How far removed are we from ancient testimony?

Popular opinion often seems to have it that our evidence on the history of the ancient world is way, way, way more direct than it actually is. Here’s an example from a social media site: a questioner is asking about the historicity of the battle of Thermopylae. Some well-meaning soul explains that Herodotus is our source (misleadingly: I can understand leaving out Ctesias’ and Trogus’ accounts, but there’s no excuse for ignoring Diodorus/Ephorus), and the questioner asks a follow-up question:

Are the manuscripts [Herodotus] authored still intact or were translations made by other verifiable sources then?

Holy high expectations, Batman! I don’t say that to bash the questioner: we learn by asking questions, and if you’re not aware of how we actually get our information about that era, it’s not automatically crazy to imagine that we might know Herodotus’ history through his own autograph copy. After all, we do have some ancient artefacts associated with well-known individuals: we have autographs by Exekias, a copy of Vergil’s Eclogues annotated in the hand of a 4th century Roman consul, a memorial for one of the Athenian generals at Marathon, and (probably) the actual body of king Philip II of Macedon. There’s no obvious way for a layperson to know that these are very exceptional things. So this isn’t a criticism: the quotation above is meant to illustrate a widespread assumption, one that is very common among lay enthusiasts.

Autograph of Exekias, the famous 6th century BCE Athenian potter and vase-painter. (Source: Wikimedia.org)

This is very reminiscent of the bizarre standards of evidence that ‘Jesus mythicists’ usually demand for the existence of a 1st century CE Judaean cult leader by the name of Jesus. A 2008 book by Dan Barker, which sounds like it ought to be reasonably conscientious -- Barker has a Wikipedia article devoted to him, and his own book decribes him as ‘one of America's leading atheists’ -- writes

Jesus supposedly lived sometime between 4 B.C.E. and 30 C.E., but there is not a single contemporary historical mention of Jesus, not by Romans or by Jews, not by believers or by unbelievers, not during his entire lifetime. ... The lack of contemporary corroboration does not disprove his existence, of course, but it certainly casts great doubt on the historicity of a man who supposedly had a great impact on the world. Someone should have noticed.

Gosh, how could we not have noticed? Well, in the real world we have only one solitary surviving contemporary historiographical source. It is Velleius Paterculus’ history of Rome: a history of an Italian city and its ruling class. It would be crazy to expect it to mention a cult leader who lived in the far reaches of the empire and had no significance outside Judaea during his lifetime. (Bear in mind that even a much more exalted figure, the prefect of Judaea at the time Pontius Pilate, is attested by a grand total of one contemporary source -- and not a historiographic source, but a badly damaged inscription.) This is very different from the Herodotus enquirer above: one is an uninformed person posing an honest question to improve their knowledge, the other is an uninformed person pretending to be an authority.

Both situations are the result of someone having very unrealistic impressions about the kind of evidence that we have. We don’t have the papyrus on which Herodotus wrote; we don’t have archives of Roman bureaucratic records. Any kind of commentary on current events is a very, very rare thing. We have much fewer surviving ancient historians than Dan Barker imagines (or seems to imply), and those historians typically deal with events decades or centuries before their lifetimes, and focus on politics and military affairs at the highest level.

There’s no contemporary historiographical source on Alexander the Great either. That doesn’t ‘cast great doubt’ on his historicity, because that would be a colossally inappropriate methodology to apply.

So if even Alexander the Great isn’t attested by contemporary sources, how can we be confident about any ‘facts’ at all in antiquity? The answer is that with different kinds of evidence, we have to adopt different methodologies. Judging ancient history by the criteria that you’d use for 20th-21st century events, or even 19th century events, is inherently nonsensical: it’d be like using carbon dating to test dinosaur fossils. It doesn’t work. It doesn’t even make sense to think about it that way.

Broadly speaking we have to tune our methodology around a two-pronged approach:

1. Chain of evidence
1. source criticism (how did surviving ancient sources get their information?)
2. transmission of sources from antiquity to the present
2. Burden of evidence
1. independent corroboration
2. contextual fit (how well does a source fit the period it’s describing, the period when it was written, and other writngs by the same author?)
3. intentionality (what is the point that our source wants to communicate, and what assumptions does it take for granted?)

(Points 2b and 2c are cribbed from Jonathan Hall’s History of the Archaic Greek world, 2nd edition 2014, chapter 2.) Taking all these considerations together, if we can work out a narrative that modern interpreters can agree on even when they’re coming to the evidence with opposing biases, then we have something that we can reasonably call a ‘fact’ about antiquity. There’s no question of ‘proving’ anything: ‘proof’ isn’t even on the table. That’s not a disaster, because ‘proof’ is something that only happens in mathematics anyway. Instead we try to find a way of agreeing on a balance of probabilities.

Even when we are looking at people like Thucydides and Cicero who wrote about contemporary events, our information still isn’t first-hand: it isn’t second-hand either. It’s more like twentieth-hand. Take a look at the following diagram:

Manuscript stemma for the text of the ‘Latin Dictys’; colour scheme and dates in the left margin added by me. (Source: W. Eisenhut, Dictys Cretensis Ephemeridos belli Troiani, Teubner, 2nd ed. 1973, p. xlvii; for full explanation see pp. xi-xlviii.)

This diagram shows the surviving manuscripts for a specific text, the Latin version of ‘Dictys of Crete’. It also shows how these manuscripts relate to one another, and the common ancestors from which each manuscript can be shown to have been derived. (For the record, ‘Dictys’ is a quasi-novelisation of the Trojan War, purportedly written by ‘Dictys of Crete’, one of the soldiers in the Greek army. Not well known, but it still has considerable importance in the transmission of Greek heroic legends from Homer all the way to the present day.)

By convention, Roman letters are used for manuscripts that still survive; Greek letters are used for ones that are lost. The methods used for constructing a ‘stemma’ like this are pretty solid and well-tested. They’re the same as those used in evolutionary biology for reconstructing the stemma of extant species of living creatures. (Or rather, biologists originally borrowed philological methods to develop their own.) There are of course always uncertainties, especially when later manuscripts are based on more than one previous manuscript: hence the dashed lines and the occasional question mark in the diagram.

So the diagram shows that we have fourteen surviving manuscripts, and one fragment. From these, the editor has deduced the existence of several lost ‘hyparchetypes’, that is to say, the common ancestors of the surviving manuscripts, represented by the Greek letters β γ ε π ρ σ τ and τ'; and, at the top, the lost archetype from which all of these are ultimately derived. (The E at the bottom represents late annotations in an older manuscript; κ represents a manuscript which is known to have existed, but has been lost.)

The thing that would probably be most startling to the Herodotus questioner, back at the start of this post, is the dates of the surviving manuscripts. Fully half of them date to the 1400s: they’re contemporary with the advent of Gutenberg’s printing press. The very earliest, E, is from the early 800s. And that’s still a long way removed from the original text of the Latin Dictys, which was probably written in the 300s.

In fact it’s even more complicated than that. Because the Latin Dictys was itself a translation from a Greek text: and that was written probably in the late 1st century CE. The Greek text doesn’t survive (except for four fragmentary ancient papyri), but we have a heap of independent testimony about it. Here’s my own reconstruction of the chain of testimony for the Greek version:

Stemma of Dictyean material in Byzantine sources; italics indicate lost sources.
(P. Gainsford, ‘Diktys of Crete’, Cambridge Classical Journal 58 (2012) pp. 58-87, at pp. 65-74)

You’ll notice the Latin Dictys is tucked away in the top left corner. And, for reference, there exist many other Greek witnesses -- the likes of Isaac Porphyrogennetos, John Tzetzes, and Konstantinos Manasses -- who are omitted from this diagram because they’re not independent sources. They don’t add anything to our knowledge of Dictys that isn’t already in the texts shown above.

What’s the upshot? Well, when we open up a copy of an ancient author, whether it’s a popular translation, a slapdash Project Gutenberg text, or a professional critical edition, what we’re looking at is a long way removed from the person who originally wrote it. We’re at least five steps removed from the Latin translator of Dictys -- probably more like a dozen steps removed, once you include the number of steps from the translator’s hand to widespread publication in antiquity. For earlier authors, like Herodotus, we can expect the number of steps to be correspondingly greater.

However, that doesn’t mean that we should have no faith in the text. We don’t need to assume that scribes were indiscriminately adjusting and fiddling with ancient texts throughout history. On the whole, both ancient and mediaeval copyists have been a very conscientious bunch. And perhaps more importantly, the ways in which they have interfered are very well understood nowadays: usually it’s possible to compensate for them, and to correct their alterations, whether intentional (very rare) or accidental (very common). There are uncertainties in modern texts of ancient authors. But a reliable critical edition will always annotate these uncertainties very rigorously.

Popular translations ... well, not so much. Some modern translations are very unreliable. Stephen Mitchell’s translation of the Homeric Iliad omits large chunks of the poem, for reasons that are controversial even among specialists; popular translations of the Bible are often directed by translators’ personal beliefs more than by fidelity to the texts reconstructed by philologists. But that’s just a fact of life: popular versions can never be completely in step with specialists.

Even staying up to date can be hard: thanks to the internet and the vagaries of copyright law, it’s much harder to get access to modern critical editions than to 19th century editions. And that can make a difference, because the modern critical editions are often quite different from the 19th century ones. The current standard edition of the Latin Dictys, by Werner Eisenhut (1973), is based on a detailed analysis of 15 manuscripts; the previous critical edition, that of Ferdinand Meister (1872), was based on just 6, and didn’t put nearly as much effort into working out the chain of influences between them. There’s always room for improvement. For some of the most popular of ancient authors we rely on only a handful of manuscripts, or even on just one: in that kind of situation there’s much more reason to be sceptical about the accuracy of the text.