8. Failure Patterns

Each pattern below is presented with its symptom, its mechanism (what actually goes wrong, as distinct from what the surface looks like), its persistence (why competent writers keep producing it), its remedy, and its exception: a published work that commits the error and succeeds.

The exceptions are the intellectual heart of the section and they are not decorative. A pedagogy of prohibitions produces writers who can avoid faults and not writers who can make choices. In every case the pattern fails because of a mechanism, not because of a surface feature, and where the mechanism is absent the surface feature is harmless.

The prologue makes this discipline unusually easy to demonstrate, because the form’s most celebrated instances break its most confident rules. Romeo and Juliet’s prologue gives away the entire plot. The Scarlet Letter’s runs to fifty pages, is largely autobiographical, opens a distance that is never closed, and is routinely skipped — and Hawthorne knew it, and kept it, and republished it without altering a word. Dracula’s is pure information delivery with no scene in it. Any account of the prologue that cannot accommodate those three facts is describing a market rather than a form.

8.1 The Info-Dump Prologue

Symptom. History, world mechanics, magic systems, dynastic genealogy or political background delivered before any person or event: the briefing in the doorway.

Mechanism. Two failures, usually simultaneous. First, per Sternberg (§3.7), preliminary exposition spends an asset: information delivered early cannot generate curiosity later, and the prologue is the most concentrated form of the most front-loaded expositional mode. Second, and more fundamental: information is not memorable without a container. A reader given four generations of a dynasty before meeting anyone who lives in it has nowhere to store the material and will retain almost none of it, which means the info-dump does not even achieve its own stated purpose.

The prologue-specific aggravation is that the separation licenses the failure. A writer who would not tolerate three pages of exposition in Chapter One will tolerate it in a prologue, because the compartment seems to make it permissible. It does not; it makes it optional, which is worse.

Persistence. The writer knows the world intimately and fears the reader will be lost without it. The fear is real; the remedy is misdirected. Readers do not need to know a world early. They need to believe the author knows it, which is achieved by confident specificity inside a scene rather than by explanation outside one.

Remedy. Run test 5 (register). If rewriting the material in Chapter One’s voice loses nothing but length, the material is information and belongs distributed through the main narrative. If the material truly requires the register — if it must be said in a chronicler’s voice — then keep the register and cut the quantity: an editorial note (type 7) delivers in a hundred words what a historical overture takes two thousand to say.

The exception. Dracula (1897) opens with an unsigned note that is nothing but information delivery: an explanation of how a set of papers was assembled, in what order, on what principle, with an assertion about their contemporaneity. There is no scene, no character, no want, no image. It is what §8.1 prohibits, and it is one of the most efficient prologues in English fiction, because the information it delivers is not about the world but about how to read the book — which is the one kind of information a reader cannot acquire from inside the narrative. See §9.7.

Tolkien’s “Concerning Hobbits” is the harder case and should be taught alongside it: unapologetic ethnography, dozens of pages, and demonstrably skipped by a large fraction of readers. It succeeds for its readers because the exposition is characterised — it has a voice, an antiquarian’s affection, a pretence of scholarship — and because the register it establishes is the one the book keeps. Info-dumping fails when it is voiceless. It survives when the dump is itself a performance.

8.2 The False-Advertising Prologue

Symptom. A hooky, violent or high-stakes prologue whose pitch the rest of the book never matches: a murder before a quiet domestic novel, a battle before a comedy of manners, a chase before three hundred pages of interiority.

Mechanism. The prologue writes a cheque the book does not honour, and the primacy effect (§3.9) then converts every subsequent page into a deviation from the promised book rather than an instance of the actual one. This is worse than Chapter 9’s borrowed hook (§8.4 there) for two reasons specific to the separated unit: the label makes the promise emphatic and deliberate, and the prologue’s privileged position (§3.9) makes its every detail load-bearing.

Note the asymmetry, offered as a conjecture across every version of this failure in the book, not as a measured finding: readers seem to forgive a book that becomes more than it promised, and not one that becomes less.

This is also one of the book’s two matching genre-contract violations. Chapter 16 (The Ending) describes the mismatched final note — a close whose register belongs to a different book. The prologue’s false advertising and the ending’s mismatched note are the same violation at the two boundaries, and a manuscript that commits one frequently commits the other, because both arise from a writer who has not settled what kind of book they are writing.

Persistence. Submission pressure (§3.6). Writers are told to hook; the fastest hook available is an event more exciting than the book’s actual material; and the mismatch is invisible from inside, because the writer knows what is coming and reads their own quiet chapters with the whole novel’s weight behind them.

Remedy. Test 10, run honestly on pitch, pace, mode and scale. Then revise the prologue, not the book.

The exception. Wuthering Heights (1847). Lockwood’s opening chapters advertise a light social comedy: a fatuous city tenant, a misjudged landlord, a joke about misanthropy, a comic misreading of a household. What follows is one of the most violent and metaphysically extreme novels in English. The mismatch is total — and it is the book’s method rather than its defect, because the distance between Lockwood’s register and the material he blunders into is what the novel is about. The reader’s discovery that the frame cannot contain the story is the first and largest of the book’s effects.

The discrimination is exact and it is the one students most need: a mismatch that the book uses is a structure; a mismatch the book has not noticed is a fraud. Test 10’s real question is not whether the pitches differ but whether the difference is legible as a choice.

A modern secondary case: Martin’s A Game of Thrones (1996) opens with a supernatural horror prologue and then withholds the supernatural for hundreds of pages of political intrigue. The mismatch is deliberate, structurally motivated (the reader must know the winter is real while the characters squabble), and paid off across a series rather than a volume — which is also a demonstration of how much longer a return may be deferred in a multi-volume work (§10.3).

8.3 The Redundant Prologue

Symptom. The prologue covers ground Chapter One will cover anyway: the same inciting incident told twice, once at a distance and once up close, with nothing gained on the second telling.

Mechanism. A failure of the second test of the higher bar (§3.10) in its clearest form. The material earns its existence — it is a real event, competently rendered — and earns nothing by being exterior, because the interior version is coming. The reader pays the double start-up cost of §1.1 and receives, in exchange, the same information twice.

The subtler version is redundancy of effect rather than of content. A prologue that establishes dread, followed by a Chapter One that establishes the same dread more efficiently, is redundant even though no event repeats.

Persistence. Drafting order. The prologue is frequently written first, as the writer’s own way into the material, and Chapter One is then written to stand alone because the writer is not yet sure the prologue will survive. Both units end up doing the same job, and the writer, having written each in a different mood, cannot see the duplication.

Remedy. Test 7 and test 9 together. Where an element appears in both, decide which position serves it better and delete the other — and note that the answer is very often the interior version, since the same material is nearly always stronger when a reader has a person to attach it to.

The exception. The Scarlet Letter’s “The Custom-House” (1850) summarises Hester Prynne’s story — the letter, the punishment, the outline of a life — before the romance narrates it. By the terms of this pattern, that is textbook redundancy: the reader is told the story and then told it again at length.

It succeeds because the two tellings are not the same act. The Custom-House version is a nineteenth-century surveyor’s precis of a document; the romance is the thing itself, in a register the essay explicitly cannot reach. The redundancy is what produces the book’s characteristic doubleness — a modern man handling a Puritan past he can summarise and cannot inhabit. Redundancy fails when the second telling adds nothing; it succeeds when the two tellings are performed by different sensibilities and the gap between them is the subject.

8.4 The Unresolved Distance

Symptom. The prologue opens a real gap — a time jump, a viewpoint switch, a mystery, a frame — and the book never returns to it. The promise is never collected.

Mechanism. Emmott’s frame model (§6.2) supplies the account. A contextual frame is built, primed briefly, and then suspended permanently. The reader holds it, at cost, for the length of the book, expecting reinstatement that never comes. The subjective experience is not confusion but a diffuse sense that the book was longer at the front than it needed to be — which readers usually report, inaccurately, as “the prologue was unnecessary”.

The mechanism explains why the complaint is retrospective. Nothing is wrong at the time of reading; the failure occurs at the end, when the reader performs an implicit audit and finds an item unclosed.

Persistence. The writer knows the connection and cannot see that it was never written down. This is the commonest form of the general problem that a writer’s knowledge of their own book is invisible to them as knowledge.

Remedy. Test 6, run as a literal inventory with page numbers. Then either build the return — preferably fragmentary (§6.6) — or delete the distance by absorbing the material into the main narrative, where no gap needs closing.

The exception. The Scarlet Letter, again, and more radically. “The Custom-House” opens a distance on all four axes at once: two centuries of time, a different town and building, a first-person narrator who is not in the romance, and an essayistic register the romance never uses. None of it is ever closed. The surveyor does not reappear. The narrative does not return to the attic. The frame is abandoned at the threshold and never resumed.

It works — insofar as it works, and there is a real critical argument here — because the non-closure is the point: the essay is about a man who cannot enter the past he is holding in his hands, and returning to him at the end would assert a continuity the book denies. Hawthorne knew the sketch was unpopular; he defended it in a preface to the second edition and declined to alter it.

Whether this is design or self-indulgence is one of the standing disagreements in American criticism, and §12.2 sets it as a seminar question rather than resolving it. That is the honest position: the exception is real, and it is contested.

8.5 The Chapter One in Disguise

Symptom. The prologue shares the protagonist’s time, place, viewpoint and register, and was labelled “Prologue” because the writer was not sure the opening earned its place as Chapter One.

Mechanism. No positional justification exists because none was needed. The label creates an expectation of distance which the material does not honour, so the reader spends the passage looking for a gap that is not there. Worse, the label instructs the reader to hold the material at arm’s length — to treat it as preliminary — which is the wrong instruction for a scene that is in fact the beginning of the book.

In this edition’s judgement this is the most frequent pattern in draft manuscripts, and it is the reason the workshop advice “cut your prologue” so often produces an immediate improvement: the advice is wrong in general and right in this very common particular case.

Persistence. The label functions as a hedge. A writer uncertain whether their opening is strong enough, or uncertain where the story begins, can defer the decision by calling the passage a prologue — which feels like a structural choice and is in fact its avoidance. Said’s distinction (§3.8) names the fault: the writer has declined to perform the act of beginning and has substituted a gesture at an origin.

Remedy. Test 3, which is also the entire cure. Relabel, renumber, and read. Nothing else is usually required.

The exception. Ellison’s Invisible Man (1952) has a labelled Prologue narrated in the first person by the book’s own narrator, in the book’s own voice, at the book’s characteristic pitch. Three of the four axes are shared: same person, same register, same consciousness. Only time differs — the Prologue is spoken from after everything, from the cellar the Epilogue returns to.

The exception is instructive because it shows that a single axis is sufficient when that axis is load-bearing. The temporal gap between the man in the hole and the boy at the battle royal is the novel’s whole architecture, and the label is what tells the reader that the voice they are hearing has already been where the story is going. Remove the label and the first chapter’s naïveté becomes ambiguous rather than framed.

The generalised rule. Chapter One in disguise is not a failure of how many axes are open. It is a failure of any axis being open. One is enough, if it does the work.

8.6 The Prologue That Spoils

Symptom. The prologue reveals the outcome — who dies, who is guilty, how it ends — and the reader proceeds knowing what the book was going to spend four hundred pages concealing.

Mechanism. Per Sternberg (§3.7), disclosure converts suspense into curiosity. Where a book’s pleasures are outcome-based — a mystery, a thriller, a race — that conversion destroys the engine, because the reader’s question (what happens?) has been answered and the substitute question (how?) is not one the book has been built to reward.

The failure is therefore conditional on genre, and that conditionality is the whole of the analysis. There is no general prohibition on spoiling. There is a prohibition on spoiling a book whose value lies in the thing spoiled.

Persistence. Two sources. Writers reach for the proleptic flash because it is the most powerful available pull-forward, without asking what it costs; and a culture in which “spoiler” is a term of abuse has produced an anxiety about disclosure that is far broader than the structural problem warrants.

Remedy. Ask which of Sternberg’s two engines the book actually runs on. If curiosity — if the book is about causes, character, inevitability, or the shape of a life — disclose freely. If suspense, do not disclose the outcome; disclose an aftermath instead (§6.3, type 1), which promises shape without answering the question.

The exception. Romeo and Juliet (c. 1595), and it demolishes the pattern as a general rule rather than merely qualifying it.

The sonnet-prologue gives away everything: the city, the feud, the lovers, their deaths, and the reconciliation their deaths purchase. The audience is told the ending in the first fourteen lines of a love-tragedy that has held the stage for four centuries. Four observations follow, and they are the whole teaching:

1. The play’s pathos is manufactured by the disclosure. Every hopeful line the lovers speak is heard by an audience who knows. Remove the prologue and the play loses its dominant emotional mechanism.
2. The operative question is changed, not removed. From what happens to how, and could it have been otherwise — which is the question the play is built to answer, at length, through a sequence of near-misses and mistimings.
3. Kermode’s concord is supplied in advance (§3.8). The audience is told there is a shape and what it is, which underwrites the tolerance a two-hour traffic requires.
4. This is not a special licence granted to drama. The same manoeuvre works in the novel: Tolstoy opens The Death of Ivan Ilyich with the death and the colleagues’ reactions; Nabokov’s John Ray, in the foreword to Lolita, discloses Humbert’s death in legal captivity and, in a single clause that most first readers pass over and no second reader can, the death of a young woman in childbirth. Both books are stronger for the disclosure, because both are books about causes.

The corrected principle. A prologue may spoil anything the book is not for. The error is never disclosure; it is disclosing the one thing the reader was going to be reading towards.

8.7 The Anonymous-Victim Cold Open

Symptom. A prologue narrated by or focalised through a character who is murdered at the end of it — usually unnamed, usually female, usually in a book about the detective who will investigate.

Mechanism. A failure in the character domain (§4.2): the reader is asked to occupy a consciousness that exists solely to be destroyed. Because the viewpoint has no future, it has no want beyond survival, no relations the book will develop, and no properties the writer has any reason to individuate. What is presented as a person is a camera position with a pulse, and readers detect the substitution immediately — not as implausibility but as a demand for feeling that has not been earned.

The secondary mechanism is exhaustion. The type has been so heavily used that the label now predicts the contents (§3.6), and a device that predicts its own contents has stopped functioning as a device.

There is also an ethical dimension that a textbook should name rather than avoid: the convention has, over several decades, produced an enormous quantity of fiction in which a woman’s terror is the standard unit of narrative interest, deployed before she has been given a name. That is not a craft objection, and it is not this chapter’s business to settle it, but a writer choosing the type should know they are choosing it.

Persistence. It solves a real structural problem. A detective novel’s first chapters are procedural and low in immediate stakes; the victim prologue supplies urgency, a face for the crime, and a reason to care about a case file. It works well enough, often enough, to remain standard.

Remedy. Two options. Individuate: give the victim a want unrelated to their death, a relation the book will pursue, and a specific that returns later — at which point they are a person and the type is no longer the failure. Or substitute (§6.4): the last ordinary day (type 14) delivers dread without a death, and the antagonist’s vantage (type 3) delivers threat without asking for empathy the passage cannot repay.

The exception. Sebold’s The Lovely Bones (2002) opens with a fourteen-year-old girl stating her name and the date of her murder — and then narrates the entire novel. The victim is not a camera position; she is the book’s consciousness, its voice, and its subject. Every objection above dissolves because the investment the opening solicits is repaid for three hundred pages.

Martin’s series prologues take the opposite route to the same result: the viewpoint character’s death becomes a declared convention, applied at the head of every volume, so that the reader reads each prologue knowing the vantage is temporary. Making the type’s cost explicit converts it from a failure into a formal signature.

The discrimination. The pattern fails when the reader is asked to invest without repayment. It succeeds when the investment is repaid — either by keeping the voice, or by telling the reader in advance that it will not be kept.

8.8 The Mythic Overture Nobody Can Follow

Symptom. A cosmogonic or legendary prologue containing a dozen proper nouns, three gods, two ages of the world and a prophecy, none of which the reader can attach to anything.

Mechanism. A comprehension failure at the level of the situation model (Chapter 9, §3.3). Readers track a small number of dimensions — protagonist, space, time, causality, intentionality — and a mythic overture typically populates none of them. Unlike the productive strangeness of an unexplained noun inside a scene, which Chapter 9 (§8.3) shows to be an asset, a mass of unexplained nouns with no scene to hold them is not withheld information; it is unstorable information. The reader has nowhere to put it.

The aggravating factor is the reader’s correct suspicion that they will be tested on it. A prologue full of names reads as an instruction to memorise, and the instruction is resented because it arrives before any reason to comply.

Persistence. The material is necessary in some books, and the overture is where it has traditionally gone. The writer’s alternative — distributing a cosmology through the narrative — is much harder and much slower to draft.

Remedy. Reduce to one. One name, one event, one image; everything else distributed. A mythic overture that establishes a single unforgettable fact — a war, a betrayal, a thing buried — buys more than one that establishes forty.

The exception. Beowulf’s exordium opens with a genealogy: Scyld Scefing, the foundling who came over the water, his funeral ship sent out laden, his descendants named in sequence, before the poem’s own hero is mentioned. It is unfollowable by a first-time reader, dense with names that will not recur, and it is one of the great openings in the language.

The reason is that the passage is not asking to be followed. It is asking to be felt — the names function as evidence that this world has depth behind it, and their unfollowability is part of the effect. This is a different mode of reading from the one modern narrative usually solicits, and it survives in the epic register.

The transferable discrimination: an overture the reader cannot follow succeeds when following it is not required, and fails when the book later behaves as though the reader was taking notes. A prologue that names forty things and then depends on the reader remembering three of them has committed the error. One that names forty things purely to establish depth has not.

Dune’s epigraph apparatus (1965) makes the same point in a modern register: excerpts from an invented future historian, unexplained, prefixed to every chapter, establishing a documented civilisation. Nobody is expected to retain them; the effect is the existence of the archive rather than its contents.

8.9 The Prologue Whose Voice Never Returns

Symptom. A frame narrator, an editor, or a distinctive prologue voice is established and then disappears without acknowledgement, the narration silently reverting to a different mode for the rest of the book.

Mechanism. This is unresolved distance (§8.4) on the viewpoint axis specifically, and it is separated out because the viewpoint axis is the most expensive to leave open (§6.1). A voice is a contextual frame with a person in it; readers bind to voices faster and harder than to places or times, and an unannounced disappearance registers as a formal inconsistency rather than as a loose end.

The failure is aggravated when the abandoned voice made a claim to authority (§4.6). If a prologue establishes that the following account was told to someone, and the following account then contains scenes that person could not know, rendered in a voice that is not theirs, the book has quietly withdrawn its own guarantee. Readers rarely name this; they report the book as “uneven”.

Persistence. Frames are easier to open than to sustain. A writer who wants the authority a frame confers may not want the constraint it imposes, and the path of least resistance is to open the frame and then write freely.

Remedy. Three options, in descending order of ambition. Sustain the frame — return to it at intervals, which also produces fragmentary closure. Close it once, at the end, converting the structure to total closure. Or remove it, which usually costs less than writers fear.

The exception. James’s The Turn of the Screw (1898). The novella opens at a country-house fireside on Christmas Eve: a company telling ghost stories, a man named Douglas who has a manuscript, an unnamed narrator who receives it. Then the governess’s narrative begins — and the frame never resumes. Douglas does not comment. The fireside is not revisited. The novella ends inside the governess’s account, with no return to the people who were listening.

This is unmistakably deliberate, and it is the source of a great deal of the story’s power. Returning to the fireside would supply an outside judgement on a narrative the story exists to leave unadjudicated; a listener saying and that was the end of it would settle a question James spent his career refusing to settle. The frame is left open because closing it would answer the book.

Compare the crux Chapter 9 (§4.6) discusses in Madame Bovary: the first-person-plural schoolboy narrator who observes Charles’s arrival and then vanishes. That is the same manoeuvre performed inside a first chapter rather than a prologue, and it remains contested — flaw, residue, or device. The comparison is worth teaching, because it shows how much of the judgement depends on whether the disappearance can be read as motivated.

8.10 The Prologue Written to Fix a Slow Chapter One

Symptom. A prologue that exists to supply excitement, tension or a hook that the opening chapter lacks. Frequently it is the last thing written, and frequently it is written after a reader has said the book “takes a while to get going”.

Mechanism. A misdiagnosis dressed as a solution. The problem — Chapter One does not compel — is real. The remedy addresses the symptom (the reader’s early boredom) rather than the cause (an opening that does not open the contracts), and it does so by adding material rather than by fixing the defective material.

Two costs follow. First, the underlying defect survives and is now harder to see, because the prologue’s energy masks it in every subsequent reading. Second, the prologue is by construction a pure reader-contract unit (§5.4’s floor): it serves one domain, and the domain it serves is the one most easily served from inside. It will therefore fail tests 1, 2 and 4 simultaneously.

Persistence. It works, in the narrow sense that early readers stop complaining. It is also cheap: writing a thousand exciting words is easier than rebuilding an opening chapter. And it is licensed by widespread advice about hooks.

Remedy. Run Chapter 9’s protocol on Chapter One with the prologue removed (test 9, question 1). If Chapter One fails Chapter 9’s tests 2 and 3 — no scene, no momentum — then the prologue is a bandage and the wound is elsewhere. Fix Chapter One. Then re-ask whether the prologue is still wanted; sometimes it is, on other grounds, and often it is not.

The exception. Frankenstein (1818). Walton’s letters exist, among other reasons, to make Victor’s long and undramatic Genevan childhood bearable. Chapters describing a happy family, an education, a friendship and a set of scientific enthusiasms are slow by any standard, and Shelley underwrites them with an opening in which a man is stranded in Arctic ice, driven by an ambition he half recognises as dangerous, and confronted with a dying stranger pursuing something across the floes.

That is a prologue buying patience for a slow opening, and it is the correct use of the reader-contract domain rather than an abuse of it. The distinction is that Walton’s letters also serve the plot domain (the unwitnessable frame), the character domain (a listener whose ambition mirrors Victor’s), the theme domain (the pursuit of knowledge past its proper limit) and the authority domain (the whole chain of transmission). The prologue that fixes a slow Chapter One fails when that is all it does.

8.11 The Prologue That Is Really an Epigraph

Symptom. A short, non-narrative prologue — an aphorism, a fragment of invented scripture, a paragraph of atmospheric assertion — that makes a thematic claim and contains no event.

Mechanism. An instrument mismatch. The material is doing an epigraph’s work: standing at the threshold, making a claim by juxtaposition, and asking to be remembered rather than followed. Labelling it a prologue makes two implicit promises it cannot keep — that it is narrative, and that the narrative will return to it — and thereby converts a functioning small device into a failing large one.

The reader’s experience is a specific mild disappointment: they brace for a story, receive a proposition, and begin the book already slightly misaligned.

Persistence. The material is often good, and “prologue” is the only label a writer feels licensed to use. The alternatives — an epigraph, an untitled italicised page, a part-title — are less familiar and feel less substantial.

Remedy. Change the label rather than the words. Most of these passages improve immediately as epigraphs, as untitled openings, or as the first paragraph of Chapter One. Where they are too long to be an epigraph and too static to be a chapter, the honest question is whether they are needed at all.

The exception. Middlemarch’s “Prelude” (1871–72). Under nine hundred words, no character from the novel, no event, no scene: a historical figure offered as a type, a question posed about later-born women whose ardour finds no adequate object, and a general proposition about the difficulty of ordinary heroism. By every criterion above it is an epigraph with delusions of scale.

It works, and its working identifies the conditions under which the pattern is not a failure:

1. It is answered by a matching unit at the far end. The “Finale” returns to the Prelude’s figure and its question. The structure is a frame, not an ornament, and the reader who reaches the last page is being paid.
2. It establishes the narrator’s characteristic move. Eliot’s narration constantly shuttles between the particular case and the general law; the Prelude teaches that habit before the first instance of it.
3. Its question is open. It does not tell the reader what to conclude. It asks something the novel spends eight hundred pages failing to settle.

The discrimination. A non-narrative threshold succeeds when it is framed — answered later, and formally acknowledged — and fails when it is merely placed. Bulgakov’s Faust epigraph and Tolstoy’s “Vengeance is mine” both succeed as epigraphs because they do not claim to be narrative; Eliot’s Prelude succeeds as a prologue because the Finale makes it one.

8.12 The Threshold Pile-Up

Symptom. The reader must pass a map, a family tree, a dramatis personae, an epigraph, an author’s note, a foreword and a prologue before reaching Chapter One.

Mechanism. Genette’s threshold, over-furnished. Each item is individually defensible and the aggregate is a wall. Three costs compound: the reader’s start-up investment (§1.1) is paid several times over before any narrative begins; the items compete, so that the prologue’s privileged position (§3.9) is diluted by everything sharing it; and the accumulation signals anxiety — a book that explains itself seven times before beginning suggests it does not trust itself to be understood.

Persistence. Each item is added separately, by a different person, for a good local reason: the author wants the epigraph, the editor wants the map, the marketing department wants the author’s note, the series requires the family tree. Nobody is responsible for the total.

Remedy. Count the pages between the half-title and the first line of narrative. Then ask, of each item, what a reader who skipped it would lose. Most books survive the loss of most of them. Where several must stay, distribute: a family tree is more useful at the back, and a map is more useful where the journey starts.

The exception. Dune (1965) surrounds its narrative with an epigraph system attributed to an invented historian, appendices on ecology, religion and a noble house, a glossary of imperial terms and a cartographic apparatus. The Lord of the Rings (1954–55) prefixes a foreword, a prologue in several parts and a note on the sources, and appends a hundred pages of chronology and linguistics. The Posthumous Memoirs of Brás Cubas (1881) opens with a dedication to the worm that gnawed the author’s corpse, a note to the reader, and a narrator writing from beyond the grave.

In each case the apparatus is not surrounding the book; it is the book — a claim about the density of a world, or a joke about the machinery of publication, sustained from the first page to the last. The pile-up fails when the items are precautions. It succeeds when they are the work.

8.13 Summary: symptom to mechanism