WEBVTT

00:00:11.288 --> 00:00:13.948
On January 26th, 2013,

00:00:13.948 --> 00:00:17.508
a band of al-Qaeda militants
entered the ancient city of Timbuktu

00:00:17.508 --> 00:00:20.258
on the southern edge of the Sahara Desert.

00:00:20.258 --> 00:00:25.768
There they set fire to a medieval library
of 30,000 manuscripts

00:00:25.768 --> 00:00:29.048
written in Arabic
and several African languages,

00:00:29.048 --> 00:00:33.038
and ranging in subjects
from astronomy to geography,

00:00:33.038 --> 00:00:34.788
history to medicine,

00:00:34.788 --> 00:00:36.528
including one book which records

00:00:36.528 --> 00:00:40.958
perhaps the first treatment
for male erectile dysfunction.

00:00:41.808 --> 00:00:43.718
Unknown in the West,

00:00:43.718 --> 00:00:47.048
this was the collected wisdom
of an entire continent,

00:00:47.048 --> 00:00:52.038
the voice of Africa at a time when Africa
was thought not to have a voice at all.

00:00:52.514 --> 00:00:55.224
The Mayor of Bamako,
who witnessed the event,

00:00:55.224 --> 00:01:00.014
called the burning of the manuscripts
a crime against world cultural heritage.

00:01:00.014 --> 00:01:02.283
And he was right, or he would have been,

00:01:02.283 --> 00:01:04.983
if it weren't for the fact
that he was also lying.

00:01:05.740 --> 00:01:08.570
In fact, just before,

00:01:08.570 --> 00:01:12.540
African scholars had collected
a random assortment of old books

00:01:12.540 --> 00:01:15.240
and left them out
for the terrorists to burn.

00:01:15.240 --> 00:01:19.510
Today, the collection lies hidden
in Bamako, the capital of Mali,

00:01:19.510 --> 00:01:22.190
moldering in the high humidity.

00:01:22.190 --> 00:01:25.770
What was rescued by ruse
is now was once again in jeopardy,

00:01:25.770 --> 00:01:27.770
this time by climate.

00:01:27.770 --> 00:01:31.260
But Africa and the far-flung corners
of the world are not the only places,

00:01:31.260 --> 00:01:32.500
or even the main places,

00:01:32.500 --> 00:01:36.980
in which manuscripts that could change
the history of world culture

00:01:36.980 --> 00:01:38.985
are in jeopardy.

00:01:39.476 --> 00:01:40.356
Several years ago,

00:01:40.356 --> 00:01:44.046
I conducted a survey
of European research libraries,

00:01:44.046 --> 00:01:46.306
and discovered that,
at the barest minimum,

00:01:46.306 --> 00:01:52.776
there are 30,000,
actually 60,000 manuscripts pre-1500

00:01:52.776 --> 00:01:54.556
that are illegible

00:01:54.556 --> 00:02:00.006
because of water damage, fading,
mold, and chemical reagents.

00:02:00.422 --> 00:02:03.026
The real number is likely double that.

00:02:03.026 --> 00:02:06.516
That doesn't even count
renaissance manuscripts,

00:02:06.516 --> 00:02:08.036
and modern manuscripts,

00:02:08.036 --> 00:02:11.586
and cultural heritage objects,
such as maps.

00:02:13.004 --> 00:02:15.774
What if there were a technology

00:02:15.774 --> 00:02:20.375
that could recover
these lost and unknown works?

00:02:21.269 --> 00:02:23.739
Imagine worldwide how a trove

00:02:23.739 --> 00:02:28.539
of hundreds of thousands
of previously unknown texts

00:02:28.539 --> 00:02:31.739
could radically transform
our knowledge of the past.

00:02:33.199 --> 00:02:37.539
Imagine what unknown classics
we would discover

00:02:38.132 --> 00:02:42.629
which would rewrite the canons
of literature, history, philosophy, music.

00:02:43.267 --> 00:02:46.767
Or more provocatively, that could rewrite
our cultural identities,

00:02:46.767 --> 00:02:50.257
building new bridges
between people and culture.

00:02:50.882 --> 00:02:53.522
These are the questions
that transformed me

00:02:53.522 --> 00:02:56.254
from a medieval scholar,
a reader of texts,

00:02:56.254 --> 00:02:58.481
into a textual scientist.

00:02:59.254 --> 00:03:01.764
What an unsatisfying word reader is?

00:03:01.764 --> 00:03:04.024
For me, it conjures up
images of passivity,

00:03:04.024 --> 00:03:06.764
of someone sitting idly in an armchair,

00:03:06.764 --> 00:03:10.984
waiting for knowledge
to come to him in a neat little parcel.

00:03:10.984 --> 00:03:14.504
How much better is to be
a participant in the past,

00:03:14.504 --> 00:03:19.508
an adventurer in an undiscovered country,
searching for the hidden text?

00:03:20.744 --> 00:03:23.744
As an academic, I was a mere reader.

00:03:24.231 --> 00:03:26.751
I read and taught the same classics

00:03:26.751 --> 00:03:29.751
that people had been reading and teaching
for hundreds of years:

00:03:29.751 --> 00:03:32.751
Virgil, Ovid, Chaucer, Petrarch.

00:03:32.751 --> 00:03:35.001
With every scholarly article
that I published,

00:03:35.001 --> 00:03:38.321
I added to human knowledge
an ever-diminishing slivers of insight.

00:03:39.900 --> 00:03:43.970
What I wanted to be
was an archaeologist of the past,

00:03:43.970 --> 00:03:45.795
a discoverer of literature,

00:03:45.795 --> 00:03:49.345
an Indiana Jones without the whip -
or, actually, with the whip.

00:03:49.345 --> 00:03:50.355
(Laughter)

00:03:50.355 --> 00:03:54.045
And I wanted it not just for myself,
but I wanted it for my students as well.

00:03:54.045 --> 00:03:57.975
So six years ago, I changed
the direction of my career.

00:03:57.975 --> 00:04:00.995
At the time, I was working
on "The Chess of Love",

00:04:00.995 --> 00:04:03.795
the last important long poem
of the European Middle Ages,

00:04:03.795 --> 00:04:05.525
never to have been edited.

00:04:05.525 --> 00:04:08.525
It wasn't edited because it existed
in only one manuscript,

00:04:08.525 --> 00:04:10.275
which was so badly damaged

00:04:10.290 --> 00:04:13.125
during the firebombing of Dresden
in World War II

00:04:13.125 --> 00:04:16.265
that generations of scholars
had pronounced it lost.

00:04:16.422 --> 00:04:19.785
For five years, I had been working
with an ultraviolet lamp,

00:04:19.785 --> 00:04:22.265
trying to recover traces of that writing,

00:04:22.267 --> 00:04:25.025
and I'd gone about as far
as the technology of that time

00:04:25.025 --> 00:04:26.245
could actually take me.

00:04:26.245 --> 00:04:28.045
So I did what many people do;

00:04:28.045 --> 00:04:31.065
I went online, and there I learned

00:04:31.238 --> 00:04:36.255
about how multi-spectral imaging
had been used to recover 2 lost treatises

00:04:36.255 --> 00:04:40.805
of the famed Greek mathematician
Archimedes from a 13th-century palimpsest.

00:04:40.805 --> 00:04:44.265
A palimpsest is a manuscript
which has been erased and overwritten.

00:04:45.530 --> 00:04:48.020
So, out of the blue, I decided to write

00:04:48.020 --> 00:04:52.020
to the lead imaging scientist
on the Archimedes Palimpsest Project,

00:04:52.020 --> 00:04:55.320
Professor Roger Easton,
with a plan and a plea.

00:04:55.320 --> 00:04:57.770
To my surprise, he actually wrote back.

00:04:59.238 --> 00:05:02.948
With his help, I was able to win
a grant from the US government

00:05:04.505 --> 00:05:08.515
to build a transportable
multispectral imaging lab -

00:05:08.515 --> 00:05:10.815
yes, this is the dirty little secret

00:05:10.815 --> 00:05:13.025
of where your tax dollars
are really going -

00:05:13.025 --> 00:05:17.805
and with this lab, I transformed
what was a charred and faded mess

00:05:17.805 --> 00:05:20.235
into a new medieval classic.

00:05:20.739 --> 00:05:23.539
So, how does multispectral imaging
actually work?

00:05:23.539 --> 00:05:25.749
The idea behind multispectral imaging

00:05:25.749 --> 00:05:27.589
is that something that anyone

00:05:27.589 --> 00:05:30.159
who is familiar
with an infrared night-vision goggles

00:05:30.159 --> 00:05:31.519
will immediately appreciate,

00:05:31.519 --> 00:05:34.629
that what we can see invisible light,
invisible spectrum of light,

00:05:34.629 --> 00:05:37.759
is only tiny fraction
of what's actually there.

00:05:38.511 --> 00:05:40.731
The same is true with invisible writing.

00:05:42.499 --> 00:05:48.920
Our system uses 12 wavelengths of light
between the ultraviolet and the infrared.

00:05:50.059 --> 00:05:54.449
These are shown down onto the manuscript
from above, from banks of LEDs,

00:05:54.449 --> 00:05:56.279
and another multispectral light source

00:05:56.279 --> 00:05:59.229
which comes up through
the individual leaves of the manuscript.

00:05:59.229 --> 00:06:03.959
Up to 35 images per leaf
are imaged this way,

00:06:03.959 --> 00:06:05.569
using a high-power digital camera

00:06:05.569 --> 00:06:08.499
equipped with a lens
which is made out of quartz.

00:06:08.499 --> 00:06:10.789
There are about 5 of these in the world.

00:06:10.789 --> 00:06:12.849
Once we capture these images,

00:06:12.849 --> 00:06:15.039
we feed them
through statistical algorithms

00:06:15.039 --> 00:06:17.269
to further enhance and clarify them,

00:06:17.269 --> 00:06:20.499
using software which is originally
designed for satellite images,

00:06:20.499 --> 00:06:25.969
and used by people
like geospatial scientists and the CIA.

00:06:25.969 --> 00:06:28.276
The results can be spectacular.

00:06:28.276 --> 00:06:30.306
Some of you may already have heard

00:06:30.306 --> 00:06:32.976
of what's been done
for the Dead Sea Scrolls,

00:06:32.976 --> 00:06:35.246
which are slowly gelatinizing.

00:06:35.726 --> 00:06:37.596
Using infrared, we've been able to read

00:06:37.596 --> 00:06:41.246
even the darkest corners
of the Dead Sea Scrolls.

00:06:41.961 --> 00:06:43.411
You may not be aware, however,

00:06:43.411 --> 00:06:45.961
of other Biblical texts
that are in jeopardy.

00:06:46.459 --> 00:06:51.519
Here, for example, is a leaf
from a manuscript that we imaged,

00:06:51.519 --> 00:06:56.518
which is perhaps the most valuable
Christian Bible in the world.

00:06:57.332 --> 00:07:03.022
The Codex Vercellensis is the oldest
translation of the Gospels into Latin,

00:07:03.022 --> 00:07:05.742
and it dates
from the first half of 4th century.

00:07:07.723 --> 00:07:12.523
As you can see, this is the closest
we can come to the Bible

00:07:12.523 --> 00:07:17.253
at the time of the foundation
of Christendom under Emperor Constantine,

00:07:17.253 --> 00:07:19.813
and at the time
of also the Council of Nicaea,

00:07:19.813 --> 00:07:23.073
when the basic creed of Christianity
was being agreed upon.

00:07:23.468 --> 00:07:26.548
This manuscript, unfortunately,
has been very badly damaged.

00:07:26.548 --> 00:07:31.258
It's damaged because for centuries
it has been used and handled

00:07:31.258 --> 00:07:34.028
in swearing-in ceremonies in the church.

00:07:34.028 --> 00:07:37.678
In fact, that purple splotch that you see
in the upper right-hand corner

00:07:38.365 --> 00:07:41.268
- upper left-hand corner.
Right-hand corner? Yes. -

00:07:41.268 --> 00:07:43.508
... is Aspergillus,

00:07:43.508 --> 00:07:49.748
which is a fungus which originates
originally in the unwashed hands

00:07:49.748 --> 00:07:51.938
of a person with tuberculosis.

00:07:52.554 --> 00:07:54.614
Our imaging has enabled me

00:07:54.614 --> 00:07:58.724
to make the first transcription
of this manuscript in 250 years.

00:07:59.749 --> 00:08:02.129
Having a lab that can travel
to collections

00:08:02.129 --> 00:08:03.859
- to where it's needed, however -

00:08:03.859 --> 00:08:05.519
is only part of the solution.

00:08:05.519 --> 00:08:08.259
The technology is expensive and very rare,

00:08:08.588 --> 00:08:11.499
and the imaging
and image processing skills are esoteric.

00:08:11.499 --> 00:08:14.749
That means that mounting recoveries
is beyond the reach

00:08:14.749 --> 00:08:18.259
of most researchers
and all but the wealthiest institutions.

00:08:18.259 --> 00:08:21.369
That's why I founded the Lazarus Project,

00:08:21.383 --> 00:08:23.779
a non-for-profit initiative

00:08:23.779 --> 00:08:25.919
to bring multispectral imaging

00:08:25.919 --> 00:08:29.919
to individual researchers
and smaller institutions

00:08:29.919 --> 00:08:31.739
at little or no cost whatsoever.

00:08:32.494 --> 00:08:34.294
Over the past five years,

00:08:34.294 --> 00:08:38.024
our team of imaging scientists,
scholars, and students

00:08:38.024 --> 00:08:40.244
has traveled to seven different countries

00:08:40.244 --> 00:08:41.534
and have recovered

00:08:41.534 --> 00:08:44.054
some of the world's most valuable
damaged manuscripts,

00:08:44.054 --> 00:08:47.184
including the Vercelli Book,
which is the oldest book of English,

00:08:47.184 --> 00:08:49.764
the Black Book of Carmarthen,
the oldest book of Welsh,

00:08:49.764 --> 00:08:53.524
and some of the most valuable
earliest Gospels,

00:08:53.524 --> 00:08:56.544
located in now
what's the former Soviet Georgia.

00:08:57.749 --> 00:09:00.759
So spectral imaging
can recover lost texts.

00:09:01.475 --> 00:09:06.505
More subtly, though, it can recover
a second story behind every object,

00:09:06.505 --> 00:09:11.275
the story of how, when,
and by whom a text was created,

00:09:11.275 --> 00:09:14.955
and sometimes, what the author
was thinking at the time he wrote.

00:09:15.727 --> 00:09:18.677
Take, for example, a draft
of the Declaration of Independence,

00:09:18.677 --> 00:09:21.247
written in Thomas Jefferson's own hand,

00:09:21.247 --> 00:09:23.727
which some colleagues of mine
imaged a few years ago

00:09:23.727 --> 00:09:25.467
at the Library of Congress.

00:09:25.467 --> 00:09:26.757
Curators had noticed

00:09:26.757 --> 00:09:30.487
that one word throughout
had been scratched out and overwritten.

00:09:30.487 --> 00:09:33.247
The word overwritten was "citizens".

00:09:33.247 --> 00:09:36.267
Perhaps you can guess
what the word underneath was.

00:09:36.995 --> 00:09:38.236
"Subjects".

00:09:38.504 --> 00:09:41.304
There, ladies and gentlemen,
is American democracy

00:09:41.304 --> 00:09:43.774
evolving under the hand
of Thomas Jefferson.

00:09:44.247 --> 00:09:48.016
Or consider the 1491 Martellus Map,

00:09:48.247 --> 00:09:50.767
which we imaged
at Yale's Beinecke Library.

00:09:50.767 --> 00:09:53.517
This was the map
that Columbus likely consulted

00:09:53.517 --> 00:09:55.287
before he traveled to the New World,

00:09:55.287 --> 00:09:57.907
and which gave him his idea
of what Asia looked like

00:09:57.907 --> 00:09:59.712
and where Japan was located.

00:10:01.218 --> 00:10:02.918
The problem with this map

00:10:02.920 --> 00:10:06.508
is that its inks and pigments
had so degraded over time

00:10:06.508 --> 00:10:08.917
that this large, nearly 7-foot map

00:10:09.495 --> 00:10:12.495
made the world look like a giant desert.

00:10:12.495 --> 00:10:16.005
Until now, we had very little idea,
detailed idea, that is,

00:10:16.005 --> 00:10:19.535
of what Columbus knew of the world
and how world cultures were represented.

00:10:20.255 --> 00:10:24.245
The main legend of the map
was entirely illegible under normal light.

00:10:24.245 --> 00:10:27.255
Ultraviolet did very little for it.

00:10:27.255 --> 00:10:29.725
Multispectral gave us everything.

00:10:30.225 --> 00:10:32.275
In Asia, we learned of monsters

00:10:32.275 --> 00:10:35.905
with ears so long that they could cover
the creature's entire body.

00:10:36.491 --> 00:10:41.141
In Africa, about the snake
who could cause the ground to smoke.

00:10:42.254 --> 00:10:44.244
Like starlight which can give today

00:10:44.244 --> 00:10:47.614
images of the way the universe
looked in the distant past,

00:10:47.614 --> 00:10:50.034
so multispectral light can take us back

00:10:50.034 --> 00:10:53.504
to the first stuttering moments
of an object's creation.

00:10:53.504 --> 00:10:57.504
Through this lens, we witnessed
the mistakes, the changes of mind,

00:10:57.504 --> 00:11:00.554
the naivetes, young censored thoughts,

00:11:00.554 --> 00:11:02.744
the imperfections of the human imagination

00:11:02.744 --> 00:11:05.784
that allowed these hallowed objects
and their authors

00:11:05.784 --> 00:11:10.175
to become more real,
that make history closer to us.

00:11:13.744 --> 00:11:15.531
So what about the future?

00:11:15.531 --> 00:11:17.781
There's so much of the past

00:11:17.781 --> 00:11:21.051
and so few people
with the skills to rescue it

00:11:21.051 --> 00:11:24.839
before these objects disappear forever.

00:11:25.541 --> 00:11:29.201
That's why I've begun to teach
this new hybrid discipline

00:11:29.201 --> 00:11:30.981
that I call textual science,

00:11:30.981 --> 00:11:35.482
a mixture between
kind of Indiana Jones meets CSI.

00:11:36.704 --> 00:11:39.004
Textual science is a marriage

00:11:39.004 --> 00:11:41.474
of the traditional skills
of the literary scholar -

00:11:41.474 --> 00:11:44.144
the ability to read old languages
and old handwriting,

00:11:44.144 --> 00:11:45.694
the knowledge how texts are made

00:11:45.694 --> 00:11:47.904
in order to be able
to place and date them -

00:11:47.904 --> 00:11:50.324
with new techniques like imaging science,

00:11:50.324 --> 00:11:52.764
the chemistry of inks and pigments,

00:11:52.764 --> 00:11:55.504
computer-aided optical
character recognition.

00:11:56.431 --> 00:11:59.031
Last year, a student in my class,

00:11:59.031 --> 00:12:01.741
a freshman with the background
in Latin and Greek

00:12:01.741 --> 00:12:03.771
was image-processing a palimpsest

00:12:03.771 --> 00:12:07.741
that we had photographed
at the famous library in Rome.

00:12:07.741 --> 00:12:12.751
As he worked, tiny Greek writing
began to appear from behind the text.

00:12:13.713 --> 00:12:15.073
Everyone gathered around,

00:12:15.073 --> 00:12:20.513
and he read a line from a lost work
of the Greek comic dramatist Menander.

00:12:21.239 --> 00:12:24.239
This was the first time
in well over a thousand years

00:12:24.244 --> 00:12:27.289
that those words
had been pronounced aloud.

00:12:27.959 --> 00:12:30.469
In that moment, he became a scholar.

00:12:31.388 --> 00:12:34.496
Ladies and gentlemen,
that is the future of the past.

00:12:34.496 --> 00:12:35.516
Thank you very much.

00:12:35.516 --> 00:12:37.156
(Applause)