The hurricane that came ashore at Galveston, Texas on the night of September 8, 1900 is described by NOAA as the deadliest natural disaster in United States history. The number it killed is still published as a range rather than a figure. The city had several things a modern planner would recognize: a federal forecast office, a telegraph line to Washington, and storm signals that were flying before the water came. That is why the failures deserve close study. The seawall and grade raising that followed still protect the Gulf face of the island, which is a narrower thing than protecting all of it.

The island in 1900: how high it stood and who lived on it

Galveston in 1900 was a cotton port and one of the largest cities in Texas. It sat on the eastern end of a long, narrow barrier island, separated from the mainland by Galveston Bay and connected to it by railroad and wagon bridges. The federal census taken that summer counted roughly 37,800 people in the city. That figure matters later in this article because it is the only firm population base anyone has, and it describes residents. It does not capture the seamen, visitors and seasonal laborers who were on the island in September.

The ground the city stood on was low. Most published histories give about 8.7 feet above sea level as the highest natural point, and much of the residential city sat well below that. A strong onshore wind combined with a high tide could put water in the streets, and residents were used to those overflows. In practice, the city’s experience of storm water was the shallow, temporary kind, which goes some way to explaining how a population can live with a hazard for decades without measuring itself against the worst case.

Texans had a recent example of what the Gulf could do to a low town. Indianola, on Matagorda Bay down the coast, was wrecked by a hurricane in 1875 and destroyed by another in 1886, after which the town was abandoned. Protective works for Galveston were discussed in the years that followed and were not built. A popular account ties that decision to a single newspaper article by the local Weather Bureau official. That link is an inference later writers have drawn, and the documents do not settle it, so I would not teach it as the reason.

What a forecaster could see in 1900, and the open Gulf he could not

The Weather Bureau of 1900 was an agency of the Department of Agriculture. Its forecasting rested on surface observations taken at fixed land stations and sent by telegraph to the central office in Washington, where forecasts and orders to display storm signals were issued. A local official, such as the man in charge at Galveston, did three jobs: he took observations, maintained the instruments and hoisted the flags the central office ordered. The authority to warn sat in Washington rather than on the coast, which shaped what any local forecaster could say publicly and how fast he could say it.

The Bureau had extended its reach into the Caribbean during the war with Spain in 1898, so observations from island stations were available when a storm crossed the Antilles. Once a storm moved west of Cuba into the open Gulf, the network had almost nothing to report from. Ship weather reports by radio were not part of routine forecasting. Aircraft reconnaissance, weather radar and satellite imagery all lay decades in the future. In early September 1900, a forecaster could know that a storm had crossed Cuba and then have essentially no information about it until the swell, the wind and the barometer at coastal stations began to change.

The science of storm surge was even further behind than the science of the track. Nobody in 1900 had a method for estimating how high the water would rise on a particular stretch of coast, given the storm’s strength, its angle of approach and the slope of the seabed offshore. Even a perfect track forecast would not have produced a figure for how deep the water would get on a given street. That is the figure a resident needs to decide whether a two-story frame house is a safe place to spend the night.

The Bureau did order storm signals displayed on the Texas coast before the hurricane arrived, so it is wrong to say Galveston received no warning at all. What the city lacked was a warning that conveyed a major hurricane and a surge that would put the entire island under water. The bridges to the mainland were the only way off, and a warning that came early enough to use them would have had to arrive days ahead, which the system of 1900 could not do.

Isaac Cline: the 1891 article and his own account of the storm day

Isaac Cline was the Weather Bureau official in charge at Galveston, and his brother Joseph Cline worked in the same office. In 1891 Isaac Cline wrote an article for the Galveston News arguing that a hurricane doing serious damage to the city was very unlikely. Erik Larson’s 1999 book Isaac’s Storm made that article widely known, and Larson quotes from it. I am paraphrasing it here rather than quoting, because the version of the wording that circulates online has passed through too many hands to trust without the original page in front of me. The article is the strongest evidence in the record that the local expert underestimated the hazard, and it was written by a careful, well-regarded professional working with the tools of his time.

Cline’s account of the storm day itself comes mainly from two places: his own report, which the Weather Bureau published in the Monthly Weather Review, and his 1945 memoir, Storms, Floods and Sunshine. In those accounts he describes seeing the tide rise during the morning even though the wind was blowing from a direction that should have held it down, and he says he reported conditions to Washington and went along the beach area urging people to move to higher ground. Larson treats the beach warnings with open skepticism because he found no independent testimony confirming them. Joseph Cline later published his own memoir, and the two brothers’ accounts of who did what that day do not agree. The fair way to state the beach warnings is that Isaac Cline said he gave them.

Cline’s wife died in the storm when their house was destroyed. He survived with his daughters and his brother, and he went on to a long career in the Bureau, much of it in New Orleans, where he continued to study tropical cyclones. Judging him by modern standards misreads what the instruments of 1900 could tell anyone. The 1891 article is the part of his record that transfers to a working agency today: a respected local specialist reasoned his way to confidence about a hazard that had not yet struck his own city, and the people around him had no independent way to test the reasoning.

Washington and the Cuban forecasters, as the record is reconstructed

Cuba had its own forecasting tradition, older in some respects than the American one. The Belen Observatory in Havana, run by the Jesuits, built a reputation in the nineteenth century for hurricane observation and warning, largely through the work of Father Benito Vines, who died in 1893. After 1898, with Cuba under American military occupation, the Weather Bureau set up its own stations in the Caribbean, and the two traditions came into contact on unequal terms.

Larson’s reconstruction, drawn from Bureau correspondence, is the account most readers know. In it, Bureau chief Willis Moore and his office regarded the Cuban forecasters as prone to alarm and as competitors to the Bureau’s authority, and the Bureau moved to restrict Cuban storm forecasts from traveling over the telegraph lines the United States controlled on the island. Larson also reports that in early September 1900 the Bureau’s forecasters expected the storm to curve north and threaten Florida and the Atlantic coast, while forecasters at Belen expected it to continue west into the Gulf. On Larson’s account, Belen’s expectation was the correct one.

The word usually attached to this episode is suppression, and the owner of this site asked me to address it, so here is how I would state it precisely. The record as Larson presents it supports two claims: the Bureau restricted the circulation of Cuban forecasts, and Bureau and Cuban forecasters disagreed about the track. Whether a Belen forecast carried to Galveston would have reduced the death toll is a counterfactual. A warning that the storm was entering the Gulf would not have told anyone where it would land or how high the water would rise, and nobody can know how the city would have acted on a forecast from a source the federal government had told it to discount.

The finding that survives without the counterfactual is institutional. A central office with the sole authority to warn treated a competent outside source as a problem to be managed rather than as information to be weighed, at a moment when its own network could see nothing over the open Gulf. Anyone who wants to go further than Larson should go to the Weather Bureau records held by the National Archives rather than to the many summaries of his book.

The night of September 8 and the ridge of wreckage

Water was over parts of the island during the day on September 8, and the bridges that were the only route to the mainland were lost to the storm. Conditions grew worst in the evening and into the night. The Bureau’s anemometer failed during the storm, so the peak wind on the island was never measured. The Category 4 classification found in NOAA’s hurricane database comes from later reanalysis rather than from a reading anyone took that night.

The mechanism of destruction is the part responders tend to underestimate. As surge and waves broke up the buildings nearest the Gulf, the wreckage did not stay where it fell. It was driven inland as a moving mass of timber, roofing, rail and whole wall sections, which battered the next row of houses and added them to the load. When the water went down, a long ridge of wreckage lay across the city. Contemporary accounts and later histories describe people who were crushed or trapped in that moving debris as well as people who drowned. For a modern planner, a surge zone with wave action is a collapse environment as much as a flood environment.

The loss at St. Mary’s Orphan Asylum, which stood near the beach west of the city, is the episode most often retold. The figures commonly given are ten Sisters of Charity of the Incarnate Word and ninety children dead, with three boys surviving. The building stood on low ground close to the Gulf, the position the surge reached first. Institutions that care for dependent people, sited for cheap land or open space, carry the same exposure today, and they rarely have the staff to move everyone at once.

With the bridges and the telegraph gone, the island could not tell anyone what had happened. The first word reached the mainland when survivors crossed the bay by boat and got to a working telegraph, so for a time the outside world knew there had been a storm and did not know its scale. From a communications standpoint, this is the pattern I have watched repeat in smaller events over twenty-five years. The area with the worst damage goes silent first, and the silence gets read as the absence of a problem rather than the presence of one.

Why the death toll is published as a range

The range most commonly published runs from about 6,000 to about 12,000, and NOAA material most often carries the storm at no fewer than 8,000. Those figures are not estimates of the same thing, which is the first reason they disagree. The lower figure of about 6,000 is generally the estimate for the city of Galveston itself. The higher figures add deaths on the rest of the island and along the mainland shore of the bay, where small communities were also struck and recordkeeping was thinner still. Averaging the two ends would combine a city estimate with a regional one, and picking the higher number without saying what it covers overstates what anyone can show.

The second reason is the population base. The 1900 census gives a resident count for the city, but nobody counted the transient population on the island that weekend: sailors on ships in port, laborers, travelers and visitors. A person nobody knew was there cannot be reported missing, and a household lost in its entirety leaves no survivor to file the report. Any list of the dead built from survivor reports will therefore come out shorter than the true toll.

The third reason is what had to be done with the dead. In the heat of a Texas September, with bodies throughout the wreckage and no means of refrigeration, the city first tried to bury the dead at sea by loading them onto barges and taking them out into the Gulf. Many came back ashore. After that, bodies were burned on pyres on the beach and in the debris, most of them without identification and without any record a modern medical examiner would recognize. Contemporary accounts describe men pressed into handling the dead, some under compulsion. Remains continued to turn up in the wreckage and along the shore for a long time after the storm, which kept the count open long after any official tally.

The Rosenberg Library in Galveston maintains collections on the storm, including a compiled list of known victims, and it is the place to go for names. That list is shorter than any of the estimates. This is what the mechanism above predicts, so it is not a sign that the estimates are inflated. A named-victim list measures the people somebody could identify and report. The 6,000 to 12,000 range is an attempt to measure everyone, and the gap between the two is a direct record of how many people died with nobody left to say so.

The seawall and the grade raising: what was built and what it covers

The recovery changed how Galveston was governed before it changed how the city was built. In 1901 the city adopted a commission form of government, with a small board of commissioners each running a department, and the model was widely copied by other American cities in the years that followed. A board of three engineers, Henry M. Robert, Alfred Noble and H. C. Ripley, was engaged to recommend protective works. Their plan had two parts: a seawall along the Gulf face, and raising the grade of the city behind it.

Construction of the seawall began in 1902. The first section, roughly three miles long and about seventeen feet high, was completed in 1904. The wall was extended in stages over the following decades to roughly ten miles. It was financed locally, with help from the state legislature. If you need the engineering specifications or the bond history for a presentation, take them from the city’s and county’s own published histories rather than from this paragraph.

The grade raising is the part people find hardest to picture. Sand dredged from the shipping channel was pumped as a slurry into the city through a temporary canal between 1903 and about 1911, and it was spread over neighborhood after neighborhood, which raised the ground behind the wall. The fill was deepest near the seawall, commonly given as up to about seventeen feet there, and sloped down toward the bay. Houses, churches and commercial buildings were jacked up so that fill could be placed beneath them, and residents lived for years on catwalks over wet sand while it happened.

The works were tested in 1915, when another major hurricane struck with a storm tide comparable to 1900. Deaths in the city were a small fraction of the earlier toll, a result generally credited to the wall and the raised grade. Hurricane Ike in 2008 showed the limit of the design: much of the flooding in the city came from the bay side, behind the seawall, which the wall was never built to stop. For a mitigation plan, that means each protective structure should be recorded together with the direction of threat it was designed against and the directions it leaves open, which in Galveston’s case is the bay shore behind the wall.

The warning chain today, and where the 1900 failures still turn up

The specific blind spot of 1900 has been closed. The National Hurricane Center tracks Atlantic storms with aircraft reconnaissance, satellites, radar and ocean buoys. Since the 2017 season it has issued storm surge watches and warnings separate from wind warnings. Coastal evacuation zones are drawn from the National Weather Service SLOSH surge model rather than from a forecaster’s judgment about the seabed. Nobody on the Gulf coast today will meet a major hurricane with the information gap that faced Galveston.

The failures that persist are the human and institutional ones, and in my professional judgment they show up in modern after-action reports more often than any technical shortfall. A public message that names a storm category without telling a resident how many feet of water to expect at their own address leaves that resident to do the 1891 reasoning alone, usually from memory of a storm they stayed through before. Experienced residents and experienced officials both anchor on the worst event they have personally seen. That is the same confidence Cline expressed in print, and it calls for nothing more sophisticated than an elevation figure and a surge figure placed side by side.

The Cuban episode carries a narrower lesson, and it needs stating carefully, because a single official voice for public warnings is correct practice. Consistency in public messaging saves lives, and the National Weather Service is the authority for warnings in this country. The failure in 1900 lay in the Bureau’s own handling of an outside forecast: it dismissed the forecast while forming its own view, at a moment when it could see nothing over the open Gulf. An emergency operations center today receives private forecasts, utility and university products, and a flood of social media, and the discipline that matters is to issue public messages from the official source while letting the planners inside the room look at everything.

The isolation problem is the one I would put in front of any communications administrator. Galveston lost its bridges and its telegraph at nearly the same moment because both crossed the same water. Many counties today route their dispatch backhaul, their fiber and their microwave paths across one river crossing or through one tower site, and the affected area falls silent first. Plan for the area with the worst damage to be unable to report for several hours, and decide in advance who goes to look when it does.

What to do at your agency

• Have your emergency manager pull the current surge, inundation or dam failure map for your jurisdiction’s worst credible water event, look up the ground elevation of the 911 center and the EOC, and write both numbers side by side into the facility section of the emergency operations plan this month.
• Ask your radio system administrator for a one-page list of every path that carries 911 and dispatch traffic out of your county, mark each one that crosses the same bridge, river or tower site as another, and bring the marked list to the next communications committee meeting.
• Have your public information officer open the pre-scripted evacuation message template, check whether it tells residents the expected depth of water or other local impact at named places rather than only a storm category or hazard name, and rewrite that one paragraph if it does not.
• Ask your communications center manager to show you the written procedure that says which sources may trigger a public alert through your alerting system, and add one sentence stating how unofficial forecasts that reach the center are routed to the EOC without being broadcast.
• Ask your county coroner or medical examiner for the written plan covering temporary storage of the dead and the record kept for each decedent when normal identification is impossible, and confirm by reading it that the plan exists in writing rather than in someone’s memory.
• Add one item to the agenda of your existing hazard mitigation planning committee asking the owner of each levee, floodwall or seawall in the plan to state in one sentence which direction of threat that structure does not protect against, and record the answers in the plan.

Takeaways

• NOAA describes the Galveston hurricane of September 8, 1900 as the deadliest natural disaster in United States history, and NOAA material most often gives the toll as at least 8,000.
• Published estimates commonly run from about 6,000, generally the estimate for the city alone, to about 12,000, which includes the rest of the island and the mainland shore, so the two ends measure different populations and should not be averaged.
• The toll is uncertain because nobody counted the transient population, because bodies were burned or lost at sea without identification, and because a household lost entirely leaves no one to report it missing.
• Galveston was not below sea level, since its highest natural ground is commonly given as about 8.7 feet, and the storm tide, usually estimated at around fifteen feet, put the entire city below the water the storm produced.
• The Weather Bureau did order storm signals on the Texas coast, but its network could not see the storm in the open Gulf and had no method for forecasting surge height, so the warning that reached Galveston did not convey a major hurricane.
• The account of the Bureau restricting Cuban forecasts comes mainly from Erik Larson’s reconstruction in Isaac’s Storm, and the claim that a Cuban warning would have reduced the death toll is a counterfactual rather than a finding.
• Isaac Cline’s 1891 newspaper article doubting that a hurricane could seriously damage the city is documented, while his account of warning residents along the beach on the storm day is his own and has not been independently corroborated.
• The seawall, first completed in 1904 and later extended, and the grade raising of roughly 1903 to 1911 protected the city in the 1915 hurricane, but Hurricane Ike in 2008 flooded much of Galveston from the bay side, where the wall gives no protection.