On the night of November 12, 1970, a tropical cyclone drove a storm surge across the low islands and coastal flats of what was then East Pakistan. Weather satellites had photographed the storm and the Pakistan Meteorological Department had issued warnings, but on islands with few radios, no shelters and no high ground, the warnings changed very little. Credible estimates of the dead run from about 300,000 to 500,000. This guide covers what the warning chain had and lacked, why the toll can only be given as a range, how the aftermath fed into the founding of Bangladesh, and the volunteer and shelter programs that followed.

The storm and the delta: what reached the coast on November 12, 1970

The cyclone developed over the Bay of Bengal in early November 1970 and moved north toward the head of the bay. At that point the Ganges, Brahmaputra and Meghna rivers empty into the sea through a maze of channels, islands and tidal flats. The storm came ashore on the night of November 12 and into the early hours of November 13, along the coast of what was then East Pakistan. The worst of the surge fell on the islands of Bhola, Hatiya and Manpura and on the chars, the silt islands that form and erode in the river mouths along the Noakhali and Barisal coasts. Much of that land sits only a few meters above sea level at most. The northern Bay of Bengal is also shaped like a funnel over a wide, shallow shelf, and that shape piles wind-driven water higher as it approaches the coast.

Published accounts give different heights for the surge, and I am not going to choose among them. They agree that the water arrived at night and swept over whole islands, and many of them place its arrival near high tide. Survivors described clinging to trees and roofs as the water passed over. Landless families had settled many of the chars only recently. The storm also arrived during the season when laborers came from other districts to harvest the rice crop. The people on the land that night therefore included many who appeared in no local record, which is one reason the death toll can only be given as a range.

The storm had no name at the time. Tropical cyclones in the north Indian Ocean were not named until a regional naming list came into use in 2004, so “Bhola” is a later label taken from the island. I was not involved in the response or in any later work in Bangladesh, and I do not read Bengali. For this piece I rely on English-language material: the World Meteorological Organization’s archive of weather and climate extremes, the epidemiological survey Alfred Sommer and W. Henry Mosley published in The Lancet in 1972, published descriptions of Bangladesh’s Cyclone Preparedness Programme, and standard histories of the 1971 war.

The warning chain in 1970: satellite pictures, bulletins and Radio Pakistan

The meteorological end of the warning chain worked better than its reputation suggests. By 1970, American polar-orbiting weather satellites were photographing the Bay of Bengal, and images of the storm’s organized cloud mass over the bay before landfall survive in the record. The Pakistan Meteorological Department tracked the system and issued warnings, and Radio Pakistan broadcast them. Forecasters in 1970 had no way to model surge height the way modern storm surge models do. Even so, they knew before landfall that a strong cyclone was headed for the delta.

The trouble lay in what the warnings said and in who could hear them. Accounts of the 1970 broadcasts describe warnings framed in a numbered signal system developed for seaports and shipping. Under that system the numbers described conditions for a harbor. They did not tell a farmer on a char what the water would do or where to go. A warning in that form assumes a listener who knows the code, owns a radio, is awake to hear the broadcast, and has somewhere to go once the meaning sinks in. On the coastal islands in November 1970, nearly every one of those assumptions failed.

The lesson I want readers to take from Bhola is the difference between links in the chain, because the same difference shows up in American warning failures. Detecting a hazard, forecasting its path, delivering the message and giving people a protective action they can actually take are separate jobs. An agency can do the first two well and still lose people at the last two. At Bhola the forecast was broadly right, and what was missing was delivery and a place to go. That is a different failure from a bad forecast, and better meteorology alone would not have fixed it.

Where the warning stopped: radios, night, open water and no high ground

Few households in rural East Pakistan owned radios in 1970, and on the outlying islands and chars the share was smaller still. The broadcast reached the towns and better-off households first and many coastal families not at all. Nothing existed to carry a warning by hand from a radio in a market town to the houses on a char. The local administration of the time had no trained volunteers assigned to the task, and it had no siren or signal flag system. Where word of the storm did travel, it traveled by chance.

A warning heard in time still needs a protective action behind it, and the islands offered almost none. Very few buildings stood high enough or were built strongly enough to survive the surge, and there was no designated shelter. Leaving an island meant crossing open water by boat in rising wind. In the 1960s a coastal embankment program built earthen embankments that protected some areas from ordinary tides. They were not built to hold back a surge on the scale of 1970, and in many places the water went over them.

Credibility probably mattered too, though it is harder to document. The coast had seen many warnings for storms that weakened or turned. People who had sat through earlier warnings without harm had reason to doubt the next one. That was especially true when the warning carried no instruction and leaving meant abandoning cattle, stored grain and a house that might be looted. I cannot measure how much of the toll that skepticism explains, and I am wary of anyone who claims to, so I treat it as one factor among several rather than an explanation.

The death toll: why 300,000 to 500,000 is the honest answer

Nobody knows how many people died, and no archive is likely to settle the question. In 2017 the World Meteorological Organization, in its archive of weather and climate extremes, listed the 1970 cyclone as the deadliest tropical cyclone on record and used an estimate of about 300,000 deaths. Figures of 500,000, and sometimes higher, appear widely in histories of the period and in Bangladeshi accounts of the disaster. I have not found the upper figure traced to a single survey I can point a reader to. Both ends of the range are estimates, and I give the range because the evidence does not support anything narrower.

Several features of the event made an accurate count impossible. Whole households died together, which left no surviving relative to report them missing. The surge carried bodies out to sea and into channels where they were never recovered. The chars had no reliable population register. Nobody on the coast counted the harvest laborers from other districts, and if they were reported missing at all, it was by families far away. Any figure for the dead depends on a figure for the living before the storm, and on those islands the second number was itself uncertain.

The most careful early attempt at measurement I know of is the survey by Sommer and Mosley, physicians working with the cholera research laboratory in Dhaka (then usually spelled Dacca). They sampled surviving households in the affected area and estimated mortality from the sample rather than from body counts. Their paper reported that deaths fell most heavily on young children, older people and women, and that healthy adult men were the most likely to survive, which fits a disaster in which survival often meant holding onto a tree through hours of moving water. I am not going to repeat their overall estimate from memory. Readers who want it should read the 1972 paper itself, which also shows how a population estimate can be built when no registry exists.

Relief from a distant capital and the charge of neglect

Pakistan in 1970 was one country in two pieces. The national government sat in West Pakistan, separated from East Pakistan by roughly a thousand miles of Indian territory. East Pakistan held the majority of the country’s population, and it had long complained that the West dominated the army and the civil service. Relief therefore had to be directed largely from the West, across that distance, into a delta with few roads, where many affected islands could be reached only by boat or helicopter.

The relief effort was widely criticized as slow and too small. Foreign governments and relief organizations sent aid, including helicopters from abroad, and foreign journalists reported on survivors waiting days for food and water. The central government defended its effort, pointing to the scale of the disaster and the shortage of aircraft and boats. I am not in a position to say how much of the delay came from decisions in the capital and how much from geography that no government of the time could have overcome quickly, and the histories I have read do not fully agree on that point either.

What is settled is how East Pakistanis perceived the response. Political leaders there, including Maulana Abdul Hamid Khan Bhashani and Sheikh Mujibur Rahman of the Awami League, publicly attacked the central government’s handling of the disaster and tied it to the region’s older grievances. The point for an emergency manager is narrower than the politics. A population that already believes it is neglected will read a slow response as confirmation, whatever the logistical explanation. The response then becomes part of a political argument, as it did in East Pakistan in the weeks before the December election.

The December election and 1971: what the cyclone can and cannot be credited with

Pakistan’s first general election under universal adult franchise was scheduled for December 7, 1970, less than four weeks after the cyclone. It went ahead on that date, although polling in a number of cyclone-affected constituencies was postponed into January 1971. The Awami League campaigned on the Six Points, the program of autonomy for East Pakistan that Mujib had put forward in 1966. It won 160 of the 162 general seats allocated to East Pakistan, which gave it an outright majority in the National Assembly.

The assembly never met to form a government on that result. President Yahya Khan postponed its session in March 1971, and negotiations among Yahya, Mujib and Zulfikar Ali Bhutto of the Pakistan Peoples Party broke down. On the night of March 25 the army began a military crackdown in Dhaka. The war that followed drove millions of refugees into India, and India entered the war in December. Pakistani forces in the East surrendered in Dhaka on December 16, 1971, the date Bangladesh marks as Victory Day.

A common claim holds that the cyclone caused Bangladesh’s independence, and that claim is an overstatement the historical record does not support. The language movement of the early 1950s, the economic disparity between the two wings, the Six Points, and Mujib’s arrest and trial in the late 1960s all came before the storm. Historians who discuss the cyclone generally treat it as something that deepened East Pakistani anger at the central government in the weeks before the vote. That is a defensible reading. Still, nobody can measure the cyclone’s share in an election result driven by grievances twenty years old, so I describe it as one contributing factor in a political crisis that was already well underway.

The Cyclone Preparedness Programme: carrying the warning the last mile by hand

For warning practice, the most important institution to come out of 1970 is the Cyclone Preparedness Programme, usually shortened to CPP. It grew out of a preparedness effort that the League of Red Cross Societies began in the region after the 1970 cyclone. After independence, the Government of Bangladesh and the national Red Cross society, now the Bangladesh Red Crescent Society, took it over as a joint program in the early 1970s. Different accounts give 1972 or 1973 as the date of formal establishment, and readers who need the exact date should take it from the CPP itself.

The design addresses the specific failure on the islands in 1970. The Bangladesh Meteorological Department issues the forecast. The CPP relays it over its own radio network from headquarters in Dhaka to offices in the coastal districts, and from there to volunteer teams organized by village unit. Volunteers carry the warning on foot, by bicycle and by boat, using megaphones and hand-operated sirens. They also raise signal flags at set points, so that people who missed the spoken warning can still see the warning level. The same volunteers are trained to help with evacuation to shelters, first aid, and search after the storm. The program has made a point of recruiting women volunteers so that warnings and evacuation help reach women in households where a male stranger would not be welcome.

The volunteer force numbers in the tens of thousands, and current figures should be taken from the program’s own published material. As I read it, the system works because the last link in the chain is a neighbor the household knows, who speaks their language and arrives with an instruction rather than a code. A volunteer at the door can answer a family’s objections about cattle or property in a way a broadcast cannot. That is why the closest American equivalent is a door-to-door notification plan staffed by local CERT members, not an additional siren or text alert.

Shelters, the 1991 cyclone and the long decline in deaths

Shelter construction began in the 1970s, but the shelters built over the next two decades covered only part of the population at risk. On the night of April 29, 1991, a severe cyclone struck the coast near Chittagong. The death toll commonly cited from Bangladeshi government figures is about 138,000, and that number carries uncertainty as well. Accounts of 1991 commonly report that the warning reached far more people than in 1970, through the CPP and wider radio ownership. They also report that the available shelters could hold only a fraction of the people who needed them, and that many who heard the warning stayed home.

After 1991 the Bangladeshi government, with funding from the World Bank and other international partners, expanded construction of multipurpose cyclone shelters. These are raised concrete buildings that serve as schools or community buildings for most of the year and as refuges during a storm, so the shelter is a familiar, maintained building rather than a locked structure nobody visits. The coast now has thousands of them. Over the same decades the country also improved its embankments, planted coastal forests, built better roads and, more recently, gained mobile phone coverage.

Later severe cyclones have killed far fewer people. Bangladeshi government figures for Cyclone Sidr in November 2007 put the dead in the low thousands. When Cyclone Amphan approached in May 2020, authorities reported evacuating more than two million people, and official counts of deaths in Bangladesh were in the dozens. These comparisons need care, because no two storms share the same strength, track, timing and tide, and the decline reflects forecasting, warning, shelters and infrastructure working together. Readers who need current shelter counts or casualty figures should take them from Bangladesh’s Department of Disaster Management.

What to do at your agency

• Have your public information officer pull the agency’s prewritten warning templates for flood, hurricane or severe weather messages and rewrite one in plain words that name the hazard, the affected area and the specific action to take, with no internal codes, then file the revised version with the joint information center templates by the end of the month.
• Have your emergency manager ask the county agricultural extension agent or the local migrant health center, in writing, which months seasonal farm workers are present, where they live and which languages they read, and write one paragraph into the existing warning and notification annex naming how those workers will be reached.
• Have your CERT coordinator draw up a door-to-door notification roster for one defined high-risk area, such as a mobile home park in a mapped flood zone, listing which trained member knocks on which doors, and put it on the agenda of the next scheduled CERT meeting for sign-off.
• Have your emergency manager check every designated shelter on the county list against the current storm surge map from the National Hurricane Center or, inland, the FEMA flood map and any dam inundation maps, and record in writing that no shelter sits inside the zone it is meant to protect people from.
• Have your radio system administrator or 911 center supervisor test one alerting path to a known receiver this month, such as an outdoor siren, a NOAA Weather Radio at a school, or a test message to staff phones in the most remote part of the county, and log whether the alert was received and understood.
• Have your public information officer add a required source line and date line to the casualty reporting template, so that no death or injury figure leaves the agency without a named source and a time attached.

Takeaways

• The 1970 cyclone came ashore on the night of November 12 across the delta coast of East Pakistan, and its surge swept low islands such as Bhola, Hatiya and Manpura and the surrounding chars.
• Weather satellites photographed the storm and the Pakistan Meteorological Department issued warnings, so the failure lay mainly in delivering the warning and in having no protective action available, not in detecting the storm.
• Accounts describe warnings framed in a port signal system built for shipping, broadcast by radio to a coastal population where few households owned radios and none had a designated shelter.
• The World Meteorological Organization uses an estimate of about 300,000 deaths, figures of 500,000 or more appear widely in histories and Bangladeshi accounts, and the honest statement is the range.
• The central government’s relief effort was widely criticized as slow, and East Pakistani leaders tied that criticism to older grievances in the weeks before the December 1970 election.
• The cyclone is best described as one contributing factor in the crisis that led to Bangladesh’s independence in December 1971, because the main grievances predated the storm by two decades.
• Bangladesh’s Cyclone Preparedness Programme closed the last-mile gap with trained local volunteers who carry warnings door to door, and the multipurpose shelters expanded after 1991 gave people somewhere to go once warned.