Episode 22: The Antikythera Mechanism & Friends

October 9, 2022

In the year 1900 a team of sea sponge divers stumbled upon a shipwreck from the late Hellenistic Era. Among the statues, coins, and jewellery, the salvage crew pulled out a small box covered in moss. Initially ignored, the contents of this box proved to be the most sophisticated mechanical device that survives from the ancient world. The Antikythera mechanism computed all the known motions of the heavens and its complexity was described by one scholar as "like finding a jet engine in King Tut's tomb."


Good evening and welcome to the Song of Urania, a podcast about the history of astronomy from antiquity to the present with new episodes every full moon. My name is Joe Antognini.

At the end of the last episode I promised a little break in the general narrative I’ve been taking through Greek astronomy. Generally speaking, the last dozen episodes have been largely biographical in nature. Starting with Thales we’ve looked at the ideas and contributions of various individuals to Greek astronomy over the centuries up until Hipparchus in the late Hellenistic Era. But at this point, before rounding out Greek astronomy with the work of Ptolemy, I wanted to spend a bit of time talking about Greek instrumentation, and in particular the remarkable device known as the Antikythera Mechanism. So, I promised you a shorter episode on that subject. Well, as you can see from the show length, I have proven myself untrustworthy once more. My intentions were genuine, I really wanted a shorter show on different instruments in Greek antiquity. But as I was researching the show, I realized I had more or less already talked about the different instruments like the diopter and the gnomon in past episodes, so there wasn’t really a lot new for me to say. And at the same time I started reading about the Antikythera mechanism, and reading about it and reading about it, and I realized that I had chosen a subject about which there was a lot more to say than I had originally anticipated. So there you go. This episode will have a typical length and it ill be exclusively on the Antikythera mechanism. So let’s get into it.

The Antikythera mechanism is undoubtedly the most sophisticated computational device of the ancient world, and the natural place to start talking about it is, of course, hygiene practices in 19th century Europe. Now today, we keep things clean with sponges, but strictly speaking, the sponges you use to clean your dishes are not true sponges but artificial sponges. They are made of polyester and polyurethane usually, materials which weren’t invented until the 1940s and 1950s. So the kinds of sponges we use to clean things today didn’t exist until the mid 20th century and earlier generations had to use other materials.

At least around the Mediterranean, a popular choice was a true sponge, that is to say, a sea sponge, a living creature with a soft, well, spongy texture. Sea sponges are animals, but really only barely. Evolutionarily, they branched off from the rest of the animals before any other kind of animal. As it happens, the waters around the eastern Mediterranean are nice and warm and are quite conducive to the growth of sea sponges. So, for millennia, the islands around Greece had supported an industry of sea sponge harvesting. This isn’t the first time we’ve encountered sea sponges in this podcast, I mentioned them in Episode 9 when I was talking about the graffiti in the public lavatory in Ostia Antica since one of them used the term xylospongium, which was a sea sponge attached to a stick with which you could wipe your bottom after going to the bathroom. At any rate, back before the invention of toilet paper in the late 19th century, having a sponge around was a useful thing if you were into hygiene.

But getting these sponges was no easy feat. They don’t really swim around so you can’t just drop a fishing line or cast a net and hope to catch them the same way you can catch fish. The traditional means of harvesting sea sponges was for divers to use a technique called skandalopetra. The idea was that the diver would be naked and use a rope to tie a stone around his ankle. The stone, in turn, would be tied by a long line back to the ship. The diver would jump out of the ship and quickly sink to the depths thanks to the rock tied to his ankle. Then he would spend as long as he could looking for and collecting sponges at the seabed, and once his breath started to run out, he would cut the rock from his ankle and climb up the rope back to the boat. Then the rock could be hoisted up as well, and after the diver had recovered, the whole process could start all over again.

This technique, centuries old, allowed an experienced diver to get to depths of about 50 meters or so, but it was limited by how long the diver could hold his breath, about one and a half to two minutes. Given the amount of time it took to dive down and resurface, he would have between 30 seconds and a minute at the bottom to find and collect any sea sponges. Given how grueling this feat was, along with all the preparation that went into securing and recovering the rock, an experienced diver could only make about 10 dives per day. So up until the late 19th century, sea sponges were certainly available in the Mediterranean cultures, but nevertheless not especially cheap.

But this ancient industry changed dramatically in the second half of the 19th century. The 19th century had seen huge advances in diving equipment, and, in particular, by the end of the century, what is called standard diving dress had become fairly practical. Standard diving dress, which was called skafandro in Greek, is basically what you would imagine in your head when you think of an old-timey diver. The diver wore a thick canvas suit with a bronze globe on their head with little circular glass portholes on the front and the sides. Then they were connected to the surface with a long tube so they could breathe. Greek sponge divers were not precious about the old ways of doing things, and as soon as this new technology came about they became enthusiastic early adopters.

The skafandro allowed the divers to attain depths of 70 m instead of the 50 m that had been possible previously, but more importantly, it hugely extended the amount of time a diver could spend at the seabed. Rather than spending barely a minute at a time looking for sponges, they could spend 25 minutes at a time. Even though with the bulky equipment they only could do two dives a day, that still allowed for five times as much gathering time as they had available previously. Now, this was still not a job for the risk averse, perhaps even more so than before. Decompression sickness was not well understood until the early 20th century, so frequently divers would die from this mysterious syndrome. Nevertheless the economic opportunity of harvesting sea sponges was so much that by the year 1900 there were about 50 teams of divers operating throughout the eastern Mediterranean, almost all of them Greek, though quite a few of these Greek crews were based out of the Ottoman Empire.

Well, in the year 1900 a crew of six or seven sponge divers captained by Dimitrios Kontos had been working off the island of Symi. As was typical they had two boats, a mother ship that contained supplies and the sponge harvest, and a smaller diving boat that would make excursions during the day. The diving season lasted from April to October and when the season ended they headed back to their home port in Tunisia. But, just as in any good Greek tale, in the course of their travels a storm blew them off course. They found themselves at the island of Antikythera, where they sheltered for a few days.

Now before I go on, I would do well to talk about the island of Antikythera itself a little bit since it lends its name to the famous mechanism that was discovered there. The name Antikythera is actually a fairly modern one. In antiquity the island was called Aigila. It’s not a big place, about 8 square miles, so it could never really support too many people. Its reputation in antiquity was for being a base for pirates. The island is strategically situated across from the island of Kythera — Antikythera just meaning opposite from Kythera, and the strait between the two is the most direct path to go from the Aegean sea to the rest of the Mediterranean. Naturally this was a convenient path for merchant ships to sail and so the island became a natural location for pirates to hang out and wait for their prey. Well Rhodes had tried unsuccessfully to rid the area of pirates for many decades, and the rise of Rome led to a further increase in piracy in the region. You see, in 167 BC, Rome was able to force the port of Delos to be duty free — any goods passing through it would be untaxed. This, naturally, made Delos an extremely attractive harbor to stop at relative to Rhodes which did have a tax. After this happened, the income Rhodes generated from its harbor tax dropped by some 85% and consequently Rhodes was unable to fund its anti-piracy operations in the region and piracy started to become a much bigger problem. Since Rome had effectively kneecapped the anti-piracy task force of Rhodes, by the mid first century BC Rome realized that if someone was going to do something about the pirates it would have to be them. Although Rome was not especially known for being a naval power, it nevertheless raised a navy and around 67 BC eradicated the pirate menace in their typically brutal fashion.

Well a few years after that a merchant ship was traveling from the Aegean Sea towards the western Mediterranean. It probably departed from one of the major ports, either Rhodes or Pergamon, and at any rate was carrying goods which had originated from a variety of locations. Where it was going we don’t exactly know, but it probably would have stopped at a number of ports and its cargo would then be passed at those major ports to a smaller ships onward to a variety of final destinations. Since the threat of pirates was gone the ship could take the more direct path near the islands Kythera and Antikythera. But tragically the ship was likely caught in a storm. Being heavily laden with cargo it quickly sank not too far off of the coast of Antikythera. We know that at least two men and one woman died, along with a fourth individual whose gender can’t be determined from its skeletal remains. Now an older theory has it that this was a ship containing loot for a triumphal procession of Julius Caesar since the ship contained so many valuable artifacts from so many different locations. But as modern scholarship has learned more about the ancient maritime trade, the consensus is that this ship is consistent with being more or less an ordinary merchant vessel of the time.

At any rate, centuries passed and by the late middle ages, after the 4th Crusade, the island came under the rule of the Republic of Venice where it acquired the name Cerigotto to distinguish it from the island opposite it which got the Italian name Cerigo. It remained a possession of the Republic of Venice of frankly not much value up until the Napoleonic Wars in the early 1800s. After the chaos of Napoleon had shaken out, the island ended up in the hands of the British government where by 1815 it had been spun out as a part of an autonomous protectorate called the United States of the Ionian Islands. Now, politically, this region was starting to get pretty restive as the 1800s progressed. By this time the Greeks had not had a state of their own for some 2000 years since the Hellenistic Era. They had by turns been governed by the Romans and the Byzantines and the Ottomans. And now some of them felt that it was high time that they govern themselves. Being fairly remote, the island of Cerigotto became a hotbed for Greek patriots who needed to get out of dodge for a little while. Being full of patriotic fervor they had no use for an Italian name for the island so they Hellenized it, turning Cerigo into Kythera, and hence the island opposite it, Antikythera.

Now in the 1820s, the Greeks had revolted against Ottoman rule. These kinds of revolts had occurred sporadically over the centuries and had always been put down by the Ottomans, and as this one progressed, it looked like the story would end the same way. But the great powers of Europe of the day, Britain, France, and Russia, spied an opportunity. In the 18th century the Ottoman Empire had been a thorn in the eastern side of Europe, so, hoping to take it down a peg, they provided major support for the Greek revolt as it was dying out and with their help they managed to turn things around for the Greeks and carve out a small independent Greek state. In a conference in London in 1830, Britain, France, and Russia created a government for this new state — the Greeks themselves would have no say of course — and settled upon a monarchy. Since they needed to preserve the balance of power between each other, they agreed that the monarch of Greece would not be of British, French, or Russian extraction. So they settled on a royal from the Bavarian line who was friendly with them all, a man who became King Otto of Greece.

Now, King Otto was not very popular with the Greeks. He imposed high taxes and as we would say today, he was not a good culture fit. He was Catholic and had no desire to convert to Greek Orthodoxy. He managed to rule for some 30 years, but his unpopularity ultimately caught up to him and in 1862 he was overthrown. At this point, the Greek elite started searching around for a replacement. Monarchy was still in favor and the Greeks understood that their small state was vulnerable and that they needed to maintain the support of the great powers if they wanted to survive as a state neighboring the much larger and more powerful Ottoman Empire. So they wanted a European royal to replace King Otto and they had a particular man in mind — Prince Alfred, a member of the British royal family and the Duke of Edinburgh. If they could get a Brit on board, not only would they automatically have the continued support of one of the greatest powers in Europe, they could grow the Greek territory by subsuming the United States of the Ionian Islands that Britain possessed. The Greeks harkened back to their ancient democratic roots and held a plebiscite in which 240,000 people voted to express their preference for their next king, and Prince Alfred got a stonking 95% of the vote. But sadly this plan was a no go. The London Conference had held that the Greek crown could not pass to British royalty, or French or Russian royalty for that matter. So after scrounging around the Greeks settled on an alternative, Prince Wilhelm of Denmark, who was friendly with the British royal family and took the name King George. But all is well that ends well and the Greeks got what they were hoping for all along. As a coronation gift, Britain ceded to King George the Ionian islands, the island of Antikythera being among them. So in 1865, after almost 2000 years, the island of Antikythera had finally come under Greek rule once again.

Well, to get back to the original narrative of the sponge fishers in the year 1900, after the diving season ended and they were blown off course and ended up at Antikythera, once the storm passed, they decided that now that they were there, they might as well do a bit more diving and pick up some more sponges before resuming the journey back home. During one of his dives, the diver Ilias Stadiatis noticed some bronze on the seabed. On his second dive he brought up an arm from a bronze statue. Evidently the divers had stumbled upon a shipwreck full of antique treasures. Now, officially that was all they did. They immediately reported their find to the Greek government which organized a salvage operation of the shipwreck. But unofficially, interviews with their descendants 60 years later suggest that they probably did a little bit of salvaging of their own first and sold some of what they found to antiques dealers in Alexandria for a bit of extra pocket change. But regardless, the scale of the wreck was large enough that they couldn’t remove very much on their own.

Now, one thing that’s worth mentioning here is that it’s actually a bit uncertain as to when the discovery of the wreck actually occurred. According to some reports, and logically, the wreck was found in late summer since this was when the diving season ended and the divers were supposedly on their way back home. But other reports say that it was around Easter, though confusingly, those same reports also say that it was at the end of diving season even though Easter is near the start of diving season. But one bit of evidence in favor of this Easter timeline is that there is a report of a telegraph dispatch from the island of Kythera around Easter 1900 of the discovery of treasure. The officials who received this message in Crete just assumed that the operator was drunk and ignored it, which was probably a pretty good Bayesian prior, but knowing what we know now, maybe the telegraph operator in Kythera was providing an accurate report after all.

At any rate, the Greek government organized a salvage operation under the supervision of the minister of education at the time, a man named Spyridon Stais. The salvaging began in November 1900 and the shipwreck immediately proved itself to be a repository of tremendous riches. It contained bronze and marble statues, jewellery, glassware, pottery, coins. The Greek press followed the operation very closely, but as the months passed the crews had found the most prominent relics and then started to recover less and less on each dive. By the summer of 1901, Stais was having trouble convincing the crews to continue working since they were coming up with so little. Probably in July one of the divers recovered a small nondescript box covered in moss. The story has it that it was so unpromising that the diver was going to just toss it off the side of the ship but was stopped by a naval officer. At any rate, this item was catalogued and put onto a shelf in the National Archaeological Museum of Athens just like all the other low-value items that had been recovered in the later stages from the wreck. The salvage operation stopped in September 1901 and the wreck wasn’t visited again until an expedition in 1953 by Jacques Cousteau.

At the time, all the interest in the find was in the statues. Among all these other valuable items, the box was completely ignored. After a year or so of sitting untreated on a shelf, the exterior of this little box had rotted away and revealed some of its inner contents. One weekend, Spyridon Stais happened to be visiting the museum and perusing its contents. He had some extra time on his hands since by this point he was no longer the minister of education. The government had collapsed just a few months after the salvage operation had ended when what are called the Gospel Riots broke out, when a Greek newspaper published the first copy of the Gospel of Matthew in modern Greek. At any rate, Stais noticed that this object contained some brass pieces and that they had geared wheels. Gears were only known from a few descriptions in the ancient literature, and in point of fact, today the Antikythera mechanism is the oldest surviving geared object in Europe. Stais realized that this object was of more than just passing interest, that it was something of some sophistication. For this reason, Spyridon Stais is usually considered to be the man who truly discovered the Antikythera mechanism rather than the unnamed diver who pulled it out of the ocean. Now, as an aside, there is some misinformation out there because Spyridon’s cousin, Valerion Stais was the curator of the Museum at the time, so a number of references mistakenly attribute the discovery of the Antikythera mechanism to him, including even that most trustworthy of resources, a Google doodle. So you have to keep your eyes peeled to make sure you have the right Stais.

At any rate, the initial interest in the object was actually in the inscription. And in particular, people really wanted to know how old it was and were less interested in what exactly it did. At the time a major open question was just how old the shipwreck was. One theory had it that it was from the late Hellenistic era, say around 100 BC, and another theory was that it came from the later Roman era, around the time of Constantine, say around 300 AD. This second theory was that this ship was full of treasures that Constantine was shipping to his palace in the new capital of Constantinople. Now, the shapes of Greek letters evolved over the centuries, so by seeing what the characters look like, you can very roughly estimate when an inscription was made. For example, a W shape for the letter omega starts to become common in the late Hellenistic era and in the Roman era, the letter epsilon changes from being written with straight lines like a capital E today, to being written with curved lines, like a capital C with a horizontal line in the middle. Likewise the letter sigma stopped being written as a capital sigma today, but as a capital C.

So it came to pass that in 1902 the philologist Adolf Wilhelm studied the mechanism at the museum along with Ioannis Svoronos who was the director of the Numismatic Museum. Based on the shapes of the characters in the inscription Wilhelm concluded that the artifact dated from the second or first century BC, and what’s more, some of the characters he could make out to read “ray of the Sun,” which suggested to him that this was an astronomical device of some sort. Svoronos, however, had been a staunch proponent of the theory that the shipwreck was from the late Roman Era and argued that the presence of serifs on the letters indicated a much later date, more like the 2nd century AD. Svoronos didn’t really know what he was talking about on this point, serifs did in fact become more common later on, but they were certainly present in earlier texts and Wilhelm was well aware of this. Nevertheless, the Greek press was following anything that had to do with the treasures discovered in the shipwreck with rapt attention and eagerly played up this scholarly controversy.

Later in the year Svoronos put together a report on the contents of the Antikythera shipwreck and assigned the section on the Antikythera Mechanism to Periklis Rediadis who was a naval officer. In this report, Rediadis presented the first theory of the device’s function. Based on the astronomical reference and the gears, he supposed that the device was a kind of astrolabe. I’ve mentioned astrolabes in an earlier episode and they principles behind them were present by the late Hellenistic era, but they didn’t really become popular until the Middle Ages. If the Antikythera Mechanism were an astrolabe, it would have been by far the oldest astrolabe known since otherwise the earliest surviving astrolabe originates from the Islamic Golden Age. I won’t go into detail about how an astrolabe works until we get to the Middle Ages, but it is basically a disc with a stereographic projection of the sky that allows its user to compute all sorts of interesting things like when and where sunrise and sunset will be, how long twilight will be for, where the stars will be on the sky, the length of shadows and so forth. The astrolabe was a sort of Swiss army knife of Medieval astronomy.

Rediadis’s conclusion that the Antikythera mechanism was a kind of astrolabe was more based on vibes than anything else. In the Middle Ages the astrolabe had been used in maritime navigation and this device was found on a boat. An astrolabe was used for astronomical measurements, and this device clearly had something to do with astronomy. An astrolabe was a fairly complex instrument, and this was also a fairly complex instrument. But this is essentially where the similarities ended. An astrolabe is based on a stereographic projection, a projection of the spherical sky onto a flat plane, and there was no evidence that there were any stereographic projections on this device. Rediadis supposed that perhaps all the gears in the device somehow computed a stereographic projection, but he didn’t really have any specific ideas as to how this would work.

Rediadis’s report noted a number of details about the device, but it wasn’t especially quantitative. In particular, the first thing that a mechanical engineer might do when presented with a device like this with a whole bunch of gears in it, is count the number of teeth on the gears and what the radius was and see if any of the gears were interlocking, and so on. But Rediadis was not so quantitative. Nevertheless, rather importantly, he took a series of photographs of the four pieces of the Antikythera mechanism which proved invaluable to inspire more research on the object.

A few years later, a German philologist named Albert Rehm had seen the photographs in Rediadis’s report and had had his curiosity piqued. He had been doing some archaeological work in the Levant, but on his way back home his travels put him through Athens and he decided to stop at the Museum of Athens and take a look at this curious artifact. He didn’t have much time before he had to leave and continue his journey home, only a couple of hours, so he resolved to focus on what is called Fragment C, which had received less attention at the time. Now by this point it was 1905 and the mechanism had been sitting in the museum for about four years. It had always been low priority relative to everything else in the haul, but by now the museum’s conservationist had finally gotten around to trying to restore it. So when Rehm got to inspect it it had been cleaned up a little bit and he noticed some inscriptions that hadn’t shown up on the earlier photographs. The text was a parapegma. I talked a little bit about parapegma back in Episode 17, but they were basically like an ancient Greek almanac, they would specify things like how many days were in the month, when certain constellations would rise, and so on. Rehm also saw the first characters of the word “Aphrodite,” which could refer to the planet Venus. And furthermore, with this newly cleaned item, Rehm found that there was a dial with the word “Pachon,” which was one of the Egyptian months. Now, by the Hellenistic Era, the Egyptian months were in common use at least for astronomers, along with the usual Greek calendar because the Egyptian months were very standard. Rather than following the phases of the Moon as the Greek months did, each Egyptian month was exactly 30 days long and there were an extra five days at the end of the year with no leap days. So the Egyptian months were very regular which made them useful for astronomical calculations. In that way they were used a little like the Julian calendar is used in modern day astronomy. The Julian calendar has exactly 365.25 days per year and doesn’t have any of the weird rules about leap days or leap seconds that the modern Gregorian calendar does, so it’s very convenient for astronomical observations over long periods of time since you don’t generally care about those things.

Well all this is to say that after Rehm had looked at this thing for a few hours it was quite clear to him that it really did have something to do with astronomy. But he wasn’t quite convinced that Rediadis’s interpretation that it was an astrolabe was correct. Instead he thought that it was a kind of planetarium that simulated the motions of the Sun, Moon, and planets. One of the fragments had a series of rings, so he supposed that a series of pointers were driven by gears so that their periods matched the periods of the planets.

Albert Rehm’s interpretation was closer to the mark than Rediadis’s, but he never published his analysis. He wrote up his work and submitted it to a prize competition, but he lost and soon thereafter became caught up in faculty administration at his university and didn’t have as much time for research. He had talked about his ideas with a few friends and they had included a description of the device and his interpretation in their own work, but sadly the world never got Rehm’s full analysis, despite encouragements from his friends to publish.

Well, the Greek world had always been much more interested in the findings of the Antikythera wreck than the rest of the world, and in the case of the Antikythera mechanism this was certainly the case. What little was published about it didn’t attract much attention abroad. By about 1910 or so interest in the object had almost completely died away, and the device lay dormant until the 1960s. Until then, if you wanted to figure out what it did there were only two ways you could do that: look at photographs of it or travel to Athens and inspect the thing in the museum with a magnifying glass.

So, about half a century passed and everyone associated with the original investigations died and the artifact remained little understood and more or less completely ignored. But this all changed in the early 1960s when an English physicist by the name of Derek de Solla Price became somewhat obsessed with the object. He had been teaching applied math but started to get very interested in the history of science. As he was reading about ancient scientific instruments he came across some brief references to this mysterious device that had been dredged up in the Aegean Sea and wanted to know more about it. Now again, at this time there were only two ways to study the device, look at photographs of it or visit it in person. Unfortunately he couldn’t afford to travel to Athens so he made a bold request to the Museum of Athens that they loan the device to the British Museum so that they could use their state of the art equipment at the British Museum’s research laboratory to study the mechanism in detail. But the Museum of Athens felt that the British Museum had quite enough Greek artifacts thank you very much and flatly denied this request and was quite clear that under no circumstances would the artifact leave Greece. So he had to do the next best thing of looking at photographs. Now this wasn’t so bad because he was able to get a new set of photographs and they were much higher quality than the original photographs from 1902. Furthermore, the device had, shall we say, evolved in the following six decades. Of course in the early 1900s the museum’s conservationist had done what he could to restore it, but also over time bits of it had accidentally gotten damaged or broken off and revealed parts that had previously been hidden. Much of this probably had happened during WWII. Since Greece was a battleground in the war, the Museum of Athens had to rapidly protect its contents. The museum’s curators dug trenches, put all the statues in the trenches, and then filled the trenches back up. For the smaller items, among which the Antikythera Mechanism was undoubtedly a part, they packed them into boxes, put the boxes into the Museum’s basement and then filled the entire basement with sand in the event that the Museum or its environs was bombed. So this frantic conservation effort almost certainly jostled the mechanism more than it had been used to and revealed previously unseen components of the device.

Well with these photographs Price became extremely intrigued by the object. Soon after he got a position in the United States and there met the historian of science Otto Neugebauer. Now I’ve really done a bit of a disservice to Neugebauer in this podcast because I’ve only mentioned him once before way back in Episode 5, but he really deserves much more than that. He was probably the greatest historian of astronomy in the 20th century, and much of what we know about Babylonian astronomy in particular comes from his work. At any rate, Neugebauer thought that Price was on to something with this device so Neugebauer helped Price to get a grant for $460 so that he could travel to Athens for a week and a half and study the mechanism.

That week and a half was longer than the couple of hours that Albert Rehm had gotten, but in the grand scheme of things it still wasn’t that much time and Price didn’t want to waste it. When he had arrived he spent 10 days taking copious notes and photographs and measuring absolutely anything that could be measured of the device, including all 20 of the gears or fragments of gears that were present.

After Price returned to the States and could start to interpret his notes he became increasingly convinced that this device was far more sophisticated than anyone had appreciated, and moreover far more sophisticated than Greek astronomy of the time was understood to be. He had a certain flair for the dramatic and described the Antikythera mechanism in the press a little hyperbolically as, “like finding a jet engine in King Tut’s tomb,” and “like opening a pyramid and finding an atomic bomb.” Based on the device Price argued that by the Hellenistic Era the Greeks had been far more scientifically advanced than anyone in the modern world had given them credit for and that they were in many ways more scientifically advanced than Europeans in the 18th century.

Now, this is an exaggeration. But it is certainly the case that the sophistication of the Antikythera Mechanism, along with a deeper understanding of ancient astronomical techniques by Neugebauer and other historians of science in the mid 20th century led to a reassessment of the state of science in the Hellenistic Era. As I hope I’ve made clear over the course of this podcast, much of what we understand about ancient Greek and Babylonian astronomy is not self-evident but came through careful readings of oblique references in a diverse set of texts. There had always been the Almagest, of course, which had been commonly read since antiquity and which set a certain baseline of Greek astronomical sophistication, but the Almagest was mostly about modeling planetary motions. All the rest of Greek astronomy, the estimates of the Earth’s size and the distance to the Moon, and Aristarchus’s heliocentric theory, and the measurements of the seasons had to come through careful study of the classical texts. So this 20th century discovery of the Antikythera mechanism fit in with this growing appreciation for what the ancients were capable of. A few decades later in the 1990s, Lucio Russo went even farther and argued that the Greeks of the late Hellenistic Age were even more sophisticated than is understood and has marshalled a good deal of circumstantial evidence to suggest that the Greeks had made major advances in optics and understood that the Earth was not a perfect sphere but was somewhat oblate. The arguments are intriguing and not implausible, but still I can’t help but feel regret at what has been lost since it seems there is just only so much that can be gleaned from the few surviving scraps from all those centuries ago.

Well, to get back to the Antikythera Mechanism, some sense of regret that more could not be known was also how Price seems to have felt in the early 1960s. He had gotten his chance to look at the Antikythera mechanism in person, he found that it was sophisticated, and believed that it had to be some sort of a computer, computers of course being quite cutting edge in the 1960s, at least the kind that used a machine rather than a person. But he just couldn’t quite piece together how the thing worked, how it all fit together. In short, Price was stuck. He had gotten to the limits of what could be understood about the device just by looking at it. There was clearly a lot going on behind the parts that were exposed to the outside, but no one dared to break it apart and see what was on the inside.

Which was for the better because a decade or so later the technology had developed to inspect the interior of an artifact in a non-invasive way with X-rays. So, until that happened, during the rest of the 1960s Price turned his attention towards other scholarly pursuits, in particular the sociology of science where he noticed a few interesting things like the fact that the number of pages in the journal Philosophical Transactions of the Royal Society grew exponentially. He also discovered what came to be known as Price’s law, which states that half of all contributions in a field come from the square root of the number of contributors to the field. So if a field has 100 people working on it, half of all the papers will be written by 10 people.

But in the 1970s radiography had gotten to the point were it was possible to ship a small X-ray machine to the Museum of Athens and do X-ray imaging of the Antikythera mechanism. This then let Price and others get the crucial information about gears that had hitherto been obscured in the mechanism’s interior and see how they all fit together. Now, although this was a breakthrough, these images weren’t the easiest things in the world to interpret. The X-ray images gave no depth information, you just got a flat projection, so there was just a big jumble of gears all stacked on top of each other with no indication as to how they fit in the depth plane. Price and his collaborators had tried to get some depth information by taking images at different angles to produce some parallax or put the film at different distances since the material would refract the X-rays differently at different depths, but in practice these ideas didn’t really work. Nevertheless, the flat images showed the number of gears, how they might interconnect, what the spacing of the teeth was, which gears had common axles, and so on, and this was enough for Price to piece together the essential functioning of the mechanism, and he published his work in a book called Gears for the Greeks. Now early on, Price had supposed that the mechanism had something to do with the planetary motions, just as Albert Rehm had thought, but with the X-ray images he rejected this idea. All the gears seemed to be related exclusively to the motion of the Sun and the Moon. Price’s interpretation of the workings of the gears was more or less correct as far as it went, but as it turned out his earlier intuition about the breadth of the device turned out to be more on the money.

In subsequent decades, radiography progressed further with the development of linear tomography and then computed tomography. The idea of linear tomography is that you put the target halfway between the X-ray source and the film and then you move the source and film towards each other or away from each other at a constant speed. Since the diffraction of the X-rays depends on the relative distance between the source and film and only at the point exactly halfway between the two is this relative distance constant, the images from other depths blur out and you get a sharp image at a specific slice through the object. By putting different depths at the halfway point, you can then build up a 3d stack of slices through the object. Computed tomography also uses the diffraction properties of X-rays, but in a more sophisticated way where a computer program can back out a complete 3 dimensional structure from the diffraction pattern. By the early 2000s CT technology had gotten good enough that a CT scanner could be shipped to the Museum of Athens to inspect the mechanism since their policy that the artifact can’t leave the country is still in force. Another clever technique that came to be used in the late 20th century and early 21st century was reflectance transformation imaging. The idea here is that you take a series of high resolution photographs of the artifact’s surface but with a light source at a series of angles all around the object. By looking at the shadow and reflectance patterns as the light source changes this allows you to build up a detailed three dimensional reconstruction of the surface of the object. These newer techniques revealed a few new inscriptions and indicated that there were evidently some missing gears, something that had long been suspected. Now, all but one of the gears was still associated with the Sun and Moon, but it was clear that the main driving mechanism was much larger than it needed to be to drive the lunisolar chain. Based on the surviving inscriptions and fragments of a plate, along with a lone gear which does not fit into the lunisolar calculation, it’s now believed that there was another gear chain that had to do with planetary calculations, even though much of it is missing.

By around the late 2000s, just over a century after its initial discovery, archaeoastronomers were able to piece together an almost complete picture of what the Antikythera Mechanism looked like, how it worked, and what its functions were. There are still a few uncertainties, particularly regarding the front side of the device, but given that essentially the entire three-dimensional structure of the surviving pieces has been reconstructed, it seems that we’re at a point where it’s not possible to make any further progress without discovering additional fragments of the device.

So then, that is the winding tale of how the mysterious Antikythera mechanism came to be discovered and more or less understood. But I’ve glossed over what exactly it did except for saying that it was a kind of astronomical computer. So for the rest of the episode I’ll give you a quick tour of the device.

Now, today the device has been broken into seven major fragments labeled A through G, and there are 75 minor fragments. So what remains of the poor thing is in 82 pieces altogether today. But when it was in its prime, the Antikythera mechanism was a rectangular box about 30 centimeters tall, 20 centimeters wide, and 10 centimeters or so deep, although the depth is a little more uncertain. On the side there was some way to drive the entire mechanism, probably a knob that was turned by hand, though Derek Price at one point made a more fantastical suggestion that it was powered automatically by a water clock.

Both the front and back of the device were used and it’s simpler to start with a description of the mechanism’s back side particularly since much more of it survives than the front so there’s less uncertainty around what it looked like. Now part of the back face rather unintentionally preserved some critical information about how the device worked. It seems that there was a sort of instruction manual that had been inscribed onto a metal plate or some other material that had been packed away against the mechanism. Over the years, these inscriptions imprinted themselves very subtly onto the back face, so even though the original material was lost, about 50 lines of text can be seen on the back face in backwards writing.

The back face itself consists mainly of two large dials. On top is a dial that is called the Metonic dial for reasons that will shortly become clear. The dial consists of a spiral with five turns, but the spiral is not quite uniform. On the right hand side the curves of the spiral form semicircles which are centered on a central point where a pointer would have been mounted. On the left hand side, however, the curves of the dial formed semicircles slightly offset from this central point, on a point a little above it called the secondary center. Now this spiral was not painted on to the mechanism’s face but was actually a groove. So the dial wouldn’t have had a pointer quite like the hand of a clock which just turned around above the dial, but it would have been a little more like a record player that had a needle which was embedded into the groove.

Each turn of the spiral was divided into 47 cells, all subtending equal angles with respect to the axial center. Since there were 5 turns, there were 235 cells in all. Now your ears should prick up when you hear this number because 235 is, of course, the number of months in a Metonic cycle, over which time the phases of the Moon line up with the times of the year. Each cell was labeled with a month and a year and on the innermost ring of the dial there was a series of numbers below some of the cells. These would indicate which day of the month was to be skipped since sometimes months had 29 days and sometimes they had 30 days. So this would allow the user to keep track of the progression of the months as time passed and incorporate all the necessary intercalations.

Within this large dial there were embedded two smaller dials, sort of like how a watch or grandfather clock might have a smaller dial inside it to represent the day of the month. Each of these two dials was divided into four quadrants. One of these incremented every year and was labelled with the year of the four-year Olympic cycle. You may recall from Episode 14 that the Greeks held the Olympic Games every 4 years, but this was just one set of games in a four year cycle. Each year of this cycle had its own game and you could use the Antikythera mechanism to keep track of where you should be going to get your sports fix for the year. The other of these smaller dials was much slower. It would only increment when the pointer on the bigger dial had travelled the entire length of its groove, five complete turns, a complete Metonic cycle. Unfortunately the dial itself does not survive, only the gears that operated it, so we don’t know how it was labelled. But because four Metonic cycles made up a Callippic cycle, this dial tracked how far into a Callippic cycle the user was.

So that is the upper half of the back panel. It basically kept track of the lunar months over the course of a 76 year Callippic cycle. The lower half of the back panel had its own large dial. It was similar to the upper dial with a spiral, but this time the spiral was continuous rather than being centered on different spots on the left and right sides and instead of having five turns this one had four. All along it the spiral was divided into 223 cells of equal angular width. Once again your ears should prick up at the number 223 because this, of course, is the number of months in the Saros cycle, so this dial is naturally called the Saros dial. I’ve mentioned the Saros cycle a few times over the course of the podcast but to remind you it’s a cycle over which the phases of the Moon match up with the times the Moon passes through its node, and as such, lunar and solar eclipses repeat from one Saros cycle to the next. So the 223 cells of the Saros dial each represented a month, and most of these were empty, but a few special cells had a label telling you that there would be a lunar eclipse or a solar eclipse or both that month. And what’s more, the inscription told you the approximate time that the eclipse was expected to occur.

Now the Saros cycle does not fit into an integer number of days, it’s off by 8 hours, so the times of the eclipses shift by 8 hours from one cycle to the next. To help you keep track of this annoying detail, the Saros dial helpfully had a little tiny dial on its interior that was divided into thirds. One of these was empty, the other said 8, and the third said 16. So to figure out the time an eclipse was expected to occur, you would look at the time inscribed in the cell corresponding to the month, and then add the number in the central dial.

A final detail about the Saros dial was that there were four ticks on the innermost loop of the spiral and these represented when the Moon was at apogee during full moon. So you could also estimate if a solar eclipse was going to be total or annular from these marks.

So the back panel of the Antikythera mechanism was able to keep track of the calendar time on the standard Greek lunar calendar and predict the time and date of eclipses. The front panel is a little bit more uncertain because less of it survives but it seems to have consisted of just a single large dial. Around the edge were two rings. The outermost ring was marked with the months of the Egyptian calendar, each month being 30 days plus 5 extra days at the end of the year. The inner ring was marked by the signs of the zodiac. Now because the Egyptian year had no leap day, it drifted with respect to the stars fairly dramatically, about one day every four years. This meant that after a century the beginning of a year would be 25 days off relative to the equinoxes, almost a full month. Consequently, the outer ring representing the Egyptian year was movable and the user could set it so that its beginning corresponded to the Sun’s location on a particular part of the sky.

Now it’s fairly clear based on the structure of this dial that there would have been a pointer to represent the position of the Sun on the sky over the course of a year. But what is less clear is whether or not the pointer just increased uniformly or if it had a variable motion like the actual Sun does. If the pointer did have a variable motion, it would produce an accurate position on the zodiac ring, but it would then tell you an incorrect day on the Egyptian calendar ring — it could be up to two days off sometimes. It’s possible that there were in fact two pointers here, one which represented the position of the Sun and one which represented the day of the year, but the pointers don’t survive so this is just speculation.

There was also a pointer which represented the position of the Moon on the sky and a piece of the gear that held this pointer survives. The Moon’s motion is non-uniform since the Moon’s orbit is eccentric and the gear chain took this into account. But what is especially cool is that right at the base of the pointer there was a circular hole where a small ball was inserted. The ball was black on one half and white on the other and rotated around to tell you the phase of the Moon at a particular time.

The most uncertain part of the mechanism is the other pointers on the front dial. There were probably five other pointers, each of which represented the position of a planet on the sky and these pointers would have accounted for the non-uniform motion of the planets, including their retrograde motions. So the pointers for Mercury and Venus would have tracked the pointer for the Sun fairly closely and the pointer for Saturn would very slowly wobble its way around the dial with there being a little retrograde motion on every year.

The knob on the side that drove the entire mechanism was fairly slow. A single turn of the knob advanced the state of the mechanism by 78 days. But because the entire Callippic cycle lasts 76 years, it would take about 350 turns to advance the dial for the Callippic cycle from start to finish. This had to be done fairly carefully, too. Because the grooves for the Metonic and Saros cycles were spirals, the needle would terminate at a certain point, at which time the operator would have to pull the needle out and reset it at the beginning. The mechanism was fairly delicate, too, so if the operator tried to force the dial when a needle had reached the end of the groove he would have been liable to break something. But although the dials on the back panel advanced fairly gradually, the front panel would have been more exciting with the Moon’s pointer spinning around three times for every turn of the knob.

One last bit that I’ll mention is that there was probably some writing on the front face of the box above and below the front dial, but because this part of the device was probably wood, it is almost completely gone. Some traces of it can be seen in the original 1902 photographs and this illustrates a sort of paradox that crops up in archaeology from time to time. Knowing what we know today, it would have been far better if the device had been immediately preserved when it was discovered rather than trying to restore it in the early 1900s. With modern radiography it’s possible to peer into the interior of the device even if it is encrusted in moss and limestone. But even though the early restoration efforts possibly ended up removing crucial pieces of information that could have been recovered with modern technology, without those restoration efforts, no one would have recognized the significance of the device in the first place and it wouldn’t have gotten the detailed attention over the decades that the Antikythera mechanism did. In the end I think we have to be grateful that as much of it was preserved as it did happen.

So that is a brief tour of the appearance and purpose of the Antikythera mechanism as we understand it today in 2022. The device compiled essentially all the astronomical models of the day into a single marvelous computational device. It was in many ways the culmination of both Greek astronomy and mechanical engineering of the time.

So who was it who constructed this wondrous device? Here we really have no idea. There have been vague speculations that perhaps Hipparchus had a hand in it, but there is not really much to go on here except that the timing sort of works out if the device had been built, say 70 or 80 years before it was transported on its ill-fated journey. And the ship on which it sank probably passed through Rhodes, which was where Hipparchus was based out of. But that is pretty thin gruel to base a theory on. Personally I like to think that it was the handiwork of some nameless otherwise astronomer or engineer who otherwise left no other record in the literature so that we are forced to judge their output solely on this one remarkable device.

So that is all I have to say about the Antikythera mechanism. The next episode will be the last in our tour of ancient Greek astronomy. We’ll look at a few of the astronomers of the Roman Era but particularly focus on the last great astronomer of antiquity, Claudius Ptolemy, who developed the most accurate model of planetary motion to date, one that became the standard model of medieval astronomy and would not be surpassed in accuracy for about 1500 years, not even by Copernicus, until the efforts of Johannes Kepler in the early 17th century.

I hope you’ll join me then. Until the next full moon, good night and clear skies.

Additional references

  • Evans, 1998, History and Practice of Ancient Astronomy
  • Jones, 2017, A Portable Cosmos
  • Lewis, 2009, Surveying Instruments of Greece and Rome
  • Lin & Yan, 2015, Decoding the Mechanisms of Antikythera Astronomical Device