Intensive Survey, 2008

Intensive Survey of Bozyer (POI05.43)

In the rolling hills of Bin Tepe a number of archaeological sites and monuments have been documented. The best known of these are the large earthen burial mounds, or tumuli, of the Lydian and Achaemenid periods. As the previous work of CLAS has shown, there are a number of occupation sites scattered among the tumuli. The majority of these sites date from between Early Bronze Age (EBA) and early modern times. Yet, there is at least one site that may have a much longer history: Bozyer (POI05.43). The first two and a half weeks of the 2008 season were spent conducting intensive survey at Bozyer. This work built to some extent on the intensive survey conducted at Kaymakçı (POI06.01) in 2007, specifically with an emphasis on assessing the density and chronological range of surface materials.

Bozyer, formerly known to us as the “Fig Orchard,” is a low mound that rises 4–15 meters from its surrounding, rolling landscape and covers around 14 ha of modern agricultural territory belonging to the town of Karayahşi (Salihli). The site appears to offer little in terms of its location: it is one of many low, rolling hills in the area; the surrounding hills preclude a sweeping view over the Hermos River plain (although there is a clear view to the south towards Sardis); and there are no caves nor other natural shelters located nearby. Furthermore, at this time, there is no reason to suspect that an ancient river or marsh would have been located immediately near the site. Cultural materials in the form of lithic and ceramic remains, however, cover the surface of the site, varying in density from place to place owing in part, perhaps, to erosion of the top and of a wide gully on its western side. Agricultural activity may be credited with further distribution of these remains, as a total of six fields cover the site (one fig orchard, four olive orchards, and one wheat field), and they are deeply plowed on a seasonal basis, churning-up new material and most likely causing damage to subsurface archaeological remains.

Based on preliminary ceramic and lithic analyses of finds collected in 2005, 2006, and 2007, we recognized that Bozyer provided the earliest known evidence for human activities in central Lydia. Lithic analyses from the preliminary work suggested possible Middle Paleolithic through EBA activities. Western Anatolia is on a route linking early populations of the Middle East to Europe, with known migratory waves of both humans and technologies during the Paleolithic through Neolithic periods. Typical Paleolithic sites, however, are usually associated with caves or other natural shelters nestled in protective locations. Bozyer shows variance from the norm in this respect: it is an open-air site with no clearly definable cave or shelter in the immediate area. Its Paleolithic remains represented an anomaly, therefore, and warranted further investigation. In addition, the Paleolithic remains were mixed with lithic finds datable to the EBA, with a variety of stone materials used to produce tools characteristic of both periods. What periods between the Paleolithic and EBA might be represented by lithic evidence, however, remained to be seen.

Furthermore, ceramic analyses from the preliminary work suggested a period of occupation earlier than most other known sites in the region, specifically during the EBA and possibly Late Chalcolithic (LC) periods. Results from the Instrumental Neutron Activation Analysis (INAA) of pottery samples taken in 2006 suggested a compositional grouping of the pottery from Bozyer separate from the pottery from other, roughly contemporary sites. This data indicated to us that Bozyer was either a local, independent node of ceramic production contemporary with other EBA sites, or that its ceramics dated from a period slightly earlier than those from other sites in Bin Tepe and around the Gygaean Lake. To answer such questions deriving from our preliminary lithic and ceramic analyses, an intensive site survey of Bozyer was conducted over a period of eight days, with an additional eight days spent processing the finds.


CLAS survey methodology was intended to provide a number of specific results: 1) to define the extent of the site; 2) to provide representative samples from collection units covering the entire site; and 3) to define use areas of the site and their distribution over time and with respect to materials and topography. In order to achieve these goals, the site was defined, a representative number of collection units were defined, the material culture from each unit was assessed, and a representative sample from each unit was inventoried and stored. First, using a rough estimation of the density of surface material as an indicator of past human activities, and with the help of GIS software (ArcGIS) and a high-accuracy Real-Time Kinematic (RTK) GPS system, a 20 meter grid was virtually superimposed over the site allowing for the precise spacing of 248 collection units. Rather than surveying the entire 400 square meters of each 20 meter grid, as had been the approach of the intensive survey at Kaymakçı in 2007, a 5% sample of each grid was defined by a 20 square meter circle drawn into the ground quickly and “on the fly” by means of rotating a 2.52 meter long rope (a “dog leash”) around a central axis located at each 20 meter grid point. Ideally, this would have resulted in a 5% sample of the roughly 14 ha site. Owing to local field conditions, however, only c. 10.3 ha (or 102,600 square meters) were surveyed, resulting in a c. 73% coverage of the entire site and an actual surveyed sample of c. 3.7%.

For each collection unit a standardized set of information was documented on forms in the field: grid name; UTM coordinates; name(s) of researcher(s); general visibility; current land use and ground cover; total number of ceramics; total number of lithics; and find totals for other, albeit usually less prevalent, material remains (e.g., loom weights, sling bullets). All finds within each dog leash were counted, while only diagnostics were kept. If, upon further investigation, certain kept finds were determined to be undiagnostic, they were returned to the site at a point recorded with the RTK GPS system. For lithics, collection focused on known tool types, items with retouch, and items made from unique or rare materials. For ceramics, collection focused on “feature” sherds (e.g., rims, bases, and handles), decorated body sherds, and distinctive fabrics (e.g., clay, temper, etc.), as well as other indicators of production (e.g., wasters, loom weights, etc.). All kept items were deposited at the Manisa Museum at the end of the season.

In order to record the current topography of the site and to assess its potential relation to the distribution of archaeological remains, Bozyer was surveyed along 2–5 meter intervals with the RTK GPS system. The result is an accurate digital elevation model (DEM). The virtual grid can be imposed over the Bozyer DEM, allowing for visual displays of all survey points and collection units. Additionally, one can then display the locations of finds by material and/or by time period with comparison to topography for understanding better the possible impact of erosion and other natural or anthropogenic forces.


Lithic analyses

The results from the 2008 work at Bozyer are promising. As suspected, a number of lithic finds are stylistically consistent with known Middle Paleolithic types; a few examples from the Mesolithic period were found; and a number of finds from the EBA further confirm the increasing population density in the region during this time. Analysis of the lithic materials from Bozyer and other sites within the CLAS study area has been conducted by Kevin Cooney (2007, 2008), a graduate student at Boston University, working under the supervision of Professor Curtis Runnels, a lithic specialist at Boston University.

Cooney’s analysis shows that Paleolithic material is abundant at the site (2008). Most of the material is flake-based, produced from large river cobbles of quartzite and white chert. A number of cores are present as well and were produced from river cobbles that show signs of large blank or expedite flake removal. Small prismatic blade cores produced from white quartz were either unipolar or bipolar and usually reduced. Discoidal and globular flake cores were also present; these were produced from quartzite, quartz and chert. Prepared Levallois flake cores and point cores produced from chert are present as well, as are amorphous Abbevillian-style cores produced from quartzite. Additionally, small, spent or almost spent cores with geometric flake removals of the Epi-Paleolithic and Mesolithic periods were found, as were blades of the EBA produced from polar and bipolar chert cores. A few examples of EBA flake technology exist also, but no tabular scrapers were found. Other stone items of the EBA include a number of sling bullets, as well as grinding stones, and a polished axe (Çilingiroğlu 2008), attesting multiple productive activities taking place at the site.

Based on this preliminary analysis, lithic material is distributed across the site, yet with highest concentrations in areas just to the southwest, northwest, and southeast of the highest point of the site, where surface soil conditions were optimal for discovery. This distribution may result as well from erosional conditions, with seasonal plowing resulting in the deflation of the top of the mound and the concentration of finds along its fringes. Notable patterns include the highest concentration of Paleolithic material in the southwest, where Cooney suggests the material may have washed out from deeply stratified deposits along the natural gully in this area. Also, it is a seeming oddity that EBA lithics were discovered in such small quantities relative to the abundance of EBA ceramics.

Ceramic analyses

Ceramic results from the 2008 work at Bozyer confirm and refine earlier findings in the area, pointing to an early EBA, potentially even Chalcolithic occupation (see Roosevelt and Luke 2007 and 2008); furthermore this work helps contextualize previous chemical analyses of Bozyer ceramics, showing a shift in ceramic production during the early EBA (see Boulanger et al. 2008). Primary analysis of the ceramic materials collected from Bozyer in 2008 was conducted by Çiler Çilingiroğlu, a graduate student at Tübingen University; the authors have added additional observations regarding ceramics from other sites in central Lydia, where appropriate.

In total 287 ceramic sherds collected from Bozyer in 2008 were analyzed: 125 rim sherds, 126 handles, 18 bases, 8 plain body sherds, 3 decorated body sherds, and 7 sherds from tripod feet, lids or water pipes (Çilingiroğlu 2008). The majority of the ceramics from Bozyer date to the Bronze Age, and to its earlier phases, based on a working understanding of our local typology and chronology, as well as on comparisons with material from Beycesultan, Ulucak, Limantepe, Panaztepe, Troy, and Elmalı-Karataş. Characteristics that point to an EBA date include their hand-made production; inverted rims; tripod feet; dark, olive-brown-grey burnishing; and a number of other stylistic conventions. Furthermore, a few rims, handles, and tripod feet strongly suggest production in the EB I or, possibly, Late Chalcolithic (LC) phase; this appears to be confirmed with comparison to material from well-known and nearby Chalcolithic sites: Tigani, Büyük Güllücek, Agio Gala Upper Cave, Gülpınar, Emporio, and Orman Fidanlığı. Of particular significance for establishing a Chalcolithic date are a sherd with pattern burnishing; a few examples of “two horned” handles; one horizontal, triangular shaped handle; and two anti-splash rims (Çilingiroğlu 2008). In addition to these early finds, there are also a small number of post-EBA ceramic finds: 11 sherds of burnished gray wares fall within the general “Gray Ware” category of western Anatolia (Bayne 2001; Pavúk 2007); 7 sherds are decorated in Lydian traditions (e.g., “streaky glaze,” “black on red”); and 13 examples of Roman red wares were found.

The distribution of pottery at Bozyer is densest in the southwestern, northwestern, and southeastern areas of the site, corresponding exactly to the highest densities of lithics. Again, this pattern appears to correspond to surface and erosional conditions. The distribution of the early ceramics alone follows the same general pattern, with the northeastern area slightly underrepresented.

General evaluation of Bozyer

Following the extensive survey of Bin Tepe conducted in 2005 and the preliminary reassessments of Bozyer in 2006 and 2007, the site was believed to be the earliest in the study area. The 2008 work aimed to verify this dating through systematic intensive survey using dog leashes located with high-precision GPS equipment on a 20 m grid. The locations of all dog leashes were surveyed simultaneously with the actual survey, allowing for relatively quick and easy survey execution in the field and relatively quick analysis of the distributional data thereafter.

The results confirm the preliminary analysis of the 2005 and 2007 grab samples: Bozyer is the earliest known site in central Lydia. Based on the lithic analysis (Cooney 2008), the area was most likely used during the Middle Paleolithic period. Limited evidence suggests at least some use during the Epi-Paleolithic and Mesolithic as well, but there is no evidence for Neolithic occupation at this time. Yet, during the LC or, perhaps, the initial phases of the EBA, the site was occupied again.

The significance of the initial occupation of Bozyer, then, is critical to our understanding not only of settlement in central Lydia, but also of the movement of early human populations across and throughout the Middle East and Europe and their lifeways. The Paleolithic finds suggest we are dealing with one of the earliest known sites in all of western Anatolia. Furthermore, the location and overall topography – a low mound nestled in the open, rolling hills of the surrounding landscape –contrast with typical Paleolithic sites, notably rock shelters and caves, and thereby offer further proof of the diversity of Paleolithic sites.

The LC and/or EB I phases of the site point to another significant period for the region: they mark the earliest occupation known in central Lydia. Based on the size of the site – c. 14 ha – Bozyer may have functioned as a large node in the east-west communication networks linking central Anatolia and peoples further to the east with populations just to the west. Furthermore, it may have played a strategic role in the migration of peoples, technologies, and other forms of material culture moving in the same directions.

After the early prehistoric activities and the possible LC occupation, the EBA phases show the continuing popularity of the location, and the range of EBA finds – ceramics, lithics, grinding stones, loom weights, and sling bullets – indicate a thriving site with multiple productive and, perhaps, defensive, activities. The paucity of EBA lithics, specifically blades, with respect to the abundance of other EBA evidence is slightly troubling. One possible explanation offered by Cooney is that the rich, surrounding agricultural area may have lured inhabitants to the fields, where blade production would have been proximal to their place of use. After the initial phases of the EBA, central Lydia became home to multiple EBA settlements, most located somewhat closer to the Gygaean Lake, the Hermos River, or seasonally flooded grass and marshlands. After the EBA altogether central Lydia became a powerhouse for Middle to Late Bronze Age fortified settlements, later the Iron Age cemetery of Lydian kings, and later still, the rural hinterland of a provincial capital through the Roman period. While Bozyer is now a key to our understanding of the earliest activities in the region, in each of these subsequent eras the site appears to have been mostly abandoned in what may have become and what remains today primarily agricultural territory.

Microtopography and Architectural Survey

As briefly mentioned above, CLAS uses a RTK GPS system for mapping topographic features as well as for surveying grids, ancient architectural features, and other remains with a horizontal and vertical precision of 1.5 cm and 3 cm, respectively. In total nine days were spent in the field conducting microtopography and architectural survey in 2008.

In addition to the work conducted at Bozyer, intensive mapping was conducted at two fortified sites of the second millennium BCE: POI07.01 and POI06.24 (Asartepe). Limited mapping at POI06.01 (Kaymakçı) was also conducted, focusing specifically on architectural features. Our purpose in surveying POI07.01 and Asartepe was to compliment the results of the intensive site surveys of POI07.04 (Kızbacı Tepesi) and Kaymakçı carried out in 2007 that illustrated the extent and complexity of those sites, the latter most likely the fortified capital of the region in the Middle and Late Bronze Age. At this time we know that at least these four, roughly contemporary and fortified sites were distributed around the fringes of the lake basin, spaced at intervals of 3–8 km. Tınaztepe, a promontory site located between Asartepe and Kızbacı Tepesi, may be an additional member of this group, and future CLAS seasons will aim to confirm its date of occupation. The 8.6 ha size and complex organization of Kaymakçı, with a series of internal walls and gates, potential storerooms, as well as at least one large megaron-like building, point to its identity as the regional capital. As the microtopography survey at Kızbacı Tepesi in 2007 demonstrated, fortifications, gates and megara are present outside of Kaymakçı, as well, and thus the internal organization of all contemporary sites in the immediate area is of importance for our understanding of site and regional development and function, particularly within the overall network of these fortified strongholds. The goal for 2008, then, was to understand better the architectural layouts of POI07.01 and Asartepe, so that these results could be compared with the layouts of Kaymakçı and Kızbacı Tepesi, resulting in as complete a record of the organization of the fortified second millennium BCE sites in the region as can possibly be determined from noninvasive means.


POI07.01 sits atop a hill that rises some 65 meters above the flood plain west-northwest of the Gygaean Lake, and just north of the Gedevre Çiftlik in the town of Hacıveliler (Gölmarmara). The site was first explored in 2007 following its discovery in a 2006 QuickBird satellite image in which the prominent circumferential terracing associated with its fortification was readily apparent. On-the-ground survey in 2007 confirmed its second millennium BCE date and its general configuration similar to that of Asartepe and Kızbacı Tepesi, with a circumferential terrace and fortification wall surrounding a central area of slightly higher elevation. Other organizational details of the 1.2 ha site remained unclear, including evidence for how it was accessed.

Using the RTK GPS system, the entire site was surveyed at a 1-meter interval over two days in order to produce an accurate DEM. Using ArcGIS, the DEM was then subjected to an artificial light source to produce a hillshaded DEM that more readily illustrates topographic changes across the site. In addition to complementing our understanding of features of the site already apparent in the QuickBird satellite image, the new hillshaded DEM reveals a number of other features as well.

First, the central area surrounded by the circumferential terrace is divided by walls that define a southeast-oriented, quarter-circle shaped space at the highest point of the site and an arcuate terrace that abuts it on the south and east. This organization and size of this complex is clearly reminiscent of the central sector of Kaymakçı, where a similarly sized, quarter-circle shaped space and accompanying arcuate terrace is oriented towards the south-southwest. These two sites, less than 5 km apart, thus illustrate an uncanny similarity in the spatial planning of presumably restricted, high-status areas. The difference in their orientations, though perhaps resulting in part from natural topography, is still to be explained, yet it seems clear that they mirror each other.

Additionally, the 2008 work revealed the probable main access to POI07.01. This access to the site was most likely on the southwest side, where a steep slope and bedrock outcrops provide a naturally defensible corridor that could be fortified easily with manmade constructions. A ramp-like corridor then leads to the circumferential terrace and central area, with some evidence for additional fortifications and/or gates at these points.


Asartepe was discovered in 2006 during survey of the northern extent of the CLAS study area. The site is located overlooking the northern fringe of the Gygaean Lake basin, roughly 155 meters above the modern town of Kılcanlar (Gölmarmara) and the ancient site of Kılcanlar Höyük. From the initial assessment of the 3.45 ha site and from the 2006 QuickBird satellite image, its date, monumentality, and general organization were clear. The site consisted of a fortified terrace surrounding a central area of slightly higher elevation. Yet, given previous successes at Kaymakçı and Kızbacı Tepesi, it was decided to clarify further the internal organization of the site with microtopography survey. Following the same protocol as at POI07.01, the RTK GPS system and ArcGIS were employed to create a DEM and hillshaded DEM that reveal many features of the site.

Most apparent in the hillshaded DEM is the circumferential terrace associated with the fortification wall. A short break on the north and an elevation break on the east may indicate where gates provided accesses to the site. Within the circumferential terrace, numerous architectural remains are readily apparent in the heaps of stone that are spread across the central area. Clarity of most of these features is lacking at this time. They appear to be most concentrated in the northern and eastern parts of the area, especially north and east of a north-south oriented ramp-like corridor that appears to extend across the central portion of the site.

Additionally, a pile of stones with c. 22 m diameter and c. 3.2 m height was mapped at the southern end of the site, just outside the fortification terrace. The core of this stone pile had been looted to a depth of c. 2.7 meters, presumably owing to its similarity to a tumulus. Although fragments of a large, rectangular, thick-bodied, and footed terracotta vessel were discovered among the disturbed stone packing (a basin with feet? a sarcophagus?), no other pottery suggested that the feature post-dated the rest of the site. Further clarification of the original configuration and purpose of the feature must await further study and, most likely, excavation.


Work at Kaymakçı in 2008 focused on defining further certain architectural features of the site. As described in previous CLAS reports, the site is defined by an 8.6 ha fortified citadel and surrounding slope and lower settlements and cemeteries. The site covers the lower part of the Kaymakçı ridge, as it extends to the southwest from Gür Dağ towards the western terminus of the Gygaean Lake, on the border of territories belonging to the towns of Hacıveliler (Gölmarmara) to the northeast, and Büyükbelen (Saruhanlı) to the southwest.

Following the microtopography and geophysical survey of 2007, the goal of our work at Kaymakçı in 2008 was to further refine our understanding of the interior organization of the site by surveying the faces of walls visible on the surface. Numerous wall faces were identified, providing wall thickness ranges for the various types of walls found on the site: wall faces apparently belonging to the exterior fortification circuit measured c. 1.7–2.5 meters thick; walls of interior fortifications measured c. 1.2–1.4 meters thick; and walls belonging to interior buildings measured 0.65–0.80 meters thick. These surveyed walls will assist not only in the creation of digital site models intended to further illustrate the complex organization of the site, but also in the planning of future excavations.

Additionally, in 2008 thought was given to the definition of the slope and lower settlements surrounding the 8.6 ha fortified circuit. At this time, we have surveyed only features located within the fortified area, yet it is clear that settlement extends well beyond the primary fortifications down both sides of the Kaymakçı ridge, and probably all the way to the modern lake shore, where the remains of second millennium BCE settlement were discovered during survey in 2006. The total site size of Kaymakçı would then far surpass the size of any known contemporary site in western Anatolia, just as the size of the fortified area is the largest in western Anatolia currently known at 8.6 ha. Plans to quantify the extent of settlement outside the fortifications of Kaymakçı are set for the 2009 season.