From Surviving Traces to a Finished Novel
About this idea
Learn how to write historical fiction by working in the gap between what the record proves and what a...
On this page
How to get started
course overview: historical fiction craft
- Topic as taught: How to write historical fiction — building a novel in the gap between what the record proves and what a reader will believe.
- Scoping call: Story writing. Historical fiction is treated as a craft course in the historical novel specifically, not general fiction craft and not history as a discipline. General story mechanics appear only where the period setting changes them.
- Audience and level: A motivated adult beginner or advanced high-school student who reads novels and wants to write one. Written at an 8th-grade reading level.
- Assumed prior knowledge: You have read novels. You can write a paragraph. Nothing else.
- Length: 12 units, roughly 18–24 sessions.
- The promise: At the end you can take a documented historical moment and build a scene inside it that is invented, defensible, and does not break the reader's trust.
- Throughline: the reader's attention → the surviving record → the gap between them → the invented mind that fills the gap → the machinery of scene, voice, and research → the promise you make on page one → the argument over whose past this is.
bedrock layer: what cannot BE faked
- Why this layer comes first: Three things are true before you write a word — how reading works, how the past survives, and the fact that your character does not know the ending. Every later choice is a response to one of them.
PHASE 1
the reader's mind IS A one-way channel
- Big idea: A story is not a picture the reader takes in all at once. It arrives one word at a time, into a mind that can only hold a few things. Every other rule in this course is really a rule about what to send and when.
- First principle: A reader receives a story serially and can hold only a little at a time, so every story is an ordered release of information.
- Essential question: What does my reader know at this exact line, and what am I making them wait for?
- Learning targets: Explain how changing the order of information changes a scene's meaning without changing any facts. Track, paragraph by paragraph, what your reader knows. Name one thing you made the reader wait for and say why the wait was worth it.
- Key terms: Serial reading — words arrive one at a time, like a song, not a photo. Working memory — the mental countertop; only a few things fit. Information release — choosing when each fact reaches the reader, like dealing cards. Setup and payoff — a nail hammered in early, a coat hung on it later. Curiosity gap — a hole in the story the reader wants filled.
- Misconception to kill: Good writing means beautiful sentences. A beautiful sentence in the wrong position does nothing; order is the first craft and prose is the second.
- Do this: Take one page of a historical novel you like. In the margin, write what the reader knows after each paragraph. Then move the third paragraph to the top and describe what stops working.
- Evidence of learning: The marked-up page plus five sentences naming the exact effect that disappeared when the order changed.
PHASE 2
the past survives only AS traces
- Big idea: The past is gone. What is left are the things that happened to survive: papers, objects, buildings, bones, stories retold. Someone made each one, for a reason. Most of what existed left nothing at all.
- First principle: Nobody can reach the past directly; every claim about it rests on a surviving trace made by someone with a viewpoint.
- Essential question: Who made this trace, why, and what did they have no reason to write down?
- Learning targets: Tell a primary source from a secondary one and say why the difference matters to a novelist. Name three kinds of people whose daily lives left almost no written record, and say why. Trace one fact you believe about your period back to the actual thing it rests on.
- Key terms: Primary source — made at the time, by someone who was there. Secondary source — someone later, retelling and interpreting the first. Survivorship — only some things last; fire and damp did the editing. Provenance — the paper trail of where an object has been. Silence — a place with no record, which is not proof of nothing.
- Misconception to kill: If it isn't documented, it didn't happen. Most of human life was never written down; a missing record is usually evidence about record-keeping, not about life.
- Do this: Pick one ordinary hour in your period — a meal, a walk to work. List every source that could tell you what it looked like. Then list what none of them can tell you.
- Evidence of learning: The two lists, plus one paragraph on the single most important thing you cannot know.
PHASE 3
they live forward, you read backward
- Big idea: Your characters do not know how it turns out. Your reader does. That gap is the engine historical fiction has and most other fiction does not.
- First principle: The character's knowledge and the reader's knowledge are permanently unequal, and the writer controls the size of the gap.
- Essential question: What does my reader already know that my character cannot, and how do I make that hurt?
- Learning targets: Define dramatic irony and point to it working in a scene. Write a line that means one thing to your character and another to your reader. Explain why hindsight in a character's mouth breaks the spell.
- Key terms: Dramatic irony — the reader knows something the character does not. Hindsight — knowing the ending and letting it leak backward. Contingency — it could have gone another way, and nearly did. Period ignorance — what nobody alive then could have known yet. Foreshadowing — an early hint that only pays off later.
- Misconception to kill: If the reader knows the ending, there's no suspense. The suspense moves from what to how and at what cost; a known ending is a resource, not a leak.
- Do this: Write 200 words set one day before an event your reader knows is coming. Your character may not sense it. Every line should be innocent to them and loaded to us.
- Evidence of learning: The scene, plus one sentence per paragraph explaining the double meaning you built.
deep structure layer: what the gaps make possible
- Why this layer is next: When an incomplete record meets a reader who needs to believe, three pressures appear — where invention is allowed, what actually breaks belief, and the trade between being accurate and being readable.
PHASE 4
the silence IN the record IS your territory
- Big idea: Fiction is not allowed everywhere. It is allowed in the gaps. Where the record is loud, invention becomes a mistake a reader can catch. Mapping the gaps is what separates a historical novelist from someone guessing.
- Core mechanism: An incomplete record divides every scene into three zones — documented, contested, and silent — and each zone permits a different amount of invention.
- Essential question: Is this gap a real silence, or just a hole in my research?
- Learning targets: Sort any scene into documented, contested, or silent. Place invented characters where the record cannot contradict them. State your own rule for when you will change something the record settled.
- Key terms: Documented — separate sources agree on it. Contested — sources disagree and historians still argue. Silent — no source speaks; you are free here. Invention — what you make up to fill a silence. Alteration — changing something the record actually settled.
- Misconception to kill: Research is done when I know a lot. Research is done when you can say which of the three zones each scene sits in.
- Do this: Take your scene list and tag every scene D, C, or S. Highlight any D scene where you are changing something. Write one line defending each change.
- Evidence of learning: The tagged scene list plus a defense of every alteration.
PHASE 5
anachronism IS A mind problem, not A prop problem
- Big idea: Readers forgive a wrong button. They reject a wrong want. Getting the objects right is the easy half; the hard half is that people then wanted different things, feared different things, and could not imagine options that seem obvious now.
- Core mechanism: Belief breaks at the level of motive and assumption, not at the level of props, because a reader tests a character against what a person could plausibly want in that world.
- Essential question: Could this person have wanted this, in these words, for these reasons?
- Learning targets: Spot a modern assumption hiding inside a period character's motive. Rewrite a modern want as the period want underneath it. Make a character sympathetic without giving them your values.
- Key terms: Anachronism — a thing from the wrong time, like a wristwatch in ancient Rome. Mentality — the assumptions a period takes for granted. Presentism — judging the past by today's reflexes. Modern implant — a character who thinks like us, in costume. Plausible outlier — a rare but real person who genuinely did differ.
- Misconception to kill: My hero has to share the reader's values to be likable. Sympathy comes from wanting something, paying for it, and staying consistent — not from agreeing with the reader.
- Do this: List five things your protagonist wants. Cross out any want that could only exist after your period. Replace each with the period want sitting underneath it.
- Evidence of learning: The before-and-after list plus a paragraph on what changed in your plot.
PHASE 6
the accuracy–legibility curve
- Big idea: A perfectly accurate period voice would be unreadable. A fully modern voice is a costume party. Every historical novel sits at a chosen point between them, and the point has to be held for the whole book.
- Core mechanism: Accuracy and readability pull against each other, so every sentence spends one to buy the other, and the reader's trust is the currency being spent.
- Essential question: What am I willing to get slightly wrong so the reader stays inside the story?
- Learning targets: Describe the trade between period accuracy and readability in your own book. Cut a detail that is true but costs more than it earns. State your voice rule in one sentence and apply it consistently.
- Key terms: Register — how formal or plain the language feels. Translated voice — modern English shaped to feel period. Detail economy — a few sharp details buy more belief than many. Research showing — when the page brags about what you learned. Fictional dream — the trance a reader falls out of when jarred.
- Misconception to kill: More researched detail means more authenticity. Past a point, detail reads as a lecture and destroys the belief it was meant to build.
- Do this: Rewrite one paragraph three ways — fully modern, aggressively period, and the version between. Read all three aloud. Pick your point on the curve and write it as one rule.
- Evidence of learning: The three paragraphs plus your one-sentence voice rule.
machinery layer: systems and frameworks writers built
- Why this layer is next: The pressures in Layer 2 are permanent, so writers built tools to manage them — a way to turn events into a plot, a set of dials for distance and voice, and a filing system that keeps fact and invention apart.
PHASE 7
the story engine
- Big idea: History hands you a sequence of events. A novel needs a chain of causes. Turning then, then, then into so, so, but is most of the mechanical work.
- How the machine works: Scenes chain by cause. Each scene has someone who wants something, an obstacle, and a change of state at the end; the reaction to that change forces the next scene.
- Essential question: Does each scene change something that forces the next one?
- Learning targets: Find the turn in any scene you read. Convert an event list into a causal chain. Spot where your real history goes episodic and fix it.
- Key terms: Scene — a stretch of time where someone tries and something changes. Sequel — the reaction and the decision after a scene. Turn — the moment the situation stops being what it was. Causal chain — this happened because that did. Episodic — things happen in a row, but nothing forces anything.
- Misconception to kill: The history is my plot. A timeline is not a story; chronology is not causality, and real events in real order can still be shapeless.
- Do this: Write your event list. Rewrite it inserting only and so or but between items, never and then. Mark every place you cannot — those are the holes.
- Evidence of learning: The converted chain with every hole marked.
PHASE 8
point OF view, tense, and the translated voice
- Big idea: Three settings decide, before you write a sentence, how much distance sits between your reader and the past: who tells it, from when, and how far back the language reaches.
- How the machine works: Point of view fixes what the reader can know, tense fixes where the telling stands in time, and register fixes how period the words sound. Set all three deliberately, then hold them.
- Essential question: Who is telling this, from when, and how much do they already know?
- Learning targets: Name your point of view, tense, and register, and say what each makes impossible. Write the same moment two ways and read the difference. Build a short list of words your narrator would never use.
- Key terms: Close third — third person that stays inside one character's head. Retrospective narrator — an older self telling their younger story. Free indirect style — the character's thoughts in the narrator's voice. Psychic distance — how close the camera sits to a mind. Diction filter — a list of words this narrator would never use.
- Misconception to kill: First person feels closer, so it suits history best. First person locks you to one mind's knowledge and one voice's period ceiling, which is often the wrong trade for a book that has to show a whole world.
- Do this: Write the same 150-word moment in close third past tense, then in first-person retrospective. Note in one line what each version makes impossible.
- Evidence of learning: Both versions plus your three locked settings.
PHASE 9
the period bible and the fact/fiction ledger
- Big idea: You will forget what you invented. The ledger is what makes your author's note honest and your revision survivable.
- How the machine works: One file holds the timeline, the real cast, the invented cast, the rules of the world, and a running list of every change you made to the record — built while drafting, not after.
- Essential question: Could I defend every date, name, and change in this book from my own file?
- Learning targets: Build a period bible with all five sections. Log an invention at the moment you make it. Catch a continuity error by checking the file instead of rereading the book.
- Key terms: Period bible — one document holding every rule your world runs on. Timeline — real events and story events on the same line. Ledger — a running list of what you changed and why. Continuity error — page 40 contradicts page 200. Open question — something still to check, tracked instead of forgotten.
- Misconception to kill: I'll sort out what's real during revision. Invention and fact blur fast; rebuilding the line later costs far more than logging it live.
- Do this: Build the file today with five sections — timeline, real people, invented people, world rules, changes and why. Backfill everything you have written so far.
- Evidence of learning: The file, with at least one entry in every section.
surface layer: the live argument
- Why this layer is last: Only once the machinery is understood can you make sensible choices about the promise you make to readers, the people your book can hurt, and the argument the genre is currently having.
PHASE 10
the reader contract
- Big idea: Historical fiction is many different promises. A historical mystery promises a solution. A saga promises time passing. Alternate history promises one changed rule. Readers punish a broken promise harder than an inaccuracy.
- So what: Your subgenre decides what the reader is waiting for, which means it decides whether your suspense lands at all.
- Essential question: Which promise am I making on page one, and do I keep it?
- Learning targets: Name your subgenre and its promises. Name three published books yours sits beside and say how yours differs. Draft an author's note from your ledger.
- Key terms: Subgenre — a shelf with its own promises. Comp — a published book yours sits beside. Author's note — where you tell the reader what you changed. Secret history — real events, invented hidden causes. Alternate history — one fact changed, consequences followed.
- Misconception to kill: Genre labels are marketing, not craft. The label sets what the reader is waiting for, so ignoring it means writing tension nobody is feeling.
- Do this: Name your subgenre, list three comps, write your page-one promise in one sentence, and draft your author's note from the ledger.
- Evidence of learning: The promise sentence and the draft author's note.
PHASE 11
writing real people and real suffering
- Big idea: The past includes people who are still alive, descendants who are still here, and events communities are still arguing about. The more recent and more real, the higher the cost of getting it wrong — and the cost lands on someone other than you.
- So what: Every invention about a real person or a real atrocity is a claim that will be read as being about them, so the work owes research, care, and open acknowledgment of what was changed.
- Essential question: Who is still alive to be hurt by this, and have I done the work to earn the scene?
- Learning targets: List every real named person in your book and what you invented about each. Explain the difference between showing a view and endorsing it. Identify a scene you are not yet qualified to write, and say what would qualify you.
- Key terms: Living memory — recent enough that people still remember it. Descendant community — people who inherit this history. Composite character — several real people merged into one. Expert read — someone who knows the ground checking your pages. Depiction versus endorsement — showing a view is not sharing it.
- Misconception to kill: It's fiction, so nothing is off limits. Fiction can go anywhere; the further in it goes, the more the work owes in research, care, and honesty about its changes.
- Do this: List every real named person in your book. For each, write what you invented and one line on why that invention is defensible. Flag anyone you cannot defend.
- Evidence of learning: The annotated list with flags.
PHASE 12
the historical novel now
- Big idea: You are not writing into a settled form. The genre is currently arguing about who may tell which past, how new research tools change the work, and whether accuracy or interpretation is the point.
- So what: Your book joins one of those arguments whether you intend it to or not, and readers will place it there for you.
- Essential question: What argument does my book join, and what is the strongest case against it?
- Learning targets: State the argument your book joins in one sentence. State the strongest objection to it without weakening it. Name the risks of your own research tools.
- Key terms: Voice appropriation — writing from inside a group you are not part of. Recovery fiction — giving voice to people the record ignored. Revisionism — retelling a story to challenge the settled version. Fabrication risk — a tool inventing a source that sounds real. Adaptation — the book's second life on screen or stage.
- Misconception to kill: If I stay accurate, I stay neutral. Choosing whose eyes we watch through is already an argument about what mattered.
- Do this: Write 300 words naming the argument your book joins, the strongest objection to it, and your answer.
- Evidence of learning: The 300 words, with the objection stated at full strength.
capstone project: one chapter with apparatus
- Write one chapter of 2,500–4,000 words set inside a documented historical moment, and submit it with its apparatus: the documented/contested/silent map for that chapter, the ledger entries it produced, one paragraph of author's note, and a one-page defense of where you set the accuracy–legibility dial.
- Rubric: Every documented element is either accurate or logged as a deliberate change with a stated reason. The reader knows something the point-of-view character does not, and the chapter uses that gap. No sentence exists mainly to display research. The protagonist's want is one a person of that period could actually hold, in their own terms. The author's note could be published unchanged.
assessment spine
- Layer 1 — Can they say what the reader knows at any line, and what each claim rests on? Check with margin annotation and short answer.
- Layer 2 — Can they sort a scene into documented, contested, or silent, and defend a voice setting? Check with a classification plus written defense.
- Layer 3 — Can they build the machinery and hold it? Check by reviewing the artifacts: causal chain, locked dials, period bible.
- Layer 4 — Can they state their promise, their exposure, and their argument? Check with a short position paper.
deliberately out OF scope
- Line-level prose craft — a full course of its own, and it changes nothing about the historical layer.
- Agents, submissions, and contracts — business, not craft.
- Any one period's actual content — the method has to survive whatever period the writer picks.
- Screen adaptation technique — a different medium with different bedrock.
- Specific archives and research databases — they change too fast to teach durably.
confidence notes
- Legal exposure for depicting real people varies by country and changes over time. Unit 11 is craft guidance, not legal advice.
- Subgenre expectations and what publishers currently buy shift; treat Unit 10's categories as stable and any market read as perishable.
- Craft terms such as psychic distance, sequel, and free indirect style are defined differently by different teachers; the definitions here are usable, not universal.
- Research-tool behavior, including fabricated sources, is changing quickly, so Unit 12's treatment will date faster than the rest of the course.
What you’ll need
- pen and paper or a notes app
- one historical novel you like
- access to a library or online primary/secondary sources for your chosen period
Recommended resources
Tools & Apps
Tutorials & Learning
Progress milestones
Track your progress with these key achievements:
1
Units 1–3
Can track what the reader knows at any line, name what each claim rests on, and write a double-meaning scene before a known event
2
Units 4–6
Scene list tagged documented/contested/silent; period wants rewritten; one-sentence voice rule locked
3
Units 7–9
Causal chain with holes marked; POV/tense/register locked; period bible with all five sections started
4
Units 10–12
Page-one promise, annotated real-people list, and a 300-word argument position paper drafted
5
Capstone
2,500–4,000 word chapter with D/C/S map, ledger entries, author's note paragraph, and accuracy–legibility defense
Common challenges
Every beginner faces obstacles. Here's how to overcome them:
Treating the timeline as the plot
Solution: Rewrite the event list with only 'and so' or 'but' between items. Every place you cannot insert those words is a hole that needs a causal scene, not another date.
Research never feels finished
Solution: Stop when you can tag each scene documented, contested, or silent. Knowing a lot is not the exit condition; zone classification is.
Putting modern wants in period clothes
Solution: List five protagonist wants, cross out any that could only exist after your period, and replace each with the period want underneath. Sympathy comes from consistent wanting and cost, not shared values.
Forgetting what you invented
Solution: Log inventions and alterations in the ledger at the moment you make them. Rebuilding the fact/fiction line in revision costs far more than logging live.
Breaking trust with research-dump prose
Solution: Cut any detail that is true but costs more than it earns. Hold one voice rule for the whole book on the accuracy–legibility curve.
Share your progress
Celebrate your achievements and inspire others:
- • Share your tagged D/C/S scene list and one-line defenses of every alteration
- • Post the three voice-curve paragraphs and your one-sentence voice rule for feedback
- • Publish or workshop the capstone chapter with its author's note and ledger appendix
- • Recommend the course throughline to someone stuck between research and drafting