10 November 2026

Colour coding the Greek gods

An odd trend has arisen in the world of AI slop: the notion that each Greek god had a specific colour associated with them.

In terms of how ancient people imagined the gods, it’s completely bogus. In the 21st century, yes, people have been discussing colour schemes for the gods in internet forums and blog posts. But they aren’t talking about ancient depictions. They’re talking about modern art design, head-canons, colours for modern Wiccan ritual use, and other modern characterisations.

And that’s fine. Colour schemes can absolutely be a thing. It’s just that there’s nothing ancient about them.

Colour scheme as art design: a relatively early example in the film Hercules (1997). Front, left to right: Hephaestus (partially concealed), Ares, Aphrodite, Dionysus. Mid-ground: Helios, Persephone, Demeter, Athena, Poseidon, Apollo. I am indebted to Ov3rReadKn1ght0wl on AskHistorians for pointing out Hercules as an example.

AI slop on this subject seems to have started appearing in 2024. At that time, I suspect, these screeds of nonsense were being generated at human request. The pace has picked up since August 2025. Some threshold must have been passed: enough people had talked about their head-canons online to persuade more recent LLM models to start generating the slop automatically.

So now it’s in a feedback loop. From now on, this bogus idea is going to be repeated ad nauseam with no human input. In November 2026, if you ask an AI whether Greek gods had a regular colour scheme, it will give a confident ‘yes’, and provide supporting links. It’s just the links will point to other AI slop. (And to a handful of human opinion posts about modern art design.)

Colour coding in ancient sculpture?

Polychromatic sculpture

Ask an AI to substantiate a preconceived idea, and it’ll clutch at straws. One favourite tactic to say that a divine colour scheme is demonstrated by coloured statues.

Yes, ancient sculptures were polychromatic. No, there’s absolutely nothing to indicate they had specific colour schemes associated with specific gods.

In the modern imagination, Athena is often associated with the colour blue (as in the Disney film Hercules). But to rationalise that by pointing to a study of polychromatism on one Roman-era statue of Athena at the MFA Boston, which found traces of a blue pigment as part of a mix with yellow to produce brown eyebrows — that’s extremely silly.

If we had more data — if we could tabulate how frequently statues of Athena were dominated by a particular colour — it’d be a different story. But we don’t. There’s nothing indicative about a single datapoint.

Fortunately, people doing actual research on pigmentation in ancient statuary haven’t succumbed to this AI slop. So far.

‘Chryselephantine’ statues

Some ancient statues were what’s called ‘chryselephantine’, where chrys- means ‘gold’. But it isn’t gold the colour: it’s gold the metal. ‘Chryselephantine’ means that ivory (Greek elephas) and gold leaf (chrysos) were used in the production process. All substances have a coloured appearance, of course, but these weren’t usually colour terms in ancient Greek. Plus, it’s a standard technique for expensive statues, not something specific to one god.

Colour themed gods in the video game Hades II (2025). Left to right: Hestia, Hermes (rear), Aphrodite, Artemis, Hera, Zeus, Poseidon (rear), Apollo, Athena, Demeter, and Hephaestus. Unlike ancient Greek poetry, colour is important in video game art design. That’s because boons from the gods are part of the game mechanics, and they need to be instantly recognisable to the player.

Epithets in Greek poetry

With poetic epithets there’s more to discuss, because there are some colour-based epithets in hexameter poetry, and some are associated with particular gods ... to varying extents Also, in some cases they only look like they’re colour-based. Here’s a list, in descending order of how often they appear in early Greek epic.

γλαυκῶπις (glaukōpis) ‘glauk- faced, glauk- eyed’

Glaukōpis is very frequently used of Athena, and only of Athena: 36× Iliad, 57× Odyssey, 33× other early hexameter.

Glaukos definitely is a colour term. It’s traditionally translated as ‘grey’, but it doesn’t have a one-to-one mapping onto English colour terms. It does refer to a single coherent region of the colour-space: it was the colour of olives; vine leaves; grapes; a clear sky; seawater; and an eye colour. So glaukos overlaps with the regions of the colour space where English uses light blue, light green, and medium grey, all low-saturation colours.

The compound glaukōpis can mean either ‘glaukos-faced’ or ‘glaukos-eyed’. There’s a traditional preference for translating it as ‘grey-eyed’, simply because we have no other reason to assign unusual colours to Athena’s face.

And, to be difficult, formally it’s just as realistic to parse glaukōpis as ‘glaux-faced’, that is, ‘owl-faced’. Glaux is the nominative form of ‘owl’; the actual stem is glauk-. So if you did want to say ‘owl-faced’ in Greek, glaukōpis is certainly what you would say. It might seem a bizarre epithet, except for two points. (1) The owl is the most standard feature of Athena’s iconography in ancient art. (2) There’s a close analogy for ‘owl-faced/owl-eyed’ in an epithet used for Hera, boōpis ‘ox-faced/ox-eyed’.

On balance, it’s more likely to be a colour term. But it’s not a no-brainer. And the most important thing about glaukōpis isn’t its literal denotation, but the fact that it sounds incredibly cool no matter how you translate it. If it is a colour term, it only refers to her face or her eyes, not to other parts of her iconography.

Athena in Lore Olympus (2017–2024) is literally ‘grey-faced’ (glaukōpis). (Image source: Lore Olympus wiki)

λευκώλενος (leukōlenos) ‘white-armed’

Leukōlenos is used of Hera (24× Iliad, – Odyssey, 11× early non-Homeric hexameter), Helen (1× Iliad, 1× Odyssey), Andromache (3× Iliad), Nausikaa (5× Odyssey), Arete (3× Odyssey), unnamed enslaved women (2× Odyssey), Persephone (1× early non-Homeric hexameter), and Selene (1× early non-Homeric hexameter).

Leukos ‘white’ is straightforwardly a colour term. We have only suspicions about why so many women get called ‘white-armed’: they’re mostly to do with the idea that indoor labour is a symbol of luxury. Or maybe the idea is that aristocratic women are concealed indoors to make them sexually unavailable. We can only guess.

The problem is that it isn’t exclusive. ‘White-armed’ is often associated with Hera, yes, but 33% of the time it’s used for other women — including two other divinities. It’s a generic epithet.

So it’s a typical term for women, which happens to be often used for Hera. It isn’t colour coding specific to Hera.

χρύσ(ε)ος (chrys(e)os) ‘golden’

Chrysē, a contracted feminine form, is used sometimes for Aphrodite (5× Iliad, 5× Odyssey, 14× early non-Homeric hexameter). In addition, outside Homer polychrysos ‘with lots of gold’ is used of Aphrodite frequently (8× early non-Homeric hexameter)

The problem is, chryseos normally denotes a metal, not a colour: ’composed of gold’, or ‘adorned with gold’. If you look at ancient discussions of pigments or colour theory, chryseos isn’t one of the terms that come up.

And the compound form polychrysos makes things perfectly clear. That means ‘lots of gold’. That would make no sense if it were about colour. It’s about gold ornaments, jewellery and the like.

In addition, the formula looks like it must be quite late. As it stands it depends on two late linguistic features: (1) the contraction χρυσέη > χρυσῆ, and (2) a short initial syllable on Ἀφροδίτη (‘Attic’ correption), a phenomenon that is exceedingly rare in mid-word. So for Homer it’s a recent development, not something seared into the marrow of tradition.

Van Beek (2022) has made a substantial case that Homeric Ἀφροδίτη with the short initial syllable directly reflects a prehistoric Greek form with syllabic r, *Ἀφr̥δίτᾱ. In theory that could address the correption on the first syllable. But syllabic r died out in Greek before the Mycenaean period; and this formula also has contraction of the final -εη, which puts it very late. Anyway, χρυσός ‘gold’ wasn’t borrowed into Greek until the Mycenaean period (Myc. ku-ru-so), as a loanword from a Semitic language; in that era its adjectival form would have to have been vocalised something like χυρῡσέᾱ (by analogy with Ugaritic ḫrṣ, Phoenician ḥrṣ), so there’s no way it could be metrically equivalent to Homeric χρυσῆ. In addition, we don’t have grounds to abandon the possibility of a Semitic origin for the name Ἀφροδίτη. On precursor forms of χρυσός see Masson 1967: 37–38; on the alternative Semitic origin for Ἀφροδίτη see Beekes 2009: 1652.

κυανοχαίτης/-χαῖτα (kyanochaitēs/-chaita) ‘black-haired’

Kyanochaita and kyanochaitēs are used occasionally of Poseidon (6× Iliad, 3× Odyssey, 2× other early hexameter); Hades (1× early non-Homeric hexameter); and Arion (2× early non-Homeric hexameter).

It’s traditional to translate kyaneos as ‘blue’. And it can mean that. Just not consistently, and definitely not in this case.

Like glaukos, kyaneos has a territory in the colour space which doesn’t line up with that of any single English term. Kyaneos covers a range from deep blues to black, usually mid-to-high saturation. In fact, ancient Greek has two terms for ‘black’: kyaneos and melas. They both mean ‘black, plus an adjacent region of the colour space’, but on opposite sides of the spectrum: kyaneos on the blue side (so it can also be used for ‘blue, navy’), melas on the red side (so ‘brown, maroon’).

Their territory in the colour-space isn’t confined to what English calls ‘black’. That suggests that when they are used for ‘black’, there’s a perception of colour-adjacency — for example, that the choice of term depends on which colours appear in the highlights of a shiny surface. Red wine in an opaque vessel looks black, but its highlights put it on the red side of the spectrum, so in ancient Greek it’s melas. Clouds have a range of blue-ish colours, so a particularly black cloud is kyaneos. Very black hair and skin are also sometimes described as kyaneos.

Comic books provide a good illustration of the idea, because artists have to make a choice of which colour to use for highlights. Famously, Spider-Man’s costume was originally designed to be red and black, but people were misled by the blue highlights into thinking it was red and blue; so it became red and blue. Black hair usually has blue highlights.

Kyanochaitēs and specularity in modern comics. Notice the blue highlights in the black hair of Dr Strange, Wolverine, Nightwing, and Catwoman. Their hair is the kyaneos variety of black. By contrast, Tony Stark’s black hair often has brown highlights, so apparently his hair is melas. Very few comic book characters consistently have pure black hair: that seems to be confined to black men with very short hair, such as T’Challa and John Stewart.

The upshot is that Poseidon doesn’t have ‘blue’ hair, because while ‘blue’ and kyaneos overlap, they aren’t the same thing. Poseidon has black hair. And so does Hades.

χρυσοκόμης ‘gold-haired’

Chrysokomēs, ‘gold-haired’, is used of Dionysos once in early non-Homeric hexameter. As we saw earlier, it isn’t usual for ‘gold’ to be a colour term, but this term must be an exception. One instance isn’t enough to constitute a trend. In addition, when a substance is used to describe appearance, there’s always a degree of metaphor involved.

In the Iliad, the harsh, unforgiving setting of a battle scene is conveyed by references to ‘iron’ at ground level and ‘bronze’ in the sky. That doesn’t mean they’re literally metallic in appearance. In Odyssey 6, Athena disguises Odysseus by giving him hair ‘like a hyacinth flower’, and the Hymn to Demeter gives Keleos’ daughters hair ‘like a crocus’. That doesn’t mean their hair colour is literally lilac, or magenta, or purple. They’re figurative. Often we can’t tell exactly what flavour the metaphor is trying to convey, because only someone of that time could tell us.

Modern colour codings for Dionysus often make him purple, because ... um, wine is purple? (A bit of advice: if you ever find yourself drinking purple wine, pause to consider where your life has gone wrong.)

‘Golden Apollo’

I don’t even need to point again to the discussion of ‘gold-haired’ above, because this one is just false. No ancient source calls Apollo ‘golden’ or ‘golden-haired’ — contrary to what I’ve seen widely claimed.

He does have some objects made of gold associated with him. As does practically every other Greek god: so, not exactly specific. I’ve seen it claimed that he has a golden bow; in poetry his usual epithet is actually argyrotoxos, ‘silver-bowed’ (11× Iliad, 3× Odyssey, 9× early non-Homeric hexameter). There’s just one isolated use of chrysotoxos Apollōn ‘golden-bowed Apollo’ in a Boeotian poet (Pindar, Olympian 14.11).

Objects associated with gods

There are other object epithets for gods, denoting a material rather than a colour. So these have nothing to do with a colour scheme. Artemis is chrysēlakatos ‘with golden arrows’; Apollo is argyrotoxos ‘with a silver bow’, and also chrysaōr ‘with a golden sword’. Aias has a bronze shield: that doesn’t mean he’s typically associated with the colour russet.

So what are we left with?

Athena has light blue-grey-green eyes — or alternatively her face is blue-grey-green, or at some point some people thought of her as a theriomorphic god with an owl’s face: these are all theoretically justifiable. Poseidon has black hair. Some gods have certain metal objects associated with them (gold jewellery, silver bows and arrows).

And that’s all. There is no other evidence for divine colour schemes.

Still, some modern divine colour schemes are comparatively influential. And there’s nothing wrong with that: just don’t claim that they reflect anything ancient. For the sake of completeness I’ll list a few here, but with a heavy emphasis on the fact that they’re exclusively modern.

God Hercules (1997) Lore Olympus (2017–2024) Hades (2020–2025)
Ares red yellow red
Aphrodite pink, sienna magenta pink
Artemis green, yellow purple green
Athena blue grey yellow
Demeter green green white
Dionysos purple pink purple
Hera magenta, orange yellow blue
Hephaistos black orange muted range
Hermes blue red various
Poseidon blue green blue, blue-green
Zeus yellow purple yellow

References

  • Beekes, R. 2009. Etymological dictionary of Greek. Leiden/Boston.
  • Masson, É. 1967. Recherches sur les plus anciens emprunts sémitiques en grec. Paris.
  • Van Beek, L. 2022. The reflexes of syllabic liquids in ancient Greek. Linguistic prehistory of the Greek dialects and Homeric Kunstsprache. Leiden/Boston. [Open-access; DOI]

30 October 2026

Sibyls and Sibylline oracles. 4. The Renaissance

1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance

The Sibyls were thoroughly christianised after Lactantius finished with them. He successfully reimagined them as virtuous pagans of the distant past. Most later readers imagined the Sibylline oracles poems were authentic predictions of Christ and the end times.

Their reputation continued to expand in late antiquity and the Mediaeval period. According to Finn Schulze-Feldmann (2017), they were integral to a wave of Christian apocalyptic literature: they were an ideal hook to hang new prophecies on.

Delphic Sibyl: Filippino Lippi, Carafa Chapel ceiling, Santa Maria sopra Minerva, Rome, 1488–1493. The scroll in the Sibyl’s hand reads propheta ex virgini (for virgine) nascetur, ‘a prophet will be born of an unmarried girl’. (Image source: Wikimedia)

Into the Mediaeval period

The first step was that the acrostic in Sibylline oracles 8 took on a life of its own. We find it quoted in the pseudo-Constantinian Speech to the assembly of saints §18 (late 300s), and Augustine gives a Latin translation (City of God 18.23, early 400s). But they seem to have known it in an excerpted form. Augustine doesn’t know the last part of the acrostic; and they both ascribe the poem to the Erythraean Sibyl — evidently not agreeing with Lactantius’ idea that a different Sibyl wrote each poem.

Bear in mind I can only pick out a handful of highlights. In the early 700s Bede wrote a short but influential treatise on the Interpretation of sibylline sayings with a condensed version of Lactantius’ account, with the novelty that he names Cassandra as the Tiburtine Sibyl. Several later writers draw on Bede’s account.

Note. Souda σ.359 (900s) also names Kassandra as a candidate for the identity of the Phrygian Sibyl. The only ancient precedent for counting Kassandra as a Sibyl seems to be that Lykophron, Alexandra 1464, compares Kassandra to Melankraira, which is one of several names given for the Cumaean Sibyl (pseudo-Aristotle, Mirabilia 838a).

Ælfric’s Christmas homily (990s) refers to several alleged pagan prophecies of Christ. ‘Sibylla’ is his first example (tr. Thorpe, pp. 18–19):

...heathen men also have set in their books concerning all these things which we have now read before you. One of these was Sibylla, who wrote in song-craft wise of Christ’s birth, and of his passion ... and yet was a heathen.
The ten Sibyls as depicted in a manuscript of the Romuleon, cod. Laurenziana Med. Palat. 156 (1464–1465). (Image source: Internet Archive)

Sibyls appear more and more frequently as the Mediaeval period goes on. Sometimes they’re cited as a religious and moral paradigm. Abelard’s seventh letter to Héloise (1130s) draws on Augustine, Lactantius, and Vergil’s fourth Eclogue, to cast ‘Sibylla’ as a model for women leading a religious life: he likens Sibylla more to an evangelist than a prophet.

In a similar vein Petrarch’s Rerum memoranda 4.30 (mid-1300s) discusses four of the Sibyls, and says the Delphic Sibyl’s prophetic power was a reward for her virginity. Incidentally, this shows a conflation of the Sibyl with the Pythia. The Pythia was a historical job title: she was a priestess who delivered oracular pronouncements on behalf of Apollo at Delphi. The Delphic Sibyl, by contrast, is purely mythical, described originally by Chrysippos in his lost book On divination (Stoicorum veterum fragmenta ii.1216). I haven’t found out who was the first to conflate them, but it must surely be earlier than Petrarch.

Hellespontian Sibyl: Jörg Syrlin the Elder, choir stalls of the Ulmer Münster (1468–1474). The inscription reads Sibilla Ellespo(n)tica in agro Troiano: felix ille dives ligno qui pe(n)det ab alto, ‘The Hellespontian Sibyl, in the Trojan country: a blessed rich man is one who hangs from a high beam.’ The second line is a hexameter verse spoken by ‘Sibylla’ in Iacopo de Voragine’s Golden legend (1260s; p. 792 ed. Graesse, 1846). I’m aware dives is a standard variant for deus — ‘blessed is that god’ — however, I like the Ulm text just the way it is. (Image source: Wikimedia)

Sometimes the interest is more antiquarian, like encyclopaedic works treating the Sibyls as historical figures. Vincent de Beauvais’ Speculum historiale 2.100–102 (mid-1200s) presents them this way, synthesising Lactantius’ account with a bundle of other material. Boccaccio combines moral and encyclopaedic approaches when he includes the Erythraean and Cumaean Sibyls in his catalogue of famous women, De claris mulieribus (1362).

On mediaeval accounts of the Sibyls, see further McGinn 1985: 18–21; Raybould 2017: 32–48; Schulze-Feldmann 2017. Boccaccio: pp. 41–42 and 50–51 in the English translation by Guido Guarino. Boccaccio also has chapters on other prophetesses of antiquity, Manto (60–61), Cassandra (70; recall that Bede had regarded her as a Sibyl), and Nicaula, the queen of Sheba, who appears in some later Sibyl-lists.

The Quattrocento

All of this leads up to an outpouring of Sibyl-related artworks in the 1400s and 1500s. These include Michelangelo’s Sibyls on the ceiling of the Sistine Chapel.

Layout of the Sibyls in the nave of Siena Cathedral (1480s). (Base image: Wikimedia; my annotations)

The most important depiction of the ancient Sibyl canon is the floor panels in the nave of Siena Cathedral (1480s). The centre aisle is dominated by panels of Hermes Trismegistus, Romulus and Remus, and the Mount of Wisdom; the side aisles have the full set of ten Sibyls, with individual iconography and prophetic ‘quotations’. (The floor is fully uncovered for only a few weeks each year, so plan your visit accordingly: timetable here.) The Siena Sibyls were completed just 20 years after the first printed edition of Lactantius came out in 1465.

A few other notable artworks contained the full set of ten Sibyls:

  • Rimini, the Tempio Malatestiano (ca. 1454): one of the side chapels, the Capella dei Martiri, is flanked by two pillars with sculptures of all ten Sibyls and two Hebrew prophets. (Selection of photographs here.)
  • Ulm, Münster: the choir stalls have wood carvings by Jörg Syrlin (1468–1474) of all ten Sibyls, eight classical philosophers and poets, and the queen of Sheba as an eleventh quasi-Sibyl.
  • Filippino Lippi, who worked in Botticelli’s workshop, painted a pair of paintings featuring the ten Sibyls (ca. 1472); the paintings are held at Christ Church, Oxford.
The Christ Church Sibyls (ca. 1472). Left to right: Persian, Libyan, Delphic, Cimmerian, Erythraean, Samian, Cumaean, Hellespontian, Phrygian, and Tiburtine. Text beneath the Sibyls (not shown) quotes various ‘sayings of the Sibyls’.

Depictions of the Sibyls normally have short textual prophecies associated with some or all of the Sibyls. The sources aren’t always easy to track down. Some come from Lactantius, others are obscure. Above, in the caption to a photo of one of the Ulm Sibyls, I noted a hexameter line quoted from the Golden legend.

There are also plenty of partial sets of Sibyls. Lippi later went on to paint four Sibyls on the ceiling of the Carafa Chapel at Santa Maria sopra Minerva, Rome (ca. 1490; image at top). And of course there are the five Sibyls in the Sistine Chapel (1508–1512). As was often customary — most notably in the case of the Orsini Sibyls, below — Michelangelo’s Sibyls are placed alongside prophets of the Hebrew Bible.

The Sistine Chapel ceiling (1508–1512), with Sibyls and prophets highlighted.

In one notable case, we have a group of Sibyls who are completely disconnected from the Varro-Lactantius canon: the Cortina Sibyls.

These are a group of five Sibyls — originally there must have been more — in an early 15th century fresco at Cortina d’Ampezzo, in the far north of Italy. The fresco was concealed behind plasterwork at the Cassa Rurale ed Artigiana and discovered during restoration work in the 1890s. The names are damaged, but the surviving portions have nothing in common with Lactantius’ ten Sibyls.

The Cortina Sibyls (ca. 1400–1420). Left to right: [Sibilla V]alve(n)sis; Sibilla [N]icaulia (the queen of Sheba); Sibilia Po(n)tue(n)sis (Po(r)tue(n)sis?); Sibilla A[grip(?)]a; and a Sibyl whose name is lost. The Cortina Sibyls wear crowns, which aligns them with older portrayals of the personified Virtues. See Raybould 2017: 67–75; Schulze-Feldmann 2017: 206. (Image source: cortina.dolomiti.org; Cornford 1903: 370 gives a useful line drawing.)

It’s not unusual for extra Sibyls to pop up here and there. Above I mentioned that Bede included Cassandra in his canon. Nic(h)aula, the queen of Sheba, is one of the Sibyls in the Cortina fresco, above; sometimes she’s added to a Sibyl-list as an extra, a quasi-Sibyl. She appears among the woodcarvings in the choir stalls at Ulm; and a 16th century chapbook, published by Simrock in the revised edition of his Deutsche Volksbücher (1869), gives a catalogue of the twelve ‘Orsini Sibyls’ (see below) but then appends Nichaula as a thirteenth Sibyl.

One important Sibyl was never part of any canon: the Apennine Sibyl. She’s more folklore than anything else. She first appears in an early 15th century book, Andrea de Barberino’s Il Guerrin Meschino (1410); later also in Antoine de La Sale’s La Salade (ca. 1440). The story is that she complained about Mary’s privileged position, thinking she herself deserved to be the mother of Christ; as a punishment, she was condemned to dwell in a cave inside a mountain until the end of time — in either Monte Sibilla or Monte Vettore, two neighbouring mountains in a part of the Apennine mountain range in central Italy known as the Monti Sibillini.

Raybould suspects the Apennine Sibyl may be one of the Cortina Sibyls. He links the far left Sibyl, the Sibilla Valvensis, to the former Diocese of Valva — Latin Valvensis — or its successor, the Diocese of Sulmona-Valva: both had their seat at L’Aquila, 50 km south of Monte Vettore. (I have my doubts: 50 km isn’t that close, and there are some big mountains in between.)

See Raybould 2017: 59–66 for an account of ‘independent’ Sibyls.

The Orsini Sibyls: prose Vaticinia, poetic Prophetiae Sibyllarum

The most influential new canon was the Orsini Sibyls. Frescoes in the entrance hall of the Palazzo Orsini at Monte Giordano, Rome, which were completed by 1434, depicted twelve Sibyls alongside twelve prophets.

These paintings came with new ‘prophetic’ texts, supposedly spoken by the Sibyls. I’ll refer to these texts as the Vaticinia, following Nerida Newbigin (2018). Copies of the Vaticinia, with descriptions of the frescoes, circulated in a variety of forms for many decades.

The Orsini Sibyls have the old Varro-Lactantius canon, plus two new additions: the Sibylla Europaea, or European Sibyl; and the Sibylla Agrippa (or Agripa, or Egrippa, or Agrippina). It’s always been suspected that the ‘Agrippan’ Sibyl originated as a textual corruption of Aegyptia ‘Egyptian’; other theories exist. The Orsini Sibyls quickly became the standard canon for 15th century literature, theatre, and music.

On the Orsini frescoes and the new Sibyl canon see Settis 1985: 99–101; Newbigin 2018: 316. On the name Agrippa: Settis 1985: 95–98 derives it from a portrait of the ancient statesman Marcus Agrippa; McGrath 2007: 268–273 is non-committal; Roth 2010: 51 n.11 follows Émile Mâle in deriving it from Aegyptia. Note that A[grip]a is a possible supplement to one of the damaged Cortina Sibyls, above.

Newbigin gives an admirably clear account of how the Orsini Vaticinia propagated. In the mid-1400s they appeared in at least two vernacular translations, most notably one by Rustici in the ‘Codice Rustici’; and in Florence they started to be used in plays.

Sacre rappresentazione were traditionally grand stagings of biblical stories, with expensive costuming and music. A common feature was to open with a procession of the prophets, or Ordo prophetarum, ending with a Sibyl who would sing a Latin hexameter poem.

Young 1922 gives texts of various versions of the Ordo prophetarum. One, at 13, includes a repeated refrain. Florence cod. Riccardiana 1271 fols. 70v–71v has a list of the Lactantius Sibyls with their Orsini Vaticinia; Marco di Bartolomeo Rustici’s Dimostrazione, an account of an imaginary pilgrimage, includes accounts of first the Varronian canon, then the Orsini canon, with vernacular translations of the Vaticinia. See Newbigin 2018 passim.

An Annunciazione play by Feo Belcari (1460s or 1470s) included eight of the Sibyls, and each Sibyl declaimed or sang their Vaticinium in a versification of Rustici’s text. Around 1470 there appeared a set of engravings by Baccio Baldini, apparently partly based on Belcari’s play, featuring all twelve Sibyls and with vernacular verse versions of the Vaticinia. A later recension of the Annunciazione play also included the full set of twelve.

See Heitz 1903 for a facsimile of the St Gallen blockbook; again, on the entire sequence of developments, see Newbigin 2018.

Around the same time as Baldini’s engravings, another set appeared in a famous blockbook at St Gallen, Switzerland, with the Latin Vaticinia but without the descriptions of the Orsini Sibyls. Here, as an example, is how the Agrippan Sibyl appears — one of the additions in the Orsini canon — with her description and Vaticinium as they appear in a 1481 book published by Filippo Barbieri.

The twelfth and final Sibyl is depicted about 30 years old, mature and serious; clothed in Roman style, wearing a tunic of red velvet; and she holds a text, ‘The Agrippan Sibyl speaks about Christ’. Her prophecy is as follows:

‘The Word will be invisible but felt; it will spring up like a root, be dried like a leaf, and its beauty will not be seen. A mother’s womb will be encompassed around him, and God will weep with eternal joy. He will be trodden by people; be born from a mother as God; pass his life like a sinner.’

The Florentine plays have her speak (or sing) the same prophecy in Italian; the Latin version, starting invisibile verbum..., appears in many artworks over subsequent centuries.

As Newbigin shows (2018: 316–317), Barbieri himself is not responsible for the descriptions of the prophets and Sibyls; he inserted them into his book from elsewhere.
Versions of the Agrippan Sibyl and her Vaticinium. The St Gallen blockbook (ca. 1470; facsimile from Heitz 1903); pseudo-Barbieri, 1481 edition; illumination in a manuscript, Paris Bibl. Mazarine cod. 412 fol. 17 (ca. 1490); pseudo-Barbieri, 1500–1525 edition with the addition of a hexameter poem, one of the Prophetiae Sibyllarum.

A further development followed in the early 1500s. A new edition of Barbieri’s book came out posthumously, including the portraits, the descriptions, and the Latin Vaticinia, but also adding verse ‘prophecies’ after the Vaticinia.

These poems may be referred to as the Prophetiae Sibyllarum. That title comes from a musical setting of them by the great choral composer Roland de Lassus (ca. 1556–1558). Each Sibyl has six lines of Latin hexameter.

The image above shows a sample, at right. The right-hand facing page has the portrait; the left-hand page has the description, then the prose Vaticinium, then the hexameter Prophetia poem.

The Prophetiae poems have only the loosest resemblance to the Vaticinia. Reinhold Schlötter and Marjorie Roth have argued that some of them were based on the Belcari-Baldini vernacular versions; that may be. More interesting to my mind is Roth’s argument that Lassus’ musical setting of the Prophetiae was designed for an aristocratic revival of the sacre rappresentazione theatrical form.

She suggests this explains the presence of Lassus’ prologue, which isn’t in any other version. Its wording is quite specific to Lassus’ setting:

Carmina chromatico quae audis modulata tenore,
haec sunt illa quibus nostrae olim arcana salutis
bis senae intrepido cecinerunt ore Sibyllae.

The songs you hear, modulated in a chromatic tenor,
are the ones that long ago gave the secrets of our salvation:
twice six Sibyls sang them with fearless voice.

‘Chromatic tenor’ describes Lassus’ setting precisely. It would be interesting to know who wrote this prologue in the 1550s: the quality of the Latin verse is of a high standard.

It’s really Lassus’ Prophetiae Sibyllarum that led me to extend this account of the Sibyls into the Renaissance. His setting is truly wild — one of the most extraordinary, imaginative, unexpected things I’ve ever heard. Give it a listen: it’s genuinely hard to find any place where you can predict what chord is coming next.

The opening of the prologue in Lassus’ setting of the Prophetiae Sibyllarum (ca. 1558); note-lengths halved for legibility. In spite of the high quality of the prologue’s Latin verse, Lassus’ setting almost completely ignores the rhythmic features of the dactylic hexameter, including elision. In 16th century choral settings of Latin hexameter, settings that use homophonic chords typically represent the dactylic rhythms rigorously; polyphonic settings simply ignore the hexameter. Lassus’ setting of Vergil’s Eclogue 1, ‘Tityre tu patulae’, is notable for switching between the two approaches in different sections.

I’ve written a piece about Lassus’ choral setting before: I had a number of small misunderstandings, as I’ve now found out in doing the research for this write-up, so I’m glad to have a chance to correct them here. The misunderstandings were over the intertwined origins of the Prophetiae poems and the prose Vaticinia; the role of Belcari’s plays; and the misattribution of the Vaticinia to Barbieri. In addition I was completely unaware of Rustici’s translations of the Vaticinia in the 1450s.

I began this series wanting to clear up in my mind the disconnect between Lassus’ Prophetiae and the ancient Sibylline oracles. The fact that it took four blog posts to get here simply reflects that the story is convoluted. At every stage in their history, the Sibyls have been ferociously esoteric. Lassus’ eerie music directly mirrors their enigmatic nature. If I’ve confused things by giving too many details, Newbigin’s lucid article will clear up any misunderstandings in Part 4, at least.

Final note: I have written ‘Sibyl’ with a capital S, contrary to the practice of treating it as a generic term. This is because it is not generic: it is part of the proper name of each Sibyl. The lower case is only appropriate in cases where it is not part of a proper name. Independent sibyls, such as the Apennine Sibyl, are edge cases.

Final note 2, two weeks later: speaking of translations, I meant to include a reference to Jane Seager’s translation of the Prophetiae poems into English iambic pentameter (1589). But I forgot. Seager’s English poems are in a manuscript dedicated to queen Elizabeth titled The divine prophesies of the ten Sibills, Brit. Lib. MS 10037. She omits two Sibyls to bring the number down to ten, as in Lactantius’ canon: she keeps the two new Orsini Sibyls, the European and Agrippan Sibyls, but removes the Hellespontian and Phrygian Sibyls. Her sequence is also different. Here’s the Persian Sibyl, in the Latin Prophetiae version and in Seager’s translation:

Virgine matre satus, pando residebit asello,
iucundus princeps, unus qui ferre salutem
rite queat lapsis: tamen illis forte diebus
multi multa ferent, immensi fata laboris.
solo sed satis est oracula prodere verbo:
ille deus casta nascetur virgine magnus.

A joyfull prince borne of a virgine chast
sytting upon an asses colt shall come:
to rayse them up that fall their synnes to cure,
though peradventure in those dayes shal be,
some that shall suffer great affliction,
allotted them by dyvers destynies,
lett it suffize, that one worde maye explane
an oracle, or prophesye divine:
this mighty god, this king as erst I sayd,
shal be brought forth by a most blessed mayd.

References

  • Cornford, F. M. 1903. ‘Fresco at Cortina d’Ampezzo.’ Proceedings of the Cambridge Antiquarian Society 10: 369–373. [Internet Archive]
  • Heitz, P. 1903. Oracula sibyllina (Weissagungen der zwölf Sibyllen). Nach dem einzigen, in der Stiftsbibliothek von St. Gallen aufbewahrten Exemplare. Straßburg. [Internet Archive]
  • McGinn, B. 1985. ‘Teste David cum Sibylla: the significance of the sibylline tradition in the Middle Ages.’ In: Kirshner, J.; Wemple, S. F. (eds.) Women in the medieval world. Essays in honor of John H. Mundy. Oxford. 7–35.
  • McGrath, E. 2007. ‘Jacob Jordaens and Moses’s Ethiopian wife.’ Journal of the Warburg and Courtauld Institutes 70: 247–285. [JSTOR]
  • Newbigin, N. 2018. ‘Feo Belcari, Baccio Baldini, and the texts of the prophets and sibyls.’ Romance philology 72.2: 313–352. [JSTOR]
  • Raybould, R. 2017. The Sibyl series of the fifteenth century. Leiden/Boston.
  • Roth, M. A. 2010. ‘Prophecy, harmony, and the alchemical transformation of the soul: the key to Lasso’s chromatic Sibyls.’ In: Wuidar, L. (ed.) Music and esotericism. Leiden/Boston. 45–76.
  • Schulze-Feldmann, F. 2017. ‘The Sibyls at the gateway to the Renaissance.’ Bruniana & Campanelliana 23.1: 203–209. [JSTOR]
  • Settis, S. 1985. ‘Sibilla Agripa.’ Études de lettres: revue de la Faculté des Lettres de l’Université de Lausanne 1985.4: 89–124. [E-Periodica]
  • Simrock, K. J. 1869. ‘Zwölf Sibyllen Weissagungen.’ In: Auserlesene deutsche Volksbücher, vol. 1. Frankfurt. 383–419. [Internet Archive]
  • Young, K. 1922. Ordo prophetarum. Madison, WI. [Internet Archive]
1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance

22 October 2026

Sibyls and Sibylline oracles. 3. The Sibylline oracles

1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance

As we saw in Part 2, some fragments of the Sibylline books were in public circulation in pagan Rome. Cicero and Dionysios of Halikarnassos accurately describe some aspects of the surviving fragments. There were also imitators.

Vergil at work: illustration before Eclogue 4 in cod. Vat. lat. 3867, fol. 9r (image source: Vatican Library)

Vergil’s fourth Eclogue is a relatively loose example. It’s framed as a ‘prophecy’ linked to the Cumaean Sibyl, though it doesn’t try to replicate the style or imagery of the Sibylline books. Vergil doesn’t give instructions for rituals, and there are no acrostics: as Parke puts it, it’s ‘a transposition of Sibylline motifs into a pastoral key’. Vergil foretells a glorious future golden age surrounding a child who is to be born in the consulship of Asinius Pollio (40 BCE). It’s hard to avoid thinking that aspect may have been partly inspired by the childbirth mentioned in the surviving Sibylline books quotation.

Parke 1988: 145. Pollio’s consulship: Eclogue 4.3, 4.11–12. Birth of a male child: 4.8 onwards. Acrostics: a few Vergil scholars have noted that the first letters of Eclogue 4.47–52 spell out cacata ‘shitty’, but nearly all scholars regard it as an accident. If Vergil had really wanted to do an acrostic, he was perfectly capable of doing something comparable to the Sibylline books, rather than putting an isolated word of doubtful implication at a random point in mid-poem.

The Sibylline oracles are closer imitations. These are a group of eleven poems and eight fragments, mostly Jewish, with an assortment of Christian additions, in Greek hexameter verse. They range in date from the 2nd century BCE to the 7th century CE. Several of the poems list off dynasties of kings and emperors, framing them as though they’re predictions. In some poems the timeline reaches all the way from the Creation (as in Genesis) up to the end times and the Last Judgement (as in Enoch or Revelation).

A sample

Here’s an excerpt: Sibylline oracles 4, lines 125–139, about various disasters ‘predicted’ to occur in the 1st century CE (trans. John Collins).

(125) A leader of Rome will come to Syria who will burn
the Temple of Jerusalem with fire, at the same time slaughter
many men and destroy the great land of the Jews with its broad roads.
Then indeed an earthquake will destroy at once Salamis and Paphos
when the dark water overwhelms Cyprus, which is washed by many waves.

(130) But when a firebrand, turned away from a cleft in the earth
in the land of Italy, reaches to broad heaven,
it will burn many cities and destroy men.
Much smoking ashes will fill the great sky,
and showers will fall from heaven like red earth.
(135) Know then the wrath of the heavenly God,
because they will destroy the blameless tribe of the pious.
Then the strife of war being aroused will come to the west,
and the fugitive from Rome will also come, brandishing a great spear,
having crossed the Euphrates with many myriads.

This covers Titus’ assault and the destruction of the Second Temple in 70 CE; an earthquake on Cyprus; the eruption of Vesuvius in 79; and the coming of one of the various false Neros, coming westward with Parthian support (probably the one mentioned in Dion Cassius 66.19.3, who came west as far as the Euphrates at some point in 79–81).

After this point S.O. 4 stops predicting ‘history’ and moves on to the end times. The references above are detailed enough that the poem should be understood to have been written not long after 81 CE.

Opening of Sibylline oracles 1 in the edition of Geffcken (1902: 6)

How many poems?

Yes yes, I know, lots of online sources say ‘fourteen’. They’re wrong. Modern editions assign the poems numbers up to 14. But the actual number of poems is eleven — or thirteen, or fifteen, depending on how you count: fourteen is the only count that’s definitely wrong.

The confusion is because the Oracles were transmitted in two overlapping collections.

The first collection, S.O. 1 to 8, has never been lost, and has been in print since the 1500s. Poems 1 and 2 are actually a single poem, but the first modern editor, Sixt Birck (a.k.a. Betuleius), decided to count them separately for no good reason, and unfortunately the numbering is now pretty much set in stone. There are indications that there was once a further poem in between S.O. 2 and 3 — the opening section of S.O. 3 is clearly from a separate poem —; also, as we’ll see below, that Lactantius had a copy of this collection, and that his copy had nine poems. So this collection can be counted as seven, eight, or nine poems.

The second collection was discovered in the 1800s. The first editor carried on the numbering from eight, and there are seven poems, so they’re numbers 9 to 15. But three of them — S.O. 9, 10, and 15 — duplicate material from the first collection. So modern editions only ever print the new ones, poems 11 to 14. (Poem 12 is a continuation of poem 11, but they were composed at different times and should be counted separately.)

So the full modern collection is S.O. 1/2; a missing poem of which fragments survive; S.O. 3 to 8; and 11 to 14. Eleven poems — or fifteen if you use Betuleius’ count, and count the duplications in S.O. 9, 10, and 15. Between them, they come to over 4,000 lines of hexameter.

You’re right to be confused. Recent commentaries on Sibylline oracles 1/2 and 3 assume you’re already familiar with at least parts of this. It’s explained in the introduction to the standard critical edition, by Johannes Geffcken, but if you can’t read Geffcken’s German, things are going to be bewildering. A much friendlier guide is one given by John Collins in The Old Testament pseudepigrapha, volume 1. Collins gives an excellent introduction to each poem; a full English translation; and detailed notes. At the time of writing, Collins’ translation is the only good English translation of the Oracles.

Critical edition: Geffcken 1902. English edition: Collins 1983. Commentary on S.O. 1/2: Lightfoot 2007. Commentary on S.O. 3: Buitenwerf 2003. On the opening section of S.O. 3 being a fragment from a separate poem, see Buitenwerf 2003: 79–94; Collins 1983: 359–360. (Note that Buitenwerf and Collins have opposing views on whether S.O. frs. 1–3 should be assigned to S.O. 3 or to another lost poem).

The contents of the Oracles

Here’s an overview. The dates are those assigned by Collins (but note that some of the dates aren’t universally agreed).

First collection (φ and ψ manuscript groups; first ed. Betuleius, 1545):

Poem Date Origin and contents
1/2 before 150 CE Phrygian Jewish original. Timeline from Creation to Last Judgement; some phases removed and replaced by a Christian section on Christ and the fall of Jerusalem (1.323–400).
3.1–92 in phases, II BCE–I CE? Fragment. Jewish (with Christian elements?). Hymn to God; rule of Rome; destruction of Temple; return of Nero, called ‘Beliar’. (93–96 = a further fragment.)
3.97–829 II BCE Egyptian Jewish. Miscellany: Tower of Babel, Titanomachy, Exodus, Exile, oracles against foreign nations, ethics, eschatology.
4 late I CE Jewish. Timeline of empires from Assyrian empire to fall of Jerusalem (600s BCE to 70 CE); end times.
5 late I CE Egyptian Jewish. Timeline, 30 BCE–70 CE: fall of Ptolemies to fall of Jerusalem.
6 unclear Christian. Short hymn (28 lines).
7 II–III CE Christian. Historical snippets; the Flood; coming of Christ; end times.
8 II–III CE Jewish (1–216) and Christian (217–428) sections. Religious perspective on world history. Christian acrostic at 217–250.

Second collection (Ω manuscript group; first ed. Mai, 1825):

Poem Date Origin and contents
9   Duplicate: = S.O. 6.1–28 + 7.1 + 8.218–428.
10   Duplicate: = S.O. 4.
11 early I CE? Jewish. Timeline from the Flood up to the death of Cleopatra (to 30 BCE).
12 ca. 235 CE Jewish; later continuation of S.O. 11. Some duplication: S.O. 12.1–11 = S.O. 5.1–11. Timeline, 30 BCE–235 CE, Augustus to Alexander Severus.
13 ca. 265 CE Jewish (with one Christian line added). Historical: ca. 240–265 CE, new Palmyrene kingdom under Odaenathus, and Parthians under Mareades.
14 VII CE Alexandrian Jewish. Jumble of references: later Roman emperors, Egypt, Rashidun conquest of Syria/Egypt.
15   Duplicate: = S.O. 8.1–9.

Why ‘Sibylline’?

These poems weren’t designed for a pre-existing collection of Sibylline material: they were composed separately from one another, over a period of centuries. So why do we refer to them as ‘sibylline’? Three reasons.

First: at the end of S.O. 3, the narrator explicitly claims to be one (or two) of the Sibyls. Lines 809–818 (trans. Collins, my emphasis):

(I say) these things to you, having left
(810) the long Babylonian walls of Assyria, frenzied, a fire sent to Greece,
prophesying the disclosures of God to all mortals,
so that I prophesy divine riddles to men.
Throughout Greece mortals will say that I am of another country,
a shameless one, born of Erythrae. Some will say that
(815) I am Sibylla born of Circe as mother and Gnostos as father,
a crazy liar. But when everything comes to pass,
then you will remember me and no longer will anyone
say that I am crazy, I who am a prophetess of the great God.

This doubles up as a claim to be both the Chaldaian Sibyl (from Babylon) and the Erythraian Sibyl. When ancient Christian writers quote from this poem, they ascribe it to the Erythraian Sibyl.

Second: there’s an acrostic in S.O. 8.217–250, mostly duplicated in S.O. 9. The first letters of each line spell out

Ἰησοῦς Χρειστὸς θεοῦ υἱὸς σωτὴρ σταυρός

Jesus Chreist, Son of God, Saviour, Cross

Incidentally there’s nothing surprising about the spelling ‘Chreist’: Koine Greek in the 2nd century CE uses ει, η, or ι to represent the sound [i], swapping between them freely.

Unlike the acrostics in Phlegon’s quotation (see Part 2), this acrostic isn’t a hexameter. Still, acrostics are a feature that was known to be associated with the Sibylline books. S.O. 9 reproduces most of the acrostic: that poem duplicates parts of S.O. 6, 7, and 8. But weirdly, it omits the first line of the acrostic, the one that provides the ι of ‘Jesus’. The preceding line (= S.O. 7.1) starts with an ω, so the acrostic is definitely incomplete. This shows, first, that S.O. 9 is a secondary version; and second, it suggests the compiler of S.O. 9 was unaware that there was an acrostic there.

S.O. 9 duplicates material from S.O. 6, 7, and 8. In this damaged manuscript, cod. Vat. gr. 1120, fol. 348v, we see the transition between them: the end of S.O. 6; one line from S.O. 7; then S.O. 8 starting at the second line of the acrostic. This MS is the oldest extant copy of the second collection (14th century; Q in Geffcken’s sigla). (Base image: Vatican Library; my annotations)

Third: ancient testimony casts the poems as ‘sibylline’. Alexander Polyhistor (mid-1st century BCE) shows an acquaintance with the earliest poem, S.O. 3, when he refers to the Tower of Babel in connection with the Babylonian Flood myth and the Sibyl (FGrHist 273 F 79).

From around 100 CE onwards, Christian writers refer to the Sibyls as authors of prophetic visions, though not yet explicitly citing the poems that we know (Shepherd of Hermas 8). The earliest citations are to S.O. 3 — which is to be expected, since that’s the only poem that makes an explicit claim to be sibylline.

Possibly the earliest citation is in Justin Martyr. Around 155 CE he refers to a prophecy that the world will be dissolved in fire. This may be from multiple sources: Justin is probably mainly thinking of Stoic ekpyrosis, but he explicitly refers to Sibylla (or a Sibyl), as well as the legendary Zoroastrian figure Hystaspes.

A couple of decades later we start to get more clearcut citations, starting with Athenagoras (170s) and Tertullian (190s). They quote verbatim from S.O. 3, and explicitly cite ‘the Sibyl’.

Justin, First apology 20, possibly alluding to S.O. 3.53–54. Athenagoras, Plea for the Christians 30; Tertullian, To the nations 2.12, both quoting S.O. 3.108–113 (in Latin in Tertullian’s case). For fuller documentation of ancient Christian citations of the Sibyls and the Sibylline oracles, see Parke 1988: 152–173.

But Lactantius is the most influential case. His works contain around 40 quotations from Sibylline oracles 3, 4, 5, 6, 7, and 8. And one passage has an especially condensed cluster of sibylline references. In Divine institutes 1.6, Lactantius

  • (1.6.7–12) reports Varro’s canon of ten Sibyls (see Part 1);
  • (1.6.10–11) tells the fable of how the Sibylline books were sold to king Tarquin (see Part 2);
  • (1.6.13) claims that each of the poems was written by one of Varro’s Sibyls, except that there is no poem by the Cumaean Sibyl because the Romans keep her books secret;
  • (1.6.13) paraphrases S.O. 3.808–813;
  • (1.6.15) quotes the Greek text of S.O. fr. 1.7, fr. 3.3–5, and fr. 1.15–16, apparently implying that they are also from the Erythraian Sibyl (i.e. they were originally part of S.O. 3);
  • (1.6.16) quotes S.O. 8.377, stating that it’s from a separate poem.

From this, we know Lactantius’ copy contained nine poems: he attributes the poems he knows to the Sibyls in Varro’s canon of ten, except for the Cumaean Sibyl. We also know it contained S.O. 3 through 8, so if it wasn’t an exemplar of the first collection described above, it was closely related to it. So it seems likely that Lactantius’ copy consisted of S.O. 1 to 8 (that is, seven poems) plus two further lost poems; that it had a fuller text of S.O. 3; and that one of the extra poems in his copy was the source of the fragment in S.O. 3.1–92.

See note above on the relation of fragments 1–3 to S.O. 3. Collins thinks fragments 1–3 come from a poem that once existed between S.O. 2 and 3, Buitenwerf things they come from S.O. 3: it depends on how you read the phrasing when Lactantius quotes from them. Collins thinks Lactantius may also have known S.O. 2, and that Div. inst. 7.15.10 paraphrases S.O. 2.21–22.

So Lactantius is the most powerful and direct reason that the collections came to be known as ‘sibylline’.

He’s also the first victim of a clump of misperceptions that have affected many people since then — the idea that the Sibylline oracles poems are genuinely related to Varro’s Sibyls; the idea that they’re derived from or related to the Roman Sibylline books. That confusion sets the tone for a lot of what we’ll see in Part 4, on developments in the Renaissance.

In recent scholarship, though — or most of it — it’s understood that the links between these things are only apparent links. The Sibylline oracles are first and foremost Jewish-Christian apocalyptic literature. They draw some motifs from pagan sibylline traditions, but their roots lie much more in 3rd–2nd century BCE Jewish apocalyptic literature, like Daniel and Enoch, than in the Roman Sibylline books.

References

  • Buitenwerf, R. 2003. Book III of the Sibylline oracles and its social setting. Leiden.
  • Collins, J. J. 1983. ‘Sibylline oracles.’ In: Charlesworth, J. H. (ed.) The Old Testament pseudepigrapha, vol. 1. New York. 317–472. [Internet Archive]
  • Geffcken, J. 1902. Die Oracula Sibyllina. Die griechischen christlichen Schriftsteller. Leipzig. [Internet Archive]
  • Lightfoot, J. L. 2007. The Sibylline oracles. With introduction, translation, and commentary on the first and second books. Oxford.
  • Parke, H. W. 1988. Sibyls and sibylline prophecy in classical antiquity. London/New York.
1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance

07 October 2026

Sibyls and Sibylline oracles. 2. The Sibylline books

1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance

Once upon a time, in early Rome, a mysterious old woman came to king Tarquin to sell him a collection of books.

By the 1st century BCE the woman in the story was identified as the Cumaean Sibyl. We don’t know if that was always the case. But, based on that identification, the books were called the Sibyllini libri (Sibylline books) or Sibyllina oracula (Sibylline oracles). For clarity I’ll consistently call them the ‘Sibylline books’, as I explained in Part 1.

19th century illustration of the Capitoline and temple of Jupiter Optimus Maximus in the Republican period. (Source: Friedrich Polack, Geschichtsbilder aus der allgemeinen und vaterländischen Geschichte, 1903 edition, fig. 61)

The story is a fable, but the books were real. Dionysios says they were held in a stone chest in the temple of Jupiter Capitolinus and curated by a body of priests — the decemviri sacris faciundis, ‘ten men appointed for sacred rites’.

There was a fire in 83 BCE, and the books were destroyed. A new temple was built. The Romans had new books made, and Sulla expanded the priests to fifteen — the quindecimviri. In 28 BCE the replacement books were rehoused in the newly built temple of Palatine Apollo. They continued to be used for several more centuries.

The Sibylline books weren’t primarily books of predictions: they were instructions for rituals, which also mentioned predictions in passing. The idea seems to have been that, when required, priests might choose a ritual from the books, then perform that ritual, to ensure purity, or ward off evil. Some extracts were made public, but a complete text wasn’t widely known.

The sale of the books

The most famous thing about the books is the story of how they were sold. Every version of the fable goes back to Varro’s book Antiquitates rerum divinarum, ‘Ancient matters concerning the gods’, written between 63 and 49 BCE (see Part 1).

Varro’s book is lost, but the story appears in Dionysios of Halikarnassos, who came to Rome around 30 BCE, and Dionysios attributes the story to Varro. Somewhat different versions appear in Aulus Gellius, Lactantius, and shorter allusions elsewhere.

Dionysios of Halikarnassos 4.62 (citing Varro); Aulus Gellius 1.19; Lactantius, Institutiones divinae 1.6. Further references: Varro, Antiquitates rerum divinarum frs. 56–60 ed. Cardauns; Cicero, On divination 2.(54).110–112; Pliny, Natural history 13.(27).88; Phlegon of Tralles, FGrHist 257 F 36.10; Servius, commentary on Aeneid 6.72.

Here’s the gist, skipping over some variations.

An old woman comes to visit king Tarquin, carrying nine books of oracular instructions. But she asks for an incredible price, 300 gold philippics, so Tarquin just laughs. She promptly burns three of the books, and offers the remaining six at the same price. Tarquin laughs even harder, so she burns another three, and offers the last three again, still at the same exorbitant price. This time Tarquin realises it’s no joke, and he agrees to the sale. The old woman departs and is never seen again.

How much of this is real? Well, Tarquin was a real 6th century BCE king — though the sources vary on whether it’s Tarquinius Priscus or Tarquinius Superbus — but no real records from that period ever existed. The books were real, but they weren’t anything like as old as that. And the Sibyl is purely mythical.

Just in case you’re tempted to think there’s something to the story, bear in mind: the price the old woman asks for is in 4th century BCE Macedonian coinage. 6th century Rome didn’t have coinage.

After the original Sibylline books were burned in 83 BCE, the Romans had replacements made. Allegedly these were pieced together from many places, including cities in Italy, but especially from Erythrai in Anatolia.

The beginning of the sole surviving fragment of the Sibylline books, in cod. Heidelberg. Palat. gr. 398, fol. 226r (image source: Universität Heidelberg library)

The surviving quotations

Later sources tell us that the Sibylline books were written in Greek hexameter verse, and they had acrostics: the first letters of each line also spell out hexameter verses. That’s what Cicero and Dionysios describe, and it’s also what we see in the surviving extracts.

An excerpt of Phlegon of Tralles’ On marvels (2nd century CE) quotes seventy lines from the Sibylline books. Phlegon doesn’t explain how the quotation came to be public knowledge. He treats it as a single quotation, but it’s clear that it’s actually two separate quotations, of 29 and 41 lines. The Greek text is most easily found in Jacoby’s Fragmente der griechischen Historiker.

FGrHist 257 F 36.10 = Hansen 1996: 40–43 (translation), 126–137 (commentary) = cod. Heidelberg. Palatinus. gr. 398 fols. 226r–227r.

The first quotation (lines 1–29) gives instructions for rites to honour Demeter and Persephone. A public coin collection is to be made to fund a ritual sacrifice performed by women, with roles grouped by age (girls, married women, and old women). No alcohol is to be involved in the ritual. The duty of the married women is to carry torches, suggesting that the ritual is supposed to take place at night. The girls sacrifice bulls to Demeter and cows to Persephone, and the old women place the offerings in a fire.

The second quotation (30–70) describes further rituals, this time an expiatory rite to avert evil: an ox to be sacrificed to Demeter, Persephone, and Hades; a goat to Apollo; and a cow to Hera. Again, the text specifies that the ritual must not involve alcohol. The second half of this quotation is an ex eventu ‘prediction’ of the Greek colonisation of Cumae, and the arrival of Trojans in Italy — suggesting that the speaker should be understood as either the Cumaean or Cimmerian Sibyl.

The temple of Jupiter Optimus Maximus as depicted in the TV series Rome (2005–2007)

The acrostics

The text has acrostics embedded in the first letter of each line, just as Cicero and Dionysios describe. Each acrostic spells out more hexameter verse. Some letters are missing in the acrostics, implying that the quotations have some lines missing. The first acrostic, in lines 1–29, goes:

<μ>οῖραν ὀπισθομα<θῶν, τίν’ ἔ>φυ πᾶς εἰς τόπον ἐλθ<εῖν>

Everyone learns too late their fate, the place they’re destined to go

This line doesn’t appear in the text Phlegon quotes: he starts with the line corresponding to the ο of <μ>οῖραν. Traditionally scholars have thought the acrostic should be understood as line 1, so that μοῖραν provides its own first letter (Hansen 1996: 129–130). It’s grammatically possible for the acrostic to act as line 1, but I have doubts: it isn’t how Cicero describes the acrostics (On divination 2.(54).112), and it isn’t what we see in the second quotation.

There, the first letters of each line are provided by lines that Phlegon quotes (though there are letters missing elsewhere):

ἵππωι ἀγαλλόμενος, πάλιν αὖ <κα>κὸν εἰς νέον ἥξει
αὐτά<ρ>. . .

Exulting in his horse, he will come back to a new evil,
then . . .

Lines 59 and 60 have to be swapped around to make the νε of νέον.

For now, at least, I’ll entertain the possibility that there’s a line missing for the μ of <μ>οῖραν. So the quoted text and the missing lines are as follows, using Jacoby’s line numbers (and my additions in parentheses). Quotation 1:

  • line 1a: the first acrostic
  • (line 1b: a missing line corresponding to μ in the first acrostic?)
  • 1–13, as quoted by Phlegon
  • 13a–g: seven lines missing, corresponding to the missing -θωντινε- in the first acrostic
  • 14–29, as quoted
  • 29a–c: three lines missing, corresponding to -ειν in the first acrostic

And quotation 2:

  • (lines 29d–e: the second acrostic; Jacoby prints it only in the critical apparatus)
  • lines 30–52, as quoted
  • 52a–b: two lines missing, corresponding to κα- in the second acrostic
  • 53–70, as quoted
  • an unknown number of lines missing, starting with the ρ of αὐτά<ρ>

Which Sibylline books?

Phlegon says the quotation comes from an occasion in 125 BCE when the senate commanded the hieromnēmones — ‘rite officials’, presumably the decemviri — to consult the Sibylline books on the occasion of the birth of an intersex baby (androgynos). If that’s right, the quotation is from the pre-83 BCE books.

Modern scholars have taken Phlegon at his word. But their focus has been on showing that the quotation is close to 125 BCE because it’s much later than the Roman regal period. It used to need argument to show that it’s Hellenistic. That isn’t a matter for debate these days.

Diels 1890: 77–103; Parke 1988: 137–138; Hansen 1996: 126–127; McInerney 2016.

But I don’t know that there’s been any consideration of the possibility that the quotations are from the replacement books that were compiled after 83 BCE. On balance I don’t think a post-83 date is compelling. But there are good reasons to take it seriously:

  • When we’re told that the books were in hexameter, that they contained acrostics, and that they were in Greek, all of that information comes from post-Varro sources.
  • No source other than Phlegon claims to describe the pre-83 BCE books.
  • Dionysios — who cites Varro — explicitly states that the acrostics are a feature of the replacement books, and that acrostics are a sign that a passage isn’t an ‘interpolation’.
  • When Dionysios describes the compilation of the replacement books, he doesn’t claim that they replicated the originals, or that replication was even the goal.
  • The quotations are very clearly not from the 6th century BCE. That’s been obvious since Diels, who dated both of them to the last decade of the 200s BCE, based on language, historical context, and metrical features. Some modern studies have considered dates as late as 83.

Just to emphasise: there are no sources telling us what the original books looked like, and there’s no real reason to assume that they looked anything like the replacement books that Varro talked about. It’s perfectly imaginable that Phlegon linked them to an incident of an intersex baby born in 125 BCE solely because lines 4–5 refer to an intersex baby.

The profile of metrical irregularities points to a date no earlier than the 200s BCE:

  • 8 cases of ‘Attic’ correption (short syllable before a mute + liquid), 5× word-internal: 10 ἀθροίσας, 34 πάτραι, 44 ἴδρις, 51 πατρίοισι, 56 πατρίοισι; 3× at word-boundary: 48 Παιήονα κρᾶτα, 54 Κυμαίδα πρόφρονες, 68 ὔμμι χρόνος
  • 1 violation of Meyer’s First Law (avoidance of a word starting in beats 1–2 and ending in beat 4a or 4b): 33
  • 7 violations of Meyer’s Second Law (avoidance of a word with the rhythm ⏑ – in beats 4b–5): 29, 40, 52, 54, 56, 63, 65
  • 5 violations of Meyer’s Third Law (avoidance of a word/phrase with the rhythm ⏕ – ⏕ – in beats 6–9): 1a, 27, 36, 41, 66
  • 2 violations of Hermann’s Bridge (avoidance of a word/phrase with the rhythm ⏑ – ⏕ – × at line-end, i.e. in beats 8b–12): 41, 51
  • (no violations of Varro’s Bridge, Tiedke-Meyer’s Law, or Wernicke’s Law)

I omit irregularities involving a word boundary preceded by a monosyllable, and those involving a word boundary followed by an enclitic.

A closer look at Attic correption shows that the rate of lengthening before mute + liquid consonant clusters is relatively high at word boundaries (16.7%), but considerably lower than most other poets within a word (50.0%). I haven’t been able to track down figures for hexameter poets, but among elegiac poets the closest fit is Mnasalkes (3rd century BCE).

For figures on the incidence of word-initial and word-internal Attic correption in elegiac poets see Slings 1993.

This profile of metrical irregularities would be as unsurprising in a 1st century BCE poet as in the 3rd century. So I think a more compelling fact pointing to a slightly earlier date is the use of iota adscript. I don’t mean iota subscripts in the Phlegon manuscript — that’s 9th century CE — but in the acrostics: the second acrostic depends on iota adscript for its first word, ἵππωι.

That still isn’t conclusive, but it does create a weak presumption in favour of a pre-83 BCE date. Prior to 200 BCE, extant papyri are consistent about including the iota in ωι, ηι, and ᾱι; later, they start to be irregular. So Phlegon’s quotation looks like something dating to the 3rd or 2nd century BCE.

The evidence is compatible with a post-83 BCE date, but I don’t see an orthographic or metrical reason to prefer a post-83 BCE date. For the record I’d quite like to assign a later date! But the evidence isn’t enough to overturn Phlegon’s statement that the quotations were used in 125 BCE.

References

  • Cardauns, B. 1976. M. Terentius Varro. Antiquitates rerum divinarum. Teil I: die Fragmente. Wiesbaden.
  • Diels, H. 1890. Sibyllinische Blätter. Berlin. [Internet Archive]
  • Hansen, W. 1996. Phlegon of Tralles’ Book of marvels. Exeter.
  • McInerney, J. 2016. ‘Phlegon of Tralles (257).’ In: Worthington, I. (ed.) New Jacoby. BNJ 257.
  • Parke, H. W. 1988. Sibyls and sibylline prophecy in classical antiquity. London/New York.
  • Slings, S. R. 1993. ‘Hermesianax and the Tattoo Elegy (P.Brux. inv. E 8934 and P.Sorb. inv. 2254).’ Zeitschrift für Papyrologie und Epigraphik 98: 29-37. [JSTOR]

Note, twenty days later: I have corrected the list of metrical irregularities after spotting two errors resulting from my own notes.

1. Sibyls | 2. Sibylline books | 3. Sibylline oracles | 4. The Renaissance