Waters for a New Rome: From Constantine to Heraclius
Research over the past two decades has revealed the complexity and magnitude of the late Roman water supply system developed in the century and a half after Constantine’s foundation of the new city of Constantinople. By the reign of Theodosius II in the early fifth century, the system could merit the claim that this was The Longest Roman Water Supply Line1, a network of channels that has now been estimated to be 494 km in length, including the pre-existing channels from the Belgrade Forest2. Unlike many great ancient cities, Constantinople was not built on a major river. The seas that surround Constantinople on three sides ensure an excellent defended situation, good communications for trade and imports, and fine fishing resources. However, there was, and remains, a major deficit of good fresh water, certainly in comparison to the variety of sources available for imperial Rome. In the early second century AD, the emperor Hadrian had sponsored a major aqueduct for the city of Byzantium, the predecessor of Constantine’s new capital, and these waters were probably drawn from the springs in the Belgrade Forest, the same region exploited and developed in later Byzantine and Ottoman times. Initially, these existing channels supplied the new city, including the new Great Palace, but as the urban area expanded on to the higher ground west of the Forum of Constantine, towards the site of Constantine’s mausoleum at the Church of the Holy Apostles (from Çemberlitaș towards the Fatih Mosque), new higher water sources were needed. In the ancient world, and until the nineteenth century, piped water could be transported over long distances only by force of gravity. It was therefore necessary to ensure that the water sources in springs or possibly rivers were at a higher level than the point of delivery, with sufficient height to provide an adequate head of water.
This chapter will provide a historical overview of the main components of the Byzantine hydraulic system over a millennium and will consider evidence for storage and distribution in the city, as well as outlining the major works in Thrace. As the new urban districts of Constantinople expanded towards the Walls of Constantine and subsequently the new circuit of Theodosius, the physical area defined by its walls and coastline exceeded all other ancient cities apart from Rome (Ward-Perkins, 2012). One problem in attempting to provide a chronology of the city’s development is that the various writings for Constantinople’s urban history date to differing eras, and often present quite distinct and often conflicting narratives. Thus in the Patria, compiled in the tenth century, almost everything concerned with the initial centuries of the city’s foundation was attributed to Constantine. This applies to the history of the main water supply, but for this key component of the developing urban infrastructure, the fourth-century source of the orator Themistius defines the importance of the new water supply by commenting that, whatever the earlier emperors Constantine and his son Constantius II had given the city by embellishing it with statues and public buildings, it remained nothing more than a sketch that was “girdled by gold but dying of thirst”3. The new aqueduct was a vitally needed element of the city’s infrastructure required for the expanding city. The works were only to be completed with the appearance of the “Thracian nymphs” accompanying the new waters from the west in AD 373. From an earlier oration it is stated how Constantius II had sent surveyors to seek new water sources in AD 357. This provides presents a time-scale of less than two decades for the completion of the new high-level water supply.
The New Water Channels
The water supply of the earlier city of Byzantium was provided by the emperor Hadrian in channels that we can estimate could have flowed along the northern flanks of the city’s hills, above the Golden Horn, at an elevation of c.35 m above sea level (Figure 1). The line of the channels was probably very similar to the known Ottoman channels of the Kırkçeşme system from Cebeciköy and the Belgrade Forest. Much of our knowledge of the new city derives from the Notitia Urbis4, lists of the city’s regions and major buildings which date from around AD 430. From this evidence we can estimate that the Hadrianic channel would have been able to provide water for the lower regions of city including six thermae (public baths) out of a total of nine, one major cistern, two nympheae (public display fountains) and a total of 98 balnea (private baths) out of a grand total of 156 for the entire city.
In the first phase of the long-distance water supply from Thrace during the middle of the fourth century, new channels were now able to enter the city at an elevation of c.56–7 m above sea level. In order to achieve the greater elevation, distant springs from the hills and forests of Thrace sourced the new water channels for the growing capital (Figure 2). By AD 373, the first line of channels and bridges bringing water from major springs at Danamandra and Pınarca was completed, consisting of a total length of 268 km, with an estimated 130 new bridges varying in size from single arches to tall, two-tiered ones. The covered water channels were all constructed from mortared stone blocks, with vaulted roofs, 1 m wide and 1.6 m in height. Within the city, the new bridge—the Bozdoğankemer or Aqueduct of Valens—with its 87 arches and length of 971 m counted as one of the longest in the Roman world (Figure 3). It led the water channel from the Fourth to the Third Hill and on to a major fountain and pool (nymphaeum) near today’s Beyazıt, although no trace of this survives. From here, water could be distributed across the higher parts of the new city and towards the Forum of Constantine (today Çemberlitaş).
But this was not sufficient, and probably around AD 400 or shortly thereafter (at about the same time as work commenced on the Theodosian Land Walls), it was decided to extend the system further to include more springs further to the west around Binkılıc, Ergene, and Vize. These new springs were almost 130 km away from the city, and by the time this second phase of the supply system was completed, around AD 450, the total length of all the aqueduct channels (including the Hadrianic line) was 494 km, comparable to all the aqueducts of imperial Rome, constructed over six centuries. The new channels were larger than those built in the first phase, up to 2 m in height and 1.6 m wide. Because of the steep hills and deep valleys along the length of the water supply line, the course of the channels was often very sinuous, and to ensure the correct gradient for water flow, they twisted in and out of narrow valleys and creeks. Many of the bridges constructed in the first phase were replaced as part of the construction of the new channels of the second phase. The grandest and largest is the great bridge at Kurşunlugerme near Gümüşpınar. Set in a deep, wooded valley, the bridge rises in three stages to more than 30 m. Constructed of an almost marble-like crystalline limestone, it was decorated with a host of Christian symbols placed on the keystones and other structurally important points intended to protect against the devastating effects of earthquakes. Five other bridges are comparable in size and grandeur to Kurşunlugerme and many other smaller bridges were constructed at the same time. These bridges within and outside the city are among the most impressive survivals from anywhere in the Roman world.
Distribution in the City
Constantinople has no technical manual comparable to Frontinus’ De aquaeductu urbis Romae; however, a sequence of imperial edicts collected in the Theodosian and Justinianic law codes provides valuable evidence for the maintenance and management of the system. From the reign of Theodosius II, an edict probably dating to c.AD 440–441 states that the waters for the Aqueduct of Hadrian were to be used only for public baths and the imperial palace, and that permissions for domestic use or irrigation were to be withdrawn5. The implication of this law is that the high-level aqueduct, which by this time extended as far as Vize and to which later texts refer as “the city’s aqueduct,” was the main provision for domestic demand, for the large number of private baths, and for the higher-level public baths and cisterns noted above. A number of the law codes make reference to problems of water theft for irrigation outside the city, and it seems likely that these edicts responded to increased demand from within the city. Certainly these higher channels also fed the three great open reservoirs located in the west part of the city, beyond the line of the Constantinian walls (Figure1).
Remnants of the distribution channels within the city are rarely encountered in excavations or building works. Wide-diameter stone pipes are known from beneath the central street of the old city and examples can be viewed in Istanbul Archaeological Museum6. Although the lines of the channels within the city are not known with certainty, we are able to establish approximately where the waters entered the city and, based on an assumed regular gradient of 1 m per km, how some of the major channels continued, given the fixed points for the high-level channels along the Bozdoğankemer. The high-level channels entered the city at a point north of Edirnekapı, at an elevation of c.59 m, and the Hadrianic, lower line from just north of Eğrikapı, at an elevation of 35 m (Figure 1)7. The projected course for the upper channel passes close by the known locations of the Aetios and Aspar reservoirs, maintaining the correct elevation to cross the Valens Aqueduct, before turning south towards the Forum Tauri, proceeding along the line of the mese (the city’s main colonnaded street), and finally terminating at the Binbirdirek Cistern, on the promontory of the Second Hill. Other clusters of cisterns lying away from this line, but situated above 35 m, must also have been supplied by a channel of this altitude. These include the cisterns around the Fethiye Mosque (Pammakaristos) and the concentration in the vicinity of the Büyük Valide Han near the Nurosmaniye Mosque.
The most recent study of the city’s cisterns8 has documented 160 cisterns within the city, yet these probably represent a small fraction of the original number. From this and earlier studies, it is apparent that cisterns were constructed throughout the life of the Byzantine city and continued to be built in Ottoman times, and that they represent the single most numerous structure dating from the Byzantine period. Some were major public reserves for storage and distribution, but others are likely to have been for private use. It is not fully understood how far some of the smaller cisterns formed part of a wider water supply network or whether they were simply an aspect of individual rainwater harvesting, where water was collected from the roofs of churches and other structures. This practice is known extensively from the southern and eastern parts of the Mediterranean and the Near East.
Roman-period cisterns in urban contexts are found most extensively in North Africa and the eastern provinces, while in Rome the largest ones were reservoirs for the great imperial thermae9. The scale of the investment in the new city of Constantinople was far greater, and was partly a response to the insecurity from barbarian peoples across the Danube. There were also other local factors such as the karst limestone catchments for many of the water sources, which meant uneven flow across the year and the need to retain storage at other times10. This principle was first asserted in Procopius’ account of the construction of the Basilica Cistern (Yerebatan Saray) by Justinian and helps to explain the vast size of the main open reservoirs and the great covered cisterns. Another such structure the Binbirdirek responded to the need to construct a very large new reservoir within a crowded urban area—the depth from the capitals to the estimated floor is 11.9 m11, while the Yerebatan Saray, which was constructed within the existing courtyard of a very large public building, both dating to the reigns of Anastasius and Justinian. These structures share the same problems as underground car parks in densely packed modern cities, and are indicative of the cost of real estate in a crowded city in the sixth century. In terms of their potential cubic capacity, they are the two largest covered cisterns known from the ancient world, far exceeding the largest surviving cisterns of Roman Italy. Public water fountains are less well known.
The chronology of the Constantinopolitan cisterns remains uncertain. None of the structures for the fourth- or early-fifth-century foundations known from texts can be identified with certainty, but the pattern of cisterns with brick-vaulted roofs supported on columns and often elaborate capitals rather than vaults was to become the norm across the city. These elegant dark interiors retain great fascination, even when they do not boast the “Disneyfied” touristic spectacle of the Yerebatan Saray. By contrast, even the grandest of Roman cisterns, such as the Piscina Mirabilis at Misenum, was a basic utilitarian structure. Roman skills in constructing vaults and arches ensured that the roof spaces could far exceed the smaller, domestic, rock-carved cisterns of the Greek and Hellenistic worlds, but they remained gloomy and uninspiring. To modern eyes, the Byzantine cisterns of Constantinople represent an apparently redundant and enigmatic elaboration of decorative motifs on bases, columns, and capitals. Some of these were most certainly recycled, but others used drafted capitals straight from the Proconnesian marble quarries, some of them completed, while others were merely roughed out. These supported the vaults of the developing hydraulic city, intended to sustain and succor the growing city above.
Past observers have commented that the forms of cisterns seen at Constantinople could be associated with and influenced by the very extensive network of cisterns known from Alexandria12. However, recent studies have demonstrated that that the two-tiered Alexandrian cisterns with columns are Islamic in date and do not belong to the Roman or Byzantine city. Elsewhere in Syria and Palestine, the pattern of cisterns followed the standard Roman form, even in late constructions such as the impressive cisterns from the Anastasian foundations at Dara and Resafa. An exception can be found at Ephesus, where a recently excavated, late antique cistern in the modern town of Selçuk has two rows of reused, second-century columns and capitals supporting now ruined vaults, presenting a clear example of the Constantinopolitan model of cistern construction13.
The scale of construction for both phases of the water supply system was vast, exceeding most of the great infrastructure projects of the ancient world. Recent analysis of the mortars and other building materials has estimated that, to build the bridges and channels, 2.5 million m3 of stone were needed. To put this in perspective, an area the size of a standard football pitch would have to be quarried to a depth of 350 m—50 m higher than the Eiffel Tower—with enough stone to build the Great Pyramid of Giza14. The great length of the system also revealed the skills of late Roman surveyors, as it was capable of leading gravity-fed channels through difficult terrain over distances never before accomplished in the ancient world.
After Justinian’s great project of the construction of the Basilica Cistern, there is textual evidence indicating that the aqueducts continued to be maintained up to the end of the sixth century. Repairs in 575–576 during the reign of Justin II “supplied the city with abundant water”15, and a later text specifically refers to both the Great Aqueduct and the Aqueduct of Hadrian, and further investment continued until the reign of Heraclius.
Iconoclast Revival to Comnene Decline (765–1180)
In 626 the city was threatened by a major Avar siege and the Aqueduct of Valens was cut, not to be restored until 765. The period from the Avar siege until the later eighth century is one of urban decline at Constantinople, followed by repeated outbreaks of plague, frequent sieges by the Arabs, and the loss of major territories in the east and the Balkans. However, during this period, there is no reason to suspect that the earlier Aqueduct of Hadrian did not continue to function, although this limited those regions with a direct water supply. The work of restoration by Constantine V on the water supply system was part of a major program of repairs to the walls of the city, and extensive reconstruction of the church of St. Irene16. The text of the chronicler Theophanes on the reconstruction of the water supply is especially important, since it provides unusual details about the size of the workforce and their places of origin. Constantine V:
Collected artisans from different places and brought from Asia and Pontos 1,000 masons and 200 plasterers, from Hellas and the islands 500 clay-workers, from Thrace itself 5,000 labourers and 200 brickmakers. He set task-masters over them including one of the patricians. When the work had thus been completed, water flowed into the City.17
Extensive structural evidence survives from many of these aqueduct bridges and it is argued that these structural phases correspond to the work described in this text18. The text of Theophanes is quite clear that the Aqueduct of Valens was restored following a period of drought, cisterns and baths did not function, and springs dried up. Restoration of the Aqueduct of Valens will have brought a new and additional supply of water into the city, supplementing the continuing supply from the Aqueduct of Hadrian, which was clearly affected by the severe drought, but had maintained a supply since the seventh century. Apart from extending the range of water sources for the city, the renewed high-level system will have enabled water to reach the more elevated parts of the city, including the church of the Holy Apostles high on the fourth hill and close to the west end of the aqueduct bridge, the Bozdoğan Kemer. The renewed hydraulic system signified the revival of the city’s fortunes in the later-eighth century. A late-ninth-century Arab description of Constantinople wrote:
There is at Constantinople an aqueduct from the country called Bulgaria. This water flows towards the aqueducts from a distance equal to a voyage of twenty days: and when it enters the city, it divides into three sections: one part goes towards the imperial palace, a second part into the prison of the Muslims, and the third part into the baths of the patricians, and the population of the town drinks from the water which is lightly salted.19
No work of restoration is attested during the tenth century, but early in the next century, c.1019, Basil II renovated the Aqueduct of Valentinian (Valens), providing “the citizens of Constantinople with a plentiful supply of water” (Skylitses History, 366). Major repairs are known from the aqueduct bridge at Ballıgerme indicate a fourth major period of work typical of the predominant style of construction known in the middle-Byzantine period.20 Work on this bridge is significant, as it shows that the water for the city was still being drawn from the distant springs near Vize and Saray. However the long-distance channels and bridges, the most extensive from known the ancient world, were beginning to show their age. And in the following century, a historian noted that in the reign of Manuel I Komnenos:
When a lack of water beset (the city) he carefully cleaned the channels. Since he noted that the arcades (stoai) which conveyed water to Byzantium were long since collapsed, and that it would be a difficult task to reconstruct them, one requiring much time.
The old bridges in Thrace were felt to be beyond repair, and Manuel undertook new works closer to the city at a place called Petra; however, the precise location remains unknown. Henceforth the main water sources lay closer to the city, from the same springs that had fed the Aqueduct of Hadrian and that were to provide major sources for the Ottoman city.21
From Byzantine Constantinople to Ottoman Istanbul
Although the major high-level watercourses from distant Thrace ceased to flow from the later twelfth century, there are still reports that water channels ran across the city’s long bridge, the Bozdoğan Kemer, probably from closer, less abundant springs at Halkalı, although the surviving Mazulkemer bridge along this line is an entirely later Ottoman construction. Theodore Metochites, a statesman and poet who endowed the mosaic decoration of the Chora monastery (Kariye Cami) between 1316 and 1321, wrote in praise of his native city:
There are reservoirs, fountains, and all kinds of great cisterns all over the city, sources of pleasant and abundant water. This comes from everywhere, both from local springs, and from afar, brought to the city by no little constraining and forcing of nature. Unavoidable necessity compels it to serve the imperial city, by all means and contrivances, some hidden, but others most obvious, which we can see and admire, as they parade the lofty flow of aerial rivers throughout the city. [There is] also the enjoyment of numerous big and beautiful baths, containing all kinds of ingenious devices, which I cannot describe, but can only urge you to experience and see them for yourselves.22
How much of this description is a nostalgic lament for past grandeur is difficult to assess. Most of the city’s great baths of late antiquity were certainly abandoned by this time, but “the aerial river” within the city survives to this day. Then, as now, it was not merely a memorial to past achievements but a continuing practical device, allowing waters to be directed to the higher parts of the city until the early-twentieth century. When Constantinople was finally captured by the Ottomans in 1453, one of the first tasks of the Sultan Mehmed II was the restoration of the water supply system. The former Byzantine rulers had provided an extraordinary legacy which reveals the ambitions of the city’s first founders and their successor’s achievements of sustaining a great urban centre for over a millennium. Until recently these marvels have been ignored and only now do we come to appreciate the exceptional monuments of hydraulic infrastructure surviving in modern Istanbul and in the forested hills of its western hinterland.
Cod. Just: The Civil Law: Including the Twelve Tables, the Institutes of Gaius, the Rules of Ulpian, the Opinions of Paulus, the Enactments of Justinian, and the Constitutions, (Çeviri ve derleme Samuel Parsons Scott), Cincinnati, 1932. [S.P. Scott’un yaptığı İngilizce çeviri internette mevcuttur: www.constitution.org/sps/sps.htm]
Altuğ, K., İstanbul’da Bizans Dönemi Sarnıçlarının Mimari Özellikleri ve Kentin Tarihsel Topografyasındaki Dağılımı, (Doktora Tezi), İstanbul Teknik Üniversitesi, İstanbul, 2013.
Altuğ, K., “Planlama İlkeleri ve Yapım Teknikleri Açısından Tarihi Yarımada’daki Bizans Dönemi Sarnıçları,” Restorasyon Konservasyon Çalışmaları 15, s. 3-22, 2012.
Bono, P., R. Bayliss, ve J. Crow, “The water supply of Constantinople: archaeology and hydrogeology of an early medieval city,” Environmental Geology, 40, s. 1325–33, 2001.
Brubaker, L. and J. Haldon, Byzantium in the Iconoclast Era, c. 680–850, Cambridge, 2001.
Çeçen, Kazım, Halkalı Suları, İstanbul: 1991
Çeçen, Kazım, Sinan’s water supply system in Istanbul [trans. of Mimar Sinan ve Kırkçeşme tesisleri], İstanbul, 1992.
Çeçen, Kazım, The Longest Roman Water Supply, İstanbul, 1996.
Crow, James, “Water and Late Antique Constantinople: ‘It would be abominable for the inhabitants of this Beautiful City to be compelled to purchase water’,” L. Grig and G. Kelly (der.), Two Romes: From Rome to Constantinople, s. 116–35. New York, 2012.
Crow, James, “Ruling the waters: managing the water supply of Constantinople, AD 330–1204,” Water History, 4, <AQ10.30> s. 35–55, 2012.
Crow, J., J. Bardill, ve R. Bayliss, The Water Supply of Byzantine Constantinople, London: Society for the Promotion of Roman Studies, 2008.
Forchheimer, P. ve J. Strzygowski, Die byzantinischen Wasserbehälter von Konstantinopel: Beiträge zur Baukunst und zur Topographie von Konstantinopel, Byzantinische Baudenkmäler 2, Vienna, 1893.
Mango, C., “The water supply of Constantinople,” in C. Mango and G. Dagron (eds), Constantinople and Its Hinterland: Papers from the Twenty-Seventh Spring Symposium of Byzantine Studies, Oxford, April 1993, s. 9–18, Aldershot. Society for the Promotion of Byzantine Studies, 1995.
Matthews, J., “The Notitia Urbis Constantinopolitanae,” in L. Grig and G. Kelly (eds), Two Romes: From Rome to Constantinople, s. 81–115. New York, 2012.
Müller-Wiener, W., Bildlexikon zur Topographie Istanbuls, Tübingen: Wasmuth, 1977
Skylitzes, History: John Skylitzes: A Synopsis of Byzantine History, 811-1057: Translation and Notes (Çev. John Wortley), Cambridge, 2012.
Snyder, J. R., A study of material technology, sourcing, and building organisation in the construction of the water supply of Constantinople and Anastasian Wall, (Doktora tezi, Edinburgh Üniversitesi), 2012
Themistius, Orationes: Themistii orationes quae supersunt / recensuit H. Schenkl; opus consummavit G. Downey, Teubner Leipzig, 1965-74.
Theodore Metochites, Byzantios (basılmamış yazma), (Vind.Phil.gr.95), fol 267v., (pasajın çev. Paul Magdalino) (Paul Magdolino’nun izniyle).
Theophanes, Chron. C. Mango & R. Scott, Theophanes Confessor, Chronicle of Byzantine and Near Eastern History AD. 284-813, Oxford, 1997.
Ward-Perkins, B., “The rise of Constantinople: old and new Rome compared,” in L. Grig and G. Kelly (eds), Two Romes: From Rome to Constantinople, s. 116–35, New York, 2012.
Wilson, A. I., “Hydraulic engineering and water supply,” J. P. Oleson (der.), The Oxford Handbook of Engineering and Technology in the Classical World, s. 285–318, New York, 2008
Wiplinger, G., “Die Wasserversorgung von Ephesos in Byzantinischer Zeit,” F. Daim ve J. Drauschke, Byzanz: Das Römerreich im Mittelalter, s. 593–613, Mainz, 2010.
* All maps and diagrams in this article were prepared and drawn by Richard Bayliss.
1 K. Çeçen, The Longest Roman Water Supply Line, Istanbul: İSKİ, 1996.
2 J. Crow, J. Bardill and R. Bayliss, The Water Supply of Byzantine Constantinople, London: Society for the Promotion of Roman Studies, 2008.
3 Orationes, XI.151a.
4 J. Matthews, “The Notitia Urbis Constantinopolitanae,” in Two Romes: From Rome to Constantinople, ed. L. Grig and G. Kelly, New York: Oxford University Press, 2012, pp. 81–115.
5 Cod. Just. 11.43.6; J. Crow, “Ruling the Waters: Managing the Water Supply of Constantinople, AD 330–1204,” Water History, , vol. 4, no. 1 (2012), pp. 35–55, p. 42.
6 W. Müller-Wiener, Bildlexikon zur Topographie Istanbuls, Tübingen: Wasmuth, 1977, figs. 303, 305.
7 Crow et al., The Water Supply of Byzantine Constantinople.
8 K. Altuğ, “Planlama İlkeleri ve Yapım Teknikleri Açısından Tarihi Yarımada’daki Bizans Dönemi Sarnıçları,” Restorasyon Konservasyon Çalişmaları 15, 2012, pp. 3-22.; K. Altuğ, “İstanbul’da Bizans Dönemi Sarnıçlarının Mimari Özellikleri ve Kentin Tarihsel Topografyasındaki Dağılımı” (Ph.D. Thesis), İstanbul Technical University, İstanbul, 2013.
9 A.I.Wilson “Hydraulic Engineering and Water Supply,” in The Oxford Handbook of Engineering and Technology in the Classical World, ed. J. P. Oleson, New York: Oxford University Press, 2008, pp. 285-318; Crow et al., The Water Supply of Byzantine Constantinople.
10 P. Bono, R. Bayliss and J. Crow, “The Water Supply of Constantinople: Archaeology and Hydrogeology of an Early Medieval City,” Environmental Geology, , vol. 40 (2001), pp. 1325-1333.
11 P. Forchheimer and J. Strzygowski, Die Byzantinischen Wasserbehälter von Konstantinopel: Beiträge zur Baukunst und zur Topographie von Konstantinopel, Byzantinische Baudenkmäler 2, Vienna: Verlag der Mechitharisten-Congregation, 1893, pp. 56-57.
12 Forchheimer and Strzygowski, Die byzantinischen Wasserbehälter von Konstantinopel, pp. 202-204.
13 G. Wiplinger, “Die Wasserversorgung von Ephesos in Byzantinischer Zeit,” in Byzanz: Das Römerreich im Mittelalter, ed. F. Daim and J. Drauschke, Mainz: Römisch-Germanisches Zentralmuseum, 2011, pp. 593-613, fig. 18.
14 J. R. Snyder, “A Study of Material Technology, Sourcing, and Building Organisation in the Construction of the Water Supply of Constantinople and Anastasian Wall” (PhD thesis), Edinburgh University, 2012.
15 Crow et al., The Water Supply of Byzantine Constantinople, p. 234.
16 L. Brubaker and J. Haldon, Byzantium in the Iconoclast Era, c. 680–850, Cambridge: Cambridge University Press, 2011, pp. 161-162.
17 Theophanes, Chron. AM 6258.
18 Crow et al., The Water Supply of Byzantine Constantinople.
19 Quoted in Crow et al., The Water Supply of Byzantine Constantinople, p. 237.
20 Crow et al., The Water Supply of Byzantine Constantinople, pp. 91-92, figs 4.2, 4.5.
21 Crow et al., The Water Supply of Byzantine Constantinople, pp. 21-23.
22 Theodore Metochites, Byzantios (Vind. gr. 95), fol. 267 v., tr. P. Magdalino