The 1906 San Francisco earthquake: a city measured by its traces

Before the flames

The rupture crossed far beyond the city

Around 5:12 a.m., residents felt a foreshock. The main shock followed 20–25 seconds later; shaking lasted about 45–60 seconds across the Bay Area. The modern consensus is Mw 7.9, calculated retrospectively; the Lawson Commission did not measure it in that form.

USGS reconstructs the San Andreas rupture across about 296 miles (477 km), from the San Juan Bautista area north toward Cape Mendocino. The familiar 8.3 and 8¼ are traditional estimates tied to later magnitude methods, not a modern Mw measurement made by Lawson.

Shaking damaged buildings; the effects varied with structure and ground. Fire then caused most of the urban destruction.

The Ferry Building clock cannot pin the moment exactly. Popular accounts say 5:12, a contemporary newspaper reported 5:16, and accounts note that the clock may have run fast. Water and gas mains left a different trace.

A city grid with a vermilion fire trace and an oblique rupture; schematic and not to scale. — Schematic drawing · lines are not to scale

Archival panorama of the 1906 San Francisco fire.

Bird’s-eye context around the Ferry Building after the 1906 earthquake.

When the water failed

A broken network changed the scale of the disaster

The earthquake broke water and gas mains and damaged the water system. With too little water for fire crews, separate ignitions joined into one conflagration. It burned through 21 April: three days of urban fire.

Fire caused most of the urban destruction; shaking also damaged buildings. The City Hall ruins make that damage visible. Contemporary studies record different outcomes by building and ground.

On the map, the fire has an outline. DataSF digitizes the boundary R.J. Waters & Co. drew in 1906: a mapped footprint, not a sequence of daily fire fronts.

San Francisco City Hall ruins after the earthquake and fire.

Read the boundary as the fire’s footprint; sources use different limits and methods.

What 80% measures

Property value is not the area that burned

A contemporary insurers’ report estimated that the fire destroyed at least 80% of the city’s pre-fire property value. It did not say that 80% of San Francisco’s land burned.

The report gave about 2,381 acres, or 3.72 square miles; other sources draw boundaries around 4.7. Across the sources, the burned area ranges from 3.7 to 4.7 square miles. The range stays visible because outlines and counting methods differ.

Dynamiting left a mixed record. Some attempts started new fires; others failed to make an effective break.

The claim that dynamiting “burned more than it saved” cannot be measured from the available evidence.

The Van Ness defense line held as the wind changed and water supplies returned. Military and naval crews also worked the hoses.

80%estimated pre-fire property value

3.7–4.7square miles · burned-area range

80% = estimated pre-fire property value. The burned footprint remains a range.

A history of the count

478 is a record floor, not a complete census

The city’s official 1907 count stops at 478 recovered or recorded deaths. The Army kept a separate tally: 498 deaths in San Francisco, plus deaths elsewhere. NOAA later called 700–800 reasonable.

Gladys Hansen and Emmet Condon argued from archival research that direct and indirect deaths exceeded 3,000. USGS commonly uses that estimate; Hansen later cited more than 3,400. These are attributed estimates, not a final list of identified people.

A 2005 city resolution set aside the 478 figure and said few, if any, Chinatown deaths had been included. It also noted suspected victims of shootings against looters.

Hansen argued that commercial interests minimized the toll to protect the city’s business recovery. That remains her explanation, not a conclusion proved by one document.

Four records, four histories of counting
Figure	Record	What it counts	Limit
478	City’s official count, 1907	Recovered or recorded deaths in the municipal record	Not a complete census; the 2005 resolution said Chinatown deaths were almost certainly missing.
498	Army report for San Francisco	Separate military record; deaths elsewhere also counted	A different administrative series, not a direct correction of 478.
700–800	NOAA assessment cited later	Range called reasonable in a later synthesis	An estimate, not a named list or municipal count.
>3,000	Hansen and Condon; commonly cited by USGS	Archival estimate of direct and indirect deaths	Attributed estimate; not a final count of identified people.

The rows use different places, methods and periods; they are not one unified list.

Geographies, methods and periods differ: the rows neither add up nor replace one another.

Shelter and return

Hundreds of thousands left; recovery was not shared equally

Sources estimate that between about 225,000 and more than 250,000 people were displaced. Before the earthquake, the city was described as having roughly 400,000–410,000 residents. The Army managed many camps, including at the Presidio, Golden Gate Park and Fort Mason.

The cottage program is often put at roughly 5,300 homes. NPS says they housed about 16,000 people at peak. The figures do not form a complete register of residents or cottages. Two Presidio cottages survive as historic objects.

Before the disaster, Chinatown had about 15,000 residents; NPS says around 400 remained immediately afterward.

Chinese residents faced segregated camps and exclusion from some aid. Officials discussed moving the district to Hunters Point.

Community leaders resisted and returned to rebuild. The later “Oriental” architectural vision took shape under racist constraints and economic pressure.

225,000–250,000+people displaced, estimates

~5,300relief cottages, common estimate

~16,000people housed at the program’s peak, according to NPS

Refugee camp in Golden Gate Park after the 1906 disaster.

Shelters built during relief after the earthquake.

Preserved Presidio cottages connected with the relief program. — Preserved Presidio cottages connected with the relief program; 2024 photograph.

View associated with Chinatown after the 1906 San Francisco earthquake.

Displacement, camp and cottage counts are estimates from different sources and periods.

A mistaken attribution

The handbill was Mayor Schmitz’s, not a Funston proclamation

Mayor Eugene Schmitz issued the handbill ordering looters shot. Retellings sometimes assign it to General Frederick Funston; the document was not Funston’s proclamation.

The National Park Service concludes that neither Schmitz nor Funston officially declared martial law. Orders overlapped, shots were fired, and legal authority was disputed. Funston’s role and responsibility for civilian deaths remain contested in the historical record.

The panel is an editorial diagram of the attribution, not a facsimile of the historical handbill.

PRIMARUL EUGENE SCHMITZ

ORDER CONCERNING LOOTING AND SHOOTINGS

NOT A FUNSTON PROCLAMATION

Attribution: Mayor Eugene Schmitz. NPS: neither Schmitz nor Funston formally declared martial law.

Rebuilding at two speeds

Downtown returned quickly; the city kept rebuilding

Shops, commercial activity and many downtown buildings returned quickly. Broader recovery continued through the following decade; the entire city was not rebuilt in three years.

Federal censuses counted 342,782 residents in 1900, 416,912 in 1910 and 506,676 in 1920. These three decennial snapshots do not isolate migration or prove that the disaster alone caused the increase.

In 1915 San Francisco hosted the Panama-Pacific International Exposition. The fair celebrated the Panama Canal and the city’s return; it does not prove recovery was complete or shared equally.

1906
The fire ends; commercial activity begins to return.
1907–1910
Downtown rebuilds rapidly; census totals also reflect other changes.
1915
The Panama-Pacific Exposition marks a public return, not equal recovery.
Following decade
The wider city continues to rebuild.

Census population counts are snapshots, not a year-by-year rebuilding measure.

The city not built

The Burnham Plan remained a vision, not the city’s full blueprint

In 1905, Daniel Burnham and Edward Bennett published a plan for monumental avenues, civic centers and a broad redesign. San Francisco did not implement it as a comprehensive plan.

The obstacles had names: cost, land acquisition, political opposition and property owners. After the fire, many owners pressed to rebuild on existing lots and street alignments. Some Burnham ideas and routes still influenced later development.

The dispute was over who would pay for change and what could be rebuilt quickly. The plan remained incomplete, though some of its ideas carried into the city that followed.

Burnham Plan · 1905Plan not implemented as a whole

Rebuilt gridExisting grid rebuilt

A schematic comparison of two urban intentions; it does not georeference Burnham’s drawing.

What stayed in the record

The disaster changed science and rules; it did not invent earthquake insurance

The California Earthquake Investigation Commission, led by Andrew C. Lawson, published its major report in 1908. Two years later, H.F. Reid offered the elastic-rebound interpretation. The disaster became a landmark in modern earthquake science in the United States; it did not invent seismology.

Lloyd’s was writing earthquake risks before 1906. The disaster reshaped catastrophe claims and insurance disputes and helped drive policy and regulatory changes. California adopted a standard fire policy in 1909.

The Palace of Fine Arts shown at the 1915 exposition was a temporary structure. Today’s building is a later reconstruction and restoration, with a related form but a different construction.

1908
The Lawson Commission publishes its major report.
1909
California adopts a standard fire policy.
1910
H.F. Reid publishes the elastic-rebound interpretation.
1915
PPIE displays a temporary Palace of Fine Arts.

Science and regulation changed in different years; neither came from a single cause.

Sources and limits

How to read the traces

This register keeps scientific studies, contemporary reports, public resolutions, historical estimates and mapped layers distinct. The seven selected photographs have verified identities, dates and rights; unselected alternatives remain research leads.

Core sources V01–V16

Additional sources

For the fire perimeter, see DataSF and the R.J. Waters & Co. lineage; for street orientation, OpenStreetMap with attribution. Modern OSM does not show the 1906 shoreline.