Santorini and its eruption in the Late Bronze Age
Chapter 2. Earth: The Minoan eruption in its archaeological landscape
(For the pdf version of this document with correct page references download the chapter 2 pdf.)
• Problems with chronology
The framework upon which archaeology is based is time, but not time as we live it now in the present, composed of hours and minutes, but rather blocks of time or periods. These periods represent broad brush-strokes of human past rather than the thoughts and aspirations of individuals. Particular periods are defined by many disparate objects, ranging from utilitarian (tools, pottery), to the existential (religion, art). The defining events which bring one period to an end and herald the next are in many instances violent and marked by the destruction of key or type-sites. This is certainly true for some periods of the chronology of the island of Crete with the periods being defined by successive destruction and rebuilding phases at the major palace sites.
Archaeological time differs from the present in its ability to ignore the actuality of time. The effect of seriation of artefacts within culture area groups leads to a disassociation with actual time, which in turn has caused a difficulty in bringing different culture areas into alignment with each other. Within its own cultural sphere time can be seen as linear and the archaeological periods are all relative to each other although they do not necessarily have an absolute calendrical date that we would recognise. Cross-referencing of cultures using seriation and typology is achieved by the elements of one culture area occurring within the cultural deposits of another. By cross-referencing, correlations can be built up between the cultural groups with which they interacted. Sometimes, because of the paucity of cultural links, there is a problem that this sample size is not large enough and is biased by such phenomena as heirlooms and the influence of special purpose burial archaeology over material abandoned in domestic settings.
Figure 2. 3 Chronological seriations for the Minoan, Helladic and Cycladic cultures. Instead of portraying the boundaries between phases as hard and fixed lines, I have tried to show them as more fluid, melding into each other. The figure also shows the architectural phase of Neo Palatial and Final Palatial.
The initial chronology of the Aegean cultures was created when the theory of diffusion of culture from the Near East was prevalent (Childe 1925) . Civilisation diffused from this area in a similar process to osmosis. The chronology of the civilisations of Mesopotamia and Egypt based on regnal years and cosmological links formed a bedrock from which other cultures could be given an absolute date. The destruction of this theory by the 14 C revolution placed studies of Aegean civilisation in a quandary (Renfrew 1973) . The Aegean cultures had some cultural links with Near Eastern centres; some of these were open to question, and they had few 14 C dates and many of these were contentious. Thus we have two types of chronology to date the Santorini eruption, the relative which will allow us to see how the eruption affects sites on Crete and the absolute which will help us to cross-reference any impact with other cultural groups. The absolute date will be based on correlations between scientific dating such as 14 C and dendrochronology but must fit in with the traditional dates of the Near Eastern cultures which are based on fixed astronomical dated events.
The three main pottery chronologies that are relevant to this discussion are the Minoan, Helladic and Cycladic, each of which has their own nomenclature. Figure 2.3 shows the relevant relative chronologies for each of these cultures. The absolute chronologies are more problematical and open to debate and so can be found in specific works on this subject (e.g. Warren and Hankey 1989; Rehak and Younger 1998; Warren 1998; Manning 1999) .
• The Minoan pottery seriation
Pottery analysis has formed the backbone of much archaeological research and many chronologies are based upon it. This is as a result of the fundamental nature of pottery in that it reflects human endeavour whilst remaining essentially a geological item which can withstand the burial process. The study of the development of form and style through a series of changes over time has a long tradition (see Renfrew and Bahn 1996, p116 for outline) . The relative chronologies of the Aegean Late Bronze Age period are based upon this type of pottery seriation. The history of the developments of pottery typology is summarised in Orton et al. 1993 table 1.1 (Orton et al. 1993) . Much of the initial work in this area was done by Evans at Knossos (Evans 1921-36) . Unfortunately Knossos is an exceptional site and has many features which are not reflected elsewhere and the neat system devised by Evans has had to be adapted to conform to other Minoan sites as the discipline has developed.
The particular phase with which this work is mainly concerned is the Late Minoan I, which was subsequently split into two separate phases, LM IA and LM IB. Both phases make up part of the Neo Palatial tradition with the destructions at the end of LM IB heralding in the Final Palatial period (following Rehak and Younger 1998, p92) . The LM IA phase of decorated fine wares does not die out completely and continues into the LM IB. There is very little variation between coarsewares in the two phases and so they are often assigned on the presence or absence of finer wares within the same context (Betancourt 1985, p124) . Many of the types, both cooking and storage vessels, carry on from MM III traditions.
Figure 2. 4 Theoretical model illustrating the development of pottery styles.
To help illustrate the problems of using pottery seriation at sites I have developed figure 2.4, which is a theoretical model and does not represent actual pottery styles. Style A represents a pot type developed in the MM III which continues into LM IA. Style B is a coarseware that is present in all phases but in varying quantities. Style C develops and is in use solely in LM IA. Style D develops and is in use solely in LM IB. The presence of Style A pottery in a context probably means that the context was in use during the MM III or LM IA periods but it could be an heirloom and may have continued in use during the LM IB phase. Because of the problem of residuality much of the pottery that is preserved in an archaeological context may be from an earlier period. The presence of Style B pottery only confirms that the material comes from a general period in the Late Bronze Age but it is not specific. Style C confirms that the context cannot be earlier than LM IA whilst Style D confirms that the context is not earlier than LM IB. It can be seen that this can generate false assumptions and that a large dataset containing many pottery styles is the best way of eradicating this. The site of Akrotiri has produced huge amounts of sherds and complete vessels of Minoan origin, none of which has been dated to the LM IB.
The LM IA decorated wares concentrate upon four types of decoration, the spiral, rosette, ripple motif and the floral band in horizontal friezes (Betancourt 1985, p128) . There is some development of the styles through time and an early and mature phase has been recognised.
LM IB decorated wares have proved more difficult to define. The initial characterisation was made by rare examples of palatial manufactured ware, such as the Marine Style or the Reed Painter. A reanalysis of material particularly from older sites has revealed a more complex scenario (Betancourt 1985, p133) . Many of the deposits come from the series of destructions that were thought to occur at the end of the period, when many sites across Crete were destroyed by fire. Betancourt's list of destroyed sites nicely illustrates the problem of creating a synthesis across sites; it shows the years when the sites were excavated varying from 1908 to 1980 (Betancourt 1985, p134) . The LM IB pottery has effectively been divided into two main traditions, the standard, which is the more conservative, and the palatial. The standard tradition follows some of the precepts of the LM IA tradition; many of the styles are so similar to LM IA that they cannot be distinguished from it. The conclusion that all sites that were destroyed by fire must represent the end of the period has been questioned, some probably occurred during, and would have contained pottery actually manufactured, in LM IA (Betancourt 1985, p139) .
The second main tradition has been referred to as the Special Palatial tradition and has four main divisions, the Marine Style, the Floral, the Abstract and the Geometric. Any of these works are highly distinctive and are possibly a result of the translation of Minoan frescoed themes to the pottery medium. So where do the excavations at Santorini fit into this schema? Marthari (1990) outlined the first detailed stratigraphical analysis from the Akrotiri excavations (see figure 2.5). This concentrated on the West House which offered the most complete sequence, and was based on nearly one thousand complete or nearly complete vessels. The majority of these were of local Theran manufacture. But there were three significant groups of imports, Cycladic, Helladic and Minoan. The Cycladic vessels confirmed interaction with islands such as Melos and Kea. The Helladic vessels, mainly Vaphio cups have a number of Middle Helladic vessels but there are enough LH I vessels to confirm that as the period of eruption. The majority of the foreign wares are Minoan however, and appear to date to the mature LM IA from central and eastern Crete . Despite the huge numbers of sherds and complete vessels excavated at Akrotiri no sherd of LM IB has been positively identified. Therefore the hypothetical link between the LM IB destructions on Crete and the eruption of Santorini can not be sustained. In terms of the relative archaeological chronology, the Minoan eruption of Santorini was not directly responsible for island-wide destruction of sites on Crete in the LM IB.
From the thousands of pottery examples from Santorini two particular vessels have emerged as being important. This is because they are involved in the debate for the absolute date of the eruption. Two different absolute chronologies have been proposed, termed High and Low which are discussed in greater detail below. The first vessel is a Canaanite storage jar that Niemeier identifies as similar to material from excavations of a tomb at Tel Kabri in the Levant (Niemeier 1990) . This very rich tomb is dated to the MB IIB and links Santorini with the Palestinian chronology. The palace at Tel Kabri also produced a floor decorated in the true fresco technique which is very unusual for Near Eastern sites and is possibly indicative of an Aegean influence (see Niemeier and Niemeier 1998, Manning 1999 for background and discussion).
The second pot was a White Slip vessel found in the earliest French excavations. The bowl was only poorly recorded and illustrated at the time of excavation and was subsequently lost (Fouqué and McBirney 1998, Plate XLII number 6; Manning 1999, fig 31 and p150-154) . This vessel ties Santorini to the pottery chronology of Cyprus and, because of the incidence of this ware at the Egyptian site of Tell el Dab'a, with Egyptian chronology as well (Bietak 1995; Bietak 1998) . This is a complex issue based upon the uneven development of pottery styles across Cypriot cultural regions. The White Slip bowl forms a central piece of the recent work by Manning who argues for a High chronological date for the eruption (Manning 1999) .
• The absolute chronology and questions of High versus Low
As stated above the relative chronology for the eruption appears to have been settled, the absolute chronology however has not been and there are two main chronologies put forward, a High and Low Chronology (see figure 2.6). The Low Chronology which is the traditional or established date would place the eruption around in the early to mid 16 th century BC and is based principally on the cross-referencing of Aegean archaeological material with other established chronologies, in particular the Egyptian absolute chronology (Warren and Hankey 1989, p137-146) . The Egyptian chronology is regarded as being one of the most robust archaeological chronologies. The High Chronology places the eruption in the early to mid 17 th century BC and is based on a number of scientific methods as well as cross-referencing archaeological material. This chronology places a greater strain on the relative chronological sequence and therefore is the more radical interpretation. Kuniholm, in the introduction to the chronology session at the third Thera conference succinctly outlined how the new evidence has built up and the recent work by Manning which focuses particularly on this issue outlines the work to 1999 (Kuniholm 1990; Manning 1999) .
Manning illustrates how the Low Chronology is based upon conventional archaeological dating whilst the High Chronology takes into account dendrochronology, 14 C and evidence from ice cores as well (Manning 1999, fig 17, p25) . Although these differing techniques do not completely agree on a specific date, statistically they indicate that the High Chronology is the most likely. At the present time the argument cannot be solved categorically and the discussion is still open. Below I discuss some of the problems associated with each of the techniques to help illustrate why there is no consensus.
• Archaeological cross-referencing
The basis of archaeological cross-referencing is that there was trade and exchange between cultures. Material from one culture is then preserved within the archaeological record of another. If one of those cultures has a good absolute chronology then it can be used to provide a framework for the other culture. In this case the main evidence comes from the Egyptian chronologies and is a complex web of facts, assumptions and art-historical developments. To illustrate this I have laid out below some of the points from a discussion article in Archaeometry on the question of the Thera date (Aitken et al. 1988) . Within the article Michael and Betancourt, and Warren presented two different opinions about three archaeological artefacts. Michael and Betancourt (p169) discounted from the debate an Egyptian stone lid with a cartouche of Khyan found at Knossos because it came from a mixed context. Warren (p176) accepted the Khyan lid because although there was mixed pottery the stratigraphic position was fixed. The el-Lisht jug, another of the examples used, is a vase decorated in the Tell el-Yahudiyah ware technique with birds and dolphins. Michael and Betancourt felt the iconography of the dolphins was from the LM I (p170) whilst Warren preferred a MM III date (p176). Since this article was written more items have been included in the debate (Warren 1998) but I am fundamentally wary of relying totally on this type of absolute dating when so much of the evidence is based on personal opinion.
• Radiocarbon
In theory Akrotiri should be an excellent case for the use of 14 C dating. The quality of the deposits and the fixed nature of their emplacement, with both long and short-lived samples being present, were seen as an ideal case. The first suite of 16 samples gave a calibrated date of 1625 BC and since then a growing number of samples have given this approximate date as the most likely result. There are, however, a number of problems with the 14 C data which means that their validity is often called into doubt.
There is the possibility of contamination by ‘old' volcanic CO 2 due to the fact that the samples were close to an active volcano. Work presented at the Third Thera conference states that although some contamination does occur this is limited to material close to the point of degassing and that such contamination would be unlikely to affect the whole suite of samples (Hubberten et al. 1990) .
14 C dating uses calibration curves to make up for the variable amounts of 14 C in the atmosphere over time. A number of these curves have been generated and effectively provide a standard used by researchers. The problem with the calibration curves as far as the Santorini eruption is concerned is that they are very flat across the dates which have been proposed for the eruption. This means that statistically both the mid 16 th century and mid 17 th century dates could be correct.
The third main area of contention lies in the use and manipulation of data. Often when suites of samples are produced by the various Radiocarbon laboratories there may be an odd result or outlier. These outliers if included cause the statistical date to move. If they are excluded then there is the possibility of massaging the results to achieve a specific aim. This manipulation of statistics in my view creates unease amongst some researchers, which in turn causes them to doubt the validity of the process as a whole.
Possibly where 14 C could be of use is in the periods surrounding the LM IA eruption, the MM IIIA, earlier LM IA and LM IB and LM II. This is the argument behind a recent article that looked to date the LM IB destruction layers on Crete (Housley et al. 1999) . The proposed date obtained from the analysis for the close of the LM IB phase was 1525 to 1490 BC which would support the High Chronology. The article did however discount some of the results that were obtained. Whilst these may be due to valid reasons such as contamination, it opens up the question of statistical manipulation and therefore means that such results are not acceptable to all.
The problem with 14 C dating can in my opinion be expressed by an idea taken from quantum physics. Heisenbergs' Uncertainty Principle states that it is impossible to completely accurately measure the position and momentum of a particle in space because by measuring its position you actually move the particle. The question for 14 C dating is whether the number of consistent results for this case will convince the majority of people of the technique's accuracy, or whether the number of outliers and spurious results will be of sufficient number to provide doubt for a significant number of the interested community.
• Ice cores
In 1980 Hammer et al. published a paper outlining the proposal that ice-cores taken from polar regions could preserve evidence of major volcanic events (Hammer et al. 1980) . In normal conditions precipitation falling out over the ice sheets of Greenland and Antarctica is very pure. A volcanic eruption can inject into the atmosphere large quantities of impurities. Despite the spread of the cloud and washout by precipitation any ‘clean-air' locations in the same hemisphere will receive a greater load of polluted aerosols than the normal background (Hammer 1977, p482). Volcanic eruptions produce large amounts of acid (H 2 SO 4 , HCl, and HF) and the frequent snowfalls over the Greenland region will scavenge some of these impurities. Layers of snow and ice are built up in this region over time and it is possible to correlate these layers on a temporal basis. Measurable peaks in acidity in these layers can then be related to the fallout from large volcanic events (Clausen et al. 1997) .
The size of an acidity peak does not necessarily correlate exactly with the size of the
volcanic eruption. Thus a large eruption does not have to create a large acidity peak. The quantity of acid aerosol created varies with the size of the eruption, the location (which hemisphere) and concentration of sulphur in the magma. Hence there are several unknown, ambient factors that can affect the size of the acidity signal.
The initial paper by Hammer et al. proposed a date of 1390 ± 50 BC for just such an acidity peak and they proposed that this related to Santorini (Hammer et al. 1980) . However this paper has since been publicly retracted, as it became obvious to the authors that there was a break in the Camp Century core that they had used. A subsequent paper published a date of 1645 ± 20 BC for a new core called Dye 3 which would appear to back the High Chronology (Hammer et al. 1987) . In 1994 a team working on the GISP2 ice core found an acidity signal dated 1623 ± 36 BC which they felt could be the Santorini eruption since it agreed well with the dendrochronology dates being put forward (Zielinski et al. 1994) .
The fundamental problem with these dates is that when they were published it could not be shown that the acidity signal shown was related to the Santorini eruption. In an attempt to get around this problem the section of the GISP2 core which contained the acidity signal was examined and a number of tephra shards were recovered. If these shards could be shown to be from Santorini then this would confirm the High Chronology date. Unfortunately the results have not been conclusive and have instead created a more confused picture; a more in-depth analysis of the results is developed in chapter 4. An initial article by the GISP2 team proposed that the tephra shards found could not be linked to Santorini (Zielinski and Germani 1998a) . This result was noted by Warren in an addendum to his article on new contributions to Aegean archaeology and he felt this removed a fundamental plank in the argument for a High Chronology (Warren 1998) .
Manning (1998) refuted the Zielinski and Germani claims and stated that it actually showed the tephra could have come from Santorini. Zielinski and Germani (1998b) rejected his arguments. The findings of the Zielinski team do not appear as clear cut as they first stated and Schmid et al. (2000) , for example, raised doubts about their methods. The most recent development is from the GRIP ice core team which states that the 1628 BC acidity peak is actually irrelevant and that the GISP2 core is missing a 1645 BC acidity peak. Thousands of tephra shards have been found from the 1645 BC GRIP signal and the early indications are that they are from the Santorini eruption (Hammer 2000) .
Despite much work in this area the ice core data can not as yet provide a definitive answer to the conundrum of the absolute date of the Santorini eruption. The withdrawal of the initial proposal put forward for a 1390 BC date has in my opinion hampered the acceptance of this dating method. The work to link specific acidity signals with tephra particles found in the cores would appear to be the right approach. However, because there is now a debate about the validity of the first results published, the data has again proved inconclusive and has provided points of argument for both the High and Low Chronology proponents.
• Dendrochronology
The evidence of dendrochronology in many ways follows a similar principle to the ice-core data. Instead of an acidity peak, an anomaly is shown in the amount of tree ring growth produced during a particular year. The change in growth is related to climatic variability caused by the volcanic eruption (Baillie 1995) . The size of the climatic variability and therefore the size of the anomaly can not be directly correlated with the size a volcanic eruption.
In 1984 a paper by LaMarche and Hirshboeck stated that there was a correlation between volcanic eruptions and small growth rings on trees (LaMarche and Hirschboeck 1984) . The lack of growth would correlate with a cold snap or frost event caused by the volcanic eruption. In particular they were using the long-lived bristlecone pine as a source for this data. In their studies there appeared to be no restricted growth in the 16 th century BC but there was one event which they correlated with 1626 BC. This was followed by a paper by Baillie and Munro which found a similar event dated from 1628 to 1626 BC in Irish bog oaks (Baillie and Munro 1988) . A study of Anatolian tree rings indicated an anomalous exceptional growth period which may correlate with the 1628 BC event, but further work still needs to be done (Kuniholm et al. 1996) .
The problem with this evidence is that although it points to a significant world-wide event it does not mean that it was Santorini. Although Santorini is the most studied volcano of the period there are many other volcanoes that could have erupted at that time to produce such a proxy record. At this stage the results of the dendrochronology work are still equivocal.
• Summary
Undoubtedly the absolute date of the eruption is an important issue but it is not the main focus of this work. Statistically the High Chronology would appear to be the most likely result on the basis of the radiocarbon evidence. At the current time I find the data from dendrochronology and ice-cores useful but not persuasive. Generally though, I find the art-historical nature of much of the Low Chronology data, some of which is obtained from dubiously excavated contexts also unconvincing. As will be shown in chapter 4 there are significant quantities of Santorini tephra preserved in sites to the east of Santorini. Further excavation combined with dendrochronology and 14 C dating from this area may hold the key to an absolute date.
| Back | Next |
This website is devoted to the eruption of Santorini volcano in the Late Bronze Age and its impact on the cultures and civilisations of the time. The website is owned by Dr. David Sewell.