Professional Documents
Culture Documents
The Building:
Its Architecture and
Archaeology
INTRODUCTION a en between the two stories, and
e present architectural each is accessible by ils own entrances,
e Despite the numerous destructions and remodel-
ue of the Bodrum Cami vepel ie we ings which the church has suffered In the course of
general conclusions regarding the history of th ae —_its history, its salient architectural features are still
buildin ryofthebuild-
ing in Byzantine times. lTheadjci
sae erected by Romanus' rdn a its entirety evident, The building is small in scale, but strictly
original intent of the wabere ane § hls palace. The organized, both in the articulation of its wall surfaces
suppo -pforthecchurch, anda die ina provide a and in the internal organization of space. Large half-
buttresses—the most disti nctiv e aspect of
The ‘ie onian phase cylindrica l
it served no relig ious func tion
was badly the exterior—punctuate the west facade and the
damaged by fire about 1200, vrre oe ably in the Latin flanks of the narthex and naos, marking the order of
Conquest of 1203-1204 Nice
Conquest, it was Sonica s ane of the Latin internal bay division (figs. 10, 11, 38). The strong ver-
ae
isch ime:the:substructii was fitted out asaburial the year 1300, at —_tical accent is, in turn, transected by horizontal cor-
a al : e —_niced string courses, dividing the lower elevation into
two registers. Emerging above this in simple, geo-
|
The church stands at the southeast corner ofahigh, metric me is the dieistrichioe fully intelligible as
flat terrace, now almost entirely enclosed by apart- to the shape which it encloses (fig. 40). A cross,
ment houses. to the west, south, and east.’ The ex- —_—_ formed by barrel-vaulted gables, is inscribed within
plorationsof Wulzinger, Talbot Rice, and Naumann? _the square of the naos below and rises out of it. Sur-
have. shown. that this terrace covers the lower walls mounting the cross at its intersection is a dome, raised
of animme nscent
fifth e.ury rotunda, 41.8 metersin on a cylindrical drum, penetrated by eight windows,
exterior diameter. The identification and original pur- and articulated between the windows by triangular
poseof this rotunda are'still a matter of speculation, —_ projecting buttresses. The clarity of the exterior ex-
but by the tenth century. it was no longer in use. At tends to the eastern elevation of the church (figs, 12
lateral
this time; simuSltan
e Te eous. with the erection of the and 39). The main, polygonal apse is flanked by
church; the entire interior enclosed by the lower walls —_ polygonal apses of the prothesis and diaconicon, in
of this rot wau n
s co reda
ved with a vaulted system each of which is a single recessed window. Originally,
carried or column Aboves:this, on the eastern half the main apse was opened by lights in each of its
ce
rsafor med, was erected the palace, three facets, set off by mullions.
of the terthu
One enters the building now through a single door
its facade facing west. The vaulted enclosure below
in the west facade of the narthex. From the narthex,
served as a cistern, The rising walls of the palace are
three doors now open into the naos, corresponding
no jonger’visible, but its foundations were fully ex-
posedin the'recent excavation of Naumann. to the nave and two side aisles. The two side passages
may originally have been blocked.
As part of this complex—early rotunda, vaulted ter-
The plan of the naos, excluding the bema and pas-
race with cistern below, and palace—the church was
tophoria (fig. 1), is laid out as a square 8 meters on
erected, much as it stands today, against the south-
a side. Inscribed in this is a cross with arms 4 meters
east wall of the rotunda, flanking the superimposed s,
is on in width, projecting in the four cardinal direction
palace, It is a two-story structure. The church into nine
is all- and subdividing the square plan of the naos
the’ same level as the terrace and palace, and cal nti _—ibays.
on. The sub str uct ure , alm ost ide
brickconstructi At the int ers ect ion of the cro ss- arm s, four piers
in’dimensions to the church, is of rough of ashlar and car ry the su pe rstructure, con-
fo rm thesame — (or igi nal ly co lu mn s)
s to cre ate a pl at axi s, ris ing
brick. Its pu rp os e wa
ou nd __ sis tin g of ba rr el -v au lt ed ga bl es in the cro ss-
ro tu nd a, th er eb y br
height as the ch to that of the palace. No internalin gi ng the gr -
ab ov e the four subsictiary corner bays. Above the in
ievel of the chur
ch
i v e s , i an igh
ei ght-t-cuspe d the apartment house to the south of the chur at No.
pe nd en ti ve s, Is 17 Mesih Pasa Caddesi permanently exposeq
12 tersection, resting on um an
carrying a dome. the so ut h fac ade , an d the ex ca va ti on s of Na
ce nt ra l sq ua re of 7" = r west
“To ie cant ac e to th e
south yD : ‘nm 1965-1966 permanently laid free the Uppe
no rt h an d th facade
is the be ma , fl an ke d to th e
un icate WI . facade and the western portion of the nor
on ic on , wh ic h co mm
thesis and di ac
s be en p The eastern part of the north facade and the west end
de co ra ti on ha
through doors. No tr ac e of of the substructure still remain obscured by later
served in the interior of the church. structures or earth fill.
Access to the interior of the substructure was o,;,.
The church stands on a rectangular platform pro- inally possible probably only through a door in the
vided for it by the substructure (fig. 28). The platform south facade which opened directly into the naos,
‘pinal dimens
is aeslightly larger (estimated orginal ¢ ions
; 13.10
(th
the This was later blocked and reopened at the beginning
x 24.10 meters) than the exterior dimensions of of the present investigation. The interior of the nag.
church (11.22 x 17.50 meters) thereby creating an €X- is a simple, rectangular hall 8 x 10.65 meters in plan
terior gallery which originally ran around the entire (fig. 7). Within this, four columns in the center of the
church ona level with its floor. To the west, the gallery
(at this point 1.30 meters in width) provided access room and two walls projecting perpendicularly from
from the palace to the main entry of the church. Along the east wall provide support for the four columns
the two flanks of the church, it was little more than and bema walls of the church above (fig. 29). At the
a catwalk, one meter in width. But to the east, it same time, they carry a system of barre] vaults (fig.
opened to a terrace 5.50 meters in width (fig. 27). The 51) corresponding to the bay division of the church:
eastern terrace is now lost and the gallery along the wide vaults in the main cross-axis, Narrower vaults
flanks badly eroded, but the western gallery, first ex- in the aisles flanking the bema and in the two sub-
posed by Talbot Rice and now permanently by Nau- sidiary corner chambers to the west. The naos interior
mann, preserves most of its original form. was lit by windows above the two doors and by three
In the west, and along the two flanks of the build- windows in the east bema wall.
ing, the gallery rests on corbels connected by barrel A single door leads from the naos into the narthex,
vaults, which arise from rectangular buttresses placed a barrel-vaulted hal], 2.65 meters in width, and iden-
around the exterior 1walis of the substructure corre- tical in length to the width of the naos. This narthex
sponding in their position to the half-cylindrical but- was originally lit by lunettes over recessed panels:
tresses of the church. The corbels are the necessary one in the center of the west wall, a second in the
solution to the problem of providing a projecting sup- south end. At a later date, access was created to the
port for the surrounding gallery and at the same time interior of the cistern within the adjoining rotunda by
maintaining the walls of the lower and upper stories cutting a hole in the north wall of the narthex and a
in vertical alignment with one another. It is important corresponding one through the wall of the rotunda.
to bear this in mind in understanding the coherent The hole in the wall of the narthex was filled in the
relation between church and substructure. course of the work of Naumann.
In the east, the gallery or terrace was originally
supported by a barrel vault in north-south axis rising Briefly stated, this is the present situation of the
from the buttresses of the substructure and carried building and the general site, together with a brief
farther to the east by a wall parallel to the substruc- identification of its original form. A general descrip-
ture. The springing of the vault and the wall still exist, tion of the archaeological findings might also be useful
and a portion of the vault was still intact at the time as an introduction to the detailed report.
of Talbot Rice’s excavation, but this in the meantime Digging was carrried out in two areas. A small
has collapsed and disappeared. sounding was made in the floor of the church at the
The substructure itself presents a utilitarian aspect, base of the northeast pier. From this it was possible
resulting from its rough ashlar and brick construction, to determine how the stylobates of the church were
and contrasting with the all-brick construction of the laid on the vaults of the substructure. Furthermore,
church (figs. 11 and 50). In its original phase, to the at least in the area where the vaults of the substructure
extent that it was visible, its most notable architectural were exposed, no evidence was found for structure
feature would have been the aforementioned rectan- having rested here prior to the present church.
gular buttresses, which subdivide the otherwise un- Mention has already been made of the earth fill,
decorated wal] surface. Subsequent progressive earth- which at the beginning of the present investigation
filling raised the exterior ground lIevel to that of the encumbered the interior of the substructure to a
church (8 meters above the original ground level of height of 3.50 meters of its 7.75 meters. Access to the
the substructure), totally obscuring the substructure. interior at this point was only possible by means of
Portions of its exterior were exposed by Talbot Rice a ladder lowered through the window in the east wail.
in 1930, but then refilled. In 1950, the construction of To facilitate excavation, the blocked south door of the
tere |
orig ji proved on the basis of pottery to be ap- eho etical reconstruction of the superstructure of
The ately half Turkish and half Byzantine in date © Palace (fig. 68). At least the plan of the building
prom Where the Byzantine fill was undisturbed and the general lines of its elevation have been ver-
ified by the recent excavation of Nauman
formed a sealed deposit, it lay in well-defined n.4
and vevealing the following sequence of events, No _ Tt will be recalled from the brief description in the
introduction to this chapter that the palace, with its
strates discovered for an original floor or floor set-
trace Unless a floor and all trace of its setting was underlying cistern, was built in and over the existing
ruins of an immense, niched rotunda. This rotunda
etal ately removed prior to the thirteenth century,
was first made known through the excavation of Tal-
. must have existed in the primary phase of the bot Rice, who was able to determine its general form
njding. A destruction of the building by fire about (fig. 67). Talbot Rice accepted Wulzinger’s analysis
1200 is indicated by the discovery of an ash layer and nica of the palace without elaborating
rectly overlaying a fine fragment of Persian Minai on it.
ware. Not long thereafter, in the late thirteenth or The most extensive investigation of the rotunda-
arly fourteenth centuries, the church was remod- palace-cistern complex was undertaken by Naumann
e The evidence for this is a deep layer, overlying
in 1965-1966. He exposed much of the perimeter of
the ash layer, of brickbats and architectural fragments the rotunda including its north and south doors,
in a loose earth matrix. The greatest concentration of down to their original ground level, and laid free the
these fragments directly beneath the windows sug- foundations of the palace above (figs. 69, 70). The
gests that they were dumped through the substruc- description here of the complex is based on obser-
ture windows from the church, and represent dam- vations made during and after the excavation and
aged liturgical furniture, opus sectile flooring, and upon Naumann’s preliminary report of his excava-
marble revetment from the primary phase of the fions,
church that could not be reused. The rotunda has an exterior diameter of 41.80 me-
Shortly thereafter, the substructure itself was fitted ters, making it, next to the Pantheon in Rome, the
out as a burial chapel (fig. 32b). The ground was lev- second largest known from the Roman period. Its
eled and a brick floor laid. Along the south wall and interior diameter is 29.60 meters, with walls 6.10 me-
in the bema, multiple tomb chambers were created ters in thickness, now standing to a maximum height
by digging out the fill down to the original ground of 3.40 meters above the original floor level. The main
level, laying rubble walls against the fill parallel to portions of the rotunda are of large, finely worked
the south wall and across the bema, and subdividing limestone ashlar, in blocks averaging 1.20 to.1.50
these areas into smaller chambers by perpendicular meters in width, set in courses averaging .50 meters
walls. The chapel bema was decorated with fresco, in height and fixed in some places with iron cramps.
and from this decor the lower portion of a panei de- The interior is opened by eight niches, four rectan-
picting a female donor kneeling before a standing gular ones in the cardinal axes and four semidrcular
Virgin Hodegetria has survived on the north bema ones in the diagonal axes. Doors open to the exterior
wall (figs. 60, 61). in the outer walls of the north and south niches.
The Turkish fill yielded the greatest concentration The exterior of the north and south of the rotunda
of pottery from the sixteenth century, somewhat less is provided with flat facades. Of these, the north fa-
from the eighteenth century and some modern, but cade (fig. 71) is the more richly articulated of the two,
little correlation may be drawn between these fills and and is presumed by Naumann to have been the prin-
the history of the building in Turkish times. cipal entry. A stylobate with two column bases i situ
precedes this facade, from which one must assume
the original presence of a colonnaded porticus. The
ROMANUS’ CHURCH south door (fig. 73) is preceded by a flight of eight
semicirculars steps which compensated for the down-
The Site, Palace, and Rotunda ward sloping terrain from north to south. Flanki
The original construction of the Bodrum Camii was the door at the western extremity of the flat facade
directly related to the siting requirements of the ro- is a wall projecting to the south, in bond with the
tunda with its superposed palace lying to the north- wall of the facade (fig. 69). Naumann reconstructs a
west, corresponding wall projecting from the eastern ex-
In a remarkable analysis of the architectural evi- tremity of the facade and suggests that both origi-
dence then available, Wulzinger was the first to dis- nally may have given rise to colonnades. A column
cern in the vaulted construction of the so-called Bod- base in sit was discovered in north-south alignment
rum Cistern the substructure of a palace (fig. 66), with the western projection, and a colunin base with
THE BULLDING
rch their blocking walls, and at the southeast to
16 lution to the problem of how to build the chu ig comment of the church substructure.
it in an
against the round wall of the rotunda, orient Following this the excavated zones could have been
t
east-west axis, make it accessible by entry in its wes filled and the site terraced downward from north to
facade, and keep this access at the same level as the
ground floor of the palace. —— consideration of the church substructure, we
will present evidence favoring such an interpretation.
There remains to be considered the general ap-
P on of its new build-
upo completi
pearance of f the the sitesite upon But an alternative interpretation of the significance of
ings, and the related question of means of access from the tipped column found by Naumann also leads nec-
palace to church. Our interpretation of the evidence essarily to the same conclusion.
relating to both of these questions differs in several If this column, together with others which would
essential respects from Naumann’s. From our earlier have accompanied it in a north-south range lying ap-
discussion of the Coliseo, we recall that Naumann proximately parallel to and three meters west of the
visualized the rotunda as standing, at Jeast in part, church substructure, was reutilized at the time of the
sufficiently above ground for him to identify it in the church construction to carry an extension of the west-
Valvassore print as the Coliseo. Moreover, he believed ern gallery outside the church, we must explain how
this extension connected to the corbel system of the
it to have been visible not only above the ground level
west face of the substructure. But the west face of the
of the palace and church, but, at least on the south
side, down to the original ground level of the rotunda. corbel system exposed by Naumann (fig. 41) is fair
and clean, with no trace that anything ever was at-
This conception emerges from his discussion of the
means of passage between palace and church, which tached to or abutted it. Moreover, had a wide gallery
he believes was effected by a bridge at the point of carried to the west by columns been intended, there
intersection of the rotunda and the northwest comer would have been no need to construct the corbel sys-
of the church substructure. At this point the upper tem in the first place, since arches connecting the
corner of the substructure overlaps the rotunda and church with the colonnade would have supported the
rests upon it, without doubt providing here a con- western gallery more easily. We must thus discard
necting bridge between palace and church (figs. 28, the hypothesis that the tipped column had any struc-
48). tural relationship to the palace and church complex,
icin Naumann’s discovery of a tipped column At the same time, the very presence of the column
approximately four meters to the southwest of the offers supporting evidence for our hypothesis of a
church substructure, and belonging originally to a largely subterranean rotunda. Naumann is certainly
colonnade flanking the entrance to the rotunda, leads correct in associating the column with the primary
him to suggest that this colonnade might still have phase of the rotunda. He found the column tipped
been standing in the tenth century, and might have to the west, with its lower end still resting on the
carried part of the connecting bridge between palace original column base, plinth, and stylobate system.
and church. Accordingly, Naumann supposes that If the column could not have been used as part of the
what survived of the south face of the rotunda and palace and church structural system, the only expla-
its flanking colonnade were exposed down to their nation for its presence is that it remained buried and
original ground level; and whatever the colonnade undisturbed in its tipped state during the construction
carried was connected to the west face of the church of the palace and church. Accordingly, the top of the
substructure at the level of the corbels, thereby wid- column would be the minimum exterior grade level
ening the western gallery outside the church. of earth surrounding the rotunda at the time of the
Our investigation of the substructure virtually ex- tenth century construction.
cludes this last possibility. Moreover, there is an al-
ternative conception of the appearance of the com-
pleted site (fig. 26) which more reasonably explains The Church
all of the internal evidence. We believe that at the There remains from the primary phase of the church
time of the construction of the palace and church, the the foundations, stylobates, portions of each of the
rotunda was found and remained largely subterra- exterior half-cylindrical buttresses, most of the inte-
nean. Once the identification of the rotunda with the rior wall surface above the lower cornice, and with
Coliseo is discarded, there is nothing to suggest that the exception of small repairs all of the interior vault-
any of the rotunda superstructure rose above the ing. Some primary structure has also survived in the
ground level of palace and church. Jower part of the eastern elevation. The small size of
If we suppose further that below this level the ro- the building, and its regular, bilaterally symmetrical
tunda was largely buried and, on the exterior, re- form make it possible to reconstruct its original form
mained so, it would only have been necessary to clear with considerable accuracy.
the interior and to excavate on the exterior at three Evidence for the original form of certain important
places: outside the north and south doors sufficiently features, however, is completely lacking. Among
THE BUILDING
18 responding supports in the substructure, to synchro- north facade at the time of our survey consisted only
nize this alignment with the interior bay organization of the restoration of 1964-1965, and it was felt that a
and to preserve the apparent identity of volume measured survey drawing of this elevation could be
among all buttresses, while, on the other hand, not omitted.
obscuring the windows in the narrower eastern bay.
The solution was using quarter-cylindrical buttresses. The west facade (figs. 10, 21a, 34) was divided hor-
The buttresses not only serve as the primary ele- izontally by a cornice into two registers and vertically
ment of exterior wall organization, but immediately by the buttresses into three bays. The cornice may be
express, as well, the nature of the internal organi- seen on pre-restoration photographs, and the lowest
zation of the building. In plan, both the major divi- courses of the three northern buttresses survive, In
sions into narthex, naos, and bema, and the subdi- the lower register, the central bay is opened by a door,
vision of each of these areas into bays, are marked the original width of which, 2.20 meters, may be de-
with their respective buttresses, termined by the lowest courses of its reveals, which
Within the church, each exterior buttress is backed survive. Over the door was a shallow segmental arch.
by a corresponding wall respond, confirming the rec- The arch has been rebuilt, and whether or not it was
iprocity of exterior and interior articulation, and giv- a true segment cannot be determined from photo.
ing rise in the interior to the supporting arches of the graphs, It sprung at an approximate height of 3 meters
vaulting. The alignment of buttress and respond in above the threshold. Two hinge sockets in the thresh-
every case is not exact (fig. 1) since slight adjustments old indicate the original presence of bivalve doors,
were necessary to reconcile their differing function. It appears that windows flanked the door in the
But these adjustments may only be discerned on two side bays, although information about them jis
measured drawings. imprecise, Their arched heads appear clearly on pre-
restoration photographs, and the surface masonry
A second notable feature of the exterior is the va- filling their lights is discernibly later. Yet it is not cer-
riety and number of openings in all facades of the tain that these were windows rather than blind ly.
building, Special difficulty exists in determining the nettes, While they were already closed in Palaeotogue
original form of doors and windows in the lower reg- times, the generous fenestration elsewhere in the
ister of the building and to a somewhat lesser extent building would suggest that windows originally ex-
in the attic and gables as well. It is evident from pho- isted here as well. The height of the sill given in the
tographs and the descriptions of other scholars, that perspective reconstruction (fig. 28) is conjectural.
even before the restoration of 1964-1965, considerable In the attic zone of the facade, lunettes crown the
modifications had been made in doors, windows, lu- door and the flanking windows, but once again all
nettes, and portions of the wall surface between the evidence for their original form has been obliterated
half-cylindrical buttresses. Some of these modifica- from the building itself. The reconstruction of the two
tions may be assigned to the Palaeologue remodeling flanking Iunettes as blind and the center lunette open
of the building, others to remodeling and repair in is based upon a clear difference in filling masonry
the Turkish period (fig. 22). These will be discussed visible on the pre-restoration photographs. The lateral
in their appropriate context below. But even the evi- fillings, discounting stone and rubble repair, are all
dence for the later modifications was almost totally brick, and follow exactly the coursing of the remain-
obliterated in the lower portion of the exterior by the der of the facade. In the north lunette, the filling even
restoration of 1964-1965. The west facade was refaced appears to be in bond with the lowest courses of the
in its entirety (figs. 36, 41). The north facade, save for surround. The filling of the central lunette, on the
two small zones, suffered the same fate. The south contrary, is of rough squared stones and brick and
facade (fig. 40) fared somewhat better, but here as shows evidence that a once open semicircular light
well, windows were inserted and walls rebuilt with the size of the surround was reduced to a smaller
little regard for their original form. semicircular light and then finally blocked. The filling
Thus, in the reconstruction and analysis, especially in two stages is clear evidence that the lunette was
of the lower portion of the exterior, it has been nec- originally open.
essary to rely heavily on photographs of the building From the pre-restoration photographs it is impos-
taken before the restoration. From these, a graphic sible to determine the original aspect of the spandrel
reconstruction has been attempted of the ground wall above the three lunettes. Most of this zone ap-
story of the west facade and the two flank walls (fig. pears to have already been repaired before the 1964-
21), for the purpose of distinguishing their structural 1965 restoration. The photographs suggest that a con-
phases, Since the eastern exterior was less affected centric frieze of some sort may have surrounded the
by the restoration, its various phases could be ade- arches of the lunettes, and presumably a coping car
quately rendered in the survey drawing (fig. 12). The ried the projecting eaves of the roof.
THY BUILDING
22 form could have been modified, The greater part of crown of the dome are in the respective ratio of 1:2:4,
the drum exterior, together with the spandrel wall Moreover, the gabled cross-arms rise to a height ex-
above the windows, dates from a rather careful Turk- actly three times their width.
ish restoration of undetermined date. We are there- But caution must be exercised here in attempting
fore inclined to accept its present form as a close re- to deduce an overall system of proportions. While
flection of the original, even if some details may not one may have existed for the building, the combj-
be entirely correct. nation of structural error, subsequent modifications,
The drum is a cylinder 5.50 meters in diameter and and our ignorance of the actual method by which the
rising to a height of 3.15 meters. It is opened by eight building was laid out, makes it extremely difficult to
roundheaded windows, the reveals of which are ra- deduce such proportional relationships except in the
dial. Set between each two windows is a triangular most obvious and simple cases, and then only when
buttress rising to the same Jevel as the window soffit it is certain that measurements may be taken from
and giving rise to dogtocth friezes over the windows. primary structure.
As in the case of the lower portions of the building, These proportions impart to the naos a marked ver-
it is uncertain whether or not the frieze reflects the ticality, enhanced both by the wall responds and the
Original form. The effect of the buttresses, on the one four vertical supports placed at the intersection of the
hand, is to enhance the already strong profiling of cross-arms. Agreement has not always existed as to
the drum, and on the other, to introduce a prismatic, whether or not the present crude square ashlar piers
octagonal element into the otherwise strictly cylin- represent the original support system, or are a re-
drical composition. Above the spandre] wall of the placement for columns which originally stood in their
windows, a dogtooth coping encircles the drum, but Jace. Those who have contended the former have
it is uncertain whether this element is original. considered the piers to be a preliminary phase in the
development of the four-column church. And even
INTERIOR. In turning from the exterior of the church among proponents of the latter opinion, it has been
to its interior, it is immediately evident that similar Suggested that the present piers may enclose the
problems exist in determining its original form and original columns.”
subsequent structural history. Repeated damage and Our investigation has left no doubt that col-
repair had already wrought substantial change in its umns—subsequently removed—were once the orig-
interior aspect before the 1964-1965 restoration. But inal support. Cleaning at the base of all four piers
the restoration had a particularly disfiguring effect revealed the badly damaged but stil] discernible oc-
(figs. 43, 45) through the crude refacing of almost all tagonal bases of the original columns (fig. 59). When
wall surfaces below the lower cornice. Detailed ob- the crude plinths of the present piers were set, the
servations regarding the prerestoration state of the original octagonal bases were filled out to form a
interior were also hindered by the fact that virtually square base, explaining why they have never been
no photographs taken beforehand were available. The observed before. In connection with the removal of
discussion must therefore be confined only to those the columns, the spandrels above each of the piers
areas unaffected by the restoration. were rebuilt (fig. 43). This is indicated not only by a
The plan and general organization of the interior difference in their masonry and by a joint near the
have already been described: a naos, square in plan, crown of each arch springing from the piers, but also
is preceded by a rectangular narthex, and appended by the absence of post holes for tie beams directly
to at its opposite eastern end by the bema and pas- above the piers, corresponding to holes directly op-
tophoria. The predominant effect of the interior is of posite in original masonry. The rebuilding of the
a compact, unified space, given form and intelligibil- spandrels was carefully executed and dates probably
ity by strong profiles and rich modeling of wall sur- from an earlier Turkish period. Leaving sufficient
faces in a tightly coordinated system of articulation. room for capitals, it may be estimated that the original
The interior (figs. 15-18) is divided clearly into hor- column shafts were approximately 3.70 to 3.80 meters
izontal zones, marked by cornices corresponding to long.
those of the exterior. Jn the naos there are four zones:
ground story, attic, gable, and drum with dome. Sim- Considerable attention has been given above to the
ple ratios govern the proportions of these vertical analysis of the exterior of the perimeter wall of the
divisions, When the heights of the cornices are re- building, and here it will be sufficient to consider
duced to their lowest whole-number ratios of 9:13:18, those aspects of the ground story interior which are
it becomes evident that the ground story rises to ex- not reflected on the exterior.
actly half the height of the interior up to the drum Notable among these is the handling of the bema
ring. Other simple ratios exist as well. Measuring and especially the pastophoria flanking it. Undoubt-
from the lower cornice, the heights of upper cornice, edly these were originally screened off from the re
intrados crown of the gable vaults, and Intrados mainder of the naga by an iconostasis, although no
THE BUILDING
24 chambers. They are carried by strong arches in trans- The terminus ante quem for the dumping of this deposit
verse axis and by the niched lunettes in longitudinal, is set at ca. 1300 from the latest pottery found in it,
thus developing the trilobed effect already observed The layer overlies an ash layer throughout the interior
for the plan into the elevation as well, of the naos indicating the destruction prior to the
The construction of al] three domical vaults is me- dumping of combustible material in the substructure.
ticulously regular. They are built up in concentric The ash layer, in turn, lay directly over a sherd of fine
courses of bricks smaller than those used elsewhere Persian Minai ware which can be dated to the decade
in vertical construction. The bricks were trimmed and ca. 1200. This establishes the terminus post quem for
cut by the mason as construction proceeded. As the the fire causing the ash layer as ca. 1200.
courses approach the crown and are laid in decreasing A further analysis of the circumstances of the de-
diameter, the bricks or brick fragments become pro- struction represented by the ash layer will be given
portionally smaller. below, but briefly stated, these confining dates would
in this context, mention should be made of the fact coincide well with a destruction of the church in the
that the domical vaulting of the church—particularly Latin Conquest of 1203-1204 and a remodeling follow-
that of the pastophoria—is identical in construction ing the end of the Conquest at the close of the thir-
to that of the substructure of the adjoining palace (fig. teenth century. By this interpretation, the architec-
76). Thus it contributes to the evidence for their con- tural fragments found in the substructure represent
temporaneous construction. decorative material from the church so badly dam-
aged as to be useless in connection with the Early
No trace of the original decoration of the church Palaeologue remodeling, and this is consistent with
remains in sift, which is not surprising in light of the the residual nature of the material found. The quan-
many vicissitudes which the building has suffered.> tity was great (261 pieces of opus sectile fragments
Yet it is impossible to conceive that a building of all- alone) but, save for a few stray intact pieces, frag-
brick construction—already a sign of affluence—small mentary and useless, indicating that it had been care-
in scale and consummate in its refinement of archi- fully sorted before being discarded. That this was a
tectural design, should not have been decorated in builder’s dump is further indicated by lime deposits
a correspondingly appropriate manner. And if the throughout the level.
building was, as we believe, built by Romanus, we From the nature of the material found, it is likely
may assume that its decoration was one of opulence that the church originally had an opus sectile floor
and splendor. made of geometric motifs in which diamond, hex-
Despite the absence of decoration jn situ, the ex- agonal, and mandorla shapes predominated. The
cavation of the substructure and the sounding in the walls were covered with polychrome marble revet-
church yielded fragmentary material, which we may ment with green Thessalian marble in particular abun-
assign with reasonable certainty to the decoration of dance, and the upper zones bore mosaic decoration.
the church. While only one piece of evidence suggests One element of the decoration deserves special
a tenth century date for the material, our excavation mention, since it is the only one found which has a
of the substructure places the decorative fragments bearing on a date of the deposit, and by extension on
as a whole to some time before ca. 1200; and there is that of the church itself. Fragments of ornamented
nothing which excludes an earlier date. polychrome revetment tile were recovered both from
The decorative material includes mosaic tesserae, the church sounding and from the architectural frag-
opus sectile, polychrome marble revetment, oma- ment level of the substructure. Unfortunately, the
mented polychrome ceramic revetment, and frag- context in which such tiles have been found elsewhere
ments of marble liturgical furniture. are not sufficiently exactly dated to determine se-
The material was found in two places. Tesserae and curely the time limits in which this method of archi-
polychrome ceramic revetment were recovered from tectural decoration was in common use.*4 In some
unstratified rubble fil] at the base of the northeast pier sites from Constantinople and from Bulgaria they are
in the church, and all elements mentioned were found found in association with tenth century buildings. In
in 2 deep stratum in the substructure excavation (fig. others not.
31), designated the architectural fragment layer (Level Their discovery at the Bodrum Camii in the archi-
3). The evidence for assigning the material from this tectural fragment level of the substructure excavation
stratum to the decoration of the church is as follows. is consistent with the hypothesis that they are part
The material lay in a stratum 1.30 meters in maximum of the damaged refuse from the tenth century phase
depth. It was deepest directly below the two windows and were dumped in the substructure at the time of
of the substructure and tapered down to the east and the Palaeologue remodeling of the church.
west, indicating that it had been dumped there There is one confirmation that such tiles were used
through chutes in the two windows from the church. by Constantinopolitan artists at least in the tenth cen-