In a single morning, Hong Kong suffered a disaster estimated to have killed around 3% of its population, leaving countless families bereaved. On 18 September 1906, a typhoon swept across the territory with devastating suddenness. The fiercest winds lasted only a few hours, yet Victoria Harbour became a scene of wrecked vessels and bodies floating in the water. Later research put the death toll at around 15,000. Compared with the population at the time, that would mean roughly 1 person in every 30. Known as the Bingwu Typhoon, the disaster left a disturbing question: how could a city whose prosperity depended on its harbour lose so many people in the very place where they earned their living?
The figure of 15,000 comes from later historical estimates. The Harbour Master’s report at the time suggested around 5,000 deaths, while acknowledging the absence of complete records. Subsequent studies proposed higher figures. A re-analysis published in 2022 considered a death toll exceeding 10,000 reasonable. Earlier Hong Kong Observatory publications also described around 15,000 deaths, or approximately 3% of the population. “1 in 30” therefore conveys the scale of the catastrophe, while the actual toll remains uncertain. Many victims lived aboard their boats. When a vessel disappeared with everyone on board, establishing how many people had been there could itself become impossible.
Hong Kong was already a busy trading port, sustained by people who worked on the water. Fishing boats supplied food. Sampans carried passengers. Small craft and lighters moved people and cargo between larger ships and the waterfront. For many families, a boat was their livelihood, their home and everything they owned. A merchant ashore could close the premises and seek shelter. A family living afloat first had to protect the vessel beneath their feet. Historical research estimated that around 90% of those killed were boat people. The harbour generated income throughout the city, but those most exposed to its dangers carried the risks of their work and their family’s accommodation together.
On the morning of 18 September, danger approached rapidly. At 8 am, the Observatory raised a black drum, signalling a tropical cyclone somewhere within about 480 kilometres to the east of Hong Kong. At 8.40 am, the typhoon gun sounded a warning of severe local winds. Conditions in the harbour were already deteriorating sharply, and vessels were scrambling for protection. Contemporary newspapers described boats rushing towards the Causeway Bay typhoon shelter, colliding along the way. Some never arrived. Between receiving a warning and reaching safety lay the time needed to cast off, get under way, navigate and find a berth. Once the wind and waves arrived first, hearing the warning offered little assurance of escape.
One of this typhoon’s most dangerous features was its exceptionally small circulation. Using historical weather observations and models, the 2022 re-analysis estimated that its radius of maximum winds was only about 7 to 11 kilometres as it crossed Hong Kong. It was a “midget typhoon”. That radius measures the distance from the centre to the band of strongest winds, showing how tightly its most destructive forces were concentrated. A compact storm might produce only slight changes at distant observing stations. By the time local pressure fell sharply and winds strengthened rapidly, its fiercest winds could already be close. For forecasters relying on sparse surface observations, this structure severely compressed the time available to identify the danger and issue a warning.
Forecasting in 1906 also depended on how observations could be transmitted. Coastal stations already exchanged weather reports by telegraph, but vast stretches of ocean remained poorly observed. Weather satellites and radar lay decades in the future. Ships might encounter the storm and record its pressure and winds, yet the Observatory lacked today’s ability to receive those observations in real time. A cyclone could cross the sea without passing close enough to a land station to reveal its full strength. Researchers could later bring ships’ logs together and reconstruct its track. Forecasters on duty had to work without those crucial reports. The information gap lay between the approaching danger and the people responsible for raising the alarm.
The destructive water also came in different forms. A storm surge is an abnormal rise in sea level caused by winds and low atmospheric pressure. Waves move across that raised surface. Together, they can drive vessels ashore and send water crashing over piers. The Observatory’s estimates based on newer research put the maximum storm surge in 1906 at about 0.39 metres in Victoria Harbour and 1.98 metres at Tai Po. The harbour figure may sound modest. Even a small rise, however, gives powerful waves greater reach, exposing boats and waterfront structures to repeated impacts. Imagining a single towering wall of water sweeping across Hong Kong obscures the destructive work of the waves themselves.
Photographs taken after the disaster preserve the consequences: large ships stranded along the shore, small boats shattered, piers damaged and waterfront buildings collapsed. Historical material published by the Observatory also includes images of workers collecting bodies, and unclaimed victims wrapped in straw mats for burial at sea. Every wrecked small boat could represent a family losing relatives, accommodation and income together. For surviving boat people, the easing winds brought only an immediate reprieve. Funeral expenses, daily necessities and the cost of replacing a vessel still had to be met. A catastrophe lasting a single morning could turn a household’s precarious finances into hardship extending over many years.
The enormous loss of life prompted demands for accountability. On 20 September, the government announced a committee to investigate whether shipping could have received an earlier warning. It concluded that, given the information available, raising the black drum at 8 am had been as early as practically possible. The 2022 reconstruction supported that judgement. This finding explained the technical limits facing forecasters, while bringing another question into focus. If a warning could sometimes provide only a short interval before danger arrived, the city needed safe places that people could reach quickly. Residents needed the practical means to turn a warning into action that could save their lives.
Typhoon shelters consequently became an important issue after the disaster. In 1909, the government passed legislation for shelter construction and advanced the Yau Ma Tei typhoon shelter project, completed in 1915. About 9 years separated the disaster from the new facility’s completion. To someone living ashore, a shelter might appear to be another harbour engineering project. To a boat-dwelling family, it determined where they could take refuge when the next typhoon approached. Disaster prevention presents a familiar spending dilemma: construction costs appear immediately in the accounts, while the lives protected become apparent during a future storm. When a city focuses narrowly on present expenditure, residents with the fewest alternatives are often the first to bear the cost of waiting.
The warning system also evolved. In 1907, loud explosive charges replaced the typhoon gun, and a black cross signal was introduced to indicate hurricane-force winds. A numbered tropical cyclone warning system followed in 1917, laying an important foundation for the numbered signals familiar today. Behind those signals lies a history of repeated efforts to communicate danger more effectively. People must hear the warning, understand the symbols and receive the message in time to act. Its ultimate value depends on whether it changes their circumstances. When vessels have somewhere to shelter and residents have time to get ashore, forecast information becomes a practical chance of survival.
The Bingwu Typhoon showed that protection from disasters requires a whole system of public provision: observations, communication, places of refuge and responsibility for vulnerable residents. The same storm can produce radically different consequences depending on people’s homes, livelihoods and access to protective infrastructure. The ability to wait out a typhoon inside a sturdy building rests on generations of investment in science and engineering. Remembering the terrible estimate of roughly 1 death for every 30 people should also prompt us to ask whether that protection reaches those facing the greatest danger. What do you think Hong Kong should remember most about this disaster?
