Sight Unseen: New Tech and Old Texts

History has inflicted many wounds on written culture, from the destruction of the libraries at Alexandria, Constantinople, and Ctesiphon to the work of modern book-burners. Humans treasure their literary heritage, which is why the discoveries of books long thought lost occasion so much interest and celebration.  Brent Seales, a computer scientist at the University of Kentucky, helps scholars to uncover such books through the use of revolutionary, custom-built imaging technology. His recent work on the Herculaneum scrolls, a collection of badly damaged Latin and Greek papyri, might well lead to major new finds. Here, he discusses how he got into the field of antiquities and where he sees the richest potential for its future lying.

Flickr. The Herculaneum scrolls could offer an unparalleled glimpse into the past.

Flickr. The Herculaneum scrolls could offer an unparalleled glimpse into the past.

The sole intact library we moderns have inherited from antiquity comprises a collection of papyri recovered from a villa thought to belong the Roman consul Lucius Calpurnius Piso, the father-in-law of Julius Caesar. This aristocrat’s house stood in Herculaneum, a town near Pompeii and a less well-known casualty of the eruption of Mt. Vesuvius in A.D. 79.

The Herculaneum scrolls, as the recovered papyri have come to be called, presented for centuries a mystery. Their carbonized condition was such that they could not be opened and read. This did not stop the antiquarians of the world from trying. They had no real success until the 18th century, when the Vatican librarian Antonio Piaggio devised a machine to unwrap them. With it, he discovered a set of works by the obscure philosopher and aesthete Philodemus, a Greek resident in Rome (and a possible protege of Piso’s): On Music, On the Stoics, On Flattery. Not a bad haul. But nothing in comparison to what many scholars hoped might be found there: the lost works of Epicurus and Euripides, Aeschylus and Aristotle. Worse still, Piaggio’s method exacted a high bibliological price. It rendered one side of every scroll he unwrapped completely unreadable. Subsequent efforts to reveal the contents of the papyri were not fundamentally any more sophisticated, all being dependent on physical intervention to some degree.

Brent Seales, the head of the University of Kentucky’s Center for Visualization and Virtual Environments and a leading expert on the scrolls, refers to these methods as “hideous.” And he would know. Seales works on a difficult problem in antiquities, one that until the late 20th century had no real technical solution. Namely, how can you decipher texts, like the Herculaneum scrolls, whose physical condition is so fragile that the very act of reading them renders them illegible? This paradoxical problem, which afflicts documents ranging from near-Eastern treasures like the Dead Sea Scrolls to the literary inheritance of medieval Europe, has been a consistent thorn in the side of antiquaries.

But Seales has found a way forward, as a series of triumphant reconstruction attempts has demonstrated. The most recent of these was the so-called Ein Gedi Scroll, a badly-burned document discovered in southern Israel by an archaeologist forty years ago and left since then within its mostly ashen case for fear of destroying it. Seales, neither an archaeologist nor a Biblical scholar, managed to reconstruct the text of the scroll, revealing it to be a copy of the Book of Leviticus dating to the sixth century A.D., a text of almost unthinkable rarity and of huge historical value. How did the affable tech wizard accomplish this miracle?

His technique grows, he told The Octavian Report in an exclusive interview, very naturally out of his career as a computer scientist. When he entered the field in the 1980s, the enthusiasm for machine learning was at a peak. But as efforts at building algorithms for cognition proved increasingly frustrating, Seales turned to a related area where advancements seemed to be coming much more quickly: perception. The explosive growth of the Internet in the 1990s added rocket fuel there, according to Seales, because it was no longer necessary to possess physical media to teach your algorithm. “All of a sudden,” as he put it, “you have this database of images that is available worldwide that you can play on, and the algorithms can sit on your computer and go out on the Internet, grab images and come back and do all that process. That idea was completely new.” This led him, in turn, to the concept of digital libraries, which were then being built up at institutions of higher learning around the world. And because being a computer scientist means being a problem addict, he found himself working on the truly difficult part of digitization, i.e. how you teach machines to read texts where the physical medium has been badly corrupted.

That gave birth to a wide-ranging effort at digitization by Seales’ current institution, which started with a collaboration between Seales and a medievalist on a highly damaged manuscript of Beowulf. And that’s when Seales knew he had found a vocation. He describes himself as being “completely awestruck” by the presence of the Beowulf manuscript and later by the oldest extant copy of Homer’s Iliad, the so-called Venetus A. This copy, which dates from the 10th century, was acquired by a Greek cardinal in the 15th century and later presented as a gift to the Republic of Venice. Seales’ efforts on the Venetus A formed part of the so-called Homer Multitext Project, a scholarly push to digitize all extant manuscripts of the Greek poet’s works so that experts can begin to chart out the both the history of the book itself in early modernity as well as the historiography of its many, many scholiasts. Seales recalls the powerful juxtaposition present and past as he worked on the Venetus: “From where we were, in the Marciana, you could hear this guy playing the piano, constantly, at the bar down below on the Plaza San Marco. La vie en rose, over and over and over again.”

The technical side of Seales’ work took a big leap forward in the early Oughts, when he first started to work with material whose damage was such that it was impossible even to turn the pages or open the scroll that contained the text. He began using a century-old technology — the X-ray — in addition to the cutting-edge algorithms and ultra-high-end custom photography equipment that had previously constituted his arsenal. X-rays reveal the relative densities of materials. That’s why your bones, being denser than your flesh, show up in strong contrast against it in X-ray images. They can do the same for the organic and metallic elements used to create the inks of antiquity, which in general exhibit a far higher density than the vellum or papyrus they have been printed on.

It was this combination of established tech and new ideas that put Seales in a prime position to begin work on the Herculaneum collection. In 2005, he got a chance to redeem the well-meant but ill-omened efforts of his predecessors and take the first images of the papyri collected without damaging the fragile scrolls beyond any hope of repair. Over the next four years, Seales and his team labored in concert with a French antiquarian to develop images of the text hidden within the scrolls. But in 2009 they faced a major setback. Despite their painstaking efforts, they were not able to generate images with enough contrast to reveal the ink against its physical background. Soon after came the shutdown of his team’s access to the vast majority of the papyri by the relevant Italian authorities, as well as contradictory claims of success from an Italian imaging team. Seales still regards these events as perhaps the signal disappointments of his career, and pointed out ruefully that the study of antiquities, no matter what the medium, requires alongside scholarly and technical abilities a firm hand in navigating the often massive bureaucracies that have grown up around a country’s cultural treasures. But he has hopes that his stunning success with the Ein Gedi Scroll will open the doors for him to work on the smaller collection of Herculaneum papyri held by Oxford’s Bodleian Library, where he is headed later in 2016 to present his case. Seales considers himself “perched” to make a real breakthrough on the Herculaneum material, and believes that if all the bureaucratic issues could be resolved at one fell swoop, he and his team stand less than a year away from revealing a complete manuscript to the world.

We have a seemingly unstoppable urge to uncover every shadowed corner of our past. It takes us to tombs long thought lost, the sea floor, and other extreme environments. Our ever-expanding mastery over land, air, and sea via technology means that true challenges to this urge are growing few and far between. The Herculaneum scrolls are clearly one such challenge. But whatever their ultimate fate, Seales has greater hopes for the technical side of his profession more broadly. One of the most persistent problems with the digital examination of manuscripts is the fact that, as noted, previous ages had recourse to means of inquiry that very often damaged the materials they were meant to expose. The problems with machine learning that initially propelled Seales into visualization, he says, are not now nearly as fearsome as they used to be. Great strides have been made since the 1980s on questions of artificial intelligence, which puts within the realm of possibility, according to Seales, technology that might allow the recovery of even the most physically fragmented texts. He also pointed out a belief, held by some scholars, that the discoveries in Piso’s villa represent less than half of the actual library, and that if excavations were to be extended, another 2,000 volumes might lie within reach. So stay tuned . . .