Chapter 3. Building History and Chronology

Regarding the original design of the Temple of Artemis at Sardis and its successive renovations, Howard Crosby Butler wrote, “we are confronted by a curious combination of certainty and doubt.”¹ After five years of excavation and many more of research, Butler was uneasy because the well-preserved plan and the “mass of details” did not seem to add up to a clear design or a construction sequence that represented several distinct historical phases.

A century later (and after further archaeological work), we are in a much better position to understand the architecture and construction of the temple and its wider cultural context. There is, however, much we still do not know. We are still baffled by the unorthodox design of the temple and its complex building history. While the east end of the temple—the east wall with its monumental door, a six-column projecting pronaos porch, and an eight-column east facade—remain nearly intact, the west end is fragmentary and confusing. Three walls—at varying distances from the firmly preserved east wall (including the original west wall, also referred to in this book as the middle crosswall, of the temple)—cross the elongated cella, and the western two show indirect evidence for centrally placed doors. These crosswalls and the two back-to-back platforms for cult images indicate that, at some time in its history, the temple had a two-cella arrangement. However, relative to the overall topography and the great Lydian/Hellenistic altar (LA 1/LA 2), the architectural features and the major level differences evident in the west end are in many aspects unclear, and they require some hypothetical proposals (see Fig. 1.2). With the exception of a few topographical benchmarks, the original landscape around the temple still presents some questions, even if we can predict the nature of the larger features and the extent of the sanctuary.

Except for the southwest corner column (column 64), none of the west peristyle columns of the temple had been started even at the foundations; other columns of the peristyle are preserved unevenly in their foundations or are missing completely. Of the west pronaos porch, at least five of the column foundations including the middle two with pedestals) are preserved at the level of their tops or column base plinths, and they probably carried columns; significant numbers of fluted drums were found in this area, too. Even at the end of its life as a pagan temple sometime in the fourth century AD, the building was largely unfinished. New, intrusive construction inside and around the cella prepared the building for its continued, partial use during the Christian era and later—its afterlife, in current parlance.

A flight of marble steps on the northwest corner is an incongruous element with respect to Greek temple design, especially since these steps are not connected presently to an architecturally defined temple platform or to a stylobate, as a regular crepidoma would be (e.g., the Temple of Apollo at Didyma). In fact, we have no stylobate proper at all. A heavy mass of mortared-rubble construction envelops the temple peristasis. Furthermore, the early monumental altar LA 1 (substantially enlarged at a later date to LA 2), located on the temple axis at the west of the building and probably once physically connected to it, complicates the question of the temple’s orientation and its relationship to the earlier elements of the larger sanctuary of Artemis before the temple proper was built.

As preserved, the temple is categorically a pseudodipteros incorporating spacious amphiprostyle porches within its ends, three intercolumniations deep. We are confident that the cella and its interior columns, and probably the columns between the anta walls at both ends, were the first to be built (but were later removed or incorporated into a new cella configuration). The peristyle belongs to a later—Roman—stage coeval with the division of the cella into two equal-length, back-to-back chambers to house the newly granted Roman imperial cult. The two major, interconnected questions before us are thus:

(1) What are the origins and architectural nature of the pseudodipteral design and its possible predecessors?; and

(2) What is the date and nature of the conversion of the temple into a double-cella structure that responded to the needs of dual cult worship?

And, one might add,

(3) What is the relationship of the temple to Anatolia’s own architect Hermogenes, if any?

These questions, and others related to them, have engaged generations of scholars, who provide differing or overlapping solutions based on their own interpretations of the evidence: archaeology and physical remains, architecture, design and historical precedent, construction and materials, economics, epigraphy, ceramics, coins, and ornamental style—and, of course, the aspects of Sardis’s social and religious culture that affected the use and design of the temple during different periods.

I will present here a summary review of the current state of scholarship on the temple, starting with its original excavator, Howard Crosby Butler. Seeing the building through the eyes of many scholars and following their analyses and proposals, past and present, exposes the many sides of the problem and elucidates their various and sometimes divergent proposals. Many of these, I gratefully acknowledge and accept. Against this backdrop, I present my own interpretations and proposals (often built upon this earlier research and subject to the same historical shortcomings and conflicts) based on the last forty years of close documentation and study of the temple. Where I can offer no consistent and satisfactory explanations, and where I am at variance with other proposals (even some contemporary archaeological ones), I am content with presenting the evidence as I know and understand it, and I defer to the judgment of present and future peers; likewise, I will represent some alternative views but enunciate what theories I prefer and why.

Howard Crosby Butler

According to Butler, there had been an older marble temple, begun in the time of Croesus, whose purple sandstone foundations once emerged just west of the Hellenistic west wall of the temple. In 1960 Gruben identified these blocks (now gone) as the foundations of the stairs and the original west door of the cella (see pp. 52-54). Butler imagined that the course of limestone blocks running across the foundations of column 79 (built over by the late west crosswall and visible on the northern end of its west face) was further evidence for this “older” temple (“probably the relics of this temple”).² The purple sandstone foundations were instead constructions of the Hellenistic era for the door and its stairs, while the limestone has been shown to belong to the foundations of the Roman west crosswall (see Fig. 3.32).³

Butler’s hypothetical “early” temple, he explains, burned down during the Ionian Revolt of 499 BC and was replaced by an all-marble temple of the Classical era (by which Butler meant the mid-fifth century BC), which in turn was damaged in the fourth century BC. Butler proposed that this second temple was a colossal dipteros planned along the same lines as the Classical Artemision of Ephesus, though never completed. The present (third) temple was, under the Seleucids (perhaps Antiochus I, r. 281–261 BC), in Butler’s view, not an independent building but merely a continuation of its fifth-century BC predecessor; it incorporated many of that structure’s elements, including column foundations and bases, and most of the capitals and fluted columns, especially those raised on pedestal bases in the middle east pronaos porch (columns 11 and 12).

Butler also speculated that the renowned Anatolian architect Hermogenes, who may have been employed at Sardis during the rule of “the great Pergamene building monarch” Eumenes II (197–159 BC), could have been responsible for the temple’s pseudodipteral plan.⁴ Butler could not substantiate this complex chronological sequence of temples—from Archaic to Roman—with any convincing evidence, such as actual building elements or datable ornament. Nor has any archaeological evidence emerged to support his thesis in the many trenches opened in the temple since his time. Although Butler was wrong in his grand scheme of the temple’s evolution, he was right in some of his piecemeal interpretations of the temple’s past and architectural usage.

According to Butler, work on the peristyle columns was nearly complete when the great earthquake of AD 17 hit Sardis. He thus believed that Roman work on the temple was not a substantial rebuilding (or a “phase” proper) but rather was restricted to repairing, replacing, and strengthening the earthquake-damaged structure, especially the bases and shafts of the east colonnade; the capitals, however, appear to have escaped the destruction and were used for a second, and even a third, time.⁵

Butler proposed that the fourth-century temple faced east, with the old Archaic altar (LA 1) “at the rear of the temple”—a situation that was seriously and justly questioned by Gottfried Gruben (see below), as it would be by most scholars today.⁶ Butler argued that the east door, whose finely carved jambs were compared favorably to early Hellenistic ornament from Priene, belonged to the original building (which I now date on stylistic grounds to the Hadrianic period, see pp. 167–174).

Butler also considered what we call the “east crosswall” to be simply a thin, nonstructural screen wall, but still an original, internal feature of the Hellenistic cella. What he thought was the “back wall” of the temple (the Roman “west crosswall”) we now know was the front wall of the west-facing temple. The space between the latter and the west crosswall appeared to him to be an independent room with a pair of interior columns (columns 77 and 78); Butler erroneously dubbed this space the “treasury.” Since the floor level of this room (actually part of the pronaos porch) was some 1.60–1.70 m lower than the cella proper, any connection between the cella and the “treasury” would have required a door and a stair, for which he could find no evidence. He admitted that this was a problem for which he had no solution, but he tentatively restored a pair of doors at the north and south ends of the wall—shown with question marks on his plan—without any stairs.⁷ However, he correctly identified the weaker rubble-masonry foundations in the middle of the west crosswall as evidence for a central door into the “treasury.” Butler did not realize that the west wall, not bonded to the north and south long walls of the temple, is a late feature, not an original wall. More surprisingly, he did not notice that this wall was laid directly over the foundation blocks of the pair of columns of the west pronaos (columns 79 and 80; see Figs. 2.64, 2.65; Plan 6), directly west of columns 77 and 78. This last oversight may explain his misinterpretation of the temple’s orientation and resulting invention of a “treasury” as a part of the original Hellenistic scheme.

Another problem area for Butler—and still for us, to a certain extent—was the marble steps that lead up to the western end of the temple from the north (“Northwest Stairs”). Believing these to be an original, early aspect of the design, Butler incorporated them into his ingenious, entirely fanciful, and grandly symmetrical restoration that also connects the great altar to the temple’s west front (Fig. 3.1). The magnificent, neo-Baroque design actually works (but cannot be substantiated by evidence) and reveals, perhaps, more about Butler’s own roots in the Beaux-Arts tradition of architecture than what the evidence on the ground supports or how Hellenistic Ionic temples were designed.⁸

Butler did not wish to attribute an independent building or rebuilding “phase” to the Roman period, except for a strengthening of the foundations by encasing them in rubble construction and “extensive repairs at the eastern end,” necessitated by the earthquake of AD 17. However, the early discovery of the colossal head of the empress Faustina the Elder in the western part of the cella (and several other colossal male and female heads inside of or in proximity to the temple; see p. 198, Table 3.1) allowed him to propose that “the old west wall of the cult-chamber was removed . . . in order to convert the cella into a double sanctuary hall.”⁹ The deified empress stood on the Roman era concrete base in the new “duplex” cella, facing west, and Artemis, on the other side of the “light dividing wall,” faced east—a theory that was at least partially correct, as far as the conversion of the cella into a “double sanctuary hall” is concerned.

Butler also claimed that the entire cella had been made into a deep cistern after the temple fell into disuse during the Byzantine era. He interpreted the brick-colored mortar in the chamber between the original west wall (middle crosswall) and the west wall as a Byzantine version of opus signinum laid over mortared-rubble construction. Since he removed this layer almost entirely, it is hard for us to evaluate this interpretation, although what little remained in the northwest corner of the original cella (as observed by the author in the early 1990s) appeared no different than typical Roman mortar for floor paving in marble. This is indeed what British Consul George Dennis, who originally exposed that part of the temple in 1882, had thought. A more professional opinion was that of architect Francis Bacon, who was also at Sardis in 1882 and reported in a letter to Charles Eliot Norton: “He [i.e., Dennis] found a terra cotta Roman pavement.”¹⁰

George M. A. Hanfmann and Kenneth Frazer

The Harvard-Cornell Expedition, under the supervision of director George M. A. Hanfmann and architect Kenneth Frazer, undertook a partial study of the temple and its precinct between 1960 and 1972. Ten test trenches were opened, in addition to the deep sondage under the original image base, in order to elucidate questions of stratigraphy and dating.¹¹ These excavations provided evidence on the use of the area as a sanctuary sacred to Artemis during the Archaic and early Hellenistic periods—and perhaps at some point in time also to Zeus—but no architectural traces of an earlier structure were found under or around the temple. Hanfmann concluded that, contrary to Butler’s belief, no earlier temple, Archaic or Classical, existed below the present one, whose builders “constructed the piers [and column and wall foundations] directly into torrent deposits” that had washed down from the Acropolis slopes.¹²

Hanfmann believed that the temple was begun during Seleucid rule (Seleucus I Nicator, r. 301–281 BC), after the Battle of Koroupedion in 281 BC. He thought that at that time it was already conceived as a two-cella temple “with Artemis [facing west] and Zeus [facing east] as protectors of the Necropolis and Acropolis respectively.”¹³ However, Hanfmann also suggested that the division of the cella actually occurred during the short reign of Achaeus, the rebellious Hellenistic general who ruled Sardis 220–213 BC.¹⁴ He supported his thesis by identifying the badly mutilated colossal head found in the temple by Butler (see below) as Zeus/Achaeus, whose cult and image, he believed, had joined that of Artemis during this short period. This head is now tentatively identified as an Antonine emperor, probably Marcus Aurelius (see Fig. 3.55 and Table 3.1). Hanfmann also maintained that the establishment of the plan as a pseudodipteros occurred during this “Achaeus” phase, 220–200 BC, an early date also supported by some in recent scholarship on Hermogenes.¹⁵ In sum, Hanfmann and Frazer, unlike Gruben, saw the cella division as a pre-Roman undertaking, but the three concurred that the adaptation of the pseudodipteral plan was a Hellenistic idea.

Hanfmann explained that the Roman reconstruction of the temple began soon after the earthquake of AD 17 and mainly consisted of the erection of the east peristyle and east pronaos porch columns over their Hellenistic foundations and bases; in this he differed from Butler, who had repeatedly asserted that the post–AD 17 Roman work on the temple was mainly a repair and was limited to the east end. However, he followed Butler’s theory that this Roman phase included strengthening the masonry block foundations of the peristyle columns using mortared rubble; he believed the actual independent ashlar foundations to be original Hellenistic work.

In Hanfmann’s scheme, during the middle of the second century AD, with the granting of neokorate honors to Sardis, the colossal images of Antoninus Pius and his wife Faustina simply joined Zeus and Artemis. The goddess and the deified empress shared the west cella, occupying the newly made image base. The extensive mortared-rubble construction, which is preserved between columns 73 and 74 (immediately west of the east crosswall and extending to the north wall), was interpreted as the foundation for this marble-clad base. The same material was used as foundation fill on the east side of the east crosswall, between this wall and the original image base (see Fig. 2.80). The temple was left unfinished, with only the major cella roofed and the fully standing columns of the east end balanced by some—if not all—of the columns of the west pronaos porch, which was approached by the makeshift arrangement of the northwest steps.

Gottfried Gruben

It is surprising that neither Butler nor Hanfmann, directors of the two American teams at Sardis, paid attention to the copious and clearly delineated techniques and construction details of the monumental building—details that provide potent and objective criteria for dating. Both were more interested in stylistic analyses of ornament or sculpture, or the greater events of local and regional history in relation to the city of Sardis, as the primary criteria for explicating the building history of the temple; the evidence of the stone itself remained unattended. We are indebted to Gottfried Gruben, the eminent German architectural historian of classical architecture, who, in a seminal publication in 1961, challenged Butler’s design and dating sequence.¹⁶ Gruben’s hypothesis regarding the building history of the Artemis temple is based on objective, though by no means conclusive, criteria: the observation of characteristic or unique masonry techniques and construction details, such as the shape and type of clamps, dowels, and lewis holes.

Gruben determined that there were two distinct construction techniques (“Technik I” and “Technik II”) used in the temple that provide reasonably secure criteria for distinguishing two major phases: Hellenistic and Roman. The early phase (Phase I) is represented by simple, bar-shaped clamps (also known as hook clamps); square dowel holes placed at joints or edges of blocks (kanten-dübel, or “edge dowels”); and a total absence of lewis holes. The Roman phase (Phase II) is characterized by butterfly clamps with or without depressed, hook ends, dowel holes with single or double channels (for pouring lead or using the channel for lead overflow), and the widely used standard-type lewis holes.¹⁷

Using primarily these differences, Gruben recognized that the thin east crosswall and the west wall were not parts of the original Phase I construction, but rather belonged to the Roman period. He was the first to point out that these walls do not bond into the north and south walls (Figs. 2.46, 2.49, 2.67), nor do their masonry courses match those of the long walls. The so-called middle crosswall, on the other hand, bonded into the north and south walls of the cella and displayed all of the typical features of the earlier, Hellenistic construction technique; this was identified as the original west wall of the cella.¹⁸ A monumental door in its middle, indicated by the telltale evidence of weaker, porous, sandstone foundations, would have been approached by stairs whose design, comparable to the door of the cella east wall, Gruben ascertained; he also correctly identified the foundations of its side or parapet walls (clearly shown on Butler’s plan but now gone). Gruben established the extent and the orientation of the west-facing original cella, thus eliminating the “treasury” theory.

Gruben proposed three major phases in the history of the temple (Fig. 3.2).¹⁹ The first, early Hellenistic period phase (300–280 BC) was represented only by the cella with its deep, west-facing, almost-square pronaos (17.91 × 17.69 m) and a short opisthodomos, one-third as deep as the pronaos (17.91 × 6.00 m). The pronaos had six columns in three pairs, the opisthodomos one pair, and the cella interior had twelve columns in six pairs. The temple was oriented to the west and separated by a spacious platform from the massive altar, the latter structure predating the temple. Drawing his influential parallels in part from contemporary dipteral temples such as the Temple of Apollo at Didyma and the second Artemision at Ephesus, Gruben proposed that the original temple was conceived also as a gigantic Ionic dipteros, but construction never advanced beyond the cella stage.

Gruben’s second phase coincided with Pergamene rule and influence at Sardis in the late Hellenistic period (190–150 BC) when the present pseudodipteral arrangement with 8 × 20 columns was adopted. The west-facing cella remained the same, as did the west porch, which he postulated was accessed by a flight of marble steps from symmetrical northwest and southwest sides. Therefore, the adoption of a pseudodipteral scheme at Sardis would be understood as a response to contemporary ideas on Ionic temple planning popularized by Hermogenes. However, Gruben noted the difference between orderly plans based on Hermogenes’s academic rules and the free and creative interpretation at Sardis, with its three-intercolumniation-wide east and west ends occupied by deep, amphiprostyle porches that precluded a continuous ambulatory of equal width as the side pteroma.

According to Gruben, these unusual porches were created by moving the pairs of pedestal columns forward from their original in antis positions between the pronaos and opisthodomos anta piers to their middle positions. Although none of the columns of the long north and south sides—and none except the previously described pair of pedestal columns of the west end—were started during this period, he believed that the six-column pronaos porch of the east side was fully finished.

More critical to the shaping of future proposals in the design of the temple was Gruben’s belief that the eight frontal columns of the east end and the last few from the east end of the north and the south sides (columns 9, 15, 14, 18, and 20) were already laid in their foundations during this later Hellenistic period. Gruben justified this view by observing and claiming that the construction techniques revealed by the tops of column foundations 9 and 15 on the north, and 14 and 20 on the south (the only positions on the east end not occupied by columns), identified them as Hellenistic (see Figs. 2.114, 2.115, 2.131, 2.133). These features are, primarily, the use of bar-type clamps instead of butterfly clamps; however, as a full study of these features indicates, bar clamps occur in Roman construction also on courses at or above ground level, and butterfly clamps occur below (as also shown by Thomas Howe, see below).²⁰ Gruben supported his arguments for a Hellenistic date for the pronaos porch by holding that the fully finished bases, and the three capitals from the “small group” assigned to these positions (capitals C, D, and G), belonged to the Hellenistic period in terms of style, perhaps as early as the first phase of the temple, which he dated to ca. 300 BC.²¹

Gruben identified the third building phase as the division of the cella into equal parts during the Roman Imperial era. To him, the inscription recording the erection of column 4 carved on the fillet of its bottom drum (see pp. 190-193 below) and the colossal head of Faustina found inside the cella both suggested an early Antonine date for this phase of extensive building and renovation of the by-then long-neglected structure. He believed that the new “double temple” with its back-to-back cellas, like the Temple of Venus and Roma in Rome, would have incorporated the cult of the deified empress and that of Artemis.²² According to the German architectural historian, in order to create two spacious chambers of equal length, the original west wall of the cella would have had to be dismantled and rebuilt, including its monumental door and stairs, at a distance of 6.01 m from the west anta piers, thus shortening the original pronaos by two-thirds and creating a pair of identical porches at the east and west ends. The east wall would also have been partially dismantled and rebuilt with a new door and stairs leading up to the east cella. While the original image base remained in this space, a new base would have been created in the west cella, whose mortared-rubble foundation is preserved as a broad platform immediately west of the wall dividing the cellas, the east crosswall.

The reorganization of the west cella would have required some ingenuity, because the western half of the room, originally a part of the pronaos, ended up being ca. 1.60–1.70 m lower than the eastern half. An even floor was achieved by filling up the lower, eastern half of the pronaos. Telling evidence of this operation is provided by a clear, horizontal line of erasure across the upper surface of the Mnesimachos inscription, which was carved on the south, interior face of the northeast corner orthostat of the original pronaos (for a full discussion of this important inscription, see p. 163 below); the area below this line, marking the new, higher floor level, was buried; since it was not visible, the bottom part of the inscription was retained (see Figs. 2.54, 2.100).

The roof of the new west cella presented a problem because its interior supports represented two different spans or systems: the four interior columns of the original cella (columns 73–76) are on a different alignment and are spaced far wider apart than the columns on the western side of the new room (columns 77 and 78). Thus, they cannot be reconstructed with a proper architrave following a continuous straight line (see pp. 203-205 below). Gruben offered an ingenious and simple solution in his suggestion that the ineffective column pair 77–78 had been demolished and replaced by a pair of piers some 5.60–5.80 m to their east. Preserved in their foundation courses directly east of the dismantled west wall (middle crosswall), these coarse “piers,” follow more or less the alignment of the cella columns (for a reconsideration of this proposal, see pp. 67-68). As Gruben explained, they carried the timber trusses of the new roof, and there was thus no need to alter the interior columns or the roof of the east cella.²³

The mid-second-century Roman building of the perip-teros would have included all eight columns of the east front (over Hellenistic foundations, Gruben believed, as did Hanfmann and Frazer), almost all of the column foundations of the north and south colonnades, and possibly the remaining five (or all six) outer columns of the west pronaos porch, making it the mirror image of the eastern one. None of the columns of the west front, except southwest corner column 64, had been started even in their foundations, although it appears that they were planned, because there was just enough distance for a row of columns west of the great altar.

In studying the metrology of the Hellenistic temple, Gruben adopted an Ionic foot of 29.52 cm (one millimeter longer than the Ionic foot of Didyma) and rounded his figures when necessary, thus making the millimeter-accuracy of the Ionic foot moot. Gruben proposed a twenty-foot interaxial module, which gave an overall interaxial length of 300 Ionic feet and a width of 150 Ionic feet (actually 148.75 feet) for an 8 × 16 temple. He noted the discrepancy between the alignments of the present peristyle columns and those inside the cella, confirming the chronological difference between them—the cella came first, then the peristyle. Gruben’s ideal, dipteral scheme based on an imaginary interaxial distance of 5.90 m (20 feet) is a theoretical construct; the actual interaxial width of the peristyle columns is ca. 5.83–5.85 m. What is more cogent and convincing in his argument is the possibility that a dipteral arrangement conforming to contemporary principles of Ionic temple design could have been achieved using simple modular ratios based on plinth sizes, interaxial distances, and column heights, i.e., a system based on geometry.²⁴

Wolfram Hoepfner

In a 1990 study devoted mainly to Hermogenes and his architecture, W. Hoepfner focused on the Temple of Artemis at Sardis as an example of the high Hellenistic pseudodipteros, predating Hermogenes by roughly half a century and providing him with a system of temple design based on whole-number ratios of lower diameters and intercolumnar distances. According to Hoepfner, this system, viewed as part of a tradition represented by the complex (“multiple”) contractions of the frontal intercolumniations at Sardis (and the late Classical Artemision at Ephesus), was passed on to Vitruvius by way of Hermogenes. Thus, an early date for the Sardis pseudodipteros was important for Hoepfner’s thesis.

Eliminating Gruben’s second Hellenistic building phase altogether, Hoepfner maintained that the temple was designed from the onset, ca. 300 BC or slightly later, as a giant pseudodipteros that was finished only during the Roman Imperial era. Instead of its present, unusual plan, Hoepfner proposed that the original temple had traditional amphiprostyle, tetrastyle porches with pairs of pedestal columns in antis and four prostyle, directly in front of the antae. Thus, in this scheme the temple displayed a more “correct” and orthodox pseudodipteral plan with a continuous ambulatory two intercolumniations wide.²⁵

During the Roman Imperial era, the two pedestal columns in antis were moved forward to their present locations by leapfrogging them over the two middle columns (numbers 10 and 13 in Hoepfner’s plan) of the proposed tetrastyle porch, as these two were dismantled and brought out diagonally to the corners of the newly created pronaos porch. Hoepfner’s interesting, eye-catching, but orthodox proposal was put to test and to rest in 1996 when we opened a deep trench inside the east porch where one of the columns of his hypothetical tetrastyle porch was expected to stand (number 13, the third column from the north in Hoepfner’s plan). The trench reached a depth of 2.50 m below the surface and exposed only the purplish clay of the bedrock (*98.25–97.76); above this was the earlier occupational layer sprinkled with Lydian sherds.²⁶ It is a pity that some recent scholarship accepts Hoepfner’s thesis—though perfectly reasonable as a hypothesis at a time when archaeological evidence was lacking—as fact without any questioning.²⁷

Thomas Howe

The most recent work on the building history and design of the Temple of Artemis was undertaken by Thomas Howe, who was the chief architect of the Sardis Expedition from 1980 to 1982. His conclusions, published in 1999, refine and simplify the framework of the chronology set up by Hanfmann and Gruben, and are largely borne out by our long-term fieldwork.²⁸

Contrary to Hanfmann and Gruben, Howe proposed two rather than three major building phases: Hellenistic and Imperial Roman. According to Howe, the temple was a Seleucid project begun a generation or two after Alexander. During the first phase (280–200 BC) the west-facing cella and its interior columns were finished, but the exterior peripteros of columns, “whether planned as a dipteros or pseudodipteros, was not begun.”²⁹ If the original building was intended to be a dipteros, the proportions of the exterior colonnade could have been a simple, modular grid based on a column base or an anta plinth, rather than one following the complex numerical ratio proportions typical of third-century BC temples.

Skipping the late Hellenistic (or “Pergamene”) period of earlier hypotheses, Howe saw the cella division and the building of the exterior peristyle as broadly contemporary activities belonging to the Roman imperial period, or the “second phase” of the temple. Following and refining Gruben’s powerful analysis of construction techniques, Howe was able to surmise that the full peristyle, including the foundations of the east colonnade, used a clamping and doweling system different from the Hellenistic construction of the temple. He argued that the evidence—including the colossal heads of Faustina and Antoninus Pius found in the cella and the second neokorate awarded to Sardis, probably under the same emperor—indicated that the undertaking to finish and redesign the temple as a joint temple to Artemis, Zeus, and the imperial cult was an Antonine project. He also evaluated the uniquely spacious pronaos porch as an arrangement closer to that of a Roman pseudodipteros, such as the Temple of Zeus at Aezane, or the early Imperial pseudodipteros known as the “Wadi B temple” at Sardis (see for Aezane, Fig. 3.74; pp. 218-219 below; Fig. 3.72), rather than to the arrangement of a Hellenistic temple with the characteristic columns in antis cella design.³⁰

Howe presented a convincing and cogent thesis that the Antonine re-creation of the Temple of Artemis at Sardis marks the end of a conservative, late Hellenistic tradition of temple design whose origins reached back to Hermogenes.³¹ However, his association of the temple with the Zeus cult, an old and intriguing idea, finds no support.

The following section presents considerations and proposals for a revised building history for the Temple of Artemis at Sardis, one partly based on the ideas of previous scholars but mainly on the reevaluation of historical, epigraphic, ceramic, and (scant) numismatic evidence. Primary to this thesis, however, is the close study of the technical details of construction and architectural documentation of the temple (undertaken between 1987 and 1998) as well as limited archaeological investigations (sporadically since 1960, but mainly in 1972, 1996, 2002, and 2010–12). These new studies allowed the formulation of new proposals for the sequence of design and construction and provided clues to the temple’s chronology. However, in a number of instances they led to diverse interpretations of the evidence, sometimes with differences in end results.

In this study I have tried to include the sanctuary’s religious and cultic associations, especially the incorporation of the Roman imperial cult; comprehensive studies of religion and cult at Sardis obviously belong to other venues. In a short section devoted to the archaeological yield from a small trench between the northeast anta and column 16, I examine varying views based on ceramic and construction evidence. A review of some six or more colossal, iconic images of the Antonine era found inside or close to the temple indicates that they were an integral part of the temple’s design and usage in the Roman period. Finally, in a separate section devoted to architectural analysis and comparison I consider the temple’s unique design in the larger context of Hellenistic and Roman pseudodipteral temples in Asia Minor, as well as its potential connection to Western and Italian temple design.

Historical Considerations

The inspiration for the Temple of Artemis belongs to the Hellenistic world of inquiry, one of ideological and materialistic expansion that was ushered in by Alexander the Great.³² The conqueror did come to Sardis when the city surrendered peacefully in 334 BC, but we have no records that connect this visit directly to the temple history. According to the Greek historian Arrian, Alexander gave orders to build a temple to Olympian Zeus on the heights of the Acropolis, but there is no mention of endowing a temple to the Sardian Artemis.³³

Stratigraphic soundings in the east cella of the temple in 1960 provided sufficient evidence for Hanfmann to assert that there had been no earlier temple under the present structure, and that conclusion has not been changed by our various investigations in the temple through 2010–12. However, evidence for continual occupation of the site during the late Lydian and Persian periods (mid-sixth to fourth century BC)—and for the presence of a sanctuary or a precinct sacred to the goddess—that predates the Hellenistic temple includes ceramics, the Archaic altar (LA 1), and the sandstone image base (“basis”) inside the cella of the temple (see pp. 163-164).³⁴

There is general agreement that the colossal temple building was started under the Seleucids, soon after the Battle of Koroupedion in 281 BC, in which Seleucus I Nicator (r. 301–281 BC), the founder of the dynasty, defeated Lysimachus (a Macedonian general of Alexander) near Sardis and gained control of most of Asia Minor.³⁵ This was the historical and political turning point that would have rendered the conditions convenient, even necessary, for such a prominent civic project. It is also an opportunity to link a great building to a “Great Man” and a great event—a satisfying theory, though not always true or expedient.³⁶ We have no hard evidence that this was so, but a late fourth-century BC date is technically possible.

The assumption that links the temple to the new Seleucid rule is, however, based on two cogent and logical considerations. First, only a strong, stable dynasty could support such a large and expensive project. Second, after the conquest, Sardis remained an important city and assumed a special position among the cities and sanctuaries of western Asia Minor as a royal center of the Seleucid reign and an official residence (or one of the official residences) for its monarchs. As the site of the western terminus of the widely traveled Royal Road, the city’s historical and geographic position was too important to be entrusted to any but the leading power-holder of the new dynasty. It was the first place where Antiochus I and Queen Stratonike stopped to burnish the Seleucid royal presence after the untimely death of Seleucus I in 281 BC, and that is where Stratonike, an exceptional royal woman with a penchant to revive or recreate cult centers with political significance, both lived most of her life and peacefully died, as discussed further below.³⁷

An equally important indicator of the city’s importance as a capital, perhaps, was the location of the royal archives.³⁸ After conquering the city and establishing peace, Seleucus I and his successors had powerful motives for founding a proper temple in the old Sanctuary of Artemis at Sardis to rival those of Ephesus and Didyma.³⁹ The former was the home of the older Artemis cult, which was related to the Sardian cult; the latter was represented by a major temple—a well-documented bequest of Seleucus I himself—comparable to the Sardis temple in many aspects of its construction and detail.⁴⁰ Appealing to local deities and cults in post-Alexandrian kingdoms and committing large sums of money for major building (that is to say, land you build on is land you own) was not simply a politically motivated measure but rather an active step in exploiting the opportunity to legitimize the dynastic claims of their rulers. In working to revive older, pre-Achaemenid local and regional cults, and rebuilding or building new major temples in major sanctuaries, it would be sensible to assume, the Seleucid rulers were calling attention to their difference from the Achaemenid rulers, who were known for establishing great temples in the East but were strangely unmotivated to do so in their western colonial centers. As Rostovtzeff observed long ago, “there was probably no Greek city in the Seleucid Empire without a cult of the reigning sovereign, his family, and his ancestors.”⁴¹

We do not have independent evidence for a specific Seleucid ruler cult at Sardis, much less in the Hellenistic temple, but there was general interest in and a political advantage to establishing dynastic cults in close association with local ones. Such a politically motivated association might have paved the way for the inclusion of the Roman imperial cult in the Temple of Artemis in the second century. The same political benefits for cultic transparency and continuity was not lost to the new Roman rulers as they assiduously preserved and encouraged these Seleucid ruler cults and their priesthoods in the Seleucid centers of Asia Minor, even long after the dynasty had become extinct.⁴²

Queen Stratonike and Sardis

Seleucus I was killed in 281 BC, only seven months after his victory over Lysimachus. If the Sardian temple was a Seleucid project, then real work on the building could not have started before the reign of his son and successor Antiochus I, and his wife, Queen Stratonike—that is, by the end of 281 BC or soon after. While firm epigraphical evidence for Seleucus’s involvement in the Temple of Artemis at Sardis is lacking, he is credited for building the Temples of Zeus at Olba and Cilicia, and one much closer to Sardis, donating lavish gifts to the near-contemporary Temple of Apollo at Didyma, which must already have been under construction early in his reign. One would expect that his generosity toward Sardis, the famed capital of the Lydian kings and now his city, where he had defeated a strong foe, would not have been anything less.⁴³ Even if he had no time to build or start such a temple, he would have bequeathed the desire and intention to do so to his son and erstwhile wife.

Stratonike, daughter of Demetrius Poliorcetes (“Besieger of Cities”) and granddaughter of Antigonus I Monophthalmos (“One-Eyed”), was Seleucus’s wife before he divorced her in 294/93 BC so that his lovestruck son could marry her.⁴⁴ It is tempting to imagine this beguiling queen—whose life story has inspired generations of artists, writers, and musicians—visiting or even residing with her royal husband in their newly acquired capital, Croesus’s golden city, and honoring the Great Goddess with a temple, as she had done earlier for the local goddess Atargatis at Bambyce (Hierapolis) in Syria.⁴⁵ She was also known to have made many offerings to Leto, Apollo, and particularly Artemis at Delos.⁴⁶

Although there is no obvious evidence that Stratonike played a leading political role in both of her Seleucid husbands’ reigns, modern scholarship views her as a wealthy patron of religion, active and involved in honoring local cults as much as enjoying being the subject of cult worship herself (as “Aphrodite Stratonicis” in Smyrna: Orientis Graeci inscriptiones selectae XI.4, 228 and 229), not a passive pawn in convenient marriage arrangements. Stratonike’s position as a “lifelong ambassador,” or bridge, between cultures and dynasties would have guaranteed her close involvement in the religious affairs of the Seleucid-controlled cities of Asia Minor, particularly Sardis, which she had made her own.⁴⁷

A Babylonian account from the reign of Antiochus I for the years 276–274 BC confirms that Sardis, once a military and administrative center of the Persian stratege and a well-known stronghold, was the official residence of the Seleucid king and his queen, Stratonike, at least during this period. According to this source, on his way to confront the Egyptian army in Syria Antiochus “left behind his [court], his wife, and the crown prince in Sapardu [Sardis] in order to strengthen the garrison [in Ebir-Nari, Egypt].”⁴⁸ Another Babylonian record, a plaque with an astrological calendar, informs us that Stratonike died at Sardis between September 24 and October 24 of 254 BC.⁴⁹ The mother queen, who had probably married her younger daughter Stratonike II to Demetrios II, the bride’s Macedonian cousin, during the same year (or a year earlier), must have been sixty-two or sixty-three years old when she died.

Curiously, barely two generations earlier, Sardis had been home to another powerful Hellenistic queen, Cleopatra of Macedon, the daughter of King Philip II and Queen Olympias, and thus the sister of Alexander the Great. After her beloved brother’s death in 323 BC, many of his generals sought her hand in marriage (which would have increased their chances in succession and become king), but in the end, none succeeded. The rivalries and intrigues among the contenders and retainers caused Cleopatra’s semi-exile in Sardis to prevent her from upsetting the balance of power.⁵⁰ Cleopatra, brave and resourceful, but isolated from her Macedonian power base, stayed at Sardis for twelve years, perhaps because she was in some ways happy there, or else because she had no options.⁵¹ After an unsuccessful attempt to escape from the city, probably to marry Ptolemy of Egypt, she was brought back and assassinated in 308 BC, at age forty-six or forty-seven, on the order of Antigonus I Monophthalmos. Anticipating the political backlash that would result from the murder of this fiery and beautiful Macedonian princess (and queen of Epirus), Antigonus put the assassins to death and gave Cleopatra a fine funeral in Sardis.⁵²

An early admirer of the power, personality, and intelligence of these queens, Grace H. Macurdy aptly commented that “like many unscrupulous women of this extraordinary northern breed . . . Cleopatra was murdered by men who feared her power and prestige.”⁵³ Stratonike, as she strolled the gardens of her royal residence at Sardis, would have known all this. As a woman of that “extraordinary northern breed” herself, she must have pondered the tragic fate of her less-fortunate kinswoman, probably with wistful thoughts about her own future—it was a hard world, and for many women, it is still a hard world.

A particularly intriguing indicator of Stratonike’s connection to the city and to Sardian Artemis is a dedication found in the Temple of Artemis, a marble ball ca. 36 cm in diameter, bearing the inscription “[Gift] of Stratonike, daughter of Demetrios the son of Antigonus.” The letter forms indicate that this is a mid-second-century BC Hellenistic copy of the original gift, which must have been made during Stratonike’s lifetime, that is, early in the third century BC and presumably soon after the city came under Seleucid power in 281 BC.⁵⁴ This may be true, but it does not alter the historical usefulness of the evidence; copies of inscriptions of cultic, archival, and political significance were common. However, there are also some doubts about the authenticity of the inscription, as Stratonike’s royal title and her husband Antiochus’s names are missing.⁵⁵ However, in inscriptions on the queen’s other known offerings the name of her husband and her dynastic titles are often omitted in favor of her father’s name. This is especially true of her lavish gifts at Delos; she never calls herself “queen” or “wife of Antiochus” but simply “daughter of King Demetrius.” One explanation could be that she made this offering before she married her husband in 294 BC; or that she was following the third-century practice of omitting the husband’s name in dedications made by Hellenistic women.⁵⁶ The more convincing answer, I believe, is to be sought not just in her well-documented devotion to her father and her clan, but also in her sense of independence.⁵⁷

One might point out that even an authentic dedication by Sardis’s queen does not in and of itself prove that the temple existed; she may have made the dedication to the sanctuary of Artemis before the temple was built; there are a number of such dedications that predate the temple but were later set up in the sanctuary. However, there is no clear information that leads to such a conclusion either. The argument here is based on the strength of Stratonike’s historical personality and connection to Sardis, as she made the Lydian capital her own; the discovery of the dedication inscription (or its later copy) is an additional indicator of this special kind of involvement. It also takes into consideration her deep sense of religiosity, particularly her devotion to Artemis, which would have been an effective supporting factor for why the temple was initiated during her reign and residence as basilissa in Sardis.⁵⁸ Despite justifiable caution, I maintain that it was Queen Stratonike who made the dedication to Artemis at Sardis, and this makes it possibly the earliest physical document we have that is associated with our temple.⁵⁹

Mnesimachos Inscription

An equally important inscription in Greek carved on the interior northeast corner of the west pronaos wall of the Temple of Artemis records in detail the mortgage obligations of a certain Mnesimachos in return for a loan he received from the temple funds (see Fig. 2.54). The display of official deeds and records on important public buildings represents a well-known ancient tradition. Mnesimachos’s extensive estate, which had been granted to him by a king (including land, villages, dwellings, slaves, and other revenue-bearing assets), is described as a kind of collateral (in the modern sense) against this loan.⁶⁰

Considerable controversy exists, however, about the nature and date of this important inscription, which naturally is central to the question of the dating of the temple itself.⁶¹ If the Antigonus mentioned in this inscription is Antigonus I Monophthalmos (r. 306–301 BC), the founder of the Antigonid dynasty of Macedon, the date of the original mortgage document, perhaps carved on a stele, would be somewhere between 300 and 250 BC, since Mnesimachos could not be expected to have survived the king by more than about fifty years.

Mainly on epigraphic grounds, many scholars consider the text carved on the temple wall to be a later copy of an earlier document broadly datable to 250–200 BC.⁶² Thus, like the Stratonike inscription, the evidence provided by the Mnesimachos inscription favors a building that was largely complete and in use by the second half of the third century BC.⁶³

Another piece of circumstantial but critical evidence for the completion and use of the temple by the mid-third century BC at the latest comes from an inscription found in Didyma. This stele records a deed of sale of land by Antiochus II in 254/53 BC to his divorced queen, Laodice. The original document was placed in the “royal archives” at Sardis, but copies were displayed in four other major sanctuaries (Ephesus, Ilion, Samothrace, and Didyma). As pointed out by Gruben, it is unlikely that the Sanctuary of Artemis at Sardis would have been selected for this honor if it did not have a functioning temple at that time.⁶⁴ Dedications, votive steles, and inscriptions carved on statue bases found among the ruins of the building provide additional epigraphic testimony in favor of the continued use of the temple through the late Hellenistic and early Roman periods.⁶⁵

Coins from the Image Base (“Basis”)

A group of 126 Hellenistic coins (54 silver coins along the east side of the image base and 72 bronze ones along the north) and one silver croeseid (in the center) were found inside the purple sandstone foundations of the base (“basis”) located in the center of the original cella (now the east cella), which was believed to have carried the image of Artemis when the temple was constructed around it. These foundations, preserved as two courses of ashlar and occupying the entire area between columns 69/71 (north) and 70/72 (south), would have been covered in marble to create a large (ca. 3.60 × 3.60 m) but low square platform, rising ca. 0.50 m from the cella floor (Figs. 2.80, 3.3, 3.4, 3.5).⁶⁶

Butler, noting the tightly fitted blocks finished with a fine flat chisel and joined by many large “pi-shaped clamps” (bar/hook clamps)—comparable to Lydian masonry at Sardis—as well as the single croeseid coin, considered the base to predate the temple. Few of these clamp cuttings are visible now; they must have been destroyed or eroded in 1914 when Butler dismantled the soft-stone blocks and opened a three-meter test pit in the center of the cella to look for earlier structures (none were found), leaving the blocks lying in heaps. In 1960 the Harvard-Cornell Expedition attempted “no more than putting the chaos of stones into some semblance of order and filling the hole in the east cella.”⁶⁷ Some 102 blocks were measured, numbered, and put in place in two courses according to a tentative reconstruction. We have no reliable drawings or photographs of the base before it was dismantled. Therefore, the way the stones of the platform have been drawn and placed on the site does not represent with certainty how they originally were.

Hanfmann and other scholars settled on a date around 220–190 BC for the construction of the base from the evidence of the reused blocks.⁶⁸ This conclusion was questioned by P. Stinson, who noted that the unusually broad placement of the cella’s column foundation blocks around the central platform must signal the builders’ attempt to accommodate this preexisting structure, although the actual spacing of the columns above them would not have been affected. Compelled by this observation, N. Cahill opted for Butler’s pre-temple date for the base. He pointed out that the Hellenistic bronze and silver coins used to date it were found “in the vertical joints between the stones of the upper course . . . as if at the foot of the pedestal of the statue,” while the silver croeseid was between horizontal courses; it can thus be linked to the actual time of construction and is compatible with the ceramic and masonry stylistic evidence.⁶⁹

Historical and epigraphic evidence, often inconclusive and circumstantial, provides a broad contextual backdrop against which our primary interest and methodology—employing the construction details of the building as a dating tool—may be measured. Careful observation of the temple construction, including the details of its marble ashlar walls and the foundations of its walls and columns, has provided us with a set of more objective criteria for distinguishing what we see as two main construction phases.⁷⁰ These technical features evidently represent the work of two major periods and were used consistently in the construction of the temple. Though these features have been described previously, they will be summarized in this section as well. Some, but not all, of the observations in the following discussion are similar to those that Gruben published in 1961 and are supported for the most part by Howe. In certain details I offer new possibilities.⁷¹

Main Walls and Foundations

Significant details that clarify the construction processes of the Artemis temple appear consistently in the long north and south walls, the east wall, the original west wall (middle crosswall), and in the foundations of the twelve interior columns of the original cella. These include the use of bar clamps (with hooked ends), square dowels without channels (used as edge dowels below floor level and in combination with leverage holes at or above the euthynteria level), and a total absence of lewis holes (see Figs. 2.9, 2.15, 2.22, 2.60, 2.62). We may identify these characteristics with the temple’s original, Hellenistic construction, or first phase.

The Roman (second) phase construction is represented by the east crosswall, the west wall, and all of the foundations of the exterior columns of the peristyle, as well as the column foundations of the east and west porches (where preserved or visible). This construction is broadly identifiable by the use of butterfly clamps (without end hooks; “wing clamps” in some publications) for the courses below the floor, or the euthynteria level, but regular bar clamps (though of a larger size than those typical for Hellenistic construction) for courses at or above the euthynteria level (as observable in the case of the foundations of columns 9 and 14: see Figs. 2.114, 2.131). Square dowel holes with single or double pouring channels are common. Standard-type lewis holes (large in size and with double slanting sides) are universally used; very large blocks, such as some of the column drums and architraves, commonly use two or more lewises. Rarely are L-shaped, T-shaped, or cross-shaped lewises employed, although these shapes may exist due to the later reworking of standard lewises.

Below is a detailed description of these Hellenistic and Roman features.⁷²

Hellenistic Phase Features

Walls identified with the first phase of construction, as readily seen in the south and north walls and the original west wall (middle crosswall), are built of larger, more uniformly shaped blocks that are well joined and fitted and display a smooth finish. Surface cuttings or construction cuttings (clamp, dowel, and leverage holes) are employed sparingly; no lewises appear except in the Roman reuse of such blocks, such as the fluted drums. The courses below the euthynteria level (-1), as seen on the western end of the south wall, are so closely and finely fitted that they required no clamps or any other cuttings except for small, square dowels. Courses at or one course above the euthynteria level (0 and +1) are joined by a row of bar clamps (ca. 12–16 cm) applied along the outer edge of the wall; courses above this level (+2 and higher) have two rows of bar clamps along both the outer and inner edges of the wall; clamps are never used to join blocks laterally across the width of the wall.

The unclamped joints of foundation blocks of the courses below the euthynteria approximate the tightly fitted, quasi-polygonal technique and must have been conceived as a structural device to retard or offset differential settlements (as visible in the connection of the middle crosswall and the south wall; see Fig. 2.60).⁷³

The technical features (mainly clamps and dowels) of the north and south walls of the cella display a roughly consistent system of distribution patterns specific to courses: edge-type dowels are used in combination with a leverage hole, often located in a row ca. 0.20–0.25 m from the edge of the wall, and without the use of pour channels. In the case of the edge dowels, lead was poured in from the edge after the dowel was put in place, and then the next block was lowered and shifted into position with the help of a crowbar. All clamps are of the bar type, with ends anchored into the stone (see Figs. 2.20, 2.21, 2.22, 2.23, 2.24); they are small or medium in size (12–16 cm) and run in rows close to the outer or inner face of the wall, or both; there is no lateral or cross joining of the blocks (see Figs. 2.15, 2.16, 2.17). These construction features and details also characterize the individual column foundations inside the original cella and the west pronaos (see pp. 60-67).

Roman Phase Features

The blocks of the Roman (second) phase, especially the column foundations of the north and south peristyles and the west pronaos porch columns (column foundations 48, 53, 54, 55, and 49), are massive, coarsely shaped, and characterized by an excessive use of clamps. Clamps above foundation level are uniformly of the bar type, while the foundation courses below often display larger butterfly clamps without end depressions.

Blocks for the east crosswall and the west wall, both Roman additions, are not as tightly or carefully fitted as the earlier Hellenistic work; smaller stones fill the wall middle (for the east crosswall, see Figs. 2.46, 2.47, 2.48, 2.49; and for the west wall, Figs. 2.67, 2.68; see also Figs. 2.9, 2.10, 4.11). While in the Hellenistic work only the outer row of blocks of a course are joined by a carefully aligned row of clamps, in the later construction almost every block is joined to its neighbor by one clamp, and often two or three. Dowel holes, when used, are accompanied by one or two pouring channels at the corners. Three of the unfluted column drums display pairs of small, square dowels of the “dry dowel” type, typically with a bronze casing to receive a dowel that would have matched exactly and did not require the use of lead; it is likely that this method was more common than the preserved evidence suggests, considering that the metal would have been an easy target for robbers.

The use of standard, rectangular lewis holes is ubiquitous; even relatively small blocks bear at least one lewis hole, and larger ones have two or more. The lewis holes are conspicuously large, deep, and coarsely cut (averaging 10.0–12.0 × 4.5–7.0 × 7.0–10.0 cm for the crosswall blocks and 12.0–17.0 × 6.0–9.0 × 10.0–18.0 cm for the north and south peristyle column foundations).

The use of masses of mortared rubble (a local variant of opus caementicium), which thoroughly encases the marble block foundations of the exterior columns, is a patently Roman construction on the site (see pp. 68-73). Although there are instances of mortared-rubble construction dating to the first century AD in Asia Minor and at Sardis, such as the Wadi B temple (ca. mid-first century AD; see pp. 217-220), the system becomes widespread during the late first and second centuries, as exemplified by large projects at Sardis such as the Bath-Gymnasium Complex and the great end walls of the theater (the date of the latter may be late first or more likely second century AD; Fig. 3.6). Like these instances, the mortared rubble of the Artemis temple appears coarsely and unevenly laid, using a surplus of rubble stones and less lime mortar.

In some cases, the consistent use of these techniques, both Hellenistic and Roman, is compromised because a wall or a part of a wall has been rebuilt using blocks from an earlier construction. This is especially true for the west wall and the preserved northern end of the middle crosswall (see Figs. 2.54, 2.58, 2.59, 2.62). Only six meters from the end of the west anta piers, the former is of “mixed construction”: some blocks show predominantly Roman features, but there are a few that bear smaller bar clamps with hook ends, even below the foundation level; this represents reuse from the earlier phase. The west wall bonds to neither the south nor the north long wall, nor does it match their masonry coursing (observable only at its northern end). It is laid directly over the foundations of the pair of original column foundations of the pronaos (column foundations 79 and 80, not on Butler’s plans). We can imagine that the late west wall was built mainly from the stones of the demolished original west wall (now called the middle crosswall) of the cella, including its central door and staircase.

The western end of the north wall also shows mixed construction, with bar clamps used on courses above ground level and butterfly clamps below ground level. The foundations of the northwest anta pier (Hellenistic) and column 48 in front of it (Roman) are connected along a joint with three large butterfly clamps, while the blocks of the anta pier above grade have bar clamps. In the same area, there are no lewises on the side of the anta pier, but many are visible below and above grade on the column side (Fig. 2.107). Differences in construction techniques are particularly dramatic at the western end of the south wall and the southwest anta pier, where the abrupt change from Hellenistic to Roman construction (and the seam between the two) can be seen very clearly on the preserved surface of the wall, 7.50–7.60 m from the reconstructed face of the missing southwest anta pier, as well as along the south face of its foundations (see Figs. 2.9, 2.10).

The East Door

Structural and Stylistic Considerations for Dating

Butler believed that the temple was meant to face east and that the east door was its original door. Hanfmann and Frazer held that the east door was a late feature, belonging to a “late Hellenistic” phase (see Figs. 2.28, 2.29).⁷⁴ However, the evidence provided by the masonry construction below the threshold and around the door opening, along with the style of ornament of the well-preserved door jambs, indicates that the east door was cut during the Roman period when the cella was divided into two separate chambers (for a description of the construction below the threshold of the east door, see pp. 46-49). In Greek and Roman architecture, construction directly under door openings is typically and intentionally left “weaker” for reasons of economy because that area supports less weight. The threshold of the Artemis temple’s east door is supported by continuous marble-block construction no different from the foundations of the main walls of the cella, a clear indication that the east wall was originally solid, and that the east door was not an original feature of the Hellenistic building.⁷⁵

The two courses visible below the threshold were left largely rough, leading one to believe that they were intentionally unfinished because the original design included a door and stairs that would have masked this part of the lower wall (see Figs. 2.28, 2.36, 2.37, 2.38). Close examination, however, reveals that these courses below the threshold were partially finished; a pair of smooth, finely dressed blocks of the second course below the threshold continues some distance under the door on both sides (see Fig. 2.43). The blocks of the middle portion of this course (for a length of 5.14 m) display a coarse, quarry-faced surface with a smoothly drawn, 6–7 cm wide “drafted edge” along their top edge. The course above this is left entirely rough with no drafted edges at the joint. The rough areas of the course below the threshold thus actually represent work in progress. Their smoothed horizontal and vertical bands are not “drafted edges” of finished masonry, but guides for trimming the surface of the rest of the block—a standard detail indicating that work progressed in stages, moving from the wings to the center of the east wall. If the original builders had planned for a door here, there would have been no finished or half-finished blocks under the threshold. Why trim masonry that would have been invisible behind the massive stairs of the door?⁷⁶ It appears then that at the time the door was cut open, finishing work and detailing on the opisthodomos walls were only partially completed, moving from the flanks to the center—a normal procedure.

Further evidence for the later date of the door can be evinced by observing the pattern of the masonry courses on both sides of the door opening. The third and fourth courses (up from the porch floor) adjacent to the south jamb of the door are 0.74 m and 0.44 m long, respectively; the fourth course on the north side is 0.60 m long (see Figs. 2.30, 2.37). These are uncommonly short blocks in relation to the temple’s scale and the normal length of its ashlar blocks. In fact, they are the shortest three of the more than 356 blocks that are preserved in situ and visible on the cella walls (typical block lengths run 1.20–2.20 m). Furthermore, they appear anomalous to the regular long–short alternating rhythm typical of Hellenistic ashlar masonry in this temple and elsewhere. This anomaly is a strong indicator that the center of the east wall was taken down to its lowest course, and that some of its blocks were either cut shorter or completely replaced in order to accommodate a large door centered on the cella axis.

While I hope that these structural considerations are sufficient to show that the east door is a late addition, they do not necessarily prove that it is Roman; Hanfmann and Frazer believed that it was part of a late Hellenistic renovation.⁷⁷ However, internal construction evidence that points to a Roman date is provided by the very large lewis holes and channeled dowels carved on the upper surfaces of the door jamb blocks, which are typical construction features of the Roman period (see Figs. 2.35, 2.36). Furthermore, stylistic considerations of the jamb and cornice ornament of the door allow us to narrow down the dating to the Hadrianic period.

Ornament

With the exception of some capitals, decorated bases, and several pieces of the unornamented inside (pteroma-facing) architraves (one fully preserved), extant Roman ornament on the temple is restricted to the jambs and lintel of the east door and the enormous consoles flanking the lintel (see Figs. 3.7, 3.27).

The presence and prestige of the Hellenistic temple, especially its superbly modeled and carved Ionic capitals, as well as the delicate torus decorations and elegant profiling of the few remaining original bases, explain its close—and on the whole very successful—Roman imitations. We have no difficulty recognizing capital C as an original piece, but it is hard to determine solely on the basis of style whether others, such as capitals A and D, are actually Hellenistic originals, or competent Roman copies (capital A is Roman, and capital D Hellenistic; see pp. 121-125).

The new ornament created for the east door—which might have been inspired by the ornament of the original west door, then moved on to the new Roman west wall—displays the persistent influence of specific Hellenistic models upon the common mid-second-century, Roman ornamental repertory of Asia Minor. In fact, the distinctive Hellenistic style or manner of the east door ornament may find signification and explanation through its links to the well-known classical qualities of Hadrianic ornament, though it also shows differences (discussed below). Compared to the refinement of the temple’s east door ornament, datable Hadrianic ornament from other sites often appears schematic and simplified.⁷⁸ It is a pity that no Roman ornament of the exterior order survives to further illustrate this process of imitation and amalgamation.

Like an architrave would be, the jamb and lintel of the east door are divided into triple fasciae, separated by bead-and-reel and cyma reversa (Figs. 3.7, 3.8, 3.9). The crown molding is composed of a bead-and-reel and an egg-and-dart ovolo (pointed darts), surmounted by a strongly projecting cavetto decorated with open and closed palmettes alternating with lotus flowers. The cornice is composed of a small, plain cyma reversa crown and a plain corona projecting over a row of bold dentils, now partially preserved. No frieze has been found, though Butler restored one for reasons of completeness and proportion. The fine quality of the carving; the round, globular beads and their sharply outlined, disc-shaped reels; the well-formed eggs deeply set in their smoothly curving casings; and the overall superb modeling and plasticity led Butler to date this ornament to the fourth century BC, which was part and parcel of his efforts to see the building as a creation of the Classical era.⁷⁹ Butler’s aesthetic assessment is understandable, yet the overall appearance of the jamb ornament is distinctively Roman

Compared to the eggs with slightly pointed ends encased in tightly fitting, narrow shells and delicate darts, all shallowly modeled, as is typical of late Hellenistic ornament of Asia Minor (as in the Hermogenes-designed Temple of Artemis at Magnesia on the Meander [Fig. 3.10] and the architraves of the Belevi Tomb near Ephesus), the egg-and-dart ornament at Sardis is deeply carved and strongly modeled, with more rounded eggs set in widely splayed and fully formed shells (Fig. 3.9). Although it displays greater stylistic affinity to the architrave ornament from the Smintheion at Chryse (Gülpınar), as compared to the thin precision and linear delicacy of its egg-and-dart crown (Fig. 3.12), the Sardis ornament has greater plasticity.⁸⁰ Interestingly, the regular, well-rounded eggs and the bead-and-reel row of the architraves of the Hermogenean Temple of Dionysus at Teos appear closer to Sardis: this may be explained by the Hadrianic date that can be assigned to a part of the temple’s entablature (Fig. 3.11). At Sardis the careful triple alignment of the palmettes, eggs, and the beads of the crown molding of the jamb, especially the superbly modeled, almost lacy palmette-and-scroll row of the top cavetto, may follow Hellenistic precedents, but the fullness of these appears more distinctive than typical Hellenistic examples, where the decorative elements are subtly, but somewhat weakly, merged into a whole (Figs. 3.8, 3.9; cf. Fig. 3.12).

Likewise, the rich leaf pattern of the open-and-closed palmettes appears sumptuous on the door jamb cavettos, setting them apart from their lean, linear, and delicate Hellenistic counterparts. The Lesbian cyma of the door, with its broadly splayed leaves, hardly appears in the Anatolian repertory before well into the second century AD.⁸¹ The same individualistic quality holds when the door ornament is compared to some early or mid-Imperial Roman examples. At Sagalassus, the architraves from the Julio-Claudian lower agora gate and the egg-and-dart of the Ionic capitals from the Temple of Apollo Clarios (whose different phases are variably dated from the late first to the early second century AD) display the same delicate but shallow carving style, especially their flattened eggs with pointed ends; this is noticeably different from the precision of the Artemis temple’s east door ornament, whose aesthetic impact is enhanced by the changing play of light on its deeply recessed, three-dimensional surfaces (Fig. 3.13).⁸²

Apart from these distant late Hellenistic/early Roman comparisons, a closer and more immediate model for the robust plasticity of the east door ornament comes from the original Hellenistic anta capital of the temple, especially when one compares the egg-and dart of the door jamb with the full, rounded, and articulated crowning egg-and-tongue of the anta capital cornice, now preserved on the southeast anta pier; this comparison perhaps illustrates the elusive concept of “Hadrianic classicism” in the plastic arts (cf. Figs. 2.293, 3.8).

A detail particular to Sardis is the elegantly formed lotus flowers that alternate with the open-and-closed palmettes of the cavetto, an elaborate combination also seen on the cavetto molding of the cella door of the so-called Serapeum at Ephesus, which is dated to the Hadrianic period (cf. Fig. 3.8 and Fig. 3.14). Considering that a special and unusual variation of this so-called lotiform motif was found in the temple,⁸³ one wonders if this strange and elegant motif was not a local preference plucked from among the rich Anatolian-Hellenistic repertory of architectural ornament in Asia Minor.⁸⁴

Taken as a whole with its strongly modeled, deeply carved, carefully aligned elements, the ornament of the east door at Sardis represents a fine balance between local style and regional—even international—conventions, as practiced by the original artists and architects of the temple, a backward glance at the delicacy and precision of Anatolia’s Hellenistic ornament mixed with or overlaid by the newly acquired, robustly articulated assertiveness and formality of the imperial presence at Sardis. These qualities of creative classical revival and renovation are commonly associated with Hadrianic architecture in general, but here we also see the inescapable influence of the temple’s Hellenistic ornament.

While it is difficult or impossible to be certain of precise dates in matters of decorative style (who can tell the difference between a late Hadrianic and an early Antonine work?), an early or middle Hadrianic date (AD 120–30) for the Sardis jamb ornament would be in keeping with the eclectic classicism typical of the Hadrianic style in Asia Minor, especially in its readiness to combine this versatile classicism with local and foreign motifs.⁸⁵ Comparing particular stylistic motifs of the Sardis east door to some of the contemporary and comparative examples from Ephesus, Nysa, Sagalassus, Miletus, and Didyma (such as the later architraves of the Temple of Apollo), and following the general development of the ornament of the Imperial era in Asia Minor, S. Pülz, F. Rumscheid, L. Vandeput, M. Kadıoğlu, O. Gülbay, and this author are among the number of scholars who firmly opt for a Hadrianic date for the Sardis door ornament—and hence, for the presence of Roman renovations at the temple.⁸⁶

It is important to note that although the east door’s ornament follows the general style and order commonly ascribed to Hadrianic ornament, such as the distinctive globular beads and the careful alignment of certain decorative motifs, few of the fairly numerous and well-documented examples of Hadrianic ornament from Asia Minor display close similarities.⁸⁷ This may underscore the special position of the Roman temple at Sardis as a deliberate imitation of its “highly visible” and prestigious Hellenistic predecessor, as mentioned earlier.⁸⁸

Common comparisons could be made to the architraves and the door jambs from the scaene frons of the theater at Nysa (Fig. 3.15); the “Antonine” temple at Sagalassus (late Hadrianic to early Antonine, Fig. 3.16); the Temple of Hadrian at Ephesus, with its distinct cavetto lotiform motifs (Fig. 3.17); and the Library of Celsus from the same city (Fig. 3.18).⁸⁹ Details of the ornament from the Hadrianic Baths at Aphrodisias (Fig. 3.19), the “Small (Hadrianic) Temple” at Cremna (mainly its door jambs), and the cornice from the late Hadrianic Nymphaeum F3 from Perge (Fig. 3.20) provide basic similarities that help underline the Hadrianic origin of the Sardis temple doors.

However, many are characterized by coarser, stenciled, and mechanical features typical of good, provincial mass-produced work, in contrast to the refinement we see at Sardis.⁹⁰ In these and other comparative examples the general carving appears flatter, even cruder: eggs are slightly elongated and not fully rounded, their shells not integrated into egg form; darts are sometimes shown as stubby and harshly shaped “arrows”; reels are elongated and simplified (some on their way to the long and “faceted” reels typical of Severan ornament); and the palmette row of the cavetto is flat and mechanical, generally a far cry from the luscious naturalism of Sardis cavetto palmettes. Even the fine and vigorously carved Hadrianic ornament (with eggs loose in over-widely splayed casings), from the entablature of a remarkable but lesser-known, Polygonal Building (a twelve-sided, domed hall) in Alexandria Troas appears somewhat mechanical when compared to the Sardian door jambs (Figs. 3.21, 3.22).⁹¹

Another cogent illustration demonstrating the considerable variety among Hadrianic ornament, from different sites in different regions under different circumstances, can be evoked by viewing the Sardis door ornament against the late Hadrianic ornament of the lower agora nymphaeum at Saga-lassus (Figs. 3.23, 3.24).⁹² In their consideration of this extensive, two-storied, three-sided aedicular ensemble enclosing a pool, Waelkens, Üner, and Richard observed that “none of [its] various types of ornaments . . . was uniform throughout the entire structure, nor did they reflect the hand of a single stonemason or a team.”⁹³ Some of the inconsistencies in style, quality, and finish of this ornament (including the ubiquitous three-fascia architrave, scroll acanthus frieze, and cornice with palmette-and-anthemion sima) can be explained by the haste with which this large project was undertaken. Yet a comparison of even the most relevant ornamental elements—from its architrave to the door jambs—with the Sardis door reveals a blatant lack of refinement in the former.⁹⁴ Often the quality and individuality of detail in ornament and construction is determined less by a period style broadly defined and more by the local narrative behind it.

The importance of individual distinction within a period style is also shown by the huge, handsome consoles with volutes (scrolls) that supported the cornice above the east door of the temple, which harken back to its Hellenistic roots (Figs. 3.27, 3.28, 3.29). The deeply carved, concave volutes with curling corner palmettes appear to have been deliberately fashioned in overall form and individual detail on the model of the famous Hellenistic capitals of the temple. However, with the oversized palmettes decorating their faces, their powerful, striking presence is more appropriate to mid-second-century imperial work. One of these consoles (now incongruously “composed” in a mock-up on site, in front of the door with other elements from it) can be compared to consoles from the door of the so-called Temple of Serapis at Ephesus, dating to the late Hadrianic/early Antonine period (Fig. 3.30). Although the similarity is striking at first glance, it is instructive to consider the simplified, mechanical, linear style of the Serapis console, with its drilled acanthus filling the scroll, in comparison with the softer, sculpted elements of the Sardis temple.⁹⁵

The peculiar ornamental style of the Roman work at the Temple of Artemis can be confidently identified as early Hadrianic. It seems to be a unique reflection of the complex and formative influence of this landmark monument’s own Hellenistic and classicizing past; its site-specific and compelling artistic tradition acts as a powerful magnet, creatively disturbing and reshaping a common style into an uncommon local amalgam.

The materials and construction techniques of the peristyle foundations also illuminate the building history of the temple. All foundations, walls, and individual foundations of the cella interior and surrounding peristyles have been described in previous sections in greater detail; here I will discuss them in the light of their structural characteristics that help to define and date the building (see pp. 68-73.

The main walls of the cella have continuous foundations in blocks of marble (see Figs. 2.76, 3.31), and the interior columns have isolated block foundations (see Figs. 2.28, 2.78). The courses of these wall and interior-column foundations step out as they go deeper. The exterior peristyle columns are also supported on individual piers of large, ashlar blocks of marble or limestone without courses that step out. Unlike the foundations of the interior columns, they are also not fully independent; instead, they are structurally connected to each other by a packing of heavy mortared-rubble, or in some cases they are encased in this material (see Figs. 2.105, 2.111, 3.33).⁹⁶ The rubble construction irregularly projects beyond the outer line of columns for a distance of ca. 0.60 m or more, probably marking the outer edge of the crepis if there was any. The block foundations of the east and what remains of the west end columns are encased on three or four sides by mortared-rubble construction and linked to each other.

Although isolated block foundations for the exterior columns of temples, stoas, and other monumental structures occasionally occur, the norm for colonnades in Greek architecture is continuous block foundations comparable to those of our cella walls.⁹⁷ The structural effectiveness of the system established as the norm in Greek and Hellenistic colonnades is underscored by F. Cooper: “[A] spread footing, running underneath the peristyle and common nearly to all Greek designs [my emphasis]” distributes concentrated loads to inhibit differential settlements, “with the additional advantage of transmitting the concentrated loads at the intercolumniations over as wide an area as possible.”⁹⁸ Variations exist, as dictated by terrain, soil conditions, and most importantly the economy. An effective alternative to the continuous-block foundation system is a “foundation grid” with short crosswalls,⁹⁹ linking the individual foundations of the exterior columns to the continuous foundations of the cella walls, creating a “waffle” system. The spaces between the columns and the crosswalls (inside the waffle) are filled with earth, gravel, or stone chips from the marble chantier, or dry, compacted rubble stone; unlike the Roman system, this is not structural, mortared-rubble work.¹⁰⁰ An excellent example of the waffle system is the Temple of Apollo Smintheus in Chryse (Gülpınar) in the Troad, dating to the mid-second century BC (Fig. 3.34), or the Temple of Apollo at Didyma, of 300–250 BC.¹⁰¹ Other variations of the system can be found in the Temple of Apollo at Claros and the Temple of Artemis at Magnesia on the Meander.¹⁰² Another elaborate example of this Hellenistic foundation system is the platform built under the Altar of Zeus on the Acropolis of Pergamon. This grid is composed of a parallel series of primary walls 1.60 m apart and connected by secondary crosswalls at 3.0–4.0 m intervals.¹⁰³

Structural use of mortared rubble alone, or in connection with ashlar masonry as seen in the exterior column foundations at Sardis, is uncommon before the Imperial era. Analysis of this construction at Sardis reveals that it is highly unlikely that mortared-rubble construction was introduced later as a separate process after the earthquake of AD 17 in order to fortify the existing Hellenistic block foundations (as suggested by Butler and Hanfmann), but rather it belongs to the same operation and phase as the building of these foundations and columns, as correctly proposed by Gruben.¹⁰⁴ The foundations of the exterior columns of the peripteros and the pronaos porches at Sardis represent a unified, contiguous system where the isolated, marble-block foundations of the columns are enmeshed with the structural mass of the lime-mortared-rubble construction; sometimes, as shown by the partially exposed foundations of columns 16 and 17, the block foundations are encircled by a rough, mortared-rubble “wall” with a mortared-rubble subfloor pad or platform extending around them, a construction type that seems to be limited to the east pronaos porch. Such mixed systems—used for foundations or upper structures of buildings—find no parallels in the pre-Roman buildings of Asia Minor.¹⁰⁵ However, they are quite typical of Roman construction in the West and stem from the basic structural logic behind opus caementicium and its numerous local variants. This method had already been used at Sardis for the foundations of the Wadi B temple, an early Imperial pseudodipteros (see Fig. 3.72).¹⁰⁶ Another contemporary and closely relevant example for the patently Roman, eminently practical nature of this mixed-foundation system of alternating strong support (under columns or piers) and weak support (in the spaces between) are the foundations of the Temple of Zeus at Euromos, a peripteros dated to the mid-second century AD (Fig. 3.35).¹⁰⁷ A similar system is also well illustrated in the foundations of the Temple of Deified Hadrian in Rome (Fig. 3.36).

Basing his judgment on an analysis of foundation conditions, Gruben hypothesized that the decision to continue the construction of the temple as a pseudodipteros belonged to what he termed the second or “Pergamene” phase (190–150 BC), when Hermogenes’s influence in the design and reorganization of Ionic temples in Asia Minor must have been impossible to ignore (see pp. 155-157 above). He believed that all of the foundations of the east end columns, i.e., the eight columns of the east facade (numbers 1–8 and 18), and the six columns of the pronaos porch (numbers 10–13, 16, and 17) were completed during this phase, although he admitted that the columns themselves were erected during the third, Roman Imperial phase of the building. Surprisingly, he ignored the distinctly Roman character of the heavy mortared-rubble work that he knew was integral to the foundations. He supported this theory by claiming that the construction details of these foundation blocks displayed his Technik I (the Hellenistic technique), characterized in particular by the use of bar rather than butterfly clamps. Since the three observable top foundation blocks unoccupied by columns (numbers 9, 14, and 15) indeed employ bar clamps to join the blocks of their top courses (course 0), his reasoning appeared justified (Figs. 2.114, 2.131). There is, however, an oversight in Gruben’s assessment of the difference between the employment of bar and butterfly clamps: bar clamps, which he identified exclusively with Hellenistic construction, are also used in Roman work for courses at or above the ground level, such as the top, or euthynteria course of column foundations. Larger butterfly clamps are restricted to courses below the ground level in Roman construction, as seen on the west end of the north wall.¹⁰⁸ Although the dowel holes are often covered on the bases by the columns that stand upon them, careful observation, especially where the edges of the bases are broken, reveals that the base plinths of most of the eight east peristyle columns have small square dowels with channels, four on each base, located diagonally on four corners.

Furthermore, the overall quality and type of the foundations of the east front and pronaos porch columns display a more-or-less uniformly coarse style and quality of masonry and a rough finish that is distinct from the carefully finished foundation work of the Hellenistic elements. The projecting top edges of the column foundation blocks in the east pronaos porch are irregularly and crudely shaped, unfinished or cursorily trimmed in pointed chisel; hastily done, they have uneven and raised surfaces (see especially the foundation of column 16, see Fig. 2.172). This kind of coarse finish (or no finish at all, as they were probably supposed to be trimmed and covered by the finished paving of the porch) is very different from the evenly shaped, smooth euthynteria of the anta piers (Fig. 2.74). Likewise, the setting and centering lines for the placement of column-base plinths are coarsely marked, unlike the precise “crosshairs” marking the corners of the southeast and northeast anta piers.

Mortared-Rubble Subfloor of the East Pronaos Porch

Investigations in 2002 showed that the floor between pedestal columns 11 and 12 (the middle columns of the pronaos porch) is filled in with mortared rubble, extending irregularly for 0.60–1.80 m beyond their west face and connecting the block foundations of these columns like a continuous concrete pad (see Figs. 2.112, 2.175; Plan 6; Plate 1, Plate 8-13).¹⁰⁹ The irregular west edge of the rubblework appears original, as it is laid directly against Hellenistic fill without any later cuts. In 2012, we explored a larger area bounded by columns 3, 4, 11 and 10 (a rectangle measuring 3.90 × 4.20 m, touching their corners) that revealed a concrete pad subfloor at least ca. 0.60–0.70 m thick, but the rubblework went deeper (see Fig. 2.113; Plate 8-13). The trench revealed similar construction in hard-packed rubble mixed with some reused fragments of brick and tile packed in lime mortar. The excavation also produced significant amounts of marble chips and fragments of fluted marble columns, suggesting that the Hellenistic columns were being broken up (and refashioned as new columns on pedestals) when the mortared-rubble floor covering this area was laid, presumably as a part of the second-century Roman construction of the temple. These limited excavations, aided by visual observation of the ground composition in hard-packed rubblework, confirm that the “corridor” between the block foundations of the east colonnade columns and the pronaos porch are linked by a massive pad constructed of mortared-rubble fill, probably reaching close to the depth of the ashlar-block foundations and laid together (as demonstrated by the rubble casing around columns 17 and 16; note also the mortared-rubble construction around columns 11 and 12). Only the north and south pteromas, and the large, open, almost-square area defined by the six columns of the pronaos porch, were simply filled with a surface layer of compacted soil, rich in marble chips (0.15–0.20 m deep). This mortared-rubble pad and the compacted soil and chips of the pteromas were probably paved over in plaques of marble, now gone or never quite started.¹¹⁰

In sum, we believe that the structural and technical considerations of the column foundations of the east end and east porch, and those of the north and south peristyles, indicate that all peripteral columns belong to the Roman era, and this was essentially the same operation responsible for the division of the cella, the opening of the east door, and the other interior cella adjustments concomitant to the new design. Naturally, the process and completion of different parts must have been spread over time, and some were never finished.¹¹¹

In order to address some of the questions concerning the ultimate transition of the Temple of Artemis from a Hellenistic to a Roman one, we need to explore the intentions behind the design of the building and its realization. A number of design elements from the earlier structure were reused in the later, either unaltered or with some changes. Foremost among these elements that appear to have played an important role in the new temple are the two columns raised on pedestals constructed in rusticated masonry, occupying the middle positions of the east pronaos porch (columns 11 and 12). Based on the discovery of many fragments of similar pedestals on the west end of the temple, the markings on their foundations and bases (columns 53 and 54), and the possible representation of these as partially preserved columns in the drawings of early travelers, we can conclude that similar columns on pedestals rose in the west pronaos porch of the temple (see pp. 23-24; pp. 60-62). These unusual supports were significant Roman period constructions that used elements of the Hellenistic columns, probably from the cella interior. Another key element is the hypothetical pair of columns between the east and west anta piers that must have existed during the Hellenistic and some of the Roman phases of the temple. Additional important evidence in the consideration of the Roman era design implications of this area, as well as the overall chronology, comes from a large area excavated in 2011 (and a smaller area in 1996) between the southeast and northeast antae, and its narrow extension between the latter and column 16 (see pp. 180-181 and pp. 184-190 below).

Pedestal Columns 11 and 12

The tall pedestals of the east porch support the only fully finished columns in the temple (numbers 11 and 12; Figs. 2.178, 2.196). They probably carried reused Ionic capitals of the smaller group, such as capitals C, D, or G. These smaller shafts and their capitals must have come from the redesigned Roman cella when a new “west cella” was created and its original columns were decommissioned (see pp. 203-205 above). The rebuilt pedestal columns of the porches can be reconstructed at 15.68–15.70 m high, which is ca. 2.16–2.19 m shorter than the height of the peristyle columns. The well-preserved eastern pair, occupying the middle two positions of the pronaos porch, represents a distinctive design within the vocabulary of classical columns (Fig. 2.181).

Flanking and emphasizing the main axis of the temple and its monumental doorway, these uprights inspired Butler to recall the “Nearer Asia[n] and biblical” notions of monumental columns that signify “the entrance of a holy or particularly important place.”¹¹² An idea similar to this but broader in implication has been rekindled by Howe, who wondered if these “enigmatic” columns in Roman guise may have evoked the ideas of triumph and triumphal procession, appropriate for a temple that accommodated the imperial cult.¹¹³ At the same time, we also note the Roman predilection for strong, axial schemes and suggest that the creation and deliberate placement of these unusual columns on the long axis of the building strengthens this axis, affirming Roman architectural traditions.

According to Butler, the outwardly tapered, roughly hewn, unfinished tops of the pedestals were intended for sculpture, comparable to the sculptured bottom drums (columnae caelata) of the Archaic and Classical Artemisia at Ephesus; the strongly projecting upper parts could accommodate the shoulders and torsos of the hypothetical figures. This idea found considerable support from other scholars, though we find the comparison imprecise. Columnae caelata denote the carving of the column shaft (well represented in the Ephesian examples) rather than a separate, carved pedestal base. Butler, however, was right to declare that they were “so unusual, so foreign to anything known in the whole range of Hellenic architecture . . . [that] we may well ask whether they were not possibly taken over from purely Lydian sources.”¹¹⁴ My own search of classical and eastern sources was no more successful in identifying clear models for this unusual architectural element, other than the generic Roman use of tall pedestals, such as the highly classicizing pedestal of the Column of Trajan in Rome. The unorthodox, even bizarre power of this motif as used at Sardis may not find ready parallels in the classical world (let alone the Lydian world), but as an appropriate marker of monumental axes and memorable and mythical landscapes, it was echoed in the architecture of later periods. Here we might cite the handsome sculptured pedestal bases for the double columns marking the main axial entrance of the Grand Palais in Paris (1897–1900; Fig. 3.37);¹¹⁵ and François de Nomé’s enticing seventeenth-century architectural capriccio Samuel Anointing Saul, a strange extraurban scene containing a triumphal arch, a stepped tomb monument, and a monumental column on a tall, rustic pedestal with carved figures (Fig. 3.38; compare to Fig. 3.40).¹¹⁶

Notwithstanding the undeniably exotic and powerful appearance of these top-heavy rustics, the reality of the origins and intended design of the Sardian examples may be more ordinary. These pedestal columns were almost wholly reconstituted during the Roman phase of the temple from parts and elements taken from the dismantled Hellenistic columns of what became the west cella (see above); the job was practical and practically done. Careful study of the fluted shafts reveals that there are minute discrepancies in the fluting (fillets and flutes often do not match by margins of 1–4 mm); therefore, the shafts must have been rebuilt from the drums of existing columns. Such discrepancies in fluting would neither have been tolerated nor possible if the flutes were newly carved on their shafts in situ, as was the custom. This view is supported by the large number of damages and repairs seen on the delicate, finished fluting of the columns; some of these were probably incurred during their transportation and reconstitution.

Five or six blocks recut and reshaped from fluted drums, similar to those used for the pedestals, were found at the west and southwest end of the temple. In the same area, there are also some fourteen or fifteen fluted column drums in a good state of preservation. As shown by Ken Frazer, each pedestal is composed of eight blocks, each face of the pedestal has three blocks with tightly fitting joints between them (with anathyrosis). The joints recede inward considerably from the rough, projecting face of the blocks. The unfinished, outward-tapering blocks of the pedestal tops were entirely recut from fluted drums of columns no longer needed in the new Roman redesign, probably the columns freed in the rebuilding of the west cella and the west pronaos porch; these recut flutes can still be seen on the pedestal (Fig. 3.39). Using these elements, Frazer reconstructed in 1993 a sketch of a full pedestal base and a column shaft surmounted by the half-preserved capital G (found on the west end of the temple) as column 53 of the west porch (Fig. 2.270).¹¹⁷ There are enough blocks and fluted drums in the field on the west side of the temple to recompose one or even two columns on pedestals.¹¹⁸

Even if the unfinished blocks were intended for sculpture, as Butler had maintained, their Roman Imperial date means that they need not have been for archaizing reliefs like the sculptured drums of the Ephesian Artemision. Rather, they could have been straightforward Roman representations in low relief, such as the assemblage of trophies decorating the huge pedestal of the Column of Trajan in Rome, or as another possibility, the classical figures of muses, allegories, racial types, or captives found on the column pedestals of Roman temples and triumphal arches.¹¹⁹ However, that is admittedly a long shot; a plainer solution is to be preferred. Although the rough top blocks of the pedestal appear to taper strongly outward, they do not project more than just 8–10 cm beyond the finished surface of the plinth course below them (when measured from the recessions of their joints) (Fig. 3.40). The intent could have been to trim down the rough blocks to an even surface, making an ordinary, modestly projecting crown molding. In either case, as Butler noted, the pedestal columns of Sardis are uncommon and eye-catching elements, inside or outside the repertory of classical architecture. Likewise, they might not be reproductions “taken from purely Lydian sources,” even though there are Lydian letters inscribed on the bottom of the shaft of column 12, which probably belonged to the column in its original, conventional role in the cella (Fig. 3.41).¹²⁰ Instead, they appear to be eclectic and creative examples of Roman bricolage, achieved by recycling some of the older and more precious elements of the Hellenistic temple, drawing attention to and accentuating the central axes of the projecting porches.

Considerations for Columns In Antis and a Pronaos Porch Open to the Sky

According to Gruben, a pair of columns with high, rusticated pedestals originally stood between the anta piers of the east and west porches (Fig. 3.2, upper plan). Gruben’s hypothesis for the existence of these in antis columns is an architecturally pleasing and structurally logical solution; these columns provided a viable cella design even when the peristyle was missing. Indeed it would be difficult to imagine how the short ends of the Hellenistic cella, presumably with its gable or pediment, would be possible without columns in these positions. Yet in terms of chronology and several other important aspects, his theory proved untenable.¹²¹

The design history of these in antis columns, now missing, is important. The depth of the pronaos porches (created by the removal and absence of these columns) strengthens the theory that they were defined by six columns (enclosing an area of ca. 17 × 13.40 m) and may have been open to the sky (Fig. 3.2, middle plan). This was the solution proposed by Butler and supported by a few other scholars mainly on structural grounds, but this theory also leads to significant architectural and spatial consequences.¹²²

Excavations in 1972 and 1996 inside these porches failed to reveal any material identifiable as foundations for these hypothetical in antis columns.¹²³ T. Asena’s 1996 trench inside the southern part of the east porch yielded large quantities of marble architectural elements from the temple: fragments of shafts, capitals, ornament, roof tiles, and the lower half of a colossal bearded head identified as the emperor Commodus (Figs. 3.60, 3.61). The pottery was predominantly Late Roman, but this mixed with some earlier material including some Lydian wares. Asena concluded that the area had been excavated in late Roman times, no earlier than the late fourth or early fifth century AD, and that this was when the imperial portrait was buried.¹²⁴ In order to clarify these speculations unsolved by the limited trench of 1996, a larger area extending between the northeast and southeast antae of the east porch (and its extension between the northeast anta pier and column 16) was excavated in 2011 (Plan 6; Fig. 3.42). This new archaeological effort, supervised by F. Can and N. Cahill, brought forward new information and new proposals suggesting major changes in the building history of the temple and its phases, mainly based on ceramic evidence. In the following discussion, our goal is to represent these views and clarify where we consider alternative readings of the evidence.

The Results of the East Porch Trench of 2011

The east porch trench of 2011 produced the first reliable evidence that the Hellenistic temple originally had columns between the northeast and southeast antae: cuttings datable to the Hellenistic period in appropriate spots in the bedrock, impressions of the stepped foundations in the mortared rubble bedding of the Roman stairs, and a large sandstone block that could have belonged to the foundations of the southern in antis column.¹²⁵ Based on the stratigraphical evidence, Can and Cahill concluded that the only digging and filling activity in this area belonged to the fourth or early fifth century AD, resulting in the removal of the in antis columns.¹²⁶ The dimensions and the location of cuttings in the trench (ca. 4.0 × 4.80 m) are appropriate for the block foundations of these columns in antis, datable to the Hellenistic period. Their spacing, reconstructed at ca. 8.35–8.40 m, matches the axial distance of the columns located between the anta walls of the west porch (numbers 77–80) but not the cella interior columns at ca. 9.40 m.¹²⁷

Additional findings from Can and Cahill’s excavations could be helpful in determining when these columns were removed. The east face of the mortared-rubble foundations of the Roman stairs retains impressions of the upper two courses of the stepped marble foundations of the Hellenistic column in antis. This observation led to the hypothesis that “after the columns were removed, the stair had sagged into the resulting void” (Figs. 3.43, 3.44, 3.44a).¹²⁸ Consequently, the removal of the supporting column foundations, which resulted in the “sagging,” must have happened some time during the Late Roman period because the pottery evidence from these trenches, with bedrock cuttings, dates predominantly to the fourth and fifth centuries AD, although it also includes Hellenistic sherds and “a significant amount of earlier residuals.”¹²⁹ Hence the in antis columns were left in place during and throughout the building’s mid-second century phase when the east door was opened and the stairs built in front of them; they were not removed until late antiquity, when the mutilated head of Commodus was dumped. In addition, these trenches of 1996 and 2011 also produced evidence that lead to the conclusion that these columns in antis were being dismantled and broken up in Late Roman times.¹³⁰

The fact that these foundation blocks must have been removed at some point in the history of the building is not questioned (though the column foundations could have remained in the ground until their final removal, while their columns were removed at an earlier date, as discussed below). Still, other questions and possibilities remain, and the “sagging” of the south end of the stair foundation is a subject for reconsideration.

First, it is difficult to declare that this sagging was unequivocally and solely the result of failing static conditions after the removal of a support. It could have happened due to natural causes, such as earth settlement or the earthquakes common to the region; there is considerable and uneven sagging in the foundations of the temple walls, as recorded in measurements for curvature; the partially measurable curvature of the east wall at euthynteria level (the center 9 m stretch could not be measured), however, revealed no distortions. Contrary to common opinion, the clay deposits upon which the east end of the temple was founded would have offered no greater stability than the gravel and sand under the western part of the temple; geologically and physically, clay is even less resistant than gravel and sand to seasonal inundations or just wetting; it is also not a stable support for shocks and movements caused by settlements and earthquakes.¹³¹ In fact, softer clays—the kind in which our temple’s foundations would have been laid—are liable to swelling and slippage.¹³²

Secondly and more importantly, hardened, mortared rubble is well known for its capacity to retain its structural form and function even after its immediate support is removed, especially when carrying a relatively minor load; mortared rubble does not “sag.” When supporting material is removed, such as the marble column foundations against and over which it was built, mortared rubble cantilevers remarkably well, and if the external forces are large enough, it cracks and eventually collapses.¹³³ In this case, the “unsupported,” “moment arm” of the undermined mortared rubble was so small that any sagging, slumping, or cracking would have been highly unlikely; furthermore, any stones taken from under the mortared rubble would have been immediately replaced by earth.

The contextual, architectural indicators for the removal of these in antis columns in the fourth–fifth century AD, rather than in the second century as a part of the larger Roman redesign of the temple, is credible but not conclusive. There is no doubt that the last large-scale pits dug in this area were in the late fourth or fifth century, filling the space where the column foundations had been removed and containing late antique material that formed “a single fill dumped from various points”; presumably this was when the marbles of the temple were broken up and carried away.¹³⁴ It is true that the pottery from the area is predominantly late antique (also containing some Hellenistic and Lydian residual material), and the rich yield of marble architectural pieces and ornament could point to a Late Roman break up of the in antis columns, but there is no absolute proof that these dozens of small, broken pieces should all belong to the two in antis columns. They could have just as easily come from other unfluted or fluted shafts of the other east porch columns, or from the fluted columns of the cella interior.¹³⁵

A third possibility bridges these divergent interpretations: the columns may have been removed during the second-century operation, but their foundations were left in place until they, too, were removed sometime in the late fourth or fifth century. In fact, even if the Roman architects had their eyes on removing the columns (which also would have allowed better access to the new stairs), retaining the foundations and filling the tight areas between the stairs and these blocks in rubble and earth would have been the easier, structurally preferable thing to do.

There is, however, one more cogent and convincing reason to support the hypothesis that the original in antis columns of the temple were not removed during the second-century Roman rebuilding but were kept instead. As an architect and historian, I am ready to say that it is simply structural logic. To remove perfectly good, functional columns that supported an architrave across the antae would have increased the minimum span of the pronaos porch columns from 7.40–7.60 m to 13.40–13.60 m—from a distance that is easy for wooden beams or trusses to one that invites difficulty. As mentioned before, this was why Butler had first proposed an open roof (see p. 181, note 122). Keeping these columns where they were would have been the logical thing to do for our Roman architect or engineer. It would also result in a structurally and architecturally conventional temple design. Therefore, the alternative hypothesis that argues for a pronaos porch open to the sky carries with it a blatant desire to explore an unconventional design in the face of structural improbability. Removing the columns was an architectural option predicated on deliberate architectural and spatial considerations, as well as a presumed preference on the part of the Roman architect for an unorthodox design—something we cannot prove.

By definition, a reconstruction must take some risks. As one scholar of antiquity put it, “there is a delicate balance in taking . . . a few steps beyond the bounds of knowledge . . . and in completing the unknown according to culturally (and architecturally) accepted rules.”¹³⁶ According to the sixteenth-century French philosopher and essayist Michel de Montaigne, a reconstruction of the ancient city of Rome was made possible by piecing together “pictures”—a mosaic of physical evidence composed of real or derived images and impressions, all with their own restrictions.¹³⁷ Walking through our temple and considering the architectonic possibilities of an open pronaos porch—against practical and conventional sense—allows such a picture to emerge, one, for all its limitations, that is too valuable to ignore; as mentioned earlier, this is an argument which should have its day in court. Thus, we wish to retain this picture, contrary not to any hard evidence that disproves it, but to the conventional thinking that discourages it.¹³⁸

Consideration of the Evidence from the Area between the Northeast Anta Pier and Column 16

The excavations of 2011 inside the east porch produced further information and dating proposals for the first major Roman rebuilding activity of the temple. As an extension of the larger trench between the eastern antae piers, this new area (AT 11.1) was small, located between the northeast anta pier and the independent block foundations of column 16 in front of it (ca. 1.20 × 2.10 m on top, narrowing to ca. 0.30 × 2.0 m at bottom; maximum depth ca. 1.0 m, with a volume of ca. 0.90 m3; Fig. 3.45). Only the top 0.30–0.50 m of the west face of the mortared rubble encasing column 16 was exposed (Figs. 3.46, 3.47). Below a top layer of soil and sand (ca. 0.20 m), the trench was filled with an indiscriminate mixture of building debris, rectangular and round bricks, roof tiles, terracotta pipe segments, chunks of rubble, and heavy lime mortar—all mixed with a rich collection of fragmentary pottery containing a few well-preserved and mendable pieces, such as a large, nearly complete lamp and fragments of four other restorable vessels. Some of the architectural debris and pottery was found cemented to the mortared rubble of the column foundation (or rather to the external face of a rubble-and-mortar wall encasing the block foundations), indicating that the rubble wall encasing the block foundation of column 16 was laid at the same time that this ceramic and construction fill was dumped. Pottery recovered included some Hellenistic and Lydian wares (4 percent and 1 percent by weight, respectively), but it was overwhelmingly early Roman, representing a chronological spread from the Augustan period to the middle of the first century AD.

The indiscriminate and varied nature of the building material filling the space between the anta and column 16 demonstrates a lack of clear, chronological layering and suggests that this mix originated from a building dump, probably from multiple sources.¹³⁹ Recognizing the heterogeneous and unstratified nature of the material, F. Can commented that “[this] deposit is brought from somewhere else, a destroyed building.”¹⁴⁰ Although dating discarded building debris is difficult, some of the fill could be late Hellenistic or early Roman; however, there was also some material broadly datable to the second century AD, such as rectangular and square bricks, two courses of a brick wall preserved as a chunk with thickly mortared joints, and particularly the crudely shaped pan and cover roof tiles.¹⁴¹ The coarse, round bricks, may also have come from late Hellenistic or early Roman contexts at Sardis where round bricks have been found, or, as likely as not, from hypocaust pilae typical of the second century AD.¹⁴²

In view of its importance, the pottery that was excavated from this trench was carefully studied by a team of specialists. As noted by E. DeRidder in her excellent final report, the overwhelming majority of the ceramics found in this small trench (some 94 percent by weight) was judged to be early Roman (Augustan to the first half of the first century AD), but a few pieces were assigned to a wider chronological range. The majority of the pottery consists of small- to medium-sized plainware body sherds, some Lydian and Hellenistic sherds, “transitional” pottery produced during the late Hellenistic and early Roman periods, lamps (Broneer types 19 and 21 and red-on-white [ROW]), finewares, transport amphorae and Eastern Sigillata B (ESB), and local imitations of ESB, as well as a few pieces of Western or Italian Sigillata.¹⁴³ Of particular interest are a large, handsome lamp found on the second east-facing foundation step of the northeast anta pier (Sardis inv. L11.22, missing only about one-quarter of its bottom); part of a one-handled jug (Sardis inv. P12.182); part of a basin or krater with interior slip (Sardis inv. P12.185); and some pseudo-Koan amphora fragments (Sardis inv. P12.180) (Fig. 3.48).¹⁴⁴ Due to their better preservation and closely datable typology, these vessels were classified as “diagnostic” and firmly dated to the first half of the first century AD. The majority of plain and utility wares, “if judged to be ‘non-diagnostic,’ were given an early Roman date.”¹⁴⁵ Consequently, the “diagnostic” ceramic deposits from this area were judged to be contemporary with the foundation of column 16, which in turn dates this foundation to the first half of the first century AD, “[built] perhaps, a generation or two after the devastating earthquake of 17 CE.”¹⁴⁶ This same argument posits that this phase of early Roman construction applies not only to the foundation of column 16, but also to all column foundations of the temple’s east end. Hence, the pseudodipteral arrangement of the temple was assigned to the same Julio-Claudian design effort, a fundamental change from what I believe to be a Hadrianic undertaking associated with the emperor’s visit to Sardis and the granting of a second neokoros to the city (see in this section below).¹⁴⁷

I agree with the dating of the great majority of the ceramic evidence from this trench (especially the larger, more complete pieces) to the first half of the first century AD; no arguments are presented here contesting that judgment. However, upon closer examination of the ceramic evidence it becomes clear to us that among the ceramics from this trench (besides the Lydian and Hellenistic material, which are earlier and viewed as “residual”), a few smaller pieces could be later, probably late first or even early second century.¹⁴⁸ Two ROW lamp fragments were tentatively assigned to the “mid-first into second century AD” by the Sardis specialists, while one expert believed they could even date later.¹⁴⁹ New evidence from Ephesus (Prytaneion and Insula 2, “Slopehouses”) established the origins of the ROW lamp type in the late Augustan period, a fact of which Outschar was aware.¹⁵⁰ Still, according to many specialists, ROW belongs overwhelmingly in later periods and contexts, predominantly Flavian to Hadrianic.¹⁵¹ The new information from Ephesus, taken at face value, indeed requires us to consider our ROW fragments within a larger chronological window, extending from the late Augustan to the Hadrianic period; however, it does not follow that the ROW fragments found in the Sardis trench must belong to the earlier, Augustan date.¹⁵²

Outschar also viewed some of the Hellenistic-type Eastern Sigillata B pottery (particularly one dish rim fragment of the thicker ESB) as local production “that could easily go on to late first century or early second century AD.”¹⁵³ It is important, however, to underline that J. Hayes, a leading expert in this type of pottery, thought that the rim fragments belonged to the first half of the first century AD.¹⁵⁴ While the first half of the first century AD is the established opinion of pottery specialists for the majority of ESB in the Sardis trench and in the eastern Aegean in general, production of ESB did not end in the region after it “peaked” ca. AD 50–60; it slowed down and “dwindled” all the way into the end of the first and early second century AD.¹⁵⁵

Although none of the ceramic fragments or construction debris from this trench must date from the later periods, it is important for our interpretation of the evidence that some could. Conversely, though some of the larger, better-preserved ceramic pieces (specially considered as “diagnostic”) must be dated to the first half of the first century, this is not true for all the ceramic material from the trench, even in the eyes of the Sardis pottery experts.¹⁵⁶ As an alternative way of looking at it, the dominant ceramic evidence pointing to the first half of the first century AD is persuasive but not probative; it cannot offer proof without a reasonable doubt. Since, a small number of artifacts from this small, self-contained trench—diagnostic or not—could be assigned a later date in the late first or early second century, in my view they provide us with a terminus post quem.¹⁵⁷

It is acknowledged that the conclusions offering a later terminus post quem for this trench are my own (sometime in the early second century AD, probably the Hadrianic period), hence the later dating of column foundation 16. This may be contrary to the conclusions of the pottery specialists, who argue that the construction of the column foundation, and by extension all other column foundations of the east end (consequently, the entire pseudodipteral design), should be dated to the first half of the first century AD. In my opinion, these conclusions are not based on a dating of all the contents of this trench (ceramic and architectural) with an absolute degree of certainty, as they are unavoidably relevant to the dating of the major construction phases of the temple as a whole. The detailed reasoning, justifications, and conclusions of the experts’ arguments—which I respect but with which I do not concur—are presented in the excellent article by Cahill and Greenewalt (2016).

Important as it is, the ceramic evidence and “ceramic horizons” on which the specialists have rightly focused, is not the only evidence or horizon for the temple. Architectural, structural, decorative, and historical horizons also require consideration, coordination, and explanation.¹⁵⁸ We should begin a consideration of our perspective on the temple’s design and history by asking an important question: If the pseudodipteral scheme had truly originated under the Julio-Claudians, what was the purpose of this fundamentally redesigned, repurposed, double-cella temple? What was the programmatic event that could explain the division of the cella and the focus on an elaborate, new east front in the early first century AD while the original west front lay incomplete and unattended? Moreover, why was the rare convention of using complex, intercolumnar contractions applied to the east peristyle (an application used only on the principal facade of a temple, based on the Archaic and possibly Late Classical examples we know; see Ch. 4, pp. 238-239) unless it was intended to be a new front? And how can we accept the east side as a new front, unless it had become the facade of a divided cella housing a new, important cult? I believe that the impetus for this ambitious new program was the granting of imperial cult privileges to Sardis under Hadrian and the consequent preparation of the city’s major temple for the new imperial masters; the new front is programmatically linked with the imperial cult. The evidence provided by six or eight full or fragmentary colossal cult images of the Antonine family (possibly also of Hadrian and Sabina, see below) establishes the Artemis temple as a joint temple subsuming the imperial cult. Other historical and epigraphic considerations surrounding this event makes the granting of the second neokorate to Sardis and the rebuilding of the temple under Hadrian a logical proposal.¹⁵⁹

Can we replace this scenario with a Julio-Claudian double-cella pseudodipteros? Could Sardis have received its imperial cult privileges under any one of the Julio-Claudian emperors and redesigned its primary temple in order to incorporate this cult? The Julio-Claudian era was a period when ties with Rome were strong. Since Sardians honored Tiberius as Founder of the City and described themselves as Kaisareis Sardianoi (“Emperor’s Sardis”), it is suggested that “already in the first century Sardians intended to dedicate this part of the building to the imperial cult.”¹⁶⁰ However, there is no evidence that Sardians were intending or planning to alter their temple in a major way as a response to the needs of a cult, nor to dedicate any part of it to such a cult at this time. They competed for the neokorate honor in AD 26 but did not get it.¹⁶¹

Archaeological evidence is often not absolute, but rather it supports particular interpretations. Here I have presented an overall interpretation of the evidence as I see it—I am prepared to accept error in some parts, but I trust that the overall picture is reliable. I believe that the strength of the evidence that supports an issue should be in proportion to the significance of that issue. In architecture, if evidence leads to fundamental, consequential (sometimes irrevocable) changes in the design and history of a building, there should be little room for doubt; when the stakes are high, the burden of proof should proportionally be high. Moreover, I feel that our job is not to provide our readers with a “proven” thesis but to present the materials and circumstances as we see and interpret them, as well as to represent the alternatives and different readings—to offer a “marketplace” of ideas and possibilities in a disciplined but open-ended discourse.

Inspired by the idea that the spirit of inquiry is that “spirit which is not too sure that it is right,” I offer the following modification scenario to bridge the gap between different viewpoints and to represent the possibility of some early Imperial construction in the east porch (a “what if” scenario, though my argument and preference for an early to mid-second century AD date remain unchanged).¹⁶² If the foundations of column 16 were begun sometime in the first half of the first century AD, this column (and its southern counterpart column 17) would have been a logical place to begin the long-delayed peristyle, no matter what the larger design was going to be, or whether it had yet been determined. This much we will accept as an alternative interpretation of the field evidence, as a technically possible alternative, but one that is not provable without declaring that the entire pseudodipteral, double-cella scheme is Julio-Claudian.

In addition to the technical, archaeological, and stylistic evidence that supports our proposition that second major phase in temple’s history was Roman Imperial (namely, the creation and implementation of the pseudodipteral plan), two major considerations, taken together, help to narrow our chronological framework to the mid-second century AD: a building inscription from one of the east peristyle columns; and the discovery of a number of colossal imperial images of the mid-second century AD inside or near the temple cella. Likewise, the stylistic evidence provided by the east door ornament strongly supports its Hadrianic date (see in this chapter pp. 167-171). The following is mainly a review of the column inscription and the imperial statues in their archaeological and historical contexts, in order to illuminate further the Roman history of the temple.

The Roman presence in and use of the sanctuary as a major cult center, attested by the large numbers of coins and votive and dedicatory inscriptions from the precinct, hardly needs explication; however, the fact that a majority of these dedications belong to the second century AD deserves emphasis. Here our short discussion will be limited to dedications from the Roman period.¹⁶³

Four marble building blocks with inscriptions on them (numbers 51–54 in Sardis VII.1) were found, along with other uninscribed, squared blocks, at the foot of the northwest steps. They appear to be building elements belonging to an extension of the temple, perhaps used as facing for the rubble wall just east of these steps, crossing the north pteroma (Fig. 4.7).¹⁶⁴ Because they are made of coarser marble and lack drafted edges, they are unlikely to have come from the cella walls. All are dedications set up by the city of Sardis, its people, and its council to honor “priestesses of the goddess Artemis” for having served her with devotion; they all date from the late first or early second century AD. Block no. 52 is dated to AD 126–27, as it names the proconsul Stertinius Quartus.

Block no. 54 honors Apphion Secunda, daughter of Menandros Lechitas, for having served “with due devotion and lavishness in a manner worthy of the goddess” (Sardis VII.1, no. 52). It is interesting that Apphion Secunda was honored also with a duplicate inscription, engraved on the pedestal of her now-lost statue (Sardis VII.1, no. 53). This tall, inscribed pedestal was set up by Butler at the northwestern end of the marble steps, where it still stands today (Figs. 2.152, 2.153).¹⁶⁵ A inscription carved on a different marble statue base honors another priestess and entreats Artemis to “ever preserve Sardis in concord through the prayers of Moschine” (Sardis VII.1, no. 50). The statue of Moschine, represented as a draped young woman (head missing), was found north of the temple in 1929 and is in the Izmir Archaeological Museum. According to Hanfmann and Ramage, the sculptural style, letter forms, and the “stress laid on ‘concord,’ suggests a late first century BC date for this statue base, possibly the period about 41 to 31 BC, when Sardis was violently disturbed by civil strife.”¹⁶⁶

There is no doubt that these dedications and offerings were representative of many others that littered the precinct around the temple and its altar during or shortly before the Imperial era; they provide evidence for considerable activity concerning the temple and the primacy of the cult of Artemis throughout this period.

The Evidence of Column 4, the “Talking Column”

The most important epigraphic document connected directly to the temple’s Roman history—and particularly the construction of its peristyle—is an inscription in Greek verse carved on the bottom fillet of column 4 in the east exterior colonnade, north of the middle axis of the temple (Fig. 2.203). The inscription informs passersby: “My torus and my foundation block are carved from a single block of stone, furnished not by the people [demos] but given by the house [of the temple]”¹⁶⁷; it proudly declares in the first-person singular, “[O]f all the columns [stones] I am the first to rise” (Fig. 3.49).¹⁶⁸

According to an alternative reading by Angelos Chaniotis, the column was “almost certainly given by the temple and not by an individual . . . given by the ‘house’ (oikos) of the (temple),” that is, the stone was furnished by the temple (“our own stones”) or temple funds, possibly alluding to and taking pride in the fact that the quarry (or parts of it) was owned by the sanctuary; this was not uncommon, especially in the second century AD.¹⁶⁹ Given the Roman habit, across the entirety of the Mediterranean region, of indulging in exaggerated literary tropes and metaphors to describe fine building stones and marbles, the clear and modest expression of pride in a locally quarried stone expressed in an inscription from Sagalassus is noteworthy and seems to echo the pride expressed by our column (and temple) at using marble not from a distant famous site, but from its own quarries located virtually a stone’s throw away.¹⁷⁰ The Saga-lassus inscription (ca. second century AD) declares that although the workmanship of a particular monument appears to be in Phrygian (Synnadic) marble, “in reality, the stone comes from this land!”¹⁷¹ The message, placed close to eye level and easily readable (the letters were probably painted red), starts in the southeast quadrant of the base and requires walking around the column counterclockwise to disclose its meaning—a movement that would have nourished kinetic memory and shown the temple to best advantage, revealing its impressive forest of columns and the ornate bases clustered at its east end (Figs. 2.223, 3.50).¹⁷²

Epigraphers specializing in Anatolian inscriptions, led by the late Peter Herrmann, suggest a broader date from the late first to the mid-second century AD, based on letter forms, epigraphic style, and content. However, the balance of opinion (particularly Herrmann) strongly favored the Trajanic/Hadrianic period, in contrast to the stylistic interpretation of Buckler and Robinson in 1932 to the “earlier half of the 1st century AD.” Hermann’s tentative date received the general approval of many of his colleagues, including Georg Petzl (conservatively), H. Malay, and C. Foss.¹⁷³ More recently, Angelos Chaniotis expressed a clear preference for a mid-second century, specifically Hadrianic date, noting not only paleographic but also textual and literary considerations.¹⁷⁴

The celebratory nature of this inscription—casting the column as a victor in a building competition as the very first column “to rise,” or the first to be completed among a cohort of others—is confirmed by the fact that the torus of the Asiatic-Ionic base is fashioned as a wreath (Fig. 3.51).¹⁷⁵ In the center and facing west, the loose ends of a ribbon tied by a broad band flutter to the sides. Pointed, overlapping, horizontal laurel leaves (unfinished except for a single leaf) encircle the torus and converge in the center of the opposite, eastern side, where eight well-preserved radiating cuttings (14 cm in diameter) apparently mark the position of a metal or gilt ornament, perhaps a medallion. In effect, the entire torus was conceived as a victory wreath (corona laurea) to honor the column and declare this honor to the goddess.¹⁷⁶

Transforming a symbol into architecture, or rather transforming a major architectural element into a symbol, is indeed a rare and special creation. The closest and earliest example that could have served as a model for the Sardis column is the most important victory column of antiquity: the Column of Trajan in Rome, commemorating his Dacian victories (Fig. 3.52). The gigantic torus of this column is decorated by delicately modeled, horizontal laurel leaves gathered by an elegant lemnisci(from the Greek lemniscus, meaning ribbon or band) spiraling around the wreath.¹⁷⁷ Although this column does not “speak” in the first-person singular, the base inscription declares the importance of the extensive leveling of the land needed to built Trajan’s forum (a massive engineering triumph orchestrated by architect Apollodorus of Damascus), evocative of the achievement declared by the Sardis column.¹⁷⁸ Thus, the assessment of the Sardis column as a “victor” is firmly anchored in the unique kinship between it and the quintessential victory column of the Roman world. If Apollodorus’s column in Rome is indeed the first shaping of a base torus as a victory wreath (as we believe it to be), it is logical that our column followed the unique, iconographic lead provided by the illustrious example in Rome; there is no earlier example of such an iconic type.

Architecture or sculpture that speak in the first-person singular are rare in the Classical, Hellenistic, and Roman periods; those that do speak are almost exclusively reserved for funerary contexts. However, such speaking objects are more common during the Archaic period, from drinking cups to kouroi when the speaking object was the “explicit preference.”¹⁷⁹ Perhaps the closest comparison our column in linguistic form and message is the monumental kouros from Delos dedicated to Apollo by the Naxians. Much in the same manner as the Sardis column, the marble colossus, nine meters high and set on a monolithic square base, boasts, “I am of the same stone, statue and base.”¹⁸⁰ The conceptual and linguistic parallelism of these two examples could not have been wholly accidental; this suggests a deliberate attempt by the Roman builders of the pseudodipteros at Sardis to recall the distant past, keeping with the archaism in the design of the original structure. A. Chaniotis noted that the verse inscription of the Sardis column was archaizing in content (though not letter forms) and described it as displaying an “archaizing ‘couleur,’ or ‘literary flavor,’ which would fit well the Hadrianic/Antonine date.”¹⁸¹

Rooted in and nourished by the Second Sophistic, the Hadrianic period was a time when the rhetorical underpinning of burgeoning urban architecture was most visible at Sardis and the cities of Asia Minor. Just as the brilliant rhetoricians of the period set the tone for the Greco-Roman koine of Asian cities, the archaizing declamation of the rising Sardis column alluded to the legitimizing past and proclaimed the rise of the city under its Roman patrons.¹⁸² Our epigraphic and literary efforts—simply to read and “translate” the column inscription in order to provide a strict chronology—should be balanced and enriched by taking a broader, longer view of history. Thus the victorious, talking Sardis column does not just commemorate one particular important occasion in the construction sequence of the temple, but in doing so it unearths its entire history and communal memory. It records an event embedded in the city’s distant past; the winner is celebrated both in relation to a specific victory at a specific time and also (as all victors deserve) in reference to all time. To quote J. C. Barrett, the inscription is “not so much a record of history as the means of creating history.”¹⁸³ Thus column 4 with its unique inscription is not simply an interesting instance among the epigraphic elements of the temple. When taken in the full context of its associations and meanings, it is a powerful tool in providing a viable date for the Roman reconstruction and explication of its architecture.

The message of this talking column and the decoration of its base allow us a glimpse into the minds of its builders, for whom the architectural challenge and subsequent gratification of raising a column of this size and scale was worth celebrating. It provides a concrete affirmation of the technological and symbolic importance of monolithic construction that underlies the nature of all monumental Roman architecture. Even though no donor’s name is given in our inscription (the column is provided by the “House of Artemis,” or most probably, the funds provided by the Sanctuary of Artemis or sanctuary-owned quarries, rather than by the people/demos of Sardis),¹⁸⁴ it still reminds us that the enormous financial undertaking represented by this column could encourage generosity in others through its celebratory gesture.¹⁸⁵ One remembers the western Anatolian tradition, from the Archaic period onward, of local individuals and royal patrons donating building columns to establish or maintain their local power and perpetuate their memory, although the practice was widely different among the cities.¹⁸⁶ It reminds us of the power of words and the mythical, regenerative stories they tell.

Colossal Heads

A number of male and female colossal heads found in or near the temple over a long period of time (1882–1996) and identified as members of the Antonine family help suggest a closer date for the major Roman redesign and rebuilding of the temple, further illuminating its history. Their existence also gives us the opportunity to study cult statuary in its original architectural setting. We have three complete or nearly complete male heads and two female heads. In addition, there are numerous fragments of colossi; at least three of these may represent separate individuals, one male and two females (see Table 3.1).¹⁸⁷

Restored at about three-and-a-half to four times lifesize, they must have belonged to acroliths; this would have limited the use of marble to the exposed parts of their bodies (face, neck, and appendages), while their torsos were completed in wood, molded sheet metal, or bronze, and painted or gilded. Some body parts could have been draped in actual cloth arranged over a wooden armature. As observed on the heads of Faustina the Elder, Marcus Aurelius, and Commodus, the area at the back of each sculpture was hollowed in order to lighten the mass (Fig. 3.61). Metal bars, wooden struts, or marble pegs must have secured them to the cella walls and stabilized them, as the heads have numerous cuttings and large dowel holes at the back. In the case of Faustina, and the head tentatively identified as Lucilla, the back might have been covered with a veil made of painted wood or metal, fitting for their goddess-like appearance in the cella. In accordance with the characteristic display of Roman heroic statuary, these standing or seated figures were not intended to be viewed in the round, but rather frontally or at an angle, placed between the columns and against the walls of the newly created east cella.¹⁸⁸

Apart from their intrinsic interest as a unique sculptural group of imperial colossi from one site and one building, we cannot overstate the importance of the images in the study of the architecture and iconography of the temple. These imposing portraits, representing six to eight members of the Antonine family (and possibly Hadrian and Sabina, see below; and perhaps others not preserved or not found), were clearly objects of worship—proper agalmata—and attest the incorporation of the imperial cult into the temple of the Sardian Artemis.¹⁸⁹ The assumption that the city was awarded its second neokorate under Antoninus Pius (as was generally believed) or Hadrian (as recent scholarship indicates) gives the reason for the division of the cella and the creation of these colossal dynastic icons; it also provides a plausible Hadrianic date for the Roman phase of the monument.

The first head, identified as Faustina the Elder, was found in 1882 in an exploratory trench opened inside the west end of the cella by George Dennis, the British consul in Izmir (Fig. 3.53). Now in the British Museum in London, it is the only fully preserved colossal head (a height of 0.91 m, chin to crown) and was a companion to the partly preserved head of her husband, Antoninus Pius, found inside the cella by Butler in 1911 (Fig. 3.54).¹⁹⁰ During the same year, a badly damaged head and neck of a bearded male (preserved from mid-nose to the base of neck, height 1.05 m) was found in the south pteroma of the temple (Fig. 3.55).¹⁹¹ Butler wisely refrained from venturing an identification, commenting that it was a “shapeless mass of marble,” but Hanfmann resolutely believed the head to be an Early Hellenistic image of Zeus whose features merged with that of Achaeus, who had briefly usurped rule at Sardis (r. 220–214 BC).¹⁹² This view served to justify Hanfmann’s primary hypothesis that the two-cella temple, intended to house the joint cults of Artemis (west) and Zeus (east), belonged to the second Hellenistic phase that he championed (see in this chapter pp. 154-155 and pp. 212-213).

Recent analysis (based on admittedly limited stylistic considerations due to the poor state of the piece) favors the head being Marcus Aurelius, the emperor who succeeded Antoninus Pius to the throne in AD 161.¹⁹³ In 1914, his last year of excavation at the temple, Butler found a colossal female head—almost intact, except for the nose and back of the head fully shorn away—immediately north of the building (Fig. 3.56).¹⁹⁴ The broad, oval face, wide-set eyes, streaming, wavy hair, plump cheeks, and overall youthful appearance (thought to be “different” from the others) triggered different opinions: Artemis; Lucilla, the daughter of Marcus Aurelius and the wife of his co-regent Lucius Verus (Hanfmann); the Younger Faustina (Buckler and Robinson); and recently again Lucilla (Smith, Burrell, and Yegül).¹⁹⁵ Near it, Butler found the fragment of another colossal female head, preserved only with “the nose and mouth and part of the brow with a bit of wavy hair.”¹⁹⁶ Identification is impossible because the present whereabouts of this fragment is not known.

A fragment of a fourth colossal female head was found in 1961 among the ruins of Butler’s expedition house.¹⁹⁷ It is represented only by a partial left eye, nose, and left cheek, and therefore cannot be the same 1914 piece mentioned by Butler above. Hanfmann tentatively suggested that it might be Artemis. Based on the soft, classical modeling of the cheek and the sharp, sensitive outline of the eye, we suggest it might be Sabina, or an idealized Sabina appearing as Artemis; it can be compared favorably to an idealized portrait of Sabina in the Museo Nazionale Romano in Rome (Fig. 3.57).

A fourth male head found near the Butler expedition house is represented by a fragment of the right side of the face: the temple, the corner of the left eyebrow, and curling locks of hair (Fig. 3.58).¹⁹⁸ Noting the fine workmanship, Hanfmann observed that it might be earlier than the head of Antoninus Pius.¹⁹⁹ Indeed, the loosely curved locks that encircle the forehead like a crown and are separated by narrow grooves, as well as the fleecy eyebrow, make the identification of this fragment as Hadrian also a possibility (Fig. 3.58).²⁰⁰

The last of the colossal heads is a bearded male found in 1996, in a trench excavated within the east pronaos porch (Figs. 3.59, 3.60, 3.61).²⁰¹ Along with large numbers of marble architectural pieces from the temple, the head must have been intentionally mutilated and buried in a deep pit dug in late antiquity (see pp. 180-181). It is judged to be roughly four-times lifesize, but only the lower half (below the nose) of the face and neck are preserved (max. height 1.19 m, max. width 0.69 m, tip of nose to chin 0.40 m). As in the other colossal heads, the back of the head and part of the neck have been carved out to lessen the enormous weight; they are furnished with deep holes for metal clamps to secure the statue to a wall behind it (Fig. 3.61). The deeply drilled, full beard with vertical corkscrew curls is coarser but more vigorous than that of Antoninus Pius; the moustache is thinner, almost sketchy, the cheeks are broader and flatter, and the neck thicker. The neckline is short and evenly rounded, suggesting that the statue was a cuirassed military type (a draped civilian figure would have had a deeper and uneven open neck). R. R. R. Smith’s identification of the head as Commodus, the last of the Antonine emperors, is convincing.²⁰²

Burrell challenged this identification by noting some of the stylistic differences between our head and typical portraits of Commodus (Commodus’s lips are thinner, and the mustache fuller and more luxuriant) and preferred to identify it as Lucius Verus.²⁰³ She cited two other reservations against a Commodus identification: Lucius Verus, as the co-regent of Marcus Aurelius, should have been in the temple; and Commodus suffered damnatio memoriae late in AD 192 or 193 (and although he was rehabilitated and deified by Septimius Severus in 195). It may be difficult to account for the two years during which the image escaped destruction. The first concern is easily explained: indeed, Lucius Verus would have been included, especially if the nearly complete female head in the Istanbul Archaeological Museum is Lucilla (Verus’s wife and Commodus’s sister). We have to assume then that the Verus head (if it existed at all) has yet to be found, or was destroyed in late antiquity along with other parts of the temple. The reason for Commodus’s head to escape destruction after 193 should be sought in mere chance, or possibly the remarkable potency and permanence imbued in cult images—the agalma—as objects of worship and devotion.²⁰⁴ Even the recycling of cult portraits of emperors was rare and undertaken with reluctance. Shorn of their pagan, religious context, many portraits survived Christianity to be venerated and honored. Naturally, there must also have been economic considerations against obliterating such costly colossal marble statues. Perhaps, the “condemned” head of the acrolith was simply ignored for a while (covered with a cloth?), or taken down in order to be recarved into the likeness of the next emperor, happily reinstalled onto its torso with the restitution of Commodus’s status.²⁰⁵

All of these heads, where sufficiently preserved, display broad, open faces, parted lips, large, deep-set eyes, and slightly dazed and distant expressions appropriate for iconic images. The hair and beards of male portraits exploit luxuriant, baroque treatment with tousled (Antoninus Pius) or corkscrew (Commodus) curls, achieved through the use of the running drill (Fig. 3.60). Dramatic twists of the neck to the left (Antoninus Pius), or right (Marcus Aurelius), and perhaps upward, too, are formulaic poses for a heroic, godlike appearance. These regularizations and generalizations typical of cult images, and their incomplete or poor state of preservation, make the identification of most Sardis heads a challenge. Furthermore, the iconography of colossal portraits is difficult to establish because they are generally outside the stereotypes and standard models created in Rome.²⁰⁶ Exploitation of the distortions created by their sheer size, height, and the sharp, unnatural angle of vision—as well as their dramatic poses and awesome appearance inside a darkened cella interior—places them in a category of their own. As pointed out by Burrell, the colossal head of Titus from the imperial cult temple in Ephesus could be a case study to show the remarkable deviation of the image from the well-known iconography of the emperor.²⁰⁷ The lack of clearly established sequences for imperial portraits from provincial workshops (our colossal heads, probably made of local marble, are almost certainly local products), further complicates the process of identification.

With the cella walls of the temple entirely gone and little of the sculptures’ bodies preserved except for the heads and necks (and a few fragmentary parts), it is difficult to envision how the statues were displayed, whether individually or as a group.²⁰⁸ It is also hard to know if all the royals, deified or mortal, were in the eastern cella or if some of the women, such as Faustina the Younger, shared the west cella with Artemis. Since it is fairly standard for all members of the imperial house to be together, often including their relevant ancestors, we should expect Artemis to have been alone in her west-facing cella (although occasionally, Roman emperors and empresses enjoyed the honor of being displayed together with a deity in the same setting and on the same podium, as is the case of Mars, Augustus, and the deified Julius Caesar honored together in the Temple of Mars Ultor in Rome).

The large platform in the east cella (the original image base, foundations ca. 6 × 6 m) could have easily accommodated two but possibly more colossal figures, though this number would depend whether the figures were standing or seated. Counting the spaces between the westernmost pair of columns (partially blocked by the central image base), there are eight spaces available for statuary between the interior columns. Each of these eight spaces (ca. 3 × 3 m) could have accommodated at least one standing or seated figure between the fluted interior columns. A hypothetical reconstruction by Cathy Alexander of the standing, cuirassed Commodus (restored to a standing height of ca. 7.90–8.0 m, i.e., ca. 4 m shorter than the restored height of the east door) gives an idea of scale, which clearly presents a tight relationship between the cult image and the interior space, though perhaps not nearly as tight as that of Phidias’s seated Zeus inside his temple at Olympia (Fig. 3.62).²⁰⁹ Thus, with eight colossi in four pairs on either side of the central base, a total of ten colossi (or possibly twelve) could have been housed—though somewhat tightly—inside the east cella.

We have extant material evidence for eight figures, four male and four female. Five of these larger works have been tentatively identified as Antoninus Pius, Marcus Aurelius, Commodus, Faustina the Elder, and Lucilla. Three fragments (two female, one male) are too small for confident identification, but two of these could belong to Hadrian and his wife, Sabina. Even if the actual division of the cella had been completed under Antoninus Pius rather than Hadrian, the latter would certainly have been given a position of honor in the space; he was the one who granted Sardis its second neokorate, and because he was the deified “father” of Antoninus Pius, the Antonine dynasty received legitimacy to rule from him. In the same way, Hadrian, whose colossal image was included along with Trajan and Zeus Philios inside the cult temple of Trajan in Pergamon, received honor and legitimacy from his deified “father.”²¹⁰ If this identification is correct, currently missing from the Antonine family are Lucius Verus, Faustina the Younger (Marcus Aurelius’s wife), and Crispina (Commodus’s wife) (see Table 3.1).

Proposals for the Location of the Cult Images Inside the Cella

It is hard to know how the four (or five) emperors and their consorts were arranged inside the cella at Sardis, because none were found in situ, not to mention that statue groups of imperial colossi are rarely preserved in their original architectural contexts. Sardis is also quite unusual in the sheer number of colossal figures that seem to have gradually accumulated inside a single interior space.²¹¹

Even though the divided east cella was a vast space by ordinary standards (interior dimensions were ca. 18.40 × 25.16 × 19 m), restrictions imposed by the architecture of the interior (basically a three-aisled basilica with a very wide nave, 6.80 m clear at floor level and wider above) would have limited the options for grouping sculptures.²¹² If Sardis had received its second neokorate under Antoninus Pius, we might have had a straightforward arrangement: the founder of the dynasty and his consort (Antoninus Pius and Faustina the Elder) occupying the central podium, with the colossi of other Antonine emperors and their consorts between the tall interior columns flanking them, facing each other across the wide nave (hypothetical alternate arrangements of the colossi inside the cella are given in the reconstruction study plan; Fig. 3.63).

Burrell suggests that Antoninus Pius was on the viewers’ left, “naked, diademed, enthroned, and [he] may have held a scepter, in the guise of Zeus.”²¹³ Standing on his left (the viewers’ right) would have been Faustina (perhaps as Hera), whose slight gaze to the right would meet the more dramatic turn of her husband’s head.²¹⁴ If the identification of the fragments suggested to be Hadrian or Sabina is correct, they would have occupied the spaces between the westernmost pairs of columns (nos. 69–71 and 70–72), Hadrian on the left and close to Antoninus Pius, Sabina on the right and close to Faustina.

Since the neokorate honor was probably received under Hadrian on the occasion of the emperor’s probable visit of Lydia and Sardis in AD 124, I suggest an alternative scenario in which Hadrian and Sabina were the main cult figures and the arrangement was reversed: instead of Antoninus Pius and Faustina, they would have occupied the central position in the cella. Naturally, this scenario would have required the reassessment of Antoninus Pius’s identification as a semi-nude Zeus. The rest of the Antonine emperors and their consorts would have taken up the spaces between the columns, probably men on the left and women on the right, in order of seniority and chronology: Marcus Aurelius and Faustina the Younger (missing); Lucius Verus (missing) and Lucilla; Commodus and Crispina (missing).

From the remains of the heads it is impossible to tell if they belonged to standing or seated figures, or whether the emperors were in toga or military dress. The figures between the columns could not have been viewed as easily and directly as the main platform figures, although slight twists in their placements, and the considerable distance across which they would have been seen, would have alleviated this problem.²¹⁵ The overall arrangement could have been reminiscent of apsidal imperial halls, typical of Italy, where dynastic images of imperial family would customarily fill the niches on both sides of a large, central apse of a basilic, often vaulted space. In some, as in the Claudian apsidal hall (“nymphaeum/triclinium”) at Punta dell’Epitaffio, in the Bay of Baiae, the deified members of a dynasty—or a mythological group with special iconographical significance—were presented as a theatrical, apsidal group.²¹⁶

A significant goal of the overall effect could have been the partial obscurity of the laterally arranged colossi between the fluted columns, allowing for a gradual, mysterious “unfolding” as the viewer progressed from the vast, columnar east porch toward the preeminent display of the cult-head on the central platform, as imagined in a digitally reconstructed sequence by R. Pellegrini (Fig. 3.64). The impressive columnar arrangement was a clear allusion to imperial presence and power. The movement forward must have been a calculated part of the cult experience—entering the wide space defined by the columns of the pronaos porch, climbing the imposing porch stairs, passing through the gigantic east door, and edging down the marble nave flanked by some of the most imposing and elegant columns in antiquity, under the watchful, distant eyes of the divine couples who occupied the space. The dim interior of the cella, illuminated mainly through the great door, would have enhanced the sense of drama and inspired devotion.²¹⁷

It is unclear how extensive the Roman rebuilding of the cella interior was, given that the room is preserved at only ca. 0.30–0.70 m below foundation level and nothing of the floor remains. Some of the fluted interior columns of the cella, as well as their bases, must have been repurposed to build the four pedestal columns of the east and west pronaos porches (columns 11, 12, 53, and 54), but we believe those four columns were taken from the west end of the original cella (columns 73–76) and the west porch (columns 77–80), when the area was redesigned as the new west cella with a new roof structure (see pp. 203-205).

Roman work would mainly have consisted of building new east and west crosswalls and obliterating the original west wall to create back-to-back cellas of nearly equal length. The new west-facing cella had a significant difference in floor levels between its east and west ends (ca. 1.60–1.70 m) that had to be resolved, as well as intercolumnar spans that did not match. This must have required a major overhaul of the interior support system and roof. On the other hand, the east-facing cella, with its original four pairs of eight columns (columns 65–72) and the “basis”—now pushed against the east crosswall—must have needed little change. The columns of the east cella, we presume, were structurally sound and visually attractive; there was no real reason for or advantage to removing them.

One of the primary archaeological arguments for a new, column-less east cella is based on the absence of evidence: Butler did not find any parts or drums of fluted columns inside the cella, while many were found at the west end.²¹⁸ Regarding a structure that was destroyed and plundered over two millennia, with the floor of its cella completely and deeply demolished, there can be no rational expectation that its elements should have remained where they had fallen. As impressive as the existing ruins of the temple are, we probably have no more than 7–8 percent of the original superstructure.²¹⁹ The near-total absence of the thousands of marble blocks comprising the walls does not imply that the walls never existed—the time and circumstances that swallowed the walls must have swallowed the 120 or so fluted drums of the east cella, too.

In a 2016 article on the archaeology of the Temple of Artemis, Cahill posits that all twelve interior columns of the original cella neatly provided the material for the four pedestal columns and “the voussoirs of a monumental arch at the western entrance to the city . . . [which] would account for the majority of the cella and pronaos column drums . . . since it took three [fluted] Hellenistic columns . . . to build one [Roman] pedestaled column.”²²⁰ Thus, in this scenario the entire temple interior was cleared of supports during the Roman rebuilding to create a large, open hall. The proposal that the short pedestaled columns of the east and west porches must have come from the cella interior is well taken. The creation of a large space uncluttered by columns for the display of the imperial colossi may also have had practical advantages. As a parallel, Cahill proposes the Temple of Apollo in Corinth; its interior colonnade was removed sometime in the first half of the first century AD and rebuilt, complete with Doric capitals, as the so-called Archaic Colonnade between the west end of the South Stoa and the Central Shops of the Roman Agora in Corinth.²²¹ It has also been suggested that the cella of the Archaic Temple of Apollo was originally undivided, and its dividing crosswall was an early Roman reconstruction.²²²

This new reading was proposed as a replacement for earlier studies going back to the time of Dörpfeld in 1886 (followed by O. Broneer, C. Williams, R. Stillwell, W. B. Dinsmoor, G. Gruben, C. Mee, and A. Spawforth), which held that the subdivided cella was an original feature of the Archaic temple, an arrangement that reflected the use of the smaller western chamber (with a platform base) as the repository of the sacred image and/or a treasury, not unlike the later arrangement of the Parthenon in Athens. C. Pfaff, who conducted the first investigations of the Corinth temple bedrock and foundations in 1993, considered the division of the cella as a part of the Roman rebuilding as a “strong possibility,” but because of the highly equivocal nature of the evidence, this was “not decisive.”²²³ In a recent study, J. M. Frey presents a stronger case for an undivided cella, mainly through a close reading of the column spacing of the relocated Archaic colonnade upon what is an early Roman stylobate. This arrangement is carried back to the cella of the Archaic temple, producing internal colonnades perfectly matching the original cella length and its own faint bedrock markings, and resulting in a proposal for an early Roman date for the subdivision. Even with the best of efforts, the evidence is equivocal, incomplete, and allows for multiple readings; still, it is a good hypothesis.²²⁴ More importantly, neither Frey nor any of the scholars who have concerned themselves with the Roman history of the Temple of Apollo at Corinth question the reason, purpose, and function of an alleged Roman double cella—why would the Romans have divided the cella in even the most cursory way?

In sum, there is much to be said about the arguments, pro and contra, concerning the possible Roman origins of the cella division at Corinth, but it is difficult to say that this case is settled with certainty. Furthermore, whether the Roman intervention and redesign of Corinth’s temple provides a cogent parallel for Sardis’s temple is not of critical importance to our question; they are very different temples with very different setups, contexts, and religious histories. The removal of the interior columns of a Greek temple at Corinth during its Roman rebuilding does not ipso facto require the removal of interior columns of a temple at Sardis during its Roman rebuilding. The more cogent and critical question yet to be addressed is whether the removal of the columns and the redesign of Apollo’s temple, with or without a cella division, was solely motivated by the desire to find cheap and convenient spolia for the building of the Archaic Colonnade in the Corinth Agora—and whether the temple was not “redesigned” as such, but just left to lie as a partial ruin without a roof (considered a possibility by Frey). Even more critically, if the Roman colonists introduced the divided cella, what was their purpose and program in doing so? This is a question for which, unlike the division of the Temple of Artemis at Sardis, we do not know the answer.²²⁵

Leaving the question of the Corinth temple aside, we see the situation differently with respect to the removal of all interior columns of the east cella of our much larger temple as a source for the Roman rebuilding of the four pedestal columns and the numerous voussoirs of the Monumental Arch. In the second-century Roman redesign of the temple with a divided cella, the area subsumed by the new west cella (which incorporated two-thirds of the original west pronaos) contained eight columns, four short columns at the upper level of the original cella (columns 73–76) and four of regular height at the lower level—what used to be the Hellenistic pronaos (columns 77–80). These eight, and possibly the four in antis columns of the east and west ends (between the east and west anta piers), were removed during the Roman rebuilding.²²⁶ Therefore, with a total of eight (or possibly twelve) fluted columns at their disposal, Roman builders had no need to dismantle the eight columns of the newly fashioned east cella.²²⁷

There are other considerations that argue against the motives for dismantling the east cella columns. Dismantling eight columns (ca. 120 drums) inside a cella with walls 18–19 meters high and cramped by other columns would require removal of the original roof—a task that would give any engineer pause. Normally, Roman construction that involved the lifting and moving of heavy loads (like the 2–4 ton drums) would have engaged many cranes and lifting devices connected to a mass of ropes and capstans stretching in all directions—hard to do inside a cella with fully standing walls. The process of lowering the delicately fluted shafts drum by drum and rolling them out of the newly constructed (or still under construction) east door would not have been impossible, but still expensive and challenging.²²⁸ Separating large column drums in good shape, connected vertically by dowels set in lead, would not have been an easy job.²²⁹ Even if all columns were removed, the new space, without any internal supports, would have needed a structurally challenging roof system. The wall-to-wall span of the cella, at ca. 18.40 m (nearly three times the ca. 6.70 m clear span of the existing nave between the columns), would have required a massive wooden truss! It would have been an undertaking not beyond the capacity of a local engineer or architect—much less our putative architect Dionysius, famed for his skill in covering large spans with roofs—but difficult, expensive, and unnecessary.²³⁰

A (rare) visitor who entered the cella would be confronted by an impressive vision: two rows of monumental, fluted marble columns carrying perhaps the most exquisite Ionic capitals of the ancient world, rising before and partly screening the vast, elegant sixteen-meter marble walls, a breathtaking experience (Fig. 3.64). In Roman architecture, columns were not mere structural expedients to carry a roof. Together with their capitals and set against a wall, they were powerful architectural elements invested with aesthetic and symbolic meanings and messages.²³¹ A row of columns marked distance, commanded perspective, created and ordered space, and imbued it with significance. Their measured architectural parataxis, reflected in the repetition and sequencing of imperial images alternating with columns inside a unified room, enhanced the pacing and kinetic perception and ordered spatial experience.²³²

The spatial and thematic iconography of columns was particularly meaningful and appropriate for imperial halls and cult spaces in Roman architecture. In spaces for imperial and religious purposes, sculpted cult figures were typically placed between freestanding or engaged columns, or in niches and aediculae defined by columns. In effect, columns and statuary exploited and enhanced reciprocal powers, unifying them into one magnificent visual experience; their combined presentation alluded to and reinforced the imperial and cultic themes such as strength, stability, probity, and continuity. In short, the empire’s far-reaching might was subsumed by the iconography of its columns. What the Roman architects found in the spectacular, columnar interior of the Hellenistic cella they inherited was much in line with the kind of space they would have wanted to create for the newly established imperial cult. If there was reason to take advantage of this powerful iconography and symbol, then why destroy it?

When the main cella was divided into two chambers a serious structural problem emerged in the support system of the roof for the newly created west chamber. While the four columns in two pairs at the east end of this chamber (columns 73–74 and 75–76) belonged to the old cella and had a uniform central span of ca. 9.30–9.40 m, the pair on the west end of the cella (columns 77–78) and the two built over by the west crosswall (columns 79–80) belonged to the original west pronaos porch; they had a narrower central span of ca. 8.40 m (Fig. 2.28). The closer spacing of these porch columns would have created a discrepancy of ca. 1.0–1.10 m, awkward and difficult to reconcile visually and impossible to reconcile structurally. The east–west architraves of the new west cella could not keep continuous, straight alignment. They would have to be narrowed down by making ca. 0.50 m right angle returns on each side in the middle of the room, and there are no intermediary supports to allow for such rebates. Furthermore, the floor of the west half of the new cella (originally the west porch) was some 1.60–1.70 m lower than the east half and thus required filling, which would have made the use of the taller pronaos columns impractical or impossible. Therefore, the Roman builders must have been faced certainly with the need for a new support system but perhaps also for a new roof and tiles for the new west cella—although, one imagines, if carefully overhauled, most of the sturdy and finely made original marble roof tiles could have been reused (Figs. 1.14, 1.15).

A pair of rather crude pier foundations (“rough piers”) made of large ashlar blocks encasing small rubble cores, located in the center of the west cella immediately east of the original west wall, seems to point to a solution, or at least give an idea of one.²³³ These foundations are preserved for ca. 0.60–0.70 m, two or three courses below the intended cella floor. They have a center-to-center central span of ca. 11.30 m and a clear north–south span of ca. 8.40 m. As first suggested by Gruben, they might have been the newly created supports for the roof, conveniently positioned more or less at the center of the room; they spanned ca. 12.20 m from pier center to the east crosswall face and ca. 13.54 m to west crosswall face, indicating that they are closer to the east wall than they are to the west by about ca. 1.30–1.34 m (Fig. 3.65).²³⁴ This difference in a room measuring 25.74 × 18.36–18.40 m is close enough; still, the columns supported by these rough pier foundations could have easily shifted toward the (submerged) west wall, resulting in equal spans toward east and west.

If a 2.40 m square column base were positioned in the center of the new west cella, it would have straddled the rough piers and about one-third of the west wall, giving equal clear spans of ca. 11.67–11.70 m toward the east and the west walls, face to face. This span could have been reduced to ca. 11.0 m if we reconstruct half-columns or half-piers projecting ca. 0.60–0.70 m from the east and west walls, though there is no physical evidence for it (Fig. 3.65 illustrates the former choice with engaged piers; Fig. 4.1, Roman phase, with half-columns). This is still a large span that would have required a timber truss system. The clear distance of 8.40 m between the piers in the north–south direction (and ca. 2.60 m from piers to north and south wall faces) would have presented no difficulties for timber beams or trusses.

Casting doubt on this hypothesis is the rather shabby appearance of these piers and the entire lack of clamps—contrary to common Roman building habit—to secure blocks to each other. However, one should note that their ashlar construction of reused blocks is fairly substantial, since what we see is all the subfloor work of some one or two courses below floor level; column plinths carried by them might have looked better. If there was a different support system, we have no evidence for it. We do not know if the supports carried by the “rough pier” foundations were masonry piers, marble columns, or wood. If they were piers they could be ca. 2.0 × 1.30 m, a fairly slender proportion for stone rising to a height of ca. 18 m or more. Regular columns would be less likely since such columns would have favored square foundations. Hefty wooden uprights carrying wooden trusses might have worked, but, at least to our eyes, this would be awkward inside such a voluminous cella intended for the goddess’s image, even primitive. In sum, this is a reconstruction hypothesis I am not completely happy about.

A system with no interior supports at all, and hefty timber trusses spanning the full width of the cella, would have not have been outside the largest timber spans Romans had achieved elsewhere (almost all in the West), offering an unobstructed view of Artemis. Indeed it would have been grand in its daring, imparting a sense of heroic achievement perfectly within the range of Dionysius, Sardis’s own heroic architect, if he ever came to Sardis, and if he were given the chance.

In the light of an important late fourth-century BC inscription from Ephesus, the so-called Sacrilege Inscription, the affiliation of the early Sardian cult to the even earlier Artemis cult of Ephesus is indicated. However, quite apart from the main cult of Artemis, well established in the sanctuary by the Archaic era, the Sanctuary of Artemis, the preeminent religious heart of Sardis, must have been the focus of its people’s ancestral cults, with overlapping, evolving syncretic beliefs and timeless traditions, its multivalent sanctity offering a welcoming ground for all that was safe, significant, and sacred—including, by the middle of the second century AD, the imperial cult.

The question of the Roman imperial cult in the Temple of Artemis is not solely a general concern about the history of religion at Sardis; it is an issue that casts light specifically on the history, date, and usage of the temple. Hence, I present a consideration of religions and cults at Sardis that is limited to their historic presence in the Sanctuary of Artemis and to their role in creating a welcoming syncretic atmosphere that could have made the inclusion of the imperial cult seem natural. There is no evidence—and thus no assertion—that these cults shared space with Artemis in the temple itself, nor can their presence in the larger sanctuary be connected directly and unequivocally with the establishment of the imperial cult. My goal is narrowly focused on cultic associations and activities relevant on the history of the temple and the sanctuary (with special consideration of Cybele cults). Religion at Sardis requires and deserves its own study.

The installation of the imperial cult in the temple, which transformed its architecture, must also have altered its meaning as a place of worship. Despite our possession at Sardis of the greatest number of imperial cult portraits (see above) from a single building in ancient Asia Minor, not all is clear about the establishment and practice of this important cult. I cannot argue for any meaningful correlation between the ancestral gods of Sardis and the cult of the emperors, but it is plausible that some earlier cults, such as Hellenistic ruler cults, continuing their traditional associations with the sanctuary, prepared the way for a smooth reception of that new cult and eventually assumed a new identity and new life by merging into the powerful presence of the imperial cult that moved into the temple itself (see below on the complex sociopolitical concerns surrounding the establishment of the imperial cult in the temple).²³⁵

The introduction of emperor worship in the cities of the Hellenized East during the early empire was slow and often required the pairing of the Roman cult with certain local cults and patron deities (a “veiled” process, as P. Talloen effectively names it), often in the same sanctuary and temple space, though never in separate chambers at that early stage.²³⁶ Even when emperor worship had become a commonly accepted, widespread phenomenon across the cities of Asia Minor, by the second century AD, the process was complex and dialectic, and always required negotiations between the local elite and their Roman overlords.

Even as we acknowledge and celebrate the firm presence of the imperial cult taking up half of Artemis’s temple at Sardis, there is no avoiding the question: What did the ordinary Sardians think of the imperial cult and imperial masters, apart from the usual, formulaic ceremonies, offices, and titles? In an outstanding cultural and administrative metropolis like Sardis, with its own unique history, there must have been persons, groups, and factions against the housing of the imperial family in the temple—baldly conceived: giving up half of the venerable temple to Roman patrons, the new faux-gods.²³⁷ The specific dynamics of this all-consuming civic and religious issue could be seen as a smaller, local aspect of the larger dynamics and negotiations for the reception of the imperial cult in general—or, even more broadly, of the viewing and understanding of the changing and overlapping identities of the plethora of cults that were, over time, a part of the welcoming religious culture of the sanctuary. The discourse undoubtedly affected Sardians of all stripes, engaging them intellectually and emotionally while being shaped by their and their city’s long-term experience and perception of religion and cult, from the ancestral Meter Oreia to Mater Castrorum, from Croesus to Constantine.

Before considering the issue of the establishment of the imperial cult in the temple, it might be useful to review, briefly and broadly, what is known about the connections of the sanctuary and later its temple with other local cults. In this review I will privilege two main belief systems for their special significance for Sardis: first, the cults of Cybele/Kybebe and Artemis; second, the cults of Zeus.²³⁸ I must reiterate and emphasize that, throughout its history, Artemis was always the principal goddess worshipped at the sanctuary and the temple. I offer the following discussion not as a proof based on hard evidence but as a hypothetical discourse on the possibilities rooted in the flexible nature of ancient religion in general and its reflection in the overlapping identities of Artemis and Cybele (or Kuvava, as she was referred to in early Lydia) at Sardis in particular. Naturally, these musings are offered to encourage discussion and are not intended to be a substitute for an in-depth study of the subject by specialists in the larger context of cult and religion at Sardis and in Anatolia.²³⁹

Overlapping Identities of Artemis and Cybele

A long-standing aspect of the cult of Artemis at Sardis is its apparent connection with the cult of Cybele/Kybebe, the great Anatolian mother and mountain goddess. In the days of the early travelers, some misidentified the temple as the Temple of Cybele (William M. Leake, 1824; Charles Texier, 1869; Gertrude Bell, 1902), and the association still crops up occasionally in popular guidebooks today.

In some of the earlier scholarship, the cults of Cybele and Artemis are conflated; however, starting with Radet—writing in the early twentieth century—Cybele came to be seen as having become the lesser element of this amalgam over the course of the Classical period (or even earlier) and was “interpreted as the alternative aspect of the main deity,” Artemis.²⁴⁰ This relationship was never clear-cut, at least in the minds of the scholars of religion who wrote about it. Kuvava, the Archaic goddess of the Lydians, maintained a presence at Sardis and enjoyed revivals during the Hellenistic and Roman periods. Views and theories about the nature and process of syncretism and amalgamation are well articulated in some of the writings of G. M. A. Hanfmann.²⁴¹ For J. Keil, studying the Ephesian Artemision, Artemis was one among many later manifestations of the original goddess.²⁴²

It is true that in their identity as ancestral female deities Artemis and Cybele/Kybebe played similar and important roles in the Lydian establishment of sovereignty and selfhood, but although overlapping, they were also different. The relationship is at its closest in the conceptualization of the religious geography of Sardis: Cybele, Mother of the Gods, Mother of the Mountains, “who dwells by the gold-bearing Pactolus,” looking upon the great mountain range surrounding the site—precisely the landscape described for Artemis’s sanctuary in Sardis.²⁴³ One might say that Artemis’s land is Cybele’s land; or, rather, that Artemis’s land, by religious, mythic, and thematic entrenchment, is embedded in Cybele’s land.²⁴⁴

Varying manifestations of Artemis and Cybele also generated focal points at different places in the city and its surrounding territory. Perhaps already in the Bronze Age, as proposed by Hanfmann, Artemis of Coloë might have had a shrine somewhere on the southeast shore of the Gygean Lake (Lake Coloë), although we do not know where this extraurban sanctuary was nor when it was active.²⁴⁵ We know that there was a separate temple, a Metroon, honoring Cybele/Kybebe in the city.²⁴⁶ As early as the sixth century BC, Cybele presided over the Lydian gold-refining area at Pactolus North (probably in her role of protecting metals), and her altar (which was somewhat later) in that precinct was restored as adorned by four lions.²⁴⁷ Further evidence for the Lydian Kuvava, with greater architectural sophistication and iconographic variety, is provided by a marble shrine found built into a wall in the Sardis synagogue (Sardis inv. S63.51; Manisa Museum, inv. no. 4029). Dated to ca. the mid-sixth century BC, it represents the goddess in high relief, standing, in the Archaic tradition, in front of an Ionic shrine and perhaps holding her lion cub. The sides and the back of the marble shrine are articulated by proto-Ionic half-columns, between which are tiers of low reliefs representing processions and activities associated with Greek Cybele and the Lydian Bacchus as well as lions.²⁴⁸

The argument for the worship of Cybele and the presence of a center city cult shrine in Sardis during the Hellenistic era is bolstered by the discovery under the cavea of the Hellenistic theater of a large number of terracotta figurines of the goddess, many accompanied by her lions. Molded in unmistakable Hellenistic style, these figurines offer, in the words of F. Gallart Marqués, “a compelling testimony to the lived cult of the goddess [Kybele] at Sardis” and the emergence of a new, Hellenized Cybele mirroring Sardis’s change from a Persian-controlled Lydian city to a Hellenistic polis.²⁴⁹ These finds underpin the durability of Cybele worship at Sardis, and the geographic spread of her cult and image.

If these figurines formed thematic groups illustrating cult ritual and activity, we are not aware of it, and they were probably votive objects at a shrine. However, another stele discovered during excavations in the Sardis synagogue, in 1968 (Sardis inv. S68.6; Fig. 3.66), that depicts Artemis and Cybele standing next to each other could represent a meaningful relationship between the two kindred deities and may possibly offer a clue to the overlapping identities of Artemis and Cybele. In this relief Artemis, on the left, holds a fawn while Cybele, slightly shorter, holds a lion cub. The goddesses are dressed alike and in similar, if not identical, poses. The pairing of the goddesses, approached by two worshippers, underscores the sense of physical, religious, and thematic closeness of the two deities but also their separateness—or at least, that is how Hanfmann and Waldbaum, who dated the relief to ca. 400 BC, saw it.²⁵⁰ The two goddesses, whose iconography was occasionally conflated in Anatolian religious contexts, have separate bodies, attributes, and identities, but they wear the same Greek dress and polos, or tall crown depicting city walls. Following the discovery of this relief, Hanfmann revised his original belief in the merging of the Archaic Cybele with Classical Artemis, reasoning that: “Our relief proves that [the goddesses] were different [i.e., they were not amalgamated].”²⁵¹ However, noting that the two small (male and female) figures of worshippers are also a part of the composition and share the same obviously sacred space under the same pediment (typically the shorthand for a temple or naos), he wondered whether the deities could have been worshipped in the same temple²⁵² and concluded against such a likelihood solely because there was no epigraphic indication that they were; inscriptions in Lydian and Greek that come from the temple mention only Artemis, not Cybele.²⁵³

Discovered in 1913, a monument north of the Artemis temple consisting of two lions (one seated and one recumbent) and an eagle, set on a broad, plain base between a pair of taller marble pedestals (probably once holding statues), invites a further consideration of the Artemis–Cybele relationship (see Figs. 3.67, 3.68, 3.69).²⁵⁴ One of the pedestal bases carries a bilingual inscription in Greek and Lydian (not necessarily belonging to the original base, the lions, or the eagle) mentioning a dedication to Artemis by one “Nannas, son of Dionysikles.” The inscription was given a date in the late fourth or early third century BC.²⁵⁵

The lions and eagles, dated to ca. 520–500 BC, were likely re-composed as a group during the Roman period.²⁵⁶ It is possible to know neither the original composition and the elements of the Archaic group nor its putative re-composition during the Hellenistic, Roman, or late Roman period.²⁵⁷ What we do know is that lions are specific and quintessential attributes of Cybele, though occasionally shared by Artemis. They also—for many cultures across the globe, including the Lydians—serve as symbols of power and kingship.²⁵⁸ Reasoning backward, the distinct presence of the lions allows one to argue for some link to Cybele in the earlier and later lives of these statue groups, and also supports the proposition that along with her lions Cybele could have been included in the arrangement.²⁵⁹

Although one of the pedestal bases of the so-called Nannas monument carries a bilingual dedication of Nannas to Artemis, the likelihood that Cybele’s cult was still honored in the sanctuary, along with Artemis, during the Roman period (or revived in the late Roman period) deserves serious thought.²⁶⁰ I concur with the view that Artemis and Cybele did not have a joint shrine in the temple, but it seems to me that a subtle conflation of the two (at least in the minds and hearts of the worshippers, like the ones shown on the stele) is possible.²⁶¹

For Hanfmann and scholars of his generation, writing five decades ago (and ostensibly “revising” the assimilation theories of an older generation of scholars, such as Radet), the insularity and separation of Artemis and Cybele (at least in the interpretation of the limited but crucial evidence provided by the two-goddesses relief [Sardis inv. S68.6]) became paramount. Despite Hanfmann’s assertion that Cybele and Artemis were separate goddesses at Sardis, I feel that it is worth returning to the question of the fluid relationship of these two deities in the light of recent and emerging sensibilities about identity and the perception thereof. The intervening years have placed us in a more flexible position to view questions of selfhood and identity in changeable, porous ways. Such multiviewing and multinaming of deities with overlapping selves would only strengthen their cults and expand their popular worship. A marble relief from Lydia (possibly from Kula, now in the Louvre; Fig. 3.71) shows a tall goddess in the center flanked by a pair of crouching lions, their paws resting on bulls’ heads; the goddess, wearing a polos, is crowned with a wreath by Nike, standing on her left, and attended by Demeter, pouring a libation at an altar, on her right—all framed within a large arch.²⁶² Decorated with crescent moons, snakes, and an eagle (recall the eagle of the Nannas monument), shared symbols all, the handsome relief is the product of just such a deliberate desire to offer a multi-viewing opportunity, and present mixed identities of Cybele and Artemis (to whom the stele is dedicated).²⁶³ For G. R. H. Horsley, who offers a critical assessment of Hanfmann’s argument for separate identities of Artemis and Cybele/Kybebe, this crude but expressive relief (dated to the beginning of the third century AD), despite its nominal dedication to Artemis, is an important example that illustrates the assimilation of Cybele and Artemis in Lydia during the Roman Imperial period.²⁶⁴

To return to the two-goddesses relief (which inspired my questions regarding identity), even though the two goddesses are “distinguishable” and have different bodies and identities, they appear to be worshipped in the same shared sacred space separately but, as the relief makes abundantly clear, together. Although Hanfmann took an uncharacteristically narrow view regarding the interpretation of the lions of the Nannas monument (believing that they could belong only to Artemis), he recognized the uncertainty of his position concerning the overall relationship of the two goddesses; he was willing to concede that, although no epigraphic evidence for a cult of Cybele has been found in the Artemis precinct, the two goddesses possibly were “ideally combined ” (my emphasis) for appeal to a greater community of worshippers.²⁶⁵

For us, to echo philosophical views prevalent in the early twentieth century (and later), the self can be described as “insubstantial” and “constructed” rather than natural.²⁶⁶ The constructed nature of body schema—which emphasizes conceptual roles of the self and the other as a game of changing places, or being in two places—may be the implication behind an exceptional representation of double Cybeles, or Cybele and Artemis, in a late Hellenistic votive relief with near-identical figures occupying separate spaces under the same pedimented votive shrine in the National Archaeological Museum in Athens (inv. no. 1540).²⁶⁷ They could be perceived to be the same in essence but different in name and notion, as illustrated by the two Artemises decorating the main alcove of cubiculum E of the Augustan Villa Farnesina in Rome: Artemis (Diana) the Huntress and Artemis the Moon Goddess = Selene/Luna/Anahita.²⁶⁸

In the same vein, insular and finite views tend to be antithetical to the protean nature of ancient sanctuaries, whose historical (though not physical) geographies emphasized the “betweenness” of place. The sacredness of sanctuary space is not invested in its permanence through exclusiveness and immutability, but rather through the opposite, inclusiveness, mutability and fluidity, and tolerance, whether for political or for spiritual reasons. Physical separateness of bodies viewed in a physical context, as in the two-goddesses relief—where each body is wed to its own physic, fenced by its own definition, opting for an “either/or” rather than “both”—can be overpowering. But in the world of religion and poetry, ancient or modern, bodies can share and fuse identities across time and place.²⁶⁹ Even when some forms of changeability and mutability (such as occupying the same space with different identities at the same time) may engender feelings of disbelief or discomfort in most of us trained in rational thought, we may be willing to accept a certain magic espoused by ancient religion and its magically inclusive settings. The acceptance of the Roman imperial cult as well as many earlier cults inside the sanctuary and the Temple of Artemis was no doubt smoothed and facilitated by just those inclusive contexts and traditions in popular as well as established religion.

Ultimately, it is hard to know if Artemis and Cybele (and a variety of other traditional cults: see below) could occupy the same space at the same time in the wonder of a Greek sanctuary. But, given the close and overlapping relationship between them and the overall religious culture of syncretism of the land of Anatolia and the Near East, that is what I suggest. To some, these changing selves and shared identities (if not supported by hard evidence) might be puzzling and contradictory, as they were for Alice—and we do not have the helpful advice of the Caterpillar about eating mushrooms, though it has been tried. We could remember that Lewis Carroll’s remarkable realm of mutable selves and meanings, disguised as Wonderland, actually reflected something deep and real about this world.

The Menogenes Inscription and the Sanctuary of Zeus Polieus and Artemis

An important inscription found in the Artemis temple in 1912 honors Menogenes, a prominent Sardian who went to Rome on behalf of the city and the koinon of Asia to congratulate the emperor Augustus on his grandson Gaius’s assumption of the toga virilis (5–1 BC). Two lines of the long document refer to “those dwelling in the Sanctuary of Zeus Polieus and Artemis.”²⁷⁰

As he was completing the uncovering of one great temple, the clarity and specificity of this passing message induced Butler to assert that there was “somewhere near another [temple] . . . waiting to be . . . brought to light.”²⁷¹ Much of the large area excavated in 1911–14 in successive terraces to the north and south of the Temple of Artemis was motivated by this strong desire. Work produced buildings and settlements in rubble stone, but no “Temple of Zeus.” Half a century later, the problem of the missing “Temple of Zeus” absorbed Hanfmann no less than it did Butler, and the precinct was explored sporadically from 1958 through 1972. Well aware that Butler had done almost everything possible in search of this elusive temple, Hanfmann was more interested in learning about the cult and worship of Zeus Polieus (“of the city”) than in discovering an independent temple (which the inscription did not name) in or near the Artemis precinct. He accepted that the sacred precincts of Zeus and Artemis were one and the same, and he believed that the cult of Zeus had been absorbed into the Sanctuary and the Temple of Artemis. Hanfmann’s dual-cult idea presented an ingenious and seemingly simple solution to Butler’s quandary and implied that Zeus might have been brought in to share the temple with Artemis.²⁷²

Hanfmann’s theory pivoted on the stylistic identification of the severely mutilated, fragmentary colossal head found in the south pteroma in 1911 (see Fig. 3.55 and pp. 154-155, p. 194 above). He viewed this sculpture as a Hellenistic work depicting Zeus, in whose battered image he recognized the features of Achaeus (rebellious brother of Antiochus III), the usurping general who held Sardis from 220 until 214 BC.²⁷³ In Hanfmann’s words, the cella was divided into two chambers and “it was to accommodate this portrait statue that the temple plan was redesigned [during the reign of Achaeus]. . . . The Zeus/Achaeus image was installed in the new east cella facing the Acropolis. Artemis continued to face the west, watching over the Necropolis.”²⁷⁴ In Hanfmann’s view, it was much later, under the Antonines, that the colossal statues of Antoninus Pius and Faustina were introduced into the temple; in effect, the imperial couple was received by Zeus and Artemis and shared the relative realms of their divine counterparts.

Recent studies of the temple refute this explanation (as well as the idea of Zeus sharing the divided temple with Artemis) on the strong evidence provided by technical details, construction methods, ornamental style, and historic probability. In concept, design, and execution, the pseudodipteral plan with peristyle columns, the crosswall dividing the cella, the opening of the east door, and the reorganization of the west cella and its new image base all belong to the Roman phase of the building, not the Hellenistic.²⁷⁵ Not many agree with Hanfmann’s stylistic identification of the colossal bearded head as Zeus/Achaeus. The contextual evidence speaks against this identification; the “Zeus” head was found in or near the temple along with six or seven similar heads firmly believed to belong to members of the Antonine family. Additionally, the head in question has been identified by R. R. R. Smith as Marcus Aurelius on specific iconographic grounds.²⁷⁶ Still, the severely damaged state of the head makes any identification a matter of educated opinion at best.

How can we reconcile the strong testimony of the Menogenes inscription (and other inscriptions found at or near the temple that also record the continuing presence of various cults of Zeus at Sardis: Lydios, Polieus, Baradates, and Olympios) for the association and perhaps conflation of the “sacred precinct of Zeus Polieus and Artemis” with the archaeological facts on the ground? There was no temple of Zeus inside the precinct of Artemis, and the Hellenistic temple of Artemis had a single cella until the Roman imperial period.²⁷⁷ The simple solution might be that the popular cult of Zeus Polieus was introduced into the Sanctuary of Artemis at some point during the late Hellenistic period and remained without an independent temple. However, we have no idea when this might have happened (the Menogenes inscription is Augustan). Less likely but still worthy of consideration is the possibility that the dual cult images of Artemis and Zeus shared the same temple cella—probably the same image podium—and were worshipped together.²⁷⁸ Two or more deities sharing a temple (without any architectural division) was not an unfamiliar scenario in Roman tradition.²⁷⁹

It is impossible to know what happened to the cult of Zeus Polieus when the temple was redesigned and became the center of the imperial cult during the Roman era, certainly by the mid-second century. It is unlikely, though not impossible, that the independent cult of Zeus (as opposed to some emperor’s identification with Zeus) was in some way conflated with imperial cult inside the crowded cella. In the cosmopolitan world of post-Hellenistic and Roman Sardis, religion could be inclusive and accommodating; the city’s sophisticated Lydian past and privileged position under the Persian satrapal administration may have made such heterogeneous religious practices unremarkable. Furthermore, conflating and overlapping sacred spaces and honors between the “old gods” and “new gods” was not a theological problem for the Romans.²⁸⁰ Apart from many other Greek and Anatolian cults, such as that of Cybele/Kybebe with its special relationship to the sanctuary, Sardis had several local, regional, and foreign cults of Zeus starting with its own venerable, Archaic Zeus Lydios, Levs. These cults tended to borrow rituals from each other and assumed each other’s identities over the passage of time due to changing perceptions, the pressures of political expediency, or mere mood and fashion.²⁸¹ In a focused study of Lydian and Anatolian cults, the late R. L. Bengisu, underlined the cogency of the cult of Carian Zeus at Sardis and observed that “Lydian identification with the cult of Zeus Karios provide[d] a broad base . . . regarding [its] applied relationship to the local cults of Zeus, Cybele, Artemis, Apollo and Mên.” She observed the continuity of these cults in the countryside, even though the broader acceptance of religious syncretism, characteristic of urban cultures of the Hellenistic (and Roman) periods, saw the decline of the Anatolian element at Sardis.²⁸²

There are several inscriptions dating from the late Hellenistic to the Roman era that bear dedications to Zeus “from the servants of Zeus” that attest to the continuing observance of some aspect of Zeus cult at Sardis.²⁸³ Apart from these, of particular interest is the so-called Droaphernes inscription, found on the east bank of the Pactolus, relatively near the temple; it is a second-century AD copy of a fourth-century BC Achaemenid original carrying a dedication of a statue to “Zeus of Baradates.”²⁸⁴ Neither the nature of this cult (except for its early date and Iranian origin), nor the name and identity of this deity are clearly known, nor is its presence in the Temple of Artemis established. However, the presence and continuity of Zeus cults at Sardis, even with a Persian “twist,” supports the larger reality of the continuing cosmopolitan, heterogeneous religious life and syncretism of imperial Sardis—a thesis supported by many scholars, including the late L. Robert and P. Herrmann.²⁸⁵

What is significant here is not so much the “linear continuity” of a particular cult, but rather the reception, sincere or superficial, of old and new ways of belief in Lydian, Persian, and a plethora of Anatolian cults and their overlapping identities, some of which were celebrated in their new incarnation through emotional rites and mysteries even under Roman rule. Thus, we do not have a simple cultural issue of “change” but a hermeneutic problem of the signification and interpretation of change—a change dictated by society at large, negotiated by its leaders and followers alike and subject to its unwritten laws and traditions. This is a change rooted in the memory-politics of the city, an “imaginary Sardis” shaped by its mythical past and remembered as it wished to be remembered.²⁸⁶ As pointed out by A. Berlin in her assessment of ceramic diversity in late- and post-Achaemenid Sardis, the strength of the system seems to have been rooted in the flexibility of cultural choices and variability in social, cultural, and religious frameworks.²⁸⁷

In a 1950 essay, Lionel Trilling, the prominent literary critic, mused that “literature is the human activity that takes its fullest and most precise account of variousness, possibility, complexity, and difficulty.”²⁸⁸ If we substitute “religion” for “literature” (a multiple view of religion and cult) in Trilling’s perceptive quote, we could come close to expressing the multiple and variable choices available to Sardis and its Artemis sanctuary throughout their memorable lives. While staying faithful to the core concepts and structures of past religions, the various and changing cults, primarily the cults of Zeus associated with the Temple of Artemis, must have been selected and scripted as relevant responses to the city’s civic and religious need to fashion an image of itself and perpetuate a plausible, flexible, and filtered memory of such an image.²⁸⁹

A relatively obscure and fragmentary inscription found at Sardis in 1914, dated around the middle of the second century AD (about the same time the temple was being rebuilt or had recently been rebuilt to accommodate the imperial cult), records a dedication in Greek by an association to honor a “priest of the Augusti and (hierophant) of the [imperial] mysteries.”²⁹⁰ This evidence is crucial not only for recording the existence of mysteries of imperial cult—well attested at other sites and in a variety of civic contexts—but as a testimony to the remarkable continuity of the mystery nature and structure of Sardian religions.²⁹¹ Could this underlying mystery structure—and this is a long shot—embedded within the “cult of the Augusti” have formed a backbone, a bridge, that facilitated the full-blown establishment of the imperial cult in the temple itself by the second century AD?

A central matter in our understanding of the Roman history of the Temple of Artemis is when Sardis was awarded its neokorate (“temple-warden”) status or honors, especially its second one.²⁹² Was it under Antoninus Pius, or slightly earlier under Hadrian? Recent evidence—epigraphic, literary, archaeological, and architectural—favors Hadrian (see below). It is a working hypothesis (and a well-published one) that the renewal of major Roman construction in our temple, including the division of the cella, represented a direct response to earning a second neokorate; the city needed to prepare Artemis’s temple for its new imperial guests.²⁹³

It is also logical that the choice of the city’s venerable Artemision as the shared seat of the imperial cult was triggered by more complex concerns about public skepticism regarding the nature of the imperial cult in the city and its relationship with Rome, rather than the simple exigency of providing an ample and appropriate space for the new cult in short order. Still, it would be economically feasible to use such an existing facility, which could, with reasonable changes, accommodate both divine functions. Those concerns bring to mind, at the everyday level, how and why the venerable goddess would vacate a good part of her temple for imperial newcomers, and in the process lose some of her prominence, at least architecturally speaking.²⁹⁴ For us, to have been able to eavesdrop upon the politics of the day, such as they might have been, would be worth far more than the masses of theories we could air and reams of sources we could quote.²⁹⁵

Considerations for an Imperial Altar

Another fundamental question related to both the architecture and the ritual of the new double-cella temple concerns the presumed presence, position, and usage of the altar for the imperial cult. If the great altar of the sanctuary served the west-facing cella of the goddess Artemis as usual, where was the altar for the imperial cult? Would one not expect a separate altar facing the east side—where none was found and no archaeological evidence observed?²⁹⁶

Dedication of altars to living and dead emperors was quite routine in Eastern as well as Western religious practice. Imperial cult liturgy typically included prayers, hymns, offerings of wine, ritual cakes, incense, and also the slaughter of animals, sometimes in large numbers.²⁹⁷ In Asia Minor the Hadrianic period was distinguished by its extraordinary enthusiasm in establishing altars for Hadrian, as well as cult worship and sacrifice, presumably at locations consecrated by imperial altars.²⁹⁸ In Roman practice, altars and altar ceremony could take place at major civic and institutional centers, and therefore imperial altars without temples were normal; but imperial cult temples without altars would have been unlikely, even unthinkable, as the altar was the focal point of imperial cult worship and ritual.²⁹⁹ Whether the Temple of Artemis at Sardis, one among Hadrian’s many imperial cult temples in Asia Minor, was a “neokorate temple of the koinon of Asia” or a municipal temple of the city gains significance in regard to the economics behind the gigantic, unfinished building. Hard evidence on this score is lacking—Burrell states that “this happy situation is rare or nonexistent”—in respect to all imperial cult temples.³⁰⁰

An important point to remember is that cult ritual and sacrifices “on behalf of the emperor” could and often did include other deities formulaically prefacing the declaration “to the ancestral gods and the lord emperors.” Or, as S. Price demonstrates with many examples, the wording could be ambiguous as to whether the offerings were to the emperors, alive or dead (sebastoi), or on behalf of them. Or, in rare examples, even the “imperial cult could be closely based upon a pre-existing cult of a god.”³⁰¹ This is a situation that could marginally explain the great altar at Sardis by assigning it a larger function, that is, receiving joint honors to Artemis as well as the sebastoi. Still, even if we imagine the dual use of the great Artemis altar, perhaps on special occasions, the routine but cogent need for an independent imperial altar for the newly established imperial cult—located somewhere at the east end of the building facing the imperial cult chamber, not one, however venerable, at the back of it—cannot be refuted.

A small altar without deep foundations (making it possible to easily disappear without leaving archaeological traces) is a plausible consideration. Such an altar would have been the first to go, completely destroyed, during the early days of Christian dominance. Unlike the lavishly decorated, massive Antonine altar at Ephesus (the “Parthian Altar”), and even smaller than the Domitianic altar (ca. 6.5 × 6.5 m) for the Flavian cult temple in the same city, this hypothetical Sardis altar could have been a simple but elegant marble structure raised on several podium steps.³⁰² Expanding on this hypothesis, it could have been conveniently and functionally located even in the wide open space within the east pronaos porch, in front of the steps of the east door.³⁰³

The First Neokorate of Sardis and the Wadi B Temple

Before we consider the question of Sardis’s second neokorate, under which mandate the imperial cult would have entered the grand rebuilt and reorganized Artemis temple, let us summarize the city’s first neokorate and an earlier temple understood to be the candidate for it.

We do not know under which emperor Sardis received its much-coveted first neokorate honors and its first provincial imperial cult temple. Since the policy was one temple per emperor, per province, until Hadrian, some emperors can be ruled out. The temples for the first three emperors—Augustus, Tiberius, and Caligula—were elsewhere. The periods of rule of Galba, Otto, and Vitellius were too short for the relatively lengthy procedure of establishing a provincial cult. Vespasian, Titus, and Domitian are possible, but the well-established pseudodipteral temple of Domitian in Ephesus, rising on a high, vaulted terrace, had subsumed the official Flavian cult.³⁰⁴ Trajan’s great temple, shared with Zeus Philios (and later with Hadrian), was in Pergamon. That leaves Claudius and Nero as possible neokorate honorees at Sardis, under whose reign an imperial cult temple may have been established.

The so-called Wadi B temple may be within our reach. During the 1981 and 1982 seasons, what turned out to be the northeastern corner of a temple was excavated on one of the steep northern slopes of the Acropolis, a commanding location in the center of Sardis, immediately west of the stadium (Figs. 3.72, 3.73).³⁰⁵ More work in the area of the temple and the large, rectangular, porticoed artificial terrace in front of it (named Field 55) in 2002, 2004, and 2005 revealed the northwest corner of the temple with three columns. Axially positioned in the center of the north side of the terrace, oriented north–south, a monumental flight of stairs gave access to what must have been a colonnaded, marble plaza surrounding the temple on three (east, north, and west) or possibly four sides. This arrangement was succinctly described by C. H. Greenewalt, jr., as an “axially-coordinated temple, terrace, and monumental approach [that] belongs to a precinct characteristic of Roman design, occupying a space of approximately two hectares [ca. 4 acres] in the heart of Sardis.”³⁰⁶

At its stylobate level, this temple can be reconstructed as ca. 23 × 39 m, with an interaxial distance of ca. 2.66 m, a plinth width of 1.35 m, and an interplinth distance of ca. 1.30 m, suggesting that its colonnades were based on plinth width.³⁰⁷ It is restored as a Corinthian pseudodipteros, probably octastyle with a pronaos porch of four columns. The architectural design and the size of the temple appear to be close to the Temple of Augustus and Roma in Ankara or the Flavian temple of Zeus at Aezane (Figs. 3.75, 3.76).³⁰⁸ The overall setting—with its large, colonnaded enclosure, approached axially and dramatically by monumental stairs—resembles the nearly coeval precinct at Aezane (Fig. 3.74).

Excavations at the temple and around its elevated terrace produced copious architectural elements, ornament, sculptural reliefs, and whole or fragmentary inscriptions, most of which must have belonged to the temple but were reused as spolia in later buildings during the late antique period.³⁰⁹ Noteworthy among them were some fragmentary and whole Corinthian capitals, one decorated by four muscular male torsos (one on each side) wearing lion-skins knotted over their chests, and with their arms lifted and holding lion cubs (or bulls), merging into the corner volutes of the capital—the figure is under study but tentatively linked with Herakles (Fig. 3.77).³¹⁰ There are also Attic-Ionic bases with guilloche and vertical, laurel-leaf torus decorations (much like the bases of the Artemis temple, though the latter has Asiatic-Ionic bases), sculptural reliefs, and floral frieze fragments, cornice pieces with modillions and lion-head spouts, and a fragment of the pediment tympanum with severely mutilated egg-and-tongue top ornament (Figs. 3.78, 3.79). Considering the use of massed, mortared-rubble construction in its foundations, refined ornamental style, and archaeological artifacts found in association with the original terrace, a date for the Wadi B temple can be suggested as at or around the mid-first century AD.

The original excavators had identified the Wadi B temple as a provincial imperial cult temple on epigraphic grounds.³¹¹ A number of the inscriptions typically honor priests and priestesses of Sardis, in the “Province of Asia and the Thirteen Cities,” strengthening the association of this temple with the imperial cult. A pair of matching architrave fragments, ostensibly from the temple but found among blocks reused for a foundation in Field 55, support this possibility. The cuttings for bronze letters on the architrave fragments have been read by Georg Petzl to refer to the Roman Senate.³¹²

In view of its imposing size, rich, high-quality ornament, and commanding position, elevated on an artificial terrace (as well as the possible reference to Heraclid decorative iconography), I concur with Ratté, Howe, and Foss, who proposed that this temple was associated with the city’s first neokorate; the emperors who were honored could have been Claudius or Nero, but less likely Vespasian.³¹³

The Second Neokorate of Sardis and Hadrian

An inscription on a statue base of Lucius Verus found in situ in the Bath-Gymnasium Complex (and likely to have been set up in honor of his return from Parthia in AD 166) indicates that Sardis had been awarded its second neokorate at the latest by the time of that victorious emperor, though the point is moot since I am convinced that the privilege had already been granted under Hadrian (see below).³¹⁴

The first appearance on coins of the title “twice neokoros” dates to the time of Septimius Severus.³¹⁵ We know, however, of an inscription of the mid-second century that indicates that the title could have been conferred earlier. Its text refers to Lucius Julius Libonianus, who was the “high-priest of Asia [and] of the temples of the Sardians in Lydia” (I emphasize the plural, which almost certainly must have included the Temple of Artemis), and who had served as strategos during the reign of Trajan.³¹⁶ Assuming that Libonianus’s position as the high priest of the Sardian temples, which would have included the imperial cult temples, could not have lasted more than twenty-five or thirty years, a time during the later years of Hadrian (AD 130–38) becomes likely.

Support for the theory that the second neokorate of Sardis was awarded under Hadrian is provided by an undated and fragmentary inscription from Field 55 that mentions the city of Sardis as the “keeper of two Koinon temples of [the] Augusti [by virtue of the] decrees [of the sacred Senate.” In his discussion of this text, G. Petzl refers to an inscription of the time of Hadrian from Hierapolis, recently published by Tullia Ritti, that lists Lucius Julius Libonianus as “High Priest of Asia, of the temples which are in Sardis.” Ritti concludes that in the time of Hadrian Sardis already been granted its second neokorate, and Petzl agrees that this is possible.³¹⁷ The city’s third neokorate came under Elagabalus ca. AD 221.³¹⁸

The presence of numerous Antonine cult portraits in the temple may still provide an argument for an Antonine rather than a Hadrianic cult. However, this argument is not nearly as strong, considering Hadrian’s almost certain visit to Sardis and the probable identification of Sabina and Hadrian among the fragments of colossi. Further consideration is due to two epigraphic hints that Sardis received the honor early in Antoninus Pius’s reign: a dedication to the emperor after his death in AD 161 as “hero” that still uses the titulature from the beginning of his rule;³¹⁹ and another dedication that continues to name him Olympios, an epithet normally used for Hadrian, his adoptive father.³²⁰ Neither can establish it as fact. In the light of the newly published inscription from Hierapolis, discussed above, my preference for a Hadrianic date for Sardis’s second neokorate remains firm.

As demonstrated in greater detail below, I believe that the second neokorate title was granted when Hadrian and Sabina visited Sardis in AD 124 during their Asian trip of AD 121–25. It is significant that Antoninus Pius never visited Sardis or Asia Minor, nor was he generous with his neokorate grants. Along a parallel line of support, we observe that Hadrian’s hand is historically and stylistically visible in the unusual design, rebuilding, and decoration of the temple. Upon Hadrian’s death in 138, the title and distinction of the Temple of Artemis as the neokorate seat of the imperial cult would have continued under his adopted son and successor, and in due time other members of the Antonine family would have smoothly and properly crowded into the east-facing cella as partners in the Cult of the Augusti.

A striking parallel of a broadly based Cult of the Augusti, which was established under Hadrian and incorporated the Antonine family, is the group of the six spectacular colossi of the young Hadrian, Antoninus Pius, Marcus Aurelius, and their wives found in 2007 and 2008 in the imperial bath complex at Sagalassus—all clearly indicating the presence and importance of neokorate honors awarded early in Hadrian’s reign when the city was recognized as center of imperial cult for all of Pisidia.³²¹ In their last incarnation, early in the fifth century AD, the six colossal images of Sagalassus were displayed neatly inside the six niches of the southern arm of the great ambulacrum-frigidarium of the bath, probably brought there from a separate honorific hall in the same complex (the so-called Marmorsaal); by then, they may have been objects of a lingering civic devotion rather than holding any active association with the imperial cult.³²²

In the next section, I would like to explore the subject further in connection to Hadrian’s probable visit to Sardis during his Asia Minor trip of AD 123–24 and his particular relationship with the Lydian metropolis.

Hadrian in Asia Minor

Cogent support of the proposal that Sardis was granted its second neokorate by Hadrian can be found in a review of the emperor’s special relationship with the cities of Asia Minor, particularly the city of Sardis and its Sanctuary of Artemis.

No emperor traveled more extensively in Asia Minor, visited more cities, or entered into more intimate and effective relationships of patronage and partnership in promoting their interests and well-being than Hadrian. The roster of temples, religious structures, and civic monuments that were started, completed, or restored under Hadrian is unmatched. O. Gülbay, author of a comprehensive study of Hadrian in Asia Minor, broadly identified two engines of a religious and administrative nature that provided the conceptual underpinnings for this exceptional relationship: first, the revival and promotion of Panhellenia as a privileged association for Greek cities (to which no less than thirty-eight Asian cities, including Sardis, belonged); second, the birth and expansion of the Second Sophistic movement in Asia Minor, reflected in the competitive promotion of philosophy, art, architecture, and urbanism.³²³ Both of these driving forces had close and fundamental ties to the promotion of imperial cult and its neokorate temples as the focus of religious and urban life.

Aware of its full political implications, Hadrian was the chief promoter of the cult and also the first to break its restrictive rules (“one temple per emperor, per province”) and establish more neokorate temples in Asia Minor than any emperor before or after him.³²⁴ Although the closest personal ties that the philhellene emperor formed were with Smyrna, Ephesus, and Pergamon—whence came the leading philosophers of the Second Sophistic, such as Polemon—Sardis, a privileged Panhellenic city with a kingly past and venerable, quasi-romantic associations, a city literally under the shadow of the Tmolos Mountains, could not have escaped Hadrian’s attention and benefaction.

Hadrian traveled widely in Asia Minor in AD 123–24 (his so-called First Grand Tour was 121–25) and 128–29, and he granted many neokorate temples dedicated to him “as god,” including the behemoths in Cyzicus and Tarsus, as well as the temples in Smyrna, Ephesus, Nicaea, Nicopolis, and Sagalassus. Smyrna and Ephesus were granted their second neokorates under this emperor.³²⁵ An inscription from Sardis documents the resolution of the “Sacred Guild of Artists” and honors Dionysus and Hadrian as the neos Dionysos.³²⁶ An important discovery in 2000 at Sardis was a tall, inscribed marble statue base carrying an eighteen-line dedication to Hadrian in Greek “from the Council and [the People] of the Sardians” and naming the superintendents and the chief strategos of the city. This is a crucial addition to other, already strong evidences of the emperor’s travels through Lydia (as I argue) in AD 123–24. The dedication and the statue almost certainly must have been occasioned by his visit to Sardis, a statement supported by the late P. Herrmann.³²⁷

To this must be added a coin from Saittae (modern Kula, ca. 40 km east of Sardis, on the Royal Road) that was struck to commemorate the emperor’s visit, depicting a laureate bust of Hadrian on the obverse and on the reverse a standing Hadrian in toga, shaking hands with Tyche, the personification of the city, wearing the mural crown.³²⁸ There is also an inscription honoring Hadrian and Sabina from Caesarea Troketta (a small hamlet near Turgutlu, ca. 30 km west of Sardis and directly on the Saittae–Sardis–Smyrna road), making it an almost foregone conclusion that because Hadrian passed through Saittae and Caesarea Troketta on the way to Smyrna, it would have been inconceivable for him to bypass Sardis.³²⁹

New studies of Hadrian’s Asiatic itinerary in 123–24, especially a new reading of the Arabic translation of Polemon’s Physiognomica—in which Sardis is included in the itinerary (named as-Srws, “the Sardis”)—also make it all but certain that Hadrian visited Sardis in his travels through Phrygia and Lydia. After he crossed from Spain to Asia, Hadrian traveled from north to south (from Phrygia to Lydia) with minor detours to Thyatira (modern Akhisar, northern Lydia) and Saittae (Kula, northeast of Sardis). Coming from the north, it is at Saittae that he would have connected with the Royal Road, visited the wondrous volcanic region known as Katakekaumene, and traveled a short distance west to reach, inevitably, Sardis.³³⁰ The path from Thyatira to Saittae by way of today’s Göl Marmara (Gygean Lake) would virtually pass through the middle of Bin Tepe, the ancient Lydian royal cemetery of “a thousand mounds,” described by the sixth-century BC historian-poet Hipponax; omitting this legendary monument would have been unthinkable for the erudite emperor.³³¹ From Sardis, Hadrian and Sabina probably turned southwest, up and around the slopes of mythical Mount Tmolos, the highest peak in the region (Bozdağ); this would have appealed to Hadrian’s curiosity about history and mythology as well as his well-known taste for climbing heights.³³² He then would have come down the northern slopes to the Hermus valley to visit Caesarea Troketta, and, then crossing the Belkahve Pass, he would have come down directly to Smyrna. The dedication to Hadrian and Sabina was found on the northwest slopes of the mountain, near the Smyrna road, and seems to support this itinerary.

It would have been fitting that Croesus’s golden city received neokorate honors, its second, from the historically minded, generous emperor. Cities of comparable size and importance, such as Pergamon, Smyrna, Ephesus, and Miletus, long had theirs—even distant Neocaesareia in the Pontus had its first neokoros temple under Trajan.³³³ On the other hand, it is all but certain that a cult center (though not necessarily an official temple) honoring Hadrian at the municipal level had been established in an unknown location at Sardis, because a Hadrianeion is mentioned in a stele reused as a water basin in the Byzantine Shops (Sardis inv. IN59.4).³³⁴ We should also keep in mind, however, that since the first neokorate honors for Sardis had been awarded before Hadrian, it would have been highly unlikely for another to be awarded under Antoninus Pius. Historical record shows that after the easy neokorate policy of Hadrian, the Antonine dynasty was tight-fisted in granting this privilege, preferring to confer the coveted title to one city per region.³³⁵ The visit of Hadrian and Sabina to Sardis, on their first grand tour, would have occasioned the second neokorate—as it did for Ephesus and Smyrna—and its grand establishment in Artemis’s own temple. Equally important, it would have provided the logic and reason for the colossal (but peripteros-less) temple’s grand rebuilding after centuries of relative inactivity.

Aside from the title neos Dionysos awarded to Hadrian by the “Sacred Guild of Artisans” of Sardis and the inscribed statue base, we do not know how Hadrian and Sabina were greeted at Sardis and what they did there. However, on this hypothetical issue we can start a new hare. We know that during his second Asian trip in AD 129, Hadrian and his entourage stayed in Ephesus for a long time—long enough to cause some victualing problem for the city, which received the unprecedented privilege of importing grain from Egypt and other gifts to make life easier for the Ephesians.³³⁶ During this long stay he was honored by a statue in recognition of his “unsurpassed gifts to Artemis. . . . [H]e granted the goddess rights over inheritances and deposits and her own laws.”³³⁷ Knowing that these major sanctuaries (including that of Sardis) were complex legal and financial entities, all of this sounds familiar. It is illogical to assume that Hadrian, known for his respect for local sanctuaries and religious tradition and his lavish gifts and privileges to the Ephesian Artemis, might not have done the same five years earlier for the Sardian Artemis and her sanctuary, one closely related to the Ephesian one.³³⁸

• 1Sardis II.1, pp. 81–82.
• 2Sardis II.1, pp. 27–28, 101, ill. 19.
• 3“[B]locks running against the Hellenistic column 79, not under it” (Cahill and Greenewalt 2016, pp. 478–80).
• 4Sardis II.1, pp. 101–4. How Hermogenes, whose work most likely falls in the late third to early second century BC, was responsible for the pseudodipteral arrangement of the temple—which also, according to Butler, was finished by the late fourth century BC—we do not know (for the most recent scholarship on the dates of Hermogenes, see p. 237, note 36). This is one of the several inconsistencies that bedevil Butler’s chronology in his handsome publication. The reader will find it difficult to follow the complex and numerous “phases” of construction, summarized on p. 101, such as: “The details of the present building . . . suggest that the earliest columns were erected not much before the middle of the fifth century and not later than the beginning of the fourth [BC]. They further indicate that some of the columns were erected in the latter part of the fourth century, and others at the end of the third or the beginning of the second.”
• 5Sardis II.1, p. 142.
• 6Sardis II.1, p. 87; Gruben 1961, p. 157.
• 7“Had there been any doorway or doorways between the two chambers [the ‘treasury’ and the ‘cult chamber’], steps would have had to be provided; but this wall . . . shows no evidence of a provision for steps” (Sardis II.1, p. 90, pl. A).
• 8Sardis II.1, pp. 85–87, ill. 97. Butler’s remarkable design for the temple’s west end, with sets of stairs and ramps incorporating the altar into one grandiose scheme, hardly noticed today, is too good to pass up without comment. As a chair-holding professor of architecture at Princeton University, Butler would have had ample familiarity with the study of historical architecture and Beaux-Arts design. We could start a hare running by pursuing another, even stronger source for Butler’s excellent design for the west end of the Artemis temple: his long association with the American Academy in Rome, a bastion of student exercises in academic classicism in the early twentieth century. Butler spent 1897–98 at the American School of Classical Studies in Rome as a fellow in archaeology. The classical school was closely connected to the Academy in those early years, and in 1913 it was moved under the same roof. Butler must have revisited Rome and the Academy during his Sardis years, and he must have been exposed to the Beaux-Arts-inspired student designs there. He was close to Gorham Phillips Stevens, an architect and the director of the Academy (a student of Charles Follen McKim, considered to be the founder of the Academy) and an avid supporter of academic classicism. In one of his Academy visits (probably 1913 or 1914), Butler even offered the Roman areas of Sardis to the Academy, reasoning that “it would be a fine thing to have one great site excavated and set in order by American enterprise.” Stevens was much interested in this little-known idea for a Academy–Sardis collaboration, but the Trustees of the Academy were not. Yegül 1991a, pp. 98, 231 n. 1.
• 9Sardis I.1, pp. 7–8, ill. 2; Sardis II.1, p. 91.
• 10We suspect that much of the rubble reported (and removed) by Butler was the Roman fill of the west cella (the original pronaos), placed there in order to raise its floor level; the red mortar above it would have been for the Roman paving of the cella. Sardis I.1, pp. 7, 52, 63–64; Sardis II.1, p. 13. A rough calculation indicates that a cistern covering the entire cella, even if the walls had been preserved only up to half their original height, would have had a capacity of 6,000–7,000 m3, far too large for the needs of what must have been a small Byzantine town. One could compare Butler’s cistern proposal to one of the large rectangular halls of the Baths of Diocletian in Rome (Hall XI, at the east corner of the bath block, 41 × 11 m), which was converted into a cistern (mainly serving the baths and its great natatio) during late antiquity. With walls covered by hydraulic cement up to five meters high and strengthened by semicircular buttresses, this cistern had a capacity of 2,000–2,200 m3, considerably smaller than the hypothetical Artemis temple cistern; Candilio 1999, p. 55; Lombardi and Santucci 2014. Furthermore, Cahill rightly observed that “since the east door wasn’t blocked or sealed (as visible in Butler photographs), it would have made a lousy cistern.” Indeed it would have (personal communication, May 1, 2017).
• 11Hanfmann and Frazer, Sardis R1, pp. 54–87; see figs. 120, 153–56, illustrating plans and sections of these trenches.
• 12Hanfmann and Frazer, Sardis R1, pp. 75, 81. Despite the clear lack of evidence inside the temple and its precinct, Frazer continued to believe that an Archaic temple with a dipteral plan resembling the early Artemision at Ephesus had been achieved at Sardis, and that an early Hellenistic temple following the hypothetical Archaic one had been built on its foundations. He postulated that the cella was divided and the major pseudodipteral scheme begun during a major Hellenistic rebuilding ca. 220 BC.
• 13Sardis R1, p. 75.
• 14On Achaeus’s rise to power in western Asia Minor, see Polybius 4.48.10–12. On his eventual defeat by Antiochus III after a two-year siege of the citadel of Sardis, see Polybius 7.16.6, 8.21.9.
• 15For an overall review of recent arguments on the “early dating” (ca. 220–190 BC) of Hermogenes’s active period (traditional “late dating” is ca. 175–150 BC), see Kreeb 1990. The majority opinion, however, favors the early second century, ca. 200–175 BC. See Bingöl 2012b; Bingöl 2013. See also p. 237, note 36.
• 16Gruben 1961.
• 17Gruben 1961, pp. 167–70, fig. 2.
• 18In fact, the northwest corner of these walls displays a monolithic, L-shaped corner block; see Figs. 2.53, 2.56.
• 19Gruben 1961, pp. 191–96. Gruben’s hypothesis concerning the three main phases in the temple’s history is largely accepted by scholars today and reproduced in surveys and popular guidebooks, such as Akurgal 1978, pp. 127–31.
• 20Howe 1999, pp. 208–9, fig. 11.7.
• 21In his stylistic judgment Gruben was keener than Butler and Gisela Richter, curator at the Metropolitan Museum of Art, who had both assigned these capitals to the fifth century BC; Gruben 1961, pp. 170–75; Sardis II.1, p. 128; Richter 1930, p. 321. We now know that Gruben was correct, because these capitals (as well as the medium-size capitals E and F) are the ones which feature the Ionian-Carian lewises, a type that originated during the Hellenistic period and was not favored by Roman builders (see pp. 121-123).
• 22Gruben, who was unaware of two other colossal male bearded heads found in or near the temple, did not consider the likelihood of the full incorporation of the imperial cult in the temple in the light of the neokorate honors received by Sardis in the second century (see pp. 215-217 below). He defined the temple during its Roman phase simply as the “Doppeltempel für Faustina und Artemis” (Gruben 1961, pp. 159, 195).
• 23Gruben 1961, pp. 181–82, 193.
• 24Gruben 1961, pp. 184–91. The conformity of the actual dimensions of the temple to measurements according to Gruben’s “chosen” Ionic foot of 29.52 cm is imprecise. For instance, Gruben proposed an interior width of 60 Ionic feet for the pronaos/opisthodomos. This dimension is not represented in the building at all. Measured at the perfectly preserved opisthodomos, the width between the projecting moldings of the anta piers is 17.58 m, and that between the anta walls is 17.96 m. Hence, Gruben takes the “perfect” number 17.71 m as the theoretical average between these two measurements (or, one might be able to “find” this figure if one chose and measured from the right pair among the multiple projecting bottom moldings of the antae). Likewise, Gruben accepted the ideal column height of 17.73 m to produce a column 60 Ionic feet tall (or near enough), but the actual height of our columns is 17.87 m, which is 60.53 Ionic feet, or 13.8 cm more than the desired number (conversely, using a 29.68 cm foot would give the exact 60 foot column!). As observed by Hanfmann, Gruben could be excused for not taking perfectly accurate measurements during his short and clandestine stay at Sardis; “his factual data must be used with caution” (Sardis R1, p. 75 n. 17). Indeed, many of our dimensions measured by total station are slightly different from his. It is an irony inherent to all metrological studies that a few centimeters’ difference—a multiple-molding profile, a wall face slightly out of the vertical, erosion, etc.—can make a mockery of one’s theoretical system, putatively based on the discovery of an ancient measurement unit or module claimed to be accurate within a few millimeters.
• 25Hoepfner 1990, fig. 2.
• 26Tumay Asena, “AT 96.1 Final Report” (Sardis Expedition field report), p. 4; Yegül, “Temple of Artemis Final Report, 1996” (Sardis Expedition field report), p. 8; Yegül 2012, pp. 100–101.
• 27Such as the representation of the plan of the Temple of Artemis at Sardis in Schulz 2012b, p. 167, fig. 1, in the very volume where my article, which illustrates why the Sardis temple was a not a traditional four-column prostyle temple, appeared! See also Schulz 2010, p. 89, fig. 89; Yegül 2012.
• 28Howe 1999.
• 29Howe 1999, p. 201.
• 30Ratté, Howe, and Foss 1986, fig. 7.
• 31Howe 1999, p. 210. Presenting a series of papers before scholarly audiences in the mid and late 1990s, this author independently arrived at the basically two-phase—Hellenistic original and Imperial Roman—solution.
• 32See the essays on Hellenistic Sardis in Berlin and Kosmin 2019.
• 33Arrian, History of Alexander 1.17.6. Butler, who wished the Temple of Artemis to be much earlier than the time of Alexander (Sardis II.1, p. 103), assumed that Arrian’s silence on the subject indicated that the temple was already finished; of course, one can assume that it was not even begun.
• 34Sardis R1, pp. 81–87; Cahill and Greenewalt 2016, pp. 491–92; Cahill 2019b, pp. 23–28.
• 35Polyaenus, Strategemata 4.9.4.
• 36For perceptive views on the traditional “Great Man” theory and approach to history predominant in the late nineteenth century (as opposed to more contextual and processual approaches), see Hall 2014, pp. 210–11.
• 37On Sardis as an “inherited Seleucid capital,” see Kosmin 2014, p. 155. Kosmin is careful to point out that certain strengths of the “inherited capital” were left significantly unchanged under the new Seleucid rulers: Sardis “was not renamed, not moved to a new site, not redesigned or rationalized, not provided with new fortifications, not expanded to incorporate new inhabitants” (p. 192). Upon careful consideration, these similarities are, indeed, excellent policy; instead of wasting money on such massive changes, the new polis could use its resources for new religious buildings, such as a theater and a Greek-style gymnasium, and, yes, for undertaking the construction of the fourth-largest Ionic temple of the classical world. On Sardis as a new city under imperial Seleucid sponsorship (especially under Antiochus I and Stratonike), see Kosmin 2019, pp. 78–82. On Sardis as a potential regional capital, see Ma 1999, p. 36; Chrubasik 2016, pp. 53, 74; and Sherwin-White and Kuhrt 1993, pp. 180–84.
• 38Wiegand and Rehm, I.Didyma, no. 492A–C, lines 44–45; Magie 1950, pp. 121, 975. An important document indicating the sale of land by Antiochus II to his divorced queen, Laodice, was recorded on five stelae erected in five major sanctuaries, but the primary copy was inscribed and kept at the “royal archives at Sardis,” as had been instructed by the king. See Austin 2001, pp. 305–7, no. 185.
• 39The interrelationships as well as the rivalry between Ephesus and Sardis and their parallel cults of Artemis is dramatically illustrated in an inscription (the “Sacrilege Inscription”) found at Ephesus. The inscription is difficult to date (L. Robert suggested a date around 340–320 BC, and D. Knibbe in the early third century BC) and thus provides little help in dating the Sardis temple, since the Sardian cult must have existed long before the temple. It does underscore, however, the strong desire on the part of the Sardians to measure up to the Ephesians (who already had their renowned temple) and the additional incentive to appeal to Seleucus I, who had been generous to Didyma (and later, his son Antiochus and Stratonike, too), for a temple worthy of his new royal city. Knibbe 1961–63; Robert and Robert 1965, p. 155, no. 342; Sokolowski 1965; Hanfmann 1987; Sardis R1, p. 179 n. 15.
• 40Inscriptions from the Didyma temple refer to the Seleucid funding of the construction and indicate that work had begun no earlier than ca. 299/98 BC Apollo’s bronze cult image was returned to Didyma with the help of Seleucus I and with the early completion of the Naiskos (Pausanias 1.16.3, 8.46.3). The adyton and the twelve in antis columns of the pronaos were largely finished by ca. 250–230 BC, around the time the cella of the Sardian temple was completed, although, like the latter, it remained an “unfinished temple.” Work evidently continued well into the Roman Imperial era, as the pilaster capitals and the “griffin frieze” of the adyton suggest an early Imperial date, while the “medusa frieze” of the main order and the columns and sculpted bases of the outer peristyle must be Trajanic or Hadrianic. See Pülz 1989, pp. 17–46; Borg and Borg 2002.
• 41Rostovtzeff 1967, pp. 431, 439. For the cult of Antiochus I established in Ilion after his succession, and a cult of Seleucus I and other instances of Seleucid rule, see Austin 2001, nos. 139, 143, 151, and 182; Orientis Graeci inscriptiones selectae, nos. 219, 212.
• 42Rostovtzeff 1967, p. 486; for examples of sacred images of Hellenistic rulers sharing honors with the established deities in temples, see Ma 2013, pp. 81–82 (“the practice here shades into that of ruler-cult for the Roman emperor”).
• 43Rostovtzeff 1967, p. 439; Magie 1950, p. 93; Orientis Graeci inscriptiones selectae, no. 214; Milet I.7, p. 283; Robert 1937, pp. 450–53.
• 44Valerius Maximus 5.7.1; Plutarch, Life of Demetrius 38; Appian, Syrian Wars 59–61; Lucian, De Syria dea 17–18; Rostovtzeff 1967, pp. 438–39; Ogden 1999, pp. 119–25. While selfless, paternal love for an ailing son and heir is the popular explanation, historically, the reasons may also be practical and political; see Carney 2000, p. 171; Taşkın 2011, pp. 116–18.
• 45For Stratonike’s dedications at Bambyce, see Lucian, De Syria dea 17–19. For the postclassical tradition of representing the popular and romantic story of Stratonike in the arts, see Stechow 1945 and Stechow 1964. I am grateful to the late C. H. Greenewalt, jr., for sharing curiosities, speculations, and bibliographical sources on “faire” Stratonike, and especially for pointing out the delightful verse play in Turkish by A. Necdet, Kraliçe Stratonike (2000). See also Macurdy 1932, pp. 78–82; Connolly 1972, p. 30; Francis 1986.
• 46Macurdy 1932, pp. 79–80; Carney 2000, p. 219; Orientis Graeci inscriptiones selectae, XI.4, 415.
• 47Carney 2000, pp. 171–72; 218–22; Kosmin 2019, pp. 79–82. Recent interest in Stratonike in the context of Seleucid royal women is evident in the contributions in Coşkun and McAuley 2016 by Algamor, Engels and Erickson, Coşkun and McAulay, Harders, and Ramsay.
• 48Smith 1924, pp. 150–59; Orth 1977, pp. 124–25 n. 6; Austin 2001, pp. 240–41, no. 141, and pp. 194–96, no. 113. The same Babylonian source mentions that during this critical period the satrap from Babylon and his high-ranking entourage visited Antiochus at Sardis and stayed in the city for at least a year because of the king’s continuing absences due to his martial involvement with Egypt on the Syrian frontier. The satrap’s wiling away his time at Sardis might have been further motivated by his desire to be first on the spot to pay homage to the king after the anticipated victory. See Otto 1928, pp. 9 n. 3, 22.
• 49The Babylonian lunar calendar confidently establishes the year and month of the queen’s death through reference to the positions of the planets, stating on line 6 that in the year 58 SÄ (= 254 BC), in the month of Tishrih (tenth month = October), “Stratonike, queen mother, met her fate in Sapardu [Sardis].” For the transcription and a German translation of the inscription and the lunar calculations for the dating, see Kugler 1922, pp. 317–20. On the life of Stratonike, see Eusebius, Chronicle, in Schoene 1866, p. 249; Justinus, Epitome 28.1.2, in Seel 1997; Beloch 1923, 3.2.93; Cadoux 1938, p. 111. See also Tarn 1969, pp. 349–52; and Macurdy 1932, pp. 80–81. Tarn thinks that Antigonus Gonatas sailed from Macedonia to Asia in 253 BC to bring home to Pella his son’s (Demetrios II’s) bride, who “he received from the hands of his sister Stratonike, whom he had probably not seen for twenty-four years” (p. 349). On his way back from Sardis, Antigonus stopped at Delos and established two festivals and vase-foundations in his and his sister’s behalf, named the Stratoniceia (p. 352). According to this testimony, Stratonike should still have been alive in the summer of 253 BC, a date contradicted by the Babylonian records; Tarn must be wrong. The marriage of Stratonike II and Antigonus’s trip back to Macedonia by way of Delos might have taken place during the summer and fall of 255 BC, but more likely in 254 BC, shortly before the queen’s death. Acknowledging some historical conditions, Tarn actually allows some flexibility in these dates: not earlier than 255 and not later than 253. Antigonus’s voyage to “Asia” to receive his son’s bride allowed him to see his aging sister—for the last time, as it so happened.
• 50Macurdy 1932, pp. 38–39, 45–48; Arrian, Events after Alexander 1.26; Diodorus Siculus 18.23.3, 20.37.3–6.
• 51For Cleopatra’s life, and the period she spent at Sardis, see Carney 2000, pp. 124–28.
• 52Diodorus Siculus 20.37.
• 53Macurdy 1932, p. 47; Whitehorne 2001, pp. 57–69. Carney, echoing Macurdy sixty-eight years later, admires Cleopatra’s “toughness and daring,” which is “precisely why she was killed rather than married” (Carney 2000, p. 128). They feared her power as queen.
• 54Sardis VII.1, pp. 91–92, no. 86; Sardis I.1, p. 43. Four other marble balls of comparable dimensions were found in the sanctuary, bearing dedications to Artemis by her priestesses, from the second and early first centuries BC (Sardis VII.1, pp. 94–96, nos. 90–93). There seem to be no known parallels to such ball offerings from other sanctuaries. See also Bringmann and von Steuben 1995, pp. 297–98. For a more recent discussion and other proposed dates, see Bumke 2011.
• 55For an overly cautious view, see the work of P. R. Franke, who prefers to see the occurrence of the names purely as accidental and argues that there is no reason to assume that the Stratonike ball is a later copy rather than a second-century BC original like the other balls; Franke 1961, pp. 200–201 (on this issue see also the next note). Franke’s skepticism appears excessive. Stratonike might have made this dedication before she assumed a royal title, or the copier of our ball might have decided to omit her titles (perhaps influenced by the political climate of the times). In a statistical sense, how many Stratonikes whose fathers were Demetrius and grandfathers Antigonus were likely to have made such dedications at the Temple of Artemis? Strongly supporting this view are Orth 1977, p. 125, no. 6, and the opinions of the late C. H. Greenewalt, jr., as expressed during numerous discussions with this author.
• 56See Wilhelm 1912, pp. 314–15.
• 57Macurdy 1932, p. 81; Ogden 1999, pp. 119–25; Bagness 1976. For Stratonike’s “surprising degree of connection to the dynasty of her birth,” and continuous choices to portray herself as an Antigonid—even after she became a Seleucid queen (and perhaps indeed because of it)—see Carney 2000, pp. 164–65, 171, 305 n. 74. A marble statue base (now lost), which carried a dedicatory inscription to Arsinoë, daughter of King Ptolemy and Queen Berenice, simply refers to Stratonike without her titles or husband’s name: “Stratonike, daughter of King Demetrius, dedicates . . .” (Orientis Graeci inscriptiones selectae, no. 14). As documented and observed by Tarn (1969, pp. 349–53): “[I]t is not difficult to avoid the supposition that Stratonike considered herself rather as the daughter of her father than as the wife of her husband.” In a personal note to the author (July 1998), C. H. Greenewalt, jr., suggested that it “might reflect her personal preference not to be identified with her husband(s)”—wise words.
• 58Although Stratonike was in the forefront of powerful Hellenistic women patrons of cult, and spent a large part of her life as queen in Artemis’s city, she was never formally connected to Artemis, as she was to Aphrodite. It has been reasonably thought that a major goddess like Artemis, or Hera, was too serious, even threatening, to be intimately connected with. Aphrodite was a less forbidding deity, especially for women. Carney 2000, pp. 221–22; Tondriau 1948, pp. 12–13, 40; Neumer-Pfau 1982, pp. 55–60. See Coşkun and McAuley 2016.
• 59I will take the opportunity to write this in a footnote, which might allow the introduction of a provisional idea (that a main-line text might shy away from), I might even carry the battle into enemy territory, and suggest that if the inspiration behind building a major temple to Artemis in her beautiful Sardian sanctuary could be traced to a person or patron, why not Stratonike, the ambitious queen of the newly conquered Seleucid land and the first denizen of Croesus’s legendary metropolis? See also Yegül forthcoming.
• 60Sardis VII.1, pp. 1–7, no. 1; Prentice 1912. For a more recent assessment of the Mnesimachos inscription and relevant social and economic issues, see Atkinson 1972. See also Levy 1976; Zawadzki 1951. My thanks to N. Cahill for bringing the following more recent studies of this inscription to my attention: Billows 1995, pp. 111–45; Débord 1982, pp. 244–51; Descat 1985.
• 61An issue relevant to the question of the dating and the architecture of the temple is the meaning of the word neopoio, used twice in the text and commonly translated as “temple warden.” It is not clear if this office required the existence of an actual temple, or if what is indicated or implied, in a general sense, is a “sanctuary.” If it did not, the association of the original document with our temple proper would not be critical. Naturally, this issue does not affect the actual version of the inscription carved on the temple wall. See Sardis VII.1, pp. 59, 71–72; Swoboda 1888; Burrell 2004, pp. 55–57. See also Olmstead 1936, p. 247.
• 62In their first detailed study of the Mnesimachos inscription, Buckler and Robinson considered the carved document on the temple wall to be the original and suggested a date within Antigonus’s rule (Buckler and Robinson 1912, pp. 22–26). These authors revised their dating of the temple inscription to ca. 200 BC in their 1932 Sardis volume, Sardis VII.1, pp. 1–7, no. 1. The same view had already been expressed by Butler in 1922 (Sardis II.1, p. 106) and by P. R. Franke, who studied the temple in tandem with Gruben and proposed a date of 220–200 BC; Franke 1961. After his recent consideration of the historical and contextual evidence, G. Petzl also thinks that the “Mnesimachos inscription [as represented on the temple wall] could not have been carved as early as 300–290 BC” (personal communication, Sept. 22, 2014). Supporting ca. 200 BC date, see also Schuler 1998, p. 160; Robert and Robert 1952, p. 173, no. 143; and Supplementum epigraphicum graecum 15.741.
• 63For a direct assessment of the inscription, see Atkinson 1972, opting for 250–200 BC as the date of the carving of the inscription. Atkinson argued that the “Temple of Cybele” was burned by Antiochus III in 214 BC as punishment for the support given by the city to the pretender Achaeus, and that the present temple was built between 214 and 200 BC Atkinson’s purely hypothetical history of the temple, which is supported by no evidence in the field (no sign of burning at all), was challenged by Hanfmann in Sardis R1, pp. 180–81 n. 44.
• 64Gruben 1961, p. 192; Wiegand and Rehm, I.Didyma, no. 492A–C; Orientis Graeci inscriptiones selectae, no. 225; Austin 2001, pp. 305–7, no. 185; Sardis II.1, p. 104.
• 65These include the previously mentioned inscribed marble balls, similar to the one set up by Stratonike, recording dedications to Artemis by her priestesses; a cylindrical marble statue base in honor of one Iollas, a civic leader who also had received many other statues in marble and bronze, is dated to the second half of the first century BC. Sardis VII.1, pp. 50–51, no. 27; Buckler and Robinson 1912; Buckler and Robinson 1913a; Buckler and Robinson 1913b.
• 66Sardis I.1, pp. 74–76, ill. 71; Sardis XI, p. v; Sardis R1, pp. 76–81, with a tentative reconstruction plan (fig. 129) and sections (fig. 130); for a plan of the coin findspots, see fig. 128. See the useful comparison by D. Van Zanten (Sardis R1, p. 80 n. 32) to the image base of Athena Polias in the temple in Priene, where the image of the goddess and its base were not set up before the mid-second century BC, although Pytheos’s temple dates to 334 BC.
• 67Sardis I.1, p. 75, ill. 71; Sardis XI, p. v; Sardis R1, p. 77, figs. 133–34.
• 68Franke gave an upper date of ca. 190 BC, which is refuted by the independent studies of Newell, Seyrig, and Hanfmann, who prefer a date around ca. 220–210 BC. Franke 1961; Newell 1941, p. 187 n. 154; Seyrig 1963, pp. 23–24. See also Le Rider 1991; BASOR 166 (1962): 34–35; Noe 1937, no. 925; Sardis II.1, p. 108.
• 69Cahill and Greenewalt 2016, p. 495; Sardis I.1, 74–76; Sardis II.1, p. 108; Sardis XI, pp. v–vi, no. 223. On Lydian ashlar masonry techniques, see Sardis R5, esp. pp. 17–46.
• 70Primary sources for Greek and Roman construction techniques that illustrate some of the features discussed here include Orlandos 1966; Martin 1965, pp. 55–57 and esp. 48–51; Durm 1892; Dinsmoor 1913, pp. 8–9; Adam 1994; Bingöl 2004; Bingöl 2012a; Nylander 1966; Aylward 2009; Pedersen 2011; Inglese 2016; Inglese and Pizzo 2014; Bonetto, Camporeale, and Pizzo 2014, pp. 1–4. See also Marta 1986; Marta 1990.
• 71See Gruben 1961, pp. 160–70; for a graphic comparison of construction techniques of the Hellenistic and Roman periods (Technik I and Technik II), see fig. 2.
• 72For comparative and selective illustrations of Hellenistic and Roman construction techniques, as seen particularly at the western end of the south wall: connections between the southwest anta pier and column foundation 49, see Figs. 2.9, 2.10, 2.11, 2.12, 2.13; the southeast anta pier and column foundation 17, see Figs. 2.72, 2.73, 2.74; the northwest anta pier and column foundation 48, see Figs. 2.107, 2.153, 2.154, 2.155.
• 73Cooper 2008, pp. 232–34; for “polygonal foundation stonework” typical of the Temple of Artemis Agoria at Demetrias, see Cooper 2008, fig. 9.2.
• 74Sardis II.1, pp. 39–41; Sardis R1, pp. 82–87.
• 75An exploratory trench in front of the east door in 1972 (“trench 3”) showed four more courses of marble-block foundations below what is visible, resting on gravel bedrock on the east side, top at *98.20. This is the same depth and same type of construction as the rest of the east wall, the southeast and northeast anta walls, and the rest of the main walls of the cella. See Sardis R1, figs. 153, 155–56.
• 76Hanfmann and Frazer believed that the east wall had to be dismantled down to and including the course below the threshold in order to install the door: “Normally, both sides of the join [top and bottom] would be dressed in one operation. The upper course is later, having been laid to replace the dismantled courses” (Sardis R1, p. 181 n. 45).
• 77As an alternate, earlier interpretation of the construction of the east door, K. Frazer posited that although the original Hellenistic temple had been planned without an east door, a door was put in during what he and Hanfmann believed to be the “second Hellenistic” period (i.e., not Roman; see Sardis R1, p. 87).
• 78See Milet VII.1, pp. 96–98, pls. 54–56 (Heroon II, Miletus); pp. 98–110, pls. 69–70 (theater stage, Miletus); pp. 122–23, pls. 83–88 (Market Gate, Miletus); also pl. 114 (Library of Celsus, Ephesus); pl. 115 (Temple of Hadrian, Ephesus).
• 79Sardis II.1, pp. 55–57, 130–32.
• 80Weber 1966; Rumscheid 1995; Bingöl 1990. Mustafa Uz clarified the difference between the architrave crown ornament of the Roman Imperial period and the few pieces found of the Hellenistic one, featuring simpler, tighter, and flatter egg-and-dart and anthemion lotus-and-palmettes. He compared them closely to those of the Temple of Artemis at Magnesia; Uz 1990, p. 58 n. 34, fig. 7; also Uz 2013, pp. 60–98. See also Rumscheid 1994, pl. 18.3, 5, 6, pl. 19.5, 6 (Chryse); pls. 78.3, 79.1 (Magnesia); pls. 185.1–6, 186.6 (Teos).
• 81Dinsmoor, who followed Butler’s chronology and believed the Artemis temple to be essentially a Classical building, was baffled by the ornamental style of the east door. He admitted that the ornament did not appear to be Hellenistic and explained the discrepancy ingeniously but unconvincingly: “The enframement of the great east doorway has Roman profiles resulting from the repairs” (Dinsmoor 1973, p. 229). There is, of course, no sign of repairs or recarving of the jamb ornament or profiles. For the classification of cyma reversa ornament into four types, see Weickert 1913, p. 10; Strong 1953. The Lesbian cyma is Weickert and Strong’s type B.
• 82For visual comparisons, see Vandeput 1997, pp. 74–75, 199–202, pls. 20.1, 20.2, 21.1, 22.3. For the delicate, linear qualities of Hellenistic ornament in general, see Rumscheid 1994.
• 83For this large cavetto fragment decorated with palmettes and lotuses growing out of female heads, possibly belonging to the upper wall molding of the cella interior, see Figs. 2.316, 2.317; see also the examples from Didyma (see Fig. 2.318) and pilaster capitals from the theater at Miletus.
• 84For the Miletus pilaster capitals, see Milet VII.1, pp. 106–7, fig. 16, pls. 65–67. See also von Mercklin 1962, pp. 45–46, 250–51; Pülz 1989, p. 51.
• 85For the general characteristics of late Hadrianic architectural ornament in Italy and its creative synthesis with the eclectic Hellenism of Asia Minor, see Strong 1953.
• 86According to Stefan Pülz, the Roman ornament of the Sardis temple (mainly the door and its overhead consoles) generally belongs to the Trajanic–Hadrianic period, but specifically the Hadrianic ornament of Asia Minor—a judgment echoed by Vandeput, Kadıoğlu, and this author; see Pülz 1989, pp. 74–77; Vandeput 1997, pp. 74–76, 85–86. For a comprehensive and comparative study of ornament from Asia Minor during the Hellenistic period, see Rumscheid 1994; for a similar general study of the Imperial period, see Köster in Milet VII.1. See also Gülbay 2009, pp. 141–43, 189–93.
• 87One should be aware of the potential inequality in the nature of the comparisons between the mere jambs and the lintel of a door. However monumental, this is limited evidence when considered alongside entire programs of ornament from large architectural ensembles in several stories, such as theater facades, nymphaea, and temple entablatures.
• 88Ornament that deliberately harks back to an earlier style is not uncommon in Asia Minor, as observed by L. Vandeput: “The tendency to give well-formed ornaments, often similar in shape to the decoration of older monuments, at highly visible positions recurs in the same way on the [Hadrianic] door-jamb of the Temple of Dionysus at Pergamon and even on the Temple of Serapis at Ephesos” (Vandeput 1997, p. 76). For the Hadrianic dating of the door jambs from the Temple of Dionysus at Pergamon, see Pülz 1989, pp. 83–85; Strocka 1988, pp. 299–300, pl. 44.1–3.
• 89Kadıoğlu 2006, nos. 240, 309, 489, 508. For Sagalassus and Ephesus, see Vandeput 1997, pp. 64–77 (esp. 75–76), 199–205, pls. 20.1, 20.3, 21.1, 22.3, 25.2, 27.2, 85.3, 86.2. For the late Hadrianic (imperial cult) temple at Sagalassus, see Waelkens and Poblome 2011, p. 107; for Ephesus, see Wörrle 1973. See also ornament from the scaene frons of the theater in Hierapolis: Gruben 1961, pp. 172–73, pl. 91.1; Hörmann 1923–24.
• 90For Aphrodisias, see Vandeput 1997, pl. 74.4; for the door frames of the Hadrianic temple in Cremna, see Mitchell 1995, pp. 6–91, figs. 28–30. For Perge, see Vandeput 1997, pls. 106.3, 107.1; Mansel 1975a, pp. 83–89; Mansel 1975b. I believe that most of the architectural ornament from the Hadrianic Baths at Aphrodisias, now exhibited on site, belongs to a third-century renovation, while the example illustrated here is probably late Hadrianic or Antonine.
• 91I am grateful to Erhan Öztepe, the director of the Alexandria Troas excavations, for sharing with me his knowledge about this very interesting Hadrianic building and its impressive architectural ornament. See Schwertheim 2018, pp. 91–170, 179–252.
• 92This handsome nymphaeum, on axis with the major colonnaded avenue, was occasioned by the neokorate honors granted to the city by Hadrian in AD 118–19. It was built and dedicated to Hadrian in AD 129–32 by T. Claudius Piso, a knight and leading citizen who came from a long line of civic benefactors. Waelkens, Üner, and Richard 2017; Vandeput 1997, pp. 89–955, 209–11, pls. 38.1–.4, 40.2, 41.3; Waelkens and Poblome 2011, pp. 108–12. See my cautionary remarks about the nature of comparisons, p. 171, note 87.
• 93Waelkens, Üner, and Richard 2017, p. 449.
• 94The same coarseness of ornamental elements that assumes visual richness when piled together as overlapping elements of an entablature and viewed from a distance (as in aedicular facades) can be detected in the fully restored, splendiferous facade of the nymphaeum on the upper agora in Sagalassus (Figs. 3.25, 3.26), given a late Hadrianic to middle Antonine date by the excavators (as judged by Vandeput: “some of the motifs . . . even seem to presage Severan decoration” [Vandeput 1997, p. 105; see also pp. 100–105, 209–11, pls. 41.1–41.4, 46.1]).
• 95These consoles are among the features that prompted Vandeput to opt for a late Hadrianic date for the Roman phase of the Temple of Artemis at Sardis, a view shared by many other scholars: “Opinion differs concerning the restoration of this temple . . . some authors date it to the reign of Hadrian [e.g., Butler in Sardis II.1, p. 131, ill. 54; Pülz 1989, pp. 74–77; Gülbay 2009, p. 143], others under Antoninus Pius around the middle of the second century AD [e.g., Gruben 1961, pp. 195–96]” (Vandeput 1997, pp. 74–75 n. 143). These useful observations were written before Vandeput was aware of our more recent work on the temple and its ornament, which fully supports her Hadrianic preference. In fact, during a field conference at Sardis in 2016, M. Kadıoğlu opted for an earlier rather than later Hadrianic date, no doubt due to the finer quality of the Sardis door jambs, able to be viewed and admired at closer quarters than other dated Hadrianic ornament from Asia Minor, such as scaene frons or theaters (Miletus, Nysa). See also Koenigs and Radt 1979; Strocka 1988, pp. 299–303.
• 96For the north peristyle individual column foundations, see pp. 70-79; for the south peristyle individual column foundations, see pp. 80-86.
• 97For a comparative discussion of Greek and Hellenistic foundations, see Martin 1965, pp. 308–22.
• 98Cooper 2008, p. 232.
• 99In French the term is murets de liason; in Turkish it is ızgara temel sistemi.
• 100The system was particularly useful in the creation of terraces for colonnades and buildings on sloping ground. The terracing for the Sanctuary of Demeter on the Acropolis of Pergamon consists of four separate walls built parallel to the hill and partially connected by crosswalls. The spaces between the main walls and crosswalls were filled with compacted earth and gravel. See Dörpfeld, Hepding, and Kolbe 1910. See also Vann 1976, pp. 35–41.
• 101Özgünel 1990, fig. 4. For Didyma, see Didyma I.3, pp. 49–50.
• 102Humann, Kohte, and Watzinger 1904, pp. 39–40, figs. 27, 29. See also Martin 1965, pp. 310–12. For Claros, see Robert 1954b; Bean 1966, p. 194.
• 103AvP III.1, pp. 12–15, pl. 2.
• 104Butler believed that “in connexion with the extensive repairs upon the temple, an attempt was made to render secure the foundation of some of the columns by embedding them in concrete” (Sardis I.1, p. 111). See also Sardis II.1, pp. 106–7; Gruben 1961, p. 169.
• 105Howe observes: “In Asia Minor, lime-mortared rubble does not appear to have been used as a structural mass until Augustan times or later, and then, because of the poor quality of the mortar, concentrated loads, such as columns or arches, are taken by ashlar piers and the rubble serves as a stabilizing infill” (Howe 1999, p. 208). See also Waelkens 1987. Gruben mentioned that a similar technique of encasing block foundations in massive opus incertum has been used in the Hadrianic Pompeion in the Kerameikos in Athens. See also Gruben 1961, p. 169; Ratté, Howe, and Foss 1986, fig. 7.
• 106Ratté, Howe, and Foss 1986, figs. 4–5.
• 107“Sütun temelleri münferit olarak inşa edilmiş, araları moloz taş dolgu yapılmış, stylobat döşemesinin sütunlar ve duvar arasında kalan bölümü (pteroma) plaka taşlarla tamamlanmıştır [Column foundations were laid individually and filled in between in rubble stones; the paving of the pteroma was completed in paving blocks]” (Serdaroğlu 2004, p. 122).
• 108Gruben’s surprising error is noted by Howe (1999, pp. 208–9): “Gruben distinguished between two clamping and doweling techniques: a Hellenistic technique with hook clamps and a Roman phase with wing clamps [butterfly clamps] and single-canted lewises. This led him to conclude that of the peristasis columns only the foundations at the eastern end were laid in the Hellenistic period and that the lateral foundations and the west end were continued in Roman times. The actual distribution of techniques is in fact a little different: both hook and wing [bar and butterfly] clamps appear in the Roman work. Wing clamps and double-canted lewises were used in the courses below ground level, hook [bar] clamps above.” (Howe is slightly wrong here: double-sloped lewises are also used at or above ground level.) “This is why the eastern columns appeared to be Hellenistic in Gruben’s scheme, because they are the only piers finished to stylobate level, and hence have hook [bar] clamps. In fact, there is a point in the new (i.e., Roman) west wall of the cella [west crosswall] where one can see both hook clamps and wing clamps in the same course: the outer face, which was to be some two courses above ground level, has hook clamps; the inner face, which was to be below ground level, has wing clamps.”
• 109Alessandra Sulzer, who supervised the 2002 trench, observed that the rubblework was laid during the Roman period, partially into the Hellenistic fill and partially into the dense, undisturbed soil. See Alessandra Sulzer, “AT 02.1 Final Report” (Sardis Expedition field report). Butler’s state plan of the temple (Sardis II.1, atlas, pl. i) does not indicate mortared-rubble construction anywhere inside the east porch, nor any linking the six columns of the prostyle porch with the rubblework that connects the foundations of the eight columns of the east row. A straight, dotted line along the east edge of this colonnade marks the presumed eastern, exterior extent of the rubblework.
• 110This is comparable to the early Antonine Temple of Zeus at Euromos, as described above in note 107.
• 111For a detailed discussion of this conclusion, see pp. 236-238.
• 112Sardis II.1, pp. 43, 93–97 (quotes p. 97).
• 113Howe 1999, p. 210.
• 114Sardis II.1, p. 97.
• 115The monumental Beaux-Arts Grand Palais on the Champs-Élysées (still used as an exhibition hall and museum) was built for the 1900 Paris Exhibition by a consortium of architects under the supervision of Albert Louvet; the grand facade and sculpted pedestal bases, which represent Architecture, Sculpture, Painting, and Music, were designed by Henri Deglane. See Boyd 1900, p. 177; Iwarere n.d.
• 116This painting at the Harvard Art Museums (obj. no. 2012.62), a seventeenth-century architectural fantasy in a dreamy urban setting, reminds us that our rusticated pedestal column, similar in appearance, might have been conceived as an architectural capriccio for its time and purpose; even if it was not, over time it became so.
• 117K. Frazer and P. Stinson, “The Pedestal Blocks: Artemis Temple Bases, Nos. 53 and 54” (Sardis Expedition field report, 1993).
• 118To add to this roster, excavations at the southeast corner of the Synagogue and the Marble Road in 2014–19 revealed no less than twenty fluted drums of the temple reshaped as voussoirs for what has been named the Monumental Arch, some three kilometers away from the temple.
• 119Although the proportions would be wrong, the sculpted pedestal bases articulating the high podium of the early Antonine Temple of the Deified Hadrian in Rome come to mind.
• 120Sardis I.1, p. 106; Sardis VI.2, pp. 39–40, no. 21, pl. ix.
• 121Uncharacteristically, Gruben missed that his pedestal blocks were recomposed from original fluted column drums. He believed that these in antis columns had stood on high pedestals that were dismantled and brought forward to the positions they occupy now (columns 11 and 12) sometime in the second century BC (his “Pergamene period”). The move and reassembly of the fluted columns and their capitals might be plausible, but this cannot be so for their unique pedestals, which we know were created by recutting and reshaping discarded fluted drums (probably from the cella interior). Gruben 1961, pp. 166–67, 181. Gruben missed the telltale marks of cut-through flutes on the pedestal blocks. The creation of high pedestals (ca. 1.60–1.70 m in height) was a typical Roman solution—not a part of the Hellenistic building—in order to facilitate the reuse of the cella interior columns, which were shorter by about that amount. The original in antis columns must have been regular, full-height specimens.
• 122Sardis II.1, pp. 92–95, ills. 101A–B; see also Robertson (1954, p. 150), who followed Butler but proposed on structural grounds the possibility of “perhaps unroofed” east and west porches surrounded by columns. The similar but larger pronaos area of Temple “GT” at Selinunte is commonly judged to have been open to the sky (Sardis: 17 × 13.40 m vs. Selinus: 20 × 17 m). See Dinsmoor 1973, p. 100; Berve and Gruben 1963, pp. 428–29.
• 123In 1972, “trench 2” and “trench 4” aimed to locate the foundations of the hypothetical in antis columns of the east and west porches. See Sardis R1, pp. 82–83, fig. 120. In 1996, the area between the eastern anta piers was reinvestigated. It had been excavated down to the level of the temple foundations by Butler, who left no stratigraphic details of the excavation except to say that no structural remains and no column foundations were found, after “the entire space within the western portico was cleared out to a depth of 3 m or more, and then filled up again” (Sardis II.1, p. 27).
• 124The trench covered an area of 3 × 5 m along the north side of the southeast anta pier and column 17, with a 2 × 2 m L-extension toward the north (covering the area of the southern of the pair of the in antis columns, “lot 2,” where the Commodus head was found). This revealed ancient bedrock at *98.40, ca. 1.60 m below the surface. Tumay Asena, “AT 96.1, Final Report” (Sardis Expedition field report), pp. 4–5, 7–8.
• 125This is a roughly rectangular block with rounded edges split in two; see Ferhat Can, “AT 11.1 Final Report” (Sardis Expedition field report), pp. 3–4.
• 126Answering this question concerning the in antis columns was one of the primary goals of the 2011 excavation in this area and led to the following conclusion: “The deposit is somewhat heterogeneous but forms a single fill dumped from various points. . . . No deposits were found in the cutting other than the Hellenistic and Late Roman fills. We conclude that a single large pit was dug in the late fourth or early fifth century CE, following the lines of the earlier Hellenistic foundation trench. This pit must have removed the Hellenistic columns in antis, together with their foundations” (Cahill and Greenewalt 2016, pp. 486–87). While I respect the archaeological and stratigraphical interpretation of the experts, and the area between the antae did produce broken pieces of column shafts and capitals, I am not entirely sure that it could not have undergone at least some partial excavation and filling at a date earlier than this.
• 127Cahill and Greenewalt 2016, p. 486.
• 128Cahill and Greenewalt 2016, p. 486 and fig. 17. Asena’s trench in 1996 also identified a larger Roman pit that could have accommodated the southern column of the in antis pair; he did not find a wide cutting in the bedrock into which the column foundation was set, nor the remains of any fill that might show that the cutting was Hellenistic.
• 129Cahill and Greenewalt 2016, p. 487.
• 130Cahill and Greenewalt 2016, pp. 486–87. Among the few numismatic finds from this area is a coin of Constantius II Caesar (ca. AD 330) found in 1972, Sardis coin 1972.1001; see Sardis M7, no. 297.
• 131From the second to the fifth centuries AD, Sardis endured at least eight massive earthquakes, any of which could have affected the buildings there: Laodiceae in 138; Bithynia/Mysia in 160; Smyrna in 178/9; Aphrodisias in 240; Ephesus in 262; Nicomedia in 270 and 358; Nicae/Bithynia in 388; Cyzicus in 460; and again in Nicomedia in 478. From late antiquity to the end of the nineteenth century there were dozens more severe earthquakes in western Anatolia. Considering only the strongest, which would have destroyed buildings throughout Sardis and the Hermus valley, these were: Gediz in 1595; Meander valley in 1653; Izmir in 1688; Kemalpasa in 1850; Bursa in 1855; Menemen and Manisa in 1880; Banaz in 1887; and Aydin in 1899. See Ambraseys 2009.
• 132For this general information on clays, earthquake behavior, and soil engineering I thank earthquake specialist Dr. M. Çelebi, Senior Research Civil Engineer and Manager at the USGS: private communication, June 29, 2017.
• 133It is normal to expect that the stepped block foundations of the column offered partial support for the wet, mortared-rubble foundations (a form of opus caementicium) of the Roman stairs when the mortar had not yet set, but not after it had solidified. The prospect of a centuries-old thick, hardened mortared-rubble mass “sagging” or collapsing once its partial support was removed is unlikely. Like the relieving arches of Roman concrete construction, the ashlar foundation blocks would have offered “an active and useful part [only] during the actual construction . . . and to a diminishing degree during the period of its drying out. Once the concrete had hardened, however, they served no structural purpose whatsoever.” After that, it would be up to land movements and earthquakes to move, crack, or fail that slab of mortared rubble. Moreover, by my measurement there is no sag at all on the trimmed stone foundation blocks of the stair, which itself is supported by the mortared rubble. For the static behavior of Roman concrete (and the above quote), see Ward-Perkins 1977, p. 99.
• 134Cahill and Greenewalt 2016, p. 486.
• 135At this point one might recall the circumstantial evidence provided by the Roman construction marks (such as lewis holes) on top of the remaining southeast anta capital. These features could not have been there if these original Hellenistic capitals were not removed, recut, and replaced at some point during the Roman era, possibly in the process of removing the original architraves across the east pronaos porch that they carried (hence their in antis columns). This would make more sense if it were a part of a deliberate architectural scheme, not a fourth- or fifth-century dismantling of the temple (see above). N. Cahill, in private conversation, also alluded to the possibility of changing the cella wall heights at this time; possible, of course, if it was not too drastic a change.
• 136Ammerman 2017.
• 137Ammerman 2006, pp. 300–303; see also Greene 1982, p. 52.
• 138Our support from architectural comparanda, however, is thin. We can count no open pronaos–associated roofs from Asia Minor (to be separated from roof openings over the cellas, or hypaethral arrangements), and there is only one from Italy, where such deep porches are common, namely the compluviate roof over the front porch of the third- or second-century BC Capitoline Temple on the Arx at Cosa; Brown 1980, pp. 3–4, figs. 60, 63. One should also consider that in all but a very few cases, the nature of the evidence for an open roof is conjectural; roofs are usually not preserved. It is possible that some of the temples we know only from remains on the ground (their footprints) had open roofs. Indeed, it is the presence of a pronaos cistern that allows us to propose an open roof for the Capitolium at Cosa.
• 139What could possibly be the source or sources of such a mixed fill, amounting to less than one cubic meter? Sardis, like all western Anatolian cities, suffered many earthquakes and destructions. Destruction debris is, in some ways, an impediment to rebuilding. Some of this debris, which would have contained household pottery indiscriminately thrown in, might have been dumped at various locations across the city and its hilly perimeter as landfill. When mortared-rubble fills for foundation and floor platforms were in high demand during the second-century phase of the temple, this material could have been reclaimed, retrieved from whatever valley or gulley it had lain in for some seventy to eighty years. That this particular batch filling our small trench overwhelmingly contained early first-century material must have been the luck of the draw. See Can, “AT 11.1 Final Report” (Sardis Expedition field report).
• 140Can, “AT 11.1 Final Report,” pp. 9–10.
• 141The nature of this debris resembles the mortared rubble fill (or foundation platform or pad) that covers most of the east end of the temple, between the columns of the east peristyle and the pronaos porch. Although the platform fill is mainly rubble and mortar, the shallow trench excavated in 2012 between columns 3–4–10–11 produced some bricks and a few tiles similar in size and typology to those from the northeast anta trench. The discovery at Sardis of baked brick from late Hellenistic and early Roman contexts is important to the larger study of the history of baked brick in Asia Minor; it deserves credit and should be better known. As a general consideration, the use of baked brick, even specially designed bricks, for column shafts (round, half-, and quarter-round shapes), engaged columns, and capitals dating to the late third and second centuries BC is well known from Greek and western Greek locations too numerous to list here; see the useful discussion in Lauter 1986, pp. 53–56.
• 142The round bricks from this trench, reconstructed ca. 36–39 cm in diameter, are almost identical in type and size to those of the hypocaust pilae of the Hadrianic Baths in Aphrodisias, which vary between ca. 34–36 and 37–39 cm in diameter. Our round bricks display “finger holes,” which are details closer to the round bricks from independent deposits from Sardis and dated earlier than the hypocaust bricks from Aphrodisias. Yet “finger holes,” representing a fairly common and sporadic practice in Roman brick and tile production, is not by itself a “type” assignable to a date or limited to those earlier bricks from Sardis.
• 143This team of specialists included E. DeRidder, J. Hayes, S. Rotroff, U. Outschar, A. Berlin, and M. Rautman. The material comes primarily from AT11.1, lot 3, baskets 19, 21, 33. See Elizabeth DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons, August 2012” (Sardis Expedition field report), with especially useful tables that summarize the results and include different points of view in analysis and dating, pp. 4–6, 8–9, tables 2 and 3; Raubolt 2018, pp. 347–60. Taking the two carefully documented groups as a whole, “Finewares” (table 2) and “Lamp Fragments” (table 3), the overall dating spread assigned by the Sardis specialists covers ca. 25 BC to the mid-first century AD, with two sherds possibly dating as late as AD 125. In the tables, individual, nonconforming views are also noted.
• 144DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons,” pp. 3–6; Cahill and Greenewalt 2016, pp. 483–85, figs. 14–15.
• 145DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons,” p. 4 n. 3.
• 146Cahill and Greenewalt 2016, p. 501.
• 147This interpretation is presented as phase “Roman 1b” of the “Phase plans of the Temple of Artemis,” Cahill and Greenewalt 2016, pp. 496, 501–2, fig. 24.
• 148See DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons,” tables 2 and 3.
• 149As noted by DeRidder: “Outschar believes this a 2nd c example of the type” (“The Artemis Temple at Sardis: Three Ceramic Horizons,” p. 9, table 3, items 5 and 6).
• 150The primary evidence for the emergence of ROW lamps in late Augustan–Tiberian contexts comes from the “Hestia Hall” of the first phase of the Prytaneion at Ephesus; see Steskal 2010, pp. 99–100. Mixed in as these lamps are with various sigillate groups, finewares, and Ephesian lamps, the dating appears somewhat vague but acceptable. The dating of ROW, even determining what really constitutes ROW, is difficult (as noted by a Sardis ceramicist, “a niche within a niche”). The ROW lamps make a stronger showing in some of the rooms of House WE4 of Insula 2, “Slope houses,” of Ephesus in contexts believed to be late Augustan (Bauphase I). Of the twelve or so lamp fragments from these contexts, six display the ROW technique and are Ephesian made. Other ROW fragments have varying suggested dates in the second half of the first century AD; others come from destruction levels, probably second–third century, or even later. Ladstätter 2005, pp. 238–39, 251–53, 260, and catalogue entries 307–8. See, in the same publication, Thür 2005, pp. 96–97; see also Vetters 1970, pp. 97–100; Strocka 1977, pp. 43, 91.
• 151Scholars and specialists I consulted include: K. Slane, “ROW lamps should safely [be] regarded as a 2nd century type . . . [and have been found] in late 1st to mid-2nd century contexts in Corinth” (Slane to Yegül, Sept. 18, 2018); R. Meriç from Metropolis, Uyuztepe, and Izmir agora, “within all my contexts ROW appears frequently during Flavian–Trajanic periods, [only rarely] ca. AD 50–60, but none before that” (Meriç to Yegül, Oct. 1, 2019); H. Williams, “I have done a lot of work on ROW and . . . I am surprised to hear of any appearance of ROW in Augustan dating. I have never seen any earlier than ca. AD 50–60, but [they are] most commonly related to later chronologies, ca. AD 70–125” (Williams to Yegül, Oct. 6, 2018); B. Wohl, Greece, eastern Mediterranean, “normal and overwhelming context for ROW is Flavian and later . . . of the half-dozen examples from Isthmia, the earliest is late 1st c AD, others are early 2nd or 2nd to 3rd c” (see Wohl 2017, p. 81, cat. nos. 51–55); Ö. Özyigit, who informs me that although some ROW lamps could begin as early as the second quarter of the first century AD [“volute-type”]: “I do not recall any from Anatolia going back to the Augustan period” (Özyigit to Yegül, Oct. 24, 2018). See Özyigit 1990. See also next note.
• 152In an interpretive article on Early Imperial Roman pottery from Ephesus (Ladstätter 2007, pp. 211–12), S. Ladstätter describes the late Augustan period (especially related to Insula 2 of the “Slope houses”) as a time when the late Hellenistic variety and diversity in ceramic forms, types, and sources were being replaced by an expansion in local production and standardization, initiating the transition to the Roman period. According to Ladstätter, the emerging ROW technique in Ephesus during the late Augustan period followed this Ephesian experimental phase of lamp production, achieving a fine, glossy appearance; hence, the development of ROW during the late Augustan period at Ephesus was a part of this experimental effort. By the second half of the first century AD, production had spread to centers outside Ephesus in a more standardized, less creative manner. If this interpretation is correct (and I find it so), it would be unlikely that a type which started in Ephesus as a special, experimental production during the late Augustan period (and which spread to other production centers only later) should be represented at Sardis at the same time as its initial Ephesian production. It should be mentioned that Outschar, clearly familiar with the earlier dating of ROW, is not willing to state that her later dating of the ROW fragments is necessarily wrong; Outschar to Yegül, Oct. 26, 2018.
• 153Outschar to Yegül, Aug. 9, 2017. Also, earlier in private conversation: “I know them [terra sigillata B, thicker local imitations] only in second-century contexts” (Aug. 7, 2012).
• 154DeRidder sums up the difference: “Hayes believes this is likely to be an early example (first half of the 1st c AD, Atalante II, Form 19), while U. O[utschar] believes that this rim fragment is 2nd c AD [Atalante II, Form 60]” (DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons,” p. 6, table 2, item 6). For a wider study of chronological and geographical distribution of terra sigillata in the Aegean and eastern Aegean regions, see next note.
• 155Other Western, Italian terra sigillata fragments were assigned a wider chronological range within the first century AD (or “second to fourth quarters of the first century AD”); see DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons,” p. 6, table 2, items 8–10. After wide consideration of ESB (“Type 1”) in the eastern Aegean and western Anatolia, P. Bes asserts that “ESB remains (fairly) common terra sigillata in several Aegean regions until the mid-2nd c AD. This pattern, if it is a pattern, requires further study.” Bes reminds us that “we cannot simply ignore this [continuation of ESB both Types 1 and 2 into the second century AD], and consider this residual” (Bes 2015, pp. 74–75, 40–41, esp. charts 24, 56).
• 156This is apparent in the chronological leeway that is acknowledged and accepted in the ceramic identification and dating tables given in DeRidder, “The Artemis Temple at Sardis: Three Ceramic Horizons.”
• 157The sherds and construction debris that could be assigned a later (or earlier) date share the same structural context as the rest (larger, better-preserved, diagnostic pottery). As observed by one consulting field archaeologist, who also works in Asia Minor, “the size or completeness of the material or the majority represented does not matter, what matters is the latest datable material . . . that is the date, at least, terminus post quem” (B. Rose, message to author, July 13, 2018).
• 158Considering the varied and strong evidence for the second-century dating of the major Roman rebuilding of the temple, Cahill accepts the possibility that actual work on the superstructure of the east end columns, the division of the cella, the east door with its Hadrianic ornament, and the testimony of the inscription around column 4 (the “talking column”) might date to the second century, even though he holds that the foundations of the columns belong to the Julio-Claudian era. Cahill and Greenewalt 2016, pp. 501–2 n. 326.
• 159This proposal, first suggested by me in 2013 (Yegül 2014, pp. 214, 218–19), is now strengthened by the careful rereading of an important inscription from Sardis, IN00.4, by G. Petzl in Sardis M14, p. 65, no. 373. See also in this volume pp. 220-221 and pp. 236-238.
• 160Cahill and Greenewalt 2016, pp. 501–2.
• 161Special honors paid to Tiberius, including the title Kaisareis Sardianoi, reflect the city’s acknowledgment of the emperor’s generosity after the AD 17 earthquake—and possibly its desire for more generosity in the future. The title and the occasion seem to be directly modeled after the actions of Tralles, a city that renamed itself Caesarea Tralles in acknowledgment of the generosity of Octavian Augustus following the severe earthquake of 25 BC (Strabo 12.8.18). Other cities that assumed similar titles include Myrina in Mysia, Cibyra in Phrygia, Hadrianopolis in Paphlagonia, and many more. At Ilion, the people and emperors were “kinsmen”; in Aphrodisias, the letter of Augustus displayed in the theater virtually shouted, “You are my favorite city of all Asia Minor.” Such politically motivated honorific titles were common among the cities of Asia Minor and their use continued unabated into the third century. Sardis’s close ties to Tiberius is a manifestation of this pattern, occasioned, even necessitated, by Tiberius’s earthquake relief, but in no way does it indicate that they had any reason to dedicate their Artemis temple to him. Even if they had been successful in obtaining the neokorate they sought under Tiberius, it does not follow that the neokorate temple was going to be the Temple of Artemis. There are no grounds to imagine that Sardians were planning or preparing their temple for the imperial cult at any time before the second century. If indeed these honorific titles indicate a special closeness to an emperor, Hadrian was twice honored at Sardis as “Neo-Dionysus.” I am grateful to B. Rose and K. Rigsby for bringing some of this material to my attention.
• 162The inspiration comes from Billings Learned Hand (1872–1961), a U.S. judge and judicial philosopher active in the first half of the twentieth century, whose famous quote is: “The spirit of liberty is the spirit which is not too sure that it is right,” from the speech he gave in Central Park, New York, May 21, 1944.
• 163For a longer discussion of votive monuments and dedications from the sanctuary, including in the earlier periods, see pp. 231-232 below. A full discussion of the Artemis precinct during the first and second centuries AD can be found in Sardis R1, pp. 53–73. Butler reports that well over one hundred coins from the Imperial period were found in the temple: Sardis II.1, pp. 108–9. For Late Roman and Early Byzantine coins, including issues of Valens, Theodosius, Honorius, Leo I, Justinianus, and Heraclius, see Sardis I.1, pp. 43–44.
• 164See Cahill and Greenewalt 2016, p. 500 for a fuller discussion.
• 165Sardis II.1, pp. 106–7, ills. 106–7; Sardis VII.1, pp. 66–69, nos. 51–54, figs. 42–44; Buckler and Robinson 1913b.
• 166Sardis VII.1, pp. 65–66, no. 50, fig. 41; Nock 1930, p. 256; Sardis R2, pp. 164–65, no. 246, figs. 426–27 (Izmir Archaeological Museum inv. no. 582).
• 167Indeed, the toruses and the spinas (but not the plinths) of all eight columns of the east colonnade bases are carved from single blocks; the bases of the east and west pronaos porches are carved in three separate pieces: plinth, spina, and torus. In the case of column 4, the bottom piece of the shaft, 30.5 cm high, is also made from the same piece of marble—a block ca. 1.04 m high and averaging 2.30 m in diameter that could weigh up to 12 tons (and equaling 4.28 m3). The monolithic construction of column 7 (spina, torus, and bottom section of the shaft) is even more massive, at ca. 15 tons.
• 168ἡ σ[̣π]εῖ̣ρ̣α χ̣ὡ̣ [ῥ]ι̣ζ̣α̣ῖ̣ος εἷ̣ς̣ ἐ̣σ̣τιν λίθος / πρῶτος δὲ πάντων ἐξ ὅλων ἀνίσταμαι / οὐ δημοτεύκτων ἀλλ’ απ’ οἰκείων λίθων (with restorations by Buckler and Robinson); see Sardis VII.1, pp. 143–44, no. 181; Sardis I.1, pp. 110–11. In Buckler and Robinson’s reading, οἰκείων λίθων / oikeion lithon is translated as “given by friends,” signifying private donors. Supporting this reading, see also Rumscheid 1999, pp. 31–32. The linguistic refinements of the text required by the subtleties of the poetic form, rich in allusion, obscure certain readings; this inscription merits further study.
• 169A. Chaniotis to F. Yegül, July 4, 2013; Yegül 2014, p. 204, n. 1; Cahill and Lazzarini 2014, pp. 36–37.
• 170Waelkens 1997, p. 241. For a typically exaggerated description—of the fine imported marbles used in the Baths of Claudius Etruscus in Rome—see Martial 6.42; discussion in Yegül 1992.
• 171σῆμα δὲ, κεἰ τέχν[α] Φρύγιον λίθον ἔργῳ ἐλέ[γχει], | ψεύδεται· ἐγ γαίης τῆσδε πέφυκε λίθος (Corpus inscriptionum graecarum 4377, lines 11–12; Inscriptiones graecae ad res romanas pertinentes 3, no. 362). The pride, in both cases, is that of an intelligent, well-considered builder, engendered not only by the preference for a local resource over distant ones, but also by the skill a local mason would use to create an imitation of these distant, famous, expensive stones by using local (should we say) inferior ones.
• 172For a special study of this inscription as a column talking in the first-person singular (objet parlant), its literary, historical, and cultural allusions, and significance, see Yegül 2014. On the nature of the objet parlant and the dynamics between the written word and voice and the identity of the speaking object and its audience, see Tueller 2008; Bal 1997.
• 173Herrmann’s opinion, unpublished but clearly expressed in a letter to the Sardis Expedition, reads: “Alle Buchstabenformen (Ѡ, Ξ, Ε, Η) sind im 1. Jhdt. n. Chr. möglich, die Häufigkeit besonders von Ѡ und Ξ nimmt aber im 2. Jhdt. zu. Ұ ist einmal von K. Buresch (A.M. 19, 1894, 104, anm.1) als Datierungsindiz für ca. 150–250 im Ausspruch genommen worden aber auch das ist zu relativieren; vgl. M. Guarducci, Epigraphica Greca I 383 Anm. 1 mit einem Beispiel aus dem 1. Jhdt. aus Sardis.” And he added that—contrary to Buckler and Robinson’s slight preference for an early first-century AD date based on the “stiffness and slenderness” of the letters—these characteristics rather point to the Hadrianic–Antonine period [“Abweichend von Buckler-Robinson wurde ich aber sagen, dass die etwas steife Form der Buchtaben wie auch ihre ‘slenderness’ eher für das 2. Jhdt. sprechen, also etwas hadrianisch-antoninische Zeit”] (P. Herrmann, “Sardis Memo, May 20, 1987”).
• Afteran independent viewing of the inscription, Hasan Malay, in support of Herrmann, wrote that “based on letter forms, especially ‘omega’ and ‘Xi’, I could not have assigned a date earlier than 2nd century CE” (H. Malay to F. Yegül, Sept. 3, 2007). It is interesting that neither Buckler and Robinson nor Butler opted for an early date in their first reading of the inscription. Buckler observed that the “’cursive’ omega” was “commoner in the 2nd and 3rd centuries than in the 1st” (Sardis VII.1, p. 144); and Butler judged it to be “in lettering of the 2nd or 3rd century AD” (Sardis I.1, p. 110). Later he somewhat inconsistently opined that the second-century inscription was “engraved upon an old base which was being reset” (Sardis II.1, p. 106).
• 174“Some letter forms, such as the Xi and the angular omega, are hardly compatible with a first-century CE date” (A. Chaniotis to F. Yegül, Nov. 6, 2012); and later: “the letter forms (though not always reliable), exclude the early date” (A. Chaniotis to F. Yegül, July 4, 2013). Taking the full historical and cultural context of the inscription, Chaniotis confirmed (during a conversation on Aug. 14, 2013): “[In fact] it is impossible before the second century CE.” I would like to extend my thanks to Angelos Chaniotis and R. R. R. Smith for sharing their specialized knowledge and insight concerning the epigraphic significance and implications of the column inscription.
• 175The meaning of the inscription, centered particularly on the translation of the word ἀνίσταμαι [ἀνίστημι] (anistemi), is crucial to the architecture and building history of the temple. Buckler and Robinson interpreted the word anistemi as “to rise again,” implying that column 4, and the other columns of the east front, were Roman reconstructions or repairs of the already-existing peristyle columns belonging to the Hellenistic period (they rose “again,” or a second time). This interpretation was accepted and encouraged by Butler because it supplied strength to the argument that the undeniably Roman work belonged to the reconstruction of the exterior columns from the existing but damaged parts of the “earlier temple,” which was probably undertaken after the devastating earthquake of AD 17—or “earlier 1st century” (Sardis II.1, pp. 142–143). This idea was readily embraced by Hanfmann (Sardis R1, pp. 75–76). The verb anistemi literally means “to stand up” but it could also imply either “to build for the first time” or “to rebuild.” There are a number of examples from the Greek period, some of which are rhetorical in nature, that support the latter meaning. In the factual, architectural context of an inscription on a column, however, the more direct and literal (and architectural) meaning of the word—as simply “to stand up”—should be favored, as Herrmann, Malay, Petzl, Chaniotis, and other epigraphists agree.
• 176For the iconography of the laurel wreath as a symbol of victory in religious and civic contexts of the ancient world, and its association with religious architecture, see Yegül 2014, pp. 205–10. See also Baus 1940; Maxfield 1981, pp. 67–96. For the use of the horizontal laurel leaf or the laurel wreath as a purely decorative motif on bases, see Alp 2008.
• 177Yegül 2014, pp. 210–13. One should underline the popularity of the type following Trajan’s column, such as the Column of Marcus Aurelius in Rome, and the columns of Constantine, Theodosius, and Arcadius in Constantinople, all with their toruses decorated by laurel wreaths wrapped in spiraling lemnisci.
• 178For a study regarding the perception of Roman engineering as a triumphal, victorious achievement worthy of celebration, see Favro, “Arch Triumphs,” forthcoming. See also Coarelli 2000, pp. 21–27; Packer 2001, p. 75. For the base inscription: Corpus inscriptionum latinarum VI.960; Cassius Dio 68.16.3.
• 179Tueller 2008, pp. 24–27. Interest in talking architecture seems to have been revived in late antiquity, as Christian monuments spoke of their deliverance from ignorant paganism; the Latin inscription on the tall marble base of the Obelisk of Theodosius in Constantinople declares with pride that once the tyrants were overthrown, “obeying the Master’s orders, I was raised in 32 days towards the upper air” (see Safran 1993; Kiilerich 1998). Here one should underline the fact that talking in the first person seems to be a habit of our temple’s columns: the base of column 17 of the east porch (a mere ten meters from column 4) bears two graffiti inscriptions, each reading MECKEAC; as convincingly demonstrated by G. Petzl, the column here proudly declares “Finish me!/Complete me!” (see pp. 106-107).
• 180Pedley 1976, pp. 21–22; Butz 2009; Bruneau and Ducat 1914, pp. 119–23. See also Guarducci 1942, pp. 115–62.
• 181A. Chaniotis to F. Yegül, July 4, 2013.
• 182Bowie 1974.
• 183Barrett 1993, pp. 236–41. In a study focusing on the epigraphy of Hellenistic Asia Minor, J. Ma emphasizes the physical and monumental nature of civic inscriptions in civic culture. Likewise, the talking column of the Temple of Artemis highlights the importance of “public writing, monumental communication, and the creation of memory in physical contexts” (Ma 2000, p. 103) in a way that is selective and purposeful to the Imperial Roman city of Sardis.
• 184Cahill and Lazzarini 2014, pp. 36–37.
• 185According to the late C. H. Greenewalt, jr., the tradition of donating or dedicating columns to the Temple of Artemis at Sardis started during the first phase of the building, long before the erection of the east colonnade. Two column inscriptions in Lydian, suggesting a late Classical or Hellenistic date, record the names of Sardians as donors. The first, in situ, is carved on the apophyge below the fluting of the shaft of column 12 from the east porch. It reads: “Manes [son of] Bakivas [grandson? of] Manes, to Artemis.” For this inscription on column 12, see Sardis II.1, pp. 106–7; Sardis VI.2, pp. 39–40, no. 21, pl. IX; Gusmani 1964, p. 259, no. 21. The second is an incomplete text on a similar apophyge fragment found some sixty meters north of the temple in 1978. It mentions one Srkastulis, perhaps denoting the name of the column donor, a resident of Sardis. This piece has been identified as belonging to one of the two pedestal columns from the west pronaos porch. For the second inscription, see BASOR 245 (1982): 24–25, fig. 26; Gusmani 1980–81, pp. 21–25. Greenewalt felt that these nostalgic Lydian “donor” inscriptions, dating from the sixth through the third century BC, represented a “tradition,” yet for some reason the practice was not continued during the Roman imperial period.
• 186For the Anatolian tradition of individuals donating columns, see BASOR 245 (1982): 25 n. 23. Examples include the famous sculptured column bases of the Archaic Artemision of Ephesus, donated by Croesus (Herodotus 1.92; Tod 1951, no. 6, pp. 9–10); the dedicatory texts in Lydian on the columns of the Temple of Athena at Pergamon (Gusmani 1964, p. 264, no. 40; Sardis VI.2, pp. 57–58, no. 40, pl. XIII); and the numerous columns bearing donor’s inscriptions from the mid-second-century Antonine temple of Carian Zeus at Euromos (Yegül 2014, p. 216, fig. 24; Bean 1971, pp. 46–47); and the inscription that records the donations of columns to a gymnasium in Ephesus (Wood 1877, app. 3, no. 18). One also recalls the donations of Seleucus I and his son Antiochus I toward the construction costs of the Temple of Apollo at Didyma, some of it from the income of the shops occupying the long portico they had built at Miletus (Magie 1950.I, p. 174; Gruben 1961, p. 192; Orientis Graeci inscriptiones selectae, no. 213); or the more modest offering of “ten squared blocks of marble” given to our temple around 130–100 BC by Moschion, the priestess of Artemis (Sardis VII.1, pp. 95–96, no. 93).
• 187Sardis I.1, pp. 7, 63–64, 66–67, 147, ills. 2, 57, 61, 164; Sardis R2, nos. 79, 80–88, 102–105, 251, and 252, figs. 196, 197, 202, 223–25, 228, 296–97, 434–35; İnan and Rosenbaum 1966, pp. 74–76, no. 40, pl. 26.1, no. 41, pls. 26.2–3. For full bibliography, see catalogue entries on the pieces in Sardis R2. See also Hanfmann 1975, pp. 72–74 n. 54.
• 188While in Greek temples the cult image could be and often was a freestanding object in center of the cella, in Roman temples the image was typically positioned directly against the rear wall of the cella, “which is open only at the front, facing towards the person and the area with the altar below”—an arrangement explicated by the augur’s position vis-à-vis the cult image and the templum. On the positioning of cult images see Rüpke 2007, pp. 183–84. See also Gladigow 1990.
• 189For a discussion of the Greek terminological and functional differences between honorific and religious images, andrias, eikon, and agalma see Price 1984, pp. 176–79; Hepding 1907, pp. 250–51. For a comprehensive consideration of the broadly conceived meaning and terminology of “cult image” in the Greek world, see Donohue 1997.
• 190Sardis inv. S61.27.15: Sardis R2, p. 96, no. 79, figs. 196–97. Another fragment of bearded colossal male, one showing the back side of a head with curly hair and wearing a diadem, might also belong to Antoninus Pius: Sardis R2, 97, no. 81, fig. 199.
• 191Sardis inv. S61.27.14: Sardis R2, p. 104, no. 102.
• 192Sardis I.1, p. 66; Sardis R1, p. 75; Sardis R2, pp. 104–5, nos. 102, 104, figs. 223–25.
• 193Greenewalt and Rautman 2000, p. 676. The same view was also expressed by the members of the Butler team; see Sardis VII.1, p. 72.
• 194Istanbul Archaeological Museum, inv. no. 4038: Sardis R2, pp. 166–67, no. 252.
• 195Respectively: Sardis R2, pp. 166–67, no. 252, fig. 435; Sardis I.1, p. 147, ill. 164; Sardis VII.1, p. 72; Burrell 2004, p. 106.
• 196Sardis I.1, p. 147.
• 197Sardis inv. S61.27.2: Sardis R2, p. 98, no. 88, fig. 201.
• 198Sardis inv. S61.27.9: Sardis R2, p. 105, no. 105, fig. 228.
• 199Sardis R2, p. 105, no. 105, fig. 228. Based on the superior finish and workmanship, Hanfmann had judged the fragmentary female and male heads (Sardis R2, p. 98, no. 88 and p. 105, no. 105) to be earlier than all other pieces of colossi. The female fragment is distinguished by the subtle modeling of the cheek and eye, recalling the kind of revived classicism that would have been appropriate in the Hadrianic period. I fully concur in this stylistic assessment. However, extra caution is necessary in both cases, since the fragments are very small. A large fragment removed in 1971 from the northwest corner of Church M’s exterior apse represents the crown of a head whose wavy hair is held by a diadem (Sardis R2, p. 97, no. 81, fig. 199). Although Hanfmann suggested Antoninus Pius because of the resemblance of the drill pattern to the better-preserved Pius head, a recent attribution to an unknown female by Annetta Alexandridis is more plausible (author in private viewing and discussion with Alexandridis in 2015 at Sardis). In addition to these heads, six fragments of colossal body parts (a hand with three fingers; a knee; two legs; a chest or shoulder; an abdomen and thigh) were found in 1961 among the ruins of Butler’s excavation house. They must have belonged to the colossi already found, or to others not yet known (Sardis R2, pp. 97–98, nos. 82–87, fig. 200). For comparative typology and dating of the colossal Hadrian head from Sagalassus, see Mägele 2013.
• 200Close and specific comparisons of hair and curls can be made with the “cuirass type” (“Panzerbüste Imperatori 32”) Hadrian in the Capitoline Museum, assigned a Late Hadrianic date, and with the so-called Stazione Termini type in the Palazzo Massimo alle Terme of the Museo Nazionale Romano, both in Rome. See Fittschen and Zanker 1985, nos. 52, 54–57, pls. 58, 59, also 60.53; for the former, see insert 22a, c; 23c, d for the latter.
• 201Sardis inv. S96.8.
• 202Greenewalt and Rautman 2000, pp. 675–76 n. 83, fig. 32. To quote the authors’ summary of features that allowed Smith’s identification: “depth of beard on neck, connection of beard and moustache, thinness of moustache, fullness and relatively wide breadth of cheeks, and tuft of hair under lower lip (which contrast with the narrower face, narrower beard in profile view, and exceptional meeting of beard and moustache in portraits of Lucius Verus; the short beard and heavy moustache in portraits of Hadrian and Antoninus Pius; the thick moustache in portraits of Hadrian’s chosen successor Aelius Verus and of Marcus Aurelius, Septimius Severus, and Clodius Albinus).”
• 203Burrell 2004, p. 106; Wegner 1939, pp. 56–65, 226–49, pl. 45.
• 204On the uncommon permanence of cult images in temples of imperial cult, see Price 1984, pp. 193–95.
• 205In an unpublished graduate student research paper, L. Pfuntner summarizes Smith’s and Burrell’s analyses of the bearded portrait and supported the former’s identification. In particular she noted the half-open mouth of the Sardis head as a unique feature “unparalleled in the portraits of Lucius Verus.” This interesting distinction, however, could have been a peculiarity of colossal icons in general, intended to enhance their appearance as distant deities, viewed dramatically from below. L. Pfuntner, “A Colossal Head from Sardis” (2008), Department of Classics, University of California, Berkeley. For comparisons, see Wegner 1939, pls. 42–43, 45, 54–55.
• 206Burrell 2004, pp. 317–18; Price 1984, pp. 172–74, 187–88; Rose 1997a, pp. 57–59; Rose 1997b; Donohue 1997; Kreikenbom 1992, p. 113.
• 207Burrell 2004, p. 319. See also Thür 2002, p. 92; Thür, ÖJh Beibl. (1972–75), 399. For a general discussion of the imperial cult in Ephesus, see Friesen 1995.
• 208Other than the individual portrait bust, there are three standard categories of cult statues of the Roman emperor: the cuirassed/military image, the nude/god-hero image, and the civilian/togate image (or, in the East, occasionally the Greek dress). See Price 1984, pp. 181–86.
• 209Strabo comments that Zeus’s seated colossal image was disproportionately large; it was reported to be ca. 13 m high in comparison to our colossal images, standing only ca. 7–7.5 m high. Even though the temple was very large, if the god rose “and stood erect, he would unroof the temple” (Strabo 8.3.30).
• 210Radt 1999, pp. 209–12, fig. 152; Burrell 2004, pp. 22–29, pls. 23–25.
• 211For the arrangement of Antonine dynastic groups, see Fittschen 1999, pls. 204–7; Kreikenbom 1992, pp. 39–51; Bol 1984, pp. 31–45, 88–89; Pekáry 1985, pp. 92–96, 104–6; İnan 1993; Rose 1997a, pp. 147–49.
• 212One could compare the double-rowed, linear arrangements of the Antonine cult figures inside the basilic cella space at Sardis to the theatrical display of their Severan counterparts inside the Temple of the Gens Septimia Aurelia at Cuicul (Djemila), where a wide, rectangular podium (11.1 × 2.85 m) occupies the entire width of the cella’s back wall. On this stage-like setting four Severan colossi were displayed, probably frontally and in a tight arrangement (Severus Alexander, Julia Mamaea, Septimius Severus, and Julia Domna). See Pensabene 1992; McCann 1968, p. 153, no. 45, pl. 53; Burrell 2004, pp. 320–21.
• 213Burrell 2004, p. 320. The fragment with curls and diadem (Sardis inv. S61.27.1: Sardis R2, p. 97, no. 81, fig. 199) was, with caution, assigned to the Antoninus Pius head, which would support the argument for the divine nudity of the emperor in the role of Zeus. For a different identification of the fragment by A. Alexandridis, see p. 196, note 199. The seated, enthroned pose is supported by the assumption that another fragment, also belonging to Antoninus Pius, shows a figure with “lower abdomen and upper left thigh at right angles to the body” (“conclusive proof that the statue was seated”); Sardis R2, p. 98, no. 87.
• 214For the “seated male/standing female” pose with Antonine examples from the theater at Bulla Regia, see Burrell 2004, p. 108; von Heinze 1982, p. 171, no. 9; Zanker 1983, pls. 19–20. See also the over-lifesize statues of the seated Poseidon and standing Demeter from the Antonine agora and basilica at Smyrna; Akurgal 1978, pl. 49; Naumann and Kantar 1950.
• 215The leftward turn of the Marcus Aurelius head would have directed the figure’s gaze (he is assigned to the left side of the cella) to the senior cult figures on the central platform rather than the person entering the cella. This might have been intentional and could indicate a level of visual interaction among the other figures (the Commodus head, however, appears to be strictly frontal).
• 216Yegül 1996, pp. 157–58; Zevi 1983; Andreae 1983.
• 217For the importance of lighting, including artificial lighting, in the celebration of imperial mysteries, see Pleket 1965; Hoepfner 2001.
• 218Another argument for the removal of the columns is Butler’s observation that the “cement” floor (I presume he meant “mortared rubble”) was continuous across the cella, while if the columns had been removed only after the temple went out of use, this would have left gaps in the floor (Butler, Sardis I, pp. 52, 62–65; Sardis II.1, pp. 13–14; see also Cahill and Greenewalt 2016, p. 504). Butler interpreted this thick layer of mortared rubble (which actually contains ground brick and tile, similar to the composition of opus signinum) as a colossal cistern (Sardis I, p. 52; Sardis II, p. 13). It might not have been a cistern but rather a late floor belonging to a postclassical use of this space.
• 219The combined cellas, with a length of over 44 m, and the walls, rising to ca. 18–19 m, would have yielded, at a rough estimate (taking average block sizes and not counting foundations), some 5,000–5,200 marble ashlar blocks. Almost all of this material is gone without a trace, except for the comparatively small portion of the cella walls that remains on the east end. For the role of railroads, which made the removal of building materials and marble blocks across long distances easy, in the destruction of archaeological remains across western Anatolia, see Greenhalgh 2013, pp. 298–320; for those effects on Sardis, which got its station (on the Izmir–Ankara line) in 1875, see pp. 319–20.
• 220Cahill and Greenewalt 2016, pp. 502–4.
• 221Cahill and Greenewalt 2016, p. 504. Another ancient instance when a cella was emptied of its columns comes from a story told by Pliny about the removal of the columns—probably the smaller interior columns—from the cella of the Olympieion in Athens by Sulla in 85 BC. This also provides a poor precedent for Sardis. If the columns removed by a victorious, conquering Sulla (interested in bringing war spoils to Rome) were to be used in the reconstruction of the Capitoline Temple in Rome (or any of the temples on the Capitoline Hill), the aim was a lofty one; the removal of columns from the cella of the Temple of Artemis at Sardis would have served no such grand purpose. See Pliny, Natural History 36.4.45; Dinsmoor 1973, pp. 280–81; Stamper 2005, pp. 81–83; Abramson 1974, pp. 17–19.
• 222See Frey 2015, and discussion in note 224 below.
• 223Pfaff’s observations were based on a series of notches in the bedrock (already observed in Stillwell’s 1932 publication of the temple in Corinth I.1) that helped to establish the outlines and alignment of the continuous foundations of the main walls and interior columns. Well below stylobate level, they do not mark the positions of the columns above them. Although Pfaff questioned whether the division of the cella into two distinct chambers could have been a part of the Roman renovation, he concluded that without “decisive evidence,” he would stay with the traditional view that the two-chamber cella was the original one. Pfaff 2003, pp. 114, 112–15. C. Mee and A. Spawforth also support the view that the “arrangement of the interior is uncertain because the Roman colonists collapsed the two rooms inside into one” (Mee and Spawforth 2001, p. 151). G. Gruben (in Gruben and Hirmer 1966, pp. 94–96) likewise preferred to regard the divided cella as original. See also Bookidis and Stroud 2004; Corinth I.1, pp. 115–19.
• 224Frey walks the reader through a series of calculations regarding the interior columns and their distances from each other, the cella crosswall, and the end walls, which are all still partly based on bedrock notches, but mainly on the application of the calculated column spacings from the Archaic colonnade (2.85 m) to the cella interior. The argument for an original single chamber (with two rows of perfectly fitting eight columns) predicates a remarkably “well-organized” arrangement of column distances, though these distances themselves are not free of metrological interpolations and speculations. Any new arrangement of the same columns with slightly different speculative distances could also give a “well-organized” arrangement for a cella scheme with a divided wall. Frey 2015, pp. 157–60; Bookidis and Stroud 2004, pp. 411–12. It is also crucial to note the slightly shallower foundations for the crosswall with respect to the longer walls; and three ashlar blocks from this foundation displaying anathyrosis facing different directions, and hence judged nonfunctional. These, in my opinion, represent false positives. In our temple, the foundation depths of the Hellenistic and Roman walls and columns vary; plenty of blocks reused in the foundations display anathyrosis (and other nonfunctional features) that face every which way. Frey also points out that the temple must have been more or less completely destroyed (no roof, no walls) during the Roman sack of the city, leaving little real evidence except for indirect, circumstantial indicators.
• 225Interestingly, neither Frey nor any of the scholars who have concerned themselves with the Roman history of the Temple of Apollo at Corinth question the purpose and function of an alleged Roman double cella—why would the Romans divide the cella, even in the most cursory way?
• 226The four dismantled short columns of the new west cella (columns 73–76) would have been the right height to supply the material needed for the shafts of the four Roman-rebuilt pedestal columns. The additional “rusticated” upper zones of the pedestals could have been provided by the columns of the west pronaos porch (columns 79–82). See also useful arguments about the dismantling and reuse of columns from the new west cella in Weber 2013, pp. 269–70.
• 227At an average of fifteen drums for the shorter and eighteen for the taller columns, the total number of drums available would have been, conservatively, about 130–32. Since it would have taken just about one-and-a-half short Hellenistic columns to build a pedestaled Roman column, the four columns proposed for the rebuilding (columns 73–76), plus any of the taller columns from the west pronaos porch (columns 77–80), would have done the job. This would have left a sufficient number of fluted drums that could be recut and reused as voussoirs in the Monumental Arch located at the western entrance to the city—if indeed this arch belongs to the second century AD and not later, as one could argue.
• 228For an excellent discussion of the daunting technical problems of a similar scenario, the removal of monolithic columns from the Temple of Apollo at Corinth, see Frey 2015, pp. 161–63. Much larger and taller, the problems of the Sardis temple could have been greater; however, their drum construction, though still gigantic, would have been a mitigating factor.
• 229S. Barker’s informative and detailed study (2010) delineating the economic advantages of the demolition and reuse of ancient buildings gives little consideration to the work involved in taking apart large marble columns composed of many drums vertically connected by dowels. His test case, the deconstruction of the Pantheon in Rome, sidesteps this issue, given that the granite porch columns of the Pantheon are monoliths, with one dowel at the bottom and one on top; deconstructing a Sardis column, with its fifteen to eighteen drums (and some twenty to thirty dowels each to unshackle), would have been a completely different matter in labor and salvage economics. See also p. 13, note 41.
• 230The spans of the timber trusses of the Basilica Ulpia in Rome, the Constantinian basilica at Trier, and the Odeion/Bouleuterion in Patara—all among the largest timber trusses recorded in the Roman world—were on the order of 22–26 m, the last praised as the work of the Sardian architect Dionysius; see pp. 246-248.
• 231Yegül 2014, pp. 215–16; Yegül 2015. See also Onians 1988, pp. 8–11, 41–58; Onians 1995–96; Rykwert 1996, pp. 170–82.
• 232MacDonald 1986, pp. 107–9, 201; MacDonald 1982a.
• 233For a detailed description of the “rough piers,” see pp. 67-68; Figs. 1.12, 2.101, 2.102, 2.103; Plan 6; Plate 1, Plate 8-13.
• 234Gruben 1961, pp. 181–82, 193.
• 235One should also consider the continuity evident among the Roman imperial cult, various forms of Hellenistic ruler cults, and cults of Roma, which were widely represented in the cities of western Asia Minor. Sardians honored Laodice, the first wife of Antiochus III, with a cult and a festival in 213 BC. There is ample epigraphic evidence from other Anatolian cities for the existence of cults honoring Seleucid and Attalid royals. For Hellenistic ruler cults as potential forerunners of imperial cult at Sardis, as well as discussions of Sardis inv. IN63.118–121, now in Petzl, Sardis M14, pp. 2, 10–14, nos. 302, 307–10, see Hanfmann, SPRT, pp. 110–13; Gauthier 1989, pp. 47–67; Habicht 1970, pp. 823–85, 91–105; Nock 1972a, pp. 143–44; Nock 1972b, pp. 213–14; Mellor 1975; Yegül 1982b, pp. 13–14. See also Price 1984, pp. 23–40, with the cautionary argument against a simple and direct connection between the imperial cult in the East and Hellenistic ruler cults, pp. 40–47; and Ma 2013, pp. 37, 119–226, esp. 202–3.
• 251Sardis R2, pp. 58–60, no. 20, figs. 78–83; SPRT, p. 221. For a broad review of Sardian cults and their relation to Artemis (and reiterating that it was Artemis who was the principal “goddess of Sardis,” not Demeter, Kore, or Kybebe), see de Hoz 2016, pp. 186–89. For an early association (and confusion) of Cybele and Artemis in the context of the Temple of Artemis, see Radet 1909. Underlining the deeply rooted presence of the two goddesses in Sardis, Roller offers a variant on the “difference” or identity issue presented by the Sardis relief: while the depiction of Artemis as taller and more substantial may indicate that “hers is the more important cult,” their near-equal representation and implied joint worship leave little doubt that the cult of Meter-Cybele lost none of her important status and presence in reference to Artemis’s newer, post-Persian cult. Roller 1999, p. 196 and her n. 6. We might add that if the language of images and gestures were to be given a voice here, it would say that the worshippers’ attention and relationship are a lot closer to Cybele than Artemis; the adoring gesture is singularly directed to her and her lion. Hanfmann and Waldbaum’s contention for a distinct separation of the goddesses and their worship is questioned by Horsley (1992, pp. 137–38).
• 252Hanfmann and Waldbaum 1969, p. 266.
• 253Privileging “written evidence,” or epigraphy, over the evidence of images, or iconography, is standard practice for classics scholars. For an art historian the situation is reversed: in the work of identification the language of images and symbols bears a cogency and strength equal to the written word (sometimes more so). The absence of epigraphic evidence naming Cybele/Kybebe from Artemis’s sanctuary is troublesome, but one might say it is to be expected, given that Artemis is the honoree of the sanctuary. Dedications to Cybele-Kybebe under a variety of names and disguises (some overlapping with Artemis) in Greece and Asia Minor have been listed by Robertson (1996, pp. 303–4 for Sardis). An interesting suggestion made at the Milas symposium on the cults of Karia (“Karia’nın Kültleri,” August 23–24, 2019) is that votive objects with graphic, sculptural reliefs were just as accessible to and popular with the less literate or illiterate segments of the population as those with inscriptions.
• 254Sardis I.1, pp. 125–27, ills. 136–38; Metropolitan Museum of Art, accession no. 26.59.9: Sardis R2, pp. 158–59, no. 235, figs. 405–406; Istanbul Archaeological Museum, inv. no. 4028: Sardis R2, p. 159, no. 236, figs. 407–9. The Nannas monument is not the earliest evidence for the pre-Roman presence of Cybele/Kybebe in the sanctuary.
• 255Sardis VI.1, pp. 38–40; Sardis VI.2, p. 38, no. 20; Sardis VII.1, p. 91, no. 85.
• 256This dating was provided by Hanfmann on stylistic grounds. Having seen the lions in Istanbul and New York recently, I suggest a slightly wider chronological spread between the two: the sitting lion (New York) can be dated later, ca. 490–480 BC, meaning that it could have been introduced to the original group a generation later—hence adding to the complication of the phases of the overall composition of the group or groups.
• 257T. L. Shear, who published the Nannas monument in a short article in 1931, reconstructed the found elements with the recumbent lion upon the long, plain base flanked by the sitting lion and the eagle on the taller, square pedestals, right and left. This is a simple, symmetrical, geometrically pleasing composition with no consideration of other figures that might have been a part of the group. It is curious that Shear restricted his rather linear iconographic search entirely to Near Eastern, Assyrian, and Hittite sources—the potential relationship of the lion and the eagle to Artemis and Cybele is not even mentioned. The rough field reconstruction in 1913 (Fig. 3.67) places the sitting lion and the eagle behind the recumbent lion, on the same long base, although there is not sufficient room, and leaves the taller pedestals empty—presumably for other possible figures. Cahill reminds me that it might be time for further field investigation of the findspot and a new restoration effort.
• 258For the myth of Herakles as the founder of a long line of Lydian kings and the symbolic significance of lions for Lydian kings, see Lucian 8.13.2, 25.10. In my opinion, it is reasonable to consider the mythical allusion of the original “Nannas lions” to royal Lydian past, but the conclusion that this rules out a specific connection to Cybele/Kuvava is not reasonable. Multiple symbolisms are the bread-and-butter of art historical interpretation. Hanfmann, reasoning entirely on the basis of the inscription, with its dedication to Artemis (not original to the monument), was reluctant to see a connection to Cybele in the lions of this group; Hanfmann and Waldbaum 1969, p. 269.
• 259My anonymous reader observes, “At this stage of preservation of the Nannas monument every hypothesis is possible,” and asks, “Were there two pedestals and two statues: one for Kybele and one for Artemis?” These hints at caution rightly hold back any attempt at reconstruction (which is not my goal). Simply to illustrate one of these very popular models, a bronze matrix of the Hellenistic period in the Metropolitan Museum of Art in New York (no. 20.2.24) can be considered. The main side (side A) depicts the mother goddess enthroned inside a pedimented Ionic shrine, her feet resting on two recumbent lions facing opposite directions (Fig. 3.70). On her right she is approached by Hermes, on her left there is a flying, wreath-bearing Nike and a standing maiden carrying a lighted torch. A crescent moon-and-stars and a tympanum complete her attributes. Based on stylistic and technical details, the origins are convincingly demonstrated to be Asiatic, probably from the region of Smyrna (Reeder 1987, pp. 423–24, 433–37). The point here is to show the unremarkable typology of the “Cybele and lions” compositions. More relevant for the presence of Cybele in the sanctuary of Artemis at Sardis may be a marble relief of the goddess found in 1958 in a late antique wall at the northern boundary of the sanctuary (NoEx 58.27). Dating to ca. the fourth century BC, it shows Cybele enthroned within an architectural enclosure, a large tympanum upon her left shoulder, holding her lion while a second one sits on her right (Sardis R2, p. 60, no. 21, figs. 84–85).
• 260For comments on the usefulness of iconographic evidence versus epigraphic evidence, see note 253. In a comprehensive discussion of the subtle but widespread conflations of many Anatolian cults (Cybele–Baki, Cybele–Dionysos, Artemis–Hekate, Artemis–Cybele–Demeter), amid the “tolerant religious climate” of Anatolia, E. Reeder (1987) underlines the critical importance of shared iconographies in identifying images.
• 261In this connection one can lightly allude to the importance to the Roman state of the Cybele cult, whose cultivation in Sardis would give the city no disadvantages (Virgil, Aeneid 4.215; 9.77–83, 107–22, 617–20; 10.156–58). Some of the royal women represented by the colossal images in the temple cella (such as Faustina the Elder) were associated and honored with the goddess Kybele-Meter and could well have retained such syncretic kinship at Sardis. One example would be the coin type on which Faustina the Younger is depicted as Augusta on the front and the enthroned Cybele, “Matri Magnae,” with her lion on the back (Coins of the Roman Empire in the British Museum 3, no. 1663). For political and ritual connections between Cybele, Rome, and Asia Minor, see Rose 2012, pp. 165–68.
• 262Buresch 1898, pp. 69–70; Radet 1904, pp. 307–8; Vermaseren 1977, p. 30, pl. 16; Vermaseren 1987, pp. 145–46; Lexicon iconographicum mythologiae classicae II.1, 701, no. 1038; II.2, 527.
• 263For M. Vermaseren the dedication to Artemis seems immaterial: the Kula relief is an exceptional exhibition of Cybele’s typical association with “famous deities” in their urban sanctuaries (multiple-deity Cybele reliefs), a phenomenon particularly true for Sardis, “[where] Cybele not unnaturally joins Artemis, to whom she is closely related and in some respects identical” (Vermaseren 1977, pp. 30–31; my emphasis); Lexicon iconographicum mythologiae classicae II.1, 701, no. 1038. I should also mention a group of similar marble reliefs from western Anatolia that depict Cybele standing or sitting enthroned within a shrine, flanked by her crouching lions (whose resemblance to Nannas lions is notable) and typically accompanied by two male deities, an older, bearded, Zeus-Jupiter-Saturn and a younger god commonly identified as Hermes. While most of these pieces originated in the Cybele sanctuary on Panayır Dağı at Ephesus, others come from Metropolis. There is little question that their distribution and appeal represented a more universal understanding of religion, embracing a pantheon of kindred deities with subtle mythic associations. For these “three-god” reliefs, see Heinzel 1999; Büyükkolanci 1999. See also Roller 1999, pp. 198–201; Naumann 1983. For multiple-deity reliefs (dated to the second century BC and later) found in two cave sanctuaries of Cybele outside Metropolis, see Meriç 2013, pp. 27–30.
• 264Horsley, in a discussion of the long history of assimilation of these goddesses in Anatolia (with many supporting examples), declared in reference to the Kula relief that it is useless to force on her an “identification as Artemis” (Horsley 1992, pp. 135–39); see also above, note 251. For arguments concerning the close representation and perception of these two deities at Sardis, see Fleischer 1973, pp. 198–99; Naumann 1983, pp. 212–13; see also, for the subtly changing and fusing identities of Cybele (“the Mother”) and Artemis, along with some other female deities, Reeder 1987, pp. 433, 440.
• 265Sardis R1, pp. 62, 64; Sardis R2, pp. 17, 20–21, 33–35 n. 26; SPRT, p. 50. For the marble relief representing the enthroned Cybele with two lions found in the sanctuary in 1958 (NoEx 58.27), see note 259. Such retention of the Cybele cult (along with other cults) at Sardis and in the Artemis sanctuary—albeit subsumed and “marginalized” but not displaced by Artemis—can be seen as a partial revival of the position taken by Radet and his generation of scholars—or, to adopt a larger view, the Hellenization of religion in Sardis and Asia Minor. Rein 1993, pp. 53–54; Raja 2017b
• 266McDermott 1986; Sartre 1956. See also Entrikin 1991, pp. 62–66. For a dialectic discourse on the semantics and logic of identity—self, sameness, and difference—see Campbell 1994, pp. 73–108.
• 267The goddesses share characteristics and display subtle differences: they have the same body type, they have long hair, they wear city crowns, they each hold a patera, and they are accompanied by lions (one has a lion seated by her side, and the other holds a lion cub in her lap and raises a large tympanum). The fact that the two figures are presented within a traditional pedimented shrine, but in separate chambers, tricks the imagination (at least, modern imagination) into believing that this is a representation of multiple identities and role-play. See Lexicon iconographicum mythologiae classicae VIII.1 (Suppl.), 753; VIII.2, p. 510, no. 42; Naumann 1983, p. 334, no. 337, pl. 30.1. On body schema, identity, and multiple dimensions in the same body image, see Campbell 1994, pp. 140 and 141–47.
• 268La Regina 1998, p. 225, figure on p. 230; see also Payne 2019.
• 269Christian beliefs regarding Christ’s worldly and godly being may still be ambiguous; variants of the Virgin Mary are famously worshipped in many Latin American countries. West African masquerades inextricably connect not only past and present, but also personalities in a spiraling confusion of being and becoming, “I am not myself!” (see Cole 1985) and mirror, in a sense, the Sufi sense of transcendence over self and the other, as in the modern Rubaiat of Ümit Yaşar Oğuzcan, the Turkish poet, “İnanma gözlerine, ben ben değilim [Do not believe your eyes, I am not I]” (see Oğuzcan 2004, p. 199). For a collection of essays that widely explore the fluid concept of the sacred in geographical, physical, and symbolic settings, see Wescoat and Ousterhout 2012 (esp. Ousterhout 2012, in that same volume), which underlines the changeable and mutable nature of the loca sancta and the universality of sanctuary.
• 270Sardis I.1, p. 114; Buckler and Robinson 1914b; Sardis VII.1, pp. 16–27, no. 8.
• 271Sardis I.1, pp. 114–115; Sardis II.1, p. 105, ill. 105. Further encouragement was provided by a large bronze coin of Sardis from the time of Elagabalus (r. AD 218–22) depicting two octastyle temples side by side; above each is shown a cult shrine with an image of the deity inside: Zeus Polieus (left) and Artemis (right); see Head 1901, pl. 27.10. The question has been posed as to whether these two temple facades are represented as such due to a special type of shorthand numismatic representation and could be conflated into one temple.
• 272There is no direct or clear evidence for this except for Hanfmann’s own interpretation of the evidence and the Elagabalus coin mentioned in pp. 220–221, note 318, which probably shows two different temples, but see note 238 above.
• 273Sardis inv. S61.27.14; see Sardis R2, pp. 104–5, no. 102, figs. 223–24. The argument for facial similarities between Zeus and portraits of Achaeus was based on the coinage of the latter. Hanfmann, SPRT, pp. 113, 120; Bank Leu 1978, no. 160B. Polybius 4.48.8–10, 7.15–17, 8.15–21.
• 274Hanfmann, SPRT, p. 120; Sardis R1, pp. 75, 80.
• 275Hanfmann was aware of the technical arguments against his proposal of a Hellenistic period “crosswall” dividing the cella, as published in 1961 by Gruben, but he considered them less convincing than his portrait identification and the evidence for a double-cult sanctuary provided by the Menogenes inscription. Hence, he unquestioningly assumed the inscription to mean a “double-cella” temple, and he expressed his reasoning thus: “The inscription of Menogenes which speaks of ‘residents in the Sacred Precinct of Zeus Polieus and Artemis,’ together with the existence of a subdivided temple [sic] . . . shows that this was a double-cult temple. . . . The fact that the wall dividing the two cellas does not bond into the long walls cannot stand up against the certainty of the above evidence” (Hanfmann, SPRT, p. 118). Although temple-sharing in the case of Zeus Polieus/Artemis (same cella), and later Zeus/Antoninus Pius and Artemis/Faustina (two cellas) at Sardis appears never to have been the case, there are numerous examples of this phenomenon during the Imperial period and under imperial policy. The Trajaneum at Pergamon accommodated Zeus/Trajan (and later Hadrian) in the same space (see p. 199, note 210); the colossal temples at Cyzicus and Tarsus honored a Zeus/Hadrian cult; the Olympieion at Ephesus honored Zeus/Hadrian; at Nicaea, Hadrian and Roma shared a temple; and Antoninus Pius shared his temple at Sagalassus with Zeus. In all of these instances, however, deities or a deity and an emperor shared the same cella space, not a divided temple with different cellas. See Gülbay 2009, pp. 67–69; Price 1984, pp. 245–60; Burrell 2003, pp. 38–39.
• 276As noted by Greenewalt, this head was also identified as Marcus Aurelius by “members of the Butler expedition.” For a summary discussion of the specific iconographic features underlying Smith’s identification, see Greenewalt and Rautman 2000, p. 676, no. 84; Sardis VII.1, p. 72. For Marcus Aurelius “Type III” portraits, see Fittschen and Zanker 1985, nos. 65–66, 70–71, pl. 75.
• 277Caution must still be advised in concluding the historic search for the alleged Temple of Zeus Polieus or Olympios. We do not quite know the ancient boundaries of the Sanctuary of Artemis; large areas, especially to the north, northeast, and southwest (eroded by the Pactolus) of the present temple, may still hide a small temple. For instance, we do not know what kind of small, Doric shrine is represented by the finely carved, right pediment cornice found just northeast of the North Building in the same precinct (some 60–70 m north of the temple). Could it be a naiskos dedicated to Zeus? I concur with Homer A. Thompson that this fine marble piece is no later than the second century BC. See Sardis R1, p. 63, figs. 81, 82.
• 278The images of Roma and Augustus shared the cella of their Augustan temple in Ankara; see p. 254 with Fig. 4.25. On temple-sharing: Robert 1975, p. 321; Bowersock 1976, pp. 112–21; Nock 1972b. On potential evidence for Roman emperors and members of the imperial family in partnership with deities and heroes (and traditional sanctuaries that came to include the imperial cult), see Nock 1972b, pp. 223–34; Talloen 2007, pp. 237–38. Epigraphic evidence on this subject is rich, but we must consider Nock’s justifiable caution on “temple-sharing,” with the significant exception of Asia Minor, where the emperor and native deities entered into “intimate associations” more readily (1972b, p. 231); see also Nock 1972a; Yegül 1982b, p. 14. The idea of a dual cult of Artemis-Zeus is also supported by Howe (1999, p. 204 and n. 169). All things considered, Cahill’s cautious and sensible comment on the subject is well taken: “The case . . . for a shared statue base [and temple] is circumstantial at best. . . . Is there any evidence left for the presence of Zeus in the temple, as opposed to the precinct? If not, why worry about it?” (Cahill, “Comments on Yegül, Temple of Artemis, April 6, 2009,” p. 14). On whether the elusive figure/deity Qλdãns, whose cult was honored together with Artemis in her sanctuary, could have been related to Zeus Polieus, see Cahill 2019b, p. 26.
• 279Not only did the primary religion of the Roman state require what in essence could be called temple-sharing among three gods (Jupiter, Juno, and Minerva, often in separate, side-by-side cellas), but minor gods and their images could be worshipped in the same cella, such as the arrangement seen on a shop-sign depicting a Roman temple with the goddesses Roma and Annona peering out between the Corinthian columns (Musei Vaticani, inv. no. 471; see Rüpke 2007, p. 146, fig. 13).
• 280Rüpke 2007, pp. 16–17; Gladigow 1988.
• 281On questions concerning the exceptionally diverse cultural and demographic nature of Sardis (“a Janus—one face a ‘free city,’ the other a royal residence, a seat of the satrap and the strategos of Lydia, and a center under the Seleucids of royal power . . . with the royal archive and the Royal Mint” [SPRT, p. 113]), see Hanfmann, SPRT, pp. 74, 90–96, 112–14, 128–35; Hanfmann 1987.
• 282Bengisu 1996. The same volume contains many essays that underline the flexible nature of Anatolian cults (with particular reference to the dynamic kinship between the cults of Cybele and Artemis). This might be the appropriate moment to honor the memory of Rose L. Bengisu, whose deep interest in the history and archaeology of Lydia, especially the Tmolos/Bozdağ region (where she and her late husband, Uğur, spent their summers), and knowledge of its many hidden gems could serve as a model for responsible and enlightened amateurs everywhere, especially Turkey. See Robert 1964, pp. 35–36; Hanfmann, SPRT, p. 98. For further evidence of the presence of Carian Zeus at Sardis, see Métraux 1971; see also de Hoz 2016, pp. 211–12; Foss 1978, pp. 21–60; 1982, pp. 178–205.
• 283Sardis inv. IN74.2, ca. 100 BC: Petzl, Sardis M14, pp. 108–9, no. 435. Sardis inv. IN91.10, late first to early second century AD: Petzl, Sardis M14, p. 54, no. 354. See Sardis VII.1, pp. 47–48, no. 22; Sardis R2, pp. 128–29, no. 161, figs. 308–9; AASOR 53 (1995): 9.
• 284Sardis inv. IN74.1: Petzl, Sardis M14, pp. 106–8, no. 434. Sardis R2, pp. 176–77, no. 273, figs. 463–64; Hanfmann, SPRT, pp. 104, 131; Mellink 1975, p. 216, pl. 42, figs. 18–19.
• 285Robert 1975; Herrmann 1996, pp. 329–35. For serious skepticism on the nature and extent of Iranian religious influence at Sardis, see Briant 1985; Briant 1998; Frei and Koch 1984, pp. 19–21. See also Chaumont 1990, p. 583; Gschinitzer 1986. E. Dusinberre follows Briant’s interpretation only in a general way and considers, at least as far as the Achaemenid period is concerned (she does not deal with the significance of the “copy” for the Roman period), that the city presented a heterogeneous picture in which multiple cultures, peoples, and religions were mixing and creating “something fully new and idiosyncratic to Sardis itself” (Dusinberre 2003, p. 118, and see also p. 233 n. 40); for the existence of similar religious dynamics at Sardis in connection with the cult of Artemis during Achaemenid rule, Dusinberre 2013, pp. 226–30. For the similar phenomenon of pantheism and syncretism of overlapping religions and cults and the proliferation of parallel, faith-based belief systems in the Sanctuary of Demeter in Pergamon during the Roman period, see Radt 1999, pp. 184–85.
• 286Hanfmann, SPRT, pp. 104, 135, 137–38; Hanfmann 1975, pp. 45, 56, 66–68, 74; Yegül 1987. On memory in Hellenistic and Roman Sardis, see Rojas 2010.
• 287Noting that at least some of the locals at Sardis had begun to enjoy table settings displaying hybrid, Lydian-Greek-Persian wares through the fourth century BC, Berlin observes that “Alexander’s advent may have spurred local citizenry to assert a new-found pride in their socio-political [and I add “religious”] identity, but this identity existed side by side with other cultural innovations. Pottery is various, just like the people who use it” (Berlin 2016, p. 358).
• 288Trilling 1950, “preface.”
• 289Yegül 2000. The intolerance of the Sardians toward the Ephesian envoy of Artemis (and being put to death for this transgression as a reverse display of intolerance!) might represent an instance when the city (or this group of Sardians) acted in a xenophobic manner toward what they considered to be an unwelcome infiltration by a foreign cult, in order to uphold their very own. This kind of behavior must have shown the other side of the cultural coin, with its “open-minded approach in adapting aspects of worship . . . from other cults or cultures” in heterogeneous post-Lydian Sardis, such as a transformed Iranian cult in a thoroughly Hellenized sanctuary; see Dusinberre 2013, p. 230.
• 290Sardis VII.1, pp. 73–74, no. 62. See also Herrmann 1996, p. 341.
• 291Herrmann 1996, pp. 322 n. 24, 329–35. Pleket 1965; Yegül 1982b. See also Keil 1923; Robert 1960, pp. 317–24. Unbound by rules of linear development, different religious or cultic manifestations might have shared a common ancestry. It is tempting to imagine that the original cult honored in the sanctuary along with Artemis and Kybebe was, indeed, the Lydian Zeus (Levs) or a version of it such as the Lydian Qλdãns (equated with Apollo, “Lord,” “Ruler and King”), which yielded to or was joined by the Persian Zeus Baradatas in the early Hellenistic era (itself a manifestation of Zeus Ahuramazda). This hybrid cult could have been transformed to Zeus Polieus in the later Hellenistic and Early Imperial periods, perhaps with a nod toward the growing urban concerns of the Roman city. Here we can revisit Hanfmann’s vision of the larger picture for Sardis—his belief that change was deliberate, eclectic, sometimes even contradictory, but mainly motivated by the city’s sense of self-determination through its choices, as cities of Asia Minor tried to recapture their past “as a source of pride and superiority toward the Romans” (SPRT, p. 135).
• 292The term broadly referred to a person, or occasionally a group, in charge of a temple of the imperial cult. On the derivation of the word neokoros and its broader meanings, either as a person or as a city awarded the privilege of establishing and maintaining such a temple at the provincial level, see Burrell 2004, pp. 3–6. The difference between a provincial imperial cult temple and a municipal one should be underlined; the former was the official seat of the cult of the emperor, a privilege voted by the koinon of the province and accepted by the Roman Senate and the emperor himself; the latter did not require such stringent official permissions and control. The establishment of municipal cult temples only reflected an extravagant form of honor and gratitude toward the emperor at the city level. Temple wardenship, as a basic and limited appellation for a city’s relationship to a particular deity, not necessarily the imperial cult, existed already by the middle of the first century AD, as evident at Ephesus, which referred to itself as the “neokoros of Artemis” in a coin of AD 65/66 and a second time as the “temple keeper of the great Artemis,” in the Acts of the Apostles 19:35, dating probably sometime in the late first or second century AD. However, these unofficial neokorate descriptions were replaced by the traditional signification of the title neokoros with the imperial cult. Friesen 1995, pp. 229–36; Keil 1919; Pick 1906.
• 293That the Temple of Artemis became an official seat of the imperial cult sometime in the second century AD is not a hard “fact” verified by written evidence, such as an inscription declaring it. The difficulty, indeed near impossibility, of the identification of such a temple, “one that makes its city neokoros,” has been recognized by Burrell: “The ideal way of recognizing such a structure would be the discovery of an inscription on it that calls it a provincial temple, mentions its designation for a particular emperor or emperors, and names the city neokoros. Unfortunately, this happy situation is rare to nonexistent. . . . Identifications of such precincts is generally based on a concordance of literary, numismatic, epigraphic, and archaeological evidence” (Burrell 2004, p. 12).
• 294In reality, of course, conservative factions in Roman Sardis might have considered such sharing outrageous, but, in fighting against it, lost.
• 295The controversial and ambiguous nature of the imperial cult vis-à-vis traditional religion, and the conflicted perception of the emperor as a new “god,” is a well-trodden subject in studies of Roman religion. Underlining the differences between everyday reality and the higher concepts invested in the divinity of the emperor, J. Rüpke observes that “constant effort was required in order to render the divinity of the emperor plausible,” such as nonstop public festivals, ludi, and gladiatorial shows associated with the cult, but also the effort to make cult temples showy, grand, and architecturally imposing; Rüpke 2018, p. 288. In the light of such everyday obstacles, the cogency of housing the cult in one of the most prestigious temples of Asia Minor, and of dedicating half of the venerable building to the “new gods”—even giving precedence to the construction and completion of the cult half (east) over Artemis’s half (west)—becomes easier to understand and appreciate. On the nature of the imperial cult and cult spaces, see Yegül 1982b, pp. 14–15, esp. n. 39; Nock 1972b, pp. 235–50; Fishwick 1978; Robert 1960, pp. 316–24; Bowersock 1973; Talloen 2007; Price 1984, pp. 156–62.
• 296One should add, however, that during the 1913 and 1914 seasons the effort to excavate the area in front of the east peristyle columns, bringing the massive earth mass down some 20 m, was vigorous—not a delicate operation with a view to uncover what might be a weak foundation for a small structure. No evidence for an altar was expected, and none was found; a Roman altar was not an issue at that time.
• 297Although these rites, performances, and sacrifices were undertaken on behalf of the deified members of the imperial family, they often addressed other, related deities as well. Oxen were typical victims offered to deified emperors, with cows being appropriate for imperial women. Livingston 2018, pp. 3–5.
• 298Price 1984, pp. 68–69.
• 299Price 1984, pp. 108–12, 134 n. 4, 216–17; see, for a short list of imperial temples with altars in Asia Minor, catalogue nos. 20, 23, 31, 108. As Rüpke pointed out bluntly, despite the widespread popularity of the cult, “constant effort was required in order to render the [questionable] divinity of the emperor plausible” (Rüpke 2018, p. 289).
• 300The theory that a koinon temple received funds from the province and a municipal temple relied on a city’s slimmer budget may favor the latter situation for Sardis, given the significantly unfinished state of its construction (although for an extraordinary large project, temple or not, being unfinished was not such an uncommon thing). Indeed, Cahill poses the question of whether the Sardis temple was indeed a provincial temple of the koinon of Asia and suggests that it might have been only a municipal temple (Cahill and Greenewalt 2016, p. 502). Yet, in that special period of generous granting of imperial cult temples under Hadrian, where the old Senate neokoros rules were in disregard, the difference between a neokoros provincial temple and a municipal one could have been blurred and minimized. Even if a temple like the Temple of Artemis at Sardis, which had its own quarries and used them for construction (and was its own economic entity) were municipal, it could have received some seed money from the generous emperor formally visiting Sardis. It should be underlined that the Sardis temple, unlike many others, appears to declare joy and pride in its ability to be a master of its money (up to a point to be sure!) and own marble from local sources (see pp. 190–193). Perhaps there was no single hard rule about these financial arrangements, and cities and temples (and other large imperial projects, such as the great imperial baths) negotiated and established their economic relationship with Rome on a case-by-case, politically inspired basis.
• 301Price 1984, pp. 210–17; Livingston 2018, pp. 4–5. For further examples of the association of the imperial cult with local Zeus or Apollo cults (or as precursors) at Adada, Selge, Termessus, and Sagalassus, see Talloen 2007, pp. 235–38.
• 302The Great Antonine Altar at Ephesus was a monument celebrating mainly Lucius Verus’s Parthian victory (AD 163–66; triumph celebrated jointly with Marcus Aurelius in 166), with rich and varied relief representations of popular imperial themes such as adoption, apotheosis, battles, personifications, select gods, and state sacrifice, but centered upon the adoption of the Antonine family by Hadrian. It was almost certainly a monumental imperial altar with a U-shaped design comparable to the Pergamene altar. Its original location (foundations) is not known, but some of its extensive relief panels were reused as a parapet for the late antique pool and fountain in front of the Library of Celsus. In its complete destruction and dislocation, probably during the early Christian era, it presents a similarity to the hypothetical imperial altar in Sardis. See Eichler 1971; Hannestadt 1988, pp. 201–4; Price 1984, pp. 158–59, fig. 8; Kleiner 1992, pp. 309–12, figs. 279, 280. For the altar of the Flavian imperial cult temple, see Wiplinger and Wlach 1996, pp. 56–57, figs. 74, 75. While we support the notion that altars were natural elements of imperial cult temples (as they were for other temples), there are in fact very few imperial cult temples with preserved altars. Nonetheless, the assumption that imperial altars existed but were not preserved is shared widely. To take one example, the description and reconstruction drawings of the Trajaneum at Pergamon include a small altar, presumably one without substantial foundations, as it would have been supported by the deep, vaulted substructures of the temple terrace. After all, smaller altars were nothing more than glorified tables. Radt 1999, pp. 213–14, figs. 157, 159.
• 303I am grateful to Professor Serdar Hakan Öztaner, the director of the Nysa Excavations in Turkey, for expounding on this idea during the discussion following my lecture on “The Temple of Artemis at Sardis: A Sacred Place Shaped by Dividing and Uniting Religions” (March 13, 2019) at Ankara University. Öztaner’s observation that such a location for an altar, bringing to mind smoke ascending to the sky during cult services, is cogent and underlines the functionality of an open roof, which is my design proposal for the area.
• 304Keil 1919. For the complicated and controversial issue of the first neokoros privilege of Ephesus with Nero and its extension under the Flavians, see Burrell 2004, pp. 60–63.
• 305Ratté, Howe, and Foss 1986. Further studies on this temple are in progress.
• 306Greenewalt 2006, p. 176.
• 307Ratté, Howe, and Foss 1986, p. 54.
• 308For the Temple of Augustus in Ankara, see Akurgal 1978, pp. 284–87; Krencker and Schede 1936; and more recently, Kadıoğlu, Görkay, and Mitchell 2011, pp. 79–108; Görkay 2012. The main publication of the Temple of Zeus at Aezane is Naumann, Krencker, and Schede 1979; see also Weber 1969; Laffi 1971; and Lyttelton 1987, pp. 44–45. While this temple was considered to be Antonine in past publications, recent work on the architrave dedicatory inscription almost certainly establishes it as a late Flavian–Domitianic project; see Jes, Posementir, and Wörrle 2010, pp. 74–82.
• 309Greenewalt 2007, pp. 743–44, figs. 1–5; Cahill 2015, pp. 421–23, figs. 9–11; Cahill 2016, p. 156; Cahill 2018, pp. 335–37.
• 310The overt musculature, knotted lionskin cape, and lifting pose of the figure favor traditional iconographies of Herakles, and indeed this powerful, hardship-overcoming god-hero has been associated with architecture, construction, and quarry work. In an honorary inscription from his hometown, the architect Perikles of Mylasa (known to have worked on important buildings in Rome) is referred to as being “equal to Herakles” (in a paper by Mustafa H. Sayar, “Milas’lı Mimar Perikles ve Eserleri,” presented at the Karia ve Karialılar Sempozyumu, Milas, August 10, 2018). For Herakles as a model of strength in architecture and building: Statius, Silvae 3.12–22. Alternative identifications of these capital figures are possible. The Wadi B temple and its decorative iconography are under study by P. Stinson and B. Yıldırım.
• 311The erudite but circumstantial evidence is teased out by Foss in his consideration of the Greek word on the tympanum of the pediment: Ἀδραμύ / τηον, a “novel spelling” (Foss) of Adramyttium, a city in Mysia (modern Edremit) that shared with Sardis the distinction of being the center of a conventus, a judicial district in the province of Asia. Foss comments that “conventus centers played a major role in the organization of the imperial cult and that this context provides a relationship between Adramyttium and Sardis in involving a temple. . . . The appearance of the name of Adramyttium in the pediment of the temple at Sardis strongly suggests that that city had a role in the dedication of the building. The most plausible occasion for such an involvement would be the erection of a provincial temple [to the imperial cult]” (Ratté, Howe, and Foss 1986, pp. 63–64). This interpretation gains support from the broken sculptural “lump” on the tympanum next to the city’s name; it must have been a representation of a patron deity or an allegorical personification of Adramyttium. Habicht 1970, pp. 67–71. See Sardis M14, p. 172, no. 572.
• 312Sardis inv. IN05.18; another portion of this architrave, reused as a foundation block, was found in 2016 (Sardis inv. IN16.18); see Petzl, Sardis M14, p. 98, no. 421.
• 313A coin of Vespasian from Sardis shows a tetrastyle temple but lacks an identification by the title neokoros; the hypothesis is questionable because the Flavian cult was well established at Ephesus. Ratté, Howe, and Foss 1986, p. 66; Head 1901, pp. 67–70; Sylloge nummorum graecorum von Aulock, Lydien, 3148; Sardis M7, pp. 251–55, nos. 246, 247. Burrell also objected to this hypothesis, mainly because of its “weak numismatic thesis.” See next note.
• 314Sardis inv. IN58.4: Petzl, Sardis M14, pp. 68–69, no. 378. See Johnson, “Epigraphic Report” (Sardis Expedition field report), p. 10 n. 4; C. Foss in Sardis R3, pp. 168–69; Supplementum epigraphicum graecum 36.1093; Burrell 2004, pp. 106–7. For a comprehensive analysis of the imperial cult and imperial cult temples at Sardis, see Burrell, pp. 100–115; Herrmann 1993, pp. 251–52. See discussion note 299 above.
• 315Sardis M7, pp. 11–12; Burrell 2004, p. 109. See also Kienast 1996, pp. 160–61.
• 316Sardis VII.1, pp. 63–64, no. 47; Hanfmann, SPRT, p. 144.
• 317Sardis M14, p. 82, no. 397; Ritti 2017, pp. 372–77. For Lucius Julius Libonianus see previous note.
• 318The third neokorate is represented by large bronze coins showing the cuirassed bust of Elagabalus on the obverse and four temples, three for imperial cult and one for the Lydian Kore, on the reverse (coin type 6: Catalogue of the Greek Coins in the British Museum, no. 171). Two almost-identical, six-column temples are shown on the bottom; one smaller six-column temple with the emperor in military dress and a four-column shrine with Kore under a “Syrian pediment” are above. No independent temple of Elagabalus has yet been found on the site, although the cult might have moved into the temple of another deity. Sardis lost its third neokorate honor when Elagabalus suffered damnatio memoriae. It was not before AD 253–60 that the city regained the title and the honor, under Valerian. Sardis M7, pp. 9–10, 12; Kienast 1996, pp. 177–79; Burrell 2004, pp. 110–15, fig. 91; Hanfmann, SPRT, pp. 145, 277 n. 81; Robert 1976, pp. 52–53.
• 319Sardis VII.1, pp. 71–72, no. 58. Compare this to the two separate inscriptions that record dedications to Hadrian as the “New Dionysus”: Sardis VII.1, pp. 31–33, nos. 13–14.
• 320Sardis M14, p. 66, no. 375 (Sardis inv. IN70.4); BASOR 245 (1971): 14.
• 321Waelkens 2002, pp. 325–27, 348–49; Talloen and Waelkens 2004. For Hadrian and the Hadrianic cult at Sagalassus see Waelkens 2007; Opper 2008, pp. 25–26.
• 322Marc Waelkens, former director of the Sagalassos Archaeological Research Project, suggested that these six cult colossi were all originally housed in the official Hadrianic cult temple (formerly known as the “Antonine Temple” on the south hill). The very small cella of this temple (6.80 × 9.40 m, with an even smaller display “podium” encircling the interior) would have made the job challenging, if not impossible. I maintain my view that these colossi originally occupied the so-called Marmorsaal of the original bath complex, a grandiose space (ca. 24 × 16 m) which would have created just the right kind of setting for the imperial images—not as an official cult center, but as an honorific hall whose associations with the imperial cult at popular level would have been wholly appropriate; this would have made the final transportation to the ambulacrum-frigidarium a lot easier; see Waelkens 2013. For the imperial cult in the civic context of Roman baths, see Keil 1929, pp. 35–36; Yegül 1982b; Yegül 1992, pp. 57, 251, 422–23, 473 nn. 9–12; Yegül 2010a, pp. 164–67.
• 323Gülbay 2009, pp. 55–64; Gülbay 2016. See also Jones 1996, p. 30; Romeo 2002, pp. 21–22; Boatwright 2000, pp. 149–50; Bowie 1970.
• 324Burrell 2003. The cities of Asia responded in turn with enthusiasm and gratitude to their benefactor, honoring him with special honors and titles such as “Restitutor Asiae,” “Restitutor Nicomediae,” “Restitutor Phrygiae,” etc. Thirty-two cities had official altars dedicated to him, while thirty-three dedicated statues. This Asiatic phenomenon was a reflection of the Panhellenic honors paid to Hadrian in mainland Greece, where at Athens alone ninety-four altars and other statues were dedicated to honor him and his flourishing cult; one of these statues was a colossal cult image. See Benjamin 1963, pp. 58–59; Jones 1996, pp. 35–36; Livingston 2018, pp. 1–3.
• 325Burrell 2004, pp. 281–84; Gülbay 2009, pp. 65–68, 185–86; Halfmann 1986, pp. 188–201.
• 326Sardis VII.1, pp. 30–32, nos. 13–14; Hanfmann, SPRT, p. 145.
• 327Upon the discovery of this inscription in 2000, Peter Herrmann informed C. H. Greenewalt, jr., that the statue base, in the form of a tall inscribed pillar (Sardis inv. IN00.4; 1.25 × 0.37 × 0.53 m), had almost certainly been occasioned by the emperor’s visit to Sardis, although the text, as far as it has been preserved, does not specifically mention this visit; few contemporary “imperial visit” statue base inscriptions do so (Herrmann to Greenewalt, Sept. 27, 2000). See also Weiss 1995, pp. 213–24. G. Petzl gives a more conservative general date for the inscription: AD 117–38, “during a visit to Lydia”; he bases his conservatism on Weiss and not on other research on the subject, including my own; see Sardis M14, p. 65, no. 373. For an important contextual study underlining the importance—and correlation to specific imperial visits— of such inscribed statues and statue bases, where the naming of city officials almost certainly denotes the emperor’s presence (and an incomplete list of dated inscriptions of statue bases from Hadrian’s travels in Asia), see Højte 2000, pp. 229–30.
• 328Coin of Saittae, Hadrian, AE31: “Tyche wearing mural crown, standing on the right, clasping hands with togated Hadrian, standing on the left” (Aufhauser auction 1992, lot 317; unpublished). See www.wildwinds.com/coins/greece/lydia/saitta/i.html (accessed April 2, 2018).
• 329Supplementum epigraphicum graecum 17.532. From Saittae to Troketta the renowned Royal Road passed right through Sardis, as does the modern Izmir–Ankara highway; Yegül 2018.
• 330Birley 1997, pp. 159, 168–70; W. Weber 1907, p. 139 and n. 504. Supporting this view see also Bowersock 1969, pp. 121–23; Weiss 1995. See also Halfmann 1986, pp. 199–200; Syme 1988.
• 331Hipponax correctly describes the relative locations of some of the larger tumuli as one travels from east to west, past Sardis toward the sea, “by the road to Smyrna.” Hipponax fr. 42: Masson 1962, pp. 129–34; Sardis M2, p. 77, no. 280. See also Herodotus, Histories 1.9.
• 332Hadrian’s curiosity “about all things” was exceptional in Greco-Roman culture, and it led to his climbing mountains for mythic interest and sheer poetic enjoyment; watching the sunrise from the top of Etna in Sicily (where he is supposed to have built a simple refuge structure: Aelius Spartianus, Life of Hadrian 13.3; Strabo 6.2.8); climbing Mount Casius near Antioch and Mount Argyrus (Erciyes) in Cappadocia for the view (Strabo 12.2.7); ascending the spectacular Mount Theches (Madur Daği, Sürmene, Trabzon) in eastern Turkey, from whose summit Xenophon’s Ten Thousand first saw the Black Sea, shouting with joy, “Thalatta! Thalatta! The Sea! The Sea!” (Xenophon, Anabasis 4.7.24; see also Arrian, Periplus 1; Birley 1997, p. 155). It would have been surprising and out of character for Hadrian to tamely walk across the foothills of Mount Tmolos, the “blest birthplace of Zeus and Dionysus” (as reported by Eumelos, an eighth-century BC Greek poet), without climbing it. The lookout and “exedra” on Mount Tmolos reported by Strabo have been shown to be actually on Kel Dağ, a lower-altitude peak on the Tmolos chain; Strabo 6.13.4–5; reporting on Eumelos, see Sardis M2, p. 9, no. 14. For recent work on Mount Tmolos and the “exedra-sanctuary” on Kel Dağ, see Bengisu 2013. See also Foss 1978.
• 333Pergamon received its second neokorate under Trajan, and Smyrna and Ephesus under Hadrian. Gülbay 2009, pp. 67–68. See also Hyde 1915; Burrell 2004, pp. 110–15.
• 334Sardis M14, pp. 24–25, nos. 319–20.
• 335Burrell 2004, pp. 284–86; Millar 1983. Perhaps it was the paucity of new neokoria under Antoninus Pius that caused the cities of Asia to boast more and more of their already existing neokoros honors and titles on coins and inscriptions. The emperor’s parsimonious nature is well illustrated in the letter he sent to the Ephesians in which he expressed admiration for local benefactors like his friend Publius Vedius Antoninus, who beautified his city with lasting and useful monuments (such as a great bath-gymnasium complex) instead of spending on cash distributions and spectacles in order to gain instant popularity. Magie 1950, pp. 1.632–33; Hicks 1890, nos. 491, 493.
• 336“[H]e provided shipments of grain from Egypt; he made the harbor navigable and diverted the river Caystros which silted up the harbor” (IvE, p. 12, no. 274). See also next note.
• 337Lewis 1974, p. 17.
• 338A colleague and historian of Asia Minor with whom I discussed this subject even suggested that Hadrian might have returned to Sardis and the Artemis sanctuary a second time during his Ephesian stay in AD 129. After all, the distance between Ephesus and Sardis could be covered on horseback in a day or two, and the land was beautiful, sublime, and full of game. However, on this issue P. Weiss, who firmly believes that “Hadrianus had visited in AD 124 the neighboring city of Sardeis [neighboring Saittae-Kula],” considers that the “assumption of another visit at a later date (AD 128–29) is unfounded” (1995).