The paint was newer than almost everything else on the page.

That was the problem.

By the time the Archimedes Palimpsest reached the Walters Art Museum in Baltimore in 1999, nobody needed convincing that the book had been mistreated. Mold had eaten through portions of the parchment. Several leaves known to exist when Johan Ludvig Heiberg photographed the manuscript in the early twentieth century were gone. A modern synthetic adhesive had been worked into the spine so aggressively that conservator Abigail Quandt later described it as tougher than the animal skin it was stuck to.

Then there were the paintings.

Four portraits of the Evangelists had been added over existing pages. They looked old enough to belong in a Byzantine prayer book. They were not. John Lowden of the Courtauld Institute established that their models came from illustrations published in 1929. The Walters has described them plainly as twentieth-century forgeries, apparently added to make the manuscript more valuable.

Under the forged saints lay a thirteenth-century prayer book.

Under the prayer book lay Archimedes.

The newest layer turned out to be the hardest one to get through.

Seven books became one

The object we call the Archimedes Palimpsest was never originally one book.

The oldest Archimedes material in it was copied onto parchment, probably in Constantinople, during the second half of the tenth century. That manuscript contained seven Archimedean treatises, including the only surviving copy of The Method of Mechanical Theorems, the fragmentary Stomachion, and the only surviving Greek text of On Floating Bodies.

Archimedes himself had been dead for more than a thousand years by the time that copy was made.

Then the book was recycled. In the early thirteenth century, parchment from the Archimedes manuscript and six other books was scraped, cut down the center, rotated ninety degrees and folded into smaller leaves. A scribe used the reused material to make a Greek Orthodox prayer book in Jerusalem. A colophon recovered under ultraviolet light dates its completion to April 1229.

This sounds brutal to modern ears because we know what was underneath. The person making the prayer book did not. Parchment was expensive prepared animal skin, durable enough to scrape and reuse. The old mathematical manuscript had become raw material.

That act of destruction is also the reason the text survived. Had the Archimedes leaves simply ceased to be useful, they might have disappeared entirely. Instead, somebody made them useful again.

A mathematician looks sideways

The prayer book eventually entered the library associated with the Greek Patriarchate of Jerusalem and was housed in Constantinople. In 1899, Athanasios Papadopoulos-Kerameus catalogued it and transcribed a few faint lines from the erased writing underneath. Those lines reached Johan Ludvig Heiberg, one of the leading authorities on Archimedes.

He recognized them.

In 1906, Heiberg went to Constantinople to examine the manuscript himself. He discovered that the faint text running sideways beneath the prayers contained major works of Archimedes. He photographed pages and worked through whatever he could decipher with the tools available, incorporating the readings into his later edition.

It would be easy to turn this into the usual lost-genius story. Reviel Netz, one of the modern scholars who worked on the Palimpsest, cautions against exactly that. Much of Archimedes was already known through other manuscript traditions. What the Palimpsest supplied was different: another early textual witness, The Method and Stomachion, the Greek text of On Floating Bodies, and diagrams that other traditions did not preserve.

The book did not give us Archimedes.

It gave us parts of him we did not have.

Then the book disappeared

Heiberg last worked with the Palimpsest in 1908. After that, its history becomes murky.

No reliable account establishes exactly how it left the church collection in Constantinople. By the late twentieth century it was being presented as part of a private French collection. When Christie’s prepared to auction the manuscript in New York in October 1998, the Greek Orthodox Patriarchate of Jerusalem challenged the sale and claimed the manuscript had been stolen. A federal judge rejected the attempt to stop the auction. The manuscript sold on October 29 to an anonymous American collector for just over $2 million including the buyer’s premium.

The ownership dispute is not necessary to resolve the manuscript’s scientific story, and the gaps in its twentieth-century provenance have not somehow disappeared.

What we can say is physical. The book that resurfaced was worse than the book Heiberg had photographed. Three pages containing Archimedes material that Heiberg had seen were missing. Mold had destroyed portions of surviving text. The forged paintings had appeared.

Four years to open a book

The anonymous buyer deposited the Palimpsest at the Walters for conservation, imaging and scholarly study. Before anybody could begin throwing sophisticated cameras at it, Abigail Quandt had to make the object safe enough to handle.

Disbinding began in February 2000. It finished in November 2004.

Four years is a remarkable amount of time to take apart a book until you see the book.

The erased texts ran through the gutters created when the old leaves had been folded into the prayer book. To image the hidden writing properly, the leaves needed to be separated. Then Quandt reached the modern glue. The twentieth-century adhesive was stronger than the parchment itself. Trying to remove it too aggressively could destroy the thing the project was trying to save.

So the work became microscopic. Wax from centuries of candlelit use had to be removed. Fragile edges were supported. Mold-damaged parchment was stabilized. Folded or stuck areas were eased open. Tiny fragments, paint chips, threads and bits of parchment were catalogued according to where they had come from.

“Technology revealed Archimedes” is true only after somebody spent four years making sure the technology did not finish destroying him.

Twelve ways of looking at the same page

The imaging effort began modestly and kept changing because the scholars kept asking for something better.

Early experiments used narrow-band filters and relatively low-resolution digital capture. The results looked impressive to nonspecialists and irritated the people who actually had to read tenth-century Greek handwriting. Three hundred dots per inch was not enough when a scholar wanted to enlarge a single damaged character across a monitor.

The scientists had initially assumed their task was to make the later prayer-book writing disappear. The scholars told them that was not particularly helpful. They needed to read what happened where the prayer text crossed directly over an Archimedes letter.

By the final imaging campaign in 2007, entire leaves were being captured at roughly 800 dpi in twelve wavelengths from ultraviolet through infrared. The team reimaged the whole manuscript in under three weeks.

Even then, the raw photographs were not the answer. Post-processing was. Different spectral responses from parchment, prayer-book ink and erased ink could be mathematically combined so faint undertext became readable.

Most of the manuscript began giving itself up.

The forged pages did not.

The forger had used paint

Multispectral imaging relies on differences in the way materials respond to light. That works beautifully until somebody covers the parchment with enough material that light does not reach what you want to see.

One of the painted pages covered part of Archimedes’ Equilibrium of Planes. The solution was to stop looking primarily at reflected light and start looking for elements inside the ink.

The old inks contained iron. X-ray fluorescence could excite atoms within the parchment and measure the characteristic X-rays they emitted. By mapping the iron signal across the page, researchers could build an elemental map rather than a conventional photograph. Where the erased Archimedes letters contained iron, the letters could appear beneath other material.

Initial experiments worked. They also worked very slowly. Recovering a small readable section beneath one forged page took roughly ten hours.

That was when the project went looking for a bigger X-ray source.

Archimedes went to a particle accelerator

Uwe Bergmann and colleagues at the Stanford Synchrotron Radiation Laboratory brought manuscript leaves to Beamline 6.

The phrase sounds more alarming than manuscript conservators generally prefer.

Synchrotron radiation could be tuned precisely enough to excite iron in the ancient ink and was vastly more intense than the laboratory X-ray equipment used in the earlier tests. It offered a way to scan larger areas far more quickly.

It also raised an obvious problem. The X-rays might damage the parchment.

Before the Palimpsest went into the beam, parchment exposure was tested so the team could establish conditions considered safe enough for the manuscript. A detector tuned to the iron signal sat at right angles to the beam while the leaf moved across the scanning stage.

The same forged page that had yielded only a small section after ten hours of laboratory XRF produced an entire column of Archimedes text in about twenty-four hours at Stanford.

A third-century BC mathematician writes. A tenth-century scribe copies him. A thirteenth-century bookmaker scrapes the copy away. A twentieth-century forger paints a saint over the parchment. A twenty-first-century physicist sends synchrotron X-rays through the saint to map iron left by the tenth-century ink.

Nobody involved in the first four steps could have anticipated the fifth.

The other books underneath

The project became more interesting when researchers stopped assuming every erased line belonged to Archimedes.

In 2002, Natalie Tchernetska identified portions of speeches by Hyperides, a fourth-century BC Athenian orator whose works were thought not to have survived in the medieval codex tradition. Five bifolia preserved roughly 320 lines of newly recovered material, including parts of previously lost speeches.

In 2005, Nigel Wilson noticed the word “Aristotle” in another difficult erased text. That eventually proved to be a commentary on Aristotle’s Categories, probably from the second or third century AD. Alexander of Aphrodisias has been proposed as an author, but the evidence does not establish that decisively.

By 2011, researchers understood that the prayer book contained parchment recycled from seven earlier manuscripts. The Archimedes Palimpsest turned out to be a misleadingly singular name. Several lost books were trapped inside it.

What Archimedes actually gained

The project also resisted the temptation to oversell what the new readings proved.

Netz specifically warns against claims that the manuscript suddenly revealed whole unknown Archimedean treatises in the modern era or that nothing important was known about Archimedes before Heiberg.

The gains were subtler and, to me, more satisfying. Modern readings corrected large numbers of details in Heiberg’s transcription and restored diagrams he had not published. A damaged passage of The Method revealed Archimedes explicitly comparing infinite collections in a way unusual in surviving ancient mathematics. Better readings of the Stomachion support the possibility that he was studying a combinatorial problem involving arrangements of fourteen pieces, although that interpretation remains scholarly reconstruction rather than an authorial label recovered from the page.

Archimedes does not need to be retroactively turned into Newton, Leibniz or a computer scientist.

His surviving work is strange enough on its own.

The digital copy

There is one final piece of the project that deserves more attention than it usually gets.

The manuscript remains privately owned.

The data do not.

The project released imaging datasets and transcriptions online. Its image bank licenses project imagery for reuse under Creative Commons Attribution 3.0. The team also developed metadata conventions specifically for the odd spatial relationships among normal photographs, processed images, X-ray fluorescence maps and transcription layers.

That feels like an appropriate response to a book whose physical history is largely a story about restricted access, disappearance and damage.

There is a natural connection here to They Thought the Books Were Charcoal. Herculaneum and the Palimpsest require very different imaging methods, but both force the same question: how do you read an ancient text when physically opening, scraping or cleaning it can destroy the evidence?

The painting

The forged Evangelist still exists on the page.

Nobody scraped it off to reach Archimedes. That would have repeated the logic that created the Palimpsest in the first place: destroy the later layer to retrieve the earlier one.

Instead, the project kept all of them. The prayer book matters. The forgery matters. The mold damage matters. Heiberg’s photographs matter because some of what he saw is now physically gone. The X-ray images matter because some surviving writing cannot be seen any other way.

And underneath the fake painting is still an iron stain left by a scribe who copied Archimedes about a thousand years ago.

The newest damage was the hardest to see through.

Sources & Further Reading

The strongest material for this piece comes from the project itself, the Walters Art Museum, and the scholars and scientists who carried out the recovery work.

• The Archimedes Palimpsest Project, manuscript history — chronology of the tenth-century copy, 1229 prayer book, Heiberg’s work, later damage and forged paintings.
• The Archimedes Palimpsest Project, conservation documentation — disbinding, modern adhesive, mold, wax and preparation for imaging.
• The Archimedes Palimpsest Project, image capture — the evolution from early narrow-band imaging to the final twelve-wavelength campaign.
• The Archimedes Palimpsest Project, X-ray fluorescence imaging — elemental mapping, forged pages, parchment safety testing and work at Stanford’s Beamline 6.
• Archimedes Palimpsest Image Bank — project images and reusable digital data.
• The Archimedes Palimpsest Project — project overview, scholarship and publication history.
• The Walters Art Museum — institutional home of the conservation and research project.