Genius at the bottom of the sea

Somewhere between 70 and 60 BCE, a Roman cargo ship was sailing from somewhere on the coast of Anatolia in modern Turkey to Rome. We don’t know where on Anatolia it started from. But wherever it was, if you look at a map, you’ll see that the ship would first have had to thread its way through the innumerable islands of Greece, strewn across the Aegean Sea like so many pearls. Then it would have to round the Pelepponese peninsula and cross the Ionian Sea. After that, through the Strait of Messina that separates the boot of Italy from Sicily. Finally, some 800km northwest through the Tyrrhenian Sea, homeward bound into Rome.

The point being, this was a long and perilous voyage. This particular ship didn’t make it. Probably hit by a storm, it sank off the coast of the little island of Antikythera, about halfway between Crete and the long fingers of the Pelepponese. Ironically, this island was the last piece of land it would have had to negotiate before Messina.

In 1900, a boat filled with sponge divers from the island of Symi was in the same waters. A storm hit them too, and while this boat didn’t sink, the divers were forced to take shelter on Antikythera. The next day, with the storm having abated, they decided not to head home to Symi—some 400km east—right away. Instead, they went diving. To their astonishment, they found this 2000 year-old shipwreck about 55m below the surface, off Antikythera’s Point Glyphadia.

Helped by the Greek Navy over the next several months, the divers brought up several objects from the shipwreck: pottery, bottles, coins, statues. These suggested that the ship was filled with treasure looted from various conquered lands, meant to feature in a triumph that Julius Caesar was planning. They also brought up a lump of bronze and wood. Over the centuries, it had got so badly worn and layered over with bottom-of-the-sea accretions that nobody paid it much attention for two more years. But when scientists started to examine it and chisel away the corrosion, there was a surprise. They found a gear wheel buried in the lump.

Imagine that. A gear is something that, even today, we associate with relatively modern, relatively complex machinery. What was it doing on this ship from two millennia ago?

Nobody paid the object much more attention for another 50 years. In 1951, a history professor at Yale University, Derek de Solla Price, started examining it—and a total of 82 fragments that together made up what’s thought to be only a third of the original machine. Two decades later, he reported his findings in a paper titled Gears from the Greeks. The Antikythera Mechanism: A Calendar Computer from circa 80 BC (Transactions of the American Philosophical Society, November 1974).

That’s right, a machine: and Price called it a “mechanism”, a “computer”. In fact, the Antikythera mechanism is now often referred to as the world’s oldest-known computer. It has an astonishingly intricate system of gears—30 bronze wheels—that were clearly designed for calendar computations, and over 19 years. It could tell you where in the sky to find the Moon and Sun and planets. It could tell you the phases of the Moon. It could even predict eclipses. How did it do all this?

A large dial on the front had two concentric scales. One marked off the year’s 365 days, labelling the 12 months. The other showed the 360 degrees in a circle and marked out the zodiac’s 12 signs. Set up to rotate around this dial were what had to be pointers that would tell the date and the positions of the Sun and Moon on that date. It also indicated the positions in the sky of the then-known five planets, Mercury, Venus, Mars, Jupiter and Saturn. There was a small ball painted silver and black. As it revolved, it showed the phases of the Moon.

The back had two dials. One had a calendar for 235 months, a little more than 19 years. Why 235? Because after that many months, the pattern of new Moons in the year repeats. The other dial specified the times of eclipses over 223 months, or a little less than 19 years. Again, this was because the ancients recognized that the pattern of lunar and solar eclipses seemed to repeat in a cycle that long. The computer even accounted for the elliptical orbit of the Moon, because of which it seems to move faster at some points in the month than others. There was a particular gear that moved another, but around a slightly off-centre axis, and that mimicked the elliptical orbit. And the whole thing was enclosed in a box about 20cm wide, 30cm high.

The ancient world—from Greece to Egypt to India—had plenty of remarkable mathematicians and scientists. But a machine, a computer, like this? There would be none quite as complex as the Antikythera mechanism for at least another millennium.

Think of all that this tells us about what our ancestors in the first century BC knew of the cosmos; spare a thought for the ancient craftsmen who designed and built computers like this. Identifying those two near-19 year cycles would have needed many decades of observations, if not centuries. There’s the varying motion of the Moon, and the thought that went into how to capture it with rotating gears. Then there are the planets. Not only were they already known, centuries earlier, as distinct from stars, but their movement through the sky was known as well. That’s quite an achievement when you consider that Saturn takes nearly 30 Earth years to complete an orbit around the Sun. What ingenuity must it have taken to understand that motion enough to build a machine to model its orbit, a machine that would also model the motions of those other celestial objects?

In March this year, a team of scientists at University College London announced that they had worked out the design of the device, and would use it to build their own, entirely modern, version of the Antikythera mechanism. (A Model of the Cosmos in the ancient Greek Antikythera Mechanism, Tony Freeth et al, Nature Scientific Reports, 12 March 2021). They remarked: “Solving this complex 3D puzzle reveals a creation of genius—combining cycles from Babylonian astronomy, mathematics from Plato’s Academy and ancient Greek astronomical theories.”

The paper is full of fascinating glimpses into what went into creating the ancient computer.

For example, there was an inaccurate way to simulate Venus’s motion using relatively small numbers, meaning relatively small gears. (The mechanism needed gears with fewer than 100 teeth.) A more accurate method involved the number 1151. But 1151 is both too large for a wheel and prime, meaning it can’t be factorized into smaller numbers. So the designers had to search for another formula for Venus that involved numbers with prime factors smaller than 100.

Or take the phases of the Moon. They were calculated “as a simple differential, which subtracts the motion of the Sun from that of the Moon to calculate the phase, displayed on a small black and white sphere.” That may not have been perfectly accurate, but there’s something almost intuitive in the idea.

All in all, the UCL scientists concluded, “our work reveals the Antikythera Mechanism as a beautiful conception, translated by superb engineering into a device of genius.”

Genius that lay on the floor of the Mediterranean for more than 2000 years.

Once a computer scientist, Dilip D’Souza now lives in Mumbai and writes for his dinners. His Twitter handle is @DeathEndsFun