Katherine Harrington, Jana Mokrišová
Team Blue, Session 1
Nezi Field Excavations
N: 1017.20 N, S: 1006.90 N, E: 263.50 E, W: 255.50 E
1-25 April, 2013

This is the final summary of the first session of excavation in 2013 in the northwest area of Nezi field. Guy Sanders (director) and Rossana Valente (field director) supervised. The blue excavation team consisted of Katherine Harrington and Jana Mokrišová (recorders), Athanasios Notis (foreman and pickman), Athanasios Sakellariou (pickman), Christos Sakellariou (barrowman), and Panayiotis Rontzokos (shovelman).

The excavation area was initially bounded by the Giambouranis House (also referred to as the Turkish house; NB 252, NB 262, 1015.90N) to the north, Wall 540 (264.20 E) to the east, Wall 366 (1006.90 N) to the south and the baulk of Nezi field (255.50 E) to the west. However, the northern boundary was extended in course of the excavation by the discovery of N-S drain to the northernmost point of the drain (1017.20 N), and after the removal of the wall 540 the eastern boundary was moved to 263.50 E.

The goal of this session was to remove final traces of Byzantine occupation in the excavation area and to further explore the Late Antique levels with the hopes of better understanding the Late Roman/ Late Antique occupational activities in the area. In addition, we also re-examined the sequence of Middle and Late Byzantine walls in the area.

Late Antique (5th-7th century CE)

We reached the Late Antique levels in most of the excavation area, except possibly any remaining fill in Cut 1040 (N-S 1016.46-1014.59, E-W 259.47-258.59), a robbing trench running N-S next to the eastern wall of the Turkish house, which may continue deeper. Additionally, we believe that there might be an extension of this robbing trench on the south side of Cut 870 in an area that was not investigated during this session (under Context 927).

An ovoid hearth, with a schist and tile bottom, was revealed in the southeastern part of the excavation area (N-S 1012.49-1011.56, E-W 262.46-261.57), but was subsequently excavated by Mohammed Bhatti and Daniel Diffendale. The fill within the hearth contained a single illegible coin (no. 2013-144), a possible minimus dated to the 5th-6th century CE, and Late Roman coarse wares.

The most significant Late Roman activity in the excavation area is the series of tile destruction fills dumped in the western side of the excavation area (immediately west of Cut 870). The earliest and the largest fill, Deposit 1080 (N-S 1015.83-1011.22, E-W 258.48-255.73), yielded a large amount of broken tile, charcoal, as well as over 200 kg of pottery and a significant amount of glass, metal, and animal bone. This is clearly a dump consisting of a large amount of cooking wares, storage vessels (e.g. amphorae), as well as glass tableware, which seems to be an assemblage of a wealthy household or households. Special objects include three bone plaques with incised concentric circles, possibly belonging to a single wooden box (MF-2013-6a-c), an iron buckle (MF-2013-4), a bronze tack (MF-2013-5), and several rare forms of African Red Slip Ware (C-2013-3, C-2013-4). Most of the 56 coins found in this layer were illegible. Some coins, however, provide a useful terminus post quem for the dating of this layer. One such coin, 2013-128, might date to after the 491 CE reform of Anastasius due to the possible presence of denomination mark. Similarly, Coin 2013-139 may also bear a denomination mark. However, the identification of these coins is tentative due to their poor preservation and will need to be verified by Orestes Zevros. Several other 5th century CE coins (nos. 2013-91, 2013-96, 2013-113) and an interesting possibly counterfeit coin of “barbaric style” dating to 383-402 CE (no. 2013-123) were also found in the context. Ceramic evidence was more useful in dating this context. Most of the pottery from dates to late 5th century CE. However, the context has very few early 5th century sherds, and the lamps and African Red Slip forms present in the context suggest a slightly later date of 500-525 CE.

Above the tile fill in the southern part of this area, two parallel curvilinear walls (S1068 and S1069), 0.45 m apart, rested on a deposit containing much dissolved mudbrick (Context 1073; N-S 1013.60-1011.79, E-W 257.53-255.71). The inner Wall 1068 was made of two courses of limestone cobbles (with the second course represented by only one stone) and a reused marble fragment (A-2013-1), while the outer Wall 1069 had only one preserved course. One section of both walls was covered with a localized concentration of mudbrick, clay, and stone (Context 1047). We originally identified this concentration as a structure, but later we concluded that it was probably a dump of building material or a dense filling between the two walls. The walls run into the western scarp, and unfortunately not enough was preserved to determine their function.

Above the northern side of the large destruction Context1080, there was another dense fill, Context 1058 (N-S 1015.65-1013.46, E-W 258.47-255.65), which was relatively tile-free, but contained a large amount of coarse and cooking pottery, many of which had joins. While this is not as clearly a destruction level since it did not contain much tile or other building material, there is no clear use-surface above or below it, and the presence of large pieces of charcoal and the joins in the pottery suggests that this may also have been a dumped destruction fill. Some rare pottery types and lamps from this context were saved, including a lamp from the workshop of Chionis (L-2013-1) dated to the late 5th/early 6th century CE, which seems to correspond well with the date based on the remaining pottery.

The fill of Context 1046 (N-S 1015.80-1013.35, E-W 255.60-257.91), yet another layer rich in tile, ceramic, and bone, was smaller in extent and thinner than Context1080, but it included a similar range of materials. The pottery forms seem to be later and are dated to the second half 6th century CE. Similarly, the glass from this deposit contained a number of tubular ring goblet feet, which were introduced in the second half of the 5th century CE and were made with new technology. These were absent from Deposit 1080, which contained only vertical stemmed goblet feet. Thus, here we have two different types of technology represented in two separate deposits. According to Athenian Agora XXXIV, however, the two types may slightly overlap chronologically. Therefore, these two deposits may either represent two chronologically separate events, or may reflect the presence of two separate, but contemporary, households who both dumped material in the same area. Significant other finds include a large ovoid marble fragment, which may be a finial or part of a balustrade.

The interpretation of the nature of the activities in this area is made difficult by the presence of large pit Cut 870 in the middle of the excavation area as well as the presence of the Turkish house to the north and the western baulk of the excavation area, all of which truncated the dumped fills. Given that the pottery dates for each of the three layers described above, it is difficult to determine whether these dumps are related to a single destructive event in the wider neighborhood or represent a series of independent events. One possibility may be that this area lay behind a N-S wall which once filled Cut 1040, which we believe to be a robbing trench. Thus, this may have been a convenient exterior dumping ground used over time. Another possibility may be that a large destructive event, such as the earthquake of 522, afflicted the area causing several households to dump their debris in one single area as a part of a general clean-up project. However, at this point we are reluctant to connect these deposits to a single historical event without further study of the pottery from these levels. Above these dumped fills were a series of leveling fills, which may have been laid down in order to reuse the area for new purposes at the end of the Late Roman period.

The remaining Late Roman activity was located in the northwestern part of the excavation area. A well preserved section of a drain, Structure 1026 (N-S 1017.17- 1014.96, E-W 260.14- 259.75), ran N-S at the northern edge of the excavation area and was parallel to the robbing trench 1040. It appears to have been truncated by the building of Wall 918. The drain dates before the middle of the 6th century CE, based on stratigraphy, but since the structure itself and the fill it rests upon were not yet excavated, this date remains preliminary. The drain may have been connected to the well—excavated in the 1960s—located just north of it, immediately to the east of the Turkish house and may have been part of the drainage system of structure related to the wall we theorize filled Cut 1040.

The latest Late Antique activity in the area is present in thin deposits preserved below Wall 540 on the east side of the excavation area. The wall, which dates to the early 12th century, was left pedestalled during the 2012 season (top H. 86.56m, bottom H. 85.98m). Excavation continued on both sides of the wall, and after its removal we were able to correlate this season’s deposits to some of those excavated in 2012.

The pedestal below Wall 540 was preserved in three sections, interrupted by two robbing trenches, Cut 980 and Deposit 985. We excavated the two northern sections and Daniel Diffendale and Mohammed Bhatti excavated the southern section. Each of the two northern sections of the baulk below the wall contained two tile destruction levels, separated by a layer of leveling fill. This clear sequence allowed us to find equivalent layers of the baulk across the robbing trench of Wall 918/992 (filled by Contexts 975 and 985), which separated our two sections of the baulk.

The lowest layers of the pedestal were excavated as Deposits 990 and 991 (990 N-S 1016.20-1014.64, E-W264.68-263.66; 991: N-S 1014.62-1011.63, E-W 264.59-263.76) and correspond to Structure 936 (dated to mid 6th- early 7th century CE), a compact floor surface excavated in 2012. They also relate to Deposit 1035, excavated by Bhatti and Diffendale. Given the narrowness of the pedestal, we cannot add much to the interpretation of this surface at this point. The possibility of further succession of floors below this level was suggested by George and Valente in 2012 report, but was explored by Bhatti and Diffendale during this season.

The earliest tile destruction layer, Contexts 989 and 984, are equal to Contexts 934 and 929 excavated in 2012, due to their roughly equivalent elevations and presence of tile. This allows us to confirm that this tile destruction layer ran under the wall, as theorized by Valente and George in 2012. Our tile layer must have been part of a larger dump of tile, possibly resulting from the clean-up of a destroyed building elsewhere, or—as theorized in 2012—from the collapse of a roof. We were not able to equate the other layers of the pedestal (including the later tile layer) to levels excavated in the previous season, unfortunately. This may be due to a disturbance from the foundation trench of the wall or other activity in the area after the Wall 540 was built, which preserved the later tile layer under the wall, but removed it to the east and west. The layer intervening between the two tile dumps and the later tile dump itself date to the late 6th/early 7th century CE, while the top-most level of the baulk (Contexts 974 and 986) dates to the 7th century CE.


Middle to Late Byzantine (802-1210 CE)

During the mid-10th to late 11th centuries CE, there was a very large pit in the middle of the excavation area (Cut 870, N-S 1014.56-1011.63, E-W 261.89-258.43), which was previously identified as a bothros by George and Valente in 2012. This pit seems to have been filled over a period of time, with the cut and first preserved fill (Context 1002) dated to the late 10th-early 11th centuries and the latest fill (Context 844, excavated in 2012) dated to the late 11th century. This extremely large and deep cut may have functioned as a disposal bothros during this period. The bothros seems to have been filled by the time that the Late Byzantine walls in the area were built, so we are currently unable to associate it with any architecture. However, it must have existed in an exterior space at its time of use.

There seems to have been a flurry of building activity in this area during the late 11th and early 12th centuries. Wall 747/1087 seems to have been erected first, in the late 11th-early 12th centuries (1087: N-S 1010.81-101.11 E-W 258.29-255.75; 747: N-S 1010.80-1010.40, E-W 264.15-263.30). During the 2012 season, the foundation trench (Cut 809) for Wall 1087 was partially excavated, but we removed the final layer of fill within the cut during this season, as Context 1086 (N-S 1011.38-1010.62, E-W 258.24-255.68), and we also exposed further foundations of the wall during cleaning. We are uncertain what space was defined by this wall in its earliest phase, but during a later phase in the early 12th century, it appears to have been abutted by the slightly later N-S Wall 540. Therefore, it seems that the space to the west of Wall 540 was divided into a northern and a southern section by Wall 747/1087. However, we cannot be certain whether these spaces were interior or exterior and we do not yet have evidence of any further walls defining the space.

Wall 540 (N-S 1015.20-1007.26, E-W 264.76-263.68), which consisted of two faces of roughly worked small boulders with cobbles and rubble in the interior, was removed during this season, allowing us to clarify its relationship with several other walls in the area. We are now certain that two E-W walls in the area, Wall 918/992 (N-S 1015.10-1014.57, E-W 266.12-259.56) and Wall 945 (N-S 1011.69-1010.89, E-W 264.50-260.95), were out of use by the time that Wall 540 was built, because Wall 540 was constructed over the robbing trenches of each of these walls. Perhaps these walls were removed in preparation for the building of Wall 540 or as part of a larger rebuilding project in the area.

Recommendations for the future:
1. Excavate any remaining fill in Cut 1040 and investigate the possibility that this robbing trench continued on the south side of Cut 870. This would shed light on our theory that the wall which once filled this cut was related to the dumping ground to its west.
2. Continue investigating the area bounded by Walls 992 and 945 to the east of Cut 870 in order to continue exploring the Late Roman levels. Special attention should be paid to any evidence of dumped destruction fills.
3. Remove Drain 1026 and the fill below it to clarify the date of the drain.
4. Consider excavating part of Wall 747/1087 for chronological purposes.
5. Clean the bottom of Cut 870 to better understand the material exposed at its bottom and take closing levels, which we were unable to do during this session.