Initial Upper Palaeolithic in south-central Turkey and its regional context: a preliminary report.

Kuhn, Steven L. ; Stiner, Mary C. ; Gulec, Erksin 等

Introduction

The earliest Upper Palaeolithic industries of southwest Asia have been documented in a relatively small number of localities widely scattered throughout the eastern Mediterranean. These assemblages share a number of features, including a distinctive approach to blade manufacture that combines elements normally considered typical of both Middle and Upper Palaeolithic technologies. Two sites located in the Hatay region of south-central Turkey, Ucagizli cave and Kanal, contain substantial early Upper Palaeolithic deposits. Technologically, the materials from these two sites resemble the 'initial' Upper Palaeolithic of the Levantine area as well as the Bohunician of central Europe. Ucagizli cave has also provided two AMS radiocarbon dates, adding to the very small number of absolute dates available for comparable assemblages in the region.

The initial Upper Palaeolithic in the Near East

The early Upper Palaeolithic in the Near East has long attracted the interest of palaeoanthropologists and Palaeolithic archaeologists. Because it is the principal land route between Africa and Eurasia, the Near East plays a key role in many scenarios for the origins and spread of both modern humans and the suite of behavioural characteristics which constitute the Upper Palaeolithic. The few radiometric dates currently available further suggest that some of the oldest typologically Upper Palaeolithic industries in the world come from the Levantine area. Moreover, some technological characteristics of the earliest Upper Palaeolithic in the Near East are strongly reminiscent of the Levallois method, more commonly associated with the Mousterian, thus raising the possibility that these assemblages document an in situ evolutionary transition from Middle to Upper Palaeolithic.

The most ancient Near Eastern Upper Palaeolithic has gone by several different names. based on early discoveries in the Jordan valley and elsewhere, Neuville (1934) and Garrod (1951; 1957) used the terms UP1 and Emiran (respectively) to refer to the first stage of the Upper Palaeolithic in the Levant. The mixture of features from Middle Palaeolithic Levallois and Upper Palaeolithic blade technologies later led some investigators to refer to these assemblages as 'transitional' (e.g. Azoury 1986). This term has been abandoned, largely because it presumes a phylogenetic relationship between Mousterian and Upper Palaeolithic based on a simple combination of technological traits. The term 'initial' Upper Palaeolithic (IUP) seems the most neutral and appropriate.

Sites yielding initial Upper Palaeolithic assemblages occur throughout the eastern Mediterranean. The southernmost IUP locality, and the best documented to date, is the open-air site of Boker Tachtit in the Negev desert. Extensive refitting of artefacts and meticulous technological studies reveal what the excavator believes to be a smooth transition between a laminar but essentially Middle Palaeolithic approach to blank production (in layer 1) to a fully-developed, Upper Palaeolithic-type system for prismatic blade manufacture (in layer 4) (Marks 1983a; 1993; Volkman 1983). Farther north, in the area of Beirut, Lebanon, lies the deeply stratified site of Ksar Akil, the key Upper Palaeolithic sequence for the Levantine area. Layers 25-21 (stage 1) at Ksar Akil (Azoury 1986; Bergman 1988; Ohnuma 1988; Ohnuma & Bergman 1990) have yielded the largest initial Upper Palaeolithic assemblages known. Similar materials were recovered from the nearby sites of Antelias shelter (layers V-VII) (Copeland 1970) and Abu Halka (IV, e-f) (Azoury 1986; Copeland & Wescomb 1965). A series of open-air sites in Lebanon have yielded assemblages, termed 'Meyroubian', which combine Middle and Upper Palaeolithic features (Copeland & Wescomb 1965). The stratigraphic contexts of the 'Meyroubian' collections are ambiguous, however, and it is not clear what exactly they represent (Copeland & Wescombe 1965; Schyle 1992). Even farther north is Umm el Tlel (El Kowm) in Syria. Here, levels II base, IIIa and IIIb yield assemblages termed 'Paleolithique intermediaire', for their stratigraphic position as well as for the combination of Mousterian and Upper Palaeolithic features. Among other things, the 'Paleolithique intermediaire' levels contain evidence of both Levallois and prismatic blade manufacture (Boeda & Muhesen 1993).

The initial Upper Palaeolithic industries of southwest Asia share at least one important characteristic, namely forms of blade production technology which appear to combine elements of both Levallois and prismatic methods. Although blade blanks predominate among retouched tools, a great many of the blades have faceted platforms. Some assemblages (e.g. Boker Tachtit, Umm el Tlel) contain substantial numbers of elongated Levallois points. It should be noted that the true extent of this apparent technological homogeneity of the IUP in the Near East has yet to be firmly established through comparative study. Initial Upper Palaeolithic assemblages certainly differ technologically from both the late Mousterian which preceded them and the Upper Palaeolithic (Ahmarian and Aurignacian) which came later. Nonetheless, future technological studies may well reveal the presence of more than one specific system of core reduction. For now, however, the initial Upper Palaeolithic industries do stand together within the technological spectrum of the late Pleistocene in the Near East.

This relative homogeneity of technological features, albeit broadly defined, is accompanied a certain amount of typological diversity. All IUP assemblages are dominated by Upper Palaeolithic tool forms (endscrapers, burins, truncations), although many of these tools were made on Levallois blanks. Middle Palaeolithic forms (sidescrapers and denticulates) may be present in some abundance as well. However, almost every site or group of sites seems to have its own unique artefact forms. At Boker Tachtit and other sites in the southern Levant, IUP assemblages contain so-called Emireh points, Levallois points thinned biderectionally at their bases. At one time the Emireh point was considered an index fossil of the first Upper Palaeolithic industries throughout the region (Garrod 1951). While it may still be a valid indicator for the southern Levant (Volkman and Kaufman 1983), the Emireh point is not especially common elsewhere. A few Emireh points have been found at sites in the Beirut area (Copeland 1986: 6-7), but here the main fossile directeur of the IUP is the chanfrein, or chamfered piece. Chanfreins were produced by a kind of fiat, lateral burin blow struck on a retouched margin, a technique which may have been employed to resharpen endscrapers. In contrast to Emireh points, which are not normally very abundant, chamfered pieces are found in large numbers in the lower part (stage 1) at Ksar Akil and in layers V-VII at Antelias shelter. The type fossil of the 'Paleolithique intermediaire' at El Kowm is the 'Umm el Tlel' point, an elongated pointed blade thinned dorsally at the base by a series of small, laminar removals detached before the piece was struck from the core (Boeda & Muhesen 1993: 56). Although they seem at first description to resemble Emireh points, Umm el Tlel points are technologically and morphologically quite distinct.

While it bears little similarity to the earliest Upper Palaeolithic of western Europe (Aurignacian and Castelperronian), the IUP of the Near East is not globally unique. Technologically, the resemblance to the Bohunician of central Europe is striking. The Bohunician is a blade-based industry which exhibits an essentially Upper Palaeolithic retouched tool inventory while retaining Middle Palaeolithic elements, especially aspects of Levallois method, in the technology of blank production (Svoboda & Skrdla 1995; Svoboda et al. 1996: 107). In Moravia, Bohunician assemblages are dominated by endscrapers, with smaller numbers of burins, leaf-points and retouched points. Middle Palaeolithic tool forms such as sidescrapers, denticulates and even Levallois points are also found in some abundance (Svoboda & Svobodova 1995; Svoboda et al. 1996). Similar industries have also been found in the eastern Carpathians. As with the Bohunician, blades and pointed flakes are often detached from bidirectional cores (Demidenko & Usik 1993). Much farther afield, the industry of layers 5 and 6 of Kara Bom cave in the Altai Mountains of Siberia also appears to be structured along similar lines, with tool blanks made by a mixture of Levallois and prismatic blade core technology (Derev'anko & Markin 1998).

Most authors agree that the initial Upper Palaeolithic of the Near East evolved in situ from some type of terminal Levantine Mousterian. According to Marks, Middle Palaeolithic assemblages rich in blades (the so-called 'early' Levantine Mousterian) persist until relatively late in the southern Levant, and the sequence at Boker Tachtit documents a direct evolutionary transition from a laminar form of late Mousterian to the earliest Upper Palaeolithic (Marks 1988; 1990; Gilead 1991: 121). At Ksar Akil there may be a more pronounced disjunction between Middle and Upper Palaeolithic (Marks & Volkman 1986), perhaps associated with an erosional interval (Copeland 1986: 6). As for relations with later industries, most researchers agree that the Ahmarian, the next Upper Palaeolithic industry to appear in the region, evolved directly out of the IUP (e.g. Marks 1990; Garrod 1957: 442; Gilead 1991: 122). Ahmarian and initial Upper Palaeolithic assemblages have been found in direct stratigraphic superposition only at Ksar Akil and possibly Antelias shelter, however, and this more recent transition has not been investigated extensively.

Despite the fact that it encompasses much of the eastern Mediterranean, the distribution of sites yielding initial Upper Palaeolithic assemblages is extremely discontinuous. Stratigraphically intact IUP assemblages have been found in only 8 sites (including those discussed here), and most of the known sites are isolated, locally unique occurrences. Only around Beirut is there what might be called a 'concentration' of IUP sites (Ksar Akil, Antelias shelter, Abu Halka). Moreover, only Ksar Akil and Boker Tachtit preserve substantial sequences of initial Upper Palaeolithic deposits. Areas generally rich in Palaeolithic deposits (e.g. Mt Carmel in Israel) have not yielded substantial, intact initial Upper Palaeolithic layers, highlighting the apparent sparseness of this 'horizon' within the Near East. It has been suggested, however, that the absence of very early Upper Palaeolithic layers in some cave sequences may be due to an erosional episode (Bar Yosef & Vandermeersch 1972).

Unfortunately, few absolute dates are available for the initial Upper Palaeolithic in the eastern Mediterranean. Most investigators assume that the earliest Upper Palaeolithic dates to sometime between 45,000 and 40,000 years ago (Bar-Yosef et al. 1996; Gilead 1991; Schyle 1992: 29-30): not coincidentally, this interval corresponds with the conventional dates for the Middle-Upper Palaeolithic transition throughout Eurasia. Level I at Boker Tachtit has yielded finite radiocarbon dates extending back to 45,000 BP, putting a lower time limit on overlying IUP layers 2-4. Layer 4, at the top of the sequence, yielded a single date of 33,105[+ or -]4100 (Marks 1983b). A late Mousterian layer (26) underlying the IUP at Ksar Akil provided a single radiocarbon date of 44,000 BP, although Mellars and Tixier argue, based on extrapolation from more recent layers, that the basal Upper Palaeolithic at the site is actually about 50,000 years old (Mellars & Tixier 1989: 767). By way of reference, dates from Bohunician layers in Moravia range between 43,000 and 36,000 BP (Svoboda et al. 1996: table 5.2), while layers 5 and 6 at Kara-Bom cave have yielded dates of 43,300 and 43,200 years BP, respectively (Derev'anko & Markin 1998).

The early Upper Palaeolithic in Turkey: Ucagizli cave and Kanal

True Upper Palaeolithic sites are surprisingly scarce in Turkey. Sites containing Lower Palaeolithic and Mousterian as well as microlithic late Upper and Epipalaeolithic assemblages are numerous (e.g. Albrecht 1988; Kokten 1960; Minzoni-Deroche 1993; Otte et al. 1995; Yalcinkaya 1981; 1986; 1995; Yalcinkaya et al. 1995), but well documented non-microlithic Upper Palaeolithic assemblages are few and far between. Recent surveys of published data (Schyle 1992; Harmankaya & Tanindi 1996) identify a number of sites as containing Upper Palaeolithic components, but many of these consist of a limited number of surface finds, and few have been further investigated or verified (Otte 1998; Ozdogan 1998: 30-31).

The Hatay region contains the best-documented early Upper Palaeolithic sites within the boundaries of the nation of Turkey. Centred on the capitol city of Antakya (ancient Antioch), the Hatay occupies the extreme northeast corner of the Mediterranean basin [ILLUSTRATION FOR FIGURE 1 OMITTED]. The northernmost end of the Rift Valley, through which the Orontes river (the Asi in Turkish) flows, extends into the eastern Hatay. Like much of the northern and eastern Mediterranean basin, the Hatay is a region of high relief, with deeply incised river valleys and areas of low-lying coastal plain separated by mountains up to 1800 m high. The region is also rich in karstic terrain, with extensive exposures of Cretaceous and Eocene limestone containing numerous caves and rockshelters.

Two sites located near the Mediterranean shore of the Hatay, Ucagizli ('three mouths') cave and Kanal, have been shown to contain substantial early Upper Palaeolithic layers. Ucagizli cave is situated on the Mediterranean coast about 10 km south of the point where the Asi (Orontes) river empties into the sea, not far north of the Syrian border [ILLUSTRATION FOR FIGURE 1 OMITTED]. The site is perched on a steep slope at about 18 m above current sea level. As it appears today, Ucagizli represents the remains of a much larger collapsed cave. Substantial in situ Palaeolithic deposits are preserved in two main areas, within a tunnel-like chamber (Locus 1) to the southwest, and both within and outside two smaller cavities (Loci 2 and 3) situated to the northeast [ILLUSTRATION FOR FIGURE 2 OMITTED]. The deposits in Loci 2 and 3 extend several metres out from the cliff face at ground surface and appear to be contiguous, but they are separated from Locus I by a steeply-eroded talus slope. In all three loci, Upper Palaeolithic deposits are present immediately below the surface. The absence of significant terminal Pleistocene or Holocene deposits may indicate that the roof of the cave collapsed sometime in the late Pleistocene, rendering it unattractive to later peoples.

Ucagizli cave was discovered and first investigated by A. Minzoni-Deroche (1992a; 1992b; 1993), who excavated within the fully enclosed southern chamber (Locus 1). In 1997, the authors further tested Locus 1 and began additional excavations in Loci 2 and 3. The sondage in Locus 2 exposed a stratigraphic sequence approximately 1.5 m deep, consisting of a series of terra rosa layers alternating with lighter-coloured deposits containing ash and calcite concretions [ILLUSTRATION FOR FIGURE 3 OMITTED]. All levels yielded artefacts and well-preserved faunal remains attributable to the early Upper Palaeolithic. The bottom of the sequence has not yet been reached: excavation was halted temporarily by a heavily cemented layer which could not be penetrated in a small trench.

Two main Upper Palaeolithic components are represented in the sequence from Ucagizli cave. The more recent of these, which is present in all three localities, has been described in previous publications by Minzoni-Deroche (1992a; 1992b; 1993). In Locus 2, the corresponding materials come from layers A-D [ILLUSTRATION FOR FIGURE 3 OMITTED]. Minzoni-Deroche classified the assemblage as 'Aurignacian' and compared it to the industry from level IV at Antelias shelter. Although the comparison is valid, the terminology is not. Along with many fiat endscrapers manufactured on blades, the more recent assemblage from Ucagizli cave (which is still under study) contains large numbers of retouched and pointed blades (el Wad points). It most closely resembles what are now called 'Ahmarian' industries such as those from Antelias, Ksar Akil (stage 2 or 4) and Yabrud shelters 3 (layer 2) and 5 (layer 4) (Schyle 1992; Ziffer 1981). The earlier, initial Upper Palaeolithic component is present in Loci 2 and 3 at Ucagizli cave but apparently not in Locus 1, as it was not reported by Minzoni-Deroche and does not appear to be represented in collections from the earlier excavations. In Locus 2, the initial Upper Palaeolithic materials come from layers G through I. The nature of the assemblage from layers E and F is more ambiguous; the artefacts bear some resemblance to those from both the overlying and underlying layers, but because available sample sizes are small, it is impossible to be sure at present whether this combination of traits reflects physical mixing or an industry that is truly intermediate.

The site known as Kanal is one of a series of caves and rockshelters in the soft limestone cliffs above the villages of Magracik and Cevlik, about 15 km north of the mouth of the Asi river. This area is the site of Seluccia, an important port city of the classical period. Many of the natural caves have been transformed into storage and habitation rooms or have been quarried for limestone, but a number of them preserve intact Pleistocene sediments. M. Senyurek and E. Bostanci conducted excavations in several of the Magracik and Cevlik caves during the late 1950s and 1960s. Three of these, Merdivenli cave (sometimes called 'First cave'), Tikali ('Plugged') cave and Kanal yielded Palaeolithic materials. The Kanal site contained both Upper and Middle Palaeolithic strata. Merdivenli and Tikali caves yielded mainly Middle Palaeolithic - described as Levalloiso-Mousterian - although a thin Upper Palaeolithic layer was reported at Merdivenli. Results from Senyurek and Bostanci's excavations have been summarized in a series of publications (Bostanci 1968; Senyurek 1959; Senyurek & Bostanci 1958a; 1958b), but the industries have never been described in detail.

Deposits at the site of Kanal are exposed in the vertical walls of a deep channel excavated during classical times. The site has been described as a cave, although if this was originally [TABULAR DATA FOR TABLE 1 OMITTED] the case most of the roof was long ago removed by ancient excavations. Contrary to some published descriptions (e.g. Minzoni-Deroche 1992b; 1993), the deposits at Kanal do not appear to be in secondary position. Despite disturbance during the classical period, the deposits appear to have retained at least gross stratigraphic integrity, as samples from successive levels of the 1967 excavations reveal very little mixing of the Mousterian and Upper Palaeolithic components.

Senyurek & Bostanci (1958a; 1958b) described the Upper Palaeolithic of Kanal as early 'Aurignacian', but this is an outdated terminology. Although these investigators were aware of the so-called 'Emiran' of the Levant, their comparisons were strictly typological in nature. They noted (correctly) the absence of Emireh points from sites in the Magracik/Cevlik area, concluding that this particular stage was missing from the Hatay area. However, as we have seen from more recent research, the Emireh point is not a very widespread horizon marker. Technologically, the materials from Kanal do bear strong resemblance to the initial Upper Palaeolithic of areas farther south.

Technological and typological aspects of the Ucagizli and Kanal assemblages

The collections from Ucagizli and Kanal were obtained using rather different recovery strategies, and the presentation of data from the two sites reflects these asymmetrical collection biases. During the recent excavations at Ucagizli all sediments were dry sieved through fine mesh (c. 2 mm.). The sample from Ucagizli is thus relatively complete, but it is also small. Senyurek and Bostanci either did not sieve sediments from their excavations or else they discarded most of the debitage. However, retouched tools, unmodified blades and Levallois pieces were systematically collected. The materials from Kanal help to provide a more complete typological picture of the initial Upper Palaeolithic in the region.

The majority of artefacts from the coastal Palaeolithic sites in the Hatay were manufactured from high-quality flint obtained as heavily rolled pebbles, probably from marine beaches. based on the sizes of artefacts, the pebbles selected appear to have averaged from 15 to 20 cm in diameter. Because raw materials were obtained in relatively small 'packages' artefacts tend to be smaller and more 'delicate' than specimens from other eastern Mediterranean sites of comparable age. A small proportion of artefacts from both Ucagizli and Kanal preserve white, chalky cortex, showing that they were obtained directly from a source of flint nodules rather than from secondary beach deposits.

TABLE 1 summarizes some of the basic technological features of the IUP assemblages from Ucagizli and Kanal. In terms of the retouched tool inventory, these are clearly blade-based assemblages. Blades outnumber flakes as blanks for tools by a ratio of about 2:1, except in the small sample from Ucagizli layers E and F. As is often the case in the Upper Palaeolithic, blades and blade fragments make up a significantly smaller proportion of the debitage, especially the fraction under 25 mm in maximum dimension. Overall, blades tend to be relatively broad and many are wider at the base than at the distal end (e.g. [ILLUSTRATION FOR FIGURES 4 NOS. 4, 6, 7, 16 OMITTED]). Again with the exception of Ucagizli layers E and F, the largest plurality of blades (both retouched and unretouched) have faceted platforms. Plain butts are the second most common variety. Platforms are relatively broad and deep and bulbs of percussion are well-defined as a rule, indicative of the use of a hard hammer.

Cores are quite scarce in the Ucagizli sample. Consistent with the morphology of the blade blanks, however, both of the specimens from the lower layers (G-I) are single platform blade cores with comparatively fiat faces of detachment, converging flake scars and faceted striking platforms ([ILLUSTRATION FOR FIGURE 4 NOS. 5, 9 OMITTED]. A larger sample of cores is available for the Kanal site. The morphologies of the Kanal specimens are quite variable, but all are blade/bladelet cores. Several of the cores from Kanal are flat, like the two specimens from Ucagizli, but pyramidal, cylindrical and even burin-like specimens are also present. Roughly 40% of the Kanal cores preserve faceted striking surfaces. All cores from both sites are relatively small and appear to have been extensively exploited prior to being abandoned.

The assemblages from both Kanal and Ucagizli are typologically Upper Palaeolithic (TABLE 2). The most abundant retouched tools are endscrapers [ILLUSTRATION FOR FIGURE 4, NOS. 1, 8, 10, 13 OMITTED], most of which are manufactured on relatively fiat blades or flakes. At Kanal, short, semi-circular endscrapers made on flakes or blade segments ([ILLUSTRATION FOR FIGURE 4 NO. 10 OMITTED] are also common. Carinated and nosed scrapers - markers of the classic Aurignacian - are present at both sites but are not at all numerous. In the sample from Ucagizli cave, burins are the second most common artefact class, while at Kanal, retouched and retouched pointed blades (el Wad points) are second in abundance to endscrapers. El Wad points are also quite common in the more recent Upper Palaeolithic assemblages at Ucagizli cave, and their presence in the Kanal assemblage could indicate physical mixing of more than one type of Upper Palaeolithic industry. Alternatively, the Kanal assemblage might indeed be typologically and chronologically intermediate between the two components at Ucagizli. Relatively rare artefact forms of special interest in the Kanal assemblages include three pointed blades with curved backs, reminiscent of Chattelperron points (e.g. [ILLUSTRATION FOR FIGURE 4 NO. 12 OMITTED]), and one possible chanfrein ([ILLUSTRATION FOR FIGURE 4 NO. 14 OMITTED]). Typical Middle Palaeolithic artefact classes, including Levallois points, sidescrapers, retouched points, notches and denticulates make up a small proportion of the assemblages from both sites. It is worth noting that the Levallois points from Kanal are not like the broad, thin points from the Mousterian deposits in the area, but instead are narrow and relatively thick (e.g. [ILLUSTRATION FOR FIGURE 4 NO. 15 OMITTED]): they probably represent unsuccessful attempts to strike blades from cores such as those illustrated in FIGURE 4.

For a variety of reasons related to both conditions of preservation and excavation techniques, almost nothing in known about the archaeofaunas associated with IUP industries in the eastern Mediterranean. Only a small sample of faunal remains was retained from the early excavations at Kanal, but a modest sample is available from the recent excavations at Ucagizli cave, where organic preservation is extremely good. The large mammal component of the fauna from layers E-I is dominated by Capra aegagrus(?), Dama sp. and Capreolus capreolus, in descending order of abundance. This mix of terrestrial species suggests broken forest conditions and exposed rocky slopes. Analyses of microfaunal remains will provide a better picture of ecological conditions. There seems to have been little or no use of marine shell for food in the earlier, IUP component at Ucagizli, a sharp contrast with the more recent Upper Palaeolithic layers in which mollusc remains are abundant. The scarcity of shellfish debris may indicate that the lower layers formed at a time when global climates were much colder and drier and the sea coast more distant than at present. While mollusc species used for food are absent, a few perforated shell beads (Nassarius gibbosula and Columbella rustica) were recovered from initial Upper Palaeolithic layers G-I.

Radiometric dates

Two AMS radiocarbon determinations were obtained from layer H of the test trench in Locus 2 of Ucagizli cave (FIGURE 3, the second layer from the bottom). The two uncalibrated determinations, 39,400[+ or -]1200 BP (AA-27994) and 38,900[+ or -]1100 BP (AA-27995), are in close agreement. No widely accepted calibration of the radiocarbon chronology exists for the age range in question. However, recent attempts at calibration-based variations in the strength of the earth's magnetic field (van Andel 1998) and carbon dating of plant materials from varved lake sediments (Kitagawa & van der Plicht 1998) suggest that radiocarbon determinations in the range of the two samples from Ucagizli underestimate the true ages by approximately 2000 years. Thus, the actual age of layer H is more likely to be somewhere around 41,000 years.

Discussion

The assemblages from layers G-I at Ucagizli cave and from Kanal fit well within the general scope of the initial Upper Palaeolithic in the eastern Mediterranean. The technology of blank production, which involved the manufacture of blades by hard hammer percussion from cores with faceted striking platforms, seems to be an especially good indicator of this phase. Typologically, the assemblages from the two sites are clearly Upper Palaeolithic. By and large, the distinctive 'index fossils' of other initial UP or 'transitional' assemblages - chanfreins, Emireh and Umm et Tlel points - are missing from Ucagizli cave and Kanal. There are several possible explanations for this fact. At Ksar Akil, the assemblage from layer 21, at the top of Stage 1, is much poorer in chanfreins than the underlying levels. Thus, the two Hatay sites could contain relatively recent versions of initial UP industries, an explanation which is certainly consistent with the AMS dates from layer H at Ucagizli. Another possibility is simply that no single index fossil serves to identify this phase or period for the entire eastern Mediterranean. As discussed above, none of the 'typical' IUP artefact forms is found in abundance at all sites: instead, each type seems to be confined to a relatively small region. Perhaps this is another illustration of the increasing 'regionalization' of material culture which appears to have begun in the late Mousterian (Kozlowski 1992; Mellars 1996: 406) and which continued throughout the Upper Palaeolithic.

The radiometric dates for layer H are within expected range of dates for the initial Upper Palaeolithic of the Near East, albeit at the younger end of that range. On the other hand, dates for most other initial UP sites are less than secure. Age estimates for Stage 1 at Ksar Akil (layers 25 through 21) were obtained by extrapolation from dated layers both above and below. Level 1 at Boker Tachtit, dated to 45,000 BP or older, is considered terminal Middle Palaeolithic by the excavator (Marks 1988:114), so we know only that the overlying Upper Palaeolithic material is more recent. The sole determination from level 4 at Boker Tachtit may actually suggest a significantly more recent age than layer H at Ucagizli cave, but the large standard deviation renders it somewhat ambiguous.

More problematic is the chronological overlap with dates for the Ahmarian, which succeeded the initial Upper Palaeolithic in the Near East. The early Ahmarian of units 3 and 4 at Kebara cave (Bar Yosef et al. 1996: 303) has provided a number of finite AMS radiocarbon dates in excess of 42,000 BP. Open-air Ahmarian sites in the Sinai (Abu Noshra II, Phillips 1994: 173) and in the Negev (Boker A, Marks 1983b: 37) have yielded dates of c. 38-39,000 BP. There are several possible explanations for this apparent overlap. It could simply be a matter of contamination. Due to the acute effects which even a tiny amount of modern contaminant can have on a sample of 'dead' carbon, some researchers consider all radiocarbon ages close to or greater than 40,000 years to be minimum estimates. The absence of a widely-accepted calibration for radiocarbon dates greater than 22,000 years could also be a contributing factor. At least one recent attempt to calibrate the radiocarbon curve back to the limits of the technique suggests that, while all determinations of between 30,000 and 50,000 BP underestimate true ages, the size of the discrepancy declines sharply for determinations of greater than 39,000 (van Andel 1998). If this hypothesis holds up to further scrutiny, then it may simply be the case that that 14C ages in the range of 39,000 to 45,000 are highly compressed in terms of actual years. Finally, one cannot rule out the possibility of a brief period of coexistence between the initial UP and the early Ahmarian. Although the latter may well have replaced the former, it is unlikely to have done so simultaneously throughout the Near East. When applied to the transition from Lower to Middle Palaeolithic or from Middle to Upper Palaeolithic, improved dating techniques have shown repeatedly that 'ancestral' forms of technology persisted in some regions long after the industries which were to eventually replace them first appeared.

Ucagisli cave and Kanal help to fill in the discontinuous map of the earliest Upper Palaeolithic cultures in the eastern Mediterranean. These two sites represent the northernmost occurrence of the Near Eastern form of IUP industry. As such, they carry us one step closer to closing the gap in archaeological knowledge about the areas between the Levant, eastern Europe and north-central Asia, where similar assemblages have been described. Perhaps the most difficult lacuna to fill will be Anatolia proper, where non-microlithic Upper Palaeolithic assemblages are remarkably few (Hours et al. 1973; Otte 1998; Ozdogan 1998). The importance of an accurate assessment of the chronology and geographic spread of the first Upper Palaeolithic cultures cannot be underestimated. Such data are crucial to understanding whether distinctive

initial Upper Palaeolithic industries are essentially local developments, or whether they represent evolutionary processes occurring more-or-less synchronously at a much larger geographic scale.

Acknowledgements. We must express our gratitude to the director and staff of the Antakya Museum, our partners in the exploratory excavations at Ucagizli cave. Ayla Sevim, Cesur Pehleven, Ismail Ozer and Mehmet Sagir were instrumental in successful completion of the fieldwork. Artefact illustrations in FIGURE 4 are the work of Kristopher Kerry. The office of the Vice President for Research at the University of Arizona provided funding for the fieldwork.

References

ALBRECHT, G. 1988. Preliminary results of the excavations in the Karain B cave near Antalya, Turkey: the Upper Paleolithic assemblages and the Upper Pleistocene climatic development, Paleorient 14/2: 211-22.

AZOURY, I. 1986. Ksar-Akil, Lebanon 1: Levels XXV-XII. Oxford: British Archaeological Reports. International series 289.

BAR-YOSEF, O., M. ARNOLD, A. BELFER-COHEN, P. GOLDBERG, R. HOUSELEY, H. LAVILLE, L. MEIGNEN, N. MERCER, J.C. VOGEL & B. VANDERMEERSCH. 1996. The dating of the Upper Paleolithic layers in Kebara Cave, Mt. Carmel, Journal of Archaeological Science 23: 297-307.

BAR-YOSEF, O. & B. VANDERMEERSCH. 1972. The stratigraphic and cultural problems of the passage from Middle to Upper Paleolithic in Palestinian caves, in F. Bordes (ed.), The origins of Homo Sapiens: 221-6. Paris: UNESCO.

BERGMAN, C. 1988. Ksar Akil and the Upper Paleolithic of the Levant, Paleorient 14/2: 201-10.

BOEDA, E. & S. MUHESEN. 1993. Umm El Tlel (El Kowm, Syrie): Etude preliminaire des industries lithiques du Paleolithique moyen et superieur, Cahiers de l'Euphrate 7: 47-91.

BOSTANCI, E. 1968. Magracik cevresinde yapilan 1966 yaz mevsimi kazilari ve yeni buluntular, Antropoloji 3: 19-45.

COPELAND, L. 1970. The early Upper Paleolithic flint material from levels VII-V, Antelias cave, Lebanon. Berytus 19: 99-143.

1986. Introduction, in Azoury (1986): 1-26.

COPELAND, L. & P. WESCOMBE. 1965. Inventory of stone-age sites in Lebanon. Part 1, Melanges de l'Universite Saint Joseph 41(2): 29-175.

DEMIDENKO, Y. & V. USIK. 1993. The problem of changes in Levallois technique during the technological transition from the Middle to Upper Palaeolithic, Paleorient 19/2: 5-15.

DEREV'ANKO, A.P. & S. MARKIN. 1998. The Paleolithic of the Altai, in A.P. Derev'anko (ed.), The Paleolithic of Siberia: new discoveries and interpretations (D.B. Shimkin & W.R. Powers US eds.): 84-105. Chicago (IL): University of Illinois Press.

DIBBLE, H. & O. BAR-YOSEF (ed.). 1995. The definition and interpretation of Levallois technology. Madison (WI): Prehistory Press.

GARROD, D. 1951. A transitional industry from the base of the Upper Paleolithic in Palestine, Journal of the Royal Anthropological Institute 80: 121-30.

1957. Notes sur le Paleolithique Superieur du moyen Orient, Bulletin de la Societe Prehistorique Francaise 55: 439-45.

GILEAD, I. 1991. The Upper Paleolithic period in the Levant, Journal of World Prehistory 5: 105-53.

HARMANKAYA, S. & O. TANINDI. 1996. Turkiye Arkeolojik Yerlesmeleri 1: Paleolitik/Epipaleolitik. Istanbul: Ege Yayinlari.

HOURS, F., L. COPELAND & O. AURENCHE. 1973. Les industries paleolithiques du Proche Orient, essai de correlation, L'Anthropologie 77: 229-80, 439-96.

KITAGAWA, H. & J. VAN DER PLICHT. 1998. Atmospheric radiocarbon calibration to 45,000 yr BP: Late glacial fluctuations and cosmogenic isotope production, Science 279: 1187-90.

KOKTEN, K. 1960. Anadolu. Maras viayetinde tarihen dip tarihe gidis, Turk Arkeoloji Dergisi 10(1): 42-53.

KOZLOWSKI, J. 1992. The Balkans in the Middle and Upper Paleolithic: the gateway to Europe or a cul-de-sac?, Proceedings of the Prehistoric Society 58: 1-20.

MARKS, A. 1983a. The Middle to Upper Paleolithic transition in the Levant, in F. Wendorf & A. Close (ed.), Advances in World Archaeologyy 2: 51-98. New York (NY): Academic Press.

1983b. The sites of Boker and Boker Tachtit: a brief introduction, in Marks (ed.): 15-37.

(Ed.). 1983c. Prehistory and paleoenvironments in the Central Negev, Israel 3: The Avdat/Aqev Area. Dallas (TX): Southern Methodist University.

1988. The Middle to Upper Paleolithic transition in the Levant: technological change as an adaptation to increasing mobility, in M. Otte (ed.), L'Homme de Neandertal 8: La Mutation: 109-23. Liege: Universite de Liege. ERAUL 35.

1990. The Middle and Upper Paleolithic of the Near East and the Nile Valley: the problem of cultural transformations, in Mellars (ed.): 56-80.

1993. The Early Upper Paleolithic: The view from the Levant, in H. Knecht, A. Pike-Tay & R. White (ed.), Before Lascaux: The complex record of the Early Upper Paleolithic: 5-22. Boca Raton (FL): CRC Press.

MARKS, A. & P. VOLKMAN. 1986. The Mousterian of Ksar Akil: Levels XXVIA through XXVIIIB, Paleorient 12: 5-20.

MELLARS, P. 1990. The origin of modern humans: an archaeological perspective. Ithaca (NY): Cornell University Press.

1996. The Neanderthal legacy. Princeton (NJ): Princeton University Press.

MELLARS, P. & J. TIXIER. 1989. Radiocarbon accelerator dating of Ksar 'Aquil (Lebanon) and the chronology of the Upper Paleolithic sequence in the Middle East, Antiquity 63: 761-8.

MINZONI-DEROCHE, A. 1992a. Ucagizli magarasi, kazilari ve cevre arastirmalri, XIII Kazi Sonuclari Toplantisi 1: 71-81.

1992b. Ucagizli magara, un site aurignacien dans le Hatay (Anatolie). Premiers resultats, Paleorient 18/1: 89-96.

1993. Middle and Upper Paleolithic in the Taurus-Zagros region, in D. Olszewski & H. Dibble (ed.), The Paleolithic of the Zagros-Taurus: 147-58. Philadelphia (PA): University of Pennsylvania. University Museum Monograph 83.

NEUVILLE, R. 1934. Le prehistoire de Palestine, Revue Biblique 43: 237-59.

OHNUMA, K. 1988. Ksar Akil, Lebanon. A technological study of the earlier Upper Paleolithic levels of Ksar Akil 3: Levels XXV-XIV. Oxford: British Archaeological Reports. International Series S426.

OHNUMA, K. & C. BERGMAN. 1990. A technological analysis of the Upper Paleolithic levels (XXV-VI) of Ksar Akil, Lebanon, in Mellars (ed.): 91-138.

OTTE, M. 1998. Turkey as key, in T. Akazawa, K. Akoi & O. Bar-Yosef (ed.), Neandertals and modern humans in western Asia: 483-92. New York (NY): Plenum Press.

OTTE, M., I. YALCINKAYA, H. TASKIRAN, J. KOZLOWSKI, O. BARYOSEF & P. NOIRET. 1995. The Anatolian Middle Paleolithic: New research at Karain cave, Journal of Anthropological Research 51: 287-99.

OZDOGAN, M. 1998. Anatolia from the Last Glacial Maximum to the Holocene Climatic Optimum: cultural formations and the impact of the environmental setting, Paleorient 23/2: 25-38.

PHILLIPS, J. 1994. The Upper Paleolithic chronology of the Levant and the Nile Valley, in O. Bar-Yosef & R. Kra (ed.), Late Quaternary chronology and paleoclimates of the Eastern Mediterranean: 169-76. Tucson (AZ): Department of Geosciences, University of Arizona.

SCHYLE, D. 1992. Near Eastern Upper Paleolithic cultural stratigraphy. Wiesbaden: Dr Ludwig Reichert. Beihefte zum Tubinger Atlas des Vorderen Orients, Reihe B (Geisteswissenschaften) Nr 59.

SENYUREK, M. 1959. A note on the Paleolithic industry of the Plugged Cave. Belleten 23: 163-175.

SENYUREK, M. & E. BOSTANCI. 1958a. Prehistoric researches in Hatay province, Belleten 22: 155-66.

1958b. The Paleolithic cultures of the Hatay province, Belleten 22: 191-210.

SVOBODA, J., V. LOZEK & E. VLCEK. 1996. Hunters between East and West: The Paleolithic of Moravia. New York (NY): Plenum Press.

SVOBODA, J. & H. SVOBODOVA. 1995. Les industries du type Bohunice dans leur cadre stratigraphique et ecologique, L'Anthropologie 89: 505-14.

SVOBODA, J. & P. SKRDLA. 1995. The Bohunician technology, in Dibble & Bar-Yosef (ed.): 279-300.

VAN ANDEL, T. 1998. Middle and Upper Paleolithic environments and 14C dates beyond 10,000 BP, Antiquity 72: 26-33.

VOLKMAN, P. 1983. Boker Tachtit: Core reconstructions, in Marks (ed.): 35-52.

VOLKMAN, P. & D. KAUFMAN. 1983. Reassessment of the Emireh point as a possible type fossil for the technological shift from the middle to the upper Palaeolithic in the Levant, in E. Trinkaus (ed.), The Mousterian legacy: 35-52. Oxford: British Archaeological Reports. International series S164.

YALCINKAYA, I. 1981. Le Paleolithique inferieur de Turquie, in J. Sanlaville & J. Cauvin (ed.), Prehistoire du Levant: 20718. Paris: CNRS.

1986. Karain cave, 1985, Anatolian Studies 36: 1-201.

1995. Thoughts on Levallois technique in Turkey, in Dibble & Bar-Yosef (ed.): 399-413. Madison (WI): Prehistory Press.

YALCINKAYA, I. et. al. 1995. Les occupations tardiglaciares du site d'Okuzini (sud-ouest de la Turquie): resultats preliminaires, L'Anthropologie 99: 562-83.

ZIFFER, D. 1981. Yabrud shelter II - a reconsideration of its cultural composition and its relevance to the Upper Paleolithic cultural sequence in the Levant, Quartar 31/32: 69-91.