February 19, 1884 — The Enigma Outbreak

[Note: Click images to view larger versions.]

The first two months of 1884 had brought nearly ceaseless rain to much of the United States. Slate gray rain clouds had cast a pall over the land, blotting out what little late-winter sun there was to be had and inundating the already soggy ground. Grass and dirt turned to sloppy mud. Rivers and streams, swollen from the relentless precipitation, bulged and spilled over their banks. Meandering creeks transformed into raging rivers, major rivers into vast inland seas. For the first time in many days, the morning of February 19 brought a welcome respite. Morning broke to fair weather and mostly cloudless skies, and many engorged rivers began to slowly recede.

A broad, deep low pressure system was tracking across the upper Mississippi Valley just south of the Great Lakes, pulling in air from the south ahead of it. A strong southerly wind drawn from the Gulf of Mexico overspread the south with warm, soupy air. In Alabama, temperatures surged by ten degrees in the span of a few hours. In Mississippi, temperatures jumped nearly 20 degrees. Dewpoints rocketed to the mid- and upper-60s as winter-weary residents headed out to enjoy the springlike weather. Several hundred miles to the northwest, the story was much different. Trailing behind the low pressure system was a bitterly cold air mass, dragged down from high in the Arctic. Temperatures dropped by as much as 30 degrees in 24 hours with the passage of the cold front, bottoming out below zero in many places.

Surface chart from 7am on the morning of February 19. Strong southerly flow is evident ahead and south of the low, setting up a broad warm sector.

Surface chart from 7am. Strong southerly flow was evident ahead and south of the low, setting up a broad warm sector. Red “++++” icons indicate tornado tracks recorded in the following eight hours.

Surface chart from 3pm. Surface temperatures ranging from 60 to 75 degrees overspread much of the eastern United States ahead of the deepening low .

Surface chart from 3pm. Surface temperatures ranging from 60 to 75 degrees overspread much of the eastern United States ahead of the deepening low. Ground observations indicate temperatures in excess of 75 degrees actually reached as far inland as Kentucky and western Virginia.

Surface chart from 11pm. Enhanced low-level winds south and east of the rapidly deepening low, along with ample, deep moisture, allowed for continued tornado production through the evening and overnight.

Surface chart from 11pm. Enhanced low-level winds south and east of the rapidly deepening low, along with ample moisture, allowed for continued tornado production through the evening and overnight.


The negatively-tilted trough can be clearly seen on the 500mb map from noon on February 19.

The negatively-tilted trough can be clearly seen on the 500mb map from noon on February 19.

A negatively-tilted trough dug southeast across the North-Central United States, while the core of an intense jet streak rapidly rounded its base. Adding to the already dangerous situation, a bulge of extremely dry air from the west began to nose in on the layer of deep moisture over the Southeast. While winds at the surface blew stiffly from the south and southeast, winds at mid-levels howled from the southwest and upper-level winds blew west to west-northwest. Additionally, wind speeds increased substantially with height throughout the atmosphere. The resulting directional and speed shear, combined with the high instability of the warm, oppressive airmass below and the dry air from the west, created ideal conditions for supercells and tornadoes over a broad area.

By mid-morning, cauliflower-shaped thunderheads began to pop up just ahead of the cold front near the Mississippi River. In the east-central Mississippi town of Winston, the southerly winds intensified and began veering toward the west-northwest. Rain came in fits, accompanied shortly thereafter by quarter-sized hail. As the precipitation abated, sharply shifting winds and a dull roar announced the arrival of the day’s first tornado. The twister destroyed two homes and a cotton gin before lifting near Louisville. Just before noon the first killer tornado barreled through the town of Columbus. Several plantations were destroyed and one woman was killed. The tornado continued its 25-mile track across the state line into Pickens County, Alabama.

At 12:20pm CST, the first violent tornado of the day touched down near Oxmoor, Alabama. The tornado tracked just southeast of Birmingham before tearing through the southern and eastern sections of the industrial town of Leeds. At least thirteen lives were lost, including four members of the Pool family, as dozens of homes were scattered in the howling wind. Newspaper reports suggest the tornado may have had a multi-vortex structure, with eyewitnesses stating that “several black shafts darted in quick succession from the cloud toward the earth.” Survivors of the tornado recounted that much of the town was left unrecognizable. Well-constructed brick homes in the area were leveled and several foundations were damaged or swept away, indicating probable F5 damage. Bodies were found impaled by various objects, and cattle and horses were reportedly skinned and dismembered. Eleven fatalities were recorded in the Leeds area. Thirty one were reportedly “gravely injured,” though the lack of subsequent reports makes it unclear how many ultimately succumbed to their injuries.

Damage from Leeds, Alabama

Drawing of an obliterated farmhouse near Leeds, Alabama. Artist unknown.


 By 1:00pm, the outbreak was nearing its peak as a broken line of supercells sprawled across much of the southeast. A small funnel descended into a field near Cartersville, Georgia. Racing to the northeast, the tornado rapidly intensified as it neared the community of Waleska. Three children were killed after they had been released from school early and taken shelter in their home. At the towns of Cagle and Tate in Pickens County, many homes were completely swept away. The majority of the affected homes belonged to prominent families and were large and well-built, indicating that this tornado, too, was perhaps of F5 intensity. The tornado and associated downbursts cut a swath nearly three miles wide through the forest, leveling homes as far apart as two miles. The bodies of several of the 22 total fatalities were reportedly carried more than half a mile. In total, the tornado and associated downbursts flattened at least 50 square miles of forest.

In East Alabama, what would prove to be perhaps the deadliest tornado of the outbreak descended from the skies just north of Jacksonville. The tornado tore through the Germania tanning yard before killing ten just north of Piedmont, then known as Cross Plains. As many as 20 lives were lost at Goshen, where a schoolmaster and six children were killed in a schoolhouse said to have been “blown to atoms.” All 19 of the remaining children were seriously injured. Large cotton bales were tossed at least half a mile, and many poorly-built homes vanished entirely. At Cave Spring, Georgia, four lives were lost and a number of well-constructed homes were scoured from the Earth. By the time the tornado had come to the end of its 35 mile path of destruction in western Georgia, it had taken more than 35 lives. Newspapers of the day made little effort to document the carnage in these rural communities, and the true death toll was likely much higher.

Shortly after 3:00pm, large hail — some three inches or more in diameter — began falling across central Georgia. The wind veered from south to northwest and the temperature began to drop. Moments later, a large multi-vortex tornado began shredding and snapping trees in Monroe County. Just northwest of Haddock, many large homes were swept away and at least eleven people were killed. Unconfirmed reports indicated at least an additional 12 fatalities. Among those killed were a mother and her four children, who had taken shelter inside their home as the storm approached. Witnesses north of Macon recounted that the tornado contained “multiple dark columns rotating about a common center.” Another 50 were injured along the 30-mile path.

At 8:30pm the final violent tornado of the outbreak began a path of destruction in southeastern Anson County, North Carolina. Two were killed and several homes were destroyed in Pee Dee. The tornado barreled through the southeast edge of the city of Rockingham. Several large homes were swept from their foundations, and large hardwood trees were reportedly debarked, denuded and snapped off near ground level. At least one victim in the area was reportedly thrown nearly a mile. The tornado intensified and widened to nearly one mile as it roared to the northeast. The small railroad community of Philadelphia, about three miles northeast of downtown Rockingham, was completely obliterated, with one observer noting that only “fragments remained of the houses and cabins.” At least 25 homes were razed to their foundations, their debris scattered over large areas. At least 15 residents of Philadelphia were killed in the storm.

In total, multiple waves of tornadoes scoured at least ten states across the Midwest and Southeast. Mississippi, Alabama, Georgia, North Carolina, South Carolina, Illinois, Indiana, Kentucky, Tennessee and Virginia were all impacted by tornadoes and destructive winds. The enigma, however, begins with the final number of tornadoes — although the official number ranges from 47 to 60, the Enigma Outbreak may actually have been among the largest in history. The extent of tornado-related damage is uncertain, but is suspected to have been several million dollars, the equivalent of perhaps $100 million today. More than 10,000 structures were destroyed. The final death toll, too, is a complete mystery. While the conservative estimate is 178 — a terrible toll in its own right — many estimates have placed the final number as high as 1,200. A further 2,500 may have been injured during the outbreak. The system produced more than 37 significant tornadoes (F2 – F5) during its 15-hour duration, ranking it still among the most violent outbreaks on record.

Map of all reported tornadoes during the Enigma Outbreak. Note that the northernmost tornado tracks may actually have been downburst damage. It is likely that there were many more tornadoes that simply went unreported, and some tracks may have been tornado families.

Map of all reported tornadoes during the Enigma Outbreak. Note that the northernmost tornado tracks may actually have been downburst damage. It is likely that there were many more tornadoes that simply went unreported, and some tracks may have been tornado families.


In the days leading up to the event, despite a recently enacted ban on the word “tornado” and a halt to nearly all research and forecasting, an intrepid scientist named John P. Finley worked diligently to record surface and atmospheric data and attempt to piece together the puzzle of tornado outbreaks. Finley developed meticulous tornado charts to track what he believed to be the most significant factors in tornado formation, including temperatures, dewpoints, dewpoint depression, wind direction and observed conditions. In his chart for the morning of February 19, seen below, he noted what he believed to be one of the most important indications of a tornado outbreak. A bulge of dry air, now referred to as a dryline bulge, was being pulled from Texas and Louisiana and forced into an area of warm, moist air overlaying Alabama and Georgia by strong easterly mid-level winds. This phenomenon is still used today to diagnose an enhanced tornado threat.

Enigma Outbreak Surface Map

“Tornado chart” created by John P. Finley on the morning of February 19. Dewpoint isolines indicate a bulge of dry air from Louisiana cutting into the existing tongue of moist air over Alabama and Georgia, boosted by a strong southerly low-level jet. Added arrows represent dry air intrusion (orange) and moist air surge (green).


Because of the extremely dynamic nature of the storm system in place, a range of severe weather effects were experienced over a vast swath of the eastern United States. The upper Midwest was gripped by an intense blizzard, with gale-force winds and snowfall rates exceeding one to two inches per hour in some areas. Much of the Ohio Valley experienced the violent winds and torrential rains of a derecho, with widespread and severe non-tornadic wind damage. Thousands of trees were “prostrated by the extreme wind,” and telegraph communications were cut off for several days. Across the Ohio Valley, the Ohio River and its many tributaries had already spilled over their banks. The additional rainfall from the storms of the 19th and 20th led to flooding which exceeded all records. Homes were swept away by deadly floods in Louisville, Kentucky; Jeffersonville, Indiana; and many other towns along the Ohio and its tributaries. At Union Point in Greene County, Georgia, hail fell in such volume that the ground was covered to a depth of three inches.

Though the true extent of the Enigma Outbreak have been lost to the ravages of time, it still stands among the most widespread and violent tornado outbreaks in history. The tornado which struck Philadelphia and Rockingham remains the deadliest tornado in the history of North Carolina. Limited newspaper resources and shoddy reporting also makes it difficult to place this outbreak historically. Damages and deaths in rural areas were very poorly reported, and the deaths of African-Americans were not counted nor recorded. Given the large African-American population in the affected areas, it’s likely the death toll was much larger than the officially accepted number. The nickname is certainly well-deserved, and the many mysteries of the Enigma Outbreak will continue to draw intrigue well into the future.


19 comments on “February 19, 1884 — The Enigma Outbreak

  1. I find this outbreak so fascinating, and loved your very well-written article. Considering the fact that less than a century’s worth of solid evidence exists, I imagine that the 1884 outbreak may have been one of several in the past few hundred years that was larger than 4/27/11. Imagine – seven, eight, ten or more EF5 tornadoes in a single day!

    • Thanks, I appreciate the kind words! Yeah, this outbreak and the 1932 Deep South outbreak (among others) are very intriguing. I’m sure there have been even more intense outbreaks in the distant past. I still wonder about some of the tornado outbreaks in Canada, too, where much of it goes undocumented because it’s so sparsely populated.

      Even today I have a feeling we’re underestimating how frequent strong/violent tornadoes actually are because we don’t have the means to detect them directly. There are a number of tornadoes in the past two years alone that I suspect were probably violent tornadoes, yet they were rated EF1 or EF2 because they didn’t impact any significant structures.

      • Even more, I imagine that more than 90% of all tornadoes capable of causing EF5 damage are not rated as such. The chances of a tornado reaching EF5 intensity while also passing over a “home of superior construction” is incredibly small. Reviewing extensive aerial images of tornado paths through forested areas on 4/27/11, I could clearly make out vegetation damage similar to that seen after the Phil Campbell storm in the paths of tornadoes rated as EF4’s or less. There were likely 10+ tornadoes on that single day that could have caused EF5 damage.

      • That’s true, too. Probably most strong to violent tornadoes are capable of EF5 damage at some point in their lives, even if only briefly. The EF scale is a huge step in the right direction, but there are still plenty of unresolved issues. I always wondered why there aren’t DIs for various vegetation and vehicles. They’re used by survey teams, but as far as I’ve seen they aren’t official DIs. Except for trees, obviously. But things like grass scouring, low-lying shrubs, vehicles tossed long distances and mangled beyond recognition, etc. Maybe there are too many variables to standardize it — the dampness of the ground, the types of vegetation and possibly time of year, type of soil and so on. But I’d think it would be useful in cases where no traditional damage was caused.

        I also agree that there were likely many more EF5s on April 27. I can think of at least three or four off the top of my head and I’m sure there were others as well. I’d love to see what the final tally would’ve been had this occurred back in ’74. I’ve been working on a post comparing the two but I seem to keep getting distracted with other posts. I probably should start sticking with one at a time rather than writing six or seven at once.

        Oh, I’m finishing up a post on the Palm Sunday tornadoes and there’s one paper I’ve been trying to get my hands on but I can’t find it. I’ve seen a few references to Fujita’s “Estimated Wind Speeds of the Palm Sunday Tornadoes” — apparently it’s one chapter in a larger study? — but it doesn’t seem to be available except from TTU’s archives. I actually have no idea whether there’s worthwhile information there or not, but it piqued my curiosity. Do you know anything about it?

  2. Along with the May 1896, 1902 Goliad Texas, 1908 Dixie, 1920 Palm Sunday, and the May–June 1917 events, this is, to me, one of the most impressive “pre-radar” and “pre-record keeping” tornado outbreaks/sequences to research and find well-written and informative articles about. Keep up the good work!

    P.S. “The Estimated Wind Speeds of the Palm Sunday Tornadoes” I do believe is Chapter 8 of this study by Fujita and Dorothy Bradbury. Click on the PDF logo on this page:


    • Thanks John. I have a few articles that are partially written, the 1920 Palm Sunday outbreak is one of them. There are several older events that I think are fascinating. I’ve been working on the 1893 Pomeroy, Iowa tornado as well since I have a ton of resources for that. I’ve gotta start focusing on one article at a time so I can get them done, haha.

      That’s what I thought it was too, but there are several references that state it as being Chapter VII of “A Comprehensive Study of the Palm Sunday tornadoes” so I wasn’t sure. Maybe it is the same thing after all. I contacted TTU but didn’t get a response. I suppose it’s not a big deal either way, there’s already a ton of information available about the outbreak.

  3. I googled “A Comprehensive Study of the Palm Sunday tornadoes” and the first result was this link to Amazon: http://www.amazon.com/Estimated-wind-speeds-Sunday-tornadoes/dp/B0007FDKWI It’s possible they could both be the same thing, just were published a few times under different titles. In the PDF link, it’s Chapter 8, which is just one number off from 7. Not sure, just wouldn’t be surprised if they were both the same document. And “VII” could be a typo for “VIII”, who knows?

    And I wasn’t trying to rush you into a bunch of articles, haha. The amount of Dixie Outbreaks really interest me, especially since information on them is rather scarce. Have you thought of doing an article that covered the history of tornadoes in “Dixie Alley”, given how severe the lack of information is on older events?

    The oldest tornado that I wish that much more info was available on is the Great Natchez of 1840. If only photography had been invented…

  4. Yeah, I actually have an idea sketched out touching on the history of Dixie Alley and comparing to the traditional Tornado Alley. Some people also include Hoosier Alley (mostly Indiana, with parts of eastern/southern Illinois, western half of Kentucky, southern Michigan and western Ohio) and Carolina Alley. The tornado frequency is marginally higher than surrounding areas there, but I don’t know that they merit their own “alleys.”

    Actually I’d like to do a post covering several aspects of tornado climatology, including the issues Max and I talked about above. It’s likely that we’ve substantially underestimated the frequency of tornadoes which are capable of causing violent-level damage, and it’s also likely that we have an incomplete understanding of the distribution of tornado frequency because of population differences.

    Anyway, if you have suggestions for any particular thing you’d like to see, feel free to send me an email. I put the link on my “About” page. I have a ton of topics to cover, but it’s nice to see which ones people are most interested in.

  5. Just some random questions:

    1. What tornadoes of the 2011 Super Outbreak do you think likely achieved EF5 intensity? I have a few potential candidates as well.

    2. Do Southern New England, the Ozarks and the St. Louis Metropolitan Area have their own little “Tornado Alleys”? Or is that just me?

    • Off the top of my head, I think you could make a case for Raleigh-Uniontown, Argo-Shoal Creek-Ohatchee, Cordova, and Ringgold all having attained EF5 intensity for at least short periods. I haven’t seen any conclusive evidence so I think it’s probably appropriate to go EF4 for each of them, but I think it’s likely that at least a few of them reached EF5 at some point.

      The example that stands out the most is the damage along Cherokee Valley Road in Ringgold. A number of homes were swept completely away and scattered, at least one of which appeared well-built. The trees in the area were twisted and snapped near ground level as well. They went conservative with EF4, but I personally think it deserved EF5. The others completely leveled homes as well, it’s just a question of how well-built they were. Some of them also threw cars very long distances.

      As for the mini tornado alleys, yeah, sort of. I don’t think I’d consider them “Tornado Alleys,” but there are definitely local minima and maxima. There’s some thought that these clusters are influenced in part by geography, which makes sense. There have been a few studies done on that, and I believe some presentations at AMS conferences.

      Parts of southern New England do have locally higher tornado frequency, though it’s harder to draw a conclusion in that region because there’s a smaller sample size. The St. Louis Metro certainly looks like something of a hot spot, though it’s difficult to say for sure because of the variation in population density. Some areas appear to be hot spots because they have higher population (and thus more likely to detect tornadoes) than surrounding areas. And Arkansas actually has a very high percentage of strong-violent tornadoes as compared to most other states.

      • “Arkansas actually has a very high percentage of strong-violent tornadoes as compared to most other states.”
        Curiously, though, it’s the only state east of the Rockies and west of the Appalachians that has never (to date) been hit by an F-5 or EF-5. Wikipedia’s list of unofficial F-5 tornadoes prior to 1950 and disputed/possible F-5’s since 1950 also shows none in Arkansas.

      • One of many oddities in the tornado record. Arkansas, for the most part, is a pretty sparsely populated state, and it’s got a lot of heavily forested and mountainous areas. That makes it less likely that an appropriately well-built structure will happen to be in the path of a tornado, while (E)F4-level damage indicators may be a bit more common. It’s still a bit strange though. It’s also a largely poor state and, while I don’t know for sure, I think it’s likely that many homes aren’t built well enough to serve as EF5 DIs anyway.

        In any event, there have been a couple tornadoes that could very well have been rated F5. The Sneed tornado of April 10th, 1929 was almost certainly an F5. There’s no way to know the construction of the structures it encountered, but many of them were so thoroughly demolished and granulated that much of the debris looked like kindling and wood chips. The Clinton tornado from the Super Tuesday outbreak also comes to mind. I haven’t seen any damage that clearly suggests it was an EF5, but it was on the ground for 122 miles and I think it’s pretty reasonable to assume it attained EF5 intensity at some point along its path and simply didn’t encounter any EF5-level DIs.

        Ultimately I think it’s more a product of the limitations and imperfections of our rating system than anything else. I think things would look quite different if we had some way to accurately gauge the intensity of all tornadoes without relying on damage surveying. Well, I can dream.

  6. “Given the slave population at the time, it’s likely the death toll is much larger than the officially accepted number”

    There would not have been a “slave population” in the South in 1884, since slavery had been abolished in 1865. That said, undercounting or not counting tornado fatalities among African-Americans continued to be a common practice in the South at least into the 1930s. I’ve read that the death tolls from the back-to-back Tupelo, Miss. and Gainesville, Ga. tornadoes of 1936 (both of which killed over 200 people) may be double the number reported since “colored” residents were not included in casualty lists.

    • Whoops, thanks for catching that Elaine. I’ll fix it. Unfortunately you’re right, and I’d say that many tornadoes actually had substantially higher death tolls than were reported. Even in the fairly rare cases where African-Americans were counted, oftentimes no one bothered to follow up with those who were critically injured.

  7. I would say this was a very major outbreak but I am intrigued by how back then as is the case sometimes now as well, the tornadoes even though there were very many of them, seemed to skip or avoid MAJOR CITIES on that day or during that outbreak. ie. Atlanta, Augusta, Charlotte, Chattanooga, Louisville, even tho the activity was all around them

    • You’d think it would happen more often, but thankfully it doesn’t. It’s mostly a product of the fact that tornadoes, even especially large ones, are a pretty small-scale phenomenon, and even large cities only occupy a very small chunk of a state’s land area. Take Nashville, for example. It’s one of the larger cities in the United States in terms of land area, but it still only makes up just slightly over 1% of the total land area of Tennessee. Even the 4/27/11 outbreak, which probably left more tornado damage in one state than any other on record, only damaged just over 1% of Alabama’s total land area. That’s an incredible amount of damage, but it’s still tiny compared to the size of the state itself.

      I guess that’s a long way of saying, although it’s hard to believe given the sheer number of tornadoes in large outbreaks like this, the probability of any given tornado striking a large metro area is extremely small. With that said, the Leeds tornado was probably only ten miles or so from downtown Birmingham. I think it’s even more incredible to think that the death toll may have been so high despite not striking any large cities.

  8. Pingback: Turning a Corner - Perfect Spring Day Today - Sticky 80s by Next Week | Paul Douglas Weather ServicesPaul Douglas Weather Services

  9. Pingback: Climate and Agriculture in the Southeast | February 19, 1884 — The Enigma Outbreak

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s