Two Epics, Seven Cities, and No Confirmed Author

What We Know

In the summer of 1935, in a village in what was then Yugoslavia, Milman Parry set two turntables spinning and recorded a farmer named Avdo Međedović singing an epic poem that ran to fifteen or sixteen thousand lines. Međedović couldn’t read. He hadn’t memorized the poem, because there was no fixed version to memorize. He built it aloud, line by line, out of a stock of traditional phrases absorbed since childhood, the way a musician improvises within a key rather than reciting a score. Parry hadn’t traveled to the Balkans out of curiosity about folk singers. He’d gone to test whether the Iliad and the Odyssey could have been made the same way.

Nobody who actually heard the Iliad performed in the eighth century BC left a signed account of who wrote it. The first person on record even guessing at Homer’s dates is Herodotus, two and a half centuries later, and his guess is explicitly a guess: he supposes Homer lived “not more than four hundred years” before his own time, which isn’t a citation, it’s an educated shrug. By then, multiple Greek cities (seven, by the number handed down, though ancient sources don’t agree on which seven) were already claiming Homer as a native son, a dispute that only makes sense if nobody actually knew.

In 1795, the German philologist Friedrich August Wolf turned that ignorance into a formal problem. His Prolegomena ad Homerum argued that the poems we have aren’t the work of one literate author but a later-assembled sequence of oral songs. It made Homer’s identity a question to be argued from evidence, not assumed. Parry and his student Albert Lord picked the argument back up in the 1930s with turntables instead of philology alone, and by the time Lord published The Singer of Tales in 1960, they’d demonstrated something no one had shown before: that oral poets really could produce work at Homeric scale, live, with no text at all. It didn’t end the argument. It changed what the argument was about.

The Roadblock

The obstacle here isn’t missing evidence in the usual sense: a lost manuscript, an unopened archive, a witness who never talked. It’s that every piece of evidence anyone has ever had is the poems themselves, and the poems support more than one reading at once. The Iliad’s structural unity, its long-range echoes, its consistent characters across twenty-four books, looks like the fingerprint of one mind. Its formulaic language (“rosy-fingered dawn,” “the wine-dark sea,” phrases repeated exactly whenever the meter needs them) looks like the fingerprint of generations of performers reusing a shared toolkit, because that’s precisely what Parry and Lord watched living singers do. Both readings fit the same text. Stylometric analysis, which ought to settle a plain authorship question, keeps finding uneven ground instead: measurable linguistic divergence between books of the Iliad, with no agreement on why. There’s no forgotten document that would resolve this, because a document isn’t what’s missing. What’s missing is a way to tell a single genius using traditional material apart from a tradition that never needed a genius at all, and nothing in the text, examined closely enough, forces that choice.

Best Guesses

One Poet Composed Both Epics

This is the oldest reading: a single individual, probably eighth-century Ionia, composed the Iliad and, later, the Odyssey. Its weakness is total absence outside the text: no inscription, no contemporary reference, no biographical fact predating legend — everything ancient sources say about “Homer the man” was written centuries later, by people who knew no more than we do. What keeps the reading alive is the poems themselves: a shared poetic language, a worldview tight enough to feel authored rather than accumulated, sustained character logic across twenty-four books, and a plot architecture too load-bearing to have been improvised by committee.

A Tradition, Not a Person

Milman Parry didn’t set out to prove Homer didn’t exist. He set out to explain how a preliterate culture could produce poetry at that scale, and having watched Avdo Međedović do it in a Balkan farmhouse, he had an answer that didn’t require an individual genius: a long-lived formulaic system, refined over generations of performance, in which “Homer” is simply the name the tradition eventually acquired. Every repeated epithet in the text is exactly this kind of tool. But the traditions Parry and Lord actually recorded are wildly multiform — no two performances of a singer’s “same” song match exactly — while the Iliad and Odyssey survive as one stable text. A fluid tradition alone doesn’t explain how anything got fixed.

Different Hands, Later Combined

Nineteenth-century Analysts read the poems’ seams — repeated scenes, minor contradictions, characters who die twice — as evidence of multiple poets welded together, a method that could never say exactly where one hand stopped and another began. Parry and Lord showed most of those specific cuts don’t survive: repetition turned out to be a formal feature of oral composition, not a scar. But the position didn’t die with them: M. L. West’s 2011 revival argues the Iliad is almost entirely one poet’s work, a poet West names only “P,” writing over a long stretch of time, but a different poet from whoever made the Odyssey, and not necessarily anyone ever called Homer. Computational stylometry now backs part of that claim: 2021–2022 language modeling found measurable, book-by-book linguistic divergence within the Iliad itself.

One Dictation Fixed the Tradition

The Greek alphabet is strange: unlike every script it was adapted from, it marks vowels with the same precision as consonants, a level of phonetic accuracy no earlier alphabet needed. Barry Powell’s theory is that someone built it that way, around 800 BC, specifically to take epic verse down by dictation, Homer’s poems included, and that this one recording event turned a fluid oral tradition into the fixed text still read today. It’s an elegant answer to a real design puzzle: no earlier alphabet bothered with that kind of phonetic precision, and this one did. Even a sympathetic reviewer found the case built more on “a willingness to believe than on demonstrable probabilities” — a compelling story, without direct evidence it happened this way.

The Verdict

Nothing here is missing in the ordinary sense: a box in an archive, a fragment still waiting to be excavated. What would move this argument isn’t a document; it’s a method precise enough that every camp accepts its verdict, and nobody has built one yet. Computational stylometry is the closest attempt so far, and it has already found real, measurable seams in the text (divergence between books, outlier passages) without anyone agreeing on what those seams mean. A single-author reading, a layered-authorship reading, and a fixed-oral-tradition reading can each currently absorb the same statistical result. What would actually settle it is a stylometric method trusted enough, across all three camps, that its conclusions stick, or, short of that, an inscription or document from somewhere near 700 BC that names an actual person and calls him the poet of these lines. Neither has turned up in two hundred and thirty years of looking. Every theory here is younger than the poems it’s trying to explain.