Ken Dark and Jan Kostenec
The Hagia Sophia Project, Istanbul 2004-8

Introduction
The purpose of this brief report is to present a summary of the results of an archaeological study of the Byzantine church of Hagia Sophia and its immediate environs, undertaken between 2004 and 2008. The project was initiated by Ken Dark (University of Reading) in 2004, and is co-directed by Ken Dark and Jan Kostenec (Charles University, Prague). The work is funded and supported by the Charles University Grant Agency, the Late Antiquity Research Group and the Andrew Mellon Foundation, and takes place with the kind permission of the Turkish authorities and of the director and staff of Ayasofya Müzesi, whose help has been invaluable.

While the Justinianic church of Hagia Sophia has been extensively studied, this project places emphasis not on that impressive structure, but on the associated Byzantine buildings around it. The underlying basis of our work is that the church can only be properly understood as the focus of an ecclesiastical complex rather than a single 'freestanding' structure. With the help of the Museum authorities, we have been allowed to examine the entire surroundings of the Byzantine church and usually inaccessible areas within the building. In recent years, removal of plaster, demolition of later structures, substantial earthmoving and other management activities, have revealed large areas of the site previously unpublished. We have generously been allowed to record and publish this new material, and to re-investigate other areas of the site.

Here, for the sake of brevity, except in headings, initials are used to stand for points of the compass (eg. S = south, N = north), and C followed by a figure for centuries AD (eg. C6 = sixth century). We use ‘Hagia Sophia’ for the whole area within present Ayasofya Müzesi and 'the church' to refer to the Justinianic building, unless otherwise qualified. Figure 1 shows the areas investigated in relation to the church.

Fig. 1 General plan of Hagia Sophia (after Müller-Wiener 1977), showing the areas discussed in the text.

The area south of the church
Previous scholars had dated the centrally planned structure immediately SE of the SW porch of the church to the reign of Justinian I and identified it as a baptistery on the basis of written sources. This identification was seemingly confirmed in the 1940s by the discovery of a marble baptismal font in its N court. However, as we have shown in a recent article in Architectura, neither this date, nor the interpretation of the structure as a purpose-built baptistery, are at all certain. It is more likely that it had originally been a reception room of the Patriarchal palace before any liturgical use and was built in the reign of Justin II1.

Fig. 2 The mosaic in the NE porch of the baptistery

In addition, our survey providedfurtherinformation about the ‘baptistery’.The fenestration of the lower level has beendramatically altered since the Byzantineperiod, when it was better lit by largeropenings evidenced on the S and E sides as well as in the W narthex. The apse is bonded into the main body of the building, proving that its present rectangular shape belongs to the original design2. In the NE porch we recorded unpublished fragments of Byzantine-period decoration. A window in the N tympanum beneath the lunette of the central bay vault of the porch is decorated with mosaic showing a leaves-and-quatrefoil pattern (Figure 2). This was later blocked by a cross-vault and the plaster over this blocking decorated with an acanthus rinceau fresco.

Previous publications erroneously connected the whole N wall of the ‘baptistery’’s porch with the SW outer buttress of the church3, but newly-visible masonry in the interior of the porch shows that the N and S walls of the porch are bonded into the E wall of the 'baptistery' and, therefore, contemporary with it. As the two parallel porch walls running E were built in the same phase as the 'baptistery' (but predate the S-W outer buttress), the buttress must be later than the NE porch. The lateral walls of the porch underwent profound changes in the Byzantine period, when two large round-headed openings were created. At the E end of the N wall of the porch interior, a fragment of an arch suggests another opening. This may imply that the C6 porch continued further to the E, perhaps as a corridor.

Fig. 3 The N porch of the baptistery, with the E pier and barrel vault seen from the NW.

Another area examined in the course of this work was the north porch of the ‘baptistery’. The lower storey of the porch is covered with a large barrel vault resting on two arches supported by two piers. This barrel-vaulted space is separated from the small court by a marble screen, the style of which suggests a C6 date for, at least, the lower storey of the porch. The W pier of the porch appears to belong to the original phase but, while the uppermost part of the E pier (Figure 3) retains its original ashlar masonry, the lower part was replaced with reused marble blocks when the W wall of the SW outer buttress was built. Moreover, on constructional grounds, it is possible that the original barrel vault of the porch was also replaced after the C6. If so, then the present vaulted upper storey of the porch – the façade of which originally had three windows beneath an arch – dates (at the earliest) to the time when the lower barrel vaultof the porch was rebuilt.

The SW outer buttress of the church was not always as solid as it appears today. First, the middle and W wall are built in a style unlike the E wall. Second, an arcade originally formed the E side of the buttress on the ground-floor level and was probably only blocked in the Late Byzantine period. Third, the E and S walls on the middle level of the buttress opened originally to the exterior with large arches resting on what seems to be Middle Byzantine masonry. Again, these openings may have been blocked in the Late Byzantine or Ottoman period. Two upper rooms of the buttress, accessible by an internal staircase, preserve fragments of their Byzantine-period decoration, but this appears to have never been studied in detail. The cruciform chapel has traces of mosaics, some of which were published by Cyril Mango4, but the rectangular room has been so far almost ignored by scholars except for a brief description by Feridun Dirimtekin in the 1960s5. It is possible to distinguish three layers in these frescoes: the earliest two have aniconic decoration but the third depicts figures of saints, one possibly Patriarch Methodios.

Fig. 5 The marble floor E of the SW outer buttres

Immediately E of the SW buttress, there is a fragment of white marble paving, originally displaying a hexagon or octagon design within a rectangle (Figure 5). This seems tohave been limited on the S by a wall. Near the S-E corner of the church, a C6 pier, or part of a wall, originally veneered with marble, is preserved in the Ottoman porch on the S side of the outer buttress. Interestingly, the C6 pier is approximately aligned with the facade of the N ‘baptistery’ porch and the wall limiting the marble paved area E of the SW buttress. Perhaps these are remains of courtyards originally encompassing the church on all sides.

In our recent paper in Architectura, we also presented structural evidence for another C6 hall of the Patriarchate, far larger than the ‘baptistery’ and the rooms above the S porch of the church and SW access ramp. Although only the N wall and part of the E wall of this hall remain above the Ottoman ablution fountains at the SW corner of the church, enough survives to reconstruct its probable layout and appearance up to the springers for the roof vaulting (Figure 6). Evidence from later seasons suggests that this hall (termed by us the ‘Large Hall’) comprised a single space, rather than having two storeys – its central bay covered either with a cross-vault or a shallow dome. If we suppose that the hall was symmetrical around this central bay, then its S façade would be approximately aligned with the S wall of the ‘baptistery’, and its internal dimensions c.10m x c.20m, with a vault reaching c.13m above Byzantine ground level.

Fig. 6 The SW corner of Hagia Sophia, showing some of the evidence for the Large Hall.

The area north of the church
Byzantine masonry survives in the E wall of the Ottoman porch in front of the door in the last surviving part of the N wing of the C6 atrium, revealing that there was once a wall running N from the atrium. This wall and that of the atrium are bonded, and so it must have been built at the same time as the church. A pier beneath the substructures of the Ottoman medrese, and W of its main court, probably again dates to the C6. This is aligned with the W limit of the door in the N wing of the atrium and so in a position corresponding with the Large Hall on the S side of the atrium.

The most surprising discovery of the whole project is, perhaps, that the whole NW porch of the church, usually assigned to the Ottoman period7, is probably a hitherto unrecognised – and previously unpublished – part of Justinian's church. Removal of plaster from its interior surfaces shows that it is built in a manner typical of C6 walls at Hagia Sophia and that its N wall is bonded to the W wall of the Justinianic access ramp that forms its E side. This raises the possibility that, although the SW porch probably dates in its present form to the reign of Justin II8, it likewise had a Justinianic predecessor. If so, the barrel-vault with the famous C10 mosaic may be the last surviving part of that porch.

We recorded further ‘new’ material between the NW access ramp and NW outer buttress. These include a largely intact, apparently C6, marble pavement and wall-revetment in the small Ottoman court adjacent to the church. Carved slabs preserved on the N facade of the church and on the E face of the access ramp suggest that the upper part of the N facade was veneered with marble.

In the early 1980s, plaster was removed and the interior of the middle outer buttress was 'cleared' to its original floor level. At the same time, a neighbouring, now open, space to the W was cleared and the Ottoman roof that had formerly covered it was demolished. Further demolition has been carried out immediately N of the buttress (in the area currently known as the 'Vezir´s Garden'), where Ottoman annexes had until recently stood, and this area 'cleared' to its Byzantine ground level. Despite the importance of these areas, only a brief report was published and there seems no prospect of any fuller publication9. Thus, the Museum kindly allowed us to record these crucial areas.

The buttress (which may date to the Middle Byzantine period in its present form) originally comprised flying buttresses, connected by a short wall on the N and covered by an inclined barrel vault. However, an earlier structure, evidenced by two large brick and greenstone piers (one with its original marble facing) - perhaps dateable on constructional grounds to the C6 – preceded the buttress. Part of the E brick and greenstone pier was incorporated into the N wall of the buttress, but the W pier remained outside it and today forms part of a wall, the upper part of which is undoubtedly Ottoman. However, the lower part of this wall shows Byzantine pure brick masonry. Another Byzantine wall faces the former across the small court and is aligned with the S stone piers of the buttress. The SW stone pier of the buttress rests directly on marble slabs of the floor inside it, suggesting that this also predates the buttress. In sum, this evidence demonstrates that a, probably C6, structure preceded the buttress.

An opening in the N wall of the buttress leads to the 'Vezir´s Garden', the site of the C4 or C5 hypogeum often shown on plans of Hagia Sophia. Previous work demonstrated that the hypogeum was later disused for burial, and a water conduit entered it from the E. This found a pier on the S wall of the interior of the hypogeum, apparently blocking one of its burial chambers, and there was also another on the N wall blocking another chamber10. The former is apparently the same as the W of two brick and greenstone piers that we recorded above ground. These piers, together with a similar brick and greenstone pier surviving on the W face of the NW outer buttress, seem to have been part of a single rectilinear structure (c.24.5m x c.13.5 m), slightly out of alignment with the church. This rectilinear structure appears to have been also associated with the marble floor inside the middle outer buttress to its south. Moreover, it may be seen in the context of a well-preserved white and green marble floor exposed in earlier work above the hypogeum. c.40 cm below this there was an earlier white marble floor11. The upper floor may relate to the same building phase as the brick and greenstone piers, as it is on the same level as the lower edge of the marble revetment preserved on the N face of the SE pier. The lower floor may be associated with the hypogeum. In addition, there is an enigmatic (c.1.9m square) brick structure on the axis of the hypogeum corridor, post-dating the upper marble floor. The E side of this is curved, and fragments of an ambo found during clearance would fit into its curved wall. This may suggest that the structure had an ecclesiastical function.

A N-S wall runs along the E limit of the hypogeum to the outer buttress. This was extensively repaired, but the original banded masonry includes smaller bricks (30-31cm long and 4.5-5 cm thick) than those in the hypogeum, resembling those in what may be the Constantinian phase of the Hippodrome12. Two – uniquely among known Byzantine bricks – bear single-line stamps on their sides. Taking into account the construction, brick sizes, and the distinctive position of the brick- stamps, we date this wall to the C4. If so, it is the earliest construction recorded in Hagia Sophia (Figure 7). Another stamped brick of this sort is re-used in the structure with the curved side and there are similar bricks in the rear wall of the pre-Justinianic portico of Hagia Sophia (see below) - the latter confirming their early C5 or earlier (we suggest C4) date.

Our work also included the Skeuophylakion and the NE access ramp. Our records of the dimensions of bricks in the original phase of the Skeuophylakion suggest an early C5, rather than C4, date for the structure (the average length of bricks is 34.5-37 cm and their average thickness is c.4.5 cm)13. In addition, we found several mason’s marks on the pavement in the small court between the Skeuophylakion and the church. Finally, on the NE of the church, examination of the structure revealed that the vestibule on the S side of the ramp is a later addition to the church. Moreover, removal of the interior plaster enabled us to identify a door in the N-S corridor in front of the ramp, perhaps for communication between the church and the Skeuophylakion14. During the Museum's restoration of the ramp we also recorded numerous mason's marks on the C6 marble window grilles, re-used stamped bricks and several almost illegible, possibly Late Byzantine, graffiti on the mortar of the ramp walls, as well as a fresco-decorated arch and a mosaic fragment in a niche beneath this arch.

Fig. 7 The N middle outer buttress and hypogeum area, showing remnants of two brick-and-greenstone piers in front of the outer buttress seen from N.

The Theodosian propylaeum
Although the bricks in the upper part of the rear wall of the propylaeum excavated by Schneider in front of the W façade of the church may date to the early C5, those in the lower part of the wall compare closely with the bricks from the hypogeum area and may, therefore, be assigned to the C4. It is possible that the lower part of the rear wall of the propylaeum dates back to the construction of the first Hagia Sophia. This may suggest either that at least the W limit of the first and second churches coincided, or that the Theodosian church re-used the W wall of the earlier church for its propylaeum.

Conclusion
This study may have identified the earliest structural evidence recorded at Hagia Sophia, and has produced new evidence for the C6 church. We have elucidated the Large Hall SW of the church and revised understanding of the so-called 'baptistery' and its N porch. This was connected to other structures along the S flank of the church by a corridor running from its NE corner. Other buildings were probably not located adjacent to the S aisle of the church but were separated from it by one or more courts paved with marble. Similar paved spaces existed along the N side of the church. The courts on the N were flanked not by buildings of the Patriarchate, but by those directly connected with the liturgical life of the church. One, newly evidenced by the massive piers and marble floor, may have been the 'Great Baptistery', which textual sources locate in this area15.

Notes:
1. K.R. Dark and J. Kostenec, ‘The Byzantine Patriarchate in Constantinople and the Baptistery of the Church of Hagia Sophia’, Architectura 36 (2006) 113-130.
2. Previous scholars reconstructed its eastern apse as originally externally curved (eg. S. Eyice, ‘Le baptistère de Sainte Sophie d’Istanbul’, Atti del IX Congressso internazionale di Archeologia cristiana (Rome 1978) 2 vols., II.257-273) or polygonal (eg. R. Mainstone, Hagia Sophia. Architecture, Structure and Liturgy of Justinian’s Great Church (London 1988) plans A2 and A9).
3. W. Müller-Wiener, Bildlexikon zur Topographie Istanbuls (Tübingen 1977) fig. 75; Mainstone (as n. 2) plan A9.
4. C. Mango, Materials for the Study of the Mosaics of St. Sophia at Istanbul (Dumbarton Oaks Studies, Washington DC 1962) 38-39.
5. F. Dirimtekin, ‘Room with frescoes in one of the buttresses of Saint Sophia’, Ayasofya Müzesi Yilliği 3 (1961) 53-56.
6. For courtyards around the church: Procopius Aedificiis 1.1.58; Paul Silentiary Descriptio S. Sophiae 610ff.; Evagrius Historia Ecclesastica 4.31.
7. For example, Müller-Wiener (as n. 3) fig. 75.
8. Mainstone (as n. 2) 29.
9. A. Koyunlu, ‘Die Bodenbelage und der Errichtungsort der I. II. III. Hagia Sophia im Lichte einer Ausgrabung’, Ayasofia Müzesi Yilliği 11 (1990) 147-156.
10. F. Dirimtekin, ‘Un hypogee dans le jardin dit “du vezir” au nord de Sainte Sophie’, Ayasofia Müzesi Yilliği 10 (1962) 109-115 (esp. 110-111 and plan).
11. Koyunlu (as n. 10) 151-154.
12. For the bricks from the Hippodrome see: J. Bardill, Brickstamps of Constantinople (Oxford 2004) 2 vols., I.105, 118-119.
13. For the dimensions of C4-5 bricks see: Bardill (as n.13) 105-106.
14. R.F. Taft, Divine Liturgies – Human Problems in Byzantium, Armenia, Syria and Palestine (Aldershot 2001) Study VII, 17-19.
15. Taft (as n.15) 7-9, 21-23.