1 00:00:00,404 --> 00:00:02,904 (eerie music) 2 00:00:08,780 --> 00:00:11,430 What does it mean to be lost to history? 3 00:00:12,280 --> 00:00:14,310 The ancient Roman town of Herculaneum 4 00:00:14,310 --> 00:00:17,693 essentially disappeared in the year 79 AD, 5 00:00:18,580 --> 00:00:21,470 buried by the eruption of Mount Vesuvius, 6 00:00:21,470 --> 00:00:25,280 destroyed by a pyroclastic surge of super hot gases 7 00:00:25,280 --> 00:00:28,020 between 400 and 900 degrees Fahrenheit 8 00:00:28,930 --> 00:00:31,233 moving nearly 200 miles per hour. 9 00:00:32,880 --> 00:00:35,150 The searing heat boiled the blood, 10 00:00:35,150 --> 00:00:37,910 and vaporized the flesh of its victims. 11 00:00:37,910 --> 00:00:42,490 Burned all in its path, leaving behind a charred city 12 00:00:42,490 --> 00:00:45,633 beneath layer upon layer of ash and volcanic debris. 13 00:00:47,410 --> 00:00:50,186 But emerging new technologies push scientific boundaries 14 00:00:50,186 --> 00:00:53,603 to recover and restore archeological artifacts. 15 00:00:56,030 --> 00:00:59,020 Taken together, advances in X-ray imaging, 16 00:00:59,020 --> 00:01:01,730 data science, and artificial intelligence 17 00:01:01,730 --> 00:01:05,550 may soon delicately and digitally unravel history 18 00:01:05,550 --> 00:01:08,017 and reveal ancient texts 19 00:01:08,017 --> 00:01:11,487 dating a few years before the height of the Roman empire. 20 00:01:11,487 --> 00:01:14,237 (dramatic music) 21 00:01:19,820 --> 00:01:22,320 Computer scientist and University of Kentucky 22 00:01:22,320 --> 00:01:26,610 professor Brent Seales is at the forefront of these efforts. 23 00:01:26,610 --> 00:01:29,210 Augmented reality is supposed to be presenting you 24 00:01:29,210 --> 00:01:31,089 with a view of your actual environment. 25 00:01:31,089 --> 00:01:35,160 There's no question that in the world of ancient material, 26 00:01:35,160 --> 00:01:39,150 this technology is gonna help us break through and read 27 00:01:39,150 --> 00:01:41,300 materials that have never been read before. 28 00:01:42,530 --> 00:01:44,510 Like the Herculaneum scrolls, 29 00:01:44,510 --> 00:01:49,340 papyrus manuscripts flash fried by Vesuvius 2,000 years ago, 30 00:01:49,340 --> 00:01:51,183 left carbonized and sealed shut. 31 00:01:52,040 --> 00:01:53,830 Mysterious Roman time capsules 32 00:01:53,830 --> 00:01:56,433 that currently drive the professor's curiosity. 33 00:01:58,470 --> 00:02:02,360 They are the canonical example of material 34 00:02:02,360 --> 00:02:06,380 from antiquity that is both too fragile to open, 35 00:02:06,380 --> 00:02:09,200 and also so valuable and mysterious 36 00:02:09,200 --> 00:02:10,840 that it has captured the attention 37 00:02:10,840 --> 00:02:12,793 of scholars for many years. 38 00:02:15,500 --> 00:02:16,670 Seales and his team 39 00:02:16,670 --> 00:02:18,720 at the Digital Restoration Initiative, 40 00:02:18,720 --> 00:02:21,080 part of the university's College of Engineering 41 00:02:21,080 --> 00:02:23,561 have pioneered archeological applications 42 00:02:23,561 --> 00:02:27,780 for technologies originally designed for medical study. 43 00:02:27,780 --> 00:02:31,183 Specifically, micro-computed 3D tomography. 44 00:02:32,910 --> 00:02:35,210 We're an imaginative people, right? 45 00:02:35,210 --> 00:02:37,706 So we see that advance, and we say 46 00:02:37,706 --> 00:02:40,130 now that I see how that works, 47 00:02:40,130 --> 00:02:42,791 there are applications that maybe no one's thought of 48 00:02:42,791 --> 00:02:45,770 that can cause breakthroughs in other areas, 49 00:02:45,770 --> 00:02:47,350 and that imagination right, 50 00:02:47,350 --> 00:02:48,590 and then that reapplication 51 00:02:48,590 --> 00:02:52,060 is part and parcel of the innovative process. 52 00:02:52,060 --> 00:02:53,370 And then how many slices do you have 53 00:02:53,370 --> 00:02:54,950 to make up the whole run of the scroll? 54 00:02:54,950 --> 00:02:57,173 I mean maybe 12,000? Right. 55 00:02:58,515 --> 00:03:00,015 That's a lot of data, right? 56 00:03:01,610 --> 00:03:03,160 And it's that imagination, 57 00:03:03,160 --> 00:03:06,003 coupled with a surefire confidence in methodology, 58 00:03:06,003 --> 00:03:09,530 which drives Seales to his most ambitious projects, 59 00:03:09,530 --> 00:03:11,263 and ultimate successes. 60 00:03:12,880 --> 00:03:15,930 Including the 2016 virtual unwrapping 61 00:03:15,930 --> 00:03:17,693 of a third century Hebrew scroll 62 00:03:17,693 --> 00:03:19,963 known as the Ein Gedi Scroll. 63 00:03:22,300 --> 00:03:25,360 Using X-ray to peer inside the badly damaged 64 00:03:25,360 --> 00:03:26,980 rolled up animal skin, 65 00:03:26,980 --> 00:03:28,850 Seales and his team laid out 66 00:03:28,850 --> 00:03:32,403 the artifact's geometric volume without ever opening it, 67 00:03:33,910 --> 00:03:35,545 highlighting the text within. 68 00:03:35,545 --> 00:03:38,295 (soothing music) 69 00:03:41,060 --> 00:03:44,130 Now he's set his sights on the Herculaneum scrolls. 70 00:03:44,130 --> 00:03:47,297 (gentle guitar music) 71 00:03:49,550 --> 00:03:52,170 Back before the eruption, the Bay of Naples 72 00:03:52,170 --> 00:03:56,520 offered a restful environment, just as it does now. 73 00:03:56,520 --> 00:04:00,306 An ideal playground for the Roman elite and well-to-do. 74 00:04:00,306 --> 00:04:03,450 (gentle guitar music) 75 00:04:03,450 --> 00:04:06,990 California's Getty Villa is an architectural replica 76 00:04:06,990 --> 00:04:09,050 of the Herculaneum Estate thought to be 77 00:04:09,050 --> 00:04:11,014 the home of Caesar's father-in-law. 78 00:04:11,014 --> 00:04:14,181 (gentle guitar music) 79 00:04:16,100 --> 00:04:20,073 Kenneth Lapatin is the museum's curator of antiquities. 80 00:04:21,230 --> 00:04:23,082 We know the Bay of Naples is where 81 00:04:23,082 --> 00:04:27,840 Romans, senators, and later emperors left the city 82 00:04:27,840 --> 00:04:29,360 in the heat of the summers 83 00:04:29,360 --> 00:04:32,550 to much more pleasant climates in the south. 84 00:04:32,550 --> 00:04:33,920 And in a way, I think it was 85 00:04:33,920 --> 00:04:36,214 like the Hamptons in the United States. 86 00:04:36,214 --> 00:04:38,964 (soothing music) 87 00:04:42,920 --> 00:04:44,655 The nearby city of Pompeii 88 00:04:44,655 --> 00:04:49,616 and its chilling body casts, death poses, 89 00:04:49,616 --> 00:04:51,890 seemingly frozen in time 90 00:04:53,260 --> 00:04:55,003 may haunt the public's mind. 91 00:04:59,660 --> 00:05:02,000 But sitting at the base of Vesuvius, 92 00:05:02,000 --> 00:05:03,991 the small town of Herculaneum 93 00:05:03,991 --> 00:05:06,690 and the lavish estates surrounding it 94 00:05:06,690 --> 00:05:08,587 were burned and buried as well. 95 00:05:08,587 --> 00:05:11,337 (soothing music) 96 00:05:20,550 --> 00:05:22,680 We know the wind was blowing south, 97 00:05:22,680 --> 00:05:25,960 so the huge column of volcanic material 98 00:05:25,960 --> 00:05:29,010 created an ash cloud, which was blown south 99 00:05:29,010 --> 00:05:32,090 and the ash gently fell on Pompeii. 100 00:05:32,090 --> 00:05:34,768 When that force of the volcano subsided 101 00:05:34,768 --> 00:05:37,250 and the ash cloud dropped, 102 00:05:37,250 --> 00:05:42,250 a very very hot, steamy, muddy volcanic gassy cloud 103 00:05:43,990 --> 00:05:47,200 rolled down the hill at high speed 104 00:05:47,200 --> 00:05:49,920 and burnt and buried Herculaneum 105 00:05:49,920 --> 00:05:51,860 very differently than Pompeii, 106 00:05:51,860 --> 00:05:54,110 burning and carbonizing 107 00:05:54,110 --> 00:05:56,740 and thus preserving a lot of the finds. 108 00:05:56,740 --> 00:05:59,230 So Herculaneum is smaller than Pompeii, 109 00:05:59,230 --> 00:06:02,053 more deeply buried, and better preserved. 110 00:06:03,753 --> 00:06:07,003 (metal tools clanging) 111 00:06:11,100 --> 00:06:14,030 Largely forgotten, Herculaneum's ruins 112 00:06:14,030 --> 00:06:16,450 were first discovered in the mid 18th Century 113 00:06:16,450 --> 00:06:19,683 under 75 feet of compacted volcanic ash. 114 00:06:21,860 --> 00:06:26,433 During the next 300 years, excavations started and stopped. 115 00:06:28,330 --> 00:06:32,413 The town still lies mostly underground, and undisturbed. 116 00:06:33,670 --> 00:06:35,870 One reason why what has been unearthed 117 00:06:35,870 --> 00:06:38,718 is incredibly rare and treasured. 118 00:06:38,718 --> 00:06:41,468 (soothing music) 119 00:06:44,410 --> 00:06:47,900 And so the Getty Villa's 2019 summer exhibit, 120 00:06:47,900 --> 00:06:51,450 Buried by Vesuvius, includes many priceless items 121 00:06:51,450 --> 00:06:53,940 recovered from its ancient prototype. 122 00:06:53,940 --> 00:06:56,810 Many on loan and transported out of Italy 123 00:06:56,810 --> 00:06:58,313 for the first time ever. 124 00:06:59,460 --> 00:07:00,373 Statues. 125 00:07:02,650 --> 00:07:03,493 Frescoes. 126 00:07:05,690 --> 00:07:08,704 And a few of those charred and scarred papyrus scrolls 127 00:07:08,704 --> 00:07:13,183 by which the original site is now known, Villa de Papryi. 128 00:07:14,100 --> 00:07:17,410 It's a small sample of an extraordinary find, 129 00:07:17,410 --> 00:07:21,230 a library of some 1,800 relics. 130 00:07:21,230 --> 00:07:24,053 There may be more as of yet un-excavated. 131 00:07:27,940 --> 00:07:29,430 We understand that when they 132 00:07:29,430 --> 00:07:31,350 first hit these papyrus scrolls, 133 00:07:31,350 --> 00:07:33,110 they looked like charcoal briquettes. 134 00:07:33,110 --> 00:07:36,000 No one knew what they were, and the story goes 135 00:07:36,000 --> 00:07:38,450 that one of them was dropped and broke open, 136 00:07:38,450 --> 00:07:41,290 and letters were found within, Greek writing, 137 00:07:41,290 --> 00:07:43,700 and then they were recognized for what they are, 138 00:07:43,700 --> 00:07:48,630 but they were carbonized and really frozen in a sense 139 00:07:48,630 --> 00:07:50,230 by this volcanic debris, 140 00:07:50,230 --> 00:07:52,963 so they can't just be easily unrolled. 141 00:07:54,250 --> 00:07:56,680 Through the centuries, laborious efforts 142 00:07:56,680 --> 00:08:00,150 to unfurl the fragile manuscripts often left scholars 143 00:08:00,150 --> 00:08:02,293 an odd collection of puzzle pieces. 144 00:08:03,910 --> 00:08:07,730 We could see a jigsaw puzzle in 3D, 145 00:08:07,730 --> 00:08:12,480 because many of the fragments are also so-called 146 00:08:12,480 --> 00:08:14,343 multi-layer fragments. 147 00:08:15,616 --> 00:08:18,930 Fabrizio Diozzi leads the Department of Papyri 148 00:08:18,930 --> 00:08:21,450 at the National Library of Naples, 149 00:08:21,450 --> 00:08:24,063 holder of the vast majority of the scrolls. 150 00:08:25,000 --> 00:08:27,503 The painstaking work to piece together the pieces 151 00:08:27,503 --> 00:08:31,590 brought to light a library largely consisting of poetry 152 00:08:31,590 --> 00:08:34,140 and epicurean philosophy detailing 153 00:08:34,140 --> 00:08:37,880 the virtues of friendship and modest pleasures. 154 00:08:37,880 --> 00:08:41,400 Writings never copied by Medieval and Muslim scribes, 155 00:08:41,400 --> 00:08:44,593 and therefore hidden from scholarly interpretation. 156 00:08:46,171 --> 00:08:49,671 (speaks foreign language) 157 00:08:53,287 --> 00:08:55,270 It's a primary source. 158 00:08:55,270 --> 00:08:58,313 We could say freshly thought, yeah? 159 00:08:59,740 --> 00:09:00,980 This is a unique find. 160 00:09:00,980 --> 00:09:04,388 This is the only library we have with its contents 161 00:09:04,388 --> 00:09:07,440 from the ancient Greco-Roman world. 162 00:09:07,440 --> 00:09:08,935 We have lots of library buildings, 163 00:09:08,935 --> 00:09:11,653 but they're empty, they're just standing shells. 164 00:09:13,000 --> 00:09:14,910 The scrolls are immensely valuable 165 00:09:14,910 --> 00:09:18,215 in the sense that it's the only library from antiquity 166 00:09:18,215 --> 00:09:21,326 that has ever been discovered in place. 167 00:09:21,326 --> 00:09:24,493 All libraries from antiquity have been lost. 168 00:09:26,290 --> 00:09:27,720 But in many respects, 169 00:09:27,720 --> 00:09:29,569 it's an unreadable library 170 00:09:29,569 --> 00:09:33,633 with some 400 to 500 scrolls still sealed shut. 171 00:09:34,830 --> 00:09:38,603 Ancient texts and thought lost in plain sight. 172 00:09:40,490 --> 00:09:42,250 What's really interesting about Herculaneum 173 00:09:42,250 --> 00:09:45,480 is that it brings out your own 174 00:09:45,480 --> 00:09:48,145 personal interests and expectations, 175 00:09:48,145 --> 00:09:52,610 because it's mysterious and intriguing 176 00:09:52,610 --> 00:09:54,910 and entirely open as to what 177 00:09:54,910 --> 00:09:57,060 each one of those individual texts will be. 178 00:09:58,260 --> 00:10:00,684 Determined to learn what lies within, 179 00:10:00,684 --> 00:10:03,730 it was his earlier work with the Ein Gedi Scroll, 180 00:10:03,730 --> 00:10:05,190 which excited those charged 181 00:10:05,190 --> 00:10:08,780 with preserving and protecting Herculaneum's heritage, 182 00:10:08,780 --> 00:10:11,923 and afforded Seales' a new scholarly partnership. 183 00:10:13,560 --> 00:10:17,070 Yes when I saw this clip, 184 00:10:17,070 --> 00:10:22,070 I think let's do with our scrolls. 185 00:10:25,090 --> 00:10:26,980 And so Seales and his team partnered 186 00:10:26,980 --> 00:10:29,440 with the UCLA School of Dentistry 187 00:10:29,440 --> 00:10:33,440 to use the powerful and sensitive X-ray equipment on site 188 00:10:33,440 --> 00:10:35,100 to scan those three scrolls 189 00:10:35,100 --> 00:10:37,639 before installation at the Getty exhibit. 190 00:10:37,639 --> 00:10:39,910 (suspenseful music) 191 00:10:39,910 --> 00:10:43,860 For two of the samples, the team designed form fitting molds 192 00:10:43,860 --> 00:10:46,113 to protect the actual antiquities. 193 00:10:48,260 --> 00:10:50,850 This 3D printed model, which is just made out of plastic 194 00:10:50,850 --> 00:10:53,333 helps us have the exact form and shape 195 00:10:53,333 --> 00:10:57,270 so that this is a lot easier to do than 196 00:10:57,270 --> 00:11:00,170 messing with the real sample, which is a lot more fragile. 197 00:11:01,560 --> 00:11:05,747 Never this papryi were so touched 198 00:11:07,460 --> 00:11:09,360 for so long time. 199 00:11:09,360 --> 00:11:11,253 So far from home. Yeah. 200 00:11:12,200 --> 00:11:14,620 It's everything new. 201 00:11:14,620 --> 00:11:16,850 Yeah, so we have to be careful. 202 00:11:16,850 --> 00:11:18,670 Okay so orientation's pretty important 203 00:11:18,670 --> 00:11:20,400 because this is a form-fitted case 204 00:11:20,400 --> 00:11:22,080 specifically for this shape, so 205 00:11:23,920 --> 00:11:27,540 the real scroll has to be oriented, right, 206 00:11:27,540 --> 00:11:29,000 in exactly this position 207 00:11:29,000 --> 00:11:31,683 so that it'll go in there just as it should. 208 00:11:33,260 --> 00:11:34,093 Are you ready? 209 00:11:35,160 --> 00:11:36,187 Yes. Yes, go. 210 00:11:36,187 --> 00:11:37,627 All right. 211 00:11:37,627 --> 00:11:38,927 Okay. 212 00:11:38,927 --> 00:11:41,510 (plucky music) 213 00:11:51,380 --> 00:11:52,653 Yeah, very nice. 214 00:11:53,600 --> 00:11:55,350 That's a perfect fit too, isn't it? 215 00:11:56,294 --> 00:11:57,127 Perfecto. 216 00:11:59,070 --> 00:12:00,193 Preservation takes priority 217 00:12:00,193 --> 00:12:02,043 throughout the process. 218 00:12:02,930 --> 00:12:05,433 Photo documentation at every step. 219 00:12:06,530 --> 00:12:07,953 Okay you have what you need? 220 00:12:09,920 --> 00:12:11,329 Soft sheets of Teflon 221 00:12:11,329 --> 00:12:13,663 add an extra layer of cushion. 222 00:12:20,740 --> 00:12:21,910 That's cool. 223 00:12:21,910 --> 00:12:24,009 This is the trickiest part of the preparation 224 00:12:24,009 --> 00:12:26,780 for the scanning because you really don't 225 00:12:26,780 --> 00:12:31,780 want to cause any damage beyond just handling it normally. 226 00:12:31,920 --> 00:12:34,883 And so the Teflon is like putting a baby to bed, you know? 227 00:12:35,880 --> 00:12:38,910 With a blanket on the top and on the bottom you know? 228 00:12:38,910 --> 00:12:40,503 You use that kind of care. 229 00:12:41,580 --> 00:12:43,230 Just like you're dealing with a little child, 230 00:12:43,230 --> 00:12:45,170 putting it to bed 231 00:12:46,040 --> 00:12:47,604 in the little form-fitted case, 232 00:12:47,604 --> 00:12:50,090 so that everybody is safe and sound 233 00:12:50,090 --> 00:12:52,540 for the time that the scanner's gonna be running. 234 00:13:02,064 --> 00:13:02,897 Oh very nice. 235 00:13:02,897 --> 00:13:03,730 No pressure. 236 00:13:05,740 --> 00:13:06,573 Awesome. 237 00:13:07,900 --> 00:13:10,210 Very nice job, Luigi, thank you. 238 00:13:10,210 --> 00:13:11,133 Very nice. 239 00:13:13,970 --> 00:13:16,650 Okay, what do you think? 240 00:13:16,650 --> 00:13:18,050 Yeah. 241 00:13:18,050 --> 00:13:18,950 Pretty stable. 242 00:13:18,950 --> 00:13:20,340 Yeah, looks great. 243 00:13:20,340 --> 00:13:21,173 Rock solid. 244 00:13:22,073 --> 00:13:23,513 Ready to go. 245 00:13:24,760 --> 00:13:27,870 Once set, the computers take over, 246 00:13:27,870 --> 00:13:32,110 performing its program, running more than 24 hours. 247 00:13:32,110 --> 00:13:35,561 Set the parameters for that scan and kick it off. 248 00:13:35,561 --> 00:13:37,350 This scanner's platform 249 00:13:37,350 --> 00:13:40,050 rotates slower than the eye can see 250 00:13:40,050 --> 00:13:43,313 in order to capture critical data from every angle. 251 00:13:44,300 --> 00:13:46,277 Here's how it works. 252 00:13:46,277 --> 00:13:50,180 The sample size of every individual point 253 00:13:50,180 --> 00:13:53,070 of this data set we collect will be very, very small. 254 00:13:53,070 --> 00:13:56,610 About the thickness of a strand of silk. 255 00:13:56,610 --> 00:13:59,410 Even better than the size of a human hair. 256 00:13:59,410 --> 00:14:01,160 We need that kind of resolution to be able 257 00:14:01,160 --> 00:14:04,343 to see everything about the detail inside the scrolls. 258 00:14:05,260 --> 00:14:06,220 What I'm seeing right now, 259 00:14:06,220 --> 00:14:07,960 I mean these views are really lovely. 260 00:14:07,960 --> 00:14:09,800 I mean we can see so much structure, 261 00:14:09,800 --> 00:14:12,430 and you see these variations in density 262 00:14:12,430 --> 00:14:14,320 and that's where the curiosity comes in. 263 00:14:14,320 --> 00:14:16,221 You know, what's that, what's causing that? 264 00:14:16,221 --> 00:14:18,172 That's really, really interesting. 265 00:14:18,172 --> 00:14:20,550 (suspenseful music) 266 00:14:20,550 --> 00:14:22,220 The data that we take through X-ray 267 00:14:22,220 --> 00:14:24,970 gives us the platform to do all of the next steps 268 00:14:24,970 --> 00:14:26,600 that allow us to open it up. 269 00:14:26,600 --> 00:14:29,597 Those steps include finding where the layers 270 00:14:29,597 --> 00:14:33,022 that are inside really are in that data. 271 00:14:33,022 --> 00:14:35,941 We call that segmentation, and it just means that 272 00:14:35,941 --> 00:14:39,773 we're finding the layers and modeling what they look like 273 00:14:39,773 --> 00:14:42,100 so that we know exactly 274 00:14:42,100 --> 00:14:44,770 what the shape is that we're dealing with. 275 00:14:44,770 --> 00:14:47,010 That modeling step is actually not that easy 276 00:14:47,010 --> 00:14:48,602 because the layers, you can imagine, 277 00:14:48,602 --> 00:14:53,140 of a wrapped up scroll, or a book that's been crushed, 278 00:14:53,140 --> 00:14:55,690 those layers and their shapes are completely unpredictable. 279 00:14:55,690 --> 00:14:57,820 We don't know in advance what they're gonna look like 280 00:14:57,820 --> 00:14:59,290 until we take the scan, 281 00:14:59,290 --> 00:15:02,480 so we have a computer algorithm that we've built 282 00:15:02,480 --> 00:15:04,850 to be able to find every one of those layers, 283 00:15:04,850 --> 00:15:07,430 identify them, and then model them. 284 00:15:07,430 --> 00:15:09,220 They've done this all before, 285 00:15:09,220 --> 00:15:10,780 so they know understanding 286 00:15:10,780 --> 00:15:13,313 the volume and layout is just the beginning. 287 00:15:14,260 --> 00:15:17,470 Identifying text might be even more complex, 288 00:15:17,470 --> 00:15:19,433 depending on the type of ink used. 289 00:15:21,410 --> 00:15:23,460 In certain situations that's direct, 290 00:15:23,460 --> 00:15:25,560 because the ink is made from iron or lead 291 00:15:25,560 --> 00:15:27,060 and we can just see it. 292 00:15:27,060 --> 00:15:29,608 In other situations, it's more subtle than that 293 00:15:29,608 --> 00:15:31,995 because the ink has to be viewed as just a layer 294 00:15:31,995 --> 00:15:35,370 of material that's very similar to what it's written on, 295 00:15:35,370 --> 00:15:37,240 and so we have to find a way to go ahead 296 00:15:37,240 --> 00:15:38,593 and do that amplification. 297 00:15:39,940 --> 00:15:43,140 The X-ray can pinpoint microscopic shapes, 298 00:15:43,140 --> 00:15:44,993 potential layers of ink, 299 00:15:44,993 --> 00:15:47,114 but then artificial intelligence 300 00:15:47,114 --> 00:15:49,620 will likely provide confirmation 301 00:15:49,620 --> 00:15:51,573 based on a pre-determined reference. 302 00:15:53,060 --> 00:15:56,456 The fact that the ink deposits in a layer 303 00:15:56,456 --> 00:15:59,130 and you can measure the thickness of that layer. 304 00:15:59,130 --> 00:16:01,531 The fact that the ink soaks into the fibers 305 00:16:01,531 --> 00:16:03,803 of the material that you're writing on, 306 00:16:03,803 --> 00:16:06,457 all of those physical phenomenon 307 00:16:06,457 --> 00:16:09,890 end up being measurable by the imaging method 308 00:16:09,890 --> 00:16:13,700 that we're using, and then we can put them together using AI 309 00:16:13,700 --> 00:16:16,993 to amplify what we see with the naked eye. 310 00:16:19,020 --> 00:16:21,540 It's not unlike learning how to do face recognition 311 00:16:21,540 --> 00:16:24,394 by sampling a million different faces 312 00:16:24,394 --> 00:16:27,400 and pointing out to the algorithm, okay that's a face, 313 00:16:27,400 --> 00:16:28,860 that's a face, that's a face, 314 00:16:28,860 --> 00:16:30,615 and then learning from those examples 315 00:16:30,615 --> 00:16:33,343 what a face looks like in an unknown image. 316 00:16:35,230 --> 00:16:37,780 In the case of the Herculaneum scrolls, 317 00:16:37,780 --> 00:16:39,550 the fragments of scrolls damaged 318 00:16:39,550 --> 00:16:42,210 by earlier attempts to unroll them will hopefully 319 00:16:42,210 --> 00:16:44,570 provide Seales with that AI detail 320 00:16:44,570 --> 00:16:46,453 to identify the ancient text. 321 00:16:47,470 --> 00:16:48,817 So I think it's really ironic 322 00:16:48,817 --> 00:16:51,040 and maybe poetically beautiful 323 00:16:51,040 --> 00:16:54,738 that the sacrifice of those early scrolls 324 00:16:54,738 --> 00:16:58,720 might provide the reference that we exactly need 325 00:16:58,720 --> 00:17:01,520 to be able to read the ones that are still inaccessible. 326 00:17:03,168 --> 00:17:04,813 So what we're doing here is we're looking at 327 00:17:04,813 --> 00:17:09,340 the 140 gigabytes of data that we collected overnight. 328 00:17:09,340 --> 00:17:12,200 What we've done here is taken over about 10,000 images, 329 00:17:12,200 --> 00:17:14,830 10,000 radiographs that we're gonna use 330 00:17:14,830 --> 00:17:17,423 to compute the internal structure of the scroll. 331 00:17:18,340 --> 00:17:19,930 With just three days to work 332 00:17:19,930 --> 00:17:21,820 with these incredible relics, 333 00:17:21,820 --> 00:17:24,160 the team compiled a total of approximately 334 00:17:24,160 --> 00:17:27,450 1.3 terabytes of data thanks to 335 00:17:27,450 --> 00:17:30,793 a combined 164 hours in the scanners. 336 00:17:31,720 --> 00:17:34,560 Algorithms will analyze the data sets, 337 00:17:34,560 --> 00:17:36,950 but given the massive amount of data, 338 00:17:36,950 --> 00:17:39,130 and Seales' detailed methodology, 339 00:17:39,130 --> 00:17:40,815 it could take six months to a year 340 00:17:40,815 --> 00:17:43,853 to reach and release any final results. 341 00:17:44,830 --> 00:17:46,950 Well you know, my expectations are always different 342 00:17:46,950 --> 00:17:47,810 from everyone else's. 343 00:17:47,810 --> 00:17:49,850 I expect things to happen quickly, 344 00:17:49,850 --> 00:17:52,760 but what's probably going to happen is that 345 00:17:52,760 --> 00:17:55,640 we will not see the complete text right away 346 00:17:55,640 --> 00:17:58,179 because there are too many computational steps involved 347 00:17:58,179 --> 00:18:01,313 in being really deliberate and systematic about our work. 348 00:18:02,250 --> 00:18:05,199 But when dealing with 2,000 year old documents, 349 00:18:05,199 --> 00:18:09,142 what's a few months mean to learn the secrets held within? 350 00:18:09,142 --> 00:18:12,740 The prospect of non-invasively, 351 00:18:12,740 --> 00:18:17,430 virtually unwrapping these scrolls and reading their texts 352 00:18:17,430 --> 00:18:19,300 is a very exciting breakthrough, 353 00:18:19,300 --> 00:18:23,000 and scholars will potentially have access to a text 354 00:18:23,000 --> 00:18:26,533 that's complete, that's whole, and undamaged. 355 00:18:27,750 --> 00:18:31,530 If I could see one of our scrolls, 356 00:18:31,530 --> 00:18:34,253 seeing the writings, 357 00:18:36,150 --> 00:18:38,483 naturally I'd jump for joy. 358 00:18:40,390 --> 00:18:41,223 What else? 359 00:18:43,050 --> 00:18:45,470 Now that the digital data is collected, 360 00:18:45,470 --> 00:18:48,120 a little patience won't dampen that enthusiasm 361 00:18:48,120 --> 00:18:49,723 and technological promise. 362 00:18:50,930 --> 00:18:53,010 So recognizing the ravages of time 363 00:18:53,010 --> 00:18:56,630 on these ancient artifacts, the Naples National Library 364 00:18:56,630 --> 00:18:58,740 has already expanded its partnership 365 00:18:58,740 --> 00:19:00,283 with Seales and his team. 366 00:19:01,460 --> 00:19:03,390 People take a look around and they say, 367 00:19:03,390 --> 00:19:06,565 we might not have this material with us for always and ever. 368 00:19:06,565 --> 00:19:09,380 (sirens blare) 369 00:19:09,380 --> 00:19:13,320 Notre Dame can burn, and that threat 370 00:19:13,320 --> 00:19:17,230 makes us reevaluate the urgency with which we want to treat 371 00:19:17,230 --> 00:19:19,303 the valuable materials that we have. 372 00:19:20,840 --> 00:19:22,400 In the end, this work 373 00:19:22,400 --> 00:19:25,020 marks an evolution in archeology. 374 00:19:25,020 --> 00:19:29,110 Unobtrusively building a modern library of ancient knowledge 375 00:19:29,110 --> 00:19:30,963 without ever turning a page. 376 00:19:31,940 --> 00:19:34,270 When the Herculaneum scrolls were discovered 377 00:19:34,270 --> 00:19:38,230 250 years ago, we were in the Dark Ages of conservation. 378 00:19:38,230 --> 00:19:39,850 We experimented with these materials. 379 00:19:39,850 --> 00:19:43,703 We as a community, and some of them were destroyed. 380 00:19:43,703 --> 00:19:47,430 I think that these new techniques for imaging 381 00:19:47,430 --> 00:19:50,240 will give us more information than ever before 382 00:19:50,240 --> 00:19:52,770 to transform how conservation is done 383 00:19:52,770 --> 00:19:55,136 so that we don't make those mistakes again. 384 00:19:55,136 --> 00:19:57,719 (somber music)