The ten tests are for revision, not composition; applied while drafting they produce a writer who cannot name a chair without justifying it. Run them on passages, not on every detail: the passages a reader found slow, the ones the writer suspects, and one sample for each main setting. Run tests 1 to 4 (Removal, Denotation and connotation, Genericness and Substitution) first, since there is no point auditing modifiers in a paragraph nobody would miss. Record a sentence, not a verdict: “this detail supplies mood and nothing else” is useful, and “fails test 2” is not. Do tests 8 (Count) and 9 (Modifiers and senses) mechanically, with a pen, because a writer’s estimate of her own descriptive density is unreliable. Each test is failed on purpose by a book worth reading.
Test 1 — Removal
Cut the detail. What does the reader no longer know, feel or anticipate?
Failure. Nothing: decoration, or a detail placed where emphasis promises more than it gives.
Exception. A reality-effect detail kept in proportion.
Test 2 — Denotation and connotation
Write one sentence about what it shows and one about what it implies.
Failure. The second sentence cannot be written: the detail means only itself.
Exception. A plain report chosen as the book’s register.
Test 3 — Genericness
Would it be equally true of another place of this kind in another book?
Failure. A category detail that confirms what the name already supplied.
Exception. A cheap orienting cue, used sparingly.
Test 4 — Substitution
Replace it with two other true things. Do they imply different books?
Failure. The choice was not doing selective work; change the object.
Exception. Interchangeable detail in a deliberately uniform world.
Test 5 — Focalisation
Whose attention selected it, and does the selection reveal them?
Failure. In close narration, a detail written over the character’s shoulder.
Exception. Declared omniscience (Middlemarch).
Test 6 — Plausibility of attention
Would this person, wanting what she wants, in her present state, notice this?
Failure. A survey performed by someone in no condition to make one.
Exception. A perceiver given a reason to wait or look.
Test 7 — Recurrence
Where else does the object appear, and what has changed each time?
Failure. One emphasised appearance (an unpaid promise), or returns with nothing varied.
Exception. Identical repetition within one passage (the fog of Bleak House).
Test 8 — Count
How many discrete details, and how many do you want kept?
Failure. Many offered, few intended: the reader discards her own choice.
Exception. A catalogue with an ordering principle.
Test 9 — Modifiers and senses
(a) Count modifiers per noun. (b) Tally details by sense.
Failure. (a) Triangulating towards a missing noun. (b) Sight only.
Exception. Nabokov’s cumulative modifiers; Robbe-Grillet’s chosen visuality.
Test 10 — Inventory
Shuffle the details. Then complete: these details, and not others, because ….
Failure. Shuffling loses nothing, and the answer is “because they were there”.
Exception. None; a passage without a principle can only be rewritten from one.
Four refinements sharpen the tests. In test 4 (Substitution), substitute objects as well as attributes: not only which true thing about the coat, but why the coat and not the shoes or the shopping list on the table; that is where the larger gains are. In test 6 (Plausibility of attention), remember that attention narrows under fear, pain, urgency, grief and desire, towards the useful and the personally charged, so a person fleeing a building does not register the cornices, and a person in grief registers two things with unbearable clarity and nothing else. If a passage needs description its perceiver would not perform, change the perceiver’s state, not the description: give her a reason to wait. In test 9 (Modifiers and senses)(b), sight dominates most drafts, by this chapter’s estimate four fifths or more of the details in a typical draft passage (a working guess, not a measurement), and the remedy is not one of each sense but the single non-visual detail that carries most, displacing a visual one. In test 10 (Inventory), acceptable completions name a perceiver’s preoccupation, a causal argument (because the rent explains the lodgers), a temporal record (because nobody has touched these since March) or a formal order (outward from the body).
Reading the results together
Failures cluster, and the cluster points to the first move. Tests 1 (Removal) and 10 (Inventory) together mean an inventory: choose a principle of selection and rewrite from it, and do not edit. Test 2 (Denotation and connotation) alone means decoration: replace the object, not the wording. Test 3 (Genericness) alone means a generic world: one specific noun and one exact quantity. Tests 5 (Focalisation) and 6 (Plausibility of attention) together mean unfocalised description: attach the passage to a perceiver with a want and a state. Test 7 (Recurrence) alone means a static motif or an unredeemed plant: vary the object’s state, or demote it. Test 8 (Count) alone means more details than a reader is likely to retain: cut to three and keep those that carry two domains. Test 2 (Denotation and connotation) passed too emphatically, with an implication longer and more interesting than the object, means a symbol announcing itself: reduce the emphasis, not the meaning.
The tests lean towards the purposeful, and applied at full strength to a whole manuscript they produce a book in which every detail justifies itself. Use them for triage, where a reader has flagged a problem, and leave the rest of the book its air.
See also: the Reference protocol, with each test’s procedure and a table of combined failures; the domain audit, which marks each detail with the domains it serves.