A Canvas City of Twenty Thousand Souls

Seen from above, the refugee settlement in Golden Gate Park was a rigid geometry of straight lines and numbered blocks, a map of military precision governing twenty thousand souls. This order, represented by ledgers and forms in the Presidio headquarters, was an aspiration born of rules that had not existed six weeks earlier. Within this grid, each tent was assigned an address that corresponded to no prior San Francisco street, a system where the details on a form could determine the length of a wait.

Block 47, Tent 12, stood at the western edge of the grid, farthest from the park’s entrance. The canvas walls had darkened with ten days of rain, and the dirt floor had packed into a hard surface that held the cold. A woman inside pinned a ration card to her coat, one of thousands distributed by the American Red Cross in partnership with the Army, while outside a sanitation squad moved between the rows with lime and shovels. The squad worked methodically, following regulations posted at the camp entrance that mandated inspection of all refuse. They wore armbands issued by the camp police, whose authority derived from General Frederick Funston’s command of the Presidio and his assertion of military jurisdiction over the city’s disaster relief. Funston had already decided that the situation required the use of federal troops, and his soldiers from the Presidio now administered this vast canvas settlement.

The Army had constructed a virtual town within the park’s thousand acres. Three thousand tents stood in formations that mimicked municipal blocks, complete with designated streets and addresses that allowed the post office to deliver mail to displaced citizens. The layout reflected the Progressive-era conviction that catastrophe required not merely charity but systematic management. Engineers from the Presidio supervised the installation of latrines and bathhouses, laundries, and communal kitchens. Water pipes ran in straight trenches beneath the temporary avenues. The regimented order suggested that chaos could be contained through proper administration, that even homelessness could be rendered legible to authority.

The scale of this operation was immense. Of the 26 official homeless encampments in the Golden Gate Park region, 21 were under the control of the United States Army. The United States Army housed 20, 000 people in military-style encampments, and 16, 000 of those refugees were living at the Presidio. Within the Presidio were four major encampments, including a camp exclusively for Chinese immigrants.

Yet the geometry on the map concealed gradients of privilege that sorted the displaced according to criteria never printed on the official forms. In the central camps, near the park’s entrance and the Presidio grounds, the Army built 5, 610 redwood and fir relief houses designed by John McLaren, the park’s superintendent. These structures, grouped in eleven separate camps, offered walls that did not flap in the wind and doors that locked. The houses rented for two dollars per month until rebuilding was completed, a sum that favored those who had salvaged currency from fireproof safes or who maintained bank accounts in Oakland. They stood packed close to each other, forming dense neighborhoods that replicated the urban fabric of the destroyed city, complete with address numbers and pathways wide enough for delivery wagons.

In the far corners of the park, beyond the reach of the water mains and the regular patrols, canvas shelters sagged on uneven ground. Here lived those who could not produce the required documentation, or whose names appeared on lists compiled by Chinese community associations, or who spoke languages that the registration clerks did not recognize. The geometric grid extended even to these margins. Officials assigned tent numbers and block designations with bureaucratic impartiality, but the services attenuated with distance from the command center. Latrines in these sections overflowed more frequently. The ration lines moved slower. Medical inspections, mandatory for all residents under public health ordinances, occurred less often and with less follow-up.

The collision between humanitarian logistics and pre-existing social hierarchies produced a settlement that mirrored the city’s fractures. White middle-class families occupied the redwood cottages near the park’s entrance, while Chinese refugees found themselves directed to segregated zones where the ground stayed muddy and the wind cut sharper. The Army maintained separate registration books for different districts, a practice that officials defended as efficient organization but that resulted in divergent rations and slower relief for those consigned to the periphery. Camp police patrolled all sections with equal frequency on the rosters, but their interventions concentrated in areas where military officers believed disorder more likely to originate.

Daily life in the canvas city operated through a bureaucracy of survival. Residents queued at dawn for identification tags that permitted them to draw rations: bread, coffee, condensed milk, meat when available. Clerks recorded names in leather-bound volumes that would later be deposited in the Relief Corporation archives, creating a paper trail of displacement that emphasized property loss over structural collapse. The forms asked for the value of goods destroyed by fire; they did not inquire about walls shaken down by seismic force. This documentation strategy served the emerging municipal narrative that would eventually dominate official memory: a story of conflagration rather than earthquake, of accidental disaster rather than systemic failure.

The sanitation squads represented another face of institutional control. Progressive reformers had long advocated for public health measures as instruments of social order, and the emergency provided authority to implement comprehensive surveillance of daily habits. Inspectors entered tents without knocking to check for cleanliness, for proper storage of food, for signs of communicable disease. They enforced regulations against alcohol, against gambling, against the assembly of large groups after dark. The camps became laboratories for theories of urban management, testing whether a population rendered suddenly dependent could be molded through regimentation into a more hygienic, more orderly citizenry.

Chinese residents faced additional layers of scrutiny rooted in decades of anti-Asian policy. Military authorities required them to register with both the camp administration and their own community organizations, creating a dual surveillance system that tracked their movements more closely than those of white refugees. When typhoid appeared in one section of the park, authorities quarantined the Chinese camps exclusively, though the bacterium showed no respect for ethnic boundaries. The quarantine lasted fourteen days, during which rations were delivered to the perimeter but no residents permitted to leave for work or to visit family in other sections. The measure protected public health by strict isolation, but it also enforced a segregation that predated the disaster and would outlast the emergency.

John McLaren’s relief houses embodied the contradiction between temporary relief and permanent exclusion. Built to accommodate twenty thousand displaced people, the structures were designed for durability, with redwood siding, fir framing, and shingled roofs, yet intended for removal once the city rebuilt. The eleven camps clustered them in patterns that separated merchants from laborers, families with children from single men, and always maintained spatial buffers between white residents and those classified as “Oriental” on the housing rolls. The two-dollar monthly rent created a financial barrier that ensured the houses served those with means to pay, while the tent cities absorbed the destitute.

At the Presidio, military discipline shaped refugee life more directly. Soldiers from Funston’s command supervised the distribution of goods with procedures developed for supply depots in the Philippines. They established reveille and taps, requiring residents to rise and retire according to bugle calls. The Army constructed barracks-style housing for officers’ families displaced from Nob Hill, maintaining class distinctions even in exile. These facilities enjoyed priority in the allocation of building materials and medical supplies, while civilian camps waited for surplus.

The registration process itself functioned as a mechanism of triage, determining who qualified for immediate assistance and who entered the slower lines. Clerks demanded proof of prior residence, proof of property ownership, proof of fire loss specifically dated to April 18 or 19. Those who had fled before documenting their losses, or whose buildings had collapsed rather than burned, found themselves shunted to secondary queues where forms required additional verification. The system privileged those who could navigate bureaucratic procedure, penalizing the illiterate, the foreign-born, and those whose entire documentation had vanished with their homes.

By late May, the camps had achieved a brutal equilibrium. The Army maintained order through a combination of logistical competence and implicit threat; the same troops who had patrolled fire lines now walked among canvas shelters with rifles slung over shoulders. The Red Cross administered charity through channels that reinforced social stratification, delivering better provisions to districts deemed more deserving. Public health officials collected data on the refugee population that would serve future urban planning while ignoring the immediate disparities in their care.

The physical landscape of the canvas city thus encoded a social blueprint. Geometric streets and numbered tents suggested democratic uniformity, but the distribution of McLaren’s wooden houses versus canvas shelters, of central locations versus peripheral bogs, mapped the city’s pre-existing hierarchies onto emergency architecture. The relief effort saved lives and sheltered thousands, yet it also demonstrated how quickly humanitarian crisis could be converted into social control, how efficiently disaster management could sort populations according to race and class while claiming to operate on neutral principles of need.

The registration clerks worked from standardized forms that had traveled east from Red Cross headquarters in Washington, their columns and checkboxes originally designed for Midwestern flood victims or Appalachian mine disasters. In the San Francisco camps, these documents encountered a population whose losses defied easy categorization. A clerk in the Presidio headquarters might spend twenty minutes with a merchant from Sacramento Street, struggling to translate a Chinese business association’s letter of property verification into the required fields of the official form. The delay rippled outward: that family received a temporary tent assignment rather than immediate placement in a McLaren cottage, their names entered in a supplementary ledger rather than the primary register. Such administrative friction accumulated across thousands of cases, producing systematic delay that appeared on no single document as discrimination.

The water infrastructure revealed similar patterns of differential investment. Engineers from the Presidio prioritized mains installation in sections where wooden houses were planned, calculating that permanent structures justified permanent utilities. Canvas districts received water from mobile tank wagons, their delivery schedules subject to military transport priorities. Residents in Block 47 learned to collect rainfall in buckets and cooking pots, a practice that sanitation inspectors noted as “resourcefulness” in their reports while simultaneously marking the same households for closer health surveillance. The inspectors’ clipboards thus recorded both admiration and suspicion, the double entry capturing the ambivalence of Progressive reform toward the poor they sought to uplift.

Camp police developed informal protocols that never appeared in Funston’s official orders. Patrolmen assigned to the Chinese sections carried bilingual summons forms printed by missionary societies, documents that assumed the recipient would appear before a camp magistrate rather than demand formal arrest procedures. The assumption saved paperwork but eroded legal standing; those who accepted the informal summons found their cases resolved through camp administration rather than civilian courts, creating a parallel justice system that persisted for minor infractions long after the emergency’s formal end. White residents arrested for identical violations, whether public drunkenness, gambling, or disorderly conduct, more frequently demanded and received transfer to municipal authorities, preserving their access to ordinary legal protections.

The ration cards themselves became objects of social differentiation. The American Red Cross issued distinct card stock for different camp sections: stiff linen paper for house residents, thinner stock for tent dwellers, a third variety for those in temporary quarantine. The visual distinction allowed distributors to sort queues at a glance, and merchants in the surrounding city soon learned to read the hierarchy. A grocer on Fulton Street might accept a linen ration card for partial payment of coffee or tobacco, extending informal credit against the card’s implicit guarantee of stable residence. The thinner cards commanded no such confidence. This secondary economy of recognition operated entirely outside official policy, yet it translated administrative distinction into market reality, deepening the material gap between camp zones.

Medical inspection produced its own archives of inequality. Public health officers maintained separate morbidity ledgers for each camp section, recording incidence of respiratory illness, intestinal complaints, the skin infections that spread through crowded canvas. The aggregated data showed higher rates in peripheral districts, findings that officials interpreted as confirmation of hygienic deficiency among certain populations rather than as evidence of differential investment in sanitation infrastructure. When the Red Cross published its official report in 1907, the statistics appeared as objective measures of community health, their production through unequal surveillance invisible in the final tables. The report’s very confidence in numerical comparison, typhoid rates per thousand and mortality ratios by district, obscured the prior decisions that had determined where inspectors looked most carefully and which cases received follow-up treatment.

The wooden houses carried their own symbolic weight beyond their practical shelter. McLaren’s design, with its front porch and gabled roof, deliberately echoed the domestic architecture of San Francisco’s residential neighborhoods, offering refugees a visual promise of eventual restoration. This architectural reassurance was itself stratified: the houses clustered near park entrances most closely approximated the destroyed homes of the middle class, while those in secondary locations, smaller, less ornamented, sometimes lacking the porch, served populations whose prior housing had been more modest. The relief architecture thus prepared residents for differentiated returns, training the eye to recognize gradations of respectability that would map onto the rebuilt city’s rental markets.

Children in the camps experienced the imposed order as normalized environment. Those born in the months following the disaster would remember no other address than their block and tent numbers, no other authority than the uniformed figures who distributed food and enforced curfew. Schooling, when it resumed in June, operated from camp buildings under military supervision, with curricula that emphasized hygiene, punctuality, and the proper completion of forms. The educational interruption thus became an opportunity for institutional capture, the emergency classroom extending camp discipline into the developing habits of a generation. Parents who resisted, who kept children home for traditional instruction in language or religion, found their ration cards subject to “review,” the administrative euphemism for temporary suspension.

The Committee of Fifty’s private deliberations cast a shadow across camp life that residents could perceive without fully comprehending. Rumors circulated of surveys and measurements, of men in business suits walking through the park with clipboards and transit instruments. These were the advance agents of the downtown plans, gathering data on population distribution that would inform decisions about street widening and neighborhood clearance. The canvas city thus served a dual function: immediate shelter and future planning resource. Its residents lived within a demographic experiment, their clustering patterns studied to determine where permanent populations might be concentrated or dispersed. The knowledge that one’s tent location might influence whether one’s former neighborhood would be rebuilt or converted to commercial use introduced a paralyzing uncertainty into daily life, a sense that even the emergency settlement was provisional in ways beyond its obvious temporariness.

By early June, the relief infrastructure had developed sufficient complexity to generate its own internal politics. Camp residents elected “block captains” to represent grievances to administration, a concession to participatory demand that also extended surveillance deeper into the settlement. The captains received small privileges, priority in ration lines and access to the camp post office outside regular hours, that created stakes in the system’s perpetuation. Some used their position to build patronage networks, trading information about registration requirements or housing availability for promises of future support. Others attempted genuine advocacy, compiling lists of sanitary complaints or documentation delays that they presented to Red Cross supervisors with the formal deference expected of petitioners. The success rate of such interventions varied inversely with distance from the park’s center, reproducing in miniature the civic hierarchies of the pre-disaster city.

The Army’s eventual construction of additional wooden houses beyond the initial plan reflected not expanded generosity but revised calculation. Funston’s command had initially projected rapid demobilization of the refugee population, expecting private reconstruction to absorb the displaced within weeks. When insurance delays, banking collapse, and political disputes over building codes prolonged dependency, the temporary structures proved insufficient for winter occupation. The expanded building program thus acknowledged the failure of civilian recovery rather than its success, the additional redwood cottages marking time while the downtown committees debated. Each new house extended the emergency’s duration, embedding the camp population more deeply into a settlement that was neither temporary nor permanent, neither fully military nor fully civilian.

The final paradox of the canvas city lay in its most successful residents: those who navigated the registration system with sufficient skill to obtain wooden housing, maintained documentation of property loss adequate for insurance claims, and preserved social connections to downtown employment. These families were precisely those most able to leave, yet their camp housing represented sufficient improvement over available alternatives to encourage delay. The relief system thus retained those best positioned for independent recovery, while the most desperate, the undocumented, the uninsured, the employed in casual labor, remained in canvas shelters that deteriorated with summer weather.

Dennis T. Sullivan had been the Chief of the San Francisco Fire Department in 1906. He was mortally wounded during the 1906 San Francisco earthquake, when a neighboring building collapsed onto the fire station that housed the Chief’s official apartment. His death left the department leaderless during the crucial first hours of the conflagration, and his absence echoed through the subsequent establishment of military authority over civilian functions. The relief camps operated under protocols developed by the Army rather than the municipal government he had served, reflecting a transfer of power from local institutions to federal military command that would shape the city’s reconstruction.

As June advanced, the camps entered their seventh week of operation. The last official refugee camp would not close until June 30, 1908, but already in the summer of 1906 the temporary city showed signs of permanence. Families planted gardens beside their tents. Children born in the park received birth certificates listing Block numbers as addresses. The Army continued to build redwood cottages, eventually constructing thousands more than initially planned, while behind closed doors downtown a different group of men drew other plans entirely.

The Committee of Fifty, later reduced to forty members, met in private offices to draft blueprints for a new San Francisco, determining which neighborhoods would be widened, which streets straightened, which populations encouraged to return and which redirected elsewhere. The canvas city and its wooden suburbs represented the immediate human cost of the disaster, a population held in suspended animation while the physical and political landscape of their future was being reimagined without their consultation.

The solidified social divisions of the refugee camps handed off a city whose human landscape was fractured, just as its physical landscape was being re-planned behind closed doors.