Streets and Public Spaces in Constantinople

Author(s): Albrecht Berger

Source: Dumbarton Oaks Papers, Vol. 54 (2000), pp. 161-172
Published by: Dumbarton Oaks, Trustees for Harvard University
Stable URL: http://www.jstor.org/stable/1291837
Accessed: 17/03/2010 02:57

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=doaks.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Dumbarton Oaks, Trustees for Harvard University is collaborating with JSTOR to digitize, preserve and
extend access to Dumbarton Oaks Papers.

http://www.jstor.org
 Streets and Public Spaces in Constantinople
ALBRECHT BERGER

his paper will mainly address the question of whether Constantinople was planned
and built with a regular street layout. The existence of such a layout has frequently
been doubted. In fact, the surviving evidence is very scanty. We have only the trace of a
single street, the Mese, longer than some hundred meters. Moreover, we should not for-
get that for a long time scholarship about Byzantine Constantinople was influenced by
the appearance of Turkish Istanbul. The Islamic street system, with its irregular blind
and winding alleys, was considered to be the symptom of a decadent culture, and since
Byzantine culture itself was widely thought to be decadent, the Islamic layout, projected
backward in time, served as a proof for the inferiority of Byzantine town planning com-
pared to that of ancient Greece. In GriechischeStddteanlagen,published in 1924, Armin
von Gerkan wrote the famous words:
Constantinople was not a Greek city any more, it ceased to be a Roman one, and did
not become a medieval one. It was an accumulation of imperial license without organic
development, which was upset more than once by the excessive growth of the popula-
tion, a pattern without any structure, without roots and without the possibility of inner
development. Despite its unique situation in the world, it wasted away because of the
basic evil of the Byzantine Empire, it died without ever having lived, and it continued to
die on through the Turkish period until the present day.

Von Gerkan ends by declaring Constantinople a "densely settled landscape" rather than
a real city.1
Since most Byzantine street alignments are unknown, the only way to ascertain a
possible street system is to check such surviving structures as churches, cisterns, the aque-
duct, and city walls. If a number of monuments lie along an imaginary line and are
oriented in the same way, a street may have led along them. However, this can serve as a
basis for reconstruction only if the monuments are sufficiently distant from each other,
since such alignments, where we can still see them, tend to be rather inexact. The
churches of Hagia Sophia and Eirene illustrate this difficulty. The old street that was
detected in 1935 during excavations in front of Hagia Sophia2 runs exactly perpendicu-
lar to the main axis of the building. Hagia Eirene, however, is turned a little more to the

Parts of this paper correspond to a more detailed study, "Regionen und StraBen im fruhen Konstanti-
nopel," IstMitt 47 (1997): 349-414.
'A. von Gerkan, GriechischeStddteanlagen(Berlin-Leipzig, 1924), 168.
2A. M. Schneider, Die Grabungim WesthofderSophienkirchezu Istanbul (Berlin, 1941), 3 f.
162 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

south, so that between its narthex and the street a slightly irregular wing had to be
inserted to fill the space. If this wing had not been preserved, the difference in orienta-
tion, which is only a few degrees, would have distorted our attempt to establish the road
path. Nevertheless, by bringing together all available information, it is possible to get a
fairly clear image of early Byzantine town planning in Constantinople.
Nine years after the publication of von Gerkan's book, Knut Olof Dalman's study on
the Valens Aqueduct, which appeared posthumously in 1933, made some more positive
remarks about the street system of Byzantine Constantinople. He wrote: "In the town
plan of present-day Constantinople, three zones can still be clearly identified that evi-
dence discernible differences in their street layouts. The first corresponds to the old town
within the Severan wall. In it no consistent street network can be recognized. The second
zone is Constantine's town, whose systematic layout can still be made clear... In the
third zone, which extends to the walls of Theodosius II, we again miss the continuous
axes of a uniform system."3 This concept of dividing Constantinople into the old, the
Constantinian, and the Theodosian town is a useful base for further investigation. How-
ever, in later remarks Dalman was sometimes led astray by the modern street system of
Istanbul, which was developing as he was writing, and he was certainly wrong when he
assumed that there was no consistent street layout in the old town.4
It is clear that when the areas of the Constantinian and Theodosian enlargements
were incorporated into the city, a number of overland roads already existed there. The
most important of these is the Mese; actually, it formed the eastern end of the old Roman
road that extended the Via Egnatia through the Balkan peninsula, ending at the gate of
old Byzantium, and was subsequently provided with porticoes on both sides. Another
road branched off to the northwest at the place where the Capitol was built under Con-
stantine.5 It is usually referred to in modern literature as the northern branch of the
Mese, but wrongly, since-as it seems-the name Mese, which means "middle street,"
was only applied to the part east of the Capitol, where the road actually lay in the middle
of the peninsula.6
It is also probable that the two coast roads along the shores of the Sea of Marmara
and the Golden Horn go back to the time before the refoundation.7 These streets will
not be treated here, but it should be mentioned that they often crossed the street grids
that were established later, following the lines of topography, and that they retained
greater importance as continuous axes than any of the newly established roads. All public
places of Constantinople outside the old town, for example, were built along the Mese
or along the southwestern or northwestern streets.

3K. 0. Dalman, Der Valens-Aquddukt in Konstantinopel(Bamberg, 1933), 53.

4E.g., regarding the hypothesis that the Roman overland road to the northwest followed the modern
?ehzadebali Caddesi.
5See C. Mango, Le developpementurbain de Constantinople(IVe-VIIesiecles),TM, Monographies 2, 2nd ed.
InoticK a B-ucavtva 8 (Bonn,
(Paris, 1990), plans; A. Berger, Untersuchungenzu den Patria Konstantinupoleos,
1988), 330 f.
6Cf. Theophanes, Chronographia,ed. C. de Boor, 2 vols. (Leipzig, 1883-85; repr. Hildesheim, 1963), 102.9,
369.29, 442.11, and ChroniconPaschale, ed. L. Dindorf, CSHB (Bonn, 1832), 1:693.22; De cerimoniisaulae
byzantinae,ed. J. J. Reiske, CSHB (Bonn, 1829), 439.3 f; see also A. Berger, "Die Senate von Konstantinopel,"
Boreas 18 (1995): 136 n. 35.
7Mentioned in nHarptaKovwavnvoTavTivo6Xo, T.
ed. Preger
1.68 in ScriptoresoriginumConstantinopolitanarum,
(Leipzig, 1907), 2:148; cf. Berger, Untersuchungen,226 f.
 ALBRECHT BERGER 163

Another problem is posed by the old retaining walls, which divided most slopes into
terraces and do not conform to the reconstructed street system as proposed below. Ernest
Mamboury and Raymond Janin believed tha thtese walls were mostly of Byzantine origin
and that where their construction appeared to be Ottoman, it was the result of later
repairs.8 Many of these walls are now lost as a result of modern building activity; however,
at the only place where major parts do survive, in the area around the former Old Sera-
glio, it is clear that they belong to its substructures and have no Byzantine predecessors.
Let us now turn to the street layout of Byzantium and early Constantinople. In the
old town of Byzantium, a street grid can be recognized quite easily (Fig. 1). It took into
account the shape of the major hills in such a way that streets ran either along slopes or
perpendicular to them, so that they were either reasonably flat or very steep and
equipped with stairs. For this reason, the streets were not always exactly parallel.
Street E ran along the northwest side of the Hippodrome and beside Hagia Sophia
and Hagia Eirene to the Acropolis. The exact line of the Acropolis walls is unknown, but
it is probable that the walls of the inner courts of the Topkapi Sarayi use them as founda-
tions.9 The only gate in these walls that leads to the northeast is the so-called Fil Kapisi,
the "Door of the Elephants." It lies in the line of the supposed Street E and may well
have had a predecessor in the Acropolis walls that was used as an exit leading to the Gate
of St. Barbara at the Seraglio Point. Street E existed at least until the tenth century10 and
may be the one called Pelargoi, or "the Storks."1'
We can guess from the position and orientation of the palace hall, now the Mosaic
Museum,12 that a parallel street existed some 120 meters to the southeast. This Street F
ran along the atrium of the palace hall, then along the east side of Hagia Sophia, ending,
perhaps after a small turn to the west, on the cropolis. It is interesting to note that in
the Byzantine period it was impossible to leave the Acropolis by using this street: in 1264,
when the patriarch Arsenios was sent into exile, he left the Great Church on the east
side, but then apparently used the street called E to reach the Gate of St. Barbara, where
he embarked upon the boat that took him to Asia Minor.13 The terraces of the old emper-
or's palace extended behind Street F to the southeast, and another Street G may be
assumed below these terraces leading to the Mangana gate. The hexagonal church
wrongly believed to have been that of the Hodegetria'4 did not lie on this street but a
little higher on the hill and had a slightly different orientation.
If we now go to the northwest, we find Street D along the west side of the Hippo-
drome.15 Since the wall of the Hippodrome turns a little bit to the east shortly before the
palace of Antiochos, the street does the same. A curved wall on the right side seen from

8R. Janin, Constantinoplebyzantine:Developpementurbain et repertoiretopographique,2nd ed. (Paris, 1964), 7 f

and enclosure VI.
9On the Acropolis in the Byzantine period, see H. Tezcan,
TopkapiSarayzve cevresininBizans devri arkeolojisi
(Istanbul, 1989).
'OH. Delehaye, Les saints stylites(Brussels, 1923), 218.
"Janin, Constantinople,405; Berger, Untersuchungen,401 f.
'2W.Jobst, "Der Kaiserpalast von Konstantinopel und seine Mosaike,"Antike Welt 18.3 (1987): 2-22.
'3George Pachymeres, Relationshistoriques,ed. A. Failler (Paris, 1984), 2:353 f.
14R. Demangel and E. Mamboury, Le quartierdesManganeset la
premiereregionde Constantinople(Paris, 1939),
93-111; on the location of Hodegetria, see Berger, Untersuchungen,378.
'5Cf. Berger, Untersuchungen,125, 159; K. Ciggaar, "Une description de Constantinople traduite par un
pelerin anglais," REB 34 (1976): 256 f.
164 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

the starting boxes is a standard feature of late Roman hippodromes.16 The change of
direction of Street D has no further implications, because it must have ended at the Mese.
However, the orientation of the Basilica and the Chalkoprateia church are known from
their existing remains, and some traces of a colonnaded street leading along their west
side were actually found in 1929.17 It seems that the changed direction of Street D was
adopted in Street C and in the remaining parallel streets B and A to the northwest. If
this is not a simple coincidence, it means that the street layout here is later than the
construction of the Hippodrome-and that is hard to believe, even if we accept that the
Hippodrome was actually founded by Severus.
The street labeled B on the sketch is interpolated, since no trace of it exists. However,
it may have touched the apse of the church whose remains, including some floor mosaics,
have been found there.18 Street A is defined by the corner of a building that was found
in 1964 close to the old walls of Byzantium.19
If we now pass to the streets that ran roughly at right angles to those described above,
we see that the corner of the building found in 1964 also establishes the direction of the
street labeled 4 on the sketch. Streets 1 to 3, which lie to the north of it, can be traced
with some greater certainty. Street 1 still exists in parts, leading through the commercial
region of Tahtakale and the southern wing of the Egyptian Bazaar. Its great age is clear
from the fact that it runs parallel to Valens' aqueduct and meets the northwestern wall
of the Acropolis at a right angle. It also runs along the Balkapan Hani in Tahtakale, an
Ottoman building on an extant Byzantine substructure from the sixth century.20Street 2
can be fixed by the early Byzantine substructure usually called the palace of Botanei-
ates,21by the remains of the church with the mosaic floor mentioned above, and by the
Chalkoprateia church. It may have turned slightly at its east end in order to conform
with streets E, F, and G, and it may have run along the north side of Hagia Sophia. How
it ended in the northwest is not clear; at any rate, it did not go beyond the predecessor
of Balkapan Hani after the sixth century, because it would have had to run across the
building.
The beginning of Street 3 can be approximately fixed by the southern side of the
Balkapan Hani. It ran straight up to the Basilica, before which it may have turned a little
in order to reach the center of the building's west side.

6My Iconclusions drawn from this curved wall in "Die Alstadt von Byzanz in der vorjustinianischen Zeit,"
Varia 2 = otciKXa BtSavnva 6 (Bonn, 1987), 8-10, cannot be sustained, as was pointed out by Mango,
Developpementurbain, 71.
17M. Schede, "Archaologische Funde," AA (1929): 358; A. M. Schneider, Byzanz (Berlin, 1936), 92, nos. 8,
9. On the Basilica and the Chalkoprateia church, see T. Mathews, TheEarly Churchesof Constantinople: Architec-
tureand Liturgy(University Park, Pa., 1971), 28-33.
sW. Kleiss, Topographisch-archaologischer Plan von Istanbul(Tubingen, 1965), 8, no. 35; R. M. Harrison and
G. R. J. Lawson, "The Mosaics in Front of the Vilayet Building in Istanbul," IstanbulArkeolojiMiizesiY1llzg 13/
14 (1967): 216 f; and N. Firatli, "Recent Important Finds in Istanbul," ibid., 15/16 (1969): 193 f; cf. also
G. Hellenkemper-Salies, RBK 4:615 f.
'9Kleiss, Plan, 12, no. 112; a sketch of the site can be found in the archive of the German Archaeological
Institute in Istanbul; see also N. Firatli, "Brief Archaeological News," IstanbulArkeolojiMiizesi Yzllzgz13/14
(1967): 226 f and pi. 63.
20C. Giiren, TurkHanlarznznGelisimive IstanbulHanlarzMimarisi (Istanbul, 1985), 85; and A. Berger, "Zur
Topographie der Ufergegend am Goldenen Horn in der byzantinischen Zeit,"IstMitt45 (1995): 162.
210n this, see W. Muller-Wiener, Bildlexikonzur Topographie Istanbuls(Tiibingen, 1977), 41; Berger, "Ufer-
gegend," 162.
 Zindan Kapi

Taurus
-
forumof
Constantine D/

land walls palace of

Antiochus e
of Byzantium

fl 1 '
0 5100000 m

1 The old town of Byzantiumand the new quarternext to the Golden Horn (c = cistern)
2 Streetsaroundthe churchof the Holy Apostles,orientedon the north-
western overlandroad. The long, solid blackline is the aqueduct.The
shortblackline is the foundationwall.

3 Streetsaroundthe churchof the Holy Apostles,orientedon the

aqueduct(the long, solid blackline). The shortblackline is the
foundationwall.
 Kapi

Kapi

Odun
L v__ O.d,
i - . .Zindan

Marcian
landwallsof
Constantine

forumof
Arcadius

500o 1000m

4 Streetsin the Constantiniantown (c = cistern)

 ALBRECHT BERGER 165

Our greatest problem in reconstructing these three streets is that they appear to radi-
ate from a single point, the so-called Odun Kapi, which in Byzantine times was called the
Gate of the Droungarios.22 Such a radial street design would be quite unusual, unattested
elsewhere at that time. The Odun Kapi also lies outside of the old walls of Byzantium,
and if these streets leading into the old town started from it, a major change in street
design must have occurred there after the Constantinian refoundation of the city.
The only major street that does not fit into this more or less rectangular pattern is
the main street that led from the western gate to the center of the city, close to the Hippo-
drome, and that, as our sources claim, already existed before Constantine. Was this street
cut at an angle through an existing rectangular grid, or did it survive from an older
phase? It is clear that the ancient center of Byzantium lay on the Golden Horn, at the
foot of the hills, and that the southern part of the town was not densely settled. In other
words, the Hippodrome and the adjoining palace were perhaps established in an area
that had previously been more or less empty. The old road to Thrace already ran on top
of the hills along the line of the later Mese, as we can see from the Roman cemeteries
along it. It is obvious that a connecting street from the gate to the old center around the
harbor must have existed, and this may well have been Street B posited above. The main
square of Byzantium, the so-called Strategion, was probably located somewhere at its
lower end.23 When the urban center shifted uphill to the Hippodrome, the Mese had to
be established. Therefore, the date of this shift depends on the date of the Hippodrome
and adjoining constructions. The Hippodrome is usually ascribed to Severus, but his role
as rebuilder of Byzantium seems in great part to be legendary.24We know that the defen-
sive walls were probably not rebuilt before 260 or 270,25 and the construction of the
Hippodrome certainly began later, perhaps not much before Constantine's time, when
Licinius was master of the city.
The Augustaion, a large public square in front of Hagia Sophia, is usually believed
to have corresponded to an entire block of the street grid. However, Rudolf Stichel will
demonstrate in a forthcoming paper that it was much smaller and comprised only the
western half of the block.
Another problem that certainly arose from the shift uphill was lack of a continuous
street from the new center to the harbor region. The foundations that have been detected
around the Basilica, the Chalkoprateia church, and Hagia Sophia were all aligned with
these buildings. In order to reach the Strategion from Hagia Sophia, the street along the
Chalkoprateia had to be followed up to the intersection with the former main artery,
Street B, where the church with the mosaic was situated.26 When Hagia Sophia was re-
built and enlarged in Justinian's time, the new atrium was extended over Street E, with
the result that all traffic from the harbor to the city center would have had to cross it. It

22Berger, "Ufergegend," 159.

230n this, see Janin, Constantinople,13, 431 f; Berger, Untersuchungen,406-11. See also Zosimus, Nea
Historia, 2.31; Pseudo-Hesychius in ScriptoresoriginumConstantinopolitanarum, ed. T. Preger (Leipzig, 1901),
1.16.
24G. Dagron, Naissanced'une capitale (Paris, 1974), 15 f; Mango, Developpementurbain, 19.
25Mango, Developpementurbain, 15. The historian Herodianus seems to have seen the city in a ruinous
state around 240 (Historiai 3.1.7).
26It may be identified with the Church of the Mother of God of Urbicius, on which see R. Janin, La
geographieecclesiastiquede l'Empirebyzantin,vol. 1, Le siege de Constantinopleet le patriarcatcecumenique,pt. 3, Les
egliseset les monasteres,2nd ed. (Paris, 1969), 207; Berger, Untersuchungen,404 f.
166 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

seems that to avoid this, a new street was later inserted into the system, leading from a
point below the Chalkoprateia church directly to the Milion.27
When the Mese was built, some adjoining areas were apparently restructured. The
upper end of the old main street B was abandoned, and another street, labeled H, was
introduced. Its alignment can be determined from three cisterns in this area: the cistern
called Binbirdirek south of the Mese,28 the 5erefiye cistern under the current town hall
of Eminonuii,and finally the great cistern to the north of the Mese, of which little has re-
mained.29
Let us summarize what we have learned so far about old Byzantium. After the foun-
dation of Constantinople in 324, a vast area had to be built up and incorporated into the
fabric of the city. It seems that a consistent plan for the whole town was not made immedi-
ately but that as a first stage only two new, separate quarters were established, one on
the hills overlooking the Golden Horn, just outside the old walls, the other around the
mausoleum of Constantine at the church of the Holy Apostles.
Of the first of these areas, the western part was incorporated into the Old Serail
shortly after the Ottoman conquest, whereas the eastern part was nearly completely cov-
ered with the constructions of the Great Bazaar. The only surviving Byzantine monument
here is the so-called Tower of Eirene, a building of middle Byzantine date.30It is oriented
exactly north-south and not to the northeast, as we would expect if the street system of
the old town had been extended into this area. The same is apparently true for a number
of cisterns in the vicinity that were published by Josef Strzygowski in 1893 but that can
no longer be found today.31 In this region a street named Uzunaarli, literally "Long
Market,"preserves the location and name of a famous colonnaded street of the Byzantine
era, the Makros Embolos, or Long Portico. This street, too, runs from south to north at
a right angle to the Mese and turns slightly east at its lower end.32
Another street running south-north can be found at the eastern end of the aqueduct.
It certainly still follows the same line it had in the early Byzantine period, since on its
west side a group of churches was found. One of them has recently been identified with
the church of the Mother of God of the Diaconissa, inaugurated in 596.33 The church
was separated from the street by a courtyard and retaining wall.

27A possible reference to it can be found in the Miraclesof St. Photeine, see F. Halkin, ed., Hagiographica
ineditadecem(Turnhout-Leuven, 1989), 117 f; A.-M. Talbot, "The Posthumous Miracles of St. Photeine," AB
112 (1994): 93 f. The approximate course of this street is shown on Fig. 1 with dotted lines.
28Itis usually but wrongly identified with the cistern of Philoxenos, on which cf. J. Bardill, "The Palace of
Lausus and Nearby Monuments in Constantinople," AJA 101 (1997): 69-83. It may have belonged to the
palace of Illus that lay close to the Hippodrome and the palace of Antiochus; on the church at this palace,
Janin, tglises, 416; Berger, Untersuchungen,560-62; on the cistern, P. Forchheimer and J. Strzygowski, Die
byzantinischenWasserbehalter von Konstantinopel(Vienna, 1893), 180 f.
29This was probably the real Philoxenos cistern, since it was closely connected to the palace of Lausos,
which has to be sought here rather than on the south side of the Mese, cf. Bardill, "Palace of Lausus," 83-89.
30Cf.Berger, "Ufergegend," 158.
31Cf. Forchheimer and Strzygowski, Wasserbehalter, nos. 21, 25, 30, 37.
32The Makros Embolos began at the Bronze Tetrapylon, see Berger, Untersuchungen,315, and ended at
the Zindan Kapl, which is sometimes called the gate of St. Anastasia, see A. M. Schneider, "Mauern und Tore
am Goldenen Horn zu Konstantinopel," Nachrichtender Akademieder Wissenschaftenin Gdttingen,philosophisch-
historischeKlasse 5 (1950): 86 n. 53; cf. Berger, Untersuchungen,446.
33N.Flrath, "Decouverte de trois eglises byzantines a Istanbul" CahArch5 (1951): 163-78. On the identity
with the Diaconissa church, cf. J. Bardill, "ACatalogue of Stamped Bricks in the Ayasofya Collection," Anatol-
ian Archaeology:Reportson ResearchConductedin Turkey1 (1995): 28 f.
 ALBRECHT BERGER 167

A former inhabitant of the area told me that around 1950, when he was still a boy,
he and some friends found a tunnel that led to the north from the Sekbanbap1Yakup
Aga Camii. This small mosque lies just on the south side of the Mese, opposite the point
where the old street began, and we can assume that this tunnel was an old waste-water
channel.
My informant told me that he entered againtethe tunnel and left it somewhere in the
vicinity of the Siileymaniye mosque. Unfortunately the place where the tunnel began is
now covered with new constructions, so there is no possibility of checking the story. We
know that the old Ottoman palace, in its first stage, comprised a vast area from the
Makros Embolos in the east to the street at the Diaconissa church in the west and from
the Mese in the south up to the beginning of the steep slope behind the mosque of
Suiileymanin the north.34 This area was subsequently reduced; the southern part was
detached already in the time of Mehmet the Conqueror's son Bayezid and served as a
building ground for his mosque, medrese, and bath; later, the mosque of Suleyman the
Great was built in the northern part. There is reason to believe that the eastern and
western boundaries simply followed old streets that had existed there since the early
Byzantine period. Because the distance between the old walls of Byzantium and the
Makros Embolos is about two-thirds that between the Makros Embolos and the Dia-
conissa street, we can assume that, in the first phase of expansion beyond the old walls,
five parallel streets were built in this part of the city, separated from each other by about
180 meters.
The quarter outside the old walls later belonged to Regions VI and VII, and it seems
that the parallel streets are mentioned where Region VII is described in the Notitia urbis
Constantinopolitanae:"Compared to the one before, the seventh region is flatter, although
it must be considered too as declining to the sea at the end of its one side [that is to the
north]. It extends from the right side of the Column of Constantine up to the Forum of
Theodosius with continuous porticoes, and with other parallel streets it declines at one
side to the sea and is thus leading down."35
There is no evidence that the parallel streets of this first expansion phase went so far
to the north that they reached the Golden Horn. The reason may be that the shore was
perhaps already settled before Constantine included it inside the city walls, and new
street planning could not be applied there. In the western part, moreover, a sharp de-
scent between the new quarter on the hilltop and the older one on the shore made the
building of continuous streets a difficult task.
Since the last century, the former extent of the Forum of Theodosius, the so-called
Taurus, has often been equated with the whole area of the old Ottoman Serail. But such
a forum, measuring 300 by 450 meters and possibly with a vast exedra on the northern
side, would have been out of proportion. Among other difficulties, it could not even have
had a completely level surface. The estimation can hardly be correct. According to the
Notitia Region VII, approximately two-thirds of which would be covered by so-large a
forum, was actually the most densely settled region in Constantinople, containing 711
houses. Over the years, the supposed size of the forum has gradually been reduced,

34E. :elebi, Seyahatnamesi,ed. R. E. Kocu (Istanbul, 1955), 28 f; see most recently A. Berger, "Zur sogen-
annten Stadtansicht des Vavassore,"IstMitt 44 (1994): 342 f.
35Notitiaurbis Constantinopolitanae,in Notitia dignitatum,ed. 0. Seeck (Berlin, 1876), 235.
168 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

reaching about 120 by 100 meters in Cyril Mango's most recent publications.36 But the
reality, I think, is different still. The only surviving part of the forum is the magnificent
Arch of Theodosius, which is usually believed to have been the forum's western en-
trance.37 Close to this arch a number of foundation walls have been found that prove
that it did not form the western, but rather the eastern entrance of an area 55 by 55
meters, with an exedra to the north. Thus, the Taurus was so much smaller than had
been estimated that even the excavators did not realize they had actually found it.38
The Taurus is the only public space of Constantinople of which material remains
exist. I suggest that there are good reasons to reduce proportionally all other fora as
well. The fora of Constantine and Arcadius both lay on the edge of a sharp slope to the
south, and if we accept that the monumental columns that still exist, whole or in part,
were located in the middle of the squares, the size of these fora cannot have exceeded
that of the Forum of Theodosius. The small size of fora in Constantinople may be ex-
plained by their having had only a ceremonial function, since their political role had
been taken over by the Hippodrome.
A second area in Constantine's new town where the layout of streets must go back to
the first years after the refoundation is that around the church of the Holy Apostles.
Unfortunately, there are no remains of the first church or of its sixth-century successor
that would allow us to define its exact location and orientation. However, a look at the
aqueduct may be helpful in this regard. Today it is usually called the Aqueduct of Valens,
since it was finished in 368, during Valens's reign, but there is reason to assume that it
was already planned and begun in Constantine's time.39As mentioned above, the aque-
duct runs parallel to one of the streets in the old part of Byzantium. Also, its southeastern
prolongation would exactly meet the main entrance of the courtyard in the Great Palace
that is now the Mosaic Museum. It is obvious that the aqueduct was planned in a clear
relationship to the street system of the old town of Byzantium.
Arches 26/27 and 52 are wider than the others in the aqueduct and were certainly
intended to serve as passages for streets.40At other points where we would expect similar
wider arches, the original construction is lost, for example, at the northwestern end close
to the church of the Holy Apostles, where the aqueduct was completely rebuilt in Otto-
man times.
Another piece of evidence is provided by the old overland road to the northwest. If
we assume that it branched off from the Mese at the Capitol, then passed along the Aetios
cistern in the Theodosian part of the city and met the walls at the Gate of Charisios, the
modern Edirne Kapi, it must have followed a straight course that was parallel not to the
aqueduct but rather to the large court of the Fatih mosque, which replaced the church
of the Holy Apostles in the Ottoman period.41
In attempting to reconstruct the Byzantine streets in this area, we therefore have two

36Mango, Developpementurbain, plans.

37A reconstruction by R. Naumann, "Neue Beobachtungen am Theodosiosbogen und Forum Tauri in
Istanbul" IstMitt 26 (1976): 117-41.
38See A. Berger, "Tauros e Sigma: Due piazze di Constantinopoli," in Bisanzioe l'Occidente:arte, archeologia,
storia. Studi in onoredi Fernandade'Maffei(Rome, 1996), 19-24.
39Mango, Developpementurbain, 20.
40Dalman, Valens-Aquadukt, 45.
41See Berger, Untersuchungen,330 f.
 ALBRECHT BERGER 169

different options. The first is to assume that the Fatih mosque stands on the foundations
of the church of the Holy Apostles and was oriented along the northwestern street (Fig.
2). The streets through the aqueduct would then have crossed at an oblique angle.42This
reconstruction, with parallel streets perpendicular to the old overland road is, however,
contraindicated by the orientation of the few Byzantine remains that still exist there. One
of these is a strong foundation wall south of the aqueduct and parallel to it, found in
1964 when an underground pedestrian passage way was built.43Also, we know of four
cisterns in this area. One is of early Byzantine date and lies under the courtyard of the
Fatih mosque,44 the other three are middle Byzantine and incorporate two substructures
of churches now lost.45 All are oriented more or less parallel to the aqueduct. It seems
reasonable, therefore, to abandon the traditional idea that the Fatih mosque replaced
the church of the Holy Apostles on the same spot. This hypothesis was introduced by
Karl Wulzingeriner 1932 and demonstrated in a quite convincing way,4 but there are a
number of objections to it. For example, if the mosque had replaced the church exactly,
the grave of Sultan Mehmet would have taken the place of Constantine's mausoleum.
This would have been an act of enormous symbolic meaning, but is impossible, since the
mausoleum must have been considerably larger than Mehmet's grave and would have
interrupted the church's eastern cross arm. The probable size of Constantine's mauso-
leum can be deduced from such similar surviving fourth-century structures as St. Cos-
tanza in Rome or the Holy Sepulcher rotunda in Jerusalem. It has also been assumed
that the mosque's courtyard was Byzantine,47 but we know that the atrium of the church
of the Holy Apostles was not that large. We should accept, therefore, that what has been
assumed as the physical and ideological replacement of the church of the Holy Apostles
by the Fatih mosque is a modern projection backward in time. The church of the Holy
Apostles was razed by the Ottomans and no part of it reused. We should reconstruct the
church as being oriented in the same direction as the aqueduct, positioning it closer
to the aqueduct than to the present building.48 In fact, the church was located by a
sixteenth-century source at the Imaret of the Fatih mosque.49
This reconstruction yields our second option for the street layout (Fig. 3), in which
only the preexisting overland road does not fit into the rectangular scheme. In the grid
as shown here, the foundation wall lies along a street, the aqueduct lies in the middle of
a number of insulae between the front and back lots, and the cisterns fit into the system

42The only reconstruction published until now, that of M. Restle, ReclamsKunstfuhrerIstanbul (Stuttgart,
1976), 256, in reality reproduces the course of the present streets, which were built in the 20th century.
43Unpublished report by W. Kleiss, archive of the German Archaeological Institute Istanbul. Restle's re-
construction ignores both cisterns and the wall.
44B. Kunter and S. Ulgen, Fatih Camiive Bizans Sarnzci(Istanbul, 1939), 16, fig. 16, 71 f.
45Forchheimer and Strzygowski, Wasserbehalter, 69, 81; and W. Muller-Wiener, "Zur Lage der Allerheili-
genkirche in Konstantinopel," in LebendigeAltertumswissenschaft. FestschriftH. Vetters(Baden bei Wien, 1985),
333-35. The first church substructure may have been that of the church of the monastery of Christ Phi-
lanthropos; cf. A. Berger, "Die mittelbyzantinische Kirche bei der Mehmet Fatih Camii in Istanbul," IstMitt
47 (1997): 455-60, the second one that of the church of Theophano; on this church, Janin, Eglises, 245,
389 f; Berger, Untersuchungen,499-501.
46K. Wulzinger, "Die Apostelkirche und die Mehmedije zu Konstantinopel," Byzantion7 (1932): 1-39.
47M. Restle, Kunstfihrer,256.
48As shown by Miiller and Wiener, Bildlexikon,21, 25, 27; Berger, Untersuchungen,520.
49Mango, Developpementurbain, 27 n. 27.
170 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

quite logically. Nonetheless, it must be admitted that the position and groundplan of
the church of the Holy Apostles proposed here is pure conjecture. Since the streets of the
old town of Byzantium and those around the church of the Holy Apostles follow the
same orientation, at some point a regular street system for all Constantinople might have
been taken into consideration, but the north-south alignment of the streets in the area
of the present bazaar shows that these plans were never actually executed.
Moreover, if this street grid around the church of the Holy Apostles did ever exist, it
was probably abandoned only a short time later. It is true that the major streets through
the aqueduct crossed it at regular intervals for a long time, until the twentieth century,
in fact, when the streets of Istanbul had to be adapted for automobile traffic.50These
streets, however, did not use the broader arches mentioned above (except in the case of
arch 52, which may be coincidence). It may seem surprising that the city streets changed
course completely, while the aqueduct passages remained in place. This can be explained
by the fact that most parts of the aqueduct were not accessible, but were covered by
houses and even churches attached to it. For example, the north church of Kalender-
hane, already in the late sixth century used some arches of the aqueduct as its northern
nave.51
If we assume that the streets through the aqueduct passages ran in a direction deviat-
ing 30 degrees from north (Fig. 4), two of them would meet gates of the sea wall on the
Golden Horn, the Cibali Kapi and the Unkapanl Kapi, respectively.52The streets would
ann
also touch a number of early and middle Byzantine nu
monuments and buildings in this
region, among them the Column of Marcian, the monasteries of Christ Pantepoptes and
Pantokrator, and the church of St. Polyeuktos. The annexed baptistery of this church
even had an oblique western wall along one of the streets.53Furthertherto e east, another
street runs alongside a modern Greek chapel and the now-disappeared, so-called grave
of the last emperor Constantine54-both structures that probably replace Byzantine
churches.55 Another street would lead exactly to the southern entrance of the Vefa Kilise
Camii, the former church of St. Procopius.56 There is, however, no indication that this
street system was extended further to the east, where the first street-grid of Constanti-
nople already existed.
The fact that the streets lead to gates of the sea walls gives us a terminusante quem
for this phase, namely the construction of these walls in 438/439.57 Now, let us assume

50Cf. the map from 1882: 19. Asirda IstanbulHaritasz,ed. E. H. Ayverdi (Istanbul, 1958), sheet C4.
51Cf."North Church," in Kalenderhanein Istanbul, ed. C. L. Striker and Y. D. Kuban (Mainz, 1997), 37-45.
520n these see Schneider, "Mauern," 77; he identifies the Cibali Kapi with the porta Putea or porta al Pozzo
mentioned by Italian sources, but this was probably the Ayakapi, see Berger, "Ufergegend," 152. As the
Byzantine names of most gates of the sea walls are not known with certainty, the Turkish names will be used
here; see Schneider, "Mauern," 65-107, and Berger, "Ufergegend," 149-65.
53R. M. Harrison, Excavationsat Saraqhanein Istanbul (Princeton-Washington, D.C., 1986), 1:16, 24.
540n this see D. M. Nicol, The ImmortalEmperor:The Life and Legendof ConstantinePalaiologos,Last Emperor
of Romans(Cambridge, 1992), 92 f. The grave was not venerated before the 19th century and disappeared
the
around 1960, when a new street layout was introduced.
55Asafe identification seems impossible. Perhaps the church of St. Stephanus of the Konstantianai lay at
one of these two places. This church still existed in 1453; see Janin, Eglises, 474 f; Berger, Untersuchungen,
471 f.
56Berger, Untersuchungen,463.
57Chron.Pasch., ed. Dindorf, 583.3 f. Whereas the sea walls around the old town of Byzantium were built
in the Roman era (cf. C. Barsanti, "Note archeologiche su Bisanzio romana," Milion 2 [1990]: 14 f), there is
 ALBRECHT BERGER 171

that the same street system also existed on the southern side of the city facing the Sea
of Marmara. We see that streets would lead to all five gates in the sea walls in the area
of the Constantinian town. These are, however, not prolongations of the streets recon-
structed in the northern part of the city but lay between them. This is quite logical,
because the distance between the reconstructed northern streets is about 220 meters, far
too great for a reasonable sized insulae, such as we know from other late antique towns.58
So, the distance between the parallel streets in this part of Constantinople was actually
about 110 meters. In Figure 4, for the sake of clarity, not all of them are indicated.
Of the gates in the Marmara sea walls only the Kumkapi, the Iron Gate of the Byzan-
tines, is known by its old name.59 The position of three other gates can be deduced with
reasonable certainty from old maps. The first gate west of Kumkapi is almost completely
invisible today and unknown in the scholarly literature, since its outer face is hidden
behind a new Ottoman revettment, and the street level has risen so high that inside only
the top of the supporting arch over it can be seen.60 The gate must have been that close
to ta Kanikleiou,from which the street went up to the church of Homonoia/Panteleemon,
mentioned by the Bookof Ceremoniesin the tenth century.61
The street system cannot be reconstructed completely, because there exist no traces
of streets at right angles to those shown here. These would be necessary to form the usual
insulae, that is, the normal blocks of eight rectangular lots arrangede in r s of four.
n two rows
The orientation deviating 30 degrees from north supports R. Brun's theory that many
monuments in Constantinople were arranged according to a pattern of large triangles,
imitating an older similar pattern in Rome.62 In fact, our reconstructed streets in the
area of the Constantinian town coincide with or run parallel to one side of Brun's cen-
tral triangle.
Between 408 and 413, the Theodosian wall was added further to the west. Properly
speaking, the newly acquired areawasaawas not considered part of the town. It not densely
settled, but contained and
cemeterenies.63
monasteries, gardens, fields, The position of the

no indication that the Constantinian town possessed sea walls before 438/39, as has been assumed, e.g., by
Schneider, "Mauern," 65.
58See A. Bryer and H. Lowry, Continuityand Changein LateByzantineand Early OttomanSociety(Birmingham,
1986), 264. The insulae of Antioch had a size of 112 58 m, in Beroia of 124 X 48 m, in Dura of 70 x 35
m, in Damascus of 100 X 45 m, in Nikaia of 100 X 50 m, in Sinope of 100 x 60 m, in Thessalonike of 100
X 50 m, and in Trebizond of 100 x 60 m. Cf. also W. Hoepfner, "Von Alexandria iiber Pergamon nach
Nikopolis," in AktendesXIII. internationalenKongressesffur klassischeArchdologieBerlin 1988 (Berlin, 1990), 279 f.
59SeeA. van Millingen, ByzantineConstantinople(London, 1899), 248 f; F. Dirimtekin, Fetihden nce Marmara
surlarz(Istanbul, 1953), 32 f.
60It lies opposite the house at Alisan Sokagi no. 121.
61Decerim.,ed. Reiske, 561.3; on ta Kanikleiou,seejanin, Constantinople,365 f; Berger, Untersuchungen,645 f.
62R.Brun, "An Urban Design Imported from Rome to Constantinople-New Rome," in BysansochNorden,
aktafr nordiskaforskarkursen i bysantinskkonstvetenskap
1986 (Stockholm, 1989): 213 f; idem, "ASystem of City
Design in the Ancient World Based on Equilateral Triangles," in InternationalCongressfor the History of Art
Washington1986, WorldArt: Themesof Unity in Diversity,ed. I. Lavin (University Park, Pa., 1989), 199 f; and
elsewhere.
63In the early period, the most prominent buildings there were the Palace of Helena and the villas of
Saturninus and Aurelianus, cf. Janin, Constantinople,317, 355, 422; V. Tiftixoglu, "Die Helenianai nebst
einigen anderen Besitzungen im Vorfeld des friihen Konstantinopel," in Studienzur Friihgeschichte Konstanti-
nopels, ed. H.-G. Beck (Munich, 1973), 79-83; Berger, Untersuchungen,362, 605 f, 629-31. On the mosaic in
Pulcu Sokagi that was found in 1994 and may have belonged to a pre-Constantinian villa, see the unpub-
lished symposium paper of Alessandra Ricci.
172 STREETS AND PUBLIC SPACES IN CONSTANTINOPLE

gates in the sea walls on both the Sea of Marmara and the Golden Horn suggest that
perhaps there were plans to extend the street system into this area but that this did not
take place except close to the seashore.
The urban development of Constantinople was always closely connected to the politi-
cal strength of the whole empire. So it is clear that possible plans for expansion became
unrealistic after the time of Justinian. Constantinople was besieged by the Avars, Slavs,
and Arabs in the seventh century: it lost most of its population, and the remaining inhabi-
tants led a rather rural life within walls that encompassed too large an area.64At this
time, Constantinople may really have become the "densely settled landscape" that von
Gerkan claimed it had been already during its peak prosperity.

Freie Universitat Berlin

64
Mango, Developpementurbain, 53-60.