Roopkund Lake is located 5,029 metres (16,500ft) above sea level at the bottom of a steep slope on Trisul, one of India's highest mountains, in the state of Uttarakhand.
The remains are strewn around and beneath the ice at the "lake of skeletons", discovered by a patrolling British forest ranger in 1942.
Depending on the season and weather, the lake, which remains frozen for most of the year, expands and shrinks. Only when the snow melts are the skeletons visible, sometimes with flesh attached and well preserved. To date, the skeletal remains of an estimated 600-800 people have been found here. In tourism promotions, the local government describes it as a "mystery lake".
For more than half-a-century anthropologists and scientists have studied the remains and puzzled over a host of questions
Who were these people? When did they die? How did they die? Where did they come from?
One old theory associates the remains to an Indian king, his wife and their attendants, all of whom perished in a blizzard some 1200 years ago.
Another suggests that some of the remains are of Indian soldiers who tried to invade Tibet in 1841, and were beaten back. More than 70 of them were then forced to find their way home over the Himalayas and died on the way.
Yet another assumes that this could have been a "cemetery" where victims of an epidemic were buried. In villages in the area, there's a popular folk song that talks about how Goddess Nanda Devi created a hail storm "as hard as iron" which killed people winding their way past the lake. India's second-highest mountain, Nanda Devi, is revered as a goddess.
Earlier studies of skeletons have found that most of the people who died were tall - "more than average stature". Most of them were middle-aged adults, aged between 35 and 40. There were no babies or children. Some of them were elderly women. All were of reasonably good health.
Also, it was generally assumed that the skeletons were of a single group of people who died all at once in a single catastrophic incident during the 9th Century.
To shed light on the origin of the skeletons of Roopkund, Ayushi Nayak of Max Planck Institute for the Science of Human History and Éadaoin Harney of Harvard Medical School analyzed their remains using a series of bioarcheological analyses, including ancient DNA, stable isotope dietary reconstruction, radiocarbon dating, and osteological analysis. We find that the Roopkund skeletons belong to three genetically distinct groups that were deposited during multiple events, separated in time by approximately 1000 years. These findings refute previous suggestions that the skeletons of Roopkund Lake were deposited in a single catastrophic event, you can download the Report on Roopkund Skulls.
They obtained genome-wide data from 38 individuals by extracting DNA from powder drilled from long bones, producing next-generation sequencing libraries, and enriching them for approximately 1.2 million single nucleotide polymorphisms (SNPs) from across the genome6,7,8,9, obtaining an average coverage of 0.51 × at targeted positions. They also obtained PCR-based mitochondrial haplogroup determinations for 71 individuals (35 of these were ones for whom they also obtained genome-wide data that confirmed the PCR-based determinations). They generated stable isotope measurements (δ13C and δ15N) from 45 individuals, including 37 for whom we obtained genome-wide genetic data, and they obtained direct radiocarbon dates for 37 individuals for whom they also had both genetic and isotope data.
Anthropologists have known about Roopkund Lake for several decades, but little was known about the provenance of its skeletons. Rockslides, migrating ice and even human visitors have disturbed and moved the remains, making it difficult to decipher when and how the individuals were buried, much less who they were. “In a case like this, that becomes impossible,” said Cat Jarman, a bioarchaeologist at the University of Bristol in England who was not part of the research team.
Genetic analysis has helped make some sense of the jumble of bones. The researchers, led in part by Niraj Rai, an expert in ancient DNA at the Birbal Sahni Institute of Palaeosciences in India, and David Reich, a geneticist at Harvard University, extracted DNA from the remains of dozens of skeletal samples, and managed to identify 23 males and 15 females.
The latest five-year-long study, involving 28 co-authors from 16 institutions based in India, US and Germany, found all these assumptions may not be true.
They found that the dead were both genetically diverse and their deaths were separated in time by as much as 1,000 years.
"It upends any explanations that involved a single catastrophic event that lead to their deaths," Eadaoin Harney, the lead author of the study, and a doctoral student at Harvard University, told me. "It is still not clear what happened at Roopkund Lake, but we can now be certain that the deaths of these individuals cannot be explained by a single event."
But more interestingly, the genetics study found the dead comprised a diverse people: one group of people had genetics similar to present-day people who live in South Asia, while the other "closely related" to people living in present-day Europe, particularly those living in the Greek island of Crete.
Also, the people who came from South Asia "do not appear to come from the same population".
"Some of them have ancestry that would be more common in groups from the north of the subcontinent, while others have ancestry that would be more common from more southern groups," says Ms Harvey.
So did these diverse groups of people travel to the lake in smaller batches over a period of a few hundred years? Did some of them die during a single event?
No arms or weapons or trade goods were found at the site - the lake is not located on a trade route. Genetic studies found no evidence of the presence of any ancient bacterial pathogen that could provide disease as an explanation for the cause of deaths.
A pilgrimage that passes by the lake might explain why people were travelling in the area. Studies reveal that credible accounts of pilgrimage in the area do not appear until the late 19th Century, but inscriptions in local temples date between 8th and 10th Centuries, "suggesting potential earlier origins".
So scientists believe that some of the bodies found at the site happened because of a "mass death during a pilgrimage event".
Chemical signatures from the skeletons indicate that the individuals had significantly different diets, adding support to the notion that several distinct population groups are represented.
If accounts of their journeys exist somewhere, none have been uncovered so far. “We have searched all the archives, but no such records were found,” said Dr. Rai.
The researchers note that Roopkund Lake is situated on a route known to modern-day Hindu pilgrims, so perhaps some of the South Asian individuals died while taking part. But that is less likely to explain the presence of individuals from the distant eastern Mediterranean.
Perhaps they weren’t actually Mediterranean migrants, Dr. Jarman said. Their genetic ancestry resembles that of present-day people from Greece or Crete, but current distribution may not apply to ancient populations. Regardless, this group came from somewhere far from Roopkund Lake, for reasons unknown.
Maybe the site held significance for groups with various religious beliefs, said Dr. Jarman. Maybe some of the skeletons were brought for burial, possibly to be left in the lake. Or maybe there were ill-fated explorers — driven by a desire to see a spectacular mountain range, killed by their own curiosity.
But how did people from the eastern Mediterranean land up at a remote lake in India's highest mountains?
It seems unlikely that people from Europe would have travelled all the way from Roopkund to participate in a Hindu pilgrimage.
Or was it a genetically isolated population of people from distant eastern Mediterranean ancestry that had been living in the region for many generations?
"We are still searching for answers," says Ms Harney.