Monday 29 June 2020

Stripping myths down to a historical core (part 1)

Part 1 | Part 2

In a few weeks the game A Total War Saga: Troy will come out, the latest installment in the long-running Total War series. The game’s director, Maya Georgieva, had this to say about it recently:
[W]e decided to draw from [classical-era Greek legend] — very carefully and concisely, by separating the impossible from the still less probable but feasible, until we could filter out a core that could be at the heart of a great new Total War game. We’re guided and reassured in this by the success of pioneers such as Heinrich Schliemann, who believed that there was substance behind those old tales and went on to prove this by excavating the sites of Troy and Mycenae ...
Maya Georgieva in interview, May 2020 (my emphasis)
A Total War Saga: Troy (2020) uses an ancient technique called ‘euhemerism’ to treat myth as a model of reality. Here, euhemerism takes the form of rationalising the mythical Minotaur as a man wearing a bull’s skull over his face.
There are some important historical concepts here. Do myths really grow from a historical core? Are there ways of stripping away the crust to reveal that core? Can a myth be treated as a historical document, to some extent?

The answers are: not normally, no, and never.

It’s imaginable that a myth might coincide with some things that actually happened. But even in those exceptionally rare situations, the myth isn’t evidence about the historical event. That’s backwards. Evidence is always external — archaeology, or historical linguistics, or documentary records. Without those things, a myth is just a myth.

As a game, Total War: Troy has no obligation to be rigorous. Verisimilitude is fine. No complaint. This verisimilitude is based on popular understandings of the relationship between the subject matter and actual history. But ‘popular understandings’ is the key phrase. My concern here is about two big misconceptions at the heart of the popular understanding:
  • The idea that there’s a historical core there to filter out.
  • The idea that Schliemann proved there was substance to a legend.
Today we’ll deal with the second one, because it’s more concrete.
Troy never needed to be ‘proved’, until 1791, when a debate temporarily obscured its location. Left: a coin from Ilium dating to ca. 165–150 BCE, inscribed ‘of Athena Ilias’: Ilium’s main civic cult was devoted to Athena. Right: a coin from Sigeium, dating to ca. 350–300 BCE, also showing Athena. Sigeium, on the west coast, was where ancient tourists could visit Achilles’ supposed gravesite.

Part 1. ‘The place is real, therefore so is the myth’

The first idea is that Schliemann’s excavation proved a legend true. All it really proves is that Heinrich Schliemann was good at PR.

It isn’t exactly unusual for a myth to be set in a real place. Troy is real? Fine! Lerna is real too, but that doesn’t mean Heracles is real and that he actually fought the Lernaean Hydra there. Nottingham is real, but that doesn’t mean Robin Hood is real. New York is real, but that doesn’t mean King Kong, Sesame Street, and the Avengers are real.

That, in a nutshell, is my response to the misconception about Schliemann.

But let’s carry on a bit: the fallacy is especially glaring, because Schliemann didn’t even discover Troy. He obfuscated a real debate. Troy was never lost.

An awful lot of people assume Ilium (to give it its strict name) was an empty hilltop from the Bronze Age until Schliemann ‘discovered’ it in the 1870s. It wasn’t. It was a sizeable city from the 700s BCE until around 500 CE, and it wasn’t abandoned until close to the 1300s. The city was in the northwest corner of Anatolia. Its citadel, or upper city, was on a hill that is now called Hisarlık, but the lower city sprawled far beyond that. There is precisely zero doubt about any of this.

Ancient people were mostly happy to regard their Ilium as the setting of the Iliad, in the same way that New York is the setting for The Avengers. Today, if you want to be a Robin Hood tourist, you’ll visit the real Nottingham: Ilium was like that for the classical Greeks. It was a tourist hotspot, visited by the likes of Xerxes, Alexander, and Scipio. As late as 1463 Sultan Mehmed II toured the ruins and the supposed gravesites of Achilles, Patroclus, and Antilochus on the coast.

But in the 1790s a debate arose, based on an alternate tradition. The 1st century geographical writer Strabo had accepted a theory that Ilium moved around a bit over the ages. Strabo thought ‘ancient Ilium’ had been a few kilometres further inland. In 1785 a French diplomatic aide named Jean-Baptiste Le Chevalier visited, armed with a copy of Strabo, and he decided that Strabo’s ‘ancient Ilium’ was on a hill next to the modern village of Pınarbaşı.

We now know that Strabo and Le Chevalier were wrong, but the argument carried on for a few decades. Ilium itself wasn’t lost. But there was a genuine doubt, at least among western Europeans, that classical Ilium was the same place as ‘Homer’s Ilios’, the fictionalised city in the Iliad. Troy was only really ‘lost’ from 1791 until 1822.

Then in the 1870s Schliemann came along. And somehow everyone got the idea that before him, no one had believed Ilium ever even existed.

The thing Schliemann really proved was that Strabo was wrong: the city hadn’t moved around, it stayed in one spot. But the impression everyone got was that he had discovered Ilium itself, out of thin air. Schliemann didn’t work hard to avoid this misconception. And, worse, he did work hard to conflate the pre-classical historical city with the city where the Iliad was set.

‘Homeric Troy’ is a slippery phrase. It gets used to mean three distinct things: the real Bronze Age city; the real city at the tme the Iliad was composed, which Homer might in principle have visited; and the city depicted in the Iliad. Schliemann exploited that confusion.

Schliemann succeeded in refuting Strabo, and that success got people thinking he was also right about blurring the lines between history and the Iliad — between the different ‘Homeric Troys’. But that isn’t right. It makes no more sense than finding out Hamelin is a real place, then thinking that means the Pied Piper story must have a historical core.
Strabo’s preferred site was a village ca. 5.5 km away (30 stadia), in the direction of Mount Ida/Kaz, that is to say, at a bearing of 119° from classical Ilium. This would imply a spot near the E87 road, about 3 km north of Taştepe. Le Chevalier got the site of classical Ilium wrong — he put it much closer to the coast, 5 km northwest of the actual site — and he put Strabo’s ‘ancient Ilium’ at Pınarbaşı, nearly 14 km from where he thought classical Ilium was.

Timeline: Hisarlık vs. Pınarbaşı

  • ca. 20. Strabo, Geography 13.1.35, reports a claim that ‘ancient Ilium’ was 30 stadia inland from the classical city.
  • 1740. Richard Pococke states that Troy was buried underneath classical Ilium.
  • 1769. Robert Wood, after visiting the Troad in 1750, notes that Homer’s descriptions don’t match the real topography.
  • 1791. Jean-Baptiste Le Chevalier, after examining the area in 1785, argues that Homer’s Troy was at a hill on the south side of the village of ‘Bunarbashi’, and that classical Ilium was just a stone’s throw from the coast, at the mouth of the Scamander (modern Menderes). He accepts Strabo’s idea that the site had moved over time, but rejects most of Strabo’s details, in the belief that he had a poor understanding of the topography.
  • 1795. Jacob Bryant criticises Le Chevalier’s argument, mainly using Homeric references. The argument continues in subsequent years, with Le Chevalier backed up by his former boss, Choiseul-Gouffier.
  • 1801. Edward Daniel Clarke visits the Troad and correctly identifies Hisarlık as (part of) classical Ilium.
  • 1822. Charles Maclaren argues that both ‘the Ilium of Homer’ and classical Ilium were at Hisarlık.
  • 1842. P. W. Forchhammer writes a survey labellng Hisarlık as ‘New Ilium’ (Ilium Novum), Pınarbaşı as ‘Troy, or Old Ilium’ (Troja vel Ilium Vetus), and labels Strabo’s preferred site separately in the hills south of Dümrek.
  • 1850s. John Brunton excavates at Hisarlık and finds a column capital and a Roman mosaic.
  • 1862. Frank Calvert, after investigating Hisarlık and Pınarbaşı, opts for Hisarlık.
  • 1864. Johann Georg von Hahn, an Austrian consul, excavates at Pınarbaşı and finds ‘nothing but the natural soil.’
  • 1865. Calvert excavates at Hisarlık and finds parts of a temple and a wall.
  • 1868. Heinrich Schliemann arrives in the Troad and investigates Pınarbaşı, until Calvert persuades him that Hisarlık is the correct spot.
  • 1871, Schliemann starts uncovering evidence of a Bronze Age city under the classical city, proving Strabo (and Le Chevalier) wrong.
For more details see Rachel B. Davies, Troy, Carthage and the Victorians. The drama of classical ruins in the nineteenth-century imagination (Cambridge, 2018), 67–74 on the 19th century ‘battle of Bunarbashi’.

Next time: Euhemerus, Phlegon of Tralles, and Lara Croft.

Tuesday 23 June 2020

Asia and Europe

The idea that Asia and Europe are distinct continents is a funny thing. They’re joined at the hip, with a land boundary more than 3,500 km long. But the names are very firmly ingrained. And, like so many traditions, the distinction comes from ancient Greece.
Count the major landmasses.
The more you think about it, the sillier it seems. Boundaries at the Ural and Caucasus mountains? Why? You wouldn’t say Oregon and Montana are on separate continents. The Urals don’t even extend all the way along the supposed Asia-Europe boundary: south of Orenburg and the Ural River they’re more gentle rolling hills.

The names aren’t going to go away, but let’s look at where the tradition comes from. We’ll finish off with a look at the linguistic origins of ‘Asia’, ‘Europe’, and ‘Africa’.

A cultural divide?

I imagine some people might take the line that ‘Europe’ and ‘Asia’ are culturally distinct, and that is a good reason to distinguish them. Of course that means ignoring the colossal cultural variation within both Asia and Europe.

And it means ignoring the linguistic links between India, Persia, Armenia, and most of Europe. It means ignoring the folktales, mythology, and poetic conventions that go along with them. Indian mythology has more in common with Greece and Ireland than with China.

Just in case you haven’t seen this kind of thing before, take a look at the verb ‘to be’ in five Indo-European languages — three Asian, and two European. Each column shows the personal forms of the present tense (‘I am’, ‘you are’, ‘she is’, ‘we are’, ‘y’all are’, ‘they are’).
Vedic Sanskrit Avestan Hittite Early Greek (selected dialectal forms) Lithuanian
asmi ahmi esmi eimi, emmi esmi
asi esi essi esi
asti asti eszi esti esti
smas eimes, emmes esme
stha este este
santi hanti asanzi enti esti
Sanskrit is from ancient India, Avestan from Iran, Hittite from Turkey. Lithuanian and Greek are obvious. The gaps are places where the relevant forms don’t appear in any surviving evidence. (Some parallels crop up in English too, but much less clearly: am/is come from the Indo-European es- stem; are is less certain. Be and was/were come from two other separate roots.)

Anyway, that’s just a taster. The iceberg goes down very, very far. And linguistic parallels come with baggage. Ancient Vedic poetry has metrical features in common with Aeolic Greek, just as Latin Saturnian verse has with Old Irish heptasyllabic verse; there are parallels between the legends of the Irish hero Cú Chulainn, the Greek Odysseus, the Persian Rostam, and the Indian Arjuna. Odysseus, Arjuna, and Rama all compete for a woman’s hand in marriage in an archery contest, with a bow that only they can wield. It’s easy to push the parallels too far, but there’s no denying they’re substantial.

This is just a long-winded way of saying that there’s no tidy cultural divide.

The history of the names

Since the divide is arbitrary, its history is really a history of the names, not of the landmasses or the people who live on them. Let’s start in the Roman era and work our way backwards.

Ancient geographical writers used the same names. The 1st century writer Strabo, Geography 11.1.1, states that the boundary between Eurōpē and Asia is at the river Don (ancient Greek Tanais), in the southwest corner of modern Russia. Ptolemy does the same in the 2nd century: he calls the region west of the Don ‘Sarmatia’ (Ptol. Geog. 3.5), and the east ‘Sarmatia in Asia’ (Geog. 5.9). (See here for a map based on Ptolemy’s coordinates.)
Strabo’s and Ptolemy’s division of Europe and Asia.
But Strabo also goes on to say (11.1.2) that Asia is split down the middle by the Taurus mountains. The Taurus range is in Turkey, or Anatolia to use the geographical name. That is: when Strabo thinks of ‘Asia’, he’s mainly thinking of Anatolia. He does include regions further afield in Asia too, east of the Caspian Sea and as far as India, but his attitude is a nice illustration. When people referred to ‘Asia’ in ancient Greek, they didn’t mean what we mean today. Mostly, they meant Anatolia.

And that makes complete sense if you think about it from the point of view of earlier Greeks, before the time of Alexander’s conquests. For them ‘Europe’ and ‘Asia’ simply meant the lands on either side of the Aegean Sea. It’s just that as their interests expanded further afield, it ended up not working so well.

This early usage, Asia = Anatolia, has stuck around in the phrase ‘Asia Minor’ (‘smaller Asia’). The scope of ‘Asia Minor’ varies depending on who’s talking. Sometimes people mean Anatolia; sometimes just western Anatolia. Modern historians writing about classical Greece often use it to refer to the Greek colonies on the west coast, as opposed to non-Greek areas further east.

Herodotus, writing centuries earlier in the 420s BCE, is aware (4.36–4.45) of a conventional division of the known world into ‘Europe’, ‘Asia’, and Africa (named ‘Libya’ in ancient Greek) — but he, too, thinks the division is a bit silly.
I can’t conjecture why the earth, which is a single thing, has three women’s names; or why its boundaries are set at the river Nile in Egypt, and at the Phasis in Colchis [the river Rioni, western Georgia, which meets the sea just south of the Caucasus] — though some put the boundary at the Tanais [= Don] and the Cimmerian strait [= Kerch].
Herodotus 4.45 (trans. Gainsford)
That is, even in his time the division was purely a convention, and people disagreed on where to put the dividing line. Even more interestingly, he pauses to consider the names’ origins:
Most Greeks say that the name of Libyē [= Africa] comes from a woman of that land, and Asiē is named after the wife of Prometheus. But the Lydians play a part in this name too: they claim that it was called after Asias, son of Cotys, son of Manes, not after Prometheus’ wife; and that the Asiad clan in Sardis was named after him too. As for Eurōpē, no one knows ... where it got this name, nor who it was that apparently bestowed the name — unless we shall say that the region got the name from Europa of Tyre, and that it had been nameless previously, like the others. But Europa comes from Asia, obviously. She never came to this land, which the Greeks call ‘Europe’. She only travelled from Phoenicia to Crete, and from Crete to Lycia [in Anatolia]. Well, all right, enough said: we’ll use their conventional names.
Herodotus 4.45 (trans. Gainsford)
Asie. Here Asie is a sea-nymph, one of the Oceanids. Herodotus makes her Prometheus’ wife; in pseudo-Apollodorus 1.2.2 she’s Prometheus’ mother.

Europa. Europa is the mythical Phoenician princess that Zeus carried off in the shape of a bull, into the sea and all the way to Crete, where he raped her. Herodotus makes it clear that that’s just people’s best guess, and he doesn’t buy it. Some of his contemporaries agreed that it makes no sense: they tried changing things up in order to get a Europa on European soil. The epic poet Antimachus (fr. 3 Wyss = fr. 3 Matthews) had Zeus take her to Boeotia instead of Crete, while Hippias (BNJ 6 F 10) decided the name must come from a different woman named Europa. They’re pretty transparently just trying to rationalise a problem away. In any case the role of the bull and parallels with Pasiphaë strongly suggest that the story is closely associated with Crete, not the mainland (Beekes 2004: 167).
Left: Europa and the bull. Right: starting in 2013, the EU decided to commemorate a fictional Lebanese person on all its banknotes, because of a false etymology that even Herodotus didn’t believe.
Herodotus didn’t invent the division, of course: he hates it. The responsibility probably lies with Hecataeus, an ethnographer writing in the late 500s BCE. He seems to have been the first person to use ‘Europe’ and ‘Asia’ in something like their modern senses. Hecataeus’ Tour (Periegesis) is lost, but surviving geographical writers refer to it frequently, citing separate sections called ‘Europe’ and ‘Asia’. And for Hecataeus, these weren’t just Greece and Anatolia: he treated Italy and Ukraine as part of ‘Europe’, and India as part of ‘Asia’.

The organisation isn’t totally clear, though. Our sources cite Hecataeus on Africa too, and when they do, they sometimes cite the ‘Asia’ section, sometimes a Tour of Libya. And the citations show some vagueness about whether places in between the river Don and the Caucasus mountains belong in the ‘Europe’ or ‘Asia’ section.
Note. ‘Europe’ = BNJ 1 F 38 to F 192, ‘Asia’ = F 193 onwards. On the structure of Hecataeus see Pownall’s commentary on BNJ 1 T 15a. On the vagueness about whether the dividing line is at the Don or the Caucasus, see Pownall on F 191. There’s some doubt about whether Hecataeus’ ‘Tanais’ really is the Don.
Some popular accounts bring up Anaximander as another 6th century writer who divided the landmass into ‘Asia’ and ‘Europe’. This one is bogus. Anaximander is supposed to have drawn up a schematic map of the known world (fr. 12 A 6 Diels-Kranz), but there’s no indication he was in the business of naming continents.

Two sea-nymphs named Europa and Asia pop up in the Hesiodic Theogony (ca. 700 BCE), lines 357 and 359, but not as a pair: they’re in a long list of names of Tethys’ and Oceanus’ children. Let’s forget them.

The closest we get in Greek sources to the original ‘Europe’ and ‘Asia’ as geographical terms is in two poems of the 7th and 6th centuries BCE. First: ‘Asia’ appears in Homer as a place in western Anatolia (Iliad 2.461). The line refers to the ‘Kaystrian river in the meadow of Asias’, or possibly ‘in an Asian meadow’. The Kaystros (modern Küçük Menderes) is a river in western Anatolia to the south of Izmir. In antiquity it had the city of Ephesus at its mouth.

And second: ‘Europe’ appears as a relatively small region in northern Greece in the Hymn to Apollo. Cynaethus put the Hymn in its final form in the 520s BCE, but it’s pretty clear he used large chunks of older poetry. The Hymn refers twice to
the people who live in the rich Peloponnesos,
and those in Europe, and in the islands surrounded by sea
Cynaethus, Hymn to Apollo 250–251 = 290–291
This checklist seems to indicate the original state of things. Cynaethus isn’t listing continents, but chunks of the Greek mainland and the islands. The Peloponnesos and Europe are the southern and northern mainland respectively.
This is how the Greeks appear to have divided up their world before Hecataeus came along and got people thinking about huge tracts of land. The scope of ‘Europe’ is shrunk down even further by the extent of Eurōpos toponyms in northern Thessaly and Macedonia: see below. The ‘Asia’ circle is centred on Sardis; the Kaystros is further south, and the Hittite ‘Assuwa league’ extended much further, including all the way to Troy in the northwest.
This much more limited scope for ‘Europe’ is encouraged by some other places where Herodotus refers to ‘Europe’. At 6.43, Mardonius crosses the Hellespont then travels ‘through Europe’ as far as Thasos; at 7.8 Xerxes states his intent to cross the Hellespont and then travel ‘through Europe against Hellas’. This strongly suggests that it doesn’t just refer to the northern mainland, but maybe just the far north.

The actual origins of the names

The origins of ‘Asia’, ‘Libya’, and ‘Africa’ are pretty clear, even if a lot of details are missing. They all come from ethnic groups, or names for specific areas. None of them comes from a person’s name, mythological or otherwise. ‘Europe’ is more difficult.

Asia. This origin story is straightforward. (For details see Corcella 2007: 614–615; Brügger et al. 2010: 135–136). We’ve got Herodotus looking at Lydian customs; we’ve got Homer putting Asia/Asios on the banks of the Kaystros; we’ve got independent evidence of a real ethnic group called ‘Asias’ at Sardis, the Lydian capital in western Anatolia; and we’ve got a Hittite name that matches with all of these beautifully.

‘Asia’ comes from ‘Assuwa’, a region within the Hittite empire in the Bronze Age. Assuwa, too, was in western Anatolia. We don’t know that it was specifically at Sardis, but that’d be a decent guess. Hittite assus means ‘good’, so the name meant ‘good land’. The derivation went like this:
Hittite Assuwa (toponym)
> Mycenaean a-si-wi-yo (male personal name), *Aswia (presumed toponym)
> classical Greek Asios (male personal name), Asia (toponym)
Europe. The linguistic origin of this name is unknown, and don’t let anyone tell you otherwise. (For details see Beekes 2004.) But we do at least know that it’s linked to the northern mainland. See above on Herodotus using the name to refer to Thrace. In addition, Eurōpos was the name of two towns in Macedonia, one town in Thessaly on the river Peneios, and a tributary of the Peneios. From the 4th century onwards the Macedonian rulers Philip II and the Antigonid dynasty were referred to as ‘rulers of Europe’ (Hartmann 2012: 40, with references). All this suggests Europe was further north than Boeotia, which is what Beekes suggests: I doubt it included even southern Thessaly. Beekes suspects the name originated with a pre-Greek goddess or demigod, as some goddesses had cults with Europa or Europia as a title. But these cults weren’t in the north, so I doubt this too.
Note. Goddesses with the title ‘Europa’: see Beekes 2004: 168, citing a 1946 book by Ninck; Peeters 2009: 79 n. 10, citing a 1937 study by Technau, and a 1940 doctorial dissertation (in Dutch). Demeter Europa at Lebadeia, Boeotia; Europa Hellotis in Crete; Hera Europia in Argos. Peeters also mentions Lucian’s ‘Europa Astarte’, but as I mentioned above, that’s a crossover under the influence of the Europa myth.
So we can roughly outline where Europe originally was (maybe). But for its linguistic origin, we can’t say anything positive: we can only rule things out.

(1) It definitely doesn’t come from Greek euru- ‘broad, wide’. If it did, the second u would have been preserved, as in the Homeric adjective euruopa ‘far-seeing’.

(2) It definitely doesn’t come from a mythical Phoenician princess. The only way it could be linked to the princess Europa is if the Greeks had transplanted her to Phoenicia from somewhere on the Greek mainland. That is a possibility: there’s a variant where her father is Phoenix (Iliad 14.321), and that name could prompt thinking of a Phoenician setting and transferring her to Phoenicia. Once Europe the landmass was seen as potentially linked to Europa the princess, then people like Lucian started drawing links between Europa the princess and Astarte (Ishtar); but that’s all post hoc. It can’t have been the case originally.

(3) It definitely has nothing to do with Akkadian erebu, Phoenician ʿereb ‘go down, set, evening, west’. That theory has found some support in Hesychius, and some modern scholars of the Near East, but no modern specialist in early Greek language thinks it’s at all likely. (And no, erebos ‘darkness of the Underworld’ doesn’t come from those words either: that one is Indo-European.) Quite aside from the implausible phonological shifts, this theory is ruled out anyway by the Eurōpos toponyms in Thessaly and Macedonia.

Libya. Libya isn’t originally Greek. (Again, for details see Corcella 2007: 614–615.) It’s a hellenised form of a name that appears as rbw in Egyptian texts from the 1200s and 1100s BCE. There it refers to a group that lived to the west of the Nile, who sometimes occupied or invaded the western Nile delta. The Greek use of Libyē to refer to the entire continent stems from the Greek colony at Cyrene, founded in 631 BCE.
The origin of ‘Africa’: the ancient Afri, in what later became the Roman province of Africa, superimposed on a map of modern Tunisia.
Africa. Africa comes to us via Latin. It’s an adjective derived from Afer, the Latin for an ethnic group in what is now northern Tunisia. (For details, see Vycichl 1985.) The core of Afer territory seems to have been between the river Bagradas (modern Medjerda) and modern Mateur, about 80 km west of Carthage. The name quickly expanded its scope in Roman thought to cover an entire province of the empire, and eventually an entire continent.

The Afri appear in Livy as vassals of the Carthaginians. Frontinus mentions an occasion when the Carthaginians repressed an Afri rebellion. Scipio’s victory at Zama in 202 BCE, a bit to the south of Afer territory, is one of the reasons for his agnomen Africanus. The most famous Afer was the comic playwright Terence, full name Publius Terentius Afer (first half of 100s BCE). ‘Terentius’ comes from his former Roman owner: most probably he was enslaved by Carthaginians, sold to a wealthy Roman at an extortionate price because of his exceptional education, then later manumitted.

The linguistic root is Berber fr, which still survives locally in the name Friguia, a valley that the Medjerda flows through. The a- is a nominal prefix that still appears in some dialects, with cognates in ifr-: most notably the Banū Īfran people who dominated parts of Algeria in the pre-Islamic period.
Note. See Vycichl 1985 on linguistic aspects, and the Afri as distinct from Carthage; Kotula and Peyras 1985 on the Afri politically and geographically. For an English-language treatment see Lipiński 2004: 199–200, who differs in some respects.


  • Beekes, R. S. P. 2004. ‘Kadmos and Europa, and the Phoenicians.’ Kadmos 43: 167–184. [Open access copy]
  • Corcella, A. 2007. ‘Book IV.’ In: Murray, O.; Moreno, A. (eds.) A commentary on Herodotus books I–IV. Oxford University Press. 543–721.
  • Brügger, C.; Stoevesandt, M; Visser, E. 2010. Homers Ilias. Gesamtkommentar, Bd. II, Fasz. 2, 2nd edition. De Gruyter.
  • Hartmann, A. 2012. ‘Europe and the Other: roots of European identity in Greco-Roman antiquity.’ In: Pinheiro, T.; Cieszynska, B.; Franco, J. E. (eds.) Ideas of | for Europe. An interdisciplinary approach to European identity. Peter Lang. 37–57.
  • Kotula, T.; Peyras, J. 1985. ‘Afri.’ In: Camps, G. (ed.) Encyclopédie berbère, vol. 2. Éditions Peeters. 208–215. [Open access copy]
  • Lipiński, E. 2004. Itineraria Phoenicia. Uitgeverij Peeters en Departement Oosterse Studies.
  • Peeters, M. C. 2009. ‘L’évolution du mythe d’Europe dans l’iconographie grecque et romaine des VIIe-VIe s. avant aux Ve-VIe s. de notre ère : de la «déesse au taureau» au rapt et du rapt au consentement.’ Dialogues d’histoire ancienne 35.1: 61–82. [Open access copy]
  • Pownall, F. 2013. ‘Hekataios of Miletos (1).’ Brill’s new Jacoby. [Subscription required]
  • Vycichl, W. 1985. ‘Africa.’ In: Camps, G. (ed.) Encyclopédie berbère, vol. 2. Éditions Peeters. 216–217. [Open access copy]

Friday 12 June 2020

Who preserved Greek literature? (Part 2)

Part 1 | Part 2

Back in December I wrote about the myth that ancient Greek texts only survived by being preserved in the mediaeval Islamic world. Some readers pointed out that I should have told the true story, as well as dismantling the myth. So here we go.

But first, I’d better repeat that it is a myth. Great Arab and Persian scholars like Averroes and Avicenna were proactive innovators, not passive pipelines for getting texts from A to B. Only in a tiny number of cases do we rely on translations for ancient Greek texts — and into a variety of languages, not just Arabic — and every now and then that number shrinks, when a Greek copy is found.

Recently I realised that, for many people, the Arabic transmission myth doesn’t just apply to Greek texts, but to Latin texts too! (Examples: 1, 2, 3.) So we’d better look at them too. We have a lot of ground to cover: make yourself a cup of tea.

For in-depth accounts, Reynolds and Wilson’s classic book Scribes and scholars (fourth edition 2013) and Pfeiffer’s History of classical scholarship (1968-1976) both cover ancient transmission up to the 1st century BCE very well; Reynolds and Wilson also tell a detailed story from then up to the 1300s. But after that date, things go a bit pear-shaped. They turn into histories of an academic field, of publication and scholarship within western Europe, rather than of the texts themselves. Pfeiffer in particular determinedly ignores any Greek people involved in the story, other than a couple of passing mentions of Manouel Chrysoloras and Ianos Laskaris — and them only for their role in ‘returning’ Greek books to Italy. Western eurocentrism is deeply ingrained in these histories.

Anyway, the short version is this. Latin texts were preserved in western Europe, especially Italy, Germany, and France. Greek texts were preserved in the eastern Roman empire, especially Greece, Anatolia, and greater Syria. Modern editions are based on manuscripts that still exist in various libraries around the world — mostly in libraries in Italy, France, Britain, and Germany, but many elsewhere too, ranging from America to Greece to Armenia to Egypt.

The loss of ancient Greco-Roman books

The ancient Greco-Roman world had a thriving book trade from the 3rd century BCE to around the 3rd century CE. The literacy rate certainly wasn’t at modern first world levels, but scribes were cheap, papyrus was recycled, and books were produced on such a scale that they were reasonably affordable.

Book manufacture was the labour-intensive part of the process. There was no such thing as copyright. When Pliny found out once that his books were being sold at shops in Lyon, he was pleasantly surprised (Letters 9.11.2): he had no reason to be annoyed at any loss of profit, because he would never have tried to manufacture books there himself.

Libraries existed all over the place. Every now and then they suffered catastrophic accidents, like fires, but so long as the local economy was doing well the libraries got rebuilt. Rome had some of the greatest libraries of the ancient world: the library of the Porticus of Octavia, the Palatine Library, and the Ulpian Library. The first seems to have disappeared by the year 200, but the other two survived into the fourth century in spite of several fires. In the east Athens, Antioch, and Alexandria continued to be centres of learning as well.
Cod. Vaticanus latinus 3867, fol. 14r: a 5th or 6th century illustrated edition of Vergil, possibly made in France or Britain. This leaf shows a portrait of the epic poet himself above the text of the Eclogues. (Image © Biblioteca Apostolica Vaticana)
But books didn’t survive by being preserved in a library. No ancient library has survived to the present day. Books survived by being copied. Books have a limited lifespan, regardless of format — scroll or codex, papyrus or acid-free paper, monograph or miscellany. Only a handful of ancient copies still exist. The books that survive are the ones that were copied, and copied, and copied again, usually many times over the centuries.

So in the story of the loss of ancient Greco-Roman literature, library fires are just a footnote. No single library had a monopoly on the classics anyway. A much bigger role was played by a format shift that affected every book, everywhere: the shift from scroll to codex. That format shift took place in the years 100 to 400 — in antiquity: most of the loss occurred before the dissolution of the western empire.

Here ‘codex’ means the standard modern book format, sheets bound with a spine. In the year 100, about 98.5% of books existed as scrolls, and only 1.5% in codex form; by 400, things had flipped round, and the codex accounted for 80% of books (Casson 2001: 127-128). After that date Greek texts continued to be lost, as the eastern empire was gradually whittled away, but with Latin texts it’s relatively uncommon to find evidence of texts surviving to that date that haven’t also survived to the present day.

Greek vs. Latin

We’d better dispose of a potential major misconception here. If you imagine that the spread of Roman domination went hand-in-hand with the spread of the Latin language, think again. Latin was never widely spoken in the eastern empire. Rome didn’t speak Greek, Constantinople didn’t speak Latin.

The actual linguistic make-up of each half of the Roman empire is a messy, complex business, with enclaves, immigrant communities, scholarly use of Greek, military and governmental use of Latin, and other languages popping up. But as a rule of thumb, ‘west = Latin, east = Greek‘ won’t go too far wrong. The only part of the empire where Latin actively displaced Greek was in the western Greek colonies, in southern Italy, France, Sardinia, Corsica, and Spain.

So the survival of ancient Latin and Greek texts is actually two distinct stories. The traditional image is of monks busily copying out books in monasteries. Monastic scribes did play an important role in transmission in both the west and the east, but to a different extent and in different ways.

In the west, the fracturing of the empire and the chaos of the 6th century devastated secular libraries, so monastic copying became the main road to a text’s survival. Lay scholars, that is non-ecclesiastical ones, began to play an important role from the 14th century onwards, but monastic transmission still mattered for a long time — even up until the Reformation, when many monasteries were dissolved in Germany.

But in the east, schools and other secular institutions played a significant role in the copying process too, especially in the 800s to the 1400s. Some emperors vigorously encouraged state libraries and non-ecclesiastical schools. But in areas conquered by the Ottomans, books that survived often did so in monastic libraries. A much more robust avenue for survival lay in the books that were moved to Italy in the 1400s: there they made their way into libraries (and by that time civic libraries were coming into existence again there), and onto the printing press.

Latin texts

The ‘fall’ of Rome in the 400s wasn’t the end of book production. The fracturing of the western empire was also a fracturing of the western intellectual economy, but major literary figures like Priscian, Boethius, and Cassiodorus continued to flourish in Italy under Theoderic (r. 493-526) — even if Boethius’ career did end with a big political hiccup.

Then everything changed when the eastern empire attacked. Justinian’s invasion in the 530s engulfed Italy in two hellish decades of war. After that came the Lombard invasion. Some Italian cities managed to remain hubs for literary production and transmission (Rome, Ravenna, Verona), but otherwise, monasteries became the safest haven for literacy and learning after the chaos of the 500s.

There were already some monastic or quasi-monastic institutions that were copying books, like Monte Cassino in Latium/Lazio (founded ca. 529), and Cassiodorus’ Vivarium at Squillace in the far south (the 540s). Like later institutions, these centres collected books, took good care of them, made new copies when needed, and wrote new books. But the wars took their toll. The first abbey at Monte Cassino was destroyed in 570, in the Lombard invasion; the Vivarium faded after Cassiodorus’ lifetime, though it hung on for another century or so.

After 535 the survival of Latin texts is mostly owed to a wave of new monastic foundations in the 600s and 700s. In Italy some of the most important centres were at Bobbio (founded 614), Pomposa (also in the 600s), and Monte Cassino (refounded ca. 718); in France, Auxerre (originally founded in the 400s), Luxeuil (580s-590s), Ferrières (630), Fleury (ca. 640), Corbie (ca. 660), and Murbach (727); in Scotland and England, Iona (563), Lindisfarne (ca. 634), Jarrow (674), and Malmesbury (675); and in Germany and Switzerland, after (re)christianisation in the early 700s, at St Gallen (719), Reichenau (724), Fulda (744), and Lorsch (764). Other important institutions followed in later centuries: St Michael’s in Piedmont, Cluny and Bec in France, Melk in Austria, Heilsbronn in Franconia, and many more.
Umberto Eco, The name of the rose (1980, English edition 1983); Walter Miller, A canticle for Leibowitz (1959)
This monastic environment is the setting for fictional depictions like those in Umberto Eco’s The name of the rose (1980) and Walter Miller’s A canticle for Leibowitz (1959). Eco’s fictional monastery is inspired by the Abbey St Michael, but its multicultural library perhaps owes more to Monte Cassino. Miller depicts a monastic order that was founded specifically to preserve ancient texts. That’s a caricature: libraries were an integral part of daily business for mediaeval monasteries, but not generally the centrepiece of their mission statement.

These places had a variety of aims and specialisations. Some of the most famous mediaeval manuscripts, the Book of Kells and the Lindisfarne Gospels, were produced in Britain in the 700s, but British institutions tended to specialise in Christian texts and producing new philosophical and historical works. For pre-Christian texts, the libraries in Italy, France, and Germany are much more important.

Lay scholars began to play an important role in the late mediaeval period, as I mentioned. Poets like Dante, Petrarch, and Boccaccio became fabulously learned in ancient literature. Petrarch coined the phrase ‘dark age’. Some Italian states funded major civic libraries (Urbino, Naples, Florence, Rome), and individuals like Poggio Bracciolini and Coluccio Salutati collected large personal libraries. The printing press arrived on the scene in the mid-1400s, but books continued to be copied by hand as well.

Monastic collections usually ended up in university or state archives, one way or another. For example, the Electoral Palatinate appropriated the library of Lorsch in the 1560s when the monasteries were dissolved: this became the core of the Bibliotheca Palatina. Then in the 1620s, at the start of the Thirty Years’ War, the collection was recaptured and most of it sent to the Vatican, where the manuscripts are still preserved. Some collections were moved wholesale; others were dispersed. More below on some of the major collections today.

Greek texts

Greek texts were primarily preserved and transmitted in Greek-speaking regions of the empire, that is (initially) in Greece, Anatolia, greater Syria, and Egypt. Things shifted as the empire gradually disappeared over the centuries.

From around 500 to the mid-800s there was a serious downturn in literary activity. It didn’t cease to exist — we have part of a parchment edition of Sappho from 7th century Faiyum, Egypt, around the time of the Rashidun conquest — but all the major centres of learning were shut down, except in Constantinople. For once, we can legitimately blame organised religion. This period saw a boom in Christian biblical literalism; most of the flat-earthers of antiquity that we know of came from Syria in this period, justifying their dogma by selective quotation of biblical passages. The waves of iconoclasm in the 700s and 800s didn’t help either.

But even in that touchy climate, considerable effort still went into textual transmission. The development of the catena, or verse-by-verse Bible commentary, in the 6th century prompted a similar trend in the secular classics. Many bodies of scholia, that is to say glosses on literary and philosophical works, seem to have been compiled starting in this period, distilled from ancient treatises and commentaries. The ancient treatises themselves are lost — the works of ancient literary critics like Aristarchus, Crates, Didymus, and Theon — but a significant amount of material survived as scholia in the margins of other texts.
The Homeric Iliad, with scholia. The end of Iliad book 8 and start of book 9 as they appear in codex Marcianus graecus 822, fol. 111v. This is one of the most important manuscripts of the Iliad, also known as ‘Venetus A’. It dates to the 900s, and is now housed at the Biblioteca Marciana in Venice. Homer’s poetry is centre-right in large lettering. In the smaller lettering in the surrounding margins are glosses or scholia derived from ancient commentaries. (Source: Homer Multitext Project)
There was a new wave in the 9th century. There was also a new format shift: books were copied from the older uncial script into the new minuscule script. This seems to have been another great filter, deciding which ancient books survived and which ones didn’t. At first the new book culture was dominated by church leaders like Leon the mathematician, who became the head of a revitalised imperial college in Constantinople, and patriarch Photios, and archbishop Arethas, who built up large libraries. But emperor Konstantinos VII Porphyrogennetos (r. 913-959) promoted scholarship outside the church as well, funding new excerpted editions of older authors. The first of the great Byzantine encyclopaedias, the Souda, was compiled in the second half of the 900s.

Byzantine book culture continued to thrive, at times erratically, until the Ottoman conquest. Paper was introduced in the 1000s. New commentaries on ancient authors appeared, some of them by scholars who worked with the great Anna Komnene. There continued to be significant book collectors, both inside the church (Eustathios, Michael Choniates) and outside (Michael Psellos, Ioannes Tzetzes). The Palaiaologan period (1261-1453), the last phase of the Byzantine era, was an even greater revival, with major lay scholars like Maximos Planoudes, Manouel Moschopoulos, and Demetrios Triklinios, who are still routinely cited by modern editors of ancient Greek plays.

After the Ottoman conquest, civic support dried up. Just like 900 years earlier in the west, it fell to monastic libraries to carry on preserving books. But in the east the survival rate was much poorer. Partly because monasteries operating under Muslim rule were inevitably much less well-equipped than those in the west; and in Turkey, Israel, and Egypt, also because of the ravages of the 20th century, when some libraries came to be neglected amid the chaos of many wars.

I haven’t been able to consult complete catalogues. Catalogues of microfilms held in Paris and Washington show thousands of manuscripts of Christian texts and Byzantine histories held in Greece, Israel, and Egypt, but the range of pre-Christian texts is limited:
  • Mt Athos, Greece, microfilms held by the Institut de recherche et d’histoire des textes:
    • Iviron: 8 codices (Aesop, Euripides, Sophocles, Synesius, Pindar, Lycophron, and Aelius Aristides)
    • Megisti Lavra: 7 codices (Thucydides, Hesiod, Pindar, Iliad, Philostratus, Aesop, Sophocles, and Libanius)
    • Vatopedi: 7 codices (Aristophanes, Sophocles, Aeschylus, Euripides, Pindar, Libanius, Dio Chrysostom, Aristides, Ptolemy, Strabo, Homeric Hymns, Callimachus, and Babrius)
  • Patriarchal Library, Jerusalem, microfilms held by the Library of Congress:
    • Holy Sepulchre (Agios Tafos): 3 codices (Aristotle, Xenophon’s Cyropaedia, Simplicius)
    • Photios II’s collection: 2 codices (Iliad, Aristotle)
    • (Also note that the famous Archimedes palimpsest was held by the Patriarchal Library until it was transferred to Constantinople sometime between the 1600s and 1840s.)
  • Sinai, Egypt, microfilms held by the Library of Congress:
    • St Catherine’s: 5 codices (Plutarch, Homer, Aristotle)
There are presumably others in various collections in Greece, but the information available online is unfortunately not very extensive. There are also many palimpsests, and I wouldn’t be surprised if a number of them have never been identified.
Monastery of Iviron, Mount Athos, Greece
Because of this shortage, modern knowledge of ancient Greek texts relies much more on the Greek books that were taken from Greece to Italy in the 1300s and 1400s. Petrarch (1304-1374) is the most famous person in this part of the texts’ story. But there are other book collectors who are even more important: Giovanni Aurispa (1376-1459), Francesco Filelfo (1398-1481), and Basilios Bessarion (1403-1472, a.k.a. Vissarion) all lived in Greek-speaking lands and brought large libraries to Italy — Aurispa and Filelfo because they were westerners who were appointed to positions in Greece, Bessarion because he was a Catholic cleric. Bessarion bequeathed his library to the city of Venice, and it formed the basis of the Biblioteca Marciana. The early 1400s saw a boom in translating Greek texts into Latin, and the rise of the printing press led to a wave of printed editions from the 1470s onwards.

The subsequent history of the texts is entangled with westerners’ idealisation of antiquity, and the fact that modern knowledge of the texts has been filtered through Italy. Westerners came to treat ‘Greece and Rome’ as something separate from, and superior to, other ancient (and contemporary) cultures. And because of the Italian filtering, they are often thought of as inextricably linked to ‘western culture’. So we need to have a word about racism here.

What about now?

The modern history of Greco-Roman texts is brutally colonialist, to an almost comical extreme. ‘Classics’ as a field is practically the archetype of a field that is white, male, and parochial. ‘The glory that was Greece, and the grandeur that was Rome’ (Edgar Allan Poe) is still a cornerstone of how these cultures are perceived: it’s no coincidence that people like Steve Bannon and Boris Johnson are enthused about Greek antiquity, and that white nationalists are obsessed with ancient Sparta.

This affects the position of ancient texts in the modern world in many ways. One obvious sign is the personnel involved. Racism plagues the field as much as ever. That isn’t because the material is an automatic switch-off for black people. Positive role models exist (though we need more of them). Just a few days ago the writer Maaza Mengiste tweeted that
Someone asked me yesterday why I liked Greek classics so much. I pointed out I could find myself in those stories. Ethiopia, the Ethiope, that was me Homer & Virgil & Herodotus were writing about. I was erased from European literature, I said. You didn't see me, so I looked away.
If you want to find university courses on colour and race in the ancient world; if you want to learn Latin or Greek from a textbook that doesn’t cast ancient slavery as comical, or even benign — well, they exist. But you’ll have to hunt around.
According to this widely used textbook for learning ancient Greek, published by Oxford University Press, slavery was benign, and enslaved people were ‘lively and cheeky characters’. (Balme and Lawall, Athenaze, 3rd edition 2016, ch. 2)
For our purposes today, the main outcome of western-eurocentrism is a widespread disregard for textual transmission, manuscripts, and scholars in any place other than western Europe. It’s seen in things like Pfeiffer’s avoidance of mentioning Greek libraries, scholars, and book-collectors. We also saw it in Part 1, in those popular articles that refuse to acknowledge transmission in Greece, let alone in eastern Europe or western Asia.

In the standard histories you won’t find a single word about Athanasios Papadopoulos-Kerameus’ monumental work in the 1880s and 1890s on cataloguing manuscripts held at libraries in northern Greece, Jerusalem, Izmir, Istanbul, and St Petersburg. It’s unsurprising that a scholar who was ready to undertake that work ended up finding a home in Russia, and not in a western country.

New editions

Popular perceptions and general introductions may keep quiet about transmission outside western Europe, but editors are more open-minded. When a modern editor wants to to put out a new edition of an ancient text, they will go where the manuscripts are — whether that’s in Italy or Greece, France or Egypt, Britain or Turkey.

The surviving manuscripts of ancient Latin and Greek texts are kept today in a wide variety of libraries. Almost all of them are available to any scholar with reasonable credentials. If you’d rather save the airfare, digitised copies are available for a moderate fee — maybe on the order of €50 to €100 if there aren’t too many pages. Some of the most important collections are the Apostolic Library at the Vatican, the Biblioteca Ambrosiana in Milan, the Biblioteca Medicea Laurenziana in Florence, the Bibliothèque nationale in Paris, the British Library in London, and many, many more.

When an editor makes a new edition of Livy or Herodotus, they regularly draw from manuscript copies made as recently as the 1700s. A modern edition will normally list the manuscripts that exist; which libraries they’re in; catalogue numbers and physical descriptions; and the genetic relationships between them, showing which manuscripts matter more than others. There will also be a set of abbreviations or sigla, which are used to scrupulously annotate variants where the manuscripts differ from one another.

Let’s take an example. A 2013 edition of Diogenes Laertius’ Lives of the philosophers, edited by Tiziano Dorandi, lists 30 Greek manuscripts. Here’s a boiled-down version of the first few:
and so on. The list carries on with manuscripts held in other collections too, in London, Leiden, Moscow, Madrid, Prague, Vienna, Cambridge, and Munich. The text of Diogenes Laertius as we know it is entirely dependent on these physical manuscripts, held at these specific archives.

You’ll notice from the links that many of these are available online. Click on them, and you’ll find yourself looking at the basis of the modern text. Some libraries are more proactive about digitising manuscripts than others. The Vatican has been making a big push to digitise as much as possible. Institutions do charge fees, because digitisation costs money, but some of them make the digitised manuscript freely available online straight afterwards.

Our texts of Vergil and Homer can’t be based on copies from the poets’ own lifetimes, because no such copies have survived. But the texts have. Don’t go thinking that their convoluted history means we can’t trust that text: scribal copying has on the whole been reasonably faithful over the millennia.

If some scribes have been dishonest or incompetent, that can often be spotted by comparing different manuscripts. Modern editors have an arsenal of techniques for analysing the differences between manuscripts and reconstructing how they came about. Once again Reynolds and Wilson’s book is a useful introduction here: they give a compact account of the principles, and the most common types of textual corruption.


  • Casson, L. 2001. Libraries in the ancient world. Yale University Press.
  • Pfeiffer, R. 1968-1976. History of classical scholarship, 2 vols. Oxford University Press.
  • Reynolds, L. D.; Wilson, N. G. 2013. Scribes and scholars. A guide to the transmission of Greek and Roman literature. Oxford University Press.