About 15,000 years ago in Gough's Cave, near Bristol in the UK, a group of people ate parts of each other.
They de-fleshed and disarticulated the bones, then chewed and crushed them. They may also have cracked the bones to extract the marrow inside.
It was not only adults that showed signs of being eaten. A three-year-old child and two adolescents all had the tell-tale marks of being nibbled on.
Some of their skulls were even modified into ornaments called "skull cups", which may have been used to drink out of.
What was going on in Gough's Cave? Was this an example of human violence between rivals, a strange kind of ritual behaviour, or simply a desperate bid for survival?
Gough's Cave was first excavated in the late 1880s. However, at the time it was intended for tourists, so there was no careful archaeological analysis. Any fossils found could well have been lost.
It was another round of excavations in the 1980s that uncovered numerous human and animal remains, with clear signs of butchery. After extensive analysis over the following decades, researchers concluded that the human bones from this cave were cannibalised.
It serves as a reminder of the dark, gruesome side of human nature
Gough's Cave is far from a lone instance. Evidence for cannibalistic behaviour goes back to at least as far as the Neanderthals – and possibly earlier.
The first evidence that Neanderthals possibly munched on their friends (or enemies) was found in France and dates from about 100,000 years ago. There are several other cases. One Neanderthal group, which lived in a cave in northern Spain 49,000 years ago, appears to have been cannibalised. Bones from a cave in Belgium also show signs that another group ate each other, according to a study published in July 2016.
Cannibalism is a topic that has sat uncomfortably with anthropologists. In part, that may be because it serves as a reminder of the dark, gruesome side of human nature. But it is also difficult to unambiguously identify.
In particular, it is difficult to tell whether cut marks come from cannibalism or from the removal of flesh from bones after death, which is occasionally done for ritual purposes.
Cut marks on animal bones are usually a clear sign of butchery. But when similar marks are found on human remains the conclusions are "not well accepted", says Silvia Bello of the Natural History Museum in London, UK.
The remains from Gough's Cave, which Bello has been analysing for several years, are covered in a great many marks. Over 65% of the bones show marks from stone tools. What's more, a lot of the bones were smashed, presumably to extract bone marrow.
They were probably chewing to suck the grease off
"It is quite symptomatic," says Bello. "You can see the same type of pattern on the other animals. You can really say they were butchering the animals in the same way, [for] the meaty bits."
Bello has even found human tooth marks. The pitted marks teeth make as they crunch into bone were easily identified.
"Some of the modification, particularly for the fingers, [shows] they were probably chewing to suck the grease off," says Bello. "The presence of human tooth marks on human bones is probably the best evidence of cannibalism."
Bello decided to try and resolve some of the controversy that arises when remains with these type of marks are discovered.
In a study published in August 2016, she and her colleagues painstakingly outlined the differences between cannibalism and "de-fleshing". This is a process where flesh is removed, but not usually eaten, from a person after they have died.
De-fleshing was often done for the purpose of "secondary burials": prehistoric people habitually took their dead with them when they moved to other areas, and to do so they first removed their flesh. The bones may have been "more significant" than the flesh, says Rosalind Wallduck, also of the Natural History Museum, who worked with Bello.
Their ancestors would have been more present in their lives
There is abundant evidence of this in Serbia between 10,000 and 5,000 years ago. "People were still moving around seasonally and taking their dead with them as they moved," says Wallduck. "Their ancestors would have been more present in their lives, and they were interacting with their remains."
The team says cannibalised human remains should now be easier to recognise, because they have a distinctive signature. "They really fit with the other butchered fauna on the site, while secondary burial has a different signature, in terms of percentage of cut marks and distribution," says Bello.
The key difference is the frequency of the cut marks. When animals or humans are butchered, there are lots of easily-observable marks, but if they have been de-fleshed there are smaller marks, on just 1% of the bones, with no tooth marks.
However, this still does not answer the question of why these ancient humans chose to chomp on human flesh.
There is no way to get a definitive answer. However, Bello says the people who lived in Gough's cave show no signs of coming to a violent death. "They were eating the dead body, but there was no killing."
Eating others' flesh may have been a cultural practice, rather than something done purely for survival
It may be that the environment in which they lived drove them to extreme behaviour. They lived in a brief respite between two very cold snaps during the last ice age: the climate was only warm for a few hundred years. A particularly cold winter may have depleted their food resources, forcing them to eat any individuals who died in order to survive.
But that does not explain the skull cups found at Gough's Cave. These skulls do not appear as if they were modified for their contents to be eaten, and similar skull cups are found elsewhere.
Instead, Bello thinks the people living in Gough's Cave made the skull cups as part of a ritual or burial practice. Quite possibly they drank out of them. If that is the case, eating others' flesh may have been a cultural practice, rather than something done purely for survival.
The age of the individuals provides another clue. If the people died as a result of conflict, there would most likely only be adult victims of cannibalism, but that was not the case. "It's not [a] warrior group," says Bello. "I think they just died naturally."
This may be so in this instance, but more recent cases of cannibalism can paint a very different picture.
Cannibalism is usually the result of violence, and the victors take the "extra step of insulting the dead", says Rick Schulting of the University of Oxford in the UK.
Schulting argues that cases of cannibalism are relatively rare, despite the high frequency of violence in our past. "We find lots of evidence of people dying in conflict in pre-history where you don't eat, butcher or dismember them," he says. "So it does imply something more extreme."
One such instance of cannibalism took place almost 900 years ago at a Puebloan site in the American South West (present-day Colorado).
It happened during a time of severe drought, but Schulting says that does not mean people ate each other for food. He thinks it is more likely that people were eating each other due to "severe conflict over resources".
"People were fighting each other and, as a warning not to come back, those who did would be horribly butchered and smashed up, possibly tortured," he says. "So these extreme acts become signals."
It is even possible that the individuals at Gough's Cave were killed after all, Schulting says. There is no observable evidence of violence, but there are many ways to kill that do not leave a mark on the skeleton.
Perhaps the lesson from all this is that there is no one-size-fits-all explanation for cannibalism. "Quite often violence does seem to be implicated when you see this kind of extreme process," says Schulting. "But even that's open to discussion and debate."
Regardless of why it happened, it is clear that cannibalism has occurred at various times throughout our story. It is another, albeit gruesome, aspect of what makes us human. And it is not obvious that it is something to be especially ashamed about, at least not when you compare it to some of the other atrocities humans have committed.
"I don't think [cannibalism] makes us any more savage," says Schulting. "We are perfectly capable of savage acts today, whether or not that involves eating somebody."
The Natural History Museum gratefully acknowledge the loan of Gough's Cave material from the Longleat Estate.
Melissa Hogenboom is BBC Earth's feature writer. She is @melissasuzanneh on Twitter.
If you liked this story, sign up for the weekly bbc.com features newsletter called "If You Only Read 6 Things This Week". A handpicked selection of stories from BBC Future, Earth, Culture, Capital, Travel and Autos, delivered to your inbox every Friday.