Pitch a Documentary Series Built From a Public Record

The deck usually looks like this: four thousand items, twelve episode slides, one document per slide. A shift book, a repair order, a letter, a photograph of a ledger. Each slide is handsome. No slide explains why the document on it would hold a viewer for twenty minutes.

The missing slide is not another document. It is a sentence that says who made these records, for whom, why, and what they cannot answer. A shift log, a maintenance entry, and a personal letter were not made to answer the same question, and they do not agree simply because they now sit in the same storage room. Once that sentence is on the wall, the episodes stop being containers for material and start being questions. That is the pitch.

What follows is a method worked through a fictional collection. Gull Point Signal Station, its shift books, its maintenance records, its letters, and every entry quoted below are invented so the steps can be shown without claiming anyone has opened a box in a reading room.

Bound the collection and explain its origin

Before you can promise an inquiry, you have to describe the thing you are inquiring into. Not "the Gull Point archive." A bounded collection map: who created each part, who it was for, when it runs, and where its coverage stops.

The exercise supposes three record groups, held together by accident of donation rather than by a shared purpose.

Shift books, 1901–1919 and 1927–1934. Kept at the station by whoever stood the watch, primarily for the station's own continuity and for the district office's periodic inspection. They record weather, signals sent and received, vessels sighted, and notable events, in the terse abbreviations of a routine form. Seven years are missing.

Maintenance records, 1911–1934. Fault reports, repair notes, and requisition forms, some completed at the station and stamped at the district office. These documents exist in part because someone needed a part or money approved. That purpose shapes them: a fault described for a requisition is described by an author making a case.

Letters, 1908–1934, outgoing. The keeper's correspondence to the district office survives; the district's replies do not. A separate bundle of family letters, 1912–1929, was written by the keeper's wife and eldest daughter to relatives inland and concerns weather, illness, wages, and the noise of the signal, not the station's official work.

Three creators, three audiences, three reasons. Now the limits. The shift books record a watch, not a person — the book does not tell you who was on the platform or how many hands it took. The maintenance records begin a decade after the station opens and continue through the years when the shift books are missing. The outgoing letters are half a conversation. Nothing here is a personnel file, a vessel log, or a district budget.

Two disciplines follow from this. First, keep the catalogue's description and your own inspection separate in the pitch. "The finding aid lists 41 volumes" is a claim about a list. "We have read 1901–1904" is a claim about your team. Decks collapse these constantly, and the collapse is what lets a later slide promise coverage nobody has verified.

Second, do not let a shared shelf imply agreement. If the maintenance log says the fog signal's regulator was defective in February 1913 and a letter says the signal ran all night, those are two statements by two authors with two purposes. Where they seem to conflict, that is a question, not a contradiction to smooth over.

Build one episode from a relationship between records

Episode one asks: how did a night watch actually operate?

It cannot be built from one document. It needs three, and it needs you to say what each contributes.

The shift book entry for the night of 14–15 February 1913 logs the fog signal in operation from 21:40 to 04:05. That is what the station recorded for its own purposes. It is a claim about the signal's schedule, made by the station, at the time or near it.

The maintenance record for the same month reports a defective regulator on the signal apparatus. A requisition for a replacement part was filed on 11 February, three days before. That document tells you the equipment was in a state someone thought worth writing down, and that the writing served a request for parts — which is also why its language may lean toward seriousness.

The keeper's letter of 19 February, written five days after the night in question, says the crew did not trust the automatic gear in thick weather and worked the signal by hand. It is retrospective, it is the keeper's account, and it is also a document with an argument in it: a man explaining why his station needs better equipment.

Put together, those three support an interpretive claim — that the logged run was probably manual labor rather than machinery — and they leave a great deal outside the claim. They do not establish that the signal was audible offshore. They do not establish that anyone heard it, or that it mattered. They do not tell you which hands did the work. A vessel's log, a wreck inquiry, or a local newspaper would be needed for the first two, and a duty roster or watch bill for the third. Neither survives in this collection.

That is the material for a close reading on screen: a camera on the entry, a camera on the requisition, a camera on the letter, and a narration that keeps each one's authorship audible. The entry can be displayed. It must not be narrated into an independently established event. "The log records the signal running until 04:05" and "the signal ran until 04:05" are different sentences, and only one of them is available to you.

One further check belongs in the pitch rather than in the edit. Whose hand wrote the shift book entries? If the same hand appears in 1901 and 1918, the books may be a retrospective compilation rather than a contemporaneous log, which changes what "recorded at the time" is allowed to mean. An archivist or a paleographer can speak to that. Your team should not assume it either way.

Change the question's scale in the next episode

Episode two asks: how did maintenance decisions change across seasons?

Its unit is a season, not a night. That shift of scale is the reason it is a separate episode rather than a longer version of the first. Episode one follows one chain of claims through one evening and worries about over-reading a single entry. Episode two looks for a pattern across years and worries about something else entirely: mistaking an administrative rhythm for a physical one.

The exercise stipulates that requisition dates cluster in the four months before the district's books closed. The tempting reading is that equipment failed more as weather worsened. The competing reading is that requests were filed when they had a chance of being funded. Both fit the same sheets. In a real collection you would count the entries and report the count — the clustering is a stipulation here, and a measurement there.

Splitting those two explanations requires records this collection does not contain: the district's expenditure ledgers, held by a different office. If they show the station's requests were routinely deferred to a particular quarter, the pattern is paperwork. If they show requests funded on merit, the physical reading gets stronger. Either way the episode's honesty depends on going to get them.

There is also a coverage trap. The maintenance records run continuously from 1911, but the shift books are absent from 1920 to 1926. For those seven years a claim about winter activity rests on the maintenance records alone. The pitch should say so, and the episode should either narrow its period or explain why the central years are argued from one record type.

Now the third question, which the assignment of a fictional collection makes unusually clean: whose perspective is absent? The shift books describe an operating state, not a workforce. The assistant keepers, the men who hauled fuel and trimmed lamps, and the fishing families along the shore appear only incidentally, because the books were never about them. That absence is a fact about the records. It is not evidence of concealment, and it is not an invitation to invent a scene to fill the silence. It is a reporting task — crew payroll, census returns, a local newspaper, a court of inquiry — and the episode is contingent on what those searches turn up. If nothing turns up, the honest move is a shorter piece folded into another episode, not a dramatized gap.

Separate existence, interpretation, and screen presentation

Three decisions run through every episode, and a pitch gets stronger when it names them in that order.

Existence. The 14–15 February 1913 entry is on a page of a volume in the collection. It exists. A close-up can show it. That is all the close-up is doing.

Interpretation. The team proposes that the night's operation was manual because the crew distrusted the automatic apparatus, supported by the maintenance record and the letter. That is an argument with grounds, and the grounds are three documents by two or three authors, one of them writing five days later with a case to make. A useful pitch states the interpretation and its ceiling in the same breath.

Presentation. The film chooses a method: a close-up of the entry, a timeline graphic, a voiced reading of the letter, a labeled reconstruction of the apparatus being operated by hand. Each is legitimate. Each can be mistaken for corroboration. A timeline graphic drawn from the shift book is the shift book, redrawn — label it as such rather than letting it read as independently verified data. A voiced reading is a performance of the letter, not a witness. A reconstruction is a reconstruction.

Attribution is part of the presentation decision, not a post-production courtesy: the item, its author, its date, and the repository that holds it, on screen, wherever a document is used. Voice-over should say "the log records" and "the letter claims," and save the flat declarative for the things the team actually knows.

For the document questions themselves — who created it, for whom, when, why, and what other evidence would help interpret it — the National Archives' "Analyze a Written Document" worksheet offers a workable question set, available on 20 September 2026 at https://www.archives.gov/education/lessons/worksheets/analyze-a-written-document-intermediate. It is a US archival-education method for interrogating a document. It is scaffolding for the questions above, and it does not verify any record's accuracy, grant access, or make an episode work.

Two dependencies sit outside that scaffolding and should never be folded into it. Access is not publication: a reading room that will let you sit with a volume has not licensed you to broadcast it, and donor terms on a family's letters can run the other way from an archive's general policy. And "public record" is a phrase about a legal category of government material, not a statement that anything in it is safe to expose — medical detail, named individuals, disputes, and later records touching living people all need their own review. Put those assessments in the pitch as separate, unresolved items with a responsible party attached. Do not let a permissions line at the bottom of a slide carry the weight of a claim in the middle.

Show an inquiry map rather than an archive inventory

The inventory slide is the one with 4,112 items and 9.6 linear feet on it. It answers a question nobody in the room is asking. Replace it with the inquiry map, one row per proposed episode:

Beside that table, three separate columns of dependency — availability, filming access, and use review — because they fail independently. A volume you can read on Tuesday may be off-limits to a camera in March, and a letter you can film may name someone whose family would rather it stayed unseen.

The map also gives you an honest answer when the material runs thin. If episode three's missing sources do not exist and episode two's ledgers are unreachable, you have two episodes and a question. Two episodes can be a series. So can one film and a written piece. What you should not do — on the strength of a strong collection and a good reading room afternoon — is promise twelve and discover the arithmetic in the edit.

The pitch ends where the inquiry does: a small map of collection to episodes, and the gaps that would change it. Who made these records, for whom, and why; the comparison that turns a document into a question; the missing piece that could sink each episode and where you would go looking. Texture is the last thing you argue for, not the first. A page can be beautiful and still not know anything.

Frequently asked questions

What should a public-record series pitch establish before it proposes episodes?

It should give a bounded collection map: who created each part, for whom, when it runs, and where its coverage stops. It should keep the catalogue's description separate from the team's own inspection—'the finding aid lists 41 volumes' is a claim about a list, while 'we have read 1901–1904' is a claim about the team. A shared shelf does not make records agree, because different creators wrote for different purposes.

How does one episode get built from a relationship between records?

Use records that contribute different things. In the example, a shift book logs a fog signal running overnight, a maintenance record reports a defective regulator and a requisition, and the keeper's letter says the crew distrusted the automatic gear and worked the signal by hand. Together they support an interpretive claim that the logged run was probably manual, but they do not establish that the signal was audible offshore, that anyone heard it, or which hands did the work. Narration should keep each document's authorship audible.

Why change the question's scale in a later episode?

Scale changes what can go wrong. A night-focused episode worries about over-reading a single entry; a season-focused episode worries about mistaking an administrative rhythm—requisitions clustered before the district's books closed—for a physical one. It also exposes coverage traps: if maintenance records run continuously while shift books are absent for seven years, claims about winter activity in those years rest on one record type.

How should existence, interpretation, and screen presentation be kept apart?

Existence is that a specific entry is on a page and a close-up can show it. Interpretation is an argument with grounds and a ceiling—for example, that manual operation is supported by certain records, one written days later with a case to make. Presentation chooses a method: close-up, timeline graphic, voiced reading, or labeled reconstruction. Each method can be mistaken for corroboration; a timeline redrawn from a shift book is still the shift book, a voiced reading is a performance, and a reconstruction is a reconstruction.

What dependencies and permissions should appear in the pitch as separate unresolved items?

Track availability, filming access, and use review separately because they fail independently: a volume readable on Tuesday may be off-limits to a camera in March, and a filmable letter may name someone whose family objects. Access is not publication; a reading room visit does not license broadcast, and donor terms on family letters can differ from an archive's general policy. 'Public record' is a legal category, not a guarantee that medical detail, named individuals, disputes, or records touching living people are safe to expose.

Further reading

• Pitch a Compilation Series From the Archive You Actually Have — Read when a collection's item count is doing the work of coverage: it shows how to inventory what the material can actually be made to say across episodes.
• Use Factual Research to Change the Series Question, Not Fill the Deck — For the moment a finding should rewrite the series question instead of padding a research slide, this shows how to turn discoveries into bounded editorial decisions.
• Build an Archive-Led Film Around the Image You Cannot Find — When an episode's missing material is the point, this compares formal ways of handling absence honestly rather than dramatizing a scene the records never contained.