Two Versions of a Historical Text Disagree. Which One Can the Documentary Pitch Quote?
Two Versions of a Historical Text Disagree. Which One Can the Documentary Pitch Quote?
Quote an identifiable version, explain a meaningful disagreement, or withhold the claim until you can support it. Do not combine the most useful words from different versions and put quotation marks around a sentence that none contains.
Before choosing, establish what you actually inspected. A manuscript image, a transcription of it, a printed edition, and a quotation in a later book are different objects. Two websites may reproduce the same object. A clean transcription may disagree with its source image. You need those distinctions before deciding what the disagreement means for the film.
This is a bounded source-comparison task, not a method for authenticating documents or recovering an author's final intention. The worked case below uses three original, explicitly fictional texts. It demonstrates how a quotation can remain attributable even when the event described remains unresolved.
Identify the objects before counting the versions
The Text Encoding Initiative's guidance on critical apparatus distinguishes the readings found in texts from the witnesses that preserve them, and provides ways of connecting each reading to its source. It also makes clear that identifying and grouping variations involves editorial judgment rather than a purely mechanical operation.[^1]
For a pitch writer, a witness is useful terminology for the particular textual source carrying a reading. It does not mean a person who witnessed the historical event. A printed summary can be a text witness without being an eyewitness account.
Start a small source ledger. Give each object a stable identifier and record its title or description, stated date, edition or holding reference where available, the exact passage, and how it was accessed. Name the object honestly: “published transcription, source image not inspected” is more useful than a broad label such as “original letter.”
Keep separate questions separate. Who wrote or produced this object? What does it say? What relationship does it have to the other versions? What does it establish about the event? Evidence answering one does not automatically answer the rest.
The fictional source set
Our constructed example concerns an inspection at a works yard. These are newly authored plain-text files, not transcriptions of real historical documents. Their apparent historical dates are part of the fiction. The separate files and construction ledger are included with the example.[^3]
| ID | Constructed object | Inspected passage | Known construction—not an inferred history |
|---|---|---|---|
| A | Yard note, dated May 18, 1911 within the fiction | A, lines 2–4 | Written first for this exercise |
| B | Office summary, dated May 19, 1911 within the fiction | B, lines 2–4 | Deliberately adapted from A, changing the inspection claim and introducing a spelling error |
| C | Later office copy, undated within the fiction | C, lines 2–4 | Deliberately copied from B for this exercise |
The construction ledger is important. We know B was adapted from A and C from B because those are the recorded acts used to make the exercise. We have not discovered a chain of copying by analyzing the wording. In a real case, the relationship would require its own evidence.
The principal readings are:
A, line 2:
The inspection did not take place on Tuesday, according to the night watchman.
B, line 2:
The inspection did take place on Tuesday.
C, line 2:
The inspection did take place on Tuesday.
Two texts agree. That does not give you two independent accounts of an inspection. Here C is a constructed copy of B. Even if their relationship were unknown, counting matching sentences would not establish whether the inspection occurred.
A is not automatically true because its fictional date is earlier. Its sentence attributes the account to a night watchman. A later date, an institutional heading, or polished prose would not make B automatically true either. The immediate result is a documented difference between statements.
Align enough text to preserve the difference
Comparison starts by identifying corresponding passages. Do not reduce a meaningful disagreement to a list of changed words before you have read the surrounding statement.
TEI by Example treats the selection and boundaries of corresponding fragments as interpretation. Its account also distinguishes recording alternatives from adopting a preferred reading.[^2] You do not need XML to use that distinction. A short table can keep the source, the difference, and its consequence together.
The rest of our fixture contains the following lines:
A, line 3: “The gate remained closed.”
A, line 4: “I did not see the inspector.”
B, line 3: “The gait remained closed.”
B, line 4: “The entry was copied from a yard note.”
C, line 3: “The gait remained closed.”
C, line 4: “Copy checked against the office summary.”
Now separate the variations rather than treating all of them as equivalent corrections.
| Point of comparison | A | B and C | Consequence for the proposed claim |
|---|---|---|---|
| Whether the inspection occurred | Denies that it took place | Says that it took place | The versions support incompatible statements about the event |
| Attribution of the account | Names the night watchman as the source | Does not include that attribution in the inspection sentence | The later wording cannot be quoted as the watchman's positive account |
| Closed entrance | Uses “gate” | Uses “gait” | An intentional shared error in this fixture; not proof of copying in a real case |
| First-person observation | A's writer says they did not see the inspector | That statement is absent | Absence of this sentence does not supply an affirmative eyewitness account |
The negation and the attribution are not decorative details. Removing “not” reverses the event claim. Removing the watchman qualification changes how the assertion presents its basis. A sentence can remain fluent while becoming substantially different evidence.
The first-person line introduces another limit. “I did not see the inspector” reports a lack of observation by that speaker. It is not equivalent to “no inspector came.” Do not convert the narrower statement into a broader finding merely because it sits near a claim that the inspection did not happen.
The shared spelling error has a different role. In our fixture, it accompanies a known copying relationship. In a real set of sources, it could motivate a question about dependence. It would not, on its own, establish which object copied which, whether both used another source, or whether the same alteration arose in another way.
Do not call unreadable text an omission
All words in this exercise are available in plain text. A real comparison may be less tidy. If a line is cut off in a photograph, you do not know that the document omits it. If a word is unreadable, you do not have a secure alternative reading. If a transcription leaves out a negation that is visible in the source image, the disagreement may concern transcription rather than two historical versions.
Record those states explicitly. “Not present in the inspected passage,” “not visible in this image,” and “reading uncertain” describe different limitations. An empty table cell makes them look alike.
The next action should follow the limitation. Request the missing page, inspect the available source image, or seek an appropriately qualified reading. Do not run an increasingly elaborate comparison on an extract whose relevant word is absent from the evidence you possess.
Reject the sentence nobody wrote
Here is the tempting composite:
The inspection did take place on Tuesday, according to the night watchman.
It sounds cautious because it names a source. But it takes the affirmative claim from B or C and attaches the qualification from A, where the watchman is associated with the opposite account. No witness in the fixture contains this sentence.
This is not harmless smoothing. It invents an attributed assertion. A footnote listing A, B, and C would not repair it: the reader would still encounter quotation marks claiming a source sentence that does not exist.
A subtler mistake would be to quote B's affirmative sentence under a heading reading “The Watchman's Account.” The words inside the quotation would be accurate, but the heading would supply an attribution B does not contain. Check the surrounding slide copy as carefully as the quotation itself.
The same applies to editorial insertions. Square brackets can make an intervention visible; they do not make every intervention justified. If the proposed insertion supplies the very identity or meaning under dispute, it needs evidence and explanation. Punctuation is not permission to resolve a factual question.
Choose a route according to the claim you need
There are several defensible ways to use this comparison. Their costs differ. Decide what the slide needs to establish before selecting the most dramatic wording.
Quote one version as that version's statement
Suppose the film is examining how the office summary describes the inspection. Then B can be quoted accurately:
The office summary records an inspection
“The inspection did take place on Tuesday.”
— Fictional witness B, line 2The yard note gives a contrary, attributed account; the event remains unresolved in this source set.
The heading makes a claim about the document, not an independent claim that the inspection happened. The accompanying sentence prevents a known contrary source from disappearing merely because the selected quotation is accurate.
The benefit is focus. The reader can see what one particular record says. The cost is that the quotation alone does not establish the event. If the deck needs to claim that the inspection happened, this route does not supply the missing support.
Similarly, quoting A would show that the yard note attributes a negative account to the watchman. It would not prove that the watchman was correct, that the writer transcribed the account faithfully, or that no inspection occurred.
Present the alternatives when the disagreement is the point
If the documentary's inquiry concerns the conflicting record, give the reader both relevant assertions. Do not make them so small that the distinction becomes decorative.
The records disagree about Tuesday's inspection
Yard note, A:2: “The inspection did not take place on Tuesday, according to the night watchman.”
Office summary, B:2: “The inspection did take place on Tuesday.”
C repeats B's inspection statement. The constructed source set does not settle which account corresponds to the event.
This route gives the disagreement room. Its cost is attention: the slide becomes about the sources rather than a single narrative beat. That is worthwhile when the difference is central to the proposed film; it is unnecessary when the slide has another job and the claim can responsibly be omitted.
A paired display is not a vote or a declaration that the accounts are equally reliable. It shows the variation the inspected material establishes. Any assessment of reliability needs reasons beyond the symmetry of the layout.
Paraphrase what the comparison supports
Sometimes the useful claim is about the state of the evidence:
The yard note attributes a negative account of the inspection to the night watchman. The office summary states that it occurred, and a later copy repeats that statement. The discrepancy is unresolved.
No quotation marks are needed because this is an editorial account of the comparison. It keeps attribution and disagreement intact.
A weaker paraphrase would be “Several records confirm an inspection on Tuesday.” It suppresses A and treats the constructed copy as corroboration. Another would be “The watchman contradicted the official account.” That may suggest a direct exchange or a securely established voice the fixture does not provide. The yard note contains an attribution; the watchman has not supplied a separate inspected statement.
Paraphrase can reduce the space needed for exact wording. It cannot expand what the sources support. Keep the ledger available so another person can check what the compression retained and what it left out.
Withhold the event claim
If the pitch depends on proving the inspection happened, none of these textual choices resolves the problem. The next task is evidence work, not finding a cleverer sentence.
Withholding can mean removing the disputed event assertion while retaining the larger inquiry. It can mean labeling the event as an open research question in development materials. It does not require filling a slide with every unresolved detail, but it does require avoiding a headline that implies the gap has been closed.
The cost is a less definitive pitch. The benefit is that the film is not being sold on a conclusion its evidence has not established.
Make a preference accountable
An editor may have a justified reason to use a particular version as the basis for a quotation. TEI's apparatus guidance and TEI by Example provide for preferred readings while making that preference an editorial choice, not a property a comparison automatically discovers.[^1][^2]
For a real project, explain the reason at the level you can support. “We quote the published edition named in this sequence because the film examines its public wording” is a bounded choice. “This is the author's true original sentence” is a much larger claim. The first does not quietly entail the second.
In the fixture, choosing A because it preserves an explicit source qualification could be useful for discussing the yard note's manner of reporting. It would not justify erasing B from a discussion of whether the event occurred. Choosing B because the scene concerns the office summary is also coherent. Choosing B because its assertion makes a better opening line is not evidence for its historical truth.
Record consequential editorial changes separately. If you modernize spelling for a particular presentation, say what you changed and retain the source reading. For this exercise, there is no need to repair “gait” inside B and C: preserving it makes the known construction visible. A reader-friendly excerpt of line 2 avoids that spelling issue without falsifying the inspected sentence.
Keep the quotation check narrower than the historical claim
For each proposed exact quotation, compare it with the named witness and locator. Then read enough surrounding text to check what the extract leaves out. Finally, inspect the headline, caption, and adjacent prose for claims the quotation itself does not carry.
A text-containment check can establish whether a string appears in a supplied file. In our fixture, the single-version quotations appear in their designated lines and the misleading composite appears in none of the three witnesses. That is a useful mechanical boundary, not historical authentication. It does not decide whether an extract is representative, whether a document is genuine, or whether its statement is true.
The human reading remains necessary because a technically exact quotation can mislead. B's inspection sentence under the watchman heading is one example. A cropped negative statement presented without its attribution would be another. Exactness is one condition, not the whole editorial judgment.
Keep the final slide short and the research trail specific. A compact witness label should resolve to the object actually inspected, its locator, and the relevant comparison notes. Avoid a footnote that opens a general archive homepage and leaves the reader to guess which version supported the sentence.
If the central word remains uncertain, identify the narrow question for a qualified specialist: whether a visible mark is a negation, whether two entries belong to the same textual unit, or whether a claimed relationship between editions is supported. Do not ask them to certify a historical conclusion your comparison has already smuggled into the brief.
For this constructed case, the selected route is to present the disagreement. A's attributed negative account and B's affirmative statement remain distinct; C is recorded as a constructed copy, not an additional eyewitness. The composite is rejected. The pitch can now say exactly what the records disagree about without pretending that their disagreement tells us what happened.
Sources
[^1]: Text Encoding Initiative, TEI Guidelines — 13 Critical Apparatus, especially 13.1.1, “The Apparatus Entry”; 13.1.2, “Readings”; and 13.1.4, “Witness Information.” Inspected September 19, 2026. Supports attributable witness readings and the editorial nature of grouping or preferring them. Encoding guidance is not authentication of a document, recovery of authorial intention, or proof of an event.
[^2]: TEI by Example, Module 7: Critical Editing, section 4.1, “Basic Organisation of an Apparatus Entry,” examples 7–11 and the accompanying notes on corresponding fragments, preference, and alternative readings. Inspected September 19, 2026. The plain-language comparison here is an original adaptation for a bounded pitch-writing task, not a scholarly edition.
[^3]: Original fictional materials created for this article: the accompanying example files, comprising the construction ledger, the separate witnesses, and the quotation checks. The dates, records, inspection, watchman, and copying relationships are invented. No historical source image, handwriting, original document, or real inspection has been authenticated.
Frequently asked questions
Can I combine wording from two disagreeing versions if I cite both?
No. Do not combine the most useful words from different versions and put quotation marks around a sentence that none contains. The tempting composite 'The inspection did take place on Tuesday, according to the night watchman' takes the affirmative claim from B or C and attaches the qualification from A, where the watchman is associated with the opposite account. No witness in the fixture contains that sentence, and a footnote listing A, B, and C would not repair it.
What should I record before comparing versions?
Establish what you actually inspected. A manuscript image, a transcription of it, a printed edition, and a quotation in a later book are different objects. Start a source ledger with a stable identifier, title or description, stated date, edition or holding reference where available, exact passage, and how it was accessed. Name the object honestly, such as 'published transcription, source image not inspected.' Keep separate questions separate: who produced the object, what it says, its relationship to other versions, and what it establishes about the event.
If two sources agree, does that confirm the event?
Not necessarily. In the fictional source set, B and C agree, but C is a constructed copy of B. Even if their relationship were unknown, counting matching sentences would not establish whether the inspection occurred. A is not automatically true because its fictional date is earlier, and a later date, institutional heading, or polished prose would not make B automatically true either. The immediate result is a documented difference between statements.
What are the routes for using a disagreement in a pitch?
Several routes are available: quote one version as that version's statement; present the alternatives when the disagreement is the point; paraphrase what the comparison supports; or withhold the event claim if the pitch depends on proving the event. Each has costs. A single quotation focuses attention but does not establish the event; a paired display gives the disagreement room but makes the slide about the sources; paraphrase can compress but cannot expand what the sources support; withholding is less definitive but avoids selling a conclusion the evidence has not established.
Does a shared spelling error prove which source copied which?
In the fixture, the shared 'gate' and 'gait' spelling accompanies a known copying relationship. In a real set of sources, a shared error could motivate a question about dependence. It would not, on its own, establish which object copied which, whether both used another source, or whether the same alteration arose in another way.