Phylakopi and the Late Bronze I Period in the Cyclades

A large mansion underlying the LH III Megaron, as well as fragments of a Minoan Linear A tablet, can also be dated to the same early phase. Santorini tephra, securely stratified, appeared first in Late Cycladic I levels, though not in the earliest ones, and their presence persisted for a considerable subsequent period. Evidence suggests that the tephra fall occurred before LM IB imports reached the site. The effect of this tentative conclusion on the different theories about the Santorini eruption and the LM IB destructions on Crete is discussed. There is no evidence for tsunami damage at Phylakopi and occupation continued without interruption at the time of the Santorini eruption which left no other sign of its occurrence than a few parts per thousand of glass sherds in the material on and between the floors.

INTRODUCTION

The purpose of this article is to summarise what has been learnt of the Late Bronze I period of the Cycladic Islands during the recent excavations at Phylakopi in Melos, and to indicate the bearing they have upon the interpretation of the 'Minoan' eruption of Santorini, and on the Aegean culture of that time.

Phylakopi, the important bronze age town on the island of Melos, was excavated in the years 1896 to 1899 by the British School of Archaeology at Athens. Duncan Mackenzie, who was later to assist Sir Arthur Evans in the excavations at Knossos, was in day-to-day charge of most of the digging and contributed to the well-documented final report (Atkinson et al. 1904). His excavation notebooks have been preserved (Mackenzie 1963) and supply details not included in the published report, as well as stratigraphic argument and reasoning which makes much more intelligible several of the conclusions in the final report which are open to discussion today.

The site was re-excavated in a brief excavation in 1911 by Dawkins and Droop. Useful new information from their unpublished excavation notebooks has been set out and interpreted recently by Barber (1974). The excavations from 1974 to 1977 were directed on behalf of the British School at Athens with the authorisation of the Greek Archaeological Service. The results allow a reinterpretation of the earlier excavations as well as the presentation of new material.

Phylakopi has, however, to be considered in relation to the other Cycladic sites of the period, notably Aghia Irini on Kea, excavated by Professor J.L. Caskey (1971; 1972) and of course Akrotiri on Thera, whose excavation by Professor Marinatos and now by Dr. Doumas has provided the impetus for the forthcoming conference. Other Cycladic sites, especially on Tenos and Naxos, were discussed by Scholes (1956) whose paper remains the best general survey of the area in the later bronze age.

THE STRATIGRAPHIC SEQUENCE

The work of Mackenzie and his colleagues resulted in the division of the sequence at Phylakopi into four main periods of occupation, namely a first, prior to full urban settlement, documented by scant traces, and the three 'Cities' whose occupation spanned almost the whole of the Aegean bronze age. The sequence, based on the findings reported by Atkinson, Edgar and Mackenzie in the final report is summarised in Table I. Mackenzie himself subdivided this sequence (Atkinson et al. 1904, 243 - 70), and his treatment has been conveniently summarised by Barber (1974, 4 - 5).

In reading Mackenzie's daybooks today one's admiration is excited by the great clarity of his thought and the systematic nature of his observations, just as the published report brought archaeological writing in the Aegean to a new standard.

But when the work was undertaken, excavations at Knossos had not begun, and they were still at an early stage when the Phylakopi report was written. The evolution of Minoan ceramic styles was thus completely unknown to Mackenzie as he dug, and the essential Mycenaean ceramic sequence of Furumark lay forty years in the future. During the excavation, therefore, the excavators did not have the benefit of a firm outline development for Minoan and Mycenaean ceramics, such as we have today, partly as a result of their own labours. It was for this reason, I believe, that Mackenzie made a number of errors in interpretation, which go back to a single difficulty.

During the recent excavations of the site we decided to excavate without any preconceptions about a sequence of three Cities. Phases were correlated with structural events observed in our own excavation trenches, as seen in Table II.

In the areas in which we dug it was possible to recognise four rather than three main episodes of building activity. There is no guarantee that the buildings of each phase on different parts of the site are precisely contemporary, and I am personally always sceptical of the "destructions" which figure so widely in historical accounts of the prehistoric period. But a fourfold division is a convenient one.

It will be observed that periods III and IV of the new scheme correspond approximately to City III of the old scheme.

On at least two occasions it seems that Mackenzie was misled by the model of three successive settlements which he himself formulated. In digging the pillar room with the frescoes in square G3 (Atkinson et al. 1904, pl. I) he observed that the deposits were in the second building complex which he came upon, counting from the surface. Mainly for this reason he attributed them to his Second City, that is to the Middle Bronze Age, whereas they can now be assigned to our period III, the Late Bronze I period. A similar argument holds for the fortification walls at the western part of the site. The re-assessment of the stratigraphic succession has important historical consequences.

PRINCIPAL FEATURES OF PERIOD III

The two main stratified deposits of this period from the 1974 - 7 excavations are from our trenches PLa and PK, located on the south side of the site, in square G5 of the 1896 - 9 excavations (Atkinson et al. 1904, pl. II). These have yielded substantial bodies of ceramic material which may be used as a point of comparison for other areas. They have been examined first by Ms. Kathleen Thorp -Scholes, and then by Dr. Jack Davis, Mr. John Cherry and Mr. Callum Macfarlane, who will be publishing the material and to whom I am likewise indebted for notes upon it. The pottery from the former context is from a very early phase of the late bronze age, with Minoan imports in predominantly dark ground styles, and the presence in some quantities of local pottery in the Naturalistic and Black-and-Red styles. Local imitation of Minoan motifs is mostly limited to ripple pattern. The second deposit reflects more fully-developed Late Minoan Ia styles.

In trench PK outside the fortification wall there is a sequence of deposits of the earlier part of the Late Bronze I period with Late Minoan Ia imports. Above this is a level, PK level 25, which (as excavated) contains sherds from the uppermost part of the deposit just mentioned and the lower part of the succeeding stratum.

It contains Late Minoan Ib import sherds, including the small open bowl, pl. A, decorated with octopodia of Miss Mountjoy's type B (Mountjoy 1974, 177).

The level above, PK level 21, contained pure Late Minoan Ib deposit. This sequence is of note, since a clear sequence of levels with Late Minoan Ia imports followed by Late Minoan Ib imports has not been observed elsewhere during the recent excavations. Another fine Late Minoan Ib import piece, again from trench PK, a baggy alabastron, is seen in pl. B and fig. 1.

There is no doubt that the pottery of our phase D, corresponding to period III of our simplified sequence, is equivalent to that of the earlier part of the Third City of the 1896 - 9 excavations, which makes its association with the early fortification wall and the frescoes near the pillar crypt (a and b below) particularly important. The early part of this phase corresponds also to the floruit of Akrotiri on Thera, although of course the later part of the phase, documented by our Late Minoan Ib imports, appears to be lacking at that site.

Naturally all these points will be documented in the final report, now under preparation, with full stratigraphic detail. For the moment only a provisional and outline account can be given.

• (a) The Fortifications.

One of the most important results of the excavation has been to document the level of activity at Phylakopi during the Late Bronze III period. It has been established that the site underwent extensive refortification during the Late Helladic IIIB 1 phase, the wall at that period being constructed of large, round boulders collected from the beach near the site. Here, however, our concern is with the earlier phase of construction of the fortifications, and it has already been suggested above that Mackenzie was mistaken in assigning the fortifications at the western part of the site to his Second City, the Middle Cycladic period.

At the southern part of trench PK we were fortunate in finding the earlier defensive wall, set some way to the south (outside) the later LH IIIB wall. Its construction (pl. C) is more regular, with dressed stones. The pottery associated with its construction and early use is unequivocally of our phase D, with import sherds of the Late Minoan I period.

In order to test this observation, soundings were dug against the defensive wall at the west end of the site in squares A3, B3 and C3 of the 1896 - 9 excavations, but unfortunately the stratified levels there had been dug away to bedrock. However, in our trench KKd (in square E5 of the earlier excavations), in situ deposits were found and followed to the foundations of the wall and to bedrock, showing that here too the wall was built early in phase D.

The inference must be that the wall further to the west is likely to have been built in the same period, although this cannot now be unequivocally demonstrated.

It would seem that the main fortifications seen at Phylakopi belong to our period III, and specifically to a time contemporary with Late Minoan I. Davis (1977, 181) has observed: "The outer face of the earliest fortifications at Phylakopi resembles in technique the Great Fortifications at Ayia Irini. It is built in rough ashlar style and in most places is founded on bedrock". The Great Fortifications at A. Irini on Kea were built rather earlier, during the later Middle Bronze Age.

But so far we have no indication at Phylakopi of fortifications prior to our phase D, although it remains entirely possible that earlier fortifications may yet be discovered on the site.

• (b) The Frescoes.

A similar revision in dating has to be applied to the fresco fragments from the site (Bosanquet in Atkinson et al. 1904) of which the most important were found in the pillar room G3, 6 and 7, and nearby. It was Mr. Sinclair Hood who suggested that, on typological grounds, he would prefer to assign these frescoes to the Late Bronze I period. In order to test this hypothesis our trench ΠS was dug in the space immediately to the east of rooms G3, 6, 7 and 11. Stratified deposits were at once found in this area, which had remained undug in earlier excavations. Numerous fresco fragments were found in levels with pottery of our phase D. The underlying levels of phase C had no such finds.

It was our conclusion (and the stratigraphic evidence will naturally be presented in detail at a later stage) that the wall-paintings in that area were painted in phase D. It seems likely that the same holds for painted fragments from elsewhere on the site. The new wall-painting fragments from Phylakopi will be published in the final report by Mr. Mark Cameron.

• (c) The Late Bronze I Mansion.

One of the objectives of the recent excavations was to date the construction of the Mycenaean megaron located in squares H and J 1 and 2. Unfortunately most of this structure was dug away without adequate record in the excavations of 1911, but the original floors were preserved in some rooms of the east wing. We consequently dug in square J 2 room 15, our trench being designated ΠA. Beneath the floor of the megaron a fill was found with much Late Bronze I pottery and a number of sherds assignable to the Late Helladic IIIA 1 period. The construction of the Mycenaean megaron/palace was thus undertaken during a developed stage of this phase.

Beneath the fill were found floor levels which may be associated with walls of a very large structure, comparable indeed in size with the Mycenaean megaron itself. Despite the virtual destruction of much of the palace and its predecessor in 1911 it has been possible to reconstruct the plan of this earlier Mansion (fig. 2).

This may profitably be compared with the balloon photograph taken for us by Professor J. Wilson Myers of Michigan State University (pl. D). Parts of this structure were incorporated into the later palace, but parts were overlain by it. The north wall of the palace is set just a little to the south of the north wall of the Mansion.

Beneath the floor levels is a fill, formed in the levelling process prior to the building of the Mansion - designated ΠA level 85. The ceramic material from this and neighbouring levels resembles that from the appropriate levels of trench PLa discussed above, although actual Late Minoan Ia imports are lacking. The fill is to be assigned to a very early stage of Late Bronze I at the very beginning of our phase D.

We have, therefore, a major structure erected early in period III. Unfortunately we cannot be certain when it went out of use, although the presence of Late Minoan Ib sherds in the fill below the Mycenaean megaron floor is suggestive.

The most significant find from trench ΠA, from a level at the very bottom of the fill preceding the palace is seen in pl. E : two joining fragments of an inscribed tablet in the Minoan Linear A script. The inscription has been studied by William C. Brice (Renfrew & Brice 1978). He notes the resemblance in format with Tylissos tablet IV 9, notably in its compartmentation by horizontal lines separating single or double rows of writing, and its contents, evidently lists of commodities identified by ligatured combinations. It is of course tempting to associate this find, which documents both literacy at Phylakopi during period III, and the use of writing for some inventory system, with the Mansion itself. It does indeed seem possible that the Mansion was the administrative and organising centre for Phylakopi at that time, and that it employed scribes and an archive, much as did the contemporary palaces of Crete.

MELOS AND THE SANTORINI ERUPTION

The archaeological finds document that Phylakopi was a flourishing settlement at the time of the floruit of Akrotiri. The question at once arises, therefore, as to the possible effects of the Santorini eruption on Melos itself. Is there any evidence of earthquake damage, or the impact of tsunamis, or indications of tephra fall? Interestingly, for the last of these there is.

It should first be pointed out, however, that the recent excavations did not produce, in any of the areas excavated, a continuous and uninterrupted sequence of occupation deposits lasting right through the Late Bronze I period. In the area of the Mansion, in trench ΠC to the north, trench ΠD and ΠA at the south-east corner, there is a fine sequence of floors (fig. 3 and pl. F) which, as we have seen, starts at the very beginning of Late Bronze I. However, the uppermost floor seems to belong to a period prior to that when Late Minoan Ib imports are first seen.

Possibly the Phylakopi period III deposits were disturbed in this area during the levelling process for the construction of the Late Helladic IlIa megaron at the beginning of Pylakopi period IV. Certainly the fill above these floors of the Mansion and below the floor of that megaron contains LM Ib import sherds as well as LM Ia.

The second main area of the recent excavations at the south of the site likewise failed to produce a complete sequence. Trenches PLa and PK are both located near the defensive walls, and the period III deposits were mainly fills and tips of debris: no very satisfactory sequence of actual floor levels was recovered in that area. So that although there is fairly plentiful Late Bronze I material, the stratigraphy does not allow of very detailed subdivision. (The stratum in PK of Late Bronze I material with clear Late Minoan Ib imports mentioned earlier was mainly outside the wall, representing debris rather than settlement levels).

Indeed it should be remarked that only in rooms H4 22, and H4 25 and 26 of the 1911 excavations (Barber 1974, 14 - 16; Dawkins & Droop 1911, pl. XI) have Late Minoan Ib imports from a good settlement context been reported in detail.

Despite these limitations it was gratifying that clear indications of Santorini tephra were found: the first time they have been documented within a clear primary context and in a well-documented stratigraphic succession outside of Thera. It is possible to document their absence from Middle Cycladic and very early Late Cycladic I levels, their appearance during that phase, and their presence on the site for a considerable period subsequently.

The work in question has been carried out by Dr. Charles J. Vitaliano and Dorothy B. Vitaliano of the Department of Geology, Indiana University, who visited the site in 1974 and took samples from a number of suitable contexts. In the light of their findings I took further samples for them in 1975. They will be reporting in detail elsewhere on the technical aspects of their work and on the wider context of their studies in the Aegean (cf. Vitaliano & Vitaliano 1974).

Here I shall simply summarise the stratigraphic significance of the results of the sampling at Phylakopi. A full and detailed discussion of the stratigraphic contexts must await the final excavation report, where the stratigraphic section for each trench and the appropriate levels synopsis will be presented in detail with comments on the pottery offered by Ms. Kathleen Thorp-Scholes and Miss P. A. Mountjoy.

It is important to stress that nowhere at Phylakopi was any deposit of Santorini tephra visible to the naked eye. Indeed no actual deposit was found. The procedure adopted was to take a large number of samples in stratigraphic succession in selected trenches where Late Bronze I material (our phase D) was found.

Samples were taken also from earlier and later levels. These samples were examined by Dr. and Mrs. Vitaliano in the laboratory, and Santorini tephra sherds were found in some, by microscopic examination, up to a frequency of around 10 particles per thousand. This clearly documents that there was a fall of ash on Melos, but does not suggest that it was of any great depth nor that it had any marked effect upon the inhabitants.

In Table III are summarised the contexts where tephra were recognised. It should be noted that 78 samples were examined in all, and that the negative results from 57 are as meaningful as the positive ones from the 21 listed here. In the full publication all of these will be fully set out: here only a synopsis is offered.

Level 23 in trench ΠC is on the penultimate floor (and below the uppermost floor) of the sequence of floors associated with the Mansion discussed earlier. Level 9 of trench ΠD occupies a comparable position. The context may be studied more clearly, however, from the results from trench ΠA. Levels 10, 16 and 65 are from the fill below the LH IIIA 1 floor of the palace, and simply testify to the continued presence of tephra on the site some time after the original fall. Levels 67 and 78 are both below the uppermost floor (floor 1) of the sequence associated with the Mansion in that area. Level 79 is immediately beneath floor 2. Lower levels, including those associated with floors 3, 4 and 5 as well as the underlying fill revealed no tephra.

The results from PLa and trench ΠS do not add to this picture significantly or contradict it. (Sample 200 from ΠS level 26 must however be discounted since this is a Middle Cydadic level: the low frequency of one particle per thousand of Bronze Age tephra may plausibly be ascribed to contamination. All the indications of higher frequencies of tephra make coherent sense stratigraphically).

The floor sequence in ΠA, like those in ΠD and ΠC, starts at the very beginning of phase D (the fill below floor 5), corresponding to the inception of Late Bronze I. It seems likely that the entire floor sequence, including floor 1, belongs to the earlier part of phase D, contemporary with Late Minoan Ia. Certainly no Late Minoan Ib imports have been found in association with these floors. But it should be stressed that very few diagnostic sherds were found with them, so that no unequivocal phasing for the tephra fall can be given. It is fair to say, however, that on the basis of our evidence it is likely that the tephra fall occurred at Phylakopi at a time before Late Minoan Ib imports reached the site.

This is certainly a significant finding since, as we have documented earlier, Late Minoan Ib Marine style imports did indeed reach the site in fair quantities. Miss Mountjoy has singled out ten pieces (which she will publish in our final report), and a complete count of all the marine pieces from Knossos is included in Mountjoy, Jones & Cherry, forthcoming. Three of our LM 1b imports indeed come from the fill below the palace floor (fig. 4) from ΠA levels 10, 18 and 65 with 66 (joining). As we have seen, tephra were detected from two of these levels (10 and 16) but it is their presence in the lower levels, stratified between floors (levels 67, 78 and 79) which is especially important.

DISCUSSION - A TITHE FOR MINOS?

While the stratigraphic context at Phylakopi of the tephra from the Santorini eruption does not give as full a sequence for the Late Bronze I period as would be desirable, several relevant conclusions emerge.

In the first place, the tentative conclusion that the tephra fall occurred before Late Minoan Ib imports reached the site has very wide-reaching implications, given that Late Minoan Ib imports did indeed subsequently arrive, as is amply documented. Those arguing that the Santorini eruption was directly responsible for the destruction late in the Late Minoan Ib period of many of the major centres of Crete have offered two alternative explanations for the absence of abundant Late Minoan Ib pottery at Akrotiri itself.

The first possibility is that the eruption took place in several phases, an early phase destroying Santorini, and a later phase, perhaps 50 years later, causing the demise of the Minoan palaces, whether through ash fall or tsunami. It should be firmly stated that there is absolutely no evidence for tsunami damage at this time at Phylakopi, a coastal site. It seems improbable that such widespread devastation by tsunami could be caused even at inland sites in Crete without evident effects at Phylakopi. Likewise we have no clear indications of earthquake activity at Phylakopi at this time. Negative evidence cannot be conclusive, but had there been such activity rather widespread destruction might have been expected at the site, such as is seen (no doubt at a later period) at Ayia Irini in Kea (Caskey 1971, 374 - 6). As concerns the ash fall, it is documented at Phylakopi in contexts apparently prior to the arrival of Late Minoan Ib pottery. The deep sea cores do not give evidence for two major ash falls separated by 50 years, and it seems likely that this ash fall at Phylakopi took place at the same time as that which buried the Akrotiri settlement.

It should be stressed that Santorini tephra are found at Phylakopi above the third in a sequence of five floors (if we count the first as earliest). Occupation continued after the tephra fall without any break in continuity, indeed without any significant indication of the Santorini eruption other than a few parts per thousand of glass sherds in the material on and between the floors.

The second possibility put forward by the advocates of the "one disaster" hypothesis is that Akrotiri was actually destroyed at the same time as the Late Minoan Ib palaces in Crete (cf. Luce 1976). The absence of Late Minoan Ib pottery in Thera would then be due to special cultural factors. But Phylakopi now shows that we do have such imports in Melos, probably all arriving after the ash fall, and this observation would be difficult to reconcile with the "one disaster" theory, unless all the Melian LM Ib material were later in manufacture than that of the Cretan palace destructions, which is manifestly not the case.

I believe instead that the Phylakopi evidence harmonises most effectively with the view of Hood (1970) and of Popham (1975), that the eruption of Santorini, occurring towards the end of Late Minoan Ia, and the destruction of Minoan sites at the end of Late Minoan Ib were entirely separate events. That Akrotiri was seriously damaged by volcanic earthquakes prior to the great tephra fall has been convincingly documented by Doumas (1974). But these disturbances could be entirely local effects, so that the first (and only) impact of the eruption in the Aegean at large would be a single, impressive fall of tephra late in the Late Minoan Ia period together, no doubt, with some modest tsunamis leaving no trace on the archaeological record. As Pichler and Schiering (1977) have suggested, there is no convincing evidence in Crete itself for the effects of the eruption, other than modest quantities of tephra which are comparable in quantity with those recovered in Melos (Vitaliano & Vitaliano 1974; Cadogan et al. 1972). Certainly I believe that the rather wild claims sometimes made (e.g. van Bemmelen 1970) for dramatic consequences elsewhere in the East Mediterranean lack any serious supporting evidence.

None of this is to deny the very wide significance of Akrotiri for the understanding of Aegean prehistory. And it is here that Phylakopi has perhaps something further to offer. For like Akrotiri it shows very substantial Minoan influence in the Late Minoan Ia period, seen in the pottery and in the frescoes (cf. Furumark 1950). But it shows also two more important features not yet recognised at Akrotiri, namely strong fortification and a central organisation supported by records in the Minoan Linear A script.

Whether Akrotiri was fortified on the landward side remains to be investigated. Certainly Phylakopi was fortified in Late Minoan Ia, and Ayia Irini on Kea re-fortified in the final Middle Cycladic period. Interestingly the walls of Kea were destroyed rather soon afterwards, so that the site was apparently undefended in the Late Minoan Ib period, allowing Davis (1977, 184) to speak of a "Pax Minoica" in that period. However, if such peace did indeed descend at that time, it was certainly not in evidence in the earlier part of Late Bronze I.

The recognition at Phylakopi of a Mansion, with the implication of an archive, must surely make very much more likely the possibility of a central palace or mansion at Akrotiri, together with an extensive archive. And of course the circumstances on Thera of being buried in warm, dry ash may have preserved these tablets well, even if unbaked. When they are discovered it may be necessary to bake them at once, before cleaning, as is routinely done on many Near Eastern sites.

When such an archive is discovered, as I am sure it will ultimately be, it may allow a distinction between the present three possibilities for the status of Akrotiri or of Phylakopi vis-à-vis Minoan Crete (Renfrew & Brice 1978). Each could have been a "colony", in the sense of having been conquered by Minoans, with the setting up of a palace administration there on the Minoan model, thereafter developing without direct administration from the mother foundation on Crete. Or, alternatively, the islands may have been conquered by a Cretan palace centre and governed directly from it. Thirdly, it may be that either Akrotiri or Phylakopi, or both, developed quite independently as palace polities, enjoying a flourishing trade with Crete and adopting a Minoan-inspired writing system, yet remaining politically without affiliation or obligation.

Between these possibilities it is difficult to discriminate at the moment, just as it was when Furumark (1950, 200) wrote, although there are times when, for Phylakopi, I favour the third. But the discovery of an archive would surely prove conclusive and perhaps document the second possibility, if such was the case. Will the Akrotiri archive, when it is discovered, solve the problem for us by documenting a tithe for Minos?

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

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