The Sun Has Perished from the Sky

Homer, Odyssey 20.356. For two thousand years readers have wondered whether a real eclipse hides in that line. Astronomy offers exactly one candidate — 16 April 1178 BCE — and the argument over it is the best kind: rigorous on both sides, and still open.

In the twentieth book of the Odyssey, the suitors sit down to what they do not know is their last meal. Odysseus is already in the hall, dressed as a beggar; the poem is one day from its slaughter. Among the company is a seer, Theoclymenus. Mid-feast, something happens to his sight; he looks at the laughing suitors and tells them what he sees:

ἠέλιος δὲ οὐρανοῦ ἐξαπόλωλε, κακὴ δ’ ἐπιδέδρομεν ἀχλύς.
“The sun has perished from the sky, and a wicked mist rushes over us.” — Odyssey 20.356–57, Theoclymenus to the suitors · Greek text and translation via Sententiae Antiquae (opens in new tab)

Either that is the most famous eclipse in Western literature, or it is not an eclipse at all. The question is two thousand years old and still open. Here is the arc, receipts on every side.

Act I — the ancient readers

The eclipse reading is not a modern projection onto Homer. Two ancient authors are cited as having read the line as a real darkened sun: Plutarch, the biographer-essayist, and Heraclitus the Allegorist, a first-century AD grammarian whose Homeric Problems survives — and who is not the pre-Socratic philosopher Heraclitus of Ephesus. Different man, different century; only one of them is famous for the river you can’t step in twice.

Even the ancient endorsement carries an asterisk, though. The working classicist behind Sententiae Antiquae argues that the Plutarch passage everyone cites — from On the Face in the Moon — may be satirizing an interlocutor who piles up over-eager poetic “proof,” not endorsing the reading himself. Act I ends the way this whole story keeps ending: a reading you can see, and a hedge you can’t unsee.

Act II — one candidate in the sky

In the late 1920s two pioneering historians of ancient astronomy, Schoch and Neugebauer, ran the sky backward. Classical chronographers put the fall of Troy around 1192–1184 BCE; the poem’s own clock puts Odysseus’s homecoming a decade later. In that stretch, one total solar eclipse — and only one — was visible from the Ionian islands where Ithaca lies: 16 April 1178 BCE. NASA-derived recomputation gives it a shadow path roughly 228 km wide and about 4 minutes 33 seconds of totality near midday local time. If the line records an eclipse at all, the sky offers exactly one candidate — and that singularity kept the idea alive for the next eighty years.

Act III — four clues, one date

In 2008 the question got its most serious modern treatment — from outside classics. Constantino Baikouzis, of the Observatorio Astronómico de La Plata in Argentina, and Marcelo Magnasco, a Rockefeller University biophysicist, published “Is an eclipse described in the Odyssey?” in PNAS (online 24 June 2008). Their move was to stop leaning on the Theoclymenus line alone and treat four other passages in the poem’s back half as literal astronomy: a new moon on the day of the slaughter — the precondition for any solar eclipse (Books XIV, XIX, XX); Venus high before dawn five days earlier, as the Phaeacian ship carrying Odysseus nears Ithaca (Book XIII); Boötes and the Pleiades both visible at dusk twenty-nine days earlier — the stars Calypso tells Odysseus to steer by (Book V); and, most conjecturally, Hermes’ errand to Ogygia, thirty-four days earlier, read as the planet Mercury near western elongation (Book V). They then searched every new moon across a 135-year window, roughly 1250–1115 BCE, for a date matching all of it. One date survived: 16 April 1178 BCE — the same eclipse Schoch and Neugebauer had found from the other direction. By the paper’s own framing — the figure is the authors’, reported in the free full text at PubMed Central — a pattern that strict recurs roughly once in two thousand years.

Now the weaknesses, which the authors carry — to their credit — on the front of the paper, not buried in a footnote. The Mercury clue is both load-bearing and the shakiest: there is no evidence anyone linked the god Hermes to the planet Mercury before Plato — roughly seven hundred years after the proposed date. The argument also has to survive something like five centuries of oral transmission before the poem was fixed in writing. And back-calculating a 3,200-year-old eclipse carries positional uncertainty of about 2–3 degrees of arc — comparable to the width of the totality path itself. Magnasco said it plainly to the press: “This is a risky step in our analysis… One may say that our interpretation of the phenomena is stretching it, but when you go back to the text you have to wonder.” He also said the team had “no clue” whether it actually happened, and that the goal was simply to get a few people “to read The Odyssey differently” (quotes via ScienceDaily, from Rockefeller’s press materials). That is not a team overclaiming. That is a team publishing a beautiful coincidence and labeling it honestly.

Act IV — the classicist answers

The formal rebuttal came in 2012, from Peter Gainsford of Victoria University of Wellington: “Odyssey 20.356–57 and the Eclipse of 1178 B.C.E.,” in Transactions of the American Philological Association 142.1, pp. 1–22. His case: the astronomy only converges if you first grant a stack of tacit assumptions, and the biggest ones fail. Chief among them — nothing in the surviving ancient Greek record suggests that Homer or his audience read a god’s travels as coded planetary motion. Pull the Hermes-as-Mercury assumption, and the four-clue lock loses a tumbler.

Antiquity supplies its own witness here too. An ancient marginal note on the very line — Scholion B on Odyssey 20.356 — flatly denies the event: “A solar eclipse did not happen but Theoklymenos sees it this way as he tells a prophecy under divine influence since the sun will eclipse for these guys” (translation via Sententiae Antiquae). A Greek scholar, centuries closer to Homer’s world than we are, reading his own language, and ruling: this is a vision, not weather. The staging agrees with him. The scene is indoors, at a feast, and nobody else in the hall so much as glances up — a strange way to write a real sun going out. Darkness washing over doomed men is stock Homeric death-imagery besides — Hades furniture, not a celestial marker. The PNAS paper itself concedes that most Homeric scholars reject the eclipse hypothesis.

Where it stands

Open. Unresolved. Probably unresolvable — because the whole question hinges on whether Bronze Age poetic imagery can ever safely be read as instrument data, and that is not a question more computation can close. And that is the better story. “Scientists prove Homer encoded an eclipse” is a headline; what actually exists is richer: a coincidence elegant enough that readers have wanted it true for two thousand years, resurrected with real rigor by two scientists who flagged their own weakest link, then taken apart with equal rigor by a classicist — every side showing its work. The sun perished from the sky at the suitors’ last feast. Whether it also perished over the real Ionian sea on 16 April 1178 BCE, the record — honestly read — declines to say.