Interesting, actually

9 October 2026 mystery 10 min read

The Avalanche That Left No Avalanche

Nine Soviet hikers cut their way out of their own tent and walked barefoot into a Ural blizzard, never to return. It took six decades, and borrowed code from an unlikely source, to get close to why.

On February 26, 1959, a search party climbing the slopes of Kholat Syakhl, a bald, wind-scoured mountain in Russia’s Northern Urals, found a tent half-collapsed under a crust of snow. It had been there for weeks. A student named Mikhail Sharavin was among the first to reach it, and what he saw was wrong in a way he could articulate but not explain: the tent was half torn down and covered with snow, empty, and all the group’s belongings and shoes had been left behind. Boots. Coats. A mandolin. Rations. Everything a group of experienced winter hikers would need to survive a Ural February — all of it still inside, untouched, while the people who owned it were nowhere to be found.

The tent did not just look abandoned. It looked fled. One wall had been slit open from the inside, not torn by wind or wildlife, but cut, as though someone inside had needed to get out faster than the entrance would allow.

What happened to the nine people who had slept in that tent became one of the most argued-over mysteries of the twentieth century — not because evidence was scarce, but because there was plenty of it, and almost none of it agreed with the rest. It took sixty years, a court record nobody could close, and, improbably, a piece of software originally written to animate a Disney movie about a girl who freezes a kingdom, to produce an explanation that finally fit most of the facts at once. Not all of them. Possibly not even most of them to everyone’s satisfaction. But enough that, for the first time, scientists had a model instead of a guess.

The hike

The group belonged to the Ural Polytechnic Institute’s hiking club, and in late January 1959 ten of them set out to ski and climb toward Otorten, a peak whose name in the Mansi language is popularly — if perhaps too neatly — rendered as “don’t go there.” They were led by Igor Dyatlov, a 23-year-old radio engineering student, and the group was experienced: several held “Grade II” hiking certifications, the level below the most senior Soviet rating. A tenth member, Yuri Yudin, turned back on the second day after knee and joint pain, aggravated by rheumatism, made the trek unbearable. He went home. The other nine kept climbing. Yudin lived another five decades, long enough to be interviewed repeatedly about the friends he never saw again, and he reportedly said that if he ever got the chance to ask God one question, it would be what actually happened to them that night.

Dyatlov had told his sports club he expected to wire a telegram once the group reached the town of Vizhai on their way back, and that he didn’t expect to arrive any later than February 12. No telegram came. For over a week, nobody worried much — long winter treks ran late often enough that a delay wasn’t alarming on its own. It was only when families began demanding a search in the last days of February that a rescue operation got underway, and it was searchers, not the hikers, who found the tent on the 26th.

What the snow recorded

From the slit in the tent, footprints led away in a single file line, down toward a line of trees roughly a mile and a half below. The hikers had clearly left in a hurry: some tracks were made by boots, but others were left by people in only socks, or a single shoe, or nothing on their feet at all, in temperatures that investigators later estimated were well below freezing. After about 500 meters the tracks vanished under fresh snow, and from there the story had to be reconstructed from where bodies eventually surfaced.

The first two were found at the treeline, beneath a large cedar, dressed in little more than their underwear. Nearby were the remains of a small fire and broken branches high up the cedar’s trunk, suggesting someone had climbed it, perhaps to look back toward the tent or to spot a route forward. Three more bodies were found between the cedar and the tent, in postures that suggested they had been trying to return to camp when they died — one of them, Rustem Slobodin, had a skull fracture serious enough that investigators debated whether it alone could have been fatal, though the official finding attributed his death to hypothermia.

The remaining four were not found until May, when the snow finally receded enough to reveal them in a ravine further downstream, under several meters of snowpack. Their injuries were markedly worse. Nikolai Thibeaux-Brignolles had a fractured skull. Lyudmila Dubinina and Semyon Zolotaryov had severe chest trauma — multiple broken ribs in patterns one investigating expert likened to the kind of blunt force produced by a car moving at high speed, with no corresponding skin wounds, as though the force had come from something broad and heavy rather than sharp. Dubinina’s body was further missing her eyes, her tongue, and part of her upper lip; accounts of exactly how many of the nine were missing eyes or other soft tissue vary somewhat between sources, a discrepancy likely owed to decomposition in running water over three months as much as to the original trauma, and the most common explanation — that the missing tissue was removed by small scavengers after death — is plausible but unproven. Investigators also found that some of the four had been wearing clothing cut from the garments of the already-dead, as though the survivors, dressed for nothing more than sleep, had stripped what they could from companions who could no longer object. Tests on some of that clothing turned up low levels of radioactivity — not dramatic by any means, but higher than background, and never satisfactorily explained; one of the group had previously lived in a town near the site of the 1957 Kyshtym nuclear disaster, and camping lanterns of the era sometimes used thorium, but both explanations remain speculative.

A Soviet investigation closed the case within a month of the final bodies being recovered, attributing the deaths to an unnamed “compelling natural force,” and sealed the files. The area was closed to hikers for three years.

Sixty years of guesses

In the vacuum that followed, nearly every theory imaginable attached itself to the case. Soviet weapons testing. Infrasound that could induce panic without being heard. A secretive encounter with local Mansi reindeer herders — ruled out early on, since the only tracks at the scene belonged to the nine hikers themselves, and nothing suggested a struggle with an outside party. Eventually, inevitably, cryptids and aliens. None of it held up to the physical evidence, and none of it explained why nine capable, well-equipped mountaineers would cut their way out of a functioning shelter into a nighttime blizzard rather than simply walk out the door.

An avalanche had been proposed almost from the start, and dismissed almost as quickly. The slope above the tent measured only about 28 degrees — shallower than the roughly 30-degree threshold generally considered necessary to trigger a slab avalanche. Searchers found no telltale avalanche debris when they reached the site weeks later. And the timing seemed wrong: the hikers appeared to have already settled into the tent, even started preparing to eat, before whatever happened, happened — not the kind of immediate, violent impact an avalanche typically delivers the moment a slope gives way.

An unlikely piece of software

In 2021, two engineers — Johan Gaume, who runs an avalanche-simulation lab at Switzerland’s EPFL, and Alexander Puzrin, a geotechnical engineer at ETH Zurich — published a new model in Communications Earth & Environment, a Nature Research journal, that tried to answer the objections rather than wave them away.

Gaume’s path to it ran through a children’s film. He had been struck by how convincingly snow moved and settled in Disney’s Frozen, and reached out to the studio’s snow-effects specialists to ask how they had done it. With their permission, he adapted the actual simulation code behind the movie’s animated snow into a physics model of how a real slab of snow could fracture and slide.

Using that model, Gaume and Puzrin proposed that when the hikers cut into the slope to carve out a flat platform for their tent — a completely ordinary thing to do, and something winter campers still do today — they removed a chunk of the snowpack’s structural support. Over the following hours, wind continued depositing snow above the cut, building a slab that the disturbed slope could no longer fully bear. Their model showed that a slab as small as roughly five meters by five meters could, under the right wind-loading conditions, release hours after the tent was pitched rather than the moment it was cut — explaining both the delay and why rescuers found no obvious debris field: a slide that small, in a group of such limited size, would have left only a modest smear of disturbed snow, one that three weeks of further wind and drift could easily erase before searchers arrived. The pair also argued that the 30-degree rule of thumb has documented exceptions, meaning the shallow slope was a complication, not a disqualifier.

To test whether a slide that size could produce the injuries recorded in the autopsies, the team turned to a very different data set: crash tests General Motors had conducted on human cadavers in the 1970s to study how bodies respond to blunt impact. Modeling a small slab’s momentum against a body lying in a tent, rather than a person flung through open air, they found it could plausibly produce the kind of localized rib and skull fractures seen in several of the hikers — without the kind of total burial or crushing that would have killed everyone outright or buried the tent completely. In their model, injured, disoriented hikers in the dark, with no light but wind and snow around them, cut their way out rather than risk the entrance, fled downhill away from a slope they feared might slide again, and succumbed over the following hours to a combination of trauma, shock, and plunging temperatures they were, by then, barely dressed to survive.

Not a final answer

The model didn’t resolve everything, and its authors never claimed it did. Gaume told reporters following the paper’s publication that several parts of the Dyatlov Pass mystery will never be explained, because no one survived to tell the story — the radiation traces and the missing soft tissue remain outside what an avalanche model can speak to, left instead to the older, unproven explanations of contamination and scavenging. Other avalanche researchers pushed back on the paper itself, maintaining that the slope angle and the lack of any debris, however small, still sat uneasily with a slab-avalanche explanation, and that the reconstruction relied on assumptions about wind loading that couldn’t be verified sixty years after the fact.

Still, the paper landed amid real convergence rather than in isolation. Russian prosecutors had reopened the case in 2019 and, independently, arrived at a version of the same slab-avalanche theory. And according to EPFL, the intense press coverage that followed Gaume and Puzrin’s 2021 study sent new expeditions out to Kholat Syakhl itself — expeditions that reportedly confirmed small, localized avalanches of exactly the kind their model proposed do occur on that slope under the right conditions, lending at least physical plausibility to a mechanism that had been dismissed outright for sixty years.

What’s left is not a solved case so much as a narrower one. Nine people died on a mountain whose name warned people away from it, for reasons that no longer require inventing weapons tests or cryptids to explain, even if every last detail — the radiation, the missing tissue, the precise order of panic in the dark — may simply have gone into the ground with them. Yuri Yudin, who turned back for a bad knee and lived fifty-four more years because of it, never got his answer either. The best anyone has managed since is a model built, in part, on snow that never actually fell — rendered first for a cartoon kingdom, then repurposed to explain a real one.

Sources