The City Walls
The late-Roman city of Selymbria, renamed Eudociopolis after the wife of Theodosius II, was one of a number of new fifth-century foundations in the eastern Thrace. The ancient city was defined by a circuit of wall approximately square in form, with the south side overlooking the Sea of Marmara in high cliffs. The city walls have received very little archaeological attention with the exception of an account by Feridun Dirimtekin, published in 1965. Photographs taken at the time of the Bulgarian occupation during the first Balkan War (1912-13), show gates and walls surviving nearly to their full height on the north and east sides and these supplement the sketches and description of the city made by Dr John Covel in May 1675 (1998). Subsequently significant parts have been demolished, especially on the east side where little now survives.
In June 2005 with the kind assistance of the Silivri Belediye, we were able to use existing city plans to trace the line of the wall on the west and the north sides. We carried out a detailed survey using a total station of the east side which is especially well documented from the Bulgarian photographs. It is clear that in the area of the Kale Park, in the south-east corner of the old city there was a now lost, domed Byzantine church distinct from the better-known church of the Apocaucos, published by Eyice (1965) and others. On the north side the walls survived well amongst modern apartment blocks. The remains of two main gates were identified from earlier accounts and a large U-shaped tower survives mid-way along the north curtain. Brick stamps and mortar impressions were found in situ adding to those already discussed by Dirimtekin (1965, 32-33). These were found on the east curtain and at the Kır Kapısı. Both could be dated to the first half of the fifth century. Parts of the north curtain are well preserved in later house walls towards the west and there is a wide brick arcade as part of the curtain wall construction. Similar bricks stamps were found dated from the fifth century, but in an additional inner arcade the stamps could be dated to the reign of Justinian, probably representing the restoration noted in Procopius De Aedifiiciis.
To the west, overlooking the harbour, the walls were constructed near the top of the hillside with a distinctive sequence of brick arcades. The curtain wall was cut into the sandstone bedrock which has been severely eroded on the downslope side. Once again two main phases of brick stamps were seen in the vaulting, the outer belonging to the early-mid fifth century and later, inner vaults dating from the sixth century. It was possible to recognise 11 continuous rows of arcades, 2.85 m wide, before later works and restoration intervened. The extensive use of blind brick arcades found on both the north and west curtain walls at Silivri was a characteristic of other major fortifications dateable to the fifth century in the Balkans - at Thessalonike and Perinthos, at Antioch in the East and on the sea-walls of Constantinople (although the date of the latter remain contested).
The Anastasian Wall
The Wall Underwater west of Silivri
In 2000 we carried out a brief survey with a total station of the submerged jetty or liman at the south end of the Anastasian Wall. With the advice of Professor Nergis Gunsenin, and with the practical help of Evren Türkmenoğlu we were able to investigate these remains off the Altinorak beach. The underwater survey confirmed the basic V-shaped outline of the feature extending 200 m at right angles to the shore line. The liman was made up from a compact mass of squared blocks encrusted with mussels. By clearing limited areas it was possible to show these were eroded limestone blocks, similar to those used in the construction of the Anastasian Wall. Amongst the blocks were fragments of broken tiles and mortar with crushed brick typical of the construction of the Anastasian Wall, there can be little doubt that this feature represents the south end of the wall, similar to the emboloi described by Procopius for the wall at Gallipoli. On the beach there was a clear spread of building material, stones and tiles continuing this line inland to the north-east where traces of the Anastasian Wall were located by Türan Gökyildirim of the Istanbul Archaeological museum on the hillside above.
From 1994-2000 we had carried out detail survey of the line of the Anastasian Wall from the Black Sea at Evcik south towards the ridge east of Gümüspinar (Derviş Kapı) see
The forest clearance in the vicinity of Nikol Dere (see below) exposed a 1 km length section of the Anastasian Wall descending the ridge in the direction of the wide valley of the Binkiliç Dere. The wall survived as a 3 m high bank with one to two course of facing blocks visible in places.
Where this ridge entered the valley of the Binkiliç Dere, above a side channel known as the Hamza Dere, we had been shown in 2004 the deep pits dug by definiciler into a section of the wall at the foot of the slope. The diggings reveal the wall standing over 2.5 m high faced with large limestone blocks and two deep pits have been excavated into the core of the wall through to the bedrock. They were over 3 m deep and will cause this part of the wall to decay and collapse as well as representing a hazard to children and livestock. These pits need to be filled as soon as possible.
There are no traces of the wall in the broad flood plain of the valley, but to the west of Karacaköy the line resumes and passes through very dense scrub and woodland. Two and a half km north-west the wall can be followed on Sayalar Tepesi where the wall survives as a high mound 2-2.5 m high. We were able to trace it for 700m where it was cut the modern road to Yalıköy and crosses the valley of the Kara Su Dere and then heads north towards Evcik and the Black Sea.
Water Supply System
Fieldwork on the water supply system was limited to revisiting Pınarca and to sites located in the villages of Belgrat and Ciftlikköy. The three springs at Pınarca are located 2.5 km south-east of the Anastasian Wall, clearance in the narrow valley of the Tek Göz Mağara showed more clearly that the sides were walled up, suggesting that the channel may have been damned to serve as a reservoir. We were also shown additional channels leading to a third cave and spring source, reinforcing the importance of Pınarca as a major source for the first phase of the long-distance water supply system.
Near Belgrat we were informed that recent forest cutting had revealed three bridges in one valley, the Nikol Dere, located 500 m north of the known aqueducts across the Maçka Dere. The presence of three bridges is most unusual and contradicts the standard practice of bridges in this part of the system: one for the high level and another for the broader, lower channel located 4-5 m apart. The upper bridge was single arched and well preserved. It clearly carried a broad channel and at the north end the tunnel could be followed and seen to turn to the north-west. This is against the line of the contours. Below this was seen the more ruined remains of a second bridge with a single arch, and constructed with long bossed blocks, different from the equilateral blocks in the top bridge. Below the second bridge was a third bridge, this was more ruined than either of the other two and was constructed with large schist rubble work. A number of the features in the Nikol Dere are unique but can be summarised as follows: the primary bridge was the middle bridge, this conforms with the pattern of high-level bridges, clearly seen last year in the Maçka Dere, close-by. The second phase broad channel can be expected at a lower elevation, in the site occupied by the bridge of schist block-work. The equivalent bridge in the Maçka Dere was also extensively rebuilt, but did retain traces of earlier work. At the Nikol Dere however the lower bridge appears to be completely rebuilt probably in the middle Byzantine period.
The problem is why there is a third, higher bridge with a broad channel. Unusually the tunnel turns into the hillside to lead north-west, not north-east along the contours as would be expected. Only 200 m to the west on the ridge above is the line of the Anastasian Wall. We found the crossing of the wall and water channels in 1997 and this is located about 750 m down the ridge to the north. It would appear that the new bridge was clearly intended for another tunnel beneath the Wall, about 300 m long, connecting the channels on the west side of the ridge and at a higher elevation along the line. The new tunnel and channel reduced the line by between 1.5-2 km and the work could be associated with the construction dated by the Longinus inscription in the reign of Justinian.
This bridge (K 25) was illustrated but not described Prof. Çeçen (1996, 175) and is situated west of Heyliye Tepe, in the district of Ciftlikköy. The bridge has a well preserved single arch 5.54 wide and two buttresses on the north (downstream) side, the stringcourse was preserved on the south side and was of the distinctive downward chamfer seen at Talas and Leylekkale. This work is characteristic of the major restoration associated with the Longinus inscription and the two examples visited this year provide us with further evidence that almost all the known bridges of the main second phase (fifth century) were extensively rebuilt in the sector from Elkafdere in the west to Büyükgerme in the east, possibly the result of very specific scheismic damage during Justinians reign. The only exception is the distinctive bridge at Luka Dere (K 17), this has similar structural features, buttresses and string courses, and is probably also sixth century in date but represents the single restoration of a bridge further to the west.
Kecigerme K 30
In his study of the bridge at Kecigerme, the Bulgarian archaeologist Oreshkov reported that he was able to see a relief carving of the busts of two figures on the keystone of the south side of the lower, central arch. We were able to confirm that this still survives but because of the dense vegetation masking the carving we were unable to improve on his rough sketch. It was clear that one figure was bearded. We could also identify the rough tabula ansata below, but there was no indication if there was an inscription with in it.
Once again it is our pleasure to be able to record our thanks to the General Directorate of Ancient Monuments and Museums for permission to carry out the survey in 2005 and to their representative Ilknur Subaşı for her support and encouragement durıng the season. We would also wish to express our gratitude to Dr Ismail Karamut of the Istanbul Archaeological Museums and the staff of the Silivri Belediye, especially Mrs Berrin Papila, for their assistance. Finally we should like to thank the villagers of Belgrat, Ciftlikköy and Pınarca for their interest and practical assistance. The survey team comprised Dr Jonathan Bardill, Dr Richard Bayliss, Alex Crow, Harry Fraser, Claire Nesbitt, Jonathan Shipley (Newcastle) and Evren Türkmenoğlu (Bilkent University).
Cecen K. The Longest Roman Water Supply (Istanbul 1996)