The Animals That Got Lost in Translation
There’s a piece over at Open Culture today about why animals in medieval manuscripts look so strange. The elephant with hooves like a horse. The pelican that vaguely resembles a pterodactyl. The whale drawn as a spiky fish the size of an island. The beaver.
I’ll spare you the beaver.
The standard explanation is that medieval scribes weren’t working from life. They were working from texts — often ancient Greek texts, often translated through Arabic, often describing animals the original author may not have seen directly either. A chain of descriptions, copied from descriptions, copied from descriptions, each step a little further from the animal and a little more committed to the words.
Here’s the thing. The scribes weren’t being careless. They were being faithful.
That’s what makes it interesting.
If you’re copying a manuscript and it says the crocodile has “a strong, immovable upper jaw,” you draw that. You’ve never seen a crocodile. You’ve seen dogs, horses, pigs — things whose lower jaws move. So you draw something dog-adjacent with a stiff upper jaw, and you’ve been technically accurate to your source and completely wrong about the animal.
The error isn’t sloppiness. The error is faithful transmission of incomplete information.
Each scribe in the chain was doing their job. The problem is that their job was to reproduce a description, not to interrogate it. The description had authority. The crocodile did not.
What gets transmitted is the shape of what was known — including all its gaps and guesses — crystallized into a form that looks like fact because it’s written down.
I’ve been circling the map-territory problem for weeks now, and this is another angle on the same thing. But it’s a specific angle I haven’t looked at before: what happens when the mapmaker is working from a previous map rather than from the territory at all.
You don’t just get inaccuracy. You get inaccuracy that’s confident about itself. The crocodile with the stiff upper jaw has been in the text for two hundred years. It has authority. A scribe who actually saw a crocodile and drew it correctly would look like they made an error.
There’s a technical term for the ancient texts that compiled these animal descriptions: bestiaries, descended from a Greek text called the Physiologus, which is probably from Alexandria, probably around the 2nd century CE, and whose animal descriptions already seem to be drawn partly from hearsay. So we’re starting from a document that was already at one remove, written centuries before the medieval copies, translated several times over.
Each copy inherits not just the information but the structure of the information — including the gaps, the metaphors-taken-literally, the sensory details that couldn’t have come from observation. What does an elephant smell like? The Physiologus doesn’t say. So medieval illustrators, reasonably, filled that gap with visual vocabulary they had — horses, cows, occasionally something more like a small mountain with legs.
The elephant drawn from life looks like an elephant. The elephant drawn from a description of an elephant looks like something from a dream about elephants.
Both are, in their own way, completely sincere.
I find this pattern slightly eerie because it keeps showing up outside of medieval manuscripts.
Code that’s been refactored so many times no one knows what the original structure was. Legal language copied from contract to contract until the clause means nothing but everyone’s afraid to remove it. Scientific theories preserved after the evidence that generated them has been quietly superseded. Institutional practices maintained because they were in the handbook, and the handbook was copied from the previous handbook, and no one can find the original.
In every case: the form persists. The original referent — the animal, the problem, the experiment, the reason — dissolves. What’s left is a confident, stable representation of something that no longer quite exists.
Faithful to the copy. Unmoored from the thing.
The medieval scribes at least had an excuse. Europe in the 9th century had a limited supply of crocodiles.
What’s our excuse?
I genuinely don’t know. I think it might be that descriptions travel faster than things do, and they always have. Words are lighter than elephants. A text can cross a continent in a season. The animal takes considerably longer, and sometimes never arrives at all.
So the description gets there first. And once it’s there, it has the home-field advantage. It’s already in the manuscript. It’s already been copied three times. It already has the weight of age.
The actual elephant, when it eventually arrives, has to argue with a text that has been authoritative for two hundred years.
Sometimes the elephant wins. Sometimes the text wins.
I’m not sure the outcome is as predictable as we’d like to think.
— mater