21
0002 uon\.'Ee1re1 soqlor{d Ydounflrso(Ios Nr I{IDIHIITOtrT{ I,{NZ flCYUIIfl fl OA III (INYfl ONIYUYX reqeFsnerell AOTOXIN IIsssA pun IIUTTIH uEJols egos 'uegeqcsuessllv\ Jep eruepalv ueqcsueElng rep umesnlN llru ln1qsul seqcstEologqorv Emqzleg lgllsrelrun rep tnlllsq seqcsr8ologrlcrv

A chronological framework for the neolithisation of the southern Balkans

Embed Size (px)

Citation preview

0002 uon\.'Ee1re1 soqlor{d

Ydounflrso(Ios NrI{IDIHIITOtrT{ I,{NZ flCYUIIfl fl

OAIII (INYfl

ONIYUYX

reqeFsnerell

AOTOXIN IIsssA pun IIUTTIH uEJols

egos 'uegeqcsuessllv\ Jep eruepalv ueqcsueElng rep

umesnlN llru ln1qsul seqcstEologqorv

Emqzleg lgllsrelrun rep tnlllsq seqcsr8ologrlcrv

A Chronological Framework for the Neolithisation of theSouthern Balkans

Luurens THISSEN

l.IntroductionThe intensive archaeological fieldwork carried out in Northwest Turkey for the last two decades has demonstrated at

least one thing: it became apparent that the area did nor fulfiI the bridge function in the neolithisation of Europe as has been

taken for granted for a long time. The excavations and surveys by M. Ozdo[an and J. Roodenberg in the Marmara region did,however, affrrm the implication made by French that this area, situated "on the border of four main environmental and

climatic zones," constitutes a rather complex blend of cultural pattems as wellr . Both Aegean and Central Anatolian influences

are, for instance, detectable in the material culture of Ihprnar, the Fikirtepe sites and the cave of Yanmburgaz. In stark

contrast to the coastal zones where these sites are located, the inland sector of Turkish Thrace seems to have tumed its back

on the sea and to have evolved as a consequence of developments in Bulgarian Thrace, without any clear link to the Marmara

area. With the Marmara region and Turkish Thrace having less and less likely played an important role in the neolithisation

ofEurope, the growing perception ofthe existence of an 'Aegean koine'-perhaps already since the early days oftuna fishing

and obsidian collecting at Melos, that is: long before the full establishment of farming sites on both sides of the Aegean Sea

-points away from NW Turkey, towards Westem Turkey. In trying to find a geographic origin for the neolithisation of Europe

(i.e. in terms of the location of a 'nuclear mother area'), it has perhaps been overlooked in our Ex Oriente Luxway of thinking,

that the Aegean part of this region, here defined by the catchment areas of the east-west flowing rivers debouching in the

Aegean Sea (the Gediz, the BUy0k Menderes and Kiigiik Menderes), and archaeologically still ill-explored, has an enonnous

environmental potential for the development of early farming. Whether or not we have here a nuclear zone in its own right, the

clear cultural affinities east and west of the Aegean (from the earliest pottery sites onwards and, by implication, probably

earlier) make the traditional labels of diffirsionism and migration ill-fitted when viewed from the perspective of this Aegean

koine, the more so if one accepts that a tradition of seafaring, automatically involving the transmission and filtering ofknowledge ofall sorts and degrees all around the coast, was indeed ofgreat ancestry

While simple pattems ofpriority and direction conceming the neolithisation of Europe probably misjudge the importance

of the Aegean Sea, masking the complexity atplay,I want here to discuss just such patterns in assessing the distribution ofearly farming sites north of the Aegean, viz. the southem Balkan region including northem Greece, the Former Yugoslav

Republic of Makedonija2 and Bulgaria. The basic assumption here is that, initially at least, the neolithisation of Europe was

indeed an Aegean phenomenon, and that it first affected Greece (Thessaly and Greek Macedonia), whence it spread northward

and eastward in chronologically decreasing steps. Although the temporal dimension in the regions concerned suggests a

'wave of advance' model as proposed by Ammerman and Cavalli-Sforza3 , I want to stress that I neither subscribe to the rigidconstant these authors see at play from SE Europe up to NW Europe, nor to the implied reiteration of the processes having

taken place in SE Europe over the whole of the continent. Acknowledging, with Nandris, that the neolithisation of Greece

and what he quite appropriately called the Macedo-Bulgarian zone,can directly be explained out of the Aegean background,

I feel, withZvelebil, thatEurope north ofthe Danubeunderwent aratherdifferentpathtowards farming and animal husbandry

not in any way related to the tell-oiented mode of life in the southa.

Limiting myself to patterns of priority and direction atplay in the Macedo-Bulgarian area, the aims of this paper are

to clariff these patterns in terms of the chronology and stratigraphy of the key sites involved, in order to check myhypothesis that - with the discard of NW Turkey as a contacl zone area - the early neolithic Bulgarian sites had their originnot in the east, but in the west.

As a tentative base line I follow Van Andel and Runnels that the process of settling in Europe beyond the area ofinitial farming that was Thessaly and that includes, as I propose, the Giannitsa Basin in Greek Macedonia as well, consisted

of selecting environments similar to the region of origin, in order to be able to maintain and continue the pattems of lifeestablished by traditions. While the data on site distribution and location for Southeastern Europe are still inadequate, itis unlikely to be a coincidence that the early sites concentrate on the edge of well-watered, small alluvial plains and in the

I French 1967,50.2 Henceforth shorteried as Makedoniia and to be distinguished from the Greek district of Macedonia.3 Ammerman/Cavalli- Sforza I 984 ; Cavalli-S forza 199 6.aNandris 1970; Nandris 1977:'Zvelebll 1986, 184; ZvelebillZvelebil 1988.5 cf. Van Andel/Runnels 1995.

t93

immediate neighbourhood of perennial springs -just as was the case in West Turkey. North of Thessaly, such territories

include the Giannitsa basin, the Ovde Polje and Kodani plain in Makedonija, the upland plains ofKjustendil and Radomir/

pernik and the Sofia Basin in Westem Bulgaria, and, finally, Thrace - the large plain drained by the Marica river in Eastem

Bulgaria.

2. The Giannitsa Basin

North of the eminent key-region for farming that proved Thessaly (exemplified by the profusion of early sites and the

large degree of continuity in time), the earliest neolithic settlements in Europe are encountered in the Giannitsa Basin. Given

the absence of a large concentration of early sites6 and the discontinuity of occupation in this area, it may be considered that

- barring conventional environmental limiting factors - different cultural conceptions regarding commitment with the

surrounding 'land' and the village, in combination with built-in inclinations to 'move on' underlay the settlements in the

Giannitsa Basin when compared to Thessaly. Consequently, the first neolithic occupation in the northern region might have

represented an event not necessarily dependent on the south. From a Thessalian perspective, the success ofthe new mode of

subsistence may not have provided an immediate reason to establish new sites away from the mother country.

If one allows incentives for settlement shift, involving migration, to have been part and parcel of the cultural luggage

of the settlers ofNea Nikomedeia, this hetps in coming to an understanding of the possible origin of the settlements in the

Ovde Polje and Kodani area further north - simultaneously explaining the affinities in material culture between both parts

of Macedonia. The fact that in both regions occupation was halted after only one or two generations of life, to be retaken

only some 300 to 400 years later, serves then as a link in spirit.

With the recent publication of the final reporl on Nea Nikomedeia also new radiocarbon dates have become available

beyond the questionable ones known for a long timeT. Although additional information conceming the stratigraphic

context of these samples is not provided, an estimation of the site's date is now possible with more accuracy. Assuming,,that the events being dated all occurred within a short period"8, the probability distributions of the calibrated dates were

combinedthroughtheOxCalprogramv2.l8,withtheexclusionofQ-655, GX-679,P-1202andOxA-1603+4280,which

showed very poor agreement within the overall pattern (Fig. 1)'.Largely stemming from short-lived sample material (bone, seeds), the combined distributions are most likely an

indication of the end ofthe settlement. The two-peaked result at the 1o confidence level being equally distributed, either

one of two is true. In view of the close cultural affinities to Anza I, to be discussed shortly, I prefer here the younger 1o

range (6090-6060 cal BC) as representing the end ofNeaNikomedeia, since more in accordance with the AnzaI Cra dates.

The beginning of life at the site, then, would have occurred a few generations earlier - probably somewhere in the second

half of the 62nd century BC.Although Nea Nikomedeia is usually incorporated in the Thessalian periodisation, and in that case dated slightly later

than the first neolithic sites therer0, this inclusion masks the possible independent nature of settlement, as was suggested

supra.Thecircumstance thatNeaNikomedeia's material culture is undoubtedly 'Aegean' (like that of EN Thessaly, like that

of West Turkey), does not make the site historically dependent on Thessaly. Consequently, where subsistence settlement in

Thessaly musthave started not much earlier than the 62nd century cal BC as well (Thissen 1997), it is wananted to view Nea

Nikomedeia as being part of a different, but culturally related as well as contemporaneous, event.

3. The Ovie Polje area

Anza,the main early site in Makedonija, is an archaeologically slightly more complex site than Nea Nikomedeia. It has

perhaps more suffered than profited from being excavated by three different teams. The excavations carried out by

Gara5anin on the central part of the mound are still not published beyond the most summary levelrl, while Gimbutas'

soundings did complicate matters beyond necessityr2 . The results of an earlier sounding, in 1960, are diffrcult to reconcile

with the lg69J0 evidence, but do confirm the sequence in rough linesr3. The importance of the site resides more in the

availability ofa large series ofradiocarbon dates than in sequentional and documentary precision.

Illustrating the confusion concerning the site's evidence is that Gara5anin and Gimbutas reached a different sequence

from their respective soundings (Table 1). It may be assumed that the stratigraphy worked out by GaraSanin is the more

reliable one, stemming as it does from a large contiguous area. The Americans, sinking small trenches in various,

discontiguous locations of the mound, failed to sound coherent arbhitectural struch{res.

6 Cf. the map in Van Andel/Runnels 1995,482.7 Pyke/Yiouni 1996, 195.8 Bronk Ramsey 1995.e The Cra dates used in this paper are catalogued in Table 8Io Cf. Wijnen 198 l, 73.I I Gara5anin 19'7 4; Garai;anin 1982, 87-93.12 Gimbutas 19761, cf. Milojdi6 1978.

'3 Koroie6/Koro5e6 1973.

194

Table 1. Anza sequence after Gimbutas and GaraSantn.

Gimbutas GaraSanin

BURNTIVbIVara

GAP

mI

GAP

IbIaIa (pits)VIRGIN SOIL

NO BURNING ATTESTED

IV (pits only)

NOGAP

III (post-wall houses)II (post*wall houses)(II) "Graben", overlaying Ic

GAP

Ic (mud-brick houses)Ib (mud-brick houses)

VIRGN SOIL

For assessing the onset of occupation of Anza, the Cra dates from Anza I do not suffer from the stratigraphic unreliabilitywhich according to Milojdi6 affects the Anza II-IV levels dug by Gimbutasrs. AnzaI is represented by two consecutive

building levels, both unique in the Balkans for the use of mud brick as primary building material. Analyzing the radiocarbon

evidence, I hav etteated Arual and Anza IHV as two separate entities, meaning that their dates were considered independently.

Assuming that the Cra samples from Anza Ia and Anza Ib each stem from events occurring within a short period, foreach level the probability distributions of the calibrated dates were combined, with the exclusion of LJ-2333 from level Ib(Fig. 2). The end of Anzalamay have occurred anlwhere between 6010-5970 cal BC. The end of Anzalb most likely falls

within the earlier I o range of 5990-5 940 calBC in view of the overlap with the Ia range. Two points can be stressed, firstlythat the occupation of Anzal came to an end before 5940 cal BC at the latest; secondly, that AnzaI predominantly dates

to the first half of the 60th century BC. As such, the site was settled two or three generations later than the end ofoccupation at Nea Nikomedeia.

Compared to Thessaly, where occupation continued uninterrupted, Anza I correlates chronologically with the

series of dates from Achilleion IIIb, representing the beginning of the Thessalian Middle Neolithic period. Both

Gara5anin's and Milojdii's correlations of Anza I to a final Early Neolithiclstart Middle Neolithic stage are possibly

correct, though they may be relegated more to technical 'nouveaut6's' (e.g., the 'scraped ware') than to really structural

affinities. The lack at Anza of such characteristic Thessalian pottery categories (angle-necked cooking pots, bead-

rimmed bowls, the 'Sesklo' painting style) may carry the heavier weight, marking the cultural distance between the

north and the south.After the burning of Anza Ib (or Ic in Gara5anin's sequence) further settlement was obstructed, as was the case also

regarding the contemporaneous occupation of Vr5nik I in the Kodani area slightly to the east. Both the Ovde Polje and the

Kodani plain were apparently discarded as a potential area for early farming. Village life in the area was only restored

during the Stardevo period some three centuries later.In the present overview I will not consider the potential candidates for early neolithic settlement lying further to the

north (Divostin, Donja Branjevina, Lepenski Vir or Gura Baciului-to mention only the more famous ones), nor such sites

to the southwest (Porodin and Veluika Tumba in Pelagonia; Podgori I, Vashtemi, Barg in Albania). All these sites have

their specific problems in terms of stratigraphy, dating and cultural origin which go far beyond the limits of this paper.

Apart from the amazing painted pottery from Podgori I with its blend of Anza I, West-Thessaly (Prodromos) and an

undefined local componentr6, the closest links to the early Macedonian style of painting are to be found in Western

Bulgaria. in Bulgaria also, the white-painting style knew its most prolific development, both in time and in space.

4. The Upper Struma area

Although the connection East Makedonija and West Bulgaria is difficult to establish unequivocally, at least during

the initial stages of occupation, a cursory categorisation of both regions' early, pottery assemblages seems to speak infavour of a certain cultural cohesiveness, and I prefer here to consider the connection Anza-Struma as a valid oner?.

Geographically, several west-east communication routes exist, linking the Vardar and Struma catchments18.

'a Gimbutas' division of Anza IV in a IVa and a IVb level is not verifiable on stratigraphic grounds, and appears to have been made forpurely typological reasons (cf. Mount-Williams 1976, 117).

'5 Milojdi6 1978..

'6 See Prendi 1 990, Figs. 6-1 1 .

ri I am much obliged to Dr. J. Pavrik and A. Bakdmska for allowing me to study a random sample of Gdlibnik pottery in some detail.r.s The plain of Kjustendil is linked to Makedonija by the Velbdzdki pass. The only other connection between the upper Struma area with

195

The one important aspect of relevance in this paper is the fact that West Bulgaria's first farming sites can be argued

to date again some generations later than the abandonment of the Ovde Polje and Kodani area. In the Struma valley,

among a cluster of sites interlinked through their material culture, the Gdlibnik site is at present the only one with a

substantial deposit covering several centuries without any serious hiatusre. In addition, Gdlabnik is the only site dating

back so early having a reasonable amount of Cra dates.

An assessment of the true age of the Gdldbnik sequence is, unfortunately, hampered by the very poor overall

agreement ofthe calibrated ages, each building horizon yielding a date range stretching from 6000-5500 cal BC (Fig. 3).

Given the stratigraphic order, at least the two early dates from Il:viii (Bln-4093 and GrN- 19783) which overlap with some

of the dates from the preceding levels may be omitted from further consideration. Assuming then that the three

remaining Il:viii dates represent a single event (i.e. - since being short-duration samples -the destruction of level

Il:viii), it is legitimate to combine their probability distributions. The result, shown in Fig. 4b, indicates that the end of

Gildbnik Il:viii either falls between 5605-5575 cal BC orbetween 55 45-5525 cal BC (at 1?). Preference is here givento the

earlier option. Gdlabnik Il:viii, showing clear Stardevo traits in the pottery and figurines, correlates well with the C'a

dates from the type-site - Stardevo - itself. In addition, the stylistic links of Gdldbnik Il:viii with Anza II-IIP0, and with

the onset of settlement at Kremenik/Sapareva Banja are confirmed by the latter site's Cla dates as we1121. The end

roughly being set at about 5600 cal BC, the beginning of village life at Gdldbnik, dating back seven building levels,

would fall somewhere between 5900-5800 cal BC (cf. Fig. 4a).

Regarding Kovadevo and Krajnici, the evidence from both sites is difficult to assess accurately, while Cta dates are

lacking. At Krajnici, the deposit containing 'monochrome' pottery was sounded in a restricted area. The abraded nature ofthe sherds and the typological similarities with the later Krajnici material form a rather weak basis for singling out the site

as having yielded unquestionable evidence ofa very early neolithic horizon comparable to Greece. Indeed, the existence

of a 'monochrome' pottery horizon, signalling an early stage in pottery manufacture (equivalent to Theocharis' EN Iperiod and Milojdii's "Friihkeramikum"), is losing ground all over Western Turkey, Greece and the Balkans, and goes in

fact back on dubious evolutionist conceptions22. The 'monochrome' and allegedly very early pottery of the Koprivec site

near Ruse - which is now grouped together with basal Krajnici - seems to correspond better with the technically outstanding

Romanian pottery as found at, e.g., Glavaneqtii Vechi in Moldavia, or at Moregti in Central-Transylvania23, assemblages

dating to a much later time frame. The assessment ofKovadevo's role in the pattern of early farming sites must await a full

publication of the finds. The presence of painted pottery at the site linked both to the south (Toumba Serres and Greek

Macedonia2a) and to the upper Struma region, would suggest a sequence encompassing several centuries - a condition as

yet not clearly apparent from the stratigraphy itself.

5. Ele5nica- SlatinaBoth Elesnica in the Mesta valley and Slatina in the Sofia Basin confirm the G[ldbnik sequence in rough lines and

again do not contradict the chronological pattem of decreasing age going from west to east. Ele5nica combines a rcther

straightforward stratigraphy, consisting of only two building levels (both burnt), with several Cra dates stemming fromthe

lower level2s. On the basis of the ceramic material, Elesnica can be rather securely correlated with the second half of the

Gdl6bnik I period, and most probably dates between 5750-5700 cal BC. Ele5nica's zigzag-gtooved ('channelled') pottery

some of its white-painted, as well as the plain ware categories correspond fully with G51dbnik26. These parallels show that

the Mesta valley was culturally strongly linked to the Struma valley in the west27. Basal Slatina conforms largely to

Elesnica's material culture, while stylistic parallels with G6l5bnik I:v-vii are equally strong2s. The Slatina radiocarbon

dates, stemming from the basal level IV only, suggest contemporaneity with the Ele5nica site2e.

the Vardar catchment goes by way of the Bregalnica river and the Deldevski pass through the Blachuna Mts., which form the watershed

between the Struma and the Bregalnica and also the border between Bulgaria and Makedonija. This pass connects immediately to

Blagoevgrad. In this respect, the excavations in the Kjustendil area at Vaksevo by S. iohadZiev and in the Blagoevgrad basin at

Bdlgardevo by L. Pernideva are ofmuch televance, but not yet published.re P.c. A. Bakdmska and Dr. J. BojadZiev; cf. also Pavrik/Bakdmska 1989.m Cf. Pardktsakdmska 1989,224f.2' Likewise J. BojadZiev, this volume.22 Todorova 1989; Todorova/Vajsov 1993,74.23 For Glavanegtii see Comqa 1978; for Moregti see Horedt 1959. Like at Koprivec, the assernblage at Glavanegtii is characterised by

chaff-tempered, red-firing slipped pottery with bumished surfaces. Cores are, likewise, non-oxidised. Nail-impresso occurs, barbotine is

also hardly present. Other correlations: biconical shapes, plain or beaded rims, vertically-pierced knob handles, tubular lugs (Comqa

1 978, Figs. 15:6-7 , I 6:9), diskbases, - even white-on-red painted occurs, present at Koprivec as well.2a Cf. Koukofli 1995,565f.2s Gorsdorf/Boj adZtev 7996, 126f.:6 Cf. M. Tao, this volume.27 The Gradevska river, a tributary of the Struma, links the Struma valley with the upper Mesta at Razlog and Bansko via the Predela pass

(ll40m).28 The Sofia basin is to the northeast of Pernik and connected with that region via a pass between Mt. VtoSa and the Ljulin Mts.2e Cf. BojadZiev 7994.

196

6. The Marica BasinEast of the Sofia basin and the Mesta catchment begins the large Marica Basin or Thracian Plain. All sites in this

catchment, from west to east, show a strikingly similar material culture as well as remarkable conformity in settlement planand building methods. This cultural homogeneity can be followed throughout the whole prehistoric period. While, stylistically,the ties with the west are not an issue, it is the chronological dimension, and, as a consequence the cultural dependence whichwill be discussed here. It is still common good to view the Thracian culture as beginning roughly contemporaneous with theGreek EN, the 'ProtoStardevo' period or the white-painted cultures in West Bulgaria3o. The decisive argument in favour has

been the Thracian white painted pottery. It should be noted that the recent pushing back of Bulgaria's early neolithic periodto a'monochrome', pre-Karanovo I stage did not question this early beginning of village life in Thrace, but simply insertedan even earlier start of the neolit[ic in the counffy3r. In view, however, of the absence of clear ties between Thrace andNorthwest Turkey during the first stages of the nebHthic there (e.g. sites such as Fikirtepe, Pendik, Ihprnar, Mentege), and

taking into account the stepwise decrease of age for the beginning ofthe early neolithic as traced from Anza to the east, it isnecessary to reassess the.early date allotted to the Thracian sites. A logical consequence ofthe west-east pattern would be

that the Marica Basin was settled much later than the Struma area, and slightly later even than the Sofia and Mesta basins.Three Thracian sites will shortly be discussed, which apart from having each a large amount of Cra dates, yield good

stratigraphic sequences with strong pattems of continuity and cultural coherence.

6.1. tavdrBecause of its location in the most westem section ofthe Marica Basin, davdar would, following our west-east model

of decreasing ages, belong to the earliest sites settled in Thrace. A cursory glance at its C'4 dates would confirm thisview32. Cavdar conforms fully to the Thracian culture concept, both in settlement lay-out, house building, and materialculture, as has been extensively argued by its excavator, Georgiev33. Only a minor part of the pottery is in the Sofia-BasinKremikovci I 'style' - characterized by a brown or wine-red decoration on a pale-rose ground3a. As is the case at TellAzmak, also in Cavdar 'channelled' decoration occurs side by side with white-painted decoration from the basal levelsonwards, although the latter seems decreasing in the later levels. On the basis ofthese correlations, it can be supposed thatthe sequence from iavdar runs parallel to the Tell Karanovo I-II and Tell Azmak periodization.

The stratigraphic sequence of levels VI-I demonstrates a high level of continuity, without any major intermptions(Table 2)3s. Both houses of level V are, for instance, built over the burnt structures of davdar VI, maintaining the samebuilding plots. Also level IV structures seems to shift only slilhtly in relation to level V below it. It appears as if house

Table 2. iavdar stratigraphic sequence.

level building method maintenance

IBURNTIBURNTru

IVVBURNTVI?vII37

'cob on posts'36

no data

no data

oven/hearth place: floor replastered two times

preserved height of the walls up to 45cm. Inside of wallsreplastered at least three times. Oven plate three times renewedno dataa domed oven shows four replasterings

walls standing to 45cm; oven plate shows three renewals

30 Cf. Champion/Gamble 1985, 114 Fig. 5.2,117 Fig. 5.5; Lichardus 1985, 354 Fig. 28; Demoule/Perlds 1993,366Fi9.2.3'Todorova 1989.32"Imallgemeinen giltdie Datenserie des Wh. 6 [i.e. CavdarVI-LT] als die fniheste flir das FnihneolithikummitbemalterKeramik (...)"- Gorsdorf/Boj adLiev 199 6, 126.33 Georgiev 1981, 99.3a Georgiev 1981, 87.35 Dennell observed, from a small sounding he made immediately south of Georgiev's dig and "through the tell at its greatest height",that "There was no evidence, either in the form ofweathered horizons or sterile layers between deposits, to indicate that the settlementwas not continuously occupied until it was finally abandoned" (Dennell 1978, 83).36 This is most probably the building method used all over the Balkans and NW Turkey (e.g. Ihprnar) during the neolithic. It diffbrs from thewattle-and-daub tephnique in that the weight-bearing structure (the posts) are connected by means ofa lattice ofhorizontal posts, not by wattle.The remaining open spaces were filled in with lumps ofmud-clay (or taufl, occasionally called 'cobs' (cf. Houben/Guillaud 1994, 188f.).3i The exact nature ofa seventh buiding level, unpublished but reported to exist (Georgiev 1981, 65f.), seems contradicted by the factthat level VI is said to have been built over "a sterile level covered by a layer consisting ofcharcoal and ashes" (ibid., 79).

197

plots at e avdar remained fixed to their original locations - rebuildings shifting only in a minor way to avoid the rotten post

stumps of the predecessor houses. It is of interest to note that this picture conforms entirely to building patterns observed

for Karanovo, Ihprnar and probably Tell Azmak as well.The observed continuity of occupation, the preservation of tradition through the choice of building plots, and the

strong coherence in material culture tightly corresponding to Karanovo I-II do not support a very long occupation ofdavdar as is suggested by the radiocarbon dates. The Cla dates from level VI range from 620G-5400 cal BC, and this span

is nearly reproduced for level III (6000-5400 cal BC). Indeed, the ten dates from level VI can be grouped into four mutuallyexcluding clusters (A-D) at the I o level (Fig. 5). Were we to consider these clusters as each representing different events,

a combination of their probability distributions would yield the following result (Table 3):

Table 3. Tentative clustering of Cavdar VI radiocarbon dates, based on a combination of the probability distributions.

Cluster aee BC at I o AgreemenFs

D

C

B

ss30 - 5430 (0.95)

ss60 - sssO (0.0s)

s695 - 561s (1.00)

s880 - s830 (0.62)

s940 - s910 (0.38)

6040 - 598s (1.00)

single date

single date

Il8.9Vo

l5l.6Vo

In theory any one of these clusters may represent the true age of iavdar VI. A similar discrepancy of dates

stemming from the same building level.is reflected in the calibrations for the subsequent level V. In addition, correct

interpretation of the davdar dates is further hampered by the poor agreement of ages stemming from the same sample. Achi-square test fails, for instance, when trying to combine the probability distributions of Bln-907 (6320+100 BP) and

Bln- 1030 (6760+ 100 BP), both dates stemming from a short-duration sample ofpeas from the bumt level III. In view ofthese

shortcomings, the davdar dates can only be used with the utmost caution.

6.2. Tell AzmakOn the basis of the architecture (location, dimensions, orientation, etc.), domestic fittings and their positions, as well

as the movable finds, the correspondence between Azmak and early Karanovo is complete (Table 4). In view ofthis total

parallelism and the proximity of Azmak and Karanovo, only 25 km further to the east, it is assumed here that both sites

have been contemporary for a large part oftheir sequenges.

Table 4. Tell Azmak stratigraphic sequence.

Level building method maintenance

II1=1'6,I:5I:4BURNTI:3I:2BURNTI:1VIRGIN SOIL

no datano datano data

'cob on posts' some walls preserved up to 70cma wall preserved for 45cm

several renewals of plaster on floors, walls and ovens

Level I:2 is mentioned to have been built directly over level I: I , using the burnt remains both as a sort of drainage

system and protection against the damp underground3e. Level I:3 "saw no marked differences in the form ofthe houses"ao.

Moreover, several finds from level I:3 cut directly into level I:2, suggesting maintenance of building plotsar. Considering

the absence of any clear stratigraphic hiatus as well as the probable retainment of house plots, and also the homogeneity

in material culture throughout the different building horizons, it is more than likely t\tt the whole sequence at Tell Azmak

was continuous. Tell Azmak levels I:1-4 (occasionally I:5 as well) are assigned to the Karanovo I period by Georgiev,

although the classic 'guide fossil' ofthe Karanovo II period, viz. 'channelled' pottery occurs throughout the deposit of

38 Agreement ishere a measure ofhow well a (posterior) distribution (which is acquired by, e.g., combining the probability distributions) agrees

with the original distribution of dates. The threshold for accepting the agreement as good is set at abo,tt 60%o (cf. Bronk Ramsey 1995).3e Georgiev 1965,7; Georgiev 1967a,94.ao Georgiev 1965,7.a' Cf. Georgiev 1967b,147-8,pt.5.

r98

Azmak I:1-5, side by side with that guide fossil of the Karanovo I stage, white-painted 'tulip beakers', which are, however,

decreasing in the upper levelsa2.

The Azmakradiocarbon dates display the same phenomenon as demonstrated for davdar, viz.theoccurrence oftwomutually excluding date clusters (A-B) within the same building level. Here, this discrepancy is restricted to basal Azmak,

the later levels yielding rather more homogeneous combinations (Fig. 6). The combined probability distributions per level

are tabulated in Table 5.

Table 5. Tell Azmak. Combined probability distributions per level.

level age BC at I o Agreement

II (I:6)

I:5

I:4

I:3

I:2

I:l/cluster B[:l/cluster A

5490 - 5280

5330 - 5060

s450 - s290 (1.00)

ss45 - ss25 (0.24)

5620 - ssTs (0.76)

ss50 - 5s30 (0.10)

5670 - ss70 (0.90)

s71o - 5600 (1.00)

6t1O - s970 (1.00)

)

)

single date

single date (too late)

I25.6Vo

66.2Vo

12l.1Vo

lL4.I7oI12.89o

It is obvious that the large spans at lo are determined by the large errors ofall the dates, all larger or equal to ti00.Most concentrated in time is the result from AnnakI:3. Counting backwards from either the 5620-5575 or the 554tr-5525 BC

span does not disqualify thel:2 span, nor the younger I:1 span (if one were to take each building level as lasting one

generation). It seems certain that life at Azmak did not begin earlier than the 57th century BC.

These calculations demonstrate, first, that the total time range for Azmak I: I to I:3 mrlst have been short (and, byextension, for Azmak I:4 to II as well), and secondly, that the two older dates from Azmak I: 1 are most likely to be too old.

6.3. TelI KaranovoAs at davdar, the first occupation at Karanovo was laid out on an obviously prepared surface, as attested by a white

ash layer covering virgin soila3. Subsequently, building horizons I.v and I.iv were built, entailing shifts in orientation, butprobably in continuous succession. In building horizon iv some additions were made on the existing plan, making a

division in an original phase iva and the later ivb necessary. Renewals of ovens on the same location also is indicative ofconsecutive readjustrnents while maintaining the house area itself. As clearly argued by the excavators, the earliest occupation

was not yet strictly aligned to a roughly N-S running 'street', which determined for a large part the location and orientation

of the structures from building level I.v onwardsaa. Put otherwise, the earliest building level differs from all subsequent

occupation in terms of planning, but not regarding building material, method or oven locations. While it is of course

aftractive to view this deviation in terms of a pristine occupation taking place on a new location, this remains mere

speculation given the restricted areathat was investigated. Occupation became slightly more dense during the latest

Karanovo I stage (iv).The Karanovo II period immediately follows upon I, starting with building horizon Il.iii. Both during Karanovo I

and II people stick to the same house plots. Rebuildings took place on the same locations, only the walls shifted slightly

as opposed to the earlier ones, probably in order to avoid the earlier foundation trenches, likely to contain incidental

rotten post stumps. An exactly similar phenomenon was observed at Iltpmar phases IX-VIII45. The pattern of building

during the Karanovo III period is hard to establish, since remains of this period were less well preserved in the new

trenches. Contrary to the 'sterile layer', as claimed by Georgiev, however, the alleged hiatus between Karanovo II and IIIhas now been refuted. Instead, the II{II transition is smooth and continuousa6. On the basis of the preliminary reports, the

Karanovo sequence may be summarised as follows (Table 6).

a2 Georgiev 1967b, 150.a3 HillerA,likolov. 1990, 25.a HillerA.trikolov 1 990, 17 ff ., 26 ; HillerA.{ikolov 199 l, 7 .

as Cf. Roodenberg 1995, 38.a6 HillerA{ikolov 1988, 24.

199

Table 6. Building history for Karanovo I-I[.period number of building levels

mII/III'transitional level'trIVIRGIN SOILTotal building levels:

+-5 including one (or two?) bumt building levelsI no archit. features5 (iiij) * two mutually shifting clusters of superimposed oven floors attested in Ql7/ll3 (vi-iv)

I3_14

The radiocarbon evidence from Tell Karanovo is remarkably similar to eavdar and Tell Azmak. Again, there are twomutually excluding date clusters in the basal Karanovo I stage (Fig. 7). The older cluster (A) deviates strongly in age fromthe younger one (B) (Fig. 7). This younger cluster, in addition, agrees with the trend in the Karanovo II and III dates -indeed, the younger Karanovo I cluster overlaps with the Karanovo II cluster. Again, one may surmise that the old dates

are too old. The overlap of dates suggests further that the duration of the individual building levels at Karanovo was short.

A date in the 54th century cal BC for the buming ofthe Karanovo III level must be considered reliable because of the

short-duration samples included (lentils and grain). Since the Karanovo II samples seem to stem mostly from the earlierpart of that phase, but the Karanovo III samples from an advanced stage of the Karanovo III period, the age gap betweenthe two ranges is justified. It is, therefore, probable that the Karanovo II period bridges the 57tW56Ih century hansition.In addition, Karanovo II would occupy most if not all of the 56th century BC. Consequently, the three building levelsmaking up Karanovo I should date back not much earlier than 5700 cal BC.

TableT . Tell Karanovo. Tentative grouping of Karanovo I-III radiocarbon dates, based on the combination of probabilitydistributions.

period/cluster age BC at I o Agreementm

IVcluster B

V'transitional'

Vcluster A

s38s - s33s (0.s5)

s430 - s39s (0.4s)

5668 - 5621 (1.00)

s640 - s59s (0.97)

5660 * s6s5 (0.03)

s840 - s720 (1.00)

590s - s890 (0.36)

s97s - s94s (0.64)

104.9Vo

93.97o

IOL.2Vo

single date (Bln-4338)

189.87o

7. ConclusionWe have observed that in all three Thracian sites with a continuous sequence and a large series of radiocarbon dates

there is an as yet unexplained error conceming the earliest data. These all fall within the 6 lst-{0th centuries BC (at 1 o). Inall three instances these very early data co-occur within the same building levels with much later dates, falling in the 57thcentury BC. In view ofthe good agreement of these later clusters with the dates from the succeeding levels, and consideringthe continuity of occupation known from all three sites, the adherence to building plots over the generations, as well as the

homogeneity of the material culture, I feel that a short chronology more likely represents the true order of things. I view,therefore, the younger date clusters from the basal levels of iavdar, Tell Azmak and Tell Karanovo as the correct age

indicators.Bulgarian Thrace seems to have been settled in 'one sweep', to put it roughly. The material culture is quite homogeneous

from west to east. A slight ancestry for davdar over the eastern sites would confirm a west-east decreasing age, but cannotbe substantiated from the Cra dates. An approximate date for the beginning of settlement in Thrace can be set at about 5700cal BC - much later than the first farming sites in West Bulgaria. It implies that the Bulgarian white-painted pottery cannotbe taken as a strong chronological marker, and suggests further that the Thracian white-painted style was stylisticallydependent on the West Bulgarian white painted potterf7. Equally, the ancestry of Westem Bulgaria over Thrace suggeststhat the technique of grooving and channelling might be first in West Bulgaria, and adopted later in Thraceas.

The west-east decreasing age for the beginning of the neolithic seems - with due reserve - confirmed going furthereast, towards the TundZa river and beyond, into Turkish Thrace. In the lower TundZa valley, Lichardus failed to discoverany Karanovo I material during his surveysae. In Turkish Thrace, the earliest sites detected during the survey work carriedout by Ozdo$an and others date to a Karanovo III contextsO. The excavations at Aga$r Prnar in Turkish Thrace yielding a

ai Recent information suggests indeed that white-painted pottery still occurred during the Karanovo III period (V. Nikolov, thissymposium, citing evidence from Tell Kazanldk).48 At Gdldbnik occurring in levels I:v-vii. Cf. J. Pavrik, this volume.ae Lichardus 1993, 86.

200

sequence of six building horizons, date back to a Karanovo IilI[ transitional phasesr. Although in a seventh level white-painted tulip beaker and lid fragments have come to light52, the restriction of this kind of pottery to the Karanovo I periodis now becoming less and less certain. The chronological framework as proposed in thispaper is schematized in Fig. S.

Table 8. Catalogue of Cra dates used in the text. Order is per level, from old to young, based on OxCal calibration [nd=nodatal.

NEA NIKOMEDEIA (n=12)53q655GX-679P-t202OxA-l605+OxA-428251OxA-3876OxA-3874OxA-l606+4283OxA-3873OxA-3875P-l203AOxA-l604+OxA-4281OxA-l603+OxA-4280

LJ-2181

I+3032LJ-2330t31LI-3186LJ-3183u-2157LI-2341LJ-2332LJ-2339I+2342LJ-2333

Bln-3580BkF3579+Bln-3579HBln-3582BlrF3581GrN-l9786GrN-I9785GrN--l9784Bln-4095Bln-4096B1ft4094Bln-4093GrN-l9783B1ft4091

8180+150

7780t2707557917399fi31370+p/0

1370l:80

732b667300180

7280807281t747220437010lf,6

727o,j140

721CE50

7t701f,07tQ!707150150

6910A907230t1701lt0rl207120x80

7100180

68401100

charcoalcharcoalcharcoalhord. vulgarebone (szs)bone (capra)lens cul.bone (ovrs)bone (sas)

charcoaltrit. dicc.trit. mon.

charcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoal

ndndndndndndndndndndndnd

ANZA Phase I (n=11)55

GALABNTK (1:15)56

IaIaIaIaIaIbIbIbIbIbIb

I:iI:ii:iI:iI:ivI:vlvii:viiI:viiI:viiII:viiiII:viiiII:viii

7120!707ll2r526950!706790+80

70701180

7020t607070+60

7U}tt50714&j806760180

7100180

6970150

676W60

charcoalcharcoalAumic acidcharcoalcharcoalcharcoalgraingraincharred woodcharred woodcharred woodwheatgrainwheat

50 Ozdopan 1985;1996; Erdogu 1997.5t P.c. dzdolanlP arzinger 7997 .s2 dzdolanlParzinger 1997, 60.s3 Pyke/Yiouni 1996, 195.5a Radiocarbon dates from the same sample were combined prior to calibration with help of the OxCal program (cf. Bronk Ramseyl 9es).5i Gimbutas 1976,.30.56 GdrsdorflBoj adliev 1996, l22f . The GrN-samples were collected by the writer with the kind permission of Dr. J. pavrik and A.Bakdmska. I am much obliged to the Dutch Foundation for Scientific Research for making this financially possible. The GrN-dates arehere published for the first time.

201

81ft409281rF3576

Bln-I583Bln-I580Bln-2108Bln-I663Bln-l582BIn-l579Bln-l581Bln-I578Blft-2662B1fF2TO7

Blft-4261Blft4l06Bln-l241+Bln-l24lABln-l162+BhFll62A+8kF1251

[NB. Not further used: Xltest fails at 5%]

BkFl l60+Bln-1160,4'

[NB. Not further used: Xltest fails at 5%]

BlrF908+Bln-998"BkF911Bln-909BkF910+BkF9l0ABkF907+Bln-1030

[NB. Not further used: X2-test fails at 5%]

Bln-906

81ft-293Bln-291Blft-292B]fFzg481ft-296Blft-295BhF20381ft-2998k-26781ft-22481ft-297Bln-298Bln-300BkF301BIn-430BkFl40A

Blft4l79Bln-4339B.1ft4177Bln-4336Bln-4338B.1ft4337Blft-3942

6710!606670L70

iavoan (n:21)51

7208i527202*557t95fi5'7070fi0

7020t457m3t457000160

6994!55

6820150

6550r507128806840+50

6891r70

(6794is7)

(6860a70)

7023836814!1206815r1006610x70(6s40fl0)

6720!t00

wheatgrain

charcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoalcharcoal+graincharcoal

charcoalcharcoal

charcoal

charcoal

charcoalcharcoalcharcoalcharcoalpea

charcoal

charcoalcharcoalgraincharcoalcharcoalgraincharcoalcharcoalcharcoalcharcoalgraincharcoalcharcoalcharcoalgraingrain

II:viiiII:viii

IV

VIVIVIVIVIVIVIVIVIVIVVv

l1I1llI:1

I:2

I:2t3[3I:3

I:3

t3I:J

L4I:4

t5II

mmmmm

AZMAK(n=16)5873031150

71581150

68781100

67681100

6779+1ffi6720+100

68701100

6812+100

6758+100

66501150

6675!IN65401100

&26*t506483f100079tl20&76t100

KARANOVOb (n=22)5e

7t34fl07090i907110i507ll0r506955*45

6810+65

682Gj50

charcoal

. charcoalcharcoalcharcoalcharcoalcharcoalcharcoal

57 Gorsdorf/Boj adliev 1996, 124ff .s8 Gdrsdorf/Boj adZiev 199 6, 133ff .5e Gorsdorf/Boj adZiev 1996, 131ff.

202 )

r 81ft4178BlrF4335

Bln-3586+Bln-3716+BkF37l6HBlft-3944B.1ft-3943

B1n-3941

Blft-346281ft-34&B1F3461Bln-3458Bln-3460Bln-3587+BkF3717+Bln-3717HBln-3459Bln-3465B.1ft-3463

BkF3585+Bln-3904

[NB. Not further used: Xa--test fails at 5%]

6730180

6710+55

6Wr346785160

61&x50675Gr50

6510160

6500150

6480+60' ffi*ffi

ffir60

charcoal Icharcoal I

charcoallhumic acid trcharcoalcharcoal

II

charcoal trcharcoal mlentil mcharredbeam mcharcoal mcharcoal Itr

grain/humic acid mffi6t34&20*ffi6410160

6350160(6233!4s)

charcoallentillentilcharcoal

mmilm

u Bln-998 is most probably equivalent to sample Bln-908A as it was called in Georgiev 1981, 108.

b The radiocarbon dates from Georgiev's excavations (Bln-l 52, I 58, 20 I and 234) have not been included, since their stratigraphic relation to the new work is unknown.

ReferencesAmmerman/Cavalli-Sforza 1984 A. Ammerman, L. Cavalli-Sforza. The neolithic transition and the genetics ofpopula-

tions in Europe (Princeton 1984).

Bojadliev 1994 J. BojadZiev. Datirane po radiovdglerodinja metod na rannoneolitno ZiliSte ot Slatina(Sofrja) (Datation au radiocarbone d'un logement du haut n6olithique a Slatina (Sofia).

- Archeolo gij a, 1994, 2, 19-23 .

Bronk Ramsey 1995 C. Bronk Ramsey. OxCal Program v2. 18 Manual (Oxford 1995).

Cavalli-Sforza 1996 L. Cavalli-Sforza.The spread of agriculture and nomadic pastoralism: insights fromgenetics, linguistics and archaeology. -In: The origins and spread ofagriculture andpastoralism in Eurasia (Ed. D. Harris) (London I 996), 5 1--69.

Champion/Gamble 1984 T. Champion, C. Gamble, S. Shennan, A. Whittle. Prehistoric Europe (London 1984).

Comga 1978 E. Comga. Contribution d I'etude de la culture Criq en Moldavie (le site de GldvAneStii

Yechi). - D acia 22, 197 8, 9-3 6.

DemoulelPerlds 1993 J.-P. Demoule, C. Perlds. The Greekneolithic: anewreview. -Journal ofWorldPrehistory7,1993,35y16.

Dennell 1978 R. Dennell. Early farming in South Bulgaria from the VI to the III millennia B.C.(Oxford 1978).

Erdo[u1997 B.Erdopu.Edirneili 1995 yrh,yizey ara$trmasr.- ArasttrmaSonuqlanToplanttst14,rr,t997,273191.

French 1967 D. French. Prehistoric sites in Northwest Anatolia I. The Iznik area.-Anatolian Studies

17,1967,49-100.Garalanin 197 4 M. Gara5anin. Die Grabung Anzabegovo und das Problem der ethnischen Zugehdrigkeit

derjungsteinzeitlichen Bevolkerung des Balkans. - Thracia 3, 197 4, 2l-28.Gara3anin 1982 M. Garaianin. The Stone Age in the Central Balkan area. - In: Cambridge Ancient

History VoL III: 1 (Cambridge 1982),7 5-135 .

Georgiev 1965 G. I. Georgiev. The Azmak mound in Southem Bul garia. - Antiquity 3 , 1965, 6-8.Georgiev 1967a G. I. Georgiev. Beitriige zur Erforschung des Neolithikums und der Bronzezeit in

Stidbulgarien . - Ar c h a e o I o gi a Aus tri ac a 42, 19 67, 9 O-l 44.

Georgiev 1967b G. I. Georgiev. Die Erforschung der neolithischen und bronzezeitlichen Siedlungshrigel

in Bulgarien. - Zeitschrift fi)r Archdologie 1, 1967, 139-159.

Georgiev 1981 G. I. Georgiev. Die neolithische Siedlung bei Cavdar, Bezirk Sofia. - In: Culturespr6historiques en Bulgarie (Izv. Arch. inst. 36) (Sofia I 98 I ), 63-109.

Gimbutas 1976 M. Gimbutas (Ed.): Neolithic Macedonia, as reflected by excavation at Anza, Southeast

Yugoslavia (Los Angeles 1976).

Gdrsdorftsoj addiev 1996 J. Gdrsdorf, J. BojadZiev. Zur absoluten Chronologie der bulgarischen Urgeschichte. -Eurasia Antiqua 2, 199 6, 1 05-l 73.

203

HillerA{ikolov 1988

HillerAtrikolov 1990

HillerA{ikolov 1991

Horcdt 1959

Houben/Guillaud1994Koro5e6A(oroiet,1973

Koukorili 1995

Lichardus 1985

Lichardus 1993

Mi1ojdi61978Mount-Williarns 1976

Nandris 1970

Nandris 1977

Ozdo$an 1985

1zdogan1996

6 zdo lanlP ar zinger 1 99 7

Panik/Bakdmska 1989

Prendi 1990

Pyke/Yiouni 1996

Roodenberg 1995

Thissen 1997

Todorova 1989

Todorova/Vajsov 1993

vanAndeVRunnels 1995

Wijnen 1981

Zvelebil 1986

ZvelebiUZvelebil 1988

204

S. Hiller, V. Nikolov. Tell Karanovo 1988. Vorbericht iiber die 5. Kampagne derOsterreichisch-Bulgarischen Ausgrabungen am Tell von Karanovo (Salzburg 1988).S. HilleE V. Nikolov. Tell Karanovo 1990. Vorbericht iiber die 7. Kampagne derOsteneichisch-Bulgarischen Ausgrabungen am Tell Karanovo (Salzburg I 990).S. Hiller, V. Nikolov. Tell Karanovo 1991. Vorbericht tiber die 8-. Kampagne derOsterreichisch-Bulgarischen Ausgrabungen am Tell Karanovo (Salzburg 1991).K. Horedt. Sdpdturile de la Moregti (r. Tg. Mureg, Regiunea AutonomA Maghiarg). -Mat eriale S i c er c et dri arhe olo gic e 5, 19 59, 83-89.H. Houben, H. Guillaud. Earth construction. A comprehensive guide (London 1994).P. Koro5e6, J. Koro5e6. Predistoriska naselja Barutnica kaj Amzibegovo vo Makedonija.Izvestaj za iskopuvanjeto vo 1960 (Le site prehistorique 'Barutnica' dans le villageAmzibegovo d Macedonien. Rapport des fouilles de l'ann6e 1960) (Beograd/Prilep 1973).Ch. Koukouli-Chrysanthrlki, I. Asl6nis, Ph. Konstantopoulou. Promachonas--:Topolnica,ena prograrnma ellinoboulgarikis synergasias. - Io archaiologiko ergo sti Makedoniakai Thraki 6 ll992l, 1995, 561-51 5.

J. Lichardus, M. Lichardus-Itten et al. Laprotohistoire de l'Europe. Le neolithique et lechalcolithique entre la Mediterran6e et la mer Baltique (Paris 1985).J. Lichardus. Die Entstehung des mittleren Neolithikums an der unteren TundZa( Siidostbulgarien). - An at o lic a I 9, 1 993, 8 5-98.V. Miloj di6. Rezension von Gimbutas 197 6. - Germania 56, 197 8, 548-559 .

L. Mount-Williams. The Anza IV (Early Vinda) ceramics. - In: M. Gimbutas (Ed.).Neolithic Macedonia, as reflected by excavation at Anza, Southeast Yugoslavia (LosAngeles 197 6), I I 7-1 58.

J. Nandris. The development and relationships of the earlier Greek neolithic. - Man,AIS. 5, I 970, 192--213.

J. Nandris. The perspective of long-term change in South-east Europe. - In: A historicalgeography ofthe Balkans (Ed. F. Carter) (London 1977),25-57.M. 6zdo[an. A surface survey for prehistoric and early historic sites in NorthwestemTurkey. -National Geographic ResearchReports- 1979 Projects, 1985, 517-541.M. 6zdogan. Tarihdncesi ddnemde Trakya. Aragtrrma projesinin 16. Yrhnda genel birdelirlendirm e. - An ad o lu Ar aS t trm al ar t 1 4, 199 6, 329 -3 60.M. Ozdolan, H. Parzinger, N. Karul. Krklareli HOytiqii 1995 yrh kazrs r. - Kazt SonuElart

Tbplanttst 18, 1997 ,1, 53-80.J. Pavrik, A. Bakdmska. Beitrag der Ausgrabung in Gdldbnik zur Erforschung des

Neolithikums in Siidosteuropa. - In: Neolithic of Southeastem Europe and its NearEastern connections. International conference 1987. Szolnok-Szeged (Eds. S. B6kiinyiu. P. Raczky) (Budapest 1989),223-231.F. Prendi. Le neolithique ancien en Albanie.- Germania 68,2,1990,39H26.G. Pyke, P. Yiouni. Nea Nikomedeia I: the excavation of an early neolithic village innofthem Greece 196l-1964. The excavation and the ceramic assemblage (London 1996).J. Roodenberg. Stratigraphy and architecture: the 'big square'. - In: The Ihprnarexcavations L Five seasons of fieldwork in NW Anatolia, 1987-91(Ed. J. Roodenberg)(Leiden 1995),35J6.L. Thissen. Greece and Westem Turkey: transaegean relations during the Early Neolithic.Paper read at the Intemational Symposium at Urla (Turkey), The Aegean in the neolithic,chalcolithic and early bronze age,13-l9th October 1997.

H. Todorova. Das Fnihneolithikum Nordbulgariens im Kontext des ostbalkanischenNeolithikums. -In: Tell Karanovo und das Balkan-Neolithikum. Gesammelte Beitriigezum Intemationalen Kolloquium in Salzburg, 20.--22. Oktober 1988 (Hrsg. S.Hiller)(Salzburg 1989),9-25.H. Todorova, I. Vajsov. Novokamennata epocha v Qulgarija (Sofia 1993).Tj. van Andel, C. Runnels. The earlieit farmers in Europe. -Antiquity 69, 1995, 481-500.M. -H. Wijnen. The early neolithic I settlement at Sesklo: an early farming communityin Thessaly, Greece (Leiden 1981).

M. Zvelebil. Mesolithic societies and the transition to farming: problems of time, scaleand organisation. - In: Hunters in transition. Mesolithic societies of temperate Eurasiaandtheirtransitionto farming (Ed. M. Zvelebil) Cambridge 1986), 167-188.M. Zvelebil, K. Zvelebil. Agricultural transition and Indo-European dispersals. -Antiq uity 62, I 98 8, 574-5 83.

\ M. Sruiver, A. Long aod R.S. Kra eds. 1993 Radiocrhn 35(1): Oxcal v2.18 cub c4 sd:12 prcblchrcnl

COMB Nea Nikomedeia

68.2% confidence

6'r70BC (0.50) 61208C

6090BC (0.50) 60608C

95.4% confidence

6180BC (1.00) 6050BC

Agreement 115.2o/o

62508C 62008C 61508C - 61008C 60508C 60008C

Calendar date

Fig. 1. OxCal combination of the probability distributions of the Nea Nikomedeia radiocarbon dates (exclusive ofQ-655, GX-679,P -1202 andoxA- 1 603+4280).

0.01Loq)

Lo)o-

a 0.005

-oo-ooL(L

0

205

l\r. Stuivel A. Long and R.S. K6 eds. 1993 Radiocarbon 35(1); Oxcal v2.18 cub r:4 sd:12 problchron]

COMB Anza la

68.2% confidence

6009BC (1.00) 5971 BC

95.4% confidence

60458C (1.00) 5960BC

Agreement 164.7o/o

_r-

LooLo)o-

=.oo-ooL(L

0.02

0.015

0.01

0.005

0

6200BC 6150BC 61008C 60508C 6000BC 5950BC

Calendar date

5900BC 58508C

0.01Loo)

Lc)o-

=.o(!-oo(L

0.005

Fig.2.

206

62008C 61008C 60008c 5900Bc

Calendar date

58008C 5700BC

OxCal combination of the probability distributions of the Anza Ia-b radiocarbon dates (exclusiv e ofLJ-2333,Anza Ib).

\

M. Stuivei A. Long and R.S. K.a eds. 1993 Radiocabon 35(1); Oxoal v2.18 cub n4 sd:l2 problchrcnl

COMB Anza lb

68.2% confidence

59908C (0.64) 5940BC

591oBC (0.36) 5880BC

95.4% confidence

60408C (1.00) 58408C

Agreement 163.4%

and R.S. Kra eds. 1993 Radiocarbon 35(1 ): Oxcal v2. 18 cub r:4 sd:'12

The GdlSbnik radiocarbon dates, calibrated individually.

Bln-3580 7120!70BP

GrN-19786 7070t1808P

Fig.3.

207

Loc)

Lc)o-

:bG-oIIL

0.01

0.005

o.02

0.015

o.ot

0.005

0

6100BC

M.stuiveiA.LongandR.s.Kraeds.1993Radio€bon35(1); oxcalv218cub14sd:12problchrcnl

6000Bc 59008C

Calendar date

58008C 5700BC

L$oLo)o-

=-o(E

-oe

TL

57508C 5700BC 56508C

OxCal combination of the probability distributions of the Gdldbnika. exclusive ofBln-3581; b. exclusive ofBln-4093 and GrN-19783.

5600BC 55508C 55008C

Calendar date

Fig.4.

208

COMB Gdldbnik l:ia

68.2% confidence

59608C (0.34) 59358C

5915BC (0.66) 58658C

95.4% confidence

M.StuivetA.LongandR.S.Kraeds.1993Radio€rbon35(1)i Oxcalv2lScubr:4sdi12probldrcnl

COMB Galabnik ll:viii

b 68.2% confidence

56058C (0.67) 55758C

55458C (0.33) 5525BC

95.4% confidence

4r

566oBC (1.00) 552oBC

Agreement 123.8%

5450BC 54008C

and R.S. Kra eds. 1993 Radiocarbon 1); Oxcal v2.'18 cub r:4 sd:12

Bln-2'108 7195t658P

Btn-1581 7000t608P

Bln-1578 6994t55BP

Bln-1241 +1241A 6891 t70BP

Btn-911 68701120BP

Bln-909 6815+1008P

Btn-910+910A 66 10t70BP

Bln-906 6720+1008P

Fig.5. The iavdar radiocarbon dates, calibrated individually. A-D denote mutually exclusive clusters.

209

and R.S. Kra eds. 1993 Radiocarbon 35(1); Oxcal v2.18 cub r:4 sd:12

Fig. 6. The Tell Azmak radiocarbon dates, calibrated individually. A-B denote mutually exclusive clusters.

210

Bln-291 7158t1508P

Bln-299 6812t1008P

Btn-300 642611508P

PHASE ll (l:6)

Bln-140A 6476t1008P

M. Stuiver, A. Long and R.S. Kra eds. 1993 Radiocarbon Oxcal v2.18 cub r:4 sd:12

179 7130t708P

177 7110!508P

-3586+371 6+371 6H 6846t348P

Karanovo lll (burnt,level)

-3587 +37 17 +37 17H 6416!348P

The Tell Karanovo radiocarbon dates, calibrated individually. A-B denote mutually exclusive clusters.Fig.7.

211

o

'b0 76-yj,

F

>N

F!()

oJo

6Y!!e

EF

!-vo!aN ll-.1

a)LOCd Cg

>FQ>

-

-:.9>,aEAa-OEl <ra

z]cod-an

Aoa

=-.i li

9a)6)c! o

)

o5l

aONF.!cr

-

ocs'omoH

oev'=

4637t)

(-)m

Fig.8. Chronological chart suggesting the step-by-step decrease of age for the beginning of the neolithic in the

southern Balkans, from south to north and from west to east.

212