Evidence of Tephra in Soil Samples From Pyrgos, Crete

Soil from a large pot, which on excavation lay on a bench presumably having fallen during or subsequent to destruction of the house, contains conspicuous pyroclasts. The glass shards and some of the accompanying suite of minerals closely match those of the Late Minoan IA tephra of Thera. All stages of alteration of shards to clay minerals occur; such argillisation may help to explain the present sparseness of unaltered glass shards in the destruction deposits and may support the pre-existence of a more extensive tephra over eastern Crete, as indicated by offshore cores. Without taking into account the apparent time-lag between the LM IA eruption of Thera and the LM IB destruction by fire of Minoan Crete, the indications of tephra, even allowing for subsequent argillisation, are too slender to support direct incendiarism by airfall ash, though superheated gas flow and lightnng strikes cannot be discounted, if they occurred at least 30 years after the LM IA eruption of Thera. However there is insufficient evidence to postulate a causal relationship between the LM IA eruption of Thera and the destruction and abandonment of Pyrgos in particular and other habitations on Crete in general in Late Minoan IB.

INTRODUCTION

The British School at Athens began to excavate at Pyrgos (35^o 00 'N, 25^o 36 'E) near the village of Myrtos (eparchy of lerapetra) on the south coast of Crete, in 1970 (Fig. 1). The history of the Minoan settlement has been divided into four periods (Cadogan forthcoming). We shall be discussing soil samples taken from levels of Pyrgos III, a period corresponding to that of the Old Palaces of central Crete, and Pyrgos IV, which is mostly or wholly Late Minoan I. It is not yet clear whether there is a phase at the end of Pyrgos III or beginning of Pyrgos IV to correspond to Middle Minoan III of central Crete.

Parts of a large country house, now assigned to Pyrgos IV, soon appeared. The house had been destroyed in a catastrophic fire, which had solidified mud bricks of the superstructure, splintered ashlar masonry and vitrified pieces of pottery. The disaster came in Late Minoan IB, on the evidence of a Marine Style jug, an Alternating Style bell-cup and a stirrup jar of LM IB shape (although decorated in the LM IA manner), which were in the debris. The School has also found a little LM IB pottery in the settlement around the country house, where we have not observed any traces of burning except in one house on the North slope immediately below the brow of the hill: what was there could easily have fallen from the country house which is only a metre or so above. In that house (in trench G2) was a LM IB globular alabastron, pierced in the base, which had been broken into over twelve hundred pieces, as if by vandals. There is other evidence which might suggest vandalism: a fine alabaster lid was found among outbuildings of the country house, as if it had been dropped there by a looter or by somebody fleeing; and a piece of a clay bowl was found a considerable distance from joining pieces, which suggests that somebody threw it away - though that need not have happened at the time of the destruction.

During the excavation samples were taken for possible pyroclastic materials. Collaboration between the archaeologists and the Institute of Geological Sciences came from discussions of the (then) Director of the Institute (Sir Kingsley Dunham, F.R.S.), the Hon. Mrs Henry Hankey and Cadogan. Harrison has been responsible for the work at the Institute.

The first samples yielded unmistakable glass shards, though in trace amount, exactly matching those of the Upper Ash layer (Late Minoan) of Thera (Ninkovich & Heezen 1965, 1967; Keller & Ninkovich 1972). Some of the accompanying heavy minerals in the Pyrgos samples could also relate to the Upper Ash layer. These preliminary results were reported by Cadogan et al. (1972) and, though the pyroclasts were present in the destruction layers of the country house in insignificant quantity, their very presence, identity and approximate contemporaneity with the Theran tephra led the authors to suggest that the ash was blown over Crete in Late Minoan IA, when the abandonment and subsequent catastrophic eruption of Thera occurred (Doumas 1974). The evidence, however, has been used to support the theory that the eruption took place in Late Minoan IB, with a date c. 1470 B.C. (Luce 1976), leading to the destruction of the Minoan settlements in Crete, as Marinatos (1939) suggested. Independent archaeological research on soil samples in Crete (e.g. Vitaliano & Vitaliano 1974) has confirmed the presence of volcanic glass shards (n = 1.509 ± .002) in trace amount in more than half of the samples collected, and associated with sanidine, plagioclase, pyroxene and euhedra of zircon. Those authors suggest that the eruption of Thera could not have occurred much later than the end of Late Minoan IA, and that little time elapsed between the abandonment of Thera and the ashfall over eastern Crete. They give few details of the archaeological contexts of their samples.

Because of the possibility of contamination of soil samples by airborne ash particles blown in during excavation, further samples were collected during October 1973, with scrupulous care taken to seal the soils immediately on recovery. Because dust collected in pithoi and other jars prior to their burial might give a more accurate record of a contemporary ash fall , additional samples were taken in 1976 of the contents of jars, which had been kept from their excavation in 1970 and 1971. This communication, therefore, places on record the petrographical and mineralogical data relating to the two more recently analysed sets of samples (referred to as 1973 and 1976) and discusses interpretations from the geological and archaeological viewpoints. We also include full data of the original (1971) group of samples.

Almost all the samples (Tables 1, 2, 3) come trom the LM IB destruction debris of the country house (Pyrgos IV). The exceptions are ENQ 2389 from a LM I level on the east side of the Pyrgos IV settlement, where there was no trace of destruction by fire; ENQ 2388, a lump of pumice, and ENQ 2391 from the MM levels on the west slope of the hill (Pyrgos III); and ENQ 2392 from just above the floor of the smaller cistern of Pyrgos III (which was later filled with river pebbles), and should then probably be assigned to Pyrgos III. The sites of the samples are shown in Figure 2 and Plates 1 and 2, and are listed in Tables 1, 2 and 3.

PETROGRAPHY AND MINERALOGY

• Sample treatment.

Immediately on collection in the field, the samples were individually labelled and sealed in polythene bags and enclosed in further polythene bags.

These were removed as each sample was examined, except for the single sample of pumice (ENQ 2388) which remained sealed until all the other samples had been examined. Each sample was treated as follows: composite grains were lightly disaggregated and subsamples treated with dilute acetic acid buffered with ammonium acetate, washed, dried and sieved through 30-, 100- and 240- mesh B.S.S. Each (-30 + 100) mesh fraction was then separated in bromoform (SG = 2.9), and all examined microscopically by immersing random micro-samples in calibrated refractive index liquids. The (-240) mesh fractions of the 1973 samples were also analysed by X -ray diffractometry to identify the crystalline phases.

• Results.

A brief summary of the analyses of the 1971 samples (Table 1) is given in Cadogan et al (1972), and the resulting constituents of the fine sand fractions are listed here in Table 4. Traces of glass shards (refractive index n = 1.509 ± 0.002) were detected in size-fractions below 0.3 mm in three out of the four samples. A trace of pale brown glass (n = 1.520 ± 0.002) was also detected in one sample (ENQ 2325). A selection of shards is shown in Figure 3. Their sharply angular, unabraded profiles and in places attenuated vesicles, are characteristic of comminuted volcanic pumice. The refractive index of the majority (1.509 ± .002) agrees with that of the LM IA eruption of Thera, though the trace of glass of higher r.i. suggests admixture from presumably the Würm eruption of Ischia (> 25,000 yr BP).

The major and minor components of the 1973 samples were similarly identified and estimates of relative abundances made visually (Table 5). The major components of the soil samples are classified as polygranular rock particles (sandstone, siltstone, compacted clay, serpentinous material), black chert, jasper, minerals (quartz, feldspar, carbonates, gypsum, hematite, magnetite, pyroxenes, amphibole, rutile, micas, clay minerals, monazite), artificial (pottery, ? mortar, tile/brick), and organic (charcoal, animal shells, fibres etc.). Of these, calcium carbonate, mainly in granular form, predominates together with gypsum, and these were probably derived from building materials, and floor coverings. Gamma calcium sulphate may be secondary. Some possible sources of these groups are discussed below; the main components are generally similar in each sample though the proportions differ. Of the trace constituents, particles of undoubted volcanic glass shards occur in the finest sieve fraction (100 - mesh: about 0.15 mm) of half of the samples. In one sample (ENQ 2387), glass shards are relatively conspicuous, but this may be of little significance in terms of locality, in view of the sampling bias involved in such minute quantities. All glass particles have a refractive index of n = 1.509 ± .001 at 20° C thus agreeing with those detected in previous samples (Cadogan op. cit. p. 312), and those identified from Crete by Vitaliano & Vitaliano (1974). The shards rarely attain 0.2 mm across, average about 0.1 mm, are very angular and unabraded showing delicate projections, principally colourless (rarely pale brown) and variably charged with elongated bubbles (Plate 3). A euhedral crystal of apatite occurred in one shard, and other shards show turbid microinclusions of possibly devitrification products. Qualitative analyses of individual shards by an X-ray energy dispersive attachment to a scanning electrom microscope show major Si, subordinate Al, minor K, Fe and traces of Na, Ca and Ti. Variation between individual shards, so far as the limitations of the technique permit shows traces of Na in four out of six particles (Table 6).

Analyses by X-ray diffractometry of the finest sieved fractions (< 240- mesh BSS) of each granular sample are summarised in Table 7 which gives approximate quantitative data for the clay and other minerals present. Clay predominates and of the clay minerals, clay mica is most abundant followed by montmorillonite, chlorite and variable kaolinite. The non-clay minerals are generally similar in types and proportions between the samples which include the ash and carbonised soil. All of the minerals are generally stable at the temperatures of open fires. Montmorillonite is a common associate of altered volcanic ash and tuff and its conspicuous presence in all of these samples might well suggest altered tephra. It has not been possible, on account of minute particle size, to analyse individual glass shards in these samples and to determine any clay alteration products which might be present.

The sources of the constituents (other than glass shards) are probably complex including tephra, local building materials, disintegration products of artefacts, other adventitious materials and local rocks. A specimen of burnt brick (ENQ 2393) is an aggregate of composite granular particles including silty, sandy and clayey micritic limestone with microfossils, hematite, calcisiltite, calcareous mudstone, ferruginous silty mudstone, calcareous lithic sandstone, chert, quartz and a trace of amphibole. But no glass shards, pyroxene or feldspar were detected either in the samples of Pyrgos III (ENQ 2391, 2392).

The fragment of pumice (ENQ 2388) is also from Pyrgos III; and one of four fragments from the same context. It is rounded (measuring 3.5 X 2.5 X 3.0 cm major axes) pale grey-buff and powdery on the exterior, but near olive-black 5Y2/1 (Colour Code of the Geological Society of America) and highly vesicular on a freshly broken surface. Vesicles and glass sporadically contain feldspar crystals (oligoclase near Ab₇₅ An₂₅), hypersthene (X = green, Z = brown), hematite and an unidentified colourless mineral of high refractive indices, low birefringence and 2V (+) near 75°. Vesicles also are lined with clay material. The glass has a refractive index of n = 1.509 ± 001 which thus matches exactly the discrete shards in the soil samples except for a pale brown colour in thin splinters. Semi-quantitative X ray energy dispersive analyses were made on random splinters of this pumice fragment, after optical checks for freedom from inclusions. Scans of six glass shards give major Si, subordinate Al, minor K, Fe, Ca and traces of Na, Ca, Ti. Inter-shard variation shows slight differences in proportions of Ca, from minor to trace (Table 6). Bearing in mind the statistical limitation of the small population analysed and the constraints of the technique, the glass of the pumice fragment appears to match very closely the shards from the soil samples. Of the latter samples, the glass shards, hypersthene, plagioclase and possibly apatite almost certainly stemmed from tephra or possibly from disaggregated pumice fragments of unknown origin. The discrete glass shards and the pumice fragment exactly match Late Minoan IA tephra of Thera and the contemporary Upper Ash layer of the Eastern Mediterranean (Ninkovich & Heezen 1967). Other constituents of the samples were probably derived from local soils or rock samples. Serpentinous and metamorphic rocks crop out in the vicinity of Myrtos (Bonnefort in Wagstaff 1972, p. 275) and these could account for serpentine, amphibole, garnet, epidote, monazite, pyroxene and possibly some of the albite detected, as well perhaps as some (if not all) of the smectite, although a primary volcanogenic ashfall origin of the latter cannot be ruled out. The serpentine and jasper may have stemmed from debris of vases etc., or from local industries producing such goods. The polygranular rock particles were mainly derived from building materials as were the carbonates and calcium sulphate. The sample of ash residue (ENQ 2391) of Pyrgos III while containing traces of the pumice-associated minerals - feldspar and pyroxene - contains no obvious glass shards. If present, the latter must be exceedingly rare suggesting that there was no concentration of tephra in this sample sufficient to cause incendiarism. A carbonized sample (ENQ 2396) of Pyrgos IV is a medium to dark gray (N5 - N6) carbonaceous and calcareous silty clay, containing burnt brick particles, charcoal, and traces of glass shards with associated apatite and pyroxene. There is no obvious concentration of pyroclastic materials, as might otherwise be expected were these the direct cause of fire.

Since it seemed possible that evidence of a contemporary airborne ashfall might well be preserved in pithoi or other vessels open and upright at the time, five samples were collected in 1976 (Table 3) by Cadogan from soils which had been kept since they were excavated in 1970 and 1971. Microsamples of the non-biological materials in each were examined microscopically. Sample ENQ 2490 consists of light buff - gray particles up to 0.8 cm across with much matrix of similar colour, and differs from the other samples. Micritic calcium carbonate predominates with trace amounts of quartz splinters, opaque dust (? carbon), and heavy minerals. Volcanic glass shards are present in trace quantity as colourless, very angular particles averaging about 0.1 mm, but rarely attaining 0.3 mm across. All stages from clear and transparent, to turbid, vesicular and highly altered (argillised) are present. Margins of many shards are devitrified and weakly anisotropic. Vesicles may attain 0.01 mm mean diameter but whether gas - or liquid - filled is unknown. The refractive index of shards is 1.509 ± .002 and little variation from this is apparent, in the other shards. Heavy minerals concentrated in the insoluble residue of acid dissolution of the calcium carbonate include hypersthene, amphibole, pyroxene, apatite and chlorite. Other minerals include probable gamma calcium sulphate and pale brown prismatic hemimorphs of unknown identity. The overall suite therefore closely matches that in the 1973 soil samples from LM IB destruction levels. An X-ray diffractometer trace (DX 1757) of the clay fraction of this sample shows major illite, subordinate smectite and chlorite, and lesser kaolinite with minor quartz and feldspar. A broad hump in the background between 17^o and 33^o 2Θ suggests that amorphous carbonaceous material may also be present.

Under the scanning electron microscope the glass shards which microscopically appear to be clear and unaltered, show extensive surface pitting and alteration (Plate 3, fig. 6). Semiquantitative analyses by X-ray energy dispersion of these selected glass shards, are compared in Table 6 with analyses of shards from the other samples which show a degree of chemical alteration with some apparent loss in SiO₂.

None of the other soils was found to contain shards as conspicuous as those in the first sample, ENQ 2490. Their main constituents are as follows:

ENQ 2491 The soil matrix to the predominant carbonised seeds consists of opaque (carbonaceous) particles up to 0.5 mm across, calcium carbonate, with traces of quartz, chlorite, calcium sulphate, fibres, and one possible glass shard.

EXQ 2492 This consists dominantly of granular and micritic calcium carbonate pellets and plates up to 1.5 cm across, with opaque carbonaceous dust. No shards were observed.

ENQ 2493 This contains predominant creamy white granular and micritic calcium carbonate, with splinters of quartz and opaque dust, but no obvious pyroclasts.

ENQ 2494 Carbonaceous particles dominate with specks of quartz and finely granular calcium carbonate. One possible glass shard (0.1 mm across) was detected amongst many thousands of grains.

DISCUSSION

The analyses of archaeological soil samples from Pyrgos confirm other discoveries of Late Minoan pyroclasts in Eastern Crete (Vitaliano & Vitaliano op. cit), which occur apparently only in trace amount. They are positively identified by glass shards whose refractive index exactly matches that of the Late Minoan IA tephra of Thera, and the general occurrence of distinctive accessory minerals (hypersthene, apatite, sodic feldspars) supports this interpretation.

However there is much admixture and dilution by materials locally derived from building materials and other sources.

The most conspicuous shards were found in the fallen jar (ENQ 2490) many of which show variable devitrification and argillisation and extreme friability.

The physical and (so far as they can be determined) chemical properties of these shards match those of the other shards in the destruction deposits. The major smectite in the clay fraction of the jar residue and in the destruction deposits indicates perhaps a major breakdown of pyroclasts to clay material, though the ultimate origin (i.e. primary airfall ash or derived) cannot be established. There remains a possibility therefore of an original tephra more extensive than than indicated by the glass shards, the ubiquitous feldspar and heavy minerals alone.

The discrete fragment of pumice (ENQ 2388), however, which exactly matches the dispersed shards as well as the Theran LM IA pumice, is enigmatic because its size rules out a primary airfall origin. The piece is possibly intrusive in its Middle Minoan context and originated from the Thera eruption of LM IA times. As is well known rounded fragments of pumice (n = 1.509 ± .002) containing alkali feldspar and orthopyroxene are very common around the beaches, and it is highly likely that continual abrasion and disintegration results in fine glass particles which strong winds could carry well inland. So the sparse glass shards may have been so carried at any time from the house's occupation to its subsequent burial but the frequent occurrence of pyroclastic heavy minerals in the destruction deposits and their presence in a jar might provide other evidence, for example of primary ashfall and/or derivation from humanly transported LM IA pumice.

SUMMARY AND CONCLUSIONS

Soil samples carefully sealed against possible contamination from Late Minoan IB destruction levels and from a Late Minoan I level have yielded unmistakable yet minute amounts of pyroclastic glass shards which match those from the Late Minoan I eruption of Thera some 100 km N of Crete. Accompanying the shards are certain heavy minerals and feldspars which match the overall suite of the Theran pumice. The clay fractions contain major smectite. Soil residues trapped in a jar showed conspicuous glass shards exactly matching the others but also showing varying stages of breakdown to smectite. Thus a more definite tephra layer may have existed.

Even if the discrete shards and other pyroclasts represent a primary airfall contemporaneous with the destruction of the Late Minoan IB building, the amount (even allowing for subsequent smectitisation) would probably be insufficient to cause ignition, as Bond & Sparks have indicated (1976). However volumes of pyroclastic materials are not themselves a priori necessary to cause ignition of property and, on the contrary, results of volcanic eruptions have, in catastrophic cases such as Mont Pelée and St. Pierre (1902), been caused by superheated gases charged with incandescent particles in a nuée ardente. In that case, the nuée deposited only a few centimetres of ash on the town itself. Ignimbritic flows, which may extend up to a distance from source of 150 km (as in Nevada) may equally be associated with superheated gas flows. So it is possible that very hot gases themselves, more or less free of significant ash particles, descended through atmospheric turbulence onto eastern Crete. Again, since much volcanism is associated with extensive lightning discharge (as in the Krakatoa eruption of 1883), a further potential source of ignition cannot be ruled out.

It is perhaps relevant that according to the report on the 1925 - 26 eruptions of Thera (Georgalas & Liatsikas 1927) which formed the new central dome of Daphni, gases and vapours were discharged during explosive phases. These nuées reached a height of 700 m after depositing their solid load of pyroclasts.

Spectroscopic analyses showed Na, H, S, C, O, Fe, N. If we extrapolate this quite minor eruption to the catastrophic explosion of Thera, which was exceptionally violent with enormous kinetic and thermal energy (Hédervári 1968; Bond & Sparks 1976), incandescent gas flow containing superheated gases might conceivably have descended over eastern Crete.

Despite the violence of the explosion of Thera, we suggest that there is not enough evidence to conclude that the explosion, in any of its stages, caused the destruction and abandoment of Pyrgos, or of any other places in Crete which suffered similarly in Late Minoan IB. Our reasons are:

1. The evidence of fire in the country house, and the lack of it in the settlement, at Pyrgos hardly fits an overwhelming natural disaster, but seems rather the work of humans. The reverse evidence at Knossos, where the palace survived and buildings in the town did not (such as the South House, or that excavated to the North of the Royal Road by Hood), supports the interpretation, just as the possible evidence of vandalism at Pyrgos, and similar evidence noted by Hood (1970) at other sites in East Crete, would do. Equally there is no archaeological evidence for a tsunami, at Pyrgos or elsewhere in Crete: at Amnisos, the one suggested case, we do not know when the large blocks were tipped askew. We do not have any geological evidence for a tsunami either (Keller & Ninkovich 1972), nor anything to tell us when and how the caldera of Thera collapsed except that it happened after the last stages of eruption (Bond & Sparks 1976). If we exclude human agency, then the 1926 earthquake in Crete may suggest that there was an analogous disaster in Late Minoan IB (Pichler & Schiering 1977); but we find it difficult to exclude humans altogether. 
2. The evidence from Akrotiri, Thera suggests that the eruption followed quickly the abandonment, re-inhabitation and final abandonment of the settlement (Marinatos 1971; Doumas 1974). The eruption took very little time (Bond & Sparks 1976). The latest pottery in the settlement is Late Minoan IA, and there is nothing reported of Late Minoan IB among the hundreds of Cretan vases, nor of the contemporary Late Helladic IIA, by contrast with other island settlements under Cretan influence which all have LM IB and LH IIA pottery in their destruction or abandonment levels. The table of offerings with dolphins from Akrotiri is quite comparable to the town's frescoes, and unlike LM IB Marine Style vases (pace Luce, 1976). The sole vase of LM IB style (Buchholz & Karageorghis 1973, I11. 948) in the Thera Museum has no context, and is not enough to set against the very many LM IA vases at Akrotiri. Finally one may not amalgamate or approximate the two styles, as Bolton (1976) wishes.

To us the evidence shows that the eruption took place in Late Minoan IA  (c. 1500 BC), which must mean at least 30 years, and probably more, before the Late Minoan IB destructions in Crete (Table 3A). There is some archaeological evidence in Crete to link with trouble in Late Minoan IA (Hood 1970).

When the volcano erupted, its effects on Crete may (Boekschoten 1971) or may not (Pichler & Schiering 1977) have been serious. They were not so serious that civilised life was impeded for long. Bond & Sparks (1976) suggest that the maximum size of particles to reach Crete was only about 2 mm; and we may speculate whether this ash was not just brushed and scraped off the fields, and farming continued. Some Theran ash continued to lie on the ground, and we have traces of it at Pyrgos in Late Minoan IB contexts of a generation or more after the eruption. We have also one shard of the 25,000 BP eruption of Ischia, which has nothing to do with the destruction of Pyrgos.

(a) Analyses of glass shards (Plate 1) (ENQ, 2387)

Refractive index (n) = all shards = 1.509 +/- 002

(b) Analyses of particles taken from a pumice fragment (ENQ, 2388)

(c) Analyses** of shards from deposit in jar (ENQ,2490) fallen on bench Shard 1234

Semiquantitative analyses* by X -ray energy dispersive analyses of

(a) shards separated from soils, of (b) particles of pumice (1973)

(c) shards from deposit in jar fallen on bench (1976)

*By Mrs A E Tresham (Institute of Geological Sciences, London )

**By Dr M T Styles (Institute of Geological Sciences, London )

From the evidence at Pyrgos and elsewhere in Crete, we can find little to suggest that the Late Minoan IB destructions were caused by volcanism. On the other hand the evidence is still slight for a Late Minoan IA destruction on Crete, contemporaneous with the eruption of Thera, or from human remains in settlements, to indicate either a natural or a humanly-induced catastrophe.

We should like to make two closing observations:

1. The Theran pumice pebble (ENQ 2388) in a Pyrgos III context does not have the same refractive index as that of the Late Glacial Bu pumice of Thera (n = 1.514 : Keller & Ninkovich 1972). It would seem to be pumice either from a minor eruption of Thera (as yet unrecorded) with similar physico-chemical characteristics to that of the LM IA eruption, or to be from the LM IA eruption and, in excavators' terms, to be an intrusion - perhaps washed down from the LM I houses higher up the slope. The soil sample from the same trench contained no volcanic ash.
2. We suggest that argillisation may have virtually destroyed any original landfall ash textures, whereas they have been preserved when deposited in the sea. There is, however, little evidence of substantial argillisation of the LM I tephra on Thera itself,- and further investigations of this theory are necessary before firm conclusions are reached.

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

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