Of all ancient myths, Atlantis is one of the most enduring. The idea of a sunken city, its forgotten knowledge and treasures just waiting to be discovered, has lost none of its appeal in the two thousand years since the Greek philosopher Plato first told the story.

But what really lies beneath the waves? I've written a feature article for New Scientist looking at some of the world's most interesting underwater towns and cities - it's published this week and you can read it here. From mysterious stone circles to ancient royal palaces, it turns out that submerged sites are yielding insights - and treasures - that rival anything to be found on land. And as today's coastal cities struggle with earthquakes, sea level rise and land erosion, these drowned settlements tell a sobering story of past communities who lost the same fight.

The sites I've looked at include Pavlopetri - a Greek town dating from the time of Homer's epics - and Abu Qir Bay, the gateway to ancient Egypt. Then there's Port Royal in the Caribbean, the "wickedest city on Earth" until an earthquake sank it in 1692, and Dunwich, once one of the largest cities in England, before it succumbed to the waves of the North Sea. Israel's Atlit-Yam reveals an incredibly detailed picture of life 9000 years ago, while the towering stone steps of Yonaguni in Japan (pictured) have sparked stories of a mysterious ancient civilisation, even though they were probably formed by natural causes. Finally, of course, there's Atlantis itself. I was once asked by a caller on a talk radio show whether the technology of the Antikythera mechanism might have originated in Atlantis. Makes a change from blaming it on aliens I suppose...

Unfortunately the online version of the New Scientist article doesn't have any photos, but click on these links for more information, including pictures, on Abu Qir Bay, Dunwich, Port Royal and Yonaguni.

Archaeologists in Rome investigating Emperor Nero's Domus Aurea, or "Golden House", have discovered a huge circular room, with the remains of spherical mechanisms installed beneath its floor.

 They think this must be what's left of Nero's legendary revolving dining room. It was written about by ancient historians such as Suetonius, but noone ever knew for sure if it really existed. Suetonius said the dining hall was huge and that it revolved "night and day in imitation of the celestial bodies". Nero used it to hold debaucherous banquets that lasted "from noon until midnight".

The Domus Aurea was a collection of party villas, that Nero built in the heart of ancient Rome after much of the city was destroyed in a fire in 64 AD. They were set in landscaped gardens with an artificial lake in the centre. One of the most impressive buildings had a great octagonal hall with a domed roof, and it had been thought that this could be the revolving banquet hall that Suetonius was talking about.

But now archaeologists, led by Mariantoinetta Tomei and Francoise Villedieu, have discovered a new, circular structure; a spectacular find that's unlike anything seen before at a Roman archaeological site. The room is more than 16 metres across, and was supported by a 4-metre thick pillar (see pic below) in its centre. The archaeologists have apparently discovered "four spherical mechanisms" beneath what would have been the room's wooden floor. They are still investigating exactly how the whole thing worked, but they think this machinery must have driven the room's rotation, probably powered by running water.

I've been meaning to post about all this for a while - the find was announced a couple of weeks ago. It got a bit of news coverage at the time (you can watch short videos here and here) though not as much as I would have expected. There's an excellent account of the find, including an excerpt from Suetonius's original description, over at eternallycool.net.

Nero was by all accounts a nightmare as Emperor. He ruled Rome in the first century AD, from 54 to 68, and he was vain, violent and probably deranged. He is said to have kicked his poor wife Poppaea to death, then castrated a freedman who was unfortunate enough to look like her, to act as her replacement. He also loved to perform in public, and once held a whole set of games just to celebrate the shaving of his beard.

Part of the problem may have been that Nero believed he was divine, in particular some scholars think that he identified himself as the Roman Sun god, Sol. So riding in his revolving celestial banquet hall would have been a very important symbolic statement. It reminds me of a theory about Rome's Pantheon, which I wrote about earlier this year (see my New Scientist article, and my blog post). This imposing temple came a little bit after Nero's time - it was completed in AD 128 - and consists of a cylindrical chamber topped by a domed roof. An opening in the top lets through a dramatic shaft of sunlight, which some believe acted as a colossal sundial, marking the timing of the equinoxes. This is thought to have involved similar symbolism, intended to elevate emperors who worshipped there into the celestial realm of the gods.

The first ever autopsy of an Egyptian mummy was carried out in 1825, by an eminent physician called Augustus Bozzi Granville. His work was ahead of its time, but it now turns out he was completely wrong about the cause of death. He also made one other chilling mistake.

The mummy was that of a well-to-do woman called Irtyersenu. She was aged about 50 when she died, and was buried in the necropolis at Thebes, in about 600 BC. Unusually, her organs were preserved intact, making her a perfect subject for scientific examination. Granville found a tumour in her ovary, and pronounced that this is what killed her. But scientists led by Helen Donoghue of University College London recently repeated the autopsy using molecular techniques and found that this tumour was actually benign. Instead they found that her body was riven with the bacterium that causes tuberculosis, a disease that was prevalent in ancient Egypt. Their results are published today, in Proceedings of the Royal Society B.

And Granville's other mistake? In 1825, when he carried out his autopsy, Britain was in the midst of "mummy mania", and he announced his results to a packed crowd at the Royal Society of London. The mummy had been coated with a strange waxy substance, which Granville concluded had been added by the mummy's makers, to help with preservation. To set the mood in his Royal Society lecture, he lit the hall with candles made from this wax.

But Donoghue's team found that the wax in Granville's candles hadn't been added as a preservation agent at all. It had been formed from the decomposing fat of the corpse.

By the way, you might think that diagnosing a 2600-year-old case of TB is impressive, but that's not even close to the oldest known cases, which were identified by Donoghue and colleagues and published in PLoS ONE last year. The victims were a woman and child who lived 9000 years ago in a Neolithic village called Atlit Yam, on the coast of what is now Israel. Sea level rise has since submerged the site, but marine archaeologists were able to scoop up the skeletons from their grave sites in the sediment, preserved in such good condition that DNA from the tuberculosis bacterium had survived within the bones.

Archaeological sites reveal countless insights into how our ancestors lived. But how can we get closer to actually experiencing what life was really like in the past?

One answer is to make computer simulations, such as the fascinating Google Earth reconstruction of ancient Rome. But Alan Chalmers of the Digital Laboratory at the University of Warwick believes that such simulations can be made far more realistic by considering how the reconstructed scenes would have been lit.

Today we're used to the glare of modern halogen lighting, and this is how artefacts such as mosaics, pottery or cave paintings tend to be illuminated when they are on display. But Chalmers argues that this bears little relation to how they would have looked to the original inhabitants.

So he's developing high-powered computing techniques that simulate different lighting conditions from the past, then work out how a particular scene or object lit that way would look. The results can be dramatically different to how we might imagine - for example candlelight contains very little blue, compared to sunlight or modern lighting. The idea is that as well as bringing archaeological scenes to life for the public, researchers can start to ask questions that haven't been possible before.

I've just written an article about Chalmers' latest project for New Scientist. He's working with students at Conimbriga, a Roman town near modern-day Coimbra, in Portugal, which is known for its magnificent House of Fountains. Many mosaics and wall frescoes from this richly-decorated villa are still intact, including a hunting scene on the floor of the Sala da Caçada or "Hunting Room".

By daylight the mosaic looks pretty bland, and computer reconstructions of the entire room give the impression of quite a harsh, dead space. Roman residential houses did not tend to have windows, however, so even during the day the room would have been lit by a series of lamps, or lucernas, fuelled by olive oil.

Chalmers' students, including Alexandrino Goncalves of the Polytechnic Institute of Leiria, Portugal, built a Roman-style lucerna using traditional methods. Then they measured the spectral properties of the light it emitted, and modelled how the hall would look when illuminated by such lamps.

They have just published their results in ACM Journal on Computing and Cultural Heritage. The resulting room has a warm, intimate feeling (see top pic) - quite different from the harsh, dead space it becomes when reconstructed using modern lighting (see bottom pic).

Chalmers has tested the approach at other archaeological sites too, with intriguing results. Modelling the dim candlelight of Medieval British homes, for example, suggests that the reason most pots from this period only had glaze on the top half is because the bottom half was almost always in shadow. And simulations from Cap Blanc, a 15,000-year-old cave site in France, hint that the prehistoric inhabitants who carved horses, deer and bison into the wall may have deliberately left the animals' legs blurred to enhance the effect of movement when the carvings were illuminated by a flickering flame.

The next phase of the work at Conimbriga is to run the simulations using state-of-the-art high dynamic range (HDR) screens, which can display images up to 30 times brighter and ten times darker than normal HD screens - crucial for recreating accurately a dim scene such as a lamplit room - and to incorporate features such as how the human eye adapts to low lighting conditions over time. Ultimately Chalmers also hopes to add authentic sounds and smells, to immerse modern viewers as fully as possible into the past.

Last week I wrote about how Mogi Vicentini's animation of the workings of the Antikythera mechanism drew spontaneous applause from the audience when I showed it during my recent talk at the Royal Institution. (By the way, a few people have said they're having trouble getting the file on Mogi's site to run - I got it to work by downloading a free VLC media player, or you can look at the version hosted by the Guardian in their blog post on my talk - lower res but still beautiful.)

Something else I showed in the talk that drew a satisfying gasp from the audience was a demonstration of the lighting technique, developed by Hewlett Packard's Tom Malzbender, that was used to illuminate the engraved lettering on the Antikythera mechanism. A metre-wide plastic dome, covered with flashbulbs, is used to take photos of an object lit from 50 different directions. The images are fed into a computer and used to make a reconstruction of how the surface of the object reflects light. Once that's done, you can ask the computer to light the object from any angle, even impossible ones like beneath its surface, or you can change how the surface reflects light - such as making the crumbling stone of a cuneiform tablet as shiny as metal (see pic). Then it's just a case of playing around to find the effect that makes the lettering as clear as possible.

You can try this for yourself on Hewlett Packard's website. Click on one of the images (my favourite is the tablet on the top left) to download the interactive demo in a new window, then move the mouse around to change the direction of the light. Or (this is the bit where everyone gasped) right click on the image to bring up a little menu, and under "effects" turn on "specular".

You can also try this with images of the Antikythera mechanism. The bottom right image, AK01a-512, is a good one to try, as it shows a piece of bronze plate from the device, that's covered in inscriptions.

News is also out in New Scientist today that Malzbender has adapted this lighting technique for use with an ordinary flatbed scanner. You can't get quite the same effects of course but the method could be useful for smoothing out creases or tears in images of old photos or documents.

Flatbed scanners use two separate lightbulbs so taking one scan gives you information from two different lighting directions. Rotating the photo or document by 90 degrees and scanning again increases that to four - more than enough for Malzbender's software to glean 3D information about the surface that's being scanned. The software highlights any pixels that aren't lying flat against the scanner (suggesting a rip or fold) and replaces them with nearby ones. Cunning.

It seems that Decoding the Heavens is causing some controversy on the web. In recent weeks, some of the researchers working on the Antikythera mechanism, as well as Anne Bromley (second wife of the late Allan Bromley, another Antikythera researcher) have posted comments expressing concern about the way that certain parts of the book are presented.

As you'll know if you have read Decoding the Heavens, the Antikythera mechanism is an emotionally-charged area of research. All of the researchers involved have devoted years if not decades of their lives to solving its mysteries, and that has resulted in a fair amount of passion and rivalry. In fact without those driving forces they probably wouldn't have reached such impressive results. I for one am in awe of what they have each achieved, and in writing about their work, I'm proud to bring it to a wider audience.

But this also means that there are many disagreements between the various researchers regarding how different parts of the story unfolded, and where the credit for various different discoveries is due. I doubt that any single account could please everyone, but as a journalist I spoke at length to as many people as possible in order to reach my own careful and independent conclusions about what happened. In writing the book I've also tried to give a flavour of the various viewpoints, with different parts of the story seen through the eyes of different people, and it was important to me where possible to portray these scientists as human - reflecting their strengths and weaknesses rather than leaving them as bland, one-dimensional "heroes". I (and my publishers) stand by Decoding the Heavens as an honest and accurate account of the Antikythera story.

The majority of the researchers mentioned in the book are happy with the end result. But of course there are different perspectives and if you are interested in finding out about these then please do look at the comments from members of the Antikythera Mechanism Research Project (you can read them here). These researchers were very helpful and open when I first started reporting on the Antikythera mechanism, but I should note that after I told them in May 2007 that I planned to write a book, some members - Tony Freeth, Mike Edmunds, Yanis Bitsakis, Xenophon Moussas and John Seiradakis - declined to speak to me further (as noted in the acknowledgments of Decoding the Heavens). They said that to do so would conflict with their own plans for Antikythera books and media projects.

Tony Freeth was named to me as the AMRP team's spokesperson on any matters regarding Decoding the Heavens. I offered him the opportunity to comment on the two chapters regarding the team's work before publication but he declined, and in the nine months since the book came out, none of the team has mentioned any concerns about its content to either me or the publisher. I am sorry to hear at this stage that they believe there are inaccuracies, and any factual errors they raise now will of course be corrected in future reprints. I think though that most of the comments they have posted come down to differences in interpretation. For example, I refer to the AMRP group as Freeth's team, whereas they point out that Mike Edmunds was the academic leader of the team, who submitted papers and so on. This is technically correct (Edmunds held an academic position whereas Freeth originally did not), but none of the sources I spoke to were in any doubt that Freeth was the real driving force behind the project.

There are also several places where the AMRP researchers seem to have misunderstood my text. For example they attempt to correct my attribution on p. 101 of a treatise on the astrolabe to Geoffrey Chaucer, saying his authorship is not certain. In fact it is his authorship of a treatise on another astronomical instrument, the equatorium, that is in dispute (as discussed on the next page). His authorship of the astrolabe treatise is well accepted.

Finally, although some comments simply repeat points already made in the book, some do contain extra clarification or information. These are generally details that I chose not to burden the reader with. For example, the researchers note that when Hewlett Packard's Tom Malzbender and his colleagues (none of whom has raised any concerns about the book) flew to Athens "with their flashbulb dome packed in a crate", the dome was taken ahead by couriers rather than being on the same plane as the researchers. However these points may well be of interest to those wishing to dig deeper into events. I hope the team do go on to publish their own books as it will be great to have other versions of the story out there.

 Regarding Anne Bromley's comments about the way her late husband is portrayed (you can read these here), it was not my intention to describe him in a negative way, and I am genuinely surprised by her reaction. The impression I got of Allan Bromley during my research was of a brilliant, lively, forceful, friendly person, who could be manipulative and competitive at times, especially when it came to knowledge and information, but who got things done and was capable of sweeping others along with his enthusiasm. I hope this is the way he comes across in the book, and multiple sources who were close to Bromley in both the UK and Australia have said that they found my account reasonable and fair.

It's good to see Decoding the Heavens provoking discussion though. Please do read the comments and make up your own minds.

Thanks to all those who came along to my talk at the Royal Institution last night, it was great to see such a big turnout. As part of the presentation I was lucky enough to be able to show a gorgeous video of the Antikythera mechanism's workings, which judging from the spontaneous applause it received was a highlight for much of the audience. This animation has just been completed by Mogi Vicentini, an Italian astronomer and computer scientist who specialises in making both physical and virtual models of astronomy instruments. Lots of people came up to me afterwards to ask for more information, so here's a link to Mogi's website, with this page dedicated to the Antikythera mechanism. You can download the animation itself from here, this is the recommended hi res version, but if you have trouble running it you could try a lower res version, here.

However much you already know about this 2000-year-old machine, the video really is breathtaking. Based on the physical reconstruction made by curator Michael Wright, it builds up the gear wheels one by one before adding the front and back dials and finally the wooden case, and it gets across the sheer sophistication of the gearing in a way that I have never seen before. So thank you to Mogi!

UPDATE 6 August 2009:

By the way, a few people have said they're having trouble getting the file on Mogi's site to run - I got it to work by downloading a free VLC media player, or you can look at the version hosted by the Guardian in their blog post on my talk - lower res but still beautiful. 

Just a quick note today to say that I'll be giving a talk on the Antikythera mechanism at the Royal Institution in London on Tuesday 28 July. It's a wonderful venue, and if you're in the area it would be lovely to see you there - more details here.

 The date works perfectly to celebrate the launch of the paperback edition of Decoding the Heavens in the UK - it is officially out on 6 August, but advance copies will be available at the event.

We've all heard about shrunken heads - a ritual practice in which people from Melanesia and the Amazon basin preserved the heads of their enemies by removing the skull, then boiling down the skin and flesh (I love the collection at London's Horniman Museum). It's thought these objects had religious significance - that shrinking the head of anyone you killed in battle stopped their soul from coming back to seek revenge.

The trophy heads of Peru's Nasca people, however, are rather more mysterious. The Nascas lived in an area called the Nazca Drainage in southern Peru, one of the driest places on Earth, in the first to eighth centuries AD. They had sophisticated ceramics, textiles and a complex system of underground aqueducts, but they're best known for carving out the Nazca lines - giant geometrical shapes and animal figures - on the desert floor. It's still not clear what these were for - perhaps for the Gods to look at, perhaps to mark out astronomical alignments relevant to the agricultural calendar, perhaps to mark the routes of cermonial processions. We can probably rule out Erich von Daniken's theory that they were landing strips for alien spaceships...

The Nascas also did an impressive line in preserved heads. This time the skull was the main feature, with a hole drilled in the front so that it could be hung or carried on a cord. Scholars have been arguing for a while over what they were for. These may have had something to do with religious beliefs surrounding agriculture and fertility (which must have been of prime concern in such a dry location) because on some pottery vessels the heads are shown with plants growing out of the top. Other experts have emphasised their role in violence and warfare. One key question was whether the skulls came from enemy warriors from other populations, or from the Nascas' own dead.

Now a team of archaeologists including Kelly Knudson of Arizona State University has measured the levels of different strontium, oxygen and carbon isotopes in the tooth enamel of Nasca trophy heads from throughout the area, and compared them to those from ordinary burials. These isotopes provide information about the diet and geographical origins of the person concerned. I wrote about a similar technique last year, used to show that most of the crew of the doomed Mary Rose came not from England but from southern Europe, suggesting that language problems might have contributed to the ship's demise.

In the trophy heads study, Knudson and her colleagues showed that isotope levels in the trophies are very similar to those from the local population, indicating that the heads came from venerated ancestors, not scorned enemies. (This got mentioned by a few blogs in January, but the paper is in the current issue of Journal of Anthropological Archaeology, so I figured that was another excuse to write about it now.)

This result doesn't mean the heads were taken peacefully, however. Some scholars believe that the Nascas engaged in ritual warfare, undertaken purely to take heads to make into trophies. The very idea sends a chill down my spine, though in such a different culture perhaps being immortalised in this way would have been seen as an honour.

It does always amaze me how differently (and how easily) past societies seem to have viewed death. When the Greek philosopher Posidonius visited the Celtic lands of Gaul in the first century BC, he came back with tales of a violent people who would hang the severed heads of their enemies in doorways. Rowdy feasts would regularly end up in fights to the death, and sometimes, after a few glasses of wine, a man would lie face up on his shield to have his throat cut for the general amusement of the party.

Greek stone inscriptions might look pretty dry and impersonal to us, but actually the lettering hides personal clues as to the identity of the scribe, just as our own scribbled notes do today. Being able to match up which inscriptions were made by the same writer helps scholars to pin down the dates of particular texts more precisely, as well as revealing links and associations between different inscriptions. This has previously taken years of training, and a fair amount of subjective judgment, but now it looks as though computers could start doing the job instead.

Epigrapher Stephen Tracy of the Institute for Advanced Study in Princeton challenged computer scientists who knew nothing about Greek inscriptions to attribute 24 ancient writing samples to their rightful makers - the texts were written by six different people between 334 BC and 134 BC. To his surprise, the team, led by Michail Panagopoulos at the National Technical University in Athens, got every single one right. Tracy reckons their technique is a "real breakthrough" that could now allow Greek inscriptions to be analysed much more quickly and objectively.

Just as in modern handwriting, ancient writers had individual quirks in the way they formed particular letters. So Panagopoulos and his team overlaid digital scans of different examples of the same letter - the As for example - then used a computer algorithm to calculate the probability that they were carved by the same person. They did this with six different letters across all 24 inscriptions, and have written up their report in IEEE Transactions on Pattern Analysis and Machine Intelligence.

Archaeologists have tens of thousands of examples of ancient Greek inscriptions, and the idea now is to build a computer database of as many of them as possible, including scans, attributions and dates. Then any new finds could be slotted into that record. Presumably the technique would also work for other ancient texts, such as cuneiform tablets from Mesopotamia. Though I wonder whether some epigraphers might get a bit defensive about this part of their work being taken over by computers.