Berkeley burned replica scrolls written by high school students to read the Herculaneum papyri

Researchers affiliated with the University of California, Berkeley, had high school students copy passages onto papyrus with reed pens and lampblack ink, then baked the scrolls in a furnace until they were charred. The point was to learn whether lead in an ink can make text on a burned scroll show up in an X-ray, and the answer was that lead absorbed as much as 25 times more X-rays than the burned papyrus around it.

The experiment, published in PLOS One on September 16, 2026, was built as a stand-in for the roughly 1,800 scrolls and fragments buried by the eruption of Mount Vesuvius in 79 CE at Herculaneum, described in the coverage as the only intact library to survive from antiquity. Earlier attempts to unroll those fragile carbonized documents by hand often destroyed them, which is why a non-invasive reading method is worth so much and why the team tested its idea on scrolls made for the purpose.

Making a scroll to burn

The lead author, Douglas Seiler, is a Berkeley affiliate and retired inventor who funded the work himself, according to the university’s news release. His team used Egyptian papyrus, a Japanese lampblack ink and reed pens, and hired high school students to write the text, which included lines from Star Wars, the Bible and a quote from the 1960s television series The Outer Limits.

Some of the ink was mixed with measured amounts of lead nitrate. The PLOS One paper gives a range of 16 to 84 micrograms of lead per square centimetre to match concentrations reported in ancient samples. The rolled scrolls were then carbonized in a low-oxygen furnace, imaged with laboratory X-ray tomography at the National Institute of Standards and Technology, and virtually unrolled by software.

Text with lead was readable. Carbon-only ink did not register at all, which is the fact that has kept Herculaneum’s ink so hard to see.

What the X-rays saw

Ink with as little as 25 micrograms of lead per square centimetre stayed detectable in the paper’s tests, and the virtual unrolling recovered a legible phrase, “the children of”. ScienceDaily’s version of the release reports that lead absorbed as much as 25 times more X-rays than the burned papyrus, which made the writing stand out dramatically. A handheld X-ray fluorescence scanner also picked up lead on the rolled scrolls, and the paper’s summary is that carbon ink alone gave no X-ray contrast while the leaded inks did, down to about 25 micrograms per square centimetre.

The handheld device is the practical prize. Berkeley’s release describes it as a simple, inexpensive way to check a scroll for leaded ink before committing it to CT scanning, a triage step that could decide which of the scrolls deserve the scarce time on a detailed imaging instrument, which the paper describes as synchrotron analysis.

The catch, and the Herculaneum problem it addresses

The Herculaneum collection sits in a strange position. Physically unrolling the fragile carbonized rolls has destroyed many of them, and the ink on most is carbon, which is nearly indistinguishable from papyrus in X-ray CT. The Vesuvius Challenge, a machine-learning competition to read the scrolls without opening them, broke through in 2023, and its Grand Prize page records $700,000 awarded to Youssef Nader, Luke Farritor and Julian Schilliger for reading text from a scroll, after organisers had put the odds of success below 30 percent. That result depended on subtle crackle patterns in the ink, not on any chemical marker.

The Berkeley paper’s claim is conditional. Its authors write that the presence of lead in Herculaneum scrolls would greatly increase the probability of successful reading. Whether the ink on the real scrolls contains lead in useful amounts is the thing a handheld scan would test, and the replica experiment cannot answer it. It only establishes that if lead is there, the signal is strong.

Leah Packard-Grams, a Berkeley papyrologist consulting on the project, said in the release that the value of the work was practically unquantifiable and the stakes the largest in the history of Greek literature. CNN’s report on the experiment covers the same argument.

Seiler explained to Gizmodo why the team chose replicas: a model can be damaged with no consequence, while a 2,300-year-old artifact cannot. Gizmodo’s account notes the furnace was chosen to mirror the volcanic heat that carbonized the originals.

If a lead-bearing scroll is found, the work moves from a furnace at a laboratory to a museum collection, and the next step is the two-stage plan the paper describes: screen by handheld scanner, then send the promising specimens for detailed imaging. The replica experiment supplies the calibration for that plan, including the 25-microgram floor below which a lead signal cannot be counted on.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview