Killer Tornadoes of the 1920s
LaGrange and West Point (Troup County), March 1920
Gardner and Oconee (Washington County) February 1921
Bleckley, Bulloch, and Candler counties, April 1929
Killer Tornadoes of the 1920s
LaGrange and West Point (Troup County), March 1920
Gardner and Oconee (Washington County) February 1921
Bleckley, Bulloch, and Candler counties, April 1929
The first week of April in 1936 has long been remembered for two of the deadliest tornadoes of the twentieth century: the Tupelo, Mississippi storm (often rated as an EF5) on April 5th and the Gainesville, Georgia storm (often rated as an EF4) on April 6th. These have often been labeled the fourth and fifth deadliest tornadoes in United States history. Not as well known, however, were EF4 tornadoes that occurred earlier in that week at Greensboro, North Carolina and at Cordele, Georgia. In Georgia on April 2nd, tornadoes also hit Athens, Dawson, Leesburg, Lincolnton, Sasser, Tignal, and Washington. On the 6th, in addition to Gainesville, Acworth, Lavonia, and Woodstock also suffered damage. This blog entry will not repeat the information that has been summarized well on other sites. Instead, I will use records from the Georgia Archives to examine other aspects of the tragedies at Cordele and Gainesville.
In recent years, a great deal of information about the Gainesville, Georgia tornado of April 6, 1936 has appeared online. An especially thorough site is the Digital Library of Georgia’s “The 1936 Gainesville Tornado: Disaster and Recovery,” http://dlg.galileo.usg.edu/tornado/. The site uses materials from several sources including the Georgia Archives, the Hall County Library System, and the Georgia State University Library’s Digital Collections and contains an interactive map, photographs, postcards and even clips of motion picture film. Morris Technology, owner of several Georgia newspapers, has posted images of newspaper front pages covering the Gainesville disaster, and links may be found at the tornado page of the Gainesville Times, http://www.gainesvilletimes.com/flat/tornado36/. Perry Williams compiled weather observations from the first week of April in 1936 and posted his analysis of the conditions that produced the tornadoes at this site: http://www.alabamawx.com/?p=43454. Unfortunately, there is not much about the Cordele storm currently online.
The Gainesville tornado of April 6, 1936 was much better documented in photographs. Proximity to Atlanta allowed news photographers to get to the scene very quickly. The visit of Franklin Roosevelt to the city a few days later also increased national interest as he promised federal assistance for the cleanup and rebuilding. The following two photographs provide a sample of the photographs in the Vanishing Georgia Collection.
Governor Eugene Talmadge dispatched the National Guard to both cities immediately. The Georgia Archives has copies of the proclamations both in the Governor’s Proclamations and in the official records of the Adjutant General. The language of the Cordele and Gainesville proclamations is identical except for the names of the officials requesting assistance: Superior Court Judge A. J. McDonald and City Manager J. R. Brown of Cordele, and Judge B. P. Gaillard and Sheriff I. L. Lawson of Hall County. The following is the Cordele proclamation.
Lindley W. Camp, the Adjutant General sent Col. Lewis C. Pope, 121st Infantry, to Cordele and Col. Thomas L. Alexander, 122nd Infantry, to Gainesville. The following is Alexander's letter from Camp authorizing him to go to Gainesville immediately.
Alexander and five companies of the 122nd rushed to Gainesville to help coordinate the search and rescue activities. Also, hundreds of federal CCC and WPA workers from all over the northeastern part of Georgia came to Gainesville. First to arrive, however, were a large number of policemen and firemen from the city of Atlanta.
I have been interested in how casualties from disasters were reported in the newspapers. Frequently numbers in early reports were inflated. In some areas the racial makeup of the casualties and survivors also made tracking casualties somewhat difficult. I was especially interested in what has become a standard observation for the Tupelo tornado. I found language like this on a number of websites: "This death toll is also inaccurate because African-Americans were not counted.” The Tornado Project site in its discussion of Tupelo has a more detailed explanation:
Since only the names of the white injured were published in newspapers, it is not possible to follow up on the fate of the black injured. This racial aspect of tornado documentation was common until the late-1940's, and occasionally present, in some form, until the mid-1950's.
In order to test this assertion, in this blog post I have evaluated the way African American and white deaths were reported in both the Gainesville and Cordele tornadoes.
I have labeled this Georgia Weather History blog as being an occasional blog. Research and pulling together information from old records still takes time even in this electronic world. I am presenting this draft now because many folks will be looking for information about the Cordele and Gainesville tornadoes during the first week of April. I still have many more sources to check and corrections to make. My evaluation [certainly not exhaustive at this point] of the press coverage of the Georgia tornadoes shows that the number of African American deaths was reported with reasonable accuracy. In Cordele, the names of African Americans were not published; but in Gainesville, the names did appear on all of the published lists. Initial reports often exaggerated the number of casualties. For example, the Associated Press on April 2, 1936 quoted the City Manager of Cordele as saying that “at least thirty-three were killed in Cordele, eight white persons and twenty-five Negroes.” The article went on to list names of five of the whites and none of the African Americans. By the next day, the AP reported the Cordele total as eighteen. The final total was twenty-one.
A source that allows us to check the casualty counts in the newspapers is now open for research. In 1919, Georgia began a system of uniform registration of deaths. By law, the certificates are closed for seventy-five years. The 1936 certificates only recently became available for public use at the Georgia Archives. I compiled a spreadsheet for the Cordele tornado deaths and one for the Gainesville deaths. I then matched up the names on the death certificates with the names in various available newspaper accounts. To ensure that I found all of the deaths, I looked at all of the April death certificates for all of the surrounding counties. The Cordele deaths were all recorded in Crisp County. Some of the Gainesville deaths were recorded in Fulton County because victims had been transported to Grady Hospital, and two deaths occurred in the State Tuberculosis Sanitarium in Alto in Habersham County where a number of injured were taken. The following is an example of a death certificate from Hall County.
The following is a spreadsheet showing the 21 dead from the tornado in Cordele. Please scroll across to see the entire entry for each storm victim. Spelling has been transcribed exactly the way it appears on the death certificates. At the far right is a column checking off the names that appeared in the newspapers.
The final death toll was 21 of whom nine were white and twelve victims were African American.
The spreadsheet below includes all of those who died in the 1936 Gainesville tornado or later from injuries received in the tornado and had a death certificate filed with the state of Georgia. If you scroll to the far right, you will find a column showing if the person was listed in newspaper accounts.
I am still working on these spreadsheets and have not yet consolidated all of the lists published in the Atlanta and Gainesville newspapers or found all of the published obituaries. The "final " death toll appeared in the papers on 4/16/1936, but the last death certificate for a storm victim listed April 23rd as the date of death. There is a discrepancy between the number of death certificates (136) and the numbers of dead reported by the Red Cross to the newspapers. It appears that the difference largely comes from the reports of the building collapse and the subsequent inferno at the Cooper Manufacturing Company or the pants factory as it was known. The Gainesville Eagle reported on 4/16/1936 in a caption under a photograph of the ruined building that over 100 lost their lives in the building. Not all of the death certificates stated where the body had been found, but over thirty certificates were for pants factory employees (almost all white women).
There are clues in the newspaper reports about why totals varied so much. Early newspaper reports around the country on the afternoon of April 6th were that at least 48 had died at Gainesville. The next afternoon, the Atlanta Journal listed 183 known dead including 119 identified whites, 17 Negroes, and 47 bodies burned beyond recognition. The paper also had a report that a Cooper official stated that 125 had been trapped and 60 to 65 were still in the collapsed building. On the 8th, the Atlanta Journal fixed the total at 188 known dead including 154 identified and 34 unidentified with about 25 still missing. The paper said that 48 bodies had been recovered from Cooper Manufacturing Company. The same day, the Gainesville News had an official total of 184 but noted that Red Cross officials admitted the list "is not absolutely accurate." The so-called "final official" total from the Red Cross that appeared in the Gainesville Eagle on April 16, 1936 was killed 149, missing, 17, and "reported dead but no death certificate" 29. The racial makeup was white 127, and "colored" 23. The list of victims by name had 152 names, two "charred bodies," and one unidentified "colored" baby found in a cistern.
The generally accepted death toll for the 1936 Gainesville tornado appears to me only a best guess. Who were these people reported dead with no death certificate? The Atlanta Journal on April 7th probably had the best explanation for the variation in counts when it described the Cooper factory's collapse:
In my brief search, I have not located any lists of the missing and presumed dead. I am sure the lists have been retained somewhere in the area. Comparing all of the available lists will allow us now to add some precision to what was obviously a difficult task in 1936.
In recent years, stories in the media and bloggers have cast doubt on the generally accepted total of 203 dead. It appears to me that the death toll was very unlikely to have been larger than 203 because that number seems to include dozens of unnamed individuals. Yet the following recently appeared on a weather blog:
203 bodies were found--however at the Cooper Pants Factory, as the tornado approached, a gas line exploded and the building was incinerated with the workers inside. In the howling winds of the tornado, the incineration was so extreme that 40 people were missing and never found. So the death toll in Gainesville GA was at least 243.
The notion I mentioned at the beginning of this post that African Americans were not included in death counts has also shown up in accounts of the Gainesville tornado. I have included an example from an Atlanta television station's story:
More than 200 of them were killed in Gainesville. But, historians say the final death toll could not be calculated. On top of that, African-American victims were not included in fatality and injury statistics at that time.
My spreadsheets and the newspaper accounts all show that African Americans were counted. In Cordele, they were the largest number killed. In Gainesville where the percentage of African Americans was smaller, African Americans were still counted and named in the official counts.
The Vanishing Georgia Collection of the Georgia Archives has some of the earliest known storm surge photographs. Vanishing Georgia contains many photographs of the storm surge and wind damage from the 1898 category 4 hurricane that crossed Cumberland Island and passed just south of Brunswick.
Earlier posts on this blog concerned tornadoes [often called hurricanes in early Georgia]. With hurricane season entering its most active stage, the time has come to discuss the tropical variety of hurricanes and how they have affected Georgians. August 27th is the anniversary of two of the deadliest hurricanes to strike Georgia: the 1881 storm reported to have killed 700 and the “Great Sea Islands Hurricane” of 1893 with reported deaths up to 2,500.
For years, meteorologists at various Weather Service offices and the National Hurricane Center as well as emergency management personnel with state operations such as GEMA [Georgia Emergency Management Agency] and local emergency management coordinators have worked to educate coastal residents about the risk to Georgia from a major hurricane strike. Since 1898, Georgians have dealt with hurricanes but none of the stronger categories–the so-called major hurricanes, categories 3, 4, and 5. Some Georgians have come to believe that the bend in the American coastline and the normal eastward curving of storms at this latitude protect Georgia from a direct hit. Because no more than a handful of today’s Georgians were alive when the Georgia coast was last hit by a major hurricane, we have to rely on historical records to help make the case for taking any hurricane threat seriously. Over the next few months, this blog will discuss some of the evidence that exists in the archival record. In this post, I want to highlight some excellent work available online that summarizes and categorizes past storms.
Researchers have mined original records extensively and have compiled extensive data on the paths of hurricanes and the deaths and damage they caused. This summer, the National Hurricane Center updated a useful publication, The Deadliest, Costliest, and Most Intense United States Tropical Cyclones from 1851 to 2010 (and Other Frequently Requested Hurricane Facts) [NOAA Technical Memorandum NWS NHC-6]. PDF and WORD versions are available for download at this site: http://www.hurricanes.gov/dcmi.shtml.
The rankings show that Georgia’s coastal areas suffered the 5th, 6th, and 18th most deadly hurricanes that struck the United States mainland between 1851 and 2010, and those all hit during the last twenty years of the 19th century. The data also show that approximately sixty percent of Georgia’s tropical systems come from the Gulf of Mexico and hit Georgia only after passing across one of our neighbors. All of the hurricanes that did strike the Georgia coast in the past 111 years have been category 2 or less. The authors of the report have a table that calculates the number of years on average between major hurricanes and lesser hurricanes hitting within 50 nautical miles of a particular point on the coast. The “return period” for Brunswick in Camden County should be 34 years for a major hurricane. The last major storm (a category 4) hit Brunswick in 1898. The return period for a hurricane of any category is eleven years, and the last storm to hit Brunswick was a category one hurricane in 1928. For Savannah, the return period for a major hurricane is 36 years; and the last major hurricane to hit there was the category three “Great Sea Island Hurricane” of 1893. Despite the calculated return period of ten years, the last hurricane of any size to strike Savannah was Hurricane David (a category 2 storm) in 1979.
NOAA’s Atlantic Oceanographic and Meteorological Laboratory (AOML), Hurricane Research Division (HRD) has been conducting a long-term Atlantic Hurricane Database Re-analysis Project to revise the National Hurricane Center’s North Atlantic hurricane database [HURDAT.] The website, http://www.aoml.noaa.gov/hrd/ has updated datasets that one can search for hurricane information.
For those interested in Georgia hurricanes, I recommend the following study. Al Sandrik and Chris Landsea in 2003 published a well-researched list of Southeastern hurricanes, Chronological Listing of Tropical Cyclones affecting North Florida and Coastal Georgia, 1565-1899. http://www.aoml.noaa.gov/hrd/Landsea/history/index.html They differentiate between storms that passed offshore or those that actually hit land. This is no easy task, and standards for what is a “hit” have evolved over time. What is especially helpful about their list is that it begins in 1565 well before the Georgia’s official founding with Oglethorpe’s arrival in 1733.
Another source of information about Georgia Hurricanes is the 2006 book published by the University of Georgia Press, Lowcountry Hurricanes: Three Centuries of Storms at Sea and Ashore by Walter J. Fraser Jr., professor emeritus in Georgia Southern University’s Department of History.
Before I move on to other topics in coming weeks, I want to respond to questions and comments readers have submitted concerning my original post on “The Tornado of April 5, 1804.”
Certainly, the use of the word “hurricane” to describe what today we call a tornado created confusion in the twentieth century. Virginia Hill Wilhoit’s history of Warren County (1933) noted that the path of the “Old Hurricane” of 1804 could be followed for many years. She mistakenly assumed that this “hurricane” had originated in the Gulf of Mexico.
I mentioned in my previous post the hurricane names that appeared on original surveys. I checked to see if the example I posted was still named Hurricane Branch. The Hurricane Branch in District 10 is now known as Little Hog Creek [USGS – Gray Quadrangle]. The hurricane name survives in Jones County, however. The major creek that empties into Falling Creek just above Dames Ferry on the Ocmulgee that was called Black Creek on the original surveys is now Hurricane Creek [USGS -- Dames Ferry Quadrangle].
The database of the U.S. Board on Geographic Names has 22 hurricane names in Georgia: 7 branches, 10 streams, 2 churches, 1 valley, 1 lake, and 1 shoal.
One of the best known places named for the 1804 tornado was Howell Cobb’s Hurricane plantation in Baldwin County near what today is the community called Howell Cobb. Cobb (1815-1868) had been Speaker of the United States House of Representatives, Secretary of the U.S. Treasury, and Governor of Georgia before becoming one of the founders of the Confederacy and a major general in the Confederate Army.
General Sherman camped at Hurricane plantation on his march through Georgia in 1864 and burned the plantation house and outbuildings when he discovered that Cobb owned the plantation.
I appreciate the suggestions to draw the path of the 1804 storm on a modern map. Transferring the path from the district plats to actual county maps was not difficult because the Georgia Archives has online in the Virtual Vault many county maps including some from the nineteenth century that show the individual land lots.
The path did not cross the future location of any major populated areas until it reached what is now the northern part of Milledgeville. I wonder if settlers later avoided building towns in the path because of stories about the storm. Deeds had phrases such as “known by the Hurricane passing through the farm” for decades after.
Georgia’s modern weather history has many examples of tornadoes taking similar paths to previous storms; the Gainesville tornadoes in 1903 and 1936 are the most famous because of the number of deaths. Athens had tornadoes in March and May of 1973 less than 2 months apart. The path of the great storm of March 20, 1875 was similar to the “Hurricane” of 1804. It too was reported to have crossed Georgia from Alabama to South Carolina. The newspaper accounts were even more exaggerated than the stories in 1804. The 1875 storm passed through the southern part of Milledgeville. I will write about the 1875 tornado outbreak at a later time. Although reporters assumed that there was a single tornado, plotting the damage reports on a map shows that there were many paths and not just a single one. Milledgeville also suffered considerable damage in the December, 1964 tornado that passed south of Gray in Jones County before striking Milledgeville.
The recent deadly long-path tornadoes in the Southeast reminded me of the “Great Hurricane” that crossed central Georgia on April 5, 1804. Until the 1830s, Georgians referred to tornadoes as hurricanes. A true hurricane did visit the Georgia coast in 1804, but it struck five months later in September.
In 1804, Georgia was sparsely settled especially in its inland areas. The western three-fifths of what became Georgia was held by small numbers of Native Americans. The question asking if a tree falling in the woods makes a sound if no individual is there to hear it certainly applies here. There were few witnesses to the great storm of April 1804; but it was documented more thoroughly than many storms a century later because of an unusual undertaking on the Georgia frontier over the next three years. The Georgia land lotteries began in 1805.
The Georgia Archives has the records from the lotteries including individual plats, district surveys, and the field notes compiled by the surveyors. The Georgia Archives has a special section of its website called the Virtual Vault for collections with images. If you click on the icon for the Virtual Vault and go to the Virtual Vault Home, you will see a list of collections. The pertinent ones for this discussion are District Plats of Survey and District Survey Field Notebooks. Clicking on either will bring up a search page. For the district plats, choose the option to search Baldwin County and a district. Districts with tornado damage are 2, 9, 10, and 11 as you can see in the map below.
Original Baldwin County contained several modern counties, but the tornado of 1804 moved across the parts that today are Jones (districts 11, 10, and 9) and Baldwin (district 2). Today, Milledgeville is in the eastern part of District one. The surveyor for that district did not include tornado damage, but it is doubtful that a tornado well over a mile wide would have stopped at the district line.
District plats provide precise evidence of the path of the 1804 storm. In the example below from the 10th district of original Baldwin County, the Surveyor [Reuben Langston] three years later was still able to show the “Traces of the Hurricane of 5th April 1804.”
Individual plats also showed the damage. Lot 28 of the second district [surveyed in July 1804] has the following wording on it, “Wholely within the ravages of the Hurricane 5th April 1804.”
Tract 21 of the eleventh district when surveyed in 1807 had the surveyor’s notation “in Hurricane” for about 40% of the lot. Several of the district plats show a path well over a mile wide at several points.
In 1802, the Creeks ceded an area west of the Oconee River in middle Georgia. In 1806, a second cession took the western “boundary” of Georgia to the Ocmulgee River. The initial land lottery in which Georgia distributed lands to “fortunate drawers” took place in 1805, and a second was held in 1807. To prepare for the lotteries that had been authorized in 1803, the Surveyor General sent surveyors to lay out 202½ acre lots (45 chains on a side) in original Baldwin and Wilkinson counties. The lots were organized into districts, and the district surveyors prepared individual plats for lots as well as district plats.
When the surveyors began their work in the summer of 1804, the scars of the April 5th tornado were still fresh. Unfortunately, field notes have not survived for the portions of Original Baldwin County crossed by the “hurricane.” As the rest of Georgia was surveyed and lotteries were conducted, the field notes mentioned hurricanes. District 5 of Original Early County has “traces of Hurricane” on lots 7, 8, 9, 29, 30, 31, 52, 53, and 68 in Joel Walker’s 1819 Field Notes. The surveyor for the 20th district of Muscogee, John G. Bostick, referred to an “old hurricane” in January 1827, but the storms mentioned in the 1819 notes and the 1827 notes do not appear to match the 1804 tornado. The long time between the initial lottery and the last one in 1832 makes it impossible to tell much about the track of the 1804 tornado through western Georgia although that didn’t stop one twentieth-century author from speculating about the path. John H. Goff in an 1964 article in the Georgia Review (vol. 18: 224-235) which was reprinted in his Placenames of Georgia (pp. 459-470) made the case for uniqueness of the 1804 storm. He called the 1804 storm “surely one of the greatest, if not actually the greatest storm that has visited the state in historical times,” and he stated that there “are records to show that this tornado passed all the way across the state. . . .” He presented a number of reports of damage along a statewide path, but we know now that early reports of tornado paths covering hundreds of miles are usually not correct. The early press reports in the April 2011 tornado outbreak mentioned a tornado on the ground from Mississippi to North Carolina. Later aerial and ground surveillance showed distinct breaks in the path.
Goff cited the observations of the commission surveying the Alabama/Georgia border in 1826. The evidence Goff cited was not convincing. The chairperson of the commission marking the boundary wrote Governor Troup that they had found the remains of the great storm that visited Sparta in 1804; but when they later discovered another path (3/4 mile wide), the chief surveyor of the commission, Edward Lloyd Thomas, wrote his opinion that this second track belonged to the storm which visited Milledgeville and Augusta in 1804. Goff undermined his own conclusion when he noted that Colonel Benjamin Hawkins in 1798 had reported numerous tornado paths near both locations. Goff does mention that various individuals reported that the path of the “old hurricane” could be followed for many years.
There is no doubt that on April 5, 1804 one or more supercell storms passed across middle Georgia with considerable damage resulting from the tornado or tornadoes that occurred. We have newspaper accounts from Augusta and Savannah in the more heavily settled part of Georgia to confirm that the storm passed to the north of Augusta and continued well into South Carolina. The Augusta Chronicle on April 7th reported the deaths of three women in one house and many persons severely wounded along with much damage to property. The storm was accompanied by hail “said to measure from 9 to 10 ½ inches in circumference.” The issue of April 21st enumerated the horrendous damage in Edgefield district in South Carolina from the “hurricane.” A Savannah paper from April 14th told its readers that they had traced the storm as far back as Hancock County and that it had passed through Warren County before crossing the Savannah River just above Augusta (Columbian Museum and Savannah Advertiser).
The accounts from the eastern part of Georgia are consistent, but Goff’s evidence from western Georgia is not reliable. There were not enough observers to document the path of a storm west of the Ocmulgee River. He also made a mistake in trying to show a longer path to the west. Goff noted that the “hurricane entered old Baldwin near the mouth of Falling Creek on the Ocmulgee.” He described the path near Falling Creek as “narrow and concentrated.” The plats drawn by the surveyor William Watson in 1807 show clearly that the storm first touched down inside original Baldwin County and did not cross the Ocmulgee as Goff indicated. In the district plat, the storm damage first appears in land lot 119 (marked with the arrow); and the initial path is narrow.
I have included a copy of the individual plat for lot 119 to show how the surveyor showed the initial touchdown in Baldwin County.
Goff’s article on Hurricane Placenames contains a large number of places throughout Georgia that had the name hurricane. He explained the common use in early Georgia of the term hurricane for what we now call a tornado. [I have seen official documents as late as 1936 calling a tornado a hurricane.] Most of these place names were not in coastal Georgia as one would expect. Instead these places are all over the state far from the usual damage path of hurricanes. Goff noted the tendency of surveyors giving the name hurricane to streams in tornado damaged areas. Surveyors had to come up with names and some used the same names over and over. For example, in the district 10 plat excerpt below, there is a Hurricane Branch flowing to the upper right out of the tornado damaged path.
Note: The district plats are all online in the Virtual Vault on the Georgia Archives website. They make up RG 3-3-24, Surveyor General, Survey Records – District Plats of Survey, 1805-1832. Individual plats are on file at the Archives in RG 3-3-26, Surveyor General, Survey Records – Headright and Lottery Loose Plat File, 1783-1909. District Survey Field Notebooks are also in the Virtual Vault. They make up RG 3-3-25, Surveyor General, Survey Records – Field Note Books, 1805-1832. The Augusta and Savannah newspapers are on microfilm at the Georgia Archives.