Revealing the invisible library

November 17, 2015 6:14 PM

Ever since the the Villa dei Papiri and its cache of at least 800 papyrus scrolls was discovered in Herculaneum in 1752, this potential treasure trove of information and insight into the classical world has fascinated scholars with what it could possibly contain. The difficulty has been in how to read them without destroying them. As John Seabrook describes for The New Yorker: "One scroll was peeled apart into many fragments; the other dried up and then, like a disaster in slow motion, split apart into more than three hundred pieces." Now, thanks to new imaging techniques, the contents of the scrolls could—slowly—be revealed.
posted by Athanassiel (20 comments total)

This is potentially really cool - there are so many lost works. Amazing that we are missing stuff by Homer, most great Greek plays, most works of history. From the Roman world there is a lot of cool stuff we don't have: including a 36 volume work called the "shitty sheet" by Catullus.
posted by blahblahblah at 7:09 PM on November 17, 2015 [3 favorites]

This is potentially wonderful, although the story ends on a bit of a downer. Who would believe that this project would be delayed because 1900-year-old scrolls' "owners" are anxious to preserve their intellectual property rights in text that they cannot read and which might be truly unreadable?

Actually, I do believe it, but it's still flabbergastingly venal.
posted by Joe in Australia at 7:21 PM on November 17, 2015 [12 favorites]

Who would believe that this project would be delayed because 1900-year-old scrolls' "owners" are anxious to preserve their intellectual property rights in text that they cannot read and which might be truly unreadable?

Without strong copyright protections what incentive do ancient greeks have to produce more scrolls?
posted by ChurchHatesTucker at 7:28 PM on November 17, 2015 [67 favorites]

It's hard to imagine a poorer application of intellectual property.

more than three-quarters of the Villa dei Papiri has never been excavated at all.

Wow. What may be must drive the scientists involved batty.
posted by Mitheral at 7:53 PM on November 17, 2015 [1 favorite]

Without strong copyright protections what incentive do ancient greeks have to produce more scrolls?

Well, you might say the same about ancient Greek urns.

("Why, what's an ancient Greek urn?" "A drachma a day.")
posted by Joe in Australia at 7:59 PM on November 17, 2015 [11 favorites]

World's Oldest Browser History.
posted by Stonestock Relentless at 8:07 PM on November 17, 2015

Totally amazing.
posted by wintersweet at 8:58 PM on November 17, 2015

The background sound of the mountain breaking on that visualization was ominous
posted by growabrain at 9:19 PM on November 17, 2015 [1 favorite]

The punchline is what killed me, that these books were maybe not quite the equivalent of Danielle Steele, but not too far off.
posted by From Bklyn at 10:54 PM on November 17, 2015

Actually, I do believe it, but it's still flabbergastingly venal.

Seriously. It's important not to jump to conclusions because the institutions involved haven't come right out and said "We're working out a deal with Elsevier" but if that's the reason, given the potential importance of this stuff, humanity would be perfectly justified in hiring a guy with a hat and a whip and a crumpled old map in Occitan to just steal the whole collection from under their noses. ("It belongs in a 'gold' open access journal!")

Shit, I bet Nicolas Cage would do it for free.
posted by No-sword at 3:17 AM on November 18, 2015 [6 favorites]

Oxford's Papyrological Indexing Network (PINAX) has a bunch of the recovered works available online here.

What may be must drive the scientists involved batty.

They worry about rain and other water sources seeping in. I say, where are the gazillionairs when you need them? (Rhetorical question, not requiring answer.)
posted by BWA at 4:15 AM on November 18, 2015

Bah. New Yorker pop-overs. Closed.
posted by clvrmnky at 5:59 AM on November 18, 2015

Okay, they say "Buy More Ovaltine" but, in their defense, the joke was original back then.
posted by dances_with_sneetches at 6:28 AM on November 18, 2015 [2 favorites]

more than three-quarters of the Villa dei Papiri has never been excavated at all.

Wow. What may be must drive the scientists involved batty.

I remember reading two guesses about archeology which were mindblowing, even if only half true, concerning the “what may be” aspect of the field. The first is that there is an archeological site in Israel so big and so deep that at current digging rates it would take over 400 years to fully excavate. Most sites in the world are not currently being dug. The next is that of the hundreds of thousands (maybe into the low millions) of cuneiform tablets currently in collections, almost none have been thoroughly read and their contents studied. About 99% of all cuneiform tablets currently in existence remain in the ground.
posted by Emma May Smith at 9:55 AM on November 18, 2015 [1 favorite]

The pictures I got from googling around for Herculaneum reminded me very strongly of the first world in The Talos Principle - don't know if it was intentional, but if you have played the game you have to agree it's a very fitting parallel.
posted by Dr Dracator at 11:38 AM on November 18, 2015

this is a definite best of web. great post, thank you!
posted by GospelofWesleyWillis at 12:54 PM on November 18, 2015

Emma May Smith: "almost none have been thoroughly read and their contents studied"

Is this a process that is resistant to automation? It would seem that in an age of Google books the tablets would at least be scanned in and some sort of automatic translation be performed.
posted by Mitheral at 4:44 PM on November 18, 2015

It would probably be amenable to some sort of automation if you had the budget to invent a device for scanning tablets, prisms, fragments, etc., and develop software capable of OCRing cuneiform. This is maybe possible but would be a LOT harder than scanning books and OCRing alphabets. (I think automatic translation is too crazy a dream given the state of most of this stuff - probably do more harm than good. Maybe some semi-automatic tentative tagging.)

However, before you started you'd need to convince a few major museums to let you run their irreplacable cuneiform collections through this experimental device. Also the whole time you were doing this conservative politicians would be watching over your shoulder saying "So, this is how you're spending our money? Cuneiform? Interesting. By the way, we're planning to abolish tenure and only give people jobs when their research is directly weaponizable. Just FYI. So, cuneiform, eh?"

Easier overall to wait for the robots to take over. Nanobots, lasers -- they'll do a much better job.
posted by No-sword at 5:05 PM on November 18, 2015

(Oh -- the conservative politicians would be there either way, of course. But they'd take a lot more notice if your budget swelled from "graduate students with cameras and iBooks" to "Google-level robotics lab.")
posted by No-sword at 5:15 PM on November 18, 2015

Is [reading cuneiform tablets] a process that is resistant to automation?

It is surprisingly hard, but also easy!

First, you have the problems associated with all OCR technology: degraded documents (tablets may be broken, abraded, or whatever) and variant letter/word shapes. But, then comes the problem that cuneiform is a writing system, not an alphabet. The same sign, contextually, may be:

• Part of a larger sign or sign-group;
• An ideogram representing one idea;
• A syllable in a Sumerian word;
• A syllable in an Assyrian word (because they basically took over the idea of cuneiform);
• A "signifier" that tells what other signs mean;
• Probably other things that don't come to mind right now.

All that being said, I really do think that a smart OCR person could mechanise the process of transcribing cuneiform to a very great extent, leaving the interpretation to humans. And almost all tablets are things like receipts ("TO UGLUK SHEEP SIXTY AND THREE FOR TWELVE SHEKELS") which can probably be mechanically identified anyway because the formats are pretty standard. So you'd be left with a smaller number of "interesting" tablets that were already partially transcribed, which would be a huge time saver.
posted by Joe in Australia at 5:38 PM on November 18, 2015 [2 favorites]