Uncovering the Archimedes Palimpsest | Will Noel | TEDxMidAtlantic

If Archimedes was around currently, or TEDx programs were around at the time of Archimedes, he as the greatest mathematician of antiquity might have been an interesting speaker. He would have had this little countdown clock over here going from 18 minutes down to zero, but still he would have had an invitation. He's not available as I understand it, but Dr. Will Knoll is, and Dr. Knoll has been in charge as the curator of rare books and manuscripts at the Walters for over 10 years now, and has some insights on Archimedes and other matters. Please give him a warm welcome. Hello everybody. This is not so much my favorite story as my only story, and it's a story of what happens if you have a big enough problem, absolutely no resources, and the ability to shout help very loudly. Archimedes died in 212 BC, but he wrote 10 treaties of truly astonishing brilliance, and he wrote them rather boringly in just three books, codices A, B, and C. Codex A was lost in 1564. Codex B was lost by the Pope in 1311, and Codex C arrived on my desk on 19th of January 1999, and this is the story of Codex C. It was written in the 10th century. It is the unique source for on-floating bodies in the original Greek. It is the unique source for Archimedes' treatise Stomachion, and the unique source for the method of mechanical theorems, all of which is rather exciting if only you can read it. The reason you might not be able to read it is because in the 13th century, the book was taken apart. All the text was scraped off. It was cut down in the middle. Imagine your Baltimore Sun cut down in the middle and stacked in a corner, and someone wanted to make a prayer book. Their prayer book was bigger than the Archimedes book, so they got six other books and did exactly the same thing to that, scraped off the text of all of those, recombined them together to create a completely new book with the text completely written off. This book is the ultimate write-off. But if you can read it, then you can find out some astonishing things. I want you to look at that. That thing there is the armpit of a goat that was grazing in the fields in Constantinople in the year 1000. It's the most important goat in the history of Western mathematics. This book was a prayer book, but it was discovered in 1906 by a man called Johan Ludwig Heiberg, who did his very best to transcribe it without any modern technology, and the guy was brilliant. His discovery made front-page news in the New York Times on July 16, 1907, which is great. You have discovered three unique treaties by the founding father of Western mathematics, and that's what you've been dedicating your entire life for. Once you found a book that's that important, you'd have thought that it was safe. This book wasn't safe. It disappeared for the entire 20th century, and it was sold to an American collector on the 29th of July of October 1998 for 2.2 million. When it was bought again, not many people thought that it would be of research interest for two reasons. One is Heiberg was brilliant, and the other is that it's now in very bad condition. That's Heiberg, the 16th page of the Method of Mechanical Theorems that Heiberg read. That's what that page looks like now. It's been painted over with a modern forgery. This was done after 1938, knowing that Archimedes was underneath it. The book's also suffered very bad from mold. This book is a write-off. Nonetheless, it was put on my desk by the owner for the purposes of exhibition, conservation, imaging, and display, so that's what we've been trying to do ever since. At the Walters Art Museum just down the road. When the news came out that we had this, we got lots of people saying, I can help, including Rasputin's great-grandson, who said it was hid. But this is Michael B. Toth, who wrote an email saying, Dear Dr. Knoll, read with interest your article in the Washington Post. It puts our work into some perspective. We here at the National Reconnaissance Office might have some equipment that might be able to help. Please get in touch. Good luck with your research. Yours sincerely, Michael B. Toth, National Reconnaissance Office. Okay, thanks, Mike. Actually, we can't use the National Reconnaissance Office because it's a private book, so we can't spend your tax dollars on it, but Mike's a volunteer, and he talks to me about cost schedule and performance. He does the Gantt charts. I run. The critical path is Abigail Quant. She's Senior Conservator of Manuscripts at the Walters Art Museum. She does the conservation on the work. Now, parchment is normally tough, but this stuff is tissue paper because of the mould, and that's probably got writings by Archimedes on it. That is the spine of the book, and this is polyvinyl acetate. It's Elmer's wood glue, effectively, and Archimedes' text is running through the spine. It took Abigail four years to take the book apart. This is a rare action shot. Everything about this book is a total nightmare. We had lots of collaboration from all sorts of people, including the Canadian Conservation Institute. We get all the little fragments, the tiny little fragments normally you wouldn't care about. You care about them if it could be Archimedes. You image them in ultraviolet light, then you stick them back together. Here is a gutter. Abigail performs what I call brain surgery. There's a before, there's an after. This is an ultraviolet after with a little circle round there saying, let me try and get that back, from our scholar in California saying, this is the earliest symbol of a circle in the history of Western mathematics. That's a good day. There were lots of bad days when we started on this project, but it was fairly clear from the beginning that what we were going to do was we were going to do multispectral imaging. The people that we got to do the multispectral imaging were three people, Roger Easton, who's Professor of Imaging Science at RIT, Bill Christens-Barry, who at the time was working for the Applied Physics Lab, but now runs Equipoise Limited, and Keith Knox, who now works for the Boeing Corporation on Hawaii looking at very small things, most of which he's not allowed to talk about, but which come from China. Multispectral imaging, you take images in different wavebands of light, and then you write algorithms to bring out what you want. That's what they did, and we wanted to bring out the Archimedes text that was underneath the overtext, and you do that, and you image the entire book in 16 different wavebands of light. There's the armpit of the most important goat in the history of Western mathematics, 16 times. Each one of those images of themselves is not only good because you can't read it, but you write algorithms like recipes, and you bring out what you want, and you try, and there's a before image, and there's an after image. That was an early image. I thought it was really cool. I think it is still really cool. The trouble is that the scholars don't think it's really cool, and so what you have to do is to invent iterative processes so that scholars and imagers and conservators are all working together, and eventually we got them all to work together. This is Reviel Nets in Stanford writing on the armpit of the goat in Adobe Photoshop. There's the armpit of the goat. This is Archimedes' method of proposition 14. He's writing out what he can see, and there he's just guessing, and this is a question to Abigail Quant asking if the tear here is original. If it's original, of course, he doesn't need to guess a letter at that particular point, but we got better at this. We combined different wave bands of light, so here you take an ultraviolet image here, and you can see an Archimedes diagram rather well, and this book is particularly important for the diagrams because it's the unique source of the Archimedes diagrams that Archimedes drew in the sand in Syracuse in the third century BC, and mathematicians' thinking diagrams. You take a UV image in which you can see the Archimedes quite clearly, and a red channel from a normal image, and you merge them together, and since you've got bright parchment in both images, that comes out bright. Since you've got dark prayer book in both images, that comes out dark, and since you've got dark Archimedes diagram here and light Archimedes diagram there, that's going to come out actually with a red tinge, and then suddenly you could start to make these things out. So here's the diagram. This is the gutter of the book, and you can begin to make out the Archimedes text up there too, running across like that, and this began to work like magic, and so we could start to read bits of the book that Heiberg couldn't read at all, including method proposition 14, and in method proposition 14, what we discovered is that Archimedes is dealing with absolute infinite sets. So we'd always thought that the Greeks could think of potential infinity, think of a number, and you can always think of a higher number than that number, but you don't think of the sum of all numbers, and we found that Archimedes was dealing in the sum of all numbers, which rather rewrites the early history of calculus, and this imaging technique got really, really useful. You zoom in, and you zoom in, and you zoom in, and you zoom in, and you zoom in, and you zoom in. They aren't pretty, these pictures, but they work, and we got better at it, so we got better at taking off the top layer as well, which is particularly important for looking at the diagrams. This is the diagram for the last proposition of Archimedes' Floating Bodies Book 2. Okay, so the forgeries. We were never going to read the forgeries by optical means, and so we had a meeting that began on April Fool's Day, 2004, and went on for three days. We came up with, we asked people from around the world to come with sort of breadboard solutions to their problems. It was sort of a TED talk for Archimedes, and then Action Steps, and three people came up with roughly the same solution. Bob Morton on the left from ConocoPhillips, Jean Hall, who's Professor of Chemistry at Rutgers, and Uwe Bergmann, who's the Staff Scientist at the Stanford Linear Accelerator Centre, and the technique they came up with was X-ray fluorescence imaging. Okay, so when you go to the doctor and you have an X-ray taken, when you go to the doctor and you have an X-ray taken, what you get is a plate back here and a negative image, because some of the electrons have been stopped by your tooth. But what if you can capture the electrons that bounce back, and you can map them? Then you get element-specific information. So this is a bore atom. Here is an X-ray coming in. It knocks off that electron. This electron on shell L makes a quantum leap into this one. Energy has to be shed, and it'll be specific to the atom that you're getting. That's called the K-alpha wavelength. What we wanted to do was to get the elemental information and to map it. We wanted to get the elemental information of iron, because the ink was written in iron. If we can map the iron on the page underneath the gold, then we could, in theory, read it, even though it had been scraped off in the 13th century. The trouble is that to do this, you need a very powerful machine. You actually need the Stanford Linear Accelerator Centre in California, or at least the synchrotron radiation laboratory. The synchrotron radiation laboratory is about the size of a football field. It was built as an atom smasher. Electrons go around one way, positrons around the other. They meet in the middle, and you discover the charm quark, which is what they did there until 1979. Now what they do is they create synchrotron radiation. The electrons go around one way, and as they go around corners, they have to shed electromagnetic radiation, lots of it, X-rays, very powerful X-rays. That's where we were going to put Archimedes. Not valid in Kansas. Use of this device or substance may require, imply, and or endorse the existence of one or more of the following, chemistry, evolution, electromagnetism, gravity, mathematics, thermodynamics, and education. Okay, so you can't read Archimedes in Kansas. Here's a leaf of Archimedes going into the beam. He's on an XY stage like that. The beam comes out here, and the X-rays bounce back, and they get picked up in this detector. Lo and behold, you can do it. You can actually do it. Now it's still very hard to read, and you have to be a nutcase to try and read it, but there are nutcases around because it's unread Archimedes text. This is a nutcase. This is Nigel Wilson of Lincoln College, Oxford. I just spent an awful lot of money trying to get him really good images on a computer, so now I'm introducing him to a computer. His wife can now manage email attachments, which is really wonderful. He does most of his work in the summer months from printouts that we do. This is Reviel Netz, who's a professor of classics at Stanford, and very different, very effective communicator, particularly with the computer. We found out lots of wonderful things, and the press and lots of people came to help us. This is one of the things we found out from the Stomachion, which uniquely survives in this book. Everyone knew that it concerned a square, and the square was made up of 14 bits, but no one knew what Archimedes was doing with this square. Now we think we know what Archimedes was doing with this square, which was trying to work out how many times you can take those bits, recombine them, and still make a perfect square. The answer is rather counterintuitive. It's 17,152. There are 17,152 ways in which you can take those things and still recombine them and make a perfect square, which is more than a game. It makes it the earliest study in the history of combinatorics in Western mathematics. That's kind of cool if you play poker. It's kind of cool if you're interested in modern information and computers as well. Right at the beginning of this talk, and I've got four minutes left, I said that other books were put together with Archimedes, so we got this person called Natalie Chernetska, who was doing her PhD on Greek palimpsests in Cambridge, to look at those. We suddenly found that we actually had a treasure trove of unique ancient checks, and it wasn't really just about Archimedes at all. One of the things we found were speeches by an Attic orator called Hyperides, who was one of the ten canonical orators of antiquity, but who otherwise hadn't made the transition from scroll to codex that happened in the first century to the fourth century AD. Hyperides' text had been found in a tomb in Thebes on papyrus scrolls in 1847 and 1897, but this is the first Hyperides text ever to have been found in a book. Four speeches from the golden age of Athenian democracy, one of which mentions Alexander the Great and Philip of Macedon, because Hyperides was defending himself in court because he'd advocated fighting against Alexander the Great, which is not a good idea. Then we found another text, which turns out to be a third century AD commentary on Aristotle's categories. I got an email from Nigel in 2005 saying that he could see this easily. I still can't see it easily, but there it is in the margin. It mentions the word Aristotle there. So we got a whole library of ancient texts, and then after looking at this book for seven years, we found who did it. Now, there are things called colophons in medieval manuscripts, and they're the first thing you look for, so it's rather unfortunate that it took seven years to find this thing, and we had to go to the Stanford Linear Accelerator to do it, but we did, and it says that this book was made on the 14th of April 1229 by Johannes Myronas, and he did this in Jerusalem, and we know that from the nature of the prayers in the book, which are specific to Jerusalem. The 14th of April 1229 was Easter Saturday in the Greek Orthodox Church. It was eight weeks after Frederick II's stupor Monday had liberated Jerusalem from Muslim control. It's precisely the time when you give a gift to the world, and it is precisely the time when you do it by creating a palimpsest, because you don't have the social infrastructure you need to kill all the goats to find the things to make a book the normal way. The scholars got more and more interested in this, so we now have a worldwide scholarly community of people working on it, mainly through our internet site, and we do a lot of our work on wiki, as far as I know. This is the first classical text ever to have been put together by wiki, and then we get students to work very hard on the transcriptions and put them into TEIP5XML, which is geospatially mapped to the images, and all of this we put up on the internet for absolutely free as a Creative Commons endeavour. This is Doug Emery, our data guy. He's looking at something really, really exciting on the computer. Data. Lots and lots and lots of data. Thank you, Doug. There are two things we know about this, and all this is funded by a private owner whose vision was to make the inaccessible accessible, as a matter of principle, and I think he did a great job. None of your taxpayers' money has been spent on this, and there are two things I want to say. One is, especially if you're in charge of special collections, one is Creative Commons. It's a moral duty. The other is raw data out there, the whole lot, everything, metadata, everything, which is precisely what we're doing with the Walters Art Museum, with passion, brains, and money, precisely what we have done with the Archimedes palimpsest, and what we will do now with our own Islamic manuscripts at the Walters, which are absolutely wonderful. Islamic manuscripts are precious. You can't touch them. You've never seen them, but now you can see them, and you can turn the pages of all of them, and this is your data now. These are people out there on Flickr who are playing with this data in the raw. You can download it at about 1,000 DPI together with technical descriptions, which are really pretty good, and yeah, so here's one. Here's a map. That's the Pyramid of Giza up there from one of our manuscripts that was an Ottoman atlas made in 1700, and I don't believe I've really got 18 minutes left, unfortunately, but so other people are using this data now. It's not my data, and my story is about to end, and the public story is about to begin. I want to show you one public story. His name is Marvin92. Marvin92 is a man on Flickr. He's got a gallery. It's called Muslim. He's got 32 of his pictures that he says are Muslim, and one of the things he's got is a Walters atlas next to the Taj Mahal, and everyone who uses this data, by the way, they always credit the Walters Art Museum. We're interested in branding. I think this is pretty good branding by Marvin92 for the Walters Art Museum. Now, I don't know much about Marvin92 because he's not as interested in metadata as I am in Doug Emery, but these are some of his friends. My mission is to bring art and people together for enjoyment, discovery, and learning, and at this point, I think I've done it. Thank you very much. Thank you.