1. Introduction

The Temple of Artemis at Sardis, the fourth largest Ionic temple in the world, is situated on the western slopes of the Acropolis, below the mass of the Tmolos Mountains in a broad valley opening into the ancient Pactolus river bed. Like the other two Artemisions of Asia Minor, the great archaic/Hellenistic temple at Ephesos and the Temple of Artemis Leucophryene at Magnesia-on-the-Meander, the principal facade of the temple was toward the west. The area might have been sacred to Cybele/Artemis from the earliest days onward as attested by the giant sandstone altar located at the west end of the temple.¹

The temple is well preserved in terms of its overall structure and clear details illuminate the process of its construction. Yet, certain portions, such as the entire west end, the north and south peristyle colonnades, and the roof structure are entirely gone, altered, or never finished. We have a single, intact, piece of the architrave, but none of the frieze, cornice, or the hypothetical pediment. This makes the understanding of the original design, and its successive renovations, very difficult. Furthermore, it appears that the architecture of the Temple of Artemis at Sardis was highly unorthodox, making it hard to place in a traditional category of Greek temple design.

2. History of Use

The layout of the cella and the front and back porches must have been conceived and developed in the century or so following the death of Alexander the Great, ca. 300-175 B.C. In the Roman imperial era the cella was divided into two chambers, and the temple incorporated the Imperial Cult. During this period, major construction was undertaken in laying out the peristyle colonnade. However, it appears that only the well preserved columns of the east end and east porch were completed. By the 4th century, Christianity had triumphed in Sardis. The temple was abandoned unfinished and a small church was erected at the southeast corner. In time, its marble blocks, paving, columns, and entablature were either systematically broken up for rebuilding the Byzantine city, or burned for lime. Parts of the cella, covered in waterproof stucco served as a large cistern. By the 9th and 10th centuries the temple was buried deeply by landslides from the Acropolis, but the tops of several columns with capitals and architraves were tantalizingly visible, and recorded by 18th and 19th century travelers.

3. History of Excavations

The earliest attempt to excavate the building was made by Robert 'Palmyra' Wood, who uncovered one of the two standing columns in 1750s.² C. R. Cockrell visited the site in 1812 and made some sketches of the standing three columns of the temple but did not publish them.³ In 1850, George Dennis, British Consul in Izmir, made several trenches in the cella and discovered a colossal head which has been identified with Faustina the Elder, the wife of emperor Antoninus Pius (AD 138-161).⁴ In 1904, Gustave Mendel, representing the Imperial Ottoman Museum in Istanbul, dug two trenches down to the bases between the columns #5 and #6, and on the south side of column #1.⁵ Proper and systematic archaeological excavations in the temple were undertaken from 1910 to 1914 by Howard Crosby Butler, the director of the Princeton Expedition. The results were published in the first two volumes of the Publications of the American Society for the Excavation of Sardis.⁶ Between 1960 and 1970, the Harvard-Cornell Archaeological Expedition, under the direction of George M.A. Hanfmann, undertook a partial study of the temple and its precinct; some ten test trenches were opened.⁷ Outside the official Sardis Expedition, the Temple of Artemis has become the subject of several scholarly studies since 1961.⁸

Since 1988, a comprehensive investigation of the temple was started by the author for the Sardis Archaeological Expedition, under the directorship of Crawford H. Greenewalt, Jr. Among the primary goals of this project are a complete architectural documentation of the building and its construction details (plan and section-elevations at 1/20, and details at 1/10 and 1/5 scales), and several computer-aided reconstruction studies. The close observation and recording of the building has resulted in new information illuminating the complex construction history of the temple and revealed a picture in some ways significantly different than previous hypotheses.⁹

4. The Building

The Temple of Artemis is a pseudo-dipteros, eight columns in front and twenty along the sides. The original pronaos (facing west), and the opisthodomos (facing east), display projecting prostyle porches with four columns in the front and two in the returns behind. Consequently, the side ambulatories (pterons) do not wrap around the ends uniformly as it is normal in pseudo-dipteral temple plans. The sides are the usual two, but the ends are three interaxial distances wide. The columnar spacing of the flanks, uniform at 4.99 m., is considerably narrower than the east and west ends where interaxial spacing increases progressively from 5.31 m. at the corners to 5.45 m., 6.65 m., and 7.08 m. at the center. Such complex contraction of interaxial distances is uncommon in Hellenistic temple design; it is an archaic Ionic practice and can be seen at the Artemision of Ephesos. The overall dimensions of the peristyle is 44.56 m. by 97.50 m.; the cella is exceptionally elongated, 23.0 m. by 67.52 m., a ratio of 1:3, another archaic characteristic. The original base or podium for the cult image, now preserved in sandstone foundation blocks, was placed centrally inside the cella (roughly 6.0 m. by 6.0 m. and 16.20 m. from the west door). A double row of twelve columns, now preserved in marble foundation blocks, helped to reduce the span and support the roof. The cella floor is ca. 1.65-1.70 m. higher than the porches and ambulatories. The original front porch (west) between antae measured 18.20 m. by 17.70 m., the back porch (east) 18.20 by 6.01 m. (a ratio of 1:3). The side ambulatories are 8.09 m. wide (10.78 m. including the colonnade); the ends, measured from the antae, are ca. 12.35 m. (15.04 m. including the colonnade). Careful measurements made in 2000 at foundation and plinth levels revealed a distinct convex curvature of the north and south walls of the cella although the full extent and consistency or the curvature has been compromised by the settlement of temple foundations.¹⁰

As it is preserved today, it appears that only the columns of the east end and the east porch were erected (although several foundation bases of the flanks might have received columns now gone). This represents 15 or 16 columns out of a total 64 required by the pseudo-dipteral plan. Most of the foundations of the side columns are in place, except those of columns #19, #21, and #43 on the north side, which are missing entirely, or were never built. Of the six columns of the projecting west pronaos porch, five are preserved in foundations, and one, #52 (northwest corner column), is entirely missing. With the exception of the southwest corner column, #64, none of the columns of the west end, even as foundations, were ever constructed. In contrast, the spectacular row of eight columns of the east end are preserved at different heights; two of them, #6, and #7, are intact including their capitals.

4.1. Building History

There is a general agreement that the inspiration for the Temple of Artemis belongs to the post-Alexandrian world of ideological and materialistic expansion, in particular, perhaps, to Alexander's visit of Sardis.¹¹ The colossal building was probably begun under the Seleucids, probably soon after the battle of Kyropedion in 281 B.C. This date is also suggested by an inscription found in the temple which mentions Stratonike, the wife of Seleucos I (301-281 B.C.). The original structure might have been conceived as a dipteros, following, in general, the great Ionian dipteroi, such as the Temple of Apollo at Didyma, or the late classical Temple of Artemis at Ephesos.¹²

A coin horde found in the foundations of the image base inside the cella, and the so-called Mnesimachos inscription on the interior of the northwest anta wall, both dated ca. 220-200 B.C., indicate that the cella and its interior columns, the image base, and possibly, the front and back porch columns inside the antae walls were completed, and the building in use, by the end of the 3rd century B.C.¹³ At this time no work had even started on the exterior colonnade.

Several scholars have suggested that the exterior pseudo-dipteral peristyle belonged to a second "Hellenistic" phase of construction during the last quarter of the 2nd century B.C., perhaps under the growing influence of Hermogenes and his renowned Temple of Artemis at Magnesia-on-the-Meander¹⁴. The hypotheses is attractive and there are general similarities between the two Artemisions (proportions of cella divisions, column alignments, and the central position of their image bases). However, the planning differences between the two temples are significant and fundamental, especially the unusual front and back porches and the deliberately archaizing, complex, interaxial contractions at Sardis compared to the orthodox plan of the pseudo-dipteros of Hermogenes at Magnesia. In fact, the Artemision of Sardis is not a true pseudo-dipteros at all.

4.2. Current Work: Construction Techniques and Details

Current investigations at the temple produced ample evidence to indicate that the entire peristyle colonnade and the dividing wall of the cella belonged to a single project undertaken during the Roman imperial period. Two structural and archaeological conditions support this conclusion. First, individual marble block foundations of the peristyle columns (ca. 1.70-1.90 m. deep) are connected to each other by a construction of heavy, lime-mortared rubble, a typical local version of Roman concrete (opus caementicium). In contrast, all of the major cella walls have continuous block foundations, ca. 2.50-3.0 m. deep. The nature and details of this construction makes it clear that it was laid concurrently with the block foundations of the columns, not added "as reinforcement" at a later period. Second, almost none of the foundations of the north and south side peristyles could carry finished bases or columns because their top courses were never finished. In these instances, the top courses (some preserved below the level of the ambulatory) are composed of very rough blocks, crudely fitted together, with uneven surfaces. These blocks do not have the usual construction features such as clamps, dowel holes, or setting marks, except for large and crude lewis holes, essential for lifting them. At least nineteen of the northern and six of the southern peristyle bases were not finished to receive columns above them.

The walls, bases, and foundations of the temple display two clearly distinguishable construction techniques that can be associated with the two major phases of building, Hellenistic, and Roman imperial. In the original, Hellenistic work, marble blocks are joined to each other by I-shaped ("hook") clamps; there are small, square dowel holes paired with "prying holes", or slits, used for crowbar leverage in setting the blocks. Dowel holes never have channels for lead runoff. Lewis holes are entirely absent. Blocks are well fitted and well trimmed. This type of construction conforms to the widespread Hellenistic masonry practices of Asia Minor. It is uniform to all four major walls of the cella, and all of the foundation blocks of the interior columns.

The second type of construction recognized as Roman work shows blocks with relatively uneven surfaces and rougher fit. Many of them reused, these blocks are generally much smaller than those of the first type. Lewis holes ("double splayed") are used widely. Large "butterfly" clamps are the norm although in a few instances I-shaped clamps are also used above foundation level. Square dowel holes have channels. This type of construction is used consistently in the middle cross wall, the new west wall of the cella (built ca. 12.80 m. west of the cross wall), and all the foundations of the peristyle columns, including those of the east and west porches.

4.3. The Greco-Roman Building

After an unusually long period of inactivity, around the middle of the 2nd century AD, major construction must have resumed at the unfinished Artemision of Sardis. The cella was divided into two chambers of equal length (ca. 25.10 m.) by a 0.90-meter thick wall; the original west wall and its door were brought forward (westward) to create a new west porch exactly the same depth as the east porch (6.07 m.). The blank east wall of the back porch was cut open and rebuilt with a door displaying handsomely ornamented jamb profiles, and approached by stairs, in imitation of the original west door of the temple. A new image base, preserved in its mortared rubble foundations, was constructed at the back of the new "west cella," against the dividing wall. The six-column prostyle porches of the east and west ends must have been created at this time by moving forward and re-erecting the two pairs of original, in antis columns (11 and 12 on the east; 53 and 54, on the west), and adding four new columns to each, a suggestion first made by Gruben.¹⁵ The prostyle porches thus created enclosed magnificent hall-like volumes, probably open to the sky due to their great spans (13.5 m. by 18.0 m.).

The two well-preserved columns of the east porch (11 and 12) are exceptional elements in classical architecture. Raised on very tall, unfinished, pedestal bases (2.17 m. high), they carry slender columns with fluted shafts -- the only instances of finished fluting in the temple. These shafts, and the finely crafted Ionic capitals they once carried, are attributed to the Hellenistic phase of the temple, while the pedestals, composed of roughly hewn, reused column drums from the original building, belong to Roman construction.

The transformation of a colossal, unfinished Greek temple into a Roman pseudo-dipteros with back-to-back cellas, posed certain problems, and produced an unorthodox architectural solution. Unlike a typical, Hellenistic pseudo-dipteros with a static and uniform peristatis around the cella, the making of unusually wide ambulatories connected to the tall, cubic volumes of the projecting prostyle porches, created and exploited a dramatic sense of space. In fact, the deep, tetrastyle-prostyle, box-like, porches of Sardis find distant parallels in the traditional Roman temples of Italy, such as the Republican Temple of Portunus in the Forum Boarium in Rome, or the Augustan Temple of Mars Ultor in the Forum of Augustus, but more strikingly, the Hadrianic Pantheon with its "pseudo-dipteral," three-aisled, porch. The Hadrianic connection is further underscored by comparison of the back-to-back cella arrangement of Sardis to the original layout of the Temple of Venus and Roma in Rome.¹⁶

It is impossible to assign a hard date for the Roman reconstruction of the temple. An intriguing inscription carved on the bottom fillet of one of the columns of the east colonnade proudly declares that its torus and the plinth are made from a single block of marble, and "of all the columns I was the first to rise." Based on letter forms and epigraphic style, a Hadrianic-Antonine date (ca. AD 140) has been suggested, and supported by circumstantial considerations.¹⁷

5. The Temple as Center of Imperial Cult

At least five, colossal portrait heads attributed to the Antonine family have been discovered inside or close to the temple. These have been tentatively identified as Antoninus Pius, Marcus Aurelius (Zeus ?), Faustina the Elder, Lucilla, and Commodus (found in 1996). Hence, the renewed construction at the temple, significantly the cella subdivision, might have been occasioned by the much coveted second (?) neocorate (the privilege to have a temple for the Imperial Cult) awarded to Sardis during the late Hadrianic or early Antonine era. In this new arrangement, Artemis must have kept the west facing cella (perhaps, even sharing it with Faustina, the Roman empress), and the emperor occupied the east facing cella (perhaps, sharing it with Zeus Polieus).¹⁸

The discovery in 2000, elsewhere in Sardis, of a statue base dedicated to Hadrian reinforces the likelihood of the emperor's visit of Lydia and Sardis in 123/24, a proposal supported by strong epigraphic evidence.¹⁹ One could suggest that the second neocorate of Sardis, and major construction and alterations on the temple in order to accommodate the Imperial Cult and imperial images, had been initiated under Hadrian although the completion of this work, and the making of the images, belonged to the Antonine era.²⁰ A late Hadrianic date for the reorganization of the temple also fits the previously discussed planning affinities between this unusual building and its close Hadrianic prototypes in Rome.