if an echo is already present, give the two units to a reader and ask what connects them. If they cannot say, the echo is precious (§8.4) and is doing no work.
Apply the §5.4 calibration before acting on the generative half. Not every join should be doing this, and a writer who applies test 9 uniformly will produce significance fatigue.
Test 10 — The consistency test
Across the manuscript, does the technique vary with what each gap requires?
This is the only test that cannot be run on a single join. Run the transition inventory (§5.5). If one technique accounts for more than two-thirds of the joins, the book has one transition and is using it everywhere (§8.3).
Its complement is worth running at the same time: does the variation have a pattern? A book whose techniques vary randomly is not better off than one that never varies; the aim is that the technique should track the gap, so that a reader — without being able to say why — comes to feel that a chapter break in this book means something different from a white-space break.
7.2 Reading the results
| Failed tests | Likely diagnosis | First move |
|---|---|---|
| 1 alone | Under-anchored arrival | §6.3: substitute a stronger anchor type in the first clause |
| 1 and 2 | An expensive gap treated as a cheap one | Add a paratextual or compound anchor, or move the gap (Chapter 6) |
| 1 and 3 | The problem is at the departure, not the arrival | Rewrite the launching line to name the destination |
| 4 and 10 | Defaulting; the manuscript has one technique | Re-cut six important joins with three different techniques |
| 5 alone | Throat-clearing (§8.2) | Delete downward from the gap until deletion causes loss |
| 6 alone | A false ellipsis, or an interval that needed a bridge | Decide whether the gap’s contents are an asset or an omission |
| 7 alone | Double anchor or recap | Delete the second orientation; trust frame recall |
| 8 alone | Unmarked simultaneity | One phrase — that same afternoon — usually suffices |
| 9 alone | Inert connective tissue, or an undetectable echo | Either find the available object, or accept a plain join |
| 2, 5 and 10 together | Not a transition problem: a structure problem | Reverse-outline the manuscript; the gaps are in the wrong places |
A warning about test 9. Of the ten tests, the echo test is the one most likely to make a manuscript worse when over-applied. It is generative rather than corrective, and generative tests reward the writer’s ingenuity rather than the reader’s experience. Run tests 1–8 on every flagged join; run test 9 on the ten joins that matter most, and on no others.