There is no such thing as the myth. There is only the version somebody wrote down, and almost nobody who wrote one down was trying to preserve it. They were entertaining a hall, equipping a corpse, defending a dynasty, training a poet, filling out a school exercise, or arguing with a rival temple. Preservation was a side effect, and side effects are unreliable.
This is the single most useful thing to know before reading any mythology, and it is the thing most retellings quietly skip. A book that opens "the Greeks believed that Zeus..." has already made a decision it is not telling you about: which Greeks, in which century, reading which poet. Once you start asking who was holding the pen, the contradictions that make mythology feel confusing stop being noise and become the most informative part of the record.
What follows is a tour of the hands the material passed through. It is not a bibliography. It is an argument about how to read.
Six kinds of people, six kinds of distortion
Sort the world's surviving mythologies by who committed them to writing and a pattern falls out immediately. Performers preserve what plays well. Priests preserve what is ritually needed. Courts preserve what legitimises. Monks preserve what is beautiful enough to excuse. Schoolmasters preserve what is good practice. Antiquarians preserve what is already dying.
Each of those motives leaves a distinct fingerprint, and the fingerprint is usually detectable. Performed material is repetitive, formulaic and full of epithets, because repetition is how a performer stays afloat. Ritual material is elliptical and allusive, because the audience already knew the story and only needed the operative words. Court material is genealogical and ends, suspiciously, at the reigning house. Monastic copies carry marginal apologies. School copies come in duplicate, triplicate, with the mistakes of tired teenagers preserved alongside the text.
So the question to ask of any myth is not "is this true", which is a category error, but "what was this document for, and what would its maker have had no reason to include?"
Homer is a name for a process
Start with the case everybody thinks is simplest. The Greek tradition reaches us principally through two poems and one systematiser, and none of the three is what a modern reader assumes.
The Iliad and Odyssey are the visible end of a long oral tradition, not the work of a man sitting down with a wax tablet. The decisive evidence came from Milman Parry, an American scholar who in the early 1930s went to Yugoslavia and recorded living illiterate singers of heroic epic — thousands of lines, composed in performance, on the spot. He and Albert Lord showed that the stock phrases which had embarrassed nineteenth-century critics as padding are actually a compositional technology: metrically pre-fitted building blocks that let a singer improvise at speed within a strict verse form. "Swift-footed Achilles" is not a description of Achilles; it is the shape that fills the end of a line.
That discovery reframed the whole Homeric Question. The argument is no longer "did one genius write both poems" but "at what point did a fluid performance tradition freeze into a fixed text, and who did the freezing." The freezing probably happened somewhere in the eighth or seventh century BCE, in the shadow of the new Greek alphabet; the final polishing happened centuries later still, in Alexandria, where librarians such as Aristarchus of Samothrace obelised lines they thought spurious and gave us the paragraphing and book divisions we still use.
Then there is Hesiod, who is a different animal entirely — a Boeotian farmer with a grudge against his brother and a lawsuit he thought was rigged, who produced in the Theogony the first surviving attempt to put the gods in order. Somebody had to decide that Chaos came first and that Uranus was castrated by Cronus in that sequence rather than another. Hesiod decided. His ordering is not the Greek belief; it is one poem's arrangement that later Greeks found convenient, which is a very different thing, and Greek local cult is full of parentages that flatly contradict him.
And the tidy handbook version of Greek myth that a school textbook draws on is later still, usually the Bibliotheca attributed to "Apollodorus", a Roman-era compilation that reads like a card index. Its neatness is editorial, not ancient.
Egypt wrote for the dead, and never for us
The Egyptian material is the most instructive case in the whole field, because Egypt is simultaneously the best-documented ancient religion and one of the worst-narrated.
The oldest religious texts on the planet are carved inside the burial chamber of the pyramid of Unas at Saqqara, around 2350 BCE. They are spells. They exist to get a specific dead king through a specific set of dangers, and they are addressed to him. Later the same technology democratises — painted onto Middle Kingdom coffins for anyone who could afford one, then written on papyrus for the New Kingdom market as the collection we misleadingly call the Book of the Dead, whose Egyptian title means something closer to "going out by day".
What this means practically is that Egypt left us an enormous, precisely dated body of writing about the gods that almost never tells a story. A spell does not need to narrate the murder of Osiris; it needs to invoke it, so it says the equivalent of "as when the thing was done to him at Nedyt", and moves on. The reader was dead, and the dead were assumed to know.
There are wonderful exceptions — the raucous courtroom farce of the Horus and Set succession dispute survives as a continuous tale on a Ramesside papyrus, and reads like satire — but the connected biography of Osiris that every modern retelling uses comes from Plutarch, a Greek priest of Apollo writing around 100 CE, in Greek, more than two millennia after the Pyramid Texts, for readers who had never seen Egypt. He is our most complete source and our least direct one at the same time. Any page here about Isis or the judgement of the heart has to hold both facts at once.
One Icelander, one library, one agenda
Now the case that should make everyone nervous. A very large share of what the world knows about Norse mythology — the nine worlds, the tidy roster of gods, the mechanics of Ragnarok — passes through a single book by a single man who was a Christian, a lawyer, a spectacularly wealthy political operator, and not a believer in any of it.
Snorri Sturluson wrote his Edda around 1220. He was twice lawspeaker of the Icelandic Althing, deeply entangled in the Norwegian court, and he was killed in 1241 at Reykholt by men acting for King Hákon. His book is a manual for poets: young skalds were losing the ability to decode the mythological kennings that court poetry runs on, and Snorri wrote a systematic guide so they could keep composing. He is therefore explaining a dead symbolic system to students, which makes him lucid, organised and quotable — and structurally inclined to tidy up.
He also opens with a euhemerising prologue explaining the gods as migrating Trojan men later mistaken for divinities, which is exactly the sort of thing a thirteenth-century Christian had to say to make the project publishable. The right way to read him is not with suspicion but with attribution. Every Norse page here marks which claims have older poetry behind them and which have only his say-so. That is why the Snorri problem gets its own essay.
The older poems themselves — the ones Snorri quotes — survive largely in one manuscript, the Codex Regius, written around 1270, which sat in Copenhagen for centuries and was carried back to Reykjavík by warship in 1971 to a national holiday. It has a gap where several pages were cut out. We do not know what was on them. That gap is a fair emblem for the whole subject.
The monks who apologised in the margin
Irish myth exists because Christian monks copied it, which is the single strangest fact in European literary history. The great manuscripts — the Book of the Dun Cow from Clonmacnoise around 1100, the Book of Leinster a century later — were produced in scriptoria by men whose faith the stories contradict.
They knew it, too. The Book of Leinster's copy of the great Ulster cattle-raid ends with a Latin colophon in which the scribe distances himself from what he has just spent weeks transcribing, calling parts of it the deceptions of demons and poetic invention, fit for the amusement of fools. Then he copied it anyway, in his best hand. Somebody thought it was worth the vellum, and vellum was calves.
The consequence is a tradition that arrives pre-filtered and pre-dated: the earliest written Welsh and Irish material is medieval, Christian and literary, whatever archaic strata sit underneath it. Anyone who tells you what the pre-Roman Gauls believed about a goddess, citing an Irish text written down fourteen hundred years later, is doing something the evidence does not support. The Celtic hub and the individual pages here are unusually blunt about this, because the popular literature is unusually bad about it.
A chronicle commissioned to settle a succession
Japan's foundational myth-texts have a completion date, a named compiler and a stated purpose, which is a luxury in this field and also a warning.
The Kojiki was presented to the court in 712 CE, compiled by Ō no Yasumaro from material recited by Hieda no Are, on a project begun under Emperor Tenmu. The Nihon Shoki followed in 720, written in Chinese, the prestige language of administration. Both run from the creation of the islands to the reigning imperial line without a break, which is not an accident of the material but the point of the commission: they establish descent from Amaterasu as a matter of record.
What redeems the Nihon Shoki as a source is a habit its compilers had no obligation to adopt. Again and again it gives the official version and then adds "in one writing it says", laying out variants it does not endorse. The birth of the sun goddess, the fate of Izanami, the killing of the food deity — the chronicle preserves rival accounts side by side, and this site follows it in that. A state-sponsored document that keeps its own contradictions in the margin is more honest than most modern retellings.
Clay, fire, and the accident that saved Sumer
The Mesopotamian corpus survives for a reason that has nothing to do with anyone's intentions: it was written on wet clay, and clay is improved by burning. Libraries that were sacked and torched baked their own archives into permanence. The scribal schools that copied Gilgamesh as a training exercise produced the redundancy that let modern scholars reconstruct broken lines from a dozen partial duplicates.
That is why the world's oldest surviving long poem can be read at all, and why the Standard Babylonian version has something close to an author's name attached — Sîn-lēqi-unninni, an editor working over much older Sumerian poems. It is also why this tradition, alone among the eight here, has a datable moment of modern rediscovery: an autumn afternoon in 1872 when a self-taught museum assistant in London made out a deluge story on a broken tablet from Nineveh and reportedly lost his composure entirely in front of his colleagues. What the flood tablet established — that the biblical account sits inside an older regional tradition — landed on Victorian Britain like a scandal, and the argument it started is not finished.
The tradition that refused to write anything down
India inverts the whole picture. The Vedic material was transmitted orally for a very long time by choice, and with a fidelity that no manuscript tradition can match, because writing was regarded as an inferior vessel for sacred sound.
The techniques were deliberate error-correction: reciting the text word by word in isolation, then in overlapping pairs, then in braided permutations, so that any drift in one recitation is caught against the others. It is redundancy engineering, achieved with nothing but trained memory, and it is why the Vedic gods can be compared against later Puranic ones with confidence about what actually changed.
The epics are the opposite case: the Mahabharata and Ramayana grew rather than being fixed, absorbing material over centuries and diverging into regional recensions that are not corruptions of a lost original but living versions with their own authority. Twentieth-century critical editions took decades to collate the manuscripts, and the result was not "the real text" but a scholarly reconstruction that no community has ever recited.
Rome, which rewrote the Greeks on purpose
The Latin layer deserves separating out, because it is the one most likely to be mistaken for the Greek one. When Roman religion took Greek myth on board it was not confusion — it was policy and literature, and it was done by identifiable authors with identifiable projects.
Ovid's Metamorphoses, finished around 8 CE, is the single most influential mythological work in Western art, and it is a witty, deliberately unstable poem by a man about to be exiled. Enormous numbers of "Greek myths" that everyone knows are Ovid's shaping and appear nowhere earlier — which is why Medusa's story looks so different depending on which century you read it in. Virgil built the Aeneid to give Rome a Homeric prehistory. Livy opened his history with material he half-signalled as legend. None of them thought they were falsifying anything; all of them were making something.
How to use this site once you know all that
The practical upshot is a reading order, not a warning label.
- Start with a story, not a god. The retellings are written at length because a myth is a narrative before it is an attribute list, and one good story teaches a tradition faster than twenty entries.
- Then read the tradition hub for Greece, Egypt, Iceland or wherever you landed, because the transmission story differs completely between them and it determines how much anyone can honestly claim.
- Use the family trees as maps of disagreement. A tree that shows a disputed parent is telling you where two sources fought; a tree with no disputed edges is telling you its maker picked winners.
- Check the equations before you trust them. The comparison pages exist because "the Roman version of X" is frequently false, and the creature pages exist because monsters cross linguistic borders more freely than gods do.
- Read the pop-culture divergences with the sources beside them, especially if a game or a film sent you here. Nothing on this site sneers at adaptations. Knowing exactly where one departs is more interesting than being told it did.
If you take one habit away, make it this: whenever you meet a confident mythological statement, ask which document it came from and what that document was for. It is a question that survives contact with almost any subject, and in this one it is the whole method.
Things readers ask once they start pulling this thread
If the oral versions are lost, how can anyone claim to know what changed?
By triangulating against material that had no reason to lie. Painted pottery, temple inscriptions, curse tablets, calendars of festivals and lists of sacrificial animals were produced for practical purposes and often disagree with the literary texts — and where an image on a sixth-century BCE vase shows an episode no surviving poem contains, that is direct evidence of a version the books lost. Archaeology cannot recover a story, but it can prove one existed.
Which single ancient work would I be maddest to lose?
Almost certainly the Nihon Shoki, on grounds of method rather than beauty. It is the only major mythographic project in this corpus that systematically records the versions it rejected. If the Greek, Norse and Egyptian traditions had been compiled by editors with the same habit, most of the disputes this site documents would be resolved facts instead of open questions.
Are there traditions where we know the writer was hostile?
Yes, and they must be read as evidence about the writer as much as the religion. A great deal of what is claimed about Continental European paganism comes from Roman generals describing enemies, or from Christian clerics describing practices they were suppressing — both with reasons to exaggerate savagery. Some of their detail is corroborated elsewhere; some has no support at all. The technique is to ask whether a claim would have helped the author's case, and to be sceptical in proportion.
Does any of this apply to religions still being practised?
It applies differently. Textual history is textual history everywhere, but a living tradition is not a set of documents — it is a community whose current practice is authoritative for its own members regardless of what a critical edition says. Pages here on Hindu material and on kami and Shinto practice are written in the present tense about present worship for that reason, and describe the classical corpus as one input to a living religion rather than as the religion itself.