Thera: The Eruption Two Sciences Cannot Agree On

The eruption ran in four main phases, documented by Heiken and McCoy in 1984 and by Druitt and colleagues in 1999. Phase 1 was Plinian: a column rising about 36 kilometers, laying pumice down across Thera and eastern Crete. Phase 2 was phreatomagmatic, water reaching the vent and driving violent explosions and base surges. Phase 3 was the ignimbrite phase, pyroclastic flows sweeping across the island and out into the sea, burying the town of Akrotiri under as much as 60 meters of pumice and ash. Phase 4 was collapse: the emptied magma chamber gave way beneath the island.

That fourth phase is the part still visible from the air. The centre of the island dropped into a caldera roughly 11 by 7.5 kilometers, now flooded by the sea. The island did not simply erupt and settle. It lost its middle.

Our own research file puts total ejecta at 30 to 80 cubic kilometers of bulk tephra, with a dense rock equivalent figure of about 40 cubic kilometers in its summary. Dense rock equivalent, or DRE, is the same material with the gas and pore space taken back out and stated as solid rock, so it is always the smaller of the two numbers. On its own figures the file puts the eruption at 2 to 5 times the size of Krakatoa in 1883 (about 25 cubic kilometers), and names Tambora in 1815 (about 160 cubic kilometers) as this eruption's only Holocene rival.

The explosivity class did not move with the volume. Sources still classify the eruption as VEI 6 to 7, most often citing VEI 7 flatly, and that has held through the whole revision history. There is a plausible reason for the hedge, and it is arithmetic rather than argument: the Volcanic Explosivity Index is conventionally keyed to bulk tephra volume, with VEI 7 requiring more than 100 cubic kilometers bulk and VEI 6 spanning 10 to 100, so a 34.5 cubic kilometer DRE figure converts to a bulk estimate sitting nearer that boundary than the older, larger DRE figures did. We offer that as an explanation of why the sources hedge, our own reading connecting two published facts, not as a finding any single paper states.

Akrotiri was excavated by Spyridon Marinatos from 1967 to 1974, a campaign that ended when Marinatos was killed in an accident on the site. Christos Doumas took over the excavation in 1975 and has directed it since.

What the deposits preserved is a prosperous Minoan town, more than 20,000 square meters of it uncovered and likely a good deal larger. Buildings stand up to three storeys. There is indoor plumbing, there are flushing toilets, and there is a sophisticated drainage system. In urban quality it is comparable to Knossos.

The wall paintings survive, and they are exceptional. Among them are the Spring Fresco, with its swallows and lilies; the Boxing Boys; the Saffron Gatherers; the Fisherman; and the Flotilla Fresco, a detailed naval procession with ships, dolphins and coastal towns in it. This is not a fragmentary record of Bronze Age Aegean life. It is a painted one.

The comparison everyone reaches for is Pompeii, and our library makes the point sharper than the nickname does: Akrotiri was buried roughly 1,600 years before Pompeii was, in 79 CE, and its preservation is comparable to Pompeii's and in some respects better. The Pompeii of the Bronze Age is an epithet that survives scrutiny.

One number should sit under everything said about Akrotiri. Only about 3 percent of the site has been excavated. The rest is still under the volcanic deposits, and every statement in this article about what Akrotiri contains is a statement about that fraction.

The major finds from the site are held by the National Archaeological Museum in Athens.

No human remains have been found at Akrotiri, and few portable valuables. The reading is that the population left before the main eruption, possibly warned by earthquakes running ahead of it.

The site itself carries the warning in its fabric. There is earthquake damage that had been partly repaired before the final eruption, and construction debris that was in the process of being cleared. Read together, that points to a run-up of weeks to months of tremors, which puts specific ground under the primary file's careful 'possibly warned by precursory earthquakes'.

People did die in this event, and in 2022 one of them was found. Sahoglu and colleagues, publishing in the Proceedings of the National Academy of Sciences, reported excavations at Cesme-Baglararasi on Turkey's Aegean coast that recovered an articulated human skeleton, a young man, together with the skeleton of a dog, inside tsunami debris tied stratigraphically to the Thera eruption. Coverage described him as the first known human victim directly associated with the Late Bronze Age Thera tsunami. The positioning of the body suggests survivors dug temporary graves that later waves then filled in.

So the precise version is this. In the roughly 3 percent of Akrotiri that has been dug, there are no human remains and few portable valuables, and that is best read narrowly: that one town got an organized, warned evacuation out ahead of the final phases. It is not evidence that the catastrophe killed nobody in the Aegean. The Turkish coast has produced a body in the tsunami debris of this same event. Akrotiri emptied. The wider disaster did not spare everyone.

Modeling by Pareschi and colleagues in 2006 and by Novikova and colleagues in 2011 gives wave heights of 9 to 15 meters on the northern coast of Crete.

Our file E_2_18 states the modeling results differently, and the difference is worth naming rather than smoothing over. Drawing on Novikova and colleagues in 2011 together with Nomikou and colleagues in 2016, it gives maximum wave heights of about 35 meters and above on Thera's own coast, roughly 5 to 15 meters on the northern coast of Crete, and measurable waves throughout the eastern Mediterranean. The Thera figure is a different coastline at a different distance from the source and does not conflict with anything. The Crete figure, 5 to 15 against 9 to 15, is our files E_1_16 and E_2_18 reporting the same modeling literature with different lower bounds, and this article does not adjudicate between them.

The deposits are the harder evidence. Layers of marine sediment, pumice and pottery debris identified as tsunami deposits have been found at multiple coastal sites on Crete, including Palaikastro, Pseira, and possibly Amnissos, the port of Knossos.

The reach is wider than Crete. Eastern Mediterranean tsunami deposits attributed to this eruption have also been reported on the Israeli coast and the western Turkish coast, including a destruction layer of marine debris lying over Late Bronze Age occupation at Cesme-Baglararasi, the same site that later produced the skeleton.

What that would have meant for a seafaring economy is not hard to state. Waves on that scale would have devastated the Minoan maritime economy, destroying harbors, warehouses and fishing fleets. What follows from that damage over the next generations is a separate and much more contested question, and it is handled at the end of this file rather than assumed here.

Modeling puts stratospheric sulfur dioxide release somewhere in the range of 10 to 200 teragrams. If the true figure is at the higher end of that range, hemispheric cooling of 1 to 3 degrees C for 1 to 3 years is plausible. That is a very wide range, our own file flags the magnitude as debated, and the primary file for this eruption carries no sulfur or cooling figure at all. This is an estimate with a factor of twenty in it, not a measurement.

One Egyptian text is repeatedly brought into this. The Tempest Stela of Ahmose I, dated to about 1550 BCE or earlier with its own date disputed, describes a great storm bringing darkness and destruction across Egypt. Karen Polinger Foster and Robert Ritner argued in 1996 that this may be an account of the eruption's atmospheric effects. The interpretation is contested: others read the stela as describing a local storm, or as metaphorical and propagandistic rather than meteorological.

A second text sits further out. The Bamboo Annals record unusual cold, frost and famine in China during the reign of King Chieh of the Xia dynasty, and this has been tentatively correlated with Thera's climatic effects. The correlation depends entirely on which dating scheme is used for the eruption, which makes it a poor witness in an argument about the dating and a fragile one on its own terms. Our own file's language is 'tentatively correlated', and this article will not upgrade it.

The high chronology's anchor is a single piece of wood. Friedrich, Kromer, Friedrich, Heinemeier, Pfeiffer and Talamo, publishing in Science in 2006, radiocarbon dated an olive branch found in place on Thera and killed by the eruption itself, and got 1627 to 1600 BCE at 95.4 percent confidence, recalibrated against the IntCal curves. A branch buried alive by the event dates the event, with no chain of intermediate assumptions between the sample and the eruption.

That date does not stand alone. It is consistent with frost rings in Irish bog oak at 1628 BCE, reported by Baillie and Munro in 1988; with a bristlecone pine frost ring at 1627 BCE; and with a sulfate spike in the GISP2 Greenland ice core at about 1642 BCE, although attributing that spike to Thera specifically is debated, since other eruptions may have contributed to it. Note that the ice-core figure is close to the tree-ring years rather than identical with them, and that the tree rings record a cold event, not an eruption's name.

The low chronology has its own anchor, and it is not a hunch. Manfred Bietak of the Austrian Archaeological Institute, excavator of Tell el-Dab'a, ancient Avaris, and Peter Warren of the University of Bristol argue for about 1530 to 1500 BCE on the basis of Egyptian pottery synchronisms: Minoan and Mycenaean pottery recovered from Egyptian contexts that are themselves dated, tying the Aegean ceramic sequence to the Egyptian historical calendar. That is a positive argument built on stratified, dated material, not merely an objection to somebody else's.

The low chronology also carries a specific technical objection to the radiocarbon dates. Malcolm Wiener argued in 2009 that samples from a volcanic setting may be skewed by a carbon dioxide reservoir effect, old carbon vented by the volcano itself contaminating the material being dated and pushing the apparent age older than the true one. Friedrich and Heinemeier contest the objection. It is not a wave of the hand at radiocarbon in general; it is a named mechanism that would bias results in exactly the direction the disagreement runs.

The reason neither side can quietly concede is that the bill comes due elsewhere. Our own file's counter-argument section states it plainly: if the high chronology is correct, Egyptian chronology for the Second Intermediate Period requires revision, with cascading effects through the entire eastern Mediterranean Bronze Age timeline; if the low chronology is correct, multiple radiocarbon dates, dendrochronological markers and ice-core signals have to be reinterpreted. Sturt Manning of Cornell has assembled the most comprehensive defense of the high chronology, in A Test of Time (1999, revised 2014). This is a discrepancy of more than 100 years, and it is the central unresolved problem in the file.

In 2018 Pearson and colleagues published in Science Advances a result that moved radiocarbon toward the archaeologists. Building an annual-resolution radiocarbon calibration series for 1700 to 1500 BCE out of calendar-dated individual tree rings from bristlecone pine and Irish oak, the team found that the existing calibration curve carried inaccuracies in exactly that window. Their revised calibration shifts the compatible eruption date to roughly 1600 to 1525 BCE, overlapping the archaeological estimate of about 1570 to 1500. That is a slightly wider window than the about 1530 to 1500 our own primary file gives for the same position, and the two are carried here as their sources state them rather than reconciled. The lead author's own stated conclusion, as the study's own institutional announcement quoted him, was that 'the radiocarbon evidence is compatible with the archaeological evidence for an eruption of Thera in the 16th century BC.' A radiocarbon study, narrowing toward the low chronology.

In 2022 Manning published in PLoS ONE a large Bayesian re-analysis that pushed the other way. It integrated radiocarbon datasets from Aegean contexts both on Thera and far from it, tree-ring wiggle-match dating at Miletos, and olive-wood samples killed by the eruption, and refined the date to 1606 to 1589 BCE at 68.3 percent probability, or 1609 to 1560 BCE at 95.4 percent. That supports the high chronology and puts the eruption in Egypt's Second Intermediate Period, a Canaanite and Levantine dominated era, rather than the New Kingdom that the archaeological chronology assumes. The same study tested the reservoir-effect objection directly, comparing radiocarbon dates from distant and presumed uncontaminated sites against Thera-proximal samples, and found no substantive difference between them.

Neither result closed the case. The dispute remains genuinely unresolved in the 2024 and 2025 literature. Manning published again in 2024, in 'Thera, the Aegean, Egypt, the Hyksos and Anatolia: Rethinking the orthodox synchronisations and histories', still arguing the high chronology, while archaeological synchronism proponents continue to place the eruption near 1500 BCE. The framing our research pass found in the current literature is that several decades of debate have followed with no clear resolution, despite wide recognition that the uncertainty undermines any ability to synchronize the civilizations of the eastern Mediterranean in the middle of the second millennium BCE.

One correction to our own file belongs here, because this article is built on it. E_1_16 states that as of 2025 the high chronology is favored by most scientists, while the low chronology retains support among some Aegean and Egyptian archaeologists. That is not false, and it is not the whole picture: the file does not engage the 2018 reconciling recalibration, the 2022 Bayesian defense, or the debate's continuation into 2024, all of which complicate a simple reading of who is winning. This article carries the fuller version, and names no winner.

The idea that Plato's Atlantis, in the Timaeus and the Critias of about 360 BCE, was inspired by the destruction of Minoan Thera was proposed by K. T. Frost in 1909 and elaborated by Spyridon Marinatos in 1939 and A. G. Galanopoulos in the 1960s. The points of comparison are real enough: an advanced island civilization with concentric harbors, destroyed by cataclysm, sinking beneath the sea. So are the points against. Plato put Atlantis in the Atlantic Ocean beyond the Pillars of Heracles, and dated it to about 9600 BCE. Neither matches Thera. The connection is evocative and it remains unproven.

The same applies, further out, to the biblical plagues of Exodus. Some writers have linked the eruption's darkness, tsunami and crop failure to that narrative. This stays highly speculative: the chronological fit is uncertain, the distance from Thera to the Nile Delta is roughly 900 kilometers, and there is no accepted archaeological evidence for the Exodus as a historical event to fit anything to.

No, the eruption did not destroy Minoan civilization instantly. The archaeological evidence shows clearly that the Minoan palaces on Crete survived the eruption by generations. The civilization's collapse around 1450 BCE had multiple causes, Mycenaean military action among them. The instant-destruction version is a misconception, and our own file's Tier 4 section names it as one.

What the record shows instead is continuity. The palaces at Knossos, Phaistos, Malia and Zakros were not abandoned after the eruption; there is evidence of continued occupation and rebuilding, the phase archaeologists call Final Palatial. The collapse came about 50 to 150 years later, around 1450 BCE, when the palaces were destroyed by fire and Mycenaean Greeks took control of Knossos.

That leaves a real question about the eruption's role, and our file's own answer is a careful one: the role is indirect rather than immediate. Destruction of the fleet, disruption of trade networks, loss of the Theran colony, agricultural damage from tephra fall on eastern Crete, and a possible psychological or religious crisis may between them have weakened the Minoan state enough for a Mycenaean takeover to become possible. May have. Every link in that chain is an argument, not a measurement.

The framework our library describes as the most widely accepted current view is explicitly multi-causal. Knappett and Nikolakopoulou proposed in 2015 that eruption, earthquake damage, trade disruption, internal social stress and Mycenaean pressure combined, which is a more moderate position than Marinatos's original 1939 thesis of direct volcanic destruction. Separately, Jan Driessen and Colin Macdonald have argued that the eruption triggered a political crisis on Crete, possibly a revolt against palace elites who had failed to protect the population.

None of that is settled, and this file does not try to settle it. What happened to Minoan Crete over the following century and a half is a larger question than one eruption, with more evidence in it than this article carries, and the honest position is that the eruption is one input to a contested multi-causal problem rather than its answer.