Report 7: The Temple of Artemis at Sardis (2020), Fikret Yegül



The Temple of Artemis at Sardis, an unorthodox pseudodipteros, is the fourth-largest Ionic temple in the classical world and one of the most impressive in its natural setting. Located on the western slopes of the Acropolis, in a broad valley opening into the gold-bearing Pactolus River and the Hermus Plain beyond, the land evoked poetic associations within the city’s plethora of ancestral cults and overlapping, syncretic beliefs. The sanctuary, which predated the temple, as attested by an Archaic altar, was sacred to Artemis, the preeminent goddess of Sardis.

It is widely believed that the temple was begun during the Seleucid rule soon after the Battle of Koroupedion in 281 BC, which gave Seleucus I Nicator control of much of Asia Minor. After centuries of Persian domination, only a powerful and stable dynasty, which assumed the mantle of a new Hellenic cultural legacy, could have undertaken such a colossal construction. Because Seleucus died at the end of 281, soon after his great victory, the project can be directly associated with Antiochus I and his queen (and one-time stepmother) Stratonike, who had established Sardis as an administrative center for the Seleucid reign, and an official residence for its monarchs. Support for associating Stratonike directly with the Sardis temple is provided by a dedication on a marble ball which bears the inscription, “[Gift] of Stratonike, daughter of Demetrius, the son of Antigonus.” Although the present ball is probably a later copy, the original must have been made during the queen’s reign and residence as basilissa in Sardis.

Facing west like the other great Artemisia in Asia Minor at Ephesus and Magnesia, the original temple with its elongated cella might have been intended as a dipteros, like the other great archaic dipteroi of Ionia. However, Hellenic construction did not progress beyond the sole cella; none of the surrounding columns were put in place. The pre-temple archaic altar seems to have been enlarged during this period and physically linked to the west-facing pronaos by a flight of steps or ramps. The next major building activity can be associated with Hadrian’s almost certain visit to Sardis in AD 123/124, and his granting the city its second neokorate, the privilege to establish and maintain an official temple of the imperial cult. The rearrangement of the temple as a pseudodipteros and the incorporation of the new cult of the emperors by dividing the cella into two nearly equal parts were the primary architectural challenges of the Roman phase. Of the sixty-four columns designed for this phase, it seems very few were actually erected, and these concentrated mainly on the new east front, dedicated to the imperial masters.

Construction toward the completion of the temple must have slowed down and ceased through the third and early fourth centuries, as the appeal of Artemis’s cult waned. By AD 400, a small church (Church M) was built on the southeast corner of the temple, using the east end of the south pteroma as a kind of entryway. In the following centuries, successions of landslides of gravel, silt, and mud from the heights of Acropolis buried the temple except for some of the columns of the east end. Through the eighth and ninth centuries extensive destruction of the temple was in progress and human occupation had ceased. When Howard Crosby Butler, the director of the first Sardis Excavations, and his team arrived on the site in 1909, two half-buried columns standing in a barley field were all they could see.

As preserved the Temple of Artemis at Sardis appears to have had eight columns at the ends and twenty along the sides. There were six columns in two rows in antis in front (west) and in the back (east), but this arrangement was changed during the Roman rebuilding, when both ends received spacious prostyle porches with four columns in front and two in the returns. The central two columns of these porches, raised on tall, rusticated (or unfinished) pedestals which carry finished bases and fluted shafts, display an unusual, even unique (some said bizarre) appearance with no known historic parallels in the ancient world; they are Roman constructs composed of reused elements from the original building. The eight-column east front was completed, although researchers do not know if these columns ever carried a pediment. The side pteromas are two intercolumnar distances wide but the ends are three; consequently, the sides do not wrap around the ends uniformly as is normal for pseudodipteral temples. The spans between the columns of the east peristyle display what is known as “complex contractions,” increasing progressively from the ends (5.32 m) to the middle (7.05 m), a rather rare archaic system applied to temple fronts only. Here at Sardis it is used anachronistically and purposefully to designate the new front for the imperial cult.

The overall dimensions of the Roman era peristyle are 44.60 x 97.60 m, not counting any sets of steps, or crepidoma, for which there is scant evidence. The cella is 23.0 x 67.51 m (in a ratio of 1:2.92). The original cella walls are 2.13 m thick at the bottom and 1.93 m at the upper, orthostat level. The base for a cult image (ca. 6 x 6 m), preserved in sandstone foundation blocks, was placed centrally inside the original cella; however, it can be shown that, along with the archaic altar, this base (“basis”) was an existing feature of the earlier sanctuary and might have been influential in determining the positioning of the later temple. The cella floor is ca. 1.60–1.70 m higher than the surrounding ambulatories and must have been reached by a flight of steps (the markings for the stair structure for the Roman era east door are visible). The interior of the original cella had a double row of twelve columns, preserved in their foundations, that reduced the central span to 6.70 m (center-to-center span is 9.30–9.40 m); this is significantly wider than the spacing for the columns inside the west and east pronaos porches, which was ca. 8.40 m. The side ambulatories (from wall to column plinth) are 8.23 m wide (their full width is ca. 10.87 m). Thanks to the two fully preserved columns of the east end (columns 6 and 7), we can establish the column height of the Roman era peristyle at 17.87 m, including their capitals. Careful measurements made at foundation and plinth levels reveal a distinct curvature of the north and south walls and the east colonnade.

The ruins of the temple, especially its majestic columns against the Acropolis and the Tmolos range, were a popular and picturesque subject for generations of travelers, artists, and scholars. Starting with Cyriacus of Ancona in 1444 (who did no drawings at Sardis), the visitors included Robert Wood and his party (his draftsman Giovanni Battista Borra left us the first graphic representation of the building) in 1750; Richard Chandler in the company of William Pars, a talented young artist (though he left no drawings) in 1765; Charles R. Cockerell, the distinguished Neoclassical architect in 1812, who left several fine pencil drawings; Charles Texier, the French antiquarian, in 1830s; and Harald Jerichau, a Danish Orientalist painter whose dramatic panorama of the temple against the snow-covered Tmolos (1878) is in the State Art Museum in Copenhagen. A smaller, more realistic view of the temple (1873) by Jerichau featuring the two iconic columns standing among camels and spring poppies, highlighted against the acropolis, appears on the dust jacket of the book.

Systematic archaeological excavations in the temple were undertaken from 1910 to 1914 by the American Society for the Excavation of Sardis under the directorship of Butler, and the results were published in 1922 and 1925, the first two volumes of a planned series. George M. A. Hanfmann resumed the archaeological exploration of Sardis in 1958 under the aegis of Harvard University. Between 1960 and 1970, some ten small trenches were opened to study the temple and its precinct. Further small-scale excavations in the temple between 1996 and 2010 undertaken by N. D. Cahill, the present director of Sardis, helped to clarify aspects of the building’s history and design. A comprehensive investigation and architectural recording of the temple, which have resulted in this two-volume final publication, was begun by the author in 1987 under the directorship of C. H. Greenewalt, jr.

Description of the Building

Chapter 2, the longest chapter in this study, is devoted to a physical description of the temple and its construction principles. The discussion starts with a consideration of commonly used construction features and methods, principally the lewis for lifting and clamps for horizontal and dowels for vertical joining. For lifting heavy blocks as well as column drums the usual method was the standard lewis, featuring a rectangular cutting with double-sloped ends. The lifting capacity for an average size lewis (ca. 12–16 cm long, 10–12 cm deep), empirically calculated, was 4–6 tons. Larger blocks required two or three lewises. Of particular interest is an earlier lewis type known as the Carian lewis, which consists of a small narrow socket with slanting sides next to a larger, square hole intended to hold a wooden block to prevent the lifting iron key in the small cutting from slipping out. This type is used for all capitals except the two largest ones, in situ on the standing columns 6 and 7, which have standard Roman lewises. Since the earliest use of the Carian lewis is the fourth century BC and the type goes out of use in the second, the smaller capitals must belong to the original Hellenistic building and can be assigned to the interior of the cella and the pronaos porches.

The primary method for joining blocks horizontally was the metal clamp, which appears in two types: the bar type consists of an iron bar whose ends were bent down (like a “hook”) into depressions in the socket; the second type, “butterfly” clamps are wing-shaped, flat with no end depressions and typically larger. Hellenistic construction is characterized by bar clamps at or above the foundation level, but rarely below, as their use in the foundation blocks of the cella interior columns. Roman construction utilized butterfly clamps below foundation level and regular bar clamps above.

The most common type of dowel in Roman construction in the Artemis temple is a small, square pin or peg (almost always metal) with one, less commonly two, pour channels for lead. Widely used in Hellenistic construction, such as the main walls and the foundation blocks of the columns inside the cella, the “edge dowel,” which has a regular square cutting on the bottom block, is paired with a matching three-sided cutting on the upper one. Another Hellenistic type dowel is the simple, round pin used to connect the fluted drums of columns; unfluted (unfinished) Roman column drums are connected by a pair of small, square dowels, placed typically on either side of a central square cutting.

Although considerable variance occurs in construction methods over time, there are enough similarities and consistencies that allow us to identify major building phases as Hellenistic and Roman. Hellenistic phase construction (cella main walls and interior columns) display bigger, carefully shaped and fitted blocks. Construction cuttings are employed sparingly. No lewis holes are used except Carian lewises. Courses below foundation have no use for clamps at all; courses above are typically joined by rows of bar clamps. Edge dowels are common. Roman construction is limited to the east and west crosswalls of the cella (built to divide the cella into two), the foundations of the exterior peristyle columns, and the partially finished columns of the east and west ends. Roman walls are made of smaller blocks; core construction is shoddier, and details are coarser. There is a consistent and copious use of lewis holes and bar clamps; the latter, when below foundations, are of the butterfly type. Dowel holes are of the small, square type and often have one or rarely two pouring channels. The outside column foundations are in individual, large blocks, ca. 2.00–2.20 m deep. They are always connected by mortared rubble (a local variant of opus caementicium) encasing the block foundations on two, three, or all four sides—no doubt to provide additional stability against earthquakes.

The Hellenistic period cella, including its all-marble roof, seems to have been sufficiently completed to be functional by the middle or third quarter of the third century BC, but much of the detailing of the moldings of the walls and some of the finer finish of marble surfaces and joints were still unfinished. The same is true for the Roman period work as seen in the different stages of completion of the east end columns, such as their Asiatic-Ionic bases and unfluted, roughed out shafts. These unfinished stages of marble work provide us with an excellent demonstration of the working methods of ancient masons and the sequence in which the decorative finishes were undertaken. The bases of the columns, especially their elegantly profiled toruses, display an eye-catching array of ornament, horizontal and vertical laurel leaves, guilloche (or “basket weave”) patterns, plain unfinished ones, and one fully finished example of oak leaves among which small critters—lizards, salamanders, scorpions—play and hide.

Perhaps the most distinctive elements of the temple are its Ionic capitals. Five of these (designated A through G), plus several large fragments, are securely identified as Hellenistic originals based on their use of the Carian lewis and the refinement of their style. Many of these were recarved with standard lewis and dowel cuttings in their Roman reuse. The best-preserved and the most elegant of these, a smaller capital (C), was probably used for display for public admiration and professional emulation, as it also is now, on display in the classical gallery of the Metropolitan Museum of Art in New York. These beautiful capitals must have served as aesthetic models for the larger Roman capitals, as capitals A and B (in situ on columns 6 and 7) indicate: they are close copies of the Hellenistic originals in style and quality. The continuing artistic influence of the temple’s exquisite ornamental style upon its Roman followers can be observed also in the well-preserved jamb ornament of the east door. Dated to the Hadrianic period, the plasticity of the deeply carved eggs and beads and the delicately modeled Lesbian cymation reveals the persistent influence of Hellenistic models with specific reference to the anta capitals of the original cella.

Building History and Chronology

In Chapter 3 I look into the contribution of a variety of evidence—historical, archaeological, architectural, technical, epigraphic—in illuminating the building history of the temple. Some of these factors, such as the primary differences between Hellenistic and Roman era construction techniques and ornamental styles are discussed earlier in the book.

In the archaeological and numismatic category are the 126 coins datable to ca. 240–220 BC recovered from the vertical joints of the sandstone base inside the cella. Since they probably slipped in during the close of construction, they offer a terminal date for the time by which the building was operational. In the epigraphic category, an inscription in Greek carved on the interior face of the northwest anta of the temple records in detail the mortgage obligations that had been imposed upon one Mnesimachos by Antigonus, probably Antigonus I (ca. 382–301 BC), the founder of the Antigonid dynasty. The text on the temple wall is a later copy whose letter forms suggest a date between 250–200 BC Thus, like the date of the coins from the sandstone base, the Mnesimachos inscription indicates that the cella was complete and in use by the third quarter of the third century. Other critical evidence for the completion of the temple around the middle of the third century comes from a stele found in Didyma that records a deed of land sale by Antiochus II in 254/253 BC to his queen Laodice. The text stipulates that the original document was placed in the “royal archives” in Sardis but four copies were displayed in four important sanctuaries, Ephesus, Ilion, Samothrace, and Didyma. It is unlikely that the Sardis sanctuary would have been selected as a royal archival center if it did not have a functioning temple at that time.

Some of the five colossal heads (and many other fragmentary ones) found inside the cella or close to the temple, identified as members of the Antonine family (Antoninus Pius, Faustina the Elder, Marcus Aurelius, Lucilla, and Commodus), provide material evidence for the Roman reorganization of the temple as a cult center. For a closer dating of the Roman phase—particularly the construction of the exterior columns—evidence is provided by an inscription in Greek verse carved on the bottom fillet of column 4 of the east peristyle, north of the middle axis of the temple. The inscription, speaking in the first-person singular, accosts the passerby, “My torus and foundation block are carved from a single stone, furnished not by the people (demos) but given by the house (oikos) [of the temple],” and proudly declares, “of all the columns, I am the first to rise.” The celebratory nature of this inscription—presenting the column as a victor in a building competition—is confirmed by the fact that the torus of the base is decorated by laurel leaves, gathered and tied by a ribbon with fluttering ends. Based on letter forms and archaizing, literary context, most epigraphists point to a date in the Trajanic-Hadrianic period.

The association of Hadrian with Sardis and its temple is strengthened by recent research which shows that Hadrian and Sabina almost certainly visited Sardis in AD 123/124, during their Grand Tour of Asia Minor. The royal visit was celebrated by a statue of Hadrian (whose inscribed base was found in 2000) dedicated by the city. This must have been the logical occasion for the granting to Sardis of its second neokorate honors as generously awarded to other cities visited by Hadrian. The existence of a grand but unfinished temple on the site would have justified the political and economic decision to finish and adjust the structure for imperial purposes. In the newly designed pseudodipteros with back-to-back cellas, the west-facing cella was retained by Artemis while the east-facing one, entered by a new door and stairs on the east wall, was given to the imperial cult, housing, over time a swelling number of gigantic cult images, including possibly Hadrian and Sabina.

Architectural Analysis and Comparisons

In the final chapter, I review the architecture of the temple as an unorthodox pseudodipteros and analyze the mixed sources for its design. I have attempted to highlight its unique and experimental design and to evaluate its artistic and cultural links to the distant Hermogenean legacy of Anatolia.

For centuries the Hellenistic temple remained a simple, stark marble box with pedimented ends and gabled roof but no peripheral columns. The cella was probably raised on a low embankment, but without a proper crepidoma; it faced a monumental altar in front and was connected to it by a system of stairs. To soften the stark geometry of the lone cella there must have been trees and plantings but also rows of inscribed stelae and dedications, some dating back to the Lydian and Persian periods. One enigmatic dedication in Lydian refers to one Qλdãns, an enigmatic figure, once thought to be Apollo, but possibly the moon god (Mên), sharing the sanctuary with Artemis. The memories of these distant cults and religious traditions of Lydia, assimilated with Greek and Anatolian beliefs, appear to have been included in the sanctuary and honored by generations of Sardians. The unfinished temple, the marble box in its glorious setting, was the shape of the sacred.

For the unknown architect of the Roman pseudodipteros, the main lines of his design were already embedded in the shape created by his Hellenistic predecessor. The elongated cella was too narrow to accommodate a facade with ten columns, unlike its colossal mate at Didyma or its Hadrianic contemporary in Rome, the Temple of Venus and Roma. An octastyle dipteros, as in Ephesus and Samos, could have worked but given the ruinous cost of creating some 128 columns, a pseudodipteros—an admired system which had been introduced by Hermogenes in Magnesia nearly three centuries earlier—was the logical choice. Yet, as already mentioned, with its unequal ambulatories along the sides and spatially deep pronaos porches at ends, this was no traditional, Hermogenean, rule-bound pseudodipteros. These wide and lofty side ambulatories (over 10 m wide, 18 m, tall and 90 m long), defined by a stark, shining marble wall on one side and the repeating rhythm of the shadows of columns on the other, were connected at their ends to deep, spacious, six-column pronaos porches set within the east and west peristyles. Possibly open to the sky like cubic towers of light, these porches were the distinctive and dramatic design elements, creative and disruptive, that made the Roman temple at Sardis unusual, special, even unique among the pseudodipteroi of Anatolia, past and future.

A simple comparison of the plans and volumes of the Sardis pseudodipteros to Hermogenes’s masterpiece at Magnesia, and its Hellenistic followers at Alabanda, Lagina, and the Smintheion at Chryse, or its Roman followers at Ankyra and Aezane, illustrates the fundamental architectonic difference between the traditional pseudodipteroi and their interpretation (re-creation) at Sardis. For the origins and development of the Sardis model we must turn to Italy, where deep pronaos porches were and remained the historic model. Even after Vitruvius’s clearly admiring discussion of the advantages of Hermogenes’s new system (3.3, esp. 3.3.8), Italy expressed little interest in the pseudodipteral temple. The sense of overwhelming frontal space attendant to the deep porches was typical and routine in late Republican temple architecture, such as the temple crowning the Samnite sanctuary at Pietrabbondante, the Temple of Portunus by the Tiber in Rome, and, the culmination of the type, the Temple of Mars Ultor in the Forum of Augustus in Rome.

The design legacy behind the Sardis temple is complex and varied: the rich patrimony of cultural resources and hybrid borrowings blur the “origins” of all historical architecture of Asia Minor. I am not suggesting that the unknown architect at Sardis was a “follower” of Rome and Italy. If one seeks a specific meaning to the “Italian connection” at Sardis, we should look into the special relation of Hadrian to Sardis, an emperor who visited the city, honored it with an imperial cult privileges—and considering his artistic inclinations—probably had a say about the hybrid design of its foremost temple. His great Temple of Venus and Roma in Rome, with its back-to-back divided cella, might have provided the closest parallel the cella design at Sardis. Not in opposition but in addition to this relationship, the architect at Sardis, as a part of the cosmopolitan world of second-century Asia Minor, must have known what was what in Rome—he would have been familiar with the Column of Trajan and the Pantheon. Inheriting an existing long, narrow, austere, archaizing cella between a river and a mountain, his choice would have been daunting as well as unique. The resulting design, reshaping an extant building into an idiosyncratic and broadly conceived variant of a pseudodipteros (if a pseudodipteros at all) benefited from sources rooted in local Anatolian practice as well as distant foreign traditions close to the heart of an architect-emperor, reflecting a creative historical trajectory from Hermogenes to Hadrian.

In conclusion, the Temple of Artemis at Sardis is an experimental and eclectic building that is not so much an “end” to the popular pseudodipteral system as a re-creation of it. Set in the magical landscape of the sacred Tmolos, “in the land of a thousand gods,” it had no equal nor a clear follower.