1The aim of the “Balkans 4000” project, as announced in the opening chapter, was to clarify the chronological and historical relationships between the different phenomena observed in Bulgaria and Greece between the end of the 6^th and the end of the 4^th/early 3^rd millennium BC, focusing on the problem of the transition from the Late Neolithic/Chalcolithic to the Early Bronze Age2. We aimed in particular to test, through a series of new radiocarbon dates3, whether or not there exists a break in the occupation as presumed by many scholars (mainly in Bulgaria, and to a lesser degree in Greece), and whether this break is patterned in any way that would possibly suggest its cause, or causes.

2The results have largely fulfilled these objectives, showing beyond any ambiguity that there is indeed a substantial break in the occupation of prehistoric sites in both countries, whose onset seems however to be more or less random, i.e. does not display any obvious patterning according to the sites’ setting, nature or previous history. The thing will be discussed in some detail further on. But we did more than that: the new dates indeed complete in several parts the existent chronological sequences, both local and regional, and refine the chronological definition of some stages that were previously poorly known.

Improvements to the late 6^th – early 3^rd millennium BC chronology

3Starting from the general chronological framework4, the following refinements can be presented as a result of the project (table 1):5

The Late Neolithic I

4The few new ¹⁴C dates from layers assigned to the stage that is conveniently described as Late Neolithic I (or LN Ia, or simply Late Neolithic) fall, without surprise, in the years between ca 6400‑6200 BP, i.e. between 5470 and 5000 cal BC at 2 sigma, thus confirming the more or less homogeneous character of the developments over a big part of the Balkan peninsula6. Their geographic situation is important, as they come from regions that were not represented so far in the radiocarbon record, namely Thessaly (Makrychori) and the Bulgarian Black Sea coast (Akladi Tseïri)7. Unfortunately, none of the sampled sites in Northern Bulgaria had layers assigned to this period. A distinction might seem further possible between an earlier stage, around 5400‑5200 cal BC, connected to the Middle Neolithic-Late Neolithic I transition, and a later, properly Late Neolithic stage, for the years after 5200; but the limit between the two is not clear cut.

The Late Neolithic II/Early Chalcolithic

5The dates assigned to the next stage (Greek Late Neolithic II-“Dimini”, or LN Ib; Bulgarian Early-Middle Chalcolithic) fall, again without much surprise, in the interval between 6000 and 5700 BP, i.e. between 4990/4800 and 4700/4500 cal BC. Several points deserve however to be underlined.

6The first concerns the start of the period, which indeed might be somewhat earlier than what is sometimes suggested. The evidence from Akladi Tseïri is at this point particularly important. Even if we exclude the date Lyon-7488 as too high8, the other two dates (Lyon-7487 and Lyon-7486), taken from the fill of a well with clear Early Chalcolithic features, place the start of it in the years around 6000 BP, i.e. very close to 5000 and in any case prior to 4800 cal BC. The earliest available dates for the equivalent phases in Mandra, Thessaly (Lyon-7648) and Agia Triada, in Southern Euboea (DEM-2097 and DEM-2183), have altogether slightly younger values (ca 5900 BP), but in both cases we are not sure that the dated layers indeed represent the start of the period9. Unfortunately, in spite of our efforts, we didn’t manage to provide any dates for this period from Northwestern Bulgaria, whose Early Chalcolithic, although culturally known, thus remains undated10.

7On the contrary, we brought for the first time evidence regarding early 5^th millennium BC inhabitation in the Rhodopes, an area which had been considered uninhabited in this period. The nature of the excavated features at the site of Varhari does not allow a conclusion as to whether the ¹⁴C dates provided apply to events that fall properly in the Early-Middle Chalcolithic or in the transition to the Late Chalcolithic11, but the fact remains that the area can no longer be regarded as an “empty space” until the end of the 5^th millennium. The historical consequences of this result have been pinpointed in the related chapter, and will also be further discussed below12.

8The third point to underline concerns the distinction between this and the next stage. Several indices indeed suggest that the broadly admitted distinction between an earlier and a later part of the Late Neolithic II/Chalcolithic (respectively corresponding to the “cultures” Karanovo V‑VI, Dimini-Rachmani, etc.) needs to be reconsidered, both in terms of content (i.e. material culture) and duration (i.e. absolute chronology). This might help deal better with some apparent “anomalies” of the archaeological and/or the radiocarbon record. Thus, in some cases, a shift seems to takes place as early as 5800 BP, i.e. towards 4600 or even 4700 cal BC, as suggested for instance by the three “early Rachmani” dates from the tell of Prodromos in Thessaly13. On the other hand, the neighbouring site of Mandra provides a similar date from a layer associated with “classic Dimini” material (Lyon-7649), and the same goes for the site of Rachmani itself, where two samples from contexts with “Dimini” pottery have also been dated to the years around 4750‑4500 BC (Lyon-7646 and Lyon-7645)14. This could mean that either one of these lots of evidence is erroneously associated with its “cultural stage”, or, more likely, that the two “cultural stages” are actually partly overlapping in time and space, and that their strictly chronological value should be taken with some reservation when applied to different sites15. Similar considerations have been expressed concerning Dikili Tash, whose sequence has undergone considerable revisions in recent years as a result of the ongoing excavations and the new series of ¹⁴C dates16. They are sustained by evidence from Kastri (Thasos), where layers broadly assigned to the KGK VI “culture”, are again dated to the years before 4500 cal BC17. Maybe the presumably too high dates from Orlitsa (ca 5700 BP, i.e. between 4700‑4500 cal BC) could be explained by a similar reconsideration of the start of the “Karanovo VI” phenomena, possibly accentuated here by an old-wood effect18. The even higher date from grave 3 at Smyadovo (Lyon-5511, 5929 ± 35 BP) pushes things further: as S. Chohadzhiev points out in the relevant chapter, life in the settlement (and therefore in the associated cemetery) should probably start at the beginning of the Late Chalcolithic, without excluding however the existence of an earlier occupation that would not be represented in the excavation trench19. A date close to the lower end of the calibrated interval (i.e. towards 4720 cal BC) would fit this scenario very well, suggesting that the site was settled a little before or right at the onset of the Late Chalcolithic properly speaking; this would agree in turn with the idea that the Varna I “culture” (with which are parallelized some of the offerings from the grave) falls at the start and not at the end of the period20.

9Further to the south the evidence from Agia Triada is not conclusive upon this point: the occupation layer 13 in trench 4, dated around 5650 BP (4579‑4360 cal BC)21, indeed provides nothing but a terminus ante quem for the local “Saliagos-Kephala” transition, which thus could be placed (also considering the dates from the lower levels in the same trench, supra) somewhere between 4700 and 4500 cal BC.

The Final Neolitlhic/Late Chalcolithic

10Whatever the date of the shift, most of the sites occupied during the previous period seem to continue uninterruptedly into the present. Among the exceptions we find some relatively “young” sites, like Vassilis in Thessaly22 and possibly Orlitsa and Varhari in the Rhodopes, which might have been abandoned before 4500 cal BC23. Conversely, some spots that were apparently not inhabited in the preceding centuries (e.g. Kosharna, Karnobat) are now settled, or re-settled24.

11The sites belonging to this period do not all evolve similarly. The new results indeed confirm the existence of several chronological “thresholds” spreading over the long interval from the mid‑5^th to the mid‑4^th millennium BC. Some of them mark simple destruction events without major consequences for the life of the settlement, whereas others mark a change of use, or the end of occupation for a shorter or a longer period. Many of these “thresholds” were already discernable in the data available25, but are now refined and consolidated, and eventually expanded to new areas.

12The first “threshold” seems to fall in the years around 5550‑5500/5450 BP, i.e. ca 4350‑4300/4250 cal BC. It corresponds to the date of abandonment – or, to be more precise, to the last attested use of graves or the last destruction event in settlements – in a number of Neolithic/Chalcolithic sites located in various areas: North-East Bulgaria26 (Smyadovo27, maybe Kosharna28), Thrace (Karnobat29, Karanovo30, Yunatsite31), but also Eastern Macedonia (Kryoneri)32 and Thessaly (Rachmani)33. Some of these settlements could actually have lived a little longer, if we accept that a small number of later occupation events might have been wiped off by erosion after the site’s final abandonment, as suggested by the evidence from Dikili Tash34. But they would not be expected to exceed one or two centuries, and probably would not extend beyond the end of the 5^th millennium BC. In terms of material culture, the dated levels are more-or-less typical “Late Chalcolithic KGK VI” or “Rachmani” accordingly. The confirmation of the end of the particular settlement at Rachmani well before 4000 BC should hopefully put an end to the misuse of the term “Rachmani culture” as a synonym of the 4^th millennium phenomena in Thessaly, and Greece in general.

13In addition to the above-mentioned sites, this chronological stage is also attested at Sadovets-Ezero, in Northwestern Bulgaria, where it does not represent however the last event35. This is important, because it establishes the chronological framework of the early stages of the KSB “culture”, proving furthermore the continuity with what follows (the KSB IV “cultural stage”, see below). Regrettably the poor coincidence between the expected stratigraphic position and the actual order of ¹⁴C dates does not allow complete exploitation of this important result36.

14This stage is not represented among the dated samples from Kastri and it is not impossible that it is indeed lacking from the settlement37. If a hiatus was confirmed in the stratigraphic sequence, it might point to a temporary displacement of the occupation (not at all unusual at Thasos, from what is seen from the regional analysis).

The Final Chalcolithic

15A second “threshold” could be placed around 5400/5350‑5250 BP, i.e. between approximately 4200 and 4000/3900 cal BC. The spectre of dated contexts is more varied than in the previous group: we again have primary destruction contexts (Sadovets-Ezero38, Akladi Tseïri39, Tatul and Dolno Dryanovo40, possibly Dikili Tash-sector 641, Kastri42, Dispilio43, Prodromos44, Palioskala45, Agia Triada46), but also some secondary deposits (Dikili Tash-sector 247, Mandra48), two undefined (Kosharna-pit 149, Sidirokastro-looter’s trench50), and even one grave (Dzhulyunitsa)51. Also, the distribution of the sites is very broad and covers the biggest part of the territory, including possibly NE Bulgaria (if we take the evidence from Kosharna to be reliable), but not Thrace, where unfortunately none of the dated samples went beyond 4300/4250 cal BC.

16The new results confirm that the stage which is called “Final Chalcolithic” in the Bulgarian terminology is not represented only in the two areas where this was already established, i.e. Northwestern Bulgaria and the Rhodopes. Its presence is now confirmed with radiocarbon dates from secure archaeological contexts in the two other regions where this was suspected (Southern Black Sea coast, Thasos52), as well as in other areas in Northern Greece where this was contested, namely lowland Eastern Macedonia53, or completely unknown (Western Macedonia). The importance of this enlargement will be discussed below. A consensus is discernable around the possible definition of this stage in terms of material culture, thanks to the fine parallelisms between the sites from the various areas: they bring together elements from the late- and post-“Karanovo VI” eastern horizons54, together with western “Krivodol-Sălcuţa-Bubanj Hum IV” features, and local particularities which become now more meaningful55. These parallelisms show two things:

• first, that the phenomena that are taking place in this period are a direct continuation of what preceded and that, whatever the changes in pottery style or other artefact categories, they should be seen as evolutions of an existent system of values and technical traditions (persistence of graphite, combination with crusted paints, sophisticated shapes, etc.);

• second, that the introduction (or emergence?) of some new stylistic traits that will be more visible in the next stage(s) is already operating in those years. The case of Dolno Dryanovo is from this point of view exemplary: the site indeed bridges, by its pottery assemblage as well as by its ¹⁴C dates, the latest manifestations of the previous system and the onset of the next.

17This stage is now clearly established for Thessaly, where concrete radiocarbon evidence had been lacking completely. The archaeologists should look for a more appropriate name than “Rachmani” in order to describe these ending-5^th millennium phenomena, which continue in some sites uninterruptedly into the following 4^th millennium (see below). In terms of material culture, we again witness this interesting merging of past and new trends, which will be further affirmed in the next stage, as seen from the Palioskala sequence56. This will also be the case further south, with the sequence from Agia Triada.

The Transitional period

18From what precedes, it is clear that the “Transitional period” has already started during the preceding centuries. From this point of view, the use of a different label might be confusing or misleading, and a prolongation of the term “Final Chalcolithic” (or simply Chalcolithic or Final Neolithic, concerning Greece), would be more convenient. The term of transition describes well, however, the process that is going on in both countries during the early 4^th millennium BC. Moreover, it takes into account some changes seen in different parts of the territory, namely the abandonment or the change of use of a number of long-lived sites. Among them, we encounter Dikili Tash and Prodromos, which show no sign of human presence57, as well as Akladi Tseïri, which is now used for inhumations58.

19Yet, the number of sites represented until the next “threshold”, towards 5000/4950 BP (= ca 3900/ 3700 cal BC), is relatively important at present59. Their geographical distribution allows the filling in of some of the previous lacunae of the radiocarbon record and establishes the chronological position of a number of potentially significant cultural traits. Thus, the series of dates from Bezhanovo and Borovan in Northern Bulgaria60 bridge the radiocarbon gap between the last “KSB” manifestations and the first presumably “Proto-Bronze” phenomena (the “Scheibenhenkel-Galatin” horizon)61. The analysis of pottery confirms that the latter do not succeed directly to the former, and that between the two a transformation process is taking place, which is reminiscent of what is going on further south in the Rhodopes (see below). Moreover, the evidence from Bezhanovo and that from neighbouring Devetaki, in Central North Bulgaria62, bridge the geographical space between the western and the eastern transitional groups (roughly the Galatin and Pevets “cultures”), confirming not only their roughly synchronous character, but also their “genetic” relationship, which seems so far to owe very little to external influences. This arouses questions about the presumed depopulation of the Eastern part between approximately 4300 and 3900/3700 cal BC: maybe life continued in this area too, like it did further to the West, in settlements that have not been preserved, or that have not been discovered yet.

20The prolongation of settlement in the Rhodopes (Yagodina63, and possibly Dolno Dryanovo64) is not a surprise. More interesting is the confirmation of the prolongation of settlement in Thasos65, based on one date from Kastri and two more from the coastal site of Agios Antonios, from layers with cultural features that are still considered to be “Chalcolithic”66. In Thessaly, both the material evidence and the ¹⁴C dates from Palioskala suggest that the transformation process toward a new reality is going on67.

21Finally, the last set of dates from Agia Triada, from layers with typical “Kephala” features, confirms what we lacked so far at the eponymous site, i.e. the persistence of settlement in the Central Aegean until the first quarter of the 4^th millennium BC68.

The Proto-Bronze stage

22None of the dates produced by the “Balkans 4000” project fall in the interval between roughly 4950 and 4800 BP, i.e. between ca 3900/3700 and 3650/3500 cal BC. The interval, already pinpointed by previous analyses69, could be ridiculously short (50 years) or tremendously long (400 years), thus leaving room for too much speculation. A unique date comes from the site of Dispilio, but unfortunately agrees badly with the expected archaeological context and thus cannot be taken, so far, as reliable evidence70.

23On the contrary, we bring now further solid evidence for the existence of a “transitional proto-Bronze” stage in the years around 4800/4700‑4650 BP, i.e. towards 3500‑3400 BC, with the five new dates from the single-layered settlement of Mikrothives in Thessaly71. This settlement indeed stands, both by its chronological position and its material characteristics, as a true “missing link” between the Neolithic/Chalcolithic and the forthcoming EBA. The presence of “Bratislava bowls” and other elements of seemingly northern origin confirms that new ingredients have been introduced into the local material assemblages during the preceding “invisible” years. However, this introduction of new elements did not necessarily imply a rapid and unidirectional population movement as suggested by some72. On the contrary, most of the pottery assemblage found in the site shows close connections with the material from earlier Palioskala and other purely “Final Neolithic” sites of the late 5^th and early 4^th millennium BC73.

24According to the archaeological evidence, this is also where the earlier phase of occupation at the site of Merenta, in Attica might be placed74. Unfortunately, all the radiocarbon dates produced by “Balkans 4000” have values between 4600 and 4400 BP, i.e. between 3400/3300‑3100/3000 cal BC, and are thus hardly different from those obtained from purely EBA contexts in other sites of Greece and Bulgaria (see below). Two possibilities come to mind: either the archaeological recognition would be false, or the dated samples would all represent later events. The latter seems more probable, considering that all the subterranean chambers where samples have been taken from were also occupied in the following EBA I period. A third option seems possible however: that the stage attested by the transitional “FN/EBA I” deposits at Merenta indeed precedes the onset of the “true” EBA, but not as much as the one attested at Mikrothives; it would be represented here by the date Lyon-7184 (4605 ± 30 BP, 3495‑3348 cal BC), whose both uncalibrated and calibrated result stands in between the Mikrothives series and the bulk of properly EBA dates75. But this is for the moment only a scenario that should be confirmed by further dating, including at other sites with similar transitional features, like the cemetery of Tsepi for instance76.

25A fourth possibility according to which all the samples from Merenta would indeed be “FN/EBA I” and dated at 3300‑3100/3000 cal BC deserves to be discussed: it would imply that the EBA I period started in Attica (and in the rest of Southern Greece?) later than in Northern Greece. The radiocarbon evidence from the area is so far too scarce to allow accepting or rejecting this hypothesis completely77; let’s say that for the moment such a delay is not sustained either by the material or by the ¹⁴C evidence, and that further data are necessary from both sides before concluding.

The Early Bronze I

26All properly speaking EBA I levels/features dated in the framework of the project gave results in the years between 4500 and 4350 BP, i.e. between 3300 and 3000/2900 cal BC. These dates agree completely with those previously available for both Greece and Bulgaria. The samples come from primary settlement contexts (Dikili Tash78, Sidirokastro79) as well as from graves (Smyadovo80): their distribution confirms that the phenomenon that is commonly called the Early Bronze Age took its shape more-or-less synchronously at least over the northern part of the investigated area, i.e. between the Danube and the northern coast of the Aegean, without any visible sign of priority or delay in any part81.

27The longest series of dates is provided by the site of Sidirokastro in Greek Eastern Macedonia: the fine statistical analysis (modelling) of the results allowed us to overcome the problematic effects of the well-known “plateau” of the calibration curve around 4500/4400 BP82 and to propose one of the two “peaks” as more probable for the start of the EBA, at least at this site. It seems indeed that the dates cluster better around the younger “peak”, i.e. around 3100/3000 cal BC, rather than around the older one (3300 cal BC)83. This statement has both a positive and a negative consequence. The “bad news” is that, at a local scale, the chronological gap between the last Chalcolithic and the first EBA event is getting bigger, and that the little we gained at the end of the 5^th millennium with the evidence for a possible Final Chalcolithic occupation of the cave (supra) is now lost again. Considered at a regional scale, however, this slight delay in the onset of the “true” EBA leaves more room for the transformation process itself, thus inviting us to expect (or hope for?) the discovery of more intermediary stages in the future.

28We will not discuss here further the dates from later Early Bronze Age, namely EBA II contexts. Let us simply note that most of them fall, as expected, in the early centuries of the 3^rd millennium BC, and that in all the cases where a stratigraphic continuity with the preceding EBA I is claimed, a smooth progression is also seen in the radiocarbon dates, without excluding however completely the possibility of small breaks. At Sidirokastro for instance more than 100 years BP separate the last date assigned to the EBA I (Lyon-7022, 4355 ± 35 BP) from the earliest one assigned to EBA II (Lyon-8908, 4200 ± 35 BP), but the calibrated intervals are practically touching each other (3084‑2900 and 2893‑2674 cal BC respectively). At Yunatsite, the unique EBA date produced in the framework of the “Balkans 4000” project (Ly-14795) falls in the years in between, thus agreeing well with the excavators’ estimation that the EBA at this site would not have started before 3000 cal BC84. The date Lyon-6029 from Kryoneri probably marks the later part of the period in Eastern Macedonia85.

For a redefinition of the “4^th millennium problem”

29The first thing that comes out of the analysis of the new ¹⁴C dates and their comparison with those previously known from the same and from other sites, is that there is indeed no site so far, either in Greece or Bulgaria, whose occupation would go on uninterruptedly from the Late Neolithic/Chalcolithic to the EBA period. This holds for sites where a hiatus is seen in stratigraphy (e.g. Karanovo, Yunatsite), as well as for those where it does not, either because the EBA remains lie directly on top of the older deposits (e.g. Dikili Tash, Sidirokastro) or because the sites are not re-occupied in the EBA (e.g. Mandra, Orlitsa, Varhari, Bezhanovo). This means that something definitely affects the lives of the Late Neolithic/Chalcolithic populations, pushing them to move away from the places that they were occupying until then.

30The results furthermore confirm that this “something” affects all kinds of settlements, independent of their nature, setting, past history, etc. We indeed see that, at one moment or another, both coastal and mountainous settlements (Kryoneri and Akladi Tseïri, but also Kastri and Sidirokastro), as well as lowland sites (Karanovo, Dikili Tash, Mandra, Prodromos…) are abandoned, whether these are tells, flat sites or caves (respectively Dikili Tash, Kryoneri, Sidirokastro), long-lived or relatively young (Karanovo, Varhari). When a cemetery is associated with the settlement, its functioning stops too (Smyadovo), although in some cases graves are installed in the area of the abandoned settlement, meaning that people are still around (Akladi Tseïri86, Mandra87).

31On the other hand, none of these topographical or cultural features seems to be directly connected to the time of the abandonment itself, no more than the geography as a whole. The new results indeed confirm the existence of some chronological “thresholds”, but show clearly that the sites representing them do not order themselves according to any of the above criteria. In other words, there seems to be no clear patterning, and more specifically no clear geographical progression of the abandonment events. This is important, because the idea of a progression is an essential component of all the scenarios advanced so far to explain the presumed depopulation of the territory at the transition from the Neolithic/Chalcolithic to the EBA.

32It is true that sites in North-Eastern Bulgaria are apparently abandoned early, towards 4350 cal BC (Smyadovo, possibly Kosharna), but at the same date several other settlements further south are also apparently abandoned – not only in Bulgarian Thrace (Karanovo, Yunatsite), but also in Eastern Macedonia (Kryoneri) and Thessaly (Pefkakia, Rachmani). As a consequence, we cannot speak any more about a priority of the Northeastern areas against the others, as a sign of hypothetical steppe-invasion movements. We cannot speak either about a priority of coastal or lowland sites either, as presumed victims of a sudden water-rise, since in many such areas we find, next to the abandoned settlements, others that continue to be occupied for a little, or for much, longer (e.g. Dikili Tash in Eastern Macedonia, Prodromos, Mandra, Palioskala in Thessaly). It is true that some regions, like Thrace, still provide no dates after 4300/4200 cal BC, but, as already said, there are indications that this could be related to taphonomy, or with the orientation of research88. Again, there is no obvious pattern in the profiles either of the sites which are abandoned, or of those that survive. The coexistence of abandoned and ongoing settlements in the same environments, together with the overall spreading of the time of the abandonment over a period of several centuries, invalidates, finally, the hypothesis of a collapse due to an abrupt climatic change. Whatever the cause of these decisions, it was by all evidence neither unique nor unidirectional.

33The other important outcome of the project concerns the start of settlement in some areas reputed to act like refuges for the inhabitants of the abandoned Chalcolithic sites. This is especially the case of the Rhodopes, on the Greek-Bulgarian border. The dominant scenario so far was that the area was first settled at the end of the 5^th millennium BC, as the result of a transfer of populations from the abandoned lowlands tells. Both these statements (i.e. the late start and the prolongation after the end of tell settlements, until 3800 or even 3600 cal BC) are used as arguments supporting the idea of the “catastrophe” scenario in the Balkans, either by natural causes or invasions89. Actually, the new ¹⁴C dates show that settlement starts in the area well before the end of the tells in the neighbouring Thracian plain, and therefore is not the work of refugees. Consequently, the two settlement patterns coexist from the early stages of the Chalcolithic period. One should note that the two sites that represent this early group (Varhari and Orlitsa) are totally flat and lay in small river valleys in the mountains: probably they would have never been discovered if not for road construction. This tells us a lot about the reliability of our corpus and raises once more the discussion of archaeological visibility. As for the second point, i.e. the prolongation of the Rhodopean sites beyond the end of the life on tells, it appears to be only half-true: some, such as the caves of Yagodina and Haramiiska, or the open-air site of Dolno Dryanovo, indeed survive until after 4000 BC, but others, like Tatul, don’t. On the other hand, some tell-sites in neighbouring plains, like Dikili Tash for instance, could also have lived as long90. Under other circumstances, Dikili Tash would have been easily placed in the group of settlements ending before 4200 cal BC, but it has now been proven that it lived until ca 4000 cal BC. During the last century (or centuries) of its existence, it shares features with some of the newly founded settlements in the neighbouring Rhodopes (Dolno Dryanovo, Tatul, Haramiiska), as well as with some pre-existent sites like Sitagroi, at ca 20 km to the north-east, and Kastri, on Thasos. We are far from the stereotypical image of a world of “prosperous farmers” replaced by a world of “poor stock-breeders”.

34This brings us back to what seems to be the most important point of our analysis: the regional continuities. Taken individually all the Neolithic/Chalcolithic sites seem to be abandoned at some point in their history, and most of them display gaps that reach or exceed a whole millennium. But considered at a regional scale, gaps are getting considerably smaller, as the individual sequences largely complete each other in most areas, indicating that the settlements are simply changing location91.

35The next question is of course, why they did so? Or, to be more precise, why did they do so at such a greater scale than before, or afterwards. A related question is, why don’t we see them more in their new locations. I am personally convinced (and much of what has been exposed in this volume seems to prove it) that the latter is merely a question of conservation and visibility, and that the archaeological record will change significantly in the next decades, now that greater attention is given (by chance or by methodological choice) to the investigation of more marginal spots of the archaeological landscape: small mountainous valleys, lowland terraces, outskirts of bigger tells, etc.

36For the explanation of change, the answer is of course more difficult. But at least we can rule out some of the previously advanced hypotheses. The temporally spaced and apparently random distribution of the affected sites points toward human rather than natural/environmental factors. By saying this, we do not suggest that environmental parameters could not have played a role, on the contrary; we just claim that the decision to abandon or to maintain a settlement in a given area was the result of human choices, and not the spontaneous response to any broad, natural catastrophic event.

37Therefore, what appeared so far as an effect of depopulation might be simply the expression of evolution or changes in settlement pattern. One could even anticipate several possible types of population movements, eventually responding to different stimuli or different kinds of needs: movements at a local scale, i.e. from one spot to another, at only a few hundred meters or a few kilometres away; movements at a regional scale, i.e. a redistribution of the population in a given area or cluster of areas according to the new conditions (environmental, social, economic, etc.); finally, maybe movements at a supra-regional scale, towards new destinations that were previously unexplored or poorly exploited92.

Continuity vs break

38While this volume was still in preparation, John Coleman published a synthetic paper, where he presents again, with additional arguments, his hypothesis about a depopulation of the Greek peninsula at the end of the 5^th millennium BC, similar to the one claimed for other parts of the Balkans, and a re-population by people who came from much more Northern areas93. I think that we can all agree with the second part of his proposal, i.e. the introduction of new elements of probably northern origin into the local repertoire by at least 3600/3500 cal BC94. But the first part of his analysis seems to stand on false grounds. New evidence – both material and radiocarbon – points now not toward a collapse but toward an overall slow transformation process, with several local break events.

39Indeed, in spite of the extended – and so far difficult to understand – abandonment of the majority of Chalcolithic settlements towards the end of the late 5^th millennium, the “Balkan koine” is not dismantled completely. Contemporary 4^th millennium sites are found at least in Thessaly and in Northwestern Bulgaria, where they seem to continue sharing a common body of habits and codes. Affinities in the material culture, as those observed, for instance, between North-West Bulgaria and the Rhodopes, most probably represent the expression of a world that still continues to live together, under the shape of a more diluted, perhaps, but still active system of cultural exchanges that allows the preservation of traditions and the spreading of innovations (in pottery, metallurgy, etc.) toward all directions. The original network did not collapse completely, but continued to function underneath during this “invisible” period. The most plausible hypothesis is therefore that more settlements of the intermediary centuries existed all over the area, and if preserved, can be expected to be discovered by future archaeological investigations.