Radiocarbon Dates from the Akrotiri Excavations

In this study we first tried to look for arguments to explain the too old dates of the reported radiocarbon dating. Investigations on the C-14 content of present-day plant material growing in the vicinity of CO₂ emanating gas sources showed that 'ages' of up to 1500 years may be artificially induced due to the incorporation of 'dead' volcanic CO₂.

Stable carbon isotope investigations on volcanic gases, air samples and plant material from Santorini revealed that the influence of volcanic CO₂ on the C-14 content of plant material may be detected through the stable carbon isotope composition of these plants.

Eleven new radiocarbon dates obtained on material from the Akrotiri excavations cluster about a mean value of 1670 BC and are therefore consistent with earlier published dates.

As the influence of volcanic CO₂ has been shown to be of a very local importance and cannot explain the overall shift of the radiocarbon dates to older values, we assume that the catastrophic eruption of the volcano of Thera took place in the 17th century BC

This assumed destruction date is not only consistent with all the other published radiocarbon dates, but is strongly supported by acidity peaks in ice cores (Hammer et al. 1987), frost events in tree rings (La Marche and Hirschboeck 1984) as well as by discussions of a revised Aegean time scale (Betancourt 1987).

INTRODUCTION

During the last scientific congress 'Thera and the Aegean World', held on Santorini in 1978, the presented radiocarbon dates disagreed with the date of the destruction of the village of Akrotiri as assumed by the archaeologists (Doumas 1978, 1980). Although broadly scattered, the radiocarbon dates indicated that the catastrophic eruption of the volcano of Thera occurred some 100 to 150 years earlier than assumed by classical pottery dating.

Due to these discrepancies, a joint project was initiated by Professor Dr A. Boettcher (Jülich) in order to shed some light on this problem. When the investigations were begun, we believed that the archaeological destruction date was correct and that the observed age differences were due to incorrect radiocarbon dates. As there has been no doubt regarding the methodological part of the published radiocarbon data, the reason for these discrepancies was assumed to be attributable to the samples.

As Santorini is a volcanic island, where CO₂ gases emerge at several sites, even today (Fig. 1), the incorporation of volcanic, i.e. 'dead', CO₂ could have shifted the radiocarbon dates to older ages (Chatters et al. 1969; Sulerzhitzky 1970). Our investigations therefore concentrated on the question as to what extent present-day plant material growing near volcanic CO₂ emanations was influenced with respect to the C-14 content. We wanted to determine if plants which were influenced by volcanic CO₂ could be distinguished from others with help of C-14 independent parameters. It was assumed that these discriminating parameters would be very useful selecting archaeological samples displaying no influence of volcanic CO₂.

To clarify this question we planned to study the following:

-     The possible influence of volcanic CO₂ on the C-14 content of present-day plant material.

-     The isotopic characterization of volcanic CO₂ emanating at various sites on the Kameni islands.

-     The determination of the ¹²C/¹³C ratio in air and plants in the neighbourhood and away from the gas emanating sources.

-     The elaboration of characteristic trace element patterns for plants growing at volcanic or non-volcanic (away from gas sources) areas of the islands.

-     The analysis of the C-14 content of samples from the Akrotiri excavations.

RESULTS

Influence of volcanic CO₂ on the C-14 content of present-day plant material:

For the analysis of the influence of volcanic CO₂ on the C-14 content of present-day plant material, we selected the Eifel area in Germany as a model location (Bruns et al. 1980). In that region, where the last volcanic activity took place some 10000 years BP, CO₂ emanating sources can be found in many places, especially in the basin of Wehr and at the lake Maria Laach where the gaseous emanations are visible as fields of bubbles. For our investigations seven samples from annual plants, especially leaves and sprouts, were taken at different distances from the bubble-emanating area and analyzed for their C-14 content. As most of the results obtained during this study are published elsewhere (Bruns et al. 1980), only the main results will be summarized here.

It has been shown for the plants from the Eifel locations, that the influence of 'dead' CO₂ on plant material causes 'pseudo ages' of up to 1500 years. The influence of volcanic CO₂ on the C-14 content of plants decreases with increasing distance from the source, but may be of importance at least over a distance of some hundred metres away from that source (Fig. 2).

Similar investigations carried out on present-day plants growing around the CO₂ emanating area at Palaea Kameni showed almost the same results, i.e. 'ages' of up to 1400 years in annual plants (Table 1).

The stable carbon isotope composition of volcanic CO₂ emanating at Palaea and Nea Kameni:

As mentioned above, emanating CO₂ gases may be observed around the Kameni islands bubbling through shallow waters in several bays (Fig. 1). The most spectacular bubble field is found within a bay at Palaea Kameni, where hot solutions which emerge together with the gases precipitate and form a thick iron-rich mud layer on the sea bottom (Puchelt 1973).

Thirty-six gas samples were collected during several field trips between 1976 and 1980 and analysed for their ¹²C/¹³C ratio. Sampling, sample pre-treatment and measuring procedures were carried out according to Puchelt and Hubberten (1980).

The results obtained on these samples are shown in Table 1 and in Fig. 3. All of the samples show a small variation around the zero point of the delta scale. No temporal variation has been observed for samples taken at the same location in different years. It is assumed that the initial carbon isotope composition of the CO₂ gases emerging at Santorini will fall in a narrow range between -1 and 0‰. The isotopically heavier values (positive isotope ratios) are explained by a kinetic isotope effect occurring due to a selective dissolution of the light carbon isotope ¹²C when the gases are bubbling through the water column. This kinetic isotope effect could be qualitatively observed in samples from the bay where an increasing water depth was paralleled by an enrichment of the heavier isotope ¹³C. This effect was subsequently determined quantitatively by Puchelt (1982) in laboratory experiments.

The mean value of the CO₂ gases from Santorini (-0.5‰), is significantly different from carbon isotope values expected in primary magmatic CO₂. Magmatic CO₂, occurring in the form of magmatic carbonates in basic rocks or carbonatites or as CO₂ gases, is isotopically significantly different (-4 to -7 ‰) from that found in the Santorini CO₂ (Fig. 3). This means that it is also different from the isotopic composition of normal atmospheric CO₂, a fact which will be of some importance (see discussions below).

If the Santorini gases do not represent magmatic CO₂ formed as a result of postmagmatic activities, the only source could be the limestones occurring underground as proved by the presence of numerous limestone inclusions in lavas of Palaea Kameni.

The isotopic composition of various Santorini limestones is shown in Table 1 and Fig. 3. These limestones, sampled from sedimentary areas of Santorini (Profitis Ilias, area of Athinios) as well as on top of Palaea Kameni (xenolithic inclusions), have almost the same isotope composition as the gases, although they are more scattered. It is therefore obvious that these gases formed either by a thermal decarbonization reaction of limestones due to the heat provided by the volcano, or by a decarbonization reaction due to the contact with a siliceous magma in the subground of the island.

The isotopic composition of atmospheric CO₂ and plant material:

Air CO₂ has been sampled at various places of the island of Thera (Fig. 1) to establish if volcanic gases may be detected in normal Santorini air. As shown above, the isotopic composition of the Santorini volcanic gases are significantly different from true magmatic CO₂. It was therefore assumed that an admixing of those gases to normal atmospheric CO₂ (δ¹³C between -7 and -11‰, Keeling 1961) could be seen in the isotopic composition.

The results obtained (Table 1 and Fig. 3) show that Santorini clean air has an isotopic composition of -9.8‰. Compared with this value, the air samples taken in the vicinity of the gas sources are shifted to isotopically heavier values due to the admixture of volcanic CO_2^. This means that plants growing in the vicinity of volcanic gas emanations are not only affected with respect to their C-14 content, but should additionally reflect the different stable carbon isotope composition of the atmospheric CO₂ in the isotopic composition of their organic carbon.

When looking at the carbon isotope composition of three plants (Table 1 and Fig. 3) it is obvious that the admixture of 'dead' CO₂ as shown by the C-14 content is paralleled by a shifting of the carbon isotope ratio towards heavier values. This means that a possible influence of volcanic CO₂ on the C-14 content of plant material may be seen by its stable carbon isotope composition and may therefore be used in the proper selection of archaeological samples.

Another attempt was made to discriminate between samples which grew some distance away or next to volcanic emanations, using selected trace elements.

Based on the assumption that plants growing on the non-volcanic part of Santorini, i.e. mainly the area around Profitis Ilias and the Platanymos ridge, were most probably not influenced by volcanic gases either today or in Minoan times, we tried to determine characteristic trace element patterns for these plants. The patterns were expected to be reflected in archaeological samples.

For this reason selected plants were collected from the volcanic and the non-volcanic part of the island and chemically analysed by various methods. The detailed results obtained will be published elsewhere (Apostolakis et al., in preparation), the most important facts are as follows. Especially with help of the alkali and alkaline earth elements Na, K, Rb, Ca, Cs and Sr and the elements Zn and Cr, plants which have grown on the non-volcanic part can be clearly distinguished from those grown in the volcanic area. Unfortunately there was no opportunity to analyse the same elements in archaeological samples. We assume, however, that even in these samples the distinction between the volcanic and non-volcanic part can be done and therefore be useful in the selection of samples for radiometric age determinations.

New radiocarbon dates of archaeological samples from Akrotiri:

A set of eleven samples from the Akrotiri excavations was provided by Professor C. Doumas for radiocarbon age determinations (Table 2). Some of these were short-lived samples such as peas or grains; however, most of them were either not specified or could not be identified.

The C-14 content was measured at the Heidelberg laboratory using proportional counters following standard methods for sample pre-treatment and cleaning procedures (Levin et al. 1980; Schoch et al. 1980). The C-14 ages were corrected with respect to a laboratory offset, then calibrated according to Stuiver and Becker (1986). The values obtained by the crosspoints of the calibration curve are given in Table 2 (method A); also shown are those obtained through the transformation of the Poisson distribution (method B) (Kromer, pers. comm.).

The dates obtained by these measurements, as shown in Fig. 4, are broadly scattered with one sample way out of the normal range (2330-2470 y BC). The broad scattering of the data, although not satisfactory, is consistent with the broadly scattered values determined by others for Akrotiri materials (Biddle and Ralph 1980; Michael 1978, 1980; Weinstein and Betancourt 1978). As most of the material could not be identified, we could not differentiate between short- and long-lived material. It is nevertheless surprising that the two short-lived samples, Nos. 6059-7967 and 6058-5519, belong to those which are farther away from the mean. With the exception of one of these samples, all the others gave dates older than 1500 BC, resulting in a mean value of 1670 BC (only sample 7161-6868 was excluded for the calculation of the mean). This mean is consistent with that reported by Michael 1978.

DISCUSSION

As mentioned earlier, the first objective for our investigation was to find a reasonable explanation for the 'incorrect' radiocarbon dates presented at the second Santorini Congress in 1978. Based on the assumption that the destruction time of Akrotiri favoured by the archaeologists was correct, the influence of volcanic CO₂ on the C-14 content of present-day plant material was studied. Such an influence, causing 'ages' of up to 1500 years in recent plants, proved to be feasible at our model location, Laacher See, as well as on the island of Santorini. This effect, however, decreased rapidly with distance (some 10 to 100 m) away from the source. A similar influence was observed by Sulerzhitzky (1970) within distances of up to 4 km away from the volcanic vents.

It has to be assumed that the contamination of living plant material with 'dead' CO₂ occurs at Santorini not only in recent times, but probably also played an important role before the catastrophic eruption of the volcano of Thera. Even though it may be impossible to reconstruct the location of the sources or to quantify the amount of CO₂ gas which was emitted during Minoan time, we tried to define criteria which would allow us to select archaeological samples which were not affected by volcanic gases. 

The stable carbon isotope composition of the volcanic gases emerging at Palaea and Nea Kameni has been shown to be significantly different from normal magmatic CO₂ due to its generation through limestone decarbonization in the subground of Santorini. For this reason, these gases are also different from atmospheric CO₂ in their isotopic composition - i.e. the admixing of volcanic gases to atmospheric carbon dioxide can be detected with help of the stable carbon isotopes. This effect has been shown for several air samples from the Santorini islands. The carbon isotope ratio of Santorini clean air, sampled at safe distances from the gas vents, has a value of -9.8‰. Air samples collected in the neighbourhood of the gas emanations are clearly shifted towards heavier isotope values (up to -5.2‰) indicating the admixture of vent gases (Fig. 3).

Plants incorporate carbon in a plant-specific isotopic equilibrium with the carbon in atmospheric carbon dioxide. This means that the shifted carbon isotope ratios of contaminated air must be reflected in the isotopic composition of the plants which have grown in such an atmosphere. Three investigated plants show this effect; those containing. C-14 from 'dead' CO₂, are also shifted proportionally with respect to the stable carbon isotopes (Fig. 3). This permits the discrimination between volcanic CO₂ affected or not affected archaeological samples with help of stable carbon isotopes.

Unfortunately it was not possible to apply this method on archaeological samples during this study.

The second attempt to discriminate between samples growing in the vicinity of gas emanations or in unaffected areas through the elaboration of characteristic trace element patterns proved to be successful. The results obtained show a clear separation of plants growing on the non-volcanic parts of the island of Santorini from those growing on the volcanic parts, especially with regard to the alkali and alkaline earth elements as well as Zn and Cr (Apostolakis et al., in preparation). The applicability of these discrimination procedures should in future be tested on organic material from the archaeological site of Akrotiri.

The new radiocarbon dates presented in this study are consistent with those published by others (Biddle and Ralph 1980; Michael 1978, 1980; Weinstein and Betancourt 1978). They show the same scattering as observed by the above mentioned authors with a mean value in the range of about 1670 BC.

When studying the earlier radiocarbon dates as well as ours it was obvious that only very few dates fall within the range which is assumed by some archaeologists to be the time Akrotiri was destroyed - i.e. between 1550 and 1450 BC (Marinatos 1971; Warren 1984; and others). When we started our investigations, we tried to explain this discrepancy by the contamination of the samples with volcanic 'dead' CO₂. Our results have shown that this mechanism may play an important role in plants which have grown in the vicinity of volcanic gas emission. On the other hand, they have also shown that this effect decreases rapidly within some 10 to 100 m from the gas source and the growing plant. For this reason we assume that the incorporation of volcanic CO₂ may have influenced some of the samples, shifting their age towards older dates. In no case, however, can this process be responsible for the overall shift of such a large sample set to the date of around 1670 BC.

We conclude therefore that the catastrophic eruption which destroyed the village of Akrotiri happened in the 17th century BC, most probably between 1700 and 1630 BC. This date is still inconsistent with the archaeological assumption, but is supported by other arguments.

By dating an acidity peak in a deep ice core from Greenland at 1644 BC, Hammer et al. (1987) concluded that the Theran eruption took place in 1645 BC with an error limit of ± 20 yr. This ice-core date fits perfectly into the range obtained by the radiocarbon dates and therefore supports the above proposed destruction time.

Another argument for a 17th century volcanic event has been published by LaMarche and Hirschboeck (1984). By studying frost damage zones in annual tree rings, these authors date the eruption of Thera between 1628 and 1626 BC. This date was subsequently rejected by Warren (1974) who tried to show that the 1626 frost ring did not correspond with the Thera eruption.

Finally, discussing the archaeological time scale, Betancourt (1987) proposed a revised Aegean chronology, which fits almost completely into the picture obtained by the radiocarbon dates.

Looking back to the starting point of our investigations, where we tried to find explanations for the 'incorrect' radiocarbon dates, we now have to conclude with an opposite point of view.

The destruction of Akrotiri by the eruption of the volcano of Thera took place in the 17th century BC; the Aegean chronology has to be revised according to Betancourt (1987).

----------------------------------------

----------------------------------------