A false belief has to be born before it can be repeated. Where does it come from? Not one place. And the place it comes from tells you the one thing that will kill it.
The companion to this page, Repeated Until True, took forty-four of this archive's record-corrections and asked one question of each: what single checkable thing settles it? A date, a primary source, a recomputation, a count, a ruler. Sorted by their cure, the deaths of myths turned out to be strangely uniform. Almost half fell to the same move: a number someone skipped.
This page asks the question from the other end. Before a claim can be corrected, it has to get born, and get plausible enough that people pass it along. A myth is not a random error. It has an etiology, and the etiologies sort into a small, recognizable set. Below are six of them, drawn from the myths this archive has already taken apart. You will know most of the claims. What you may not have noticed is that the way each was born is legible, and different, and that it predicts the way each one dies.
Here is a claim everyone has heard, and the one-line truth under it. Before you read on: why did this false belief survive? Pick the origin you think fits. There is a defensible answer, drawn from the layer's own verified page, and it comes with the check.
Put every specimen in a grid. Down the side, the six ways it could have been born. Across the top, the five discriminators that could kill it, the same five Repeated Until True found. Tap a dot to read the case.
| Ia date | IIa source | IIIa recomputation | IVa count | Va ruler | |
|---|---|---|---|---|---|
Column IV, a count, is empty here on purpose. It is not unpopulated in the archive: Repeated Until True fills it (the myth of many Eskimo words for snow, how many humans have ever lived). None of these nineteen births happen to die that way. The matrix is a slice of a larger gallery, not the whole of it.
Read the grid as a picture and one thing jumps out. The births fan out; the deaths collapse in. Six origins spread down the rows, but the dots pile into a single column: a recomputation, the same tool that settled nearly half of Repeated Until True's larger sample. All the variety is in how a myth is born. The way it dies is, more often than not, just arithmetic that was skipped the first time.
And the birth predicts the death. The map is tightest at its two corners. A belief that was a story someone told, a wartime cover for radar, an Econ-101 tale about barter, a romantic biographer's invented flat-earth, always dies to a date: find when the story was constructed and the claim that it was “always known” falls apart. A belief that is the wrong question, Everest as “tallest,” the tomato as fruit-or-vegetable, was never false at all, only underspecified, and only a ruler settles it: name the measurement you mean and both answers are right. Those two are near-bijections. The corners are clean.
The middle is where the cures converge. A real finding stretched past its scope (Mehrabian's 93%, coffee and dehydration), a coincidence read as a cause (the full moon, sugar and children), an untested obvious (the five-second rule, bulls and the colour red): three different births, and all of them die to the same thing, someone finally doing the sum over the whole sample instead of the memorable part of it. That convergence is not a coincidence. It is why the companion portal could sort forty-four corrections into just five cures. The deaths were always going to be few, because most false beliefs are, underneath, a number no one checked.
One myth in this archive resists the whole scheme, and honesty asks that it be named rather than filed. Spicy food does not cause stomach ulcers: the cause is a bacterium, H. pylori, and Marshall and Warren proved it, Marshall drinking a culture in 1984 to force the point, the work taking the 2005 Nobel Prize. That belief was not a wartime lie, nor a phantom statistic, nor an untested intuition. It was the best medicine of its day: taught, published, believed by doctors, and then overturned by a discovery. Call it a seventh way to be wrong, the belief that was once true enough to teach. It is rare, and it is the honourable failure mode: not a myth so much as a hypothesis the evidence eventually retired. The six above are the ordinary births. This one is the edge of the taxonomy, shown so you can see where the edge is.
This page is one half of a pair. Repeated Until True is the other: the same corpus of myths, sorted by the check that kills each one, opened with the illusory-truth effect (repetition alone makes a claim feel truer). Birth here, death there. Between them runs the middle of a myth's life, the repetition that carries it from the one to the other.
Every claim, its origin, and the discriminator that settles it. The eight marked ✓ are already sorted in Repeated Until True by that same cure; the rest extend its gallery by the same test. This portal builds no new fact: every figure lives on the linked layer's own verified page. What is new is only the reading, the sorting-by-origin and the shape it makes.
| The claim (false) | Its origin | Its cure | In R.U.T. |
|---|
What is lifted and what is new. Lifted: every claim, correction, date, and figure, from the nineteen linked layers and their verifiers, plus the five-cure scheme from Repeated Until True. New (and therefore a reading, not a measurement): the six-origin taxonomy, each specimen's placement in it, and the observed pattern that origins diverge while cures converge. The origin of a belief is softer to classify than the text-test that refutes it, so a few specimens carry a second, weaker origin as well; the placement is a lens on the corpus, not a theorem about it. Where this page assigns a cure to a myth not yet in the companion gallery, that is an extension by the same test, marked above.