The St. Francis Dam Disaster: Why the 1928 Collapse Happened and What It Changed

Just before midnight on March 12, 1928, a concrete dam in a quiet canyon about 40 miles northwest of downtown Los Angeles gave way. Its reservoir, roughly 12.4 billion gallons of water meant to carry the city through a drought, poured down San Francisquito Canyon and the Santa Clara River Valley for 54 miles to the Pacific. By dawn the flood had killed at least 431 people. It is still one of the deadliest civil engineering failures in American history, and the reasons it happened changed how dams are designed, reviewed and regulated.

The short answer: the St. Francis Dam stood on poor rock on both sides of the canyon, was made taller during construction without being made wider at the base, had almost no protection against water pushing up from beneath it, and was never checked by anyone outside the city water bureau that built it. Investigators still argue over which side broke first, but nearly all of them agree on that list. The failure led directly to California’s state dam safety program in 1929 and shaped how Hoover Dam was reviewed. Here is how the dam came to be, how the night unfolded, what experts agree on and still debate, and what you can see at the site today, with the centennial less than a year and a half away.

Why Los Angeles built a dam in San Francisquito Canyon

By the 1920s, Los Angeles depended on William Mulholland. The Irish immigrant had started as a ditch tender in 1878, taught himself engineering and geology, and rose to run the city’s Bureau of Water Works and Supply. His Los Angeles Aqueduct, opened in November 1913, carried water 233 miles from the Owens Valley by gravity alone. The city kept growing anyway, from just over 100,000 people in 1900 to 576,673 by 1920, and drought cut the aqueduct’s flow. Mulholland wanted a reservoir big enough to hold about a year’s supply, in case of drought, sabotage by angry Owens Valley farmers, or an earthquake that broke the pipeline.

He chose a narrow spot in San Francisquito Canyon, along the aqueduct about 10 miles north of present-day Santa Clarita, between two of the city’s hydroelectric powerhouses. Work began in 1924, the first concrete was poured that August, and the curved concrete gravity dam was finished on May 4, 1926. Water began filling the reservoir on March 12, 1926, exactly two years before the collapse. When full, the lake stretched about 3 miles up the canyon and held about 38,000 acre-feet.

The engineering problems built into the dam

In hindsight, the warning signs were in the ground, in the drawings and in the way the project was run. A peer-reviewed case study by the Association of State Dam Safety Officials (ASDSO) compares the dam with a widely used 1923 engineering text and finds it fell short of the standard practice of its own day in almost every way that mattered.

Bad rock on both sides: the eastern wall of the canyon and the canyon floor were mica schist, a flaky, layered rock that was cross-faulted and laced with talc. The upper western hillside was a reddish conglomerate and sandstone with veins of gypsum, which softens badly when soaked. The contact between the two ran along the inactive San Francisquito Fault, right under the dam’s western end. Mulholland himself had described the unstable schist in a 1911 aqueduct report. The site exploration was thin: according to the ASDSO case study, crews dug no test pits and drilled only four or five shallow borings, all on the western side.

Taller, but not wider: as the city grew, Mulholland raised the planned height twice, by 10 feet each time, so the finished dam stood 185 feet above the streambed and about 205 feet above its deepest foundation. The base was not widened to match. The extra height also forced the bureau to add a 588-foot wing dike along the ridge beside the western end to keep the bigger reservoir in.

Uplift, the hidden force: water seeping under a concrete dam pushes upward, which makes the dam effectively lighter and easier for the reservoir to slide or tip. Engineers had been warning about this uplift for years, especially after a Pennsylvania dam failure in 1911. St. Francis had no grout curtain to slow seepage through the foundation and drainage wells only under its central section, leaving both ends without that relief.

No outside review: California had a civil engineering law by 1917, but city engineering departments were exempt. Nobody outside Mulholland’s bureau had the authority to check the design, and nobody did.

Cracks, leaks and the last inspection

The dam leaked and cracked from the start, though not dramatically. Temperature and contraction cracks appeared as the reservoir filled in 1926, and seepage began where the water reached the fault line on the western side. In 1927 two long diagonal fractures opened on opposite sides of the dam. Mulholland judged them contraction cracks and had them packed with oakum and grouted. Geological engineer J. David Rogers, who has studied the failure for decades, argues in the PBS documentary Flood in the Desert that sealing cracks on the downstream face was the worst possible fix, because it trapped water pressure inside the dam.

Heavy rains in early 1928 brought the lake almost to the top. On March 7 it stood three inches below the spillway, and Mulholland ordered that no more water be sent in. On the morning of March 12, the dam keeper, Tony Harnischfeger, found a new leak on the western side running muddy water, a classic sign that a foundation is washing out. He called Mulholland, who drove up with his assistant, Harvey Van Norman. They spent about two hours at the dam. Van Norman traced the mud to a freshly cut access road rather than the leak itself, and both men left convinced the dam was safe.

The night of March 12–13, 1928

About 11:50 p.m.: Ace Hopewell, a carpenter at Powerhouse No. 1, rode his motorcycle past the dam and saw nothing unusual. About a mile and a half up the road he heard a rumbling like rocks rolling down a hillside, stopped for a cigarette, assumed it was a landslide, and rode on. He was the last person known to see the dam standing and survive.

11:57:30 p.m.: the voltage on the city’s power lines dropped sharply and a transformer blew at Southern California Edison’s Saugus substation, as power lines on the hillside above the dam snapped. The dam broke apart in seconds. Only the center section was left standing, a tall slab reporters soon called "the Tombstone." Pieces weighing up to about 10,000 tons were carried as far as three-quarters of a mile downstream.

The first minutes: a wave roughly 140 feet high swept over the dam keeper’s cottage a quarter-mile below the dam. Harnischfeger and his young son were never found. About five minutes after the collapse, a wall of water still about 120 feet high reached Powerhouse No. 2, a mile and a half downstream, and destroyed the plant and the cottages around it. Of the 67 workers and family members living there, 64 died.

Around 1 a.m.: the flood, now about 55 feet high, had left the canyon, turned west down the Santa Clara River and wrecked the Saugus substation, cutting power to the valley. Near the Ventura County line it hit an Edison construction camp where workers were asleep in tents. Eighty-four of them died.

About 1:30 a.m.: a valley telephone operator learned the dam had failed. She and other operators began calling homes in the flood path, while motorcycle officers and a deputy sheriff raced through the valley shouting warnings. In Santa Paula, a descendant of one family told PBS, officers rode through the streets calling "Agua, agua!" Those warnings are credited with saving hundreds of lives in Fillmore, Bardsdale and Santa Paula, though the water still tore through all three towns.

About 5:30 a.m.: roughly five and a half hours after the collapse, the flood reached the ocean south of Ventura, nearly two miles wide and carrying bridges, orchards, oil, livestock and bodies. California’s Department of Water Resources says it destroyed about 900 houses and swept away 24,000 acres of farmland. Victims washed ashore along the coast as far south as the Mexican border.

What investigators agreed on

At least a dozen investigations started within days. A commission appointed by Governor C. C. Young, led by the dam engineer A. J. Wiley, filed its report on March 24, just eleven days after the collapse. A committee for the Los Angeles City Council, a county coroner’s inquest and independent experts followed. They disagreed on details, but several conclusions held up.

The foundation was the problem: every major panel blamed the ground under the dam rather than the strength of the concrete. The governor’s commission wrote that the western conglomerate, "when wet, became so soft that most of it lost almost all rock characteristics."

Water got under the dam: seepage through weak, fractured rock and the uplift pressure it created were central to every serious explanation. The dam’s own automatic water-level recorder showed the lake level beginning to fall shortly before the collapse, a sign that water was already escaping somewhere.

One man had too much say: the coroner’s jury found "an error in engineering judgment" in choosing the foundation and the type of dam, placed responsibility on the Bureau of Water Works and Supply and its chief engineer, and cleared Mulholland and his staff of criminal blame. Its recommendation became the disaster’s lasting lesson.

The construction and operation of a great dam should never be left to the sole judgment of one man, no matter how eminent, without check by independent expert authority. (Verdict of the Los Angeles County coroner’s jury, 1928)

What is still debated

Which side broke first: the governor’s commission said the western end, on the soaked conglomerate, gave way first. The geologist Bailey Willis and the engineers Carl Grunsky and his son argued for the eastern end, and Willis spotted an old landslide in the eastern hillside. Rogers, building on Charles Outland’s 1963 book Man-Made Disaster, has argued in detail that the ancient landslide on the eastern side reactivated as water saturated the schist, and that version is now the most widely accepted. Historians Norris Hundley Jr. and Donald C. Jackson accept most of his geology but put more weight on how far the design fell short of 1920s practice.

How long the dam was failing: Grunsky read the water-level chart as showing about 23 minutes of falling water before the break; most engineers read it as 30 to 40 minutes. That gap matters, because it bears on whether a slow leak grew into a blowout or a sudden landslide did the damage.

How many people died: the official count in August 1928 was 385. Remains kept turning up for decades, including one victim found deep underground near Newhall in 1992, and many migrant farmworkers who lived along the river were never identified. Historian Ann Stansell’s victim roster puts the toll at at least 431. ASDSO lists more than 432, and the state landmark marker says over 450.

How much Mulholland knew: at the inquest he said, "The only ones I envy about this thing are the ones who are dead," and took the blame in words usually quoted as "If there is any error in human judgment, I was the human." He also insisted he had seen nothing dangerous on March 12. Historians still argue over whether that was honest misjudgment, overconfidence after decades of success, or both.

What changed after St. Francis

California’s dam safety law: on August 14, 1929, the state created a dam safety program and ended the exemption for cities with their own engineers. The state could now review and supervise almost every non-federal dam over 25 feet high or holding more than 50 acre-feet. That program became today’s Division of Safety of Dams, which oversees more than 1,250 dams. The same year, California created a state board to license civil engineers.

Mulholland’s other dam: people living below the nearly identical Mulholland Dam above Hollywood demanded action. Review panels found it too lacked adequate protection against uplift. The city kept the Hollywood Reservoir permanently drawn down and in 1933 and 1934 piled about 330,000 cubic yards of earth against the dam’s downstream face, hiding it behind a grassy slope.

Hoover Dam: the collapse came while Congress was weighing the Boulder Canyon Project, a dam on the Colorado River that would hold back a reservoir about 700 times larger. President Coolidge signed the bill on December 21, 1928. According to PBS, where St. Francis had been largely one man’s creation, Hoover Dam was examined by teams of experts at every stage of its design and construction. If you have visited it, our guide to the Hoover Dam and our look at daily life for the workers who built it pick up that story.

The wider profession: the ASDSO case study notes that the disaster added urgency to the founding of the International Commission on Large Dams that summer, now the main global body for dam safety standards. Mulholland retired by the end of 1928 and died in 1935, two months before Hoover Dam was dedicated. The dam was never rebuilt; the Tombstone was dynamited in May 1929 after a sightseer fell from it, and Bouquet Reservoir in a neighboring canyon replaced part of its storage in 1934.

Visiting the St. Francis Dam site today

What is there: broken blocks of gray concrete, rusted handrail and the pale scar of the old landslide on the eastern hillside. The site is California Historical Landmark No. 919, and its bronze marker stands on the grounds of Powerhouse No. 2 beside San Francisquito Canyon Road, about a mile and a half below the dam site.

How to reach it: after storms wrecked the original road in 2005, San Francisquito Canyon Road was rebuilt on a new route in 2009. The old road past the dam is closed to cars and open to walkers and cyclists; the paved stretch is about 1.6 miles end to end. There are no services, little shade and sections where the old road has crumbled into the creek, so bring water, wear sturdy shoes and stay off the concrete blocks.

The national monument: Congress created the Saint Francis Dam Disaster National Memorial and National Monument on March 12, 2019, the 91st anniversary, covering 353 acres of the Angeles National Forest. The Forest Service signed a management plan in December 2023, and a design by retired Disney Imagineer Greg Wilzbach, with a 50-foot sculpture shaped like the surviving section and a wall of victims’ names, won a public competition that April. There is still no visitor center; the St. Francis Dam National Memorial Foundation is raising money to build one, with the centennial on March 12, 2028.

The bottom line

The St. Francis Dam did not fail because of one bad decision. It failed because weak rock, a design stretched taller than it was drawn, missing defenses against water pressure and an engineer nobody could overrule all lined up in one canyon. The rules that came out of it, independent review, state oversight and licensed engineers, are why most people living below a dam today never think about it. Like the Tacoma Narrows Bridge collapse a dozen years later, it is a disaster engineers still study precisely because the lessons were so expensive.

Related on aglena: The Great Molasses Flood: How a Wave of Syrup Swept Through Boston, The Halifax Explosion: How a Harbour Collision Became the Largest Blast Before the Atomic Age

Further reading