Megan Skiendel: David, hey — I've been in a weird headspace all week, honestly. I pulled out an old journal I kept in 2008 and I'm reading it and I'm thinking: is this evidence? Like, is this a document, or is it just me complaining about my commute?
David Sterling: Well, both. That's the tension, right there.
Megan Skiendel: Which is exactly where we're going today — primary sources, secondary sources, what counts as evidence, what gets trusted. And the thing that keeps pulling me back is: what happens when the evidence is wrong? Not misread. Actually, deliberately wrong.
David Sterling: The Donation of Constantine.
Megan Skiendel: Eight hundred years, David. Picture this — a monk in a medieval scriptorium, candlelight, copying out this document with total reverence. The word of Constantine himself, transferring authority over the western Roman Empire to the Pope. That's the document. And the thing is, a primary source is basically a witness statement — it tells you what someone at the time saw, believed, did. The closer they were to the event, the more we trust it. That's the everyday click. That's all it is.
David Sterling: Right — but the document was forged.
Megan Skiendel: Forged. Completely fabricated. And it shaped European church-state relations for eight centuries before a Renaissance scholar named Lorenzo Valla looked at the Latin in 1440 and said — wait, this grammar doesn't exist yet. The language in this document couldn't have been written when Constantine was alive.
David Sterling: Wait — how long did that actually take to unravel? Because 1440 is not the end of it.
Megan Skiendel: That's the part that honestly floors me. Valla catches it in 1440. The political use of the document doesn't stop immediately. It kept functioning as a legitimate claim for decades after that. So the forgery and fabrication in the historical record — it's not just about the moment of exposure, it's about how long bad evidence keeps doing real work in the world.
David Sterling: And every secondary source built on it in the intervening eight centuries was executing its method correctly. On corrupted data.
Megan Skiendel: Which is the problem wearing a methodology costume, yes.
David Sterling: And that's the structural problem, not just the historical embarrassment. Secondary source is derivative by definition — its validity depends entirely on the primary evidence it draws upon. Corrupted input, corrupted output. The method was sound. The data wasn't.
Megan Skiendel: Which is the CFO analogy. Analyst writes a technically correct report. Filing is accurate on its face. But the quarterly call was a lie.
David Sterling: Right — and the secondary work isn't wrong in method. It's wrong in output. That's a different failure mode than sloppy scholarship.
Megan Skiendel: So what actually saved medieval history? Because historians clearly didn't just throw up their hands after 1440.
David Sterling: Triangulation. Cross-referencing multiple independent primary sources so credibility accumulates through convergence — not reliance on any single document. The Donation was exposed, and historians still reconstructed accurate pictures of medieval power because they had other sources that weren't forged. The system self-corrected. Slowly.
Megan Skiendel: Slowly is doing a lot of work in that sentence.
David Sterling: Eight centuries of bad primary evidence doing real work — yes, slowly. But the triangulation held. That's the mechanism. No single source is treated as definitive.
Megan Skiendel: Okay, but — I mean, that works when you have multiple sources pointing the same direction. What about cases where the record is just thin? Or worse, one-sided by design?
David Sterling: Anne Frank's diary and the Nuremberg Laws. Hold both at once. The diary is a firsthand account — what it felt like to be subject to a regime. The Nuremberg Laws, enacted in 1935, show how that regime legally justified itself institutionally. Neither alone tells you the full story.
Megan Skiendel: Wait — so the diary can't tell you what the Nazi regime believed it was doing.
David Sterling: Exactly the gap. The Nuremberg Laws reflect only the perspective of those who held power. The diary reflects only the experience of those subject to it. The structural interdependence is literal — remove either source type and the argument is either ungrounded or blind.
Megan Skiendel: And the bias in what survives — whose archive makes it to the Bancroft Library — that's already filtered before any historian opens the box. The secondary work inherits that silently.
David Sterling: That's the error tolerance question I can't answer cleanly. How many secondary arguments are built on compromised primary foundations and never discovered? The Donation got caught. Most forgeries probably don't.
Megan Skiendel: Which is what keeps pulling me toward the bigger structural problem — because the forgery thing is almost the clean version. The messier version is that the boundary between primary and secondary isn't even fixed. Like, imagine a librarian in 1820 pulling Edward Gibbon's Decline and Fall off the shelf. Published 1776. She's holding a secondary source on ancient Rome, right? Gibbon wasn't there. But she's also holding a primary source on 18th-century European intellectual life — what an educated mind in that era thought Rome meant, what anxieties it was carrying.
David Sterling: Same text. Opposite category.
Megan Skiendel: Simultaneously. That's not a flaw in the system — that's actually how the system works. The research question determines which thing you're holding.
David Sterling: Doesn't that create a loop, though? If the question predetermines what counts as evidence, you've already built the answer into the method before you've read a word.
Megan Skiendel: That's — yeah, that's the real danger. And I'll concede it. But the antidote isn't pretending the category is fixed. Actually, Cathleen Sarti at Oxford published something on this in 2024 — source categorization as a discipline, not just a sorting exercise. Her argument is that you have to name what you're treating a source as before you draw conclusions. Not after.
David Sterling: Declaration of intent before interpretation. That's the discipline.
Megan Skiendel: Right — but most people don't do it. The U.S. Constitution is the case that makes me sit with this. 1787. Primary source for constitutional history, obviously — what authorities mandated, how the society was legally structured. But the moment you ask what the framers couldn't see or wouldn't say — the silences, the assumptions they didn't flag — now it's an object of analysis. Same document, different epistemic status.
David Sterling: The framers weren't writing a confession. They were writing law.
Megan Skiendel: Exactly. And what they omitted is as evidentially loaded as what they included. That's the historiography move — treating earlier scholarship as primary evidence of the intellectual moment it came from. Gibbon becomes data. The Constitution becomes data. Secondary collapses into primary the second you shift the question one degree.
David Sterling: Which is either elegant or catastrophic, depending on whether the categorization is explicit. If Sarti's right — name the frame first — the loop is disciplined. If she's wrong, or if scholars skip that step, you've got an infinite regress. Every source is simultaneously everything.
Megan Skiendel: No, I don't buy infinite regress. The discipline holds. The bias and selectivity in the archive is the prior problem — whose version made it into the record shapes what's even available to categorize. That filtering is invisible and it predates methodology entirely.
David Sterling: The record is pre-curated. I mean — that's not a methodology problem you can solve with better categorization. That's a power problem.
Megan Skiendel: And that's actually what makes the formal mechanics so important — because once you're inside the method, with Thomas Jefferson's letters, with whatever's physically sitting in the Bancroft Library, the question of how you actually apply source criticism before you interpret anything — that's the part we haven't pulled apart yet.
David Sterling: Right — and that's exactly the entry point. The formal name for it is source criticism. The University of Basel Department of History codifies it as a two-step sequence: criticism first, interpretation second. You cannot draw conclusions from a source before you've interrogated the source itself.
Megan Skiendel: What does that actually look like on a Tuesday afternoon?
David Sterling: Graduate student. Bancroft Library at Berkeley. She's been given a folder of Thomas Jefferson's handwritten letters — published collections, canonical primary sources for colonial U.S. history. Before she quotes a single sentence, the Basel framework requires her to ask: who wrote this, to whom, when, why, and — critically — what did Jefferson choose not to say.
Megan Skiendel: The omission is evidence.
David Sterling: The omission is evidence. That's the step most readers skip. Proximity is not the same as reliability — the closeness is actually where the bias lives. Jefferson writing to a political ally about a vote isn't neutral testimony. It's strategic.
Megan Skiendel: And the Bancroft itself — I mean, the archive she's sitting in wasn't built neutrally. Which institutions had the resources to acquire and preserve those letters? That's not random.
David Sterling: No, it's not. But — and this is the point I'd push back on — source criticism is precisely what makes that visible. You can't correct for a power structure you haven't named. The two-step forces you to name it.
Megan Skiendel: Okay, but naming it doesn't recover what isn't there. The letters that didn't make it into the Bancroft folder — that gap is permanent.
David Sterling: Fair. That's where material culture actually earns its place. A Roman aqueduct — no author, no intent to deceive, no strategic omission. It's primary source evidence by temporal proximity alone. The medium doesn't determine the category.
Megan Skiendel: Wait — so the aqueduct clears a hurdle the Jefferson letter can't?
David Sterling: Different kind of hurdle. The aqueduct can't lie to you. It also can't tell you what anyone believed. The letter can lie — but it can tell you what Jefferson wanted someone to think he believed. Both are primary. Neither is sufficient alone.
Megan Skiendel: That's actually the triangulation point landing in a different place than I expected. It's not just — check multiple sources for consistency. It's that different source types are filling structural gaps the others literally cannot fill.
David Sterling: Which is the load-bearing thing. The archive being pre-curated is real — Megan's right that it's a power problem. But the method's answer to it isn't to refuse the archive. It's to enter it knowing criticism precedes interpretation, every time, without exception.
Megan Skiendel: The discipline doesn't fix the gap. It just makes you honest about what you're standing on.
David Sterling: That's actually where I want to land. The monk copying the Donation of Constantine — he wasn't lying. He believed it.
Megan Skiendel: He believed it completely. And that's — I mean, that's the thing that keeps shifting on me the longer I sit with it. The document was fabricated, yes. Bad evidence of what Constantine actually did. But it's... it's almost perfect evidence of what medieval Europe needed him to have done. What a society in that moment required to be true in order to hold together.
David Sterling: Two sources. Same document. Both doing their job.
Megan Skiendel: Which is — honestly, that's what historians are actually reconstructing, right? Not 'what happened' as some fixed retrievable fact. What people at the time knew, believed, acted on. The monk's belief is data. The forgery's staying power is data. The provisional reconstruction is more honest than the retrieval fantasy ever was.
David Sterling: Provisional isn't weak. That's the method being honest about what it can do.
Megan Skiendel: Yeah. I think I came into this wanting the forgery story to be a cautionary tale. It's not, really. It's closer to a proof of concept.
David Sterling: Frankly, same. Good one to sit with.
