Northeast of the temple, the excavation of the fourth trench measuring ca. 24 x 16 m (where in 2014, we had exposed the corner of a monumental structure with a semi-circular exedra and later walls to the north) continued with the exploration of deeper horizons inside and outside the building. Inside the building, the excavation started by removing the terracotta pipes (Context 3012) which were found last year near the north wall (Context 3006), following their photogrammetric and architectural drawing, so that we could dig deeper. The pottery found in the layer of the pipes dates to the 6th century CE. Between the north Wall 3006 and the south Wall 3002 two aligned stone bases are preserved in situ (Contexts 3072 and 3073), 4.7 m apart. The east base had been exposed in 2014 but the west one was exposed this year with the extension of the western side of the trench by 1 m. The location and size of these stones refer to column bases, in this case of an inner colonnade of the building. In order then to investigate the lower part of the north wall as well as the thickness and the foundation of the square stone bases we dug a 5 x 3 m section between Wall 3006 and the south side of the stone bases. The excavation of Context 3026, which is a hard packed clayey soil for the most part empty of finds, brought to light a terracotta pipe (Context 3074) of northwest-southeast orientation. In order to dig deeper in this area without removing the pipe, we opened a second, adjacent trench to the east, 3.5 x 3 m. The excavation of this trench revealed the foundations of Wall 3006 and of the eastern stone base (Context 3073). The base, consisting of two adjoining stones, measures 1.06 x 1.25 m, 0.64 m in depth, and rests on a hard red soil. On the same soil rests Wall 3006 as well, albeit at a higher level. With the exposure of the inner side of this wall down to the foundation level a linear recess came to light, some 30 cm higher than the bottom of the wall, which may correspond to the floor level. However, we did not identify traces of a floor at this level during excavation of Context 3026.
On the east side of the building, i.e., to the west of its eastern wall (Context 3003) we continued the excavation of Context 3033, that we had begun in 2014, and its underlying deposit 3046. To the east of Wall 3003 we continued the excavation of Context 3019 as well as of two overlaid deposits (Context 3064 and 3077) in order to investigate the outer side of the complex towards the east where we had noted through geophysical prospection the presence of architectural structures. Indeed, the excavation of these deposits was crucial for our understanding of the architecture and building phases of the complex.
Along the west (inner) side of Wall 3003, but at a lower level, the inner side of the initial east wall of the building came to light, featuring the same construction technique to the later wall (rubble and mortar). In addition, along the east (outer) side, Wall 3003 rests on an earlier wall of similar orientation (Context 3066) made of ashlar blocks projecting 15 cm from the line of the later wall. The north end of this wall is preserved beyond the outer corner of the building, which proves that it was not part of this building. Most likely it was originally related to another large structure which extends to the east and which we had detected through geophysical prospection. To the same structure must belong a small section of a wall (Context 3065), of east-west orientation, 0.69 m wide, perpendicular to Wall 3066. The initial east wall of our building was built adjacent to Wall 3066. If we assume that Wall 3066 had the same width than Wall 3065, then we must accept that the width of the rubble and mortar masonry of the initial Wall 3003 was about 1 m.
At a later date, the northern end of this wall was cut by the construction of a basin 2.50 x 2m (Context 3064) with walls made of layers of tiles and mortar, and with their inner sides lined with thick mortar. The east side of the basin is preserved to a maximum height of ca. 35 cm which surely does not correspond to its initial height given that the later Wall 3003 stepped on it. The basin was most likely fed by the terracotta pipes that we had excavated immediately to its west (Context 3012). During a third phase, at least the west side of the basin was destroyed from the construction of yet another basin (Context 3058), 1.40 m wide and 0.40 m deep. The floor of this basin consists partly of the leveled west wall of Basin 3064 and partly of stone slabs which appear to extend underneath the west side of the basin, and therefore must belong to an earlier building. The remains of the north side of this small cistern were traced to the north of Wall 3006 which cut it and up to the south side of Wall 3045. Therefore the length of Cistern 3058 must have been ca. 3.3 m. The erection of the north wall (Context 3006) which eradicated the cistern as well as of the later east wall (Context 3003) with which Wall 3006 bonds, represents the fourth (in a row) building phase that we were able to identify.
The surviving semi-circular apse (Context 3007) belongs to an even later building phase, given that its extremities are built against Wall 3006 as we had already noticed in 2014. However the continuation of the excavation to the north of the north wall of the complex revealed traces of an earlier apse, which are partly preserved at a lower level and in physical connection to the western side of the Apse 3007. This apse (Context 3050) features similar masonry to its successor, namely rubble and mortar, but had a smaller radius. The visible width of its wall is 0.66 m and its maximum surviving height reaches 0.73 m. In this case, too, the wall of the apse does not bond with the northern wall of the building, which suggests that it was built after this wall but when exactly is not yet clear. Two terracotta pipes go under the surviving part of Apse 3050, one of which (Context 3054) most likely connects to the one found last year at the northwest corner of the apse (Context 3028) as well as to the south of Wall 3006 (Context 3012). Apse 3050 did not cut this pipe as it did with the second pipe (Context 3052) but its builders seem to have taken this pipe into account. If this pipe is connected with the supply of Cistern 3064, then it means that Apse 3050 came later than the cistern in question.
To the north of the apse and due west of Wall 3017 excavation progressed to a maximum depth of 0.58 m, removing layers of mostly natural fills which resulted from the erosion of the slope of the acropolis (Contexts 3029, 3039, 3040/3069) and had accumulated here because of the existence of Wall 3017. These deposits contained mixed material dating up until the Late Roman period, contrary to the underlying Context 3057, which is made of packed red soil and extends below the level of the architectural structures, containing very few finds. At this level was built the terracotta cylindrical pipe (Context 3062) of eastern orientation that we were able to follow beyond Wall 3017 over a total length of 11.15 m.
The continuation of the excavation to the north of the apse, to a depth of 0.40 m, revealed the base of Wall 3022 that we partly exposed last year, built of ashlar blocks, spolia, and smaller stones, and preserved to a maximum height of half a meter. The wall rests on a thin layer of pebbles, gravel, and tile fragments, which is partly preserved to the north and the south of the wall (Context 3055). The overlying layers (Contexts 3042, 3043, 3044, 3055, 3056, 3071) contained, with the exception of Context 3042, only few artifacts. In Context 3042 we found mixed pottery, Hellenistic as well as Roman, with the latest dating to the Early Roman period. The pottery from the underlying layers has not been examined yet. To the north of Wall 3022, we found a small section of the terracotta Pipe 3052, which is better preserved to the east of the apse. The function of Wall 3022, which was eliminated by the construction of the apse, has not been understood yet, however it is significant that it aligns with Wall 3045 which we discovered to the east of it.
The excavation to the east of the apse and Wall 3017, and to the north of Wall 3006, extended northwards as far as Wall 3016. Within the first layers that we removed from this whole area (Contexts 3035, 3041), ca. 30 cm thick, we found pottery dating from the Hellenistic to the Late Roman era (the latest sherds belong to the late 6th/early 7th century CE), as well as 11 bronze coins that have not been cleaned and identified yet. Having taken out these layers, we discovered Wall 3045 in the line of Wall 3022 but of different masonry style, and the continuation of the terracotta Pipe 3052, as well a stone channel (Context 3053).
By continuing the excavation to the north of Pipe 3052 (Context 3051) we encountered the eastern continuation of the second terracotta pipe (Context 3062) that we had located to the west of Wall 3017, and which connects to the stone Channel 3053 through a rectangular cut on the west side of the channel. From the layers that we removed to the south of Pipe 3052, and particularly from Context 3060, we collected interesting artifacts, among which a complete unguentarium, parts of moldmade bowls and various small finds. With the excavation to the north and the south of Wall 3045 (Contexts 3061, 3077-3079), the traces of the north side of the Cistern 3058 described above came to light, as well as a second, lower course of Wall 3045 which appears to continue to the west towards the Apse 3007.
The upper course of Wall 3045, of an east-west orientation, extends eastwards beyond the limits of the excavated area, but towards the west it stops 2.6 m east of the Apse 3007. It is made of ashlar blocks, 1 m long, 0.80 m wide and 0.25 m high, of which the westernmost, which is also the highest, bears traces of mortar on its south side. This is explained by the fact that part of Wall 3045 served as the north side of the Cistern 3058. The stone floor of the basin that we had observed to the south of Wall 3006 was also encountered between Walls 3006 and 3045.
A cut on the upper surface of Wall 3045 is in alignment with the stone channel (Context 3053) oriented southwest-northeast; we exposed it to a total length of 4.12 m. This channel has a total width of almost 1 m and is 0.32 m high. Its walls consist of various stones, some of which are spolia, so that its inner width does not exceed 15 cm. The position of the cut on Wall 3045, which is higher than the floor of the Basin 3058, allows us to assume that we are dealing with an overflow device. The two terracotta pipes, Contexts 3052 and 3062, which were located further to the west, appear to end to the same stone channel. Pipe 3052 consists of sections 0.47 m long and of an outer diameter of 9.5 cm, whereas Pipe 3062 has sections 0.61 m long and 10 cm in diameter.
Among the structures that came to light to the north of the building complex, Wall 3045 must be the earliest. Its function is not clear yet, but it may be part of the southern retaining wall of a central street of the city that led to the theater and which we had identified by geophysical prospection. Within the presumed boundaries of this street, we did not find traces of a roadbed but a network of pipes of different orientations. The dating of these structures is still unclear, however, at least the stone channel could hardly coexist with a road surface.
In sum, excavation in the fourth trench yielded abundant information on the architecture of this area and its building phases. We revealed the foundations of a large, stoa-like structure of the Early Roman or even Hellenistic date (Contexts 3065 and 3066), while it is possible that we have the southern retaining wall of one of the major arteries of the city (Context 3045). In both cases more excavation is needed, particularly in the area to the east of the Late Roman complex. Regarding the Late Roman remains, we identified six successive building phases, while excavation to the south of the temple confirmed that the large Π-shaped complex encompassed an area of some 2,500 m². The finds so far are not particularly revealing with regard to the function of this complex, however its layout and size bring to mind large urban villas of Late Roman times.