Christian scribes interfering with ancient texts?

Are texts like Herodotus’ Histories and Petronius’ Satyrica the same as when the authors wrote them? Or have they been irreparably messed around with by the dozens of scribes who copied them and re-copied them over the centuries?

How likely is it that mediaeval scribes with a Christianising agenda haven’t meddled with ancient texts? (Image: cover of Bart Ehrman’s book Forged, 2011)

Modern readers of ancient literature often don’t spare a moment to think about the many steps separating modern editions from the ancient ones. When they do, though, sometimes they’ll begin to have serious doubts about how well they match the originals. They may even completely lose trust in the textual tradition.

Passages like the following don’t exactly inspire confidence.

At this time came Jesus, a wise man. If indeed it is right to call him a man: for he was a wonder-worker, a teacher of people who delight in receiving the truth, and he attracted many Jews, but also many of the Greek (world). This man was the Messiah [Christos].

And when he was accused by the leading men amongst us, Pilate sentenced him to crucifixion. Even then, those who had loved him at the beginning did not stop doing so. For he appeared to them on the third day, alive again. The divine prophets had announced these things and countless other marvels concerning him. And even now the tribe named after him, of the Christians, have not yet disappeared.

Josephus(?), Antiquities of the Jews 18.63-64 (§3.3)

This passage appears in manuscripts of the 1st century historian Josephus. But it’s basically impossible to accept that Josephus really wrote it. He was a romanised Jew, and definitely no Christian. The passage is so notorious that it has its own nickname: the ‘Testimonium Flavianum’. (‘Flavius’ is the Roman name that Josephus acquired when he defected to the Romans.)

It’s nearly certain that a Christian scribe or scribes interfered with the text of Josephus at this point. Origen knew book 18 of the Antiquities, and he flatly states that Josephus did not accept Jesus’ messiahship. That was in the 200s. By the 300s, though, the damage had been done: Eusebius quotes the corrupt passage in full, three times, as an example of a non-Christian Jew accepting Jesus’ divinity.

Note. Origen on Josephus: Against Celsus 1.47 (referring to Ant. 20.200, but also with knowledge of book 18); Comm. Matthew 10.17 (referring to Ant. 20.200). Eusebius on Josephus: Demonstratio evangelica 124b-c (3.5); Ecclesiastical history 1.11; Theophany 250 ed. Gressmann. Later witnesses to Josephus also quote or paraphrase the Testimonium.

There’s room for disagreement over the extent of the scribal misdeeds here — some scholars think it’s a complete fabrication, maybe even forged by Eusebius himself; others think that maybe only the bits in bold have been tampered with — but there’s no doubt that tampering did happen.

The Testimonium Flavianum as it appears in cod. Ambrosianus F 128, 11th century. (Base image: Roger Viklund’s blog)

When we look at this, shouldn’t we just give up all hope of ever reading the texts that ancient writers actually wrote? Shouldn’t we take it for granted that ancient texts are riddled with forgeries? — and that the Christian church, which transmitted ancient texts through the mediaeval period — well, Latin texts, anyway — sanitised everything it could, and destroyed anything it couldn’t?

Well, I have good news and bad news. First, the bad news. We know that there are loads and loads of ancient forgeries — that is, ‘forgeries’ in the sense that we can be sure they weren’t really written by the author that they’re linked to. We know that Euclid didn’t write large chunks of the Elements, that Simonides didn’t write the famous epigram about the Spartan dead at Thermopylae, that Euripides didn’t write the Rhesus, that Seneca didn’t write the Octavia, and that Matthew, Mark, Luke, and John didn’t write the Christian gospels of those names.

This is called ‘upward attribution’. When people don’t know who the real author of a book is, they very often end up assigning it to a famous name. You can see the same process in recent stuff too. When people assign spurious quotations to Albert Einstein, or any Japanese animé to Hayao Miyazaki, or the ‘Windows 95 song’ to Weird Al Yankovic, that’s upward attribution. It’s not usually a deliberate deception. It’s a natural consequence of a group of people having an interest in a book, but imperfect information about it. Errors can often go viral in that situation.

And now, the good news. Authorial attributions may sometimes be doubtful or wrong, but the texts themselves are comparatively robust. Did you notice, when I introduced the Testimonium above, I said ‘passages like the following’? That was intentionally misleading. In fact there is nothing like the Testimonium. As an example of religiously-motivated tampering, which has completely displaced the original text, it is unique.

We know of other examples of Christian interference, but none where that interference was anything like as successful as it was in the case of the Testimonium, and hardly any of a comparable size. (If Christian scribes had really censored and re-written ancient texts that way, how would we ever have got the bawdy humour of Aristophanes or Petronius? And no, Sappho isn’t a counter-example. Ancient moralists sometimes took issue with her sexual reputation, both pagan and Christian, but no one took a decision to destroy her poems. There was still an edition of her poems in use in Egyptian schools as late as the 7th century. Sappho wasn’t lost because she was a woman or because of her sexuality: she was lost because she was a lyric poet, and all early lyric poets ended up being lost.)

Critical edition of Herodotus, with the apparatus showing the omission of Histories 1.199 in three of the twelve manuscripts that the modern editor used for this edition. (Base image: Herodoti Historiae, ed. C. Hude, Oxford, 1908)

The biggest parallel I’m aware of is in Herodotus. Some manuscripts leave out a chapter, Histories 1.199, that describes sacred prostitution at the temple of Ishtar in Babylon. But I’d better repeat: some manuscripts. The bowdleriser’s limited success only goes to show how unrealistic it was for a misbehaving scribe to expect to taint the entire manuscript tradition.

You see, scribes were on the whole a conscientious lot. Sure, there were exceptions, like the assholes who tampered with Josephus and Herodotus. But there was no Mediterranean-wide conspiracy. Dishonesty and ulterior motives have always existed, but scribes in different places and times acted independently of one another. The ulterior motives that drove scribes in one generation, or in one place, had no guarantee of persisting into the next.

As a result, when scribes spotted problems in a text they were copying, they very often took pains to correct them. It’d be going a little far to say that the system was self-correcting, but the system did overall aim at faithful reproduction and restoration. That’s why Herodotus’ bowdleriser didn’t get away with it. In Josephus’ case, the forger got luckier — but also bear in mind that there we aren’t talking about a mediaeval forger, but an ancient one, probably earlier than the 300s.

Nearly all transmission errors that we see in mediaeval copies of ancient texts are accidental. Where we do see damage caused by a scribe’s deliberate action, it’s normally because the scribe is trying to fix it, but botching the job. Most ‘deliberate’ errors are caused by scribes spotting a problem, trying to correct it, but not getting it quite right; or else spotting something that looks like a problem but is actually correct.

Here’s an example of a Christian scribe ‘correcting’ something that looked like a mistake, when it was actually right. Maybe, once you know the details, you’ll agree with me that the scribe was conscientious.

Petronius’ bawdy Satyrica (1st century CE) is a novel about the hapless Encolpius, his love-slave Giton, and friend Ascyltos, and their misadventures in the Greek parts of southern Italy. At one point they attend a party where many of the guests can barely speak Latin. The language jokes make this an exceptionally difficult text, especially for mediaeval scribes who normally didn’t know any Greek. According to the sole surviving manuscript, a character says

aut ego non me novi, aut non deridebis, licet barbam auream habeas. Sathana tibi irata sit, curabo ...

‘Either I don’t know who I am, or you won’t get the last laugh, even if you have a golden beard! I’ll make sure Satan is angry at you ...’

Petronius, Satyrica 58

Now, Satan clearly doesn’t belong in a 1st century novel written by a Roman aristocrat. Petronius probably never even heard of Satan. He must originally have written Athana, not Sathana: the Greek goddess Athena, but with her name in the Doric dialect. The episode is set near Naples, and there was more than one temple of Athena nearby: one on the cape south-west of Sorrento, and another in Paestum a little further south.

The scribe who introduced the error had a source text that didn’t look like the one I printed above. The manuscript he was copying from would have looked more like this:

...LICETBARBAMAVREAMHABEASATHANATIBIIRATASIT ...

Latin manuscripts usually don’t have word breaks or punctuation. Now, our scribe knew perfectly well that ATHANA wasn’t a Latin word or name. But he didn’t know Greek, and he certainly didn’t know Doric! So he didn’t realise it was a deliberate solecism: he very reasonably inferred it was a copying error. He imagined that Petronius had originally written HABEASSATHANA, with a name that was very familiar to the scribe and which happens to work very nicely in the context: ‘Satan’. At some point, he guessed, another scribe had accidentally missed out the second S. And so he added it back in.

Critical edition of the Petronius passage, with the apparatus showing editorial repairs to the manuscript readings. Notice also the obeluses or ‘daggers of despair’ on the following line, indicating a place where the text is so corrupt that it can’t be restored with confidence — not because of scribal malevolence, but once again because there’s too much Greek (ἀλογίας and maybe more), and the scribe couldn’t understand it. (Base image: Petronii Arbitri Cena Trimalchionis, ed. Martin S. Smith, Oxford, 1975)

This is what actual Christianising interference usually looks like. It’s an error, but it arises out of the scribe’s conscientiousness, as an attempt to restore the text, not erase it. This kind of thing is much more typical than the intentional damage that we see in Josephus, anyway!

In a sense, Antiquities 18.63-64 is a precious treasure: it’s one of the very, very few cases of deliberate forgery that was so comprehensively done that we actually can’t be confident in restoring the correct text. Irreparable textual forgery is a real rarity! Texts that have been attributed to the wrong authors are common; textual mistakes are very common. But malicious textual inaccuracies — not so much.

Where we have multiple manuscript traditions, they’ll normally provide enough evidence for a modern editor to fix malicious alterations. The entire process of modern textual criticism is devoted to teasing out transmission errors and forgeries, and repairing them. When even that isn’t enough, when it just isn’t possible to restore the original text — as with the Testimonium Flavianum — any responsible critical edition will highlight that fact very transparently. Because that’s the entire point of critical editions.

Further reading

• Olson, K. A. 1999. ‘Eusebius and the Testimonium Flavianum.’ Catholic Biblical Quarterly 61.2: 305-322.
• Reynolds, L. D.; Wilson, N. G. 1991 [1968]. Scribes and scholars: a guide to the transmission of Greek & Latin literature, 3rd edition. Oxford: Clarendon Press.