Analogies and Conjectures
Dating the Eruption of Thera: Minoan or Greek
Abstract
© Dr M D Magee
Contents Updated: Thursday, 06 July 2006
Santorini
Why do people think the eruption of the volcanic island of Thera (now Fira) in about 1600 BC was so catastrophic? Because it has been linked with the fall of the Minoan civilization? Because it has been linked with the destruction of Atlantis? Possibly both, but what is the evidence?
Santorini is an island with an enclosed elliptical lagoon measuring about 12km by 7km and 400m deep in the center, and surrounded by 300m high steep cliffs. Roughly in the center is the present volcano Nea Kameni. The collapse and sinking into the sea in the Minoan age of a volcano estimated to have been 1500 feet high must have been a terrible event accompanied by a huge tsunami, but it is certain that the Thera eruption did not destroy the Minoan civilization. Ash lies thick on Thera itself but only lightly on Crete—a few inches. Minoan remains appear on Crete above Theran ash layers. The Minoan palaces were burnt in about 1450 BC, long after the date of the explosion, now determined, and the Egyptians made no special note of it. It seems to have had surprisingly little impact on life for such a supposedly immense explosion. Yet, the eruption of Krakatoa in Indonesia in 1883, similar but smaller than estimates for Santorini, produced a tsunami which killed 36000 people and a noise heard 4500 km away. The eruption of Thera, considered as catastrophic, is thought to have happened in one day, or for just a few weeks. Perhaps really, it took a lot longer.
|
When the eruption happened on Thera, the caldera already existed much as it is seen now, though more complete being a nearly continuous ring. It had formed a few hundred thousand years ago by an earlier huge eruption which spewed out vast amounts of magma and ash until it had emptied the magma chamber and, unable to support itself, it collapsed. The superstructure crashed into the empty magma chamber, leaving the caldera, a hole where the magma chamber had been. The hotspot beneath the crust then causes more hot material to push up forming a new magma chamber and repeating the process at intervals. That the shape of the island before the Minoan eruption was similar to to-day is suggested by stromatolites and gastropods, within the caldera before the Minoan eruption, as dated by radiocarbon. The caldera bay where they grew was mainly shallow, but certainly in the enclosed portion to the north of the active cone.
If the whole island had been a volcanic cone which was blown out except for the caldera ring, the volume of ejected material must have been around 60 km3 (Druitt and Francaviglia, 1992), twice the present guesses of 27-30 km3 (Pyle, 1990). Such a huge discrepancy between the collapse and eruption volume is difficult to explain. So, the Minoan eruption was not of the whole island but of a volcanic island within the ring, roughly where the present small island in the caldera is. The magma chamber filled and exploded from time to time, and between those times Surtsey-like activity punctuates dormant periods creating the islands within the caldera that evidently explode violently at irregular intervals. An island emerged from the sea in 197 BC, and Nea Kameni in 1704, and it erupts periodically—three times in the twentieth century, the last being in 1950.
The Minoan Eruption
|
A 75 feet thick deposit of white tephra, 90% pumice, still lies on top of the pre-eruption ground level. Its three bands show the eruption happened in three phases, the first of which was a mild powdery eruption leaving a thin ash layer, 3 cm thick, in which winter rain has washed runnels but had not washed it away. It suggests this initial venting was less than a year before the main event, enough to warn the residents to evacuate the island. The next two layers are thick light coloured tephra, the topmost one being almost white. No bodies other than older burials have been found. People had had time to escape and to take their valuables with them too.
The eruption caused a tsunami that impacted the north coast of Crete, less than 100 miles off, knocking walls out of alignment. The wave cannot have been too powerful, though, especially at such a close distance. It was scarcely devastating, and seems to have been accompanied by fire. Marinatos, the original excavator of Santorini, could “not reconcile the fire with a terrible inundation caused by the sea”. Effectively, the ring of the caldera stopped the collapse of the magma chamber from disturbing the sea and surrounds as much as it might have done.
H Pichler and W Friedrich noted that the state of the geology suggests a two stage collapse of the island, the two stages considered to have been immediately one after the other. Maybe the second stage of the collapse was actually 400 years later. L Pomerance associates the general havoc c 1200 BC in the whole Mediterranean world with the final collapse of the chamber. Either the second stage was an eruption of new magma, or the volcano was left after the first collapse in a metastable state until another seismic event—Santorini was also struck by a devastating earthquake in 1956—triggered the second collapse, perhaps with sea water breaking into the magma chamber.
|
In the first, the 1600 BC collapse, after a period of up to a year of mild venting, about a sixth of the area of Thera exploded, ejecting 5 to 8 km3 of collapsed material, probably blowing out the Surtsey-like island (c 3 km3) inside the caldera and some of its support column. Sparks and Wilson showed the eruption was increasing in violence with time. It was a continuous dry gas driven eruption, and shows no signs of involving any seawater interaction with the magma. It supports the vent being on the caldera island and not submerged. Only in the final phase (2-40 cm) of fine white ash suggests a reaction with sea water entering the vent. The collapse of the inner island meant the pyroclastic flow and the upheaval of sea water were contained by the caldera.
So, the tsunami associated with the original collapse was modest, its power largely absorbed by washing up and over the caldera wall. The tsunami on open water was slight. Interestingly, a mural found in Akrotiri was of a large blue wave rocking boats and throwing people from them, leaving them helpless, or even dead, in the water together with sticks or oars from the boats. A tsunami had been experienced just before the eruption, because the job was unfinished and the artist's mortar and materials were found. While the tsunami evidently had not fazed the people, soomething did, and it was probably the begining of the volcano explosively spewing out pumice.
Such fall out of ash as occurred was directed away from the eastern Mediterranean and Egypt, and no doubt helped by the prevailing winds, blew towards Asia and the Black Sea. Ash layers in cores drilled from the seabed and from lakes in Turkey show that the heaviest ashfall was towards the east and northeast of Santorini. Ash found in Crete was from the warning phase of the eruption, months earlier. No ash deposits found in the Nile delta are from Santorini.
Signs of a large tsunami appear in the Ionian and Sirte abyssal plains, and on the slopes of the Calabrian Rise and the Mediterranean Ridge. Some pumice strata in Cyprus date c 1200 BC. Analogous deposits have been found at Amnisos, Anaphi, and on the coast of Palestine). Navigation will have been impossible with the floating pumice on it. Even when the volcano erupted mildly in 726 AD, it was impossible to pass the pumice rafts for a long time. No remains of settlements dated from 1500 to 1200 BC have been found on Santorini. It suggests that the later collapse (around 1200 BC?) was more catastrophic, with the settlements that must have formed in between times, when the tuff had consolidated and sprouted some vegetation, being literally washed off the island. The whole of the contents of the caldera, a mud of sea water and tephra, frothed up and over the rim carrying ash and large rocks from the collapse of the magma chamber.
Now sea water entered the magma chamber producing violent explosions that pulverized the magma and ejected large stone blocks. The end of the eruption seems to have been a pyroclastic flow down the steep slopes slowing and finally settling on the flatter areas. Geologists think this final layer was perhaps reworked by the conditions. It was in these final stages that about 30 km3 of material were ejected, agreeing with the size of the magma chamber and the volume of tephra found. It broke open sections of the caldera ring, creating two channels to the open water instead of just one. Deposits of ejecta from the explosions within the caldera show that the caldera existed before the eruption, and so does the state of erosion of some cliffs which show their antiquity. They were not newly formed. Indeed, Mount Profitis Ilias in the southeast of present island is made of ancient limestone.
Dating the Eruption
Dendrochronologists found tree ring growth in America interrupted in 1629-1628 BC suggesting a catastrophic climatic event, probably an eruption. A 9000-foot Greenland ice core (GRIP) has evidence of 400 volcanic eruptions, in the past 7000 years, severe enough for the ash to reach Greenland, about 3500 miles away from Santorini. A prominent ash layer at a depth corresponding to 4803 BC may have come from the eruption in Oregon that destroyed Mount Mazama, leaving the giant caldera that is now Crater Lake. Among the cores was a candidate dated to 1644 +/- 20 BC years. Scholars suspected Santorini. Then analysts found tests of the tephra in that layer of the core did not match the characteristics of Santorini. The methods used to match the Greenlandic tephra with Thera were flawed and geochemical data suggest the tephra is not from Thera. Another ice-core sample suggested an eruption in about 1623 BC. This might have been Thera, but the earlier one, it is thought, was in Alaska. It could still have affected climate, if it was that big, and therefore the growth of Anatolian trees. Growth anomalies at Porsuk in Anatolia might have occurred around 1645 BC not 1628 BC.
The widespread traces of the eruption could be immensely valuable for dating except that no one could date the eruption itself. Now carbon dating and dendrochronology have dated the event to between 1630 and 1600 BC, about 100 years before the date reckoned by ancient historians of around 1500 BC. Thera's culture matched Cretan Late Minoan IA (LMIA), and fragments of nine Syro-Palestinian Middle Bronze II (MBII) gypsum vessels offer another match. The culture also matches Late Cycladic I (LCI) and Late Helladic I (LHI) but precedes Peloponnesian LHI.
|
On 28 April 2006, the journal Science published two research papers arguing that new radiocarbon ages required an eruption date between 1627 to 1600 BC. S Manning et al established a chronology for the initial Aegean Late Bronze Age cultural phases (Late Minoan IA, IB, and II) by using 14C dates from the surrounding region—127 samples of wood, bone, and seed collected from various locations in the Aegean, including Santorini, Crete, Rhodes and Turkey. They sent the samples to three separate labs—Oxford, Vienna, and Heidelberg—to reduce any systematic radiocarbon dating errors, and analysed the results by Bayesian statistics. Manning's research offered a broad dating for the Thera event between 1660 to 1613, stretching Late Minoan IA, IB, and II by almost a century, challenging conventional dating of Egypt and the Aegean and Cyprus in the mid–second millennium BC.
W Friedrich et al, narrowed the time-line for the eruption of Thera by radiocarbon wiggle-matching a carbon-14 sequence of tree rings to 1627-1600 BC. Scientists separately date annular rings in the wood by the radiocarbon method then match them to a sequence by statistical methods. An olive branch, found on Santorini buried in situ by the pumice and ashes, was living until the hot ash carbonized it. So the date of its outermost annular ring was the year of the eruption. It was dated to c 1613 BC. The trunk of the tree unfortunately had long ago fallen down the cliff and been lost to the sea.
A question mark over the method is that outgassed carbon dioxide from the volcano contains old carbon containing no 14C. The plant growing in an atmosphere low in 14C, must yield a radiocarbon date too old. Since Santorini is a windy island, it seems unlikely to be a major factor unless the caldera was acting as a vessel to hold the carbon dioxide, which is a dense gas, and the plant was growing inside the caldera ring in a sheltered spot. Its situation now is near the top of the easter inner side of the caldera rim, not sheltered, but its situation at the time might have been different. To date some carbon in the topmost light coloured tephra would also be useful, just to decide whether it was laid down in the same eruption or a later one. On the two phase theory, it might be an eruption 400 years later.
Archaeologists, Peter Warren and Manfred Bietak, doubt the new date. They prefer a conventional interpretation of the absolute ceramic chronology, derived from correlations with Egypt. In fact, the archaeology is a sequence and not a set of dates, so they cannot take priority over a definite date. The eruption was before the end of the LM IA period, but late MM IIIB ware had reached the island. So Minoan ceramics testify to an eruption in the ceramic LM IA period. The problems are the lack of certain LM IA pottery from datable contexts.
These radiocarbon dates on short-lived samples from Santorini might provide the best evidence yet available, but all possibility of systematic errors like the presence of old carbon, sample selection, and the statistical validity of the wiggle matching method cannot allow the luxury of certainty. In earlier attempts, blatant sample manipulation had occurred, whereby a core set of samples were accepted and “outliers” rejected. That is not science unless there is a sound scientific basis for the selection. It is rarely promising when classicists glibly use sophisticated statistical methods. Curiously, the uncalibrated date is about 1300 BC! Are the calibrations really sound? The whole needs repeating critically. If the dates are right, the chronological adjustments needed are not great but they would put the mockers on any reduction of the dates, and that is where most of the somewhat later data point.
