On the Relation Between the Thera Eruption and the Destruction of Eastern Crete, c. 1450 B.C.

The principal obstacle is an apparent difference of some decades in the dates of the pottery in the destruction-levels in the two places; and a number of different solutions to the problem have been propounded.

The present paper summarises these solutions and gives reasons for rejecting all but one of them. It is maintained that the positive evidence in favour of this one is satisfactory, and that certain objections alleged against it are illusory.

About 1450 B.C. the prosperous civilisation of the eastern half of Minoan Crete came to a sudden end. Palaces, towns, and country mansions were not only destroyed but also abandoned. A few sites were re-occupied on a smaller scale for a generation or two; most have remained uninhabited from that day to this.

Prolonged desolation of a prosperous region is abnormal, and it is natural to try to relate it to another abnormal catastrophe - a huge and more or less contemporary eruption at Thera. But a theory of cause and effect is not easy to reconcile with a discrepancy between the ages of the pottery in the destruction-levels in Thera and in Crete. The Theran pottery is judged to be a generation or two older than the Cretan; in particular, a distinctive style of decoration, the Marine Style, already highly developed in Crete, is almost if not quite wanting in Thera. The basic question is therefore whether a satisfactory explanation can be found for this apparent discrepancy.

Six different answers have been given. I summarise them, briefly stating my reasons for rejecting all but the last.

- I -      That the eruption occurred in two phases at least a generation apart, Thera being buried in pumice and ash c. 1500 B. C., Crete being devastated by ash-fall c. 1450 B. C.

The rejection of this theory by the vulcanologists at the First Thera Congress appears decisive ⁽¹⁾ and I say no more abouf it.

- II -     That there was an interval of at least a generation between the volcanic eruption and the formation of the caldera, the former burying Thera and the latter creating flood-waves sufficient to cause havoc in eastern Crete (Doumas 1974, 112 - 113; Vitaliano 1971, 42 - 93; Hiller 1975, 49).

This theory said by Thorarinsson to be 'very improbable in itself' (1971, 231; cf Hays 1971, 134) is exposed to fatal objections. First, the condition in which buildings, and objects in and near them, were actually found by the excavators of such exposed north-coastal sites as Mallia, Nirou Khani, and Gournia, proves beyond question that these places were never struck and submerged by tall flood-waves at high speed. I do not now repeat the evidence which I have given elsewhere in some detail ⁽²⁾ . Secondly, if the eruption occurred c. 1500 B.C., its effect upon Crete must have been negligible; for there is no indication, let alone proof, of damage from flood or ash-fall at that time, nor was any place abandoned. On the other hand the abandonment of eastern Crete c. 1450 B.C., which can no longer be referred to the ash-fall, must now be ascribed to human causes, presumably the systematic destruction of all places and the annihilation or expulsion of all inhabitants, by force of arms. I have given elsewhere (1970, 12) reasons for rejecting absolutely so improbable an explanation of the prolonged desolation of eastern Crete.

- III -     The same objections apply to the third answer, that there was no connection whatsoever between the eruption of the volcano, which occurred c. 1500 B.C., and the desolation of eastern Crete some fifty years later (Hood 1973, 111 - 118; c.f. Hiller 1975, 46. 61. 70).

It is indeed certain that some places in eastern and central Crete were damaged by tectonic earthquake in L.M. I A (Hood 1971, 380 - 383), but this theory has to concede that the eruption, if it occurred at that time, had no discernible effect upon Crete. The ash-fall must have been negligible, for no place was abandoned; flood-waves did no visible damage. ⁽³⁾ The desolation of eastern Crete fifty years later has to be referred to human agency. The most serious objection to this theory, as to the preceding, is the denial of the presumable effect of the ash-fall. Eastern Crete lay under the path of the ash-cloud ⁽⁴⁾ at a distance of 110 - 150 km. from the volcano. Now the much smaller eruption of Krakatoa deposited ash on the deck of the vessel Norham Castle at the rate of 45 cm. in an hour, 110 km. to windward of the volcano (Symons 1888, 269); the Tweed measured 17 ½ cm. at 600 km (Symons, 264 f. 447f).

The ash-fall from Katmai in 1912 lay 20 cm. deep at a distance of 160 km. and 'caused extreme damage to buildings and crops' (Thorarinsson 1971, 231), and tephra from Tambora in 1815 lay 60 cm. deep over western Sumbawa at 200 km., where 38,000 people died from starvation or disease (Van Padang 1971, 57. 60). Thorarinsson has told us that in Iceland a depth of 20 cm. in highlands and 30 - 50 cm. in more fertile lowlands has caused abandonment for decades (1971, 213; cf. Hiller 1975, 54) and that the soil would probably regain fertility more slowly in Crete than in the climate of Iceland (1971, 232).

In the light of the above examples it is practically certain that the deposit on eastern Crete, which lay under the ash-cloud from the stupendous eruption of the Santorini volcano, cannot have been less than 30 - 50 cm. deep, and must have caused abandonment of the region for at least a decade or two. 

I am of course aware that the presence of Santorini tephra in an apparently sealed L.M. I. A deposit at Zakro may be used as an argument against me. If the desolation of eastern Crete is to be related to the ash-fall, Santorini tephra should be found in the destruction-levels of L.M. I. B, not in layers of L.M. I A; and so indeed they have been - in L.M. I B strata at Pyrgos (Cadogan 1972, 310 - 313), Zakro, Gournia, Mallia, and Vathypetro. ⁽⁵⁾ In layers which are certainly L.M. I A, on the other hand, two samples have been identified at Zakro; one of these comes from a layer which 'appears to have been sealed' before the L.M. I B phase began, and both are believed to have been 'lying essentially where they fell' (Vitaliano 1974, 24). There is, however, good reason for the caution which Cadogan and the Vitalianos recommend. The samples from archaeological sites are very small and very few, and there is a built-in obstacle to their use as evidence : if you find Santorini tephra in a L.M. I B context, you must allow that it may have fallen in L.M. I A, remaining in situ or being trodden in by foot or blown from one place to another by the wind; if you find it in a L.M. I A context, you must allow that it may have sunk from a higher to a lower level by natural or artificial disturbance of the surface. It is not clear how any layer can be so 'sealed' as to preclude the possibility of the movement of microscopically small particles in the course of 3,500 years.

- IV -      That there is no problem, because there is no chronological gap to be filled. This theory has two quite different forms :

(a)      That the absence of Marine Style signifies merely that Thera was behind the times, not keeping pace with developments in Crete. ⁽⁶⁾

The principal objection is that contemporary Cretan settlements overseas, notably in Cythera and Ceos, were not behind the times and did keep pace; why should Thera, a prosperous community and very close to Crete, be the only one which lagged so far behind? I doubt if this theory has many, if any, supporters nowadays, and I say no more about it.

(b)     That the Marine Style, whose alleged absence has caused most of the trouble, was in fact familiar in Thera, and that the basic problem is therefore illusory.

The best evidence for this is a table for offerings found on a window-sill in West House at Acrotiri (Marinatos and Hirmer 1973, 62, pl. XLIV; 19 col. pl. C; Hiller 1975, 59, pl VI). Marinatos described it as in 'a pure and already advanced Marine Style', which shortens, though it does not close, the gap between Thera and Crete. The discovery of this table should serve as a warning. More objects in this style may yet be found, and they may be of a character to narrow further or even to close the gap. Let us remember that sixty years elapsed before the first pure deposit in the Marine Style was found at Cnossos itself (Hood 1961-2, 25). Moreover, much of the local L.M. I. A pottery at Acrotiri is decorated in the rich floral style characteristic of East Crete and perhaps persisting there into the period of L.M. I B (Coldstream 1969, 150).

This answer would be likelier than any other, if it were not ruled out by the last solution, to which I now turn.

- V -      That there was an interval of at least one generation between the destruction of Acrotiri by tectonic earthquake ⁽⁷⁾ and the volcanic eruption which buried Thera and devastated eastern Crete. (Galanopoulos 1971, 195; Money 1973, 50 f; cf. Hiller 1975, 46. 50).

Four episodes in the tale of the doom of Acrotiri have been clearly defined :

1. The town suffered relatively minor earthquake-tremors. All over the site there is plain evidence of the removal of stores and utensils to places thought to be proof against shock (Marinatos passim; Doumas 1974, 110).
2. The town was very severely damaged by an earthquake of great violence. As no bodies have been found at Acrotiri, and as almost all valuables have been removed, it is inferred that the inhabitants evacuated the town before the great earthquake. The severity of the earthquake may be illustrated by a few pictures from a single area: Marinatos 1968, fig. 51 (and a good photograph in Luce 1969, Plate VIII) : a two-storey building has been reduced on this frontage to a couple of courses of ashlar above ground; several courses of ashlar, and the whole of the superstructure, collapsed. Marinatos 1972. Plate 12 : the stone staircase behind this threshold has cracked down the middle. Luce 1969 Plate IX : the side-wall of the whole of the upper storey and of part of the lower storey collapsed. There was no pumice or ash amid or under the debris; it is proved here and elsewhere that the worst of the damage ⁽⁸⁾ occurred before any tephra fell.
3. Later occupants (I avoid the common term 'squatters') lived on part of the site for a short time. The town was already wrecked by earthquake when they arrived. Let one of many illustrations suffice: Marinatos 1970, Plate 41 : the bath was deliberately placed (presumably to catch water) on top of the debris of a two-storey building already reduced to a heap of rubble on the ground-level floor.
4. The town was buried under an immense mass of pumice and ash. Now the later occupants clearly prove that there was an interval between the great earthquake and the great eruption. The question remains, how long an interval? Our problem would be solved if it were (say) thirty years; is there any reason why it should not have been, and is there any evidence that it was?

Two objections have been alleged against an interval of such length ⁽⁹⁾. First, that if there was a long interval, the inhabitants would have returned and rebuilt their town. This objection, in itself rather weak, takes too little account of the severity of the damage, which renders any notion of re-building impractical.

The mass evacuation together with all valuables does not suggest any intention to return. ⁽¹⁰⁾ If the inhabitants returned after the great earthquake, they would find their town for the most part a heap of debris resting on ground-floors or ground-floor ceilings. The later occupants lived uncomfortably in spaces cleared amid the rubble, and soon gave up the attempt. Secondly, it is agreed that the later occupants stayed only a short time (months rather than a year or two), (Marinatos 1970, 66) and this fact is said to indicate that the interval between earthquake and eruption was brief (Hood 1970, 105; Renfrew 1971, 46 - 47).

The weakness of this argument is apparent: it would only be valid if we knew that the later occupants came shortly after the earthquake and left shortly before the eruption, and we have no reason whatsoever to suppose either of these things.

The later occupants may very well have been immigrants - pirates, perhaps, or a ship-load of refugees or merely shipwrecked or storm-bound sailors - arriving a few years or even a few decades after the earthquake. And, whether they were local or immigrants their reason for abandoning the attempt to reoccupy the site was probably the most obvious one - the discovery that the difficulties were practically insuperable and the discomforts extreme.

It is imprudent to make any general statement about the later occupants. Acrotiri was probably a large town; it may have extended all the way from the present excavation-site to Balos (Marinatos 1976); and the excavation-site is relatively very small. We do not know whether what we observe in this limited area is characteristic or exceptional.

In the area excavated, the presence of later occupants is so far clearly attested only in the southern sector close to the shore; if a ship's company landed here, this is the area which they would presumably first occupy.

The principal traces of their activity are as follows :

 - Marinatos

             1971c, 28 : in section Delta 14, a layer of animal-bones on the floor, 'left-overs of meals'.

              1976. 20-1, Plate 29 b : in section Beta 2, a hearth roughly constructed outside a wall.

              1970. Plate 38.2 : in section Gamma 1, a room converted into a workshop.

              1970 30-1 fig. 15 : between sections Beta and Gamma, a street cleared of debris, which was piled up round a corner at the north end of the street.

              1970. 47, Plate 42 : in section Gamma, a doorway roughly blocked up.

              1970. 43, Plate 35.2 : in section Gamma, a provisional doorway.

              1970. fig. 27, Plate 26.2 : in section Gamma, a window converted into a door.

              1970. 45, fig. 26 : in section Gamma 2, filling of a gap caused by collapse of the upper flight of a staircase.

              1970. Plate 32.1; in section Gamma 4, a rough wall built across a room.

              1976. 22, Plate 32a: in section Epsilon, a workshop made out of the ruins of the south-east corner of 'Xesté 3'.

The usual signs of prolonged occupation are everywhere wanting. The furnishing of the improvised workshops is 'of a hasty and provisional character' (1970, 38); pottery is 'absent or scanty' (30). Some tools are suitable for the demolition of unstable masonry (Doumas 1974a, III), but there is no sign of rebuilding (Marinatos 1970, 31), as opposed to rough repair. The work done, so far as it is at present definable, is within the scope of a dozen men employed for a couple of months if not less. ⁽¹¹⁾

The fact that the later occupants stayed a short time has no bearing whatsoever on the question how long a time elapsed between the earthquake which preceded their arrival and the eruption which followed their departure. We need a few decades; and that need has now been supplied.

In 1972 Mr J.H. Money ⁽¹²⁾ noticed an apparent layer of humus between the upper surface of the earthquake-debris and the first deposit of pumice. Dr I.W. Cornwall (Money 1973, 51f) analysed four samples taken from this layer and identified humus in all of them. Two of them he judged to be soil in situ requiring 'probably several decades' for its development.  

This evidence is decisive. It cannot be explained away. It is vain to object that the humus-layer is not observed uniformly over the site; ⁽¹³⁾ it is in fact quite often present, and the evidential value of one sample proving the lapse of several decades is not in the least affected by the absence of such proof in other samples, or indeed by the absence of samples in some places. It remains certain that the interval between earthquake and eruption was long enough to allow humus to develop 'probably for several decades' along the upper surface of a wall.

In summary, there is no obstacle to, and there is objective proof of, a satisfactory answer to our basic problem. It follows that ash-fall from the Theran eruption may be accepted as the primary cause of the desolation of eastern Crete in L.M. I B. The cause of the fires which destroyed most of the palaces, towns, and villas in central and eastern Crete remains uncertain, but there is no longer any reasonable doubt about the cause of the sudden abandonment of this wide and prosperous area; the ash-fall, and that alone, explains the desolation of Crete.

- (1).   Healy 1971, 182; Thorarinsson 1971, 231; cf. Vitaliano 1974, 21 and Hiller 1975, 51. On the Theran eruption, see now especially Bond and Sparks 1976.

- (2).   Page 1970, 38. Luce (1969, 41-2) believes that Gournia stands too high (120 - 154 feet above sea-level) to be reached by the flood-wave; if so, my point is proved; so impotent a wave would have only a slight and transient effect on the north-eastern coast of Crete (much of which is mountainous).

- (3).   Hood (1973, 117) suggests that tsunamis may have been the cause of the abandonment of certain houses in Gournia in L.M. I A.

- (4).   See the diagram in the fundamental publication by Nincovich and Heezen (1965). The diagram is reproduced in Page 1970, 36, and Hiller 1975, 55.

- (5).   Vitaliano 1974, 19ff. Apart from the archaeological sites, Santorini tephra was found in 50 % of a large number of samples from road-cuts and other surfaces on the north coast from Heraklion to Itanos.

- (6).   Marinatos 1970, 68; but he later withdrew (Marinatos & Hirmer 1973, 61). See especially Hiller 1975, 48 n. 17.

- (7).   Tectonic, not volcanic (Padang 1971, 57). It was presumably the same earthquake as that which shook Crete in L.M. I A (Acta 381).

- (8).   Not all of it; see Doumas 1974, 112.

- (9).   Hood 1970, 104. His refutation of a further argument based on certain cracks filled with tephra is wholly convincing, and I say no more about it.

- (10).   Galanopoulos (1971, 195 n.2) compares the seventy-year desolation of Yalta in the Crimea after the earthquake at the end of the fifteenth century.

- (11).   Doumas adds further evidence of 'squatter' activity, but nothing to justify the statement that 'it could not have been done by the squatters', or that it was 'part of a general plan to re-build the city'. The jars containing mortar for wall-plastering in the West House were presumably left there at the time of the main evacuation.

- (12).   Money 1973, 51. Warren (1973, 175) had already in 1970 noticed and stressed the importance of a layer of 'debris and natural earth' below the pumice, shown to him by Doumas (Hood 1970, 97).

- (13).   Doumas 1974, 111. His other objections (that the humus-layer might be 'part of the debris shovelled after the removal of the stones' or of 'earth used for the construction of walls', and that the charcoal in the humus-layer is in some way inconsistent with Dr Cornwall's findings) are decisively refuted by Money in his appendix to Doumas' article.

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