The Archimedes Codex PDF

9780306815805_FM.
qxd 8/22/07 10:38 AM Page i
THE
A RC H I M E D E S
CODEX

9780306815805_FM.qxd 8/22/07 10:38 AM Page ii
This page intentionally left blank

9780306815805_FM.qxd 8/22/07 10:38 AM Page iii
THE
A RC H I M E D E S
CODEX
How a Medieval Prayer Book
Is Revealing the True Genius
of Antiquity’s Greatest Scientist
REVIEL NETZ
WILLIAM NOEL
Da Capo Press
A Member of the Perseus Books Group
9780306815805_FM.qxd 8/22/07 10:38 AM Page iv
Many of the designations used by manufacturers and sellers to distinguish

their products are claimed as trademarks.Where those designations appear
in this book and Da Capo Press was aware of a trademark claim, the
designations have been printed in initial capital letters.
Copyright © 2007 by Reviel Netz and William Noel
All rights reserved. No part of this publication may be reproduced, stored

in a retrieval system, or transmitted, in any form or by any means, electronic,
mechanical, photocopying, recording, or otherwise, without the prior written
permission of the publisher. Printed in the United States of America.
Production Services by Chris Grison for Outbox Creative Partners

Set in 11.5 point Bembo by Outbox Creative Partners
Cataloging-in-Publication data for this book is available from the Library of Congress.
First Da Capo Press edition 2007

First published in Great Britain in 2007 by Weidenfeld & Nicholson.
ISBN-10 0–306–81580–X
ISBN-13 978–0–306–81580–5
Published by Da Capo Press

A Member of the Perseus Books Group
www.dacapopress.com
Da Capo Press books are available at special discounts for bulk purchases in the United
States by corporations, institutions, and other organizations. For more information,
please contact the Special Markets Department at the Perseus Books Group, 2300
Chestnut Street, Suite 200, Philadelphia, PA 19103, or call (800) 255-1514, or e-mail
[email protected].
1 2 3 4 5 6 7 8 9—10 09 08 07
9780306815805_FM.qxd 8/22/07 10:38 AM Page v
Contents

List of Illustrations vii
Preface 1
1 Archimedes in America 3
2 Archimedes in Syracuse 27
3 The Great Race, Part 1: Before the Palimpsest 65
4 Visual Science 87
5 The Great Race, Part II:The History of the Palimpsest 117
6 Archimedes’ Method, 1999, or The Making of Science 139
7 The Critical Path 159
8 Archimedes’ Method, 2001, or Infinity Unveiled 183
9 The Digital Palimpsest 205
10 The Stomachion, 2003, or Archimedes at Play 233
11 New Light on an Old Subject 261
Epilogue: “The Vast Book of the Universe” 281
Acknowledgements 295
Further Reading 299
Index 305
v
9780306815805_FM.qxd 8/22/07 10:38 AM Page vi

9780306815805_FM.qxd 8/22/07 10:38 AM Page vii
List of Illustrations

1a The Archimedes Palimpsest. (John Dean)
1b The Archimedes Palimpsest open. (John Dean)
2a The Euryalus fortress, Syracuse, Sicily. (John Dean)
2b Hagia Sophia, Constantinople (Istanbul). (John Dean)
3 A medieval scribe. (Walters Art Museum)
4a Constantinople (Istanbul). (John Dean)
4b The Monastery of St. Sabas, Palestine. (John Dean)
4c Leaf of the Palimpsest, Cambridge University Library.
(Cambridge University Library)
5a Abigail Quandt, Senior Conservator of Manuscripts and Rare
Books at the Walters Art Museum. (The Archimedes
Palimpsest Project)
5b Conserving the Palimpsest. (The Archimedes Palimpsest Project)
5c Disbinding the Palimpsest. (The Archimedes Palimpsest Project)
6a Heiberg’s photograph of folio 57r of the Archimedes
Palimpsest, Rochester Institute of Technology, Johns Hopkins
University. (The Archimedes Palimpsest Project)
6b An illustration in H. Omont’s 1929 publication of Greek
manuscripts in the Bibliothèque Nationale. (The Archimedes
Palimpsest Project)
6c Folio 57r of the Archimedes Palimpsest as it is now. RIT and
JHU. (Copyright: Owner of the Archimedes Palimpsest)
6d An X-ray Fluorescence image of folio 57r. SLAC. (Copyright:
Owner of the Archimedes Palimpsest)
7a A detail of the Archimedes Palimpsest before it was disbound.
(The Archimedes Palimpsest Project)
vii
9780306815805_FM.qxd 8/22/07 10:38 AM Page viii
THE ARCHIMEDES CODEX
7b The Palimpsest in normal light. It is very difficult to discern

any undertext. RIT and JHU (Copyright: Owner of the
Archimedes Palimpsest)
7c An early experiment, this is a highly processed image of the
same area. RIT and JHU (Copyright: Owner of the Archimedes
Palimpsest)
8a Setting up a leaf of the Palimpsest to be imaged. (The
Archimedes Palimpsest Project)
8b Roger Easton, Professor of Imaging Science at the Rochester
Institute of Technology. (John Dean)
8c Bill Christens-Barry of Equipoise Imaging LLC. (John Dean)
8d Keith Knox, Chief Scientist at the Boeing Corporation, Maui.
(John Dean)
9 Diagram to Archimedes’ Spiral Lines proposition 21, in ultra-
violet light. RIT and JHU. (Copyright: Owner of the Archimedes
Palimpsest)
10a A detail of the Palimpsest in normal light. RIT and JHU.
(copyright: Owner of the Archimedes Palimpsest)
10b The same page in pseudocolor. RIT and JHU. (Copyright:
Owner of the Archimedes Palimpsest)
11a Reviel Netz has annotated this ultraviolet image in green.
11b Nigel Wilson. (The Archimedes Palimpsest Project)
12a Will Noel and Reviel Netz study a leaf of the Archimedes
Palimpsest. (John Dean)
12b A pseudocolor image of the Hyperides text in the Palimpsest.
RIT and JHU. (Copyright: Owner of the Archimedes Palimpsest)
12c A detail of the third unique text in the manuscript in natural
light. RIT and JHU. (Copyright: Owner of the Archimedes
Palimpsest)
12d A pseudocolor image of the same detail as 12c, with the word
“Aristotle” circled. RIT and JHU. (Copyright: Owner of the
Archimedes Palimpsest)
viii
9780306815805_FM.qxd 8/22/07 10:38 AM Page ix
L I S T O F I L L U S T R AT I O N S
13a Bob Morton, Abigail Quandt, and Gene Hall. (The Archimedes
Palimpsest Project)
13b EDAX scan. (The Archimedes Palimpsest Project)
13c Outside Beamline 6-2 at the Stanford Linear Accelerator
Center. (The Archimedes Palimpsest Project)
14a A leaf of the Archimedes Palimpsest in the beam at SLAC.
14b Uwe Bergmann, Reviel Netz and Will Noel. (The Archimedes
Palimpsest Project)
15a Abigail Quandt inserts a forgery into the beam at SLAC. (The
Archimedes Palimpsest Project)
15b A regular image of one half of a forged page. RIT and JHU.
(Copyright: Owner of the Archimedes Palimpsest)
15c An “iron map” of the same page, taken at SLAC. SLAC.
(Copyright: Owner of the Archimedes Palimpsest)
16a Uwe Bergmann, Abigail Quandt, Keith Knox, Reviel Netz
and Roger Easton at SLAC. (The Archimedes Palimpsest Project)
16b A detail of folio 1v of the Palimpsest, taken in normal light.
RIT and JHU. (Copyright: Owner of the Archimedes Palimpsest)
16c An “iron map” revealing the colophon of the scribe of the
prayer book. SLAC. (Copyright: Owner of the Archimedes
Palimpsest)
Endpapers: front The Archimedes Palimpsest, folios 16v–17r in

natural light; back The Archimedes Palimpsest, folios 16v–17r
processed to reveal the text of Archimedes’ On Floating Bodies.
ix
9780306815805_FM.qxd 8/22/07 10:38 AM Page x

9780306815805_FM.qxd 8/22/07 10:38 AM Page xi
This book is dedicated to Lynn,

to Maya, Darya, and Tamara,
and to Ioannes Myronas
9780306815805_FM.qxd 8/22/07 10:38 AM Page xii

9780306815805_FM.qxd 8/22/07 10:38 AM Page 1
Preface

icetas Choniates, the brother of the Archbishop of Athens, wit-
N nessed the greatest calamity that ever befell the world of
learning. In April 1204, Christian soldiers on a mission to liberate
Jerusalem stopped short of their goal and sacked Constantinople, the
richest city in Europe. Nicetas gave an eyewitness account of the
carnage.The sumptuous treasure of the great church of Hagia Sophia
(Holy Wisdom) was broken into bits and distributed among the sol-
diers. Mules were led to the very sanctuary of the church to carry the
loot away.A harlot, a worker of incantations and poisonings, sat in the
seat of the Patriarch and danced and sang an obscene song.The sol-
diers captured and raped the nuns who were consecrated to God.
“Oh, immortal God,” cried Nicetas, “how great were the afflictions
of the men.”The obscene realities of medieval warfare crashed upon
Constantinople, and the hub of a great empire was shattered.
The looted city had many more books than people. It was the first
time that Constantinople had fallen in the 874 years since
Constantine the Great, Emperor of Rome founded it in AD 330. Its
inhabitants still considered themselves Romans, and the city held the
literary treasures of the ancient world as its inheritance. Among the
treasures were treatises by the greatest mathematician of the ancient
world and one of the greatest thinkers who had ever lived. He
approximated the value of pi, he developed the theory of centers of
gravity, and he made steps toward the development of the calculus
1,800 years before Newton and Leibniz. His name was Archimedes.
Unlike hundreds of thousands of books that were destroyed during
the fall of the city, three books containing Archimedes’ texts survived.
1
9780306815805_FM.qxd 8/22/07 10:38 AM Page 2
Of the three books, the first to disappear was Codex B; it was last
heard of in the Pope’s library in Viterbo, north of Rome, in 1311.
Next to disappear was Codex A; it was last recorded in the library of
an Italian humanist in 1564. It was through copies of these books that
Renaissance masters such as Leonardo da Vinci and Galileo knew the
works of Archimedes. But Leonardo, Galileo, Newton and Leibniz
knew nothing about the third book. It contained two extraordinary
texts by Archimedes that were not in Codices A and B. Next to texts
such as these, Leonardo’s mathematics look like child’s play. Eight
hundred years after the fall of Constantinople, this third book, the
Archimedes Codex, technically known as Codex C, walked on stage.
This is the true and remarkable story of the book and the texts it
contains. It reveals how these texts survived the centuries, how they
were discovered, how they disappeared again, and how they even-
tually found a champion. This is also the story of how patient con-
servation, cutting-edge technology, and dedicated scholarship
brought the erased texts back to light.When they started in 1999, the
members of the team working on the book had little idea of what
they would uncover. By the time they finished, they had discovered
completely new texts from the ancient world and had changed the
history of science.
2
9780306815805_01.qxd 8/22/07 9:03 AM Page 3
1
Archimedes in America

Archimedes for Sale
NEW YORK, NEW YORK

elix de Marez Oyens. What a great name! I don’t know him, but
F I saw him on TV once. His name and demeanor together seemed
tailor-made to suggest a distinguished and international pedigree, a
pedigree that quite naturally produced deep learning, refined taste,
excellent judgement, and total integrity. He clearly had a vast
knowledge of books, and he was extraordinarily good at selling them.
That’s why he was the international director of the Books and
Manuscripts Department for Christie’s auction house in New York.
Thursday, October 29, 1998 was an exceptionally busy day for
Felix. Most of it was devoted to the auction of the final part of the
phenomenal collection of books on science and medicine from the
collection of Haskell F. Norman. Among the 501 lots were some
treasures. In the morning, he sold Marie Curie’s doctoral thesis,
which she had signed for Ernest Rutherford, the man who dis-
covered the nuclear structure of the atom; a first edition of Darwin’s
On the Origin of Species; and a copy of Einstein’s 1905 publication on
Special Relativity. In the afternoon, further extraordinary books were
under the hammer: a copy of the first edition of James Clerk
Maxwell’s Treatise on Electricity and Magnetism, which had been won
as a prize by J. J. Thompson, the man who discovered the electron;
Wilbur Wright’s first published account of the trial flights at Kitty
Hawk, North Carolina; and Nicolai Lobachevskii’s On the Principles of
3
9780306815805_01.qxd 8/22/07 9:03 AM Page 4
Geometry, the first published work on non-Euclidian geometry. Great

books, all of them, and a great day for Felix.
Sandwiched between the morning and afternoon sessions of the
Norman sale was a separate mini-auction devoted to just one book.
It was not a printed book but instead a handwritten one, and it had
not belonged to Norman. In fact, the impressive catalogue that Felix
had prepared for the occasion, with the splendid sale code “Eureka—
9058,” didn’t record to whom it did belong. It didn’t even look like
a great book. It was charred by fire, devoured by mold, and it was
almost illegible.To make matters worse, just the day before, the Greek
Orthodox Patriarchate of Jerusalem had sought a restraining order
against Christie’s in the US District Court, Southern District of New
York, Judge Kimba Wood presiding.The patriarchate argued that the
manuscript had been stolen from one of its libraries. Christie’s suc-
cessfully defended their right to auction the book the next day, but it
was clear that the case of the rightful ownership of the book would
be pursued after the sale. Even with the smart catalogue, the book
itself was going to be a hard sell. Who would want an illegible man-
uscript, in appalling condition, with an ongoing court case attached
to it? Nonetheless, at 2 p.m. on that day Felix was determined to sell
it for an astronomical sum, and he set the reserve price for the man-
uscript at $800,000.
Felix hoped that the book would be worth that much because,
barely visible underneath thirteenth-century Christian prayers, were
the erased words of an ancient legend and a mathematical genius:
Archimedes of Syracuse. Incomplete, damaged, and overwritten as it
was, this book was the earliest Archimedes manuscript in existence. It
was the only one that contained Floating Bodies—perhaps his most
famous treatise—in the original Greek, and the only versions of two
other extraordinary texts—the revolutionary Method and the playful
Stomachion.You could barely read them but, as Felix was very quick
to point out, there was the possibility that the most modern imaging
techniques might help.There were other erased texts in the book too,
but they were almost invisible. No one could read them, and no one
4
9780306815805_01.qxd 8/22/07 9:03 AM Page 5
ARCHIMEDES IN AMERICA
had given them much thought. What mattered was that this book
contained the extremely battered material remains of the mind of a
very great man. If this was a big day for Felix, it was a huge day for
the history of science.
The auction room was in Christie’s offices on the corner of Park
Avenue and 59th Street in New York City. The room was lined by
large contemporary paintings, which provided the splendid visual
setting that the manuscript could not. The manuscript itself was
strapped to a book cradle and secured inside a dramatically lit cage to
the right of the auctioneer’s podium. Reporters arrived as the
minutes before the sale counted down.They stood at the back of the
room with their cameramen who trained their lenses on the book
and tried in vain to make it look as photogenic as one of the
paintings. The rows furthest from the podium were full, but mainly
with academics like the Professor of Mathematics at West Point, Fred
Rickey. He was passionate about the manuscript and deeply inter-
ested in its fate, but could not possibly afford it.The seats at the front,
where one might expect the most seriously interested customers,
were still alarmingly empty. Felix may have been a little worried. But
Felix was lucky. His lucky number was two, because the market value
of an object is always determined by how badly more than one
person wants it.
One of the people who wanted the book badly was Evangelos
Venizelos, the Minister of Culture of Greece. He wanted it for his
country. He had publicly broadcast that it was Greece’s moral, his-
toric, and scientific obligation to acquire the manuscript. At the last
minute, he organized a consortium to buy it and the Greek Consul
General in New York, Mr. Manessis, was sent to the auction. He sat
in the front row, together with an associate, on the left side of the
room.
Just behind Mr. Manessis was a man hoping to disappoint him—
Simon Finch, a high-profile book dealer from London. If your idea
of a bookseller is a bespectacled and tweedy English gentleman, then
think again. Finch is nothing like that. About 45 years old, he looks
5
9780306815805_01.qxd 8/22/07 9:03 AM Page 6
more like a rock star than a book man, and he sells books to rock stars
quite as frequently as to libraries. Finch is the sort of man who can
normally be found at book fairs wearing Vivienne Westwood suits
and sporting designer stubble and disheveled hair. He actually owns a
pair of blue suede shoes. Finch is a romantic and that’s why he is in
the book business. If you don’t think that the combination of great
history and supreme quality that books can provide is romantic, then
he will tell you that that’s because you’ve never turned the pages of a
great book. Five minutes later, you might be a customer. When he
went in to bid for the Palimpsest containing the treatises of
Archimedes, Finch had more than his usual air of mystery. No one
knew for whom he was acting, and no one knew quite how much
that person was prepared to pay for the Archimedes Palimpsest.
At 2 p.m. the duel started, with Christie’s Francis Wahlgren on the
podium. The reserve price of $800,000 was quickly reached, and the
auction headed over the million-dollar milestone. Every time the
Greeks raised their paddle—number 176—high into the air, Finch
would respond with his—number 169.The Greeks were on the tele-
phone, taking instructions, and each time the price went up it would
take them slightly longer to raise their paddle. Each time, Finch
would top the new price. The Consul General answered to the call
of $1,900,000 from the podium. Finch responded quickly to a call for
$2,000,000.Wahlgren looked to the Consul General for a response to
his request for any offers over $2,000,000. The Greeks were on the
phone, desperately raising money. After what seemed like an eternity,
Wahlgren brought down the hammer. “Two million dollars it is,” he
said. “Paddle 169.” The Greeks had failed; the book had gone to
Finch’s unknown client. With the buyer’s premium, the Archimedes
Palimpsest had sold for $2,200,000.
This one book made just under half the amount of the combined
total of all 501 lots of the Norman sale. No wonder its story hit the
presses. The next day, Finch’s role was printed on the front page of
the New York Times to an expectant world. He was the front man, not
for a university, not for a library, but for an individual. But Finch
6
9780306815805_01.qxd 8/22/07 9:03 AM Page 7
would not tell all; he admitted only that the buyer was an American
citizen who was “not Bill Gates.” Felix de Marez Oyens had shown
the book to Finch and the buyer before the sale. Felix had called it
“an old, dirty book” and took it out of a brown-paper bag in his
desk. This was not Felix’s usual sales pitch, but it had worked.
Whoever the individual was, unlike many eminent institutions, he
wanted the book badly enough to take on a national government and
a religious leader. He was also prepared to pay top dollar for the priv-
ilege of owning a moldy, illegible, legally contentious old book. Was
he a nutcase, intent on keeping secret knowledge to himself? Felix
might have been happy, but many were outraged. If the Palimpsest’s
past was obscure, its future appeared dangerously uncertain.
BALTIMORE, MARYLAND
My name is Will Noel, and I am a curator at the Walters Art Museum
in Baltimore, Maryland. The Walters, as it is always known, is a great
American museum modeled on a Renaissance palazzo in Genoa.
Think grand marble staircases and a central courtyard surrounded by
columns and you get the picture. It stands together with a number of
other noble edifices around Mount Vernon Place in downtown
Baltimore. In the center of the square is a tall pillar surmounted by
George Washington. If this were London, the square would be crowded
with tourists, street musicians, and students. But, located in inner-city
Baltimore as it is, Mount Vernon Place is normally quite empty of
people which lends it a sense of moody suspended animation that the
passing traffic doesn’t quite relieve. Inside the building is the superb
collection of two individuals—father and son, William and Henry
Walters. In a great act of civic philanthropy the collection was given by
Henry to the City of Baltimore in 1934. Few people visit it, but the
museum houses fifty-five centuries of art and in many areas its holdings
are truly fabulous. Thomas Hoving, the director of the Metropolitan
Museum of Art in New York, said of it: “Piece for piece it is the greatest
art museum in the United States.” It is my job to research into, teach
from, and exhibit the Walters collection of manuscripts and rare books.
7
9780306815805_01.qxd 8/22/07 9:03 AM Page 8
They are the stuff of legend and the fabric of history. They range in
date from 300 BC to 1815, from an Egyptian Book of the Dead to
Napoleon’s memoirs. Most of them are medieval and sumptuously
illuminated with images. Among the Walters’ other holdings are
massive Roman sarcophagi and paintings by Hugo van der Goes,
Raphael, El Greco,Tiepolo, and Manet.
Gary Vikan, the director of the Walters, is my boss. Several weeks
before the sale I had told Gary about the Archimedes Palimpsest. It is
part of my job to follow the New York sales and Gary has a particular
interest in medieval manuscripts. It struck a chord. When I walked
into work the day after the sale, he hailed me down the grand
stairway of the house that was once the Walters’ home and, bran-
dishing the New York Times, he said, “Will! Why don’t you find out
who bought the Archimedes Palimpsest and see if you can get it for
exhibition?”
I thought it was a bad idea. After all, the Walters is an art museum.
It is concerned with what things look like.You cannot even see what
is interesting about the Archimedes Palimpsest. I sent Gary a memo
asking him if he really wanted me to do this. A couple of days later I
got my memo back with a characteristic directorial scrawl: “NOT
WORTH MUCH WORK.” It was clear to me that at least I had to try. I
didn’t have any more leads than anybody else did. Simon Finch was
the only name I had and so I asked Kathleen Stacey, the head librarian
at the Walters, to find his email address on the web. This she did, and
I sent Finch the following email:
Dear Mr. Finch,

I am the manuscripts curator at the Walters Art Museum,
Baltimore. The Walters has 850 medieval manuscripts, 1,300
incunables, and another 1,500 books printed after 1500. Most
of these books are illustrated and Henry Walters collected
them between about 1895 and 1928.
We have an active acquisitions program, although our
funding is limited. We have, for example, recently purchased a
8
9780306815805_01.qxd 8/22/07 9:03 AM Page 9
deluxe sixteenth-century Ethiopian manuscript from Sam

Fogg. In general terms, therefore, I would be most interested
in receiving your catalogues and would be grateful if you
would add me to your mailing list.
However, I do have a more specific reason for writing. The
Director of the Walters, Dr. Gary Vikan, is a specialist in Greek
material, and was fascinated (as am I) by the Archimedes
Palimpsest. Dr.Vikan wondered if there was any possibility of
displaying the manuscript at the Walters for a short period of
time. I do not know whether the purchaser of the volume
would be at all interested in this idea. But if you think he
might be, I would be most grateful to you if you could pass
on the suggestion.
The Walters does have an active exhibition program. We are
currently putting on a show of works from the Vatican, Monet
came earlier in the year, and the Arts of Georgia are coming
in 1999. If the owner of the Palimpsest is interested in putting
the manuscript on view, he might consider that the Walters is
an appropriate place.
Please excuse this cold call. It is just a thought, but from our
point of view an exciting one, given the extraordinary cultural
importance of the codex. Whatever you think of this I would,
as I say, look forward to hearing from you and receiving your
catalogues. With many thanks for your time,
William Noel
Curator of Manuscripts and Rare Books
I moved my cursor to the top left of the screen: send. By the next
minute I had dismissed it from my mind. Frankly, the chances of any-
thing resulting from this were remote in the extreme, I didn’t even
much want anything to happen, and I had labels to write for an exhi-
bition of Dutch illuminated manuscripts. Still, I had done my job.
Emails are short on ritual. There is no walk to the mailbox, no
looking at the stamp, no slicing the envelope, no guessing the
9
9780306815805_01.qxd 8/22/07 9:03 AM Page 10
handwriting. They just pop up unbidden on your computer screen

while you are engrossed in your daily business. Some of them, like
little electronic terrorists, can blow your mind and change your
life. Three days after my email to Finch this happened to me. I was
happily writing an exhibition label for a book illuminated by the
Masters of the Delft Grisailles when my computer went PING.
You’ve got mail. Sam Fogg. Left Click:
Dear Will,
I am writing with reference to your letter to Simon Finch
on the subject of the Palimpsest. I think the buyer of the
Palimpsest is very sympathetic to the idea of sending the
Archimedes to the Walters. I have already suggested to him
that we visit the Museum in January. Perhaps we could discuss
this and the Archimedes on the telephone soon.
Best wishes,
Sam Fogg
I sat motionless in my chair—eyes shut, hands cradled behind my

head, rocking gently, my stomach slowly turning to wax. Then I
picked up the phone and dialed a number. It was a number I knew
almost by heart. Although I hadn’t been expecting to hear from him,
I knew Sam Fogg well. As an unemployed postgraduate eking out an
impoverished existence in Camden Town, London, I had once done
some research on his behalf (for which you should read “he
employed me”), and when I became a curator in America, I was in a
position to acquire the odd manuscript from him. Sam is one of the
art world’s most colorful characters. Famous for having sold ceiling
panels of Henry III of England’s painted bedroom at Westminster to
the British Museum; for having sold a leaf of Jan van Eyck’s miniature
masterpiece, the Turin Milan Hours, to the J. Paul Getty Museum; and
for having bought a Rubens for £40,000,000. Sam is savvy, suc-
cessful, and very smart. I don’t remember the conversation well, but
10
9780306815805_01.qxd 8/22/07 9:03 AM Page 11
Sam must have told me that Simon Finch had called him because I
had mentioned Sam to Finch in my email.
I arranged a flight to London. Before I caught the plane I discussed
strategy with Gary. He thought that Simon Finch and Sam Fogg
might actually be the same person and that I was being given the
runaround. I didn’t think so. Two days later I could prove it: I had
lunch with both Simon Finch and Sam Fogg in Brown’s restaurant
on Maddox Street in London. It was only at this lunch that I dis-
covered who the owner of the Palimpsest actually was. He had in fact
been present at the auction, unnoticed by the competition, and
unrecognized by the press. He still likes to tell the story. Moreover, he
had known exactly the liability that he was trying to buy and had
bought it on the assumption that he would deposit it somewhere for
conservation and scholarly study. His anonymity was important to
him and hereafter he became known in any written correspondence
as Mr. B. We agreed that Sam and Mr. B would visit the Walters in
January.
This was just great. The trouble was I didn’t really know anything
about Archimedes or his book. My brother Rob had written a story
about a dog-eared palimpsest once and so I had the vague, romantic
notion that palimpsests could harbor secret knowledge that you could
only understand if you were really smart. But that was all I could
remember. I needed a few facts and a map of the Mediterranean. It
was November. I had two months to learn enough not to look like a
total idiot.
At about eleven o’clock on the morning of Tuesday, January 19,
1999 Mr. B and Sam arrived at the museum. I met them at the
entrance. Sam was a laugh a minute, as he always is; Mr. B was com-
pletely silent. Nervous to begin with, I took them up to the manu-
script room, a climate-controlled vault that serves as my office as well
as the repository of hundreds of medieval treasures. I entertained Sam
and Mr. B for an hour or so, before taking them to have lunch with
Gary. I couldn’t get a measure of the man. All I knew was that he was
11
9780306815805_01.qxd 8/22/07 9:03 AM Page 12
retiring, rich—richer than Croesus—and that he liked food. I knew

he liked books too, but I wasn’t learning anything more.
I had arranged to have lunch at a Baltimore institution, Marconi’s,
which is about four blocks from the Walters on West Saratoga Street.
A slightly down-at-heel survivor of Baltimore’s elegant past, it serves
wonderful food in a beautifully proportioned, white wooden-paneled
room. On the way, Sam walked in front with Gary, and I walked
behind with Mr. B—a nervous puppy trying to come to grips with
the biggest fish of my little career. I remember congratulating him on
his exciting new acquisition and saying that it was extremely generous
of him to even consider putting his great new treasure on deposit at
the Walters. His reaction to this was my first lesson in the mind of Mr.
B. He said that he had already left it on deposit with me. I did not
understand. I asked him to say it again. He said that he had left it in a
bag on my desk. I swallowed hard. As a museum registrar will quickly
point out, this does not conform to standard museum protocols for
the transportation and documentation of objects worth several mil-
lions of dollars. I went with the flow. Great, I told him, and I reassured
him that I had locked the door of my office on the way out.
Lunch was cordial, but a little odd to me. As I have said, Mr. B
enjoys his food, and he also likes to take his time. I wanted to go back
to the museum and look at the manuscript. I was quite happy with
one course; Mr. B wanted his chocolate sundae. I could barely sit in
my seat, and I couldn’t get Mr. B out of his. Eventually lunch was
over and the check was requested. Gary tried to pay with a credit
card. This is Baltimore. Marconi’s doesn’t take American Express. I
paid cash. Back we walked to the museum. I made my excuses on the
way and ducked away to buy a pack of cigarettes. I hadn’t had one for
three hours, and I smoked two in five minutes—pacing nervously. I
caught up with them in time to turn the key to the manuscript room
where my desk sits.
A lightweight blue bag was on my desk. Stamped in white on its
side was a pair of scissors and, underneath them, the words GIANNI
CAMPAGNA, MILANO. I unzipped the bag and pulled out a brown box.
12
9780306815805_01.qxd 8/22/07 9:03 AM Page 13
On the spine, in gold letters, was written: “THE ARCHIMEDES

PALIMPSEST.” I called my colleague Abigail Quandt, who is the
Conservator of Manuscripts at the Walters.We opened the box. Inside
was a small, thick book. The cover was made of battered leather and
was badly stained. On the upper cover there was a flash of red paint
and an odd silver-looking stud. Abigail placed the book between two
velvet-covered blocks of wood on the table.The blocks prevented the
manuscript from opening too far and placing unnecessary strain on
the binding and the pages. She opened the book just far enough so
that we could see inside. She kept it open by gently draping book
“snakes” over the edges of the pages. (These “snakes” are actually
curtain weights that you can only get at John Lewis, a department
store on Oxford Street in London; they work really well for keeping
your place in a medieval book.) Mr. B, Gary, and I all peered over her
shoulder. At first I saw nothing. Only slowly did my eyes adjust. And
then the awesome thought dawned on me. I was looking at the
unique key to the mind of a genius who had died 2,200 years earlier.
I could barely see it to read it, and I would not have understood it if
I could, but there it was nonetheless.
After a few minutes, I grasped the fact that the time for gawking
was over. Proper looking would have to come later.The museum reg-
istrar, Joan Elisabeth Reid, prepared a receipt for the book, which I
handed over to Mr. B. I took his email address. (Email was, and still
remains, his preferred form of communication.) We said our good-
byes at the front entrance of the museum on North Charles Street,
and then I dashed back upstairs to the manuscript room where Sam
was still waiting. I gave him an enormous and excited hug, forgetting
for a moment that we were on live video feed and that the Walters’
security staff was monitoring our every move.
Two days later I received a letter from Mr. B that contained a
check made out to the Walters. It was big enough to get the insti-
tution’s attention and me a pay raise.
*
13
9780306815805_01.qxd 8/22/07 9:03 AM Page 14
Help for Archimedes

Mr. B told me that he had bought an ugly book. Since he’d paid over
two million dollars for it, I took this with a pinch of salt. But no. Now
that I had it in my hands, I could see that he had played it straight this
time. It was ugly. It was small—about the size of a standard bag of
Domino sugar. When I opened it, I saw that the pages were mottled
brown in color. Matching tide lines caused by water faced each other
across page openings. The pages tended to be brighter in the middle
than around the edges where they were more deeply stained. In fact
right on their edges the pages were black, as if they had been in a fire.
Overlaid upon the brown of the pages was stitched a grid pattern of
slightly darker-brown Greek letters, which were all jumbled up. The
monotony of the pages was only slightly relieved by the speckled red
of the odd capital letter and occasionally by purple stains of mold.
When I turned pages, I could, just once in a while, make out the
circles and straight lines of things that looked like diagrams that
would, most inconveniently, disappear into the spine of the book
from the inner margins. Compared to other manuscripts I had
handled, the pages didn’t flex very easily, and they were contorted.
Sometimes, as I was turning a page, it would suddenly “pop” into a
slightly different shape. Once in a while a whole page would just
come out of the book in my hands. As I went through the book from
start to finish, four pages stood out as having a certain charm because
they had paintings on them, but overall it was a deflating experience.
And then, toward the end, the pages looked so fragile and so moldy
that I shut the book in alarm. This book, for which Mr. B had paid
so much, was on its last legs.
That is not a very helpful description, so let me describe the book
etymologically. It is a manuscript book or, more technically, a manu-
script codex. Derived from the Latin words manu (by hand) and
scriptus (written), a manuscript is entirely written by hand. It is fun-
damentally different from a printed book in that it is not one of a
large number of books printed in an edition. It is unique. Other man-
14
9780306815805_01.qxd 8/22/07 9:03 AM Page 15
uscripts might contain some of the texts in it. All I knew for sure at
this point was that no other manuscripts contained Archimedes’
Method, Stomachion, or Floating Bodies in Greek. Secondly, this manu-
script is a palimpsest. Derived from the Greek words palin (again) and
psan (to rub), this means that the parchment used to make it has been
scraped more than once. As we will see, to make parchment you need
to scrape the skins of animals. If you want to reuse parchment that has
already been used to make a book, you need to scrape the skin again
to get rid of the old text before you write over it. This palimpsest
manuscript consisted of 174 folios. Derived from the Latin folium
(leaf), a folio has a front and back—a recto and a verso—that are
equivalent to modern pages. The folios were numbered 1 through
177 but, mysteriously, three numbers were missing. I hoped Mr. B
knew that he was missing some folios.
The manuscript is now called the Archimedes Palimpsest, but this
is a bit confusing. Make no mistake: the manuscript is a prayer book.
It looks like a prayer book, it feels like a prayer book, it even smells
like a prayer book, and it is prayers that you see on its folios. It is only
called the Archimedes Palimpsest because folios taken from an earlier
manuscript containing treatises by Archimedes were used to make it.
But remember the Archimedes text had been scraped off. Note, too,
that the scribes of the prayer book used the folios taken from several
other earlier manuscripts as well as the Archimedes manuscript. At
the time of the sale, nobody had a clue what was on these folios.
They didn’t look like folios from the Archimedes manuscript, and
they didn’t look as if they were all from the same manuscript. For
example, while the Archimedes text was laid out in two columns, the
texts on other palimpsest folios were laid out in one column; others
had a different number of lines per folio; and the handwriting on the
other folios, when it wasn’t invisible, was sometimes very different.
Mr. B had bought several different books in one. Basically, I con-
cluded that the Archimedes Palimpsest was only called the
Archimedes Palimpsest because no one could identify the other texts
in the manuscript and because the Archimedes texts were considered
15
9780306815805_01.qxd 8/22/07 9:03 AM Page 16
so much more important than the prayer book that was on top of
them.
But how important, really, was this “Archimedes Palimpsest?” I
began to ask around and Mr. B’s book got decidedly mixed reviews.
Even though it had commanded $2.2 million at auction, the truth
was that only three parties had put up a fight for it: the patriarchate,
the Greek government, and Mr. B. None of them knew all that much
about Archimedes. How come, I asked? Was there no academic insti-
tution sufficiently interested in it to enter the fray? I found out that
many well-informed scholars were skeptical that we could learn
much more from the book. Everybody kept mentioning that
someone named Heiberg had discovered the manuscript and read it
in 1906. And Heiberg, apparently, was something of a god in classical
studies.They said that it was unlikely that he would have missed any-
thing important. Mr. B, they told me, had bought a relic, not a book
that would reward much further research.
Still, Mr. B had entrusted his relic to me, and I had no choice but
to take his new possession as seriously as he did. His book clearly
needed three things: first, since it was literally falling apart, it needed
conservation; second, since no one could see the Archimedes text in
it properly, it needed advanced imaging; third, if by any chance
Heiberg had missed a few lines, then scholars needed to read it. I
knew that Mr. B would require the best. This was good, because his
book was such a wreck that it needed the best—the best conser-
vators, the most advanced imaging, and the most highly qualified
scholars. I was none of these things, and I wondered whether I was
the right person to be looking after Mr. B’s book. My expertise is in
Latin manuscripts, not Greek ones; religious books, not mathematical
ones; beautiful books, not ugly ones; and certainly legible books, for
goodness sake, not invisible ones.
That Mr. B chose me, of all people, to look after his book seemed
more than a little absurd, I thought. But Mr. B knew my limitations.
My job, as he saw much more clearly than I did at the time, was not
16
9780306815805_01.qxd 8/22/07 9:03 AM Page 17
to do the work, but to get the right people to do it. But how was I
going to do that?
THE PROJECT MANAGER

On Friday, July 16, 1999, the Washington Post published an article on
the Palimpsest. Abigail and I received many emails in reaction to it.
Some are among the most zany I have ever received. (To the unac-
knowledged grandson of Rasputin, I can only say that I have not yet
found any corroboration of your pedigree in the Archimedes
Palimpsest.) Let’s concentrate on the ones we found helpful. Here’s
the best of them.
Dear Drs. Noel and Quandt:

I read with interest the article in the Washington Post.
Congratulations. It certainly puts our work in perspective. We
in the intelligence community have equipment that may be
able to help. We also have a wide range of contacts in the
imaging community that could prove useful to you. If you
would like to discuss this further, please do not hesitate to get
in touch. Whatever the case, it sounds like a fascinating
project. Good luck in your endeavors.
Yours sincerely,
Michael B. Toth
National Policy Director
National Reconnaissance Office
The National Reconnaissance Office (NRO) is not a secret any

more, but it was for a long time. Mr. B told me that the only reason
it was forced to become public is that people could not understand
why hundreds of cars were disappearing into a small office building.
The answer was that most of it was underground and that it was the
unacknowledged nerve center of the US reconnaissance satellite
program. Now, however, you can find details about NRO on the web.
17
9780306815805_01.qxd 8/22/07 9:03 AM Page 18
Working with the CIA and the Department of Defense, it can warn
of potential trouble spots around the world, help plan military oper-
ations, and monitor the environment. Its mission is to develop and
operate unique and innovative space-reconnaissance systems and
conduct intelligence-related activities essential for U.S. national
security. As an avid John le Carré reader, I have always been
enthralled with the world of espionage.
I phoned Mr. Toth. I was tempted to say, if he hung on just a
moment, I would take the book up to the roof of the museum and if
he could just fly a satellite over it, we would all be finished in a few
minutes. More soberly, I invited him up from Washington to
Baltimore. I was still hoping that he would have a gadget, maybe in
his back pocket, maybe disguised as a watch, that could help me with
my problem. Much to my disappointment, it soon became clear that
no government agency could help us with the imaging of the
Palimpsest. Since it was private property, the tax dollars of the
American public could not be spent on it. Mike said that he would
nonetheless be happy to help us as a volunteer. Deprived of his toys,
I was not sure how he could, but it seemed unwise to annoy this man,
and he seemed pretty certain that he would be useful.
Mike, it turned out, was an expert at managing highly technical
systems, including imaging systems and particularly in assessing some-
thing called “program risk.” This was an amazing stroke of fortune.
Apparently, I had found someone who was professionally trained to
tell me exactly the magnitude of my trouble. But, more importantly,
he was willing to help me. I am a scholar who specializes in illumi-
nated liturgical manuscripts from Canterbury, England from the early
eleventh century. I have a few skills. I can, for example, recite the
Book of Psalms backward and the kings and queens of England
forward from Hengist through Henry VIII. But these skills are not
particularly well suited to running an effective integrated project at a
reasonable cost, to the correct level of performance, and on a prac-
tical schedule in order to produce value for the owner and an
Archimedes text for the world. I needed someone like Mike, a tech-
18
9780306815805_01.qxd 8/22/07 9:03 AM Page 19
nical consultant, and preferably one who, I liked to believe, had

pressed the “go” button to launch a space shuttle.
Mike, like so many people who would ultimately help with
Archimedes, was a volunteer. He didn’t want money, and he didn’t
want his government service celebrated by the press. In fact, his work
on Archimedes was all done through his father’s company, R. B.Toth
Associates, and that was how we introduced him to people. With
Mike on board everybody else seemed to get a cover, too. Mr. B
became the “source selection authority” (that is, he decided every-
thing); Abigail became the “critical path” (that is, everything
depended on conservation); the scholars became the “end-users”
(that is, they defined what was best); and the imagers became the
“value added” (that is, they made the difference). And me? I was
given the very grand title of “project director.”
THE SOURCE SELECTION AUTHORITY

I know the owner of the Archimedes Palimpsest. I know him very
well. If you don’t know him by now, you don’t need to know him.
To the press, I say that he’s of more use to you as an enigma; to the
curious I say mind your own business. To those who do know him,
he is a loyal, generous, thoughtful, and enlightened man. His email
style is a bit short, but you get used to it.
When the Archimedes Palimpsest was sold, some scholars were
outraged that the book had returned to a private collection. But if
Archimedes had meant enough to the public, then public institutions
would have bought it. Archimedes did not. Public institutions were
offered the book at a lower price than it actually fetched at auction,
and they turned it down. If you think this is a shame, then it is a
shame that we all share.We live in a world where value translates into
cash. If you care about what happens to world heritage, get political
about it, and be prepared to pay for it.
The practical reasons why it might have been a “bad thing” if the
manuscript was maintained in a private collection are that the book
might have been poorly handled and the right scholars might not
19
9780306815805_01.qxd 8/22/07 9:03 AM Page 20
have gotten to see it. Someone could have just tossed it into his or
her attic. As we will learn, given the state in which the book came
out of its last private collection, these were valid concerns. I hope by
the end of this book, if not by the end of this chapter, to demonstrate
that this manuscript has been cared for extremely well and that the
right people have cared for it.Another reason why it might have been
a “bad thing” is that its future was uncertain. This still remains true.
When the work is done the manuscript will go back to its owner, and
I do not know what will happen to it then. But the best predictor of
future behavior is past behavior and over the last eight years the
owner has behaved responsibly, thoughtfully, and generously.
What do I mean by this? Well, Mr. B is extremely interested in the
Archimedes Palimpsest and greatly concerned with the project and
its goals. He is knowledgeable about books, he cares about them, and
he has a superb library. He makes all the important decisions
regarding the book, but he does so after carefully listening to us and
reading proposals that I have forwarded to him. And what’s more, he
pays for all the work that needs to be done.The project has never suf-
fered from a lack of money. Manuscript scholars, classicists, and math-
ematicians owe a great deal to the owner of the Archimedes
Palimpsest.
THE CRITICAL PATH

The first task was to secure the well-being of the manuscript.
Whatever else happened, the manuscript had to stay safe. I didn’t have
to do anything and have done nothing about this since the book
arrived at the Walters. Abigail Quandt has done it all. Abigail has an
international reputation for the conservation of medieval manu-
scripts. She has worked on some of the world’s most famous manu-
scripts, including the Dead Sea Scrolls and one of the greatest
masterpieces of the Middle Ages—the Book of Hours of Jeanne
D’Evreux. Abigail received her training in Dublin with Tony Cains,
Head of Conservation at Trinity College Dublin and in England with
20
9780306815805_01.qxd 8/22/07 9:03 AM Page 21
Roger Powell who rebound the Book of Kells. She has been at the
Walters much longer than I have—since 1984.
Abigail was integral to the planning of Archimedes’ future. In any
of the decisions concerning the well-being of the manuscript—and
there would be many—Abigail’s voice was the strongest. I didn’t just
have a great colleague; I was convinced that Archimedes was in the
safest possible hands—hers. I could rest assured that I wouldn’t make
the situation worse for Archimedes, and I was able to concentrate on
other things.
THE END–USERS
I received many offers, by a variety of enthusiasts, to help with the
decipherment of the Palimpsest. Some of these offers were rather
forceful (this is an understatement). I tried not to be offensive while
I worked out a strategy. The manuscript was so fragile that I could
not let just anyone have a crack at it. I needed to get the two or three
people who could best edit the texts so that they could be published.
The question was which two or three?
Gary Vikan immediately advised me to get in touch with Nigel
Wilson of Lincoln College, Oxford. He was an obvious choice for
two reasons. The first was that he knew the book better than anyone
else having just contributed a great deal to the catalogue Christie’s
produced for the auction. Christie’s asked him to catalogue it for the
same reason that I wanted him to work on it: he is without peer in
scholarship on the transmission of classical texts from antiquity
through the Middle Ages and his paleographical (script-deciphering)
and philological (text-analyzing) skills are legendary. I wrote to him
on Monday, January 15, 1999 and explained that if we were to do
justice to the manuscript, we needed a distinguished scholar who
knew about the subject to be our advisor. If he was willing, he was
in this respect uniquely qualified to help us. Ever since then Nigel has
been helping us. He has become far more than an independent
advisor.
21
9780306815805_01.qxd 8/22/07 9:03 AM Page 22
Next I phoned my very discreet friend Patrick Zutshi, Keeper of

Manuscripts and University Archives at Cambridge University
Library and spoke to him about my problem. He advised me to get
in touch with Patricia Easterling, who was Regius Professor of Greek
at Cambridge University. This was pretty grand for me, but not, I
thought, for Archimedes. So I phoned her and said, “Can you please
tell me who is the best person to study the Archimedes Palimpsest?”
I met her in early March 1999 in the tea room of the University
Library, and she suggested I get in touch with Reviel Netz who was
translating Archimedes into English for Cambridge University Press.
Netz, she said, would be more interested in the book than most.
While many were skeptical about the discoveries that could be made
from the text, all agreed that the manuscript was important for its dia-
grams. Netz had a distinct interest in the diagrams. (More about this
later.) Netz was at the Massachusetts Institute of Technology. I wrote
him an email, and then we talked in detail on the phone. Pat
Easterling was right:
“Yes. I need to see the diagrams, particularly for Sphere and
Cylinder.” I think were the first words out of his mouth. I am still not
sure because he has a rather thick Israeli accent. That’s a bit pushy, I
thought, and I tried to put the brakes on. So I spoke slowly and
painted a broad picture of what our work might be and how he
might fit into it. And, if he was interested, perhaps, just perhaps, he
should come to Baltimore in the fullness of time.
When I met him at the airport terminal a couple of days later, I
understood immediately that his pushiness was induced by his fear
and his excitement. I did my best to calm his fear: yes, the Walters was
a center of excellence; no, the Palimpsest wasn’t here for a passing
visit; yes, he could look at it—tomorrow, even—but he had to be very
careful; no, I didn’t plan to show it to just anyone. By the next day, I
understood where he was coming from. He knew better than anyone
that the box containing the Palimpsest was a time machine to
Archimedes of Syracuse in the third century BC. He explained the
importance of the diagrams to me as no one else ever has. Having
22
9780306815805_01.qxd 8/22/07 9:03 AM Page 23
convinced himself that I understood the grave responsibility that was

on my shoulders, he looked at me with sympathy. He knew that I was
going to do my best for the book, even though I did not understand
it and it would be a long, demanding task that would take me away
from my own research for years. Good. He was on my side, if only
because I was on Archimedes’ side. Now, eight years later, we find
ourselves writing this book side by side and chapter by chapter.
Unlike me, Reviel has never thought of the Palimpsest as ugly. He
doesn’t care about its looks; he simply regards the Palimpsest with
awe. His initial reaction was to feel daunted by the task ahead of him.
His doubts were gone, though, when he heard that he would be
working side-by-side with a colleague of the stature of Nigel Wilson.
Reviel had another suggestion, too. He thought it was important to
get someone to work just on those folios of the Palimpsest that con-
tained texts not by Archimedes. He wanted to know who kept
Archimedes company in this prayer book. I thought this was a good
idea. Even if the text of Archimedes was well understood, there was
the chance that we could find out more about the other palimpsested
texts.
The name Reviel suggested was Natalie Tchernetska, a Latvian
who was doing her PhD on Greek palimpsests at Trinity College,
Cambridge. Pat Easterling was her supervisor. Small world. I met her
in Pat’s office at Newnham College in the summer of 1999. She was
helpful in assessing the images, and we will have reason to look at
some of her work later. This was the core of the academic team that
was going to paint an entirely new picture of the greatest mathe-
matician of antiquity and to reveal the world’s greatest palimpsest.
THE VALUE ADDED

One day in August 1999 I sat down beside Abigail in my office and
faced Mike Toth. We had to find the right people to image the
Palimpsest. This was intimidating. I felt overwhelmed with the
thought of how much work I would have to do, but I didn’t even
know precisely what this work was. Mike thought that we should
23
9780306815805_01.qxd 8/22/07 9:03 AM Page 24
arrange a competition for people to image the Palimpsest. I thought

this was a bad idea; it seemed like a lot of work. Mike gently insisted.
It would greatly increase the number of imaging procedures that we
could perform on the book and it would give the participants the
incentive to reduce costs and increase performance in the hope that
they would be rewarded with the commission for imaging the entire
volume. This was merely sensible, he said. It sounded like rocket
science to me.Then he told me for the first time about a Request for
Proposals. An RFP is quite standard to me now. It is a document in
which you outline the problem and ask for a solution.
Abigail wrote the RFP. It is one of a number of thorough and bril-
liant documents that she has written throughout the history of the
project. It started with a goal: to digitally retrieve and preserve for
posterity all the writings in the 174 folios of the Archimedes
Palimpsest. It mentioned all the constraints: because the manuscript
was very fragile, all the handling of the manuscript would be under-
taken by Abigail and personnel that she designated. It outlined the
phases of work: after the competitive phase, the selected contractor
would image the entire manuscript in a disbound state. The whole
proposal ran six pages. In response to the RFP, we received six pro-
posals. Of the six, we submitted three to Mr. B and of the three Mr.
B selected two for the competition.
One team consisted of Roger Easton, a faculty member at the
Chester F. Carlson Center for Imaging Science at the Rochester
Institute of Technology, and Keith Knox, who was the Principal
Scientist at the Xerox Digital Imaging Technology Center also in
Rochester. (He now works for Boeing in Hawaii.) Keith, together
with Brian J. Thompson, had achieved fame years earlier by devel-
oping and patenting a method—the Knox-Thompson Algorithm—
that recovers images from telescopic photographs that have been
degraded by the atmosphere. More recently, Roger and Keith had
formed a team together with the late Robert H. Johnston to image
degraded texts including a palimpsest in Princeton University Library
and several of the Dead Sea Scroll fragments.Their work had already
24
9780306815805_01.qxd 8/22/07 9:03 AM Page 25
been celebrated on the BBC and on American TV. They had already
done some work on the Palimpsest, because Keith’s sister-in-law
knew Hope Mayo, who had worked with Nigel to prepare the cata-
logue for the Christie’s sale. Some of their images are actually in the
catalogue. Roger, Keith, and Bob Johnston were a known quantity
and a safe bet.
The other team was from Johns Hopkins University and was, in
effect, one man, William A. Christens-Barry. Bill is not an imaging
scientist, still less a photographer; he is a physicist. At the time we met
him, he was working at the Applied Physics Laboratory of Johns
Hopkins University. APL employs nearly three thousand engineers,
information technologists, and scientists. It works primarily on devel-
opment projects funded by federal agencies. Foremost among these
are the US Navy and NASA. Scientists at APL participate in the
entire range of data collection and analysis activities of interest to its
sponsors, including data from air-, ocean-, and space-borne recon-
naissance and imaging platforms. Work on non-defense, non-space
projects constitutes a secondary activity of the laboratory. Most of
Bill’s research pertained to problems in biological and medical
science, particularly in relation to cancer. Impressive place; impressive
guy. His proposal was full of ideas that no one else had even con-
sidered.
THE PROJECT DIRECTOR

All these people had well-defined roles, but I wondered about my
role. I was to be Archimedes’ factotum. I did the talking, and I did the
arranging. As Mike put it, I kept an awful lot of plates spinning on
their poles. And I was going to have to do it for a long time. By the
end of the year, I had talked to the right people and had arranged a
lot. I had a plan in place and the key players were on board. I could
say what I was doing to anybody who called. I just couldn’t really say
why I was doing it. If you had called me up and asked me why any
of us wanted to do this work, I would immediately have referred you
to Reviel Netz.
25
9780306815805_01.qxd 8/22/07 9:03 AM Page 26

9780306815805_02.qxd 8/22/07 9:05 AM Page 27
2
Archimedes in Syracuse

rchimedes is the most important scientist who ever lived. This
A conclusion can be reached as follows.The British philosopher A.
N. Whitehead once famously remarked: “The safest general charac-
terization of the European philosophical tradition is that it consists of
a series of footnotes to Plato.” This judgment may sound outrageous,
but in fact it is quite sober minded. Plato’s immediate followers, such
as Aristotle, tried above all to refute or to refine Plato’s arguments.
Later philosophers debated whether it was best to follow Plato or
Aristotle. And so, in a real sense, all later Western philosophy is but a
footnote to Plato.
The safest general characterization of the European scientific tra-
dition is that it consists of a series of footnotes to Archimedes. By
which I mean, roughly the same kind of genealogy that Whitehead
meant for Plato applies to Archimedes. As an example, we need only
to look at one of the most influential books of modern science,
Galileo’s Discourses Concerning Two New Sciences. This book was pub-
lished in 1638, by which time Archimedes had been dead for exactly
1,850 years—a very long time indeed.Yet throughout it, Galileo is in
debt to Archimedes. Essentially, Galileo advances the two sciences of
statics (how objects behave in rest) and dynamics (how objects behave
in motion). For statics, Galileo’s principal tools are centers of gravity and
the law of the balance. Galileo borrows both of these concepts—
explicitly, always expressing his admiration—from Archimedes. For
dynamics, Galileo’s principal tools are the approximation of curves and
the proportions of times and motions. Both of which, once again, derive
27
9780306815805_02.qxd 8/22/07 9:05 AM Page 28
directly from Archimedes. No other authority is as frequently quoted

or quoted with equal reverence. Galileo essentially started out from
where Archimedes left off, proceeding in the same direction as
defined by his Greek predecessor.This is true not only of Galileo but
also of the other great figures of the so-called “scientific revolution,”
such as Leibniz, Huygens, Fermat, Descartes, and Newton. All of
them were Archimedes’ children. With Newton, the science of the
scientific revolution reached its perfection in a perfectly Archimedean
form. Based on pure, elegant first principles and applying pure
geometry, Newton deduced the rules governing the universe. All of
later science is a consequence of the desire to generalize Newtonian,
that is, Archimedean methods.
The two principles that the authors of modern science learned
from Archimedes are:
• The mathematics of infinity
• The application of mathematical models to the physical
world
Thanks to the Palimpsest, we now know much more about these two
aspects of Archimedes’ achievement.
The mathematics of infinity and the application of mathematical
models to the physical world are closely interrelated. This is because
physical reality consists of infinitesimal pulses of force acting instan-
taneously. As a consequence, to find out about the outcome of such
forces, we need to sum up an infinite number of “pulses,” each infin-
itesimally small. This is surprising. We might think that the mathe-
matics of infinity is some kind of flight-of-fancy with no practical
application. After all, we might think that there is no infinity to be
met with in the ordinary world. But it turns out that the mathematics
of infinity is the most practical tool of science, so important that it is
often called simply “the calculus.” The application of mathematics to
the physical world via the calculus is in a formula, modern science.
Newton, in particular, used the calculus in implicit form to work out
how the planets behave. It is a beautiful result and the inspiration for
28
9780306815805_02.qxd 8/22/07 9:05 AM Page 29
A R C H I M E D E S I N S Y R AC U S E
all later science. It is also, at its core, the application of Archimedean

insights.
And so, since Archimedes led more than anyone else to the for-
mation of the calculus and since he was the pioneer of the application
of mathematics to the physical world, it turns out that Western
science is but a series of footnotes to Archimedes. Thus, it turns out
that Archimedes is the most important scientist who ever lived.
Archimedes’ influence was not confined to the contents of his
science.There is a special quality to his writings. Again and again, his
readers are shocked by the delightful surprise of an unexpected com-
bination. Elegant, unanticipated juxtapositions were Archimedes’
staple.The main reason later scientists were so influenced by him was
that he was such a pleasure to read. Later mathematicians, directly or
indirectly, all tried to imitate Archimedes’ surprise and elegance. Our
very sense of what a mathematical treatise should aim to be is shaped
by Archimedes’ example. In the following chapters, I will try to
explain not only the contents of Archimedes’ works—his contribu-
tions to the calculus and to mathematical physics—but also his style.
Both are equally worthy of our admiration.
I gradually came to appreciate both of these aspects of Archimedes’
achievement while working on the Archimedes Palimpsest. A major
discovery made in 2001 made us see, for the first time, how close
Archimedes was to modern concepts of infinity. Another major dis-
covery made in 2003 made me rethink our entire perception of
Archimedes’ style. Such was the work on the Palimpsest. I would
laboriously pore over a manuscript page (or more often over its
enhanced image on my laptop screen); the letters forming into words,
into phrases; usually nothing new; occasionally discoveries, sometimes
of important historical significance; and then, twice, discoveries that
shook the foundations of the history of mathematics.
I never thought that I would ever find myself laboriously poring
over manuscript pages.The work of editing major texts from antiquity,
based on the transcription of medieval manuscripts, was mostly com-
pleted in the nineteenth century. Of course, one could always make
29
9780306815805_02.qxd 8/22/07 9:05 AM Page 30
small improvements or one could edit minor authors, but not many
people do this kind of work today. This is not only because the more
interesting authors have already been edited, but also because the
intellectual climate today is very different from that in the nineteenth
century. Nowadays, people are less interested in the dry details of texts
and more interested in the syntheses based on those texts. A PhD
thesis in Classics today is usually some kind of theoretical reflection
upon the established texts rather than an addition to the texts them-
selves. “Theory” is what people want. Putting it bluntly, you’re not
likely to get a job if your intellectual output is made up only of textual
editions. Nor is this necessarily a bad development. Nineteenth-
century scholarship was very impressive, and we owe it a great deal.
But it does sometimes make for very boring reading (often in Latin,
at that), and it is even occasionally naive in its lack of critical and the-
oretical reflection. Our understanding of the ancient world was made
much richer and more profound by the application of insights from
cultural anthropology, for instance, or from general poetics and lin-
guistics. My own PhD thesis, prepared at Cambridge under the super-
vision of Sir Geoffrey Lloyd, the doyen of Greek science, was very
much part of this modern tradition. I was greatly inspired by Geoffrey
Lloyd’s application of anthropology to the study of Greek thought, as
well as by his comparative method (where he places Greek science
side by side with its Chinese counterpart). My first book, The Shaping
of Deduction in Greek Mathematics:A Study in Cognitive History, involved
specifically the application of insights from cognitive science (or the
other way around: my hope was that cognitive scientists would find
something to learn from what historians had to tell them). My
objective throughout was to uncover the mathematical experience:
how does it register in the mind’s eye? To get a sense of this, I was per-
suaded, one must first be able to read the mathematics in accurate
translation, which carefully follows the author’s formulations, because
they convey to us how the ancients themselves thought about their
science. The most important of them all was never translated into
English. For Archimedes there existed only T. L. Heath’s poor para-
30
9780306815805_02.qxd 8/22/07 9:05 AM Page 31
phrase published in 1897, which simply ignores Archimedes’ mathe-

matical language. I, therefore, decided to produce a new translation
with a commentary that incorporates my own theoretical angle on
Greek mathematics.
I was going to do more than just translate Archimedes. I am one
of a number of scholars who have, only recently, begun to pay
attention to the visual aspect of science. I mentioned earlier that
nineteenth-century scholarship may appear, in some respects, out-
dated, and here is one respect having to do with the editing work
itself. The scholars who edited mathematical texts in the nineteenth
century were so interested in the words that they ignored the images.
If you open an edition from that era, the diagrams you find are not
based upon what is actually drawn in the original manuscripts. The
diagrams represent, instead, the editor’s own drawing. I was shocked
to realize this and began to consider whether I should produce, for
the first time, an edition of the diagrams. I knew that this would
involve travel to the major libraries housing Archimedes’ various
manuscripts. I researched where those manuscripts were located. It
turned out that they were in Paris, Florence,Venice, and Rome.Well,
why not? I decided it was a good idea.
This was a very ambitious project and not an altogether likely one.
There are some 100,000 words of Archimedes to be translated.
Difficult 100,000 words.Worse, as friends kept pointing out to me—
what was I going to do when the text was uncertain? How was I
going to decide, given that the most important manuscript was no
longer available?
Because, you see, there it was—the Archimedes Palimpsest, the
unique source for Floating Bodies, Method, and Stomachion and a
crucial piece of evidence for most of the other works—and no one
knew where it was. It had been studied at the beginning of the twen-
tieth century and then it disappeared. Nor did I expect it to resurface,
which was my reply to my friends: since the manuscript is likely to
remain unavailable, let us just proceed as if it did not exist, otherwise
we will never do anything regarding Archimedes.
31
9780306815805_02.qxd 8/22/07 9:05 AM Page 32
Pat Easterling, the Regius Professor of Greek at Cambridge and an

expert on Greek manuscripts, closely followed my project, teaching
me the basic skills of paleography. One day I received a letter from
her. The letter said that Christie’s was asking permission to photo-
graph a certain leaf kept at Cambridge University Library, because
this leaf was believed to have been taken out of the Archimedes
Palimpsest, a manuscript that they were about to sell.
I mentioned this casually to my colleagues in ancient science,
assuming they had known about this all along. No one did.This letter
from Pat Easterling was a bombshell. The news of the imminent sale
broke in the community of Archimedes scholars, and the rest is
history. Will has already mentioned his own meeting with Pat
Easterling and his email to me. As for my reaction to this email, that
is, my own wild, childish, embarrassing cries of jubilation; of this I
prefer not to speak. Let us speak of Archimedes.
Who Was Archimedes?

The Second Punic War (218–202 BC) was, to antiquity, much like
World War II was to the modern era. It was a cataclysmic catastrophe
of unprecedented proportions, turning the geopolitics of the
Mediterranean upside down. For a moment, it appeared as if
Hannibal might conquer Rome.Yet, Rome survived, triumphant, and
so powerful at the end of the war that the entire Mediterranean was
at its mercy. The independence of Greek states was gone; the civi-
lization that Archimedes represented was humbled. One of the major
turning points of the war came as Syracuse fell. This, the leading
Greek city in the western Mediterranean, had made the wrong
strategic decision of allying itself with the Carthaginians. In 212 BC,
following a long siege, its defenses—set up by Archimedes and unde-
feated in battle—succumbed to treachery. We do not know how, but
Archimedes died.
The above, in point of fact, sums up what we know about
Archimedes as a historical figure. It should be stressed that we are
32
9780306815805_02.qxd 8/22/07 9:05 AM Page 33
lucky to know even that; indeed we should be amazed that we can

date events in antiquity at all. After all, no one in antiquity jotted
down: “Archimedes died in 212 BC!” The way ancient dates are
derived is fundamentally as follows. We are lucky to possess several
historical documents from antiquity arranged as annals. These docu-
ments detail events year by year. (The Roman author Livy is a famous
example.) Their dating system was different from ours, but occa-
sionally such authors provide us with astronomical data, eclipses, in
particular.We can then apply Newtonian physics to calculate the date
of those events, and with the results of such calculations, we gain
footholds into ancient chronology, constructing the basic equiva-
lencies between ancient dates and modern ones. Without such astro-
nomical data, no chronology could be fixed with any certainty. Even
calculating the date of Archimedes’ own death is based on his science.
The siege of Syracuse was a major event etched into ancient
memory. It was listed in all the annals, and we know very well when
it ended. The figure of Archimedes, as the chief Syracusan engineer,
was of great fascination to his contemporaries, and he appears again
and again in ancient accounts. (This, once again, is a little like World
War II: think of the way in which Einstein was etched into the public
imagination as “the father of the atom bomb.”) It is safe, therefore, to
assume that the date of his death is correct. But other pieces of evi-
dence are much less reliable. Archimedes’ dates are often reported in
encyclopedias as 287–212 BC. We know where the date 212 comes
from. But what about 287? This is based on a Greek author who
mentions that Archimedes died “as an old man, 75 years old.”
Unfortunately, the author in question, Johannes Tzetzes, lived in the
twelfth century AD! What he has to say about Archimedes comes
from a gossipy, fanciful poem. This is also the main source for the
story about Archimedes inventing mirrors that burned enemy ships.
Surely, Archimedes’ contemporaries would have reported such a
thing had it happened and Tzetzes was a Byzantine, whose navy was
famous for its ship-burning feats. In short,Tzetzes’ story is just that—
a story—and he makes his Archimedes an old man just for literary
33
9780306815805_02.qxd 8/22/07 9:05 AM Page 34
effect. Probably Archimedes was quite old (so says the reliable
Polybius), but nothing more is known.
Here is the problem. Archimedes was so famous that legends clung
to him. And now, how are we to separate history from legend? This
is the historian’s problem. Up until the nineteenth century it was
common to accept ancient stories as reality. Since then, skepticism has
reigned. Perhaps historians today are too cautious, but we tend to
dismiss nearly everything that is said about Archimedes. Did he cry
“Eureka”? I doubt this myself, and let me explain why. Let’s take the
most famous version of this story (also the earliest) told by
Vitruvius—the date and author already give room for doubt.
Vitruvius writes some two hundred years after Archimedes’ death
and, in general, is not a very reliable historian (his book is a manual
of architecture, which he spices up with historical anecdotes).
Here is the story. Archimedes is lost in thought contemplating the
problem of a crown. The crown is supposed to be made of gold, but
is it pure? Then Archimedes notices the water splashing out of his
bath . . . and immediately he runs out crying “Eureka, eureka.” Eureka
what, precisely? According to Vitruvius, eureka is the observation that
the volume of water displaced by a body immersed in it is equal to
the volume of the body itself: so place the crown in water, measure
what may be called the “splashed” quantity—and you have the
volume of the crown. Compare this to an equally heavy mass of gold:
does it make the same splash? The heavier it is, the smaller the splash
it makes. So now you can conclude whether the crown has the spe-
cific gravity of gold or not.The method is sound, but it is based on a
trivial observation. Essentially, the observation is “bigger things make
bigger splashes.” This is so trivial that it was not even mentioned in
Archimedes’ treatise on Floating Bodies.
To me it appears that Vitruvius or his previous source knew that
Archimedes discovered something about bodies immersed in water.
They were also familiar with some trivial, pre-scientific observations
such as “bigger things make bigger splashes.” They invented the story
to tie the two together. Vitruvius clearly knew nothing about
34
9780306815805_02.qxd 8/22/07 9:05 AM Page 35
Archimedes’ science.This is the pattern with all of the stories dealing

with Archimedes, from Vitruvius to Tzetzes.They appear to be urban
legends. Sorry.
Some pieces of real evidence can be put together, providing us
with the outlines of a fascinating story. As we will see again and again
throughout this book, the pieces of evidence are extremely
minuscule and often call for lots of interpretation. This holds for the
most important piece of biographical evidence that we have about
Archimedes.This piece of evidence is an aside made in the course of
one of Archimedes’ more surprising works, the Sand-Reckoner. In this
treatise, Archimedes notes various estimates that were offered for the
ratio of the sun to the moon: Eudoxus, for instance, said that the sun
was nine times bigger and, as the manuscript reads, “Pheidias
Acoupater” said the sun was twelve times bigger. There is no such
name or place as “Acoupater.” The text reads literally pheidia tou
akoupatros.You must bear in mind that, until late in the Middle Ages,
Greek was written without spaces between the words.This gives rise
to the following conjecture. If we separate the words differently,
which is permitted when interpreting ancient Greek texts, and
change just one letter, the gibberish “acoupater” makes perfect sense.
The reason we are allowed to make corrections such as these is that
mistakes were certainly made by scribes as they copied their texts; our
manuscripts are full of such scribal errors. So the suggestion was
made, by several editors in the nineteenth century, to reread it with
the “k” changed to “m,” and space inserted as follows: pheidia tou
amou patros. This translates to “Pheidias, my father.” You can see that
this is a thin thread, but a very robust one. The text has to be cor-
rected, and the correction offered is so elegant and straightforward
that it seems as if it has to be true.
On this thin thread hangs the entire family biography of
Archimedes, which should give us a sense of how important—and
how difficult—the detailed study of manuscripts is. All of our
knowledge of the ancient world is derived from the patient, laborious
piecing together of jigsaw puzzles. The previous example does not
35
9780306815805_02.qxd 8/22/07 9:05 AM Page 36
seem like much, but it does tell us that Archimedes’ father was an
astronomer and that his name was Pheidias.
This is a fact I find very meaningful. I have studied the name
“Pheidias” in antiquity, and I ask you to bear in mind two facts. Art,
as well as craftsmanship in general, were not highly appreciated by
ancient aristocrats (who generally speaking looked down on anyone
dirtying their hands). And, Pheidias is the name of the most famous
artist in antiquity—the master sculptor of the Parthenon in the fifth
century BC. Now, with these two facts in mind, I ask you to consider
the following observation: elsewhere when we determine what a person
named “Pheidias” did, it typically turns out that he was an artist of some sort.
The conclusion is quite simple. The name Pheidias would be
bestowed upon your son—as a proud prophecy—only in artists’ fam-
ilies. Otherwise, why give a name that had the lowly associations of
craftsmanship? So let us note this fact, as well: Archimedes’ grand-
father was an artist.
We have not yet exhausted the quarry of names. What about the
name Archimedes? This is, in fact, unique—and uniquely appropriate
to Archimedes. As is very often the case with Greek names, it is com-
prised of two components: arche or “principle, rule, number one,”
and medos or “mind, wisdom, wit.” The name means, if read from the
beginning to the end, “the number one mind,” which is a very good
description of Archimedes. But it was probably meant to be read from
the end to the beginning, as is more often done with Greek names.
It is a unique name, but it has a parallel in another name, Diomedes,
with “Dio” (a variant of “Zeus”) instead of arche. The name
Diomedes means “The Mind of Zeus,” and therefore the name
Archimedes means “The Mind of the Principle,” which sounds a bit
strange but makes perfect sense. Greek philosophers in the genera-
tions before Archimedes, starting with such figures as Plato, gradually
evolved a kind of monotheistic, scientific religion in which they wor-
shipped not so much the anthropomorphic gods of Greek religion
but rather the beauty and order of the cosmos, its “principle.” The
name Archimedes suggests that Archimedes’ father, the astronomer,
36
9780306815805_02.qxd 8/22/07 9:05 AM Page 37
subscribed to such a religion and worshipped the beauty and the

order in the cosmos. So we can tell quite a bit about Archimedes’
background from tiny pieces of evidence and lots of interpretation.
The grandfather was an artist; the father was a scientist, more specif-
ically an astronomer who turned to the new religion of beauty and
order in the cosmos; and the son who created works in which art and
science, beauty and order, all worked together in perfect harmony.
Those works are of course the key to understanding Archimedes.
The stories may be urban legends, but the works exist for us to read
and the surprising thing is that, dry mathematical pieces as they might
seem to be, they actually burst with personality. In his pure science,
Archimedes keeps splashing out of the bath. Art and science, beauty
and order: let us begin to see how they come together in Archimedes’
works.
Science before Science

When we say that “Archimedes was a scientist,” we may be tempted
to imagine him wearing a white coat, contemplating vials with pur-
plish liquids. Well, this is not what he did. He was wearing a tunic,
contemplating diagrams drawn on sand. We may also be tempted to
imagine him as a very earnest man, dedicated entirely to the cause of
impersonal truth. This, would be as wrong as the purplish vials.
Archimedes was not a modern scientist. His was a different kind of
science; a science from before our own professionalized, capital letter
“Science.”
Perhaps the best introduction to the man is what he tells us in the
introduction to one of his treatises, On Spiral Lines. The introduction
is presented as a letter to a colleague, Dositheus, and Archimedes
begins by reminding Dositheus of previous letters.You will recall, says
Archimedes, that I put forth a number of mathematical puzzles. I
announced various discoveries and asked for other mathematicians to
find their own proofs of those discoveries. Well—(notes Archimedes,
somewhat triumphantly)—no one did! And now, continues
37
9780306815805_02.qxd 8/22/07 9:05 AM Page 38
Surface of segment b Surface of segment a
Segment b Segment a
b a
FIGURE 2.1 Archimedes’ hoax: A sphere divided into two segments
Archimedes, it is time to reveal a secret: two of the discoveries were

“poisoned.” For example, Archimedes had announced his “dis-
covery” that, if a sphere is cut into two segments, and the ratio of the
surfaces is a:b, then the ratio of the volumes is a2:b2 (see fig. 2.1).
I wish to stress that there is no doubt, on the internal evidence of
his own writings, that Archimedes was indeed aware, from early on,
that these two claims were false. He is not trying to save face retroac-
tively. He really did send out “poisoned” letters hoping to trap his
fellow mathematicians. As he puts it, he did this “so that those who
claim to discover everything, producing no proof themselves, will be
confuted, in their assenting to prove the impossible.”
Archimedes, you should note, did not have a gentle character—nor
did he have an earnest one. “Playful” is one term that comes to mind,
“sly” is another. It is not for nothing that historians keep debating the
precise meaning of Archimedes’ discoveries: he meant for his readers
to be puzzled. So, incidentally, he would most likely have thoroughly
relished the future history of his writings.That the effort to read him
is so tantalizing, so difficult; is precisely as Archimedes wanted.
The entire structure of scientific activity during Archimedes’ time
was radically different from anything we are familiar with.There were
no universities, no jobs, and no scientific journals. It is true that, about
38
9780306815805_02.qxd 8/22/07 9:05 AM Page 39
a century before Archimedes’ death, a number of “schools” were

founded in Athens, but these, too, were quite different from modern
scientific institutions. They were more akin to present-day clubs,
where like-minded people come together to discuss issues of impor-
tance to them (usually philosophical rather than scientific). In
Alexandria, the Ptolemaic kings set up a huge library. There were
other libraries, as well, but they were not part of research institutions,
instead simply marks of enormous wealth and prestige. So, quite
simply, there was no career in science. Nor was there much glory in
it. After all, few people could even read science.The real path to glory
was then—as always in the pre-modern world—via poetry. If you
wanted to make a name for yourself, to win some kind of eternity,
you would write poems—which, after all, was what everyone read
(starting, in early childhood, with the Iliad and the Odyssey).
How would one become a mathematician? You would have to be
exposed to it by chance—say, by your father if he happened to be an
astronomer . . . And then, you were hooked.This was a rare affliction.
I once estimated that in the entire period of ancient mathematics,
roughly from 500 BC to AD 500, there were, perhaps, a thousand active
mathematicians—one born every year, on average. I should make
clear right now that earlier figures, such as Pythagoras and Thales,
were not mathematicians at all.The name “Pythagoras’ theorem” is a
late myth. Mathematics began in the fifth century BC—the age of
Pericles and of the Parthenon—but very little is known about the
authors. Perhaps the most important was Hippocrates of Chios. (This
should not be confused with the doctor of the same name, from Cos).
All we know of such authors comes from late quotations and com-
mentaries. In the fourth century BC not much more is known:
Archytas was Plato’s friend and a great mathematician, but only a
single proof survives by him. Even that does not survive from
Eudoxus, later in the century; but Archimedes mentions him twice,
with admiration. Apparently Archimedes considered Eudoxus to be
his greatest predecessor; but the works of this predecessor are now all
lost.
39
9780306815805_02.qxd 8/22/07 9:05 AM Page 40
This is not so with the works of Euclid, writing, perhaps, early in

the third century. They survive in plenty. However, Archimedes
wouldn’t think very highly of them, as they consist mainly of basic
mathematics. Archimedes was an advanced mathematician, writing
for people who knew much more than just the contents of Euclid’s
Elements. And there must have been very few of these. I believe that
Archimedes may have had an “audience” consisting of a few dozen
mathematicians at most, spread throughout the Mediterranean, many
of them isolated in small towns, impatiently waiting for the next
delivery of letters from Alexandria (the exchange center)—is there
anything new from Archimedes?
When Archimedes’ introductions start with a letter sent to an indi-
vidual, this should be understood in a very literal way. They were
private letters—sent out to people in Alexandria who had the con-
tacts to deliver the contents further. Everything depended on this
network of individuals. Archimedes keeps lamenting in his introduc-
tions the death of his older friend Conon (who was an important
astronomer). He was the only one who could understand me! In
most of Archimedes’ letters there is a faint note of exasperation.There
was no one to write to, no reader good enough. (There would be, in
time. Archimedes would eventually be read by Omar Khayyam,
Leonardo da Vinci, Galileo, and Newton; these were Archimedes’ real
readers and the ones through whom he has made his real impact. He
must have known that he was writing for posterity.)
Many of the works are addressed to Dositheus, of whom very little is
known.We do know one thing, which, yet again, is based on his name
alone. It turns out that practically everyone in Alexandria at that time
named Dositheus was Jewish. (The name, in fact, is simply the Greek
version of Matityahu or Matthew.) This is very curious: the correspon-
dence between Archimedes and Dositheus is the only one known from
antiquity between a Greek and a Jew. It is perhaps telling that the arena
for such cross-cultural contact was science. In mathematics, after all,
religion and nation do not matter.This, at least, has not changed.
40
9780306815805_02.qxd 8/22/07 9:05 AM Page 41
Squaring Circles
And what mathematics it was, sent to Dositheus! First came a
treatise on the Quadrature of the Parabola. Then two separate books
on Sphere and Cylinder. Then a book on Spiral Lines. (The one in
which the hoax was revealed.) And, finally, a book on Conoids and
Spheroids.There might have been more, but these are the five books
that survive. The five works form a certain unity, as together they
constitute the cornerstone of the calculus. However, this is
probably not how Archimedes would have thought of them. To
him, they were all variations on squaring the circle. That is: time
and again, Archimedes takes an object bounded by curved lines and
equates it with a much simpler object, preferably bounded by
straight lines. Apparently this task—squaring, or measuring, the
circle—was, for Greek mathematicians, the Holy Grail of their
science.
The very idea of measurement depends on the notion of the
straight line. It is not for nothing that we measure with rulers. To
measure is to find a measuring tool and apply it successively to the
object being measured. Suppose we want to measure a straight line.
For instance, suppose we want to measure your height, which is
really saying that we want to measure the straight line from the
floor to the top of your head. Then what we do is take a line the
length of an inch and apply it successively, well over a sixty times,
but probably less than eighty times to measure your height. Since
this is very tiresome, we have pre-marked measuring tapes that save
us the trouble of actually applying the length successively, but, at
the conceptual level, successive application is precisely what takes
place.
When measuring area, we do the same successive application, but
instead of a straight line, we use a square. This is why floor plans are
literally measured by square feet. Cubes similarly measure volume. Of
course, not all objects come pre-packaged in squared or cubed units.
41
9780306815805_02.qxd 8/22/07 9:05 AM Page 42
First, any area – however complex in

shape – is easily divided up, as in the
picture, into triangles.
Second, every triangle – no matter its

shape – is exactly half the rectangle
enclosing it, as a consideration of the
two symmetries in the figure serves
to show.
Third, every rectangle can be easily

transformed into an equal square, by a
proportional reduction and enlargement;
reduce the length, and enlarge the width,
by exactly the same ratio, so that length
and width become equal.
FIGURE 2.2 How to measure an area bounded by straight lines
However, Greek mathematicians from very early on came up with

three important discoveries (see fig. 2.2):
• Every area bound by straight lines can be divided into tri-
angles.
• Every triangle can be made equal to half a rectangle.
• Every rectangle can be made equal to a square.
The combination of these three facts means that it is possible to
measure any area bounded by straight lines as a sum of squares. The
42
9780306815805_02.qxd 8/22/07 9:05 AM Page 43
Parabola
Triangle
FIGURE 2.3 The area of the parabola is four-thirds the triangle
same, analogously, holds for solids divided into pyramids, which are
then made equal to cubes. It is all this straightforward.Take any object
bounded by straight lines. It can be conceptually difficult—a Rubik’s
cube or a many-spangled snowflake—but its measurement always
follows the same principle and it is truly straightforward. Take,
instead, an object as apparently simple as a baseball—just the most
ordinary sphere—and measurement suddenly breaks down. It is
impossible to divide the baseball into any finite number of pyramids
or triangles.The baseball has an infinitely complex, infinitely smooth
surface. Archimedes would measure such objects again and again,
pushing the most basic tools of mathematics.
In the Quadrature of the Parabola, Archimedes measured the
segment of a parabola: it is four-thirds times the triangle it encloses
(see fig. 2.3). A very striking measurement, given that the parabola is
a curved line, so this is rather like squaring a circle. He also, in the
same treatise, introduced a certain daring thought experiment: to
conceive of a geometrical object as if it were composed of physical
slices hung on a balance.
The two books on Sphere and Cylinder directly approach the
volume of the sphere. It turns out that it is exactly two-thirds the
cylinder enclosing it.What is its surface? It turns out it is exactly four
times its greatest circle (see fig. 2.4). This recalcitrant object—the
sphere—turns out to obey some very precise rules. In the second
book, remarkable tasks are achieved. For instance, finding the ratio
43
9780306815805_02.qxd 8/22/07 9:05 AM Page 44
Sphere
Greatest circle
FIGURE 2.4 The surface of the sphere is four times the area of its greatest circle
between spherical segments, which was the substance of the hoax

mentioned previously.
In both Spiral Lines and Conoids and Spheroids, Archimedes is not
content to measure known objects. Instead he invents a new curved
object—a complex, counter-intuitive object—and then measures it.
The spiral line—invented by Archimedes—turns out to enclose
exactly one-third the area enclosed by the circle surrounding it (see
fig. 2.5). As for conoids, which are hyperbolas or parabolas turned
around so as to enclose space, and spheroids, which are ellipses turned
around in similar fashion—these have more complex measurements.
All were obtained, with precision, by Archimedes (see figs. 2.6, 2.7).
This is a major feature of all of his works. Archimedes starts out
promising to make some incredible measurement, and you expect
him to fudge it somehow, to cut corners. How else can you square the
circle? And then he begins to surprise you. He accumulates results of
no obvious relevance—some proportions between this and that line,
some special constructions of no direct connection to the problem at
hand. And then, about midway through the treatise, he lets you see
how all the results build together and, “By God!” you exclaim, “he is
actually going to prove this precisely, no fudges made!”
44
9780306815805_02.qxd 8/22/07 9:05 AM Page 45
Spiral line
Circle enclosing
the spiral
Area enclosed
by the spiral line
FIGURE 2.5 The area of the circle is three times the area enclosed by the spiral
FIGURE 2.6 A conoid is a solid created by rotating a parabola or a hyperbola on its axis
Each of these works was of a completely different order of origi-

nality and brilliance from anything ever seen before. In all of them
Archimedes was furthering the mathematics of infinity.
Imaginary Dialogues
In his measurements Archimedes adopts a surprising, circuitous route,
which was always his favorite way of approaching things.The general
45
9780306815805_02.qxd 8/22/07 9:05 AM Page 46
FIGURE 2.7 A spheroid is a solid created by rotating an ellipse on its axis
plan is as follows: apply a combination of “indirect proof” and

“potential infinity.”
Both indirect proof and potential infinity are best considered as
imaginary dialogues. Indirect proof is easier to understand, and you
have probably engaged in some version of it yourself.You try to con-
vince someone of the truth of your position. Let us say, for instance,
that you want to convince your interlocutor that, when you draw a
straight line joining two points on the circumference of a circle, all
the points on this straight line must fall inside the circle. Everything
you tell him about this line fails to persuade him. And so you resort
to indirect proof. You assume the opposite of the truth, as if pre-
tending to agree with your interlocutor.
“Let us assume that some point E falls outside the circle,” you
concede (see fig. 2.8). And now you follow the logic of this situation,
until you draw the following conclusion: line DZ is both smaller than
DE and bigger than it. But a line cannot be at the same time both
smaller and greater than the same given line. “See,” you now turn to
your imaginary interlocutor, “I have conceded your claim, but the
result was an absurdity; therefore your claim must be false. I have now
46
9780306815805_02.qxd 8/22/07 9:05 AM Page 47
A D The line AB must never get out of the circle.
B
Imagine it does, as the ‘line’ AEB.
DE is greater than DZ, because it contains it:
DE > DZ. DZ is equal to DB (both are radii
of the circle), while DB, in turn, is greater
than DE. (This is because in a triangle such as
A D D
Z
A E B
E
B the external side DB is greater than the internal
line DE.) As a consequence, DZ is greater than DE:
DZ > DE.
Both DZ > DE and DE > DZ Contradiction!
FIGURE 2.8 Indirect proof:Why does a line never get out of its circle?
proved this indirectly.” This type of argument is one of the hallmarks

of Greek mathematics.
Archimedes did not invent potential infinity, but he made it his own
in a series of original applications. You remember the fundamental
problem with measuring a curved object: it could not be fully divided
into triangles; with any finite number of triangles in it, there is always
some piece of the curved object “left out.” Now let us concentrate on
the size of this piece that is left out. What Archimedes does is develop
a certain mechanism, capable of indefinite extension, of packing tri-
angles (or their like) into the curved object. This is best seen, once
again, as an imaginary dialogue between Archimedes and his critic.
Let us say that he has packed the curved object in such a way that
a certain area has been left out, an area greater than the size of a grain
of sand.
47
9780306815805_02.qxd 8/22/07 9:05 AM Page 48
A critic comes along and points out that there is still a difference—
the size of a grain of sand.
“Is that right?” exclaims Archimedes. “All right then, I shall apply
my mechanism successively several more times.” By the end of this
operation the area left out is smaller than the grain of sand.
“Wait a minute,” says the critic, not yet satisfied. “The area left out
is still greater than a hair’s width.”
Archimedes, unfazed, applies the mechanism once again, with the
area left out becoming smaller than a hair’s width.
“No, no!” the critic squabbles again; “the area left out is still bigger
than an atom.”
The critic may think he has had the last word, but Archimedes just
goes on applying his mechanism. “See,” he returns now to the critic,
“the area left out is now even smaller than the atom.” And so it goes
on, the difference always becoming smaller than any given magnitude
mentioned by the critic.
This dialogue could go on indefinitely.This is what philosophers refer
to as potential infinity.We never go as far as infinity itself in this argument.
There is no mention, at any point, of an area which is infinitesimally
small, merely of areas that are very, indefinitely small. But we allow our-
selves to go on indefinitely. And this, taken together with indirect proof,
allows Archimedes to measure the most incredible objects.
Squaring the Parabola

Three times in his career Archimedes proved that the parabolic
segment—a certain curved object—is exactly four-thirds the triangle
it encloses. This was his favorite measurement. Later on we shall see
his most spectacular measurement, which transcends geometry itself.
But before we can follow such flights of imagination we must first
acquaint ourselves with Archimedes’ geometrical method, which is
based on the combination of indirect proof and potential infinity. It
is an extraordinarily subtle argument, one that even professional
48
9780306815805_02.qxd 8/22/07 9:05 AM Page 49
FIGURE 2.9-1 The parabola encloses a jumble of triangles.This can be made to

approach the parabola as closely as we wish.We assume its difference from the parabola
is less than a grain of sand
mathematicians have a hard time unraveling. It is like an affirmation

based on a double double negation. And this is how it works.
Since what we are going to prove is that the area of the curve is
four-thirds the triangle, how shall we start? By assuming, of course,
that the area of the curve is not four-thirds the triangle! This, after all,
is how indirect proof works. Let us assume that the curved area is
greater than four-thirds the triangle, by a certain amount:
1. The curve is greater than four-thirds the triangle, by a
certain amount. Let us say, it is greater by a grain of sand.
For exactly such occasions, Archimedes has a special mechanism up
his sleeve. He fills up the curve with triangles so that the difference
between the triangles and the curve is smaller than a grain of sand!
We therefore now have two objects side by side. One is a curve.
The other is the product of Archimedes’ mechanism—a complex
jumble of triangles whose difference from the curve is known to be
smaller than a grain of sand:
2. The curve, with a grain of sand removed, is smaller than the
jumble of triangles (see fig. 2.9-1).
At this point Archimedes leaves aside the results obtained so far.
The following piece of reasoning involves, instead, a separate piece of
49
9780306815805_02.qxd 8/22/07 9:05 AM Page 50
geometrical ingenuity. Remember that the jumble of triangles is an

object bounded by straight lines. Four-thirds the enclosed triangle is
also an object bounded by straight lines, i.e. both are objects that can
be precisely measured by ordinary means. It is therefore not a surprise
that through the application of geometrical ingenuity one can
determine a definite measurement comparing the jumble of triangles
and four-thirds the enclosed triangle. What Archimedes would come
up with—applying his geometrical ingenuity—is as follows:
3. The jumble of triangles is smaller than four-thirds the
enclosed triangle (see fig. 2.9-2).
Now, recall result 1. It was: “The curve is greater than four-thirds
the triangle, by a grain of sand.” Or, putting differently:
4. The curve, with a grain of sand removed, is equal to four-
thirds the triangle.
Put this alongside result 2: “The curve, with a grain of sand
removed, is smaller than the jumble of triangles.”
The same object is equal to four-thirds the triangle, but is smaller
than the jumble of triangles. In other words, the jumble of triangles
is the greater. It is greater than four-thirds the triangle, which we can
put as:
5. The jumble of triangles is greater than four-thirds the
enclosed triangle.
This makes no sense if we put it side by side with result 3: “The
jumble of triangles is smaller than four-thirds the enclosed triangle.”
Results 5 and 3 directly contradict each other. There is no way to
reconcile them. The jumble of triangles cannot be both smaller and
greater than four-thirds the enclosed triangle. In other words, we are
left with only one option: concluding that our original assumption
was wrong.The curve is not different from four-thirds the enclosed tri-
angle. The curve is therefore four-thirds the enclosed triangle. The
50
9780306815805_02.qxd 8/22/07 9:05 AM Page 51
One - third the triangle + the triangle =
The jumble of triangles is smaller than

four - thirds the enclosed triangle.
FIGURE 2.9-2
result is therefore obtained. Indirect proof and potential infinity

together have brought us the answer.
Beyond Potential Infinity

Now let us put this into historical context. In the seventeenth century,
mathematicians found a way to apply this technique of Archimedes in
51
9780306815805_02.qxd 8/22/07 9:05 AM Page 52
a more general fashion. Instead of finding, ingeniously, this or that

strategy for this or that object, they had a general recipe for measuring
all curves.This was the calculus, which, as mentioned previously, is the
foundation of modern science. The persons most directly responsible
for this were Newton and Leibniz, and if you wish to know which of
the two deserves more credit, you are not alone.The battle for priority
between Newton and Leibniz was the most famous—as well as the
ugliest—in the history of science. Most scholars today think it should
be considered an honorable draw. Both, in a sense, followed
Archimedes. More than this, both Newton and Leibniz managed to
bring the calculus to a great height—on shaky foundations. The
underlying logic of handling potential infinities was not clearly
worked out by the inventors of the calculus. It was put in order only
at the beginning of the nineteenth century, especially by the French
mathematician, Cauchy, who essentially reverted to the Archimedes
method of implicit dialogue (“You find me an area X; I’ll find you a
smaller difference.”). In every step along the way, our calculus, as well
as our understanding of potential infinity was Archimedean. So much
for potential infinity. As for actual infinity, the contemplation of an
actual infinite set of objects, it was not mastered by Newton or
Leibniz, but was put in order, if it can be called order, in the late nine-
teenth century by authors such as Cantor.
And here comes the shocking surprise. In 2001, it was discovered
for the first time—against all expectations—that Archimedes knew of
actual infinity and used it in his mathematics. This was discovered
through the Palimpsest and it is, without any doubt, the most
important discovery made through its reappearance.
Proofs and Physics

The new discovery was made by reading a passage of Archimedes’
Method that had never been read before. This work, the most fasci-
nating of all the works by Archimedes, survives in the Palimpsest
alone. It is his most fascinating work because, more than anywhere
52
9780306815805_02.qxd 8/22/07 9:05 AM Page 53
else, he brings together his two interests: the mathematics of infinity,

which we have seen already, and the combination of mathematics and
physics, statements of pure geometry with those of the physical
world. It all comes out of the balance. Archimedes was the first to
prove, mathematically, the law of the balance: objects balance when
their weights are exactly reciprocal to their distances from the
fulcrum. In the Method, he pushes forward a surprising technique. He
takes geometrical objects and creates a thought experiment in which
they are both arranged on a balance. He then uses their weights (that
is, their lengths and areas) and their distances from the center, to
measure some purely geometrical properties. The law of the balance
becomes a tool of geometry instead of physics.
This work was not among the series sent to Dositheus.This may be
because Archimedes valued it so highly. It was instead sent to the most
influential intellectual of Archimedes’ time—Eratosthenes. This
polymath wrote on everything from Homer to astronomy, from prime
numbers to Plato. As a result he was nicknamed “Beta,” because he
was the number two on everything . . . Archimedes, who clearly saw
himself as number one in his field, approached Eratosthenes with great
apparent respect, but he almost seems to be teasing him, as if saying
“See if you can catch me!” The work is so interesting, in part, because
it is the most enigmatic. Archimedes suggests that he has discovered a
method of finding mathematical results which is very powerful, but
which does not quite constitute a proof.Yet he never explains what
this method actually is or how it falls short of actual proof. He leaves
it as an enigma to be worked out by the reader—first Eratosthenes and
then, since the discovery of the Method in the twentieth century, every
historian of ancient mathematics. Everyone has a theory about the
Method.We will return to this enigma, perhaps to understand it better,
based on the new readings of the Palimpsest.
Of course, Archimedes’ claim to be the founder of mathematical
physics does not rest on the Method alone. Of his studies in this field,
two major works are still extant—Balancing Planes and On Floating
Bodies. In Balancing Planes, which we will examine later, Archimedes
53
9780306815805_02.qxd 8/22/07 9:05 AM Page 54
finds the center of gravity of a triangle, which is one of the key results
of the science of statics. On Floating Bodies sets the stage for another
science—hydrostatics. This work provided the foundation for
Vitruvius’ nice but silly story about Archimedes’ splashing in the bath.
He may have splashed, he may have run naked; but he certainly did
not cry Eureka over such a trivial observation as “bigger things make
bigger splashes.” The deduction in On Floating Bodies is much more
subtle and sophisticated.
This is it. In a stable body of liquid, each column of equal volume
must also have equal weight otherwise the liquid would flow from
the heavier to the lighter (this is why the face of the sea is even). The
same must hold true even if a solid body is immersed within such a
column of liquid. In other words, if we have a column of liquid with
a solid body immersed in it, the aggregate weight of the liquid and
the body must be equal to that of a column of liquid of the same
volume. It follows that the solid body must lose some of its weight.
Archimedes performs a complex calculation that demonstrates that it
must lose weight equal to the volume of water it has displaced.
This explains why we feel lighter in the bath. Indeed it tells us pre-
cisely by how much we should feel lighter in the bath. Now that’s
something to cry eureka about! Because, you see, by the power of pure
thought alone, Archimedes is capable of saying what must happen in
the physical world! This power of mind over matter is what is so fas-
cinating about Archimedean science. This is what Galileo and
Newton tried to imitate and, incredibly, succeeded in doing. In this
way, ultimately Newton discovered, by the power of pure thought—
as well as by the calculus—how the planets must move. And, with this
Archimedean achievement, Newton set the stage for all later science.
Puzzles and Numbers

Newtonian science was sober-minded; Archimedes’ science was not.
Archimedes was famous for hoaxes, enigmas, and circuitous routes.
These were not some external features of his writings; they charac-
54
9780306815805_02.qxd 8/22/07 9:05 AM Page 55
terized his scientific personality. Science is not—mathematics is

not—dry and impersonal. It is where one’s imagination is allowed to
roam freely. And so, Archimedes’ imagination roamed to childs’ play,
and to a puzzle called the Stomachion or Bellyache (because of the
difficulty of solving it). The game consists of a tangram puzzle of
fourteen pieces made to form a square. Archimedes wondered: what
was the underlying mathematics of this puzzle?
This, in fact, was a puzzle to modern scholars, as well. Since 1906,
we have known that Archimedes wrote about this Stomachion
puzzle. But just what was he trying to do? We had available to us only
a single bifolio of the Palimpsest and it was one of the worst pre-
served. Heiberg had made little sense of the Greek and none of the
mathematics. Digital technology finally allowed us to make further
readings and, in 2003, I was finally able to offer an interpretation of
the Stomachion—the first ever in modern times. I argued that
Archimedes was trying to do the following: calculate how many ways
one can form the square given the original fourteen pieces. There is
more than a single way, as figure 2.10 illustrates. In point of fact, there
are 17,152 distinct solutions.
The most striking thing about this interpretation is not the very
big number itself, but something else. If true, and most historians
think I am probably right, then this would make Archimedes the first
author, ever, on combinatorics—the field of mathematics calculating the
number of possible solutions to a given problem.This was the second
important discovery made through the Archimedes Palimpsest.
The field of combinatorics is at the heart of modern computer sci-
ences, but it had no such application in Archimedes’ time. It is very
different in character from the kind of geometrical study we see so
often in Greek mathematics. Indeed, the Stomachion appears like a
pure flight of fancy. You look for a number because you know it is
there. Along the way you generate a fantastic array of complicated cal-
culations. Archimedes, after all, was a master of this game—this hunt
for big numbers and surprising combinations. In the Sand-Reckoner, he
calculated how many grains of sand it would take to fill the universe.
55
9780306815805_02.qxd 8/22/07 9:05 AM Page 56
2
9
5
3
8
14
12
7
6 4
11
13 10
Two other solutions
FIGURE 2.10 The Stomachion
(For this, you recall, he needed an estimate of the size of the universe
and he mentions his father’s estimate.) And then again, most famously,
he offers a fantastically precise approximation of the ratio of a circle to
its diameter, which is known today as the number pi. He managed to
56
9780306815805_02.qxd 8/22/07 9:05 AM Page 57
FIGURE 2.11 A ninety-six-sided polygon within a circle.The inner circle shows a mag-
nified detail of the ninety-six-gon within the outer circle.
determine that this ratio is smaller than that of 14688 to 46731⁄2, but
greater than that of 6336 to 20171⁄4. He then simplified it, losing a tiny
bit in precision but gaining much more in clarity, to the ratio of less
than three and a seventh and greater than three and ten-seventy-
oneths! This amazing calculation is based on a method not unlike that
of the treatment of potentially infinite series, but, in the case of the cir-
cumference of the circle, precise calculation is impossible so approxi-
mation works best. So Archimedes calculated not the circumference
of a circle but the circumference of a polygon with ninety-six sides,
which, to sight, is nearly the same (see fig. 2.11).
Perhaps the most strikingly playful calculation made by
Archimedes is that of the cattle of Helios. His readers would have
known the context from their memories of Homer’s Odyssey. In
book twelve Odysseus’ crew reaches the island of Thrinacia, which
was sacred to Helios. Against Odysseus’ advice, his crew members
57
9780306815805_02.qxd 8/22/07 9:05 AM Page 58
slaughtered the cattle of Helios and feasted lavishly for seven days.
Throughout the remainder of the Odyssey, they are horribly punished
for this transgression. Tradition has identified the island as Sicily, so
the story could be turned into a poetic tribute to Sicily’s power and
a warning against interfering with the island. Archimedes created a
riddle—a calculatory puzzle couched in a poem:
Measure for me, friend, the multitude of Helios’ cattle,
Possessing diligence—if you partake of wisdom:
How many did once graze the plains of Sicily,
The Island Thrinacian divided in four herds,
In color varied . . .
The text goes on for about three pages, with many mathematical
constraints for instance:
. . . The white bulls
Were of the black a half and then a third
And then the whole of yellows, friend, do know this . . .
In short, Archimedes constructed an arithmetical problem with
eight unknowns (four herds: black, white, yellow, and many-spangled,
each divided into both bulls and cows); seven equations (for example,
the one above, [white bulls] = 5/6 [black bulls] + [yellow bulls]); and
two complex conditions in which the solutions were integers (there
are no half-cows). To try and solve this problem, it turns out, is a
transgression as fateful as the original slaughter. Do it at your own
peril. Modern mathematicians have proved that the smallest solution
involves a number written out in 206,456 digits.
This was a game. For you see, the presentation above, with its
short, metrical lines, was not whimsical on my part. Archimedes wrote
out this problem in verse. A poet-mathematician! The thought seems
absurd, but it was natural for Archimedes whose entire science was
based on a sense of play and beauty, on hidden meanings. In this case,
the hidden meanings were, among other things, political. Archimedes
was trying to suggest that one should not interfere with Sicily. Many
58
9780306815805_02.qxd 8/22/07 9:05 AM Page 59
did, in his time, and he did his best to stop them. And here, finally, we
return to the historical facts of Archimedes’ life.
Death and Afterlife

Syracuse was the leading city of Sicily—the pivot between the
eastern and western Mediterranean. Previous invaders had tried to
conquer it. Most famously, in the year 415 BC, Athens tried to force
the outcome of the Peloponnesian War by gaining the riches of Sicily.
The crashing failure of that expedition marked the end of the
Athenian Empire.
So it was with some realistic hope that the Syracusans were waiting,
in 214 BC, for the Romans to come. For a generation Syracuse, nom-
inally free, had been tightly enmeshed within the Roman sphere of
influence. Undoubtedly Archimedes—like most of his fellow cit-
izens—was eager to shake off this indirect Roman control. Hannibal’s
recent victories over the Romans appeared to make this finally pos-
sible. Syracuse openly sided with the Carthaginians. Unless the
Romans could contain and reduce this city, their own fate would be
sealed. Through Sicily, Hannibal would receive fresh supplies. The
Romans, their power depleted, would not be able to withstand the
siege for years on end. So the fate of the Mediterranean rested on this
question: could Syracuse take it for long enough?
The previous century had seen a military revolution. From the
wars of hoplites clashing on the battlefield a new kind of warfare had
evolved—the siege. In the ancient world’s version of an arms race, all
cities were building up their walls and all the military powers were
building up their arsenals of catapults—the machines that drove the
military revolution. In principle, a catapult is no more than a huge
spring that, once released, serves to propel a rock to smash into walls
or people. It can be surprisingly effective, knocking down, in time,
very sturdy defenses. But one needs to get the catapult within range
of the walls where it will also be within range of the catapults from
inside—and so on, with the arms race.
59
9780306815805_02.qxd 8/22/07 9:05 AM Page 60
The Romans fully expected the Syracusans to spring rocks on

them.Yet they were in for a surprise. I now quote from Polybius, a
very sober historian who wrote not long after the events described
and the best source for Archimedes’ life. Here is the surprise sprung
by Syracuse on its invaders:
But Archimedes, who had prepared machines constructed to
carry to any distance, so damaged the assailants at long range, as
they sailed up [the first attempt was by sea] with his more
powerful catapults as to throw them into much difficulty and
distress; and as soon as these machines shot too high he con-
tinued using smaller and smaller ones as the range became
shorter and, finally, so thoroughly shook their courage that he
put a complete stop to their advance . . . [The Romans gave up
the assault, and so Polybius sums up:] Such a great and mar-
velous thing does the genius of one man show itself to be . . .
The Romans, strong as they were both by sea and land, had
every hope of capturing the town at once if one old man of
Syracuse were removed; but as long as he was present, they did
not venture even to attempt to attack . . .
What did Archimedes do? After all, catapults were well known
before him. What seems to have taken the Romans completely off
balance was the careful sighting and ranging of the catapults. There
were no “blind spots,” which is a crucial consideration. “Blind spots”
gave the attackers safe spaces.They could jump from one to the other,
largely disabling the scheme of defense.
How do you avoid blind spots and sight your catapults? This is a
formidable problem, not at all capable of solution by means of simple
trial and error.There had to be some principle of constructing a catapult
to order. That is, one which would allow a catapult, positioned at a
given point, to cover a precise range.
At the most basic level, this involves a deep problem of geometry.
The propulsive power of a catapult is roughly equivalent to its mass,
which determines the physical force it exerts.This, in turn, is roughly
60
9780306815805_02.qxd 8/22/07 9:05 AM Page 61
equivalent to its volume. Now, how do we measure volume? The

same way we measure surfaces: by multiplying the dimensions (two
in the case of surfaces, three in the case of volumes). Since a solid has
three dimensions, its volume is equivalent to the cubic power of its
linear dimension. So, let us say, that you have a catapult whose length
is a yard, and you wish to transform it into another catapult that is
twice as powerful. Making it two yards long would be wrong. A two-
yard-long catapult would be not twice as powerful, it would be eight
times more powerful. How do you create a catapult that is twice as
powerful, then? For this, we must find the cubic root of two (which is
roughly 1.26) and extend the length, as well as all the other linear
measurements, by that ratio.The finding of cubic roots is not an easy
task. In fact, it calls for very powerful mathematical techniques. Greek
mathematicians had already tackled the problem, but there were few
solutions allowing any practical application. None appear in
Archimedes’ name, but there is no doubt that he had found a tech-
nique and applied it in the year 214 BC.
However, I suspect more than this. Here is a pure flight of fancy
on my part. Archimedes could tell where a missile would land by fol-
lowing in his imagination the curve traced by that missile. So, why
wouldn’t he concentrate his attention on the problem of representing
projectile motion as a geometrical curve? Wouldn’t he have found the
curve—his beloved parabola—and used it to place his catapults in
position? After all, Archimedes’ pupils Galileo and Newton did.They
based their ideas on mathematical techniques not so different from
those used by Archimedes. Galileo traced the motions of projectiles
and Newton traced the motions of planets using parabolas.
Archimedes would have had interesting problems to ponder, surely,
constructing those machines!
And so legend has it: picture Archimedes in the year 212 BC, pon-
dering his problems.The Roman siege is nearly defeated, Archimedes’
genius triumphing over Roman power. Complacency has set in. The
Syracusans are celebrating a festival and a deserter informs the
Romans that the sentries, drunk, have deserted their posts. The
61
9780306815805_02.qxd 8/22/07 9:05 AM Page 62
Roman general, Marcellus, quickly dispatches a group of soldiers to

occupy positions on the wall and, as ever in this kind of warfare, once
a breach is made the game is up. Soon the city is overtaken by the
Romans. They have little reason to be merciful. Syracuse had been
counting on Rome’s downfall and now Rome could thoroughly
relish the reversal of its fortune.The looting is unprecedented in scale,
even for the Romans, who now take away everything that can be
removed.
We are told that the booty included a huge planetarium, a marvel
of science, produced by Archimedes. We are also told that Marcellus
wished to capture and bring home Archimedes in person, but with
the pillage came a spree of senseless murder.The legend is famous and
reported by Plutarch:
He was by himself, working out some problem with the aid of
a diagram, and having fixed his thoughts and his eyes as well
upon the matter of his study, he was not aware of the incursion
of the Romans or of the capture of the city. Suddenly a soldier
came upon him and ordered him to go with him to Marcellus.
This Archimedes refused to do until he had worked out his
problem and established its demonstration, whereupon the
soldier flew into a passion, drew his sword, and killed him.
So much for the legend. (Plutarch himself reviews several other
alternatives, and the truth is likely be different still.) But it is an appro-
priate legend. Several ignorant hands nearly brought an end to
Archimedes’ heritage by destroying a scientific legacy that they did
not understand.Yet Archimedes survived.
For we may conclude with sober historical fact, instead of legend.
Let us follow Cicero in the year 75 BC. Archimedes has been dead for
137 years. Sicily is a Roman province, part of a Mediterranean that
has been thoroughly subdued. Cicero himself is a quaestor, a high
official, on the island. He is also a cultivated man, with deep respect
for the Greek scientific heritage. He knows about the old tomb of
Archimedes and is capable of finding it again, despite its having been
62
9780306815805_02.qxd 8/22/07 9:05 AM Page 63
lost for all those years. And on the tomb the old engraving (requested
by Archimedes himself) is still there: a sphere, and a cylinder exactly
enclosing it.
Archimedes did prove that the first was always two-thirds the
second—a masterpiece of reasoning that got him as close to squaring
the circle as is humanly possible. The diagram on this tomb was
immortal. Archimedes did find the first deep, revolutionary truths. In
time, they would give rise to our science. But first, his works had to
survive—to cross the seas of history so that, on the other shore,
modern science could be born.
63
9780306815805_02.qxd 8/22/07 9:06 AM Page 64
64
9780306815805_03.qxd 8/22/07 9:09 AM Page 65
3
The Great Race, Part 1
Before the Palimpsest

r. B sent me another letter, but it wasn’t a check this time. It
M contained transcripts of the court case concerning the
Palimpsest, which continued after the sale. In the transcripts, the
lawyer for Christie’s said that plans for an exhibition of the manu-
script were already being explored through a major museum in
Baltimore. Clearly, Mr. B had the idea of an exhibition even before I
did. Since the book would not have made much of an exhibition by
itself, I decided to make a film that explained why it was so
important. I got in touch with John Dean, a filmmaker and now a
dear friend. He bought the airplane tickets, and we flew around the
Mediterranean making the movie, with John singing and making
friends along the way. We had to pack the story of 2,200 years into
two weeks of filming.
The film told the story of the book as if it were taking part in a
race. The race lasted for centuries and took place all over the
Mediterranean world. It was a race for survival, and it was an epic.
Archimedes rode on a donkey—the concern of the scholar and the
care of the scribe. Arrayed against Archimedes were the mighty thor-
oughbreds of destruction: war, indifference, and the second law of
thermodynamics. If his works were to survive, Archimedes would
have to stay ahead of his opposition throughout the race; his treatises
had to be rewritten more times than they were destroyed. Ancient
authors were all in the same race and facing similar odds. But for most
of them, thanks to the printing press of Johannes Gutenberg, the race
was effectively over by the end of the sixteenth century. From 1454
65
9780306815805_03.qxd 8/22/07 9:09 AM Page 66
onward, these authors dismounted their donkeys and climbed onto

Pegasus, the winged horse. Even destruction’s thoroughbreds found it
difficult to overtake the printing press. But Gutenberg’s invention was
still 1,666 years away when Archimedes died. By a series of truly fan-
tastic circumstances, Archimedes’ race is not even over in the twenty-
first century, and the final lap is still being run in the New World.
A Letter Is Written
You might think of it as a soccer ball kicked by Italy. But Sicily isn’t
a sphere; it’s a triangle. And that’s how the ancients thought of it. John
Dean and I landed in Palermo, which is on the western angle of the
triangle. We drove through its center at Piazza Armerina, then down
toward the south-east angle, leaving Mount Etna on our left.We then
arrived at Archimedes’ home town—Syracuse. John and I might have
stayed in Winston Churchill’s favorite hotel, but we walked the streets
that Archimedes trod, sat in the theater where he watched plays,
visited the altar at which he worshipped, and followed the city walls
that he defended. High above the city, on the Epipolae Plain, the
impressive remains of the Euryalus fortress still stand. In April 1999,
it looked stunning with its white ramparts emerging from a sea of
wild flowers. The view was magnificent, and we could see the port
below us to the east.
From Syracuse, before the Second Punic War, Archimedes had
written a letter to a friend. The letter began:
Archimedes to Eratosthenes: greetings! Since I know you are

diligent, an excellent teacher of philosophy, and greatly inter-
ested in any mathematical investigation that may come your
way, I thought it might be appropriate to write down and set
forth for you a certain special method . . . I presume there will
be some among the present as well as future generations who
66
9780306815805_03.qxd 8/22/07 9:09 AM Page 67
T H E G R E AT R AC E
by means of the method here explained will be enabled to

find other theorems which have not yet fallen to our share.
It is easy to accept recorded history as a full account of the past.

However, it really isn’t. If we think of it this way, we not only misun-
derstand what history is but also miss marveling at what it is that we
know. Nothing is inevitable about the fact that we know that a great
man in the third century BC wrote a private letter to a friend; it is
absolutely extraordinary that we know this. Amazingly, we also know
a lot of what he wrote in the letter and even what it looked like.
The letter was made of papyrus sheets and wrapped around a
wooden core. In other words, it was a roll. Papyrus is a fibrous plant
that used to grow plentifully in the Nile delta, and papyrus rolls were
the stationery of choice in the ancient Mediterranean. Strips taken
from the lower stem of the plant were laid down parallel and slightly
overlapping each other. Then another set of strips was laid down at
right angles to the first.The two layers were bashed with a mallet and
they would stick together to make an excellent writing surface.
Sections could be glued together to make rolls of different lengths.
The rolls were made in Egypt and distributed throughout the ancient
world from the greatest trading port at the time—Alexandria. To
write on the roll, Archimedes used a reed pen and wrote on just one
side of the sheet—the side on which the papyrus leaves were laid
horizontally. He wrote in narrow columns of capital letters, not down
the length of the roll but instead down its width. He did not leave
spaces between his words, and there was virtually no punctuation as
we understand it today. His diagrams, which he regarded as integral
to his text were placed within the columns of his text, following the
text to which they referred.
Once he finished his letter, Archimedes took it to the port and
arranged for it to be shipped to its destination. He was sending his
roll on a perilous sea voyage right back to where it had come from—
67
9780306815805_03.qxd 8/22/07 9:09 AM Page 68
Alexandria. If he kept a copy of what he sent—and this was standard

practice—then no trace of that copy survives. Perhaps it was
destroyed, along with Archimedes himself, in the siege of 212 BC.
Once the boat left the harbor, the fate of his letter was no longer in
his hands. It was no ordinary letter; it was his Method.
In the Library
The town of Aswan, in Egypt, lies on the Tropic of Cancer. This
means that at noon on June 21 the walls of its buildings cast no
shadows. The walls of Alexandria, about 485 miles farther north, do.
At the same date and time a small shadow lines their walls’ northern
edges. Even though the walls of both towns are vertical, and the sun’s
rays are nearly parallel, the walls stand at different angles to those rays.
The vertical walls of Aswan are at a seven-degree angle from the ver-
tical walls of Alexandria. It was obvious to Eratosthenes that the
surface of the Earth itself, the ground upon which the walls of both
towns stood, was curved. By measuring the distance from Aswan to
Alexandria, which accounts for 7 of the 360 degrees of a sphere, one
can estimate the circumference of the Earth. Eratosthenes thought
this through in the third century BC and came up with the value of
250,000 stadia, in which 1 stadion is 125 paces or about 625 feet.This
was astonishingly close to the value accepted today—24,900 miles. A
smart calculation, and because of feats like this Archimedes addressed
the Method to Eratosthenes. It may have been because Eratosthenes
also knew a lot about a lot of subjects that he was appointed director
of the library at Alexandria in 235 BC.
Alexandria was a young city in Eratosthenes’ day. It was founded
on April 7, 331 BC by Alexander the Great, and it soon replaced
Memphis as the capital of Egypt. From 305 BC it was ruled by
Ptolemy I Soter, of Greco-Macedonian ancestry, and the dynasty that
he founded was to rule Egypt until Cleopatra’s suicide in 30 BC.
Under the Ptolemies, Alexandria became a great center of Greek
culture. By 280 BC a Temple to the Muses—the world’s first
68
9780306815805_03.qxd 8/22/07 9:09 AM Page 69
museum—had been created. Built in the palace complex, it was a set

of buildings, including a large dining hall, a covered walk, and an
arcade with recesses and seats for the use of a scholarly community.
It was there that Eratosthenes and other scholars spent their days,
pacing the covered walk, speculating on such subjects as the circum-
ference of the Earth, and occasionally receiving letters from their
friend Archimedes.Their library constituted the greatest collection of
texts in the ancient world. It had grown rapidly in the previous fifty
years, because scholars had set out to systematically record the sum of
world learning as it then existed. In the middle of the third century
BC a catalogue of the library was attempted by Callimachus of
Cyrene, entitled “Catalogues of the Authors Eminent in Various
Disciplines.” It was a monumental achievement. It took up 120 rolls.
It was divided into categories, and within each category authors were
discussed alphabetically based upon the first letter of their name.
It would not have taken more than a couple of weeks for the ship
bearing Archimedes’ letter from Syracuse to arrive at the port of
Alexandria, because it had a tail wind. After making a sacrifice to
Poseidon for a safe voyage, someone would have taken the roll to its
intended recipient—the safest of all hands, the learned and respon-
sible lover of rolls—Eratosthenes. Eratosthenes probably placed his
letter from Archimedes in a section of the library devoted to science
texts, and it rested there beside other rolls containing different trea-
tises by the same author. It is more than likely that Eratosthenes had
copies made of the letter. Judging by Archimedes’ hope that later
generations would read the letter, this was most certainly his expec-
tation. One of the copies may have been stored nearby in the
precincts of the Temple of Serapis. The scrolls in the Serapeum, as it
was called, were copied from those in the museum and were available
not only to scholars leading their sequestered lives but also to
members of the public. Eratosthenes lived to be about 80 years old.
In his later years, he was blind and it is believed that he committed
suicide by voluntary starvation. But he had done all he could for
Archimedes, and it was enough—just.
69
9780306815805_03.qxd 8/22/07 9:09 AM Page 70
We know that Archimedes’ letter to Eratosthenes arrived in

Alexandria only because we know that it was read there. In the first
century AD someone retrieved a copy of it. We even know the name
of this person—Hero. Hero wrote a treatise, Metrica, which mentions
the Method. “The same Archimedes shows in the same book [the
Method] that if in a cube two cylinders penetrate whose bases are
tangent to the faces of the cube, the common segment of the
cylinders will be two thirds of the cube.This is useful for vaults con-
structed in this way . . .”
Hero was interested in Archimedes’ treatise because it could help
in the construction of a groin vault, which consists of cylindrical
spaces carved out of rectilinear masonry. This was a specialty of
Hero’s, and he also wrote a treatise on vaulting.We will see again and
again, while tracing the survival of Archimedes’ treatises, that they
were of interest to people who wanted to apply his knowledge to
real-world technical problems.
Note, however, how thin the trail of evidence is by which we can
trace the history of Archimedes’ works. Hero’s Metrica survived only
in one manuscript, and we can place him in Alexandria in the first
century AD only because the Sun, the Moon, and the Earth have
extremely regular movements in relation to one another. In another
treatise, Dioptra, Hero gave an eyewitness account of an eclipse of the
Moon, which he states took place on the tenth day before the vernal
equinox beginning at Alexandria in the fifth watch of the night. Otto
Neugebauer, the great figure of ancient mathematical astronomy,
observed that this corresponds to an eclipse in 62 AD and to no other
eclipse for centuries on either side of this date. And for centuries on
either side of this date, we hear nothing further of the Method.
A Change of Medium
Nothing is more dangerous for the contents of old documents than
an information-technology upgrade, because mass data transfer has to
take place and somebody has to do it. The transition from the roll to
70
9780306815805_03.qxd 8/22/07 9:09 AM Page 71
the codex—the book format we know today—was a revolution in

the history of data storage (see fig 3.1).
The introduction of codices was gradual. It started in the first
century AD yet was only more or less complete by the end of the
fourth century. To me, the great surprise is that it took so long. The
genius of the codex is that it contains knowledge not in two dimen-
sions, like a roll, but in three. The roll has height and width; the
codex has height, width, and depth. Because it has depth, it doesn’t
need to be nearly as wide. A codex with 200 folios (400 pages), 6
inches wide, has the same potential data-storage area as a roll of the
same height that is 200 feet long. Since each leaf of a codex is so
thin, you save an awful lot of width for a very small increase in
depth. What’s more, to access data on a roll you have to travel
through the width dimension. To access data in a codex, you only
ROLL
CODEX
FIGURE 3.1
71
9780306815805_03.qxd 8/22/07 9:09 AM Page 72
have to travel through the depth dimension, which is just a couple

of inches thick. There is a big difference between “unrolling” and
“thumbing through.” If you are consulting a catalogue listed alpha-
betically getting to Archimedes is not a problem. But what about
Zeno? With a codex you could just flip to the folios at the end and
then close the codex.You would have to unravel nearly the whole
roll in order to find a few lines about Zeno. This did not happen, of
course. If the catalogue were of any length, this literally could not
have happened. This is why Callimachus’ catalogue consisted of 120
rolls. If Callimachus’ rolls were ever transcribed into codices, there
would have been a lot less than 120 of them.
Ancient texts that didn’t make the transition from roll to codex
simply disappeared. The ancients shunned their rolls for the same
reasons that we neglect our 78 rpm vinyl records: they had become
an outmoded storage system. Only decades ago, 78 rpm records were
the preferred medium for music recordings. Now they are more often
found in trash cans than on turntables. Similarly, the fragmentary
remains of ancient texts can now be found in the trashcans of the
ancient world. If Archimedes’ letter to Eratosthenes had stayed in its
roll, it would have been at first neglected, then abandoned, and finally
it would have crumbled to dust. In fact, the copy (or copies) that
stayed in rolls did just that, though no fragments of Archimedes have
been found in trash cans.
Just because Archimedes was famous, it doesn’t mean that his
works were an obvious priority in the IT upgrade. In truth, even
though he was legendary, Archimedes was hardly ever read.The most
important of his results, like his approximation for the value of pi,
became well known and well used, but few people actually read his
arguments.They were simply too difficult. And here Archimedes was
at a particular disadvantage compared to other great thinkers in
antiquity.The Homers, the Platos, and the Euclids of this world were
recognized at the time as not only great but also fundamental. They
were frequently used and, in due course, copied into codices.
Archimedes was too difficult to be fundamental; very few people
72
9780306815805_03.qxd 8/22/07 9:09 AM Page 73
could understand him. His genius actually worked against him. His
texts were very often left unrolled, and they were always going to
have had a tough time getting into codices.Three hundred years after
Hero another mathematician, Pappus, discussed a treatise by
Archimedes on semi-regular polyhedra. Of this treatise no trace
remains. Perhaps it never made it into a codex. It lost the race against
destruction right there.
The person who did more than anyone else to ensure the survival
of Archimedes’ treatises through this decisive period was named
Eutocius. Eutocius was born in Ascalon, Palestine in about 480 AD.
He didn’t just read Archimedes’ treatises; he researched them and
explained them. Eutocius traveled widely among the great centers of
learning at the time, including to Alexandria, where he must have
met a teacher named Ammonius. Eutocius dedicated his first work on
Archimedes, a commentary on Sphere and Cylinder I, to Ammonius,
and clearly held him in high regard. In his preface, Eutocius says that
he would write commentaries on other treatises by Archimedes if
Ammonius approved of this one. Ammonius must have approved,
because Eutocius went on to write three additional commentaries—
on Sphere and Cylinder II, on the Measurement of the Circle, and on
Balancing Planes. Eutocius had to struggle to find Archimedes’
writings. Regardless of whether they were already in codices or still
in rolls, there weren’t very many of them. At one point in Sphere and
Cylinder II, Archimedes promises to prove a mathematical point but
never does. Eutocius therefore went on a search. He writes: “In a
certain old manuscript (for we did not cease from the search of many
manuscripts) we have read theorems written very unclearly (because
of the errors), and in many ways mistaken about the diagrams. But
they had to do with the subject matter we were looking for, and they
preserved in part the Doric dialect Archimedes liked using, written
with the ancient name of things.” Eutocius then included an account
of this text in his commentary.
Eutocius’ treatises survive together with the works of Archimedes
that they comment upon. And this is an important point. Eutocius,
73
9780306815805_03.qxd 8/22/07 9:09 AM Page 74
like everybody else, clearly saw the advantages of the new IT and
exploited it. Just as a CD can store a great many more Bach cantatas
than a 78, so a single codex can contain many more of Archimedes’
treatises than a roll. Eutocius, it seems, prepared an edition of several
of Archimedes’ treatises, together with his commentaries, and had
them bound within wooden boards. From the sixth century onward,
we should imagine a treatise by Archimedes inside a handy
parchment codex, placed safely within wooden covers, and nestled
comfortably with other letters of a similar nature.
The Gathering Storm

Archimedes’ letter to Eratosthenes might have been comfortable in
codex form, but it was by no means safe. Times were changing, and
they were not changing to Archimedes’ advantage. One by one the
great cities of the ancient world, which held the ancient schools of
learning and the books upon which they depended, were pillaged by
invaders. Rome was sacked by the Goths in 410 AD, Antioch by the
Persians in 540 AD, and Athens by the Slavs in 580 AD. There may
have been many copies of Archimedes’ letters outside of Alexandria
in the third century AD; there were hardly any by the end of the sixth.
Things were not much better in Alexandria. In about 270 AD,
Emperor Aurelian, during his war against Zenobia, damaged a large
portion of the palace complex, which contained the Museum. In 391
AD, Theophilus, the Archbishop of Alexandria, destroyed the
Serapeum, the sister library of the Museum. In 415 AD, the distin-
guished female mathematician Hypatia was torn to pieces by a fanatic
and ignorant Christian mob. Archimedes’ letters had to get out of
Alexandria before they suffered a similar fate.
As the ancient world disappeared, its gods went with it. And as
Christianity became the official religion of the Roman Empire,
many classical texts, if they were not condemned as dangerous, were
dismissed as irrelevant. It is not that Christians willfully destroyed
them very often; they just ceased to copy them. Scribes put their
74
9780306815805_03.qxd 8/22/07 9:09 AM Page 75
energies into Christian texts. The Christian curriculum did include

some ancient authors—Homer was necessary for Rhetoric and
Euclid for Geometry. But Archimedes was not included in the cur-
riculum of Salvation. Fewer people than ever before had the
resources to read him, and even fewer would have read him, if they
could.
In the fifth and sixth centuries, for every donkey that Archimedes
could ride on, there were hordes of thoroughbred barbarians. The
race against destruction for all classical texts was getting increasingly
desperate. The only question was where could they run? The only
answer was Constantinople.
Into the Ark

John Dean and I flew to Constantinople, present-day Istanbul, in
search of Archimedes. The city was founded on the Bosphorus by
Constantine, the first Christian Emperor, on Monday, May 11, 330
AD. He founded it specifically as the capital of the eastern Roman
Empire. Constantinople was a relative newcomer to the
Mediterranean world of learning. Successive emperors poured the
resources that only they could marshal into making the city worthy
of the empire that it inherited. Certainly the emphasis was on
Christian works. Constantine ordered fifty complete copies of the
Bible to be written. But the classics were also a concern in
Constantinople. On Wednesday, January 1, 357 AD, in an address to
the Emperor Constantius, the philosopher Themistius described a
plan to guarantee the survival of ancient literature. He proposed a
scriptorium—a writing center—for the production of new copies of
the classics. This would ensure that the new capital of the empire
would become a center of culture. The plan may have actually been
put into effect. In 372 AD, an order was issued to the city prefect,
Clearchus, to appoint four scribes skilled in Greek and three in Latin
to undertake the transcription and repair of books. In 425 AD, the
Emperor Theodosius II established an imperial foundation for literary
75
9780306815805_03.qxd 8/22/07 9:09 AM Page 76
and philosophical studies. Even more importantly, in 412 AD he built

massive walls around the city.
John Dean and I didn’t have to look far to find Archimedes. His
geometry is indelibly stamped on Constantinople. Between 532 AD
and 537 AD, the Emperor Justinian crowned the “New Rome” with
one of the greatest buildings the world has ever seen—the Church of
Hagia Sophia. This is how a contemporary, Procopius describes it:
A construction of masonry rises from the ground, not in a
straight line, but gradually drawing back from its sides and
receding in the middle, so as to describe a semi-circular shape
which is called a half-cylinder by specialists, and this towers to
a precipitous height. The extremity of this structure terminates
in the fourth part of a sphere, and above it another crescent-
shaped form is lifted up by the adjoining parts of the building
. . . On either side of these, columns are placed on the ground,
and these, too, do not stand in a straight line, but retreat
inward in a half-circle as if making way for one another in a
dance, and above them is suspended a crescent-shaped form.
Hagia Sophia is an astonishing building on many levels, but the
important point is that it was designed with diagrams and by
numbers. It was the work of mathematicians. One of them was
Anthemius of Tralles. Anthemius wrote texts on Burning Mirrors and
Remarkable Mechanical Devices. The second architect was somewhat
younger than Anthemius. His name was Isidore of Miletus and he
wrote a commentary on Hero of Alexandria’s treatise on Vaulting.
Anthemius and Isidore were masters of Archimedes’ discipline and
the plan of Hagia Sophia might well resemble the figure that was
inscribed on Archimedes’ tomb.
The mathematical world was small in the sixth century, and it was
getting smaller all the time. It is not surprising that Anthemius and
Isidore were well acquainted with Eutocius. Eutocius dedicated his
commentaries on the works of Apollonius of Perga to Anthemius.
Isidore knew Eutocius’ works on Archimedes extremely well. The
76
9780306815805_03.qxd 8/22/07 9:09 AM Page 77
text of Eutocius’ commentary on Sphere and Cylinder I is preserved

because it was copied by one of Isidore’s students. When he had fin-
ished his copying, this student wrote: “The commentary of Eutocius
of Ascalon on the first book of Archimedes On the Sphere and the
Cylinder, the edition being collated by the Milesian mechanical
author, Isidore, our teacher.” Thus, Isidore was preparing an edition
of Archimedes’ works, together with Eutocius’ commentaries, in
Constantinople.
Archimedes made it to Constantinople in the nick of time. For
three hundred years after the time of Isidore, his writings, like most
of the classics, disappear from recorded history. The empire became
embroiled in internal strife about holy images and was subject to
external threats from barbarians and Arabs. When texts were read,
they were read only to bolster particularly contentious aspects of
Christian doctrine. Constantinople did the one thing it had to do for
Archimedes, and for so many ancient authors—it survived. It was the
only city of the ancient world of any consequence to survive unmo-
lested into the Middle Ages. Constantinople served as the ark for
ancient literature, and the Noah of the classics was the Emperor
Theodosius. A hundred years before Isidore built his great church,
Theodosius had already constructed the city’s massive walls to
weather the Dark Age storm.
The Byzantine Renaissance

On Saturday, July 26, 811, Krum, the Bulgarian Khan, slew
Nicephorus, the Byzantine Emperor, at the battle of Pliska.The Khan
turned the Emperor’s skull into a wine cup. Not a good start to the
ninth century for Constantinople. On the surface, things didn’t look
much better thirty years and six emperors later when Michael III,
“The Drunkard,” ascended the throne. But, in fact, the intellectual
climate was improving, and it got even better when Basil I assassi-
nated Michael in 867 AD. Under Basil I, Constantinople quickly
became the capital of the greatest empire in the Mediterranean
77
9780306815805_03.qxd 8/22/07 9:09 AM Page 78
world. The Macedonian dynasty that he founded could boast both

scholarship and mettle. While Constantine VII wrote a book on the
administration of the empire in 1014 AD, Basil II took 14,000
Bulgarians prisoner and blinded ninety-nine out of every hundred;
the lucky one got to guide his comrades home. Constantinople had
entered a golden age, if not an enlightened one.
The famous Byzantine “rebirth” of the ninth and tenth centuries
produced impressive buildings and consummate works of art. John
and I could still see and film the Imperial Palace, and when we got
back to Baltimore we could still photograph Henry Walters’ fabu-
lously illuminated Byzantine manuscripts. But the most important
thing about this cultural revival was that as scholars started reading
the classics that had been lying neglected in their libraries, they
started copying them again. The most voracious reader of them all
was a distinguished civil servant, and twice the Patriarch of
Constantinople, Photius. His Bibliotheca was a compilation of works
that he had read; it contained a summary of the contents, style, and
biography of each author. As Nigel Wilson describes it, Photius had
invented the book review. The Bibliotheca is invaluable for many
reasons, but prominent among them is that it gives us a good idea of
the extraordinary variety of classical texts that were still extant in
Constantinople in Photius’ day. Admittedly, scholars have thought
that some of Photius’ claims were a bit much. He claimed, for
example, to have read works by Hyperides, an ancient Greek orator.
Since no one else in Constantinople even mentioned whole texts by
Hyperides, and since none survive in codex form, this is considered
unlikely. Nonetheless, the statistics gleaned from Photius are
impressive. For example, of the thirty-three historians that he dis-
cussed, the works of twenty are now unknown.
When ninth-century scribes copied classical texts they wrote in
a fundamentally different script from that used in Isidore’s day.
Before the ninth century, texts were written in capital letters, tech-
nically called majuscules. Generally speaking, after the ninth
century, they were written in a new script, technically called
78
9780306815805_03.qxd 8/22/07 9:09 AM Page 79
FIGURE 3.2 Majuscules (left) and Minuscules (right)
minuscule, whose letters could be joined up in order to take up less

space (see fig 3.2).
The origins of minuscule script were probably in the letters, docu-
ments, and accounts of the civil service based in Constantinople.
Minuscule script was faster to write, its letters were easier to form, and
more words could fit on a folio. By the middle of the ninth century,
it was also used for religious and scientific texts. Many codices, written
in capitals during and before the sixth century, were systematically
transcribed into lowercase script. This change in how texts were
copied was just as important of a hurdle for Archimedes’ texts as the
transition from roll to codex. So few majuscule codices from the fifth
and sixth centuries survive, because many ninth-century scholars may
have destroyed their majuscule manuscripts after they made minuscule
copies. The majuscule codices may have been increasingly difficult to
read, and after they had been transcribed, there was no need for them.
The texts of nearly all the ancient Greek authors were dependent on
79
9780306815805_03.qxd 8/22/07 9:09 AM Page 80
just a few minuscule manuscripts copied in Constantinople during the

ninth and tenth centuries. Archimedes was no exception. He was
actually dependent on three manuscripts.
And this is the fundamental point. All that stuff that Reviel has
talked about, and will go on to talk about, survives because of just
three physical objects, of which the codex on my desk is one.
The ABC of Archimedes

These three objects are called Codex A, Codex B, and Codex C.
They had some texts in common. All three contained Balancing
Planes; A and B contained Quadrature of the Parabola; A and C con-
tained Sphere and Cylinder, Measurement of the Circle, and Spiral Lines;
and B and C contained On Floating Bodies. Codex A was the unique
witness to Conoids and Spheroids and Sand-Reckoner; Codex C is the
unique witness to Method and Stomachion.
It seems very likely that all three codices were the fruit of a revival
of interest in Archimedes’ texts early in the ninth century. When a
scribe finished copying Archimedes’ Quadrature of the Parabola in
Codex A, he wrote a little note of adulation, but not to Archimedes.
He wrote: “Leo the geometer, may you flourish—May you live many
a year, dear friend of the Muses.” This Leo was almost certainly the
Leo who was giving private instruction in Constantinople during the
820s. Known as Leo the Philosopher, he was evidently a talented
teacher. In 830, the Arabs captured one student, who had read Euclid
under his supervision. His report of Leo’s learning was sufficient to
cause the Caliph to invite Leo to Baghdad. Thankfully, Leo did not
go. Leo was clearly something of a polymath and a practical one at
that. He built fire signal stations between Constantinople and the
border of the empire. If there was an emergency on the border north
of Tarsus, a message could reach the capital in less than an hour. In
the late 850s, Leo’s skills were rewarded. He was appointed director
of a school in the Imperial Palace. He must have played a prominent
role in choosing the other professors. One of them was Theodore, a
80
9780306815805_03.qxd 8/22/07 9:09 AM Page 81
geometer. We are safe to assume that Archimedes was being studied

and copied in the school of the Imperial Palace, and that the texts in
Codex A were copied in that school. The accession of Basil I in 867
ensured that Archimedes’ treatises could be studied in safety, at least
for the time being.
We will trace the histories of the three manuscripts that contained
Archimedes’ treatises in more detail later. But for now, it is important
to know that Codices A and B survive no longer. We have only
copies and translations of them. As a result, Codex C is not only the
unique source for the Method and Stomachion and for Floating Bodies
in Greek, it is also the oldest surviving manuscript of Archimedes’
treatises in Greek by over four hundred years. Codex C is on my
desk, and now it is time to see how it was made.
Codex C
Codex C, like most medieval manuscripts, is not written on paper; it
is written on the backs of animals. The skin of an animal is such a
refined product of natural selection that it is hard to see how it could
be used for much else. But skin has two great qualities: it is supple,
enabling movement, and it is tough, allowing animals to sustain all
types of knocks. Skin is well suited to life on earth outside of fire and
too much water, and, with some treatment, the same properties make
it an excellent, durable writing surface. With that treatment, it is
called parchment.
Parchment was invented at Pergamum in Asia Minor—or so
legend has it. King Eumenes II wanted his library to match that of
Alexandria, so the Ptolemies put an embargo on the export of
papyrus from Egypt at the beginning of the second century BC.
Parchment was Eumenes’ home-grown substitute. With the intro-
duction of the codex, parchment came into its own. While papyrus
certainly had tensile strength, it fractured more easily than parchment
when it was folded. Since codices consisted of folded sheets, those
made of parchment survived better than those made of papyrus.
81
9780306815805_03.qxd 8/22/07 9:09 AM Page 82
Parchment is more durable than paper, which is why certificates and

awards meant to last are sometimes still inscribed on sheepskins.
Making parchment is not everybody’s idea of fun. Certainly not
Reviel’s; he’s a vegetarian. This is what you do. Kill the animals and
drain the blood from their veins. Flay them; slit their underbellies; cut
off their extremities; and peel back the skin. Place the skins in a vat
containing a weak solution of lime, which you can make by heating
limestone. The lime solution is destructive to organic tissue, so it
breaks down the epidermis and the subcutaneous fat and weakens the
bonds that attach the hair to the skin. The inner layer of the dermis
alone remains intact. This layer is mainly made up of collagen.
Collagen is a protein made up of three chains of amino acids that spiral
around an elongated straight axis. The chains are staggered and the
resulting fibers have no definite terminal limits. Collagen is the crucial
constituent of parchment; it is what makes it tough. After several days,
remove the skins from the vat, place them over a beam, and scrape
them down with a dull blade. Once you have removed the worst of
the fat and the hair, attach the skins to wooden frames. As they dry
they will contract and become taut. Once they are taut, scrape them
again, this time with a very sharp half-moon-shaped blade. Cut them
from their frames, and you have produced parchment.
Imagine a set of newspapers stacked in a pile, sandwiched between
wooden boards, the whole lot stitched together, and you actually have
a good first impression of the physical construction of a tenth-
century Byzantine manuscript. The newspapers, called quires, usually
consisted of four nested double sheets called bifolios (see fig. 3.3).
This makes up eight sheets of a newspaper (sixteen pages).
Remember that manuscript people call the sheets folios. Since you
could get two bifolios out of each sheep, twenty-four sheep were
needed to make enough bifolios for the surviving parchment in the
Archimedes manuscript. To make the quires you had to cut the skins
to size, rub them with a pumice stone to raise a nap, and give them
a transparent glaze mainly consisting of egg white. This was the
support that housed the texts of Archimedes and the others.
82
9780306815805_03.qxd 8/22/07 9:09 AM Page 83
Quire
FIGURE 3.3 A Quire
And now to the ink. This is more fun to make.You start with a
solution of gallic acid. Gallic acid is present in oak galls, which are
growths on oak trees resulting from infections from insects and mites.
It is made up of carbon, hydrogen, and oxygen, and it has the power
to contract organic tissues such as collagen. This allows the ink to
etch into the parchment and remain in place. Crush the galls and boil
them in water. To this solution add ferrous sulphate, which is some-
times called green vitriol or copperas. This will supply most of the
color to the ink. It is a compound of iron and sulphuric acid that you
can frequently find together with pyrite. You then need to add a
thickening agent to this solution. By a process called gummosis, trees
belonging to the Fabaceae family, produce gum from their bark if
they are under attack. Gum Arabic is produced by the acacia tree,
which grows in Africa. Gum tragacanth, produced by several shrubs
of the genus Astragalus, principally astragalus gummier, can be found
in Asia Minor. If you are making a manuscript in Constantinople, you
might find it easier to get hold of gum tragacanth. It is still used today
to coat pills. To produce it in a quantity sufficient to satisfy the phar-
maceutical industry, incisions are made into the bark and wooden
wedges are driven into the incisions.The chemical makeup of gum is
complicated and varied, but it contains carbon, hydrogen, oxygen,
83
9780306815805_03.qxd 8/22/07 9:09 AM Page 84
and metals such as calcium, magnesium and potassium. Since the

resulting mixture darkens slowly as it oxidizes on the parchment, you
might want to add carbon black to it. This way you will be able to
see what you are writing as you are writing it. Combine the ingre-
dients and give them a good shake, and you have the ink that the
scribes used to write their texts.
And so, about a hundred years after the death of Leo, a scribe pre-
pared himself for a job.We can imagine this scribe, because pictures of
scribes writing survive from this period. When we see him, he has
already gone through familiar procedures. Guided by a ruler, he has
made lines on the parchment to help him keep his columns of text
straight, he has sharpened his reed pens on a stone, and cut the nib
down the middle to help the ink flow. He has prepared his ink and
placed the inkwell on his table, and he has beside himself a knife so
that he can refine his pen or scrub out a mistake should he make one.
He has prepared the tools of his trade, the mystic implements of the
human voice. We picture him seated in his chair, ready to write. He
doesn’t have a writing table, but then he doesn’t need one. He is going
to write on his lap, resting the parchment on a board. In front of him,
on a stand, is a codex. This is the codex that he is going to copy.
Before we look at what he wrote, let’s think a bit about the codex
on that stand in front of him. Did the codex on the stand look like
the codex that the scribe was about to make? Was it a sixth-century
codex from the time of Isidore or was it written in minuscule? Was
it one codex containing the same treatises in the same order or
should we be thinking about several codices sequentially placed upon
the stand? We do not know enough about the parent manuscript (or
manuscripts) that our scribe used. Judging from the text, he did not
copy any of his treatises from Codex A. It is certainly possible that he
copied a sixth-century manuscript, but the evidence is not con-
clusive. This is perhaps the most important unanswered question
about our manuscript.
Be that as it may, the scribe did his job and wrote his text. Each
folio that he wrote measured about 12 inches by 7.5 inches. He wrote
84
9780306815805_03.qxd 8/22/07 9:09 AM Page 85
his text in two columns and wrote thirty-five lines in each column.
He made a codex with generous margins, so together the columns
were 9.5 inches high and 5.5 wide. Of course, Archimedes’ letter to
Eratosthenes—the Method—was just one of the texts he copied. The
manuscript currently starts toward the end of Balancing Planes and is
followed by Floating Bodies and the Method.The Method is followed by
Spiral Lines, Sphere and Cylinder, Measurement of the Circle, and finally
by one folio of Stomachion. Our scribe was an expert, writing in a
minuscule script characteristic of the third quarter of the tenth
century. Nigel Wilson says that his handwriting is similar to that in a
manuscript, with the date 988, which is now in the Monastery of St.
John the Theologian on the island of Patmos. The scribe didn’t
understand what he was copying but, as Reviel will explain later, this
was a good thing. We can assume that he worked for no longer than
a few months. We can only guess at the true extent of the original
codex, because we are now missing the beginning, the end, and
several chunks of the middle. In fact, it is perfectly possible that the
manuscript originally contained more of Archimedes’ treatises.
The codex our scribe wrote was actually a typical product of the
Byzantine Renaissance of the ninth and tenth centuries.As with most
Byzantine manuscripts, we do not know who commissioned it or
even who read it. In fact, judging by the lack of marginal comments,
it does not seem to have been used very much. But none of this
matters. It is the unique source for the Method, Stomachion, and On
Floating Bodies in Greek. If our scribe did nothing else in his long life,
his was a life well spent.
This is a story of survival, achieved against all the odds, through an
extravagant process by which creation just managed to outrun
destruction. Many rolls have crumbled and many codices have
burned. I have brought you to the earliest surviving text of
Archimedes—the thoughts of Archimedes preserved in a highly
ordered arrangement of flesh and iron. But it was made in the tenth
century, closer to our own time than to the time of Archimedes.
Tenth-century parchment codices look nothing like third-century BC
85
9780306815805_03.qxd 8/22/07 9:09 AM Page 86
rolls. As Archimedes’ letters were copied, they were transformed and

the treatises that survive no longer look anything like the letters that
Archimedes wrote.Archimedes would not have recognized them, and
could not have even read them.This is important, but to explain why,
we need again to consult the expert.
86
9780306815805_04.qxd 8/22/07 9:13 AM Page 87
4
Visual Science

here is a lot one can learn from manuscripts. For one thing, we
T can pick up what’s written in them. We can find out what
thoughts Archimedes had, in Syracuse, in the third century BC. We
can find out how such thoughts came to influence all of later science.
And we can do more: we can use manuscripts to find out not just
what thoughts past scholars had, but also how they came to think
such thoughts. How did Archimedes think through his mathematics?
How did his readers? Such questions are raised by a recent cognitive
turn in science studies. To answer such questions we must turn to
manuscripts, because they provide us with the unique source of evi-
dence for this fundamental question: how does science register in the
mind’s eye?
In fact, we all have, nowadays, a rather clear image of science in our
mind’s eye. Let us consider the following experiment. In fig. 4.1, I show
two pictures of open pages from books.Take a look for a minute: even
though the illustrations are too small to be read comfortably, they allow
us immediately to make certain judgments.We just know that the left-
hand page is scientific. It is, in fact, from an introductory text in the cal-
culus (the subject pioneered by Archimedes, to which we shall return).
The right-hand page is from James Joyce’s Finnegans Wake (and is, inci-
dentally, much more difficult to read than the left-hand page).
When publishers say they are afraid of publishing popular science
that “looks technical,” what they mean is that they want their pages to
look like the right-hand, not the left-hand page.What are readers afraid
of ? They are afraid of equations.With good reason: they were force-fed
87
9780306815805_04.qxd 8/22/07 9:13 AM Page 88
FIGURE 4.1
such equations for several terrible years of their childhood and adoles-
cence.The result is that we tend, first, to hate equations and, second, to
consider them to be the natural format of science. Both assumptions
are wrong. Equations are a great invention—they should be respected,
if not loved—and they are not natural. Instead, they are a historical
invention whose origins lie in such documents as the Palimpsest. The
Greeks did not use equations. Archimedes did not use equations.Their
science looked nothing like the left-hand page in the figure.
Before Equations
Equations make logic visible. Suppose you say: “The first together
with the second is equal to the third; therefore the first is equal to the
third minus the second.”
Give it a moment’s thought and you see why this is true. But that’s
the trouble: you must give it a moment’s thought; meanwhile your
88
9780306815805_04.qxd 8/22/07 9:13 AM Page 89
V I S UA L S C I E N C E
attention has wandered away and the thread of the argument may
have been lost. Write instead:
A B C, therefore
ACB
and it works effortlessly: we see how the argument runs. Now no
attention is thrown away, and we can go on following the argument
with great ease.
We see how the same information acts differently through its dif-
ferent interfaces. The different media are important not only for the
survival of science but also for its very nature. Indeed, we can hardly
understand Greek science without first understanding its essential
interface. This, however, was not the equation. It was the diagram.
Greek Mathematics Was a Visual Science

The starting point, as always, should be at Syracuse. In particular, we
begin with the story recounted by Cicero: the tomb of Archimedes.
We recall Cicero finding, after much effort, a desolate, neglected slab
of stone and on it the message that Archimedes chose as his symbol—
a diagram showing a sphere and a cylinder. This was the self-chosen
symbol of science. Once again, we make the inevitable comparison
to Einstein. What was Einstein’s symbol, the emblem that immedi-
ately comes to mind? No, I don’t mean the protruding tongue; I
mean this:
E mc2
You have seen this emblem endless times. It has become a kind of
symbol not only of Einstein but also of science in general. And it is,
of course, an equation. So here is the starting point, in a nutshell:
modern science is a science of equations; ancient science was a
science of diagrams.
In the ancient exact sciences—not only mathematics but also
astronomy and mechanics—and in many other fields, such as music
89
9780306815805_04.qxd 8/22/07 9:13 AM Page 90
theory, diagrams always occupy center stage.The text is made of indi-

vidual “propositions,” each making a point, proving that so-and-so is
the case. The late British historian of Greek mathematics David
Fowler used to say that each proposition is “drawing a figure and
then telling a story about it.” Everything is about those figures and
done for the sake of those figures.
Diagrams, of course, are also used in modern science; but there is a
big difference. In modern science diagrams serve as a kind of illustration;
they are there to make the experience of learning science somewhat less
traumatic for the student, but they are not part of the logic of the
argument itself. In modern science it is considered crucial to make sure
that no information depends on the diagram, otherwise one could end
up with a false argument such as is demonstrated in figure 4.2.
The area of a triangle is the product of its two sides, divided by two.
Proof: we draw a triangle ABC. The two smaller sides are AB, BC.
On the longer side AC, we apply another triangle, exactly identical to
the triangle ABC, namely the triangle ACD. The result is a rectangle
ABCD. The area of this rectangle is obviously the product of the two
sides AB, BC.The triangle ABC is obviously exactly half the rectangle
ABCD (after all, the two triangles ABC, ACD are identical). So, the
area of a triangle is the product of its two sides divided by two, QED.
What is wrong with this proof? Well, it assumes something just on
the basis of the diagram, even though there is no basis in the text for
making such an assumption. In the diagram we happened to have
A D
B C
FIGURE 4.2 The right-angled triangle ABC

90
9780306815805_04.qxd 8/22/07 9:13 AM Page 91
drawn a right-angled triangle. In a right-angled triangle the claim of

the proof follows. But, it does not follow in other triangles (see fig.
4.3, where the product of the two sides AB and BC is clearly more
than double the area of the triangle!). In short, we thought we were
talking about triangles in general, but inadvertently we slipped into
only talking about right-angled triangles, all because we had taken
the diagram on faith. And so modern philosophers and logicians are
adamant: do not rely on diagrams!
The logical and philosophical issue is deep: language is general, but
a diagram is a particular thing.You see, you cannot draw a diagram
without it having some particular properties. Suppose I wanted to
draw a triangle so that its angle is neither right-angled, nor acute-
angled, nor obtuse-angled, but just a “general” angle. How would I do
that? I can’t do it. On the page, I have some definite triangle drawn,
and because it is a definite triangle, it also has some definite angle.
Language, on the other hand, is more forgiving. I can say, “Let there
be a triangle,” and because I did not say which triangle—just “a tri-
angle”—I am allowed to think of it as right-angled, obtuse-angled, or
acute-angled.And so the modern philosophers and the logicians insist:
to make sure the logic of the proof works in its full, most general
form, we must rely on the language alone and never on the diagram.
B
A C
FIGURE 4.3
91
9780306815805_04.qxd 8/22/07 9:13 AM Page 92
A E
B H ∆ Z
FIGURE 4.4 Two isosceles cones
This is precisely what Greek mathematicians did and, incredibly,

without making any logical errors. It is one of the great puzzles of
Greek mathematics: it is thoroughly diagrammatic and also thor-
oughly precise.The Greek mathematicians never made mistakes, such
as the trivial one mentioned above, not even in subtle and indirect
ways. Greek mathematics is as precise as modern mathematics. How?
In a few pages, I shall try to provide an answer.
First of all, let us see how Greek mathematicians did in fact rely on
the diagram. The following is from the first book on Sphere and
Cylinder—the work eternalized on Archimedes’ tomb. As already
mentioned, Archimedes had a sly manner of writing—clever and
playful, always hiding the main point of attack from the reader until
the last moment. It should therefore come as no surprise that, until
late in the book, spheres and cylinders are hardly mentioned. Instead,
Archimedes keeps referring to cones. Here, for example, is proposition
17 (see fig. 4.4):
Let there be two isosceles cones ΑΒΓ, ∆ΕΖ; let the base of the cone
ΑΒΓ be equal to the surface of the cone ∆ΕΖ; and let the height ΑΗ
be equal to the perpendicular ΚΘ drawn from the center of the base
Θ, on one side of the cone (such as ∆Ε). I say that the cones are equal.
92
9780306815805_04.qxd 8/22/07 9:13 AM Page 93
The proposition makes its claim in terms of the diagram. This is

the only place where the points and lines of the proposition are pro-
vided with meaning.This is done through alphabetic labels in exactly
the same way as we do it today. In fact, in this sense we follow a
Greek invention. (The Chinese had a different method—each line
was labeled as if it had a different color. But then again, their alphabet
is very different.)
Now, one of the most difficult things for a scholar is to notice
those things that are obvious. The things that lie “under our noses”
are often the most difficult to see, but when you notice them they
may be the most rewarding. I had one such moment when consid-
ering such simple passages as these in Greek mathematics. I made this
observation in the first chapter of my PhD thesis and, quite frankly,
it is the one thing most of my peers know about my work. I will
probably die and still be mentioned as “the guy who made that
observation on Greek diagrams,” which I find quite annoying, seeing
that this was just about the first thing I ever did as a scholar. (I do like
to think it was not all downhill from there!) Still, it is an important
observation because it definitely shows that Greek mathematicians
did not work the way modern philosophers and logicians wish that
they had. They most certainly relied on the diagram.
For example, in an expression such as “the cones ΑΒΓ, ∆ΕΖ” we
may easily guess that the points ΑΒΓ, ∆ΕΖ each stand for the vertices
of a triangle cutting through the cone (see fig. 4.4). But how are we
to know the individual distribution of the letters? In each cone, two
letters must stand on the base and one on the top, but which is
which? This is what makes this observation so difficult.Visual infor-
mation is so powerful that the moment we are in front of a diagram
we immediately “read off” the information and establish that ΒΓ, ∆Ζ
are bases and Α, Ε are tops; and we even fail to notice that the text said
no such thing. In fact, this is the general rule throughout Greek math-
ematics: the identity of objects is not established by the words but by
the diagrams. The diagrams are there not as some kind of illustration
to make the reading experience more pleasant; the diagrams are there
93
9780306815805_04.qxd 8/22/07 9:13 AM Page 94
to provide us with the most basic information. They tell us the who’s
who of the proposition—which letter stands next to which object.
Ancient diagrams are not illustrative, they are informative; they con-
stitute part of the logic of the proposition. And so, Greek science was
a visual science.
How did it come about, then, that Greek mathematicians did not
make trivial mistakes based on the information in the diagram? How
did they keep their logic perfect? The reason has to do with a very
special interface used in Greek mathematics: the subtle, clever way in
which diagrams are used.
The Sands of Syracuse

What did Archimedes’ diagrams look like? As we can see, following
Will’s explanation of the history of manuscripts, this is a question for
which we have only very indirect evidence. The earliest evidence we
have, in fact, is in the Palimpsest itself. One’s first reaction might be
that of despair. If our evidence is so far from the original, what chance
do we have of ever getting there? How can we realistically hope to
know what ancient diagrams looked like? Indeed, this is a difficult
question. In principle, nothing guarantees that we can answer it. It
could be that medieval scribes simply invented their own diagrams
instead of faithfully copying their ancient sources. After all, modern
editors very clearly do just that: they invent their own diagrams.When
I set out on my study of the medieval diagrams of Archimedes, I
couldn’t tell whether the medieval scribes did the same or not. My
greatest fear was that I would get to Paris, Rome,Venice, and Florence,
and each time, open an old book and find a completely different
diagram. If this had been the case, my conclusion would have been
that the ancient diagrams just could not be reconstructed.
Instead, city after city, page after page, and diagram after diagram—
I opened books and they all showed effectively the same figure. Errors
crept in here but not there. Corrections were made in some manu-
94
9780306815805_04.qxd 8/22/07 9:13 AM Page 95
scripts, but not in others (suggesting the original might have con-
tained an error detected by some scribes). But it was clear that the
diagrams were related. They were copied, not invented. In short, one
could apply the philological method. To do so, we take separate manu-
scripts and compare them. If two separate manuscripts possess the
same text or diagram, this means that there must have been a
common source for both. This allows us to go back and infer an
earlier form. And while we can never be sure that this earlier form
dates all the way back to Archimedes, it is still very important to try
to push our evidence back as far as possible.
Now this must be stressed. Readers are sometimes dissapointed to
hear that not all of Archimedes’ works are represented by the
Palimpsest alone. Some of the works are represented in the Palimpsest
and also in the various descendants of Codex A.This is not some kind
of blemish on the value of the Palimpsest. To the contrary, for the
philological method, it is of the utmost importance to have more than
one source.Taken alone, the Palimpsest can only tell us about the year
AD 975. But when we can compare it to other independent medieval
sources, it suddenly tells us much more. Whenever both the
Palimpsest and another independent medieval manuscript tell the
same story, we can push back the dates probably to a source from
Eutocius’ time or before, following which the two traditions
diverged. And this already makes us much closer to the world of
Archimedes himself.
This work is of a complex character. After all, Codex A itself is no
longer extant, so we have to apply the philological method twice
over. My original project of studying the Archimedes manuscripts
involved the descendants of Codex A alone. (The Palimpsest,
remember, was not available when I started my work.) So I looked at
those manuscripts.
There are some 250 figures in the works of Archimedes, but let us
review one example. In fig. 4.5, we can see the various variants of the
diagram for Sphere and Cylinder I, proposition 38. On the basis of
95
B B
H
9780306815805_04.qxd
Z H
Z
A A
8/22/07
E K E
K
9:13 AM
96
Page 96
B B
H
H
Z
Z
A
E E
K K
FIGURE 4.5 Variants of diagram for Sphere and Cylinder 1, proposition 38

9780306815805_04.qxd 8/22/07 9:13 AM Page 97
B
H
FIGURE 4.6 Lost diagram of Codex A
these variants, I was able to reconstruct the lost diagram of Codex A

(fig. 4.6).You can see that the similarity between the various descen-
dants is such as to make my reconstruction quite safe. There is only
one point of detail—two of the codices have a line AB drawn and the
others drop it. Since the line AB is not required by the text, I can guess
that it was in the original figure.The more alert scribes simply did not
copy it.Two of the scribes did not think about what they were doing
and just copied what they had in front of their eyes. For this reason,
they are the more trustworthy witnesses.This is a well-known paradox
of the philological method, known as lectio difficilior (“the more dif-
ficult reading”). A bad piece of text is likely to be the original one.
Now, this in itself involves a fair amount of time travel. The
descendants of Codex A are from the fifteenth and sixteenth cen-
turies, while Codex A itself (like the Palimpsest) probably derived
from the tenth century. The philological method has already gained
us some 500–600 years. I had traveled from the Renaissance to the
Middle Ages. But I wanted to continue my time travel. I needed a
time machine to get me from the Middle Ages to antiquity.
97
9780306815805_04.qxd 8/22/07 9:13 AM Page 98
B
H
FIGURE 4.7 The Palimpsest diagram
The email from Will Noel, inviting me to look at the Palimpsest,

offered me just that. This is why I was so excited. I just had to look at
the diagrams. Here, once again, was a crucial deciding point. Either the
figures would be largely identical to those of Codex A, in which case I
could reconstruct ancient diagrams, or they would be different, in which
case my quest for “the original Archimedes” would abort at around the
year AD 975. Byzantine scribes—it might turn out—merely invented
their diagrams and did not copy them from their originals.What I saw
at Baltimore during my first visit—faintly, but the traces were familiar
enough—was the same figure. Indeed, with digital imaging I recon-
structed it with confidence. I took the reconstructed figure from Codex
A and put it side by side with the figure from the Palimpsest (see fig.
4.7). I could see that they are nearly identical.This discovery was among
the most important ones made through the Palimpsest. It is the corner-
stone to the reconstruction of the figures of Archimedes.
Let us continue to follow the philological method. The Palimpsest
does not include the line AB, and it adds a letter A at the bottom of the
98
9780306815805_04.qxd 8/22/07 9:13 AM Page 99
circle. Now, it is easy to see how a scribe can, in haste, forget to copy a
single letter. I therefore chalk the absent letter A in Codex A down to
scribal error and assume it was present in the common archetype. As
for line AB, this is less clear. It is there in error, and since there is only
a single manuscript bearing this error—Codex A—it may well be only
the error of the scribe of Codex A. Of course, it could also be an earlier
error, that was corrected by the scribe of the Palimpsest, but not by that
of Codex A. But then again, I now know enough about the scribe of
the Palimpsest to tell you that he did not, in general, correct geomet-
rical errors. He clearly understood nothing of mathematics, judging
from some absurd errors he made. In other words, I believe that the line
AB was not there in front of his eyes; it was not part of the common
archetype of the two codices, A and the Palimpsest. And so, having
completed my philological detective work, I argue that the Palimpsest
preserves the ancient diagram of Archimedes’ Sphere and Cylinder I, 38
in fig. 4.7. Thus, we can go all the way back to Syracuse. I now move
on to consider the deep conceptual significance of this.
The Logic of Greek Diagrams

Look at the diagram again. I just noted that I believe it is identical to
the one drawn on the sands of Syracuse by Archimedes himself. And
I believe it represents the most crucial fact about Greek diagrams—
one that holds the key to their great success as cognitive and logical
tools. It explains why ancient diagrams did, indeed, contribute to the
proof, against everything said by modern philosophers and logicians.
I need first to tell you something about the lines ΑΖΗΒΘΛΓ. In
the diagram as it stands, the lines appear like a sequence of arcs, very
much like the rounded edges of the drum of an ancient column. But
what do they represent, geometrically? They represent a polygon—a
sequence of straight lines. Indeed, in figure 4.8 you can see how the
modern editor has chosen to represent the same figure. Instead of arcs
he has used straight lines. He has preferred to call a spade a spade. If
it’s a polygon, let it appear like a polygon. This is not so with
99
9780306815805_04.qxd 8/22/07 9:13 AM Page 100
B
H
FIGURE 4.8 A modern editor’s representation of the same figure
Archimedes, whose position appears to be that one may well draw a

series of circular arcs to show a polygon—who cares how it looks?
This is not an isolated example.We can compare the diagrams that
we could reconstruct, for example, for proposition 30 (see fig. 4.9).
There are altogether fourteen examples of the same type of figure in
this treatise.The circular arcs form a principle of the drawings in this
treatise. And it is of deep significance.
First of all, I believe no one would dare introduce such a radical
convention against the manuscript authority. Suppose you were a
scribe and paid to copy diagrams from the original. The original has
polygons. Well, you would copy them as polygons. You would not
invent circular arcs instead. And this reason—that no one would
introduce such a convention against the source—can be repeated
again and again for each stage of the transmission. The only way to
account for such a convention is to assume that it is due to the author
himself. And so, this convention brings us to the shores of Syracuse—
face to face with Archimedes. Let me confess: this thought does inspire
me with awe. There is something particularly “tangible” about dia-
grams. Words are conceptual, but drawings are physical—they are
bodily.This is how he, Archimedes, traced his figure, turning a stick in
his hand. If indeed I am right, and I have succeeded in reconstructing
100
9780306815805_04.qxd 8/22/07 9:13 AM Page 101
Z
K
M
B
A
E
FIGURE 4.9 Diagram for proposition 30
the diagrams of Archimedes, then I have reconstructed some kind of

extension of his body. Those are the traces left by him, personally.
And now, we progress to the next conceptual point.What does the
convention of representing polygons by circular arcs mean? It is part
of a very wide phenomenon I can identify in the diagrams of
Archimedes (and in other diagrams in medieval manuscripts).
Namely, the diagrams are non-pictorial.You draw a polygon, but you
do not make your drawing look like one. Instead of being a picture,
the ancient diagram is a schematic representation.
Here is another example, once again using figures from both the
Palimpsest and the modern edition. This time we need to look at a
figure for which the only contemporary evidence is from the
Palimpsest—the first figure of the Method. It is in some ways the most
important piece of visual evidence contained in the Palimpsest. Of
course, since we only have a single source, we can no longer apply
the philological method. We cannot compare this figure to those of
other medieval manuscripts and in this way derive an original source.
However, by looking elsewhere, we have already gained confidence
101
9780306815805_04.qxd 8/22/07 9:13 AM Page 102
T
Z
H M
E
K
X
N
B
O
A ∆
A modern illustration
M
E
T
K
N
O
B
H
A ∆
The Palimpsest diagram
FIGURE 4.10 Archimedes’ Method 1
in the Palimpsest.We have seen that, whenever it can be compared to

other sources, the Palimpsest contains diagrams that appear to be
authentic. And so our best guess is that the diagram here, too, is close
to Archimedes’ original spirit.
102
9780306815805_04.qxd 8/22/07 9:13 AM Page 103
Finally, we need to compare the two drawings in figure 4.10. The

modern one is “correct.” The lines ΤΗ, ΖΑ, etc. should indeed all be
exactly parallel; line ΖΑ should be exactly bisected at point Κ; and
ΘΓ should be exactly bisected at the same point Κ. The curve ΑΒΓ
should indeed be a subtly curved, parabolic segment. In short, it is a
pictorial diagram—one that is true to the object. The figure of the
Palimpsest, on the other hand, is schematic.The line ΖΑ is not exactly
bisected at point Κ, nor is line ΘΓ.The curve ΑΒΓ is drawn freehand
as a kind of circular arc. Such are schematic drawings: they merely
suggest the object without drawing it precisely at all.
Why did Greek mathematicians produce non-pictorial drawings?
Why did they find the schematic satisfactory? Now, do not think for
a moment that this was because ancient draftsmen were in any sense
deficient. They could very well draw spectacular pictorial represen-
tations. The great discoveries of the Renaissance masters—per-
spective and illusion—were already made in antiquity. We know this
mostly in a roundabout way. When the Romans pillaged such cities
as Syracuse, they gained an appetite for Greek art, and they did their
best to imitate it. In such lost cities as Pompeii, wall paintings tell us
a lot about the level of ancient art, and they show a clear under-
standing of the geometrical principles of draftsmanship. Take for
instance the rendering of depth and foreshortening in figure 4.11.
The vanishing points are in place; the illusion is compelling. On a
wall in Pompeii—and there were hundreds of such walls in Pompeii
alone—one can admire the Greek understanding of the optical
principles of painting. And there is no doubt that such an optical
understanding existed in antiquity.We have extant several treatises in
optics, one of which by Euclid even contains a theorem specifically
on pictorial foreshortening—a wheel, seen from the side, looks like
an ellipse and not like a circle! In other words, the wheels in the wall
painting from Pompeii go back to knowledge shared by Euclid
himself.
Yet, paradoxically, nothing of this splendid draftsmanship is in evi-
dence in Greek mathematical diagrams. Greek mathematicians chose
103
9780306815805_04.qxd 8/22/07 9:13 AM Page 104
FIGURE 4.11
to avoid the pictorial on purpose and instead preferred “free,”

schematic figures that do not represent their object. Why is that?
The reason is that those strange, counter-intuitive diagrams were
the solution found by Greek mathematicians to the philosophical
problem of using diagrams within proofs.This is a subtle, deep point.
It deserves our close attention and admiration.
Remember what the philosophical trouble was with diagrams:
namely, that they were particular.Your objective is to make a general
point about triangles in general (how are triangles in general
measured?). However, you can’t draw a triangle in general; you must
draw a particular triangle. If you happen to draw a right angle, then
you might be misled into believing that the area of a triangle, in
general, is the product of the smaller sides divided by two.You come
to rely on the particular properties of that particular diagram.
But is it really necessarily the case that a particular diagram sug-
gests a particular property? This is the subtle point. I could draw a
green triangle, a blue circle, and a red square. But if I wanted to make
a geometrical argument about them, I wouldn’t refer to their color.
In the Western tradition, we do not think of colors as geometrically
significant.The color appears to be merely incidental; it is not part of
the drawing at all as far as the geometry is concerned. It is there
merely because it is impossible to draw a triangle without choosing
some kind of color. Usually, of course, we use black. But this does not
make our geometry “the geometry of black figures.” Color is simply
irrelevant.
104
9780306815805_04.qxd 8/22/07 9:13 AM Page 105
Now imagine a tradition where the same is true for such prop-
erties as the size of an angle. So that, for instance, a polygon may
be represented by a series of circular arcs and no one thinks that
there is anything wrong. Because, you see, such properties as the
precise angles are simply irrelevant. This is not what a geometrical
figure represents. The precise angles are rather like the color. So
that, when we draw a right-angled green triangle it is no more
right-angled than it is green. Of course, it happens to be right-
angled just as it happens to be green, but both the color and the
precise angle are irrelevant and are discarded by the sophisticated
reader. Only a naive child would see a “green” triangle. And only
a naive modern reader—untrained in ancient diagrams—would see
a right-angled triangle.
To put this in the most general terms: ancient diagrams are
schematic, and in this way they represent the broader, topological fea-
tures of a geometrical object. Those features are indeed general and
reliable; a diagram represents them just as well as language represents
them. And so, ancient diagrams can form part of the logic of an
argument which is perfectly valid.
We have learned, therefore, something crucial and surprising about
Archimedes’ thought process, about his interfaces. He essentially
relied on the visual; he used schematic diagrams that can be used in
perfect logical rigor without danger of error based on visual evi-
dence. When Archimedes gazed at his diagrams along the Syracusan
seashore, he saw figures largely similar to those that we can reproduce
today based on the Palimpsest. And I know that what he saw there
was a crucial part of his thought process—one of the most basic tools
that made Greek science so successful.
Mathematics Is Beautiful
It was for good reason, then, that Archimedes had a diagram put on
his tomb. His reasoning inherently involved diagrams. And those dia-
grams were used in a clever, subtle way—very different from that of
105
9780306815805_04.qxd 8/22/07 9:13 AM Page 106
FIGURE 4.12
modern illustrations—so that they could serve as part of the logic of

the argument.
Indeed, I think I may have a guess concerning the shape of
Archimedes’ tomb. I think it was very simple. Greek geometrical dia-
grams avoid the complex effects of perspective and three-dimensional
illusion. How to represent a sphere and a cylinder, then? Simply by a
square enclosing a circle. I believe this may have been all that there
was (perhaps with the statement, inscribed underneath, that the
cylinder was one and a half the sphere). A simple, austere figure. The
ancients often inscribed epigrams—short, suggestive poems of
farewell and regret—upon their tombs.This diagram served as such a
succinct, effective visual epigram. Perhaps it was something such as
figure 4.12.
Beautiful visual epigram that it was, it would have been resonant
with meaning. It would have suggested many other objects, many
other discoveries. The same figure of a square enclosing a circle
would first of all bring to mind Archimedes’ achievement of the
measurement of a circle—his remarkable approximation of Pi. And
the square and the circle could equally have suggested rectangles and
parabolas, referring to Archimedes’ many important discoveries con-
cerning parabolas and other conic sections in the Quadrature of the
106
9780306815805_04.qxd 8/22/07 9:13 AM Page 107
Parabola, On Conoids and Spheroids, On Floating Bodies and, of course,

in the Method itself. Indeed, a circle within a square would quite
directly represent the theme common to most of Archimedes’
works—his obsession: measuring curved objects. Of course, I offer
merely a guess.Yet I find this reconstruction of the tomb attractive. It
combines simplicity of form and complexity of meaning, a visual
epigram and a work appropriate to the subtle storytelling genius of
Archimedes. It is in this sense that Archimedes’ science was beautiful.
Of course this is not the only kind of beauty one can imagine.
Greek mathematical diagrams were austere. Other ancient pictures—
we can once again remind ourselves of Pompeii (see fig. 4.11)—were
quite the opposite. The beauty of Pompeian paintings was lavish—as
no doubt was that of many mansions in Syracuse in the year 212 BC.
Nor is mathematics necessarily austere. In the seventeenth century,
for instance, Archimedes took a very different form. Rivault’s edition
of Archimedes—incidentally the one used by Newton—was pro-
duced in Paris in 1615, and it represented the rich tastes of French
monarchs (to whom it was dedicated). The figures of the sphere and
the cylinder were lavishly executed with the aid of three-dimensional
perspective. The images of Rivault are ravishing, but they have little
to do with the mathematical significance of the works of Archimedes.
Indeed, by suggesting that diagrams are precise illustrations, Rivault
destroys the specific achievement of ancient diagrams—their austere,
abstract precision as topological, schematic drawings.
The beauty inherent in such austere drawings can be quite com-
pelling, purely on visual grounds. Archimedes’ study Spiral Lines is
very markedly visual in this respect. Almost all of its figures are
arresting and one has the impression that Archimedes studies the
spiral partly because of an aesthetic, visual fascination. The
Palimpsest figure of proposition 21 (see fig. 4.13) is one of the most
beautiful. (It is, of course, partly obscured by the prayer-book
writing.) It is deeply austere, deeply non-pictorial. Look carefully,
and you will notice that the spiral is not drawn as a true spiral,
smoothly curved, but rather as a sequence of arcs of different circles.
107
9780306815805_04.qxd 8/22/07 9:13 AM Page 108
FIGURE 4.13
The small straight lines are especially telling. They form, indeed, the
exact analogue to the curved polygons we saw in Sphere and
Cylinder. Once again, we see the non-pictorial character of Greek
diagrams. Those small straight lines each stand for a small arc. (In
Sphere and Cylinder arcs are drawn to represent straight lines, and in
Spiral Lines straight lines are drawn to represent arcs.)
This, of course, is Archimedes in the year AD 975. It is largely
faithful to the ancient Archimedes, I believe, not only in its diagrams,
but in its entire visual impact. The narrow columns, for instance, are
significant: they hark back to the writing on papyrus rolls (which
were written in a sequence of very narrow columns). As mentioned,
we have no exact parallel to compare them with from antiquity itself,
but there are, of course, a number of scientific works that have sur-
vived in papyrus form. They are not as important as those by
Archimedes or by Euclid, but still they tell us a lot about the
appearance of ancient science.The earliest of them all is a very minor
108
9780306815805_04.qxd 8/22/07 9:13 AM Page 109
astronomical text known as the Ars Eudoxi, which is extant in Paris.

It is kept at the Louvre Museum, where it is known simply as “Pap.
Gr. 1,” i.e. Greek papyrus number one. It is a very ancient piece of
papyrus indeed—from the late third century BC. Possibly, this piece of
papyrus could have been inscribed while Archimedes was still alive.
Its narrow columns, rough, schematic figures, and clean, yet unorna-
mented, writing are all suggestive of what we have since learned
about the manuscript tradition of Archimedes himself. I believe this
is what ancient science looked like.
It would not have been in book form, of course. As Will Noel has
mentioned, this is the major difference from the Palimpsest. Ancient
writings took the form of a roll, not a book. Instead of leafing
through Archimedes, you would roll him out. Will has pointed out
how inconvenient this is for extracting information from, say, a dic-
tionary. But I think the roll is, in fact, quite handy for the continuous
reading of geometry.The reason Will is right is that, at about the time
people were making the transition from roll to codex, they were also
making the transition from a culture where continuous reading mat-
tered most, to one where extraction of information was paramount.
This was the world of the ultimate reference work—the Bible.
The roll is more convenient for geometry, as the following con-
sideration suggests. Have you ever read a piece of geometry that
extended beyond a couple of pages? You will recall the experience,
then, of leafing back and forth while you were reading, from text to
diagram and back, forgetting the diagram while reading the text and
then forgetting the text while reading the diagram. The roll in this
respect is a much better interface. (I say, books on geometry should
be printed on rolls!) You roll out the book so that you have in front
of your eyes, perfectly, the entire text together with the figure. Such
was the ancient mathematical roll—a polished piece of spare design.
Think of it as the equivalent of an elegant Italian coffee machine—
simple lines joined together to make a perfect product.
The simplicity would be not only in the drawings but also in the
writing itself. This is a major consideration in the history of
109
9780306815805_04.qxd 8/22/07 9:13 AM Page 110
writing—not only for mathematics but for any writing whatsoever.

In time, writing diversifies. Ancient writing was simpler than ours.
Instead of the many fonts we use, and especially instead of the
division we have between uppercase and lowercase characters, the
ancients knew just one font, just one case—effectively, the upper one.
THIS IS WHAT ANCIENT WRITING LOOKED LIKE. This is
what the Ars Eudoxi looks like, and this is how Archimedes’ original
writings looked. Which is simple and elegant, in its own way.
As writing gets diversified, the interface changes. In the Middle
Ages—i.e., in the Palimpsest—the writing used to copy Archimedes
would already have been different. And this, too, is important for the
history of mathematical interfaces. I have already said that, today, the
emblem of mathematics is the equation, the arrangement of symbols.
And this is the product of a long historical process whose roots lie in
the Middle Ages. Once again, let us turn to the Archimedes Palimpsest.
The Medieval Origins of Mathematical Symbols

The Palimpsest is an important piece of evidence not only for the
year 225 BC but also for the year AD 975. The scribe who copied the
works of Archimedes perhaps did not do much to advance the con-
cepts of mathematics. He was no Archimedes. In fact, I am sure the
scribe knew no mathematics at all. But he did make a contribution
to the history of the interfaces of mathematics. His choices in the
manner of writing down the words, arranging them on the page, and
putting the works together all made a contribution to the way
Archimedes would be read by later mathematicians. This is true
throughout history, all the way from Archimedes in Syracuse down to
the present day. Scribes, typesetters, and publishers make a silent con-
tribution to the history of science, sometimes as important as that of
the scientists themselves.
Silently, scribes have invented the mathematical symbol. By doing
so, they paved the way for the modern equation—the most powerful
tool of modern science. While faithfully copying Archimedes’ dia-
110
9780306815805_04.qxd 8/22/07 9:13 AM Page 111
FIGURE 4.14 Sphere and Cylinder II: A phrase from proposition 2
grams, the scribe from the year 975 AD was already preparing the way
for the equations of today’s science.
Let us look at a passage from the Palimpsest. Figure 4.14 is from
the second proposition of the second book on Sphere and Cylinder.
Just as we saw with the Greek diagrams, the writing in the original
Greek produced by Archimedes had no frills at all. ARCHIMEDES
WROTE LIKE THIS, or, more precisely, ARCHIMEDESWROTE-
LIKETHIS (word division, too, is a medieval invention). In particular,
Archimedes’ text had no abbreviations: he spelled every word out
fully.
A scribe’s work is very tedious—copying, word after word, char-
acter after character. It just makes so much more sense to abbreviate.
If a word is repeated often, why not invent a symbol to represent this
word directly, instead of copying it again and again? Of course there
are aesthetic disadvantages to it.With too many abbreviations the text
may no longer look like Greek and begin to look like stenography.
With a work of poetry, for example, produced for a high price, you
would not use many abbreviations. But for a technical work, such as
111
9780306815805_04.qxd 8/22/07 9:13 AM Page 112
mathematics, which was probably not very highly paid, you might.
The Archimedes Palimpsest is an example of a fine, polished piece of
scribal craftsmanship, but it is not a luxury manuscript. No one would
have stopped the scribe from using scribal abbreviations.
And so we go back to the text of Sphere and Cylinder. Here is a
translation of his text:
(1) Therefore as the [line] ΚΘ to the [line] ΘΕ, the [line] ΘΕ
to [the line] ΕΓ, and therefore as the [square] on Κ∆ to the
[rectangle contained] by ΚΘ∆, the [square] on ΑΓ to the
[rectangle contained] by ΑΕΓ
You will notice first of all that my English translation takes up
much more space than the Greek text in the manuscript. There are
two reasons for this, one due to Archimedes, the other due to the
medieval scribes. The first is the square brackets. Archimedes did not
write in such words as “line,” “square,” “rectangle,” letting the
reader infer them from the context. In this way, he was capable of
writing a very spare text. The language used by Archimedes was a
piece of polished, minimalist design on par with his skeletal, polished
diagrams. He used very few words, because the readers already knew
what kinds of things he was talking about (just as the readers could
“read” the minimalist diagrams correctly, because they could under-
stand their nature as mathematical diagrams). Since Archimedes used
only uppercase letters and no word division, his text looked like this:
(2)THEREFOREASTHEKΘTOTHEΘETHEΘETOEΓ
ANDTHEREFOREASTHEONΚ∆TOTHEBYΚΘ∆
THEONΑΓTOTHEBYΑΕΓ
This, indeed, is somewhat challenging as an interface.
Medieval scribes, at this point, took some crucial steps in the
invention of more effective typesetting interfaces. The use of several
cases—an uppercase alongside a lowercase—is of great value. It allows
us to separate the letters referring to the diagram (which remain in
uppercase) from the rest (which is now in lowercase). Word division
112
9780306815805_04.qxd 8/22/07 9:13 AM Page 113
was another major invention.The two together bring us to a text that

is more familiar to us:
(3) Therefore as the ΚΘ to the ΘΕ, the ΘΕ to ΕΓ, and therefore
as the on Κ∆ to the by ΚΗ∆, the on ΑΓ to the by ΑΕΓ
At this point the medieval scribes introduced yet another major
invention—their own contribution to making the text compact—
abbreviations.They did not do this because of any deep, sophisticated
mathematical reason.They did this because they were lazy. Instead of
copying out the word pros (the Greek for “to’) time and time again,
they simply inserted a symbol that looked somewhat like our capital
sigma, Σ.There are other abbreviations or symbols, such as for “as” (a
bit like ω), for “and” (a bit like Κ), and “therefore” (a bit like ε), all
marked in figure 4.14. So this is what the text finally looks like:
(4) ε ω the ΚΘ Σ the ΘΕ, the ΘΕ Σ ΕΓ, Κ ε ω the on Κ∆ Σ

the by ΚΘ∆, the on ΑΓ Σ the by ΑΕΓ
Note also that the Greek word “the” can often be written out in
one or two characters (it changes according to gender and case), instead
of the three characters used in the English “the.” I represent this by an
abbreviation, t’, to allow us to get the full flavor of the original writing:
(5) ε ω t’ ΚΘ Σ t’ ΘΕ, t’ ΘΕ Σ ΕΓ, Κ e ω t’ on Κ∆ Σt’ by

ΚΘ∆, t’ on ΑΓ Σt’ by ΑΕΓ
This, too, may well appear rather confusing, and you may even
prefer the full form of example (1) above to the abbreviations in
example (5).You may feel that example (5) is written out in hiero-
glyphics. But this is all a matter of habit: one simply needs to learn
this particular notation just the way we learn our own modern nota-
tions. For instance, a modern mathematician may well put down the
same text as follows:
(6) ΚΘ:ΘΕ::ΘΕ:ΕΓ
→ Κ∆2:ΚΘ∗Θ∆::ΑΓ2:ΑΓ∗ΕΓ
113
9780306815805_04.qxd 8/22/07 9:13 AM Page 114
which is exactly as hieroglyphic as example (5). Archimedes himself

would have had no idea what example (6) was talking about, just as
a modern reader, faced with example (5), has no idea what it’s about.
One has to learn the notation and then the hieroglyphs make perfect
sense as a symbolic rendering of the contents.
For this is the deep point. All the examples—from (1) to (6)—
contain exactly the same meaning, merely with a change of pack-
aging. The difference is in the interface. But what a difference an
interface makes! Indeed, the invention of abbreviated notations is one
of the key steps in the growth of modern science.
The history of this invention still needs to be charted. Scholars
have only very recently started to look at medieval manuscripts, not
as containers of information about antiquity, but as interesting docu-
ments in their own right. How did scribes come to invent a system
such as that in example (5)? We still do not know the full answer; we
are still collecting the evidence. As the earliest extant manuscript of
Archimedes, the Palimpsest will form one of the key pieces of evi-
dence for this research.
The broad outlines of this history—the history of the scientific
interface—are, however, clear. The transition was made from the
science of diagrams to the science of equations. Indeed, these may be
seen as two different ways of utilizing human visual skills within this
highly conceptual field of mathematical thought. From one kind of
visual science—the Greek one based on the diagram—we have made
the transition to another kind of visual science—the modern one
based on the symbol and the equation.
The Archimedes Palimpsest stands midway: as the best evidence we
can gain, indirectly, on the old science of diagrams; and as a major piece
of evidence for the new (nascent) science of symbols and equations.
The Mathematical Experience

Everything I have referred to—the nature of the mathematical
diagram, the beauty of the mathematical page, the invention of math-
114
9780306815805_04.qxd 8/22/07 9:13 AM Page 115
ematical symbolism—leads to a single point: mathematics is a matter

of experience. Of course, mathematics is a highly conceptual, abstract
discipline. But even an abstract content has to be mastered by a
human person, somehow. It has to be sensed through the eyes. As
humans, we are capable of understanding abstract concepts, but we
can only understand them via our experience.The most abstract con-
cepts must have some sensual packaging, in the sounds of language
and in the artifacts of vision. For humans, to understand is, first of all,
to see and hear.
In recent decades, this is the emerging consensus among philoso-
phers, logicians, historians, and cognitive scientists: cognition and
logic, the abstract and the concrete, are ultimately inseparable.Which,
in a sense, is something paleographers—the scholars of ancient
writing—knew all along.
In the study of ancient manuscripts one is used to questions like:
How is the text written? What are the visual tools invented by the
scribe? How is the page meant to work? When studying a manu-
script, one is led to a study of both content and form.The ideas con-
veyed by the text may be abstract, but their physical form is not.This
is indeed a physical object—as Will Noel calls it, Archimedes’ brain
in a box. This is the purpose of it all: by the study of the cognitive
history of diagrams and symbols and of pages and manuscripts, we
may gain an understanding of Archimedes’ own brain as it worked
back then, at Syracuse.
Only, when I met Will Noel in the spring of 1999, one could
barely glimpse any of this evidence.The manuscript made in 975 AD
had been nearly obliterated by the millennium separating us from
Byzantium’s heyday. It is time to rejoin John Dean and Will Noel in
their journey across the Mediterranean in order to understand how
this came to pass—how this manuscript was changed beyond recog-
nition; and yet, time and again, managed, against the odds, to survive.
115
9780306815805_04.qxd 8/22/07 9:13 AM Page 116

9780306815805_05.qxd 8/22/07 9:18 AM Page 117
5
The Great Race, Part II
The History of the Palimpsest

Disaster Strikes
ack in Constantinople, John Dean and I climbed up the Galata
B Tower and looked out. Beyond the Golden Horn, the glorious
panorama of Constantinople was spread before us. Hagia Sophia and
the Blue Mosque dominated the view. The mosque was a reminder
that Constantinople fell to the Ottoman Turks in 1453. This is often
heralded as a great tragedy, but the really disastrous sack of
Constantinople had already happened 250 years earlier. It was perpe-
trated by Christians from Western Europe.
In 1204, the Fourth Crusade, sanctioned by Pope Innocent III, had
to get from Europe to Egypt and from there, in theory, to the Holy
Land.The problem was how to get to Egypt.The Doge of Venice was
prepared to provide a fleet for 4,000 knights, 9,000 squires, and
20,000 foot soldiers, but at the price of 86,000 marks. The crusaders
agreed, but they were 34,000 marks short when they were ready to
sail. So they agreed to recapture the Dalmatian city of Zara for the
Venetians and their portion of the loot would make up the dif-
ference. The crusaders trashed Zara, but after they had pillaged the
town their loot was still not enough to pay the debt. The crusaders
could not, in honor, default on their debt to the Doge. There was an
imperative to recover it. How to do this? Politics supplied the answer.
Isaac II, Emperor of Constantinople, had been ousted by Alexius II in
1195, blinded and thrown in a dungeon. Isaac’s daughter was married
to Philip of Swabia, and his son, Alexius Angelus, was also at Philip’s
117
9780306815805_05.qxd 8/22/07 9:18 AM Page 118
court. Alexius Angelus agreed to pay the crusaders and the Doge of
Venice 200,000 marks if they would install him on the throne of
Constantinople. The Pope would be happy because Alexius Angelus
agreed that the city would become Catholic; the Doge would be
happy because he would get his money and trading privileges, and
the powerful Philip of Swabia would have a puppet on the throne of
Constantinople. Even modern politics doesn’t get much grubbier
than this.
The realities of medieval conquest do. The crusaders succeeded in
toppling Alexius II. Alexius Angelus was made co-emperor of
Constantinople with his father Isaac II. But with their puppet in
place, the crusaders’ debt still remained. Constantinople was in no
position to pay the money that Alexius had promised.While the cru-
saders were waiting for their money, a few of them started attacking
a mosque. A fire broke out in the chaos that followed. It spread
quickly and soon great tracts of the city stood in flames. The fire
lasted for eight days killing hundreds and destroying a strip three
miles wide running right through the middle of the ancient city. Still
the money was not forthcoming. Alexius, not surprisingly, lost the
support of Constantinople’s beleaguered inhabitants. He was
strangled and his father Isaac II died of grief. Hostilities broke out
again. On Monday, April 12, 1204, the crusaders breached the ancient
walls of Theodosius. That same night another great fire broke out.
The next day Constantinople surrendered. But it was only then that
the full horror—the horror recorded first hand by Nicetas
Choniates—began. As a result, the cash went into the coffers of the
Doge, the city went into the hands of the crusaders, the Catholic faith
was imposed upon the Orthodox, and the classics went up in flames.
This was truly a cataclysmic event for the texts of the ancient
world. The ark of the classics was burned. This is how and when
twenty of the thirty-three historians discussed by Photius disap-
peared. Who knows how many copies of Archimedes’ treatises? The
future of these treatises was not in Constantinople. The copies that
118
9780306815805_05.qxd 8/22/07 9:18 AM Page 119
T H E H I S T O RY O F T H E PA L I M P S E S T
survived in the thirteenth century would be found elsewhere.

Codices A, B and C were flotsam upon the waters of the
Mediterranean world. Let’s see where they washed up—first Codices
A and B, and then C, the book on my desk.
Archimedes in Italy
In 1881 a scholar named Valentin Rose came across a manuscript in
the great Vatican Library. It was written by William of Moerbeke, a
Franciscan friar and a great translator of Greek texts including several
works of Aristotle. This particular book was his translation of the
works of Archimedes from the Greek into Latin. He finished writing
the book on Tuesday, December 10, 1269. Since William became a
chaplain and penitentiary of Pope Clement IV at Viterbo, Italy, some
time in the 1260s, and was still there in 1271, he must have translated
Archimedes’ treatises there.
But from which manuscripts did William translate, and where did
he get them? There were two of them, and they were listed in a cat-
alogue of manuscripts belonging to the Pope at Viterbo in 1311.They
were the manuscripts that today we call Codex A and Codex B.
Number 612 was Codex A. Even in 1269 it couldn’t have been in
great shape because it was already missing its cover. In the catalogue,
the codex is recorded as Angevin. This probably means that Charles
I of Anjou gave it to the Pope after the Battle of Benevento in 1266.
Number 608 was Codex B. Since Codex A did not contain Floating
Bodies, we can assume that William must have translated this treatise
from Codex B.
So Codices A and B washed up in Italy. Codex B, however,
didn’t last very long. It has been missing since 1311. Codex A, on
the other hand, became one of the most highly sought after
codices of the Italian Renaissance. In 1450 it was in the possession
of Pope Nicholas V who commissioned Jacopo of Cremona to
translate it again. In 1492 Lorenzo de Medici—Il Magnifico—sent
119
9780306815805_05.qxd 8/22/07 9:18 AM Page 120
Politian on a search for texts that he didn’t have in his own library.
Politian found Codex A in the library of Giorgio Valla in Venice,
and he had a copy of it made. This copy is now housed in
Michelangelo’s architectural masterpiece, the Laurentian Library, in
Florence.Valla thought Codex A was so precious and rare that he
would not permit it to leave his library. He even declined a request
to borrow it from Ercole d’Este, Duke of Ferrara. Alberto Pio of
Carpi bought Giorgio Valla’s library. When Pio died in 1531, the
manuscript became the possession of his nephew Ridolfo Pio, who
died in 1564. No one has seen Codex A since then.
Even though they disappeared, Codices A and B did their job: they
transmitted Archimedes to the modern world. The reception of
Archimedes has been meticulously documented in the monumental
work of scholarship by Marshall Clagett, Archimedes in the Middle Ages.
Whether it was directly through Codex A or through the Latin trans-
lations by William of Moerbeke and Jacopo of Cremona,
Archimedes’ treatises came into the hands of the most talented men
of the Renaissance. The Renaissance was, of course, well disposed to
receiving the works of this great man.
The Archimedes of legend had already become a byword for bril-
liant inventors and mathematicians. Filippo Brunelleschi, for
example, was heralded as a “Second Archimedes” after building the
magnificent dome of Florence Cathedral early in the fifteenth
century. Renaissance figures soon learned that the Archimedes of
the treatises far surpassed the legendary figure. Leon Battista Alberti,
the great Florentine author, architect, and painter knew about
Floating Bodies and utilized it in his exposition of the “Eureka” story.
More impressively, as James Banker showed in 2005, Piero della
Francesca, whose paintings reveal astonishing subtleties of geometry,
had actually transcribed the full text of Jacopo of Cremona’s trans-
lation. Regiomontanus, the German mathematician whose work
was so important to Copernicus, also copied Jacopo’s translation
after the Pope had given it to Cardinal Bessarion. By hook or by
crook the great artistic and mathematical minds of the Renaissance
120
9780306815805_05.qxd 8/22/07 9:18 AM Page 121
got their hands on Archimedes’ treatises. In 1544, the first edition of

the works of Archimedes was printed in Basle. For many of
Archimedes’ treatises, the race against destruction was over, and they
had won. Galileo and Newton would read them, and modern
science would be born.
The Book Leonardo Never Knew

It might strike you that I have neglected one of the greatest minds of
the Renaissance—Leonardo da Vinci. We have established that from
Hero onward, Archimedes was of interest to the foremost mathe-
maticians and architects of the day—intellectuals who had not only
mastered higher mathematics but who also wanted to apply their
knowledge. It is, therefore, no surprise to learn that Leonardo was also
anxious to obtain copies of Archimedes’ works. In his notebook he
writes: “A complete Archimedes is in the hands of the brother of the
Monsignor of Santa Giusta in Rome. He said that it had been given
to his brother who was in Sardinia. It was first in the library of the
Duke of Urbino but was taken away at the time of Duke Valentino.”
Somehow Leonardo must have succeeded in acquiring some
Archimedes manuscripts. His notebooks reveal knowledge of the
Measurement of the Circle, Spiral Lines, Sphere and Cylinder, Floating
Bodies, and Balancing Planes.The last named of these treatises particu-
larly fascinated Leonardo as it dealt with finding centers of gravity.
Leonardo used it to demonstrate how to find the center of gravity of
a triangle. (In the next chapter, with Reviel, you will be doing the
same.) Being Leonardo, however, he did not stop with what he dis-
covered from Archimedes. He used Archimedes’ work as a platform
for his own calculations. In Balancing Planes, Archimedes had only dis-
cussed how to find the center of gravity in plane figures. Leonardo
went beyond this. Applying Archimedes’ techniques, Leonardo
attempted to find the centers of gravity in solid ones, too. He ended
up developing a theorem for finding the center of gravity in a tetra-
hedron. It was a remarkable achievement by this Renaissance giant
121
9780306815805_05.qxd 8/22/07 9:18 AM Page 122
and an example of the way that Renaissance scholars built on the

work of Archimedes.
However, there was one treatise that Leonardo did not know
about. Consequently, he could not have known that Archimedes had
gone way beyond him 1,700 years earlier. In the Method, Archimedes
had already found the centers of gravity for much more complicated
solids than the tetrahedron—solids with curved surfaces. In his letter
to Eratosthenes, Archimedes had calculated the center of gravity for
a paraboloid, a spherical segment, a segment of an ellipsoid, and even
a segment of a hyperboloid. It’s not that Leonardo researched his
subject inadequately; Leonardo couldn’t have known about this text.
It wasn’t part of Codex A or Codex B, the only two Greek manu-
scripts by which Archimedes was known during the Renaissance. It
was part of Codex C. Or rather, it had been.
A Write-off
A thirteenth-century scribe prepared himself for a job. He had gone
through the procedures many times before. He had already prepared
his reed pens, his ruler, and his knife. He sat down in his chair. Beside
him was a small table with an inkwell filled with black ink. He took
the first sheet of parchment from a stack nearby. With a hard point,
he incised lines on the parchment upon which he would shortly
write his letters. The parchment now rested in his lap on top of a
board. In front of the scribe on a stand was the codex that he was
about to copy. He was poised to write. Are you experiencing déjà vu?
Excellent. Look at the scene again. This time we are not particularly
interested in the codex on the stand. It is the parchment that the
scribe is about to use that should be the object of our curiosity. The
parchment was going through a process it had been through before.
You have guessed it, of course.This scribe’s parchment was Codex
C, the Archimedes manuscript from Balancing Planes to Stomachion,
Method included. It had been taken apart and its folios erased of text.
The scribe had more prayers to write than he had Archimedes
122
9780306815805_05.qxd 8/22/07 9:18 AM Page 123
FIGURE 5.1 How to make a palimpsest
parchment to write on, but this didn’t stop him. He simply reused
parchment from other codices as well—at least four of them.
The palimpsesting of Archimedes and of all the other unidentified
texts in the prayer book was a ruthless operation. The manuscripts
were taken from their shelves, their bindings were cut off and dis-
carded, and the stitching between their quires was undone. This was
quick and easy to do. Once the codex was in pieces, the bifolios were
scrubbed with some kind of natural acid. There are no Greek texts
telling us how this was done, but Theophilus, writing On Various Arts
in Western Europe in the twelfth century, suggests that by using
orange juice and a sponge it is quite easy to erase the letters perfectly.
There is no doubt that some kind of acidic mixture was used, but the
operation on the Palimpsest was much more severe than the one pre-
scribed by Theophilus. Abigail found holes on the edges of the
Archimedes bifolios that appear to have been made by nails, which
held the pages under tension. This would be consistent with damp
bifolios being tacked down to a board to prevent shrinking as they
dried out. Abigail further noticed that there were scratch marks on
top of the Archimedes text. After the bifolios had dried, it appears
that they were further rubbed with a pumice stone. It’s done.
Archimedes is gone. The skins on which his texts had been written
were removed from their wooden frames and stacked in a corner.
The first thing the scribe did upon picking up an Archimedes
bifolio was to cut it in two down the fold, thereby separating it into
123
9780306815805_05.qxd 8/22/07 9:18 AM Page 124
two folios. He did not trim these folios further and this is lucky for
us; on each surviving folio none of the residual traces of the
Archimedes text were trimmed by the maker of the palimpsest. The
scribe then took the two folios, rotated them ninety degrees, and
folded them in the middle so that they became two nested bifolios in
the prayer book. The folios of the prayer book are therefore exactly
half the size of the original Archimedes folios.
However, when the scribe picked up separate Archimedes bifolios,
the bifolios were already highly disordered, so different bifolios of the
Archimedes text are now found widely separated from each other in
the prayer book. They are also found interspersed with palimpsested
parchment of the other manuscripts that the scribe used. The
Archimedes manuscript formed the overall skeleton of the
Palimpsest; the parchment of other manuscripts fleshed it out.
It was standard practice in making palimpsests for the scribe to cut
the bifolios in half and rotate them, and it made good sense.The great
advantage of the procedure was that the scribe did not have to
contend with the distracting remains of a palimpsested text because
he was writing at right angles to it. It is far easier to write over a text
at right angles than it is to follow its path. Of course, the scribe could
have rotated the scrubbed bifolios without cutting them down the
middle and simply folded them in half the other way. But, the result
would have been an extremely tall, thin, and unwieldy codex. The
procedure followed by our scribe was carefully designed to produce
new codices effectively and economically. For this reason,
palimpsested codices are nearly always half the size of the codices
from which they were made. Of course, since a folio of the
Archimedes manuscript became a bifolio of the prayer book, it could,
and often did, constitute both the first and last folio of a quire in the
new manuscript.As a result, the middle of each of the old Archimedes
folios passes right through the spine of the prayer book.
If he knew what he was about to copy over, the scribe did not even
give it a second thought.The first piece of parchment in his new codex
contained On Floating Bodies. He covered it with a blessing for loaves
124
9780306815805_05.qxd 8/22/07 9:18 AM Page 125
at Easter. Further into the codex, he wrote over a different section with
a prayer for repentance. He wrote over the beginning of the Method
with a prayer of marriage. Over a later section of the Method, he wrote
a prayer recited at the foundation of a church. And, note this, over
Archimedes’ critical proposition 14, he wrote a prayer for the dead.
For a short section of the prayer book our scribe worked with a
colleague. He was probably glad to get the help because it was a long
job. Indeed it was so long that the Archimedes manuscript could not
by itself supply all the necessary parchment, and other codices had to
be used too. Since no one who had investigated the Archimedes
Palimpsest was very interested in these codices, it is entirely appro-
priate that over a folio of one of them the scribes wrote a prayer for
those unreasonably excluded.
It may have made good sense to them, but the scribes of the prayer
book had really stitched up Archimedes. Think about it. If some odd
duck ever wanted, for some strange reason, to read any one of the
palimpsested texts, it would be great fun to watch them try. For
example, if someone was interested in the text of Method, proposition
14, they would have trouble finding it. It starts on column 1 of folio
110 recto of the Palimpsest. To read it, they would need to turn the
codex ninety degrees and read through the prayer book text—the
prayer for the dead—to decipher the erased text beneath.Very soon
they would get stuck, because the column disappears into the gutter.
They would then need to find where it reappears. In this case, it
appears five folios further back on folio 105v. On this folio, they would
not be able to read at least two of the lines of text hidden in the gutter.
If they persisted, they would have to read the second column of text
too. Now things get yet more complicated.They would have to rotate
the codex 180 degrees and read column 1 of folio 110v and then
rotate it 180 degrees to read the bottom of this column on folio 105r.
To finish this folio, they would have to repeat this operation to read
the second column. Having read as much of this folio as they could,
they would then need to find the next folio of Archimedes text. It
could be anywhere in the codex. Actually it is on folio 158, more than
125
9780306815805_05.qxd 8/22/07 9:18 AM Page 126
fifty folios further along in the prayer book. Then the whole process
would start all over again. A truly interactive user experience. But
nobody would ever do this, would they? I mean, it’s a mug’s game.
So what was the name of the Christian scribe who did this? And
what, if any, are the mitigating circumstances that the defense can
summon up before we pass judgment on him for obliterating
Archimedes? Having no idea how to answer these questions, I
roundly condemn an anonymous medieval scribe and move on. The
book moved on, too, and when we can trace it next, it is three
hundred years later and on a different continent.
Buried in the Desert

John Dean and I left Constantinople behind and flew to the Holy
Land. We arrived in Tel Aviv, hired a car, and headed for Jerusalem.
We went to the Wailing Wall, which is the foundation of Solomon’s
Temple. We saw the Dome of the Rock where the prophet
Muhammad ascended to Heaven and met with Allah. We visited the
Church of the Holy Sepulcher where Christ was buried. The next
day we headed south from Jerusalem, through the West Bank security
checkpoint, and into a different world. We drove to Bethlehem and
got lost. We missed our turn, and we couldn’t speak Arabic. John’s
smile was a sign of friendship and my agitation an international signal
of distress. The combination persuaded a very patient Palestinian to
get us back on the right road.
We turned left, to the east, and into the desert. The road ended and
we got out of the car. It was the early evening and the sun was low in
the sky, but it was still extraordinarily dry and hot.To our left a boy on
a donkey passed us, whacking his stick, and driving his goats back to the
village. Apart from the bells around the necks of the animals, it was
silent.We could see for miles across the Judaean desert.The sky was blue
over Jordan; the land, everywhere, was burnt ochre. Below us, about a
mile down a gravel path, I could see two towers. I had seen them before
in a black and white nineteenth-century print above the desk of my
126
9780306815805_05.qxd 8/22/07 9:18 AM Page 127
friend, Patrick Zutshi, in Cambridge.They looked exactly the same, and

I knew them to be the towers of the Monastery of St. Sabas.
The print I knew was by the Royal Academician David Roberts.
He had arrived at the Monastery on Thursday, April 4, 1839 with the
Reverend George Croly. They had initially approached the
monastery from a different angle, from the east. Croly records:
The immediate approach to the convent is striking . . . It was
night when after having descended into the bed of a ravine,
where the Kidron passes to the Dead Sea, and arriving at the
foot of the Mountain of St. Saba, we saw the convent above us,
by the uncertain light of the moon. It looked a lofty and colossal
structure, rising in stories and terraces, one above another, against
the sides of the mountain to its summit, and there crowned with
clouds. An old white-bearded monk, leaning on his staff, was
toiling up the side of the hill leading a long procession of
devotees. Below, apparently growing out of the rock, was a large
palm tree said to have been planted by the hands of the Saint in
the fourth century. History, and probably legend, contributed its
share to the effect. In a chapel behind an iron grating in one of
the grottos was a pile of skulls.The tradition of the convent said
they were those of hermits who, to the amount of several
thousand, had been slaughtered by the Osmanlis [i.e. Ottomans].
We ascended the flight of steps, climbed up a ladder, crept
through a small door only large enough to admit one at a time,
and found ourselves in an antechamber, surrounded by above a
hundred Greek pilgrims . . . It was Passion Week.The monks
receive strangers with courtesy, and they not merely permitted
the artist to sketch their chapel, but as their service was
beginning before he had finished his design, they would not
suffer him to lay aside his pencil.
John and I went through the main door, which was in a round
archway and painted deep blue.We were received warmly by the only
monk in the community of thirteen who admitted to speaking
127
9780306815805_05.qxd 8/22/07 9:18 AM Page 128
English. His given name was Lazarus, and he had come to the
monastery from San Francisco. He showed us around the complex
including the cell of St. Sabas and the chapel of St. Nicholas where
the skulls of the departed members of the community are indeed
housed. It is still the most extraordinary, beautiful, and spiritual place
despite the political upheavals that constantly surround it. Everything
was exactly as Croly had described it. It was as if time had melted
away. Brother Lazarus had found peace at St. Sabas. He missed the
Grateful Dead, but he was reminded of them by the insistent ringing
of the semantron, a crescent-shaped metal bar by which a fellow
monk, even at that moment, called him to prayer. Before he left us,
he pointed to the taller of two towers that crowned the assemblage
of churches and cells. St. Justinian’s Tower, he said, contained the
library. John and I had reached our destination. In 1834, there were
more than 1,000 manuscripts in the library of St. Sabas. One of them,
one of the least prepossessing, was the Archimedes Palimpsest.
We only know that the Archimedes Palimpsest was at St. Sabas
because a Greek scholar, Papadopoulos-Kerameus, described the
manuscript in 1899. He said there was a paper quire in the book,
which was added in the sixteenth century, and in which there was an
inscription indicating that the book belonged to the monastery. The
manuscript doesn’t have this quire any more and this inscription no
longer exists. It is only thanks to Papadopoulos-Kerameus that we
know how the Archimedes Palimpsest survived the centuries.
The Palimpsest contains prayers that the brothers in the monastery
would have used on an almost daily basis. It includes a prayer said
when something unclean falls into a vessel of wine, oil, or honey;
there is St. Gregory’s exorcism for unclean spirits; and John
Chrysostom’s prayer for Holy Communion, to name just a few of
them. The Palimpsest shows every sign of frequent use. The codex is
charred at the edges, as if has been scorched by the desert heat or
even burned in a fire. Many of the folios are covered with wax
droplets, which would have fallen on the manuscript as its prayers
were recited by candlelight. There are many emendations and addi-
128
9780306815805_05.qxd 8/22/07 9:18 AM Page 129
tions to the text and in some places the prayers have been traced over
in order to make them more legible. Moreover, either through
damage or because the prayers were no longer considered relevant,
approximately sixty folios from the manuscript went missing while it
was at St. Sabas. That is about a third of the entire codex.
The Monastery of St. Sabas provided a temporary respite for John
and me and a more permanent sanctuary for Brother Lazarus. But, it
was a tomb for Archimedes. The monks had every reason to read the
prayers in the Palimpsest, but absolutely none to read what was beneath
them. Abstract mathematics is not a priority at St. Sabas. Archimedes
was effectively buried at the monastery for at least three hundred years.
Unlike the texts in Codices A and B, those unique to Codex C
remained unknown to the Renaissance and the Scientific Revolution.
Somehow, like Brother Lazarus’s biblical namesake, the Archimedes of
the Method and Stomachion would have to be raised from the dead.
Signs of Movement
One of the last stops on my journey with John Dean was Lincoln
College, Oxford to see a great scholar and a gentleman, Nigel Wilson. I
have already spoken of Nigel, but it was only when I met him in Oxford
that I got to know him. The first thing that struck me was that he was
honored to meet us. This was only in part impeccable civility. It was
mainly because I bore responsibility for the Palimpsest, work on which
he would later describe as “one of the most fascinating scholarly projects
imaginable.” The Palimpsest really mattered to Nigel and this accounted
for his remarkable patience when we filmed him over and over again in
the college library repeating some of the simplest statements that he has
ever contrived to utter. One sound bite he gave us was: “Constantinople
was the one place in the ancient world with an unbroken tradition of
copying and studying ancient texts.” Another was: “I went to
Cambridge, saw the leaf and said,‘That’s it; that’s Archimedes.’”
In truth, in 1971, acting on the suggestion of his friend G. J.
Toomer of Brown University, Nigel set out from Oxford to
129
9780306815805_05.qxd 8/22/07 9:18 AM Page 130
Cambridge to see a palimpsest fragment containing a mathematical

text that had been catalogued by Pat Easterling. Nigel found it easy
to read and he recognized it as Archimedes based on a technical term
used in the text. It came from Sphere and Cylinder and it fits between
folios 2 and 3 of the Palimpsest.
The fragment was in Cambridge University Library, and it had the
number Add. 1879.23. The University Library logs its acquisitions
and this was one of forty-four fragments that were sold to the library
on Wednesday, February 23, 1876 by the executors of an estate. The
estate had belonged to the German scholar Constantin Tischendorf.
Twenty years earlier, Constantin Tischendorf had made the
greatest manuscript discovery of all time. This was not the
Archimedes Palimpsest. It was the earliest surviving complete copy
of the New Testament, together with substantial portions of the Old
Testament in Greek. It is now known as the Codex Sinaiticus. It was
written between about 330 and 350 AD, and it may be one of the
original fifty copies of the scriptures commissioned by the Roman
Emperor Constantine after his conversion to Christianity. It was
written in the same type of majuscule script in which Isidore of
Miletus would have written.Tischendorf found it in the ancient and
secluded monastery of St. Catherine’s in the Sinai Desert.
Tischendorf negotiated with the monks to borrow the codex and
conveyed it to the Russian Tsar Alexander. In return, the Tsar gave
Tischendorf the title “Von” before his surname, thereby making the
son of a German physician a Russian nobleman, and paid the monks
9,000 rubles for the codex. Good deal.
Tischendorf was, among other things, a very great biblical scholar.
He had no difficulty recognizing the importance of the Codex
Sinaiticus. But why was he in possession of a folio from the
Palimpsest, and how on earth did he get a hold of it? Actually, he
nearly tells us himself. In 1846, he published a book titled Travels in
the East. In it he recounts a visit to the Metochion of the Holy
Sepulcher in Constantinople where he found nothing of particular
130
9780306815805_05.qxd 8/22/07 9:18 AM Page 131
interest except for a palimpsest containing mathematics. In 1899, we

know that the Palimpsest was at precisely this location because
Papadopoulos-Kerameus then catalogued it there. Clearly it was
already in the Metochion in the 1840s and clearly Tischendorf had
come away from his visit with a folio torn from it.
Of course we do not know how the manuscript moved from St.
Sabas back to Constantinople. When John Dean and I had tried to
visit the Metochion of the Holy Sepulcher there was no one to ask.
It was Easter and all the monks had gone to Jerusalem to their
mother institution—the patriarchal Monastery of the Holy
Sepulcher in Jerusalem.The manuscripts of St. Sabas had been incor-
porated into the library of the Greek patriarchate early in the nine-
teenth century. It is not hard to imagine, therefore, the circumstances
in which a useful prayer book ended up back in the city in which it
had been made seven hundred years earlier.
Back from the Dead

The front page of the New York Times on Tuesday, July 16, 1907
records a sensational discovery: Professor Heiberg, from Copenhagen,
had discovered a new Archimedes manuscript in Constantinople. A
certain Professor Schone had brought to Heiberg’s attention the
description of a codex in the 1899 catalogue by Papadopoulos-
Kerameus. Papadopoulos-Kerameus didn’t have tenure and was paid
by the page for his work. Perhaps this is why when he catalogued
manuscript number 355, he not only described the contents of the
prayer book in detail but he also transcribed a section of an erased
text that had been written over. Heiberg recognized the transcribed
erased text as the work of Archimedes. He first tried, through diplo-
matic channels, to have the manuscript sent to Copenhagen, but this
failed. So, during his 1906 summer vacation, he traveled to
Constantinople and met the librarian, Mr. Tsoukaladakis, of the
Metochion who allowed him to study the manuscript. This is when
131
9780306815805_05.qxd 8/22/07 9:18 AM Page 132
he discovered the staggering truth: Heiberg had found a sleeper, con-

taining the unread greatest thoughts of a mathematician of genius.
Heiberg published Archimedes’ letter to Eratosthenes, the Method,
in an academic journal called Hermes. Between 1910 and 1915,
Heiberg completely re-edited the works of Archimedes to incor-
porate his readings from the Palimpsest. His edition is ultimately
based on three codices: Codex A (now lost), which was number 612
in the Pope’s library in 1311; Codex B (now lost), which was number
608, and Codex C (now found), which is the Archimedes Palimpsest.
Heiberg’s publications are the work of an extraordinary scholar,
but they are also the work of a man limited by a number of factors.
First, the physical constraints of the bound prayer book. As we have
seen, the scribes of the prayer book constructed their manuscript in
such a way that two or three lines in the middle of every folio of the
original Archimedes manuscript were hidden from view in the
gutter. In such places, Heiberg simply had to guess what was written.
Secondly, he had to work with the technology of his day; he did not
even use ultraviolet light, which is now a standard procedure for
people reading faint texts. Thirdly, the intellectual framework within
which he operated limited him. Heiberg was a philologist. A philol-
ogist is a lover of language, not of drawings. He paid no attention to
the diagrams in the codex. For his Hermes publication he had a math-
ematical colleague named Zeuthen reconstruct the diagrams from
the Archimedes text. But as Reviel jumped up and down to tell me,
ancient mathematicians didn’t think in text; they thought in dia-
grams. The Palimpsest was the unique source for the diagrams that
Archimedes drew in the sand in the third century BC and they had
never been studied. Finally, Heiberg was only really interested in
Archimedes. He did mention that there were other works in the
Palimpsest. He read just a few words of one of them, but he was only
excited by the palimpsested pages that originally belonged to Codex
C. Despite the work of the great Johan Ludwig Heiberg, there was
still a great deal to be learned from the Archimedes Palimpsest.
132
9780306815805_05.qxd 8/22/07 9:18 AM Page 133
So there was a lot of work that still needed to be done on the

Palimpsest and scholars throughout the twentieth century knew it.
However, they couldn’t go and see it.The Archimedes Palimpsest had
disappeared.
Lost in Paris
By 1938, the Metochion’s manuscripts had been moved to the
National Library of Greece in Athens. It was done under the noses of
the Turkish authorities who had specifically forbidden such exports.
Certainly this was safer than to have the books stay at the Metochion,
because life there had become very unpleasant.
At the end of World War I, an English and French military
presence in Constantinople supported the Sultan of a crippled
Ottoman Empire. Mustafa Kemal—later Ataturk—left the capital and
rallied Turkish nationalists to found the modern state of Turkey. In
1923, the Allies and the Sultan were ousted from Constantinople. In
the process, Ataturk defeated the Greeks who had rashly invaded
Turkey in 1921. In an early example of ethnic cleansing, hundreds of
thousands of Greeks living in Turkey were forcibly transferred to
Greece. In 1925, Ataturk abolished religious orders and hanged the
Greek Patriarch of Constantinople.
It was in this atmosphere that the books in the Metochion were
surreptitiously moved to Athens. There are no records of how it was
done, but it was done very quietly. And the veil of silence that sur-
rounded the Metochion manuscripts in the twenties and thirties must
have been just too tempting for someone, because the Palimpsest was
one of a number of spectacular manuscripts that never made it to
Athens.
The Metochion’s manuscripts that did not make it to Greece are
now in various institutions, including the University of Chicago; the
Cleveland Museum of Art; the Bibliothèque Nationale in France;
Duke University; and the Walters Art Museum. Henry Walters
133
9780306815805_05.qxd 8/22/07 9:18 AM Page 134
bought one—a beautiful Gospel book, now Manuscript W.529. The

Palimpsest was not nearly as beautiful as these books, but someone
had done something about that. In one very striking respect the book
looks very different from the one that Heiberg saw.The four painted
pages that I thought were charming when I first looked at the book
were not there when Heiberg studied it. The catalogue says:
Four leaves, all now detached, are illuminated with full-page
portraits, presumably intended to represent the Evangelists.
Some of the colors look strangely modern . . . Neither
Heiberg nor Papadopoulos-Kerameus in his description refers
to them, so they must be relatively recent, presumably a mis-
guided attempt at the Metochion to embellish the manuscript
and enhance its value in the eyes of a prospective purchaser.
The pictures have been painted over both upper and lower
scripts. All four leaves are listed by Heiberg as containing text
by Archimedes . . .
In other words, the pictures were forgeries. Gold, lead, copper,
barium, zinc, and a whole host of other elements had been plastered
over the flesh and iron that encoded the unique text of the letter that
Archimedes had sent to Eratosthenes! As if the scribe of the prayer
book had not done enough to obliterate Archimedes, his greedy suc-
cessors had poured insult on to injury and painted over his corpse.
The auction catalogue merely says that the book left the
Metochion and ended up in a private collection in France. However,
the court case over the rightful ownership of the manuscript meant
that further explanation had to be given. The result was a number of
documents that Mr. B sent to me after the Palimpsest was left in my
care.The most revealing of them is the sworn affidavit from someone
named Robert Guersan. Robert Guersan, was the son of Anne
Guersan, who had owned the manuscript before the sale. He believed
that his grandfather, Marie Louis Sirieix, had acquired the Palimpsest
in the 1920s and had kept it in his home in Paris.
134
9780306815805_05.qxd 8/22/07 9:18 AM Page 135
Sirieix had served in Greece during World War I and had traveled
in Greece and Turkey in the early 1920s. This was presumably when
he acquired the manuscript. He had lived in Paris, served with dis-
tinction in the French Resistance during World War II, and left for
the South of France in 1947. That is when he left the Palimpsest in
the care of his daughter, who had moved into his apartment. He died
in 1956.
In the 1960s, Anne Guersan began investigating the book she
inherited. She sought the advice of Professor Bollack, a neighbor in
Paris, and Professor Wasserstein from Leicester. In 1970, she left a few
detached leaves of the codex with Father Joseph Paramelle at the
Institut du Recherche et d’Histoire des Textes of the Center
Nationale de la Recherche Scientifique in Paris. At that point, she
knew what she had. In 1971, she took the book to the Etablissement
Mallet “to remove fungus stains from a few of its pages and otherwise
to preserve its condition.” Then she tried selling it. In the 1970s, a
short brochure was produced and it was discreetly offered for private
sale to a number of individuals and institutions. All of them declined.
Anne Guersan finally turned to Felix de Marez Oyens of the
Manuscripts Department at Christie’s.
The Palimpsest arrived on my desk on January 19, 1999 even
before the legal issues, raised at the time of its sale, were resolved.
While John Dean and I were on our jaunt around the
Mediterranean, Christie’s and the patriarchate were still thrashing
things out in court.We had more fun than they did.They did not dis-
agree over the facts. Instead, the case turned on the interpretation of
the law, and Judge Kimba Wood ruled in favor of Christie’s.
According to French law, which she judged applied in this case, if
Anne Guersan had owned the codex publicly, peacefully, continu-
ously, and unambiguously for thirty years, she had the right to sell it.
The burden of proof rested with the patriarchate’s lawyer to demon-
strate that she had not owned it in these circumstances and the patri-
archate’s lawyer did not come up with this evidence. Judge Wood also
135
9780306815805_05.qxd 8/22/07 9:18 AM Page 136
noted that, should New York law be deemed to apply, she would still
rule in favor of Christie’s, but on a different principle—the principle
of laches. Generally, the principle of laches is applied when it is clear
that a plaintiff unreasonably delayed in initiating an action and a
defendant was unfairly prejudiced by the delay. Judge Wood must
have thought that bringing an action the night before the auction
was an unreasonable delay. The case was finally dismissed on
Wednesday, August 18, 1999 by which time the exhibition was
already open at the Walters.
I had learned a lot in five months. Sure, there were plenty of holes
in the Palimpsest story, though I knew enough to think that it was
tragic. I did not know the name of the scribe who had obliterated the
Archimedes text, and I did not know when, why, or where he did it.
But John had enough to make a movie, and I had enough for an
exhibition. I could give it a happy ending by promising to reveal the
erased texts despite all that had happened to the manuscript. “Eureka:
The Archimedes Palimpsest” opened on Sunday, June 20, 1999 and
traveled to the Field Museum in Chicago in the fall. The Palimpsest
was opened to a folio on which visitors could just barely make out
the diagram that accompanied proposition 1 of Archimedes’ letter to
Eratosthenes.
The exhibition started with John Dean’s movie. The movie tells a
strange story—ideas originate in the head of a man living on a tri-
angle in the middle of the Mediterranean in the third century BC.
These ideas are uniquely preserved today on a manuscript written in
Constantinople twelve hundred years later. They survive the rise and
fall of empires, the sacking of cities, and numerous changes in writing
technology. And even though these ideas are scraped off and written
over, they are still there. It is an astonishing journey. The letter
beginning “Archimedes to Eratosthenes, Greetings,” starts right at
the top of folio 46r column 2 of Codex C and there alone.
Delightfully, you can see the beginning of the letter and the name
136
9780306815805_05.qxd 8/22/07 9:18 AM Page 137
“Archimedes” quite clearly before the column gets swallowed up by

the text of the prayer book.
FIGURE 5.2 The beginning of the Method
137
9780306815805_05.qxd 8/22/07 9:18 AM Page 138

9780306815805_06.qxd 8/22/07 9:22 AM Page 139
6
Archimedes’ Method, 1999
or The Making of Science

was there in June 1999. What a sensation it was to see the
I Palimpsest opened to the first diagram of the Method! I had always
dreamed of seeing this diagram. That it was partially hidden from
sight, disappearing into the gutter, only added to its mystery. I saw all
the visitors coming into the museum to gape at this modest-looking
page, and I knew that they were looking at the only surviving evi-
dence of Archimedes’ greatest achievement.
The Method survives in the Palimpsest alone. There is no trace of
it elsewhere—no other Greek manuscript, no Arabic version, no
Latin translation. The Palimpsest is the only physical object in the
universe to bear witness to this achievement of Archimedes. The
Method is unique not only among the Archimedean works but also
among all other mathematics produced prior to the sixteenth
century. Back in June 1999, we already knew—thanks to Heiberg’s
transcription—that here Archimedes came closest to the modern cal-
culus. We also knew that Archimedes came closest to revealing his
method by which physics and mathematics can be brought together.
These are the two keys to the science of Archimedes: the calculus,
which is the mathematics of infinity, and the application of mathe-
matics to physics. Mathematics, infinity, and physics: this triple com-
bination is all present in the Method. We can see how by following
two great mathematical proofs.
The first proof, an example of the application of mathematics to
the physical world, is Archimedes’ discovery of the center of gravity
of a triangle. It is a result found outside the Method, but it is crucial
139
9780306815805_06.qxd 8/22/07 9:22 AM Page 140
for the understanding of how the Method works. The second proof is
an example of the triple combination: mathematics, physics, and
infinity. It is the first proposition of the Method, in which Archimedes
finds the area of a parabolic segment.This brings us to the very height
of Archimedes’ achievement—collecting along the way the tool kit
required for the making of modern science.
The Center of Gravity

The first tool we need for modern science is the minuscule size of a
point, and it is of vast significance. Science cannot be made without
it. It is the center of gravity.
Let us put ourselves in the place of a physicist—say, in Newton’s
place.We wish to consider the motions of heavenly bodies under the
influence of gravity.There is a fundamental problem: stars and moons
are large bodies; they possess structure. Let us apply this to the fol-
lowing example: the dark side of the Moon is further away from the
Earth during a full moon than the bright side of the Moon. And so,
the Earth’s gravity acts less powerfully on the dark side, because it is
further away, than on the bright side. If we wish to be precise, we can
say that each point of the Moon has a slightly different gravity acting
on it.There are an infinite amount of points on the Earth each exer-
cising gravity differently on an infinite amount of points on the
Moon. How many combinations of gravities are there? The answer:
infinity times infinity. The problem has the complexity of infinity
multiplied by infinity!
Yet, Newton was capable of calculating his gravities. He dealt with
the motions of heavenly bodies based on the assumption that each of
them acts as a single point. In Newtonian physics, for most purposes,
the Earth is a single point and the Moon is a single point. There is
only one point—the Earth—exercising gravity on only one point—
the Moon. Such points are the centers of gravity. That is, we look for
the point that is the “average” of the weight or gravity of the Earth
and the point that is the “average” of the weight or gravity of the
140
9780306815805_06.qxd 8/22/07 9:22 AM Page 141
THE MAKING OF SCIENCE
FIGURE 6.1 Hanging a parallelogram
Moon. We then treat the Earth and the Moon as if they are all con-
centrated on those single points. It can be proven mathematically
that, for most calculations, once we find the center of gravity, we can
calculate with a single point instead of an entire object. Physics
cannot exist without the center of gravity. And it is, once again, an
invention of Archimedes.
The idea of “center of gravity” is best understood with a planar,
two-dimensional object. For instance, let’s use a circle. Our objective
is to balance the circle and hang it from the ceiling so that it remains
stationary. So, where do we tie the string? This is the easiest case: we
tie the string to the center of the circle.Tied anywhere else, the circle will
collapse. To keep it stationary, the circle must hang precisely from its
geometrical center. In this easiest case, then, the geometrical center
and the center of gravity coincide.
A square also hangs stationary when the string is tied to its exact
center.The same is true for all parallelograms, as we will demonstrate.
Simply find the point where the two diagonals meet, and you will
find the center of gravity of a parallelogram (see fig. 6.1). But the
question begins to become truly difficult when we approach more
complex objects. The key to all of them is the triangle. The triangle
has no obvious center in the way that a circle, a square, or even a par-
allelogram does. But once we find the center of gravity of a triangle,
we will be able to find the centers of gravity of all other rectilinear
141
9780306815805_06.qxd 8/22/07 9:22 AM Page 142
objects. As we have already seen, all rectilinear objects can be

measured by dividing them into triangles. To find the center of
gravity of any rectilinear object, then, we need to first crack the
problem of finding the center of gravity of a triangle. The rest will
easily follow.
The following, then, is the key to the science of centers of gravity:
we cut a paper triangle and hang it from the ceiling. How do we hang
it so that it remains stationary? How do we go about answering this
question? How do we pursue the science of centers of gravity?
You might wish perhaps, at this point, to conduct an experiment.
You might want to take several paper triangles and hang them from
the ceiling from various points, in this way finding out where the
center of gravity happens to be. This approach makes sense. After all,
you can’t tell how the world behaves without checking it out for
yourself.Your mind cannot dictate to the world how it should behave
and so, purely through thought, you are not going to find out how
objects hung from the ceiling are going to behave. Science is about
hard evidence, not about pure speculation.
Not quite. Much of the time science is about pure speculation.
Archimedes invented the concept of the center of gravity and found the
center without ever conducting an experiment—doing it all in his head!
Let us look at the process of finding the center of gravity of a tri-
angle. It is worth our while to follow this in detail. We shall see
Archimedes’ mind in action. What follows is the already advanced,
thirteenth proposition of Archimedes’ book On Balancing Planes.
PROOF ONE: HOW TO BALANCE A TRIANGLE,

OR MIND OVER MATTER
As you may recall, the language Archimedes uses for his science is
beautifully spare. For this reason, it is also a very difficult language to
read, whether in the original Greek or in translation. Let me explain
it in my own words—closely adhering to Archimedes’ own line of
thought—how Archimedes balances a triangle. As usual, this involves
some twists and turns.
142
9780306815805_06.qxd 8/22/07 9:22 AM Page 143
B D F C
FIGURE 6.2-1
In figure 6.2-1, we find the triangle ABC, which we will ultimately

balance. We will find its center of gravity—the point from which we
will attach a string that will keep the triangle stationary.The line BC
is divided into two at D (so that, BD = DC). The resulting line AD
is what is called a median in the triangle. Archimedes is going to
prove that THE CENTER OF GRAVITY OF A TRIANGLE MUST FALL SOME-
WHERE ON A MEDIAN LINE. This is not finding the exact point; it is
merely finding the line on which the point lies. But, bear with
Archimedes. In geometry, one needs patience.
First of all, a piece of logical ingenuity: we are going to assume the
opposite of what we want to prove. We are going to assume that the
center of gravity does not fall on the line AD. In other words, we are
going to assume that the center of gravity falls on some other line, such
as AF. Let us assume, then, that the center of gravity is the point T falling
on the line AF.This assumption will lead to an absurdity and, therefore,
we will know that we were wrong.We will know that the center does,
after all, fall on the median line. As we have seen already, this is a logical
technique greatly beloved by Archimedes, known as “indirect proof.”
So, initially, we assume that the center of gravity does not fall on
the line AD but on some other point, T.
Now, we need to introduce another complex piece of geometrical
ingenuity (see fig. 6.2–2). We add the points E and Z to the triangle.
E divides the line AB in two (so that, AE = EB) and Z divides the
143
9780306815805_06.qxd 8/22/07 9:22 AM Page 144
T
E Z
B D C
FIGURE 6.2-2
T
E Z
M
K L
B D C
FIGURE 6.2-3
line AC in two (so that,AZ = ZC).We then connect the three points:
D, E, Z. Now, inside the big triangle ABC, we have four smaller tri-
angles. If you were a Greek mathematician, it would not be difficult
for you to prove the following fact: ALL FOUR SMALL TRIANGLES ARE
SIMILAR TO THE BIG TRIANGLE, AND THEY ARE EQUAL TO EACH OTHER.
Similar triangles are identical to each other in every way except size.
Remember that we have assumed that T is the center of gravity in
the big triangle. So, the centers of gravity of the smaller triangles will
have to be similarly situated. Let us trace the centers of gravity in two
of the smaller triangles (see fig. 6.2–3).
The centers of gravity of the smaller triangles are to be the points
L and K. L is the center of gravity for the bottom-right triangle and
K is the center of gravity for the bottom-left triangle.What about the
144
9780306815805_06.qxd 8/22/07 9:22 AM Page 145
T
E Z
M
K N L
B D C
FIGURE 6.2-4
T
E Z
M
B D F C
FIGURE 6.2-5
two remaining small triangles? They, in fact, taken together, constitute

a parallelogram, and so a simple consideration of symmetry shows
that their combined center of gravity must lie at the point where the
two diagonals of the parallelogram meet—the point M.
Now, let us create the line KL (see fig. 6.2–4). If we consider the
two smaller triangles—the one at the bottom right and the one at the
bottom left—as a single geometrical object, it is clear where their
combined center of gravity must lie. It must lie on the line con-
necting their respective centers of gravity at its exact middle. That is,
the center of gravity of the two small triangles must lie at the exact
middle of the line KL. Let us call this middle point N.
We are now ready to conclude our proof (see fig. 6.2–5).We need
to create the lines AT and MN. Now, M is the center of gravity of
145
9780306815805_06.qxd 8/22/07 9:22 AM Page 146
two of the triangles, and N is the center of gravity of the remaining

two. The combined center of gravity of all four triangles, then—that
is, the center of gravity of the big triangle—must therefore fall
exactly in the middle of the line MN. In this diagram we can clearly
see that this is not where point T lies.This is actually a bad argument
to follow and a classic example of why one should not rely on dia-
grams too much.The question is this: how do we know that the point
T can never lie on the line MN?
This is how. In any triangle, for the point T to lie on the line MN,
the two lines MN,AF must intersect at some point. Indeed, they must
intersect at the point T.
And, they can’t. It would be an easy task for Archimedes to show
that the lines AT, MN must always be parallel. Therefore, they can never
intersect. We have asked the point T to lie on the line MN and by
doing this we have required two parallel lines to cut each other! This must
be wrong. No matter where we place the initial point T, as long as it
is on a line such as AF and not on the median line AD, we will always
derive the same absurdity of two parallel lines being required to cut
each other. So, we know that the true position of the center of gravity
of any triangle is on its median line.
In any triangle there are, of course, not one but three median lines.
It is proved that when we draw all three median lines of a triangle—
any triangle—they meet at exactly one point. In figure 6.2-6, we see
this in the triangle ABC. Its sides are bisected by the median lines AD,
BZ, CE. All three lines, AD, BZ, CE, meet each other at the point X.
The point X lies at an exactly defined position: it is one-third the way
along the median line. DX is one third of AD; ZX is one third of BZ;
and EX is one third of EC.This is where the center of gravity must lie.
So, Archimedes could have suggested the following experiment.
Create a paper triangle. Draw a median line. Find the point one-third
of the way along the median line. Attach a string to this point and
hang it from the ceiling.The triangle will remain fixed and stationary.
How does Archimedes know this? Oh, that’s easy; Archimedes
146
9780306815805_06.qxd 8/22/07 9:22 AM Page 147
E Z
B D C
FIGURE 6.2-6
explains. It is because lines, divided in two, give rise to four equal and
similar triangles. It is because a certain line happens to be parallel to
another. It is because of geometry. Follow logic and you see for
yourself. We turn away, in disbelief. But, Archimedes is right.
The products of pure thought, which at first glance have nothing
at all to do with each other or the physical world, are brought
together. Before you know it, pure speculation binds the physical uni-
verse and forces it to behave in a particular fashion. No experiments
were necessary to find this out. Mind rules over matter because, ulti-
mately, even brute matter must follow logic.
This is rather like the magician telling us—without even
looking—about the contents of our wallets. Archimedes has told
us—without even looking—how the world must behave, where a
triangle must balance.
Follow this a bit further. We start at Syracuse in the third century
BC when all we can do is hang a triangle from the ceiling. But, we
follow the line of thought long enough and we are able by the twen-
tieth century to launch a rocket to the Moon and to explode an
atomic bomb. All the way, following the same principle, we apply the
power of reasoning to the universe, and the universe follows logic.
This is the principle discovered by Archimedes. This is science in
action.
147
9780306815805_06.qxd 8/22/07 9:22 AM Page 148
The Law of the Balance

There is another complementary act of magic. Following the magic
of mind-over-matter—pure mathematics discovering a physical
fact—comes another act of magic no less spectacular. It is matter-
over-mind—physics discovering a mathematical fact. This is done in
the Method. Most historians of mathematics consider this the most
amazing act performed by Archimedes. Besides being a piece of
physics-over-mathematics, it also introduces infinity in a puzzling,
strange way. In the next few pages, we will follow this.
We need tools for this act. The first we have already: the center of
gravity of a triangle. The other is another physical fact, proved math-
ematically by Archimedes in the treatise On Balancing Planes. It is
called the Law of the Balance, which we have already mentioned. It
may also be called the Law of the Lever.While the two machines may
do different things, they work by exactly the same mathematical rule.
Archimedes relies on a balance for his measurements in the Method.
However, he was equally familiar with the lever and the law was most
famously and succinctly expressed by him as follows: “Give me a
place to stand and I shall move the Earth.” That is: “Give me a lever
long enough, and I can move any object whatsoever.” Why is that?
It is because of a principle of proportion. Let me now explain this,
first with a balance.
We can take any two objects and put them on a balance. One arm
of the balance has Object One, whose weight is, say, 10 pounds. The
other arm of the balance has Object Two, whose weight is, say, 2
pounds. The balances are of the moveable type, so that we can make
each object nearer or further away from the fulcrum. The question:
at which distances will the objects balance? The answer is as follows:
the ratio of the weights is 5:1, and therefore the ratio of the distances
should be reciprocal—that is 1:5. The distance of the lighter object
should be five times the distance of the heavier object, and then they
will balance. The rule: weights balance when they are reciprocal to
their distances.
148
9780306815805_06.qxd 8/22/07 9:22 AM Page 149
Ellipse Parabola Hyperbola
FIGURE 6.3 The three conic sections.Take a cone and cut it in three ways. If you cut
from both sides you create an ellipse; if you cut it parallel to one side you create a
parabola; and if you cut it moving away from one side you create a hyperbola.
If, instead of a balance, we use a lever, the same principle still holds.
The object that is five times more distant is capable of balancing an
object exactly five times heavier. Make it even more distant and the
lighter object will even move the heavier object. All of this
Archimedes proved in On Balancing Planes, of course by pure
thought. By standing in the realm of pure thought, Archimedes
moved the Earth.
So I repeat: first, the center of gravity of a triangle is one third
along on the median line and second, objects balance each other
when their distances are reciprocal to their weights. These are two
facts about the physical world. With their aid, we will measure the
area of a segment of a parabola.That is, once again, we will find how
a curved figure is equal to a rectilinear one. We have already seen
Archimedes obtain this result in one way. In the Method, he obtains
this in another, much more spectacular way, which in itself is quite a
surprise. Who would think that triangles and balances would have
anything to do with parabolas?
149
9780306815805_06.qxd 8/22/07 9:22 AM Page 150
A C
FIGURE 6.4-1
The Parabola
The very notion of a parabolic segment is very abstract. Parabolas
belong to a family of curves invented by Greek mathematicians as an
act of pure geometrical fancy, having no physical significance in mind.
We take the surface of a cone, and we cut it by a plane. Depending on
how we produce this cut, we derive one of three sections: hyperbolas,
parabolas, or ellipses (see fig. 6.3). Circles, squares, and triangles make
sense: we more or less meet them in daily life. This is not so with
hyperbolas, parabolas, and ellipses. Their interest is mainly in the fact
that—as it turns out—there are all sorts of nice geometrical propor-
tions that arise with the combination of conic sections.
Conic sections are best considered as toys invented by geometers to
aid in their geometrical play. I keep returning to the irony of mathe-
matics-over-physics, of how pure thought turns out to rule the physical
universe.This is one of the most remarkable ironies: the conic sections,
which were invented as geometrical toys, turned out to be the curves
defining motion in space. Electrons orbiting around the nucleus of an
atom, a rocket launched to the Moon, a rock thrown by a catapult—
all such motions obey the curves of conic sections. So that this study
is, in fact, one of the major routes leading to modern science.
PROOF TWO: THE AREA OF A PARABOLIC SEGMENT,

OR MATTER OVER MIND
We follow on this route, concentrating on the area of the parabolic

segment (see fig. 6.4–1). By a “parabolic segment,” we mean the area
150
9780306815805_06.qxd 8/22/07 9:22 AM Page 151
E
K
A D C
FIGURE 6.4-2
intercepted between a parabola and a straight line crossing through it,

such as ABC. (ABC is the parabola; AC is the straight line).You will
notice, of course, that a parabolic segment is a curvilinear object.This
is the great mystery that constantly exercised Archimedes’ mind: how
to measure curvilinear objects; how to reduce them to rectilinear
objects. Soon we will see.
We move to the next figure and note certain facts (see fig. 6.4–2).
First, each parabola possesses an axis of symmetry. In this case, it is the
line BD, around which the parabola is “the same,” to the left and to
the right.
To explain the following facts, I need to add a few bits to the con-
struction. We need to draw a tangent to the parabolic segment at the
point C, namely the line CZ. We also need to draw a line parallel to
the axis, passing through the point A, namely the line AZ. The
tangent and the parallel meet at the point Z. We have enclosed the
parabolic segment within a triangle: the segment ABC is enclosed
within the triangle AZC. We also need to extend the line DB to
reach the point E, and the line CB to reach the point K.
An interesting series of geometrical relations now depends on
the fact that the axis cuts exactly through the middle of the
151
9780306815805_06.qxd 8/22/07 9:22 AM Page 152
T Z
H
M
K E
N
B
O
A X D C
FIGURE 6.4-3
parabola, as well as on some other properties of the parabola. They

are as follows:
• The point K is exactly at the middle of the line AZ.
• The point B is exactly at the middle of the line DE.
• The triangle AKC is exactly half the area of the triangle AZC.
• The point B is exactly at the middle of the line KC.
• The triangle ABC is exactly half the area of the triangle AKC.
Taken together, all of the above means, among other things, that:
• The bigger triangle AZC is four times the area of the
smaller triangle ABC.
This is rather like the division of one triangle into smaller triangles
that we saw with the previous proof, and it will be used by
Archimedes later on.
Now I need to mention one of the funny facts about parabolas.
We draw a line parallel to the axis of the parabola BD (see fig. 6.4–3).
There are infinitely many such parallels, and we take one at random,
say the line MX. So, MX is parallel to BD. Here comes the funny fact,
which, as is typical, takes the form of a proportion:
152
9780306815805_06.qxd 8/22/07 9:22 AM Page 153
• The line MX is to its smaller section OX as the base of the

diameter AC is to its smaller section AX.
To the algebraically inclined, this can be expressed symbolically as
follows: (MX:OX::AC:AX).
So we return to the line chosen at random, MX. We add a few
more details to the diagram. First we extend the line from C, through
B, onward. It cuts the line MX at the point N, and it cuts the line ZA
at the point K. We further extend it to the point T, so that KT
becomes equal to KC, and the point K is exactly in the middle of the
line TC.
We then do something very unorthodox. It will take us some time
to see why we do this. But what we do is take the segment OX and,
in our minds, transport it to a new position SH. Its middle point is
now T, lying at the end of the line KC, extended. This is a thought
experiment then.We imagine a piece of geometrical line transposed.
This is already sensational, because Archimedes has just moved a geo-
metrical object and treated it as if it were physical—a piece of wood
that one can carry about.
I now note a few consequences. First of all, I recall in the original
proportion, the funny result for parabolas:
• As MX is to OX, so is AC to AX.
Second, because the lines MX and ZA are parallel (this is how we
constructed the line MX), the ratio is preserved:
• As AC is to AX, so KC is to KN.
These are “the same” ratios, simply sliding along parallel lines.
Combine the two above, and you can immediately see (eliminating
the middle term, as it were) that:
• As MX is to OX, so KC is to KN.
The parallel line taken at random, MX, is to its smaller section, OX,
just as the base AC is to its smaller section AX and also as KC is to
KN.
153
9780306815805_06.qxd 8/22/07 9:22 AM Page 154
Further, remember that K is the exact middle point of TC (TK =

KC). So whatever is true of KC, will also be true of TK. Therefore,
it must also be true that:
• As MX is to OX, so TK is to KN.
The parallel line taken at random, MX, is to its smaller section, OX,
as TK is to KN.
Notice what Archimedes has achieved. His ratios until now have
all been “packed in,” with a line relating to its own segment. This
final ratio of TK to KN, however, unwraps the package and trans-
forms the relation to one between two independent line segments
touching at one point only. This will become useful later on.
Finally, we may bring in another consideration.Whatever is true of
OX must also be true of SH. This, after all, was our original thought
experiment: to transpose OX so that it becomes SH. The two are
identical.We noted above that the parallel line taken at random, MX,
is to its smaller section, OX, as TK is to KN.
Now if I exchange SH for OX—since the two are identical—we
can say the following:
• As MX is to SH, so TK is to KN.
The parallel line taken at random, MX, is to its smaller section—now
SH—the same as TK is to KN.
This last proportion is the one we have been looking for. The
magician is about to perform a trick.We have already engaged in one
thought experiment—imagining the line OX as a physical line and
transporting it to occupy the position SH.We now need to engage in
another much more radical thought experiment. No one said any-
thing remotely similar to this prior to Archimedes.
What we imagine now is that the lines MX and SH are lying on
the arms of a balance with its fulcrum at K.We treat them as physical
objects possessing weights, which are a reflection of their lengths.The
154
9780306815805_06.qxd 8/22/07 9:22 AM Page 155
two lines possess centers of gravity, obviously at their exact centers—

that is, respectively, at N and T.
You see what we just did: we considered geometrical objects as if
they were physical. I repeat: no one ever did this prior to Archimedes.
Just as he invented the mathematical treatment of physics, he has also
invented the physical treatment of pure mathematics.
Remember the previous result: “MX is to SH as TK is to KN.” So
what is the ratio of weight between MX and SH? It is the ratio of
their lengths; the ratio of the line MX to the line SH. This, we have
seen, is the same as the ratio of the line TK to the line KN.Therefore,
the ratio of the lines is reciprocally the same as the ratio of their dis-
tances from the fulcrum.
Apply the Law of the Balance and derive the beautiful observation
made by Archimedes: the two lines MX, SH will balance with K as
their fulcrum.
Dizzying? Hold your breath.We now move onto another thought
experiment, even more dizzying. Fresh from this trick, the magician
now prepares yet another one.
The random line MX exactly balances its smaller section OX,
around the fulcrum K, when the smaller segment is transposed so that
its center becomes T. But we chose the line MX randomly. No matter
which other parallel line we choose, the same will hold true. The
ratios will change, but they will remain respectively proportional.
In other words,
Each parallel line inside the triangle AZC balances its
respective section from the parabolic segment ABC (posi-
tioned at T), around the fulcrum K.
If you agree to that, you must agree to the following:
All parallel lines, inside the triangle AZC, taken together,
balance all their sections from the parabolic segment (posi-
tioned at T), taken together, around the fulcrum K.
155
9780306815805_06.qxd 8/22/07 9:22 AM Page 156
Or, better still:

The triangle AZC balances the parabolic segment ABC (posi-
tioned at T), around the fulcrum K.
How can it be otherwise? We slice the triangle and the parabolic
segment, parallel line by parallel line, and each time we do the slicing,
we find the same balance at the same fulcrum. So, when we take the
entire triangle and the entire parabolic segment, they must obey the
same law of the balance: the entire triangle and the entire parabolic
segment must balance each other at exactly the same fulcrum.
So I repeat: the triangle as a whole balances the parabola as a
whole, with K as the fulcrum.
We know where the center of gravity of the transposed parabolic
segment is: it is at the point T. This, after all, is our thought exper-
iment. Parallel line by parallel line, we have transposed the parabolic
segment so that the center of gravity of each line is T. If each line
taken separately has its center of gravity at T, then taken altogether
they will also have a center at T.
We can then say that the triangle, set at the position of the
diagram, balances the parabolic segment, with the center of gravity of
the parabolic segment at T and the fulcrum at K.
What about the center of gravity of the triangle? Well, we previ-
ously worked very hard on this question.The center of gravity of the
triangle is at the point one-third the way along the median line—that
is, at the point one third of KC.
However, one third of KC is also one third of KT.That is, the dis-
tance of the center of gravity of a triangle from the fulcrum K is one
third the distance of the center of weight of the parabolic segment.
The parabolic segment is three times as far from the fulcrum as is
the triangle.Therefore, the triangle must be three times the weight of
the parabolic segment, and the area of the triangle must be three
times the area of the parabolic segment.
We can make this result even more elegant. Consider the triangle
ABC, which, as we recall, is exactly one-fourth the triangle AZC. In
156
9780306815805_06.qxd 8/22/07 9:22 AM Page 157
other words, the parabolic segment ABC is four-thirds the triangle

ABC.
Put simply: a parabolic segment is four-thirds the triangle it
encloses.
This was a moment of magic. Consider that each treatise by

Archimedes contains at least one such moment of magic and you can
begin to see the measure of the man. In the Method, each proposition
is just as magical. No wonder Heiberg was so excited in 1906.
Note the complex route leading to this magic.We had the thought
experiment—considering geometrical objects as physical ones. And
then, something even further. We had a result for pairs of slices—pairs
of random lines taken of the triangle and the parabolic segment. We
then moved to the triangle itself and the parabolic segment itself, each
taken as a whole.
In other words, we took a proportion involving four lines and
turned it into a proportion involving infinitely many lines—all the
infinitely many parallel lines constituting the triangle or the parabola.
Are we allowed to do that? This question, from then on—from
Archimedes’ time to our own—became the central question of math-
ematics. The Method, by bringing together mathematics, physics, and
infinity, raised the most fundamental questions of science. It antici-
pated Newton’s calculus, but it also anticipated the conceptual diffi-
culties of that calculus.
How much did Archimedes know about infinity? In June 1999, we
did not know. The question on everyone’s lips: what are we going to
find, further, in the Method—if anything? We needed to look inside
the Palimpsest and be able finally to read it. But in June 1999, Abigail
had not yet unwrapped the book. The brain was still caged inside its
box, as I waited, impatiently, for it to be freed.
157
9780306815805_06.qxd 8/22/07 9:22 AM Page 158

9780306815805_07.qxd 8/22/07 9:26 AM Page 159
7
The Critical Path

onservators don’t like being the center of attention, but that’s
C just where I had put Abigail Quandt—in the public eye and
subject to its scrutiny. If you work on Leonardo’s Last Supper,
Michelangelo’s David, or the unique witness to the thoughts of
Archimedes, you’d better not slip up. Everyone tells you what you
should be doing, but only you can do it. And no one had any idea
about the problems that Abigail faced. They do now, but they didn’t
then. Hers was not only the critical path, as program manager Mike
Toth characterized it; it was also the most important and the most
onerous. Like Reviel, you are going to have to wait until you get
more of the Method. This is Abigail’s story.
Abigail is not your normal book conservator. Most book conser-
vators work with paper books, very few work on parchment manu-
scripts. There are good reasons for this. First of all, there are many
more paper books in the world than there are parchment ones.
Second, in general, paper books need conservation treatment much
more often than parchment ones. This is particularly true if they are
printed on bad paper with a high acidity level. As I write, such books
are literally self-destructing in libraries across the globe. Many are the
paper conservators who fight this battle. Parchment does not have
this acidity problem, and it is much tougher than paper. One essential
difference between parchment and paper, is that parchment is much
more sensitive to changes in temperature and humidity—it is skin,
after all. If you lay a sheet of parchment over your sweaty hand it will
quickly curl. Actually, it will curl into the shape it had on the back of
the animal from which it came. With finely illuminated manuscripts,
159
9780306815805_07.qxd 8/22/07 9:26 AM Page 160
such as those at the Walters, this can have serious repercussions. The
pigments in the illuminations do not change shape with the
parchment as humidity changes and after a while the pigments flake
off. Abigail had been working on parchment with this kind of
problem for more than twenty years. She is a parchment expert, and
very few people have her skills.This was why she was uniquely qual-
ified to work on the Palimpsest.
Normally, the best thing to do with a historic object is absolutely
nothing, which is what conservators do most of the time. Don’t
touch it; secure and monitor its environment. After all, a codex that
has survived a thousand years is unlikely to degenerate much further
if it is not handled and if it is not subject to pollutants or extremes of
climate. In the past, even well intentioned treatments resulted in per-
manent damage and the loss of important historical evidence. In the
nineteenth and early twentieth centuries many palimpsests were
wrecked by the treatment they received. Scholars routinely read
palimpsests by applying chemicals to them. In 1919, English novelist
and manuscript scholar M. R. James wrote that erased text could be
revived by the dabbing (not painting) upon it of ammonium
bisulphide, which, unlike the old-fashioned galls, does not stain
the page. Dabbed on the surface with a soft paint-brush, and
dried off at once with clean blotting paper, it makes the old
record leap to light, sometimes with astonishing clearness,
sometimes slowly, so that the letters cannot be read till next
day. It is not always successful; it is of no use to apply it to
writing in red, and its smell is overpowering, but it is the elixir
of paleographers.
There were other elixirs, too. The most powerful was Gioberti’s
tincture—successively applied coats of hydrochloric acid and potassium
cyanide. I’ll just repeat that—successively applied coats of hydrochloric
acid and potassium cyanide. Needless to say, ammonium bisulphide,
too, has a severely detrimental effect on parchment. Working in the
twenty-first century, Abigail couldn’t apply chemicals to reveal text in
160
9780306815805_07.qxd 8/22/07 9:26 AM Page 161
T H E C R I T I C A L PAT H
FIGURE 7.1 The spine of the Palimpsest
Mr. B’s book.There was little that Abigail could do to make the erased
text appear.This would be a challenge for the imagers.
However, it was not an option for Abigail to do simply nothing

with the Palimpsest. Despite all the lessons of history, Mr. B gave the
go-ahead for her to perform radical surgery on the manuscript. It was
a brave decision, and we hope that history will say that it was the
right one. The reasons for it certainly sounded excellent at the time.
The only way that scholars would be able to read the Archimedes
text now hidden in the book’s binding and the only way that imagers
would be able to take the scientific images that were needed, was for
Abigail to take the entire manuscript apart.
The Path Is Blocked

On Monday, April 3, 2000, Mr. B, Reviel, Natalie, and Mike gathered
at the Walters. It was a historic day. Abigail was going to disbind the
Palimpsest. All went smoothly at the start. Just before the sale, in
order to make the book presentable, Scott Husby from Princeton
had carefully put a temporary binding on the codex that was
designed to be easily taken apart. Abigail quickly took this binding
off, leaving the parchment text block naked, without its covers. It
161
9780306815805_07.qxd 8/22/07 9:26 AM Page 162
was a pitiful sight, though it took Abigail some time to make clear
that it was also a tragic one. And here begins the worst part of the
story of Mr. B’s book. If, to Reviel Netz, the Palimpsest was the
unique source for the diagrams that Archimedes drew in the sand, to
Abigail Quandt it was a conservation disaster zone.
The spine of the book was covered in glue. This was a post-
medieval practice, which helped to secure the structure of the codex
but which obviously presented problems for Abigail in taking it apart.
If you look carefully (see fig. 7.1), you can see that the spine seems to
have two separate colors. The darker color is hide glue, made from
animal skin, and Abigail could remove it with reasonable ease. This
section forms the second half of the codex from folio 97 onward.The
real problem was the light glue from folios 1 to 96. According to
Abigail, “The other half of the text block has been coated with a
transparent adhesive, probably a type of poly(vinyl) acetate (PVAC)
emulsion. While PVAC will swell in contact with water and/or
alcohol, there is no way to dissolve it once it has formed a dried film
on the surface of an object. Attempts to remove this adhesive from
the spine folds of the Palimpsest have proved to be extremely risky,
since the glue is stronger than the parchment.” In other words, this
is stock-in-trade wood glue. Precisely those lines that Heiberg could
not read, because they were hidden in the spine of the bound codex,
were now coated and stuck together with commercial glue.
This was not the worst problem. Let’s look at the unique surviving
folio of Archimedes’ Stomachion (see fig. 7.2). Think of it as a cross
section of the brain of a great man. It is, quite literally, in pieces and
large parts of it are simply missing.The rest of it is covered in an awful
purple color. Now, up until this point, I have insisted that parchment
is tough. Its basic constituent, after all, is the stuff that makes up your
shoes. There are only two ways in which you cannot only wear out
your shoes, but destroy them. One is to burn them. But, the
Palimpsest had survived a fire at St. Sabas. Another is to throw them
into a bucket of water and then expose them to air. Pretty soon they
will get moldy. And that is, more or less, what happened to this folio
162
9780306815805_07.qxd 8/22/07 9:26 AM Page 163
FIGURE 7.2 Archimedes’ Stomachion!
and in varying degrees to the entire Palimpsest. If you immediately

take care of your wet shoes, then you can clean them up reasonably
well. But that’s not what happened to the Palimpsest—far from it.
The Palimpsest was left to stew and the mold was left to grow. And
mold grows by eating what it is growing on. The mold actually
163
9780306815805_07.qxd 8/22/07 9:26 AM Page 164
digested the parchment. All the folios of the Palimpsest suffered in

some degree from the mold. Normally, the side of the parchment that
was on the inside of the animal is in a poorer state than the side that
the animal presented to the outside world.This is because the skin has
evolved—or been intelligently designed—to resist microbe attacks
from the environment. But in the face of the mold attack that the
Palimpsest was subjected to even the hair side was often heavily
damaged. Sometimes it didn’t look so bad, and the folios seemed
strong enough. But Abigail showed them to me backlit over a light
box and the light shone through like stars in the night sky.
So, the leaves were stuck together, and they were extremely fragile.
And then, of course, four of them were painted over with forged por-
traits of the Evangelists. Worse, the forger did not just scrape off the
prayer-book text and paint on top of it. He also “distressed” the folio.
For example, he made incisions into folio 57r and took nicks out of
it to make the picture look older after he had finished it. It is often
tempting to display pictures in manuscripts independently of the
codex itself. Perhaps this is why there was a nice curved rust stain at
the top of the folio. Abigail believes it was made by a paper clip. Not
formal enough for you? Maybe. Then why not take a piece of Blu-
tack (a blue putty-like substance) and stick it to its backing that way.
There’s plenty of Blu-tack on the back of this folio—just try reading
through it. What a way to treat a book—any book. What a tragedy
that it was this book.
The disbinding of the book, upon which everything depended,
came to a virtual standstill. The Archimedes Project took what can
best be described as a duck dive. Reviel went back to Stanford,
Natalie to Cambridge. The lights went out in the studio that had
been especially prepared to image the book, and Mr. B went home.
Abigail spent days just looking at the book, thinking, and docu-
menting. The days turned into weeks and the weeks into months. I
dodged press inquiries with breezy comments about the importance
of the book and some small conservation issues. Abigail and I stopped
talking—always a bad sign.Then she asked for what seemed to me to
164
9780306815805_07.qxd 8/22/07 9:26 AM Page 165
be an extravagant amount of money to employ her colleagues at the

Canadian Conservation Institute to begin a whole battery of tests.To
my surprise, Mr. B wrote the check. Mike Toth sent me dark emails
warning of slippage in the schedule early in the project and the
dangers of pure research taking us away from our well-defined and
limited objective. There was nothing I could do about it. At the
Walters, as in many American museums, curators cannot tell conser-
vators what to do. This is ultimately a very good thing, but occa-
sionally also an incredibly frustrating one. It just didn’t seem that
much of a problem to me, and yet it brought my whole system to a
standstill. Never mind about me. Picture, if you will, Assistant
Professor Netz, newly installed in the Department of Classics at
Stanford, trying to get tenure. The one manuscript that he needed
had almost miraculously appeared. He had seen it; he had even
touched it; he knew better than anyone the secrets that it contained.
But now, he was reduced to begging for glimpses of only parts of it.
Noel was out of leaves; Netz was out of luck.
The world was watching, the scholars were poised to read, and the
scientists were ready to image. But everybody was waiting for
Abigail—not all of them patiently.
The Heiberg Photographs

Reviel made himself busy, stirring every bush he could about the
Palimpsest and digging into its history. He studied Heiberg’s work
over and over again. He read it cover to cover. And then, in the intro-
duction, he noticed that Heiberg had said that he took photographs
of the manuscript when he was in Constantinople. Since the manu-
script itself had disappeared into private hands for most of the twen-
tieth century, the photographs would have proved extremely useful to
anyone who could find them.The trouble was that no one had.There
is a large archive of Heiberg material in the Royal Library of
Denmark, but repeated searches in the archive had not revealed the
photographs. Reviel, however, thought he would give it another shot,
165
9780306815805_07.qxd 8/22/07 9:26 AM Page 166
so he wrote an email to a colleague of his, a Danish historian of

ancient science, Karin Tybjerg. Karin prevailed upon the Keeper of
Manuscripts at the Royal Library, Erik Petersen, to have another
look, and Erik had an idea. The Heiberg archive had been deposited
at the Royal Library after Heiberg had died. But what if Heiberg had
given the photographs to the library before he had died? Heiberg
would have understood the importance of the photographs and, as a
good humanist, might well have wanted to make them publicly
accessible. If so, where would they be? In among the photography
collection, of course. And Erik found them. They were Ms. Phot 38
in the Royal Library. In June of 2000, Reviel and I visited
Copenhagen to see the photographs. Erik warmly welcomed us and
we were handed a plain album full of photographs of a manuscript.
There were sixty-five in all, and they were the smoking gun.
The photographs give a clear picture of what the manuscript
looked like in 1906. If you compare the leaf containing Archimedes’
Stomachion now, with its condition then, you can see that they are
barely recognizable as the same folio.The Stomachion page was whole
in 1906. It is a moldy fragment now. Of course, some folios in the
Palimpsest are worse than others. The tragedy is that their condition
is almost in an inverse relationship to the importance of the texts that
they contain. Archimedes has been extremely unlucky: the folios
containing the Method and the Stomachion are in the worst condition
of all. When I tell you that Abigail marked the Stomachion folio as
“very poor,” you will understand that she is a master of the under-
statement.That’s not my style. I tell you: the body of Archimedes may
have died by the sword at the hands of a Roman soldier in the third
century BC, but the genius of Archimedes was eaten by mold over
two thousand years later.
There is, of course, no sign of the forgeries in Heiberg’s photo-
graphs. On the photograph of folio 57r Heiberg had written
“M16”—it was the sixteenth folio of the Method. Now, a picture was
on this page. A balding man with a white beard, wearing a green
robe, sits in a chair with a high, curved back with his feet resting on
166
9780306815805_07.qxd 8/22/07 9:26 AM Page 167
a blue stool. In his right hand he holds a pen and in his left a scroll
upon which he is writing. In front of him is a desk with the instru-
ments of his trade and before him at eye level upon a lectern is the
book from which he is copying. He is set within an archway, and the
whole picture is framed with a border.The background is gold.There
is hardly even the faintest trace of the prayer-book text, let alone the
writings of Archimedes.
The Heiberg photographs are dramatic and unequivocal evidence
that most of the damage to the book occurred in the twentieth
century, after it had been revealed as the unique source of treatises by
Archimedes.Abigail was stuck with an old book, but one that had only
recently sustained such serious damage.Who was responsible for it?
The Forged Miniatures

In early March of 1999, I stopped in to see John Lowden in his
rather cramped office on the top floor of the Courtauld Institute,
where he is the Professor of Medieval Art. I showed him a picture
of one of the forgeries. To my astonishment, he immediately cried
out “Snap,” and pulled out a publication that he had written on a
manuscript in Duke University Library, North Carolina. The codex
at Duke contained four Evangelist portraits very similar to the ones
in the Palimpsest, and John had demonstrated conclusively that they
were modern forgeries.
He told me to go downstairs into the library and pull out a pub-
lication by Henri Omont, published in 1929, of Greek manuscripts in
the Bibliothèque Nationale in Paris. I quickly found what he knew I
would. The figures of the Evangelists in the Palimpsest, like those
from the Duke manuscript, were copied from illustrations in that
book. They were not exact copies; their backgrounds were much
simpler than the backgrounds of the pictures in Omont’s publication.
But the figures of the Evangelists themselves, their chairs and their
writing desks were the same. Abigail established that the drawings
were actually traced on a scale of 1:1 from the publication.
167
9780306815805_07.qxd 8/22/07 9:26 AM Page 168
Following the Christie’s catalogue, I had been cursing the scribe of

the prayer book and the greedy and careless monks at the Metochion
for the appalling state of the book. The Duke codex, too, had
belonged to the Metochion. Since both books had paintings by the
same forger, I was more sure than ever who was to blame for the
books’ condition.
But I was wrong. In May 2001, Abigail received back a monu-
mental work of research and scholarship.The Canadian Conservation
Institute delivered their report on the Archimedes Palimpsest. It’s an
impressive but depressing read. It is also full of useful information
about the appalling condition of the book. Among a whole range of
pigments that they chemically identified in the forgeries, one was
particularly revealing—phthalocyanine green. This color only
became commercially available in Germany in 1938.As we have seen,
by 1938 there were no manuscripts at the Metochion. In fact, the for-
geries were made at least fifteen years after Marie Louis Sirieix was
supposed to have acquired the codex.
The realization that I had entirely misunderstood the story of the
Palimpsest made me angry, and I wanted to set the record straight. I
wanted to know who was responsible for the catastrophic treatment
of the book after it left the Metochion. Now it seemed likely that
Marie Louis Sirieix was to blame. But the story told by Robert
Guersan was short on particulars. Perhaps Sirieix had not even
known that the book contained the unique letters of Archimedes;
Anne Guersan seems to have recovered this information. Without
further documentation, it would be impossible to be sure.
The Willoughby Letter

In May 2006, I walked into work and found placed upon my desk,
without ceremony, a copy of a letter. It had been left there as an early-
morning gift by my good friend Georgi Parpulov, a Bulgarian scholar
of Greek manuscripts with a wry smile and an extraordinary ability
168
9780306815805_07.qxd 8/22/07 9:26 AM Page 169
to uncover the deeply buried. Georgi had dug this letter out of an
archive of photographs, the Harold R. Willoughby Corpus of New
Testament Iconography at the University of Chicago. The letterhead
made it clear that it was from an antiquities dealer living in Paris. It
read: Salomon Guerson, Rare Carpets, Antique Tapestries, 169
Boulevard Haussmann, Paris, and it was addressed to Professor Harold
R. Willoughby at the University of Chicago, Illinois:
February 10, 1934
Dear Professor Willoughby

Pursuant to our correspondence of 1932 with regard to a
manuscript that I had shown to you and a folio which was
identified through your mediation by the curator of the
Huntington library as being the manuscript of Archimedes
described by J. L. Heiberg in Hermes vol. 42, page 248, I would
like to let you know that I wish to sell this manuscript.
I have shown it to M. Omont of the Bibliothèque
Nationale as well as to the Bodleian Library and both have
made me offers that I found insufficient.You would greatly
oblige me if you would let me know whether this manuscript
interests you or at any rate if you would write to me to whom
I could offer it with a chance of selling it. I am asking $6,000.
In expectation of receiving news from you, I ask you, dear
professor, to be assured of my highest respect for you.
S. Guerson
The evidence of this one letter meant that the twentieth-century

history of the Palimpsest needed to be rewritten. Not only were the
forgeries painted after the manuscript left the Metochion, but the
manuscript was identified in 1932 as the Palimpsest and Sirieix had
not owned the manuscript until after 1934. How had the manuscript
169
9780306815805_07.qxd 8/22/07 9:26 AM Page 170
gotten from Constantinople to Paris, and who was responsible for the
forgeries?
A New History
The person who first recognized the Archimedes Palimpsest for what
it was after it left the Metochion was the curator of the Huntington
Library in Los Angeles in 1932. I wrote to my colleague Mary
Robertson, curator of manuscripts at the Huntington Library, and
although she could not come up with definite proof, she thought it
most likely that the curator in question had been Captain Reginald
Berti Haselden. Between 1931 and 1937 Haselden had been in cor-
respondence with Professor Edgar Goodspeed of the Department of
New Testament Studies in Chicago over some palimpsested material.
He was particularly interested in ultraviolet photography and wrote
a book in 1935 entitled Scientific Aids for the Study of Manuscripts. This
was exactly his cup of tea.
It seems that Haselden only had the opportunity to identify one
folio of the manuscript, not the entire codex, and that this one folio
had been transcribed by Heiberg on page 248 of his article in Hermes.
This is folio 57. How it was that Haselden identified this folio alone
is something of a mystery. Maybe he saw a photograph of just this one
page. Just possibly it had already been separated from the manu-
script—as it is now—and Haselden merely studied this single leaf. Be
that as it may, it is further proof that the forgeries were added after
Salomon Guerson wrote his letter, and therefore after the book had
been identified as Heiberg’s Codex C. Folio 57 is now covered with
a forgery. Not even Haselden, with his interest in scientific aids, could
have identified the Archimedes text through the forgery. The for-
geries were indeed done after 1932, and after Salomon Guerson had
acquired the manuscript.
Even before Georgi had unearthed the Willoughby letter, John
Lowden had already suspected that the Guerson business was
involved in the history of the Palimpsest. He had his own very good
170
9780306815805_07.qxd 8/22/07 9:26 AM Page 171
reason for suspecting that they were indeed responsible for the for-
geries. He discovered that the Guersons owned a leaf from a
Byzantine manuscript that was exhibited in a famous exhibition of
Byzantine art in Paris in 1931. The unusual way in which the figures
in the Palimpsest forgeries were framed was exactly like the leaf that
the Guersons owned and exhibited in the 1931 exhibition. The
Willoughby letter was an extraordinary confirmation of John’s
insight. It demonstrated that not only did the Guersons own the
manuscript, but that they also had known Henri Omont, from whose
publication the Evangelists in the forgeries had been traced.
John had also made substantial progress in determining how the
Palimpsest might have traveled from the Metochion to Paris.
Salomon Guerson certainly knew one of the most famous dealers of
the twentieth century, Dikran Kelekian. And it might well have been
through Kelekian that they had acquired some of their manuscripts.
In 1931, Kelekian owned two miniatures taken from the same book
as the miniature that the Guerson business had displayed in Paris in
the same year.The manuscript from which they all came is known to
have been perfectly intact in a convent in Constantinople as late as
1922. By 1931, Kelekian had inserted his two miniatures into yet
another manuscript, which came from—guess where—the
Metochion. The Guersons had good access to manuscripts from the
Metochion. The circumstantial evidence that the Guerson business
was responsible for the forgeries in the Duke manuscript, as well as
the Archimedes Palimpsest, is compelling.
But one small thing doesn’t add up. The Guerson business was a
respectable and successful one on Boulevard Haussmann. The
Willoughby letter shows that Salomon Guerson knew what he had,
and he knew it was worth a lot of money. Conservatively, $6,000 is
$70,000 today, which was a lot of money for a medieval manuscript
in those days. Salomon Guerson thought the book was valuable pre-
cisely because he knew it contained the writings of Archimedes.The
Canadians, for their part, had demonstrated in their report that the
forgeries could not have been done until after 1938. So we are left
171
9780306815805_07.qxd 8/22/07 9:26 AM Page 172
with the situation that the Guersons hung on to the book for at least
seven years, waiting patiently for someone to pay the appropriate
price for it, before suddenly making forgeries out of the mathe-
matician’s letter to Eratosthenes after 1938. Something doesn’t add
up. Salomon Guerson may have been a little unscrupulous in his
treatment of Byzantine manuscripts, but he didn’t have sufficient
motive for this particular crime. We needed to find one.
The Casablanca Hypothesis

On Friday, June 14, 1940, the Germans entered Paris. They wore
gray; the Norwegian Ilse Lund wore blue. Czech resistance hero
Victor László was sick in a freight car on the outskirts of Paris.
American freedom fighter Rick Blaine stood on a station platform in
the rain with a comical look on his face because he’d been stood up
by a girl. Rick got on the train and left Paris, along with three and a
half million others. He eventually made it to Casablanca, where he
made a handsome profit at his Café Américain, which was mainly
populated by once-wealthy Europeans, most of whom were selling
family treasures for a song in order to bribe their way to safety. This
is the plot of a movie starring Humphrey Bogart and Ingrid
Bergman. Reviel and I have a similar plot to account for what hap-
pened to Archimedes. It is just as short on hard facts as the movie and
should be similarly understood as fiction. It has a much darker plot
than Casablanca. Here is the summary:
On Friday, June 14, 1940, the Germans entered Paris. Salomon
Guerson and Archimedes do not leave town, at least not on the same
day as Rick Blaine. Salomon thinks that he can stick it out in Paris.
Forty-eight hours later he isn’t so sure: all Jews are ordered to reg-
ister at a police station. But he is still there on Wednesday the
twenty-sixth when Hitler arrives. Salomon is thankful that Hitler
heads down the Champs-Elysées from the Arc de Triomphe rather
than up the Boulevard Haussmann, but he can hear the noise down
172
9780306815805_07.qxd 8/22/07 9:26 AM Page 173
the street from his shop, which he has closed for the day. Salomon
never reopens his shop. Its contents are plundered by the Nazis.They
take any artwork of value to the Jeu de Paume to be sorted and then
ship it back to the motherland. Salomon goes into hiding, with just
a few of his possessions. One of them is the Palimpsest. It is small,
portable, inconspicuous, and, he thinks, valuable.
Two years later, Salomon gets increasingly desperate. On
Wednesday, July 16, 1942, the Vichy police begin deporting Parisian
Jews, rounding them up at the Winter Velodrome. Their more per-
manent deportation camp is at Drancy, from which, in the next two
years, 70,000 people, including many of his friends are shipped to
Auschwitz and disappear. Salomon is struggling to stay alive. He looks
at his remaining assets; he is reluctant to part with the Palimpsest, but
eventually he must. He cannot sell it himself, of course; if he tries, the
book will simply be confiscated. He decides to give it to a friend to
sell on his behalf. But Salomon finds himself short of friends and the
friends he does have find the book a hard sell, at any price. Finally
Salomon turns to Marie Louis Sirieix. He is hopeful of a good
reception. Sirieix is a Resistance hero and his daughter, Anne, is
married to someone with an extraordinarily similar surname—
Guersan. Sirieix is sympathetic, and he even believes Salomon when
Salomon tells him that it is the unique key to the mind of
Archimedes. But Sirieix also says that no German will believe that it
is Archimedes, and anyway, the Nazis are not interested in ugly books.
They are interested in art. The Germans have been systematically
looting art from Jews in Paris for months. Sirieix is a freedom fighter,
not an intellectual, and he takes the pragmatic approach that if only
the book had pictures in it, then it would have real currency. Indeed,
it would be more valuable than gold.
Salomon Guerson leaves with a seed planted. The screen goes
black for a moment and resumes with Salomon, a few days later,
returning to Sirieix. He says that he hadn’t noticed before, but there
are several pictures in his book. Sirieix is suspicious but generous. He
173
9780306815805_07.qxd 8/22/07 9:26 AM Page 174
is much more impressed with Salomon, and he agrees to buy the

book. Archimedes’ parchment letter to Eratosthenes becomes
Salomon’s letter of transit, and it is now covered in pictures. Salomon
successfully escapes Paris; Sirieix returns to fighting the Germans,
confident of an eventual victory. He has never been particularly inter-
ested in the Palimpsest and hides it in his damp basement.The credits
roll over a backdrop of the Palimpsest slowly gathering dust and
being devoured by a purple mold.
A Plea to the Reader

Abigail now had the data she needed to start work. She had all the
documentation from the Canadian Conservation Institute, she had
undertaken her own research, and others had helped too. It was clear
that once Anne Guersan paid attention to the book that she had
inherited, she took steps to have it restored. But she may have only
contributed further to Abigail’s problems.The PVAC glue that locked
the Palimpsest pages together was widely used for the spines of man-
uscripts in the sixties and seventies. Perhaps this is another example
of someone trying to do their best for a book and making the
problem worse. Someone also seems to have paid particular attention
to the forgeries, keeping them separately and mounting them with
Blu-tack, which only came on the market in 1970.
There were, however, several leaves that Abigail could not do any
work on.These were the three that I had noticed were missing when
Mr. B first left the book with me. They were present when Heiberg
looked at the codex, and he even took a photograph of one of them.
Abigail had found traces of pigment transferred on to the facing folios
of these now-lost leaves. It was safe to assume that the forger had
painted these folios as well. Some forgeries made from the book might
have been successfully sold and they might be decorating the walls of
an apartment in Paris, in Germany or, perhaps more likely, in the
United States. Look out for them. If you see them, turn them over. If
they have two texts on the back, one much fainter than the other, then
174
9780306815805_07.qxd 8/22/07 9:26 AM Page 175
please let me know.They are very valuable and not because they have
paintings on them. Remember: value translates into cash. You can
contact me through http://www.archimedespalimpsest.org.
Intensive Care
If caring for Mr. B’s book sounds like a daunting job, then I have suc-
ceeded only in hinting at the true magnitude of the task facing
Abigail and her colleagues in the book and paper conservation lab at
the Walters. Every aspect of the material characteristics of the codex
has been investigated. The Canadian Conservation Institute took a
microscopic core sample from a folio containing the Archimedes
text. Analysis of the sample only reinforced the fact that the collagen
was breaking down and that the surviving palimpsested text was
extremely thin—a mere stain engrained in the parchment. Before any
work was done on a folio, a color-coded map was made recording its
condition—the tears, the drops of wax, the mold stains, the rust, and
the Blu-tack. The manuscript was also comprehensively pho-
tographed. Each bifolio of the prayer book had its own written con-
dition report, its own treatment proposal, and its own treatment log.
If the treatment proposals had been for patients in hospital, then those
patients would have been in the intensive-care unit. And intensive
care is exactly what they were given. As she worked, Abigail saved
everything.To this day there is a box of carefully bagged fragments of
the Palimpsest. Each bag is labeled, telling us which bits of thread,
glue, wax, pigments, and paper came from which folio of the prayer
book. The myriad tears in the parchment were given tiny mends, so
that more bits did not fall off when they were imaged.
It was not until Saturday, November 8, 2003, that Abigail started to
treat the beginning of Archimedes’ letter to Eratosthenes. I have
edited her treatment log, which might not be a great read but which
does demonstrate the intensity with which she worked on the only
surviving copy of the letter.
175
9780306815805_07.qxd 8/22/07 9:26 AM Page 176
On that Saturday, Abigail separated it from the bifolio that was

wrapped around it. It took all day to do this, because the two
bifolios had been stuck together with PVAC. She removed
loose debris from the gutter. The next day, Sunday, Abigail
relaxed the parchment at the spine by lightly brushing it with
a mixture of isopropanol and water. Then she made a tracing
of both sides of the bifolio, showing the damaged areas and
those that had been obscured by glue and paint. This took two
hours. Among other things that Abigail documented on her
tracing was a paper reinforcement that had been attached to
the spine-fold with PVAC and which obscured Archimedes’
text. Abigail also plotted on her tracing several small fragments
of parchment and a small deposit of purple paint. She con-
cluded that one of the forgeries had once been stuck in the
book next to this leaf. Abigail then comprehensively pho-
tographed the bifolio. Then she applied more isopropanol and
water to the paper reinforcement. After 15 minutes she began
to take it off. By the end of the day, she had removed it com-
pletely. Monday was a day of recuperation, not for Abigail, but
for the bifolio. On Tuesday, November 11,Veterans’ Day,
Abigail looked at the other side. Around the spine-fold, there
were large gritty deposits, colored fibers, white fibers, black
accretions that might have been hide glue, and white crys-
talline particles that she thought might have been silica gel,
put there possibly in 1971 by the Etablissement Mallet when
they were trying to stop the mold. She also noted further
blobs of PVAC on the bifolio. She began cleaning the residual
adhesive and accretions from the spine-fold. The fold itself was
very cockled and creased; Abigail tried to flatten out these
creases. All the residues and accretions that she removed that
day were saved. She did not return to the bifolio again until
Sunday, November 16. In the spine-fold, around the fourth
sewing station, Archimedes’ text was obscured because the
parchment was torn and crushed and a loose flap of it was
176
9780306815805_07.qxd 8/22/07 9:26 AM Page 177
FIGURE 7.3 A section of the Palimpsest, before treatment, after treatment,

and in ultraviolet light. Note Netz’s circled “kuklos”
embedded with the PVC and white paper fibers. Repeated

applications of ethanol and water freed the flap, which was
realigned. The area was then dried under pressure. The next
day Abigail started work on several flaps of parchment that
were curled up in a severely degraded and perforated area of
the bifolio. She applied tiny amounts of ethanol, the flaps
gradually relaxed, and Abigail secured them back in place. The
area was then dried under pressure. On that day Abigail also
worked on areas of parchment that were folded over around
the edge of the leaf and these she reinforced with Japanese
paper. Abigail never attempted to remove wax droplets on this
leaf, because it was simply too fragile. When she had fully pre-
pared the leaf for imaging, she sent a sample of the PVAC for
analysis at the Canadian Conservation Institute.
All the other folios of the manuscript got this same level of care. I
show you one untypically spectacular example. Abigail mended the
gutter of one of the Archimedes folios. Here, a flap of parchment had
been scrunched up and broken and it had to be unfolded so that the
text could be read. Abigail performed brain surgery that morning.
Later in the day, we took an ultraviolet photograph of the folio and
177
9780306815805_07.qxd 8/22/07 9:26 AM Page 178
sent it as a jpeg attachment to Reviel Netz.This is the email I got by

return:
From: Reviel Netz
Sent: Sun 4/15/2001 10:14 AM
To: “William Noel”
Dear Will,
The attached is your recent AQ picture, fantastic. Circled is
the self-obvious symbol for the Greek word Kuklos, meaning
“circle.” First time I’ve seen this symbol in the Archimedean
manuscript tradition, and this has consequences for working
out the relationship between the branches of the tradition, as
well as of course for the history of mathematical symbolism.
Kudos to AQ.
In his enthusiasm, Reviel was actually wrong; Heiberg had noticed

the kuklos. Luckily Reviel only told me this a lot later—thank
goodness. It was one day when our spirits rose and those days were
rare and precious, because, basically, preparing the Archimedes leaves
for imaging was a time-consuming nightmare.
Most of Abigail’s work occurred under the microscope and most of
it is now not apparent to the naked eye. The flash of a camera and a
clever algorithm, as we shall see, can transform a page. A brilliant
scholarly insight can transform our understanding of Archimedes. But
Abigail and her colleagues always knew that this was not what their
work was about. Except to the very careful observer, other than the fact
that they are disbound, the folios of the Palimpsest look little different
after Abigail’s treatment from how they looked before she started work.
I had originally hoped to see a dramatic change in the forgery
pages. In fact, I wanted Abigail to scrape the forgeries off. But other
people had very different views. John Lowden, for example, con-
sidered that they were an important part of the history of the codex.
What was for me merely incidental graffiti that obscured the writings
of a genius was to John a record of twentieth-century attitudes toward
178
9780306815805_07.qxd 8/22/07 9:26 AM Page 179
the Byzantine past. This was a view that Abigail shared. And she
brought two more observations to the table. First, if she tried to take
off the images, she might destroy the Archimedes text underneath.
Second, even if we did not have the technology to read through the
forgeries now, the technology might exist in the future. We could
always wait. After all, Heiberg waited. He had not taken the codex
apart, and he had not painted it with Gioberti’s tincture. He must have
been sorely tempted, but, for the good of the codex, he didn’t.
Eventually I saw Abigail’s point. More importantly, so did Mr. B.
Yet it was here, truly in the emergency room, that the tide of the
project almost imperceptibly changed. It was upon Abigail’s patient
work that the later triumphs of the project were built. For, by hook
or by crook, she did do it. She took the thing apart. The disbinding
started on April 3, 2000. The final folios were disassembled on
Thursday, November 4, 2004. On average, one palimpsested folio was
liberated from the prayer book every fifteen days. After she had pre-
pared them as best she could, the folios were mounted in specially
prepared mats so that the scientists could image them.
Discoveries
Of course, the scholars did not wait until the scientists had taken their
images of these folios. Their view was that, if they were in good
enough shape for the scientists to image, they were in good enough
shape for them to try to read. Once Abigail had started to disbind the
leaves, Natalie, Reviel, and John Lowden came and studied them.
The first scholarly discovery came quickly, on April 3, 2000. On
the very day that Abigail took the book apart, Reviel and Natalie
were scrutinizing parts of the Palimpsest with ultraviolet lamps.They
sat next to each other conferring. Understandably, the first folio came
under their gaze quickly. I have explained already that this page is in
very bad condition. But as Reviel stared at it under UV, he thought
he saw underwriting, and he thought it was underwriting by the
Archimedes scribe. He discussed it with Natalie. Natalie, too, studied
179
9780306815805_07.qxd 8/22/07 9:26 AM Page 180
the leaf. “Yes,” she said. They had just discovered a new page of
Codex C. The very first page of the codex contained Floating Bodies
text in Greek that Heiberg simply hadn’t noticed. On the first day of
disbinding, on the first day of reading, and on the first page of the
codex we had discovered a whole new page of Floating Bodies in
Greek. It was a major triumph for the project.
It was slowly becoming apparent that Heiberg had not known the
manuscript quite as well as people thought.This became even clearer
when Reviel and I went to see the Heiberg photographs.There were
sixty-five photographs in all, and they were marked up with notes by
Heiberg, who identified folios of the Archimedes text as he went
along. Naturally, all the photos were of folios that contained
Archimedes text. We could see the way he worked. He labeled those
folios containing the Method “M,” Stomachion “St,” and the rest he
labeled by reference to his own previous edition, which he had pub-
lished in 1880. However, there were only sixty-five photographs. Of
these, thirty-eight were openings, the rest were single folios. Heiberg
had photographs of only 103 rectos and versos of folios out of a
codex that contained, in his time, 354. One photograph in particular
caught Reviel’s eye. It was of the right-hand side of an opening. But
I noticed that Reviel was not looking at this folio; rather he was
looking at the little bit of the preceding folio, on the left-hand side
of the original opening, which was by chance included in the pho-
tograph. It contained only three lines of Archimedes text. Reviel
looked, and looked again. “This is Floating Bodies,” he said. Reviel
knew instantly that he was reading a section of Floating Bodies in
Greek and for the very first time. Heiberg had overlooked the folio
on the left. He had not taken a photo of it, and he had not transcribed
it. Looking at the photographs now, one can only admire Heiberg’s
skill. It is extraordinarily difficult to read the Archimedes text from
the photographs, and he had read them. Nonetheless, there were
whole sections of Method and Floating Bodies for which he hadn’t
taken photographs, and which he had left largely unread.
180
9780306815805_07.qxd 8/22/07 9:26 AM Page 181
The best day for Abigail and me in this challenging period was
Saturday, April 13, 2002. John Lowden was in Baltimore to look at
the Palimpsest. I knew that he had to fly to London at 3 p.m., so I
went down to the conservation laboratory at about midday to see
how he was doing. He popped his head out of a black curtain to tell
me that the Palimpsest had been presented to a church 773 years ago
. . . exactly. With the help of an ultraviolet lamp John had looked at
the very first folio of the Palimpsest. The first folio of any codex is
normally in worse shape than those that follow, but the first folio of
the Palimpsest is a wreck. Bookworms don’t actually like parchment
folios, they like the wooden covers in which the folios are bound.
But they don’t have a very good sense of direction. The first folio of
the Palimpsest is covered in wormholes. Also, the outside two inches
of the folio are stained very dark by the oils in the leather of an old
binding. John gave it another look. In the bottom margin, right in
the stained area, he discovered an inscription, technically called a
colophon. He could not decipher it completely, but it was clear that
the prayer book had been given to a church by the scribe on April
14, 6737. But this is in Greek Orthodox time. In the thirteenth
century, time was not calculated from the birth of Christ, but from
the origin of the world. As we all know, the world was formed on
September 1, 5509 BC. To get the modern date for April 14, 6737
one must therefore subtract 5,508. The answer is April 14, 1229.
Seven hundred and seventy-three years later, we knew when the
Palimpsest was made.
Reflections
As this story unfolded, a picture of the history of the Archimedes
Palimpsest was emerging in my mind, which is at odds with the one
I have previously recounted. The roles of those who had played their
part in the history of the book now seemed extraordinarily different.
I began to question my hasty condemnation of the scribe who had
181
9780306815805_07.qxd 8/22/07 9:26 AM Page 182
erased the Archimedes text, to rethink my description of St. Sabas as a

tomb for Archimedes, and to feel shame for my mistrust of the monks
of the Metochion. It was because the monks of the Metochion had
their manuscripts documented that the Archimedes texts were redis-
covered; St. Sabas is better characterized as a safe house for Archimedes
than as a tomb. If the price for that safety was the Christian disguise
that the scribe of the prayer book provided, it was a price well worth
paying. And if it was love of mathematics that had ensured the survival
of Archimedes’ letter to Eratosthenes for the first thousand years, it was
love of God that ensured its survival to the twentieth century.
The scribe was the unwitting savior of Archimedes and not his
nemesis.The Palimpsest was the creation of religion, not its victim. It
was the victim, rather, of two world wars and the art market. It was
the damage sustained in the twentieth century that had led most to
believe that the Palimpsest was now a battered relic of little research
interest. Reviel was convinced that this was a false assumption and
the breakthroughs that we made only increased his insistence that he
be allowed to see the book. Once a few leaves were disbound he
arranged a trip to Baltimore together with his friend Ken Saito.They
were going to come the first weekend in January 2001, the sixth and
the seventh. This was the moment of truth. Could Reviel and Ken
really get more Archimedes text out of the book than Heiberg? Every
time I spoke to the press I said, “Yes”; every time I looked at the
wrecked folios of the book I thought, “No.” In the end, it didn’t
matter what I said or what I thought. It merely mattered that the
book now had its chance and that Reviel was flying in from
California to pass judgment.
182
9780306815805_08.qxd 8/22/07 9:27 AM Page 183
8
Archimedes’ Method, 2001
or Infinity Unveiled

t was nearing the end of 2000. The Palimpsest had been available
I for research for almost two years, and yet, there was very little to
show for it. I was driven back to the traditional routines of library
work: visiting the manuscript; holding a magnifying glass in one hand
and a UV lamp in the other; and looking at the manuscript intently,
one character after another. I envisaged myself doing just this—slowly
and painfully going through the manuscript. To be honest, though, I
wasn’t sure I would be able to read much more this way than Heiberg
did. Would it all be just a waste of time?
Will insisted that we had to make priorities. Only a few pages
could be made available for my next trip in January 2001, so which
ones would I really would like to have? Only a handful of the
potential folios were from the Method and of those only one had a
substantial gap left by Heiberg’s own transcription.
I therefore asked for the bifolio 105–110 to be ready for our visit.
I wouldn’t be alone: I had a guest with me, a tourist or, more pre-
cisely a pilgrim. For now, while the Palimpsest was there, this was
how historians of mathematics would think of a trip to Baltimore—
as a pilgrimage. I knew Ken would value the experience and, besides,
I like talking to him. Who knows? We might even get to find out
something about Archimedes.
Professor Ken Saito teaches at the University of Osaka, and he is
one of the best historians of mathematics at work today. I have always
admired his early study of the way in which Euclid’s results were used
in the theory of conic sections. He is a master of the logic of Greek
mathematics: when he reads a text, he sees precisely where it comes
183
9780306815805_08.qxd 8/22/07 9:27 AM Page 184
from and where it leads to. If anyone could work with me on the
Palimpsest, I knew this was the man.
Saito first came to visit me at Stanford. It was his first time in
America, and I thought he would like to visit my advanced Greek
class. I had my students translate Euclid and Archimedes (most of the
Greek language students at Stanford can easily handle the mathe-
matics), and I enjoyed showing my students off. We also spent a day
in San Francisco, which Ken enjoyed a lot, although I think he just
couldn’t wait to travel, at long last, to Baltimore.
We had a long flight ahead of us—enough time to prepare for our
visit. As we were flying to Baltimore, Ken and I discussed some of the
perennial questions of the history of mathematics.To what extent did
Archimedes anticipate the calculus? How much did he know about
its conceptual difficulties?
Here, in outline, is the history of mathematics as it was known,
back then, in January 2001. The Greeks invented mathematics as a
precise, rigorous science. They avoided paradox and mistakes. In
doing so, they also avoided the pitfall of infinity. Their science was
based on numbers that can be as big as you wish, or as small as you
wish, but never infinitely big or small. Numbers that are as big or small
as you wish are known as “potentially infinite,” instead of actually
infinite. The Greeks did not use actual infinity.
In the scientific revolution of the sixteenth and seventeenth cen-
turies, scientists such as Galileo and Newton incorporated new tech-
niques into mathematics by employing actual infinity. They utilized
magnitudes that were in fact infinitely small or infinitely big. This
allowed many important breakthroughs, but there was a price to pay:
infinity introduced the paradoxes and errors that follow upon it.
Mathematics became more powerful but less precise.
In the nineteenth century, mathematicians created new techniques
for dealing with infinity. Gradually, a new mathematics evolved where
infinity was introduced and tamed, so to speak. One could deal with
infinity, without any paradoxes or errors. The precision of Greek
mathematics was regained, on a new level with infinity being used as
184
9780306815805_08.qxd 8/22/07 9:27 AM Page 185
INFINITY UNVEILED
a precise mathematical tool. This allowed the great explosion of

mathematical discoveries—and therefore of scientific discoveries—in
the nineteenth and twentieth centuries.
In a nutshell, the Greeks had precision without infinity.The scien-
tific revolution had infinity without precision. Modern science, since
the nineteenth century, has had both precision and infinity.
What was the potential infinity used by Archimedes? Recall the
imaginary dialogue. Archimedes packs a curved object so that a
certain area has been left out, an area greater than the size of a grain
of sand. A critic comes along and says: “There is still a difference
greater than the size of a grain of sand.” “Is that right?” exclaims
Archimedes. “All right then, I shall apply my mechanism successively
several more times,” and then the area left out is smaller than the
grain of sand. “Wait a minute,” says the critic, “the area left out is still
greater than a hair’s width.” Archimedes goes on, and so on, and on
and on it goes.The difference always becomes smaller than any given
magnitude mentioned by the critic.This dialogue goes on indefinitely.
This is potential infinity.
Let us take another example. First consider the collection of whole
numbers using only the principle of potential infinity. Then say that,
for each whole number, no matter how large, we can think of
another one bigger than it.This is another imaginary dialogue, a kind
of auction.You say a million, I say two million; you say a billion, I say
a trillion. This bidding sale has no end. But no one is allowed to say
infinity itself; there is no such number allowed.
But then, the guidelines are altered, and we can use actual infinity.
Suppose someone enters the auction and says, “I have a number,
which is even bigger than all the numbers you have mentioned. This
is the number of all whole numbers. It says how many whole
numbers there are.” The auction ends with a bang: actual infinity has
brought it to an end.
There are clearly more whole numbers than a million, a billion, or
a trillion. The number of all whole numbers is infinity. And it gives
rise to all the paradoxes of infinity.
185
9780306815805_08.qxd 8/22/07 9:27 AM Page 186
Suppose, for instance, you wish to compare the number of whole

numbers to the number of even numbers. We may put them side by
side in two rows.
1 2 3 4 5 ...
2 4 6 8 10 ...
For each number in the top row there is a number in the bottom
row (its double). The bottom row does not get exhausted, ever. For
each whole number there is an even number and vice versa. The
number of whole numbers is the same as the number of even
numbers. In the case of whole and even numbers, we find that they
are the same size even though there are clearly, in some sense, twice
as many whole numbers than there are even ones. In infinity,
“normal” concepts collapse: a collection may be equal to its half. And
so we cannot count on ordinary rules of addition and summation.
Infinity gives rise to too many paradoxes.This is why it is such a dif-
ficult tool to handle.
In the nineteenth century, mathematicians found the techniques to
calculate with infinity. (The main insight came from Archimedes’
imaginary dialogues.) The Greeks never made this step (it is a big
one). They had collections that were “as big as you wish,” but never
collections that were actually infinite.
Even in the Method—so we thought, back in January 2001—
Archimedes did not break this rule. He played, dangerously, with
infinity. But he did not speak of “the collection of all the parallel lines
in the triangle.” All he said was that, since each parallel line balanced
its paired section with a certain fulcrum, so would the entire triangle.
Whether this was true or not depended on techniques of summing
up infinitely many objects. But Archimedes never explained what he
was relying upon for his summation. Even in his most radical exper-
iment actual infinity was avoided. It was left for Galileo and Newton
to uncover it.
So went our conversation. During breaks, Ken Saito would return
to the book he had brought with him: a copy of Heiberg’s edition of
186
9780306815805_08.qxd 8/22/07 9:27 AM Page 187
INFINITY UNVEILED
Archimedes’ Method. True to form, Saito was immersing himself in

the ancient text.We were about to read a hitherto unknown piece of
the text. For some reason—we didn’t know why—Heiberg had a gap
in his edition. What could Archimedes have written there? To make
sure that we got to the bottom of it, Saito wanted to know every-
thing about the context.
This, then is the overall structure of the Method. Archimedes begins
with an introduction addressed to Eratosthenes. It is a very flattering
introduction: “You are such a mathematician that you are capable of
passing a real judgment on my method.” As we were talking, I sug-
gested to Saito one of my hobby horses. Wouldn’t it be like
Archimedes, I said, to be ironical in this introduction? That is, did
Archimedes not intend, perhaps, to unmask Eratosthenes? I suggested
that we think of the Method as a puzzle sent to Eratosthenes and
intended to defeat him. It is, after all, a very puzzling text—could it
be intentionally so?
Perhaps, nodded Saito, and he returned to the text.
The Method is indeed a puzzle. We have followed the first propo-
sition with its remarkable combination of physics, mathematics, and
infinity. Proposition 1 has two striking properties: the application of
physics to mathematics and the summation of infinitely many lines.
The same combination is repeated throughout the first thirteen
propositions of the Method.
Archimedes promises in his introduction that, at the end of the
treatise, he will repeat the proofs for some of the results, deriving them
now in an “orthodox,” standard way. Most of this disappeared either
when the palimpsester discarded parts of the original manuscript, in the
year 1229, or at some later time, when a few of the prayers for reasons
we do not understand got cut out of the prayer book. Either way, much
of the end of the Method was lost. But proposition 15 survives in part,
and it is indeed an orthodox, standard proof based on the imaginary dia-
logue of: “I shall find you an even smaller magnitude.”
Proposition 14 is different. It is neither an orthodox proof, nor is
it like the first thirteen propositions of the Method. It does not rely on
187
9780306815805_08.qxd 8/22/07 9:27 AM Page 188
the combination of the application of physics to mathematics and

infinite summation. Instead, it is based on infinite summation alone.
Not that people have paid much attention to it throughout the twen-
tieth century. Propositions such as proposition 1 appear to be puz-
zling enough.Why bother with this one as well? Especially since this
proposition survives in fragmentary form only. Heiberg could read its
beginning and its end, but not its middle. The writing was too faint.
Could it be read now?
This was precisely why Saito was preparing himself. We were
going to look at the middle of proposition 14, the part that Heiberg
had not read. Abigail had just unwrapped it, and the scientists were
about to produce a digital image.
This would be the first major trial for the project. Either we make
readings where Heiberg had failed or we give up, content with
Heiberg’s edition. Perhaps he had done as much as could be done.This
would be a pity in terms of the Greek edition, where every word
counted. But I doubted how much interesting material could be
revealed for the history of mathematics. In truth, I thought I could use
the Palimpsest to further the understanding of the cognitive history of
mathematics—looking at such questions as the nature of diagrams and
abbreviations. In terms of the traditional concerns of the history of
mathematics, I doubted that the Palimpsest could teach us much that
was new. Perhaps we would be able to read something, perhaps not. But
it would not be of much consequence to the history of mathematics.
After all, the general outlines of Archimedes’ principles of sum-
mation were clear enough from the first thirteen propositions of the
Method. Heiberg had not read the middle of proposition 14, but he
had read enough of other propositions to develop a clear enough
guess of how Archimedes would proceed here. The object of the
proof was clear, from its beginning and end. Archimedes was meas-
uring the volume of a cylindrical cut.That is, we take a cube (see fig.
8.1). We enclose a cylinder in it. We cut the cylinder (and the cube)
by an oblique plane passing through the middle of the base of the
cube and an edge of its top. What is the volume of the cylinder that
188
9780306815805_08.qxd 8/22/07 9:27 AM Page 189
INFINITY UNVEILED
FIGURE 8.1
has been sliced off by this cut—the fingernail-like shape that is

shaded in figure 8.2? This is a very strange figure, framed by a com-
bination of a semi-circle, a semi-ellipse, and the contours of a cylin-
drical surface. This is the entire point for Archimedes. A very strange
and unwieldy figure is going to be measured, very precisely, in terms
of a rectilinear figure. Let us see how Heiberg understood this meas-
urement. (For the more geometrically inclined readers, please note
now that I am treating the enclosing figure as a cube. Archimedes
himself approaches the problem in the more general terms of any
parallelepiped, but we gain a lot in simplicity by considering the case
of the cube alone, which is then very easy to transfer to the general
case treated by Archimedes.)
Once again, then, we witness the measurement of a curvilinear
object. And once again—as everywhere else in the Method—
Archimedes uses some kind of slicing by parallels. Heiberg was capable
of reading that a random plane is drawn, parallel to the vertical edge
189
9780306815805_08.qxd 8/22/07 9:27 AM Page 190
FIGURE 8.2
of the cube (see fig. 8.3). This results in various cuts being sliced off
from the original cube and the original cylinder, as well as from the
bases of those figures. Archimedes considers certain planes and lines
and derives certain proportions (we will look at these in more detail).
Heiberg followed all of this. And then—the gap in the argument.
Heiberg could read no further for a long stretch and then, when he
picked up the text again, he was already near the end of the propo-
sition. There Heiberg found Archimedes’ conclusion, which was that
the cylindrical cut was exactly one sixth of the entire enclosing cube.
How had Archimedes got there? Had he actually proved his result?
Heiberg couldn’t read the relevant passage. Everyone since Heiberg
has assumed that it involved the same kind of implicit summation used
in proposition 1. That is, having obtained a proportion for the ran-
domly chosen slice,Archimedes implicitly transferred that result to the
entire cylindrical cut—the way Archimedes moved in proposition 1
from the randomly chosen parallel line to the triangle and parabola
taken as a whole. This was everyone’s guess; and my guess, too—
nothing new would come out of this proposition. Saito, meanwhile,
was immersing himself in his text: perhaps we were wrong after all?
190
9780306815805_08.qxd 8/22/07 9:27 AM Page 191
INFINITY UNVEILED
FIGURE 8.3
The Volume of a Cylindrical Cut

Saito and I finally sat there, in front of the bifolio 105–110.We turned
quickly to the area that Heiberg had left as white lines, remarking in
a footnote: “quid in tanta lacuna fuerit dictum, non exputo” or “I shall
not speculate as to what could have been written in such a large gap.”
In order to follow the mathematical argument in more detail, let
us concentrate on just the object that interests us. In figure 8.4, we
“slide out,” as it were, a section of the cube—the triangular prism cut
off by the inclined plane. From now on we will concentrate just on
this triangular prism. We now draw three further diagrams, recon-
structing the text as far as the lacuna. These diagrams (see fig. 8.5–7)
are simply three different views of this triangular prism. It is so dif-
ficult to perceive that one needs to view it from certain angles, simul-
taneously, to gain a sense of it. (Archimedes, however, visualized the
object clearly enough on the basis of figure 8.7 alone!)
191
9780306815805_08.qxd 8/22/07 9:27 AM Page 192
FIGURE 8.4
In figure 8.5, we see the entire complicated triangular prism with

a randomly chosen plane running parallel to the upright edge of the
cube.
Figure 8.6 shows the randomly chosen plane from the side.The ran-
domly chosen plane cuts off a triangle from the original triangular
prism. It also cuts off a smaller triangle from the original cylinder.
Thus, figure 8.6 shows two triangles, one enclosed within the
other—the bigger triangle, from the triangular prism, and the smaller
triangle, from the cylinder. The smaller triangle is especially
important, because the strange figure that we are about to measure—
the fingernail-like figure cut off the cylinder—is made up of the col-
lection of all such triangles cut off from the cylinder. (These triangles
get bigger and bigger as the randomly chosen plane gets removed
further away from the edges of the cube.)
Figure 8.7, finally, offers a bottom view. The randomly chosen
plane creates not only triangles from the triangular prism and the
192
9780306815805_08.qxd 8/22/07 9:27 AM Page 193
INFINITY UNVEILED
FIGURE 8.5
Cut off from

the half-cube
Cut off from

the half-cube
Cut off from the Cut off from the circle

half-cylinder
Cut off from
the rectangle
FIGURE 8.6 AND 8.7
193
9780306815805_08.qxd 8/22/07 9:27 AM Page 194
cylinder but also line segments on the bottom of the cube. The
bottom of the half-cube is a rectangle. This rectangle is the footprint
of the entire triangular prism. Within this rectangle we also find a
semi-circle.This semi-circle is the footprint of half the cylinder—the
half whose slice we are about to measure. To really spice things up,
Archimedes then drew another curve inside this rectangle, and—
what do you know—this figure happened to be a parabola! And so
we have, in figure 8.7, a rectangle, inside it a semi-circle, and inside
that a parabola. And then, a straight line passes through them all. The
line is the footprint of the randomly chosen plane.
The randomly chosen plane creates a line through the rectangle.
This line is the base of the greater triangle of figure 8.6, the one that
is cut off from the triangular prism. The randomly chosen plane also
creates a line through the semi-circle within the rectangle. This line
is the base of the smaller triangle of figure 8.6, the one that is cut off
from the cylinder. It also creates a line through the parabola. This
time, the line has no three-dimensional meaning in terms of figure
8.6 and functions in the context of figure 8.7 only. Once again, then,
we have several lines encased inside each other—one that cuts off the
rectangle, one that cuts off the semi-circle, and one that cuts off the
parabola.
We have now presented our cast, and we may concentrate on just
four actors.
The first two are from figure 8.6: these are the greater triangle and
the smaller triangle of figure 8.6. Let us call them “the triangle of the
prism” and “the triangle of the cylinder,” respectively.
The next two are from figure 8.7. We actually need only two out
of the three mentioned above.We need the line that cuts off the rec-
tangle, and the line that cuts off the parabola. Let us call them “the
line of the rectangle” and “the line of the parabola,” respectively.
These four cast members will now participate in a four-part
arrangement of proportion.
What Heiberg managed to read, prior to the gap in his reading,
was already a magnificent result. Archimedes, through tremendous
194
9780306815805_08.qxd 8/22/07 9:27 AM Page 195
INFINITY UNVEILED
geometrical ingenuity, had succeeded in proving that in such an

arrangement:
• The area of the triangle of the prism is to the triangle of
the cylinder as the line of the rectangle is to the line of the
parabola.
A triangle is to a triangle, as a line is to a line: two dimensions as
one dimension.
Here came the long gap in the text.The text picked up again with
Heiberg reading the following:
• The volume of the triangular prism is to the volume of the
cylindrical cut as the area of the entire rectangle is to the area
of the entire parabolic segment.
The triangular prism is to the cylinder as the rectangle is to the
parabolic segment: three-dimensional figures are to each other as
two-dimensional figures are to each other.
Archimedes’ geometrical discovery was that the triangles in figure
8.6 related to each other as certain lines do in figures 8.7. This, in
other words, is a statement about a random slice, exactly the same
type of statement that we saw in the first proposition of the Method.
It appears, therefore, that Archimedes was once again making the
move, implicitly, from a statement about random slices to a statement
about the entire objects from which the slices were taken. The first
triangle is a random slice of the triangular prism, the second triangle
is a random slice of the cylinder; the first line is a random slice of the
rectangle, the second line is a random slice of the parabola. And since,
in each random slice, the first triangle is to the second triangle as the
first line is to the second line, so, with the entire objects, the trian-
gular prism will be to the cylinder as the rectangle is to the parabola.
So the text states, with Archimedes appearing to make an implicit
move. But the thing is, there was, after all a gap—a space for Archimedes
to make his argument explicit.
What could that explicit argument have been?
195
9780306815805_08.qxd 8/22/07 9:27 AM Page 196
We could see why Heiberg had not made much progress.The page
was largely illegible. Once again, we found ourselves admiring
Heiberg for what he had been able to read. Even with the UV light
the gap seemed hopeless, and we turned instead to the passages
Heiberg had read already, trying to verify them. But even those we
could hardly read. How had Heiberg ever made it?
We looked again at the conclusion to the proposition. Having
shown that the half-cube was to the cylindrical cut as the rectangle
was to the parabola, Archimedes went on with a quick calculation.
We remember from the first proposition of the Method that a para-
bolic segment is four-thirds the triangle it encloses. Everyone knows
that the rectangle is twice the triangle it encloses. So how much is the
rectangle to the parabolic segment? It is as “two” is to “four-thirds”
which (to simplify a bit) is as “six-thirds” is to “four-thirds.” It is as
six to four or as three to two. The rectangle is therefore to the para-
bolic segment as three to two or, more comfortably put, the parabolic
segment is two-thirds the enclosing rectangle. We’ve got it! The
cylindrical cut is two-thirds the triangular prism that encloses it or,
better put for our purposes, it is four-sixths of the triangular prism
that encloses it. It will take a while to show the following result: the
triangular prism enclosing the cylindrical cut is exactly one-fourth the
original cube as a whole. The cylindrical cut is four-sixths the trian-
gular prism; so it is one-sixth of the original cube as a whole. The
strange, fingernail-like object is exactly one-sixth the cube.We’ve got
it! Yet another curvilinear object successfully measured using a recti-
linear object.
Another elegant result by Archimedes and no application of
physics, it appears this time.The triangles and lines are not put on an
imaginary balance.They are simply summed up: infinitely many pro-
portions being summed up into a single proportion. How does
Archimedes do this? Does he simply ignore the paradoxes, the errors
of infinity?
We would not give up. We went back to the gap in the text. We
swapped positions, first Ken looking at the page, then me. I was more
196
9780306815805_08.qxd 8/22/07 9:27 AM Page 197
INFINITY UNVEILED
trained and quickly it became clear that I should be doing the

looking and Ken should be writing down what I saw.
Will left the room, counting on our good behavior. After many
minutes of frustration, I did something I shouldn’t have done. I
slipped the bifolio out of its plastic encasing. Unprotected by the
plastic, the reflection from the ultraviolet lamp was clearer. I stared
hard into the area on the page—those white lines left by Heiberg—
trying to find some traces of Greek characters.
I thought I did see something. At first I discarded it, because it did
not make sense in the context. There was no reason for Archimedes
to be using such a word. But I did think I saw the three characters in
sequence—epsilon–gamma–epsilon, or ege.
“I think I see ‘ege,’” I finally told Ken. “Probably something to do
with megethos, the Greek word for ‘magnitude.’ This does not make
much sense.”
Because, you see, Archimedes was talking about certain concrete
geometrical objects—a cylinder, a triangle, and a parabola. In such a
context, a Greek mathematician would not make the transition into
speaking about magnitudes in general terms.The word “magnitude,”
with its generality, is appropriate not in a concrete, geometrical
context, but in a more abstract context such as in a study on the
theory of proportions or of magnitudes. It was as if, in the middle of
a calculation with concrete numbers, the text moved to a discussion
of the principles of calculation as such.
“Oh, that’s very interesting,” said Ken. I should explain that Ken
is a well-behaved Japanese scholar.That was an expression of extraor-
dinary agitation. I know he wanted to ask me whether I was sure, but
he probably thought such a question would be impolite.
“I am quite sure,” I said, looking at it again. Indeed there was less
and less doubt the more I looked at it. In fact, I began to see the
traces of a theta, immediately following the epsilon–gamma–epsilon.
It was epsilon–gamma–epsilon–theta. There was no doubt that
Archimedes was talking about a megethos. He was talking about
abstract magnitudes.
197
9780306815805_08.qxd 8/22/07 9:27 AM Page 198
This process of gradual certainty is so typical in the reading of the

Palimpsest. By getting acquainted with the page, you gradually learn
how to discard the noise; you learn to concentrate on the signal. It’s
a little bit like tuning through the airwaves.The radio signal is at first
noisy, but then you settle on it and it begins to transmit. I now
actually saw this “magnitude.” I stared at it, dumbstruck.
“Archimedes must be applying result 11,” said Ken. At first I didn’t
quite hear, looking as I was, so closely at the page. And then I began
to be even a bit annoyed. How could Ken be telling me what was
happening through the lacuna, based on a single word? I was still
looking for more words. Ken went on.
“In the introduction to the Method, Archimedes mentions that he
will use certain basic results. One of these results is proved elsewhere,
in the treatise on Conoids and Spheroids. It deals with general magni-
tudes. I always thought Archimedes meant this result to be used only
in the later parts of the Method, where he uses orthodox, geometrical
methods. But apparently he uses it in proposition 14 as well.”
Now this got me interested. I put the page back into its plastic
casing and picked up the copy of Heiberg’s edition that Ken had been
reading the whole time. By God, what he said did make sense!
We were now both feverishly going through the possible
argument—drawing figures, sketching proportions, seeing for our-
selves how this result could be used for bridging the gap from the
random slices to the entire objects. The result had to do with the
summation of proportions. It did make sense. Archimedes could have
been summing up proportions. So it was not an implicit move after
all. Archimedes must have had an argument.
“But wait, Ken; there’s a problem.” I stopped, tearing myself away
from the figure we had just drawn. “If this is right, then Archimedes
must be summing up a collection made of infinitely many magni-
tudes. This does not sum up. It becomes infinite; you can no longer
calculate with it.”
Ken agreed. Something was still missing. The basic result used by
Archimedes was based on a proposition proven in Conoids and
198
9780306815805_08.qxd 8/22/07 9:27 AM Page 199
INFINITY UNVEILED
Spheroids. In this work, it was clear that it could be proven only for
sums involving a finite number of magnitudes, because otherwise you
would have to speak about an object made up of infinitely many
magnitudes, which made no sense. That was actual infinity. How
would Archimedes even go about talking about it?
“If one thing is clear, it is that the Greeks did not use actual
infinity. There is something wrong here. Or else, something very
new.”
This was clear, indeed. It was January 2001, and we knew that we
had hit on a major discovery for the history of mathematics. Just
what, though?
Could we simply be mistaken? I trusted Ken’s insight into Greek
mathematics. It all made sense. And I was sure I did see that word,
those four letters epsilon–gamma–epsilon–theta.Well, at least the first
three were certain. . . . But could we base a new interpretation of
Greek mathematics (a new twist to the entire trajectory of Western
mathematics) on this evidence alone?
That evening Ken and I explained to Will that we really, really
wanted to see digital images of this page. An entire chapter in the
history of science was waiting to be written based on those images.
The Method, March 2001

I kept going in and out of my office, checking my mailbox. When
would it come? It took the imagers just over a month and then, early
in March, the CD-ROM arrived. I knew this CD-ROM contained
high-resolution, sharp digital images, made with UV, of a single piece
of the Palimpsest. It was a small piece of the Palimpsest, no more. But
this piece included one side of the bifolio 105–110. The side I truly
needed to read.
For the rest of the day, I put all my other work aside. I was looking
at a jungle of digital traces, enlarging them so that the pixels blew up
and then reducing them again to see the entire picture. I was getting
my mind tuned off the noise and tuned in to the signal.
199
9780306815805_08.qxd 8/22/07 9:27 AM Page 200
I could easily pick up the εγε again. It was clear as daylight in the
digital image. Indeed the theta was very clear as well, and, looking for
it, I could quickly pick up several other appearances of the word
“magnitude.” Undoubtedly this was what Archimedes was talking
about. A few additional words became visible as well, referring to
certain geometrical objects—a cylinder here, a rectangle there.
Archimedes was probably applying the general principles of sum-
mation of proportion to the concrete, geometrical terms of the figure
at hand. And there was no doubt: the digital images made a world of
difference.
After making those first inroads, the reading stalled. This was once
again a typical part of the cycle of reading. Once you gained your
first, easy conquests, there would come a pause.There were no longer
any “easy” words to read, even with the digital images.We needed to
do some work. Looking and thinking: what could those traces mean?
I looked in this manner for a couple of hours, making little further
progress. I needed to clear my head a little, so I went out for a walk,
after which I took another look. Just out of curiosity, I stared not at
the line of writing itself but a little bit above it. Something there
arrested my attention. It was not just a smudge of digital noise; it had
the texture and consistency of real ink. Blowing up the pixels, I saw
it—the kind of trace that I would normally skip as being too incon-
sequential but one that, in such a fragmentary text, could be mean-
ingful. It was an accent mark, an acute sign above the line, like this: ´.
And because of my acquaintance with the scribe, I knew it was the
type of acute sign the scribe used on top of an iota. It is rather like
reading the dot of an “i” and identifying the “i” on that basis.
More than this, I knew which “i” it was. It was an iota with an
acute accent, and it could belong only to a handful of words where
an iota is stressed this way. One likely candidate, in a mathematical
context, would be the word ísos. Archimedes could have been
speaking about this being equal to that, right? And I could indeed see
a sigma, now that I was looking for it.
200
9780306815805_08.qxd 8/22/07 9:27 AM Page 201
INFINITY UNVEILED
What kind of “equal” then? Looking ahead, I thought I saw

another one of those general words of proportion theory. However,
this time it wasn’t “magnitude” but “multitude.” The couple of
words fell into place: “equal in multitude.” Isos plethei.This was good
mathematical Greek. This was “equal in multitude” to that. I looked
further and further. The text was peppered with “equal in mul-
titude.” It appeared that this was what Archimedes was doing
throughout this passage. He was showing how the result pointed out
by Ken applied to the case at hand. He was showing how this was
equal in multitude to that. It was all about such-and-such magnitudes
being equal in multitude to other magnitudes.
How I wished Ken was with me then! It was just too good to be
true. The expression “equal in multitude” is used in Greek mathe-
matics when discussing the numbers of objects in two separate sets.
Suppose I have a set of three triangles and a separate set of three lines.
A Greek mathematician would say that the two sets are “equal in
multitude,” meaning that they are each made up of three objects.
This is what Archimedes was doing. He was saying that, with the
infinitely many slices produced in the cube, after all the random slices
were made and the cube was entirely cut, the triangles, produced in
the cube by all the possible random cuts, were “equal in multitude”
to the lines in the rectangle.You see? In each random slice there was
a triangle in the cube, standing on top of a line in the rectangle. And
Archimedes pointed out that the number of triangles of which the
prism was made was the same as the number of lines of which the
rectangle was made. Surely he meant this to be verified by the fact
that there was a one-to-one relationship. Each triangle stood on an
individually separate line and each line was at the bottom of an indi-
vidually separate triangle.
Archimedes repeated this type of statement three times: he went
through the various configurations produced by the slices, showing
which set was equal in multitude to which set.And once those equal-
ities of multitude were secured, the result pointed out by Ken
201
9780306815805_08.qxd 8/22/07 9:27 AM Page 202
applied. This was typical of Archimedes. He did not actually refer,

explicitly, to his result; he did not even quote it. He set up the con-
ditions under which the result could apply; and this he did by
showing, in detail, which equalities of number applied in the config-
uration at hand.
Only, of course, those equalities of number were like nothing else
we ever knew from Greek mathematics. The objects Archimedes
counted—the sets of triangles and lines—were all infinite. Archimedes
was explicitly calculating with infinitely great numbers.
More than this, Archimedes was making his calculations based on
a sound principle. He apparently was stating that this infinite set was
equal to that infinite set, because there was a one-to-one relationship
between the two sets. He did not say so in so many words, but
Archimedes was never an explicit author. He always left much of the
work for the reader.
Note the following fact. Archimedes could have assumed, in prin-
ciple, that just because the two sets were infinite, they were also equal.
This would have been a very natural assumption to make. But the
very fact that Archimedes found it necessary to state that particular
sets of infinitely many objects are equal, shows that he avoided this
naive assumption. Instead, he must have assumed that infinite sets
could be said to be equal only when a special argument could be
made for their equality. And this leaves only one possible argument
for this equality—the argument of one-to-one correspondence.
It so happens that the tool of one-to-one correspondence is the
one by which the concept of infinity was finally structured in the late
nineteenth century. It is no less than the cornerstone for modern Set
Theory. And so we can sum up the lessons learned from pages
105–110 of the Method.
First, we find that Archimedes did not merely make an “implicit”
move from a random slice to the object made up of those random
slices. He relied instead on certain principles of summation. This
means that he was already making a step toward the modern calculus
and was not merely anticipating it in some naive way.
202
9780306815805_08.qxd 8/22/07 9:27 AM Page 203
INFINITY UNVEILED
Second, we find that Archimedes calculated with actual infinities

in direct opposition to everything historians of mathematics have
always believed about their discipline. Actual infinities were known
already to the ancient Greeks.
Third, we see that with this concept of infinity—as with so many
others—the genius of Archimedes pointed the way toward the
achievements of modern science itself. Back in the third century BC,
at Syracuse, Archimedes foresaw a glimpse of Set Theory, the product
of the mature mathematics of the late nineteenth century.
Mathematics, Physics, Infinity—and Beyond

It appears that there is some kind of complementary structure in the
Method. Thirteen propositions apply both physics and the implicit
summation of infinitely many objects. In proposition 14 physics is no
longer applied. The summation of infinitely many objects is not
implicit, but explicit, based on a rule of infinite summations. It
therefore appears that, for Archimedes, the application of physics
could act as some sort of short cut. When this short cut was not
available, one needed instead an explicit, mathematical rule for
infinity. It is as if Archimedes thought that, in the physical world, the
summation of infinitely many objects was not such a problem. After
all, physical objects are made up of infinitely many parts. But when
one moved to abstract, mathematical objects, there was a need for a
special, mathematical principle for doing such a summation.
Archimedes, after all, did not produce the science of the physical
world that Galileo and Newton would later produce. This is even
though he had assembled—as we have just seen—the tool kit for the
making of such science. I think I understand why. The reason was
that, for Archimedes, the combination of physics and mathematics
was important not for the sake of physics but for the sake of mathe-
matics itself. Archimedes’ great desire was not to find out about the
motions of planets but to measure curvilinear objects. It so happens
that, in our universe, mathematics, physics, and infinity are so closely
203
9780306815805_08.qxd 8/22/07 9:27 AM Page 204
tied together that, looking to advance pure mathematics, Archimedes

also laid the foundation for modern science.
Whichever interpretation we choose, it is clear that our under-
standing of the historical relationship between mathematics, physics,
and infinity will now have to be drastically revised in light of propo-
sition 14. But more than this. We will need to revise our under-
standing of the Greek treatment of infinity. I like to sum this up with
the phrase: “it’s not that they couldn’t.” They could very well
envisage actual infinity; they could even operate with it. For various
reasons, in most contexts, they preferred to avoid it. But this
avoidance was a conscious decision, not some kind of reflection of a
shortcoming on the Greeks’ part. They were ahead of the infinity
game. And so the same goes, in my view, of science. I think
Archimedes was capable of producing the kind of science of physics
that Galileo and Newton produced. He made the decision not to—
other things occupied his mind.
So much for the broad picture of the history of science and math-
ematics. Another thing became clear now—of great interest to
everyone involved with the Archimedes Project. The work mattered.
There were important passages to still be read.Already, with 105–110,
my reading managed to move substantially beyond Heiberg’s. Other
pages appeared to be even more difficult, but now it became even
more crucial to make them visible. The pressure was now on the
imagers to produce a completely new kind of product—an image to
make the invisible come to light. Would it be possible?
204
9780306815805_09.qxd 8/22/07 9:28 AM Page 205
9
The Digital Palimpsest

bigail had disbound the Palimpsest, and Reviel and Ken Saito
A had rewarded her work with an insight that blew apart the
boundaries of Greek mathematical thought. But it was clear from the
start that Abigail’s work was merely a step on the way toward a more
radical transformation of Mr. B’s book. As she was taking the prayer
book apart, I was asking the scientists to put all the palimpsested
codices that it contained back together again just as they had been
before the year 1229.
I did not want the scientists to reproduce the Palimpsest; I wanted
them to replace it. I wanted them to make something that was so
much better than the codex that scholars would no longer need to
make the pilgrimage to Baltimore. I asked them to make the invisible
visible, to make it available on desktop computers around the world,
and to make it appear in its correct order. Archimedes first, of course,
but then the palimpsested texts from the other codices. This was a
utopian fantasy. After all, we didn’t even know how many other
codices there were, let alone what was in them! Yet the result in 2005
exceeded everyone’s expectations. Today, scholars can read texts that
they literally had not dreamed of reading in 1998.They had been
unable to read the texts directly from the manuscript, but now they
can read them on a computer. This success was hard won, and it was
a long time coming.
It was clear from the beginning that both imaging teams—the one
from Johns Hopkins headed by Bill Christens-Barry and the one
from the Rochester Institute of Technology headed by Roger Easton
and Keith Knox—would be putting most of their efforts and most of
205
9780306815805_09.qxd 8/22/07 9:28 AM Page 206
their faith into a technique called “multi-spectral imaging.” I needed

to understand what “multi-spectral imaging” involved, and my guide
was the only teacher among them, Roger Easton, Professor of
Imaging Science at RIT. I thought of images as shapes produced by
artists. Roger thought of images as numbers produced by light. Not
surprisingly, it took him a little time to explain his view of things to
me.
Light
Light, Roger told me, be it from the sun or from light bulbs, comes
in waves of electromagnetism, which themselves consist of tiny
energy bundles called photons. The photons can be characterized by
the distance between their peaks—their wavelength. Some photons
come in long wavelengths, such as radio waves, microwaves, and
infrared waves. Some come in much shorter wavelengths, such as
ultraviolet waves, X-rays, and gamma rays.Visible light forms a very
small part of the entire electromagnetic spectrum, between infrared
and ultraviolet. The shorter the wavelength of a photon, the greater
its energy. But all photons travel at exactly the same speed in a
vacuum: the well-known speed of light—186,282 miles per second.
Photons interact with matter, which is made up of atoms. More
specifically, they interact with electrons that take their places at
various distances from an atom’s nucleus. Not all photons interact
with all electrons. Crucially, the interaction depends upon their
respective energies; they have to resonate with each other. If they do,
a photon will change the energy state of an electron and in response
the electron will itself emit a photon. The photon emitted by any
given electron will have a precise wavelength and a precise energy; its
wavelength will depend upon the energy an electron needs to shed,
which in turn depends upon an electron’s place in the composition
of an atom.
The human eye uses photons to make all the colors of the rainbow.
This is how it does it. With its lens, the eye focuses photons emitted
206
9780306815805_09.qxd 8/22/07 9:28 AM Page 207
T H E D I G I TA L PA L I M P S E S T
by the electrons on to the photosensitive cells of the retina. The

photons induce chemical changes in these receptor cells.The changes
in the cells depend upon the wavelengths of the received photons.
When your cells receive photons with a wavelength of about 400
nanometers, they will change in such a way that they generate an
electric current that will travel via the optic nerve to the visual center
of the brain, which will interpret the current as a color: blue. When
your cells receive photons with a wavelength of about 700
nanometers, the same process will occur, but the chemical change
will be slightly different as will the resulting current and the color. In
this case, you will see red.We read by recognizing patterns of intensity
and color generated in the visual center of our brain as letters. The
problem is that many of the letters in the undertext of the Palimpsest
cannot be read, even in bright sunlight.
“What could be a better light source than the Sun?” I asked
Roger, “and what could be a better receptor than the eye?” The
trouble with the Sun as a light source is that it gives off photons at all
sorts of different wavelengths. The image that your eye sees under
sunlight is the sum of images created at all of the visible wavelengths.
If you can create a source that emits light over a relatively narrow
band of the spectrum, then the resulting image will just carry the
information from that one wavelength, which will not be over-
whelmed by light at the others.
Consider, for example, images created using ultraviolet lamps.
Although photons from these lamps have wavelengths shorter than
those that the eye can detect, they have a remarkable effect on the
parchment that they hit. They energize the atoms and molecules in
the parchment, which absorb some of the energy and re-emit the rest
as photons with a wavelength that happens to be in the blue section
of the spectrum visible to humans. While the parchment re-emits
visible photons, the ink on the parchment obscures them. As a result,
the ink is effectively “backlit” by the soft blue light from this “fluo-
rescence,” the contrast of the faint undertext increases and thus the
text becomes more readable. Ultraviolet fluorescent light has long
207
9780306815805_09.qxd 8/22/07 9:28 AM Page 208
been used by scholars reading palimpsests and with great success.

Reviel and Ken used it to read Method proposition 14. But you
cannot use an ultraviolet lamp effectively except in a dark room;
photons at other frequencies completely obscure what they do.
The eye itself is such an amazing piece of machinery that it is hard
to imagine how a manmade version can be better. But the eye has
plenty of limitations that we normally do not notice because it has
evolved to suit our everyday needs. Its limitations become more
apparent when you try to do something extraordinary, though.
Looking at planets is difficult, explained Roger, because the size of
the image on the retina is so small that it covers only a few of the
eye’s sensors. Since each sensor “sees” a large part of the planet, the
eye cannot see (“resolve”) the fine detail. Here, telescopes come in
handy. Or try another problem: I find it difficult to see my cat Gracie
after dark because the cells of the human retina do not respond to the
waves emitted by warm-blooded animals; these infrared waves have
much longer length than the light we can see. The human eye is
responsive to only a tiny part of the electromagnetic spectrum. But
modern cameras can detect infrared waves and can find warm-
blooded animals in the dark. This is the basic reason why we use
cameras rather than our eyes to read the Palimpsest. Unlike the
human eye, modern cameras are sensitive to light outside the visible
spectrum, and thus can “see” information to which our eye is
“blind.”
In short, you can get very different results using narrowband illu-
mination captured by a camera from those you get when you look at
an object under the Sun. The successes of imagers around the world
in revealing hidden text by using cameras under different narrowband
lighting conditions are remarkable. For example, a team at Brigham
Young University has attained extraordinary results by imaging the
carbonized rolls of a library that was buried in Herculaneum under
the volcanic ash of Vesuvius in the early afternoon of Tuesday, August
24, 79 AD. When viewed in normal light, you cannot see any text
written on many of these rolls at all. But when imaged at a specific
208
9780306815805_09.qxd 8/22/07 9:28 AM Page 209
wavelength, the text “pops out” in the most remarkable way. We did
not think that imaging Archimedes would produce such clear-cut
results, primarily because the Palimpsest is, physically and chemically
speaking, a much more complicated object. The text in the
Herculaneum rolls had not been scraped off, it had not been over-
written, and its support hadn’t suffered the mold damage that the
Palimpsest had. The rolls had been subject to just one catastrophic
incident that had changed the chemical composition of the rolls
themselves and the text on them.We were right, too; there is no one
wavelength at which the Archimedes text pops out. But this is where
multi-spectral imaging comes into its own.
Numbers
Roger told me that multi-spectral imaging was a relatively new tech-
nique that had become widely available only since the arrival of
computers and digital-imaging technology. Computers turn all the
information they receive into numerical values—digits. Actually just
two “binary digits” (“bits”) are used—0 and 1—but they are com-
bined in a great variety of ways. For example, your laptop computer
converts your taps on the keyboard into different combinations of 0
and 1, which it can store and use as instructions to make certain pat-
terns on your screen.When you digitally record music on your com-
puter, the loudness of sound at each time interval is again interpreted
as a number.When you take a picture with a digital camera, the light
that hits the camera sensor is turned into numerical values. Each
“piece” of the image, each so-called “picture element” or “pixel,” is
given a number made out of 1s and 0s. Many images are “8-bit”
images, and the numbers attached to these pixels are made out of
eight-figure combinations of 1s and 0s. So, for example, the number
10101010 actually has a value of 170. The number 11111111 has a
value of 255. This is the highest value that an 8-bit number can have
because, including 00000000, there are only 256 ways in which 0s
and 1s can be combined in an eight-figure series. To extract this
209
9780306815805_09.qxd 8/22/07 9:28 AM Page 210
numerical information you need a software package—a series of

instructions to your processor that sorts the numbers and presents the
information in useful ways. And you had better have the appropriate
software package. Beethoven’s Ninth would not make a pretty
picture, and the Archimedes Palimpsest is unlikely to sound any
better than it looks.
One of the great advantages of digital technology is that it is pos-
sible to combine the numbers from images in different ways.You can
instruct the computer to adjust the numerical values in the image—
to suppress numbers that are too high or too low and to amplify small
differences if you decide that they are important. This is how com-
puters get rid of the red-eye caused by the flash out of your family
photos. But another advantage of digital technology is that you can
overlay one set of numbers with another.You can, for example, add a
backbeat behind the voice of a rock star. More importantly for our
purposes, you can combine an image taken at one wavelength of light
with an image taken at another one to make some feature in that
scene more visible. If you take images at many different wavelengths
and stack these different images in order of wavelength, one upon
another in a computer, you produce a “data-cube” of digital infor-
mation in which each is seen in different wavelengths of light. Do not
imagine this data-cube as a hologram; imagine it as a sea of numbers
containing patterns—or curves—that reflect the characteristics of the
area imaged. By writing computer algorithms (recipes for retrieving
data in a certain way), scientists can carve up the data-cube to manip-
ulate the values of the numbers, accentuate certain curves, and extract
the information they want. Much more information can be extracted
from a digital data-cube created using narrow bands of light than can
be retrieved from the Palimpsest under any single lighting condition.
The most basic procedure to extract information from a data-
cube, Roger explained, was “principal components analysis.” You ask
the computer to make a set of pictures from weighted combinations
of the numerical values of images taken at each wavelength. The
images in this new set are based on the amount of difference between
210
9780306815805_09.qxd 8/22/07 9:28 AM Page 211
numerical values of pixels that are close to each other. As a result they
do not show patterns of color, but patterns of contrast.The first image
in the new set highlights those areas where contrast between different
features is greatest; the second image shows the next greatest contrast,
the third image the next, and so on. By this process, you start out with
a set of images of the same area in different wavelengths of light and
end up with a set of images that combine the wavelengths of light to
show the different objects of the image. Obviously, in the Palimpsest
the first principal component shows the image feature with the most
contrast, which is the prayer-book text with its nice dark ink outlined
against the light-brown parchment around it. But the second prin-
cipal component is indeed, in large part, the Archimedes undertext.
Yet another principal component image might show the mold. Once
you have separated out the components, you can make them as bright
or as dark as you like by manipulating the numbers.
Modern science has turned light into numbers, and modern sci-
entists can change the numbers. But the skill is in how you change
the numbers and this is as much an art as it is a science.
Digital Cooking
The two teams of imagers started their competition in June 2000, and
they worked with five leaves that were already detached from the
binding of the codex when it arrived at the Walters.
Bill Christens-Barry took his images with a Kodak digital camera.
This is a standard type of camera used by professional journalists the
world over. It couldn’t make a very big data-cube, but it could create
images with a high spatial resolution. Bill and his colleague Joanna
Bernstein imaged at 600 dots per inch. Bill called his best shot at
manipulating his data his “cookie-cutter” technique. He chose a set
of images from the ultraviolet range of the spectrum in which he
could see both the prayer-book text and the Archimedes text rea-
sonably well.Then he separated out the principal components of the
images he took in normal light and selected one that just showed the
211
9780306815805_09.qxd 8/22/07 9:28 AM Page 212
prayer-book text. He then played with these two pictures in the

computer; he subtracted the image of the prayer-book text from the
ultraviolet images that showed both texts well and was left with just
the Archimedes text.
Keith and Roger’s camera made Bill’s look like Fred Flintstone’s.
To select their wavelengths of light they didn’t use glass filters placed
in front of the lens; they used the latest technology, a “liquid-crystal
tunable filter” (LCTF), by which they could select the wavelength of
the incoming photons by turning an electronic knob. The camera
even had a tiny electrical refrigerator to keep its sensor cool.With this
camera Keith and Roger built data-cubes at thirty-five different
wavelengths across the spectrum—a much bigger data-cube than
Bill’s camera could make.The only disadvantage was that this camera
could only image at 200 dots per inch. It had much greater spectral
resolution, but less spatial resolution than Bill’s camera.
Keith and Roger processed very differently from Bill. They
examined each folio that they were looking at and determined pixels
that belonged to three different classes of object: pixels that were def-
initely parchment, those that were definitely prayer book, and yet
another class—the important one—pixels that were definitely
Archimedes. They then located the corresponding pixels in the
images that they took—all thirty-five of them—and made a com-
puter evaluate the vital statistics for each of the pixels.Then, the com-
puter calculated the degree of likelihood that any given pixel was a
prayer-book pixel, an Archimedes pixel, or a parchment pixel. If the
computer was sure that a pixel was Archimedes, it would be very
bright; if it was less sure, it would be dimmer. The computer then
combined the results from the different wavelengths. This technique
is called “matched spectral filtering.”
It looked as if Bill had come to a gunfight with a knife, or at least
with a point-and-shoot camera. But, actually, I looked at the images
from both teams and thought that they were marvelous: Bill’s images
were just as good, in my eyes, as Roger and Keith’s. In the image
212
9780306815805_09.qxd 8/22/07 9:28 AM Page 213
showing the Archimedes text, I could see diagrams where previously

I had seen nothing. I could see Archimedes text appear from
nowhere. And I couldn’t see the prayer-book text. It had all but dis-
appeared into the parchment background. I thought that we had
cracked it already. I held out the hope that with these pictures we
could recreate the Archimedes manuscript as it had been even before
it was palimpsested.This would be the Jurassic Park of medieval man-
uscript studies, and the resurrection of Archimedes. If you looked at
one of these pictures, you would understand my excitement. I
thought that both teams of imagers had done the job that we had
asked and that the only real problem we had was how to choose
between them.
On Friday, October 20, 2000, a segment on the Palimpsest was
broadcast on ABC’s news program World News Tonight with the late
Peter Jennings. It detailed the remarkable efforts of imaging scientists
to uncover the erased texts of Archimedes in an ancient manuscript
in Baltimore. Suddenly the imagers were stars.Three days later, on the
Monday, they were to present their results to Natalie and Reviel. We
were all in for a shock.
Bad Recipes
Reviel could not make the review meeting; he had pneumonia. Natalie
Tchernetska, however, voiced complaints for both of them. In her
words, the photographs of both teams, but particularly Keith and
Roger’s, were “out of focus.” They had all sorts of unexplained white
spots on them.They were not of sufficient resolution. Getting rid of the
prayer-book text had not helped at all in reading the Archimedes text.
Plain old high-resolution photographs and photographs just taken in
ultraviolet light were much better, she said, than these processed images.
What had gone wrong? As it turns out, it is not easy for imaging sci-
entists and medieval paleographers to understand each other. So let us,
like the imagers, take each of Natalie’s complaints in turn.
213
9780306815805_09.qxd 8/22/07 9:28 AM Page 214
Her first complaint was that the images were out of focus. Actually
they were not out of focus. The problem was one that all multi-
spectral imagers face. To get images of different wavelengths of light
they had had to change the filters on their camera. Because the light
going through different filters refracted at slightly different angles, the
resulting images were of slightly different sizes. Since they had taken
images at many different wavelengths, and these images had not “reg-
istered” properly, the result was that the processed images looked
blurred. Now, this doesn’t matter much when you are imaging large
tracts of ground from space, trying to find a coca field in the Amazon
rainforest, which is normally how this technique is used. But it does
matter—very much—when you are trying to read the niceties of tiny
Greek script from the tenth century. Clearly, Roger, Keith, and Bill
were going to have to use fewer wavelengths or find another way to
get around this “registration” problem.
Her second complaint was that the images had lots of white spots
on them, which looked as if they were supposed to be Archimedes
text but were not. Imaging scientists call these spots “artifacts.” The
imagers had in fact found imaging Mr. B’s book to be extremely dif-
ficult. As a result, they had had to write very complicated algorithms
to extract the Archimedes text. Now, every time you manipulate an
image you are playing with data.You might be bringing out the text
that you want, but you are also, inevitably, adding noise just by stirring
the ingredients. Again, in most applications of multi-spectral imagery
this doesn’t matter, at least not very much. But in trying to read the
Archimedes text, it does matter, very much. The scientists had to
come up with simpler algorithms.
Roger and Keith took images at 200 dots per inch—about 8 pixels
per millimeter.This was a perfectly sensible thing to do. It is, more or
less, the resolution of the rods and cones of the eye if the page is
viewed at normal viewing distances. It allowed a complete single
folio of the Palimpsest to be imaged in two sections with the available
digital camera. They did not make enlarged images of the folios,
214
9780306815805_09.qxd 8/22/07 9:28 AM Page 215
which would have required a much higher resolution.We simply did

not know that Reviel and Natalie wanted to read magnified images
to see all of the critical features in the text. If possible, they would
have wanted a single Archimedes character to fill up the entire com-
puter screen and still not appear pixilated; Natalie would have loved
to see the image as though through a microscope. In reading
palimpsest texts, size matters after all.This was another lesson that the
scientists had to learn and another way in which the camera could
potentially improve upon the eye.
But the most revealing and unexpected complaint that Natalie had
was that the imagers had taken away the prayer-book text. They
wanted it back again. What on earth had we been doing and why?
The scientists had actually succeeded in separating the Archimedes
text from the prayer-book text and then in eliminating the prayer-
book text. Now the scholars were saying that this didn’t help. The
reason that it didn’t help is actually quite straightforward. The scien-
tists had made the prayer-book text disappear by making it exactly
the same color as the parchment.The trouble was that now, when the
Archimedes characters disappeared beneath a bit of prayer-book text,
the scholars didn’t know why. It was no longer clear to them whether
the Archimedes letters were invisible because they in fact did not
exist or because they were actually hiding underneath the letters of
the prayer book.The scientists had created images with characteristics
that the scholars simply did not value, however impressive I thought
them to be.
The whole day was a litany of complaint. I was as confident in the
afternoon that the results were useless as I had been in the morning
that they were a triumph. Mike Toth, Abigail, and I met in a closed
session at the end of the day. And then, to my amazement, Mike
insisted that nothing at all had gone wrong. In fact, he explained, this
was how experimental imaging projects worked.
If you ask scientists to come up with a solution to a difficult
problem, you will make errors in defining that problem and they,
215
9780306815805_09.qxd 8/22/07 9:28 AM Page 216
more likely than not, will fail to get the best solution the first time.
Really difficult problems, said Mike, get resolved in incremental
steps. These steps begin with criticism and end with under-
standing. Mike said that it was quite normal in such imaging
projects for scientists to produce a misconceived product. We were
just at the beginning of a long process by which the imagers come
to understand fully what the scholars need and through which they
could refine their techniques. Furthermore, Mike insisted, the
imagers had done well. They had succeeded in separating the
Archimedes text from the rest of the manuscript and there were
signs that they were pulling out Archimedes text that could not be
seen at all under normal light conditions. Actually, he went on,
instead of firing the imagers, we should make them join forces and
hire them all. In other words, Mike thought that Mr. B should pay
for all three of them to work on the project; we could combine
Bill Christen-Barry’s experimental approach with the processing
skills of Keith and Roger.
I didn’t actually think that Mike was nuts, because I knew that he
had a vast amount of experience in judging the results of technical
projects. But I could not see the way forward, and I dreaded to think
what the reaction of the “source selection authority” would be when
I emailed him. His reply was, typically, far briefer than my wordy
missive. His verdict: “OK.”
The First Words

Throughout the taxing period up to March 2001, Reviel and
Natalie had been trying to transcribe the Archimedes texts from the
scientists’ images. I was copied on emails that reveal like nothing
else the difficulty of their task. Here is a typical one from Reviel to
Natalie:
Natalie, I’m making progress!
216
9780306815805_09.qxd 8/22/07 9:28 AM Page 217
Take a look at 48v col. 1 line 6, after the easily readable word
perile/psomen. Heiberg is surely wrong to get the rho immedi-
ately after without a gap—there is surely a one character gap;
furthermore his undotted eta is a very bad eta.This scribe tends
to have a small foot of the eta flexed inside a little bit, like a
knee reacting in a knee-jerk, but this foot is very smooth, a
continuous parabolic curve; in fact, this is more like the scribe’s
kappa than like his eta. Now, the character just preceding the
rho is faint, but does suggest an alpha. Heiberg’s concluding to/s
seems likely, and so we may have ark[2-3 characters]to/s. How
about arkounto/s? Then the immediate couple of words is
perile/psomen arkounto/s—“we shall include,” “sufficiently.”
The entire passage could be made to read, e.g., kai allo/n
pleiono/n (homoio/n touotois) theo/roumeno/n ta (pleista) ou
perile/psomen, arkounto/s gar ho tropos hupodedeiktai dia
to/n proeire/meno/n. I bracket words that are truly speculative,
though there is some trace of a lambda for pleista, and the
famous “moi” at the beginning of line 5.
This transcription was one that Reviel made from one of the trial
images. The transcription might have been helpful to Natalie, but it
was of absolutely no use to the imagers. Reviel found his own way
to show what text he could decipher and what still needed work. He
drew pictures.
Working mainly from the ultraviolet images, Reviel would write
in green what he could read and in red what he could only guess.
There are alarming amounts of red in these pictures. Sometimes he
would send images with questions on them. One particularly
important passage seemed to be on folio 105. Reviel writes in what
he sees but frankly admits to total guesses on the folio. It seems like
an extraordinary struggle. And indeed it was. It was worth it, of
course, because eventually we would discover that Archimedes knew
about “actual” infinity. But we couldn’t go on like this indefinitely.
217
9780306815805_09.qxd 8/22/07 9:28 AM Page 218
Making Light Work

Roger, Bill, and Keith had a lot to prove. But they not only learned
from the criticism of their efforts; they came up with a new concept
for the imaging of the Palimpsest based on their early results. They
would address the resolution problem and image not at 200 dpi but
at 600. To address the registration problem, they were not going to
filter the light at all. Instead, the images would be collected under
three different lighting conditions, with low-wattage tungsten lights
(which give off a very “reddish” light), with Xenon strobe lights
(which give off short flashes of bright, white light), and with “long-
wave” ultraviolet lamps that emit most of their light at 365
nanometers, which is just barely shorter than the short-wavelength
limit of the human eye.They would also take the images with a pro-
fessional color digital camera, the type now used by every profes-
sional photojournalist in the country. There is no point in using the
latest technology if the latest technology doesn’t help you. Although
Bill’s Kodak camera didn’t have the spectral precision of Roger and
Keith’s, it could get the spatial resolution that the scholars wanted; it
meant that the registration problems would be less severe; and Bill
had demonstrated in his experiments that more processing was not
needed to separate the prayer-book text from the Archimedes text.
So, the knife won out over the gun after all. The Proof of Concept
imaging took place in early 2001. I had to wait several months for the
imagers to come up with a processed product.
Success was achieved by trial and through error, but this time the
imagers had good data and a much better idea of what the scholars
wanted. The imagers were, of course, playing with numbers. But in
what follows I explain their solution visually, in terms of colors. It is,
quite literally, easier to visualize this way.
Up until this point, the scholars had found the UV images to be the
most useful. The scientists therefore looked at what it was that the
scholars didn’t like about the UV images. They had two serious defi-
218
9780306815805_09.qxd 8/22/07 9:28 AM Page 219
ciencies. Firstly, they were rather “soft”; they seemed to lack defi-
nition. Secondly, they were essentially monochromatic. They were
shades of blue: the parchment was bright blue and the ink was dark
blue.Although the texts stood out better than they did in natural light,
it was harder than ever to distinguish between the prayer-book text
and the Archimedes text. Keith took the UV images as his starting
point. He wanted to make it clear which text was Archimedes and
which text was from the prayer book with a minimum of image pro-
cessing. He also had to restore the sharpness that the UV image lacked.
In the Proof of Concept imaging, the imagers noticed that there
was a big difference in the appearance of the manuscript when it was
imaged using white strobe lights from when normal tungsten lights
were used. Low-wattage tungsten light is, as I mentioned, very red
compared to strobe light and in tungsten light the Archimedes text
was much fainter. The image consisted of red, green, and blue
“channels,” and they saw that in the red channel the Archimedes text
almost disappeared completely.To me this was a bad thing, but not to
the imagers. They had two simple, unprocessed images of a page and
these images were completely different. By combining them they
could come up with a different, synthetic image.
So Keith made a new picture altogether. He started with a blank
“digital canvas.” On to this canvas he could insert his images. He had
three digital channels in which to do this—red, green, and blue. In
the red channel, he put the tungsten-red image. In the blue channel,
he put the ultraviolet-blue image. And in the green channel, he
simply put the ultraviolet-blue image again. The important point for
Keith was not that the Archimedes text disappeared in the tungsten-
red image; the important point was that both the parchment and the
Archimedes text were red. So in the red channel of his picture, he had
bright Archimedes, bright parchment, and dark prayers. In the blue
and green channels, he had dark Archimedes, bright parchment, and
dark prayers. By combining these elements into one picture, he got
bright parchment, dark prayers, and dark Archimedes with a red tint.
219
9780306815805_09.qxd 8/22/07 9:28 AM Page 220
This was a very elegant solution. It involved far less processing than
the images produced in the initial trials. It clearly differentiated the
prayers from the Archimedes text by color, and it gave the
Archimedes text a greater clarity than the UV image. The images
were really just what Reviel wanted. They had a resolution of more
than 600 dpi; they made a clear color difference between the
parchment, the Archimedes text, and the prayer-book text; they had
few artifacts; and they were not blurry.The process had another great
advantage: it worked well over relatively large areas of the
palimpsested texts and little local processing was necessary. In fact, the
processing could be automated.An entire day’s worth of images could
be processed overnight on Keith’s laptop computer as it ran on the
desk in his hotel room. We called them “pseudocolor images,” the
method that produced them “pushbutton processing,” and Keith’s
package of software code for making them “Archie 1.1.” By
September 2001, we had the key to unlock the secrets of the
Palimpsest.
A New Box for the Brain

Writing books in the Middle Ages was a laborious business. A scribe
named Raoul working in the Monastery of St. Aignan in France,
wrote: “You do not know what it is to write. It is excessive drudgery;
it crooks your back, dims your sight, twists your stomach and sides.
Pray, then, my brother, you who read this book, pray for poor Raoul,
God’s servant, who has copied it entirely with his own hand in the
cloister of St. Aignan.”
Roger, Keith, and Bill became twenty-first-century Raouls, and we
should spare a thought for them. They created text just as surely as
Raoul did, and although their procedures were very different, they had
exactly the same feelings about the process. From 2001 onward, they
would visit the Walters every six months or so and, for ten days at a
stretch, image the fifteen folios of Mr. B’s book that Abigail and her team
220
9780306815805_09.qxd 8/22/07 9:28 AM Page 221
had most recently liberated. And while Abigail liberated the leaves, I
placed the scientists in their very own cell—a bare white-painted
cinder-block room with no windows that cannot have been much
bigger than the average medieval monk’s living quarters. And I had to
lock them in it. Frequently they were working after hours in a museum
containing thousands of priceless treasures, and they would have to call
me to let them out if they so much as wanted to use the restroom.
Each time they came, the imagers filled their cell chock-a-block
with equipment, which Roger drove down from Rochester. Roger
had made a special gantry for the imaging of the codex: the cameras
were mounted above a motorized X-Y stage, upon which each
bifolio was placed and imaged. The imagers to this day have not
touched the Palimpsest.A conservator wheeled in each folio from the
conservation studio about fifty feet away. Each folio was mounted in
its own bespoke mat. A conservator would place it carefully on the
X-Y stage. Once it was on the stage, a computer moved everything.
To turn the leaf, the imagers had to make a phone call to the con-
servation studio and someone would turn the leaf for them.
Roger was in the driver’s seat, literally. He drove the X-Y stage,
and he took the pictures with a click of the mouse. Each side of each
leaf was photographed thirty times.To get a resolution of 600 dpi, ten
separate pictures of each folio had to be taken and in three different
lighting conditions. Keith was “the lights.” He flicked the switch that
turned on and off the strobe lights, the tungsten lights, and the ultra-
violet lights. Bill recorded every move on spreadsheets. We now have
over 15,000 records. For each image we recorded the following: the
folio; the side of the folio; the position on that side of that folio; the
date that the image was created; the camera make; the camera serial
number; the lens brand; the lens serial number; the lens size; the
wavelength of the illumination and whether it is fluorescent or
reflective; the make, serial number, and wattage of the illumination
source; the size of the aperture on the camera; the shutter speed; the
resolution; the pixel X count; the pixel Y count; the camera incident
221
9780306815805_09.qxd 8/22/07 9:28 AM Page 222
angle; and the distance of the camera from the folio. There are more
columns than this, and some of them, even today, I do not under-
stand. But the scientists needed to document everything thoroughly,
not just for their own records, but for posterity. There is always the
possibility that someone might use this data to make better images
with more effective processing algorithms in the future.
If you think this is boring, you are not alone. Ten days sitting in a
cell-like room, taking pictures in the bright light, and then in total
darkness. Bill called it “trained-monkey work.” It was unbelievably
boring. It was also frustrating.Things broke and they had to be fixed;
there were long pauses when leaves needed to be delivered; and the
worst thing was the “whirr” of the X-Y stage as it moved from one
section to the next.They couldn’t oil the cogs. And for Keith it didn’t
stop. Each night he would take the data collected back to his hotel
room and make his marvelous creations—new images using the
ultraviolet and low-wattage-tungsten images as ingredients.This is an
important conceptual point. These are not images of the Palimpsest;
they are synthetic creations made from images of the Palimpsest.They
are works of art. Well, they work, anyway. And that’s the point.
Still, at this stage, we had no product. All the images had to be
assembled so that scholars could access them. First, the ten individual
shots of each folio had to be “stitched” together and this had to be
done for the strobe, ultraviolet, and pseudocolor images. Roger
Easton and his graduate students at RIT performed 5,520 stitching
operations. They then had to devise a way that the scholars could
access the images easily. The browser that Roger and his students
designed is the mechanism by which the scholars access the texts in
the Palimpsest. It is infinitely more flexible than the Palimpsest itself.
If they want they can read the prayer book with the leaves falling in
the right order. At the click of a mouse, however, the images magi-
cally reorder themselves so that they appear in their Archimedes order
just as they were before they were palimpsested.The scholars can also
choose whether they want to see the pages in normal light, ultraviolet
light, or in synthetic pseudocolor.They can see the Archimedes pages
222
9780306815805_09.qxd 8/22/07 9:28 AM Page 223
in more detail than the eye can see the original. they can zoom in on
a section of a page and “blow it up” without losing any resolution.
New Text
Of course, the imagers did not understand the text they were cre-
ating. It was up to Nigel Wilson and Reviel to read the Archimedes
text. Nigel and Reviel are now working on completely new editions
of Method, Stomachion, and Floating Bodies.Theirs is, in many ways, an
ideal collaboration. Nigel has greater familiarity with the tran-
scription and decipherment of tenth-century Greek cursive than
Reviel. On the other hand, Reviel understands Archimedes’ mathe-
matics so well that he can guess words that are no longer visible in
the codex. When you make a scholarly transcription of a text, you
need to note what it is that you can see and what it is that you can
guess. For a word to be really solid in our transcription, therefore,
both Nigel and Reviel have to see it. They work independently of
each other—Reviel in Stanford, California and Nigel in Oxford,
England. They confer when they have completed a passage and
compare notes. Here is a typical example, from folio 105v, which
contains Method, proposition 14. Nigel writes:
Dear Reviel,
In col.2 line 4 I think the reading PhANERON hWS
EIRHTAI does not fit the spaces as exactly as we should like,
and my suggestion, based on staring a long time at the image, is
that we read PhANEROI TO SKhHMA.This introduces a verb
which A. does not use much if at all elsewhere, but it is good
enough as Greek. Have another look and see what you think.
To which Reviel replies:
I definitely see now your Chi, which makes SKhHMA a very

attractive reading. Looking at it further, I wonder whether I
223
9780306815805_09.qxd 8/22/07 9:28 AM Page 224
do not see a nu after all at the end of PhANERON. How

about TOUTO GAR PhANERON TWi SKhHMATI, with
the scribe substituting, as he does so often, omicron for
omega, and then, following upon those two neuter accusatives,
he can find no fault in SKhHMA? I am not sure TOUTO
GAR PhANERON TWi SKhHMATI is very good Greek,
but it is less radically deviant than TOUTO GAR
PhANEROI TO SKhHMA (which, if correct, would be
rather exciting). Perhaps, if we ask for a super-high-resolution
X-ray of this particular point, this may serve to clarify the pos-
sible value of the technology.
Occasionally Reviel writes to me when he is excited. The first

folio of Archimedes’ letter to Eratosthenes was imaged on Thursday,
November 20, 2003, four days after Abigail had finished preparing it;
but it wasn’t until Wednesday, October 12, 2005, that Reviel tran-
scribed it. He left the most difficult pages to the end, because by then,
he could read faint traces of script with much more fluency:
I’m making some progress here.Was a funny sensation to tran-

scribe the introduction to the Method, rather like a Shakespeare
scholar transcribing the manuscript text for “To be or not to
be.” Surprisingly many subtle changes, of some significance (e.g.
Eudoxus was the first not to “discover” a result, but to “publish”
it, etc.). It will help morale to notify everyone that reading of the
forgery pages 57–64 is absolutely crucial. And I so want this
border paper to be removed by Abigail.Talk to you—Reviel.
I already own two printed results of the digital Archimedes. The

first is a beautiful book, printed by Nigel. Like Archimedes’ letters this
book is private for Nigel’s friends. It exists in only fifty copies. It is
his transcription, made from the pseudocolor images of On Floating
Bodies propositions 1 and 2, including the diagrams. It also includes a
transcription of folio 81v, which Heiberg overlooked and which is
224
9780306815805_09.qxd 8/22/07 9:28 AM Page 225
now on the back of a page containing a forgery, covered in glue.

There are no gaps in Nigel’s transcription that I can see. However,
some of it is in Latin, based on Moerbeke’s text, because Nigel
printed this book in 2004 and at this stage he didn’t have a pseudo-
color image of folio 88r.
The second is the monumental first volume of three by Reviel
entitled The Works of Archimedes and published by Cambridge
University Press. This is the first proper translation into English of
Archimedes’ Sphere and Cylinder and Eutocius’ commentary on the
same. It is peppered with diagrams.These, I have no doubt, more accu-
rately recreate the designs manufactured by Archimedes on the sands
of Syracuse than any yet made. That’s why Reviel’s first words to me
were: “Yes, I need to see the diagrams, especially of Sphere and
Cylinder.” Clearly, we gave him what he wanted, and we gave the world
a better understanding of the most important scientist who ever lived.
The digital Palimpsest is encased in a silver box—a 300-gigabite
external hard disk drive that you can plug into your computer.
Scholars no longer read Archimedes by looking at iron marks made
with a reed pen on animal skin. Archimedes’ treatises are now digitally
stored as 1s and 0s in a computer. Archimedes has received his latest
IT upgrade. Only Nigel Wilson does not use this computer disk. To
the amazement of the imagers, Nigel prefers to use hard-copy prints
of the pictures. Nigel is an end-user; he gets what he wants. Even now,
he does most of his transcription work from these prints, in the
summer months, when the light is good, using a magnifying glass.
A New Voice
A great codex that had already revealed most of its secrets. As you will
recall, that’s what most of the experts thought when the Palimpsest
was auctioned at Christie’s. Given the reputation of Heiberg as a
philologist of ancient texts, and given the treatment that the
Palimpsest had received since Heiberg’s time, this skepticism seemed
well founded. Even Reviel thought that his work would actually be
225
9780306815805_09.qxd 8/22/07 9:28 AM Page 226
mainly on the diagrams. The discovery of two unknown folios of

Floating Bodies and the new reading of Method proposition 14,
changed Reviel’s view of things. It didn’t take long for the rest of the
world to be convinced of the importance of the Palimpsest. On
Wednesday, December 6, 2000, I got a call from Will Peakin. He
wanted to write an article on Mr. B’s book for the color magazine of
the Sunday Times of London. The June 17, 2001 cover of the mag-
azine has an image of the Palimpsest and the words “EUREKA: It’s just
a few lines of scrawled Greek text, but new technology has identified
the hand of Archimedes—and the results are rewriting history.” But
we hadn’t seen anything yet. It was just the start.
Up until the summer of 2002, we had all been working for
Archimedes. But that was about to change. About thirty folios of the
Palimpsest do not contain Archimedes’ treatises. They come from
other palimpsest texts. It was Natalie Tchernetska’s task to look into
them. She started with one particular page on which Heiberg had
read just a little phrase that no one subsequently had been able to
identify. When she received pseudocolor images of this page, she
painstakingly transcribed just a few more lines.Then she tried to find
a match for these lines in Byzantine texts. One particularly rich
source for Byzantine texts is called the Suda. It is a massive tenth-
century encyclopedia of authors ancient and modern. Eventually
Natalie found a close match: it was a quotation of a lost speech by an
ancient author named Hyperides. A few days later, on Saturday,
October 19, 2002, Natalie sent me the following email.
Dear Will, In the course of further exploration of the non-

Archimedes folios, I recently deciphered the text of a Greek
orator, unknown otherwise. I could identify parts of lost
speeches by Hyperides: ff. 135–138 contain a fragment of his
lawsuit speech “Against Timandros;” ff. 136–137 a fragment of
a political speech, possibly “Against Diondas;” ff. 174–175
possibly a fragment of the same political speech. Kind regards,
Natalie
226
9780306815805_09.qxd 8/22/07 9:28 AM Page 227
Natalie had never heard of Hyperides; nor had I. He sounded like

a character out of Asterix, perhaps a close cousin of Ekonomikrisis,
the shrewd Phoenician merchant in Asterix the Gladiator. But no. It
dawned on us quickly that this was truly a sensational discovery.
Hyperides is, in fact, one of the ten canonical orators of antiquity. He
was born in 389 BC, five years before Aristotle. Like Aristotle,
Hyperides lived in Athens, and he was a politician in the world’s most
influential democracy.
In the ancient world, seventy-seven speeches were attributed to
Hyperides who was celebrated for his style and his wit. His most
famous speech is lost. It concerns Phryne, a prostitute famous for her
beauty. In fact, legend has it that her body was the model for
Praxiteles’ famous statue of the goddess Aphrodite at Cnidus. But
Phryne was also Hyperides’ mistress, and when she was accused of
offending the Eleusian mysteries, Hyperides defended her. He wasn’t
doing very well, so he ripped open her robe and exposed her breasts
to the jury. It worked, and she was acquitted. But despite the style and
subject matter of Hyperides, his speeches suffered from the transition
from roll to codex. In 1998, László Horváth, from Budapest, valiantly
searched for a codex that was mentioned as containing Hyperides’
speeches in the sixteenth century, but he never found it and its con-
tents were never known. Indeed, until the nineteenth century,
Hyperides was only known through quotations from later authors.
Then, in 1847, papyri containing his texts were discovered in a tomb
in Thebes, Egypt. The last big discovery was in 1891. But, in 2002,
Natalie had discovered a new Hyperides text, and what’s more she
had found it in a codex. If we can read all of these folios, we will have
added more than 20 percent to the surviving work of this great figure
from the golden age of Athenian history.
Hyperides was outspoken in his support of resistance to the mil-
itary might of Philip of Macedon and his son Alexander the Great.
When Alexander died, in 323 BC, he advocated full-scale rebellion
against the Macedonian overlord.The rebellion failed. Hyperides had
his tongue cut out to make a mockery of his oratory and then was
227
9780306815805_09.qxd 8/22/07 9:28 AM Page 228
executed. Plutarch, in his Lives of the Orators, writes of Hyperides:

“His monument is now altogether unknown and lost, being thrown
down with age and long standing.” Not so. Natalie found ten folios
of the Palimpsest that contain his speeches. But retrieving the literary
legacy of this great figure is appallingly difficult. The pages of the
Palimpsest containing his speeches are even more difficult to read
than the Archimedes pages. As I write, an international team of
scholars, including Natalie, Pat Easterling, Eric Handley, Jud
Hermann, László Horváth, and Chris Carey are working collabora-
tively to provide a critical edition of the texts.
One of the speeches that Natalie identified mentions some of the
great historical figures from antiquity—Demosthenes, the well-
known orator, as well as Philip of Macedon and his son, Alexander
the Great. It also mentions a figure far less well known—Diondas.
Natalie boldly suggested the circumstances in which this speech
might have been delivered. Philip of Macedon’s military might had
been growing, and Athens needed to react. Demosthenes was par-
ticularly hostile to Philip, calling him “the pestilent knave from
Macedonia,” and he successfully negotiated an alliance with the city
state of Thebes. Hyperides was delighted and supported a proposal
that Demosthenes receive an honorific crown for his diplomatic
triumph. But in 338 BC, despite their alliance, the Athenians and the
Thebans lost disastrously to Philip’s forces at the Battle of
Chaeronea. At this point, Diondas indicted Hyperides, because, he
argued, Hyperides’ support for Demosthenes was unconstitutional. It
seems to have been a blatant and cynical political move to damage
both Demosthenes and Hyperides who had spearheaded that anti-
Macedonian sentiment in Athens. We know that Hyperides wrote a
speech in his own defense and that he was acquitted. This, Natalie
deduced, was Hyperides’ lost speech. It not only sheds light on
Athenian politics in the grim days after Chaeronea, but it also pro-
vides new context for one of the greatest speeches of antiquity—
Demosthenes’ own speech, “On the Crown.” These pages will be
studied for years, but already great progress is being made. László
228
9780306815805_09.qxd 8/22/07 9:28 AM Page 229
Horváth is deciphering one of the most difficult of the Hyperides

leaves. He emailed me to tell me that on his page, when Hyperides
discusses previous alliances between Athens and other Greek city
states, he differs from the great historian Herodotus. The difference
is in number of ships that Athens contributed to the Greek fleet at
the great Battle of Salamis when the Greeks, led by Themistocles of
Athens, triumphed over the Persians who had threatened to overrun
them under Xerxes in 480 BC. Hyperides gives the total as 220 ships,
while Herodotus says that the Athenians provided 180. Since
Herodotus’ total for the number of ships provided by all the city
states does not tally with the figures that he gives for individual
cities, László thinks that Hyperides’ speech might be crucial in
assessing the details of one of the most important battles of Western
civilization.
There seems to be no end to the secrets of the Palimpsest. On

Monday, June 11, 2005, I received an email from Nigel Wilson, in
which he wrote that he had identified several further leaves of a
philosophical text and on one he “read the name Aristotle clearly
enough.” There are seven folios of this text in the Palimpsest, and this
text has yet to be transcribed and identified. I passed this information
on to Reviel, who transcribed more words. He could not match
them in any search engine for Greek texts. This sounded familiar.
Perhaps it is a yet unknown commentary on Aristotle. Since Nigel
thinks this manuscript was written in the late ninth century, this
would make it a commentary from the ancient world. It is not dif-
ficult to come up with suggestions. Perhaps the most convincing so
far is a suggestion by Marwan Rashed, a French scholar whom Reviel
contacted. He suggests that it might be a text by an early Christian
author criticizing various Greek philosophies, including the
Pythagoreans, for their failure to take into account the possibility of
creation out of nothing. As such, we might have uniquely preserved
in the Archimedes Palimpsest the views of an early Christian author
on the inadequacies of the pre-Christian worldview.
229
9780306815805_09.qxd 8/22/07 9:28 AM Page 230
We now know that Mr. B’s book is not really the “Archimedes
Palimpsest” at all; the Archimedes codex is only one of the important
manuscripts wrapped up in it. Mr. B’s prayer book houses a small
library of unique ancient texts. In addition to the Archimedes man-
uscript, it contains five leaves that uniquely preserve speeches by one
of Athens’ greatest orators and seven leaves that uniquely preserve
ancient views on Aristotle. It also contains some Byzantine texts: four
leaves from a late tenth-century book of hymns partly in honor of St.
John Psichaites, an abbot in Constantinople who rebuilt his
monastery after it was destroyed in 813 BC by the Krum, the
Bulgarian Khan with the curious wine cup, and two leaves from a
saint’s life. Seven leaves, from at least two separate manuscripts, have
not been identified at the time of writing.
The Palimpsest might not ever reveal all its secrets, but I will make
one prediction. I think it is very likely that Reviel and Ken Saito will
be the last people to discover new text from the Palimpsest, because
since their visit the texts in the Palimpsest have undergone another
transformation. In the twenty-first century, if you want to read what
Archimedes had to say to Eratosthenes in the third century BC and
what Hyperides said to the Athenians in the century before that, you
should not make the pilgrimage to the codex in Baltimore.You can’t
read it there.You need one of Roger Easton’s little silver boxes.
Parenti’s Twin
We know the scribe of the prayer book used several different codices,
but exactly whose library was he recycling? It must have been
extraordinary. John Lowden once said to me, in jest, that it was the
library of Photius. It could not have been his, of course.With the pos-
sible exception of the Aristotle commentary, the palimpsested texts
were written long after Photius died. Still, Hyperides was one of the
authors mentioned by Photius. No modern scholar believed that
Photius had actually read Hyperides, but now it seems that Photius has
230
9780306815805_09.qxd 8/22/07 9:28 AM Page 231
been telling us the truth for a thousand years. Like Photius’ library,
these texts must have been collected together in Constantinople. But
this does not mean that they stayed in Constantinople. Books traveled
with their owners. So where were these texts when they were turned
into a prayer book?
Understandably perhaps, in assembling the scholars to work on the
book, I had concentrated on those who could help with the
palimpsested texts. It was not until the legendary liturgist Robert Taft
got in touch with me that I paid much attention to the texts of the
prayer book. He suggested that I give photos to an Italian scholar,
Stefano Parenti. Stefano noted that the prayer book contained certain
very rare texts, including one for the purification of a polluted con-
tainer and another for the storing of grain.These prayers, and others,
Stefano also found in a manuscript that can almost be described as
the twin of our prayer book. It is in St. Catherine’s monastery in
Sinai, the same place where Tischendorf found the Codex Sinaiticus,
and it was written by a priest named Auksentios in 1152–3. Some of
the prayers are in the very same order. Others, such as a group of
prayers at the elevation of the host and a prayer for the consumption
of the leftover gifts of the presanctified liturgy, Stefano knew to be
specific to Jerusalem in the Middle Ages. Finally, Stefano noted that
there were frequent references to prayers “for this city” in our prayer
book. It seems unlikely, therefore, that the prayer book was made at
St. Sabas, even though it ended up there in the sixteenth century. But
it is eminently likely that it was finished in Jerusalem, just fifteen
miles away, on April 14, 1229.
We do not yet know how the palimpsested texts made it to the
Holy Land and perhaps we never will. The problem is not that it is
so unlikely, but rather that there are so many ways that books could
travel there from Constantinople in the thirteenth century. The
reason for this is that the Holy Land, at this time, was a destination of
choice for Christians from Europe both for pilgrimage and for
crusade. Jerusalem was a particularly interesting place to be in 1229.
231
9780306815805_09.qxd 8/22/07 9:28 AM Page 232
Frederick II, the Holy Roman Emperor as well as the King of Sicily,
Cyprus, Jerusalem, and Germany, the wonder of the world for his
energy, learning, and religious skepticism had finally fulfilled his vow
to go on crusade. On Sunday, February 18, 1229, less than two
months before the date in our book, he liberated from Muslim
control all of Jerusalem, except the Dome of the Rock, as well as
other cities, including Nazareth, where Jesus grew up, and
Bethlehem, where Jesus was born. This indeed was something for
Christians to celebrate.The scribe of the book had joy in his heart as
he wrote his prayers. Now that I knew that we could recover the texts
in the Palimpsest, I had nothing but understanding for the scribe and
thanks for the fact that he had used such treasures to write his book.
Indeed I wanted to thank him personally. The trouble was, I didn’t
even know his name.
232
9780306815805_10.qxd 8/22/07 9:30 AM Page 233
10
The Stomachion, 2003
or Archimedes at Play

A Package from Mr. Marasco
t was September 2003; and I was just back from my summer
I vacation; and a Mr. Joe Marasco had sent me a present. There was
a funny-looking package waiting for me together with the rest of my
mail. Its sender described himself as a fan of Archimedes, which, to
be honest, was rather worrying. I had fewer nuts calling me than Will
did, but I did get my share of them. (And no, I didn’t discover
Rasputin in the Palimpsest either.)
I opened it, cautiously, to find a truly gorgeous toy—large pieces
of red glass cut in all sorts of shapes and fitted together to make a
square. Nice, I thought. A pity, of course, that the pieces were fragile
and sharp, because we were expecting our first child and I was
shopping for toys. But I could keep it at my office and show it off as
an example of the funny things you get when you are an Archimedes
scholar.
I understood Mr. Marasco’s point. He was sending me a replica
of the Stomachion. I vaguely remembered that it was an ancient
game in which the objective was to use fourteen pieces to create
some form. But this just about summed up my knowledge of this
strange toy. I knew there was an obscure fragment by Archimedes
dealing with the Stomachion, but no one had really studied it. The
standard view was that Archimedes used the game as a starting point
to motivate a geometrical discussion of some kind. No one had
233
9780306815805_10.qxd 8/22/07 9:30 AM Page 234
even guessed the nature of the geometrical discussion. I knew why.

Our knowledge rested on a small fragment preserved in a mutilated
form.
I also knew that this fragment was right here in my office.The new
hard drive from Roger Easton had recently arrived. It contained,
among other things, the digitally processed images of folios 172–7.
Well, I thought, Mr. Marasco made a nice gesture. The least I could
do as recompense was try and see if I could read anything from the
Stomachion.
This would be tough. Some time in the sixteenth century, or even
before, the Euchologion manuscript had lost its final folios, 178–185.
They had been replaced by a paper supplement, inserted into the
manuscript in the sixteenth century. Folio 177 became the last folio
and, therefore, the one least protected from mold and other damage.
I did, of course, ask to look at this folio when I visited Baltimore,
but it immediately became clear that nothing could be gained by the
naked eye.The parchment was so worn down that in places it literally
had disintegrated. There were holes in the parchment—Greek words
gone forever. It was not even a well-defined piece of rectangular
parchment. It was made up of precarious, unwieldy pieces—an eaten
surface that only barely held together.
But this was just part of the problem. Even where the parchment
existed, the writing was so faint, so thoroughly covered with mold
stains that nothing could be made of it. To the naked eye, it was as if
this folio had never been palimpsested.The first thing that struck the
eye were large blobs of an ugly, blackish substance—the remains of
the mold. Other than this, you could see—with difficulty—the top
layer of text, but the underlying text had disappeared completely. I
approached the folio gingerly. This was not anything I was going to
extract out of the plastic coating, so I lit it with the UV lamp. Not
much more came up.
Well, that’s it, I thought at the time.We weren’t going to make any
progress with the Stomachion. A pity but, then again, it wasn’t such an
important treatise. And, even if we were able to read more, would it
234
9780306815805_10.qxd 8/22/07 9:30 AM Page 235
A R C H I M E D E S AT P L AY
help at all? The text was just too fragmentary.We would never under-
stand the Stomachion, and it was best for me to invest my time more
profitably elsewhere.
In all of this, I was merely following the traditional response.
Heiberg was only able to read some fragments of the text and he
didn’t venture an interpretation of it. Dijksterhuis, a great Archimedes
scholar, wrote a careful commentary on each of Archimedes’ treatises,
but he had practically nothing to say about the Stomachion. Indeed,
we can see his growing impatience. He began his commentary with
speculation: “[the treatise] may indicate that [Archimedes] studied the
game from a mathematical point of view . . . [he] discussed some of
the properties of the so-called Stomachion.” But then he lost his
confidence: “In the Greek fragment, however, we do not find much
about this investigation.” Dijksterhuis concluded: “It can no longer
be ascertained whether this result was the object aimed at or whether
it played a part (and if so, what part) in the investigation as originally
announced.”
The fundamental point is that this single bifolio of the Palimpsest
is nearly all we had to go by. On this bifolio, Archimedes concluded
his treatise on the Measurement of the Circle and began a new treatise
whose title was barely legible but may have been “Stomachic” or
“Stomachion.” There were some words of introduction, a single
simple proposition, and the beginning of another one. Both, obvi-
ously, were mere preludes to the real action of the treatise. None of
the substantial mathematics was left. Essentially, when the maker of
the Palimpsest chose which folios to use out of the original
Archimedes book, he threw away all of the Stomachion folios except
for this single bifolio. It is easy to see why. The Stomachion was the
final treatise in the original Archimedes book, and as we have already
learned, the important rule of manuscripts is that the end is always in
the worst shape. The parchment on which the Stomachion was written
was probably already in bad shape in the thirteenth century.Thus the
treatise was simply thrown away—not good enough to serve even as
recycled parchment. The maker of the Palimpsest probably reasoned
235
9780306815805_10.qxd 8/22/07 9:30 AM Page 236
that this particular piece of animal skin would not survive another
round of scraping.
There were some additional pieces of evidence, without which
our position would have been even more precarious. The references
were from antiquity to a game called the Stomachion or “Bellyache.”
The game was supposed to be so difficult that it made your belly
turn. (The difficulty of putting the pieces together was a constant
theme throughout the history of the Stomachion.) It involved fourteen
pieces, put together to form a square. So much was implied by the
ancient testimony on the game itself. The evidence suggested that
Archimedes did not invent the game but that he did make some
mathematical reflections upon it. (In the same manner, contemporary
mathematicians sometimes use the Rubik’s cube to introduce ideas
from Group Theory.) These mathematical reflections became so well
known that some people called the game “Archimedes’ Box.”
However very few people ever read Archimedes’ actual treatise. The
only surviving Greek manuscript containing the text of the
Stomachion was in front of us—the original Archimedes book serving
as foundation for the Palimpsest. And so, in the year 1229, when the
maker of the Palimpsest threw away the bulk of the Stomachion, he
threw away the only evidence the world had in Greek.
There was one additional piece to the puzzle. Just like the
Palimpsest, this evidence was ignored for years. In this case, obscurity
was the result not of the ravages of fortune—as with the Palimpsest—
but of scholarly neglect.The manuscript in question was in plain sight
for all to read, but it remained unread for generations simply because
too few scholars read Arabic. Only in 1899 did Suter, a German scholar,
come across an Arabic manuscript from the seventeenth century that
mentions the “Stumashiun of Archimedes.”
Indeed, much of the Greek heritage survives only in Arabic.
(Much of it is still unpublished, because of the same scholarly
neglect; quite possibly, there are more works by Archimedes that
remain unnoticed in Arabic manuscripts.) Starting in the ninth
236
9780306815805_10.qxd 8/22/07 9:30 AM Page 237
century, translations of Greek texts were made in Arabic centers of

learning, such as Baghdad. Typically, however, the Arabic versions
are far removed from the original. Arabic mathematicians were
very good scientists. They added a great deal that was original and
not present in Greek science. They would often rewrite the
sources, abridge them, rephrase them, and so on. This clearly was
the case with the Stomachion. The manuscript found by Suter is
unfortunately no more than an Arabic abridgement of a small part
of the original Archimedean text. The Arabic manuscript is a very
brief text indeed—a couple of folios long—and, once again, fur-
nishes us with very little information. But it does one crucial
thing: it discusses the construction of the Stomachion as a square
divided into fourteen pieces. Based on the Arabic text, we can
reconstruct the precise shape of the Stomachion puzzle (see fig.
10.1). This is a famous diagram—everyone who knows anything
about the Stomachion is familiar with this figure of a square
divided into fourteen pieces, the canonical form of the
Stomachion puzzle. This is what Archimedes was playing with. So,
Marasco’s model was essentially a copy of a diagram contained in
a seventeenth-century Arabic manuscript—with a certain
proviso—more of which later.
This, then, was the sum of our knowledge. There was a treatise by
Archimedes dealing with a certain game, the objective of which was
to construct some shapes from fourteen given forms. This was all we
knew.
No one had even bothered extending this knowledge any further.
When the sale of the Palimpsest was made everyone was excited that
we might get new readings of the Method. Literally no one suggested
that we might get new readings of the Stomachion. This treatise was
the poor relation, the one everyone kept forgetting. Partly, this was
because there was so little left with which to work. More impor-
tantly, the thinking was that it was just a game. It could not possibly
rank with the significance of Archimedes dealing with such major
237
9780306815805_10.qxd 8/22/07 9:30 AM Page 238
FIGURE 10.1
issues as infinity and the application of mathematics to the physical

world. It was Archimedes at play, an Archimedean pastime. It was a
pity we knew no more about it, but at the end of the day, we could
survive without it.
Such was the feeling—my feeling, too—when the Palimpsest resur-
faced. So it was only with reluctance that I turned on my external
hard drive to try and see what the pseudocolor images might teach
me. It would be hard, grinding work, probably for little gain. But I was
going to try nonetheless. At one point or another, I reasoned, I must
look at it and if I did it now I would have something interesting to
tell Mr. Marasco when I wrote him my thank-you note.
Making Sense of the Stomachion

One would have hoped, in principle, that the remaining text we
had would furnish us with enough clues. We had no more than a
single Greek bifolio, but after all this was a crucial one. It was the
238
9780306815805_10.qxd 8/22/07 9:30 AM Page 239
first in the treatise, so it included the introduction. Surely the intro-

duction would give us a sense of the goal of the work. Yet what
Heiberg had managed to read was fragmentary and obscure. He
translated, more or less completely, the first paragraph and it read
as follows:
As the so-called Stomachion has a variegated theory of the
transposition of the figures from which it is set up, I deemed it
necessary: first, to set out in my investigation of the magnitude
of the whole figure each of the figures into which it is divided,
by which [number] it is measured; and further also, which are
the angles, taken by combinations and added together; all of the
above said for the sake of finding out of the fitting together of
the arising figures, whether the resulting sides in the figures are
on a line or whether they are slightly short of that but so as to
be unnoticed by sight. For such consideration as these are intel-
lectually challenging; and, if it is a little short of being on a line
while being unnoticed by vision, the figures that are composed
are not for that reason to be rejected.
This paragraph may be obscure, but it does tell us something.
Archimedes would look at the measurements of the various pieces
that composed the Stomachion. He would look at the angles to see
which pieces, put together, would fit in combination to create, for
instance, a straight line. So the treatise was some kind of study of the
ways in which the pieces of the Stomachion could be made to fit
together.
Now, here comes an important consideration. The people who
most influenced Heiberg’s interpretation of the Stomachion were the
Roman grammarians of the late Imperial period who wrote many
centuries after Archimedes’ death. It so happens that this group of
authors was fond of a certain cliché: they compared the many expres-
sions one can form with just a few words to the many ways in which
one can make different figures with just a few basic shapes.Thus, they
said, one could take the Stomachion pieces and fit them together so
239
9780306815805_10.qxd 8/22/07 9:30 AM Page 240
FIGURE 10.2
as to make an elephant, a warrior, or a bird; the possibilities were

unlimited (see fig. 10.2).
So, the pieces could fit together in a free game of creativity. What
must be stressed is that there is no limit to the number of shapes one
can make in this way. This is because, to make an elephant or a
warrior, one must be allowed to put together pieces in a free manner,
not necessarily setting a vertex next to a vertex. In figure 10.3, we see
a close-up of the “elephant,” and we see how several of the pieces
are placed “loosely” next to each other and do not meet exactly
vertex to vertex. Now, if this is the rule—that pieces may be placed
“loosely” next to each other—then the logic of infinity applies. It
means the pieces can be placed continuously along any edge. One is
240
9780306815805_10.qxd 8/22/07 9:30 AM Page 241
FIGURE 10.3
allowed to attach piece X one-half the way from the end of the edge,
or one-third the way from the end, or one-fourth, or one-fifth . . .
There are literally endless ways of positioning one piece next to
another. The number of different elephants one can make with
fourteen pieces is literally infinite. We are reminded, once again, of
how ubiquitous infinity is in mathematics.
Heiberg, with his huge learning, was aware of this Roman gram-
marian cliché. And so, as he turned from the first paragraph to the
second, he thought he had an idea of what Archimedes meant. He
thought Archimedes was talking about the boundless plurality of the
elephants. Now there was very little Heiberg could read at this
point—the writing becomes nearly illegible—but he thought he
could reconstruct some traces of meaning: “So it is possible . . . many
. . . with the same shapes . . . moved around. . . .” So Archimedes was
saying, Heiberg thought, that there were both elephants and warriors
to be made and many of them too.
What was the point? Heiberg did not know and nor did we, fol-
lowing him. If the point was that there were many elephants to be
put together, then there was no interesting mathematical question to
ask. How many elephants and warriors? Infinitely many or, better
241
9780306815805_10.qxd 8/22/07 9:30 AM Page 242
put, as many as you wished.What on earth was Archimedes trying to

determine? Perhaps, we thought, he was just making some random
comments on the geometry of the fourteen shapes. Not an important
treatise, for sure.
So I still reasoned. I was in good spirits, though, studying the
images on my hard drive.The pseudocolor worked. Indeed, it worked
incredibly well. The manuscript had deteriorated terribly since
Heiberg reviewed it and still, with the pseudocolor technology, I
could go through the lines and occasionally read them as if they were
written in plain ink. In places, it was like waving a magic wand; I
looked at the naked-eye image and saw nothing and then, after
turning on the pseudocolor, Archimedes’ Greek was plain to see. I
quickly confirmed Heiberg’s readings and made a few additional
completions, to make better sense of the first paragraph. Heiberg had
not read everything, but he certainly got the drift, and I could prove
this based on Archimedes’ Greek.
Still, I could not see Archimedes’ point.To make some better sense
of it, I decided to take a step back and read Heiberg’s text again for
whatever it was worth. This was a standard move in such research.
Before taking a plunge into the deciphering of a text, it was good to
try and gain some kind of understanding, however imprecise, to guide
the reading. I turned off the computer and picked up Heiberg,
reading first through the little he could make of the Greek, and then
through Suter’s text.
Just to make sure I was following the text, I compared the diagram
Heiberg provided—the canonical diagram of the Arabic manu-
script—with the model I had from Marasco. Anyway, it would be
more fun working with the model than with the diagram!
At this point, I became furious with Marasco. His model did not
match the diagram. There was something wrong. Was he, after all, a
fraud, someone who couldn’t even read a diagram correctly? Or had
I misread the diagram? I looked at it again and began to wonder if
perhaps something had gone wrong by accident. Perhaps Marasco
242
9780306815805_10.qxd 8/22/07 9:30 AM Page 243
had prepared the model from the correct diagram and then his shapes
had gotten mixed up through some kind of error?
But wait, I wondered: was it possible to fit the pieces together in a
square by placing them in some arrangement other than the original
diagram? I mean, surely there was no more than one way of fitting all
fourteen complex shapes together? It seemed like quite a complicated
arrangement. But then again, maybe there was more than one way of
fitting the pieces together.
Well, it had to be clarified. I certainly was curious. I checked the
figure, piece by piece. Marasco’s model was, indeed, the original
diagram with the pieces of the square arranged differently. Of course,
I saw it now. There were multiple ways in which the diagram could
be rearranged. There was certainly more than one way of fitting the
fourteen pieces together into a square.
And then, all of a sudden, my throat went dry.
Could this have been Archimedes’ point? Were there many dif-
ferent ways in which the same square could be fitted together with the same
pieces? This would be too exciting. Let me explain why.
Improbable Combinations
The significance of my new thought was that, finally, we had a mean-
ingful problem. We were no longer dealing with the continuously
changing, infinite number of arrangements of elephants and warriors.
We were dealing with a certain finite number of ways in which a
square could be made from the given pieces. I always imagined the
number to be 1. That is, my intuition was that the accepted diagram
represented the only way that the square could be fitted together.
Now, thanks to Marasco, I saw that my intuition was plain wrong. I
still had to show that there were many ways. If there were only a
handful of ways of arranging the square, then it wouldn’t have been
an interesting problem for Archimedes to solve. I also needed to show
that the number could in principle be calculated, and that it did not
243
9780306815805_10.qxd 8/22/07 9:30 AM Page 244
involve a huge calculation beyond Archimedes’ means. And so I

needed to study, in pure mathematical terms, the problem of the
Stomachion square: how many ways were there to fit the given pieces
together to form a square? Also, of course, I needed to go back to the
hard drive, read more of the introduction’s second paragraph, and see
if it fitted into the new hypothesis.There was lots of work to be done.
We were going to uncover the prehistory of combinatorics.
Combinatorics is essentially a simple science. As its name suggests,
it is the study of combinations. Suppose, for instance, that we need to
make a choice: we have three candidates for the presidency. How
many combinations do we have? Obviously, the answer is three. Now,
we need to make it slightly more difficult: imagine that we are
choosing not a president but a Roman-style pair of consuls of equal
powers. We need, therefore, to choose two consuls out of three can-
didates. How many options do we have? This may appear tricky at
first sight, but actually the answer, once again, is three. Choosing two
out of three is really the same as choosing one out of three, because
each time we are choosing a single candidate to be left out. To choose
A and B is the same as to leave out C; to choose A and C is the same
as to leave out B; and to choose B and C is the same as to leave out
A. This exhausts, once again, our options.
Now imagine that we are selecting consuls not of equal power but
instead a president and a vice-president. How many options do we
have? This is somewhat more complicated. Essentially, each of our
previous choices bifurcates. Each of them can be turned into two
choices—president and vice-president. If we choose A and B then we
have two president and vice-president pairs to form out of this
selection—A as president and B as deputy or vice versa. In short, for
each choice of consuls we have two choices for a president-and-vice-
president pair. In other words, the number of options becomes
3 2 = 6.There are six ways of choosing a president-and-vice-pres-
ident pair out of three candidates.
This is all suggestive of the nature of combinatorics. In some ways,
indeed, it is a simple science. Many of its questions, even the more
244
9780306815805_10.qxd 8/22/07 9:30 AM Page 245
interesting ones, can be approached without any complicated tools.

This is related, however, to the main drawback in combinatorics—
there are very few short cuts. There is no surprising theory on the
basis of which we can easily solve everything. Rather, it is almost as
if for each new problem we need to invent a new, ingenious
approach. Combinatorics is a science of endless ingenuity, of endless
puzzles and games.
Where did this science come from? This in itself has always been
something of a puzzle. Most scholars think it emerged out of games.
After all, this was what gave rise to the science of probability. In the
seventeenth century, card games were introduced in Europe. Quickly
enough, Europeans everywhere were busy playing card games.
Everyone was betting: which hand would be dealt next? When you
bet fortunes on hands, your mind tends to focus on some very well
defined questions. How likely am I to get an ace? How likely am I to
get a joker? The answers to these questions essentially involve com-
binatorics.You need to calculate how many combinations of cards are
possible and then how many of them contain an ace. If there are a
million possible combinations and a hundred thousand of them
contain an ace, it means that the odds are one-in-ten that an ace will
be dealt. Its worth betting if the return is over one-to-ten and not
worth it otherwise.
Now this is very useful to know. In the long run, the card player
who has combinatorics up his sleeve is bound to win. He is no more
likely to win each individual bet, but he is certain to place his bets in
such a way that, in the long run, he will end up the winner. Which
is precisely why casinos in Las Vegas are prosperous. They apply the
science of combinatorics against people who fail to apply it. Science
wins.
Fermat, known for his Last Theorem, and Pascal, known for his
deep theological observations, were among the first to apply this
science.They did not make a fortune with their bets. (Historical evi-
dence suggests that mathematicians—unlike Las Vegas casinos—are
not very good at following through with their scientific knowledge.)
245
9780306815805_10.qxd 8/22/07 9:30 AM Page 246
Instead of making a fortune, they created the science of combina-

torics, quickly using it to calculate the probabilities of events not only
in card games but in many other domains as well.The calculations of
combinations are much more than a trifling game, it turns out—they
serve as the foundation for the science of probability.
Probabilities now serve as one of the cornerstones of science. And
that’s the reason why combinatorics is so important. Indeed, today’s
physicists believe that the universe is governed by quantum
mechanics, which is essentially probabilistic in character.There are no
rules saying that this or that must happen; physics merely asserts a
certain probability of events happening. Einstein, famously, differed. He
passionately refused to concede to “God plays with dice.” The evi-
dence, so far, seems to suggest that on this Einstein was, for once,
wrong.
Ancient Combinatorics?
There is a certain puzzling, intangible quality to combinatorics. It’s
frequently a very abstract science. There are often no diagrams to be
drawn.You just go through the problem in your head, considering the
various options and possibilities. It is a fun subject—but, generally
speaking, not a visual one.
Now this non-visual character of combinatorics makes a world of
difference. We have already seen many diverse problems discussed by
Archimedes, but even with all their diversity, the great majority them
still dealt with geometry. After all, the diagram was the key tool of
Greek mathematics. Even though Greek mathematicians made inter-
esting discoveries in, say, number theory (for instance, showing that
there are infinitely many prime numbers), their main field was the
visual, concrete science of geometry. Calculating how many ways
there were to make certain selections and combinations? This would
be just too abstract, too non-visual. And for this reason, we didn’t
think combinatorics was a field that Greek mathematicians were
likely to tackle. The standard opinion was that problems of pure cal-
246
9780306815805_10.qxd 8/22/07 9:30 AM Page 247
culation were not made an important part of mathematics before the

seventeenth century.
This view changed in 2002 during the international meeting of
historians of Greek mathematics held in Delphi, Greece. It happened
following a presentation by Fabio Acerbi. Fabio studied physics,
earning a PhD, but then decided it was not his field after all. He
became a high school teacher so that he could concentrate on his
love of the ancient world. (He is the graduate of an Italian public high
school where students study not only the sciences but also Greek and
Latin.) He quickly produced a series of original, inspired studies of
ancient mathematics, combining both his mathematical and linguistic
gifts.The first to make a real splash was the one he presented to us at
Delphi. Its topic was Hipparchus’ numbers.
Here, again, was a question that few had paid any attention to at
all. Plutarch mentions (in the course of an otherwise unrelated philo-
sophical discussion) an ancient quarrel between a philosopher and a
mathematician. The philosopher, the Stoic Chrysippus, said that by
the rules of Stoic logic one could combine ten assertions in more
than a million ways. The mathematician Hipparchus countered
stating that the correct number was either 103,049 or 310,954,
depending on how the number was defined but, either way,
Chrysippus was wrong. Hipparchus was a great mathematician and
astronomer. (Among other things, he was the first to produce a cata-
logue of all the stars visible to the naked eye—a remarkable
achievement by any standard.) But this reads more like some kind of
private joke that doesn’t need any special attention paid to it. And
indeed, historians of mathematics have never tried to make any sense
of the numbers.
Let me now move back to the year 1994 when David Hough, a
graduate student of mathematics at George Washington University,
leafed through a textbook of combinatorics. He came across
Hipparchus’ numbers, which were mentioned as a kind of curiosity.
He also happened to consult a handbook of important mathematical
numbers. This handbook contained, among other things, what are
247
9780306815805_10.qxd 8/22/07 9:30 AM Page 248
known as “Schröder numbers.” The tenth Schröder number was

103,049—the same as the smaller Hipparchus number.
Now this is some coincidence, thought Hough. He consulted with
the author of the combinatorics textbook, Richard P. Stanley, an MIT
professor of mathematics. And in 1997 they published a small notice
in the American Mathematical Monthly suggesting that Hipparchus
could have produced some genuine combinatorics. Lucio Russo, an
Italian historian of mathematics, came across this notice and sug-
gested to Fabio Acerbi that he should look into it. In the summer of
2002, Fabio had a theory of how Hipparchus’ problem could be
defined and how the two numbers—103,049 and 310,954—could be
seen as the correct solutions to this problem. In addition, he could
show, by means available to an ancient mathematician, how the
numbers were obtained.
In essence, one of the possible interpretations of a Schröder
number was the number of ways a sequence of characters could be
put inside brackets. For example, the four characters, abcd, could be
put inside brackets in various ways:
(a(bcd)), (ab(cd)), ((a)(b)(cd)) etc . . . .
The fourth Schröder number was eleven. That is, there are eleven
different ways that the four characters, abcd, can be put inside
brackets. (This is surprisingly high, as is so often the case with com-
binatoric problems.) Acerbi showed that, according to Stoic logic, the
problem of combining ten assertions was analogous to putting ten
characters inside brackets. He then developed a method for solving
this problem within Hipparchus’ means. He showed that with one
extra condition—allowing not only to “assert” claims, but also to
“negate” them—the number becomes 310,954, which confirms the
second number reported by Plutarch.
At the conference, we were skeptical at first. Everything went
against our hard-won intuitions. But the more we looked at Acerbi’s
evidence, the more we became convinced. The numbers just could
not be a coincidence. We were not going to hit upon the tenth
Schröder number by sheer accident.The only way Hipparchus could
248
9780306815805_10.qxd 8/22/07 9:30 AM Page 249
have come up with his numbers was the way Acerbi did—by doing
the math. And so, even though the brief mention by Plutarch tells us
almost nothing, it is sufficient to prove, beyond a doubt, that ancient
combinatorics did exist.
This was a stunning discovery. The study of pure calculation—of
counting the number of possible combinations—was invented by the
Greeks, and it was brought to a high level of sophistication by the
time of Hipparchus.
Hipparchus lived in the second century, which makes him fifty
years or more younger than Archimedes. But there was now nothing
unlikely in assuming that Archimedes was engaged in combinatorics.
It would make him the first person—as far as we can judge—who
ever produced a study of combinatorics. Indeed, it would make
perfect historical sense. Archimedes would be at the inception of a
tradition whose culmination would then be the work of Hipparchus.
The pieces fit together. My interpretation of the Stomachion could fly.
Just to be sure, I sent a quick email to Acerbi—was he familiar with
anyone ever suggesting that the Stomachion was a study in combina-
torics? I then went feverishly back to work on the transcription.
Another email went to Nigel Wilson, alerting him to the significance
of folios 172–7 and asking him to make as much progress as he could
on the reading of those folios. I knew I needed his expertise to
confirm my own guesses.
I also sent another email, to my colleague Persi Diaconis in the
Mathematics Department at Stanford. Persi is a magician-turned-
mathematician. He still likes to perform tricks and one of his favorite
pursuits is the application of mathematics to games. He is famous for
his proof that one must shuffle a deck of cards at least seven times to
make it thoroughly remixed. More recently, he studied the flip of a coin
showing that it is not truly random after all. About 51 percent of the
time coins actually end up landing on the same side as they started. He
likes all sorts of surprising combinations. I knew he would like my
problem, and I also knew that he was a distinguished combinatoricist.
Most importantly, he was a friend. He would not laugh at me for asking
249
9780306815805_10.qxd 8/22/07 9:30 AM Page 250
such a trivial question. And so I put the question to him: how many
ways are there of fitting together the Stomachion square?
Fitting the Pieces Together

The first reply I got was from Fabio. He was quite sure that no one
had ever considered the possibility that the Stomachion was a combi-
natoric study. He pointed out, quite rightly, that until recently no one
had considered the possibility that any ancient treatise was dedicated
to combinatorics. I replied quickly, sharing with him the few readings
that I had and suggesting that he join Nigel Wilson and me in the
writing of an article on the Stomachion. I liked the spirit of teamwork
we had when we wrote the publication concerning infinity and the
Method, and I relished the prospect of another collaborative effort.
However, in this case, the collaboration would be based on email
alone.To this day, Fabio has never laid eyes on the physical manuscript
bifolio of the Stomachion.
The team would expand further. I did not hear back from Persi for
quite a while. It turns out that he doesn’t use computers. Eventually
I left him a note and the next day he showed up at my office telling
me that they were working on it. He had given the problem to his
students. His wife, Susan Holmes, a distinguished statistician was also
hooked on the problem. A number of colleagues, hearing what I was
working on, were sending me emails with calculations. Everyone was
trying out combinations with the fourteen pieces and a square. And
no two solutions were alike. Clearly the precise calculation was much
more complicated than it had appeared at first sight. We were all
coming to terms with the mathematics of the Stomachion.
The easiest way of visualizing the various possible combinations of
the Stomachion puzzle is by thinking of them as the results of sub-
stitutions and rotations. Suppose you start with the original
arrangement of the Arabic manuscript (see fig. 10.4).You could then
take the triangle BZE (composed of the four pieces ZLF, LFEHT,
TKH, KHB) and substitute it for the triangle ZDG (composed of the
250
9780306815805_10.qxd 8/22/07 9:30 AM Page 251
A Z D A Z H D
M M
L O L K
F N F “F”
T C “C” C “T”
K
Q O “L”
Q
B H E G B E G
FIGURE 10.4
three pieces ZDOC, ONC, NCG). The result is a new arrangement.

This is an example of a substitution. Let’s call this substitution S
(above right).
Alternatively, take the triangle AGB (composed of the seven pieces
AMB, MLB, KHB, TKH, LFEHT, FQE, QEG) and rotate it around
an imaginary axis passing through the points F and B. The result is a
new arrangement (see fig. 10.5).This is an example of a rotation. Let’s
call this rotation R.
It would be the easiest if we could just count all the possible substi-
tutions and rotations and then multiply them to get the number of pos-
sible arrangements.This was the basic approach everyone was taking at
the outset. But it would not do.This is because there are very complex
ways in which the various substitutions and rotations interact. For
example, when you’ve applied substitution S, you can no longer apply
rotation R. Substitution S destroys the line AG in the triangle ABG.
There is no longer a triangle to rotate. Reciprocally, once you’ve
applied rotation R, you can no longer apply substitution S, because
rotation R destroys the triangle ZBE.There is no longer a triangle con-
gruent with the one at ZDG. In short, there is a very complex pattern
of ways in which substitutions and rotations can combine and ways that
they cannot. It becomes a second-order combinatoric problem, above
251
9780306815805_10.qxd 8/22/07 9:30 AM Page 252
A Z D A Z D
L
L O O
F N E F
T C C N
K
Q Q
T
H
“L”
K “M”
B H E G B G
FIGURE 10.5 Rotation R

A Z D A N O D
M M
C
L O Q
F N F
T C T Z
K K
Q
L
B H E G B H E G
FIGURE 10.6 Rotation R*
and beyond fitting the fourteen pieces together, the problem is fitting
the substitutions and rotations together. This kind of complexity, with
combinations and combinations-of-combinations, very often arises in
discrete mathematics.
There is another complication.We saw that some substitutions and
rotations rule each other out, but others cancel each other out. To see
this, let us consider a very simple case. One possible rotation, as we
saw, is rotation R: turning the triangle ABG around the imaginary
252
9780306815805_10.qxd 8/22/07 9:30 AM Page 253
A D
Rotation R** R otation R
B G
A D A D
B G B G
Rotation R*
FIGURE 10.7 The three rotations R, R* and R**, combined, end up canceling each
other: we’re back to the original square
axis FB. Another possible rotation, of course, is what we may call

rotation R*: turning the triangle AGD around the imaginary axis FD
(see fig. 10.6). What happens if we apply both rotations R and R*?
We end up having rotated the entire square. Nothing in the internal
arrangement has changed. In this sense, the two rotations cancel each
other out. Worse still, if we add in another allowed rotation, call it
R**, where we rotate the entire square around the imaginary axis
DFB, then the effect of the combination of the three rotations R, R*
and R** is to cancel each other out. Having applied them all, we
return to exactly the same starting position (see fig. 10.7).
253
9780306815805_10.qxd 8/22/07 9:30 AM Page 254
This is once again a typical situation in finite mathematics. It is a

situation most closely studied by a branch of mathematics known as
Group Theory. Group Theory is, essentially, the study of various ways
that permutations add up and cancel each other. It is this theory that
is demonstrated by the Rubik’s cube, and we find that it can also be
demonstrated by the Stomachion. The simple game contains in itself
an introduction to finite mathematics.
For our immediate purposes, though, the following is what mat-
tered. A simple multiplication of all the allowed substitutions and
rotations would provide us with what is known as an “overcount” of
the Stomachion solutions. This happens for two reasons: some of
these substitutions and rotations rule each other out and cannot be
combined; while other combinations are allowed but then end up
canceling out.There are fewer Stomachion solutions allowed than the
simple multiplication suggests.
How many solutions were there? I was on tenterhooks. I needed
the number to be high enough. If I ended up with some twenty or
thirty solutions, it would be an obvious anticlimax. Surely
Archimedes wouldn’t have bothered with such a treatise unless the
number was big enough.
The weeks passed. The mathematicians were still working at it.
Meanwhile, I kept going back to my hard drive, adding to the tran-
scription, each time a few more characters. Gradually, word by word,
it made more and more sense. In a way, it was the mathematicians
who helped me the most in the reading. This is how readings take
place: you can make a reading only after you have formed some kind
of guess as to the possible meanings of the text to be read.This, above
all, was why Heiberg had not succeeded in reading the infinity
passage in the Method or in reading the Stomachion. He had expected
neither actual infinity nor combinatorics.
I began to make sense of the small theorem we had, right after the
introduction. The mathematicians made clear that a major simplifi-
cation resulted from the fact that certain pieces were “glued
together.” It could be proved, geometrically, that no substitution or
254
9780306815805_10.qxd 8/22/07 9:30 AM Page 255
rotation could ever separate, for instance, the two pieces AMB, MLB.
There is no legitimate way the two can fit inside the square unless
they are glued to each other along the side MB. In effect, it is as if we
had just one piece—ALB—with the line MB forming a kind of dec-
orative pattern, no more. Applying such reasoning in two other
places, it could be shown that the problem amounts to a puzzle with
eleven effective pieces, not fourteen. This, indeed, was a major sim-
plification. And it became clear that the first small theorem was likely
to make a contribution toward this kind of geometrical analysis.
Better still, the analysis of the structure of substitutions and rota-
tions seemed to make sense of the second, and last, paragraph of the
introduction—the one that Heiberg had been unable to read. I could
now, finally, offer a reading and one that was even supported by Nigel
Wilson. This was crucial, Nigel had not been privy to the mathe-
matical discussions.While those discussions were indeed requisite for
my own formulation of the text, it was important to see that the
reading was there even without any knowledge of the mathematics.
And so the text that we put together went like this:
So then, there is not a small number of figures made of them,
because of it being possible to rotate them into another place of
an equal and equiangular figure, transposed to hold another
position; and again also with two figures, taken together, being
equal and similar to a single figure, and two figures taken
together being equal and similar to two figures taken together—
then, out of the transposition, many figures are put together.
It appeared as if Archimedes was discussing precisely this phe-
nomenon of rotations and substitutions.
But at an even more elementary level, the most crucial thing for
scholars, at this point, was that the new reading was inconsistent with
the “elephant and warrior” interpretation.This treatise was not about
how many different figures could be composed. We know this, now,
because of the repeated insistence on congruity of different pieces and
piece-combinations. For the sake of moving pieces about, as you do
255
9780306815805_10.qxd 8/22/07 9:30 AM Page 256
when putting together a warrior or an elephant, such insistence was

irrelevant. But it was absolutely to the point, if the goal was to
compose different combinations within the same square.Those different
combinations arise precisely from the fact that one can substitute one
piece for another (or one combination of pieces for another) because
the two are congruent.
We were thus confident that Archimedes’ treatise dealt with the
problem of the combinations for constructing the square from the
given fourteen pieces. And we could say even more. The clear
emphasis of the introduction was on one small statement to which
the first paragraph led and from which the second paragraph took
off. It appeared that this statement asserted the very point of the
treatise. So here was what Heiberg could not read: “ . . . There is not
a small number of figures made of them . . .”
This is what Archimedes was doing in this treatise—counting a big
number. The operative word—“number”—is in fact the very same
word, plethos, which was so crucial in the reading of the infinity
passage in the Method. (In the context of abstract proportion theory,
plethos is usually translated as “multitude,” but in this counting-based
context the best translation is “number.”) In both cases, Archimedes,
surprisingly looking at big numbers—those of infinity, in the infinity
passage, and those of combinatorics, in the Stomachion.
But was the number really that big? This we still did not know. I
knew that the various back-of-the-envelope calculations people were
producing at first all involved an overcount because of the problems
previously mentioned. How many remained? I just did not know. As
the days passed, I began to worry that my problem, which at first I
thought might have been too trivial, might in fact be too compli-
cated. If modern mathematicians couldn’t solve this straight away,
perhaps I was wrong in believing Archimedes would have tackled it.
I finally scribbled my thank you note to Marasco, mentioning that
the Stomachion might be even more interesting than it appeared at
first sight. He promptly showed up at Stanford. It turned out that
Marasco was a retired businessman from the computer industry with
256
9780306815805_10.qxd 8/22/07 9:30 AM Page 257
a PhD in physics. He understood very well the mathematical issues,

and he also had the business experience and the contacts to suggest
one further avenue for progress. We needed to spread the word
among computer scientists, offering a small reward for the first person
that came up with a solution to the Stomachion puzzle. He thought
$100 would be enough, which he then put up. It became the
informal Marasco Stomachion Award.
My mathematical friends had now decided that they needed to get
their minds around it, too. Persi Diaconis and Susan Holmes had
already brought in another renowned combinatorics couple, Ron
Graham and Fan Chung from UC, San Diego.They had already been
conversing by phone and email for a few weeks, but at some point
they decided they needed to take a more hands-on approach. Persi
and Susan traveled to San Diego and, for a long weekend, the four of
them did nothing but draw diagrams and study the combinatoric
principles underlying the Stomachion puzzle. At the end of it, they
came up with what they were quite sure was the final count.
At the same time, Bill Cutler, a computer scientist from Illinois
found a way of defining the problem in computer algorithm terms.
He described to the computer how to put together a Stomachion
square and then wrote a program that went systematically through all
the potential arrangements. Many of them aborted. The program
then counted all potential arrangements that did work. In this way, it
counted all of the actual solutions to the Stomachion. The software
came in first: Bill Cutler won the Marasco Award!
The answer was 17,152. Specifically, there were 17,152 different
ways of arranging the pieces and still getting a square. In figure 10.8,
you can see part of the printout produced by Cutler’s software. And
to think, I had been surprised to find out that there was more than a
single solution!
Mercifully, the group of mathematicians found the same number.
And while they could not claim first place, they did something extra
that was crucial for our understanding of the problem. Cutler’s software
was based on counting all the possibilities one by one, a process that in
257
9780306815805_10.qxd 8/22/07 9:30 AM Page 258
FIGURE 10.8 Some of the solutions found by Bill Cutler’s software
principle can only be done by a computer. Archimedes certainly did

not do it that way. However, the group of mathematicians found their
number with “pencil and paper,” which would have been Archimedes’
method (only he would have used papyrus and a reed pen). They lit-
erally did not use any computers for their work. Nor did they rely on
any high-powered mathematics that were not available to Archimedes.
258
9780306815805_10.qxd 8/22/07 9:30 AM Page 259
FIGURE 10.9 A typical family of substitutions between Stomachion solutions, found by

the group of mathematicians that solved Archimedes’ puzzle
Instead, what they did was produce an ingenious “map” of the possi-
bilities (see fig. 10.9).The various solutions to the puzzle were arranged
into twenty-four basic “families,” depending on a certain arrangement
of the main constituents. Inside each of the twenty-four families was a
list of the basic possible solutions drawn with lines connecting any two
solutions that could be transformed into each other by simple substi-
tutions and rotations. At this stage the study had produced 536 basic
solutions. Finally, certain simple rotations, which did not involve any
substitutions, were applied independently to the rest to generate from
each basic solution thirty-two rotations. This finally gave rise to the
17,152 solutions found by the mathematicians.
259
9780306815805_10.qxd 8/22/07 9:30 AM Page 260
We now had everything in place. We had the historical context,

thanks to Fabio Acerbi’s pioneering work. We had the reading, con-
firmed by Nigel Wilson.We had the mathematical solution, produced
by a distinguished group of mathematicians and confirmed by com-
puter software. We also knew that this solution was well within the
means available to Archimedes. The pieces fitted together. We had an
watertight argument showing that, with the Stomachion, we have the
earliest evidence, anywhere, of the science of combinatorics.
It was early December 2003. I announced our new discoveries at
a conference at snowbound Princeton. Gina Kolata, science corre-
spondent for the New York Times, was present and two weeks later we
found ourselves on the front page of the Sunday New York Times. “In
Archimedes Puzzle, a New Eureka Moment,” read the title. Which,
indeed, it was. We had now gained a completely new understanding
of Archimedes, as well as the process of the making of Western
science. Once again, we had rewritten the history books. And we had
done something extra, more precious than everything else. We had,
quite by accident, stumbled upon Archimedes at play.
260
9780306815805_11.qxd 8/22/07 9:34 AM Page 261
11
New Light on an Old Subject

he autumn of 2003 might have been a high point for Reviel,
T but it was a low point for everybody else. The new interpre-
tation of Stomachion was thrilling, but you could still look at the
Palimpsest 17,152 ways and, even after three and a half years, it was
not completely disbound. And however many ways you looked at
the images, they were just not good enough—at least not good
enough for Mr. B. He was not satisfied with our progress, and he
told me so in no uncertain terms. Worse than that, Reviel agreed
with him. There were passages on folios already imaged that he still
could not read. And they were important passages. The very first
folio of the Palimpsest, for example, was Floating Bodies. Heiberg
had not read it; it had never been transcribed; and Reviel and Nigel
were making very little headway with it. Most challenging, of
course, were the folios containing the forgeries. The pseudocolor
and ultraviolet were of almost no use on these folios. And very little
progress was being made on the pages of the Palimpsest that con-
tained texts from other palimpsested manuscripts. The pressure to
decipher these grew to a new pitch after Hyperides was discovered,
but we were having little success.
I told them I was happy to look for other solutions. I was lying. I
was exhausted, and I really didn’t think we could do any better. Mike
Toth told me that there was no point in sending out a new Request
for Proposals across the country. Since September 11, 2001, most
imagers had been working on lucrative government contracts devel-
261
9780306815805_11.qxd 8/22/07 9:34 AM Page 262
oping systems to locate and identify terrorists. He should know, I

thought. Mike and I have a “don’t ask, can’t tell” policy. I don’t ask,
because he can’t tell. Just staying the course was a daunting prospect
for me, for Abigail, and for Roger.The forgeries, in particular, seemed
entirely beyond us. I had no idea what to do.
A Meeting of Minds
Inevitably it was Mike who came up with ways forward. His first idea
was to get us to the heart of the intelligence effort at the CIA, to the
wizards of Langley. Mike even got me into CIA headquarters, which
was quite a feat so soon after the attacks of 9/11.We toured the CIA
museum with its curator and met Charlie the Catfish. Charlie really
looks like a catfish, and he swims like a catfish, too. He’s actually a
mechanical gadget and his mission is still classified. Not sure he will
be of much use in Afghanistan or Iraq. I also saw a bug, literally and
figuratively—a remotely controlled dragonfly that can steal sound.
Unlike Charlie, she had never been operational. She was too easily
blown off course by the wind. After leaving the museum we went
into the elevator and up several floors. I was introduced to Dr. Don
Kerr, who was the Deputy Director of the CIA for Science and
Technology. I presented him with images of Archimedes’ Floating
Bodies to hang on his wall. It was, after all, the government that had
invented multi-spectral imaging. The imagers spent two hours in
briefings and conversations with the experts at the CIA; they clearly
knew far more than they were allowed to say. It was interesting, but
we did not come away with the radically new approach to our
problem that we so clearly needed.
Program managers sometimes take a bad rap as unimaginative bean
counters. Not mine. Turning 180 degrees away from his secret
resources, Mike decided to take full advantage of the press.
Archimedes had always found his friends through the press. It was
because of the Washington Post that Mike had gotten in touch with
262
9780306815805_11.qxd 8/22/07 9:34 AM Page 263
NEW LIGHT ON AN OLD SUBJECT
us, and Keith, Roger, and Bill had known the manuscript was at the
Walters. The subsequent success of the project had secured more
airtime. Within a few days of the publication of Will Peakin’s article
in the Sunday Times, I had a chat and a quick bite with John Lynch
of the BBC at Washington Union Station. John told me that he had
produced a science program that recounted the story of Andrew
Wiles’s single-handed effort to prove Fermat’s Last Theorem. I had
seen and marveled at this program. I agreed to make a documentary
with the BBC’s flagship science series Horizon. The director of
“Archimedes’ Secret,” as the program on the Palimpsest was called,
was Liz Tucker. It aired on March 14, 2002 and attracted 2.9 million
viewers—over 13 percent of the UK viewing audience that evening.
The project to retrieve the unique texts of Archimedes was now
world famous. Surely, ambitious and cutting-edge imaging scientists
would want to be in on this act? Let’s cut to the chase, Mike said. Let’s
have these guys thrash out solutions to the most difficult imaging
problem in the history of science.
Mike told me that I should not waste anybody’s time—least of all
mine. I didn’t have to bother with a full scale RFP. I just had to send
out a short summary of our problem: we needed to read text that had
been written on animal skin in about 970 AD; erased shortly before
April 14, 1229; written over; scraped off again; and covered in
paintings. Anyone who responded to the challenge would not have to
write thorough, wordy proposals just a brief summary of 500 words.
To any credible proposal, I would give a good sample of our data, as
well as information on how we had collected and processed it, and tell
them that they were welcome to improve upon it. Mike told me to
make it clear that I was not looking for science experiments, but rather
practical propositions that could be made to work within six months.
He then told me to invite the best ten suggestions to a summit at the
Walters.This was, Mike insisted, an efficient and quick way to explore
a variety of new approaches. It would also be inexpensive. Mike
insisted on this point; we do not pay these people anything. The best
263
9780306815805_11.qxd 8/22/07 9:34 AM Page 264
of them, he said, would not come for money; they would come for
Archimedes.
And come they did. This was thanks in no small part to Keith,
Roger, and Bill, who searched for anyone who might help. Kirk
Martinez, whom Bill and Keith had been to see in London a year
earlier, came from the University of Southampton. From the
University of Rutgers, Bill conscripted the delightful Professor of
Chemistry, Gene Hall, and from Bartlesville, Oklahoma, he pulled
in Bob Morton and Jason Gislason who both work for
ConocoPhillips. Andy Johnston, who worked on the Archimedes
database, brought in the late John Hillman from the University of
Maryland and his colleague Bill Blass from the University of
Tennessee. They had recently imaged the “Star-Spangled Banner.”
Abigail found Emanuele Salerno. Emanuele came from Pisa and
represented the Easyreadit consortium, a European advanced
image-processing collaboration with representatives in the
Netherlands, Italy, the United Kingdom, and France. She also con-
tacted Mike Attas and Doug Golz from the University of Winnipeg,
Canada. Finally Uwe Bergmann came from Stanford. His mother
Ingrid, who lives in Karlsruhe, Germany, subscribes to GEO mag-
azine. Uwe was visiting her from California, where he is a scientist
at the Stanford Linear Accelerator Center. Although Ingrid did not
know much about his work, she thought he might be interested in
an article on the placebo effect in the magazine. So she left it on his
bedside table. He picked it up, read it, and happened to look at the
next article too. It was by Katja Trippel, and it was an excellent
piece on the Archimedes Palimpsest. It caught Uwe’s imagination,
and he thought he could help. He emailed us at exactly the right
time.
The summit started on Thursday, April 1, 2004, which I thought
was most appropriate. In my undying skepticism, I asked Mr. B to
attend. I did not want to try to convince him of something I almost
264
9780306815805_11.qxd 8/22/07 9:34 AM Page 265
certainly would not believe in and probably would not understand. If

he decided to support anything that a scientist suggested, he needed
to know precisely what he was getting into, both in terms of time and
money. He came and watched the conference unfold. Everybody
played his or her part, and it was intense. We didn’t have the time to
be genteel. We wanted clarity, we wanted proposals, and we wanted
them both fast. Tempers flared as eminent scientists advocated their
proposals and criticized their opponents, their equipment, and their
competence. Under these circumstances, my generous hospitality
budget became important, as drinks were passed well into the night.
On Sunday morning Mr. B, Abigail, Roger, Mike, and I met in closed
session. Then and there Mr. B approved the funding of three new
approaches to the task.
New Approaches
OPTICAL CHARACTER RECOGNITION

Could a computer guess Greek characters better than Reviel Netz
and Nigel Wilson? Mr. B thought that it would be an interesting
experiment to try. So did Reviel. I think, like the chess player Gary
Kasparov, he wanted a computer to beat. I kept my thoughts to
myself. I wasn’t particularly concerned. Computer processing would
not interfere with the campaign as a whole. It would not involve the
manuscript itself, because the computer would work best from the
pseudocolor images that we had already generated. In terms of
workflow, Mike assured me that Optical Character Recognition by a
computer was a breeze.
There were several impressive OCR proposals, but they all
seemed a long way from implementation given the time period
required. Mike, once again, came up with an idea to get the desired
result. We would hold a competition. The goal of the competition
265
9780306815805_11.qxd 8/22/07 9:34 AM Page 266
would be for the imaging scientists to produce something that

would help Reviel read the text. The machine that got closest to
Reviel’s transcription of two designated pseudocolor folios would
triumph, and Mr. B would pay $10,000 to the winner. Whether or
not you think $10,000 is a lot of money depends upon who you
are. To design a machine for Reviel, a professor might not think it
is a lot of money. And clearly they didn’t, because we didn’t get
machines from them. But to Derek Walvoord, a graduate student
of Roger Easton’s at RIT, $10,000 seemed like a lot of cash. Six
months later, he delivered his product.
Derek set himself the task of identifying characters by com-
paring them to a known alphabet. His machine was effective and
very simple to operate. It worked on any PC. You pulled up a
pseudocolor image of the folio you wanted to work on and
selected a “Region of Interest,” normally a partially obscured
character. Then you ran the software and watched as the machine
did its number crunching. It produced a list of characters in order
of likelihood. It worked, too. You could click on a partially
obscured theta, and the machine would display a theta as the
closest matching character. Amazing. The only problem was that
we already knew it was a theta. The letters we wanted the com-
puter to recognize were precisely the ones that were so obscured
that the human eye could not recognize them.
Derek’s machine was affectionately called DEREK. DEREK was
impressive. But it wasn’t Deep Blue, and it couldn’t beat the ten giga-
neurone computer inside Reviel’s skull. But it showed enough
promise for Mr. B to give the go ahead for DEREK II. DEREK II is
much more powerful, because it combines optical recognition of the
characters with a statistical approach to the Greek alphabet and
Archimedes’ vocabulary.This will be useful in helping the scholars see
possible combinations of letters and words in the parts of the text that
have been completely eaten away by the mold. It is being tested as
this book goes to the press.
266
9780306815805_11.qxd 8/22/07 9:34 AM Page 267
EL GRECO
There were several multi-spectral-imaging proposals at the con-
ference.We could, in theory, have invited any of these contributors to
image the Palimpsest. Emanuele Salerno had worked extremely hard
with the data that we had sent him, but he concluded that no better
results could be achieved with that data. Like other participants, he
wanted to collect more data.There is virtually no limit to the amount
of slices that you can add to your data-cube; if you image with
enough wavelengths, you are called a hyper-spectral imager. But Mr.
B concluded that it was not worth the investment to essentially revisit
techniques that we had already tried.
Bill Christens-Barry was on the inside track, of course. He knew
that Mr. B was no longer seriously invested in multi-spectral imaging.
Bill came to the conference with a very inexpensive method to
address all the problems that the scholars had found in the early trials.
His solution would give a far more finely sliced data-cube than was
possible with the push-button processing Keith had used to process
the pseudocolor images. So, he put his idea forward. A year or two
earlier, he and Keith had visited the National Gallery in London to
see a multi-spectral imaging apparatus called VASARI. Giorgio Vasari
was a sixteenth-century Italian painter, but now VASARI was an
acronym for Visual Arts System for Archiving and Retrieval of
Images.VASARI interested Bill and Keith because it did not filter the
light in front of the camera. Instead it used light sources of very par-
ticular wavelengths, and thus avoided filters altogether.This approach
largely eliminated the registration problems that beset Keith and
Roger when they compiled their dense data-cubes. The problem
then was that narrowband light sources of sufficient intensity were
very expensive. But technology is always on the move and in 2004
Bill realized he had an extraordinarily cheap way to generate light at
specific wavelengths. He could use Light Emitting Diodes or LEDs.
LEDs are the lights that you can find all over the dashboard of your
car, and have been able to for some time. But only in recent years
267
9780306815805_11.qxd 8/22/07 9:34 AM Page 268
have they become available at many different wavelengths. LEDs are

so cheap that they are almost throwaways. Bill’s thought was to attach
them to fiber-optic cables and illuminate the parchment in various
wavelengths of light.
Mr. B agreed to fund Bill’s proposal, so Bill put his machine
together and we tagged the experiment on to our last production-
imaging session at the end of 2004. Roger brought the monochrome
scientific camera from RIT that he had used with Keith in his initial
experiments. To achieve the same resolution as the Kodak camera,
forty images needed to be taken of each folio. Bill’s machine looked
like a wonderful gadget, and Bill had automated the controls so that
it could take a great number of images quickly. LEDs are easy to build
into electronic circuitry and have no moving parts, so the automation
of the system was very simple. This was effective hyper-spectral
imaging on the cheap. Although the idea stemmed from VASARI,
Bill’s imaging machine looked nothing like it and a different tech-
nology was used. We christened it EL GRECO, the nickname of the
great painter Domenico Theotokopoulos, as a nod to the Greek text
that it was designed to capture.
Using EL GRECO, Bill avoided the registration issues that had
plagued the initial trials, and he had the narrowband light that he
needed to slice his data-cube more finely. Bill found that the most
effective post-processing technique for the EL GRECO data was the
algorithm that Keith had written for the push-button processing. In
fact, time and again during this experimental phase, we realized how
effective the push-button processing was at retrieving Archimedes’
text. The EL GRECO images of the Archimedes text were slightly
better than the standard images that we had been distributing to the
scholars. More importantly, EL GRECO gave us the chance to tailor-
make our wavelengths for the different texts in the manuscript. The
standard pseudocolor worked better for the Archimedes text than it
did for the Hyperides and the Aristotle commentary. By using dif-
ferent LED sources for the various palimpsested codices, we hope in
the future to make significant strides.
268
9780306815805_11.qxd 8/22/07 9:34 AM Page 269
EL GRECO was an improvement, but it was no better at reading

through gold than the standard techniques that we had been using.
To do this we needed a very different technology.
X-RAYS
Gene Hall, Professor of Chemistry at Rutgers University, calls himself
the “Paper Detective.” He specializes in identifying and dating for-
geries of all sorts, but particularly letters and banknotes. He does this
by examining their chemical composition using X-ray fluorescence.
X-rays, just like visible light, consist of photons, but the photons of
X-rays have a much shorter wavelength (hundredths of nanometers,
rather than the hundreds of nanometers in visible light) and a much
greater energy. The human eye cannot see them but other detectors
can.These detectors can convert the information into a form that we
can see.We are all familiar with X-rays because of the dentist. But the
X-ray images of our teeth are generated by transmitted X-rays. X-
rays are zapped through our jaw and received by an emulsion plate
on the other side. Gene wasn’t interested in the transmitted X-rays.
He was interested in the X-rays that do not get through. These
interact with the material that stops them, and they cause this
material to send out other X-rays at very particular wavelengths.
These emitted X-rays contain crucial information—if you can get it.
Now here is the important part.While the photons of visible light
give you color information, X-ray photons give you elemental infor-
mation. This is because they interact with atoms differently. In the
early 1920s, Niels Bohr and his colleagues thought of the atom as
containing a nucleus of protons and neutrons. Electrons were found
around this nucleus, orbiting in shells at various distances from the
nucleus. This may well be how you think of an atom; it is one of the
very last creations of classical physics and it will serve our purposes.
Bohr labeled each shell with a letter, the one closest to the nucleus
being letter “K.” (The reason that these distances were designated
with letters from the middle of the alphabet is simply that when sci-
entists first probed the makeup of the atom, they were not sure how
269
9780306815805_11.qxd 8/22/07 9:34 AM Page 270
many shells they would find so they left room at either end of the
sequence.) Photons of visible light interact with electrons found in
Bohr’s outer shells. Since they are further from the nucleus, less
energy is needed to change the state of these electrons. Higher-
energy, shorter-wavelength X-ray photons interact with electrons
found on Bohr’s inner shell, K, where more energy is needed to
change the state of the electron. When I say that X-ray photons
change the state of electrons on the inner shell, I actually mean that
they knock them from the shell entirely. However, at the same
moment that the electron “on” shell K is displaced, an electron from
the next shell, L, replaces it.The electron on shell L makes a quantum
leap to the inner shell.As it does so, it loses a lot of energy and it emits
an X-ray photon. Since the atoms of each element have their own
distinctive arrangement of electrons, the precise wavelength of this
emitted X-ray corresponds to the energy difference of the electrons
involved. Hence it will be specific to the element of the atom hit by
the incident X-ray. If you can detect this emitted X-ray, you can
determine which element it came from.
Gene’s thought was that his instrument should be able to detect
the X-ray photons given off by the iron in the ink of the palimpsested
texts. It was a clever idea, and it was the idea that Gene had very
briefly put into practice in his lab with a forged leaf from the
Palimpsest before the conference. But the results were inconclusive.
Another participant in the conference was convinced that Gene’s
idea was a good one. His name was Bob Morton, and he was a
research scientist at the petroleum company ConocoPhillips. Bob is
not normal. He does not have a measurable IQ because he does not
belong to the population for which the test was devised. He is one of
the most intelligent, alarming, funny and inventive people I have ever
met, and I say this without ever having been to one of his fabled
Fourth of July parties in Bartlesville, Oklahoma. He came to the con-
ference with his minder, Jason Gislason, who interpreted Bob for the
rest of us until we got used to him. Bob’s presentation was, frankly,
270
9780306815805_11.qxd 8/22/07 9:34 AM Page 271
amazing. It wasn’t about Archimedes at all; it was about fossils. Using

the same machine as Gene, Bob had looked at fossils from the famous
Burgess Shale. He had mapped the distribution of elements in the
fossils. More than this, he had mapped the elemental composition of
the stone beside the fossilized bones, so that he could determine the
chemical makeup of the soft tissue of the fossil. He called the results
EXAMS—“Elemental X-ray Area Maps.” His images of the fossils
were far clearer than normal photographs. His final image stole the
show. He called it a SEXI—a “Stereo Elemental X-ray Image.” It was
a 3-D image of the fossil Marrella splendens, made out of various
EXAMs in silicon, iron, and potassium. He had made this by taking
two EXAMs of the same fossil at a slightly different angle—actually
7.5 degrees.This is the difference in angle at which your two eyes see
the same object at a distance of 4.2 feet. He then overlaid the two
images on top of one another and color-coded them so they could
be seen separately by your eyes when wearing 3-D glasses.The result
was truly amazing. Not only did you see the fossil in startling 3-D
clarity, you also saw its elemental composition. Not surprisingly, I
wanted Bob and Jason on my X-ray imaging team and so did Mr. B.
Both Gene and Bob used an EDAX Eagle II Micro X-ray
Fluorescence imaging instrument. Inside a chamber is an X-Y stage
(a calibrated moving platform) controlled by a computer. The stage
moves the sample beneath an X-ray generating tube and an X-ray
detector. The software on the EDAX machine is very clever. The
detector actually picks up a wide range of X-rays.The result is a data-
cube of information analogous to the data-cube collected by the
multi-spectral imagers. However, since this is a data-cube of X-rays,
it contains elemental information rather than color information. As
the sample is scanned, the computer automatically produces EXAMs
extracted from the data-cube, each of which displays the distribution
of a particular element across the area of the scan. EDAX probes cost
in the low hundreds of thousands of dollars. Mr. B was all for buying
one, but we thought we had better try it out more vigorously first.
271
9780306815805_11.qxd 8/22/07 9:34 AM Page 272
So we got in touch with EDAX, specifically Tara Nylese and Bruce

Scruggs.We asked if we could come to their offices in New Jersey for
a week and try out their machine with two folios of the Palimpsest.
So Abigail, Bob, Gene, and I went to their office in New Jersey and
occupied it for a week. Tara, Bruce, and the entire EDAX team
opened their doors.
We brought with us one of the toughest of all the challenges, folio
81. We intended to image the recto page. It was not a particularly
important part of the Palimpsest. It contained part of Archimedes’
Equilibrium of Planes, the text of which was well known through
Codex A. It was also in good physical shape. But it was almost entirely
covered by a forgery. Our first scans were not very successful. But
with Bruce, Bob, and Gene in consultation, we refined our param-
eters. We doubled our dwell-time (the amount of time that the
detector stayed over a particular area), so that we could pick up a
bigger signal.We also increased our resolution (the granularity of the
recorded image in dots per inch) and reduced the area that we
scanned. We concentrated on just one line where we thought
Archimedes text might be present. Fifteen hours later, we had a
whole bundle of maps.We had a large number of maps that contained
elements from the forgeries. We had a gold map (blank across the
folio), a zinc map, a barium map, and a copper map, all of which
brought out different parts of the forgery. But we also had an iron
map. I emailed the iron map to Reviel; he could read the words: para
eutheian. We had read through the gold.
There was only one fly in the ointment. In fifteen hours, we had
scanned half a line of Archimedes text. There were approximately
thirty-five lines of text on any given forged folio. If we worked on all
four forgeries, it would take us 4,200 hours. Bob had always warned
me that time was the biggest factor in X-ray imaging. If we went
ahead with this procedure, I would be retired by the time we had fin-
ished imaging the Palimpsest.
272
9780306815805_11.qxd 8/22/07 9:34 AM Page 273
Beamtime
The conclusion was clear. Since we didn’t have the time, we needed a
more energetic source for our X-rays.This was when Uwe Bergmann
showed his stuff. He also gave a presentation at the conference sug-
gesting that X-rays could be used to retrieve Archimedes text. While
Gene and Bob proposed to do their work on an EDAX Eagle machine,
which is the size of a small refrigerator, Uwe Bergmann proposed to
use a machine the size of a soccer field—the SPEAR (Stanford
Positron Electron Accelerating Ring). It is part of the Stanford Linear
Accelerator Center or SLAC in California. SPEAR was built as an
atom smasher—more technically, a synchrotron, which is an oval par-
ticle accelerator. The particles, electrons, and their positively charged
equivalents, positrons, are accelerated to very, very, very nearly the speed
of light. The electrons travel in one direction around the ring and the
positrons travel in the opposite direction. When they collide, they
create new particles and particle physicists analyze the results. It was at
SPEAR that Burton Richter discovered the charm quark in 1974 and
Martin Perl the Tau lepton in 1976. This is very cool. It is, in fact, so
very cool that I was rather determined to get Archimedes there, even
if it meant transporting him across the United States.
The SPEAR is not used as an atom smasher anymore, and anyway,
we didn’t want to hit Archimedes with particles traveling at
99.999999986 percent of the speed of light. Rather, we wanted to hit
it with light itself, and now SPEAR is used as the world’s greatest
light bulb. To explain this, we need to go back to two of Isaac
Newton’s famous Laws of Motion. His first law states that every
object in a state of uniform motion tends to remain in that state of
motion unless an external force is applied to it. Even though the elec-
trons in the synchrotron are traveling at a uniform (extremely high)
speed, their state of motion is not uniform; they are not traveling in
a straight line. Actually, they are bent by very powerful magnets. The
273
9780306815805_11.qxd 8/22/07 9:34 AM Page 274
third law states that for every action there is an equal and opposite
reaction. So what happens when the highly energetic electrons are
swerved? What’s the reaction? Well, electromagnetic radiation—lots
and lots of it—spun off the ring like tomatoes off the back of a truck
turning a corner at high speed.
To the particle physicists this synchrotron radiation was wasted
energy, an inconvenient by-product of the atom-smashing process. But
one day in the mid-1970s someone summoned up the courage to ask
the particle physicists if they could literally “tap” the ring and capture
the synchrotron radiation that it was emitting. For several years at
SPEAR, X-ray scientists, like parasites, harnessed the synchrotron radi-
ation provided by the ring that was primarily there to serve the atom
smashers. Eventually the high-energy physicists moved on to bigger
machines. Since 1990, SPEAR has been dedicated to the generation
of synchrotron radiation. A synchrotron X-ray beam is intense (there
are an awful lot of photons), collimated (all the photons point in the
same direction), and polarized (the electromagnetic field of all the
photons swings in a well-defined plane). In other words, you have a
colossal army of X-rays, all marching to the same drum, and the
experimenter can call the tune. The Stanford Synchrotron Radiation
Laboratory (SSRL) is one of the most advanced light sources in the
world. Today, more than fifty synchrotrons are operated around the
world and many more are under construction. They have names like
BESSY, Boomerang, Diamond, Soleil, SPring-8, and SPEAR3—the
newest upgrade of the ring at the Stanford Synchrotron Radiation
Laboratory.
A number of “beam lines” run off the synchrotron to little inde-
pendent labs. We were designated beam line 6–2. Many beam lines
have two hutches, so that while one experiment is being run,
another can be set up. There is no downtime for beams at SSRL,
because “beamtime” is a precious commodity. The hutches are
lead-lined and while the experiment is running, no one can enter
the hutch. It is not a good idea to be zapped by the beam. This is
274
9780306815805_11.qxd 8/22/07 9:34 AM Page 275
all you get when you get beamtime at SSRL—a beam line and a
hutch. While the EDAX Eagle probe is a commercial machine
with an awful lot of software attached, designed for a wide variety
of applications, the synchrotron is just a light source. Uwe had to
build the machine.
Unlike the EDAX machine, which sends out X-rays at many dif-
ferent wavelengths, Uwe could tune his beam precisely to the best
wavelength to look at iron or any other element. Uwe and Abigail
got Greg Young of the Canadian Conservation Institute to do
exhaustive tests on an old parchment document of Abigail’s to make
sure that his experiment would not damage the parchment. Having
conducted these tests, Uwe realized that he could indeed raise the
intensity of his beam at the wavelength that responded to iron. Uwe
attenuated his beam using especially designed filters. He fine-tuned it
with a filter of Reynolds Wrap, which he assured me was good
enough for the job. He designed his X-Y stage and carefully calcu-
lated the distance between the sample and the detector. He con-
structed a humidity chamber, so that the humidity would remain
constant and the folio of the Palimpsest would not change shape
while the scan was running. All the computers and workstations were
positioned outside of the hutch. Each of the several computers did
different things One computer recorded the position of the beam and
another recorded the position of the sample on the X-Y stage. If the
stage stopped moving, the experiment automatically shut down so as
not to damage the parchment. Another computer recorded the data
in the scan and a final computer was used to convert the files into a
format that we could use in the post-processing software and dis-
tribute to the scholars.
To help him, Uwe had Martin George write the software for the
computers, which had to meet two very different criteria: it had to
be advanced enough to precisely capture the data and it had to be
easy enough for Mike, Abigail, and me to use. The three of us would
have to take turns with Uwe running the scans, keeping an eye on
275
9780306815805_11.qxd 8/22/07 9:34 AM Page 276
the Palimpsest, and even fine-tuning the beam.The reason we had to

do this was that the experiment would run for seven days, twenty-
four hours a day, and in shifts.
It didn’t look like a professional set-up. Inside the hutch the guts
of the machinery were scattered all over the place and outside of
the hutch it looked like an electronics junk yard. I soon realized
that this is precisely what serious professional operations look like,
because looks don’t matter. Essentially, new machines are tailor-
made for very different operations every day in this extraordinary
place.
Uwe estimated that the scanning of one of the two columns of
folio 81r, the forged page, would take thirty hours—approximately
seventeen times faster than at the EDAX machine. Abigail put the
folio on to the X-Y stage and the scanning started. It was utterly
mesmerizing. Back and forth the scan would go, and slowly the
iron map would appear that brought out the Archimedes text. We
had to be constantly watchful for a fading signal, because the
position and strength of the beam changed over time. (The elec-
trons got topped up three times a day). If it did change, we had to
“nudge” the beam. We couldn’t scan large sections because the
resulting files would be too big. So Mike wrote down the novice’s
guide to the synchrotron and stuck it on a computer: “Press
STOP; Open hutch; Turn on light; Check humidity; check
ARCHIE; Turn off light; Close hutch; flick SAFETY switch; press
EXIT PLOTTER; Press EXIT RASTER; Check FILE save in
Dir/*.*; Select RASTER; press RETURN; Change XY coordi-
nates; press APPLY; flick SHUTTER 3 switch; select RASPLOT;
choose pixel 1; press START.”
Thirty hours later we had a column of text to show Reviel. And
Reviel read it. We had achieved our objective. We assured Uwe that
we would be back, and when we came we would bring the most
important pages of text containing the most difficult challenges.
276
9780306815805_11.qxd 8/22/07 9:34 AM Page 277
March 2006
We returned to SLAC for two weeks in March 2006. This time we
took more people with us.We needed more people to staff the beam,
but we also needed the talent. Uwe spent all of his time optimizing
the experiment that only he understood. Bob lent his years of expe-
rience on imaging with X-ray fluorescence. Keith and Roger
processed the images. Jennifer Giaccai, the conservation scientist at the
Walters, joined Abigail. And Mike and I lent a hand when we could.
This time we also brought out the very first page of the manu-
script.This was the page that Reviel and Natalie had first recognized
as On Floating Bodies in April 2001. It was the one that also contained
the inscription by the scribe of the prayer book with the date April
14, 1229. The page really was a wreck and the pseudocolor images
had revealed nothing.
From the moment the scanning started, it was clear that something
extraordinary was happening. The charred, stained, and worm-eaten
parchment in the hutch appeared on the screen as a dense lattice of
Greek characters. I knew that we were seeing, pixel by pixel, line by
line, at the Stanford syncrotron, a map of the iron on the page that
would give us the previously unknown Greek version of Archimedes’
On Floating Bodies. Keith Knox sent the first images to Reviel by
email. He received this reply:
From: Reviel Netz
Sent: Mon 3/13/2006 12:32 AM
To: Keith Knox
Cc: Nigel Wilson; Mike Toth; Uwe Bergmann; Roger Easton;
William Noel
Subject: folio 1v col.1
Thanks Keith for the images.
XRF for 1v col. 1 is sensational. I attach the transcription of
277
9780306815805_11.qxd 8/22/07 9:34 AM Page 278
1v col. 1 lines 2–11. Previously, with very hard work, I

squeezed some 3.5 lines out of the old pseudocolor, but now I
fairly easily read effectively the entire text, noting a couple of
errors, too, in my old pseudocolor based reading.
Reviel
This leaf comes from the very long final proposition of On Floating
Bodies, which is, by common consent the most complex proposition
ever written by Archimedes. It concerns the conditions under which
a conic section, somewhat similar in appearance to the hull of a boat,
will or will not be stable when immersed in water. It is significantly
different from the Latin text that Heiberg derived from Moerbeke
and a group of diagrams appear where none were suspected. This
unique surviving Greek text of Archimedes’ On Floating Bodies was
finally revealed on March 13, 2006—777 years after it had been
erased and overwritten.
The Gift-Giver
The colophon from which John Lowden had so painstakingly
extracted the date on which the scribe of the prayer book had dated
his work as April 14, 1229, was also on this same page. I sent out the
following email:
All: We attach two images of the lower section of fol. 1v.

“Before” was taken when the Palimpsest arrived at WAM.
“After” was taken today: it’s an X-ray fluorescence image
taken at SLAC; after Abigail Quandt conserved it. It contains
some text written on April 14, 1229. But can anyone now
give us any more details?
My friend Georgi Parpulov was the first to reply:

From: Georgi Parpulov
Sent: Tue 3/14/2006 4:39 AM
278
9780306815805_11.qxd 8/22/07 9:34 AM Page 279
To: William Noel

Subject: colophon
Hi, Will,
+[This] was written by the hand of presbyter John Pogonatos
(?) on the 14th day of the month of April, a Saturday, of the
year 6737, indiction 2.
Wait till you hear from Nigel Wilson: he will be able to read it
with much greater precision.
Before we heard from Nigel, we heard from John Lowden:
From: Lowden, John

Sent: Thu 3/16/2006 10:31 AM
To: William Noel; Georgi Parpulov; Nigel Wilson, Reviel
Netz
Subject: RE: colophon
Have just received this on return from Dublin. The improved
legibility is astonishing.
First impression is that the name (refining GP) is Iw(annou)
iere(os) tou Murwna
I would check Ioannes Myronas iereus as scribe.
But maybe I should check first and communicate after!
Yours (too hastily?)
John
And then finally, on the Sunday, we got confirmation from Nigel:
From: Nigel Wilson

Sent: Sun 3/19/2006 7:32 AM
To: William Noel; Georgi Parpulov; John Lowden; Reviel
Netz
Subject: Re: colophon
279
9780306815805_11.qxd 8/22/07 9:34 AM Page 280
Dear Will, John, Reviel et al.,

I agree with John’s suggestion and that Myronas is probably
the name; the last letter could be alpha and has an accent. I
have asked a Greek student of mine to check in the telephone
directories to see if it is still a name in Greece. (Mylonas is
well attested.)
Best wishes,
Nigel
There we are then.The case is nearly closed. Finally we know who
preserved the texts of Archimedes, Hyperides, and the rest.The priest
Ioannes (John) Myronas finished his work on April 14, 1229. In that
year, April 14 was the day before Easter Sunday.Traditionally, it was a
day upon which people made gifts to religious institutions for the
good of their souls. What an extraordinary gift this was. Ioannes did
not just redeem himself. On the anniversary of Christ’s Resurrection,
Ioannes Myronas gave the world its greatest palimpsest and saved the
secrets of Archimedes.
280
9780306815805_RM.qxd 8/22/07 9:36 AM Page 281
EPILOGUE
“The Vast Book of the Universe”

Moonlighting for Archimedes
his is not typical university work. The Archimedes Palimpsest
T seems to insist on being unique and the project of its deci-
pherment has few parallels. Remarkably much has been achieved in
less than ten years. Remarkably, too, nearly all of this was achieved by
weekend warriors, summoned by the thrill and glory of working for
Archimedes. We all had our day jobs. Will Noel was curating manu-
script exhibitions at the Walters, I was teaching Greek science at
Stanford, while Roger Easton was teaching imaging science at
Rochester, and Nigel Wilson was editing the works of Aristophanes
for the Oxford Classical Texts series. I am not sure what Mike Toth
was doing. In fact, only one individual got the Archimedes Palimpsest
as a day job—and this underlines the priorities of manuscript studies.
Abigail Quandt put aside most of her other obligations to concen-
trate, day after day, on the disbinding and conservation of the manu-
script. Her hands were the busiest.
And all of us did this for one simple reason: we are in awe of one
individual who lived, some 2,250 years ago, on a triangular island in
the middle of the Mediterranean. That we succeeded in doing so
much for him is, in my view, due to three individuals. They deserve
our thanks.
281
9780306815805_RM.qxd 8/22/07 9:36 AM Page 282
The Patron
That so much has been accomplished so quickly is primarily a tribute
to the owner of the Palimpsest. The team of scholars and scientists
working on the project have something which is, today, very rare: our
work is led by a rich patron.There was a time when this was standard.
Science in Alexandria—as well as in Syracuse—was pursued under
the patronage of Hellenistic kings. No doubt a rich patron commis-
sioned the Archimedes manuscript in the tenth century. Most
Renaissance artists and scholars worked for rich patrons. However,
since at least the Middle Ages, scholarship has often been pursued
within public institutions. The Church is the most obvious example,
and it is due to her that most manuscripts survive. Today most man-
uscripts are held by a different kind of public institution—the state or
its universities. Almost all the important manuscripts of the world can
now be found in such institutions and, at first, everyone’s feeling was
that this manuscript should belong to the public. We were proved
wrong. In retrospect, it was a stroke of luck that the manuscript found
itself in private hands. No public institution could have acted so
flexibly, with such generous and well-thought application of
resources. Think of it; the owner did something rather outrageous.
He entrusted it to Will Noel, who is by now a world expert on the
Archimedes Palimpsest but eight years ago could not tell Archimedes
from Pythagoras. The owner then more or less told Will Noel to do
with the manuscript as he pleased.The owner implicitly promised to
pay along the way as necessary. (I say “implicitly” because, I am told,
the owner is not a man of many words.) It turns out this was the
clever thing to do. Had the manuscript been housed at a university,
the academic politics of its research would have been much more dif-
ficult and each expense along the way would have had to be
accounted for in a much more tedious, haphazard, and time-con-
suming fashion. In short, the owner saved us the disadvantages of
public institutions. Not that private owners are to be preferred in
282
9780306815805_RM.qxd 8/22/07 9:36 AM Page 283
“THE VA S T B O O K O F T H E U N I V E R S E ”
principle. In my own view, private owners are, in general, not the best
custodians of world treasures. After all, it was the Greek Church that
saved the manuscript for a millennium and then it was private owners
that, through the course of the twentieth century, nearly destroyed it.
With the current owner we have been lucky. He has not merely done
well by Archimedes; he did all that he could.
The Philologist
I do not know the owner of the Palimpsest all that well, however Will
Noel does. Almost daily, he is in email correspondence with him dis-
cussing the way to move forward with the Palimpsest project. My
own daily correspondence is even more virtual. It is all in my head.
And it is with another great benefactor of this project without whom
all of this would have been impossible. In my thoughts, I always con-
verse with Johan Ludwig Heiberg.
Right down to the last pages, we have been critical toward him
throughout this book—the gaps he left, the false guesses he made, the
diagrams with which he never bothered. Now is the time to admit
the truth. Without Heiberg, we could have never made it. We would
look at the text and see at first just a jumble of meaningless traces.We
would interpret a few of them. We would conjecture a sense. We
would end up at a dead end.Then we would check Heiberg and, low
and behold, he had already made sense of it. He had even read
further! Only then, looking back at the page, do we see those traces
that provided Heiberg with his reading. And then, finally, based on
Heiberg’s foundations, we can go further and add to his readings.
The transcription project was rather like an expedition to a lost
island.You believe you face what no one has ever witnessed. And then,
time and again, you have the same uncanny experience.You suddenly
realize that the previous explorer—Heiberg—had already been there. I
was excited when I saw symbols for circles emerging out of Abigail’s
treatment of the Palimpsest. I took it for granted that they were new.
283
9780306815805_RM.qxd 8/22/07 9:36 AM Page 284
But no, Heiberg had seen them too and had noted them in his critical
edition. Again and again, Heiberg took me by surprise.
Let us put it this way. There is a long tradition of readers of
Archimedes, from scholars such as Hero of Alexandria, Eutocius of
Ascalon, and Leo the Geometer from Byzantium, to the present day.
No one in this tradition will ever rival the authority of Heiberg. We
are extraordinarily lucky that he, and no one else, was in Istanbul in
1906 and studied this manuscript for that brief historical moment.
No one else would have made so much out of the manuscript.
Heiberg almost single-handedly saved the text of Archimedes. It is
only thanks to the most modern technology that we can now go
beyond his reading. And for this, I suggest, our gratitude should go to
Archimedes himself.
The Founder’s Tools

I started this book by saying that Archimedes was the most important
scientist who ever lived. We can now see how: in the tools he
created—and in the way in which later science was shaped by
Archimedes’ blueprint for science. Archimedes, more than anyone
else, shaped the history of the calculus—the essential study for the
measurement of curves. And he also, incredibly, was the founder of
combinatorics—the science underlying our own theory of proba-
bility. These two—the calculus and the theory of probability—
underlie contemporary imaging science. The imaging scientists
working on the Archimedes Palimpsest applied a science that was
fundamentally Archimedean.
To illustrate this, I now concentrate on a relatively standard tool
used by imaging scientists—the equalization of probability curves.
While standard, it is a useful introduction, because it brings into focus
the main concept of imaging science, namely that of information.
We have often mentioned the term “information” in this book.
We have discussed the way that books move from one information
storage system to another and how changes in information tech-
284
9780306815805_RM.qxd 8/22/07 9:36 AM Page 285
nologies impact the transmission of knowledge. It is time to come

clean and explain that “information” is not some kind of vague,
metaphorical concept. Information is a technical term possessing a
clear, subtle mathematical definition. Most importantly, in contem-
porary science, information can be measured.
The fundamental intuition can be articulated as follows. We look
at an array of numbers and we ask, “how predictable is it?” Suppose
the array of numbers is as follows:
255, 255, 255 . . . 255
All of the numbers are exactly 255. There is a clear intuition that
this is very predictable and so it is also very uninformative. On the
other hand, an array that is as follows:
127, 45, 254, 11, 6, 189 . . . 39
is much less predictable and contains more “information.”
Now, let us review the arrays of numbers:
255, 255, 255 . . . 255 127, 45, 254, 11, 6, 189 . . . 39
What do they mean in terms of image science? As Will Noel has
already explained, imaging scientists do not think of images as faces
or flowers. They think of images as two-dimensional (sometimes
many-dimensional) arrays of integer values. Each of the integers in an
array represents the properties of a pixel. Typically, they consider the
grey level of a pixel in a black-and-white picture. In a black-and-
white picture each pixel is assigned a level of grey that usually ranges
from 0 (least light—that is, black) to 255 (most light—that is, white).
The 255, 255, 255 . . . 255 array, therefore, is correlated with a per-
fectly blank image—one that is totally white.The 127, 45, 254, 11, 6,
189 . . . 39 array, on the other hand, is associated with a complex
pattern of light and shade. The perfectly blank image is completely
predictable (totally white), therefore totally uninformative. The
complex pattern of light and shade is much less predictable and
therefore also much more informative.
At this point, it can be mathematically shown that: the most
informative image is the one in which all the levels of grey are
equally probable.
285
9780306815805_RM.qxd 8/22/07 9:36 AM Page 286
In the perfectly blank image, a single level of grey was the most
probable—the totally white one. All of the rest had no probability of
occurring. In the complex pattern of light and shade, however, all
levels of grey were equally probable. And this is the underlying math-
ematical reason why the complex pattern of light and grey is the
more informative.
The goal of imaging science is to make images as informative as
possible so that, for instance, scholars can use them to read the words
of Archimedes. The mathematical result implicitly suggests a certain
possible technological application. In order to make an image more
informative, let us equalize its distribution of probabilities.We should
try to make all the levels of grey equally probable.
How do we do this? We need a new mathematical conceptual-
ization.We need to return to the image and consider it not merely as
an array of numbers but, instead, as a curve. We draw a two-dimen-
sional matrix with the familiar x and y-axes; x horizontally and y ver-
tically.We make the x-axis stand for the possible grey levels—from 0,
the perfect black, to 255, the perfect white. For each of the 256 levels
of grey, we draw on the y-vector each times it occurs. So, for instance,
the perfectly blank image has a very simple appearance in such a
matrix; it is empty everywhere except for a single, tall column
standing at 255 at the rightmost end of the x-axis (see fig. 12.1). A
complex pattern of light and shade, on the other hand, has a more
complex appearance (see fig. 12.2).
To simplify it a bit: most images appear as a kind of bell-curve,
with most pixels somewhere in the middle between black and white
and the rest somewhat less probable (more black or white) as we
move away from the center.
Now recall our aim—to make the image as informative as pos-
sible—is to make the distribution of probabilities as equal as possible.
That is to say since the most informative image appears as a “flat”
curve or rather like a rectangle—one where all levels of grey occur
equally often (see fig. 12.3)—we wish to take a curve such as the bell-
curve in the figure and turn it into a flat, rectangular shape.
286
9780306815805_RM.qxd 8/22/07 9:36 AM Page 287
0 32 64 96 128 160 192 224 255
FIGURE 12.1 A distribution of pixels where all pixels have 255 level of light corre-
sponds to the perfectly blank image.
0 32 64 96 128 160 192 224 255
FIGURE 12.2 A normal image has associated with it a curve that has a roughly bell-
like shape.
Now Archimedean science kicks in again. For this operation, the

transformation of a bell-curve into a rectangle, is simply the meas-
urement of a curvilinear object by a rectilinear one. When
Archimedes was measuring his parabolas, showing how they were
equal to two-thirds a given rectangle, he was doing precisely the kind
of operation we need to do right now. (Indeed, some of the curves
we need to measure may take the form of a parabola.) And what con-
temporary scientists do, at this stage, is apply the tools of the cal-
culus—i.e. of the science arising out of Archimedes’ measurement of
curvilinear objects—in order to “flatten” the curve correctly.
287
9780306815805_RM.qxd 8/22/07 9:36 AM Page 288
0 32 64 96 128 160 192 224 255
FIGURE 12.3
We applied the theory of probability to develop the notion of

“information” and to discover that the most informative image is the
one associated with an equal distribution of probabilities of grey
levels. We then applied the calculus to transform the curve of the
actual distribution of probabilities into the desired “flat” shape asso-
ciated with the equal distribution of probabilities.The end result was
no more and no less than the simultaneous application of probability
and the calculus. This is just an example of what imaging scientists
do, but also a representative one (if you wish, an informative one)—
though, of course, to develop the images used for reading the
Palimpsest, much more than this was required. But this is precisely the
type of mathematical technique applied. Probability and the calculus
are what imaging science is made of. And so, without Archimedes, we
would not have the science to read him.
Archimedes’ Blueprint for Science

Even more important than the contents of Archimedes’ science was
his spirit—his blueprint for science. After all, his combinatorics were
lost when Ioannes Myronas decided not to use all of the parchment
in front of him and to only use a single leaf of the Stomachion. It
remained for seventeenth-century mathematicians to reinvent the
science of combinatorics and therefore to create probability. They
288
9780306815805_RM.qxd 8/22/07 9:36 AM Page 289
would have done so even without Archimedes having written the

Stomachion. And yet, without Archimedes’ example, I doubt that we
would have the kind of science we have today.We can see how it goes
back to Archimedes’ invention of applying mathematical, abstract
models to the physical world.
Let us take as our starting point, once again, the mathematical
notion of information. Imaging science and the computer sciences, as
well as many other disciplines of the digital revolution, are all essen-
tially based on this concept. As mathematical concepts go, this is a
very recent one. A mathematician working at Bell Laboratories
named Claude Shannon introduced it in the year 1948. He was
working on making telephone lines function better, and it occurred
to him that there could be a mathematical theory associated with the
amount of information traveling through such lines. His inspiration
came directly out of physics, and the concept of information, as
found by Shannon, is inherently a concept of mathematical physics.
What Shannon did was to take the concept of entropy, defined in
mathematical physics, and to apply it to the flow of information in,
for instance, telephone lines. What is entropy? It is a measure of how
“probable” a given physical state is. A physical state can be very likely,
in which case its entropy is high, or it can be very unlikely, in which
case its entropy is low. We can therefore see where Shannon took his
inspiration: information is (to simplify a bit) reverse entropy.
One of the deepest observations ever made by science is that—
hold your breath—probable things happen more often. So that, more
often than not, physical systems move from unlikely states—with low
entropy—to more likely states—with high entropy.Wait long enough
and the amount of entropy in the universe must increase (or, as we may put
this after Shannon, the amount of information in the universe must decrease,
which is also the reason why reception on our cell phones is so bad).
I am completely serious about saying that this is one of the deepest
observations ever made by science. It is a beautiful example of how,
by the power of pure thought, we can work out how the universe
must behave. It is a tautology, that probable things happen more often;
289
9780306815805_RM.qxd 8/22/07 9:36 AM Page 290
and because of this tautology, which we came to by pure thought

alone, we can see also that the amount of entropy in the universe must
increase. This is known as the second law of thermodynamics, and it
ranks as one of the most fundamental discoveries of physics.
This becomes especially significant the moment we can calculate
which physical systems have more entropy and which have less. This
was the reason the concept of entropy was introduced in the first
place in 1872 by a German physicist named Boltzmann. He produced
a mathematical approach for measuring the amount of entropy in a
physical state. Suppose we take as our physical system a certain gas
composed of many gas molecules. Then Boltzmann could show that
the more rapidly, on average, the gas molecules were moving the less
entropy the system had; or that the slower, on average, the gas mole-
cules were moving the more entropy the system had. In short,
Boltzmann could show that higher entropy is associated with the
slower motion of molecules. Based on the second law of thermody-
namics—showing that entropy must increase—Boltzmann could also
show that gases must eventually move from faster states to slower
states.
Now, it is also established that what we call “heat” is really a meas-
urement of the speed of molecules in the physical system. A “warm”
system is really one where the molecules move faster, a “cool” one is
where they move slower. And so Boltzmann shows, based on the
second law of thermodynamics, that all systems must, eventually,
became colder.
This is magic—belonging right up there in the pantheon next to
Archimedes. Applying pure thought alone, Boltzmann proved in
1872 that everything must, eventually, become colder. But the comparison
with Archimedes can be sustained further.
Why did Boltzmann produce his mathematical theory in 1872?
The behavior of heat was the urgent scientific problem of his time.
So much else in science was already understood in mathematical
terms but not heat. In the two centuries prior to Boltzmann, scien-
tists were working hard on extending Newton’s achievement.
290
9780306815805_RM.qxd 8/22/07 9:36 AM Page 291
Newton determined, on purely mathematical grounds, how the

planets must behave. The universe was made of points—centers of
gravity—exerting the force of gravity on each other. This was a
unified theory of motion, where everything was reduced to basic
tools of geometry and calculation.The theory was published in 1687
in Newton’s Principia. And from 1687 onward, scientists tried to
emulate Newton’s achievement—to produce mathematical theories
to which one could reduce various physical phenomena. In the early
nineteenth century electricity followed gravity, analyzed by mathe-
matical techniques somewhat comparable to those of Newton. By
1872 the central physical phenomenon still resisting mathematical
treatment was heat. Boltzmann, in his study, made a fundamental con-
tribution to the mathematization of physical science. He was essen-
tially completing Newton’s program.
Only it wasn’t Newton’s—it was Archimedes’ as Newton would be
the first to admit. Newton, in 1687, was himself heir to a long tra-
dition. His great predecessor was Galileo. Both Newton and Galileo
aspired, above all, to return science to its Archimedean heights. They
wished to take Archimedes’ mathematical tools and to make such
tools deduce as much of physics as one could. The Newtonian
program of reducing physical systems to geometrical representations
obeying mathematical laws was taken from the Archimedean blue-
print for science.And so, without Archimedes, there would be neither
Galileo nor Newton. And for that matter, there would be no
Boltzmann or Shannon either. Nor, for that matter, contemporary
imaging science.
The “Vast Book”

“Philosophy is written in this vast book, which lies continuously
open before our eyes (I mean the universe). But it cannot be under-
stood unless you have first learned to understand the language and
recognize the characters in which it is written. It is written in the lan-
guage of mathematics, and the characters are triangles, circles, and
291
9780306815805_RM.qxd 8/22/07 9:36 AM Page 292
other geometrical figures. Without such means it is impossible for us

humans to understand a word of it, and to be without them is to
wander around in vain through a dark labyrinth.”
So Galileo wrote, in 1623, recovering the spirit of the science of
Archimedes. This metaphor of the vast book of the universe is still
with us. We do think of the universe as a “book” whose secrets we
try to uncover, and we still use mathematics to do so.The importance
of Archimedes to the history of science is his having shown how this
metaphor could literally work. The book of the universe was first
deciphered by him—and it was found to be written in the language
of mathematics.
In 1623, when Galileo came to write down these words, all the
manuscripts of Archimedes were already gone. Codex B was lost
some time in the fourteenth century; Codex A was lost some time in
the sixteenth century, probably when Galileo was still a child. Only
one copy remained but it was hidden from sight. The monks who
used it never did learn to read its triangles, circles, and other geo-
metrical figures.
In 1687, when Isaac Newton’s Principia was published, this
codex—the Archimedes Palimpsest—was still in the Holy Land, still
hidden from sight. It was worlds away from the British scientist clois-
tered in his lodgings at Trinity College, Cambridge.
In 1872, when Ludwig Boltzmann published his study on the
second law of thermodynamics the Archimedes Palimpsest was
already in Istanbul. It was about to be discovered, however briefly, by
Heiberg. By 1948 it was lost again. When Shannon produced his
mathematical definition of information, the manuscript was probably
already lying mutilated in a Paris apartment.
Fifty years later it walked on stage.And now, science was ready.The
science inspired by Archimedes came full circle to be able to recover
nearly all of his words. And now, finally, we get a sense of the full
measure of the man.
A final word of caution should be added. We have not gotten
everything. There are still gaps in our reading. But we remain opti-
292
9780306815805_RM.qxd 8/22/07 9:36 AM Page 293
mistic. Even now, as I write, I am studying the latest SLAC images of

the forgery pages, and I can see that Heiberg’s transcription of Method
proposition 1 needs significant revision. The science inspired by
Archimedes never stands still. The process extends without end:
science steps back from physical reality to consider its mathematical
underpinnings and, this way, more and more is always found.
Archimedean science keeps making progress, and, in time, it will
catch up with Archimedes.
293
9780306815805_RM.qxd 8/22/07 9:36 AM Page 294

9780306815805_RM.qxd 8/22/07 9:36 AM Page 295
Acknowledgements

eading the Archimedes Palimpsest was a far more complicated
R undertaking than the narrative encompassed in these pages
might indicate. Indeed, we do not know all the people who helped.
Choosing just a few may alienate many, but the contributions of
some have been so substantial that we cannot conclude this book
without listing them. Knowing that this list is incomplete, we
nonetheless extend our thanks to all who have so generously con-
tributed to the project. A lot of the work was done at night, on
weekends, and during vacations, and we have also to thank the
project’s many widows and orphans, particularly Carol Christens-
Barry, Dale Stewart, Daniel and Donald Potter, Elisabetta Gaiani and
Sofia Bergmann, Hanneke Wilson, and Lucretia Toth. Deep gratitude
is owed to Uwe Bergmann, Serafina Cuomo, Patricia Easterling,
Roger Easton, Jr, László Horváth, Geoffrey Lloyd, Abigail Quandt,
Ken Saito, and Nigel Wilson for their expert assistance in the writing
of this book. All mistakes of fact and interpretation are our own.
Many friends helped us make ourselves clearer, including Richard
Ash, Christopher Collison, Charlie Duff, Susan Elderkin, Guy
Deutscher, Kathryn Gerry, Richard Leson, Amanda Mann, Audrey
Scanlon-Teller, and Jean-François Vilain. Success depends upon good
editors, and we had the wonderful Francine Brody at Weidenfeld and
Nicolson, the sagacious Robert Pigeon at Da Capo, and the heroic
William McLean at the Walters.
295
9780306815805_RM.qxd 8/22/07 9:36 AM Page 296
MANAGEMENT AND SCIENCE AND IMAGING

ADMINISTRATION Allyson Aranda
Ken Dean Mike Attas
Barbara Fegley Uwe Bergmann
Kirstin Lavin Bill Christens-Barry
Richard Leson David Day
Griffith Mann Charles Dickinson
Amy Mannarino Roger Easton, Jr
Joan Elisabeth Reid Alex Garchtchenko
Harold Stevens Martin George
Mike Toth Jason Gislason
Gary Vikan Douglas Golz
Lynn Wolfe Gene Hall
Tom Hostetler
CONSERVATION AND Keith Knox
HANDLING Matthew Latimer
Kevin Auer Bob Morton
George Chang Nick Morton
Jane Down Tara Nylese
Gil Furoy Emanuele Salerno
Jennifer Giaccai Bruce Scruggs
Paul Hepworth Derek Walvoord
Erin Loftus
Amy Lubick DATA AND INFORMATION
Maureen McDonald TECHNOLOGY
Mike McKee Martina Bagnoli
Elizabeth Moffatt Diane Bockrath
Elissa O’Loughlin Doug Emery
Abigail Quandt Cathleen Fleck
Jane Sirois Andy Johnston
Scott Williams Joe McCourt
Gregory Young Carl Malamud
Anthea Zeltzman
296
9780306815805_RM.qxd 8/22/07 9:36 AM Page 297
AC K N OW L E D G E M E N T S
SCHOLARSHIP Stephen Menn

Fabio Acerbi Tamás Mészáros
Colin Austin Stefano Parenti
Chris Carey Georgi Parpulov
Persi Diaconis Erik Petersen
Patricia Easterling Marwan Rashed
Mike Edwards Peter Rhodes
Zoltán Farkas Ken Saito
Eric Handley Robert Sharples
Jud Herrman Richard Sorabji
Susan Holmes Natalie Tchernetska
László Horváth Stephen Todd
John Lowden Nigel Wilson
Gyula Mayer David Whitehead
Henry Mendell
297
9780306815805_RM.qxd 8/22/07 9:36 AM Page 298

9780306815805_RM.qxd 8/22/07 9:36 AM Page 299
Further Reading

Those interested in finding out more about the Archimedes Palimpsest,
its imaging, conservation, and scholarly study, should visit the website
www.archimedespalimpsest.org, and follow the links. It is our hope to
present all our data on the web at www.archimedespalimpsest.net, and
we have already made a start at this. Other than this, readers might like
to consult the following publications.
ENCYCLOPEDIAS
Gillispie, C. C. (ed.), Dictionary of Scientific Biography (New York, 1975).
Hornblower, S. and A. Spawforth (eds.), Oxford Classical Dictionary
(Oxford, 1996).
The Catholic Encyclopedia, at www.newadvent.org.
ANCIENT MATHEMATICS
Those interested in learning more about ancient Greek science
would best start with these very readable books:
Lloyd, G. E. R., Early Greek Science:Thales to Aristotle (London, 1970).
———, Greek Science after Aristotle (London, 1973).
Those interested more specifically in the achievements of Greek

geometry should start with:
Knorr, W. R., The Ancient Tradition of Geometric Problems (New York,
1986).
299
9780306815805_RM.qxd 8/22/07 9:36 AM Page 300
ARCHIMEDES
The best general book on Archimedes’ scientific achievement is likely
to remain for many years to come:
Dijksterhuis, E. J., Archimedes (1956; revised edn, Princeton, 1987).
The following is a three-volume publication, written in Greek, with

Latin translation and introduction. It may be difficult to read. Still, we
thought we should mention this; we have referred to its two authors
quite frequently:
Heiberg, J. L., Archimedes, Opera Omnia (Leipzig, 1910–15).
Those interested in the early history of the calculus and its concepts
should still read:
Boyer, C. B., The History of the Calculus and its Conceptual Development
(New York, 1959).
HYPERIDES
The speeches of Hyperides known before their discovery in the
Archimedes Palimpsest are edited with a translation in the Loeb
Classical Library:
Burtt, J. O., Minor Attic Orators, vol. II (Cambridge, MA, 1954).
MANUSCRIPT TRANSMISSION OF THE CLASSICS

For the transition from roll to codex:
Roberts, C. H. and T. C. Skeat, The Birth of the Codex (London, 1983).
For those interested in Greek scripts, the following are good intro-
ductions:
Barbour, R., Greek Literary Hands AD 400—1600 (Oxford, 1981).
Easterling, P. and C. Handley (eds.), Greek Scripts: An Illustrated
Introduction (London, 2001).
Metzger, B. M., Manuscripts of the Greek Bible: An Introduction to Greek
Paleography (New York, 1981).
300
9780306815805_RM.qxd 8/22/07 9:36 AM Page 301
F U RT H E R R E A D I N G
*
For a broad survey of the history of writing, readers might try:
Sirat, C., Writing as Handwork:A History of Handwriting in Mediterranean
and Western Culture (Turnhout, 2006).
There are many technical studies on the making of manuscripts. A
useful basic text, with bibliography, is:
Brown, M. P., Understanding Medieval Manuscripts: A Guide to Technical
Terms (Malibu, CA, 1994).
Not at all relevant to Archimedes, but for readers who would like to
know more about the wonderful world of medieval manuscripts, the
best general introduction available is:
De Hamel, C., A History of Illuminated Manuscripts (London, 1987).
For those interested in the transmission of ancient texts through to

the age of printing, the following are indispensable:
Reynolds, L. D. and N. G. Wilson, Scribes and Scholars, 3rd edn
(Oxford, 1991).
Wilson, N. G., Scholars of Byzantium (London, 1983).
A major scholarly feat, focused on the history of the text of

Archimedes in Latin-speaking Europe, is:
Clagett, M., Archimedes in the Middle Ages (Madison, WI, 1964–84).
IMAGING AND IMAGING PROCESSING

Roger Easton recommends:
Baxes, G. A., Digital Image Processing: Principles and Applications (New
York, 1994).
Falk, D. R., D. R. Brill and D. G. Stork, Seeing the Light: Optics in
Nature, Photography, Color,Vision, and Holography (New York, 1986).
For advanced light sources, such as the Stanford Linear Accelerator

Center, check out http://www.lightsources.org.
301
9780306815805_RM.qxd 8/22/07 9:36 AM Page 302
THE PALIMPSEST
The main publications on the Archimedes Palimpsest since
September 1998 are listed below, alphabetically by author:
Christens-Barry, W. A., J. R. Bernstein and M. Blackburn, “Imaging
the Third Dimension of the Archimedes Palimpsest,” Proceedings of
IS & T PICS Conference (Montreal, 2001), pp. 202–5.
Christie’s, New York, “The Archimedes Palimpsest,” sale catalogue
9058, Thursday, 29 October 1998.
Down, J. L., G. S. Young, R. S. Williams and M. A. MacDonald,
“Analysis of the Archimedes Palimpsest,” in V. Daniels, A.
Donnithorne and P. Smith (eds.), Works of Art on Paper, Books,
Documents and Photographs, The International Institute for
Conservation, Contributions to the Baltimore Congress, 2–6
September 2002 (London, 2002), pp. 52–8.
Easton, R. L., Jr, and W. Noel, “The Multispectral Imaging of the
Archimedes Palimpsest,” Gazette du Livre Médiévale, 45, 2004, pp.
39–49.
Handley, E., “Eureka? The conservation, imaging and study of the
Archimedes Palimpsest,” exhibition pamphlet, Trinity College,
Cambridge, 21–2 and 25–9 July 2005.
Knox, K., C. Dickinson, L. Wei, R. L. Easton, Jr, and R. Johnston,
“Multispectral Imaging of the Archimedes Palimpsest,” Proceedings
of IS & T PICS Conference (Montreal, 2001), pp. 206–10.
Lowden, J., “Archimedes into Icon: Forging an Image of
Byzantium,” in A. Eastmond and L. James (eds.), Icon and Word:The
Power of Images in Byzantium (London, 2003), pp. 233–60.
Netz, R., Archimedes:Translation and Commentary, with a Critical Edition
of the Diagrams and a Translation of Eutocius’ Commentaries, vol. I:
“The Sphere and the Cylinder” (Cambridge, 2004).
———, Archimedes:Translation and Commentary, with a Critical Edition
of the Diagrams and a Translation of Eutocius’ Commentaries, vol. II:
“Advanced Geometrical Works” (Cambridge [forthcoming]).
302
9780306815805_RM.qxd 8/22/07 9:36 AM Page 303
F U RT H E R R E A D I N G
———, Archimedes:Translation and Commentary, with a Critical Edition

of the Diagrams and a Translation of Eutocius’ Commentaries, vol. III:
“The Mathematical-Physical Works” (Cambridge [forthcoming]).
———, “Archimedes and Mar Saba: a Preliminary Notice,” in J.
Patrich (ed.), The Sabaite Heritage:The Sabaite Factor in the Orthodox
Church: Monastic Life, Liturgy,Theology, Literature, Art and Archaeology
(2002), pp. 195–9.
———, “The Origin of Mathematical Physics: New Light on an Old
Question,” Physics Today , June 2000, pp. 31–6.
Netz, R., F. Acerbi and N. Wilson, “Towards a Reconstruction of
Archimedes’ Stomachion,” Sciamus, 5, 2004, pp. 67–99.
Netz, R., K. Saito and N. Tchernetska, “A New Reading of Method
Proposition 14: Preliminary Evidence from the Archimedes
Palimpsest (Part 1),” Sciamus, 2, 2001, pp. 9–29.
———, “A New Reading of Method Proposition 14: Preliminary
Evidence from the Archimedes Palimpsest (Part 2),” Sciamus, 3,
2002, pp. 109–25.
Noel, W., “The Archimedes Palimpsest, Old Science Meets New
Science,” Proceedings of IS & T PICS Conference (Montreal, 2001),
pp. 199–201.
Parenti, S., “The Liturgical Tradition of the Euchologion ‘of
Archimedes’” Bollettino della Badia Greca di Grottaferrata, IIIs. 2
(2005) [but actually 2006], pp. 69–87.
Quandt, A., “The Archimedes Palimpsest: Conservation Treatment,
Digital Imaging and Transcription of a Rare Mediaeval
Manuscript,” in V. Daniels, A. Donnithorne and P. Smith (eds.),
Works of Art on Paper: Books, Documents and Photographs, The
International Institute for Conservation, Contributions to the
Baltimore Congress, 2–6 September 2002 (London, 2002), pp.
165–70.
Tchernetska, N., “New Fragments of Hyperides from the
Archimedes Palimpsest,” Zeitschrift für Papyrologie und Epigraphik,
vol. 154, 2005, pp. 1–6.
303
9780306815805_RM.qxd 8/22/07 9:36 AM Page 304
Wilson, Nigel, “Archimedes: the Palimpsest and the Tradition,”

Byzantinische Zeitschrift, 92, 1999, pp. 89–101.
———, “The Archimedes Palimpsest: A Progress Report,” in “A
Catalogue of Greek Manuscripts at the Walters Art Museum and
Essays in Honor of Gary Vikan,” Journal of the Walters Art Museum,
62, 2004, pp. 61–8.
———, “The Secrets of Palimpsests,” L’Erasmo, 25, 2005, pp. 70–5.
———, Archimedes’ “On Floating Bodies” I.1–2, edited with an
English translation (Oxford, 2004).
Young, G., “Quantitative Image Analysis in Microscopical Thermal
Stability Measurements,” Canadian Conservation Institute
Newsletter, 31 June 2003, pp. 10–11.
304
9780306815805_RM.qxd 8/22/07 9:36 AM Page 305
Index

abbreviations 111–2, 113 “Archimedes’ Box” see Stomachion
Acerbi, Fabio 247, 248, 250, 260 Archytas 39
Alberti, Leon Battista 120 Aristotle 27, 119, 227, 229, 230
Alexander the Great 68, 227, 228 Ars Eudoxi 109–10
Alexandria 39, 40, 67–70, 73, 74, 282 Athens 133, 227, 228–9
Ammonius 73 atoms 269–70
Anthemius of Tralles 76 Attas, Mike 264
approximation of curves 27
Arabic manuscripts 236–7 Balance, Law of the 27, 53, 148–9,
Archimedes 1–2, 4, 27–63, 65–6 155–6
centers of gravity 140–1, 142, Balancing Planes 53–4, 80, 85
144–5 centers of gravity 140
Constantinople 76–7 Eutocius’ commentary 73
diagrams 94–8, 99–101, 105–9 Leonardo da Vinci 121
infinity 51–2, 184, 185, 199–203 Basil I 77–8, 81
influence on science 26–8, 279, Bergmann, Uwe 264, 273, 277–8
283–6, 287–8 Bernstein, Joanna 211
Law of the Balance 53, 148–9 Bibliotheca 78
Leonardo da Vinci 121–2 Bibliothèque Nationale 133, 167, 169
Method 66–70, 85, 139–40, 157, Blass, Bill 264
183–7, 199–201 Bohr, Niels 269–70
minuscule manuscripts 78–80 Boltzmann, Ludwig 290, 291, 292
parabolas 150–3 Brigham Young University 208
Renaissance Italy 119–20 Brunelleschi, Filippo 120
Stomachion 233–60 Buonarotti, Michelangelo 120, 159
survival of manuscripts 72–4, Byzantine texts 77–8, 85, 171, 226,
80–1, 85–6 230
tomb of 62–3, 76, 89, 92, 105–7
volume of a cylindrical cut calculus 1, 29, 41, 52, 139, 284
188–99, 201–2 imaging science 284, 286
305
9780306815805_RM.qxd 8/22/07 9:36 AM Page 306
Newton 27, 52, 54, 157 Italian Renaissance 119–20

principles of summation 202 loss of 292
Callimachus of Cyrene 69, 72 Codex B 2, 79, 80
Cambridge University Library 22, Heiberg’s edition 131
129 Italian Renaissance 119–20
cameras 208, 209, 211–2, 221–2, 267 loss of 292
Canadian Conservation Institute 165, Codex Sinaiticus 128–9, 227
168, 174, 175, 177, 275 codices 71–3, 79, 80, 81
Cantor, Georg 52 colophon 181, 278–80
card games 245 combinatorics 55–7, 244–51, 254,
Carey, Chris 228 256–8, 284, 288
Casablanca hypothesis 172–4 computer technology 209–10, 265–6
catapults 59–60 see also digital imaging
Cauchy, Augustin Louis 52 cones 92–3, 150
Central Intelligence Agency (CIA) see also parabolas
18, 262 conoids 44
centers of gravity 1, 27, 54, 140–7 Conoids and Spheroids 41, 44, 80, 107,
Leonardo da Vinci 121–2 198–9
parabolas 155, 156 Conon 40
Choniates, Nicetas 1, 118 conservation 16, 159, 174–9
Christens-Barry, William A. (Bill) 25, Constantinople 75–80, 230
205–6, 211–2, 264 Heiberg’s discovery of Palimpsest
EL GRECO 267 131–2
imaging problems 213–6, 218 Metochion manuscripts moved
imaging process 218, 220 from 133
VASARI 267 return of Palimpsest to 131
Christie’s 3–6, 32, 65, 135–7 sacking of 1, 116–9
Chrysippus 247 Wilson on 128–9
Chung, Fan 257 court cases 4, 65, 134–5
CIA see Central Intelligence Agency Croly, George 127, 128
Cicero 62, 89 crusades 231–2
circles cubic root 61
Archimedes’ tomb 105–6 Cutler, Bill 257–8
pi 56 cylindrical cut 188–99, 201–2
squaring 41
Clagett, Marshall 120 data-cubes 212, 267
Codex A 2, 80–1, 95 Dean, John 65, 66, 75–6, 115,
diagrams 97, 98–9 126–30, 135
Heiberg’s edition 131 Demosthenes 228
306
9780306815805_RM.qxd 8/22/07 9:36 AM Page 307
INDEX
DEREK 266 Eureka! myth 34, 54, 120

Diaconis, Persi 249–50, 257 Eutocius 73–4, 76–7, 95, 225, 284
diagrams 22, 31, 67, 89–109, 115 Evangelist pictures see forgeries
beauty of 105–7
digital imaging 213 Fermat, Pierre de 241–2, 258
Heiberg’s publications 132 Finch, Simon 5–7, 8–11
schematic representations 101, Floating Bodies see On Floating Bodies
102, 103–4 Fogg, Sam 8, 10–12
The Works of Archimedes 225 folios 15, 81, 122
digital imaging see imaging forgeries 132, 163–4, 165–6, 167,
Dijksterhuis, E. J. 235 175, 256
Diondas 226 “distressed” folios 160
disbinding process 159–64, 176 X-ray imaging 267
Dositheus 37, 40, 41, 53 fossils 266
draftsmanship 103 Fowler, David 89
Duke University Library 133, 167, Frederick II, Holy Roman Emperor
168 227–8
Easterling, Patricia 22, 23, 32, 130, Galileo 2, 27–8, 40, 121, 203, 291
228 infinity 182, 184
Easton, Roger 25, 205–6, 211–2, 264 on mathematics 291–2
EL GRECO 267 movement of projectiles 61–2
imaging problems 213–6, 218 pure thought 54
imaging process 218, 220 geometry 48, 53, 143, 246
VASARI 267 diagrams 104
EDAX machine 271–2, 273, 275, 276 Newton 28
Einstein, Albert 33, 89, 246 parabolas 147
EL GRECO 267 volume of a cylindrical cut 191
elemental information 271–2 George, Martin 275
ellipses 150 Giaccai, Jennifer 277
entropy 289–90 Gislason, Jason 264–5
equalities of multitude 202–3 Golz, Doug 264
equations 87–9, 111–2, 114 Graham, Ron 257
Equilibrium of Planes 272 Greek Orthodox Patriarchate of
Eratosthenes 53, 66, 68–70, 134, 174, Jerusalem 4
187 Group Theory 232, 250
Euclid 40, 72, 75, 80, 103, 183–4 Guersan, Anne 134–5, 168
Eudoxus 34, 224 Guersan, Robert 134, 168
Eumenes II 81 Guerson, Salomon 169, 170–3
307
9780306815805_RM.qxd 8/22/07 9:36 AM Page 308
Hagia Sophia 1, 76 Archimedean science 284–8

Hall, Gene 264, 269–70 complaints about results 213–6
Handley, Eric 228 digital imaging 24–5, 199–200,
Haselden, Reginald Berti 170 205–25
heat 290 multi-spectral imaging process
Heath, T. L. 30–1 206–11
Heiberg, Johan Ludwig 16, 174, 178, UV images 218–20
217 indirect proof 45, 51, 143
discovery of Palimpsest in infinity 27, 28, 45, 51–2, 139, 157
Constantinople 131–3, 292 history of mathematics 184–6
forgeries 134 Law of the Balance 148
Method 132, 157, 183–8, 194–5, magnitudes 199
254–5, 293 Method 186, 187–8, 199–204, 254
On Floating Bodies 278 potential 45, 47–8, 51, 52, 184,
patience 179 185
photographs 165–7, 174, 180 Stomachion 240–1, 254
reputation of 225 volume of a cylindrical cut 196–7
sections of manuscript overlooked information 284–8, 289
by 180–1, 224–5 ink 83–4
Stomachion 54, 231, 235, 237–8, Isidore of Miletus 76–7, 130
250–2, 253 isosceles cones 92–3
tribute to 283–5 Istanbul see Constantinople
Hermann, Jud 228
Hero of Alexandria 76, 284 Jacopo of Cremona 119, 120
Herodotus 229 James, M. R. 160
Hillman, John 264 Jerusalem 126, 131, 231–2
Hipparchus 247–8 Johnston, Andy 264
Hippocrates of Chios 39 Johnston, Robert H. 24, 25
hoaxes 41, 54
Holmes, Susan 250, 257 Kelekian, Dikran 171
Homer 53, 57, 75 Kemal, Mustafa (Ataturk) 133
Horizon 263 Kerr, Don 262
Horváth, László 227, 228–9 Khayyam, Omar 40
Hough, David 247–8 Knox, Keith 24–5, 205–6, 277
Huntington Library 169, 170 EL GRECO 267
hydrostatics 54 imaging problems 213–6, 218
hyperbolas 150 imaging process 218, 220
Hyperides 78, 226–9, 230, 261 VASARI 267
X-ray images 277
imaging 18, 23–5, 199–200, 205–25 Krum, Khan of Bulgaria 77, 230
access to images 221–2 Kuklos symbol 178
308
9780306815805_RM.qxd 8/22/07 9:36 AM Page 309
INDEX
language 91, 112 measurement 41–4, 284

Law of the Balance 27, 53, 148–9, parabolas 43, 48–51
155–6 potential infinity 47–8
LEDs see Light Emitting Diodes Measurement of the Circle 73, 80, 85,
Leibniz, Gottfried Wilhelm 1, 2, 52 121, 235
Leo the Philosopher (Geometer) 80 Method 4, 15, 52–3, 80, 137, 180
Leonardo da Vinci 2, 40, 121–2 analysis of digital images 223–4
letters 112–3 centers of gravity 122, 139–40
Lever, Law of the 148 deterioration of folio 166
light 204–7 diagrams 101–2
Light Emitting Diodes (LEDs) 263–4 Heiberg’s transcription of 132,
Lloyd, Geoffrey 30 186–7, 293
logic 247, 248 infinity 157, 186, 187–8, 199–204,
Lowden, John 179, 230 255
colophon 181, 278, 279 Law of the Balance 53, 148
forgeries 167, 170, 178 letter to Eratosthenes 66–70, 85,
Lynch, John 263 136–7
parabolas 149, 157
magnitude 197–9 prayer written over 124
Marasco, Joe 229–30, 233, 234, volume of a cylindrical cut 191–9
238–9, 256 Metochion of the Jerusalem Holy
Martinez, Kirk 264 Sepulchre (Constantinople) 129,
mathematics 130, 131–2, 165, 168, 182
application to physics 29, 139, 148, Metrica 70
155, 203, 289, 290 Michelangelo see Buonarotti
Euclid 40 minuscule script 79–80, 85
experience of 114–5 Morton, Bob 264, 270–2
Galileo 290–1 motion 27, 273
Greek 30–1, 39, 41, 42, 55 mold 162–1, 234
combinatorics 246–7 “Mr. B” 11–6, 19–20, 157, 161, 261,
diagrams 90, 92, 93–4, 103–4, 282–3
246 multi-spectral imaging 206–11, 214,
“equal in multitude” 201 262, 267–8
indirect proof 46 multitude 201, 256
infinity 184–5, 199 Myronas, Ioannes 279, 280, 288
parabolas 150
Group Theory 236, 254 National Library of Greece 133
history of 184–6, 188 National Reconnaissance Office
infinity 184–6, 202 (NRO) 17–9
symbols 110–4, 115 Neugebauer, Otto 70
see also calculus; geometry Newton, Isaac 1, 2, 27, 40, 121, 203
309
9780306815805_RM.qxd 8/22/07 9:36 AM Page 310
calculus 27, 52, 54, 157 parchment 15, 80–1

infinity 184, 186 conservation of 159–60
Laws of Motion 273 deterioration of 161–3, 234,
movement of planets 54, 61–2, 235–6
140, 290–1 Parenti, Stefano 230
NRO see National Reconnaissance Parpulov, Georgi 168, 278–9
Office Pascal, Blaise 245
Nylese, Tara 272 Peakin, Will 226, 263
Petersen, Erik 166
Odyssey 39, 57 Pheidias 35–6
Omont, Henri 167, 169, 171 Philip of Macedon 227, 228
On Floating Bodies 4, 15, 53, 80, 85 philological method 95, 97, 98–9
absence of Eureka moment 34 Photius 77, 230–1
analysis of digital images 223–4 physics
Leonardo da Vinci 121 application to mathematics 148,
prayer written over 124–5 187, 188, 203–4
Renaissance scholars 120 entropy 288–90
sections overlooked by Heiberg mathematics application to 27,
180, 224 139, 150, 155, 204, 289, 291
X-ray scanning 277–8 Method 139, 140, 157
one-to-one correspondence 201–2 Newtonian 140–1, 273–4, 290–1
Optical Character Recognition quantum mechanics 246
265–6 synchrotron radiation 274
owner of Palimpsest (“Mr. B”) see also science
11–17, 19–20, 161, 165, 216, 261, pi 1, 56
265–66, 282–3 Piero della Francesca 120
Oyens, Felix de Marez 3–5, 7, 135 Pio, Alberto 120
Plato 27, 36, 53, 72
palimpsest definition 15 plethos (multitude) 201, 256
Papadopoulos-Kerameus, A. 126, Plutarch 62, 229, 247, 248, 249
131–2 Politian 120
Pappus 75 Polybius 60
papyrus 67, 81, 108–9 polygons 57, 99–101
parabolas 44, 149, 150–7, 287 Pompeii 103, 107
Archimedes’ tomb 106 potential infinity 46, 47–8, 51, 52,
conic sections 150 184, 185
geometrical method 48–51 principal components analysis
projectiles 61 206–7
volume of a cylindrical cut 191–3 probability 245, 246, 284, 286, 288
parallelograms 141 Procopius 76
310
9780306815805_RM.qxd 8/22/07 9:36 AM Page 311
INDEX
proportion 27, 148, 194, 196, 197 schematic representations 101, 103,
pseudocolor images 220, 238, 256, 104–5
267, 277 “Schröder numbers” 248–9
puzzles 35–6, 54–8 science 37, 38–9, 55, 87, 147, 204
see also Stomachion abbreviated notations 114
pyramids 43 Archimedes’ influence on 26–8,
Pythagoras 39 284, 288–91, 292–3
cross-cultural contact 40
Quadrature of the Parabola 41, 43, 80 diagrams 89–90, 114
Quandt, Abigail 13, 17, 19, 20–1, 24 equations 87–8, 89, 114
disbinding and conservation Hellenistic patronage 282
process 159–67, 175–82, 205, imaging 284–8
220–1, 281 infinity 184
forgeries 167, 179 probabilities 246
palimpsesting process 123 speculation 142
X-ray scanning 275, 276, 278 see also physics
quantum mechanics 246 “scientific revolution” 28, 129, 184,
185
Rashed, Marwan 229 scribes 84, 85, 181–2, 220, 232
rectangles 42 abbreviations 112, 113
Archimedes’ tomb 106 Christian texts 74–5
volume of a cylindrical cut 194–6, diagrams 94–5, 97, 98
201–2 palimpsesting process 122–6
Regiomontanus 120 symbols 110–1
Renaissance 119, 129 scripts 78–9
Rickey, Fred 5 Scruggs, Bruce 272
Roberts, David 127 Serapeum 69, 74
rolls 67, 69, 71–2, 73, 109 Set Theory 202
Royal Library of Denmark 165–6 Shannon, Claude 289, 291, 292
Rubik’s cube 236, 254 Sinai 130, 231
Russo, Lucio 248 Sirieix, Marie Louis 134, 168–9,
173–4
St. Catherine’s monastery 130, 231 Sphere and Cylinder 41, 43, 80, 85, 130
St. John Psichaites 230 diagrams 22, 92, 95–7, 99, 108, 111
St. Sabas Monastery 127–9, 131, 162, Eutocius’ commentary 73, 77, 225
182, 231 Leonardo da Vinci 121
Saito, Ken 182, 183–4, 186–8, 190, Reviel’s translation of 225
191, 205, 230 symbols 112
Salerno, Emanuele 264, 267 spheres 43
Sand-Reckoner 35, 56–7, 80 spheroids 44
311
9780306815805_RM.qxd 8/22/07 9:36 AM Page 312
Spiral Lines 41, 44, 80, 85 Toth, Michael B. 17–9, 23–4, 165,
diagrams 107, 108 261–5, 281
Leonardo da Vinci 121 imaging problems 213–4
squares 42, 106–7 Optical Character Recognition
Stanford Linear Accelerator Center 265–6
(SLAC) 264, 273, 278–9 X-ray scanning 275, 276, 277
Stanford Positron Electron triangles 42, 90–1, 104
Accelerating Ring (SPEAR) center of gravity 54, 121, 139–40,
273–4 141–7
Stanford Synchrotron Radiation parabolas 43, 49–51, 52, 149–55
Laboratory (SSRL) 274 volume of a cylindrical cut 191–9,
Stanley, Richard P. 248 200–1
Stoic logic 247, 248 Trippel, Katja 264
Stomachion 4, 15, 54–5, 80, 85, Tybjerg, Karin 166
233–60, 288 Tzetzes, Johannes 33–4
creative shapes 240–3
deterioration of folio 159–60, 164 ultraviolet (UV) images 177, 179,
Heiberg photograph 165 206–7, 211–2, 213, 217–9
rotations 250, 251–4, 259
solutions 55, 251–2, 256–9 Valentin, Rose 119
substitutions 251–5, 259 Valla, Giorgio 120
Sunday New York Times 260 VASARI 267, 268
Sunday Times 226, 263 vaults 70
Suter, H. 236, 237 Venizelos, Evangelos 5
symbols 110–14, 115 Vikan, Gary 8, 9, 11–3, 21
synchrotron radiation 273 Vitruvius 34–5, 54
Syracuse 32, 33, 59–60, 62, 66, 89 volume 41–2
Taft, Robert 231 Walters Art Museum 7–13, 133, 136,

Tchernetska, Natalie 23, 179, 214–6, 165
216–7, 226–8 Walvoord, Derek 266
Temple of Serapis 69 Washington Post 262–3
Temple to the Muses 68 Whitehead, A. N. 27
Thales 39 William of Moerbeke 119, 120, 225,
Themistius 75 278
Theodosius II 75–6, 77 Willoughby, Harold R. 169, 171
Theophilus 123 Wilson, Nigel 21, 23, 85, 129–30, 281
thermodynamics 290 Aristotle commentary discovery
time 27 230
Tischendorf, Constantin 130–1, 231 Bibliotheca of Photius 78
312
9780306815805_RM.qxd 8/22/07 9:36 AM Page 313
INDEX
collaboration with Netz 223–5, writing 81–2, 108–10

250
colophon 278–80 X-ray scanning 269–72, 274, 277,
digital imaging 224–5 278–9
Stomachion 249, 250, 254, 260
The Works of Archimedes 225 Young, Greg 275
World News Tonight 213
World War II 172–4 Zutshi, Patrick 23, 127
313

The Archimedes Codex PDF

Uploaded by

Copyright:

Available Formats

The Archimedes Codex PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Archimedes Codex PDF

Uploaded by

Copyright:

Available Formats

9780306815805_FM.

qxd 8/22/07 10:38 AM Page i

This page intentionally left blank

Many of the designations used by manufacturers and sellers to distinguish

Copyright © 2007 by Reviel Netz and William Noel

All rights reserved. No part of this publication may be reproduced, stored

Production Services by Chris Grison for Outbox Creative Partners

First Da Capo Press edition 2007

Published by Da Capo Press

This page intentionally left blank