by Decoding the Heavens- Solving the Mystery of the World's First Computer (retail) (epub)
Hadland assigned almost all of his R&D staff to work on the new machine. They needed to double the voltage of X-Tek’s existing microfocus X-ray source, so they decided to place two of these sources back to back, making one positive and one negative, so that the resulting electron beam would be accelerated into the tungsten target with twice the energy.
It sounded simple, but there was a lot of work to be done. The company already had a negatively charged X-ray source that would do the job, but positively charged ones are much more difficult to make. Seven engineers, including Hadland himself, would have to develop a new electron gun, a new high-voltage generator, and a new controller that could operate each of the generators both individually and in tandem. Another three people would work to update the computer software needed to reconstruct three-dimensional images from the collected data.
With September ticking ever closer, Tony Freeth waited impatiently. Since his team had made public its plans to study the fragments, Michael Wright had started publishing his work. Slowly but surely, Wright was solving one part of the mechanism’s structure after another. Freeth was convinced that Wright had been doing everything he could to block his team’s access to the fragments. And he was seriously worried that by the time he got to throw all of his technology at the problem, there wouldn’t be anything left to discover.
He did have one ace up his sleeve, however. A member of staff at the museum, Mairi Zafeiropoulou, had been gathering together the Antikythera fragments for the team to image them. The holdings of the National Museum had still never been fully catalogued, so the precise contents of its storerooms were always something of a mystery, even to the staff. The largest Antikythera fragments – A, B and C – were easy enough to find, because they were on display. But Zafeiropoulou had no idea where Fragment D was. After much searching, armed with Price’s photos, she eventually found it in an unmarked wooden case. She also located lots of tiny scraps, which seemed to have been scraped off the larger fragments as they were cleaned, which might contain vital letters from the mechanism’s inscription.
She also found Fragment E, which had been discovered in the stores by the curator Petros Kalligas in 1976. And then, wonderfully, lying in a box alongside an old piece of wood, she found a green, corroded lump that didn’t match any of the old photographs. It was a few centimetres across and completely untreated, covered in limestone with worm tracks still winding over its surface. Underneath it was just possible to make out portions of concentric circles – matching those on the back dials of the Antikythera mechanism. She called the new piece Fragment F. Imagining it would provide essential clues that no previous researcher had seen.
September arrived, and X-Tek’s machine still lay in pieces on the workshop floor. But Tom Malzbender, along with colleagues Dan Gelb and Bill Ambrisco, flew across the sea from California with their flashbulb dome packed in a crate.
Once at the National Museum, they were taken down to a basement to do the imaging. It was a bit of a shock after the well-equipped labs of Hewlett-Packard – the room was basic and bare, with antiquated wiring running exposed around the wall, and the lack of air-conditioning made it stifling. Malzbender was the only one allowed in the room with the Antikythera fragments (watched over by a museum official), so he rolled down the shutters and turned off the lights, while Tony Freeth and the others waited eagerly next door. Malzbender felt privileged to be alone with the ancient mechanism, but for once he was just a little nervous. He had travelled a long way to make this happen. It was no time for his equipment to go wrong.
Over the next five days Malzbender took more than 4,000 images of 82 fragments; so many, in fact, that it nearly wore out his camera. His dome hung vertically, so that when each piece was in place, he had to stoop to line up the shot before pressing the shutter to release 50 flashes. Then he downloaded the images on to a memory stick and walked to the next room where Dan Gelb had his laptop and the software needed to reconstruct the images. At first, the composite pictures appeared on the computer screen as high-resolution colour photos; a series of sharp splotches of dull green and beige.
Tony Freeth hovered in the background, leaning over Gelb’s shoulder as he coaxed the software to work its magic. With a few keyboard strokes, Gelb took the fragments out of this world and into one not bound by the laws of physics, transforming them into beautiful drops of mercury that hung in the black background of space and glinted silver in the distant Sun. These mercury asteroids were covered in bold, clear, impossible writing: a message through time from one civilisation to another.
At the end of the week Malzbender and his colleagues packed up to go home, but Roger Hadland and his team were still working late nights in Tring. Fortunately, Freeth had managed to persuade the reluctant museum staff to give them another slot with the fragments in October. So the X-Tek team burrowed themselves away in the development lab, while customers’ requests fell on deaf ears and one order after another was lost. If the Antikythera project didn’t work, Hadland thought, he wouldn’t have much of a company to go back to.
With just a week left until the final deadline, he was starting to panic. The high-voltage generator wasn’t working. It was meant to conjure a voltage of 225 kV – nearly a quarter of million volts – but the circuit was only registering a measly tenth of that. Eventually, in desperation, the engineers unhooked the cable from its X-ray source and switched it on to inspect the generator’s workings more closely. They weren’t concerned about the disconnected cable – 20-odd thousand volts is low for equipment like this, and should be quite safe in the open air.
Famous last words. There followed an almighty crack, like a pistol shot, as a huge spark, almost a foot long, leapt from the cable’s end. Then a shocked silence as each man present briefly considered his own mortality – if it had hit someone, such a huge electrical discharge could have killed them. But after a few seconds, Hadland started to grin. That proved the generator was working fine. Someone must have put the wrong resistor in the circuit, causing the voltage to read ten times too low.
Unfortunately, the explosion had just destroyed the computers they were using to run the machine. Then the second deadline came, as did the lorry booked to carry the machine to Athens. The X-Tek team worked on and the Greek lorry driver had to wait in Tring for two days, sleeping in his cab while they finished rebuilding the system.
Late in the evening on the second day, they were finished. The huge lead-lined cabinet contained an X-ray source that was smaller and more powerful than any other in the world. Hadland called it BladeRunner, after the turbine blades he hoped it would one day be used to test.
Pride soon gave way to exhaustion. It took a couple of hours to pack up the machine, literally building a crate around the cabinet with timber and plywood, then they had to load it on to the lorry in the early hours of the morning. The cabinet alone weighed nearly nine tons, and the forklift truck they had hired couldn’t take the weight without tipping over. Putting another ton of lead on the back of the forklift solved that problem, but then it didn’t have enough power to move forwards, so they had to use a second forklift to push it along. The various X-ray sources and detectors added another couple of tons to the lorry’s load, and finally it set off for Athens.
It would take the driver five days to get there, heading slowly south through Italy to the port of Brindisi, before rolling on to a ferry bound for Greece. Meanwhile Roger Hadland and Tony Freeth jumped on a plane, meeting the driver just outside Athens. The lorry was 20 metres long – perhaps not quite as big as the Antikythera ship, but still, it became apparent that it wasn’t going to fit through the narrow streets of the city. Its precious cargo would have to be transferred to a smaller truck. It took them all day to find a company with forklifts that were up to the job, followed by a second late-night loading exercise.
The next day a police escort ensured that they reached the National Archaeological Musuem without further mishap. The trees outside the pretty, white building left dancing shadows in th
e early morning sun, but Hadland’s attention was focused solely on the workers he was directing; he felt like a nervous parent as the plastic-wrapped crate was inched down from its truck and through the museum’s black iron gate. By the time it reached a gentle ramp into the building itself, the sun was high overhead. It took the rest of the day and three forklifts chained together – two pulling and one pushing – to shift the enormously heavy box up the slope. Hadland wondered if it had been this much work building the pyramids.
The rest of the week was spent getting everything wired up, taking some conventional X-ray images, and calibrating the system. Everything was still at the prototype stage, so a mass of scrappy wiring connected the equipment in the cabinet to a couple of industrial PCs, sat on packing cases for tables. The rest of the room was filled with spare generators, X-ray sources, repair kits, cabling – everything Hadland could think of to bring, just in case any of the components failed.
One of the first pieces of the Antikythera mechanism to be examined was the almost-lost Fragment D with its single cogwheel. It should provide a simple test of the CT. They started up the equipment and a cone of invisible X-rays splayed out from the source, through the fragment and on to the detector behind. Rather than risk damaging the fragment by stopping and starting for each image, it was fixed to a slowly rotating turntable. It inched around once in an hour like a clock’s minute hand, while the computer recorded ten images for every degree – more than 3,000 over the entire 360-degree rotation.
When the circle was complete, it took about an hour for the computer to assemble all of the data into a three-dimensional volume. Finally, X-Tek software engineer Andrew Ramsey pulled up the images on his computer screen, with a crowd of people including Freeth, Hadland, Moussas and various curious museum staff jostling behind him. One face missing was Mike Edmunds, who had stayed in Cardiff.
There was silence. The surface images from Malzbender’s team had been stunning, but everyone knew that for the project to be a success they needed to see inside; they needed to see the internal workings. Ramsey scrolled down through the depth of the fragment. At first all they could see was a blur, but then a crackling sharp gearwheel emerged from the fuzz, as if being hauled up out of grey sand. It was better than any of them had dared hope. The letters ‘ME’ had been scratched into the side of the wheel. It was like a signal from the past, an ‘I WOZ ERE’ from 2,000 years ago. Suddenly, they felt a direct, almost physical connection with this ancient machine, and with whoever had carved those letters so long ago.
Then Freeth started to laugh. ‘Somebody e-mail Mike and tell him we’ve found a gearwheel with his initials on!’
After that they worked through the fragments one by one, working up to the biggest Fragment A, with its characteristic four-spoked wheel, learning how to get the best results out of the machine as they went, and all the time getting more and more excited by the details they found. It became clear that there was hardly any intact bronze left inside the fragments – something that Michael Wright could have told them if they had asked, after his experience 15 years earlier when the delicate zodiac scale had snapped in his hands. But at that moment Wright was alone in his workshop preparing for his own trip to Athens, desperately crafting the finishing touches on his model of the mechanism.
The state of the fragments meant that there was less dense metallic material inside than the team had thought, making them much more transparent to radiation. That meant the pictures were even sharper than the researchers had hoped, with resolution in some places down to just a few thousandths of a millimetre. Every surviving part of the mechanism – every tooth, peg, shaft and pin – could be seen in pure, crisp, breathtaking detail.
The team worked as long as the museum staff would let them each day (although that was still nothing to how hard Michael Wright had laboured in that cramped darkroom, Eleni Magkou thought, as she watched the team’s colourful comings and goings). Each evening the group would meet in the lobby of their hotel and they’d go for dinner in jubilant mood, raising their glasses to the cry of ‘More gears!’
As an engineer, Roger Hadland was especially dumbstruck by how familiar the components inside the mechanism were. It was like opening a textbook that displayed many of the techniques and arrangements of gearwheels that he saw around him every day, in automatic devices from car windows to camera shutters. The other surprise was that the CT revealed lots of hidden inscriptions, which had been invisible even to Malzbender’s team, because they were completely covered by limestone or the products of corrosion. The hollows of the engraved letters were less dense than the corroded metal around them, so as Andrew Ramsey scrolled just beneath the surface, even the tiniest inscriptions showed up beautifully.
Progress was steady, although they had trouble finding enough computer hard drives to store all of the data. Each scan incorporated 3,000 images of 12 megabytes each, adding up to 36 gigabytes per scan. Overall the team produced nearly a terabyte of data. That’s enough to fill a library, and to print it all out on paper would require pulping a small forest of trees.
Time was also in short supply. The National Museum had given the team a strict slot of two weeks to finish their imaging and get out of the building. As the finish date drew closer, they had their data on the larger fragments, but realised that there wouldn’t be enough time to image all of the smaller scraps. These might not even all come from the mechanism, but without putting them in the scanner it was impossible to know which ones contained a crucial clue. Xenophon Moussas disappeared off to a little art shop down the road and came back with several polystyrene cylinders, out of which he gouged cradles that fitted the pieces snugly. If they fixed the cylinder to the turntable, they could scan several fragments at once. It was clumsy, but it worked; they got the images they needed just before they were thrown out. As a determined Michael Wright completed his Athens lecture and carried his precious model home to London, BladeRunner began its slow lorry ride back to Tring.
9
A Stunning Idea
It’s an absolutely unbelievably stunning and sophisticated idea. I don’t know how they thought of it. We’re just following in the tracks of the ancient Greeks.
— TONY FREETH
THE CENTRE FOR History and Palaeography is on the first floor of a grey, stone office building, squeezed into an Athens back street just a couple of minutes walk from the old Roman marketplace and the Tower of the Winds. It doesn’t look like much from the outside, but within its doors the shelves overflow with piles of old books and manuscripts, and humidifiers fill the air with moisture, so that the delicate paper doesn’t shrink or curl.
In the corner a dried goatskin is stretched taut over a wooden frame – a demonstration for visitors of how parchment is made – and on one wall a series of posters explains how the physicist Yanis Bitsakis decodes ancient palimpsests. These are old texts on parchment that has been washed and scraped clean, so that a new text can be written on top – a common practice in the Middle Ages when parchment was a scarce raw material. Sometimes these scraped-away traces are all that’s left of important ancient documents. A tenth-century copy of several of Archimedes’ mathematical writings, for example, was recently found beneath the words of a medieval prayerbook.
Yanis Bitsakis uses his technical expertise to uncover stories where others wouldn’t even bother to look. To reveal the hidden layers of ink on a page, for example, he photographs it with a camera that detects different colours – or wavelengths – of light. Different inks reflect characteristic combinations of wavelengths, meaning that when he displays the images on a computer screen he can turn the different colours up and down – even those not normally visible, such as ultraviolet and infrared – looking for the perfect mix to muffle the overlaid writing and bring out the lost messages beneath.
Now he has a new subject. Tony Freeth has hired Bitsakis to work through the thousands of computer images of the Antikythera mechanism from Hadland’s BladeRunner machine and Malzbender’s flashbulb dom
e. With his long dark hair tucked behind his ears, the physicist checks each separate slice of the CT data, and painstakingly tries out all of the lighting conditions he can think of on Malzbender’s photos, looking for every last trace of visible lettering. Then he passes the most useful images on to the epigrapher Agamemnon Tselikas, the director of the centre, who does his best to read them.
Agamemnon Tselikas (Memos to his friends) is a large but gentle bear of a man, excellent company and fond of life’s pleasures. But for three months, he spends his evenings alone with images of the Antikythera mechanism, working in silence every night from around eleven until the early hours of the morning. The letters are tiny, some less than two millimetres high, and all squashed together without spaces with no clue as to where each word starts and finishes. He drinks thick, black Greek coffee as he scrolls from one slide to another, trying to get inside the head of the machine’s maker, so that he can decipher his words.
The results start to flow almost immediately. The very first inscription that the two men read – on the back of the mechanism quite near the top – is ‘ELIKI’, which means ‘spiral’. It appears within the phrase: ‘. . . the spiral divided into 235 sections . . .’ In London, Tony Freeth nearly jumps out of his armchair when Bitsakis phones to tell him. The snatch of text – clearer than any that Derek de Solla Price was able to translate – shows that the inscriptions do contain operating instructions. And it confirms what Michael Wright has been saying about his measurements of the upper back dial. Here is independent evidence that the dial was a continuous spiral rather than a series of concentric circles, and that it was divided into the 235 synodic months of the 19-year lunisolar cycle. Whereas the front dial gave the day of the year, this back dial allowed the user to track months and years over much longer periods of time.
