Back to the table

107 Intermediate Rh Red Herring Plot & Beats

Red Herring

A false clue that points readers the wrong way.

What it is

A red herring is a clue that points the reader toward the wrong answer. Most people assume the clue has to be untrue. The best ones are the opposite: a fully true fact, a real secret or a real lie or a real coincidence, that fits an innocent reading as well as a guilty one, and that the reader interprets the way you hoped. That one distinction is what separates a clever ending from a cheap one. A reader who was misled by a true clue closes the book satisfied with their own effort. A reader who was misled by an invention the author later takes back feels handled, and says so in reviews. Every false lead you plant is also a promise to explain it later. So the whole craft comes down to one rule. Never plant a falsehood. Plant something true, aimed wrong, and make sure that afterward the reader can see exactly where they slipped.

The textbook definition, information that steers the reader to a wrong conclusion, turns up everywhere and is close to useless. It names the effect and skips the mechanism. Worse, it implies a lie that gets retracted, and a red herring built that way fails. The device's bad reputation, cheap, mechanical, a cheat, comes from writers building it exactly like that.

So start with the one distinction that underpins everything else here. A lie is an untrue statement made by the book. The maid says she saw nobody on the stairs, and later it turns out she did. The timetable says the last train leaves at ten, and later it leaves at eleven. In both cases the reader's logic was fine. The evidence was confiscated. There was nothing to solve, and the author overruled the reader instead of outwitting them.

A red herring is true material whose meaning the reader misreads. Take Netta Burrage, the bookkeeper in the invented case that runs through the rest of this explanation. She was at the mill at midnight: true. She told the police a falsehood about it: true. She burned a page from a ledger: true. Every one of those facts is still true on the final page. What shifts is what the facts refer to. She has spent two years selling flour off the books to pay for a brother in care, and she burned the page as discovery closed in. A frightened thief behaves exactly like a killer. Set the two cases side by side. In the herring case the reader reasoned soundly from real evidence and was beaten by the fact that people carry more than one secret, which is the intended pleasure of a mystery. In the lie case the reasoning never mattered, because the author was never competing.

Here is a working definition you can build with. A red herring is evidence the book gives truthfully, and that an innocent explanation accounts for. Your job is to provide the evidence and let innocence look like guilt for as long as the story needs. No lying is required. One refinement: a character may lie, and the lie still counts as true evidence. A lie is an actual event with an actual cause, and the cause is what the reader has to solve. When Helen Fielding's Daniel Cleaver gives Bridget his version of events, the words are false and the clue is true: a man said this, for reasons of his own. The rule, then, is that honesty belongs to the narration, whatever the dialogue says. Characters may deceive one another at will. The book may never deceive its reader.

Everything so far concerns the fairness of a single clue. Whether the whole twist plays fair, the contract between writer and reader and the test of rereading, belongs to Plot Twist. For one clue, four conditions settle it.

Condition one: the clue must be explicable. There has to be a real, specific cause you could state. "To mislead the reader" is the one cause that disqualifies. If you cannot write the innocent explanation in one sentence before you plant the clue, you are planting a dangling thread, not a herring.

Condition two: the truth must be reachable. A reader with a notebook could have worked it out. Could, not would. Every piece of data they would need has to be on the page, even if the pieces sit three chapters apart and nothing connects them. A low bar, and the one most failures miss.

Condition three: the narration stays silent on the false reading. A character may voice it. The detective may voice it and be wrong. Readers may reach it themselves, which is the best of the three options, because nobody disputes their own conclusion. What the narrative voice may never do is state the false reading as fact. The moment it does, the clue has become a lie.

Condition four: the reader must eventually learn what the clue really was. No speech is required, and no great detail. But no explanation at all is the most reliable way there is to cheat a reader.

Test one and four first. They catch most faults. Two and three catch the rest.

Nearly every red herring in print works in one of four ways, and knowing which one you are using tells you what it needs. One invented case will serve for all four. Eustace Trenholm has drowned. His body was found in the wheel race at Scarrow Mill, a tide-driven mill that still operates on the Gaddon estuary. Dinah Crake, the coroner's officer, is investigating.

The first mechanism is an innocent person with a guilty-looking secret, and of the four it lasts best. Netta Burrage's alibi fails. She has a motive she refuses to discuss. She goes white when the police arrive. All of it is real. The cause is the flour theft, which started when her brother went into care. She acts guilty because she is guilty, of something else. The reader needs no push at all.

The second is a true fact in a false frame. Here the clue is a pattern or a sequence, a shape the reader imposes on the facts. Scarrow Mill runs only on the ebb, for roughly five hours either side of low water. Crake has four weeks of tide data and the names of everyone who was on the causeway that night. With them she shows that Trenholm went into the race while the wheel was turning, which fixes the time of death to the window when it turned. Her deduction is correct and the case rests on it. The unproven assumption everyone makes is that he was at work. In truth he was waiting for a visitor arriving by boat, which was possible only on that tide. The frame, "this is a work accident, or a murder staged as one", was built by the reader out of accurate data. When it breaks, nothing becomes false. Everything simply means something else. Of the four, this one is the most elegant and the hardest, because you have to know the true shape of the story before you can build the false one.

The third is a coincidence mistaken for design. Romilly Halsey was on the causeway at a quarter past two in the morning, an absurd hour. She was there to meet a married man from the next village, which has nothing to do with Trenholm, and she will tell nobody. The reader sees presence at the scene plus the hour and infers intent, because fiction has trained them to expect reasons. That makes the coincidence unusually strong and unusually risky. A coincidence nobody acknowledges reads as carelessness. The fix is to have a character remark on it. If Crake says aloud that two things happening at once are either the whole case or nothing at all, the reader can accept "nothing at all" when it arrives, because the option was put on the table openly.

The fourth is the suspect the reader wants convicted. Rafe Pledger, a developer, wants the mill site for twelve houses. He was seen arguing with Trenholm in a car park. He was rude to a woman in a shop in chapter four. Nobody accuses him. The reader convicts him on character and enjoys the prospect of his punishment. This mechanism runs on appetite rather than evidence. Of the four it is the easiest to set up and the most resented when it fails, because the reader's taste was played rather than their reasoning. Done well, it hides the culprit behind the reader's own preferences, which none of the other three can do. Done lazily, it gives you an unpleasant suspect who is nothing but unpleasant, serves no other purpose, and gets cleared in a paragraph. For fairness, Pledger needs a real role in the plot beyond being disliked, and his unpleasantness should cost him something besides the charge.

The types combine. The commonest strong false lead is a type-one character inside a type-two structure. They are not interchangeable, though. A single disclosure clears a type one or a type three. A type two requires re-explaining the whole shape of the book, so it belongs at the climax, never two-thirds of the way through.

If you take one technique away from all this, take the secret that is not the crime. Of the four it is the most useful, and it costs almost nothing in honesty. The recipe: give a character a real secret, serious enough to hide and unrelated to the main crime. Then let the plot put pressure on them. Everything else follows by itself. They evade. They lie about where they were, because the true location incriminates them for the wrong reason. Their alibi collapses, because it was covering the secret. They behave guiltily all the way through, and they are guilty: of adultery, of theft, of a lapsed qualification, of an abortion, of an immigration problem, of drinking, of a secret child, of money taken from a charity account eleven years ago. The reader deceives themselves by sound logic, and at the reveal they can blame nobody but themselves.

Five things make this mechanism clean. There is no authorial lie, because every scene truthfully records fear. It survives rereading, because fear of exposure replaces fear of arrest and the scenes get better. You get characterization for free, because a real secret brings history, inner life and pressures with it, so the herring doubles as supporting cast work. It clears itself, because exposing the secret dissolves the lead with no separate "and she was innocent all along" scene needed. And the clearing scene is worth having regardless. A decent person confessing the lesser, shabbier truth to escape the greater charge works in any genre.

The only requirement is proportion. The secret must justify concealment under suspicion of murder, and that is a high bar. A parking fine is not worth risking a murder charge to hide. A secret that would not survive police questioning makes the concealment unbelievable. The false lead turns into irritation, and the character looks stupid purely for the author's convenience.

A red herring gets confused with four of its neighbors. A red herring is not a plant. A plant is an early detail that makes a later turn feel prepared rather than arbitrary: the true clue, working ahead, visible. A herring is a plant aimed at the wrong target. The construction is identical. Foreshadowing owns plants, Chekhov's gun, placement, repetition, and how to make a detail memorable without making it conspicuous. A red herring is not the reveal, either. Learning the answer, structuring it, and surviving a reread all belong to Plot Twist. The herring feeds the reveal and stops before it. Nor is a red herring a subplot. A false lead can look like one, with chapters of its own and a small arc. But a subplot is a second storyline with its own stakes and its own resolution, and a herring is a misreading of the main line. The two can overlap, as when an innocent suspect's secret carries real emotional weight, and then you are writing both. Pacing a subplot is that page's business. Last, a red herring is not merely a secret. Every interesting character withholds something. A secret becomes a herring only when the plot invites the reader to link it to the central question. The device is the link. The secret is the raw material.

How many should a book have? Too many train the reader to disbelieve, and once that happens the true clues fail to register. You can watch it happen. Theorizing is participation, and participation is most of the pleasure of the form. Each time a theory is demolished, the reader builds a revised one. After two or three demolitions they are still revising. After six or seven they stop, having correctly judged that theorizing is futile. Passive reading follows. They wait to be told. No surprise is possible then, because there is no expectation left to overturn, and the force of your true clues has been spent on the false ones.

So, as a guideline and never a law: in a full-length novel, three to five false leads that the reader really chases, each with its innocent explanation named and a page where that explanation arrives. Under three feels thin. One suspect gets cleared and the culprit is found by elimination. Over five or six brings on the give-up effect. Short crime fiction has a budget of one or two. A long multi-strand book can hold more, with each strand budgeted separately.

Counting is crude, though, so here are two sharper checks. The first is a ratio test. Tally the false leads. Then tally the true clues that survive to the end and bear the solution. If the false ones run beyond roughly twice the true ones, the book is mostly misdirection, and the solution will feel unsupported however well it is made. The true clues are what the reader is meant to end up holding, so supply enough of them. The second check is the theory test, and it is the best of all. Ask a beta reader to note a suspect and a reason every few chapters. Accuracy does not matter. The signal is that they keep answering. A note at the halfway mark that says "no idea, I've stopped guessing" proves the ratio is wrong. Beta Readers covers how to ask without leading.

Every false lead is a debt. The reader carried it, pondered it, and perhaps bored their family with it. Payment is due. The failure has a name, "never mentioned again", and it is exact and frequent. A lead is planted, pursued, loaded with significance, then dropped. No scene, no line, no nod. The book proceeds to the true answer and the lead vanishes. Readers call this a plot hole. The label is imprecise, since the plot is intact, but the instinct is correct: a promise was broken.

Here is why it stings. Told the innocent explanation, the reader replays their reasoning and finds the wrong step. That replay is a pleasure, like being shown how a card trick was done. Untold, there is no replay. The reader is left with a loose fact and a suspicion that the author forgot, a suspicion that is usually justified and hard to rebut.

Payment scales with the weight the lead carried. A single-scene oddity costs half a sentence or one line at the end, something like: "Romilly Halsey had been on the causeway to meet a married man from the next village, which she would rather have been charged with murder than admit." A suspect the reader chased for chapters costs a scene or a proper exchange, with a price paid by the character. A flat explanation in reported speech is paid on paper and emotionally a shrug. A structural misreading, the type-two kind, costs a re-explanation inside the climax, never a post-climax wrap-up, because re-seeing the shape of the book is itself the payoff.

Explanations may be dull. Sometimes they are dull on purpose, and dullness is often the point. They need not come through the detective. The best come from the innocent party's own mouth, which turns paperwork into a scene with someone's dignity at risk. All of this gets checked during Revision. The audit under "Try it yourself" below is that pass.

Why it matters

Readers who are only told are passive, and passive reading is slow. A reader who is theorizing leans forward. The dramatic question supplies the wanting-to-know, and the false leads supply the doing. That two-way engagement is what separates a book that is merely read from a book that is devoured until two in the morning.

Misdirection protects the ending. You have known the solution for months. Your reader has read the genre widely. Unless something distracts them, they reach the obvious answer by page sixty. Misdirection is not an ornament laid over a finished plot. Without it, there is no way to delay the answer until it can land as a discovery.

In an investigation plot, the dread starts here. No candidate for suspicion means no dread, because dread needs someone to be suspicious of. A suspect who might be guilty frightens more than a murderer the reader has not met yet. The tension page builds on this.

Honest misdirection sells the next book. Series readers buy an author's way of handling false leads more than they buy any one book. Honest misdirection earns forgiveness for a slack middle. One withdrawal of evidence changes how that author is read from then on, and by fewer people, because that is the complaint reviews actually record. Reader Promise makes the larger case.

"Mysteries require them" is true, and far too small. Most writers reading this work in other genres, where misdirection is already at work without a name. Outside crime, the usual subject is motive rather than fact: the events are right and the reasons are wrong. Horror misleads about what the threat wants. Romance misleads about a rival's reappearances. Literary fiction misleads about why a family broke, and the false reading is commonly the family's own. The mechanism is identical. Only the hidden thing changes. The examples below include non-mystery cases on purpose, because the mystery version dominates the craft literature and is not the only one.

The clearing scene can beat the misunderstanding it clears. A short invented sketch shows the move outside crime. Vesta Pinnock has moved back to her hometown. Her ex, Ferris Mowbray, keeps turning up: at the gas station, at the school gate, at her mother's house. On the evidence he looks like an ex who cannot move on, and that reading is reasonable. The truth is eight chapters of failed attempts to tell her that her mother has been admitted to the hospital twice, news the mother swore him to keep from her. Every sighting is real. Every one means the reverse of how it looks. And the scene where Vesta learns this beats the misunderstanding that led up to it.

Real-world examples

William Cobbett's Weekly Political Register (1807): the word's own origin story is a red herring

Every account of the phrase opens the same way. Hunters, poachers or hound trainers dragged a smoked fish across a trail to throw the dogs off. Good story. There is no evidence it was ever done. What is on record is the fish itself, smoked to a reddish-brown and powerfully scented. The plain sense of "red herring" is very old: Anglo-Norman verse in the late thirteenth century, Middle English in the early fourteenth. Early references to the fish and hounds do exist, and none is about diversion. Thomas Nashe's 1599 pamphlet, written in mock praise of the herring, claims the fish's skin is the best thing there is for getting hounds onto a scent. That describes a training drag, the reverse of a diversion. Around 1697 Gerland Langbaine's horsemanship tract used a red herring as a stand-in lure when teaching horses to follow hounds. That passage was later bound, without credit, into an edition of Nicholas Cox's The Gentleman's Recreation, where Cox's framing turned it into advice on training dogs. Diversion is absent from all of it.

The figurative sense has one source and one date. William Cobbett, the radical journalist, wrote in his Weekly Political Register of February 14, 1807 that as a boy he had dragged a red herring across a hare's trail so the pack lost it. He used the picture against the English press, which had chased a bogus report of a French defeat instead of the real news. He told the story again in 1833. He probably made it up for the argument. The founding instance of the phrase is itself a red herring.

The scholarship is settled. Professor Gerald Cohen and Robert Scott Ross traced the figurative sense to Cobbett in two articles in Comments on Etymology in October 2008. Michael Quinion, the etymologist behind World Wide Words, backed the finding, and the Oxford English Dictionary accepted it. Quinion checked half a dozen fox-hunting histories and found no mention of a red herring laid as a false scent. The supposed custom looks reverse-engineered from Cobbett's metaphor, mistaken for reportage, and relayed in good faith as "fact" ever since.

Two lessons for a writer. First, treat inherited craft lore with suspicion. A great deal of it is confident assertion with nothing behind it, and a tool's own etymology is a cheap place to practice checking. "Cliffhanger" is the parallel case: its supposed literary origin is an offhand remark by a twentieth-century critic, as the Cliffhanger page explains. Second, Cobbett's picture misrepresents the technique. A dragged fish is an alien object inserted solely to deceive and never explained afterward, which is the exact kind that fails. Effective herrings are hares: real creatures with real reasons to run, that merely are not the quarry.

Silver Blaze (1892): the shifty stranger who did nothing worse than his job

Arthur Conan Doyle's "Silver Blaze" appeared in The Strand Magazine in December 1892 and was collected in The Memoirs of Sherlock Holmes the next year, 1893. For the first mechanism, the innocent person with a guilty-looking secret, there is no cleaner short example in English, and you can read it in one evening.

A racehorse is missing. The trainer, John Straker, lies dead out on the moor. A stranger, Fitzroy Simpson, is under arrest. He was seen near the stables that night. He had been asking about the horse. He had visited both yards. Nothing about Simpson is false. He is a bookmaker's agent on a disreputable errand, and he is evasive because the errand is disreputable. There was no murder.

The solution has been common knowledge for well over a century, but skip the rest of this paragraph if you want to meet it fresh. There is no murderer. Straker led the horse onto the moor intending to lame it for a betting swindle. He struck a match behind the animal, and a kick from the horse killed him. The victim is the criminal. The survivor under arrest is a small-time crook in the wrong place at the wrong time. The misdirection is made entirely from Simpson's real circumstances. Doyle hides nothing about him and asserts nothing false. A shabby secret beside a corpse, and the police and the reader do the rest.

The same story holds the dog that did not bark, the best-known true clue in the genre. A negative clue, fairly given, in plain view, free for anyone to notice. One tale with a true clue nobody sees and a false one everybody chases makes an instructive pair. That clue belongs to Foreshadowing. One irony of dates: thirty-six years later, in 1928, Van Dine's rule 20 listed ten devices beneath any self-respecting writer in the genre, and the silent dog was among them. Invention to cliché inside a working lifetime. Whatever you plant today may age the same way.

Van Dine (1928) and Knox (1929): the two rulebooks, stated in full and weighed

Golden-age crime fiction produced two celebrated sets of rules. They are always quoted piecemeal, so here they are stated accurately, dated, and then judged.

S.S. Van Dine published "Twenty Rules for Writing Detective Stories" in The American Magazine in September 1928. Van Dine was the pen name of Willard Huntington Wright, an American critic and the creator of the detective Philo Vance. His premise was that detective fiction is a sport, and the writer owes the reader fair play. Four of his twenty rules matter here. Rule 1: "The reader must have equal opportunity with the detective for solving the mystery. All clues must be plainly stated and described." Rule 2: "No willful tricks or deceptions may be placed on the reader other than those played legitimately by the criminal on the detective himself." Rule 10: "The culprit must turn out to be someone who has played a reasonably prominent part in the story: a person the reader knows and has an interest in." Rule 15: "The truth of the problem must at all times be apparent, provided the reader is shrewd enough to see it."

Rule 2 is the most useful sentence ever written on the subject, and it sits close to the distinction this explanation opened with. Deception is licensed to the criminal, inside the story, against the detective. There is no equivalent license for the writer, who sits outside the fiction and faces the reader. Murderers may plant false clues. Novelists may not.

Ronald Knox followed a year later. Knox was a Catholic priest, a Bible translator, and a detective novelist. His ten rules appear in the introduction he wrote for The Best Detective Stories of the Year 1928, an anthology he edited with H. Harrington, published by Faber & Gwyer in London in 1929, and in New York by Horace Liveright as The Best English Detective Stories of 1928. The essay is dated February 28, 1929, and was later reprinted as "A Detective Story Decalogue". Here are the ten in short form. Rule 1: "The criminal must be someone mentioned in the early part of the story, but must not be anyone whose thoughts the reader has been allowed to follow." Rule 2: "All supernatural or preternatural agencies are ruled out as a matter of course." Rule 3: "Not more than one secret room or passage is allowable." Rule 4: "No hitherto undiscovered poisons may be used, nor any appliance which will need a long scientific explanation at the end." Rule 5: "No Chinaman must figure in the story." Rule 6: "No accident must ever help the detective, nor must he ever have an unaccountable intuition which proves to be right." Rule 7: "The detective must not himself commit the crime." Rule 8: "The detective must not light on any clues which are not instantly produced for the inspection of the reader." Rule 9: "The stupid friend of the detective, the Watson, must not conceal any thoughts which pass through his mind; his intelligence must be slightly, but very slightly, below that of the average reader." Rule 10: "Twin brothers, and doubles generally, must not appear unless we have been duly prepared for them."

The problem Knox was addressing was real. In the 1920s detective fiction sold in enormous numbers, and its appeal was competitive: reader against detective. The contest means nothing if the author can pull out a twin never mentioned, a poison unknown to science, or evidence the detective kept to himself in chapter three. Rules 6, 8 and 10 share one purpose, a fair contest between reader and detective. Rule 4 is different but just as practical. A solution that needs a page of chemistry is unreachable, and so unfair even when nothing was hidden.

Now the judgment. These are period documents, and treating them as law harms books. Good books break several of them. Knox's rule 2, no supernatural, excludes every modern paranormal mystery and much of the horror-crossover crime shelf. Van Dine's rule 3 forbids a love interest, the job being to convict someone rather than pair anyone off. Today's crime fiction regards that as plainly wrong. The detective's private life is one of the main reasons readers come back. Van Dine's rule 11 bars a servant as the culprit, which is a statement about which characters count, a class assumption rather than a fairness rule. And Knox's rule 1 had already been broken, three years before he wrote it, by Agatha Christie, in a book now routinely voted among the best crime novels ever written. That argument lives on the First Person page, which owns the unreliable narrator.

One rule is plainly racist. Knox's rule 5 is indefensible as worded. His own commentary supplies context: he said it was aimed at a cliché rather than at people, the sinister Oriental villain of the pulp thrillers, whose appearance signaled that plotting had been abandoned, and that there was "no reason in the nature of things" why such a figure should spoil a detective story. Context does not redeem the sentence. A useful side note: roughly half of the ten, four or five of them, amount to a catalogue of devices Knox was sick of, with fairness not involved at all. That kind of rule dates quickest and ugliest.

The rules also carry more authority than they ever had. They are routinely called the bylaws of the Detection Club. The club was founded in 1930, with G.K. Chesterton as its first president. Its initiation oath, written by Dorothy L. Sayers, binds the members' detectives to rely on their wits and forswear divine revelation, feminine intuition, mumbo-jumbo, jiggery-pokery, coincidence, and act of God. Members treated Knox's principles as guidelines, not statute. Some ignored them. The fair-play rules relaxed considerably over time.

The famous attack came from Raymond Chandler in "The Simple Art of Murder", in The Atlantic Monthly of December 1944. His complaint was not unfairness but lifelessness. Obeying the rules produced books of contrivance and formula, with no way past either. He rated Philo Vance the most asinine character in the genre, and he took apart A.A. Milne's The Red House Mystery at length as a puzzle that collapses once you read it as human conduct. Chandler was arguing for his own American school, so he is not impartial. His objection still stands, though. Tidy deductions do not make a novel.

So are the rules binding? No. Nobody serious has treated them that way for seventy years. Two of them still earn their keep, and both are about clues rather than devices: Van Dine's rule 2, on who may deceive whom, and Knox's rule 8, the detective's clue is the reader's clue. The rest are 1920s taste, interesting as history and harmful as instruction. The Cliffhanger page applies Knox's rules 1 and 9 to one narrow matter, the honesty of a chapter break that withholds something. Used that way they are a precise tool for a specific problem, never a code.

The A.B.C. Murders (1936): a true pattern in a false frame

Agatha Christie's The A.B.C. Murders was published by the Collins Crime Club in London on January 6, 1936, and by Dodd, Mead in New York on February 14, 1936. The second mechanism, a true fact in a false frame, is stretched across the entire novel.

The premise: victims die in alphabetical sequence. Alice Ascher in Andover. Betty Barnard in Bexhill. Sir Carmichael Clarke in Churston. A railway guide is left at each scene. All of it is true, and it stays true. The pattern is real. There is no error, no coincidence, no hoax. The pattern is the red herring.

The twist is among the best known in the genre, but skip this paragraph if you have not read the book. The alphabet exists to disguise one ordinary-motive killing as the motiveless series of a maniac. Franklin Clarke murdered his brother Sir Carmichael to inherit, and the other deaths are camouflage. The book runs a human false lead too: Alexander Bonaparte Cust, a traveling salesman with blackouts from a head wound, set up with the typewriter, the railway guides and the weapon, and so unsure of his own movements that he half believes he did it.

Watch how the two leads work together. Cust is a standard frame-up and unremarkable on his own. The pattern is the real feat, because it conceals a category rather than a fact. The reader misses no evidence. They simply ask the wrong question, "who is doing this", which has no answer because its premise is false, instead of the question that solves the case, "who wanted this person dead". The whole plot rides on a frame the reader assembled, unasked, from accurate facts.

This mechanism is the hardest and the most rewarding, and it dictates where the clearing can happen. Not in the third quarter. Re-seeing the shape of the book is the climax. Christie's re-frame comes last because no other place would hold it.

We Have Always Lived in the Castle (1962): loyalty running the wrong way

Shirley Jackson's We Have Always Lived in the Castle, published by Viking in 1962, is horror, and it runs one red herring through the whole book without ever posing as a detective story. Six years before it opens, arsenic in the sugar bowl wiped out most of the Blackwood family. Constance, the elder sister, was tried and acquitted. The village is certain she did it, and the reader broadly agrees. The narrator, Merricat, the younger sister, so visibly adores Constance that her defense counts as devotion rather than testimony.

Every supporting fact is true. Constance washed the sugar bowl. Constance never took sugar. Constance was charged. Constance is protecting someone. Near the end, after a fire, with the villagers hunting the pair, Constance admits she had always known Merricat did it. Merricat confirms it instantly. She chose the sugar for the exact reason that Constance never touched it. Nothing is withdrawn. The reader had the loyalty running the wrong way.

Bridget Jones's Diary (1996): a false account instead of a rival

In romance, the harmless rival is the stock false lead. The version that works is a false account, and it needs no rival at all. Helen Fielding's Bridget Jones's Diary, published by Picador in 1996, borrows its structure from Austen. Daniel Cleaver, the charmer, gives Bridget his own account of his rift with Mark Darcy, and the version casts Mark as the villain. The underlying facts are real: a betrayal, a marriage, two men who stopped speaking. The blame is reversed. Daniel slept with Mark's wife. Bridget's feelings change because something old is re-explained, not because something new happens.

The ancestor is Wickham's account of Darcy in Pride and Prejudice, which the Theme page uses for a different purpose. Imitate the re-attribution rather than the triangle. A real grievance pointed at the wrong person is honest, and it buys you an entire act.

Sharp Objects (2006): a false lead that is a complete answer to another question

Gillian Flynn's first novel, Sharp Objects, from 2006, takes the first mechanism to its limit. Camille Preaker, a journalist, returns to her Missouri hometown to cover the killing of two girls and moves back into the house of her mother, Adora. The convention is that the false suspect's secret is bad but legal. Adora's is criminal, and graver than the crime being investigated.

The resolution is widely known, the 2018 HBO adaptation included, so skip this paragraph if you need to. Adora has Munchausen syndrome by proxy. Years earlier she poisoned Camille's younger sister to death. She has been poisoning Amma, the surviving daughter, and she poisons Camille. Adora is a murderer. She is not the killer of the two girls. Amma is.

Here is the payoff. When the reader learns the truth about Adora, they hold an explanation that covers everything: the deaths, the atmosphere, the family. And it is true. Nothing is withdrawn and no clue is retracted. The reader banks a real, enormous revelation as the solution, which leaves room for the true answer to arrive. The false lead was a complete answer to a different question, never a dead end. The lesson reaches beyond crime. A false lead that readers cannot shrug off has to be a real, weighty, fully resolved truth that answers a different question, never a misunderstanding. Readers forgive a wrong guess about the killer. They do not forgive three hundred pages spent on someone who turned out to be nothing.

Everything I Never Told You (2014): the outsider a family needs to blame

Celeste Ng's Everything I Never Told You, published by Penguin Press in 2014, is literary fiction. It opens with the death of a teenage girl, Lydia, and spends the book on why. Jack Wolff, the neighbor, is suspected by the family and by the reader. He has a reputation. He was close to Lydia. He is evasive. Her father is sure he is responsible. All of it is accurately reported, and the evasion and the secret are both real. His love was for Lydia's brother.

The novel's real subject is the motives the family cannot voice. The false lead is there because a family under that much pressure has to blame someone outside itself. The misdirection is self-inflicted by the characters more than installed as a puzzle device. It would occur with or without a novelist having a use for it. Writers outside crime should take that as the point. You do not need to install herrings. They are already present wherever people misjudge one another. The craft is spotting which misjudgments the reader shares, then going back to correct them.

Everyone in My Family Has Killed Someone (2022): total honesty, complete misdirection

Benjamin Stevenson's Everyone in My Family Has Killed Someone is recent, easy to find, and argues the subject openly. It was published in Australia in 2022 and in the United States by Mariner Books on January 17, 2023, and the series has continued. The narrator, Ernest Cunningham, writes how-to guides for would-be mystery novelists. He opens by invoking Knox's decalogue, quotes commandments eight and nine, and binds himself to both: every clue produced, no thought hidden. His pledge is only truth, or what he believed true at the time. He goes further than Knox and lists, up front, the chapters in which someone dies.

Why this matters: the pledge barely restricts him. The title proves it before page one. Everyone in my family has killed someone sounds like rhetoric and is a literal statement of fact. The whole book is made of sentences like that, each one checkably true and each one almost certain to be misread. The narration never lies. The reader builds the misreading out of ordinary assumptions about how crime-novel sentences work.

The book is worth more than a dozen essays on fair play. Total honesty sworn, the rulebook printed up front, the bodies announced in advance, and the misdirection still complete. False leads that depend on the author lying are therefore the weaker kind. The book is also the clearest recent proof that the golden-age rules are now raw material. Stevenson uses Knox rather than obeying him. The novel breaks most of the other eight rules, and it was published, read and enjoyed anyway.

An everyday scenario: the herring that outshines the solution

Say you have finished a first draft. The killer is the business partner, and the motive is a quarrel over a share in a lease. Your chief false lead is Netta Burrage, the bookkeeper: two years of stock theft to fund her brother's care, a false statement to the police about the night of the death, and a chapter near the end, in her kitchen with the kettle boiling, where she confesses all of it.

Three beta readers come back with the same verdict in different words. The ending is fine. The bookkeeper's chapter is the best thing in the book.

Read that as a warning, not a compliment. Her lead has a motive with a price on it, a brother, a wage, two years of small daily dishonesty, and a decent person speaking her shame. The solution has a lease. The herring has become the emotional core of the book, and the main plot has become the admin the reader sits through to get back to her.

There are two real fixes, and weakening Netta is not one of them. The first is to raise the solution to the herring's weight. If the lease really is the subject of the book, give the partner a stake with a price equal to hers. The second, which is more often the right one, is to ask whether you wrote the wrong book. If the liveliest figure in it is the innocent one, she may be the protagonist of the real book, with the murder demoted to a subplot.

Catch this in revision, never after launch. Readers never diagnose it this way. They report a slightly flat ending, then praise the bookkeeper. Joining those two remarks is your job.

Try it yourself

The exercise is an audit. On a full novel it takes about an hour, and it works on an outline, a partial draft, or a complete one.

  1. List every false lead, one line apiece, in the order the reader meets them. Include anything a reader was meant to suspect even briefly: a character, an object, a time, a pattern, a coincidence. A single entry suggests a thin book. Nine means keep going.
  2. Build the herring ledger. Five columns, no blanks. Head them "The false clue", "Page it lands", "Who or what put it there", "The innocent explanation, in one sentence" and "Page the reader learns that explanation". Column five carries the whole exercise. Every blank in it is a review asking "but what about the ledger page she burned?" If you prefer a block per lead, paste this one and fill it in; the mechanism and narration lines get their answers at steps 4 and 7.

False clue: Lands on page: Mechanism: Who or what put it there: Innocent explanation (one sentence): Reader learns it on page: Does the narration ever assert the false reading? Y / N

  1. Separate explicable from merely unexplained. Read each innocent explanation aloud. If what comes out is "that was just to put the reader off", you have no herring. Supply a real cause or delete the entry.
  2. Label each entry with its mechanism. Use one of the four names: secret, frame, coincidence, appetite. If every label reads appetite, your misdirection is powered by dislike and will feel manipulative. If no label reads secret, you are working harder than you need to.
  3. Run the ratio. Count the false leads. Count the true clues that survive to the end, the ones the answer stands on. If the false ones run above roughly twice the true ones, add true clues. Do not cut false ones. The usual real problem is an under-evidenced solution, not excess misdirection.
  4. Flag your favorites. Mark the two entries you like most. They are the likeliest to be overbuilt, because enthusiasm inflates word count. Check whether either one beats the solution for interest, with the scenario above as your model.
  5. Audit the narration. Find each lead's strongest passages and read only the narrative sentences, skipping the dialogue. Reporting, describing and withholding are allowed. Claiming is not. "She had been at the mill to destroy the evidence" is a lie. "She had been at the mill, and she had burned something" is a clue.
  6. Sweep the coincidences. Every entry labeled coincidence needs a character who notices it as one. If nobody does, insert one line from the book's thinker. Unremarked, a coincidence is a mistake. Remarked, it becomes a possibility that can later be dismissed without insulting anyone.
  7. Pay what is missing. For each empty cell in column five, choose a line, an exchange or a scene, according to the weight the lead carried. Place the payment while the question is still in the reader's memory. An explanation that lands forty pages after the reader stopped wondering is paid to no one.

Common mistakes and misunderstandings

The clue no one would have chased. A detail is dropped in. Nobody reacts to it, nobody pursues it, and then at the end it is cleared as if it had been live all along. The result is misdirection that never misdirected, and the resolution pays in pace for a debt that was never incurred. The test: halfway through, ask a reader to name three suspects and whatever is nagging at them most. A clue absent from every list is idle. Strengthen it or delete it.

A suspect is not a herring. A book can have eleven suspects and zero red herrings, if nothing makes the reader prefer one over another. What converts a suspect into a false lead is evidence, meaning a fact that points the wrong way. A line-up of declared suspects is a cast list, and a cast list contains no misdirection at all. Supporting Cast covers the cast's other purposes.

Adding without counting. The commonest structural fault in an amateur mystery is a surplus of herrings, never a shortage. The cause is one-at-a-time additions in revision, each reasonable on its own. Six reasonable additions produce a reader who quit guessing. List what you have before adding number seven.

The character who exists only to be suspected. Some suspects have no function beyond looking guilty and being cleared, and the reader senses the seam. A character with nothing to do is furniture, whatever their dialogue count. Give them a plot reason that outlasts the suspicion: they know a true thing, they want something opposed to the protagonist, they are required in the final chapter. Antagonist handles the kindred fault, the villain whose sole function is being wrong.

The coincidence nobody admits. Readers presume meaning. Two things happening at once read as connected. Denying the connection without ever acknowledging it reads as an authorial shrug. Have someone say it. One sentence turns a liability into a possibility that was raised and dismissed.

The favorite is the problem. Misdirection you enjoyed writing gets overbuilt, and the enjoyment shows as length, detail and jokes. A herring with better jokes and better prose than the solution throws the book off balance and reveals, truthfully, where your heart was. A harsh fast test: given the choice, would you rather be the author of the clearing scene or of the climax?

Dropping a lead is not retiring it. A lead that stops being mentioned has been abandoned, not resolved. Abandoned clues stick in memory far harder than paid ones, because attention clings to unfinished business. The ledger in step 2 catches every one of them.

"Fair" is not "guessable". Some writers take the golden-age rules to mean that readers should be able to solve the book, so they make the solution transparent and the ending becomes a formality. The requirement is reachable, not reached. Van Dine's rule 15, the truth apparent to a reader shrewd enough to see it, does not mean the truth is obvious. The reader to aim for says "it was there and I missed it". The reader to avoid says "I knew by chapter nine".

A herring more sensible than the solution. Sometimes the innocent story beats the true one on plausibility, on motivation, or on emotional satisfaction, and then the ending feels wrong no matter how well it was clued. More foreshadowing cannot help, because the fault is comparative. Write the two motives together on a single page. If the false one is better, the true one is unfinished.

Thinking of this as a mystery-only tool. Writers of romance, fantasy, horror and literary fiction skip the element, and a revision later their readers are still guessing the ending. Outside mystery, what gets misdirected is motive rather than fact, and the misdirection is already there in the characters' misreadings of one another. The task is to find which of those misreadings the reader shares, then go back and settle them.

Related elements