When the Shaking Stopped
The fires were already burning before most people reached the street.
At 5:12 in the morning on April 18, 1906, the San Andreas Fault ruptured along 296 miles of its length, from north of San Juan Bautista to Cape Mendocino. The shaking lasted approximately 45 to 60 seconds. In those seconds, chimneys collapsed, gas mains cracked, and water pipes fractured beneath the streets. By the time the motion stopped, the tools needed to fight a fire had been destroyed alongside everything else.
San Francisco in 1906 was a city of roughly 400,000 people. It was prosperous, densely built, and proud of its recovery from earlier disasters. The earthquake that morning was not especially long. In a city with an intact water supply, it might have been manageable. But the water mains broke in the first seconds. More than 50 fires ignited across the city that day. Firefighters arrived at hydrant after hydrant and found nothing.
The Fault Beneath the City
The San Andreas Fault had been mapped by geologist Andrew Lawson in 1895 and named by him for a small lake in the Coast Ranges. Its significance as a seismic hazard was understood in outline. What no one had measured was what it could actually do.
The 1906 rupture answered that question definitively. Ground displacement along the fault reached a maximum of about 20 feet near Tomales Bay. The shaking was felt from southern Oregon to south of Los Angeles, and inland as far as central Nevada. In the San Francisco Bay area, the quake's intensity was highest where buildings sat on soft sediment-filled valleys rather than bedrock. This would later become foundational knowledge in earthquake engineering. In 1906, it simply meant that entire neighborhoods built on landfill or bay mud shook violently longer and harder than nearby hills.
The magnitude is now estimated at 7.9 on modern scales, though historical estimates have ranged from 7.7 to 8.3. The number matters less than what the rupture triggered. The earthquake lasted under a minute. The fires it ignited burned for three days and destroyed roughly 2,800 acres across 490 city blocks.
The Fire Chief Was Already Dying
San Francisco's fire chief, Dennis Sullivan, had spent years warning city officials about exactly this scenario. The water system was inadequate for fighting a major fire. The cisterns were too few and too small. A large earthquake, he argued repeatedly, would break the water mains and leave the city defenseless. He had proposed an independent auxiliary water supply fed by saltwater from the bay. The city had not funded it.
At 5:12 AM on April 18, Sullivan was asleep at his quarters at Fire Station 3 on Bush Street. The ornamental dome of the adjacent California Hotel broke free and smashed through the roof, crashing through three floors and carrying Sullivan down with it. He was pulled from the wreckage gravely injured. He died four days later, on April 22, without recovering consciousness.
The man who had prepared the most detailed plan for fighting a catastrophic fire was gone before the fires were an hour old. Command fell to officials who had never prepared for anything like what they now faced.
The Ham and Eggs Fire
Not all of the fires started the same way. Some came from ruptured gas mains. Some from overturned stoves and broken chimneys. One of the largest had a particularly ordinary origin.
A woman at 395 Hayes Street lit her kitchen stove to cook breakfast. She did not know that the earthquake had cracked her chimney. The heat built up where it was not supposed to go. The fire spread into the walls. From Hayes Valley, it grew into what became known as the Ham and Eggs Fire, one of the most destructive blazes of the three-day catastrophe. It eventually consumed much of the Western Addition and merged with other fires moving south from the city center.
The name was not meant as a joke. It became shorthand for a hard fact: the fires that destroyed San Francisco were not caused entirely by dramatic infrastructure failure. Some of them started because people were hungry and cold and trying to make breakfast, not knowing their homes had already become kindling.
"I wake up about 5 o'clock, feeling my bed rocking as though I am in a ship on the ocean. I look out of my window and see the buildings toppling over, big pieces of masonry falling. From the street below I hear the cries and screams of men and women and children."
— Enrico Caruso, eyewitness account, 1906
Dynamite and the Spreading Fire
With the hydrants dry, the acting authorities made a decision that would make things worse.
General Frederick Funston mobilized U.S. Army troops and ordered buildings dynamited along Van Ness Avenue to create a firebreak. The idea was sound: if a wide enough gap could be cleared, the fire would have nothing left to consume. But the work required skill that most of the men carrying it out did not have. Improperly handled, a dynamite blast does not simply knock down a building. It can scatter burning debris, open gaps in walls that feed fresh air to adjacent fires, and ignite structures that were not yet burning.
In some places, the firebreaks held. In many others, inexperienced crews set new fires in neighborhoods that might otherwise have been saved. The destruction of the city's Western Addition and parts of the Mission District owed something to the fires that started with the earthquake and something to the fires started by soldiers trying to stop them.
Three Days
The fires burned from the morning of April 18 into April 20. By the time they were out, 490 city blocks had been destroyed. The burned area covered approximately 2,800 acres, six times the area consumed by the Great Fire of London in 1666. Around 25,000 buildings were gone. Three-fifths of the city's homes and lodgings no longer existed.
More than 200,000 people were left without shelter. Refugee camps spread across Golden Gate Park, the Panhandle, and the Presidio. Tents and improvised shelters held families for months. The military administered food distribution. Conditions varied from difficult to miserable. The social hierarchies of the city did not disappear inside the camps. Wealthier families found better resources. Chinatown's residents, whose neighborhood had been entirely destroyed, were initially directed toward segregated camp sections or displaced toward the edges of the city altogether.
Enrico Caruso Had Sung Carmen the Night Before
On the evening of April 17, the Metropolitan Opera had performed Carmen at San Francisco's Grand Opera House. Enrico Caruso, the most famous tenor in the world, had sung Don José. He was sleeping on the fifth floor of the Palace Hotel when the shaking began.
Caruso looked out his window and saw masonry falling into the street below. He heard screaming. He dressed, gathered what he could carry, and made his way outside. A photographer named Arnold Genthe, whose own studio had been destroyed, encountered Caruso near the entrance to the St. Francis Hotel. The tenor stood on the pavement in a fur coat thrown over his pajamas, smoking a cigarette. "Hell of a place," he said. "Hell of a place."
Later, Caruso was particular about one detail of his account. Some newspapers had reported that he collapsed outside in hysterics, sat weeping on his luggage in the street. He denied it firmly. He said he was frightened, as many others were, but he had not lost his head. Within hours he was on a train south, away from the city. He never returned to San Francisco. He never performed there again.
The Death Count That Was Never Meant to Be Accurate
The official death toll issued in the weeks after the earthquake was 478. Later official counts settled at around 700. For decades, these figures stood as the accepted record of what the disaster had cost in human lives.
They were not accidental undercounts. They were a policy.
City historian Gladys Hansen began researching the 1906 death toll in 1963 and spent the following decades documenting individual casualties from contemporary records: coroner's files, hospital admissions, military reports, and personal accounts. She co-authored Denial of Disaster in 1989, concluding that more than 3,000 deaths could be attributed to the earthquake and its fires. Modern estimates range as high as 6,000.
The discrepancy was not difficult to explain. Chinatown, which was entirely destroyed, had hundreds of deaths that were not formally counted. The deaths of recent immigrants living in informal housing on soft ground in the South of Market district were similarly omitted. City and business leaders actively promoted lower figures to protect property values and reassure outside investors that San Francisco was recovering and safe. Fire damage was insured; earthquake damage often was not, so framing the disaster as primarily a fire event served the financial interests of property owners and the city's commercial recovery.
In 2005, the San Francisco Board of Supervisors formally adopted the figure of 3,000 or more as the official death toll, nearly a century after the fact.
Fire Not Earthquake: The Insurance Logic
The framing of the 1906 disaster as primarily a fire event was not simply a matter of public relations. It had a direct financial logic.
Standard property insurance policies in 1906 covered fire damage. Most did not cover earthquake damage. The distinction mattered enormously to property owners trying to rebuild and to insurers trying to minimize payouts. A building that fell because the ground shook was an uninsured loss. A building that burned because of fires spreading through the city was a covered one.
Some property owners went further. Reports emerged of people deliberately setting fire to their already-ruined structures after the earthquake, to convert an uninsured loss into an insured one. The result was a sustained official effort by city leaders, business associations, and some insurers to describe the disaster as a fire that happened to coincide with an earthquake, rather than a fire that the earthquake caused. That framing shaped public memory, delayed serious seismic safety reform in California, and contributed to the suppression of the death toll that lasted nearly a century.
What the Earthquake Taught Science
The 1906 earthquake produced the most important early study of a major seismic event. The California State Earthquake Investigation Commission, chaired by geologist Andrew Lawson, published its findings in 1908. The Lawson Report remains a foundational document in seismology. It established the San Andreas Fault's geometry, documented the pattern of ground displacement along the rupture, and demonstrated the relationship between local geology and shaking intensity.
Grove Karl Gilbert, who worked on the investigation, had already articulated a principle that the 1906 event confirmed: areas built on unconsolidated sediment shake harder and longer than nearby bedrock. The lesson would be relearned, at great cost, during the 1989 Loma Prieta earthquake, when the same zones of soft bay mud that failed in 1906 failed again, and the elevated Cypress Street Viaduct collapsed onto the Interstate below.
Harry Fielding Reid, working from Lawson Commission survey data, published the elastic-rebound theory of earthquakes in 1910. The idea was that strain accumulates in rock on either side of a fault over decades or centuries, and releases suddenly when the accumulated stress exceeds the friction holding the rocks together. The theory explained the surface displacements measured along the San Andreas and provided the first coherent physical model of how earthquakes work. It remains the principal model of the earthquake cycle today.
San Francisco was rebuilt quickly. By 1915, the city hosted the Panama-Pacific International Exposition to demonstrate its recovery. The seismic hazard that had been documented in careful scientific detail was not reflected in the building codes that governed the reconstruction. The Lawson Report had been read by scientists. It had not been read by the politicians who wrote the rules.