It was perhaps the most destructive event in Hawai’i in memory. And it was raised again in the blog comments recently. It turns out we have published a post on it in the past, written by Hector: https://www.volcanocafe.org/mauna-loa-from-1852-to-1868-part-2-the-1868-eruption-and-disaster/ Still, I had some bits together for a post on it so wrote it up anyway. Not surprisingly, it has quite a bit of overlap with Hector’s and I recommend you read them both.

The eruption began in the morning of 27 March, with a cloud rising above the summit of Mauna Loa. 12 hours later, earthquakes began and tremor was felt. So far, perhaps not that unusual. Eruptions on Mauna Loa often start at the summit but quickly move to the rift zone, so magma dikes commonly come after the onset of eruption. But the earthquakes grew stronger and more frequent. The next day, there was a massive shock, measuring magnitude 7.1 or thereabouts (the magnitude scale was not yet invented) which caused damage across the Ka’u area of Hawai’i. Over the next few days, the earthquakes continued, but less strong: the strongest was a magnitude weaker, at M6. By now, the earthquakes appeared to have been aftershocks of the strong event: the largest aftershock is typically one magnitude than the main one. It is as if the earthquakes changed from being magmatic to being tectonic, the reverse of the more common case!

On 2 April 1868, events came to a head. At 4pm, the strongest earthquake yet known from Hawai’i hit, with a magnitude estimated at 7.9. The damage was widespread across the island, with landslides. Destruction at Ka’u, near the epicentre, was complete.

Five days later, the eruption resumed with a vengeance. As typical for Mauna Loa, this was now from the rift zone. It came far down the southwestern rift, close to the southernmost reach of highway 11. The fountains were as high as on the recent Kilauea epsidoes, but the lava did not stay confined in a caldera. It flowed down the steep slope in several streams, and within hours reached the coast.

USGS wrote a recap of the story in March 2018, for the 150-year anniversary. Within months, there would be a rerun with an M6.9 earthquake in the same region, followed by a larger eruption, albeit this time from Kilauea instead of Mauna Loa. Volcanoes don’t do reruns, but they do do sequels. So did USGS, with another post in 2025.

It is difficult to add anything new to this series of excellent posts. But it is still a very interesting event, one that one can only hope will not repeat itself.

Earthquake

Earthquakes were first felt in the morning of 27 March, a few hours after the onset of the summit eruption. One report from Kona mentions these. The earthquakes grew stronger overnight.

The first major event was on 28 March, in the early afternoon. It caused considerable damage, including the collapse of the walls of several stone churches.

As an aside, these were so-called ‘apana churches, local congregations of native Hawai’ians, largely independent of the main ‘settler’ churches as they were too remote for regular visit by missionaries. The buildings were often made with walls of lava stones, although wood was preferred in earthquake-prone regions. Thatch was used at first but became too open to arson – sometimes by white people: settler violence is nothing new! The native churches often had one entrance towards the ocean and one towards the mountain, honouring the produce of both. The native church communities declined in the late 19^th century, as the number of native Hawai’ians dwindled and they moved to work in agriculture elsewhere on the island.

Sara Lyman and her family in Hilo kept a diary of felt earthquakes that ran from 1833 to 1917. It notes the time of the earthquake and how strong it felt; typically, it contained 1-10 events per year. The following page comes from that diary (from USGS bulletin 2027)

Hilo, of course, was far from Mauna Loa and the diary would miss many quakes. Closer to the mountain, there are reports of earthquakes in the evening of 27 March, some 12 hours after the onset of the summit eruption. On 28 March things became more severe, and the diary’ author wrote

“The earthquakes commenced on Friday night (Mar. 27); some say we had thirty or more shocks. Saturday morning I awoke about daylight, and up to 1 P. M. I counted ninety-seven earthquakes; none were very severe, but their frequency frightened us. While at dinner we had a heavy shock crockery jingled and the house creaked like a ship in a storm. I left the table and went over to _ and while talking with them we had a fearful shock […] walls were falling down with a thundering noise, and the air was filled with dust, the earth still quivering. In less than twenty-four hours we had over two hundred shocks!” (from Hector’s post, https://www.volcanocafe.org/mauna-loa-from-1852-to-1868-part-2-the-1868-eruption-and-disaster/)

The earthquakes continued over the next days, making sleeping difficult. But the strongest shock by far came at 4pm on Thursday, 2 April. This one is known as the Great Kau Earthquake, an M7.9 event.

Frederick Lyman, a sheep farmer in the region, wrote “.. did not stop — shook East and West, North and South, round and round, and up and down—lessen, then increase in violence. It was impossible to stand; we had to sit on the ground, bracing with hands and feet to keep from rolling over. […] when the shake came, a storm of mud and a great bank of earth twenty feet high came flying all about, mixed with rocks, logs, and ferns.”
The latter referred to a mudflow that came down the flanks for Mauna Loa, one of many landslides on the island. This one buried a settlement and killed some 30 people.

Reverent Titus Coan later reported “The crust of the earth rose and sunk like the sea in a storm. The rending of rocks, the shattering of buildings, the crash of furniture… the fright of men and animals, made throughout the southern half of Hawaii such scenes of terror as had never been witnessed before. The streams ran mud, the earth was rent in thousands of places; and the very streets of Hilo cracked open.”

Minutes after the main shock, the ocean began “surging and boiling furiously.” A series of tsunami waves, more than 10 meters high, which “wiped out the villages along the coast, 75 people swallowed up in it.”

The damage was widespread. Even Hilo was affected: “I saw several fissures in the earth near Wahiawa River, of from eight inches to one foot in width, which were caused by the earthquake of April 2nd, and run in the direction of Mauna Loa.” 

This was written by William Hillebrand, who was in Hilo and visited the area of Kau a few weeks later and wrote a report in the Hawaiian Gazette, May 6, 1868. He reported about the Kau region:

“The locality in question [where the earthquake occurred] is that comprised between the ranch station of Messrs Reed & Richardson, on the east, and Mr. F. Lyman, on the west, a distance of five miles. The government road connecting these two places runs through a fine grassy plain, which has a very gentle fall towards the sea, its elevation being about 2,000 feet. [..] The ground around Reed & Richardson’s station is torn up into numerous small cracks and fissures, running in every direction. Some are large enough to engulf, horse and rider, a fact which actually occurred a few days after the earthquake. A large cistern, built in solid masonry and covered with an arched stone roof, was rent to pieces, and the roof entirely broken away. Not a single stone fence is standing; their places are indicated by flat belts of stone on the ground. The dwelling house—a good wooden framed one—exhibits a wrench across its roof, so that the gutters empty themselves in the sitting room; the cookhouse is thrown off its foundation; other out-buildings are completely overturned; and of the grass houses some are smashed down, others, greatly inclined. [..] we could survey the extent of the land-slides on the opposite side of the hill, which were considerable. From this place, our guide pointed out to us a human figure in the distance, moving slowly over the dreary field. It was a husband searching for the body of his wife.”

“All the people from near the beach had taken refuge on higher lands near the upper road. My professional services were called for by many people who had been injured by the great oceanic earthquake waves. The great wave rose to a height of 25 feet, and according to reliable information, portions of the coast-line have subsided considerable. In some places cocoanut trees formerly out of water are now a foot deep in the sea. Every village along the coast of Kau and part of Puna has been swept away. The whole population of Waiohinu I found encamped on a high hill to the east among the ferns.” 

The event had been frightful, and the aftershocks seemed to have been continuous:

“From that Thursday to Sunday the earth constantly rocked and swayed, the hills seemed to alternately approach and recede. Most people became seasick. Strange roaring and surging noises were heard under the ground. When the ear was applied to the earth it would often receive a distinct impression as if a subterranean wave struck against the earth’s crust. The prevailing direction of the earthquake waves was said to have been from N. E. to S. W.”

There are reports of a further large earthquake on April 4. Less is known about this, and it is not mentioned in the Sara Lyman diary. This is thought to be a magnitude-7 earthquake affecting the Kilauea region.

The number of felt earthquakes remained high for ten years, declining from 40 events in 1869 to 10 by 1878. These were aftershocks strong enough to be felt (without instruments) in distant Hilo.

Eruption

The summit eruption was seen (from a distance) around 6am on 27 March. It was also heard as a roaring sound: it must have been an impressive outburst. The eruption may only have lasted a few hours, as the sound stopped later. It is common for Mauna Loa’s summit eruptions to be short-lived, when the eruption begins to move to the rift zone. Earthquakes were felt from the evening onward. These earthquakes fit in with this pattern, with a dike moving down the southwest rift zone. The earthquakes were rather strongly felt: it must have been a substantial dike, perhaps pushing through reluctant rock.

Strangely, there was no clear eruption for another ten day, and although smoke was seen along the rift zone, lave was not reported. It seems that plumes went where lava didn’t. After two days, the plumes arrived at or around the future eruption site. This may show that the dike went deep, and was not close enough to the surface to break through – one may guess sea level or below.

On April 6, there appears to have been a small eruption, based on overnight fine ash reported from a ship. This was the ‘throat clearing’. The lava burst out on April 7, along three separate fissures along the southwest rift zone. The main surface rift, seen weeks later, pointed directly at Mauna Loa. Now the lava came down the slope fast, split into several branches over a flatter area, before flowing down towards the ocean, which it reached within 4 hours. The speed of the lava approached walking speed, and several people escaped only just in time. The lava fountains from the fissures are estimated at 200 meters tall. The eruption ended on April 11 or 12. The volume of the erupted lave is estimated 0.12 km3, but Hector has here argued for double that size, based in part on submarine flows.

The sequence of events suggests that the summit eruption was only a side show. The shallow magma reservoir erupted into the caldera, but the real action was several kilometers deeper where a deeper feeder reservoir had received fresh magma. Mauna Loa normally would transfer the pressure upward, into the shallow regions. But the mountain could not contain the stress, and during the 27^th, a dike began to move out from the deeper reservoir. (This is speculative but supported by the fact that the lava from this event was unusually primitive.) It is reminiscent of the onset of the 2014 Bardarbunga eruption, but that was in a much smaller mountain and next to a rifting zone which provided a low-stress route out. It was a similar type of event, but a different order of magnitude.

The plumes that were seen may have been mainly from heated water in the rock. The magma was too deep to itself come up, but the dike caused surface extension and a depression – this was the rift that was later seen to point straight at the summit. The dike reached the eruption site by 30 or 31 March, but no eruption followed. Again, this resembles Bardarbunga where the dike would stop for days, before breaking through a blockage and moving on. At this particular point, the blockage may have been from a change in the rock. The currently active southwest rift ends near this location: the 1868 eruption was about as low on the mountain as eruptions come, in these waning days of Mauna Loa when it is nearing the end of it shield eruptions. Beyond this point, the rocks are older and colder, from 100,000 years ago when the mountain was more active and still growing. And so the magma dike sat there, growing and pushing but stalled by the solid bar.

Within days, the big earthquake would happen, taking the pressure off – for a while. After this, the magma was happy where it was, but eventually the buoyancy and gas pressure overcame the weight above, the magma moved up and the eruption began.

Collapse

Hector in his earlier post on these events suggested that the Mauna Loa summit experience collapse. Last year, this sentiment was echoed by HVO.

The summit eruption was seen only from a distance so details are uncertain, but it seems typical for Mauna Loa. The lava normally flows into the caldera and provides some resurfacing. But documented visits to the summit were infrequent. The caldera was mapped in 1841, and in 1864 it was found to have changed little apart from some resurfacing. But a map from 1872 shows a new depression, not evident in the earlier mapping.

The comparison between the two visits is shown here, in a figure taken from HVO’s volcanowatch of March 2025. The new depression near the centre of the caldera is estimated at 1.5 km³ in volume. (It is not known how accurate the second map is, though.)

When did this depression form? The only real candidate is the 1868 eruption, causing a collapse in the caldera from magma withdrawal. The volume is similar to that of Kilauea in 2018, but happened in days rather than months. The series of earthquakes between March 28 and April 2 could be due to this collapse. They reached magnitude 6 (estimated), while Kilauea peaked at magnitude 5 and Bardarbunga in 2014 managed magnitude 5.5. In a collapse event, at first the surface sags without strong seismic signatures. Once the sagging becomes large, the edge of the zone fails and a stepwise collapse begins, often alternating between two sides. These were the magnitude 4-5 earthquakes from Kilauea in 2018 and Bardarbunga in 2014: Kilauea had 61 separate such events. At Mauna Loa, the hours of delay between summit eruption and seismicity may have been such a sagging phase. After that, the actual collapse began, shaking the Kau region.

This leaves one small problem. The lava volume is around 0.25 km³ (accepting Hector’s value), whilst the collapse volume is 1.25 km³. A cubic kilometer of magma is missing! Where is it?

The only realistic option is that it remained behind in the dike. But that is quite a lot! For a 30 km long dike, and assuming a height of the dike of 4 km, it needs a width of 8 meters. You can’t just hide 8 meters of extra width: the whole side of the mountain would have to move. But perhaps that is exactly what happened.

Kilauea

Let’s first do a few detours. Kilauea had been highly active in the months before the Mauna Loa events. That changed now. In a report by William Hillebrand:

Thursday, April 2d, at a few minutes past four, p. m., the big earthquake occured, which caused the ground around Kilauea to rock like a ship at sea. At that moment, there commenced fearful detonations in the crater, large quantities of lava were thrown up to a great height; portions of the wall tumbled in. This extraordinary commotion, accompanied with unearthly noise and ceaseless swaying of the ground, continued from that day till Sunday night, April 5th, but from the first, the fire began to recede. On Thursday night, it was already confined to the regular lakes; on Saturday night, it only remained in the great south lake, and on Sunday night there was none at all; Pele had left Kilauea. The noises now became weaker, and were separated by longer intervals. By Tuesday, quiet reigned in Kilauea. 

This was unusual for the time, as Hillebrand comments:  “Kilauea has been dry once before, after the great flow in 1840.”  Source: https://nupepa-hawaii.com/2018/05/03/eruption-150-years-ago-1868/

While Mauna Loa erupted, Kilauea ceased to erupt. However, it can also be viewed as both summit eruptions ending and magma moving into rift zones. Mauna Loa’s rift took ten days before it erupted. Kilauea was quicker with a fast eruption at Kilauea Iki. There were also reports of lava having been seen in the Kilauea southwest rift, but this seems unconfirmed. But this was on a much smaller scale than the eruption of Mauna Loa.

Changes

Back to Mauna Loa. Mauna Loa’s eruptions are fed from a shallow magma reservoir and a deeper recharge reservoir, with the shallow magma a bit more evolved. Eruptions begin with the more evolved magma (lower MgO fraction) before delving into the newer, hotter magma. The rift eruptions also commonly start by pushing out left-overs from previous eruptions, as was seen in the 2022 Mauna Loa eruption. Apart from that, the composition of the lavas is quite uniform but has varied over time.

Since the 1850’s, Mauna Loa’s lavas have become a bit more evolved and a bit cooler, by about 50^oC. The eruptions in both 1852 and 1868 were picritic, extremely ‘young’ with MgO as high as 13%. These magmas were exceptionally fresh. Since that time, the eruptions have carried a larger fraction of stored magma. This indicates that the supply rate of fresh magma is not as high, and it is sometimes taken as an indication that the 1868 earthquakes damaged the main conduit of the mountain. However, the magma supply rate may also just cycle, currently in a low phase. For now, activity at Mauna Loa remains below the historical average.

Failure

So we can understand the eruption sequence but have a problem with the missing magma. In fact the tsunami is also related to this problem. Can we piece together what actually happened?

The scientific papers, mainly from the USGS and from the last century, are all hidden behind paywalls and are not currently generally accessible. The main information therefore comes from popular USGS articles. The important pointers are the two large but distinct earthquakes several days apart, the tsunami which following the second – but not the first – event, the apparently delayed rift eruption, and the major swarm of earthquakes.

Earthquakes of magnitude 7 and above are not magmatic, and are also too larger to come from caldera collapse. Neither can a dike by itself cause them. Applying the usual relations, a magnitude-7 earthquake correspond to a rupture length of some 60 km and a width of the rupture zone of 10 km. Magnitude 7.9 comes with a 400-km rupture length and 20 km width. The length of an M7 rupture is actually similar to the length of the Mauna Loa southwest rift zone. But this is not a transform fault, and therefore the width of the earthquake zone would need to be interpreted differently. The length of an M7.9 rupture can not be accommodated anywhere on the island of Hawai’i – this was a different type of event.

Hawai’i is basically a pile of lava sitting on a sea floor. The weight of the pile has bend down the sea floor considerably, and so now the sea floor bends down towards the centre of the island.

In this depiction (courtesy HVO, of course), green is the mantle, the bricks represents the oceanic crust, brown the lava pile, and the top of the brick layer is the old sea floor. The depth is indicated on the left, with zero corresponding to sea level. The oceanic crust is moving north, at around 4 cm per year, towards a future oblivion. The mountain is carried with, but the hot spot is not. So effectively, the volcano should be moving south relative to the crust, in order to remain above the hot spot. It is a losing battle. This is the reason why Mauna Loa is uncomfortably close to the southern shore line.

Hawai’i is unstable. The mountainous pile tends to slide towards the depth of the southern ocean. The slide happens on the plane between the lava and the crust: the mountain slides across the sea floor, at a depth of around 10 km below sea level. It slides on an upward slope, but this is fine because it is being pushed by a far higher mountain behind.

The slide has to overcome the friction of the sea floor, at the so-called decollement layer. It now become very much like a transform fault, but one which is horizontal rather than vertical. The layer gets stuck while the top of the pile slides slowly. This puts stress on the pile, and eventually the decollement layer overcomes the friction, fails and begins to slide. Once it begins to move, the friction reduces and the movement accelerates. This is the earthquake. The behaviour is similar to that of a thrust earthquake.

There were three separate events. The first was the M7 earthquake, which came in response to the forming dike near the summit. This dike pushed the pile oceanward, and this started a failure, but the movement was arrested before the entire block failed. Still, an M7 is similar to the failure that accompanied the 2018 Kilauea eruption it was a substantial move which involved much of the Mauna Loa rift zone.

But it was not complete. The slide had been arrested but this increased the stress elsewhere. And the growing dike was pushing more. During this time, the earthquakes were a mixture of the dike moving downrift, and aftershocks of the M7. It lasted five days. Then, the arrest failed. The block south of the Mauna Loa rift failed in its entirety. The top had already moved quietly over the years and centuries, but the bottom had not. This caused buckling. The ground level had gone up above the blockage. Now that suddenly gave way.

So far, no different from the previous event but it extended much further towards -and into- the ocean. The surface went down underneath the sea. It pulled the water down with it. New water raced in to fill the gap. This set off the tsunami. This is also how we known this event went further, because the previous M7 had not set off tsunamis.

There was one more. But it was not only the Mauna Loa rift that gave way. The Kilauea rift block also moved. But this is bordered by a different rift! Did both fail simultaneously? Or are the report of an M7 near Kilauea two days later correct, and did this fail at that time? We don’t know. After an M7.9, an M7 is also within range of possibilities for normal aftershocks.

There is another possible scenario, namely that the early M7 was the southwest rift of Mauna Loa failing, taking that side of the block with it, and the M7.9 5 days later was the entire rest, in effect the entire south side of Kilauea. The block there extends further into the sea, and this can cause the large tsunami. An argument against this is that the earthquakes felt in Ka’u after this came from the northeast, and that is the direction of Mauna Loa, not Kilauea. But as HVO wrote, we “will probably never be able to completely reconstruct the events of 1868, and some controversy and difference of opinion about what happened will certainly continue“.

The south flank of Kilauea remained quiet for a century. On 27 November 1975, it broke again in a M 7.2 earthquake (or M7.7 moment scale), an event that also included a large tsunami. This was followed by the 4 May 2018, M 6.9 earthquake during the early phase of the Puna eruption. Both events overlapped with the 1868 Ka’u Earthquake in location but the epicentre was much further east. The 1975 event was over a fault length of 40 km, a width of up to 30 km and a slip of 4-5 meter.

It seems that large earthquake east of Kilauea are fairly frequent (albeit not as frequent as posts on the 1868 events), at decades to a century recurrence time. But failure at Mauna Loa is less frequent. We don’t know the recurrence time for those. But id it happens, there is a fair chance that it takes part of Kilauea with it.

There are two more points to make. First, the movement does not always occur in large earthquakes. There are also ‘slow slip’ events where a large number of micro-slips add up to a big one over the course of a few days. This happened in 1924, when the east rift of Kilauea moved by 10 meters, even allowing the sea to move in-land by a kilometer.

Second, the 2018 eruptions had many similarities with 1868. It started with a collapse event, albeit at Pu’o’o rather than Kilauea, which pushed magma down the east rift zone and cause an eruption. This triggered the M6.9 earthquake, and this opened up the floodgates from Kilauea, leading to flood of fresh magma coming down the rift, invigorating the eruption and causing a second collapsed, now at Kilauea. You couldn’t make it up. But perhaps it shows what else may have happened in 1868.

Albert, June 2026