Richard Bayliss and James Crow
Water Supply of Constantinople 2002

This research project, incorporating fieldwork as a major component, seeks to advance our understanding of urbanism in Constantinople throughout its history by investigating the provisions for water supply. Completed by the emperor Valens in AD 373 and supplemented by additional channels in the 5^th-6^th centuries, this is the longest known Roman water supply system, the main branch from Vize being in excess of 250km. Archaeological and hydrogeological research is being carried out on the channels and bridges outside the city and the cisterns and reservoirs within.

Our fieldwork in 2001 had focused on the channel from the Ballıgerme Aqueduct eastwards and also on the major Danamandıra (Papu) tributaries. We were able to establish the physical relationship between this source and the main supply line and demonstrate that the springs of Danamandıra had provided a substantial source for the original supply line in the 4^th century (Bayliss, Crow, Bono, 2001). East of Ballıgerme two principal phases of the system have been identified: a c.1m wide channel representing the primary 4^th century phase and a slightly lower and larger (1.6m wide) channel representing a massive supplement to the system, probably dating from the 5^th century. The lower channel required new bridges to be built and in many cases the old bridges were abandoned, with both wide and narrow channels subsequently carried across the new, larger structures. In September 2002 we undertook our second full season of fieldwork on the aqueducts outside the city, in addition to a short investigation of the city’s cisterns earlier in the Spring.

The principal objective of the 2002 summer season was to investigate in more detail the evidence for the water supply system on the long (110 km) stretch between Vize and the junction with the Danamandıra tributary.

The archaeological remains in this region are more fragmentary than further east where the central region is heavily forested. Nevertheless it is possible on this long stretch to observe the remains of both narrow (0.6-1m) and wide (1.5+m) channels, which seems to represent two different phases of the same supply line. Numerous bridges were visited in this area and several expeditions were made through the forests where new channel observations were made.

Between Binkılıç and the Karamanoğlu Dere the channel was seen on several occasions in tributaries of the İstranca Dere (Manganez Dere, Cineviz Dere, Babadar Dere and Elmalı Dere) and was found to be uniformly 1.5-1.6m wide. Further to the west, in the catchment of the Ergene Dere north of Saray, reconnaissance between the Ayvacık Dere and the Galata Dere revealed that the main channel was only 90cm-1m wide, although near the village of Kavacik an additional channel, 1.6m wide, was found some 22m below this narrow channel.

In addition to this reconnaissance detailed surveys were carried out on the Ballıgerme and Büyükgerme aqueduct bridges. The former crucially lies close to the junction between the Vize and Danamandıra lines and both bridges preserve evidence of late repairs. Using a Trimble DR200+ Reflectorless Total station we produced accurate plans and elevation drawings of these two bridges. At the same time Jonathan Bardill (Newcastle University) undertook full study of the mason’s marks on the bridges and was able to identify some significant correlations.

On the kind advice of Gökhan Çağlayan and Dr Caroline Finkel we made an investigation of the area to the north of Subaşı and located a small channel with a pedimented roof situated in the hillside above the village. Further reports would suggest that this forms part of a major tributary channel with its sources close to the village of Pinarca, which we investigated and reported in 1998 (Bayliss, 1999). Based on the topography of this region we can suggest that this tributary would have joined the main channel south of Kalfaköy.

In Spring 2002 Crow and Bayliss were also able carry out field survey on behalf of the Silivri Belediye, establishing the line of the of the Anastasian Wall to the north of the city in advance of future developments. As part of this survey we also discovered three new types of brick stamps associated with the Wall. This completed our investigation of the remains of the Wall in this area, which were begun in 2000 with a campaign of topographical and geophysical survey. Our findings on the location of towers were substantiated by evaluation excavations undertaken by Turan Gökyıldırım of the Istanbul Archaeological Museum in May 2002.

Finally a further investigation was made of the Ottoman supply line which runs 15km to the west of the city to sources in the vicinity of Halkalı. Here the Ma’zulkemer aqueduct bridge is of particular interest as the only apparent evidence of an earlier Byzantine or Roman system on this line (Mango, 1995). Considerable effort was made to gain access to this bridge, which is located within a military zone. With the support of Dr Alpay Pasinli and the assistance of the Topkule Kışlası personnel we were able to study the water bridge in great detail. We have concluded that this bridge, while showing several phases of modification, is most likely Ottoman in date.

Following on from our previous work on the open-air reservoirs of the city we spent 10 days in Spring 2002 studying the city’s smaller Byzantine cisterns. Many of the cisterns recorded by Forchheimer and Strzygowski at the end of the 19th century (Forchheimer, Strzygowski, 1893) are now either unknown inaccessible or destroyed, making a reappraisal of their work somewhat problematic, but we were still able to study over 30 cisterns in some detail.

A primary objective was to investigate the topographic relationships between the cisterns and to look specifically at water distribution within the city. The main premise of the working model was that Byzantine water channels are known to have entered the city at two different heights. The long-distance supply line from Danamandıra and Vize entered the city near the Edirne Kapı at a height of 63-4m, from where it was able to supply much of the city. In addition, a second channel is known from the Belgrade Forest to the north of the city, which entered the Theodosian Walls near the Eğri Kapı, at a considerably lower level of 33-5m. This model has very significant implications for water provision to different areas of the city in the Byzantine period, particularly relating to the distinctions that can be drawn between the 2nd-century “Aqueduct of Hadrian” (reported in 6^th-century Byzantine sources) and the 4th-century long-distance water channel completed under Valens. A substantial paper on our discoveries is in preparation and is planned for publication in 2003.

It is with much gratitude that we acknowledge the assistance of Mustafa Demirel of the Antalya Museum and Neslihan Güder from the Monuments and Museums General Directorate who acted as representatives of the Directorate for the Spring and Summer seasons respectively. We also wish to thank the Directorate and its Director, Dr Alpay Pasinli for permitting and facilitating the continuation of our research in 2002. Once again we were kindly received by our colleagues at the Istanbul Archaeological Museum and in particular wish to thank the Director, Dr Halil Özek.

Mehmet Akif İşin of the Tekirdağ Museum was as ever an enthusiastic host and we were delighted that he was able to join us on a day of fieldwork in the vicinity of Kavacık. We also thank him for his time in the preparation of the documents for the sinter sample from Ayvacık. BaşkanıSelami Değirmenciand his staff in the Silivri Belediye made us feel most welcome in the city and we particularly thank Mr Özcan Işıklar (Deputy Director) for his great efforts to assist our research. We were fortunate to be joined for three days by Kristin Romey, managing editor of Archaeology Magazine, who has since produced an impressive full-colour article on our project (Romey 2003).

Funding this year was gratefully received from the British Institute of Archaeology at Ankara and the Leverhulme Trust. We thank all responsible for their continued support. The Spring season was undertaken by Richard Bayliss, Jonathan Bardill and James Crow. In the Summer season they were joined by Tom Crow (Geologist) and post-graduate students from Newcastle University: Naomi Belshaw, Aimee Lawrance and Claire Nesbitt.

Bayliss, R 1999. Anastasian Wall Project + Water Supply of Constantinople. http://museums.ncl.ac.uk/long_walls/index.html.

Bayliss, R, Crow, J, Bono, P, 2001: 'The water supply of Constantinople' Anatolian Archaeology 7: 18-20

Çeçen, K, 1996a: The Longest Roman Water Supply Line. Istanbul

Çeçen, K, 1996b: Sinan's water supply system in Istanbul. Istanbul

Forchheimer, P, Strzygowski, J, 1893: Die Byzantinischen Wasserbehälter von Konstantinopel. Wien

Mango, C, 1995: 'The Water Supply of Constantinople' in Mango C, Dagron G and Greatrex G (eds) Constantinople and its Hinterland. Aldershot: 9-18.

Romey, K. 2003. "Lifeline to Byzantium", Archaeology Magazine 56(1), 24-31.