Radiocarbon Dates from the Site of Akrotiri, Thera, 1967-1977

When the first radiocarbon date from Thera - wood from a trunk of a small carbonized tree from the Phira mines - was published in 1968 (Marinatos 1968, pp. 55 - 60), it created mixed reactions. With the relatively crude correction factors of the day all that could be stated was that there existed about a 67 % probability that the date fell between 1520 B.C. and 1750 B.C. More recently determined correction factors (Ralph, Michael and Han 1973) refined the date to 1870 - 1770 ± 40 B.C.

Up to this writing (May, 1977) all of the published radiocarbon dates of samples from the island of Thera, save one (L - 362), have been processed in the Radiocarbon Laboratory of the University of Pennsylvania and all have been calibrated on the basis of the 1973 tables of correction factors developed in the same laboratory. The dates were calculated with the generally accepted 5730 half-life.

Altogether, 27 samples from Thera, were obtained by the Pennsylvania laboratory. Of these four could not be processed because they were either too small or contained little or no carbon. Of the remaining 23 samples, two were not from the site of Akrotiri: P - 1401, the carbonized tree mentioned above, and P - 1697, carbonized "fava" beans from a mining installation between the villages of Akrotiri and Meghalochori. The latter sample was found in a large jug buried beneath a layer of pumice at the edge of the preciptous side of the caldera. It dated to 1420 - 1400 ± 60 B.C. The only sample not processed by the Pennsylvania laboratory, L -362, also came from a pumice quarry at Phira and dated to 1750 - 1710 ± 100 B.C. The elimination of the above samples leaves 21 belonging to the site proper (see Table I).

The Akrotiri samples were collected over the past nine years by the late Spridon Marinatos and by Christos Doumas. They lend themselves to division into two major classes:

(1)     Long-lived, such as charcoal or sizeable pieces of wood. As a rule these are expected to date earlier than the archaeological context in which they were found;

(2)      Short-lived, such as beans, grains, or small branches of shrubs. These are expected to be contemporaneous with the archaeologically determined dating.

The objective of this study of sample dates is to try to resolve the dating of the destruction level at Akrotiri, that is, of the Thera volcanic eruption about which there is some controversy (Acta 1971). For this reason we feel justified in eliminating those samples which obviously do not belong to the destruction level.

There are four such samples:

(1)      P - 1601, charcoal (Olea sp.) from Trench Arvaniti 3, found in association with "pure" Middle Cycladic III - Late Cycladic I context. The C-14 date is 1730 - 1690 ± 60 B.C.;

(2)      P - 1602, charcoal (Pinus sp.) found in the same context as P - 1601 and dating to 1870 - 1720 ± 40 B.C.;

(3)      P - 1891, charcoal from pit dug for modern roof pillar no. 11 within structure B. Its date of 2000 - 1960 ± 70 B.C. is consistent with the archaeological picture under the destruction level;

(4)      P - 1893, charcoal (Pinus sp.) from 2 meters below the destruction level in area E of Room 4 in Structure Delta. The date is 2600 ± 70 B.C.

Thus we end up with 17 samples which come into play in our attempt to date the destruction which traditionally is cited as having taken place in the period 1550 - 1500 B.C. (Marinatos 1939; Marinatos 1971; Hood 1971, et al.). Of the 17 samples, 7 are long-lived and 10 are short-lived (Table II).

Unfortunately, all of the long-lived samples, save one, were undersized, with the result that their statistical tolerance is large since they had to be processed in small counters. This makes the date less reliable. Thus the long-lived samples tell us little, since they are also inconsistent within themselves, viz., three dates support a later chronology for the destruction (P -1619, P -2562, P -2566), three support the traditional date (P - 1599, P - 2563, P - 2564), and one an earlier chronology (P - 1890). As a matter of fact, with its large sigma, P - 2563 supports both the traditional and the early chronology.

Turning to the short-lived samples we find a more consistent pattern, yet one with grave problems. When the first six dates were presented last year (Michael 1976), five of the six were found to be internally consistent and supported an early chronology. Seeking a confirmation of this chronology, we acquired additional samples from Dr. Doumas in 1976. These appear in Table II as the 2500 series. The scattered, inconsistent dates of the long-lived samples were noted above. The four new short-lived samples divided equally into two supporting an early chronology (P - 2559, P - 2565) and two being disconcertingly early by several hundred years (P - 2560, P - 2561). With eight of the ten short-lived samples supporting an earlier chronology by some 100 - 150 years - their average comes to 1660 B.C. - one is tempted to give considerable weight to this result. However, there are several facts which deter one from doing so. First, the lessened reliability of the undersized long-lived samples. The one date of a full-sized sample that does support an earlier chronology, P - 1890, could be challenged on the basis that the charcoal may have been derived from the inner rings of an old tree - a common occurence. Second, 50 % of the short-lived samples were shrubs. They were collected from the floors of several buildings in the Akrotiri complex. Since the ceilings of Theran buildings were probably constructed with a wooden framework filled with compressed shrubs, perhaps fascines, and rubble, and covered with clay, it could well be that the shrub samples were derived from the collapsed ceilings. Thus their radiocarbon dates may refer to the construction of the ceiling rather than the destruction of it. Third, there are the two anomalous dates of grains found in jars on the destruction level - P - 2561 and P - 2560. It has been noted that gaseous emanations from volcanoes influence radiocarbon dates of vegetation in their path (Chatters et al., 1969).

In Hawai, for instance, the age error introduced by vented gasses varied with the distance from the vent or fumerole of the growing vegetation. Thus the emission of volcanic carbon dioxide ("dead carbon") on Thera over relatively long periods of time may have depressed the radiocarbon levels of the plant when incorporated during its photosynthesis - with resultant older radiocarbon dates.

Yet, even this suggestion does not resolve the problem fully. All of the P - 2500 series of samples were excavated during 1976 and came from various storage jars in the same room. We have seen the diversity of the eight dates. Could it be, however, that grains stored in one jar came from a field affected by gaseous emissions, whereas those in other jars were grown on unaffected fields? These and other questions will have to be resolved before a final evaluation of the Thera dates can be made.

Because of the geological history of Thera, unusual local conditions - perhaps affecting only part of the island - may have existed sometime during the middle third of the second millennium B.C. To document this will take a careful, comparative study. And since the chronological periods we wish to establish and differentiate are, or seem, relatively close to each other, only short-lived samples should be used. They should be collected under carefully controlled conditions, preferably by both archaeologist and radiocarbon specialist.

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

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