Professor J. Crow and Professor D. Mektav
Survey in Thrace August-September 2008

This year’s field survey in Thrace (2008) marked the final year of fieldwork concerned with the aqueducts and water channels outside the city. The recent publication of the monograph (Crow et al. 2008) draws on fieldwork up to 2005, but since 2007 we have been collaborating in a joint project with Professor Derya Maktav of ITU and supported by TUBITAK and the British Academy. This research programme (2007-08) has utilised satellite and digital map data to develop further research into the Byzantine water supply system (Çeçen 1996) and to provide an extensive digital terrain model and GIS to document the complex system of Byzantine hydraulic monuments in the region west of Istanbul. This short report will outline the results of surveys in 2008.

Kurşunlugerme K 20
This bridge has been visited in every year of our survey and on this occasion our aim was to locate suitable mortar samples for future analysis and to acquire additional GPS data to map the channels. Fig 1 shows the results of the surveyed line compared with the known Orthophoto draped over the DSM available for Trakya. No further damage was noted to this bridge since last year. We also revisited the bridges at Büyükgerme K29 to collect mortar samples and take additional GPS points including a line east from K 29 towards Merdiven Dere, and further points at Kecigerme K 30.

Kumarlıdere K 31
Kumarlıdere is the third of the long stone bridges and measures 130 m in length with a surviving height of 30 m (Crow et al. 2008). However, examination of the IKONOS satellite images of the forest shows that there are two long straight lines leading to the bridge from north and south, indicating the position of two embankments carrying the water channels towards and away from the stone bridge (Figure 2-3). The total length of the embankments and stone bridge has been measured at 320 m using the Trimble GeoXT. The combined embankments and bridge create a remarkable man-made surveyed line across the forest cover and whilst it is possible to see the 60 m width of the embankment on the ground, the

Figure 1 Orthophoto draped over DSM (digital surface model) showing the valley at Kurşunlugerme looking towards the east. The spots mark GPS points taken along the line of the lower channel, with one marking the line of the higher channel, the vertical bridge is less clearly defined from the horizontally derived image.

satellite image draws attention to the scale of the entire structure. Each of these major bridges was a huge undertaking for a pre-modern society and they represent an enormous investment in manpower and materials. Through the accurate identification of the full extent of the ancient works it allows scholars to recognize the scale and significance of the water supply project for the ancient and medieval city of Constantinople. As a further aid to understand the scale of these great bridges we propose to create CAD visualisations of the bridge at Kumarlidere to draw attention to the size of the bridge and its embankments when first constructed (see Bayliss (2003) for previous study of the Anastasian Wall). Using the information now available through the GIS it will be possible to set this visualisation within an accurate topographic setting. One concern at this and other bridges was the growing evidence for deliberate damage to the stonework of the bridges, especially the carved crosses and other symbols (see Crow (2008) 158-80).

Figure 2: Photo of the Kumarlıdere Bridge (K. Çeçen, ITU Archive)

Figure 3: GPS points on IKONOS image. (Reproduced with permission of ITU).

Following the survey at Kumarlıdere we were able to follow the line of the channels from the south-east end of the bridge through the forest to the south towards Kalfa Köy. We were able to map more precisely the channel bridges using GPS. The first bridge was located at Kanlı Dere K32. The channel was very clear on the north-west side of the bridge, 1.80 m wide. The mortar in the channel side was preserved up to 1.50 m in height with evidence for a thin sinter. At the next valley Kaşıkçı Dere K33 the remains of two bridges were located, followed by a third valley at Ayazma Dere K34 where foundations were visible. South of the village the line of the wide channel could be followed as an embankment for the wide channel in the forest although it was also possible to find traces of the narrow channel running parallel to it. East of the road to Akalan the forest was extremely dense and traces were seen in few places along the Derinçatak Dere. At one place we were shown a section of the wide channel measuring 1.40 m in width, the most easterly occurrence of the line yet recorded before the ridge at Haydut Tepe (see Crow et al. 2008, 75). We were able to return to the remains at Balık Germe which were now extremely overgrown.

Channels between Dağyenice and Tayakadın
Beyond the ridge at Haydut Tepe the line of the channel (including now the line from the spring at Pınarca) turns north-east of the village of Dağyenice (see map 7 in Crow et al. 2008) where traces of channels are seen towards Kemer Tepe. On the Kemer Tepe ridge a line of trees and bushes masks the remains of a ruined arcade leading across a short col. Traces of a narrow tunnel were noted to the west of the ridge and immediately north of the arcade traces of ceramic and tile indicate a settlement of Byzantine date. To the south of the village we were shown the traces of a channel and limestone quarry representing the line of the channel from Pınarca. In a field close to the forest edge traces of ceramic and tiles suggest a settlement of Byzantine date.

As the aqueduct system approaches the city of Istanbul in places the forest is less dense and the channels can be traced as a line in fields or in clearings. Remains of narrow channels 0.85-0.90 m wide were seen north of Boyalık in Erikli Ormanı. Further traces were located around Tayakadın village including the remains of the entrance to a narrow tunnel (see Crow et al. 2008) north-west of the village. Beyond the tunnel the line of channel follows the Alibey river towards the city, and the bridge at Büyük Kemer K35 was revisited, with new evidence for stone robbing. The channel crossed the next shallow valley to the south-east (between K35 and K36; see Crow et al. 2008, map 9) on a raised wall c. 6 m broad; the channel was clearly visible at a width of 1.20 m. It was of great significance that there was no trace of a second parallel channel at this point and this observation confirms the single channel noted from Dağyenice eastwards. Also of great significance was the cross section through the channel base revealing a sequence of five distinct mortar floors representing different phases of construction and repair throughout the tunnel’s life (Fig 4).

Fig 4 Detail of the base of the water channel south-east of Büyük Kemer showing 5 layers of channel floor (J. Crow).

References
Çeçen, K. The Longest Roman Water Supply Line (Istanbul 1996)
Crow, J., Bardill, J. and Bayliss, R., The Water Supply of Byzantine Constantinople (JRS Monograph 11, 2008)

Acknowledgements
We are very grateful to the General Directorate for permission to carry out our survey in 2008 and to our representative Günay Karakaş of the Edirne Museum for her interest and assistance. Staff at the Ingiliz Arkeoloji Enstitusu were of great help in arranging the programme. In addition we would wish to thank team members; Celal Kolay of ITU and students from ITU, Marmara and Edinburgh Universities, Irfan Akar, Cihan Uysal, and Riley Snyder, who assisted us. Finally we wish to thank many of the villagers throughout the region for their interest and enthusiasm in helping us to locate the many monuments we were able to visit this year.