Ancient Computer

Ancient Computer

PBS Airdate: April 3, 2013

TONY FREETH (Mathematician): If it hadn't been discovered when it was, in 1901, no one would possibly believe that it could exist, because it's so sophisticated.

ALEXANDER JONES (Historian of Ancient Astronomy): This mechanism would be remarkable, even if it was a less clever thing than it is.

NARRATOR: This is the story of one of the most extraordinary finds in history. This corroded bronze object is a machine that can look into the future. It was built 2,000 years ago, in ancient Greece.

YANIS BITSAKIS: Somebody, somewhere in ancient Greece, built an extraordinary machine that was actually a mechanical computer.

NARRATOR: A hundred years ago, a group of divers chanced upon a wreck, full of the largest horde of ancient Greek treasures ever found. Among the priceless ancient Greek bronze sculptures is another bronze object, no bigger than a modern laptop. It's known as the Antikythera Mechanism.

As a team of scientists tries tries to unravel the secrets of the Antikythera Mechanism, we're taken on a journey that charts the fall of one great ancient empire and the rise of another.

TONY FREETH: An ancient Greek scientist had done a truly remarkable thing. He'd found a way, using bronze gearwheels, to track the complex movements of the Moon and probably all the planets as well. It was a mechanism of truly staggering genius.

NARRATOR: This is the story of the world's first computer.

If it hadn't been for a storm, on the rocky Greek island of Antikythera, a hundred years ago, one of the most bewilderingly complex objects ever to emerge from the ancient world might never have been found.

After they'd sheltered from the storm, a team of sponge divers decided to try their luck underwater.

The sponge diver hadn't discovered a graveyard. He'd come upon a heap of marble and bronze sculptures. It was part of the biggest hoard of Greek treasure ever found. It had come from an overloaded Roman galley, sunk 2,000 years ago, as Rome's empire began to grab Greece's overseas colonies in the Mediterranean.

By accident, the divers had rescued some of ancient Greece's most beautiful artifacts. But among the bronze and marble statues was perhaps the most important object of all: Item 15087 in the Athens Museum. It had soon split into several badly corroded lumps of bronze. Then, remarkably, researchers noticed tiny gearwheels in the machine. Much later, the Antikythera Mechanism, to the amazement of scientists, would be revealed as, "The World's First Computer," built 2,000 years ago by a Greek genius.

MIKE EDMUNDS (Astronomer): You're just amazed by the quality of the workmanship. And then suddenly, you look, and you see there are tiny Greek characters engraved into the actual metal, itself. The shock they must have had when they first saw this and saw these gearwheels... They knew wooden gearwheels were used in Greek mills and so on, but nothing was known like these precision, metal-engineered gears.

NARRATOR: It was against the background of this Greek mystery that, in the year 2000, a team of international scientists was formed by astronomer, Professor Mike Edmunds, to investigate the puzzle. As a group, they were an odd mixture of astronomers, historians of science and mathematicians.

TONY FREETH: The Antikythera Mechanism is an incredible puzzle, probably one of the most fiendish puzzles in history. We had no confidence, ourselves, that we would actually be able to solve it.

NARRATOR: Tony Freeth, a mathematician, coordinated and led some of the team's major investigations.

TONY FREETH: So when we started, we thought we would try to put the basic information together. Maybe the objects of the wreck would help us piece together some clues. How old was it? Where did it come from? We had many, many questions.

NARRATOR: In 1976, another expedition to the Antikythera wreck by the famous French diver Jacques Cousteau, had given them their first clues. Cousteau discovered much cargo left after the initial find in 1900. There was more pottery, many more amphorae, some of the original timbers from the ship and bronze figurines. Cousteau believed it was a Roman galley.

Significantly, his divers brought up bronze and silver coins, an archaeologist's dream for dating a treasure site.

MIKE EDMUNDS: These are the coins that came from the expedition by Jacques Cousteau, in 1976?

PANAGIOTIS TSELEKAS (Coin Expert): Yes, a lot of silver and bronze coins.

MIKE EDMUNDS: Thirty-six silver coins, plus some bronze coins.

PANAGIOTIS TSELEKAS: Some bronze coins.

MIKE EDMUNDS: Okay. So this has a basket on it?

PANAGIOTIS TSELEKAS: This has a basket, introduced by the kings of Pergamon. Most of the coins were struck in Pergamon—around 30 of them—and the rest were struck in Ephesus.

MIKE EDMUNDS: But what do you think they can actually tell us about where the ship came from?

PANAGIOTIS TSELEKAS: They can tell us the ship came from Asia Minor.

MIKE EDMUNDS: So what date would this be?

PANAGIOTIS TSELEKAS: These are dated in the decade from 70 to 60, before Christ.

TONY FREETH: These coins, from the cargo, had given us our first clues to the likely date and route of the ship's last voyage.

MIKE EDMUNDS: So what were these Greek treasures and the strange Antikythera Mechanism doing on a Roman ship 2,000 years ago?

NARRATOR: Since the time of Homer, the Greeks had been great sailors, forging settlements and remote colonies in places such as Pergamon, in Asia Minor, and further north in the Black Sea. And everywhere the Greeks went, they left giant temples to the gods that protected them.

But by the time the galley sank, in the middle of the first century B.C., these far-flung Greek settlements had become vulnerable to a hostile new power in the Mediterranean: Rome.

TONY FREETH: But as we looked closer at the Antikythera cargo, we began to realize this wasn't a marauding Roman warship plundering Greek colonies, but something more unusual.

DIMITRIS KOURKOUMELIS (Archaeologist): It was quite a big ship, probably a huge ship, for this period, for these Roman times, one of the probably biggest trade ships. Only a few harbors can receive this kind of ship. That is probably Delos, Pergamon and Ephesus...that region; Rhodes.

TONY FREETH: Tell me about where these amphorae come from?

DIMITRIS KOURKOUMELIS: Look, these amphorae, these five amphorae, they're coming from Rhodes, from the island of Rhodes. These two of them, they're coming from the island of Kos. They were containing wine.

TONY FREETH: Tell me what you can say about the date of these amphorae and therefore the date of the shipwreck.

DIMITRIS KOURKOUMELIS: These amphorae, they are between 65 and 50 B.C.

TONY FREETH: These dates were almost identical to those we'd found on the coins from the wreck.

MIKE EDMUNDS: So, we believe that the galley started its journey somewhere in Greek colonies in Asia Minor—possibly Pergamon, possibly Ephesus—but somewhere along this coast. The ship would have sailed down, probably calling in at Kos, then on to Rhodes, where it loaded more amphorae and then set out, heavily overloaded, on the trade route back to Rome, where it met its fate in a storm at the island of Antikythera.

NARRATOR: Some of the ship's cargo had been dated, very accurately, to between 70 B.C. and 50 B.C. Perhaps this was now close to the vessel's last voyage.

But the group couldn't be entirely sure the bronze mechanism really was 2,000 years old. Perhaps it was a modern machine that had been dropped from a passing ship, by chance, coming to rest on the wreck the sponge divers had found. So the team decided to scrutinize previous work done on the machine.

MIKE EDMUNDS: The team were drawn to the work of Derek de Solla Price. He was an English-born physicist. He started work in the 1950s, first in Cambridge and later in Yale.

He was the first to really examine the pieces in very great detail. And he started doing radiographs and being able to see the insides of the mechanism. Suddenly, out of those radiographs, they realized there were 27 gears inside this thing. It was seriously complicated.

TONY FREETH: Price also was the first person to count the gear teeth with Karakalos and Karakalos' wife as well.

MIKE EDMUNDS: And they did it by drawing around the gears and, literally, just counting. So it wasn't surprising they didn't get it entirely accurate all the time.

NARRATOR: But all the two-dimensional X-rays of the gears were overlapping, making this task formidable. Price realized, if he could find the exact tooth count on a gearwheel, it might begin to unlock the mechanism.

TONY FREETH: So Price had identified a 127-tooth gear in the X-rays of the mechanism.

MIKE EDMUNDS: He also had the number 235. And these two numbers are very important in ancient Greek astronomy.

NARRATOR: Price wondered whether an ancient astronomer might be using the 127 tooth gearwheel to follow the movement of the moon. This was a revolutionary idea.

TONY FREETH: Price was beginning to have sleepless nights, worrying about the authenticity of the mechanism. If the ancient Greeks scientists could produce these gear systems 2,000 years ago, the whole history of Western technology would have to be rewritten.

NARRATOR: The team believed such sophistication was surely beyond the great achievements the ancient Greeks had made, more than two millennia earlier. They are regarded as some of the most creative people the world has ever known. Two and a half thousand years ago, they began a revolution in thinking, followed by technical advances comparable to those of the industrial revolution.

This peaked in the fifth century B.C., when Athens, the largest city-state, produced magnificent art and architecture that is still revered today. In nine years, they built the huge temple of the Parthenon, to their goddess Athena, on the sacred Acropolis rock that dominates Athens.

Public discussion of ideas and oratory led to larger public events like theater. This huge theater at Epidavros could hold 14,000 spectators, and it had superb acoustics. Drop a coin on the stage and it could be heard in the back row.

The ancient Greeks developed astronomy, which they treated as a branch of mathematics. They were able to plot how heavenly bodies moved in space and calculate their distances and know the geometry of their orbits.

Now, with the mysterious gearwheels, the team suspected ancient astronomers were trying to mechanize the movements of the sun, moon and planets. Could they put astronomy and complex mathematics into a device and program it to follow the motion of their closest neighbor, the moon?

ALEXANDER JONES: The phases of the moon were a fact of central importance for the ancient Greeks to plan ceremonies, such as the ones that were held in the Parthenon, up on the rock above me.

NARRATOR: The number 235 that Price had found was the mechanism's key to computing the cycles of the moon.

ALEXANDER JONES: The Greeks knew that from one new moon until the next was a time averaging about 29 and a half days. But that made a problem for their calendar, if they were going to have one with 12 months in every year. Because 12 times 29-and-a-half makes only 354 days, 11 days short of the solar year, the natural year. So the natural year, with its seasons, and the calendar year would quickly go out of sync.

But the Greeks also knew that 19 solar years almost exactly equals 235 lunar months. That means that if you have a cycle of 19 calendar years, then your calendar, in the long term, is going to stay perfectly in line with the seasons.

Through Price, the mechanism was beginning to yield one of its secrets.

NARRATOR: And on the back of the mechanism was the remains of an upper dial, with Price's 235 divisions representing the 19-year cycle. The Greeks called it the Metonic calendar. Why had these ancient astronomical numbers and a dial been built into the machine?

The phases of the moon were immensely useful to the ancients. These told them when to plant crops, when to fight battles, the timing of their religious festivals and whether to travel at night.

ALEXANDER JONES: The 127-toothed gear had given Price a clue to another one of its functions.

The ancient Greek astronomers realized that, while it takes about 29 and a half days for the moon to catch up again with the sun, it takes only 27 and a third days for it to get back to the same star in the sky. So, that's the length of time it takes for the moon to go once round the earth. If you do the arithmetic, what you find is that in that 19-year cycle, this means there are 254 orbits of the moon around the earth.

But that's a lot of teeth to put on a small gear, so what the designer did instead was to take half of that number. Half of 254 is 127. And he used other gears to multiply its effect up to 254. Price had discovered the 127-toothed gear—one of the other main functions of the mechanism—was a display of the moon's revolutions around the earth.

TONY FREETH: We now had two prime numbers, 19 and 127, which both had crucial functions. Were there more?

After 20 years of intense research, Derek Price thought he'd solved the puzzle of the mechanism, but he hadn't used up all the gears, particularly a large gear at the back, which he thought had 222 or 223 teeth. So, we became sure that Derek Price hadn't solved the puzzle of the mechanism.

We were desperate to see inside the Antikythera Mechanism. One day, I was looking at a science journal, and I saw this exquisite X-ray picture of a goldfish. Another picture I saw was of a locust, with all this fine detail of the internal structure. Could we use these techniques to look inside the fragments in three dimensions?

NARRATOR: Tony Freeth took a chance; he phoned a U.K. X-ray company, X-Tek Systems, now part of the Nikon group. Could he convince them of the importance of the mechanism?

ROGER HADLAND (X-ray Engineer): Well, to begin with, I wasn't interested. My colleagues told me it was some ancient bronze calcified lump. So, really, I determined to ring up Dr. Freeth and say, "I'm really sorry, but we're unable to do this, because we haven't got a system powerful enough to penetrate the sample." But after an hour's conversation, I was pretty much convinced this was something we had to do.

TONY FREETH: Roger decided to build a special prototype machine for us to X-ray the Antikythera Mechanism. It was an extraordinary decision.

ROGER HADLAND: The financial director was absolutely furious. He stormed out of our meeting, declaring I was going to bankrupt the company.

TONY FREETH: The Antikythera Mechanism is extremely fragile. So Roger agreed to take this machine—this eight-tonne machine—to Athens, to study the fragments. When they got to Athens, the police cleared the streets so that we could get the lorry through. It was an X-ray machine the size of a small van, basically, but weighing eight tonnes. With the help of three forklift trucks, we managed to shoehorn the X-ray machine through the entrance, into the basement of the museum.

You could describe it as a very high-tech machine to crack one of deepest and most extraordinary conundrums that have come from the ancient world. This thing is remarkable beyond belief.

The Antikythera Mechanism, itself, is extremely fragile, and since it was brought here, more than 100 years ago, it's never moved from this museum. So we had to bring the technology to the mechanism, rather than take the mechanism to the technology.

We'd made this huge team effort. We arrived there with the machine, without any confidence that we would find out anything new.

And the way that the technique works is that you put a fragment on a turntable, and you rotate the fragment in front of the detector and take maybe 3,000 different X-ray projections. A computer then puts this all altogether so you've got 3-D X-rays.

And when we saw the first image of Fragment A, it was absolutely amazing. It was like a new world really. It was almost like going into an unknown underwater world. Now, we actually had the data to enable us to really tackle the problem of how it worked, for the first time.

I decided to make a digital model of the Antikythera Mechanism, in order to try and understand better how it worked. What is remarkable is how much is crammed into such a small space. All the gears in the main fragment are packed together in layers, almost touching each other. We found 27 gears. Probably in the complete mechanism there were 50 or 60 gears.

It upsets all our ideas about what the ancient Greeks were capable of. It rewrites the history of technology. It tells us that things were going on in second century B.C. Greece, which we had no idea about.

Would our new data from the X-rays solve the puzzle of the largest gearwheel? It was broken and had either 222 or 223 teeth. Two-hundred-twenty-three was another prime number. The two prime numbers we already knew about were 19 and 127, in the gear that Price had found. Our plan was to follow this trail of prime numbers to see if they would unlock the astronomical secrets of the mechanism.

NARRATOR: But all this time, the international research team had a rival. Michael Wright had been working, in London, on the riddle of the mechanism, for 25 years.

MICHAEL WRIGHT (Mechanical Engineer): And slowly you get the metal to work round the instrument, until it gets nearly symmetrical, and then you can put it on there. Where it's tight, you...

NARRATOR: Michael Wright makes things, from musical instruments to a working model of the Antikythera Mechanism. He was formerly an expert on engineering at the Science Museum, in London.

Previous research showed the bronze fragments, with their gearwheels, had once been fitted into a wooden box that hadn't survived. So Wright had built his multi-geared machine into a box, powered by a handle on the side.

Michael Wright's own model included a radical addition to that of the international team's. Using earlier ideas by Price, he'd built a very complex and highly ingenious planetarium on the front of his mechanism.

MICHAEL WRIGHT: It's very obvious there is a lot of mechanism lost from the front of this, which is the big fragment. There was a pattern of pillars on this wheel. It had structure, some sort of structure, that revolved like a merry-go-round. And what I ended up with was models of the planets.

This is a model of the Greek cosmos, geocentric. You can think of this cover plate, in the middle, as representing the earth; and everything goes round it. The easy one to spot is the moon, because it's the fastest moving thing on the dial, and it's the front pointer.

The night sky was the ancient Greeks' television. What else were you going to look at, at night? People were much, much more aware of the sky. The calendar was organized according to the moon: official positions changed, debts became payable on the new moon. You had to have a calendar that, in some way, reconciled the year controlled by the sun, with the month controlled by the moon. They're tricky numbers, and they are built into the model.

NARRATOR: These numbers had to be translated into gearwheels with awkward teeth counts like 53 and 127. How did the ancient engineers do this? Michael Wright had the simple answer.

MICHAEL WRIGHT: There is nothing difficult in any number. Fifty-three is no harder than 54. We think that the Greek mechanic started with a prepared sheet of metal. He didn't have a hacksaw, so he had to cut it out with a hammer and chisel.

What I usually do, for almost any number, is to divide the wheel into six, to begin with, by stepping the radius round. Now, I ought to be fitting eight and a large fraction into each of those six divisions. So now, I'm going to attack it with a file and make teeth.

That's 53 teeth. That's ready to go into my Antikythera Mechanism. I suppose my Hellenistic workman friend would have taken about half an hour to make that.

NARRATOR: Meanwhile, as Tony Freeth created his own digital model, he was suspicious of Wright's use of gearwheels with odd numbers like 53 teeth.

TONY FREETH: Fifty-three, 53 teeth? Why 53 teeth? It seemed to have no function—odd prime number, no function. And I thought 54 was a much more likely number. You know, divisible by two, divisible by three, several times. So, in my model, I changed Wright's 53 to 54. And it proved to be a huge mistake.

NARRATOR: This tiny change would escalate into a major problem in the future. But there was a more pressing problem: none of the investigators working on the machine had ever been able to explain the function of the large wheel, the one with either 222 or 223 teeth, at the back of the mechanism.

TONY FREETH: We were desperate for more data. We knew that the surfaces of many of the fragments were covered in inscriptions, which were incredibly hard to read. And then I read about this technique, invented by a guy called Tom Malzbender, of Hewlett Packard. It was a brilliant technique for enhancing the surfaces, the surface details.

One of the ways this technique had been used is on paintings, for example, at the National Gallery in London, to look at the surfaces of paintings. Just looking at brushstrokes, the fingerprints, the essential form. This can reveal things that are under the surface.

Here, for example, is a painting by Franz Hals. And if I move the virtual light, I can see the brushstrokes on the painting. I can see how the painter has applied all the brushstrokes. And we thought it would be an absolutely brilliant technique to use on the Antikythera inscriptions.

I got on very well with Tom, and I then set about persuading him to come to Athens to use his technique on the Antikythera Mechanism.

Tom had an absolutely brilliant insight, which is that you can look at a surface by taking a series of still photographs, 2-D still photographs, with lighting from 50 different angles. Flashlights, arranged in a dome, flash, and a picture's taken with the lights at all these different angles. And this means a computer can put all this information together, and it can take away all the confusions of the surface coloration and the surface texturing and show the essential form of the surface.

You can actually look, in detail, at an inscription, say, and the clarity of the image leaps out at you.

NARRATOR: The new data from the inscriptions produced a significant breakthrough.

XENOPHON MOUSSAS: (Reading) "Sigma Kappa Gamma." It is there.

TONY FREETH: What is that? How much is there?

XENOPHON MOUSSAS: Two-twenty-three.

TONY FREETH: Two-twenty-three is there?

XENOPHON MOUSSAS: The line below.

TONY FREETH: Definitely 223?

XENOPHON MOUSSAS: Yes. Yes. 223.

NARRATOR: Did this mean the large gearwheel at the back of the mechanism actually had 223 teeth? If so what was its function?

Then, with a chance discovery in the museum, the mystery began to unravel.

MARY ZAFEIROPOULOU (Senior Archaeologist): I went to the stores of the bronze collections, and I start looking at the places where the Antikythera were. I found a tray, and there were eight boxes. And then, I was looking around and have seen some other small fragments in the boxes. Altogether we have 82 fragments.

MIKE EDMUNDS: One of the new fragments, in particular, stood out. Mary had labeled it as "Fragment F," but it also appeared to contain part of a curved dial.

NARRATOR: Measuring only a few centimeters, Fragment F would turn out to be the key to the big wheel and the entire mechanism.

A couple of weeks later, Tony Freeth started to examine the 3-D X-rays taken of Fragment F.

TONY FREETH: I suddenly realized, with mounting excitement, that Fragment F formed a new scale on the lower back dial on the Mechanism. In order to understand this new four-turn spiral dial, I wanted to count the scale divisions round the whole dial, make an estimate of them. So I started to enter all the data into an advanced computer program. And the results seemed to come out as 220 to 225. So I became sure that it must be the magic number 223.

We found the importance of the number 223 at the British Museum, in London. Three centuries before the Golden Age of Athens, around 700 B.C., Babylonian astronomers had made the breakthrough. Over hundreds of years they'd written thousands of astronomical tablets, many recording the huge significance of the number 223. The Babylonians called it the 18-year period.

John, tell me what all these tablets are here?

JOHN STEELE (Historian of Babylonian Astronomy): We have maybe 3- or 4,000 astronomical tablets, ranging from reports sent to the king by scholars, advising him on astronomical matters, to how to interpret omens and what they meant for his kingship.

TONY FREETH: Tell me about the 18-year period.

JOHN STEELE: So, the 18-year period describes a cycle of 223 months used to predict eclipses. This is what, today, we would call the Saros cycle. By looking at past observations, they saw that after 223 months, the 18-year cycle, eclipses of the same kind of say lunar eclipses, would repeat, with very similar appearances.

TONY FREETH: Tell me, John, how the kings of Babylon reacted to an eclipse prediction?

JOHN STEELE: A substitute, usually a criminal or someone like that, would be officially appointed to the throne, and the real king would officially abdicate. This is an example of one of the tablets, and, again, it's a letter describing this substitute king ritual. It mentions, down here, the substitute king ascended the throne and took these bad omens, these signs, onto himself. Once they were deemed to have done their worst, the substitute king would be killed, and the real king would come back to the throne unharmed, unscathed by what had happened.

TONY FREETH: Since the substitute king was killed, the omens were clearly correct then.

JOHN STEELE: Indeed. Of course.

NARRATOR: This solved the mystery of the large gear at the back. It must have 223 teeth to turn the pointer of the 223-month Saros dial. The team had found something remarkable. They'd uncovered a machine that could look into the future. It could predict eclipses.

TONY FREETH: What we realized was that the ancient Greeks had built a machine to predict the future. It was an extraordinary idea that you could take scientific theories of the time and mechanize them to see what their outputs would be, many decades hence. It was essentially the first time that the human race had created a computer.

NARRATOR: The gearwheels in the mechanism programmed the computer, but where was the output data displayed? The clues were in Fragment F.

TONY FREETH: When you first start looking at the X-rays from Fragment F, there's nothing much there. And then the scales emerge, as you go down through the layers. And not only scales, but you see these little scale divisions, blocks of characters here? These looked to me a little bit like Egyptian hieroglyphs, so I called them "glyphs."

The glyphs must be the eclipse predictions. I soon realized that the first letter here is a Sigma: Sigma standing for the Greek letter S, standing for Selene, the Goddess of the Moon. That must indicate a lunar eclipse.

I next realized the letter Eta, H in the Greek alphabet, must stand for Helios, the sun. So this must indicate a solar eclipse.

Now, what became really tough was to try to decode the next symbol, the sort of anchor-like symbol. And this took me a long time. By chance I found a book on Greek horoscopes. Amongst a whole mass of symbols, I found this symbol in the document. It was short for the word "hora," meaning hour. So what that told us was that, not only did this mechanism predict eclipses, but it predicted the hour of the eclipse as well.

As the hand sweeps along the scale, here, it's just reaching a lunar eclipse. Here, then, is a solar eclipse in this month, followed by a lunar eclipse in that month.

Our research group pooled our resources in a discussion on the internet. And Yanis Bitsakis, in Athens, found the phrase, "the color is black," in the eclipse inscriptions. Alexander Jones, in Toronto, found this really exciting, and he then discovered the phrase, "the color is fire red."

ALEXANDER JONES: We were discovering a very sophisticated machine that not only was predicting eclipses, decades in advance, and what time of the day or night they were happening, but even the direction the shadow was going to cross and what color the eclipse, sun or moon, was going to be.

we're watching a total eclipse of the moon, in Athens.

For the ancient astronomers, eclipses had a special significance, but for most Greeks, eclipses could have a much more dire and ominous significance.

NARRATOR: In 413 B.C., a lunar eclipse led to a fatal maritime disaster for the Athenian fleet at Syracuse. Athens was engaged in a long war with Corinth and its colony Syracuse, in Sicily. More than 130 Athenian triremes and 130 transport ships were blockading the harbor at Syracuse. And on the night of 27th August, 413 B.C., there was a total lunar eclipse. Nicias, the superstitious Athenian admiral of the fleet, consulted a soothsayer on board, who said the red eclipse of the Moon was a bad omen. The fleet should not put to sea for "thrice times nine" days. In the ensuing battle Nicias lost half his ships from arrows fired from the shore.

The computer predicted eclipses through the Saros dial on the lower back of the mechanism. It was dependent upon gearwheels following the repeating cycles of the moon. But there was a new puzzle. The team knew that the moon moves round the earth in an elliptical motion. That means it moves fastest when it's closest to Earth and slower when it's furthest away. How could the mechanism's designer possibly make gears that tracked this fluctuating path of the moon?

TONY FREETH: Michael Wright made an absolutely brilliant observation from his X-rays. He discovered that one of the gears at the back of the Mechanism, mounted on this large 223-tooth gear, had a pin in it.

MICHAEL WRIGHT: This is the pin and slot mechanism. There's a wheel in the back of the instrument. You can see with the naked eye that it's got some kind of slot in it. The next thing you see is this round ghost in the slot. There is a pin.

TONY FREETH: Now you might think, "Well, they'll just turn together, and that's a completely useless device." But he made this other very critical discovery: he discovered that the gear with the pin turns on a slightly different axis than the gear with the slot. So this mechanical device induces a variability in the motion of one of the gears.

NARRATOR: And, amazingly, the pin-and-slot device, built into the mechanism, plotted the variable motion of the moon.

MICHAEL WRIGHT: So you get a variable motion, which I can show you best in the model.

When the wheels are like this and the pin is at the inside of the slot, the driven wheel on top is going fast. When we come round here, and the pin has moved to the outside of the slot, the driven wheel is going slow. And this is modeling the way that the moon's speed in the sky actually varies.

NARRATOR: But when Wright originally found the pin-and-slot device, he failed to understand what it was for.

TONY FREETH: And he threw the idea away. It's a tragedy for him in a way, that he had this absolutely, the most brilliant observation in the history of the mechanism...

MICHAEL WRIGHT: I world's know what it was doing there. I even came to wonder whether it was a sort of mechanical fossil.

NARRATOR: But there was yet another lunar complication the machine had to deal with.

TONY FREETH: The ancient astronomers were fascinated with the motions of the moon, and nothing is easy about the motions of the moon. It's very, very complicated stuff.

In modern terms, we know the moon goes round in an elliptical orbit. But that ellipse isn't stationary; it rotates round in a...very slowly in a period of about nine years.

Based on cycles I knew were part of the mechanism, the Metonic and Saros cycles, I deduced the ancient Greeks had calculated this annual rate of rotation to be 0.112579655 to nine places of decimals.

I thought, "Well, surely the mechanism can't calculate to this degree of precision. That would be virtually impossible!"

I puzzled over how the designer could have possibly built gears to track the rotation of the moon's ellipse once every nine years.

In my model, I'd driven the large gear...223 tooth gear that helped track the moon's movements, with a 27-toothed gear. I'd changed Wright's 53-tooth gear to 54 teeth, exactly double the 27 of my input gear.

Then I had my own eureka moment. I was on the plane to Athens. I was playing around with the figures. I knew the input gear had 27 teeth. So I put that into my calculator and calculated the result. And, to my deep disappointment, it was too big.

So I thought, "Maybe it's only 26 teeth." So I put that into the calculator, and it was too small. Now, I was a mathematician, and mathematicians often have crazy ideas, so I tried putting in a gear with 26 and a half teeth. And I pressed the key on the calculator and the result was 0.112579655: exactly the right answer to nine places of decimals.

It just hit me like a thunderbolt! Twice 26-and-a-half is 53. Michael Wright had been correct about the 53-tooth gear.

The riddle of the 53-tooth gear had been solved, at last. It turns the large 223 tooth gear with just the right nine-year rotation, so that the pin and slot exactly models the ancient Greek theory of the moon's variable motion.

We'd followed the trail of clues in the prime numbers. The numbers 19, 127, 223 and then, finally, 53, to understand how the mechanism worked. And it was just an amazing moment, when everything came together.

We knew now what all the numbers in the tooth counts were, which had been a complete mystery before and not understood. And it was just quite an incredible moment.

NARRATOR: The team's next quest, however, was just as formidable. Who'd invented this extraordinary machine 2,000 years ago? This investigation would throw up many more surprises.

MIKE EDMUNDS: We thought the answer to the question, "Who'd made the mechanism?" might lie inside the mechanism, itself.

NARRATOR: Tony Freeth, in London, and Alexander Jones, on the other side of the Atlantic, were deciphering Greek month names on the Metonic upper back spiral. Then the breakthrough happened.

ALEXANDER JONES: Every Greek city-state had its own distinctive calendar. Four of the month names stood out as being really quite rare. There was one called "Lanatropios," another, called "Dodekateus," a third, called "Pysdreus," and a fourth one, called "Phoinikaios," which was the first month of the calendar.

I realized that these four months belonged to the calendar of ancient Corinth. And so they had to be coming from either Corinth, itself, or from one of the colonies that Corinth had founded, for example Syracuse, over the sea, in Sicily.

NARRATOR: In the third and fourth centuries B.C., Syracuse was the second largest city-state in the entire Greek world. Founded centuries before, by poor Corinthian immigrants from the Greek mainland, it had prospered remarkably.

ALEXANDER JONES: Significantly, Syracuse was the home of the most brilliant of all the Greek mathematicians and engineers, Archimedes.

MIKE EDMUNDS: As an astronomer, Archimedes determined the distance to the moon. As a mathematician, he showed how to calculate the volume of a sphere and how to calculate that fundamental number, pi.

TONY FREETH: We believed that only a mathematician of Archimedes' status could have designed the Antikythera Mechanism. As a brilliant inventor, he'd designed screw devices to lift water. And he'd designed machines with grappling hooks that could grab enemy ships out of the water.

NARRATOR: Archimedes lived in Syracuse in the third century B.C. At that time, Rome was challenging the power of Greek cities in southern Italy. If rich Syracuse could be taken, all of Sicily would come under Roman control.

Led by the Roman General, Marcus Claudius Marcellus, Roman legions laid siege to Syracuse in 214 B.C. Archimedes is believed to have designed cranes to pull Roman ships out of the water. Then, after two years of siege, by trickery, Roman soldiers got inside the city. General Marcellus gave orders that the city should be sacked but Archimedes' life be spared

But, according to the historian Plutarch, a Roman soldier came upon an old man drawing circles in the dust. When he refused to obey an order, the legionnaire ran Archimedes through with his sword.

Syracuse was stripped and its treasures taken to Rome. Just two valuable objects were personally taken by General Marcellus. He states they were machines belonging to Archimedes. These, the team believes, might be early versions of the Antikythera Mechanism.

One hundred fifty years later, in Rome, the formidable orator and consul, Cicero, writes of a sighting of one of Archimedes' machines in the house of Marcus Marcellus, grandson of the victorious General Marcellus.Cicero writes:

CICERO VOICE: Archimedes had thought out a way to represent accurately, by a single device, those various and divergent movements of the five planets with their different rates of speed. Thus the same eclipse of the sun would happen on the globe, as it would actually happen..."

NARRATOR: Were the rotating planets Cicero wrote about 2,000 years ago, the final clues to the construction of the mechanism? Could Michael Wright's complex planetarium on the front of the mechanism be simplified to match the design genius of the original?

ALEXANDER JONES: And I'm going to start off by passing around some pretty pictures, and then I'll explain why I'm passing them. These are medieval pictures, showing the Cosmos, but the way that it's shown is an ancient Greek way. What you have is the earth at the center, and you have a set of rings, which are supposed to be spherical shells in which each of the planets going from the moon up to Saturn are. The front display must have been a picture, like these medieval ones, that showed the Cosmos in cross-section.

TONY FREETH: Now, I began to think, "How can you mechanize these planets so it exactly predicts their positions in a way that's simple and not overly complex?"

Mars is what I started with. You'll notice it's got four gears. It's got a pin and slot and it looks, in principle, almost identical to the moon mechanism. And you finally end up with all the planetary mechanisms in here. I was just amazed. They all fit to create the cosmos on the Antikythera Mechanism, just like on Michael Wright's planetarium model, but much, much simpler.

That coincides exactly with Alex's picture of the cosmos.

I'd love to take credit for discovering these things, but I think I was rediscovering what the ancient Greeks did.

MIKE EDMUNDS: But all this leaves a major unanswered question: "What happened to the brilliant Greek technology that produced the world's first computer? Why was it never developed? Why was it lost from the western world?"

NARRATOR: As first the Greek world declined, followed by the collapse of the Roman Empire, historians believe Greek scientific texts were passed East. By the fourth century A.D., information on the mechanism perhaps went first to the Byzantine world and then to Arab scholars.

Michael Wright has a clue suggesting some of the mechanism's Greek technology would later become available to Islamic science.

MICHAEL WRIGHT: In 1983, a man came into the museum. He was a collector of astrolabes. He bought this from a dealer, we think in the Lebanon. We believe we can date this instrument to about 520 A.D. That makes it the second oldest geared instrument that we know of, after the Antikythera Mechanism. The gears in the back connect to this wheel, which shows the phase of the moon. That's new moon, waxing crescent.

MIKE EDMUNDS: So it's likely that the ancient Greek knowledge of gearing was kept alive in the Byzantine world, then, by the Arabs. It was reintroduced into Europe in the 13th century, when the Arab Moors came up through Spain.

NARRATOR: Then, during the Renaissance, in the 14th century, highly sophisticated gear trains suddenly appeared in clocks all over central Europe. They all used the complex gears found in the Antikythera Mechanism.

The original mechanisms coming from Archimedes' workshop are likely to have been much larger. But as the Greek engineers grew more confident, over several generations, they were able to minimize their technology and bring it down to the size of a box. And that box was almost certainly the most prized object on the Antikythera treasure ship.

TONY FREETH: The Antikythera Mechanism was small, light and portable. They'd managed to cram nearly all their knowledge of astronomy into this small-geared device. It was "the theory of nearly everything" in a box, very similar to today's modern laptop computer.

ALEXANDER JONES: Here, we believe, is the complete and intricate machine. On its rear face, Greek scientists of 2,000 years ago fashioned a computer mechanism that displayed a calendar that followed the moon, that predicted eclipses. While on the front they'd reproduced the universe, as they understood it, with the five planets, the sun and the moon, performing the complicated steps of their dance through the heavens. Here was Greek genius at its height, the great and divine cosmos, represented through mechanism, by scientists who wished to show there was no mathematical challenge beyond their abilities.

TONY FREETH: We know that this society was the birthplace of the art, architecture and culture that is the foundation of our modern world. Now, we also know it was the cradle of advanced technology.

NARRATOR: But if it hadn't been for two storms in the Mediterranean, we might never have known about this mechanical wonder. The first storm, around 70 B.C. sunk an overloaded Roman trading ship, carrying the precious mechanism. Then, in 1900, another storm drove a team of sponge divers to shelter off the Island of Antikythera. Without these two events, the most important scientific discovery to emerge from Ancient Greece might have been lost forever.