Sardis, 2011

In the summer of 2011 the Sardis Expedition undertook excavation, geophysical survey, conservation, restoration, and publication research at the ancient city site and at its necropolis of Bin Tepe. Bahadır Yıldırım was again assistant director, Crawford H. Greenewalt, jr. was director emeritus, and Seval Konak, of the Izmir Archaeological Museum, was the representative of the Ministry of Culture. We are grateful to Seval Hanım for her help and support, her intelligence, warmth and good cheer.

Excavation was conducted in the temple of Artemis and on two adjacent hills in the center of the ancient city, called ByzFort and Field 49 (figs. 1, 2). The two hills in the city center were both terraced in the Lydian period, forming a raised area of perhaps 6 ha, separated from the lower city by monumental limestone terrace walls (fig. 3). One goal of this season’s excavations was to investigate a Lydian terrace wall built of limestone blocks near the east edge of ByzFort, which had been uncovered by erosion the previous year.¹ Upon excavation, this proved to be three different terrace walls belonging to two phases, both probably dating to the first half of the sixth century BC (figs. 4, 5, 6, 7). The earlier phase consists of two parallel terraces stepping up the hill. These are built from relatively small blocks, clamped with lead butterfly clamps, one of which still remained in situ. In a later phase, also apparently dating to the first half of the sixth century BC, part of this wall was dismantled and rebuilt on a new orientation. While the earlier walls were parallel to the main terrace wall and probably reflect the natural topography, the later wall was on a different orientation, parallel to the Lydian terrace at the north end of the hill, and, indeed, to walls over the whole hill, revealed by geophysical survey of last summer (figs. 2, 4). This suggests that the whole hill was reorganized in the Lydian period to conform to a single system.

On the adjacent hill, Field 49, excavation revealed more of a Lydian terrace wall built of limestone ashlar blocks and some reused limestone and marble blocks, also dating to the first half of the sixth century BC (figs. 4, 8). Excavation in 2010 had shown that this wall was built on the stub of an earlier terrace wall built of large polygonal boulders, dating to perhaps the 7th c BC; and that it had been dismantled and rebuilt in the early Roman period, and again in the later Roman period, remaining in use for more than 1000 years. Excavation on top of the hill did not reach undisturbed Lydian levels. At the lowest level reached, however, in a context with mostly Lydian pottery of the sixth century BC, was found a jasper sealstone on a bronze wire mount, with the image of a grazing wild goat with long, curving horns (figs. 9, 10). This is the first such sealstone found in the Harvard-Cornell excavations at Sardis, and the first from occupation layers, rather than from graves.

The Lydian terrace wall may have remained in use into the Hellenistic period, when substantial buildings were constructed, only preserved in foundations. These walls, and the Lydian terrace, were apparently destroyed in the early 1st c AD, perhaps in the earthquake of 17 AD, and the terrace then rebuilt, reusing the same limestone blocks, but set in a substantial mortared rubble backing. A number of early Roman phases are preserved, one of which included a partly-exposed furnace. Later Roman phases include a room with a large subterranean plastered basin and other hydraulic features, and another room set into the side of the terrace against the Lydian and early Roman terrace wall. This latter room had a bench, beneath which was found a hoard of about 123 Late Roman bronze coins (fig. 11). The latest archaeological phase was a series of graves, probably Byzantine in date and now numbering fifteen in a small area (fig. 12). As in previous years, finds from these graves were very sparse and the dating is uncertain.

We are just beginning to understand the long history of this area, but the continuity is striking. The terrace wall remained in use from 7th or 6th c BC until the area was finally abandoned more than 1000 years later. Later buildings revealed by excavation and geophysical survey follow the alignment set up in the Lydian period. Moreover, a Lydian house excavated in the cavea of the theater in 2006 follows this same alignment. This region of the city was probably therefore structured and planned as early as the Lydian period, as was the adjacent ByzFort hill.

Prof. George Hanfmann suggested already in 1977 that ByzFort was a possible site of the Palace of Croesus, and excavations on this hill and the adjacent Field 49 support this hypothesis.² Monumental terrace walls of limestone ashlar masonry are non-domestic, appropriate for public structures such as gateways and palaces, or for a sanctuary.³ The finds from excavations in the 1980s and 1990s include many fragments of chalcedony, a material used for royal and elite tableware at Persepolis, Daskyleion, Uşak and elsewhere; these and the sealstone suggest an elite, rather than a domestic use for this area, and are less appropriate for a sanctuary.

The goal of excavation in the temple of Artemis was to investigate the columns in antis at the east end of the temple (figs. 13, 14, 15). Although most scholars assume that these columns existed, excavation in the west porch in 1911 and in the east porch in 1972 and 1996 failed to recover any trace of them or their foundations.⁴ Most authors follow Gruben’s suggestion that the columns were dismantled in the Roman phase, which he and others date to the second century AD, and set up again on pedestals as part of the new porches of the temple. This would have left a wide space between the east door and the new pedestaled columns, which is usually thought to be open to the sky.⁵

A long trench was excavated from one anta to the other, down to the natural clay bedrock on which the temple was built (figs. 15, 16, 17). This exposed ancient cuts in the bedrock in the appropriate places for foundation trenches for columns in antis, and one block of the foundation apparently still in situ. This confirms that the Hellenistic temple did have columns in antis, although even their foundations have been almost entirely removed.

The foundations for these columns were still in place in the Roman period, because the foundations of the stairway leading up to the east door were constructed around the column foundations. The columns and their foundations were removed by digging a massive pit from one anta to the other, which was backfilled with earth containing many fragments of column capitals, flutes, and other architectural fragments, presumably the remains of the columns in antis themselves. A colossal head of Commodus, one of half a dozen colossal statues that were set up somewhere in the temple, had been found in this same fill in 1996 (figs. 18, 19).⁶ The stair foundations then slumped into the void when the column foundations were removed; at that point the stairs must not have been in use (fig. 20).

The fill of this pit forms a single deposit dating to the Late Roman period, in the late 4th or early 5th c AD (fig. 21).⁷ The presence of column fragments suggests that not only the foundations, but the columns in antis themselves were removed and broken up at this time. We conclude, therefore, that the Hellenistic columns in antis stood in place until the Late Roman period, when they were dismantled together with their foundations. The pedestaled columns therefore cannot be the columns in antis, but must come from the cella, which was about 1.6 m higher than the peristyle, thus explaining the pedestals. With its columns in antis intact, the porch could have been roofed, or intended to be roofed, during the Imperial Roman period.

The reason for removing these columns, together with their foundations, in the late Roman period remains unclear. It cannot be simply stone-robbing, when so many other blocks of the temple were much more easily available for reuse. These columns and their foundations seem to have been deliberately targeted in a Late Antique modification or conversion of the temple, perhaps associated with the nearby chapel, Church M.

Lying on the foundation blocks of the Hellenistic northeast anta and continuing to the foundations of the Roman porch column in front (no. 16) was a mass of pottery, masonry debris, bricks, and rooftiles, mortared into the foundations of the column and therefore contemporary with its construction (fig. 22). Pottery from this deposit includes at least five largely complete objects: a large lamp of Broneer Type 21 (fig. 23), a local krater and a pitcher, a pseudo-Coan amphora, and a cooking pot (fig. 24). Sherds from the deposit included early forms of Eastern Sigillata B1, fragments of molded bowls and three molds, and thin-walled ware, typical of the first century AD, and the lamp and ESB are unlikely to be later than the middle of the first century. The deposit contained nothing that must be as late as the second century AD, when the Roman phase is usually dated. This suggests that the initial construction of the Roman phase of the temple may have begun significantly earlier than usually argued, in the Julio-Claudian period rather than Hadrianic or Antonine.⁸

Christian Hübner and Stefan Giese of GGH Solutions in Geoscience conducted geophysical prospection at the city site and at Bin Tepe. Their survey of the hills in the city center helped clarify the overall orientation of buildings in this region, adding to our understanding of the organization of this region and its long history (figs. 4, 25). Georadar was able to penetrate up to 1-2 m, and identify a few deeply buried structures. It confirmed that buildings on the ByzFort hill were oriented to the Lydian terrace on the front of the hill, while those on the neighboring Field 49 were oriented to the Lydian terrace on its west slope. Magnetometer survey was much less informative.

Geophysical survey at Bin Tepe focused on the large mound known as Karnıyarık Tepe. This colossal tumulus, 220 m in diameter and 53 m high, was explored through tunnels by the Sardis Expedition between 1962 and 1966, and through geophysical prospection and coring in 1993 and 1995.⁹ The tunnels had uncovered an early, unfinished crepis wall that had been buried when the mound was expanded, but failed to locate a chamber in the center of the mound. Electrical Resistance Tomography and magnetometer survey was conducted over the surface of the mound, but its large size prevented the instruments from locating even the tunnels dug in the 1960s. Prospection within the tunnels located another anomaly, but this too may be an artifact of the geology or tumulus fill.

2011 was the second year of a three-year project to conserve the Lydian Altar, which lies adjacent to the temple of Artemis.¹⁰ The focus of work was the later phase of the altar (LA2), which enclosed the early phase restored in 2010 (LA1). Nearly all the marble stair blocks on the front of LA2 had been robbed out in antiquity, and sandstone foundations and the perimeter wall had deteriorated since the building was excavated in 1910. Remaining blocks were consolidated with silane and acrylic consolidants, and restored walls to the state they were found in 1910. The foundations of the stairs are too fragile to leave exposed, and rather than rebury them, the robbed-out marble stairs are being replaced with new travertine blocks, which will both protect the building and make it more intelligible to visitors (fig. 26).

Work on the mosaics of the Synagogue also continued, repairing damage caused by frost heave, weathering, and exposure (fig. 27). Revetment mortars and deteriorated modern repairs were also cleaned and consolidated. Other conservation projects included the repair and study of a round building model found in the Synagogue in 1963. Burned and badly damaged when found, this resembles a smaller version of the Choregic Monument of Lysikrates in Athens, and is one of a number of earlier objects found reused in the Late Roman Synagogue (figs. 28, 29, 30).

The study of two Lydian electrum coins found in 2008 and 2009 (figs. 31, 32) continued with the help of Doç. Dr. Bülent Önay of Dokuz Eylül University. Like other Lydian electrum coins, these were made from about 53% gold and 44% silver. Numismatists have argued since the 19th century that the natural gold from around Sardis contained a significant but unpredictable amount of silver (about 27%), and that the earliest coinage somehow guaranteed not only the weight, but also the purity of the metal.¹¹

We therefore analyzed samples of natural gold from the Pactolus river and streams around Sardis by SEM-EDS. This showed, unexpectedly, that the natural ore around Sardis was almost pure gold, with only trace amounts of silver. This then raises the question, where did the Lydians obtain the electrum for their coins, if not from the Pactolus? As early as the reign of Gyges, the Lydians controlled Mysia and the Troad, an area famous for gold in antiquity and today. Gold mining in this area is attested as early as the Lydian period, and the gold from this area does contain a large amount of silver.¹² This region, therefore, rather than Sardis, may be the source of precious metal for the world’s first coinage; and the invention of coinage may be related to the rise of Lydian imperialism in the 7th c BC, rather than control of local resources.¹³

On May 4, 2012, we lost a great archaeologist and friend (figs. 33, 34). Crawford H. Greenewalt, jr. excavated in Turkey every summer from 1959 until 2011, working at Sardis, Gordion, Old Smyrna, and Pitane. He was director of Sardis for almost 30 years, from 1976 until 2007, and continued to work on his beloved Lydian pottery until shortly before his death. Greenie was a scholar and an artist, and many Anatolian archaeologists will remember his broad smile, his extraordinary knowledge and wisdom, his generosity and great modesty.

(Adapted from Kazı Sonuçları Toplantısı 34: 147-160)

• 1Ratté 2011, 102-7.
• 2Hanfmann 1977.
• 3Ratté 2011.
• 4Gruben 1961, 166-7; Hanfmann and Waldbaum 1975, 82; Greenewalt and Rautman 2000, 674-5; Cahill and Greenewalt 2016.
• 5Gruben 1961, 166-7; Howe 1999, 204-5; Yegül 2012, 105, fig. 8.
• 6Greenewalt and Rautman 2000, 675-6.
• 7Dated by African Red Slip pottery (Hayes Form 59 and 61) and other ceramics; a coin of Constantius II Caesar (324-330 AD) was discovered in 1972 (Coin 1972.1001, see ⁠Buttrey et al. 1981, no. 297). Andrea Berlin, Elizabeth DeRidder, John Hayes, Ulrike Outschar, Marcus Rautman, and Susan Rotroff kindly examined all the pottery in 2012.
• 8On the date of the Roman phase, see Gruben 1961, 182; Howe 1999, 204; Yegül 2012, 108-9. The deposit and the date of the temple are still under study.
• 9Ratté 2011, 74-77, with previous bibliography.
• 10Hanfmann and Waldbaum 1975, 88-103; Ratté 2011, 123-5.
• 11E.g. Head 1887; Wallace 1987; Le Rider 2001; Kroll 2001; Kroll 2008.
• 12Yılmaz 2002, 1837-1838.
• 13See Cahill, Hari, Önay and Dokumacı forthcoming.