On a warm November afternoon, archaeologist David Kilby walks along a remote, cactus-studded ridgeline in West Texas, near the border of Val Verde County, and imagines a scene that unfolded there thousands of years ago. He points to a low hill, then a narrowing funnel of boulders leading to the lip of a rugged canyon. Here, researchers believe, bands of prehistoric Texans drove herds of bison toward a notch in the cliff wall, where the animals plunged ninety feet to their deaths.

“Think about the people standing behind this hill, and others popping up out of that arroyo waving blankets,” Kilby says, sweeping one arm wide as he envisions the scene.

The bison, panicking as they tumbled into the gorge, couldn’t stop as more creatures crowded in behind them. The noise must have been terrible—animals bellowing in fright, crashing onto the rocks below, and moaning as they lay dying. Excavations at the Bonfire Shelter, a rocky overhang beneath that notch, have raised as many questions as they have answered over the past half century. “Bonfire Shelter doesn’t give up her secrets easily,” Lee Bement, an archaeologist who led an excavation there in the eighties, once told Kilby.

Bison jumps were a hunting method once used by ancient peoples across North America; there are about 150 other known sites, mostly in Canada and Montana. The technique took advantage of the animals’ herding behavior: hunters drove them toward an unseen cliff or drop-off, over which they plunged. Other hunters waited below to dispatch the injured animals. In some places, hunters used rock cairns or brush to create lanes that steered the bison toward the jump. No such features have been found at Bonfire Shelter, but that doesn’t rule out the possibility of their use, or that of fire, to control the direction of the herd. 

Bonfire Shelter is by far the southernmost jump that’s been discovered, and it just might be the oldest—by some six thousand years. This would suggest that some of Texas’s earliest native hunters were even more organized than researchers thought.

Kilby, a professor of anthropology at Texas State University and director of the Ancient Southwest Texas Project, has spent the last four years sifting through layers of burned bison bone and sediment, looking for clues to Bonfire’s mysteries. He sees the most recent excavation, which he’s leading along with Marcus Hamilton of the University of Texas at San Antonio, as a sort of cold-case investigation into Texas’s earliest history.

In the 1950s, local lore held that broken bones found scattered in a crescent-shaped rock shelter on a private ranch in western Val Verde County were the remains of a lost herd of cattle. But when Michael Collins, a teenager with a budding interest in archeology whose family knew the landowner, saw them, he suspected they weren’t just from wandering cattle. (Collins went on to become a noted archaeologist and director of the Gault School of Archaeological Research.) The landowner, who has asked to remain anonymous to protect the site, contacted the University of Texas. Then, in 1963, a team led by archaeologist David Dibble, together with paleontologist Dessamae Lorrain, began the first excavation of Bonfire Shelter. The team burrowed into the mound of sediment that had formed beneath the notch, exposing a dense, nearly three-feet-thick layer of burned bison bone, along with a scattering of prehistoric spear points. They estimated that an astonishing eight hundred animals were driven off the cliff and butchered here, although archaeologists now believe that number is much smaller. 

Almost immediately, questions arose. The researchers couldn’t tell if hunters had burned the carcasses intentionally after the kill, perhaps to quell the stench or stop them from attracting scavengers, or if they burned naturally. The latter could’ve resulted from spontaneous combustion (decay creates heat, and piles of carcasses can burst into flame) or a grass fire.

The archaeologists dug deeper into the mound. Beneath that uppermost layer of bone, now called Bone Bed 3, they uncovered a second, much older layer of bone, this one less dense and less burned. They estimated that the 120 animals in this layer, called Bone Bed 2, were bigger: an Ice Age version of bison that stood eight or nine feet tall at the hump and weighed about two thousand pounds each. They also found spear points known to be between twelve thousand and thirteen thousand years old.

There’s little debate among archaeologists that the bones in the top layer resulted from an intentional bison jump about 2,500 years ago. It’s the second layer that sparks controversy. If this deeper layer was also an intentional bison jump, Bonfire Shelter would represent the oldest bison jump in North America—by a lot. Not only that, as Dibble and Lorrain wrote in scientific articles, this second layer represented three different killing events.

Finally, Dibble’s team opened another pit at the back of the shelter, where they discovered the oldest bones of all—remains of ancient horses, bison, camel, antelope, and mammoths. What they didn’t find in this deepest layer, known now as Bone Bed 1, were stone tools. That led to more questions. How did these ancient bones get here? How did the animals die? They couldn’t tell for certain, but they strongly suspected humans were involved.

In 1983, two more University of Texas researchers returned to the site, determined to wrestle more information from it. Solveig Turpin and Lee Bement, who is now a leading researcher at the Oklahoma Archaeological Survey, wanted to know if humans had a hand in the scattering of Pleistocene bones found in Bone Bed 1, the deepest and oldest layer. 

Ultimately, Turpin and Bement came to the same ambiguous conclusion as Dibble and Lorrain: they couldn’t determine how the animals, whose bones were radiocarbon-dated to about 14,000 years ago, got there. Although they believed it was possible that humans had killed or butchered them, they found no stone tools or other definitive evidence to prove it. All they found were puncture wounds and gnaw marks on the ancient bones.

Then, more intrigue. Back in 1978, archaeology theorist Lewis Binford had declared that the middle layer of bones at Bonfire Shelter looked more like a butchering site than a primary kill site. That prompted Southern Methodist University archaeology professor David Meltzer, who had excavated other bison jump sites, to test the claim with a new study of Bonfire Shelter in 2003. Meltzer and several graduate students reanalyzed bones collected during the 1960s excavation and took a fresh look at the site. They noticed several things. 

First, the bones in this second layer were widely scattered, not in a dense heap. That could indicate that the bison had been killed on the canyon floor and dragged to the shelter for butchering. But, they added, data mapping suggested that Bonfire Shelter would be an ideal location to run bison over a cliff.

“If you get a bison herd moving at high speed, they’re not going to see the jump point until they’re right on top of it,” Meltzer says. “It’s a fabulous place to have a jump kill.”

Yet they couldn’t find hard evidence to prove that any of the jumps dated to 14,000 years ago. At a primary kill site, researchers expect to find skulls and other discardable parts. A butchering-only site would contain a different set of bones, such as ribs and upper limbs—the parts heavy with meat. He also noted that although the shelter is located high off the canyon floor today, thousands of years ago it wouldn’t have been so hard for hunters to drag their kill there. Repeated flooding has lowered the canyon floor’s level.

Bonfire Shelter Bison Jump Site
The canyon wall at Bonfire Shelter in West Texas, where researchers believe that Native Americans hunted bison en masse. Courtesy of David Kilby
Bonfire Shelter Bison Jump Site
Marcus Hamilton of UTSA excavates the jawbone and ribs of an Ice Age bison using a bamboo tool and brush. Courtesy of David Kilby

“The bottom line is I don’t know, because there’s too much ambiguity,” Meltzer says. 

Archaeologists continued to debate the mysteries of Bonfire Shelter in academic journals until Kilby finally picked up the case. Armed with a battery-operated Makita mini leaf blower to quickly clean up exposed bone and artifacts, Kilby descended into the site’s existing pits in 2017 to examine them once more. He focused his research on the two deeper layers of Bone Bed 1 and Bone Bed 2. “We’ve got two cold cases we’re looking at here,” Kilby said.

First, how did the Pleistocene bones in the deepest layer get there, and did humans have a hand in it? And second, were the bison in the middle layer driven over the notch in the cliff intentionally by humans, or (as happened at other sites) were they killed elsewhere and dragged to Bonfire Shelter for butchering? 

Unlike his predecessors, Kilby had some new tools packed in his detective kit.

“Forty years have passed, and we can do some cool stuff now,” he says. Radiocarbon dating has become more exact, and there are new techniques including three-dimensional modeling and micro morphology, in which scientists inject plastic resin into sediment, let it harden, and then cut it into narrow slices with a rock saw to enable microscopic examination of sediments in place.

Crews are now refilling the excavation pits at the shelter, burying the scene of the killings to protect it from the elements. In a Texas State lab, Kilby and his team are examining the thousands of samples they have collected—enough bags of sediment to fill a bedroom, plus ten storage boxes packed with stone tools and bone fragments. 

It will take several years to measure, study, and record the evidence. The findings will be important, and Kilby thinks he has some good clues. He’s found articulated body parts, jaws, and tails in Bone Bed 2. If it was only a butchering site, he says, the hunters probably would have left those undesirable parts in the arroyo instead of hauling them into the shelter to butcher.

Breakage patterns in bones provide hints too. “If it’s a drive over the rim, we would expect to see breakage patterns in the animal bones consistent with falling from a great height,” Kilby says. “And that’s what we see . . . This to me is clearly a jump. It’s concentrated under a funnel in a big pile with broken bones. That’s a bunch of circumstantial evidence to overcome if you’re trying to prove a jump wasn’t the weapon.” 

If that middle layer of bones at Bonfire Shelter does represent an intentional bison jump, it represents one that occurred thousands of years earlier than anywhere else, Kilby says as he points out ancient bison teeth and bones embedded in one excavated wall of the debris cone. 

That would mean that some of North America’s earliest hunters were even more savvy than scientists believed. “There’s a level of social organization on the part of twelve-thousand-year-old Paleo-Indians that we didn’t know about before,” he says. “The coordinated cooperation of multiple groups required to pull this off is not seen before in the archaeological record.”

And because two types of projectile points, from slightly different time periods, were found in this middle layer, Kilby says he leans toward interpreting the site as more than one hunting event—maybe two or three.

As for Bone Bed 1, the deepest layer? In the lab, Kilby’s team will look for the most minuscule signs of man-made tools. They’ll water-screen sediment, searching for fragments the size of a grain of sand that would indicate humans were here nearly 15,000 years ago. If that happens, he’ll be convinced that Bonfire is one of the earliest known occupied sites in North America.

“So far we haven’t found anything that can’t be explained by animal behavior,” Kilby says. “If this is a cold case, the saber-tooth cat is one of our suspects. I’m less convinced it’s the work of humans; I think it’s a predator den.”

This is a crucial distinction, in part because bison jump sites are relatively uncommon. The ones archaeologists know about are in Montana and Canada and are not nearly as old as the middle layer of bones found in Bonfire Shelter. “This is thousands of years older and hundreds of miles farther south,” Kilby says. 

It will take two years to analyze the samples, then several more to summarize the findings in articles. Eventually, Kilby hopes to write a book—no doubt a whodunnit—about the project. In the meantime, the scientific community is eager to review Kilby’s findings. “I think everyone would just like to know what the evidence supports,” Kilby says.

Meltzer, the archaeologist from SMU, is among those paying close attention. “Ten thousand years ago, with the possible exception of Bonfire, [humans] were not running bison over cliffs,” he says. “They were maneuvering them into arroyos and dispatching them in sand dunes, but not doing the whole stampede thing.” That doesn’t mean Bonfire Shelter isn’t an anomaly, Meltzer says, but he’s not yet fully convinced: “I’ll be interested to see the evidence Kilby presents.”

Back on the cliff above Bonfire Shelter, Kilby thinks once more about the significance of this bleak and prickly spot, and the importance of what happened here. And he hopes that his latest research will help solve a cold case that dates back thousands of years.