Case Study · Landslide Dam · 1925 / 1927
June 23, 1925: A landslide dammed the Gros Ventre River in Wyoming. A 7-mile lake formed. Engineers and geologists arrived and declared the dam not just safe but permanent. Residents who had wisely been sleeping on high ground went home. May 18, 1927: A 50-foot wall of water destroyed the town of Kelly. Landslide dams are made of loose debris. Every one of them will fail. The only question is when.
Teton County, Wyoming · June 1925 & May 1927
At approximately 4:00 PM on June 23, 1925, Rancher Guil Huff was riding on the north side of Sheep Mountain in Teton County, Wyoming, when the mountain began to move. The Buckrail account of the event draws on his firsthand account: he and his horse fled at a full gallop as the slope came down. Behind them, approximately 50 million cubic yards of Sheep Mountain — rock, soil, and debris — slid off the north face and fell 2,100 feet into the narrow valley of the Gros Ventre River below. The slide moved fast enough that Huff barely escaped. It hit the valley floor with enough force to fill the canyon, rise 300 feet up the opposite bank, and block the Gros Ventre River entirely. The river had no outlet. A lake began to form behind the debris dam.
The town of Kelly was located 4 miles downstream from the dam. For months, its residents hauled their bedding up to higher ground at night, fearing the dam would fail. The Jackson Hole History account of the Kelly Flood documents what happened next: engineers and geologists from across the country surveyed the dam. On July 16, 1925 — three weeks after the slide — water levels were still rising, noted at 5 feet below the dam's crest. The experts declared the dam not just safe, but permanent. In the Buckrail account, a ranger recorded their conclusion: "safe, in fact permanent." Forest Ranger Charles Dibble, stationed at a nearby ranger station, was less convinced. The Buckrail account documents that Dibble "would regularly inspect the rubble for evidence of leaks, reporting that water regularly ran through it." But the experts had spoken. The residents of Kelly, tired after months of sleeping on high ground, eventually went home and resumed normal life. The dam that was declared permanent continued to have water running through it.
June 23, 1925
Landslide Date
0
Deaths (1925 slide)
6
Deaths (1927 flood)
2 Years
Between Events
50 ft wall
Flood Height
The winter of 1926-27 was harsh. Heavy snowfall accumulated in the mountains above the Gros Ventre Valley. The following spring brought rain on top of the snowmelt. By May 17, 1927, the Gros Ventre River was rising fast — water was reported to be spilling over the dam. By May 18, Dibble's worst fears were confirmed when pieces of farming equipment began floating down the river — equipment that had been behind the dam. He jumped into his Model T and raced toward Kelly. He met a 50-foot wall of water crashing down the river corridor. The Jackson Hole History timeline documents the sequence: noon on May 18, a 5–6 foot surge followed by a 50-foot wall of water broke through as the weakened dam failed. Within 15 minutes the entire town of Kelly was flooded and entirely destroyed, leaving only three buildings standing: the church, the rectory, and the school. Six people died. Approximately 75 buildings were destroyed. The Cowboy State Daily account of the event quotes the Jackson's Hole Courier from May 19, 1927: "Gros Ventre River Flood Takes Huge Toll in Life and Property." The dam that experts declared permanent in 1925 had lasted less than two years.
The Science
Think of the difference between an engineered dam and a landslide dam this way: an engineered dam is built with controlled drainage, compacted fill graded for strength, and designed spillways that can safely pass a calculated design flood. A landslide dam is built from whatever the mountain was made of — rocks, soil, roots, organic material — deposited in an uncontrolled pile. It has no drainage design, no compaction, and no spillway. The Penn State geology course materials on the Gros Ventre slide describe it precisely: "Such a dam of loose debris is not very strong; water flowing through its porous spaces or over it can remove rocks and weaken it greatly until it collapses catastrophically." The Gros Ventre dam did exactly what its physical composition predicted it would do. Dibble saw the water running through it for two years. The engineers who declared it permanent were wrong not because they were incompetent, but because no loose debris dam is permanent in the long run.
The mechanism by which the Gros Ventre dam failed was likely "piping" — the progressive erosion of a channel through the interior of a porous earth or debris dam by seeping water. Piping begins when the hydraulic gradient (the pressure difference between the upstream and downstream faces of the dam) is sufficient to move individual particles through the dam's porous structure. Once a small channel forms, the flow through it increases, which accelerates erosion, which enlarges the channel, which further increases flow in a positive feedback loop. The seepage that Dibble documented throughout 1925-27 was the observable symptom of this process. The catastrophic failure on May 18, 1927 — the sudden release of the full lake through a dam that took 15 minutes to destroy Kelly — is the endpoint of a piping failure that had been underway for two years.
The Gros Ventre dam held for approximately 23 months before failing. In that 23 months, it held through at least one full snow-melt season (spring 1926). The engineers who visited in July 1925 and declared it permanent had observed 3 weeks of stability. Their assessment, that the dam was "safe, in fact permanent," was extrapolating from 3 weeks of holding to an indefinite future — without modeling the piping process or calculating the dam's capacity to withstand a high-snowmelt season. The USGS account notes that the geological conditions that caused the 1925 slide — saturated weak shale beneath more permeable sandstone — are the same conditions that created the dam's structure. A dam built from such material is inherently vulnerable to internal erosion during high-flow periods.
Timeline
01
June 23, 1925, ~4:00 PM: 50 million cubic yards of Sheep Mountain fall 2,100 feet into the Gros Ventre River valley. Rancher Guil Huff escapes at a full gallop. The debris dams the river, rising 300 feet up the opposite bank. Lower Slide Lake begins forming — eventually ~7 miles long. No deaths from the slide itself. Kelly residents begin sleeping on high ground at night.
02
July 16, 1925: Water still rising behind dam, 5 feet below crest. Engineers and geologists from across the country survey the dam and declare it "safe, in fact permanent." Residents of Kelly, tired of sleeping in tents on high ground, return to their homes and resume normal life. Ranger Dibble, unconvinced, continues inspecting — "water regularly ran through it." He reports this but the expert consensus is permanent safety.
03
Winter 1926–27: Harsh winter, heavy rains, severe spring snowmelt. May 17: River rising, water spilling over dam. May 18: Farming equipment floating downstream — it had been behind the dam. Ranger Dibble races toward Kelly. 5–6 foot surge, then a 50-foot wall of water. Noon: Dam fails. Within 15 minutes: entire town of Kelly destroyed. Only 3 buildings left standing. 6 deaths, ~75 buildings destroyed.
04
1927–1959: The Gros Ventre lessons are documented but not yet formally institutionalized. 1959: M7.3 Hebgen Lake earthquake triggers Madison Slide in Montana, creating a new landslide dam and lake (Earthquake Lake). Army Corps of Engineers — knowing the Gros Ventre precedent — rapidly constructs a spillway to lower the lake level and prevents a repeat of the Kelly Flood. The 1927 disaster directly saves lives in 1959.
Human Decisions
What almost worked
For months after the 1925 slide, Kelly residents hauled their bedding up to higher ground each night. This was the correct behavioral response to an unengineered landslide dam of unknown stability blocking a river above their town. The Buckrail account documents it: "For months, Kelly residents hauled bedding up to higher ground in fear that the newly formed earthen dam would fail in the night." The people closest to the threat had the most accurate instinct about it. They were overruled by expert opinion that proved to be wrong.
Ranger Dibble documented seepage through the dam throughout the two-year period between the slide and the flood. His observations were correct — the piping process he was observing was the mechanism of eventual failure. While his warnings were insufficient to change the expert consensus that prevented Kelly from remaining evacuated, his observations were documented. And the National Park Traveler account of the USGS 1925 centennial piece confirms: the Gros Ventre lessons were directly applied in 1959, when Army Corps engineers rapidly constructed a spillway for the Madison Slide dam to prevent a larger disaster than Kelly.
What failed
The engineers and geologists who surveyed the Gros Ventre dam in July 1925 and declared it "safe, in fact permanent" were applying their best expertise to a novel situation. Landslide dam assessment was less formalized in 1925 than it is today. But the declaration of permanence — as opposed to a more cautious assessment acknowledging the inherent instability of loose debris dams — enabled the return of Kelly residents to their homes below a dam that was visibly leaking. The PSU geology course summary captures the physics: "Such a dam of loose debris is not very strong." The expert declaration of permanence was an overstatement of confidence in a structure that was inherently and predictably impermanent.
The 1959 Madison Slide response — Army Corps engineers constructing a spillway to lower the lake level and controlled drainage — is the engineered answer to an uncontrolled landslide dam. At Gros Ventre in 1925, no such intervention was made. The dam was surveyed and declared safe; no spillway was constructed, no drainage design was installed, and no controlled outlet was created to reduce the water level behind the dam and decrease the hydraulic pressure on its structure. The Penn State geology account notes directly that the Corps' knowledge of landslide dam dynamics, informed by Gros Ventre, was what enabled the 1959 intervention to succeed.
The cascade lesson
The Gros Ventre case study teaches two things simultaneously. First: landslide dams are not permanent. They are made of loose, unsorted debris with no engineering, no controlled drainage, and no spillway. Water flows through them from the moment they form. That seepage progressively erodes their structure until, under sufficient hydraulic load — typically during a high-snowmelt or high-rainfall year — they fail catastrophically. The question is never whether a landslide dam will fail; it is when and how much water will be released when it does. Second: the residents of Kelly who slept on high ground for months had the right instinct. They were overruled by expert consensus that was wrong. When a landslide dams a river above your community and authorities debate whether the dam is stable, the correct individual preparation is not to wait for the expert consensus — it is to take the precaution that Kelly residents originally took and that the Kelly Flood proved was warranted.
What You Can Do Now
The Gros Ventre lesson applies any time a landslide, earthquake, or any other event creates an earthen dam above a community — and any time authorities are deciding whether that dam is stable enough for downstream residents to return.
A landslide dam is not a temporary nuisance that goes away — it is a structural hazard that will fail at some point. The time between formation and failure can be days, months, or years. If a landslide or earthquake-triggered slide dams a river upstream of your community, treat the downstream risk as ongoing and evolving until the dam is either engineered (spillway constructed, lake level lowered) or the dam fails. Don't treat initial stability as evidence of permanent safety.
Landslide dam resourcesLandslide dam failures can produce walls of water moving faster than people can run. The Gros Ventre flood destroyed Kelly in 15 minutes. At a walking speed of 3 mph, a person in the path of a flood moving at the Kelly event's peak speed would have no chance of outrunning it on foot. When any dam failure warning is issued — whether for a constructed dam or a landslide dam — evacuate to high ground perpendicular to the drainage immediately. Don't wait for visual confirmation.
Dam failure evacuation guideEarthquakes frequently trigger landslides. The 1959 Madison Slide in Montana was triggered by a magnitude 7.3 earthquake. If a major earthquake occurs in a mountainous region near you, check whether the USGS or state emergency management has identified any landslide-dammed rivers upstream of populated communities. The USGS and Army Corps of Engineers monitor for this — their rapid response to the 1959 Madison Slide was informed directly by the Gros Ventre precedent.
Earthquake-triggered landslide guideThe Gros Ventre dam held through the relatively normal snowmelt season of 1926 but failed in the high-snowmelt season of 1927. Landslide dams fail when the hydraulic load exceeds the dam's erosion resistance — and the highest hydraulic loads occur during high-snowmelt years and exceptional rainfall events. If a landslide dam exists upstream of your community, spring snowmelt season and major precipitation events are the highest-risk periods. Emergency managers tracking a landslide dam will typically issue heightened watches during these periods.
Flood seasonality guideThe Kelly residents who slept on high ground for months were right. The experts who told them the dam was permanent were wrong. Ranger Dibble, who observed water seeping through the dam and filed reports accordingly, had the correct assessment. When you observe warning signs of geologic instability — seeping water through a dam, cracking on a slope, unusual sounds or movement in soil — and those signs are contradicted by official reassurance, the safest individual decision is to take the precaution the warning sign suggests, not to defer entirely to the reassurance. The Gros Ventre case is the clearest historical documentation of what happens when observable warning signs are overruled by institutional confidence.
Landslide warning signs guideLandslide case study series
Oso 2014 covers documented risk that never became restrictions. La Conchita 2005 covers the same slope failing twice. Bingham Canyon 2013 covers zero deaths through monitoring. Montecito 2018 covers post-wildfire debris flows. Together, they document every major failure mode — and one major success — in the landslide preparedness record.
Full landslide case study seriesSources