by Andrew Ramage and Nancy H. Ramage
Chapter 2. Introduction to Lydian Trench—Sector HoB
Sardis, the capital of the kingdom of Lydia, became one of the great cities of antiquity. Lying about 80 kilometers (50 miles) inland from the Aegean Sea (Fig. 2.1), in the valley of the Hermus River, it boasted a high, steep, and defensible hill that served as a fortified Acropolis (Fig. 2.2). To the west lay a more rounded massif of sandy conglomerate that would become the hill of the Necropolis. Its position below the mountains of ancient Tmolus (modern Bozdağ) to the south and the Hermus River plain to the north allowed the Lydians to command varied and fertile lands that supported the agriculture needed to feed a growing population (see Fig. 2.7). While Sardis originated as one of a number of humble villages in the area, it expanded to ever-greater size and wealth, famed in the Greek literary sources of the sixth and fifth centuries B.C. for its luxury and gold, its textiles and horses.
The Lydians as a people were little known before the investigations of the Harvard-Cornell Expedition that began in 1958. But the site of Sardis has been recognized since Cyriacus of Ancona visited in the mid-fifteenth century and recorded the inscriptions that he found there.1 Other early travelers, such as Robert Wood, Giovanni Battista Borra, C. R. Cockerell, and Thomas Allom, also wrote about their observations and made drawings over the centuries;2 but Howard Crosby Butler, of Princeton University, was the first to dig systematically at the city. Excavating between 1910 and 1914, he concentrated on the Temple of Artemis and the tombs of the Necropolis.3 It was under George M. A. Hanfmann of Harvard, however, that digging in the earlier levels of the city began, work that continues to this day.
The excavation that is the subject of Part I of this volume was undertaken in a plot of land named for a large building, the House of Bronzes, found nearby (Fig. 2.2: no. 4 and Fig. 2.4).4 The Lydian levels in this area were excavated in most seasons between 1960 and 1970, and, in smaller campaigns, in 1983, 1984, and 1985. In this book the term “House of Bronzes” without other qualification will be reserved for the actual Late Roman house. Digging through the Late Roman building served as the impetus for excavation in the Lydian levels when earlier remains were found beneath the House of Bronzes. The Lydian finds discovered in this area were immediately recognized by the excavators as important features, and the acronym “HoB” (as an abbreviated sector name) was used to denote workings in the general area that had been started with the excavation of the actual House of Bronzes, but that now referred to the earlier levels of the “Lydian Trench” at HoB.
Once it had been determined that Lydian occupation levels lay below the House of Bronzes,5 there was great interest in finding out how far back in time that habitation went. Because the open ground to the south of the House of Bronzes was free of Roman buildings, it offered an excellent opportunity to lay bare not only a wide exposure, but also a stratigraphic sequence that would help to clarify the chronology of the sporadic finds of early Lydian pottery that had so far been discovered.6 The Lydian Trench at HoB came to take the form of a rough rectangle with boundary lines from E20 to W42 and from S80 to S130 on the site grid plan (see Fig. 2.4).7
Physical Description of the Site
The sector HoB forms part of the gentle slopes that extend like a wide skirt about 1,200 meters to the north of the Acropolis, below the precipitous cliffs and large spurs of original Tertiary conglomerate that forms the bedrock beneath the city of Sardis (see pp. 26–31). HoB actually lies roughly 900 meters from the Acropolis heights and about 300 meters to the east of the present bed of the Pactolus stream.
At the east side, the boundary of the sector is formed by a low hill, rising about five meters above the general level of the ground. For a long time it was thought to be a natural ridge, but investigations since 1976 have shown that it is formed from the core and fallen superstructure of the monumental mudbrick city wall of the seventh century B.C. (Fig. 2.2: no. 64a).8 The wall runs southwest, across a low saddle and into a steep hill to the south of sector HoB. This hill rises a total of about 20 meters and is the result of Roman terracing operations that were undertaken to accommodate several rows of slope houses, referred to as Middle Terrace East and Middle Terrace West (visible in Fig. 3.2).
Extensive sectional trenching into this hillside failed to reach bedrock, and revealed only mixed Roman fill that appears to have been purposely dumped, since there were unusually large quantities of bones and potsherds in distinct levels. Digging at the top of the ridge beside the Late Roman city wall nearby (Fig. 2.2: no. 9)9 also revealed nothing but Roman fill. We assume that there was a rise in the ground in Lydian times too, but the degree of the incline certainly became exaggerated as a result of the tenacity of the deep Roman foundations and massive Roman dumping.
The surface before excavation was relatively even, and sloped downward only about two meters toward the north, over about 300 square meters, from about level *103 to *101.10 It had been ploughed and was being used as a melon field when first acquired in 1958. The area of the trench proved to be surrounded on three sides by substantial Late Roman buildings but did not itself have any structure built upon it that was later than the third century B.C. We do not know why this large area of more than 60 by 50 meters should have remained open and unpaved during the Roman period, since it was neither a cemetery nor obviously a storage area for merchandise or other materials. Certainly the area of HoB was an accessible and important location, as it lay near the Roman Marble Avenue that ran through town along the east-to-west axis, a main road that was in use until recent times. The chief signs of Roman activity within the specific area are the terracotta water pipes that run across the trench and that were in places immediately above the Lydian remains (Fig. 2.3).11 We do not yet know whether they serviced the entire lower city to the northeast or if most of them were destined to supply the Bath-Gymnasium Complex to the north.
Topography of Sardis
The Lydian levels in sector HoB were situated in a fairly low but important place in the town, just outside the official boundary defined in the late seventh century by the city wall.12 This wall ran along the north slopes of the Acropolis and extended into the Hermus valley, encompassing the large area that was the city center. But since the section of the city wall nearest to HoB contains a gate (Fig. 2.2: no. 63), the area was well placed for travelers and traders moving along the main route between the coast and the interior.
The identification and subsequent exploration of the massive Lydian city wall upset many of the previous assumptions made about the siting of the city, as it became clear that all the sectors that had already been excavated lay outside the boundary of the Lydian city. This was not critical for the interpretation of many of the buildings in sector HoB itself because they had been built and occupied before the creation of the wall; it did, however, make a difference to our understanding of the rationale behind the siting of some other features. For instance, the Hellenistic and earlier Roman graves alongside the east–west road leading toward Magnesia-by-Sipylos (Manisa) and Smyrna (Izmir) are no longer an anomaly when one realizes that they were outside the formal boundary that lasted until the Late Roman wall was built to encompass a new area at the west as far as the river (Fig. 2.2: no. 6).
For the most part, the Lydian levels of HoB illustrate growth and enhanced activity in the handicrafts and daily life of the community from the Late Bronze Age into the sixth century B.C. A decline in the intensity of occupation in the later sixth century indicates that something changed after the sacking of Sardis by the Persians in 547 B.C.13 Any new construction happened elsewhere, and the site was neglected and left open for much of the fifth century. It seems to have been regarded as an unofficial dumping ground with only sporadic occupation and a pronounced lack of substantial walls.
The Course of the Excavations in Sector HoB
The northern boundary of HoB was formed first by the Roman House of Bronzes and, to the west of it, by the south wall of another large building or enclosure, still unexplored. At the east, the boundary was defined by the west wall of the Roman Building R, which ran at an oblique angle to the trench edges, northeast to southwest (see Fig. 2.4). To the south, a late terrace or enclosure wall with Roman residential buildings framed the excavated area. A trace of the seventh-century Lydian city wall was found beneath the Late Roman city wall14 as it climbed a ridge toward the Acropolis. Its disintegrated remains would have formed a fine uniform basis for holding in the Roman terracing, since its massive and well-prepared construction would have easily resisted the weight of the buildings. Only at the west edge have no late buildings been found, but exploration here has been hindered by the presence of the modern village.
It is clear that the Roman boundaries have nothing to do with the Lydian uses of the space. The Lydians took the ground as they found it and built accordingly; the Romans altered the ground to suit their needs. Besides the disturbed remnants of Lydian buildings found within the House of Bronzes, their presence is implied in the area of Building R (see Fig. 2.4), where large deposits consisting of almost pure Lydian pottery were found in what would otherwise be Roman levels. This suggests that while cutting their foundation trenches, the Romans encountered the Lydian occupation levels, and that material was then deposited at the contemporary ground level. What only slowly became apparent is the extreme depth and unevenness of the Lydian deposit, and the way in which the combination of Roman buildings and erosion altered the topography.
In 1958, the first year of work by the Harvard-Cornell Expedition at Sardis, excavations were begun along the former Izmir–Ankara highway, not far from the visible remains of a Roman gymnasium (Fig. 2.2: no. 1). The aim was to investigate a Late Roman building, soon dubbed the House of Bronzes for the bronze vessels that were found there. The excavations were carried on here in the summers of 1958 and 1959.15 Already in these first seasons, sondages (small soundings) beneath the floors of the Roman house uncovered Lydian pottery and walls, and were nicknamed the “Lydian Shop” and “Lydian Room” (see Fig. 2.4). After excavating the so-called shop and an adjacent area of later tombs to the west, the excavator, Donald P. Hansen, resumed operations a little to the south in the early part of the 1960 season. His intention was to capitalize on the exciting discoveries of 1958 and 1959, but then he took on other tasks. At that point Gustavus F. Swift, Jr., was put in charge of the area of the Lydian Trench, and excavation continued under his supervision until 1970.
The only writing on sector HoB that Swift left at the time of his death in 1976 was an extensive outline of eleven pages that he had prepared in 1970, as well as the annual season reports for each year of excavation. These normally do not attempt to synthesize the accumulated material, since they purport to summarize the finds of one particular season.16 Throughout this volume, where a description in Swift’s words seemed the most vivid and accurate, passages from his writings are included verbatim. His preliminary conclusions, stated or implied in the season reports and the published BASOR reports,17 have been in most cases confirmed by Andrew Ramage’s studies. Where there is a difference, the reasons for the revisions, some of which derive from later work in HoB itself, have been stated.
The incentive to expand the Lydian Trench was provided by the discovery of Lydian pottery in beds of sandy soil within the first explorations of 1960. The stratigraphic situation was similar to that of the Lydian levels encountered beneath the floors of the House of Bronzes in 1958 and 1959,18 except that it was now possible to gain a much wider exposure. The areas in the northeast corner, nearest to the House of Bronzes, were opened first, and in subsequent seasons the trenches were extended to the west and south.
In the course of these extensions, several deep soundings were dug to test the stratigraphy and develop a chronology for the Lydian levels. They were called “deep pits” in preliminary reports, but, wishing to avoid possible confusion with ancient pits, the excavators came to refer to them as “soundings” or “test trenches.”19 The first, Deep Sounding A, was dug in 1960; two larger and deeper ones, Deep Soundings B and C, followed in 1962 and 1966, respectively (see Fig. 3.1). B and C were contiguous at the outset, but the trenches diverged as the levels descended and the soundings narrowed. The finds and stratigraphy of all three deep soundings, although not identical, are closely related. Later, in the 1980s, several smaller and less ambitious sondages were undertaken to answer particular questions.
Structures uncovered in the Lydian Trench were given identifying letters running from A to H and J to O (see the composite plan in Fig. 2.4). Roman Building R forms part of another series. To some extent the identifying letters given to the Lydian structures reflect the movement of the excavations from east to west; thus, Building A is in the eastern part of the trench, and Building N is at the extreme west.
In 1960 the east end of the Lydian Trench and some of the Roman Building R were dug, as well as Deep Sounding A. There was considerable disturbance in the area from Roman pipes, which ran north–south toward the House of Bronzes. Building A, the first complete example of the small rectangular buildings typical of the Lydian Trench, was cleared during this season. In 1961 the trench was extended to the north and west; the digging exposed Building B, the eastern end of Building C, the northern part of Building D, and the Stone Circle industrial area just east of Building C. The rest of Building C and an area to the south of it were excavated in 1962, when Deep Sounding B was also undertaken.
In 1963 a large area was exposed adjacent to the 1962 trench on the south and west. Buildings D, E, and F were added to the list of Lydian buildings. Investigations of Lydian remains in 1964 concentrated on a trench that revealed the stratigraphy in the southwest corner of the area. No additional Lydian buildings were discovered, but a comprehensive east–west section at S120 illuminated the occupation levels of what we call the South Side, being at the southern end of the trench (see Fig. 3.3). Additional work was undertaken in the Roman levels at the south, in Middle Terrace East and Middle Terrace West (Fig. 2.2: no. 5), where two long north–south cuts were made into the hill at the south of the area. These revealed the colonnaded street and a series of Late Roman houses stepping back in terraces, much like the now-famous slope houses (Hanghäuser) at Ephesus.
Lydian levels formed the focus of the work in 1965, largely in the central and western part of the trench. We concentrated on opening a level of the late eighth or early seventh century B.C., characterized in the annual reports as “Kimmerian burning” (although that designation is problematic and will be discussed later, pp. 57–58).20 In 1966 we had the double aim of clarifying further the later seventh-century walls found at the west side of the trench in 1965—formerly referred to as the “Lydian Bazaar” precinct, and now considered to be a group of houses. We also exposed deep levels (Deep Sounding C)21 over a much larger area than before on the eastern side of the trench.
There was no digging at HoB in 1967. The excavations of 1968 were largely concerned with extending our knowledge of the seventh-century phase, the Lydian II levels, although there was some work in Roman areas near Building R and the adjacent street. The most notable discovery there was an early Byzantine inscription that implied that the street should be thought of as the Hypaipa Road.22
There was again no digging at HoB in 1969, but work was taken up again in 1970. The targets were seventh-century levels to the west, and the earlier burned levels at the south. Almost all the area within the trench had by now been dug to a depth of two to five meters, although the deep soundings, of course, were substantially deeper, up to 11 meters below the original grade. Additional work in the Roman colonnaded street brought to light an arched façade of brick and reused marble blocks, which may correspond to a tetrapylon referred to in the inscription mentioned above.
After that time there was no excavation here until 1980, when a project to complete work on the House of Bronzes was initiated. In 1983 three stratigraphical soundings were sunk beneath the Lydian II levels at the west end of the trench, and in 1984 another was started, continuing in 1985, just to the south of Deep Sounding C, in an attempt to get more detailed information from the Iron Age levels.
The intermittent excavations of each level over more than a decade produced a huge quantity of pottery, most of it local, that was discarded with minimal comment upon its implications. The remaining sample has been necessarily skewed because more attention was focused upon Greek and painted pieces that would offer chronological, if not commercial, connections. The bias may be most acute for the early levels, but it is built-in everywhere. Although much domestic coarse and plain pottery was discarded, we need to put back, as best we can, the weight of domestic activities that must have been present, even in buildings that may also have served as workshops. Some of the extant pottery might well have served purposes for both the household and light industry.
Estimating the extent to which the Lydians used organic materials is a puzzle, where accepting the general proposition that they used perishable materials still leaves us shy of the actual quantities required for comfortable (or more probably uncomfortable) existence. Certain soft goods like textiles are easy enough to imagine, and are confirmed by ancient texts. Harder items like wood and bone are more difficult; some pieces last, but how many more should be imagined for a full accounting? Evidence for wooden furniture is absent except for iron hardware that must have served as “brackets.” Such items, which turned up frequently, should be connected with furniture because of imprints of wood grain on the corrosion product.23 Very few lumps of charcoal to represent the actual wood have survived, although patches of crumbled charcoal were widespread. Similarly, there was hardly a trace of leather.
Alluvial Deposits and Stratigraphy
The general position of the site in its geological and geomorphological setting has been described previously,24 and the soils of Sardis were made the subject of a special study in 1970.25 Essentially, the Acropolis and the Necropolis form a narrow layer of conglomerate set against the metamorphic mass of Mount Tmolus (Fig. 2.5). The discontinuity, which runs roughly east–west, is about two kilometers south of the Temple of Artemis (Fig. 2.2: no. 17) and can be strikingly observed in a fifteen-minute walk from the temple (Fig. 2.5).
The erosion of the loose conglomerate of the Acropolis accounts for the many deposits of gravel and soil that washed down the hill and into the area of the Lydian Trench of HoB. Some of these inundations probably occurred when the inhabitants stopped maintaining terraces uphill that had been built to avoid such events. Periodic catastrophic inundations caused by slope destabilizations also buried the Temple of Artemis to the west of the Acropolis.
A description of the landscape may be quoted from the publication of the first archaeological expedition to Sardis in the early twentieth century (Fig. 2.6):
At the northern edge of the marble [of the Tmolus range] a non-conformity marks the beginning of a new geological series. This series consists of unfolded beds of alluvium exhibiting three distinct phases. The first, or lowest, phase, is composed of massive homogeneous beds of fine red clay. These thin out very much at the edge, so that no adequate measurements can be obtained of their thickness which is approximately 30 m. This material bears every resemblance to deep-water marine deposits, being evenly distributed, perfectly sorted so that it contains no coarse material, and stained an even dark-red color by oxide of iron.[br] The second phase consists of grey or pale green beds, in all about 200 metres thick, of fine sandy clay interbedded with gravel containing pebbles of quartz and occasionally of schist rarely more than 2 cm. in diameter. The fine beds vary in thickness from 5 to 50 cm. and the coarse beds from 10 cm. to 5 metres. Their texture is frequently quite loose, though more often the material is solidified by lime cement. The beds dip slightly toward the north because of having been deposited against the ridge. They are evidently shoal-water deposits; the succession of coarse and fine layers indicates a seasonal variation in the streams bringing down the material deposited and an absence of currents to sort it out. That these are subaqueous deposits is shown by the continuance of the beds; their light colour is probably due to a growth of reeds which leached out the iron oxides.[br] The third phase, by far the most conspicuous because it has most resisted erosion, is exposed in the bare cliffs of the Acropolis and Necropolis Hills. It is composed of fine grains of kaolin, muscovite, biotite and probably other ferro-magnesian minerals, and a little quartz, with quantities of pebbles; the flat pebbles of schist usually about 10 cm. in diameter but frequently much larger or smaller, irregular pebbles of gneiss often slightly angular, but always water-worn, quartz and pebbles of all sizes up to 20 cm., but usually not over 10 cm., in diameter.26
It is the interaction of the erosion processes on these two rock formations that has produced the Sardis landscape. The most active erosion agent was water: first through the direct action of the rain and the creation of smaller and larger streams; and then by the cutting and filling action of the Pactolus and, to some degree, of the river Hermus.27 Since the poorly cemented conglomerate is particularly susceptible to erosion, the frequently heavy rainstorms of the Lydian winter (which were probably no less heavy in antiquity) contributed to its movement in great quantities at one time. Periodic earthquakes may also have caused large masses of conglomerate to spall from the Acropolis, which would then disintegrate and flow downhill, leaving massive deposits of sand and gravel.
These processes have generated several large alluvial fans in the valley, where small streams like the Pactolus spill from the steep confines of their narrow valleys into the broad, flat plain of the Hermus. Besides sand, gravel, cobbles, and other components of the conglomerate, these fans contain much larger pieces of the matrix and inclusions of the Tmolus bedrock, such as gneiss, schist, and quartz. The field or river stones, mostly smaller than 25 centimeters, provided (and still do) a basic building material for inhabitants of the valley. Cut stone is so far rarely attested in Lydian vernacular architecture.
Another possible source of gravel deposits in Lydian and prehistoric times is the overflow from the main course of the Hermus. Until the middle of the last century it was not unheard of for the whole valley to be inundated, and for villages to be cut off, but the building in 1960 of a large dam at Demirköprü (Fig. 2.7) removed that threat. This kind of flooding would have deposited fine sand or silt, however, rather than gravel or cobbles, because Sardis is at the edge of the valley; floodwaters here would flow more slowly and deposit the lighter materials.
The Pactolus fan was probably wider and further south in ancient times than it is now,28 and some of the levels in HoB may have resulted from its buildup if the river was turning rather more to the east than at present. We know that the river has changed its course since antiquity because it has cut into ancient structures, both Lydian and Late Roman, at several places on each bank between the Temple of Artemis and the Roman bridge (Fig. 2.2: no. 6). The extent of the alluvial fan in Lydian times, however, is rather obscured by modern buildings and cultivation near the bridge and crossroads. The Late Roman city wall also had an important effect in altering the topography and, presumably, later local drainage patterns, and may have contributed to the deposition of sediments. For the area of HoB, the stretch of city wall between sections 30 and 2929 (Fig. 2.2: no. 9) would have acted as a barrier and encouraged the buildup of debris beside it in late antiquity, once the pattern of city life had been interrupted. In addition, the sections to the west and north formed part of an embankment on the east side of the Pactolus and will have contributed to the regularization of its channel, causing the fan to be moved further downstream. The Roman bridge and its abutments would also have contributed to the channeling of the Pactolus. Without much more extensive digging or a considerable program of systematic coring, this theory will remain a hypothesis.
From the first, the excavations in the Lydian Trench of HoB produced a consistent pattern of natural stratigraphy. Again and again the excavators reported finding strata of sandy gravel in different grades and thicknesses, between which were beds of clayey soil. These beds have provided most of the evidence for human occupation in the area: remains of buildings, floors, great quantities of pottery, and small artifacts of various materials. Much of the soil matrix is probably derived from the decomposition of mudbrick, which was the standard material for the upper parts of Lydian walls. At first this was only a suspicion, following traditional wisdom, but finding isolated bricks in the excavation (described at the time as “potter’s clay”)30 and later bricks in situ on the wall socles made it certain.
In general the gravel layers were lacking cultural remains, except for one or two waterworn potsherds and the occasional object. Every so often there might be an isolated patch of clayey material within the gravel, which was otherwise not connected to anything else. Such flooding must have been particularly devastating in periods when the terraces further up the slopes of the Acropolis were not maintained. It is as if there were a range in the intensity and spread of the flooding, such as what insurance agents refer to when they speak of a “100-year flood.” In any case, while requiring attention in the course of excavation, these isolated patches add little to our understanding of the course of Lydian occupation in the area, other than to suggest that there were more frequent attempts to establish structures than most of the record shows.
To avoid oversimplification and exaggeration of the contrast between water-laid gravels and evidence of human occupation in clayey or earthy levels, certain facts must be kept in mind. To begin with, although the gravel layers were sparsely provided with potsherds and other small objects, they were never (even those below the Late Bronze Age material at level *90.00) totally devoid of them. Such sherds had worn surfaces and were rubbed and rounded, and might well have originated from other parts of the settlement upstream. The occasional beds of pure fine sand were, on the other hand, regularly sterile or nearly so. Architectural remains were confined to the clay layers, and features such as hearths and mouths of pits were only occasionally found on gravel surfaces. One could imagine rebuilding on parts of the alluvial fan after a thin skin of soil and vegetation had reestablished itself, and the vagaries of the weather would determine the length of successful occupation. Again, small intermediate bands and patches of the clay material occurred among the gravel beds, but with only a few artifacts for content. Finally, some deposits clearly consisted of an unstratified mixture of gravel, earth, and artifacts and could have been deposited by either natural or human agency. But the major beds of the yellowish clayey material have defined the stratigraphic and cultural sequence throughout the excavations.
With the aid of the section drawing and general view of the South Side (see the Table of Chronology with Elevations, and Figs. 3.2 and 3.3), we can get an idea of the succession and relative thickness of the different strata that were contained within the Lydian Trench. Drawings and descriptions of the deep soundings help us arrive at a composite picture of the levels to ca. *90.00. This is, of course, a generalized construct and varies, both absolutely and in degree, at some points in the trench, but overall the picture is remarkably uniform.
It goes without saying that the ancient people of Sardis did not erect monumental buildings, or even residential quarters, in a streambed, where they would have been constantly threatened with destruction by flood. Given this area’s tendency to become overwhelmed by inundations of gravel, one is driven to the conclusion that the site of the Lydian Trench in HoB was valued because of its position near lines of communication and its proximity to other parts of the settlement (see Fig. 2.2). In addition, access to the northern slopes of the Acropolis is much gentler and more open from this area than from higher up the Pactolus valley. But it is clear that the uses to which this place could be put were limited by the risk of loss of permanent constructions or moveable property of any great intrinsic value. That an apparently unpromising site was in fact usable is supported by the evidence from the excavations, which confirm its use for commercial, domestic, and industrial purposes from the end of the second millennium through the middle of the sixth century B.C.
Şek. Table of Chronology with Elevations
Stratigraphic and Chronological Summary
The results of the excavations in sector HoB, set out in detail in the following chapters, reveal a long tradition of occupation at the site, as well as consistency in the manner of living and the use of materials. The thick layers of gravel show that there were occasional catastrophic floods, which would have interrupted life in the area and destroyed many buildings. It would have taken time to rebuild them, but no significant gaps showed in the stratigraphic sequence from the Late Bronze Age through the Early Iron Age and the time of the Lydian kingdom.
The evidence produced from sector HoB for Late Bronze Age occupation is one piece of a small but growing body of information about life in this region in that period. The long-standing project has elucidated many aspects of ancient Sardis, but only recently have other levels been found within the city that are as early as those found in sector HoB, described below, dating back to the Late Bronze Age.31
At the site of Sardis itself, occupation levels of the Early Bronze Age have been recognized only recently, even in the deep soundings (see below). But earlier, at Eski Balıkhane, about 30 miles (48 kilometers) to the north of Sardis, on the shores of the Gygaean Lake, an Early Bronze Age habitation site had been found; and at Ahlatlı Tepecik, also by the lake and not far away, an Early Bronze Age cemetery had been excavated.32 To the north of the lake lay another site, Kılcanlar. These places in central Lydia (Fig. 2.7) show that there was a considerable scatter of such sites and that small settlements had been established in many locations.
Since 2005, survey work in central Lydia, led by Christopher H. Roosevelt and Christina Luke, has identified a number of larger sites around the Gygaean Lake.33 One of these, at Kaymakçı, is much larger than the rest. Satellite images show the destroyed remains of three concentric rings of fortification walls enclosing about 8.6 hectares. This is by far the largest circuit in western Anatolia or the Aegean area, including mainland Greece and the Ionian coast. Surface surveys over several years have indicated that the Late Bronze Age was the floruit of this site. This discovery is most important for the local picture in Lydia. Here, then, is the ideal predecessor to Sardis as the center, or core, of Lydia. Excavations now in progress will test the validity of this idea, but whatever turns up, one must address the question about the mechanism involved in a sudden change of the regional center of gravity from Kaymakçı to Sardis.
Description of Major Occupation Levels
The occupation levels in the Lydian Trench of HoB are divided from one another by obvious bands of gravel and sand, as mentioned above, some as much as a meter thick. The main levels in the excavated area are summarized in Table 2.1, which gives approximate levels that vary from place to place depending on the slope of the land.34 The descriptive table of periods and levels set out here is meant to provide a general framework and a set of working definitions to be used in the main text of the following chapters. Of course, one must remember that the sloping topography and uneven surfaces have caused variations in elevations, and the range of levels must not be considered inflexible.35
This table differs in several respects from tabulations published previously36 and from opinions implied in the preliminary reports. The Lydian periods defined here are in principle derived from the major stratigraphic levels, usually surfaces or floors, with reasonable indications of occupation; they are only secondarily reflections of changes in pottery styles. As shown in Chapter 1, most styles of Lydian pottery do not belong exclusively to any one level but instead overlap. At issue is the relative quantity of a style within a level, and the introduction of certain new forms and styles over time, or the gradual disappearance of others.
While this chronological division may be satisfactory for the present, work continues, at Sardis and elsewhere in western Anatolia, that may call for revisions or provide a much richer picture of Iron Age cultures in the area. The first use of “Lydian” in the chart, for the early historic levels, is not meant to suggest that Lydians only came into being or arrived at Sardis in the ninth or eighth century. This is, however, the period of their rise to preeminence in western Anatolia, when they became politically powerful, as suggested by textual sources.37 The structure of their language and its relationship to Hittite and other Anatolian languages38 indicates that the Lydians were already a separate people early in the second millennium B.C., but we cannot be sure that they are the same group that inhabited the shores of the Gygaean Lake at that time.39 Indications that the Lydians’ occupation goes far back into the Bronze Age can be found in the stories of the dynasty descended from Lydos himself, before the accession of the Heraklid dynasty.40
Exposure of the Late Bronze Age stratum was small, and the principal finds, consisting of the floor of a small round hut and a cremation burial (in Deep Sounding B, pp. 46–47), tell little about the use of the area at the time. The hut was presumably a dwelling, but the single cremation burial does not indicate the existence of a cemetery. No other burials of this period are yet known from the Sardis region, although, as mentioned above, several inhumation burials from the Early Bronze Age were found together at Ahlatlı Tepecik. Those burials clearly formed a cemetery, but there were few signs of contemporary occupation close by.41 At present it can be said only that the hut and accompanying potsherds indicate settlement with at least one nearby burial.42
We were able to uncover only very small areas in HoB that went back as far as the Late Bronze Age, but by the time of the Early Iron Age and Lydian IV we are on firmer ground, given the large numbers of pithoi and the quantity of monochrome pottery. In the case of Lydian IV, the occupation has been inferred from the finds. The sub-Mycenaean fragments and a variety of the Protogeometric style of pottery seem to have flourished concurrently at Sardis: at least they are found together on one floor, so that their value for labeling chronological periods at Sardis, following the Greek model, is dubious (see pp. 43–44).
The principal floor of the Early Iron Age, or Lydian IV, stratum (at level *94.75), especially in the area of Deep Sounding C, is more illuminating, even though the only structure consisted of a flimsy wall (Fig. 3.8). Much of the soil in the thick stratum above the floor was burned red, while other parts had many particles of charcoal, suggesting the presence of some unidentified industrial activity nearby. This is an inference from the character of the fill, which seems to derive from activity on a larger scale than the finds themselves would indicate. There were fragments of large pithoi throughout, as well as a complete but oddly broken example of one in situ with its neck found inside it (HoB 216; see Figs 3.6, 3.7). However, the burning was not intense enough or sufficiently widespread to bear an interpretation of purposeful destruction.
In addition to the large numbers of pithoi, Lydian IV produced a few objects of daily life such as a whetstone (HoB 240), a pounder (HoB 239), a quern (HoB 103), a piece of flint, and some knives (HoB 37 and HoB 101).43 Although identifying activities that might have been undertaken during this period is difficult, it can be surmised that one focus required the presence of considerable storage capacity, presumably of staple foodstuffs. Since almost nothing in the way of structures was found here, it may be that it was a seasonal or outdoor working space, or that the structures were made of perishable materials.44
This general account can be set beside an analysis of the relative proportions of pottery types in the different levels, which shows an overall decline in gray or buff monochrome wares, changing from over 80 percent of all pottery finds in Lydian IV to 20 percent or less in Lydian I. A corresponding change is the dramatic rise in the proportion of painted to plain pottery, also in an increasing variety of shapes. These changes are particularly marked between Lydian III and II, and rather less so between II and I.
The architectural remains of Lydian III were both scanty and fragmentary. A small furnace or oven (see Fig. 5.14) was found45 with an associated three-sided structure too small for occupation, and no use of the area other than commercial or industrial can readily be seen. Several bothroi (rounded pits; see Fig. 5.1) that were dug into the clay covering of Lydian III from above, some of them lined with a limey plaster, probably originated in Lydian II, or they may have been connected with the laying of an intermediate clay surface directly over the destruction debris. These were more likely intended for an industrial purpose rather than for storage of food, given the probable dampness.46
The buildings of Lydian III are distinctly flimsier than those of Lydian II, and the density of structures is much less, even if one reconstructs walls out of almost every heap of stones. The finding of several human skeletons or disiecta membra that had (presumably) not received any last rites adds to the impression that the area was not highly prized or subject to ideas of tidiness or propriety—a feeling that has endured to this day in connection with factory yards or mining dumps. Because this level shows much evidence of having been the site of a battle, we call it the Destruction Level.
In Lydian II, we ought to be able to rely on a resumption of the flow of better-documented imports (or imitations) to delimit the chronological boundaries, but in fact, because we have been restricted to pots that continue the Protocorinthian tradition, which is in itself difficult to disentangle, this period is harder to pin down than Lydian III. In most situations evaluating individual potsherds is not so much the problem as balancing the evidence of several.
Lydian II is the first level in which coherent architecture is preserved, including a series of rectangular buildings connected by an enclosure wall (Fig. 2.3). Each one had a single-room plan in which built-in furnishings could be distinguished. Similar fittings are found in those of the succeeding level (Lydian I), sometimes rather better preserved. These structures were previously referred to as shops, following the identification of this area as Herodotus’s Lydian Agora, but we do not consider them equivalent to the retail establishments of the modern world, nor even to the later Byzantine Shops to the north of HoB (Fig. 2.2: no. 3), where goods were displayed in the street outside the premises.47 Rather, we now look upon these small structures as spaces that served as both houses and workshops. A distinguishing feature of Buildings G, H, and K was the number of iron objects they contained, and objects from Lydian II produced several indications of industrial activity.
We may restore Lydian II to include craftsmen in a place that combined dwellings and workshops selling new things but also repairing old ones. Some outsiders might have come to exchange foreign goods for local ones, but the use of coins had not yet begun. It is unlikely that the small-scale artisans would make use of pre-weighed bullion either, but our knowledge of the day-to-day details of ancient commerce at this period is slim. Most of the premises in the Lydian Trench could have been used as living spaces too, although one or two seem much too small (unless some kinds of temporary structures made of perishable materials were tacked on), even if one allows for a quite different understanding of space requirements and the number and type of household goods necessary for a family.
The evidence for specialized activity was most plentiful in Lydian I. Buildings A, B, and F (and M and N as far as excavated) and the “Lydian Room” and “Lydian Shop” were all small, rectangular single-room units, apparently scattered over the area at random, usually ten meters or more apart from each other. Inside and between them, several finds suggested shopkeepers’ or a homeowner’s stock: at one point ten unused lamps (HoB 579); at two others, concentrations of fired and unfired loom weights (HoB 645 [twenty-six loom weights] and HoB 716 [six loom weights]). Clearly a repair workshop was here. Sherds showing drill holes at their broken edges were not unusual, and excavators found the small lead strips used as cramps to repair pottery. The “Lydian Shop,” found in 1958, which belongs in Lydian I, is of the greatest importance. Here were found the corner of a kiln (Fig. 7.20) and a stock of Waveline amphora necks and other vases, some having drill holes for repair (HoB 524–HoB 533).48 In the area of Buildings A and B, another place where there must have been an industrial workshop, many simple bone pins were found (HoB 590)49 as well as a blank for a bone pin (HoB 593) and several other examples in various stages of manufacture.50 In the area of Building B, dies used in jewelry-making turned up, including HoB 602 and HoB 603, and two more came from nearby, just outside Building F (HoB 604, HoB 605). From south of Building C came a small, flat square of copper alloy, an intaglio die that was used as a pattern for impressed ornaments of gold foil (HoB 734). Thus, at various places from the periods of both Lydian II and Lydian I, much evidence supports the idea that these buildings housed small industries, even if they were also, probably, the simplest kinds of living spaces.
In general, the dating of the levels has been set rather loosely, because no external framework for the chronology of Lydian pottery yet exists. Imported pottery, where available, offers clues, but the dating and attribution of imported Late Geometric and Orientalizing wares found in East Greece and the islands of the Aegean are not really precise enough to do more than provide a broad indication.51 Furthermore, Sardis lacks a widespread sampling of figural Protocorinthian pottery, which might have allowed us to fix the chronology for the critical years between 700 and 650 B.C.52
After the Persian conquest of about 547 B.C., little of note was found in the Lydian Trench of HoB. This situation is mirrored in the sector Pactolus Cliff (PC), and one might wonder whether Sardis became a backwater then; but many graves containing fine jewelry, seals, and other luxury items from that time were excavated by the Butler expedition, and sector Pactolus North (PN), not far from sector PC, produced significant levels of the Persian period.
Possible explanations for this scarcity of remains in HoB and PC in the Persian period are that these areas’ function in relation to the surrounding district was altered so that the commercial activities were moved elsewhere or were drastically reduced. Another possibility is that the buildings of this period in HoB and PC were so insubstantial (as Herodotus describes them [5.101]) that they left no trace in the archaeological record and only ceramics of the period survive here. It is also possible that some buildings now dated to the Lydian or Hellenistic period actually belong, perhaps at least in part, to the intervening Persian era.53
Although the detailed discussion in the following chapters ends with the Achaemenid conquest of Sardis ca. 550 B.C., a quick overview of the Achaemenid and later levels in HoB is included here for the sake of completeness. Occupation continued in the fifth and much of the fourth centuries B.C., and post-Lydian structures clearly show the resumption of industrial use of the area. The large Building C, which roofed over a single enclosed space (perhaps with half-height walls) of 8 × 18 m, is best understood as a workshop or warehouse. The Stone Circle, Buildings E and F, and the well next to F were also in use in this period. In the later fourth century, the well was partially filled and the area essentially abandoned. Sometime in the mid- to later third century the well was completely filled in, and for the duration of the Hellenistic period the excavated area of HoB seems to have been a vacant lot where people camped and trash accumulated. The site again lay unoccupied in the later Hellenistic and early Roman period, during which time parts of Building C were robbed. This gap in occupation may be explained by the growth of the late Hellenistic and early Roman cemetery that was situated beside the main road nearby toward the north and west (see Fig. 2.4).
Later construction within the boundaries of the Lydian Trench of HoB itself amounted only to the laying of more than a dozen Roman terracotta water pipes having a general flow from south to north, and in a few cases from west to east. One major group of pipes ran through the east edge of the trench and another through the west edge (see Fig. 2.3); could they mark the lines of small Roman streets or tracks? It is quite fortunate that the laying of the pipes did not require deeper trenches that would have disturbed the seventh-century Lydian levels. A contemporary ground surface, not more than one meter higher than the pipes, could be traced by plotting hard-packed earth surfaces found during excavation. All that these show, however, is that the Lydian Trench area appears to have been composed of an open space among the surrounding buildings during later Roman times.
The overlying two meters of Late Roman fill, mixed with earlier materials, may be attributed to occupation wash that was carried downhill after the decline of the city in the seventh century A.D. or later. At any rate, the latest materials in this fill are not evidence for occupation in the Lydian Trench of HoB, but rather for the occupation on the slopes above, from which the material must have originated.
In the remainder of Part I, the chapters present the evidence that both supports the descriptions given in this introduction and justifies the general conclusions offered about the growth of a grand urban center. Starting with the Late Bronze Age and the Early Iron Age, the finds from sector HoB shed light on the early eras at Sardis, as recorded in the Lydian Trench, followed then by the four periods of Lydian occupation, from Lydian IV through Lydian I. The excavated area of HoB is large and deep; it has produced rich and revealing remains of Lydian life at Sardis on the one hand and, on the other, many incomplete and tantalizing architectural fragments and artifacts without meaningful contexts. The final chapter of Part I, then, offers a cultural and historical summation based on the artistic and technological connections of the finds.
In the discussion that follows, we present the objects and their contexts according to their levels, beginning with the earliest, an arrangement that in general terms follows the development of occupation at the site rather than the sequence of discovery; however, within the deep soundings, we follow the levels encountered as we were digging down. Chapter 3 thus begins with the deep soundings that were required to reach the early levels, where in each section we describe the stratigraphy from the Destruction Level (at ca. *97.00) and proceed downward to the level at ca. *90.00, which represents an unknown time in the Late Bronze Age (thirteenth to twelfth century B.C.).
The large and small finds presented here have been selected from the much greater number of inventoried objects from this sector (and the even greater number that were discarded or never inventoried) in order to present a general picture of the uses of different spaces and to provide a general chronological framework for the different levels of sector HoB. The following five chapters contain the discussion of the periods, from the Late Bronze Age to the middle of the sixth century. Then comes Part II, with a discussion of the finds from Pactolus Cliff; and finally, Part III of the volume presents a catalogue of the finds from HoB and PC to illustrate and back up our claims, accompanied by a second volume with plates containing photographs and drawings of each object.
- 1Butler, Sardis I.1, p. 4.
- 2Scott 2001; Hanfmann and Waldbaum, Sardis R1, pp. 1–2; Greenewalt 2003.
- 3Butler, Sardis I.1, and Butler, Sardis II.1.
- 4Hanfmann, “Sardis 1958,” pp. 22–27.
- 5Hanfmann, “Sardis 1958,” pp. 27–30; Hanfmann, “Sardis 1959,” pp. 30–32. More details are to be found on pp. 24, 90–92, where the Lydian Shop and the Lydian Room are described, as well as other exploratory test trenches below the House of Bronzes.
- 6Early Lydian pottery was not well known at the time, largely because many of the Lydian finds from Butler’s excavations came from graves not earlier than the seventh century.
- 7On the grid system in use at Sardis, see Sardis R1, pp. 8–10.
- 8The observation that this hill was artificial, and made of mudbrick, was first put forward by Andrew Ramage while on an afternoon promenade with Nancy Ramage. “Sardis 1976,” p. 64, and preliminary reports since; see also Greenewalt 2006; Cahill 2010b.
- 9Hanfmann and Waldbaum, Sardis R1, pp. 1, 44, figs. 13, 25; and “Sardis 1959,” p. 20.
- 10The Harvard-Cornell Expedition made use of several datum systems for referencing the levels at which archaeological features and objects were discovered; for sector HoB, this is the “B datum,” which references a specific point assumed to have an arbitrary elevation of 100 meters. This system and others used at the site are described in detail in Sardis R1, pp. 10–16.
- 11There were six pipes at the west side (ca. W30–35) and three or four at the east (E5–10), all running south to north, and one in the middle, running west to east at ca. S117.
- 12Cahill 2004a.
- 13Cf. Cahill 2019a; A. Berlin, “The Post-Lydian Occupation in HoB” (Sardis Expedition field report, 2015). Persian and Hellenistic activity in the sector is under study and will be the subject of a future volume on post-Lydian remains.
- 14On the Roman city wall, Sardis R1, pp. 35–52, fig. 11. On the Lydian wall south of HoB, Greenewalt, “Sardis 1999,” pp. 416–17.
- 15“Sardis 1958,” pp. 22–27; “Sardis 1959,” pp. 22–28.
- 16It is customary at Sardis for the individual excavators in charge of sectors to prepare typed descriptions of their progress halfway through the season, and again at the end. These are called mid-season and final reports for the particular year. To avoid confusion with the final publications of the sectors and various other annual reports, they will be referred to here as “mid-season reports” or “season reports.” Specialists on-site prepare similar reports, e.g., for bones or conservation.
- 17Reports describing the annual Harvard-Cornell excavations at Sardis appeared in the Bulletin of the American Schools of Oriental Research and its supplements from 1959 until 1990.
- 18“Sardis 1958,” pp. 27–30; “Sardis 1959,” pp. 30–32.
- 19In this book, “pit” is to be understood as an ancient feature for rubbish, storage, or industrial purposes, not as a modern investigation.
- 20“Sardis 1960,” p. 12.
- 21As laid out, Deep Sounding C covered a rough quadrangle within the area of E5–W9 / S98–109. By the end of the excavation in C, the much-reduced area of the trench covered from E3–W4 / S103–107.5.
- 22IN68.19 and IN68.20; Foss and Hanfmann in Sardis R1, p. 31; Foss, Sardis M4, pp. 44–45 and sources 18–19; “Sardis 1968, 1969,” p. 29; “Sardis 1985,” p. 20. See Hypaipa on the map in Fig. 2.7.
- 23A bracket was found in HoB at W6–15 / S90–95 *98.9–98.2. Others were found but not saved.
- 24D. F. Belknap in “Sardis 1978”; Donald Sullivan in “Sardis 1981, 1982,” pp. 53–57; W. Warfield in Sardis I.1, pp. 175–80; Sardis R1, pp. 17–18; Hanfmann and Mierse, SPRT, pp. 2–3. Brinkmann 1971. See also Cahill 2016.
- 25Olson 1970 and Olson 1971.
- 26Warfield, “Report on the Geology of Sardis,” in Sardis I.1, pp. 176–77.
- 27Cf. Cahill 2000, pp. 183–84 and n. 11.
- 28Opinion of the Turkish geologist Cengiz Saran, cited in Letters from Sardis, p. 144. See also Sardis R1, p. 26; and SPRT, pp. 20, 26.
- 29Ca. W200–E100 / S200–250. See Sardis R1, p. 52, table 4.
- 30This is not so absurd as it sounds: the clay of a great many of the bricks used for the face of the city wall was quite plastic and contained very little sand or straw.
- 31Excavations in 2017, 2018, and 2019 on Field 49 at Sardis revealed occupation levels of the Late Bronze Age, and below them, fill, perhaps terrace fill, apparently dating to the Early Bronze IIIB period. See Cahill 2019b, p. 124.
- 32SPRT, p. 17.
- 33Roosevelt 2009; Roosevelt and Luke 2009; Roosevelt 2010; Roosevelt et al. 2018.
34See note 10 above for the system used to record the elevations listed in Table 2.1.G. F. Swift’s original version is included below because it is implicit in the BASOR excavation reports and other publications, and a new version should not be offered without explanation and comparison. He distinguished eight main trench levels in his 1970 season report, which reflected his overall conclusions about the interpretation of the trench and his intentions for publication. These can be set out as he wrote them, from later to earlier periods:
- 1. Late Roman mixed fill
- 2. Roman: water pipes and traceable ground surface
- 3. Hellenistic: *100.00–99.00
- 4. Lydian level I (end of seventh century): ca. *99.00 (floors)
- 5. Lydian level II (later seventh century): *98.50–97.50
- 6. Lydian level III, Destruction Level (early to mid-seventh century): ca. *97.50–96.50
- 7. Protogeometric (eleventh or tenth century): *94.75–94.25
- 8. Late Bronze Age (pithos burial, etc.): *91.00
- 35For instance, some objects from Lydian III come from as high as *98.2 (HoB 360); and from Lydian I, up to *99.45 in the floors south of Building C (HoB 728).
- 36Ramage, Sardis M5, p. 3; Hanfmann and Mierse, SPRT, p. 26.
- 37Herodotus I.7 (Pedley, Sardis M2, no. 26).
- 38On the Lydian language, see Gusmani, Sardis M3; Melchert 2010 with bibliography. See also Beekes 2002; Beekes 2003.
- 39Mitten and Yüğrüm 1971.
- 40Herodotus I.7 (Pedley, Sardis M2, no. 26).
- 41Hanfmann, Mitten, and Ramage, “Sardis 1967,” p. 7; J. Spier in Hanfmann and Mierse, SPRT, pp. 17–20; Mitten and Yüğrüm 1971; Mitten and Yüğrüm 1974.
- 42An Early Iron Age child burial was found in a trial trench dug in 1985. See p. 44. Greenewalt, Rautman, and Cahill, “Sardis 1985,” p. 84 and fig. 32.
- 43Including grindstone S62.58; flint F62.3.
- 44Compare Lydian IV from Pactolus Cliff, pp. 117–118.
- 45At W12 / S97.
- 46It must be admitted that the practice of storing grain in pits was continued until not long ago in central Anatolia, but these were so different in construction and circumstance that they support our suggestion that the Sardis bothroi were not used to store food. See Schmidt and Krogman 1933, p. 115, and the massive underground silos at Boğazköy that were able to contain thousands of kilograms of grain: Seeher 2006.
- 47Crawford, Sardis M9, pp. 86–87 and 90–91.
- 48Ramage 2008.
- 49Also found here were HoB 606–HoB 609.
- 50HoB 591, HoB 592, HoB 610.
- 51The authors follow Coldstream for Geometric chronology (Coldstream, GGP, p. 330 [repeated in Coldstream 1977, p. 385]), but there is still plenty of room for overlapping styles or debate over where in a series any particular piece falls. See Cook 1960, pp. 118ff., and more recently Kerschner and Schlotzhauer 2005. Special thanks go to Michael Kerschner and Nezih Aytaçlar, who helped identify pottery during a series of productive meetings at Sardis.
- 52According to Judith Schaeffer’s report (Sardis M10, p. 6, table 2), all but one of the Early and Middle Protocorinthian pieces found at Sardis are linear kotylai. It is difficult to distinguish between specific periods when one is dealing with small sherds of these styles rather than whole profiles. Late Protocorinthian kotylai are represented by two pieces only (Sardis M10, cat. Cor 26 [P65.64] and Cor 37 [P87.82]), one of which has a scale pattern with added red (from another sector, ByzFort). Several of Schaeffer’s chronological judgments have dates that are too low by current standards, in the opinion of Kathleen Lynch.
- 53See Cahill 2017; Cahill 2019a; Dusinberre 2003.