The Mississippi valley had once been densely inhabited, but it was so no longer. The Indian cultures had been decimated by disease; their lands were emptied and their villages deserted. The area had just become US property, and American settlers were moving in but it was still early days, here on the western frontier, in 1811. To the north, St Louis was home to some 1200 people; to the south, Memphis did not yet exist: its area remained part of the Chickasaw nation, themselves fairly recent (15th century) migrants. Yet the river was already heavily trafficked and boats were frequent; the age of the steamboat was only days away. Turning one of the many bends of the river, the boats would pass the town of New Madrid. It consisted of some 400 log cabins and two brick buildings, and was similar in size to St Louis. The name recalled a checkered origin; the town had been settled during the Spanish period in 1778, the Spanish traded it to the French in 1800 (in secret), who sold it on to the US in 1803 as part of the Louisiana purchase, in a fairly broad definition of Louisiana which extended from the Gulf to beyond the Dakotas, even covering a small part of what is now Canada. The trade included the land with all its beauty and faults, as well as the population, Indian, white migrants, and black people (some slaves, some free) alike. The town would finally end up in Missouri. New Madrid still exists, and is now a town of 8000 people, having been far outgrown by both St Louis and Memphis. But it is not old New Madrid: that lies lost underneath the river. The current town dates from 1812, a new New Madrid, risen from the ruins of a disastrous year.
The earthquakes began on Monday, 16 Dec, 1811, 2:15 am, on a dark and moonless – but not stormy – night. The first shock lasted one to two minutes. People on boats on the river said the shock lasted about 40 seconds – people on land felt it for longer. A second, less severe shock followed 30 minutes later, one of a very large number of aftershocks. Another strong quake happened around 6:30, and around 7:15 there was a very strong quake which in places may even have exceeded the first; in one report the last event was violent but lasted only about 20 seconds and was followed by a large spout of water rising from the river to a height of 6 meters. At this time the land was seen to move in foot-high waves, coming from the west.
New Madrid was left badly damaged and many houses had lost their roofs and chimneys. But the town was not destroyed. Further south, the damage was worse. The day after the earthquake, on Dec 17, the very first steam boat arrived on its inaugural journey. Seeing the devastation, the New Orleans did not stop but continued its journey south. Many accused this new-fangled and dubious invention of having caused the earthquake!
The numerous aftershocks slowly decreased over the weeks. But another strong earthquake hit on 23 Jan 1812. Two riverside villages were flooded by the Mississippi: Little Prairie and Point Pleasant – luckily the first town had been abandoned in December, its people having fled to New Madrid finally arriving on Christmas Eve, leaving a single infirm (and black) person behind, and the population of the second town, about 100, managed to escape on foot, in the dark. Again the aftershocks came and declined. A lesser earthquake came on 4 February, and the final destructive one, possibly the strongest, happened on 7 February, at 3 am, lasting 90 seconds. This time the epicentre was close to New Madrid, and this did destroy the town. The seven weeks of shocks left widespread devastation. The quakes were felt over much of the young nation, and even shook the White House. Casualties were probably light, due to the sparse population. Most people who died had been on the river. Casualties among the Indian villagers may have been higher but these were not recorded.
The Mississippi plains had now changed beyond recognition, so much so that captains lost their bearings. Many islands in the river had sunk, disappeared, or changed in outline. On one ex-island only the top of the trees were above the river. Land slides had eroded the banks of the river. Numerous little mounds of sand dotted the previously featureless plains. Trees had snapped or toppled and numerous floating trees obstructed the river. Large areas of land had turned to water, a result of liquefaction, which downstream also greatly increased the flow rate of the Mississippi. In other places the land had been thrown upward. The February quake dammed the river and famously caused the Mississippi to flow backwards for a few hours. New rapids made the river almost impassable for a while, until erosion had removed the new obstacles of clay. New lakes formed – and some of these still exist.
News of the disaster slowly made it to the populated areas, where it was viewed as sensational but unimportant. The headlines, like yesterday’s news, were quickly forgotten. This was after all the frontier, where strange things happened but where few people went – and fewer came back. The New Madrid earthquakes became a legend rather than actual history; and like legends, the telling grew in exaggeration and decreased in believability. That has continued until today, even now that science has rediscovered the event. Popular writings try to outdo each other in describing what happened. The size became an 8+, than an 8.5, in one book even an 8.8. It could have split the US. Who cared? It wasn’t real history anyway. The middle ground between denial and doom was lost, the kernel of truth overgrown like badly written software. Let’s see whether we can dig out the hidden truth.
Eyewitness and local reports
Many reports were published in newspapers and magazines, over the days, weeks and years following the events. Some were detailed, some provided overwhelming impressions but less factual information, and some reported sensational second-hand stories. A few of the reports give wrong dates; the accuracy is not always beyond doubt.
Godfrey Lesieur is one of the better examples. He was in New Madrid on that December day but only wrote down his recollections 50 years later. Although his recollected dates were not fully correct, details of his report agree well with others:
The first shock of the earthquakes was on 2am o’clock on Dec 17, 1811, and was very hard, shaking down log houses, chimneys, etc. It was followed at short intervals, half to one hour apart, by comparatively slight shocks until about seven o’clock in the morning, when a rumbling noise was heard apparently in the west resembling and not unlike distant thunder. In an instant the earth began to shake and toter to such a degree that no persons were able to stand or walk. This lasted perhaps for one minute, and at this juncture the earth was observed to be as it were rolling in waves of a few feet in height, with visible depressions in between. By and by these waves or swells were seen to burst, throwing up large volumes of water, sand and a
species of charcoal, some of which were covered in part with a substance of which, by its peculiar odor was thought to be sulphur. When these swells bursted wide and long fissures were left running north and south parallel with each other for miles.
[…] A very large extent of country on either side of Little River, also on both sides of the St. Francois river, in this state and Arkansas, and on the Reel Foot Bayou in Tennessee was sunk below its former elevation about ten feet, thus rendering that section of country entirely unfit for cultivation. It is a remarkable fact and worthy of notice that so few casualties occurred, among the citizens there were but two deaths, both victims being women: Mrs. La Font died from fright and Mrs. Jarvis received injury from the fall of a cabin log from which she died.
[…] In many conversations with the old men of several tribes, Shawness, Delawares, and Cherokees, all said that they had no traditionary account that earthquakes had ever visited the country before.
[…] The water thrown up during the eruption of the “land waves” was luke warm, so warm indeed as to produce no chilly sensation while wading and swimming through it.
Mathias Speed was on the river during the final, February 7 quake. He wrote the following for the Bardstown Repository (1812):
About 3 o’clock, on the morning of the 7th, we were waked by the violent agitation of the boat, attended with a noise more tremendous and terrific than I can describe or any one can conceive, who was not present or near to such a scene. The constant discharge of heavy cannon might give some idea of the noise for loudness, but this was infinitely more terrible, an account of its appearing to be subterraneous.
As soon as we waked we discovered that the bar to which we were tied was sinking, we cut loose and moved our boats for the middle of the river. After getting out so far as to be out of danger from the trees which were falling in from the bank – the swells in the river was so great as to threaten the sinking of the boat every moment.
We stopped the outholes with blankets to keep out the water – after remaining in this situation for some time, we perceived a light in the shore which we had left – (we having a lighted candle in a lantern on our boat,) were hailed and advised to land, which we attempted to do, but could not effect it, finding the banks and trees still falling in.
[…] We landed at New Madrid about breakfast time without having experienced any injury – The appearance of the town, and the situation of the inhabitants, were such as to afford but little relief to our minds. The former elevation of the bank on which the town stood was estimated by the inhabitants at about 25 feet above common water; when we reached it the elevation was only about 12 or 13 feet –
There was scarcely a house left entire – some wholly prostrated, others unroofed and not a chimney standing – the people all having deserted their habitations, were in camps and tents back of the town, and their little watercrafts, such as skiffs, boats and canoes, handed out of the water to their camps, that they might be ready in case the country should sink.
I remained at New Madrid from the 7th till the 12th, during which time I think shocks of earthquakes were experienced every 15 or 20 minutes- those shocks were all attended with a rumbling noise, resembling distant thunder from the southwest, varying in report according to the force of the shock. When I left the place, the surface of the earth was very little, if any, above the tops of the boats in the river.
There was one boat coming down on the same morning I landed; when they came in sight of the falls, the crew were so frightened at the prospect, that they abandoned their boat and made for the island in their canoe- two were left on the island, and two made for the west bank in the canoe – about the time of their landing, they saw that the island was violently convulsed – one of the men on the island threw himself into the river to save himself by swimming – one of the men from the shore met him with the canoe and saved him. –
This man gave such an account of the convulsion of the island, that neither of the three dared to venture back for the remaining man. The three men reached New Madrid by land.
The man remained on the Island from Friday morning until Sunday evening, when he was taken off by a canoe sent from a boat coming down. I was several days in company with this man – he stated that during his stay in the island, there were frequent eruptions, in which sand and stone, coal and water were thrown up.
The violent agitation of the ground was such at one time as induced him to hold to a tree to support himself; the earth gave way at the place, and he with the tree sunk down, and he got wounded in the fall. – The fissure was so deep as to put it out of his power to get out at that place – he made his way along the fissure until a sloping slide offered him an opportunity of crawling out.
He states that frequent lights appeared – that in one instance, after one of the explosions near where he stood, he approached the hole from which the coal and land had been thrown up, which was now filled with water, and on putting his hand into it he found it was warm.
William Pierce, also traveling on the river, provided a good description of the size of the area affected by the 16 December quake (although it would be easier if more of the names he described still existed on the maps!)
At New Madrid, 70 miles from the influence of the Ohio, and on the right hand, the utmost consternation prevailed among the inhabitants; confusion, terror and uproar presided; those in the town were running for refuge to the country, whilst those in the country fled with like purpose towards the town. I am happy, however, to observe, that no material injury has been sustained.
At the Little Prairie, 103 miles from the same point, the shocks appear to have been more violent, and were attended with severe apprehensions. The town was deserted by its inhabitants, and not a single person was left but an old negro man, probably too infirm to fly: everyone appeared to consider the woods and hills most safe, and in these confidence was reposed. Distressing, however, as are the outlines of such a picture, the latest accounts are not calculated to increase apprehensions. Several chimnies were destroyed, and much land sunk, no lives however have been lost.
A little below Bayou River, 103 miles from the same point, and 130 miles from the spot where we lay, the ruin begins extensive and general.
At Long Reach, 146 miles, there is one continued forest of roots and trees, which have been ejected from the bed of the river.
At the near Flour Island, 174 miles, the destruction has been very great, and the impediments in the river much increased.
At the Devil’s Race ground, 193 miles, an immense number of very large trees have been thrown up, and the river is nearly impassible. The Devil’s Elbow, 214 miles, is in the same predicament; below this the ruin is much less, and indeed no material traces of the earthquake are discoverable.
Some other reports need to be read with some degree of skepticism. A report from North Carolina mentions a second-hand story of tremendous noises in the mountains hours before the big shock, accompanied by flashes of fire issuing from the side of the mountains, ending with a loud snap or crackle. The credibility of this story is not clear. Reports that church bells rang in Charlestown, Boston, are not supported by newspapers published in Boston itself. They may be confused with Charleston, South Carolina – much closer to the epicentre, and in an area where residents reported feeling like they were adrift on rough oceans during the quakes.
There are no good reports on how the earthquake affected the Indian population. The Shawnee chief Tecumseh had just failed to unify the various tribes against the settlers, and in some ways he was seen as a prophet of the destruction. Did it change their culture, their philosophy, their politics? That wasn’t recorded – it wasn’t ‘American’ history.
The many reports allow the story to be pieced together. There were three main episodes of earthquakes, shown in the following table. Each consisted of one or more major events, with a large number of aftershocks.
The most detailed information came from people who were on the river. The Mississippi was already a major transport route; in contrast, the local population was sparse and lived in isolated settlements. The lack of detailed reports on the January quakes is related to this. At that time, the upper Mississippi was fully frozen and all river transport had ceased. The lack of detail does not mean this was a lesser event, but it did affect fewer people. The December quakes led to many landslides along the river, and toppled many trees. In addition, there were a number of islands within the river and these were badly affected, in some cases fully submerging. But only in the February earthquake did new rapids form. This final earthquake had a dip-slip component which pushed up the land, and crossed the river in three places. The fault causing the December quake either was pure strike-slip, or did not cross the river.
A significant point is the widespread liquefaction. Water came up everywhere, in places spouting from the ground, deep enough to swim in. Often it also brought up sand, forming new sand hills. The sand blows (also called sand volcanoes) and sand dikes could be hundreds of meters long – one covered the abandoned town of Little Prairie to a depth of 2 feet. The shape of the land had changed dramatically after the earthquakes. The cartoon below, from USGS, shows how these sand blows worked.
Liquefaction is a common hazard of earthquakes. It happens when the soil is loose and granular, saturated with water, and subjected to strong shaking. The shaking separates the granules from each other, and this causes the saturated soil to loose its strength and expel some of its integral water content. Suddenly the soil flows like a liquid. It can happen during or immediately after the earthquake, and may be seen hundreds of kilometer from the epicentre. Conditions in the Mississippi valley, with its sediments and high water table, were ideal for this. A famous example of liquefaction (not earthquake related) is the Norwegian disaster of 1978: see this program. It gives a good idea how devastating and dangerous it can be.
Why was the water luke warm? There was no obvious source of heat – no baby volcanics here. But remember that it was winter, and the temperature was around 5 C. The Mississippi would have been rather chilly. The water that came up from the ground wasn’t ground water. It had been an integral part of the soil and clay, and was released at depth by the shaking. It would have had the same temperature as the soil. At a depth of a meter, the soil approaches the average temperature over the year, which in this area is 17 C. As long as the water came from soil deeper than this, it would have been at this temperature: it would feel comfortable, but not hot, to people accustomed to winter temperatures. There are some report of much hotter water, in one report at 142 F, but that was in North Carolina, far from the epicentre, where liquefaction did not occur. This was water released from local hot springs. The report of ‘warm’ water after the February earthquake can be understood by the fact that the Mississippi had recently been frozen, upstream. It would have been very cold water, and in contrast water at 17C would have felt shockingly warm!
A number of the reports mention the haze (or darkness) and sulphurous smell. The haze was no doubt due to dust shaken into the air. The cause of the smell is less obvious. In the old sand boils, driven to surface from meters deep, you can find small nodules, akin to hardened asphalt. Scratch them and they smell like petrol. Organic matter, buried below the sediment in anoxic decay can smell pretty bad! The lights seen by the man trapped on the island sound like Will-o’-the-wisps, also called Jack-o’-lantern – swamp lights, related to escaping methane and other volatile hydrocarbons.
The area worst affected in the December quakes ran from New Madrid to a place called Flour Island. The name Flour still exists on the map, and is about 40 km north of Memphis. South of Flour Island, there was little destruction; north, the river was almost impassable due to debris and other hazards (including captains getting lost as the river bed had changed so much!). This makes the total length (as the crow flies) of the quake zone roughly 80 kilometer.
The magnitudes of the earthquakes are very poorly known. Various papers have suggested values of low M8’s. But recently, estimates have come down, and mid M7’s is now more commonly quoted. Given the length of the fault system involved, even this may be high. The re-assignment of the church bells of Boston to Charleston has also drastically reduced the required magnitude. In 2010, Susan Hough argued that the shaking maps are consistent with about M7.0. This is a dramatic reduction, but it brings the earthquakes within range of other intraplate tectonics.
Using an approximate relation between the area of the rupture (length times depth, in square kilometer) and the magnitude of an earthquake, M = 4.07+0.98*log(Area), and taking the length as 100 km (an overestimate) and depth as 10 km (perhaps an underestimate), the earthquake size comes out as M = 7.0 . It is difficult to get much more out of the rupture zone. This provides another argument against the very high older estimates in excess of M8: the New Madrid events have most likely been around M7.
Many web sites state that the New Madrid earthquakes were among the largest earthquakes to hit the US in recorded history. This is overstating the case. There were large and destructive, but there is no evidence they were as huge as that.
An American fault
The local people all confirmed that there had been no earthquakes felt before these events. People from the local tribes also had no traditions of earthquakes in the area. It seemed a tectonic dead water. The Mississippi brought enough action for anyone, in its meandering and constant threat of changing its channel. But in fact the Mississippi has a hidden history. It hijacked a much older valley, which does have a tectonic origin.
For this, we need to go back a long way. Before the Atlantic Ocean, before Pangea, there was the Iapetus Ocean. Its closure formed the Appalachian mountains. But the Iapetus also had a beginning. Before it formed, Europe and America were linked in another supercontinent. The inevitable splitting involved rifting, and one of those rifts started but never finished. It became a failed rift. One or two arms of the triple junction started an ocean, but this arm did not. Magma intruded and solidified (but did not make it to the surface). Much, much later, marine and river sediment filled in much of the depression. This is the Mississippi embayment; New Madrid is in its northern part. The depression is still visible, even if deeply filled in: it is now called the Reelfoot rift, a wide valley which the Mississippi has claimed for its own floodplain.
Recently (geologically speaking), the area has re-activated: new faults have developed. The figure, taken from USGS, plots moderate earthquakes (M>2.5) since 2001. These clearly delineate a fault. In fact, three fault segments are visible, one from Blytheville to Tiptonville (called Cottonwood Grove), a perpendicular one from Dyersburg running north-northwest (the west-dipping Reelfoot thrust), and a less clear segment from New Madrid northeast-ward. The first and last segments may trace the same fault, but with an offset; the second, perpendicular segment appears to be the fault doing the offsetting. This is a complex region! The complexity, the short length, and the limited offset all suggest this fault system is still young. It seems two or all three segments ruptured separately in 1811/12, giving rise to the earthquake sequence. The perpendicular fault is visible on the surface as the Reelfoot scarp, a step-change of as much as 8 meter, which dates in part from the 1811/12 events.
Recent GPS measurements confirm the faulty nature. They show right-lateral slip of 5 mm per year. That would allow for 5 meter of slip every 1000 year, and this should be roughly the recurrence time. Is there any evidence for past earthquakes here? In fact, there is, but you have to look hard. The most obvious remnant of the 1811/12 earthquakes were the numerous sand hills. But some of the sandy deposits have Indian artefacts above them. As the Indians had largely left the area by 1811, these artefacts must predate the earthquake, and therefore those sand deposits must also be older than 1811. It seems there must have been very similar earthquakes here before. The latest one before 1811 has been dated to around 1450 A.D., and earlier ones to 900 A.D., 300 AD., and 2350 BC, give or take a century or so. All of these were outside of the Indian oral tradition. For a recurrence time of 500 year, the slip of the 1811/12 quakes would have been around 2.5 meter.
The total offset between the two fault fragments at New Madrid is about 20 km. Presumably these segments where originally one fault, which was split in two; the offset gives an age of 4 million year. For comparison, along the San Andreas fault there has been over 100 km of cumulative slip. New Madrid indeed is a young tear in a much older land.
The image, kindly created by Geolurking, shows some of the features of the region. The earthquake was one of the usual ones near New Madrid, a recent aftershock of 1811. The fault lines are from USGS shapefiles, and the pluton shapes are from several different papers (green hatched shapes). The light blue dashed line is the extent of the Mississippi embayment. The red concentric circle in Tennessee is Wells Creek crater. The three faults which combined to destroy New Madrid can easily be recognized.
There were four main shocks in the sequence, and incessant aftershocks.
The first shock, at 2:15am Dec 16, was a major strike-slip on the Cottonwood Grove, with a displacement of perhaps 3-4 meter. During the morning quake on that day, 5 hours later, the land waves came to New Madrid from the west as stated by Lesieur. One model has it as a small slip on the Reelfoot fault but this seems inconsistent with the direction of the surface waves. However, there is also a small extension of the Reelfoot fault running to the west from its end point, towards Malden, and this might be consistent with the report from Lesieur. The 7 February earthquake was the one that caused the rapids to form in the river. Therefore, this was the major thrust, on the Reelfoot fault. The aftershocks from this event came to New Madrid from the southwest.The three quakes occurred on three adjacent faults, curving round poor New Madrid.
The question of Illinois
This leaves the enigmatic 23 January event. Models have difficulty reproducing it: there isn’t an obvious area which could have caused it. One suggestion is that this event wasn’t near New Madrid at all. There is another potential area of activity 200 kilometers away, in Illinois, where small earthquakes are frequent. A local land owner, Yearby Land, experienced the 1811/1812 quakes as a boy of 9. Many decades later his stories were passed on by his nephew. Land told how the ground would shake and then rock and roll in long waves. After a short quiet spell, there would be another shock and roll. In these long continued rollings, the tall trees would weave their tops together, interlock their branches, then part and fly back the other way, and when they did this the blossom ends of the limbs would pop like whip lashes; and the ground was covered with broken stuff. About two miles east of his father’s house, a big crack was made in the ground, and you could not see to the bottom of it. The ground on its south side had sunk down about two feet. By 1858 this slope (at 38.07° N, 88.11° W) could still be traced over a length of two miles. The slope is reminiscent of a fault, and has led to suggestions that this was the epicentre of the 23 Jan 1812 earthquake. In support of this, there is evidence on the ground for liquefaction and sand volcanoes there too. Land said that piles and piles of pure, snow white sand were heaved up from the size of a bee-gum too three or four wagon loads.
Image: The small star shows a possible epicentre of the 23 Jan 1812 event; the large star shows the location of the 1968 Southern Illinois earthquake. Black circles correspond to locations where liquefaction was documented during the 1811–1812 sequence. The easternmost and largest circle identifies the location of the White County, Illinois, event, the account of which describes substantial liquefaction as well as surface cracking.
An analysis of the shaking reports of this quake finds some support for this location, as shown in the image. They suggests that the 23rd Jan earthquake, which they estimated at M6.8, may have occurred on an east-west oriented fault, triggered by stress from the first Mew Madrid earthquake. This can not be considered conclusive. We do not know whether the crack traced a fault or a land slip, and sand boils can occur some distance away from the epicentre. This is a known area of earthquake activity, but the quakes are quite wide distributed and at low level compared to New Madrid. Also, the lack of report from further south may just be from the lack of river traffic rather than a deficit in shaking. More research is definitely needed.
Here and now
New Madrid is a damaged area in a big land. What are the implications? Almost depopulated in 1811, now as many as 4 million people live within the New Madrid Seismic Zone. How safe are they?
The recurrence times of the earthquakes are such that a major event within the next century on the New Madrid fault is less likely. A smaller earthquake could still be damaging, though. There were strong aftershocks in the 1850’s and 1890’s, and there are still numerous small tremors, typically a few each week. The chances of an earthquake up to M6 over the next 50 years are considerable. Such an earthquake could do significant damage. The I-55 connecting Memphis and St Louis passes through the main liquefaction area and could quickly be destroyed. Three major pipelines pass through the region. And the earthquake resilience of buildings in Memphis has never been tested for real! One day it may be.
But there is an even deeper issue. What did re-activate the fault, and is this the only area that has been thus affected? Or are there similar re-activated faults nearby that we don’t know about just because they have not failed in living memory? As to the first question, it has been suggested that stress building up in the continent (there is always stress in the US) can re-activate an ancient line of weakness, and do so only over a small region. Once it gives, the stress may than be transferred to another fault fragment. For the second question, soundings of the Mississippi basin over the past decade have revealed that there are such old faults also further south, which were active perhaps 100 million year ago. But in places, the ruptures from these faults extend into the top sediment, suggesting events within the last 10,000 year. These are the unknown faults which could do a new New Madrid.
Both New Madrid and Illinois are regions of significant on-going micro-seismicity. This is interpreted as aftershocks from the 1811/12 earthquakes, rather than precursors to the next event. But remember there were no felt earthquakes for many years prior to the 1811 shock. Before the event, the fault went silent. This is important: we need to hear the silence. The silent faults need to be mapped, especially in regions where stress may be building up. Risk management begins with recognizing the dangers.
Perhaps one day, another town with no tectonic history will give its name to an unforeseen event, moving from a small locality to historic significance. Thus is the danger of silence.
The Louisiana purchase transferred a land with hidden faults. Perhaps the US should have paid for the extended warranty.
Arch Johnston and Eugene Schweig, 1996: The enigma of the New Madrid earthquakes of 1811–1812. Annual Review of Earth and Planetary Sciences Vol.24, pp.339-384
Karl Mueller, Susan Hough & Roger Bilham, 2004: Analysing the 1811–1812 New Madrid earthquakes with recent instrumentally recorded aftershocks. Nature, 429, 284-288 (paywalled)
Updated 11 March 2017, to include Geolurking plot of the Mississippi geology