The excavation of the palaestra of the gymnasium, an imposing monument of the ancient city, which extends over two, successive terraces of an overall surface area of 4,600 m², took place in 1932, from 1934 to 1939, and from 1952 to 1953, and its results were reported briefly in the corresponding Praktika volumes of the Archaeological Society. From the early 1950s until today the monument, exposed to the elements, has suffered from the erosion of the hillside above it, and the collapse of the stucco layers that covered its retaining walls, the stone channels running in front of the stylobates of the upper and the lower terraces and the two fountains. In addition, the restored (by A. Orlandos in 1937) section of the south fountain is heavily damaged, mainly because of the oxidization of the iron clamps that were used at the time. Clearly the monument was in need of conservation, consolidation and protection. For this reason, in the summer of 2020 we proceeded with cleaning and drawing it in detail, as well as with recording all the surviving pieces of its columnar orders that lie on the site. In places where the fill covering parts of the stylobate and the ancient floor level was considerable, particularly along the west side of the upper terrace, the cleaning was done based on excavation contexts of various shapes and dimensions. In addition, we opened two small trenches on the north side of the lower terrace, in order to address stratigraphic and chronological uncertainties.
We started cleaning the palaestra from the west side of the upper terrace, which has suffered the most from the erosion of the slope of the acropolis, resulting in the covering of the toichobate with debris to a depth that in places reached a meter. The superimposed Contexts 5500 and 5501, together almost half a meter deep, were oblong layers of earth, ca. 25 m long and 1.5 to 1.8 m wide, that covered the area between the west wall and the vertically cut bedrock of the hillside. The lower context (5501), more compact and with clearly more pottery and tiles than the topsoil, was dug down to the level of the upper surface of the highest preserved course of the west wall of the palaestra. The composition and the content of this context suggest that it was part of the artificial fill of a corridor that must have existed there during the construction of the west wall. More undisturbed layers of this artificial fill can be sought at lower levels (down to the toichobate) but such excavation was beyond the goals of this year and the permit we had from the Ministry of Culture.
After removing these contexts we exposed a) a low wall built on top of the west wall of the palaestra, b) a cut in the bedrock of cylindrical section, and c) a retaining wall aligned with the rock-cut section to the west of the upper terrace. The low wall (Context 5515), of north-south orientation, was exposed to a total length of 14 m but appears to continue further to the south. In its masonry, we observed remains of the fourth course of the west wall of the palaestra, along with rubble and mortar. It is either a repair of the west wall or, a later addition, perhaps after the abandonment of the palaestra, which is more likely judging by its poor quality. It is possible that it was part of the east side of an open conduit which followed the slope of the hill in a northern direction. During its excavation, we observed a section of a cylindrical pipe that crosses the wall and seems to have been placed there in order to direct some of the water eastwards.
The semi-circular cutting of the bedrock, ca. 70 cm in diameter and 1.65 m in visible depth, is rather a remnant from the extraction of a column, the final diameter of which must have been ca. 50 to 55 cm, if we take into account the need for an extraction channel ca. 10 to 15 cm wide and for a protective mantle (apergon), a few centimeters thick, around the final surface of the column. Alternatively, this cutting could belong to the shaft of a well, which would have gone out of use when the upper terrace of the palaestra was built. The third feature observed, namely the retaining wall, is 6 m of visible length and 2.05 m of maximum visible height, and preserves four courses of ashlars. It was built in order to hold the earth of the hillside which is quite steep at this location and therefore subject to landslides.
Among our finds from this area, we single out a fragment of Doric geison with guttae, of fine quality and craftsmanship, similar to the one found by Orlandos in 1934. These two fragments probably belong to the first building phase of the west terrace of the palaestra, which appears to be contemporaneous with the construction of the lower, eastern terrace of the complex, and therefore dating to the 3rd century BCE.
The next step was the excavation of a mound of earth, almost rectangular measuring ca. 16 (north-south) by 4 m (east-west axis) and rising some 2 m from the floor level of the palaestra. It appears that this mound had been left undug by A. Orlandos, even though he writes that during the last excavation season (in 1953) the thick fill of the upper terrace was removed and the entire stylobate of the west side along with the water channel were exposed. On the figure though (no. 6) that accompanies his report we can discern in the background the presence of the mound that we removed this year. We began the excavation by its northernmost part, by opening a trench 2 x 4 m (Context 5502). During the excavation, down to a depth of 1.5 m, we did not distinguish any change in soil color or composition. The finds were sparse, a few tile fragments and sherds of different periods, that indicate that his layer was formed by the accumulation of soil that had been washed down the slope. The underlying layer (Context 5503) was clearly darker and with more inclusions, but still with very few ceramics or other finds. This allowed us to remove the rest of the mound with the use of a JCB, always under close supervision and inspection of the earth as it was dug out. The last layer, 20 to 30 cm above the level of the stylobate, was dug manually. The final surface (Context 5510), relatively compact and with few inclusions, lies approximately at the level of the stylobate.
The absence of a layer of roof tiles within a space that was meant to be roofed, since it was part of the west stoa of the upper terrace, suggests that at a later phase the building had lost its roof, and that its roof tiles were subsequently removed. It is perhaps to this later phase that we should assign the low wall that was built on top of the west wall of the palaestra.
The SW corner of the upper terrace also needed clearing of the fill, accumulated during the last seven decades, down to the level of the toichobate that projects by 5 to 7 cm from the inner line of the wall. Thanks to the removal of the soil that covered the entire west side of the upper terrace, we exposed the toichobate throughout its length and brought to the fore the maximum preserved height of the west wall, ca. 2 m, corresponding to four courses.
The stone channel with its ellipsoidal basins that runs in front of the stoa had also been covered with soil, especially along the west and south sides, while in three spots column drums were placed in the channel obstructing its cleaning. We lifted these drums, one of which had caused severe dislocation of the channel, and we cleaned the channel and the corresponding sections of the stylobate (Contexts 5505 for the south side, 5509 for the west side, and 5511 for the north side). The channel preserves its exterior fine mortar coating only in some places, but elsewhere only the underlying coarser layer is preserved with its characteristic punch holes made for the application of the final layer. From the soil covering it we retrieved fragments of two antefixes (Context 5505) and a piece from the drapery of a marble statue (Context 5509). On the inner side of the stylobate, next to the southwest corner of the peristyle, we distinguished a relatively clean, whitish and compact surface (Context 5506), lying ca. 15 cm lower than the stylobate, and which is likely a remnant of the floor of the stoa. In the same area, more specifically at a spot missing the slab of the stylobate, we identified a possible robbing trench (Context 5507) filled with a reddish soil containing tile fragments (Context 5508), which we did not remove.
The stylobate is preserved only partially along the west side of the peristyle. In order to uncover its foundation course and extract information on its construction and dating, we opened a trench 1.30 m wide along the west edge of the channel (Context 5517). In many places we identified traces of the robbing trench of the slabs of the stylobate as well as remains of the whitish layer possibly originating from the floor of the stoa. The exposure of the stylobate and its foundation course allowed us to make various observations regarding the architecture of the upper terrace and its building phases. The south side seems to represent the original phase of the building with the krepis of the stylobate regularly constructed, projecting ca. 20 cm from the line of the stylobate as one would expect. As a rule, the stones where the columns rest are square whereas those lying between the columns are rectangular. On their surface, pairs of dowel holes are visible, spaced 36 to 38 cm apart, for the fastening of the columns. Based on these pairs we calculate the interaxial spacing to 2.05 m and the intercolumnar distance to 1.24 m, and thus we can reconstruct eight columns on this side with a base diameter of 0.61 m.
The west side represents a second phase of the complex, when the original stylobate was repositioned ca. 40 cm eastwards from its original location, thus projecting ca. 20 cm from the front of the top course of the krepis. The empty space thus created was filled with rubble stones and mortar. The reason this happened is not clear yet. On the surviving stones of the stylobate we observed setting lines that do not correspond to the expected pairs of dowel holes, as well as a few pry marks and dowel holes that indicate that the stones are reused. The traces observed on the southwest and northwest slabs of the stylobate point to the same conclusion. The southwest slab, which follows the repositioned phase of the west stylobate, bears four dowel holes in a row and three setting lines. The first pair of dowel holes defines the position of the original location of the column, and the second pair its position during the later phase. The northwest slab of the stylobate bears a pair of dowel holes that are unrelated to the expected position of the column, and a square dowel hole at the center which seems to be in the right position with the line of the colonnade. In fact, it appears that in this second building phase it was just the corner columns that were fastened to the stylobate, whereas the other columns were erected without dowels.
On the north side, which preserves only the westernmost section of the stylobate over a length of less than 8 m, the distance between two square dowel holes is 2.45 m, which may well correspond to the interaxial spacing of the second phase of the upper terrace. This spacing divides satisfactorily the western stylobate into 24 interaxial spacings with 23 columns, and the northern stylobate into eight interaxial spacings with seven columns, not taking into account any corner contraction. It is also important to note that the level of the stylobate on the north side is ca. 60 cm lower than that of the south side, and the same difference in altitude was measured at the corresponding water channels.
The architectural members, columns and parts of the entablature that lie on site, belong to the second building phase. The columns have a lower diameter of 0.61 to 0.62 m, with their lower sections left unfluted and their capitals rather roughly dressed. There are also large parts of geisa, of similarly rough dressing, and triglyph and metope blocks which, at an even later phase, when manifestly the stoa had collapsed, were placed on top of the south wall of the upper terrace in order to protect the staircase. On the contrary, the two Doric capitals and the geisa fragments that were found and attributed to the first phase of the palaestra are products of clearly finer craftsmanship as recognized by A. Orlandos.
At the lower (eastern) terrace, we focused on the west and south sides, that had undergone the heaviest filling episodes. Along the retaining wall on the west side of the terrace, earth washed from the upper terrace had covered the toichobate and a portion of the bottom course. The removal of the soil throughout the length of the retaining wall exposed the entire masonry preserved to a maximum height of four courses. We paid particular attention to the cleaning of the fountains. The thickness of the fill inside the south, partially restored fountain, was relatively small and its careful cleaning highlighted the pebble mosaic floor of the basin. During the cleaning of the porch of the fountain, in the interstices of its stone slabs, we found a bronze coin from Corinth dating to 50/51 CE. In the north fountain, the fills were substantial, more than half a meter thick, having covered the loose architectural members that had been placed inside the basin. During the removal of the fill, we lifted these members and placed them on wooden pegs in front of the fountain. They are mostly epistyle blocks from the front of the basin and the porch that had allowed A. Orlandos to graphically reconstruct the façade of the fountain.
The fill of the rooms along the south side of the lower terrace was also heavy because of the erosion of the hillside. Following its clearance, the line of the toichobate was exposed as well as the foundation of the wall running in front of the rooms. This wall along with the southern rooms were eradicated when the staircase was built, directly on the axis of the front wall. Given that the first step of the staircase is 62 cm above the level of the toichobate, we can easily deduce that when the staircase was built, the ground level at least on the southern side of the palaestra had been raised by more than 40 cm. In addition, the incorporation of parts of the frieze of the colonnade of the upper terrace in the wall to the south of the staircase, reinforces the hypothesis that when the staircase was built the palaestra had already ceased to function as such, and that the staircase served a different purpose still to be determined.
The north side of the lower terrace shows at least two phases as suggested by the parallel rows of ashlar blocks, one in front of the rooms and the second 80 cm further to the south. The southern row of blocks extends westwards to the retaining wall where an anta was carved in relief. The main questions concern the relationship of this southern row of stones to the entrance to the complex, preserved at the northwest side of the lower terrace, as well as to the rooms along the north side. In order to investigate the relationship to the rooms and their cross-walls, namely whether this southern row of stones belongs to an earlier line of the rooms’ front later moved 0.80 m towards the north, we opened two small trenches, between the two rows of foundations and on the axis of two cross-walls of the rooms. The excavation of the west trench, measuring 2 x 1.80 m, progressed to bedrock, which appeared at the depth of ca. 26 cm. Under the topsoil, we encountered a layer (Context 5520) that we interpreted as construction fill, ca. 20 cm thick. During its removal, the euthynteria of the toichobate came to light, projected by 19 cm, as well as a Sikyonian trichalkon of the first half of the 2nd century BCE.
The excavation of the east trench to a maximum depth of 59 cm exposed again the bedrock and the euthynteria of the toichobate, which here projects by 14 cm. In none of the trenches did we have clear indications that the corresponding cross-walls of the rooms once extended further to the south than the actual front of the rooms, and therefore the hypothesis that this southern row of stones may have corresponded to an earlier front of the rooms cannot hold. On the other hand, at least some of the stones along the front of the rooms are in secondary use and seem to come from a stylobate since they bear characteristic dowel holes for the fastening of the columns. This subject as well as the changes observed on this side of the palaestra need further investigation.
The cleaning of the stone channel and of the stylobate around the three sides of the stoa, wherever the latter is preserved, allowed us to observe the surface of the stones, to measure the interaxial distances, and to calculate the total number of columns with a base diameter of 0.65 m. With a standard interaxial of 2.12 m, we estimate that the north and south sides of the peristyle, each bore 10 Ionic columns besides the corner ones, and that the east, long side had 25 columns. The attachment of the bases to the stylobate was secured by pairs of opposite dowel cuttings as well as by a central dowel hole. As we were cleaning the krepis of the stylobate on the north side of the palaestra, we found another Sikyonian trichalkon of the first half of the 2nd century BCE.
At the conclusion of these operations we proceeded with a photogrammetric rendering of the monument using a drone. The orthophoto of high definition that was produced was based on 824 shots, while the 3-D model was based on 1,317 high analysis photos. This analysis enabled us to produce a photogrammetric plan at a scale of 1:10 and to observe the architecture of the monument in great detail.
The architectural study also included recording of all loose architectural members that belonged to the Doric and Ionic elements of the building. Overall, we recorded 75 members, mostly column drums and bases, but also geison fragments, parts from the epistyle and the frieze of the two terraces, and a few anta parts. This corpus is supplemented by related members that are displayed in the museum, and those were grouped to the west of the large temple of the agora. The catalogue of these members, that come mostly from the old excavations at the site, namely of the gymnasium, the large temple, the south stoa, and the bouleuterion, numbers 124 entries in total. Our goal, now, is to complete the identification of these members, the drawing and photogrammetric rendering of the most important ones, and their attribution to specific monuments.