The oldest original synagogue building in the Diaspora: the Delos synagogue reconsidered

The oldest original synagogue building in the Diaspora: the Delos synagogue reconsidered

Monika Trumper


The original function of the synagogue on Delos has long been contested, and can be determined only through analysis of the architectural history of the building. In this article, the author reconsiders the history of the building’s construction on the basis of fieldwork to date. Five phases of construction are distinguished: two predate 88 B.C. and the remaining three date between 88 and the end of the second century A.D. The structure’s most characteristic features–a large hall, a water reservoir, orientation toward the east, and an isolated location on the eastern seashore–are not consistent with those of a private house, a meeting place for an association, or a pagan cult building, but rather confirm its function as a synagogue from the earliest phase onward.


The synagogue on Delos is the earliest known to date, either in the Diaspora or in Palestine. (1) Its identification has long been the subject of heated debate. During the last 30 years or so a consensus has emerged that the building was an assembly hall for Jews or Samaritans, but this consensus pertains only to the building’s very last phase of use. In several recent publications it is argued that the building was first erected either as a private house or as a pagan meeting place for an association. It is argued here, however, that the building was conceived and used as a synagogue from the beginning, that is, from the time of its initial construction, in the period before 88 B.C.

The building in question, GD 80, is situated on what was the eastern side of the city of Delos, far from the main harbor and the city center with its sanctuaries, public places, and commercial areas. It now belongs to the area known as the Quartier du stade (Fig. 1), which is largely unexcavated and, today, comprises mostly residential buildings, a building that was used as a meeting place for an association (GD 79a), and an important complex of agonistic buildings (gymnasium, xyste, stadium, GD 76-78). In contrast to all these constructions, GD 80 is located immediately on the shore, at the southern end of a large bay that served in antiquity as a small, well-protected harbor, accessible even during rough weather. (2) The maximum exterior dimensions of the rectangular structure are now 28.30 m (N-S) x 30.70 m (E-W), the full eastern extent being unknown. GD 80 consists of three parts (Fig. 2): two large rooms, A and B, similarly equipped with marble benches, in addition to a throne in room A (a lime kiln was installed in room A in a later phase); area C, essentially the east half of the building, which includes a marble stylobate running north-south, marble benches preserved in the northwest corner, and an entrance in the south wall; and the D-complex, consisting of several small rooms (D1-D7), one of which (D1) gave access to a large water reservoir.


Although the stadium was established in the period of Delian independence (314-167/6 B.C.), the major development of the Quartier du stade dates to the following period, after the Romans conquered the island, declared it a free port, and gave control to the Athenians. Delos soon became a booming cosmopolitan trade center, and the city grew considerably in order to house the numerous merchants from across the Mediterranean world who established themselves in the tax-free port. Even though the island was sacked twice, in 88 B.C. and 69 B.C. by, respectively, the troops of Mithridates VI and then pirates under Athenodoros, its abandonment and decline during the first century B.C. can be attributed more to the influence of increasingly successful Roman ports such as Puteoli and Ostia.

After Andre Plassart excavated GD 80 in 1912-1913, he published only a short report accompanied by a schematic plan. (3) Some 50 years later, Philippe Bruneau dedicated a much more detailed study to the building, providing an accurate plan, sectional drawings, and an epigraphic register. (4) Recently, L. Michael White and Donald D. Binder have offered new theories regarding the history and function of GD 80, based on their own examinations of the archaeological remains. (5) In addition, GD 80 is addressed to a greater or lesser degree in all studies on ancient synagogues, especially in works on early examples and the evidence from the Diaspora. (6)

Despite its importance and early date, however, GD 80 has yet to be published comprehensively. Therefore, much information has been unavailable until now, including, for example, an extensive record of all the architectural elements of the building; a catalogue of all material discovered; a detailed reconstruction of the construction history based on the architecture itself and on excavations since 1913; extensive documentation with photographs and drawings; and an examination of the immediate surroundings and the quarter in which the building is situated.

Ideally, such documentation should be accomplished by an interdisciplinary team consisting of scholars familiar with the specific architecture and archaeology of Delos and specialists in Jewish culture and history. As the completion and publication of such a comprehensive study seems still to be well in the future, however, the only alternative is to pursue research more limited in scope and with more modest goals. In this article, therefore, I address just one major issue not resolved in previous studies: the history of the edifice according to the evidence visible above ground, and the building’s nature and probable function in the different phases of that history. It is by examining principally the extant remains of GD 80 in Delos that I hope to provide new information and insights for the ongoing study of early synagogues.

As the precise function and use of early synagogues in the Diaspora and in Palestine remain the subject of much controversy, (7) treatment of the architecture and the development of GD 80 should be independent of the issues of its identification and function. Consequently, the first section of this article is limited to an examination of the architectural remains. On the basis of the materials, building techniques, and types of bonding used in the construction of the walls and rooms, (8) different complexes of walls can be identified, each of which distinguishes particular parts of the building.

Whether these wall systems correspond to single phases, and whether the sequence of such phases can be established with any degree of certainty, is discussed in the second section, which is not limited to the reconstruction of the architectural development, but also includes reflections on the character of each phase and the possible function of the building at that time. In order to make clear to what extent my reconstruction of the history of the building differs from those previously published, these two sections are preceded by a short summary of the most important reconstructions to date, those submitted by Bruneau, White, and Binder. (9)

In the last major section, I address the question of when the building was first considered a synagogue and used as such–whether from the beginning or only in a later phase–and what criteria allow us to determine its function in the different phases. The value of the available information (from the architecture, equipment, comparanda from other synagogue buildings, and literary and epigraphic sources) is discussed briefly in this context. A short summary and suggestions for further research conclude the article.


In his first study of GD 80, Bruneau distinguished two phases, the second comprising only the division of the main hall into two rooms (A and B). (10) He submits that, in light of the fact that three inscribed statue bases and one block with a dedication and a graffito, all from the nearby gymnasium (GD 76), were integrated into the east wall of the main hall (A/B), the building could have been erected only after the destruction of the gymnasium by the troops of Mithridates VI in 88 B.C. From the beginning, it could have been conceived as a Jewish assembly hall.

This view is amended slightly in Bruneau’s later article with the insertion of an intermediary phase consisting of the reconstruction of the east wall of rooms A and B with the reused material from the gymnasium. (11) Accordingly, the original building can be dated to the period before 88 B.C., that is, sometime in the second century B.C. Although clear evidence for its original function is conspicuous in its absence, the measures taken in the second phase, after 88 B.C., most probably corresponded with the structure’s use as a synagogue, either refurbished or otherwise transformed from a predecessor of unknown nature into a Jewish building. The last discernible alteration of the architecture (the third phase, involving the division of the hall) could have been realized at any time until the end of the second century A.D., when GD 80 fell out of use.

White argued that the building was originally a private house, erected in the second century B.C. and defined by gneiss walls that delimited the large hall (A/B) and the complex of rooms to its south (D). (12) Renovations of this dwelling then occurred in two stages, in the late-second and mid-first centuries B.C. The first renovation, the conversion of the house into a synagogue edifice, included the modification of the portico in area C to create a three-winged portico entrance on the side that fronts the sea, and perhaps the partitioning that resulted in rooms A and B. In the second stage of renovation, dated to after the Mithridatic destruction of 88 B.C., the east wall of rooms A and B was rebuilt with reused blocks and the rooms themselves may have been refurbished. Even though White’s description is extensive and illustrated by a plan indicating four different types of walls, some questions concerning the construction history remain. (13)

White was not the first to identify the original building as a domestic residence, but he was “the most vigorous proponent of this viewpoint.” (14) It is interesting to note, however, that most scholars writing about synagogues after the publication of White’s article in 1987 follow his hypotheses, and usually reprint his schematic plan instead of Bruneau’s much more detailed field plan. (15)

Both Bruneau and White agree that GD 80 is situated in what was a residential quarter and was surrounded by buildings, but they do not try to establish a relative chronology between the building and its neighbors. (16) They also agree that a thin wall in D (Fig. 3, “Predecessor(?) granite wall”), which was made entirely of granite and is now razed to the ground, does not fit into any phase of the building’s history and must, therefore, belong to an earlier structure that stood on the same site.


Recently, Binder has tried to rehabilitate the construction history proposed by Belle D. Mazur in 1935, which was challenged by Bruneau in 1970 and again in 1982, and successfully, as no one has supported Mazur’s ideas since then. (17) Mazur came to the conclusion that the building as she reconstructed it could never have functioned as a synagogue, yet Binder has no problem identifying it as a Jewish, or rather Samaritan, assembly hall, at least in its second phase. According to him, following Mazur, the original building comprised the large hall A/B, the D-complex, and a courtyard of enormous size (28 x 28 m) with a full peristyle of eight columns on a side. He then conjectures that rooms to the north and to the south of the courtyard, which are not yet excavated, also could have been part of the building. Later changes (second phase) consisted of some additions in the D-complex and, primarily, the division of the large hall and the renovation of its east wall (both incorporating reused marble blocks from the gymnasium and, therefore, dating after 88 B.C.). Although Binder does not completely reject the hypothesis that the building was constructed as a synagogue, he favors an alternative scenario: the building was erected as a cultic hall by a pagan association in the second century B.C., but after 88 B.C. it was taken over by Jews (or rather Samaritans) who renovated it and transformed it into a synagogue. (18)


The field plan (Fig. 2) and sectional drawings published by Bruneau provide much useful information for the reconstruction of the building’s architectural development, such as the materials used, sizes of the walls, wall joints, preserved stucco decoration, and elevation drawings. Some important details are missing, however, and others do not correspond to the archaeological evidence; (19) since the documentation of neither White nor Binder (including plans and photographs) supplies the missing data, a close examination of the actual remains is indispensable.

Five wall systems, which denote different parts of the building, are distinguished here, and special attention is given to the walls of neighboring edifices and to the features of Binder’s “courtyard” C. (20) It is of paramount importance to distinguish clearly between the different wall systems and to determine which walls are to be associated with one another. Furthermore, in each case it must be carefully considered which criterion is the more important and revealing: the material and the technique used to build the wall, or how it was bonded (joined) to another wall. Even though the analysis of wall bonding is, in general, a valuable criterion in the study of architecture, bonding must be evaluated with caution, because walls of different phases can be joined without leaving evidence of their different ages. (21) The disregard of wall systems has led to misinterpretations in previous publications, as will be demonstrated with relevant examples.


Solid thick walls made of regular gneiss blocks delimited a large hall (A/B) that measured 16.80 m (N-S) x 14.40 m (E-W), (22) with three entrances in the east wall (Figs. 2, 3) that were, from the beginning, provided with marble thresholds (see below). The exterior of the north facade was covered with red, waterproof stucco, but the external face of the east wall may have been decorated with finer white stucco (see below). The floor of the hall was probably surfaced with a simple waterproof pavement of marble chips that was not subdivided or decorated in any way. The interior walls were covered with stucco most likely in “Masonry Style,” of which only the red plinth course is preserved (Figs. 4, 5). (23)


Roofing such a large room must have been challenging and might have required interior supports. Because of later changes to the building and its current state of preservation, the issue of supports cannot be addressed, but it is highly improbable that such a room was left open to the sky, that is, built as a rather expensive, luxurious courtyard. (24) The roofing was presumably realized as a simple ridged roof with the ridge beam oriented north-south. Sufficient lighting of the hall would have been admitted by the three doorways and could have been ameliorated with additional windows in the north and west walls, as neighboring buildings had yet to be constructed.

The nature of the extension of the gneiss building to the south and east is unknown. The west wall of D1 becomes considerably thinner some 0.30 m south of the northwest corner, and although no clear joint can be discerned, the material and construction technique change visibly (Fig. 6). In addition, traces of stucco with a white finishing plaster are preserved on the south face of the north wall, in the joint with the west wall. Since this stucco begins at a height of about 1.40 m above today’s floor level, originally it must have decorated the jamb of a window in the west wall whose dimensions cannot be reconstructed (Fig. 7). Both the change in building technique and the blocking of the window are, most probably, due to the renovation of the west wall in a later phase, when the building was extended to the south. Because the window was blocked, it does not seem likely that mere technical reasons were responsible for the reduction of the thickness of the west wall of D1. (25) This view is confirmed by the character of the east wall of D3, which also becomes thinner and was, judging from the material used and the construction technique, certainly remodeled at some point, though not at the same time as the west wall (see below, pp. 534-535).


In short, even if issues involving the extension and layout of the southern part of the building remain open, it is certain at least that the building continued farther south and thus included the water reservoir under rooms B and D1. (26) In addition to the foregoing evidence, expansion is suggested by more fine white stucco on the exterior face of the south wall of B; (27) most probably this decoration dates to the first phase of use or at least to an early phase, because the later partition wall of D2/D3 abuts it (Fig. 8).


The position of the reservoir, partly under hall A/B and partly under D1, is quite unusual. Whether the location was a deliberate choice with regard to a specific function of the reservoir or was selected for merely practical reasons (28) cannot be determined with certainty, particularly because the exact size and layout of the original building and the possible influence of external factors (29) are unknown. In theory, the foundation of the predecessor wall in D1-D3, situated immediately south of the reservoir, might be identified as remnants of the original south facade; but given the reduced width, different material (granite), and the fact that the wall was not parallel to the north wall, such a reconstruction would hardly be convincing. (30)

The matter of the expansion of the building to the east is also open to speculation. It seems rather unlikely that the large hall was entered directly from some public space, street, or place. The idea of a colonnade in front of the building finds many parallels in Delos, but none involves such a long portico or is equipped with a stylobate, and certainly not a stylobate of marble. (31) A courtyard belonging to the building must have been defined and closed off by walls, but as the preserved north and south walls of courtyard C were certainly established in a later phase, only further excavation can clarify whether they simply replaced predecessors or were part of a later enlargement of the building.


According to the relative chronology, two different sets of walls, or wall systems, were added to the first gneiss configuration. One consisted of mixed gneiss and granite walls that were thinner than those of the large hall but strictly parallel and orthogonal to them (Fig. 3). They define large parts of the visible D-complex, which was extended to the south and divided into several rooms when it was completely remodeled in a second phase.

The inner partition walls of the D-complex (Fig. 9) are not sufficiently preserved to permit the identification of entrances or establish their relationships in all cases. (32) The west facade (of D1, D4-D6) is certainly bonded with the north wall of D5/D7, which itself, running east-west, divides the entire D-complex. The wall between D1 and D2 is bonded with this major east-west wall, but the corresponding wall between D2 and D3 abuts the east-west wall at its south end and, as noted above, the stucco on the north wall of D1-D3 (Fig. 8).


Other walls that were not bonded with neighboring walls also suggest that a differentiation of the D-complex took place after its extension south. (33) It is certain that the south and east facades of the D-complex were renovated in a later phase, because they differ in material and orientation from the mixed gneiss and granite walls and are not bonded with them (see below, under “Granite Walls,” and Figs. 18, 19). Yet the original accessibility of this sector cannot be determined with certainty, and any discussion of it must be limited to the visible remains of the remodeled building: today, there is no doorway in the south wall and only a gap in the east wall leading from courtyard C to the largest room of the complex, D7. Although a threshold, well-made jambs (i.e., those built with larger, well-cut stones), doorposts (independent elements fitted against the jambs as parts of a door frame), and any other evidence for such features are missing, this gap probably corresponds to an ancient entrance. (34) According to this hypothesis, the D-complex would have been quite a secluded area with highly restricted accessibility. It is open to speculation whether the original east wall of D3/D7 was provided with more or larger doorways. As far as can be determined from its current state, room D7 must have functioned as a distributional space, probably giving access to D1-D3 and D6. If the wall between D1 and D2 continued to the north wall, access to D1 might have been limited to a kind of narrow corridor that remained after the (subsequent?) creation of D4 (Fig. 9).


Without the–entirely hypothetical–reconstruction of windows in the external and partition walls, the lighting of the D-complex seems problematic, because one single doorway to courtyard C would not have provided sufficient light for all the rooms. One of the rooms could have served as a courtyard or a light well, but clear indications (e.g., waterproof pavement, drain) of such a function are absent.

The possible expansion of the building to the south will have had implications for any construction to the east, but, as stated above, all discussion of the latter remains hypothetical.


The second wall system to be added to the gneiss complex comprised only a single wall, the east wall of the large hall A/B (Fig. 10). It was made of gneiss and marble spoils, including inscribed blocks (statue bases and an inscribed dedication base) and plain blocks. (35) Both Bruneau and White ascribed the visible state of the wall to a renovation after 88 B.C., but Bruneau forthrightly admits that his hypothesis is fragile, “nee moins d’un reexamen de la ruine que d’un souci de vraisemblance historique, celui d’accorder la date de l’implantation des Juifs a Delos et celle de leur Synagogue.” (36)


In theory, the east wall could have been conceived as an imposing facade distinguished by its remarkable size and the decorative integration of marble blocks employed in its initial construction. The fact that an arch in the south wall of room B, which served as a relieving arch for the water reservoir, is made of reused marble blocks seems to support this possibility because the arch certainly dates to the first building. Evidence contradicting this view, however, is provided by the marble thresholds (Figs. 11-13). Plassart alone noticed that the steps of all three thresholds are each equipped with two sets of pivot and bolt holes. According to his hypothesis the inner, or east, sets of holes were carved when the threshold blocks (whose original contexts are unknown) were reused, in the first phase of GD 80’s history. (37) The reuse of all three threshold blocks seems quite reasonable because, in Plassart’s opinion, this wall of gneiss and marble spoils belongs to the first phase of the building’s life, which would mean that there was no remodeling of the east wall, but an extensive integration of reused material from the outset.


On the other hand, the prospect that there were two successive east walls suggests that the two sets of pivot and bolt holes–which, it should be noted, are remarkably similar–in each threshold step correspond with two phases of the wall’s history, or that they testify specifically to the existence of two successive versions of this wall. But a detailed analysis of the threshold blocks shows that they had to be adapted to an enlarged east wall and that, in opposition to Plassart, the outer, or west, sets of holes are later than the inner sets.

The majority of known Delian threshold blocks are laid out in such a way that the pivot and bolt holes do not lie within the thickness of the wall, or between the planes of the doorjambs, but just outside both in the interior, lower step of the threshold itself, so that the leaves of the door would not abut the doorjambs but the wall instead. (38) Whereas the inner, east holes of the threshold blocks in question are situated between the planes of the jambs, the outer, west ones correspond to the usual arrangement, being positioned just west of the wall (Figs. 11-13). In addition, parts of the visible wall cover some of the inner pivot holes. Finally, the outer holes are more crudely cut than the inner ones and are carved right next to or even into the west edge of the interior step of the threshold. Given the considerable depth of the thresholds, this seems unnecessary and puzzling as an original arrangement. (39)

For these reasons, the inner, east set of holes must belong to a previous phase of use, the outer set having been used together with the wall visible today. (40) Two conclusions can be drawn from this evidence. First, the elaborate marble thresholds were originally set in a wall of lesser thickness, the later thickening of the wall (Fig. 10) with massive marble blocks requiring a shift of the pivot and bolt holes as far west as possible. (41) Second, the central threshold (Fig. 12) must have been conceived as passable like the other two and used for some time before being blocked by the partition wall of A/B; therefore, this partition wall could not have been built either before or at the same time that the east wall was renovated. (42)

Yet to what extent, and why, exactly, was the east wall remodeled? The nature of the wall’s construction might answer both questions, or at least the first. That the marble spoils were conceived as a deliberate embellishment (which might provide evidence for a new function, or at least a heightened importance, of the building) is unlikely for several reasons. The spoil material is not distributed merely for ease of incorporation or for maximum decorative effect; on the contrary, it is used with great care to stabilize the wall as much as possible. The first visible layer or course consists almost entirely of marble slabs laid as headers that span the entire thickness of the wall. In this way they form a kind of euthynteria for the wall. In the following layers or courses the reused marble blocks are set up vertically and cover only about half the thickness of the wall. The remaining spoil blocks are concentrated in the central part of the wall, framing and stabilizing its critical areas–the doorways. In short, the wall is built with a kind of framework, the marble spoil material serving as frame and studs, and the gneiss blocks as infill. The wall was not simply repaired in its upper parts, but completely remodeled down to its foundations.

To stabilize the wall further, fine white stucco was probably applied, which would have rendered the marble blocks invisible. This supposition depends on the extent of remodeling ascribed to this phase, however, as stucco is preserved only in the joint between the east face of this wall and the abutting north wall of courtyard C (Figs. 14, 15). This exterior corner of the building is only partially excavated; the exposed area includes two marble blocks, separated vertically by an intervening stone, that clearly served to stabilize the outer corner (Fig. 16). But did these blocks belong to the first building, constituting anomalies in the homogenous gneiss walls, (43) or to the renovation? Was the first gneiss facade or the marble spoil wall revetted with stucco? Neither the interior northeast corner of room A nor the southeast corner of room B provides clear evidence for either possibility. The bonding of the north wall of A with the east wall could have been effected easily during the remodeling, but, alternatively, the interior faces of the north and east walls of A might have been retained, the renovation being limited to that outer corner (Fig. 10). (44) The southeast corner of B and the northeast corner of D3 are not preserved sufficiently to permit a determination of how the three walls meeting there were bonded to one another (Fig. 17). Since a large marble statue base of Sosilos (IG XI iv, 1087) occupied the exact point at which the three walls intersect, the remodeling of the east wall was certainly extended to the corner of room B; (45) accordingly, the remodeling might have included the northeast corner of room A, which, as just mentioned, was stabilized with marble blocks like the rest of the east wall. The date of the latter is of major importance with regard to the construction sequence of the whole, as is demonstrated below.


But why was the east wall remodeled in such a careful manner? With its three doorways it was certainly the most fragile wall of A/B and, therefore, most prone to collapse. Obviously, the enlargement of the wall is a response to this danger and not the result of a change in roofing–the other walls were not reinforced at the same time, so the roof could not have been raised appreciably. It is equally unlikely that the addition of a roofed portico required the enlargement of the east wall, for such a portico would have extended farther south (see below, pp. 562, 564) to front the D-complex, the east wall of which seems not to have been enlarged at the same time. It is usually assumed that the wall was destroyed, for example, during the Mithridatic raids and, therefore, needed repair. But this cannot be proven because further evidence for destruction is missing, and the extent of the devastation effected by Mithridates’ troops in Delos is unknown. (46) Thus, the reason for the remodeling must remain uncertain; only its date and the manner in which it was done can be determined.


The granite wall system comprised several walls that were almost entirely composed of irregular large granite blocks:

1. The north wall of courtyard C, which abuts stucco at the northeast corner of A (Figs. 14-16). (47) If this corner belongs to the first gneiss building, the north wall could have been added in a second phase; otherwise, it must have been built after the installation of the marble spoil wall.

2. The east wall of the D-complex (D3/D7), which seems to have been completely rebuilt (Fig. 18). (48) Even though it is clearly bonded with the south wall of D7, the way in which it was joined with the southeast corner of B cannot be judged (as above, Fig. 17). As in the case of the north wall of courtyard C, whether this wall was established before or after the remodeling of the east wall of A/B remains an open question. The north wall of D7, which is bonded with the west facade of the D-complex, abuts the east wall, but not the stucco that is preserved on the west face of that wall in D3. Therefore, the north wall of D7 is not later than the east wall, but contemporaneous or–because the two walls are not bonded–earlier.

3. The south wall of the D-complex (D6/D7). The wall is bonded with the south wall of courtyard C and the east wall of D7, but seems to be abutted by the west wall of D6 (Figs. 19, 20). (49) That its orientation differs considerably from that of the other walls can only be noted, not explained. (50)


4. The south wall of courtyard C, including an entrance that was flanked by two monolithic granite jambs (Figs. 20, 21). (51) The granite threshold is barely visible today because one of the jambs has collapsed and now blocks the doorway. Two shallow cuttings on its north edge suggest that a block was added to form a lower step that would have had a set of pivot and bolt holes. In this way, the door could have been closed properly from the courtyard, clearly indicating that the wall was the south limit of the building. This was doubtless an entrance from the outside, from some public space into the building. (52) Yet its modest construction suggests that this was only a side entrance, conveniently placed in proximity to the possible “service” area D1-D7. A carefully planned hierarchy of doorways within this building is attested by the fact that the only two preserved posts of the doorways to A/B and between A and B are of marble. (53)


Despite their modest construction, the granite walls are of paramount importance to our understanding of the building because they define its actual form, especially that of the courtyard in the east. While they can safely be dated later than the first gneiss wall and the mixed gneiss and granite walls, establishing the sequence of the granite walls and the marble spoil wall is highly problematic.


The last wall system to be discussed involves only the partition wall A/B, which divided the hall into two separate rooms of virtually equal size. Its remains are even thicker than those of the east wall of A/B and consist primarily of large regular granite blocks, small gneiss blocks, and reused marble blocks. (54) Its remarkable thickness might have been associated with an alteration of the roof and ceiling, or it could be attributable simply to the apparent absence of a proper foundation, and to the fact that it is not bonded with the adjoining walls. Of the three doorways, two are still equipped with a marble threshold; the better preserved of the two shows pivot and bolt holes for a door, thus confirming that the entrances between A and B could have been closed by double-leaf doors. Whereas the central entrance in the east wall of A/B was certainly blocked by this huge partition wall, the north entrance, which gave access to room A and was found blocked at the time of excavation, could have been walled up at any time. (55) Once closed to courtyard C, room A was accessible only from room B, which certainly would have had consequences for its use and lighting. Indeed, without additional windows (56) room A would have been rather somber, in any case darker than B and the former hall A/B. The small niche in the west wall of A might have been created for this reason; its reduced size, low position above the floor, and modest, improvised construction with spoil material suggest a function as a lamp niche (Fig. 22). Contrary to White’s view, the off-center position of this niche with regard to room A does not prove that it belonged to an earlier phase; its position is equally off-center with regard to the large hall, the size and lighting of which did not require a lamp or a lamp niche. (57)


Even if the arrangement of the visible equipment with marble benches and a marble throne (see below) is clearly dependent on the establishment of the separate rooms A and B, it cannot be disproved that the furniture had been used in an earlier phase and was adapted to the new situation. (58) The benches (and especially their supports) are partially made of reused material, which could have been integrated from the beginning or added after a destruction or rearrangement of the building. Similarly, there is no evidence that the famous marble throne was carved specially for this edifice and not taken from another building; indeed, several scholars are convinced that this throne was originally set up in the theater, which is a considerable distance from GD 80. (59) The current arrangement of the benches along the north and west walls of A and the south wall of B need not necessarily correspond to the original layout; according to Plassart, the visible marble benches, and several inscribed marble blocks found on and under the benches or at the bottoms of walls, escaped the lime kiln that was later installed in room A only because the neighboring walls had collapsed and covered them completely. Therefore, benches could also have been set up against other walls, for example the west wall of B or even walls with doorways. (60) As previously mentioned, the pavement with marble chips and the stuccoed wall with its red plinth course that certainly served to decorate the large hall by this time were partly destroyed and crudely repaired for the installation of the partition wall, but otherwise were used continuously after the subdivision of the hall.

It is unlikely that the partition wall was inserted for purely practical or technical reasons, because, for example, pillars or columns would have sufficed to support a roof in danger of collapse. The intent must have been to differentiate space or create more rooms, which could be used separately. This view is supported by the fact that benches were set up not only in rooms A and B, but also in courtyard C. As all three spaces were equipped in the same way in this regard, the possible activities performed in them might have been limited and required primarily seating. But the spaces were probably used for different occasions or by different groups.


Although the immediate surroundings of the edifice have not been excavated, the remnants of many walls belonging to neighboring buildings are visible (Figs. 1, 2). No plan has ever been drafted that shows all the visible walls, which would allow us to reconstruct the layout of the Quartier du stade much more precisely than has been possible to date, so only the immediate neighboring walls of GD 80 can be taken into account here:

1. A wall that abuts the northwest corner of room A and extends west (fFig. 23). This wall was certainly erected after the construction of GD 80, perhaps even as soon as the first phase, with gneiss walls, was completed. (61) The size, extent, and position of the building to which it belonged, however, are unknown.


2. A wall that is set against the stucco on the north exterior surface of the northeast corner of room A and leads north (Figs. 15, 22). Its course can be followed for about 45-50 m, to the hypothetical prolongation of the Rue meridionale, which is the southernmost visible east-west road of the Quartier du stade (Fig. 1). This wall formed the east facade of an insula and was equipped with three, probably four, entrances. The entrances corresponded with three or four plots of fairly equal (north-south) extent, which suggests a preconceived layout and a simultaneous execution.

3. Several walls north of room A. One is parallel to the north wall of A, defining a corridor about 2 m wide between the two walls that is blocked by a wall at the west abutting the north wall of room A (Fig. 22). A third wall seems to abut the east end of the east-west wall and extend north from it.

4. A wall that leads north from, and forms a right angle with, the preserved east end of the north wall of C (Fig. 24, foreground). Although the interior corner of the two walls was covered with stucco (Fig. 24, right foreground), it is not clear whether these walls are bonded and therefore might have been built at the same time. (62) If they were erected simultaneously, this would mean that GD 80 extended farther north, or that its extension to the east was effected together with a transformation (e.g., enlargement) of the northern neighbor. Furthermore, it would suggest that this corner marked the east limit of GD 80. But only further investigation and excavation can clarify whether these walls are really bonded and corresponded with the seaward facade of the respective buildings.


In short, GD 80 was obviously integrated into an extended insula, probably marking its south end and serving as the starting point for the development of its eastern part. There was, most probably, no neighbor to the south, at least not south of courtyard C. As there are no doorways in the extant west and north facades of the building, the walls to its west and north certainly did not define rooms that belonged to it, (63) but in all likelihood were constructed for separate neighboring edifices.

Even though the orientation of the building differs considerably from that of all fully excavated buildings in the Quartier du stade, it corresponds with the layout of the Rue meridionale and the many walls visible south of that street, that is, to the south of Ilot II (see above, Fig. 1). Because the history and urban development of this quarter have not been examined to date, it must remain open whether its construction proceeded from south to north or vice versa; whether GD 80 influenced the orientation of its immediate surroundings or was itself conditioned by a preconceived urban layout for this quarter; and whether the building had neighbors to the west and southwest. (64)


Although the visible form and extent of courtyard C can be assigned safely to the granite wall system, the constructions within it could have been established independently, that is, together with another wall complex or even several wall complexes. As mentioned previously, the reconstruction of the eastern part of GD 80 is highly problematic, starting with the first gneiss building.

From the existing evidence, one is moved to reconstruct a courtyard bounded by at least three porticoes. Foundations a and d on Bruneau’s plan (see Figs. 2, 3) are so wide and so clearly parallel to, and equal distances from, the north and south walls, respectively, that they must have delimited corridors or ambulatories. These spaces were nearly as wide as the distance between the existing north-south stylobate and the west wall, and had to be roofed as well because the benches along the north wall of C continue east of the line of the stylobate and were, most probably, not exposed to the open sky. (65)

Yet two details seem to defy such a reconstruction: (1) the northernmost block of the marble step below the west stylobate is worked on its east face with the intention that it should be seen, and (2) it displays on its north face an anathyrosis that extends from the west and bottom edges of the block over most of the block’s width and height (Figs. 25, 26). These points suggest that the west colonnade continued north and that the abutment of a north colonnade with the west colonnade would not have been planned. (66) Two explanations can be offered for this phenomenon. First, the northernmost slab of the marble step is just spoil: the anathyrosis on its north face would not have been visible in another context, such as a peristyle with three (or four) wings, and its east face had been fully worked for the original context but was only partly visible when reused here. (67) Second, the block in question testifies to two phases in the history of the west colonnade: the colonnade originally continued farther north–and maybe also farther south (Fig. 27)–and was only subsequently incorporated into a three- or four-winged peristyle, such a transformation requiring that it be shortened to join with the new north and south colonnades. Further excavation might reveal whether one of these two solutions is correct, or if there could be others. (68)


If a layout with porticoes at the north, west, and south seems possible (indeed, quite likely), (69) it remains to be considered whether the following can be reconstructed more precisely: (1) the extension of the courtyard toward the east; (2) the design of the east facade; and (3) the design of the porticoes (the number of columns, their height, and order).

1. The nature of the extension of the building to the east cannot be determined with any degree of certainty, and the formation of the east limit of the building is entirely open to speculation. Foundation a extends about 6.5 m to the east, and the corresponding north wall of C continues for another 1.7 m (Fig. 2). At present, it seems most likely that the east limit of the building was aligned with the east facade of the neighbor to the north. There is no evidence for the reconstruction of a significantly extended courtyard, although for neither the north nor south wall of C is an outer corner or a clear terminus preserved (Figs. 24, 28). (70) It is equally unlikely that three cuttings in the natural rock at the east end of the north portico, just east of the supposed east limit of the building, are to be associated with the facade of the building (Figs. 24, 29). The largest cutting, which is rectangular, is not symmetrically situated in relation to the north portico and it has a different orientation. The layout of the two flanking round cuttings confirms that whatever they accommodated, on either a fixed or a temporary basis, would not have been oriented either parallel or perpendicular to the building. Therefore, the cuttings certainly were not intended for use associated with a column or supports for a railing that belonged to the building. (71) A satisfactory explanation of these cuttings must await further investigation and excavation. (72)


Since the supposed east limit of courtyard C was only 15 m from its west wall, the restoration of a fourth colonnade at the east is impossible. The restoration of a courtyard bounded by three porticoes is then the only plausible solution. (73)

2. Three configurations of the east facade can be posited: (a) the north and south porticoes ended in columns, were completely open and freely accessible, and with the west portico, formed a three-winged portico (see Fig. 38:a, below); (74) (b) the north and south porticoes were terminated by walls at the east (with or without doorways) and still formed, with the west portico, a fully visible and accessible three-winged portico (Fig. 38:b, below); or (c) the entire complex was closed off by a wall running from the northeast corner to the southeast corner of the courtyard, with access via doorways, or no access at all (Fig. 38:c, below). (75) In the latter case, the colonnades would have defined a truncated three-winged peristyle inside the building and within an actual closed courtyard.


It need not be stressed that the form of the east facade is most important for the reconstruction of the accessibility, openness, and visibility of the complex and the amount of privacy it afforded its users. There might have been a promenade, quay, beach, or path in front of the east facade, because in antiquity the sea was farther east than it is today. (76) Could such a feature have been closed off efficiently and legally and restricted to private use? (77) Could anyone passing by cast a glance at the three-winged portico, or were they confronted with a high, blind wall and diverted to the south facade to find the only, and rather modest, entrance to the building? Did the owners and users of the building seek complete privacy and seclusion or were they, on the contrary, interested in presenting to the public their ambitious, extended, three-winged portico as proof of their wealth and social standing? Or did they just desire a pleasant view of, or direct access to, the sea?

3. The number and sizes of the slabs of the west stylobate can be fairly well determined on the basis of cuttings in the marble step below (which is completely preserved) for vertical dowels, and according to pry cuttings in the gneiss foundation west of this step that were used to position the stylobate blocks. According to these cuttings, the stylobate comprised 10 or 11 slabs that were joined to the step with dowels (Fig. 30). (78)


The number of columns to be restored depends on the bottom diameter and axial span, and in many cases, evidence on the stylobate gives clear indications of these dimensions. However, the single preserved block of the stylobate (L. 2.40 m, W. 0.725 m; block 4 in Fig. 30) exhibits no clear marks for the setting of a column, except for two rectangular sockets with pour-channels that could have served to anchor a column with a bottom

diameter of at least 0.45 m and at most 0.725 m, but probably about 0.60 m (Fig. 31). (79) If no column was fixed on that block, the shortest possible axial span would have exceeded about 3.10 m. (80) A comparison of the colonnades of private and public buildings in Delos demonstrates that such a span was very rarely achieved. Table 1 presents the dimensions of some four-winged peristyles and porticoes.


According to this comparative table, it seems necessary to restore a column on the stylobate block mentioned above, which was neither worked with much care nor very expensive. (81) Since the axial distance between a column positioned on this slab according to the evidence of the pour-channels and the conjectured position of a column on the southernmost block of the stylobate measures about 4.55 m, another column must be inserted between them, which creates an interaxial distance of 2.275 m. To the north, the axial space between the same restored column and the conjectured northernmost column measures about 13.20 m, which could be divided into six intercolumniations of 2.20 m. This would accommodate a total of nine columns, all set up somewhere near, or at, the midlength of the slabs of the stylobate (Fig. 32). Even if the stylobate was not extended 0.25-0.30 m to the north, as hereby proposed, a reduction of the number of columns to eight, with an axial spacing of approximately 2.60 m between the five northern columns, is highly unlikely because, in that case, four columns would have rested directly on the joints between the stylobate blocks, which was usually avoided. (82)


If the north stylobate continued to the supposed east limit of courtyard C, it would have measured about 9.70 m from the northwest corner to the east end; if it abutted an east wall it would have been about 0.60 m shorter. The longer stylobate would have offered space for five columns with an axial spacing of about 2.25 m and the shorter one for four columns with similar placement, as the east wall would have taken the place of the fifth column. (83)

Today several column drums and capitals are to be found in GD 80. The five fluted drums in room A and the Corinthian capital in courtyard C (Figs. 10, 22) were most probably stored here to feed the later lime kiln in room A; they could not have been set up on the stylobate because the lower diameter of the only preserved bottom drum is about 0.90 m, which exceeds by far the width of the preserved stylobate block (0.725 m). In addition, one plain column drum of white marble is today situated near the stylobate in the west portico. Its dimensions, with a lower diameter of 0.58-0.60 m, would allow it to be set up on the stylobate. (84) A plain Doric capital of white marble is just north of this drum, suggesting that the two belonged to the same column (Fig. 25). A more elaborate Doric capital of white marble, with flutes on the necking, is stored west of the building, to the west of the marble throne in room A (Fig. 22). (85)

Clearly, the restoration of the colonnades remains hypothetical. In theory, the columns could have had a bottom diameter of 0.45-0.725 m, but probably it was about 0.60 m. (86) According to Vitruvian standards (Vitr. 4.1.8, 4.3.4), (Doric) columns would have measured between 3.15 m and 5.075 m in height (4.20 m for a bottom diameter of 0.60 m), including the capital. For columns with these diameters in diastyle (Vitr. 3.3.1, 4), the intercolumniation would have measured between 1.80 m and 2.90 m (2.40 m for a bottom diameter of 0.60 m). Vitruvius’s rules, however, are rarely found to have been applied precisely in Delian architecture and cannot, therefore, be used as reliable guides for reconstruction. (87)

Having outlined the conditions that a reconstruction of courtyard C must take into account, can we cite any parallels to support one of the alternative conjectured restorations? For a truncated peristyle closed on its fourth side, parallels can be found in some private houses in Delos, but none involves a space as large as courtyard C. In addition, their builders were operating under certain restrictions: most were additions to existing courtyards with dimensions that were not convenient for the construction of a peristyle. (88)

The symmetrically arranged, open, pi-shaped portico (see Figs. 38, 39, below) would be unique in Delos, as was correctly emphasized by Binder. (89) Neither the porticoes with short projecting wings, such as the Portique d’Antigone (GD 29), which opens onto a large public/sacred space, nor the large pi-shaped portico in the Sanctuaire des Dieux syriens (GD 98), which frames the cult theater, nor the large pi-shaped portico to the south of the Samothrakeion (GD 93), can really serve for comparison. Such a structure as part of GD 80 would have been a rather ambitious monumental entrance with three wings, for which, even outside of Delos, exact parallels are scarce. One could, however, compare it with similar arrangements in some of the Hellenistic sanctuaries that are laid out on several terraces, such as the sanctuaries of Athena on Lindos, Asklepios on Kos, or Fortuna in Palestrina. (90) In addition, the scheme of a three-winged portico recalls the many coastal porticoed villas that appear on small square or horizontal oblong panels in Roman wall painting. (91)

The architectural repertoire of the Hellenistic and Roman periods provided an extensive spectrum of colonnades, stoas, and porticoes that was drawn on with great variety: some were pi-shaped, some had short projecting wings, some framed open spaces, some served as vestibules or entrances to buildings, some (except colonnades) were freestanding buildings, and so on. (92) But with regard to GD 80, the absence of precise local parallels might be attributed to the late date of this pi-shaped portico, which was probably constructed in Roman times, more precisely in the Imperial period (see below, p. 565). Perhaps this would also explain the discrepancy between the portico’s apparent aspirations to monumentality and the modest quality of the entire C-complex.

In the space within the colonnades, no pavement has survived and it is likely that it never existed (Fig. 2). Since a waterproof pavement with drain may be considered standard equipment for courtyards of even very modest houses, (93) its absence here requires explanation: (1) it is simply not preserved; (2) the modest construction of complex C could not include such an expense; (3) it was not necessary, because the open space bordered by the colonnades never functioned as a courtyard, but was freely accessible from a similarly unpaved path on the side adjacent to the sea, which would have allowed rainwater to be drained easily toward the sea; or (4) vegetation was planted in the space, forming a kind of garden. (94) None of these solutions requires a grand staircase with three steps leading from the courtyard into the west portico. (95) Although the original level of the space is unknown, it seems unlikely that the foundation (or even perhaps the euthynteria) of the stylobate was visible. Therefore, the lower step of the two-step stylobate was entirely sufficient if one wanted to enter the portico directly from the open space of C. (96)

To sum up, together with the granite wall system, which defines courtyard C in its visible limits at the north, south, and southwest, a three-winged portico or peristyle was established within its confines (Figs. 3 and 38, 39, below). On the basis of an architectural analysis, restoration of the following elements is proposed: nine columns on the west with an axial spacing of 2.20-2.275 m; four or five columns each on the north and south with an axial spacing of 2.175-2.25 m (Fig. 32); and possibly plain marble columns of the Doric order with a bottom diameter of 0.60 m and a height of about 4.20 m.

The design of the east facade is unknown, but it involved either (1) a rather ambitious monumental exposure with a pi-shaped portico that opened onto the sea and framed an open, freely accessible space; or (2) a simple blind wall that concealed a three-winged peristyle within a closed courtyard. The pi-shaped portico or three-winged peristyle might have been built as part of a single construction project that included the integration of reused material, or perhaps it was merely an ornamental enlargement of the complex. In the latter case, the north and south colonnades would have been added to the existing west colonnade, which in turn would have to have been shortened in order to accommodate the new layout. The added colonnades could have been constructed using material of lower quality, in correspondence with the north, south, and southwest walls of courtyard C, and therefore might not have been preserved as well as the west colonnade.


A lime kiln was installed in room A in an unknown period (Figs. 2, 3). Its position in the middle of the room seems strange, but the room was probably already largely destroyed and filled with debris by the time the kiln was built. In fact, rubble might have served as support for the structure; lime kilns were ideally built into slopes “in order to take advantage of an efficient constant temperature and easy access to the lower part for the fire and to the upper part for loading and unloading.” (97) If the doorway between room A and courtyard C, which was found blocked when excavated, had been blocked before the installation of the lime kiln, the latter must have been loaded and fired from room B. In any case, the kiln’s position suggests that the building was a convenient source of marble that could be extracted from the building itself and the surrounding area. The fact that almost nothing is left of the colonnades clearly supports this suggestion.

Several walls cannot be ascribed to any of the described wall systems, nor can their functions be ascertained, namely the “late” walls built at what had been the west ends of the north and south porticoes, and the walls on foundations b and c, whose west ends rested on the euthynteria of the west colonnade (Figs. 2, 3). All four were certainly erected after the installation of the colonnades and could well have been part of one program, the precise nature and date of which is unknown, however. Whereas the walls in the north and south porticoes seem to have blocked traffic to and from the west portico, the function of walls b and c cannot be determined. They might have flanked a kind of ramp that facilitated the transport of heavy items (to feed the lime kiln?), but such a ramp would not have ended opposite the door to room B. Moreover, even the installation of a mere ramp may be considered an unlikely expense in association with a simple lime kiln.

In conclusion, the last use of GD 80 can be only partly reconstructed. Whereas the kiln was operated in a period when the building was certainly abandoned, walls b and c, and the “late” walls that extended north and south from the west stylobate, could mark a phase when the building was reduced in size–the eastern part no longer being in use–but still quite intact and inhabitable.


The different wall systems and the constructions to which they belong have been defined and described, but they have not yet been assigned to specific phases. This is the objective of the following section, which is not limited to a mere enumeration of the separate measures: the character and purpose of each phase, as well as the possible function of the building, are analyzed and discussed.


The gneiss wall system can be attributed safely to the first phase; it formed the original nucleus of the building to which all further wall systems were added (Fig. 33). This edifice was conceived as a freestanding building and comprised the large undivided hall A/B, with three doorways equipped with marble thresholds, and the large water reservoir, which was accessible from the south. Although the building was without doubt extended to the south, the exact limit and structure of this southern part (the number and character of the rooms) cannot be reconstructed. The reservoir, however, was most probably integrated into the building. With regard to the eastern part, we can only speculate. Given the monumental size of the hall and the fine quality of its construction, one would expect a courtyard, a monumental entrance, an elaborate facade, or some combination of such elements. Several alternatives are possible:


a. A “public” colonnade, freely accessible from all sides and opening onto public space (Fig. 34:a). Initially, this reminds one of the numerous public rues a colonnes in Delos, but their function as support for a projecting upper floor and their modest character (mostly granite columns, without stylobates) differ considerably from the visible remains of the north-south colonnade of GD 80. (98) An open marble colonnade in front of a building that is raised above the walking level of the corresponding street or space would be unique in Delos but is, nevertheless, conceivable as a slightly simpler version of the following alternative.


b. A colonnade closed off at the south and north ends that clearly would have belonged to the building, decorating its facade and serving to create a kind of vestibule (Fig. 34:b). It would, nevertheless, have been easily accessible from the public space to the east. The preserved north-south colonnade is conceivable in such a context, with an extension toward the north corresponding with the north wall of A and an unknown extension at the south. Yet it must be stressed that nothing remains of a north wall that is to be associated with this colonnade. On the contrary, the visible stuccoed exterior northeast corner of room A clearly speaks against such a possibility if this corner really was part of the first building. (99) Comparable colonnade-vestibules are found in several public and sacred buildings, opening into a space within a temenos or a public space. (100) In either case, such a colonnade would have distinguished the building as public or sacred.

c. A simple courtyard without a colonnade and with unknown extent and accessibility (Fig. 34:c). This alternative requires a predecessor of the visible north wall, which creates the same problems encountered in alternative b. Given the size of the hall, for which parallels are cited below, such a plain courtyard is very unlikely.

d. A courtyard with a single colonnade in front of the western rooms (Fig. 34:d). As this is a modified version of alternative b, with an extension of the framing walls eastward, the pros and cons need not be discussed again. As with alternative c, however, convincing comparisons for the combination of such a large hall and a courtyard with a single colonnade are missing in the Delian record.

Since important objections have been raised to each of these alternatives, none of them can be safely and convincingly supported. Despite all the factors that weigh against it, it cannot even be entirely ruled out that the large hall was entered directly from public space. For now, alternative a is perhaps the most plausible and attractive one. This choice is supported by the fact that the long wall that later abutted the north face of the northeast corner of GD 80, and acted as the east facade of an insula, is aligned with the east wall of the large hall. Therefore, the latter could indeed have been the first facade of the building, distinguished at most by a colonnade in front of it. Both GD 80 and its northern neighbor later would have been extended eastward in correspondence with each other. (101)

Regardless of how the eastern part is reconstructed, the features that are safely assignable to the first building clearly exclude it from identification as a private house, which has, only recently, also been argued extensively and compellingly by Binder. (102) Even though all Delian houses include one large room, an oecus maior, that could have been equipped with three doorways, the dimensions of the hall of GD 80 far exceed the standards of private dwellings. (103) Further, the private oeci maiores are all integrated in a well-defined domestic context: they open onto a courtyard, usually provide access to one or two annex rooms, and are surrounded by additional living and service rooms. (104) By contrast, the large hall of GD 80 has convincing parallels in the large assembly rooms of buildings that served for the meetings of associations, such as the Etablissement des Poseidoniastes (GD 57, Fig. 35). These are also accessible via three doorways, have no annex rooms, open onto large peristyle-courtyards, and were clearly conceived to meet all the needs of the particular association, including assemblies and banquets (or symposia). (105) GD 80, however, lacks some features that have been identified as essential components of meeting places in Delos (none of which is situated on the shore): a large peristyle-courtyard; a latrine; commercial space such as shops, magazines, or workshops; and “sacred” space or objects such as shrines, niches, altars, stelai, and figurines that attest the veneration of gods. (106) In addition, the orientation of the large hall under discussion is remarkable because it does not open to the south, as is usual in Delos and elsewhere. (107) In comparable cases a differing orientation is due primarily to external factors, which seem to be missing here, since the building was established before its neighbors and without visible constraints regarding urban layout or topography. (108)


The urban layout being unknown, it is, moreover, difficult to determine whether the large hall was deliberately erected over the natural gap in the rock that was transformed into a water reservoir. Is this arrangement –clearly unfavorable within the Delian context–attributable only to the predefined position of the plot of land?


Two wall systems were added to the first gneiss building, the mixed granite and gneiss walls and the marble spoil wall (Fig. 3). One of them may belong to the second phase of reconstruction. The mixed granite and gneiss wall system is identified here with the second phase because it is the most similar in technique, material, and quality to the gneiss wall complex (Fig. 36).


This phase probably comprised an extension of the building to the south and the installation of several rooms, some of which seem to have been subdivided subsequently, thus providing more rooms that could be used separately. As none of the features of these rooms (pavements, stucco coating, drains, installations) or finds are preserved, it is impossible to judge their purpose and character–service rooms, storerooms, or simple living quarters? Because of their sizes, forms, and positions, especially in comparison with the large hall A/B, the possibility that they functioned as luxurious assembly or banquet rooms can be largely ruled out.

The reconstruction of the area to the east remains problematic, as discussed with regard to the first phase (Fig. 34). In theory, a colonnade could have been added at this point. If such a colonnade existed from the first phase onward, it would have to have been extended farther south during the second phase, in correspondence with the conjectured extension of the south room complex.


In the sequence proposed here, remodeling the east wall of hall A/B by integrating reused marble material is assigned to the third phase of the building’s history (Fig. 37). In theory this modification could be assigned to the second phase, but the remarkable differences between the respective wall systems do not support such a possibility. Whereas the renovation of the east wall undoubtedly took place after the installation of the gneiss wall system (the first phase), establishing the sequence in which the marble spoil wall and the granite wall system appeared has proven to be particularly difficult, as described above. Both features are of paramount importance to our understanding of the construction history, the spoil wall because of a terminus post quem of 88 B.C. for its realization, the granite system because it–finally–clearly defines the problematic eastern part of the complex. Since the alternative favored here is that the renovation of the east wall of A/B involved the entirety of the wall, from the northeast corner to the southeast, the granite wall system is assigned to a later phase. (109)


The marble spoil material was distributed with great care and consideration, in such a way that it formed a kind of stabilizing framework for the new wall. In addition, the wall was probably completely covered with stucco, which would have rendered the supposedly precious and prestigious reused marble material invisible. (110) Therefore, the intention in this phase was not to embellish the building (111) and increase its prestige, but merely to rebuild and carefully stabilize a wall that was probably destroyed or partly collapsed, perhaps as a consequence of the Mithridatic raids in 88 B.C. Nor did this measure alter the essential character and appearance of the building.

Again, nothing can be said with any degree of certainty about the eastern area. The north-south colonnade may also have been erected during this phase, but, in contrast to the remodeling of the east wall, this clearly would have been a result of the desire to embellish the building and enhance its prestige, and the problems with this possibility have already been discussed with regard to the first and second phases. Additionally, since the stuccoed exterior northeast corner of room A is attributed to this third phase, the reconstruction of a wall bounding the west portico at the north (Fig. 34:b-d) certainly must be excluded. (112)


The chronological sequence of the wall systems again offers two alternatives, because both the granite wall system and the huge granite-gneiss-marble wall that divides hall A/B into two separate rooms were built after the marble spoil wall had been completed. The remarkable difference between them speaks against a contemporaneous realization of the two, but there is little evidence for giving primacy to either one. The granite wall system is chosen here because the similar furnishing of A, B, and C with benches is dependent on, and therefore subsequent to, the bisection of hall A/B, and presupposes the existence of courtyard C in its visible form (Fig. 38:a-c). (113)

This phase involved minor changes in the south, conceivably as part of the renovation of the south and east walls of the D-complex that was presumably motivated by a partial collapse of the respective walls or rooms. The principal result of the measures taken in this phase was, however, an enlargement of the building to the east. This comprised either the complete installation of a pi-shaped portico or a three-winged peristyle, or the transformation of a preexisting north-south colonnade into a pi-shaped portico or a three-winged peristyle. Since at present it seems most likely that the tripartite structure was not closed by an east wall but opened onto the sea, allowing free access to, and a view toward, the building, it must have served as a monumental, grand facade and thus certainly increased the building’s prestige.

Despite this considerable change in appearance, the general function and character of the building were not fundamentally altered. The large hall and the southern room complex still could have been used as they had been in the previous three phases. In addition, numerous activities could have taken place within the porticoes, such as shaded promenades, the gathering of small assemblies, outdoor relaxation (possibly with a view of the sea), the setting up of votives and basins, and washing. Whereas the terrain north of GD 80 was certainly built up and still in use in this phase, the entrance in the new south wall of the courtyard suggests that there were no neighbors immediately to the south.

Since the third phase occurred sometime after 88 B.C., the fourth phase must be dated accordingly, and could be assigned either to the last days of the Roman Republic or to the Imperial period. A late, Imperial, date might be supported by three antefixes with palmettes that were found in the water reservoir. (114) Bruneau compares them to examples of the first half of the first century A.D., and even to some from the middle of the third century. If they belonged to this building, the earlier date is more likely, (115) and they would testify either to a repair of the existing roof or to the construction of a new roof, for example, for the porticoes, in this period.

According to current research, Delos was largely abandoned in the Imperial period and was reduced to a small settlement that occupied only a minor part of the former Hellenistic city–the western coast and the quarters immediately south and north of the Sanctuary of Apollo. (116) The Quartier du stade is usually thought to have been destroyed and deserted after 88 B.C. and, accordingly, must have been completely isolated by the end of that century. Therefore, the people who frequented GD 80 would not have lived nearby, but instead somewhere in the western part of the island.

This supposition requires comment. Two possible patterns of use can be conjectured. The first is that the users of the building in this phase (and probably in the previous phase) had owned it before 88 B.C., even from the beginning, and continued to frequent it because of its special importance and function. The second is that the original owners abandoned the building after its partial destruction or the decline of the quarter, the new proprietors choosing it deliberately because of its isolated location and transforming it according to their needs. It must be stressed, however, that some construction took place in the area to the north at the time of or after the completion of the granite walls. Also, the small harbor immediately north of GD 80 still could have been in use. One of the buildings on the shore might even have been fronted by a colonnade with a purely functional or decorative character. (117) For these reasons, the vision of a vast deserted quarter on the eastern shore of Delos must be considered with caution, as it may be attributable to the limited excavation of this quarter or to the modern interpretation of the archaeological evidence, or both, and need not fully correspond with the ancient reality.

Although building activities in Imperial Delos were largely confined to the reoccupation and remodeling of extant buildings, some new buildings were constructed, at least two Roman baths, for example. (118) Even if, as the finds suggest, commerce in Imperial Delos was mainly oriented toward the eastern Mediterranean, (119) there must have been some Roman influence, tangible for example in the presence of Roman thermal edifices and associated bathing customs. Within this larger context, the ongoing construction at GD 80 is more easily comprehensible, especially the strange amalgam of ambitious scale and modest construction that characterizes this phase. For the persons who owned and used this building it still must have been worthwhile to establish a three-winged peristyle or a monumental pi-shaped portico that dominated the small harbor. The unusual form of the three-winged portico might have been a result of Roman influence.


The last appreciable alteration of the still fully usable building was the division of the large hall into two rooms by means of the large granite-gneiss-marble wall (Fig. 39). The consequences of this measure have already been discussed, but it should be emphasized again that the differentiation of the space did not necessarily provide two rooms that could be used entirely separately with the mere closing of the connecting doors. After the doorway between the west portico and room A had been blocked–either at the same time as the installation of the dividing wall or at some later date–room A was accessible only from room B. The two rooms would have formed a kind of interconnected room-suite, with B as a better-lighted large vestibule or front room and A as a darker, more secluded back room.


Such a layout suggests a clear hierarchy of use that is further supported by the presence of the marble throne in room A. Since the furnishing of the two rooms seems to have been identical, with the only apparent exception being the marble throne, the activities performed in these rooms cannot have differed considerably. Even if it cannot be ruled out that the visible furniture was introduced in one of the previous phases of use, it is evident that part of the furniture consisted of reused material and that not all of it was carved specially for this building–including the marble throne, which is likely to have originally been set up in the theater.

This conclusion is supported by historical circumstance: although all phases after 88 B.C. (third through fifth) could have followed shortly one after the other, the fifth phase can quite safely be assigned to the period after 69 B.C. (or to a much later date), when heavy reuse of marble elements was normal in Delos. It would be interesting to know when the owners of GD 80 obtained permission or could simply remove the furniture of such a prominent public building as the theater; in any case, they put considerable effort into transferring this throne from its original location, in the Quartier du theatre south of the city center, to their edifice on the eastern coast, presumably because they had designated it for a special purpose.

The remaining materials that have been found in this building, such as votives, a sundial, basins, lamps, and other finds, could have been used in the last phase. Yet some objects date from the end of the second century B.C. through the first century B.C., and therefore might have been used in previous phases or transferred from elsewhere and reused.


Some constructions, such as foundations b and c in courtyard C and the “late” walls crossing the north and south porticoes at their west ends, could have been effected before the abandonment of the building and, presumably, would have altered the nature and use of the courtyard. For example, the north and south porticoes were blocked and an installation was constructed within courtyard C that abutted the west stylobate. Alternatively, some of these walls may have been erected in association with the installation of the lime kiln in room A, which was certainly added after the abandonment and partial decline of the building. Hereafter, this last, heterogeneous, phase will not be taken into account.

As a result of the foregoing reconstruction of the history of GD 80, some major points and problems have emerged:

1. The details of the extension eastward, the formation of the east facade, and the matter of accessibility (with regard to a main entrance), cannot be determined with any degree of certainty for any of the five phases (Figs. 34, 38, 39).

2. The edifice was conceived as a freestanding complex, but was successively surrounded by buildings in the north and west, the construction of which could have begun shortly after the first phase and continued into the fourth phase, at least.

3. The two most characteristic features of the building are the large hall, which dominated the building from the first through the fourth phases, and the water reservoir, which was extant and usable in all phases.

4. Two fixed points help determine the absolute chronology of the phases. The first two phases can be assigned to the period before 88 B.C., the third through fifth phases occurring sometime between 88 B.C. and the end of the second century A.D., when the building was abandoned. Without further research, however, it cannot be determined when exactly the building was constructed, nor how long each phase lasted. Does the important and formative fourth phase, for example, date to the first century B.C., the first century A.D., or to the second century A.D.?

To this point, the building’s construction history has been described and analyzed without reference to its possible owners and users. Even if it is now generally agreed that the visible building, that is, the building of the fifth phase, was used as a Jewish or Samaritan assembly hall, it remains to be discussed when this function was assigned to the building. Was the edifice constructed as a synagogue or was it transformed into one, and, if the latter, when?


The identification of a Jewish or Samaritan use for GD 80 is based on epigraphic and architectural evidence. (120) However, Table 2 clearly demonstrates that inscriptions and furniture have always been given priority over the architectural elements and design. It is sufficient to cite the opinions of just three scholars–Bruneau, White, and Binder–because no other substantially differing views have been presented in the literature. (121) If the informational value of these respective criteria–inscriptions, furniture, and architecture–is examined, however, it becomes apparent that their order of importance must be changed.

Five small inscribed votives were found within the building. Four of them include vows to a [TEXT NOT REPRODUCIBLE IN ASCII.], a “God Most High.” Although the identity of theos hypsistos and the nature of his cult are still discussed, it is generally agreed that this epithet was certainly, though not exclusively, used by Diaspora Jews (and also Samaritans) to refer to their god. These inscriptions are regarded as primary evidence for the identification of GD 80 as a synagogue. (122) Even though the earliest two votives are dated to the first century B.C., they cannot testify with certainty to such an early Jewish or Samaritan use of the building because they, like the other three, are small and movable and might easily have been transported from one building to another. Therefore, the possibility that the two oldest votives were first set up in another building and were transferred to GD 80 only in the last (fifth) phase of its use cannot be ruled out.

There is, indeed, an ongoing discussion about three other Jewish and Samaritan inscriptions. One was discovered in a private house nearby, in the Quartier du stade, and the other two, on two stelai, were found in an unexcavated area some 90 m north of GD 80. Did these inscriptions originally belong to GD 80, being displaced in a later period, or were they discovered in their original contexts, thus bearing witness to Jewish or Samaritan ownership of the respective buildings? (123)

Similarly, the use of the benches and throne in the last phase can be proved, but remains hypothetical for all previous phases. This holds equally true for all other movable furniture that has been found in the building, for example several water basins (Figs. 40-42). (124) According to the construction history proposed here, the fifth phase does not correspond with a formative restoration after a possible destruction in 88 B.C., but follows much later in the construction sequence.


The artistic embellishment suggesting a Jewish or Samaritan provenance is to be regarded with extreme caution. Palmettes, which appear on the marble throne, on antefixes, and on a marble lintel (of the third century A.D.), and rosettes, which decorate an inscribed votive offering, might be among the most prominent motifs of Jewish and Samaritan art, (125) but they were certainly no less prominent in non-Jewish and non-Samaritan pagan art. (126) Therefore, it is difficult to decide whether the decorated objects were, at least to some degree, specially made for Jewish or Samaritan use, whether they were deliberately chosen out of a large stock of spoil material for Jewish or Samaritan reuse, or whether no special meaning can be assigned to their presence in this building because of the extensive diffusion of these motifs throughout the ancient world.

For Binder, the major architectural change that marked the transformation of the building into a synagogue, after 88 B.C., is the subdivision of the large hall, probably to create two rooms for the segregation of the sexes. (127) That this occurred, most probably, in the very last phase need hardly be drawn to our attention. Contrary to Binder and White, Bruneau’s decisive dating of the architectural alteration to some time after 88 B.C. involves only the embellishment of the east wall of the large hall, leaving the latter’s bisection (and characteristic furnishing) to an even later date. But if the architecture did not change at all during Bruneau’s first and second phases, why then could the original building not have been conceived as a synagogue? In other words, what exactly identifies the building of the second phase as a synagogue?

Such uncertainty with regard to the original edifice is even more surprising because Bruneau argues compellingly and extensively that the water reservoir was conceived and used for ritual bathing. Since this reservoir, with its remarkable design, belongs most certainly to the very first phase, its identification as a Jewish ritual bath necessarily requires Jewish ownership and use of the building in the first phase. (128) Clearly, before discussing further evidence for the identification of the building in its first four phases, it must be determined whether the hypothesis of a ritual bath can be supported.


What requirements had to be met by a Jewish ritual bath, and how can a potential installation for ritual bathing be recognized? The Jewish laws provide an elaborate system of regulations regarding ritual purity. The required state of purity could be obtained by immersion in a natural body of water–a spring, river, lake, or the sea–or in a pool specially designed for this purpose. Only the latter, which had to be carved into the bedrock or built directly into the ground, was called a miqveh; it had to be filled with at least 40 se’ah (250-1000 liters, according to various calculations) of undrawn water, which was either collected rain water or spring water delivered by an aqueduct. Steps occupying the entire width of the miqveh, or narrow staircases on one or two sides of the basin, led down into the pool. (129) To date, the earliest known miqva’ot are located in the palaces of the Hasmoneans in Jericho, on the Temple Mount in Jerusalem, and in synagogues in Jericho and at Gamala, Masada, and Herodion. Reich, the authority on miqva’ot, has stated that, “frequently used in the Second Temple period in Judea (Judah) and Galilee, miqva’ot were absent from the Late Hellenistic and Early Roman world.” (130) Therefore, if the water reservoir in question was a miqveh of some kind, it would be one of the earliest examples in the world, the earliest associated with a synagogue building, and the only one outside Palestine.

How, then, was the water reservoir of GD 80 constructed, and how is it to be judged within its local context? In Delian architecture, wells, cisterns, and large water reservoirs were usually located in a courtyard. Small ordinary wells were carved into the natural rock, built up to the sufficient height with gneiss or granite blocks, and had a small opening that was framed by a stone border or curb (puteal). Larger reservoirs and cisterns were also cut into the rock and built up, but had to be roofed with the aid of arches or beams. Whereas the arches were made of poros or granite, the beams were wood or stone. The walls of cisterns and larger reservoirs were covered with waterproof stucco, which was not deemed necessary for the narrow wells. It is self-evident that, for reasons of stability, the placement of walls or stylobates and colonnades on top of the arches or beams was avoided. (131)

Within this context, the water reservoir of GD 80 has both normal and abnormal characteristics. (1) A natural gap in the rock with a total length of 6.08 m was partly built up and roofed with a poros vault that extended 4.00 m beneath the floor of room B. According to Bruneau, the remaining 2.08 m, visible in room D1 (Fig. 43), was open to the sky and freely accessible, and the marble arch in the south wall of B would have been constructed intentionally for this purpose (Fig. 44). However, this arch was necessary because of the enormous size of the large hall A/B and the weight of its roof, which rested on the walls, among them the south wall of B; thus, it served simply as a relieving arch. Furthermore, it certainly would have been walled up and covered with stucco in room B and, most probably, also in D1, and thus would not have been visible at all. (132) (2) The large opening in D1 is framed by the natural rock, which forms a completely irregular border. This border would have been highly inconvenient for any use of the reservoir, either the drawing of water or bathing, which suggests that the opening was originally stabilized and framed by walls and probably even covered with beams of stone, or, more preferably, wood. Like the floor on top of the vault in room B, this construction could have collapsed or, if it was wooden, even burned. (133) (3) Finally, there are no steps leading down into the reservoir, the walls of which are very steep and high. Bruneau imagines a wooden ladder or stairs, which would have been quite a simple and makeshift arrangement, especially when compared with the elaborate and costly construction of the poros vault and the marble relieving arch. If human access to the water reservoir was intended from the beginning, or in one of the subsequent phases, it would not have caused too many problems either to cut steps into the rock or to build a simple staircase of stone. (134) On the whole, the regular and frequent use of the water reservoir for bathing or immersion is to be seriously doubted.


By Jewish standards, the Delian water reservoir would not have qualified as a miqveh; it could, at best, have served as natural body of water because it was filled by groundwater like a well. (135) Given the proximity to the sea, which could have served the same function, the identification of the water reservoir as a bathing pool becomes even more questionable. It is more likely that the water was used like it was in many other buildings, that is, for all activities that required fresh water: drinking, cooking, washing, cleaning, and so on. Moreover, the several marble basins that were found in the building, and that might have formed part of its furnishings in the last phase and perhaps even in previous phases, could have held water for ritual ablutions. (136)

In sum, to base the identification of the entire building on the possible function of this water reservoir is not particularly convincing, and might easily provoke quite justified protest. It is strange that this water reservoir and its conjectured purpose have gone largely unaddressed, and that even those scholars who have identified it as a miqveh (137) have never discussed it in full detail. If the reservoir was not conceived as a special facility for a specific function, its strange position might simply be due to geologic and topographic circumstance. In addition, the area around its opening could have formed a small courtyard in all phases. (138) In conclusion, the water reservoir of GD 80 is not considered here to have been a miqveh.


Another element that can be assigned to the original building with certainty is the large hall. Its size speaks of public, but not necessarily Jewish or Samaritan, use. Its three doorways might be typical of synagogue architecture, as argued by Bruneau, but since many large rooms in Delian architecture–in private domestic buildings as well in those of a “semipublic” nature used for the meetings of associations–are provided with three entrances, this arrangement can hardly be regarded as decisive evidence for the identification of Jewish or Samaritan use. A large room requires a large number of openings, preferably more than one doorway and windows, for technical reasons such as lighting and easy access for possibly numerous users, (139) and for symbolic reasons such as the need or desire to mark its importance and prestige.

Even though the orientation of the large hall toward the east is remarkable, it need not necessarily have been chosen with regard to specific cultic practices or requirements, for example as an intentional imitation of the inner court of the Temple at Jerusalem or of that at Gerizim, or “to direct focus in the general direction of the Orient.” (140) It might be attributed merely to the wish to have a pleasant view of the sea or to present a prestigious facade to the sea. (141)

Whereas the design of the large hall speaks clearly neither for nor against Jewish or Samaritan use, two other factors do seem to testify to Jewish or Samaritan initiative with regard to the original purpose of building: (1) the overall plan, structure, and elements of the edifice, and (2) its location on the seashore. The opponents of the domestic use theory have compared the original building with buildings used for the meetings of associations and, indeed, identify it as the cultic hall of a pagan association. (142) Yet there were different kinds of associations in Delos; the fact that the Etablissement des Poseidoniastes (GD 57, Fig. 35) is cited as the only architectural parallel makes clear which type of association the supporters of the meeting-place theory have in mind: not one connected with the cultic thiasoi, which gathered mainly for the veneration of their gods, but one of foreign merchants with a broad range of common interests involving commercial, national, cultic, and social aspects. (143)

A comprehensive comparison of GD 80 with all constructions in Delos that might be attributed to use by an association is beyond the scope of this article. The following comparison is limited to meeting places that are similar to the Etablissement des Poseidoniastes and might, therefore, have been built and used by a similar association. A comparison with such buildings has been presented above and has shown that GD 80 lacks some features characteristic of this type of structure in Delos: commercial space (shops, workshops, magazines), sacred space (chapels, niches) and specific equipment, (144) a latrine, and, most probably, a large peristyle-courtyard with a proper vestibule.

As the fittings (such as pavements, stucco, drains or other installations, furniture, equipment, and small finds) of GD 80 in its first phase cannot be reconstructed, comparisons must be restricted to the architecture. Common traits shared by meeting places and GD 80 are limited to a large hall, a water supply, and some additional rooms; a colonnade might have embellished the facade of the building but certainly would not have compensated for a closed, fully usable peristyle-courtyard.

If the original hall A/B, whose fittings and decoration are only partly known, (145) was used in a way similar to that of its counterparts in pagan meeting places, it was multifunctional and could have served for banquets, symposia, and assembly meetings, but nothing hints at a pagan cultic use of this room. (146) With its unusual size, it must have been conceived for large meetings from the beginning; it easily could have offered, for example, space for 20-22 or even more couches set up on the north, west, and south walls, and it certainly could have accommodated many more than 40-44 symposiasts stretched out on couches, if gatherings included seating on benches. (147)

The limited repertoire of separately usable rooms in GD 80 did not change considerably in phases 2-4 if these phases involved only the addition of some small rooms in the south and an open three-winged portico in the east (Figs. 36, 37, 38:a, b). But with the possible addition of a closed peristyle-courtyard in the fourth phase (Fig. 38:c), the building would have become similar to local meeting places for associations and would have offered space for a new range of activities.

In sum, before the fourth or even the fifth phase, there is no evidence for a remarkable alteration of the building that would clearly attest a change of ownership or use. (148) Further, the architecture suggests that this edifice was erected by a group whose needs were different from those of the associations that used the known association buildings in Delos. This view is supported by the unusual location of GD 80.

Although the position on the shore has often been noted and claimed as specifically apt for Jewish or Samaritan use, nobody has explained it with regard to the supposed non-Jewish or non-Samaritan, or private or semipublic, use of the first building. (149) The adherents of the private house theory could find some parallels on the western coastline, but the supporters of the meeting-place theory would have considerable difficulty in finding convincing comparisons. (150) Given the clearly commercial aspects of the cited association meeting places, the remote position of GD 80 in a mainly residential, noncommercial quarter does not support the thesis that a pagan association with commercial interests deliberately chose this place for its meeting place. (151) Therefore, it seems most likely that the building was planned and realized by a group that gathered neither because of common commercial interests nor in order to venerate pagan gods in temples or chapels and with sacrifices–and this very well could have been a Jewish or Samaritan group.

Indeed, three arguments have been advanced for a specific Jewish or Samaritan choice of the location on the seashore: (1) the proximity of water for ritual ablutions; (2) the proximity of water for reasons of purity (worship near water ensures the necessary purity of the land); and (3) the marginalization of Jews and Samaritans in the society, which meant that a location far from the city center was required. (152) The third argument can be ruled out because there is no evidence for an intentional official separation of the Jews or Samaritans within Delian society, or for a restriction of Jewish and Samaritan rights before the 40s B.C.; (153) moreover, the structure’s integration into an insula within a larger urban plan might have been intended from the beginning, even if it was realized only subsequently. But the first two arguments could well have induced a Jewish or Samaritan group to build its synagogue in this place, which, in addition, was probably much quieter than the commerce-oriented western shore and in an “idol-free” quarter with no pagan sanctuaries. (154)

The preceding analysis of GD 80 has shown that it was, in all probability, a synagogue from the time of its construction in the second century B.C. until its abandonment at the end of the second century A.D. But three questions remain: (1) Can a comparison with the structure and function of other, safely identified synagogues further strengthen this theory? (2) To what extent are the surviving fittings specifically Jewish or Samaritan? (3) Does the archaeological evidence, that is, the construction history of the building, correspond to the history of the Jews and Samaritans in Delos as understood from the literary sources?


Since the synagogue of Delos has always been regarded as the oldest known example of a synagogue, no precedents or precise models can be cited or compared with it. The shift of the date of its original construction from the first century B.C. to the second now distances it even further than before from other examples. (155) To further complicate matters, the existence and the function of synagogue buildings before the destruction of the Temple in Jerusalem in A.D. 70 have remained the subjects of controversy in recent scholarship. (156)

A consensus has finally emerged, however, that synagogues certainly did exist before A.D. 70 in Palestine and the Diaspora, and that they were multifunctional, serving both sacred and profane needs. They accommodated all the different aspects of Jewish communal life–political, liturgical, social, educational, judicial, and spiritual–within their walls. Consequently, the buildings were used for various activities, such as ritual washing; the teaching and study of scripture, especially the communal reading of the Torah and the expounding upon and study of it; corporate prayer and worship; customary feasts and communal dining; storage of sacred funds and writings; the exhibition of votive offerings and dedicatory and honorific inscriptions; archiving legal transactions and decrees; court proceedings; meetings addressing community needs; and housing guests. (157)

Whereas the synagogues in Palestine are usually identified on the basis of characteristic architectural features (first and foremost a large hall with columns or pillars as roof supports, benches along the walls, often a miqveh next to or near the hall, and sometimes a niche or a room for the storage of sacred scriptures), Diaspora synagogues can be recognized only if they are provided with specific Jewish (or Samaritan) symbols, elements, and inscriptions, because their forms are thought to have been shaped more by local architectural customs and influences than by Palestinian concepts of synagogues. In this sense Diaspora synagogues form a heterogeneous group, (158) but they nevertheless served to house all activities that “declared the distinctiveness of the Jewish community and the adherence to its traditions,” as well as “to entrench a sense of collective identity.” (159)

As in synagogues in Palestine, the dominating feature of the building in Delos is the large hall, which might even have included supports for the roof and could have served most of the functions enumerated above; it was presumably used in a more flexible way, however, with moveable furniture, as there is no evidence for benches built along the walls. Therefore, those participating in common meals could have reclined on couches according to local customs, whereas for more religious/liturgical activities such as prayer or reading the Torah, and for larger civic meetings, benches could have been set up in the hall. The secondary rooms in the south might have been conceived, and indeed enlarged and partitioned, to meet additional needs such as the storing of sacred funds, scriptures, archives, and furniture, or the preparation of communal feasts, or the lodging of guests. (160) Votive offerings and dedicatory or honorific inscriptions were set up by preference within courtyards and in close proximity to shrines, altars, or cult images; by contrast, large assembly halls, without specific features such as niches, apses, or podia, offered no suitable space for such activity. Therefore it must remain uncertain whether, and if so how and where, the first users of GD 80 honored their god and benefactors with offerings.

The best parallel for GD 80 from the Diaspora is furnished by the synagogue of Ostia, which was situated outside the city and in proximity to the sea, was conceived as a monumental synagogue from the beginning, and was remodeled several times during its use from early in the second half of the first century A.D. until the fifth century. (161) The original building comprised a large assembly hall with benches, a podium, and an entrance construction that was square with a column at each corner; a separate, unusually large triclinium with built benches; and water facilities. It is worth noting, however, that the permanent furnishings for specific functions and the strict spatial differentiation between prayer hall and dining room might be attributed to the date of origin–about 200 years later than that of GD 80 in Delos. In the second half of the first century A.D. the functions and importance of synagogues might have changed or shifted slightly, for example, from being multifunctional community centers to promoting a sharper focus on religious aspects and activities, which would be reflected in the permanent equipment of the large hall in Ostia. In addition, the design of the synagogue in Ostia might have been inspired predominantly by local architectural styles, but despite some general similarities with meeting places, there is no exact parallel in any part of Ostia, and the synagogue does not fit into the known typology of such meeting places. Most interestingly, the original synagogue buildings in Delos and Ostia both lack the peristyle-courtyard that was obligatory for the larger meeting places in Ostia and the cited examples in Delos. (162)

Although the architecture of each of these synagogues is fairly comprehensible within the larger local context, it is at the same time notably different and without precise parallel. Given the remarkably monumental scale of both buildings, the absence of features such as a courtyard or shops was, most probably, not due to a shortage of money or to external factors such as the limited size of the building plots, but to an intentional adaptation to the needs of a specific group, the respective Jewish or Samaritan community. One might extend this comparison to the question of the extent to which Jewish or Samaritan communities resembled voluntary associations in the Graeco-Roman world, and to what degree differences and similarities in the organization and structure of these communities were reflected in the form and function of the respective synagogues and meeting places, but such a comprehensive sociohistorical study is beyond the scope of this article. (163)

According to the interpretation of GD 80 presented here, the small Aegean island of Delos provides a second, and even earlier (than that at Ostia), example of a Diaspora synagogue that was originally constructed as such, thereby offering the rare opportunity to study the design and development of an early synagogue. Its design corresponds to the specific nature of Jewish life in the Diaspora as recently described by Gruen: The building could well be interpreted as a negotiation between integration and segregation, between adaptation of local customs and consideration of Jewish needs. With their synagogue the Delian Jews or Samaritans might well have fitted within the world of Hellenistic and Imperial Delos while expressing and maintaining their own distinctiveness. (164)


It is generally agreed that for Second Temple synagogues no specific fittings with Jewish symbols or elements are to be expected, especially not in the Diaspora, where Jews adapted to the regional and local culture, customs, and language. (165) Some features of GD 80, however, have always attracted special attention from supporters as well as opponents of its identification as a synagogue, namely the throne, the niche in room A, the votive offerings, and small finds such as lamps.

The marble throne (Fig. 22) is usually identified as the Cathedra of Moses; its importance would have been emphasized by placing it in the center of the west wall of A, facing the single east entrance (and Palestine). (166) If, however, the doorway between A and courtyard C was blocked from the beginning of the fifth phase, this privileged position never would have been visible from the outside (the courtyard) and probably would not have been readily evident from the interior of B, as none of the entrances to A from B offered a direct axial view of the throne. Yet regardless of its position, the probably prominent provenance of this throne, the distant theater, suggests that it was intended to serve a special function. (167)

The small niche in the west wall of A, 2.20 m north of the throne’s current position (Fig. 22), is rarely mentioned in the literature. Binder conjectures that it might have served to store the sacred scrolls used in Sabbath services. (168) Within the Delian context, however, this niche, which was obviously created subsequent to the construction of the wall and is rather crudely made, is best explained as a place for a lamp. Since the lighting of this room could have been problematic in the evening and during the darker seasons, this function seems reasonable. (169)

Because of the concentration on the interpretation of the previously mentioned inscriptions, the function or functions of the votives and honorific bases that bear them has often been neglected. Two of the six bases that were either found in or attributed to GD 80 were discussed by Mazur as evidence against the synagogue theory. They show cuttings and the remains of lead in the top surface that might have served for the soldering of bronze figurines (“idols”), except that their form and size do not correspond to the usual bases for statuettes in Delos; both are much higher than the average bases for votives and statuettes and could, therefore, have supported larger objects, such as basins or tables. (170)

Three of the remaining four small votives resemble strongly the numerous “altar-incense burners” known from Delos, were certainly modeled after them and, most probably, were used accordingly. (171) This is quite remarkable because it is usually thought that the sacrifice of animals and the offering of vegetables and incense was confined mainly to the Temple in Jerusalem. In addition, such sacrifices were part of the sacred rituals in some early temple-synagogues in the Diaspora, where a gradual shift from animal sacrifice to the offering of vegetable and incense took place in the Hellenistic period. But the synagogue in Delos is normally counted among the examples with a more recent history of Jewish presence that focused from the beginning on the reading of the Torah and other synagogue rituals. (172) The evidence for possible incense offerings allows the synagogue building in Delos to modify this picture slightly, the more so because the votives were obviously kept–and presumably used–until the abandonment of the building. The burning of incense conforms to local practice because the required items have been found in large numbers in sanctuaries and houses all over Delos.

Unlike the votives, the two inscribed Samaritan stelai (173) that were found north of the synagogue, but that might originally have been set up in the building, had the function of only documenting and perpetuating the honors that a grateful group of Samaritans bestowed on their benefactors.

It remains to be considered where and how the different votives and honorific bases were presented within the context of the building. Some bases and stelai were not worked on the back and were, therefore, clearly conceived for a position in front of a wall or column. (174) They could have been set up along the walls without benches in rooms A and B, or in the porticoes. This holds true also for the freestanding examples that are worked on all sides, because a position in the middle of A or B does not seem very practical for circulation within, and the regular use of, these rooms.

According to Bruneau, the two Samaritan stelai were inserted into a wall and faced the interior of a second synagogue, but this hypothesis is not very convincing. In such a case the stelai, of remarkably different ages (250-175 B.C. and 150-50 B.C.), either must have been integrated into the wall when it was built, and/or the wall must have been partly demolished before it could receive them. If both stelai were integrated into the wall at the same time, then the earlier one must have been set up elsewhere originally, and thus could not be used to date the wall or the building where it was found. If they were not integrated at the same time, the later one must have been inserted subsequently, requiring a partial destruction and rebuilding of the wall. The parallel that Bruneau cites as evidence for incorporation into a wall was certainly an instance of subsequent and rather crude insertion. (175) Therefore the possibility cannot be ruled out–and seems, indeed, more likely–that the stelai in question were set up in front of a wall, with or without bases. (176)

A complete list of the small finds is not available and probably could never be compiled now, given that the excavation was conducted so long ago. According to the first excavator, “les menus objets recueillis parmi les ruines sont ceux-la memes qu’on pouvait s’attendre a rencontrer.” (177) For the selected objects that are listed in the publications, no exact findspots are given. It is therefore difficult to use the finds to reconstruct the precise use of the building and its rooms, or the process of abandonment and later reuse. Parts of the building seem to have been destroyed by fire, but the extent and impact of such a catastrophe cannot be determined. (178)

Numerous lamps were collected near the walls and mostly from under the marble benches; these positions could correspond with the actual use of the lamps, in which case the synagogue would have been abandoned during an instance of use, (179) or they could be the result of a subsequent shifting of the debris, for example when the lime kiln was built. It has often been observed that the Jewish community clearly obtained its lamps exclusively from Roman merchants, and that it was not particular about the decorative motifs on them. (180)

If the seventeen fragments of fine glass dating to the Imperial period belonged to the equipment of the synagogue, perhaps there was a set of luxurious tableware for feasts or other occasions; in any case, this equipment would testify to a certain affluence and standing of the Jewish or Samaritan community. (181)

In conclusion, there is nothing specifically Jewish or Samaritan about the fittings of GD 80 except the wording of the Greek inscriptions on the votives. Even the arrangement of benches along the walls finds several parallels within the Delian context, for example in the nearby gymnasium (GD 76), in private houses, in sanctuaries and public buildings, and in the Agora des Italiens (GD 52). (182) It is instead the remarkable combination of the throne with the architectural layout that distinguishes GD 80 from other constructions in Delos.


As several authors have already extensively discussed the history of the Jews and Samaritans in Delos, (183) only the most important facts with regard to the construction history of building GD 80 will be summarized here.

The earliest document attesting the presence of Samaritans in Delos is an inscription in honor of their benefactor Menippos from Herakleia, which is dated to the period 250-175 B.C. on the basis of paleography. (184) While the translation of the clausula [TEXT NOT REPRODUCIBLE IN ASCII.] is contested, it is agreed that this inscription testifies to the existence of a Samaritan synagogue well before 167/6 B.C. (185) Yet several problems arise: Were there two synagogues on Delos, one for the Orthodox Jews and one for the Samaritans? Did these two buildings exist contemporaneously? Did these two hostile groups really build their assembly houses in close proximity to each other (see Fig. 1)? Or was the Samaritan synagogue already out of use when the Jews constructed their synagogue, GD 80? Or was GD 80 not the Jewish synagogue but rather a Samaritan building? And if so, did the Orthodox Jews, whose presence on the island is attested in the literature, have a synagogue of their own at all, and, if so, where? (186)

At the present time, none of these questions can be answered with any degree of certainty. Two remarks will be made, however. First, it does not seem very likely that the two Samaritan stelai were ever integrated into a wall; rather, they were just stored coincidentally at their findspot, that is, they were probably displaced. (187) In this case, there would be no evidence that the wall adjacent to the stelai belonged to a synagogue. A trial trench might finally resolve this issue. Second, if GD 80 is identified with a Samaritan building, the Samaritan inscription honoring Menippos might have implications for the date of construction. The original building is usually dated to the second century B.C., regardless of its supposed function, but it is not specified anywhere whether this means before or after 167/6 B.C. (188) Most buildings in Delos, and the development of the Quartier du stade, are usually attributed to the period of the free port, so the existence of a synagogue in this quarter before or around 175 B.C. would be surprising, but not impossible. Again, only further excavation and investigation might clarify the absolute chronology of GD 80.

In light of the uncertainty of ownership–either Jewish or Samaritan–it is not possible to link safely the diverse historical data with building GD 80. Accordingly, the following discussion of the available historical and archaeological data is to be understood as a distinctly hypothetical exercise. (189)

In 139 B.C. the Roman consul Lucius sent a letter of friendship on behalf of the Jewish people to different addresses, among them Delos. (190) The Jewish community was obviously important enough to be included on the list and might already have built its first synagogue. It is likely that Jews and Samaritans frequented the island or even lived on it for the same reasons that many other foreigners from all over the Mediterranean world did: for profitable commerce in the booming cosmopolitan trade center. Inscriptions show that Jews participated in Delian life at the end of the second century B.C. and the beginning of the first, but nothing is known of the years around 88 B.C., when the synagogue was perhaps partly destroyed. (191)

Two decrees of the 40s B.C. testify to privileges and rights granted to the Jews in Delos and eventually Paros. The first, issued in 49 B.C., provided an exemption from military service for Jews who were Roman citizens. (192) The second, promulgated at about the same time, was written in response to complaints from the Delian Jews about restrictions on their ritual practices. It specifies that Jews should be allowed “to live in accordance with their customs and to contribute money to common meals and sacred rites.” (193) This decree conveys several points: (1) it attests the presence of a Jewish community on the island–which must have been quite important and active–defending the interests and rights of Jews in a period when Delos was already in full decline; (2) Jewish religious activity had obviously been limited for some time (but why and for how long cannot be determined); and (3) the Jews in question held common meals, probably regular banquets and/or special feasts.

According to Binder, the third point “suggests the existence on Delos of a synagogue with an ancillary banquet hall used to hold feasts on sacred days.” (194) But it is far from certain that this decree was referring to the Jews in Delos. It is addressed to the “magistrates, council, and people of Parium,” and mentions [TEXT NOT REPRODUCIBLE IN ASCII.].

Parium is usually identified with the island of Paros and not with the city of Parium in the Troad, but this does not explain if and why the regional magistrates in Parium were responsible for the Delian Jews. (195) Regardless of the location of the privileged Jews, it seems fairly safe to imagine that sacred rites and customs of Jews on neighboring Cycladic islands did not differ considerably, and that common dining played an important role in all of them.

Because the bisection of the large hall (fifth phase) and the permanent furnishing of the resulting new rooms, A and B, were most likely not realized as early as 50-40 B.C. but instead much later, there is no need to search for a separate dining room in GD 80. The undivided hall could well have served for common meals, with its participants seated on benches or resting on couches. It is readily conceivable that the Jewish community, as attested in the 40s B.C., still needed an assembly hall that they maintained, repaired, and adorned with care. The partial, stabilizing rebuilding sometime after 88 B.C. (third phase) and the embellishment and enlargement (fourth phase) of GD 80 would fit well in this picture.

Despite the information provided by literary evidence, a number of questions remain unanswered. Why did the Jews stay on Delos when the profitable international trade had shifted to successful rival ports such as Puteoli and Ostia? How did they earn their living, and where did they obtain the means to repair and sustain their synagogue and finance the different aspects of communal life, for example the common meals? Did they enlarge their synagogue (fourth phase) in connection with the confirmation of their rights, or later? In other words, was their social and financial position stable over decades or even centuries, thus enabling them to initiate major building activities as late as the first or second century A.D.? Or did the Delian Jewish community consist mostly of people of modest means who eked out a living on a nearly deserted island? (196) If a dedication by the Athenian people and the inhabitants of Delos in honor of Herodes, tetrarch of Galilee, in A.D. 6 was really motivated by the presence of Jews, the latter must have had at least a modest number of influential, wealthy patrons. (197)

Further epigraphic and literary evidence is limited to what is found on a few modest votive offerings, the lettering of which suggests a date late in the first or even in the second century A.D.; these could have been offered by either Jews or Samaritans. (198) The five inscriptions found in GD 80 itself imply an active use and votive praxis for two or even three centuries; if the other inscriptions, especially the two rather early Samaritan ones, really referred and belonged to this building, such a period could have been even longer. (199) The extent of the votive and honorific activities performed by Delian Jews and the location and arrangement of the respective objects cannot be determined due to the scarce evidence and, presumably, to the installation of the lime kiln in their synagogue.

No sources give any hints with regard to the possible date and historical background of what is here termed the fifth phase, which had considerable consequences for the use of the rooms and for the entire building. The changes might reflect a shift in the activities performed in the building, for example the abandonment of dining (or at least a change in the dining customs) in favor of liturgical tasks such as teaching, reading, and studying the sacred texts. The setting up of benches in rooms A and B and in the west portico suggests that the community was still quite large, (200) but it is uncertain whether the use of spoil material for the furniture indicates the community’s impoverishment.

Since it is much contested whether the destruction of the Temple in Jerusalem in A.D. 70 had a major impact on the evolution and function of synagogues, especially in the Diaspora, (201) it is neither safe nor necessary to date any of the phases of the synagogue in Delos in accordance with this catastrophe. With regard to function, it is interesting to note that the synagogue of Ostia was not provided with a separate triclinium with built benches during its second phase (the first half of the second century A.D., probably in the Hadrianic period). The demolition of the first large triclinium, and the reinstallation of a triclinium in the third phase (not before A.D. 306), in a different setting and with a different design, were certainly both intentional measures taken to meet the changing needs of the Jewish community. (202) A provocative explanation for this archaeological evidence is offered by Klinghardt, who argues for a significant shift of activities performed in synagogues after A.D. 70 that included, in part, a distinct separation of prayer and communal dining. (203)


The main objective of this study has been to elucidate the construction history of building GD 80 and to ascertain whether in its initial phase it was used as a synagogue or served some other function. The results can be summarized as follows:

1. Building GD 80 has a complex construction history that comprises five phases of active use and later reuse.

2. Its most characteristic features are a large hall, A/B, a large water reservoir, orientation toward the east and the sea, and an isolated location on the eastern shore of the island, far from the main harbor and the city center. All these features are constitutive, that is, they were integral to the design of the original building. Furthermore, there was a remarkable continuity in design and–most likely–in the use of space from the first through the fourth phases.

3. GD 80 was conceived as a freestanding building and was only subsequently surrounded by neighbors to the north and west. Thus, it was probably the southeast starting point for the development of an insula.

4. The first two phases can be dated to the perio