About 3000 years ago, thousands of warriors fought on the banks of the Tollense river in northern Germany. They wielded weapons of wood, stone, and bronze to deadly effect: Over the past decade, archaeologists have unearthed the skeletal remains of hundreds of people buried in marshy soil. It’s one of the largest prehistoric conflicts ever discovered.

Now, genetic testing of the skeletons reveals the homelands of the warriors—and unearths a shocker about early European diets: These soldiers couldn’t digest fresh milk.

Searching for more insight into the battle, researchers sequenced the DNA of 14 of the skeletons. They discovered the warriors all hailed from central Europe—what is today Germany, Poland, and the Czech Republic. Unfortunately, their genetic similarity offers little insight into why they fought.

“We were hoping to find two different groups of people with different ethnic backgrounds, but no,” says study co-author Joachim Burger, a geneticist at the Johannes Gutenberg University of Mainz. “It’s disappointingly boring.”

However, two of the 14 skeletons were women, suggesting a more complex scene than archaeologists had reconstructed.

The study, published today in the journal Cell Biology, turned up a different surprise, too. None of the warriors had the genetic mutation that allows adults to digest milk, an ability known as lactase persistence that’s common in many Europeans.

Other studies have shown lactase persistence was common in parts of Germany by 500 C.E., and widespread across the region by 1000 C.E. So the gene must have spread before that time, but after the battle just 2000 years earlier. That means that within about 100 generations, the mutation had penetrated populations across Europe. “That’s the strongest selection found in the human genome,” Burger says.

The finding only deepens the mystery of lactase persistence. In a 2007 study, Burger showed that Europe’s first farmers, living more than 8000 years ago, weren’t lactase persistent either. At the time, he argued that the mutation gradually spread along with the development of agriculture and herding, a theory supported by signs of milking and cheese- and yogurt-making in Stone Age Europe. People able to digest milk, the argument went, would be able to get more calories from their herds than those without, and more of their children would survive to pass on the gene.

But the Tollense skeletons show that at least 6000 more years went by before the gene for lactase persistence caught on. The DNA results also quash the theory, first proposed in 2015, that the gene for lactase persistence was imported to Western Europe at about 5000 B.C.E. by cow-herding nomads from the steppes of modern-day Ukraine and Russia, the Yamnaya people.

The results leave scientists more puzzled than ever about exactly when and why Europeans began to drink milk. “Natural genetic drift can’t explain it, and there’s no evidence that it was population turnover either,” says Christina Warinner, a geneticist at Harvard University and the Max Planck Institute for the Science of Human History who was not involved with the study. “It’s almost embarrassing that this is the strongest example of selection we have and we can’t really explain it.”

Perhaps something about fresh milk helped people ward off disease in the increasingly crowded and pathogen-ridden European towns and villages of the Iron Age and Roman period, Burger speculates. But he admits he’s baffled too. “We have to find a reason why you need this drink.”