Shortly before midnight on March 12, 1928, the St. Francis Dam in San Francisquito Canyon, in northern Los Angeles County, gave way. That same morning its builder, William Mulholland, had inspected a new leak and judged it safe. The flood ran roughly 54 miles down the Santa Clara River valley to the Pacific. By the most widely used modern count it killed at least 431 people, although the true number is uncertain. This guide covers the foundation rock, the day of the failure, the flood path, the inquest record and the 1929 California law that put city-owned dams under state supervision.

Mulholland’s dam in San Francisquito Canyon: who built it and how it grew

The St. Francis Dam belonged to the City of Los Angeles. The city’s Bureau of Water Works and Supply built it, and that bureau was the forerunner of the Los Angeles Department of Water and Power. Its chief engineer and general manager was William Mulholland. He had come to Los Angeles as a young man without formal engineering training and learned the water business on the job. He went on to direct construction of the Los Angeles Aqueduct from the Owens Valley, which opened in 1913. By the 1920s he was one of the best-known water engineers in the country, and inside his own bureau his judgment on a dam was effectively final.

The reservoir sat along the aqueduct line in San Francisquito Canyon. Its purpose was to hold a reserve of water close to the city in case the aqueduct was cut. That was a practical worry, because the aqueduct had been dynamited more than once during the conflict with Owens Valley landowners in the mid 1920s. Work on the dam began in 1924 and it was completed in 1926. It was a curved concrete gravity dam, which means it relied mainly on its own weight to resist the push of the water.

During construction the planned height was raised twice, by about 10 feet each time according to most accounts, to enlarge storage to roughly 38,000 acre-feet. Published accounts generally agree that the base was not widened in proportion. The bureau did its design and review in-house. As most histories describe it, the California dam law in force at the time exempted dams built by cities that had their own engineering departments. As a result, no outside engineer and no state official had authority over the design, the foundation work or the decision to raise the dam.

What follows comes from the published record. That includes the report of the commission appointed by Governor C. C. Young, the record of the Los Angeles County coroner’s inquest, and the later historical work. The main books are Charles Outland’s Man-Made Disaster (1963), Norris Hundley Jr. and Donald C. Jackson’s Heavy Ground (2015) and Jon Wilkman’s Floodpath (2016), alongside the papers of engineering geologist J. David Rogers. These sources disagree on several points, including how the failure began and how many people died, and I say so where they do.

The foundation: schist on one side, red conglomerate on the other

The dam was keyed into two quite different kinds of rock, which met along a fault under the western part of the site. The eastern abutment and much of the canyon floor rested on Pelona Schist. This is a mica schist that splits readily along its layering, and on the eastern wall those layers ran roughly parallel to the slope, an arrangement that geologists recognize as favorable to sliding. The western abutment rested on a reddish conglomerate and sandstone. The investigating commissions in 1928 found that this rock was weak and lost much of its strength when wet.

The commissions put weight on simple tests in which samples of the red conglomerate broke down after soaking in water. Their account of the failure followed from that finding. Reservoir water seeped under and around the western end of the dam, softened the conglomerate and carried it away until the western section lost its support, and the failure then spread across the structure. The governor’s commission concluded that the dam failed because of its foundation. It did not find a defect in the quality of the concrete.

J. David Rogers later reexamined the site and the record, and in the 1990s he published a different sequence. He argued that the eastern abutment sat on an ancient landslide mass in the schist, which the geological practice of the 1920s was unlikely to have recognized. In his account, saturation of that slope and water pressure pushing up under the dam, called uplift, destabilized the eastern side first. He also argued that the design did too little to relieve uplift. His work is widely cited, though other writers continue to emphasize the western conglomerate. I treat the order of failure as an open technical question. The point both accounts share is that the rock beneath the dam failed.

For emergency managers, the inquest’s concern matters more than the geological argument. That concern was that a foundation in mixed and faulted rock had been accepted by the builder’s own judgment, with no independent geological review, and that the dam had then been made taller on that same foundation.

The morning of March 12: a muddy leak and the builder’s inspection

The reservoir filled through 1926 and 1927 and stood close to its spillway level in the days before the failure. Cracks and seepage had appeared during filling. Mulholland described them as the ordinary contraction cracks found in large concrete dams, and they were sealed. The damkeeper, Tony Harnischfeger, lived with his family in the canyon below the dam and was responsible for watching it day to day.

On the morning of March 12, Harnischfeger found a new leak near the western abutment and telephoned the bureau. By the accounts given afterward, the water looked muddy to him. Muddy seepage at a dam is a warning sign, because it can mean the water is carrying foundation material out from under the structure. Mulholland drove up that morning with his assistant chief engineer, Harvey Van Norman, and the two men examined the leak. They concluded that the water was clear where it emerged and had picked up its color by running across loose material from a newly cut access road. Mulholland later testified that he considered leaks of that kind normal for a dam of that size, and the two men left the site.

A number of accounts describe the evening that followed. They come from people who said they drove or rode past the dam in the hours before the collapse. Some reported nothing unusual, and others recalled noises or a sag in the road near the eastern abutment. Several of these accounts were recorded long after the event, and none of them can be checked independently today. They are reports, and they do not establish what the dam was doing that night.

The structural problem with the morning inspection was who made the decision. The engineer judging whether the leak was dangerous had designed and built the dam. He had approved raising it and carried the reputation that went with it, and no second engineer had the authority to disagree and lower the reservoir. The coroner’s jury later addressed that arrangement directly.

Just before midnight: the collapse and the section left standing

The dam failed at about 11:57 p.m. on March 12, 1928. The time is fixed largely by interruptions recorded on the electric transmission lines that ran through the canyon, because those lines failed as the dam and the flood took them out. That timing evidence is unusually good for a disaster of that era. It is one of the reasons the later investigators could reconstruct the flood’s progress downstream with some confidence.

Most of the dam was destroyed, but a central section remained upright in the canyon. Sightseers who came to see it nicknamed it the Tombstone. The two wing sections broke apart, and large blocks of concrete were carried downstream, some of them a considerable distance. The positions of those blocks became evidence in the argument over which side of the dam gave way first. The reservoir drained rapidly into a canyon that had no warning system and few people awake.

The Harnischfeger family, living below the dam, were among the first people killed. The damkeeper’s body is commonly reported as never recovered. The standing center section was demolished with explosives in 1929. The dam was never rebuilt, and the city later developed storage elsewhere, at Bouquet Canyon. In 2019 Congress designated the site a national memorial and monument under the administration of the U.S. Forest Service. The Forest Service is the place to check for current information about visiting it.

The flood path: Powerhouse No. 2, Kemp, Santa Paula and the sea

The first place the flood reached was Powerhouse No. 2, a hydroelectric plant of the city’s Bureau of Power and Light about a mile and a half downstream. Plant workers and their families lived there, and the flood destroyed the plant and killed most of the people at the site. Estimates of the wave height in the upper canyon run well above 100 feet. Those figures are reconstructions from scour lines and debris and were never measured as the wave passed, so I treat them as approximate.

The water left San Francisquito Canyon, entered the Santa Clara River near Castaic Junction and turned west into Ventura County. Near the county line, Southern California Edison had a construction camp at a place called Kemp, where roughly 150 men were sleeping in tents. Published accounts commonly give 84 of them as killed. Below Kemp the flood passed Piru and spread across the farmland and orchards around Fillmore and Bardsdale. It then reached Santa Paula and the lower valley, widening and slowing as it went. It reached the ocean between Ventura and Oxnard around dawn, by most accounts roughly five and a half hours after the failure.

No warning system existed for any of this. Word spread instead through utility employees who noticed their lines failing and through telephone exchange operators who began calling subscribers. Sheriff’s deputies and police went door to door or drove through low ground. In Santa Paula, motorcycle officers, among them Thornton Edwards, are credited with riding through the neighborhoods nearest the river to rouse residents. The towns farthest downstream had hours of notice, and the people at the powerhouse and at Kemp had essentially none.

From a communications point of view, the flood crossed a county line between a city water bureau and agencies that had no prepared relationship with it. The warning that saved lives in Ventura County depended on individuals who happened to know what was happening and could reach a telephone. That is the same gap that downstream notification lists in a modern dam emergency action plan are supposed to close.

The death toll: at least 431 and why it stays a range

The figure most often published today is at least 431 dead. That number traces largely to victim-by-victim research by historian Ann Stansell, who compiled names from death records, cemetery and newspaper sources and other documents. Earlier totals, in 1928 reporting and in later books and articles, ran lower, with many in the high 300s and around 400. Some writers have argued that the real number was higher still, and estimates of 450 and above appear in print. I give the toll as at least 431, with the understanding that it is a floor based on documented victims and that the true number is unknown.

The count is uncertain for several reasons. The flood carried bodies and debris all the way to the ocean, and some victims were never found. Others were buried in silt across miles of farmland, and remains were still turning up years later. A large share of the people living along the lower Santa Clara River were farm workers, many of them Mexican nationals or Mexican Americans. Some were seasonal or living in camps that kept no reliable roster, so a missing person might have had no one locally to report the loss. The construction camp at Kemp raises the same problem, since a transient workforce is hard to count after the fact.

The disaster is often described as one of the deadliest in California history, commonly placed second after the 1906 San Francisco earthquake and fire. Anyone using that comparison should know that the 1906 toll is itself disputed and has been revised upward substantially by researchers. In a briefing or a training slide, “at least 431 people, and probably more” is accurate and defensible. A single tidy number presented as final is wrong.

The inquest and the commissions: what they found and whom they held responsible

Several bodies investigated the failure within weeks. Governor C. C. Young appointed a commission of engineers and geologists, and the City of Los Angeles sought its own technical review. The Los Angeles County coroner’s inquest heard testimony from engineers, geologists, survivors and Mulholland himself. The governor’s commission reported quickly and placed the cause in the foundation, as described above.

Mulholland’s testimony at the inquest is the part of the record most often quoted. As widely reported, he accepted responsibility, saying in substance that if there had been an error of human judgment, he was the human, and that he would not try to place it on anyone else. The wording varies between secondary sources, so I am giving the substance rather than a verbatim quotation. Anyone who needs the exact words should go to the inquest transcript or to a scholarly history that cites it, such as Heavy Ground.

The coroner’s jury found that the failure was caused by the foundation rock. It placed responsibility on the Bureau of Water Works and Supply and its chief engineer for errors of engineering judgment, and it did not find criminal culpability. In the jury’s view, the geology of the site was not adequately understood by the science of the time. Its most important recommendation concerned oversight. The jury said, in substance, that the construction and operation of a great dam should not be left to the judgment of one man, however eminent, without independent expert review. Mulholland retired from the bureau later in 1928 and died in 1935. The city paid claims to victims and families through a settlement process, and I am not giving dollar figures here because I have not confirmed them.

At the time, Mulholland and others raised the possibility that the dam had been sabotaged, against the background of the aqueduct bombings. That sabotage theory is a myth. It found no support in the investigations and should be described as a rejected explanation. It still surfaces in casual retellings, and the correction belongs in the same sentence that mentions it.

The 1929 law: how municipal dams came under state supervision

In 1929 the California Legislature passed dam safety legislation that ended the practical exemption for city-owned dams. It placed non-federal dams above certain size thresholds under the supervision of the state engineer. As the law is generally described, the state gained authority to approve plans before construction and to inspect dams during construction and operation. It could also require repairs or limit how much water an owner stored. That program continues today as the Division of Safety of Dams within the California Department of Water Resources, and its current jurisdictional thresholds and procedures should be checked there.

The same year California enacted registration of civil engineers, and many histories connect that law to the St. Francis failure. The connection is commonly drawn, though the registration movement had roots before 1928. The legislative history is the place to confirm how directly one led to the other. Taken together, the two measures reflected the jury’s conclusion that public safety could not rest on any single engineer’s reputation, and that plans for dams capable of killing people downstream needed review by qualified people outside the owner’s organization.

The rest of the country moved more slowly. State dam safety programs varied widely for decades. A national inventory and inspection effort came only with the National Dam Inspection Act of 1972, which followed other deadly failures that year. Today, who regulates a given dam depends on who owns it and on its size and hazard classification. Federal agencies, states and tribes each have roles, and some small dams fall under no one’s regular inspection. The Association of State Dam Safety Officials maintains information on state programs, and the state program itself is the authority on which dams it regulates.

For a working agency, the 1929 law implies a question worth answering for every dam upstream: who has the legal authority to review its condition and order a change, and how quickly can that person be reached? St. Francis had no such person in 1928, and some dams still effectively have none.

What to do at your agency

• The emergency management coordinator asks the state dam safety program for its list of regulated dams whose inundation areas reach the jurisdiction, and produces a one-page table showing each dam’s owner, its regulator and its most recent inspection date.
• The hazard mitigation planner checks the dam section of the current hazard mitigation plan and adds the regulating agency for each listed dam, marking any dam that no state or federal program inspects.
• The communications center supervisor confirms that CAD or the dispatch protocol contains a call type for a dam owner or operator reporting seepage, cracking or an abnormal condition, with the notification list attached, and builds that call type this month if it is missing.
• The communications center supervisor makes a test call to the 24-hour number of the next county downstream on each major river corridor and records the date and the answering position in the notification list.
• The emergency management coordinator sends an overnight test through the jurisdiction’s mass notification system to an internal staff test group and keeps the delivery report showing how many messages arrived within ten minutes.
• The training officer adds an inject to the next scheduled tabletop in which a dam owner reports a leak at mid-morning that its own engineer calls normal, and finished means the after-action notes record who the county would call for an independent technical opinion.

Takeaways

• The St. Francis Dam, built by the Los Angeles Bureau of Water Works and Supply under William Mulholland and completed in 1926, failed at about 11:57 p.m. on March 12, 1928.
• The dam stood on Pelona Schist on the east and a weak red conglomerate on the west, separated by a fault, and investigators placed the cause in that foundation.
• The 1928 commissions emphasized the western conglomerate softening when wet, while J. David Rogers later argued that an ancient landslide on the eastern abutment and uplift pressure began the failure, and the sequence remains debated.
• Mulholland inspected a new leak on the morning of the failure and judged it safe, and no independent engineer had authority to review that decision or lower the reservoir.
• The flood traveled about 54 miles through San Francisquito Canyon and the Santa Clara River valley to the ocean, and warning downstream depended on telephone operators, utility workers and officers rather than any planned system.
• The death toll is commonly given as at least 431 based on Ann Stansell’s research, earlier counts were lower, and the true number is unknown because many victims were never found and some farm workers likely went uncounted.
• The coroner’s jury found no criminal culpability, held the bureau and its chief engineer responsible for errors of judgment and recommended independent review of dams, and the sabotage theory raised at the time is a myth.
• California’s 1929 dam safety law brought municipal dams under state supervision, and the question of who regulates each dam upstream of a jurisdiction is still worth answering in writing.