There is little doubt that the occupation of the atrium in Phase IX was ended abruptly by a destructive natural phenomenon, an earthquake. The entire interior of the church, the area of the porticoes in the atrium, and western and northern rooms were found strewn with collapsed stone. Generally, the total depth of the loci that contain massive stone tumble varies considerably, from more than 3 m of the barely differentiated massive collapse in Room I to about half a meter in the center of the church. A notable exception is the very center of the atrium which contains very little stone, but it is suspected that the area could have been partially cleared in Phase XI.

During the excavations, particular attention was directed toward discerning the pattern of collapse, and separating the layers of stone material. This attempt was partially successful in the aisles where separation between stone collapse episodes was possible, due to the greater depth of the deposits. Occasionally, layers of silt lamination were clearly discernible between stone tumbles. In other cases, an evidently changed pattern of cultural material within loci suggested inter-collapse activities, occasional robbing, or human redeposition, as in case of Room I. The pattern that emerges of the entire post-Phase IX accumulation in the church proper, up to the modern surface. strongly indicates that the site may have experienced more than a single earthquake. Yet, it is certainly difficult to associate particular sets of loci with later earthquakes, except in a general sense. At least in some areas, stone collapse layers presumably resulted from continuous natural deterioration and decay rather than from a earthquake related phenomena.

Therefore, while the results of the earliest destructive seismic event may be reasonably recognized, their separation from later events, including long lasting natural deterioration, was not always successful. For example, the first (=lowermost) seismic related deposits in the church proper are usually well defined by the presence of column drums and capitals, but this is not always the case, since drums occur also in upper tumbles. That may indicate that the first tremor was substantial enough to cause the majority of columns to break and collapse, but perhaps not strong enough to destroy all of them. As such, some surviving stumps of already damaged columns, if not entire shafts, could have fallen in later destructive events. Many drums and capitals, obviously heavier than ordinary ashlars, fell into the strata representing the fire, or into deposits relocated in Phase IX, and often embedded themselves deeply into these layers. The first earth tremor caused the colonnades (and presumably the parts of the walls) of the church to fall in a northern direction. To be exact, the common slight deviation from the north evidenced in the collapse is the well attested alignment, NNW to SSE or NNE to SSW, of the fallen drums and ashlars with a general directional patterning, i.e. fanning out towards the north. This earliest architectural collapse in the aisles is also characterized by uneven distribution in random piles, as opposed to the upper stone tumbles which, wherever preserved, resemble more uniformly deposited fields of stone. This general pattern, not unlike that observed in the Civic Complex Church in Pella, which suffered two earthquake destructions, may also reflect scavengers rummaging among the stones after the first earthquake. - Fiema et al (2001:105)

For the purpose of a better stratigraphic control, some loci of considerable depth were subdivided during the excavations. Some strata associated with the early architectural collapse are generally deprived of finds, but most of them contain considerable amounts of iron nails, wall plaster, and roof tiles, as well as wall mosaic fragments. That indicates that the remaining portions of the burnt-out roof collapsed during this disaster, as well as most of the clerestory walls above the colonnades. Occasional finds of burnt wood or ash lenses in these strata are thus not surprising. Curiously, numerous marble fragments also occur in some tumble strata. The ceramic material found in the first collapse is far from homogeneous, but 5th-6th century types are most common, with some 7th century sherds. Thus the value of the ceramic dating for these strata is debatable Probably, both marble and the mixed pottery presence may be associated with later human churning of these layers. These shadowy and poorly documented phenomena certainly occurred in the later history of the site, and their influence should not be overlooked. - Fiema et al (2001:105)

The appearance of the deposits of drums and capitals, was particularly dramatic in the nave, notably in the western and central parts. Square B4 featured a field of densely packed stone material which extended vertically from ca. 899.5 m up to 900.5 m, including the collapse loci 07 and 03 (Fig. 116). Several column drums which preserve a rough NE or NW pattern of collapse were located in locus 07. Locus 03, atop of it, contained mainly ashlar collapse. This distinction should not be related to separate episodes. Rather, the collapse of columns of the south row was followed directly by the fall of the clerestory walls. These loci contained very little cultural material, but notable amounts of roof tiles. These were also characteristic for neighboring boring B3, where loci 05 and 04 displayed a similar pattern of deposition i.e. the column drums in the lower locus (Fig. 3, section f-f'). Farther east, Square A4 preserved the earliest stone collapse in loci 07 (upper) and 06. Locus 07 (lower) is probably a silt layer containing some gravel and charcoal. This presumably accumulated before the earthquake. Both loci 07 and 06 also yielded considerable amount of roof tiles. However, what set them apart from the deposits already mentioned was the abundance of cultural material including nails, mosaic chunks, glass, wall plaster, iron objects, and a great quantity of marble fragments, the latter located close to the edge of the chancel platform. A Nabataean coin was found in A4.07 (upper).

On the northwestern side of the nave, stone deposits were also considerable. Loci J1.06 (upper) and 04 were a very dense tumble with several drums and capital fragments. These rested in a matrix of sandy soil together with large quantities of cultural material. The tumble in 04 was of considerable depth in the northern part of the square, but petered out in the southern half, being supplanted there by locus 05 (Fig. 3. Section f-f'). That locus, at the same level as J1.04, contained few stones but an abundance of cultural material including roof tiles. A coin of Tiberius II Constantine, struck in Nicomedia in A.D. 580-81, was found in that locus. A more complicated stratigraphic situation was encountered in neighboring J2 (Fig. 117). There, loci 08 and 07 seem to be the lowest deposits of the structural collapse, which contained abundant cultural material. The tumble in these loci was of medium intensity, consisting primarily of scattered fragments of capitals, and drums. Two columns had collapsed into that square. Both were substantially scorched by fire, especially the capitals. Upon collapse, the latter shattered like glass into dozens of pieces which can be found even in locus 09, directly underneath. The major high-density collapse stratum with numerous stones and drums was locus 04. Altogether, the average depth of loci 07 and 04 was from ca. 899.7 to ca. 900.2 m. - Fiema et al (2001:105-106)

In the north aisle, the early collapse was well represented, and in squares such as J4, H2, and H3, separable from the later falls. This collapse was very uneven in terms of absolute levels, because of the considerable difference in depth of the material (including the pavers) relocated there in Phase IX. Well-defined early collapse was noted in Square G4, where Loci 23 and 21 contained drums and capitals, among the ashlars, as well as numerous wall mosaic fragments. Locus 21 was sealed by a silt lamination (locus 19), on top of which rested a later, equally impressive tumble. A similar situation occurred in the western part of the aisle. Loci J4.11 and 10 (early collapse) were well separated from the later collapses (J4.05, 02) by the silt layer, J4.08 (Fig. 3, section i-i'). A substantial stone tumble was observed in H2. Loci 09, 07, a total of ca. 0.6 m in depth, cover the area of the nave and the north aisle. Several drums were floating in these loci. Judging from the lack of any pattern, the drums could have been pushed aside from the top of the bema in the following phase. The presence of a reasonably well-defined upper stone tumble here was also noted. That upper collapse also contained column drums which may indicate that some columns or their stumps had survived the first tremor. The early collapse was poorly represented in Square H4, although loci 24 and 21 may be reasonably associated with that episode. This square featured an easily recognizable upper tumble (H4.15, 14) with several column drums and large ashlars (Fig. 2, section d-d'). Evidently, not all columns had fallen during the earthquake of Phase X.

Most problematic was the pattern of collapse in Square H3. It preserved an entire column - the easternmost in the north colonnade - which had fallen in a row of several aligned drums (Fig. 118). The column was at ca. 900.2 m on the bottom of locus H3.11, a tight packed tumble, the top of which reached 900.8. Directly south, was the tumble locus H1.07 which contained the continuation of the column from H3.11 as well the remains of the column next to it. The drums in H1.07 were more randomly dispersed, without an aligning pattern. According to stratigraphic observations from the neighboring Square G4, the collapse of the entire column in Squares H3 and H1 should be associated with later destructions which, in G4, rest on the silt lamination locus 19 (supra). This proposition, based upon comparative absolute levels, seems possible, although it faces difficulties. Loci below the fallen column in H3.11 indeed contained some stone material (H3.15, 16) which may represent a rather insubstantial, early collapse. On the other hand, locus H1.07, where the remaining drums of the collapsed column were deposited, is evidently associated there with the lower part of a major collapse as opposed to the upper one represented by H1.04, 03. The excavator of H1 was of the opinion that all these tumble loci in the bema area and directly north of it would have resulted in a simultaneous, single collapse. The history of the collapsed column in H1 remains thus unresolved, with the possibility that it might have survived the initial destruction. It may also be that the disturbances in the north apse area which happened in the following phase, are responsible for the ambiguous stratigraphic relationship between Squares G4, H3, and H1.

In the south aisle, drums and the capitals, were much rarer. A concentration of several drums was found in B1.07, but with no evident pattern of collapse. These drums probably belong to one of the southern columns which, during the tremor, twisted slightly. Its upper part collapsed northward while the central and lower sections fell backward. Neighboring B2 contained a drum deposition of the same character in loci 06 and 05. However, the squares of the south aisle did not lack stone material of other types. Numerous ashlars and the rubble material from the interior of Wall A were densely packed in loci A1.14, 13, A2.06, B1.07, 06, B2.06, 05, and C1.13 (upper) (Fig. 2. Section h-h'). The fallen stones often preserved a fanning-out pattern spread from Wall A, to NE and NW. The depth of these deposits did not exceed 0.5 m, and usually ranged from ca. 899.9 to ca. 900.5 m. It is noteworthy that the collapse layers tended to slope downward from Wall A. The reason for that was the depth of the pre-fall deposits, which represented installations against the wall and the redepositions of Phase IX. Three coins were found in A1.14-13, none dated beyond the 4th century A.D.. Generally, the recognition of two or more major episodes of architectural collapse in this entire area is difficult. The uppermost layers seem to be better related to the continuous natural deterioration of Wall A. It may be that the large part of that wall had fallen already in Phase X. An exception may be Square Al, which appears to display two separate major collapses (loci 14, 13, and 09. 08). - Fiema et al (2001:106-107)

Uncertainty was already expressed as for the manner and removal of the bema's and central apse's marble floor. The nature of deposition of the lowermost layers in this area (e.g. A3.18, 16, H1.10, 08) which cover the mortar bedding of the removed floor is also unclear. Directly above these strata were stone tumble loci A3.17, 05 = H1.09, 05 = G2.14, 12 = F4.14, 12, which occupied the major part of the chancel platform and its western steps, from ca. 900.2 m to ca. 900.6 m (Fig. 2. section a-a' and Fig. 3. section b-b'). These substantial stone deposits tended to slope to the west and south. All contained very high numbers of roof tiles, and wall mosaic fragments as well as loose tesserae. The fragments of the marble furnishings of the chancel were equally numerous everywhere, forming a solid layer of marble within locus H1.05. All the aforementioned loci were, in turn, covered by an equally massive tumble at a higher level. These loci included H1.04, 03 = A3.04, and G2.08, 07, 06 = F4.08, 06, the latter set also covering a substantial part of the apse's interior. The depth of the combined lower and upper collapse loci did not exceed 1 m, and the top of the upper collapse was at about 901.1-.2 m. Although the upper collapse could reasonably be distinguished from the lower one, there were no clear intermediate layers of tumble-free soil between them. This could lead to the conclusion that all collapse layers mentioned here represent a simultaneous deposit. However, this situation may not be as obvious as it appears. Farther east, in the apse area (G2,F4), the evidence allowed the separation of stone deposits into different episodes, and also to note possible disturbances. Square F4 contains the intact southern half of the synthronon while the other half had been totally removed from the area of its northern counterpart (G2). The apse's semicircles was filled with a deposit (F4.10=G2.10) of debris including stone, mortar chucks, window glass and wall mosaic fragments. Locus F4.10 rested directly on the mortar bedding (F4.15) of the removed apse's floor. In the north, G2.10 covered not only the remains of the floor's mortar bedding (G2.15), but also appeared to spread occupied by the northern part of the synthronon. There, G2.10 was over locus 13, which in turn superseded locus 17. Locus G2.17 featured stone blocks in a clayish soil which created the elevated surface for the entire semicircle of the apse, while locus 13, above it, probably represented debris and rubble from the actual removal of the synthronon. Both loci produced mosaic fragments, tesserae, glass, mortar and plaster, but no fire-related debris.

This entire area, up to the middle steps of the preserved part of the synthronon was covered with tumbles F4.08, 06 = G4.08, 07, 06 which also filled the easternmost strip of the bema along with the apse. The bema (=western) section of these tumbles was a relatively dense accumulation of stones, mostly ashlars. But the eastern section - the apse, the steps of the synthronon, and the area where that installation was not preserved - presented a substantially different image.Although both sections seemingly maintained a stratigraphic relation, the eastern section's layers contained more sandy deposits, some irregular rubble material, practically no ashlars , and were generally of such low density as not to warrant a tumble designation. Yet, these loci contained substantial quantities of wall mosaics and tesserae. That phenomenon is probably due to several factors: the nature of deposition in this area; the general impact of the first earthquake in this spatially restricted area; and the probability of later human interference. It is obvious that a stone collapse from above would not deposit easily on the narrow steps of the synthronon in Square F4, but would largely tumble farther west toward the bema where it finally rested. However, if the northern part of the synthronon had been removed before the first earthquake, and the semidome completely collapsed during the first tremor, the area of G2 would provide a good horizontal surface for stone material to accumulate in dense deposits. This is evidently not the case in G2.

First, it seems improbable that the initial earth tremor caused the complete collapse of the apse's semidome. If it had, the mosaic fragments would have been largely restricted to the lowermost, relatively dense collapse layer, not only on top of the bema but in the apse as well. Instead, all stone tumble loci mentioned so far in the bema and in the apse area yielded substantial amounts of wall mosaic fragments and separate tesserae. Pockets of ash and charcoal pieces were also observed. Therefore, while some damage to its structural integrity is possible, the apse area must have survived largely intact. What followed was a series of collapses, some probably major, others presumably associated with the gradual deterioration of the structure.

The second part of the issue - whether the northern half of the synthronon was still in situ when the calamity befell in Phase X, or it was removed prior to that event — cannot be resolved with certainty, as the nature of deposition in Square G2, especially in Loci 13 and 10 is ambiguous. A total absence of burnt remains in G2.13, 10, which are otherwise typical for lowermost loci in the neighboring F4, seems to eliminate Phase VII as a candidate for the removal time. Had the north synthronon been removed in Phase IX, the presence of marble fragments in locus G2.10, over the place where the removed part had been, is difficult to explain. If, however, the northern half of the synthronon was still intact, and, like its southern counterpart, had survived the initial tremor in Phase X, the confusing deposition noted in the place of its removal becomes more understandable. Presumably, only in Phase XI, some rubble from the initial damage which had accumulated on its steps and in front of it was cleared away into the eastern bema, before the actual removal of the installation. The depth of stone tumble there, as well as their density, is indeed considerable. Finally, the removal of the northern section of the synthronon seems to fit into the type of activities conducted in the north apse and Room I, both of which certainly happened only after the initial architectural collapse. It then follows that both deposits G2.13 and 10 in the northern half of the apse may be secondary, i.e. resulting from these human disturbances, and the tumble above them should be of a later date.

The proposed preservation of the central area, and probably of the lateral apses as well, through the first earthquake may relate to the specifics of other construction. Firstly the considerable thickness of Walls D, CC, and VV, and the sheer mass of relatively regularly laid stone fill behind the apse's walls (MM, E, AA) might have resisted the initial tremor. The central apse, being basically a separate, tower-like section of the church, was thus more flexible during the tremor, since it was also supported on the western side by the still-complete synthronon. Such amazing preservation of church apses and semidomes in the seismic arras of Syria-Palestine into modern times is not unique. For example, all three apses, including then semidome, arches, and sometimes even the upper rooms over the side apses are still preserved in the North and South Churches at Mampsis, and the South Church in Sobata Shivta. An opinion was voiced that inscribed apses are generally more resistant to the effects of earth tremors than protruding ones. On the other hand, the impact of the initial tremor on the bema area in the Pella church would have been more observable. Even if not all of the parallel pairs of the easternmost columns collapsed at that time, the effect of the tremor would still have been particularly devastating. The eastern clerestory walls above the arches would have fallen with a mass of stone deposited on the bema and adjacent areas of the aisles.

Deposits F4.08, 06 = G2.08, 07, 06 in the apse were, in turn, covered by the medium density ashlar tumbles F4.05, 04 = G2.05, 04, which also covered the uppermost preserved sections of the synthronon, and extended westward into the bema area. If the hypothesis above is correct, the apse sections of deposits F4.08, 06 = G2.08, 07, 06 represent natural accumulation and the slow deterioration of the apse area during the following Phase XI and beyond. The eastern bema parts of these loci, together with the central and western bema upper stone deposits already mentioned above (H1.04, 03 = A3.04), and the upper tumbles in the apse (F4.05, 04 = G2.05, 04) presumably represent later structural collapse(s), seemingly associated with the final fall of the semidome (Phase XII). The amount of wall mosaic fragments and tesserae decreased noticeably compared to the synthronon's lower tumbles, and practically no burnt deposits were found. - Fiema et al (2001:107-108)

The image of the initial seismic destruction in the south apse is far from uniform. Locus F2.17 (lower), noted above, contains great quantities of material which could have been redeposited during Phase IX. One element of the original marble furnishing of the apse - the marble colonnette (F2.37) which had supported the altar — survived intact and continued to stand through the postulated acts of vandalism in Phase VIII and a potential redepositing in Phase IX. The colonnette is ca 0.95 m tall, with its top at 900.25 m. Its survival was possible only because the area around it was already filled up when a major collapse came. Otherwise, the colonnette would have been knocked down. The top of F2.17 was at ca. 900.2 m, which is just high enough to preserve the colonnette intact. Only a few stones came loose during the initial tremor, collapsed and became embedded in deposit F2.17 (lower). The loci above, F2,13 and 10, which contained relatively large ashlars, represent either a gradual deterioration of the structure of the apse and/or later tremors. The tumble continued farther up through loci 08 and 07. Large quantities of glass tesserae and wall mosaic chunks were found in all these loci The extant evidence does not warrant the suggestion of a complete collapse of the apse's semidome during the initial earthquake. Since the tumble loci did not display a clear subdivision, it is postulated that this area experienced a series of minor collapses related to several tremors or/and gradual decay.

The sequence of the pre-Phase Xl deposition in the north apse cannot be easily reconstructed. The early deposition inside the apse, which should also include the by-products of the initial earth tremor, was substantially disturbed by activities in Phase XI — mainly the excavation through the mosaic floor (infra). Locus 29, the lowermost sandy but compact material in the apse and which contained some burning, was ca. 0.05 m thick. This locus partially represents the primary deposit related Phases VIII and IX, but it was substantially disturbed in the following phase. Locus G4.26 was directly upon it: it was a very low density tumble with no clear pattern. Locus G4.26 filled the apse at the same level as G4.25 in the front of the apse, loci being separated from each other only by the much distributed remains of the chancel screen. Yet, the composition of these loci differed considerably. While each contained many glass tesserae and wall mosaic fragments, locus 26 lacked the traces of burning, charcoal and decayed plaster characteristic of G4.25. The difference is so striking that while stratigraphically equal, these two loci may represent two separate and unrelated episodes. Perhaps locus G4.26 reflects the process of natural decay of the apse's structure which happened after the initial, probably insignificant debris of the destruction in Phase X had been cleared during Phase XI. Had the apse's structure, including the semidome, indeed collapsed in Phase X, one would expect a multitude of wall mosaic fragments associated with a particular tumble, and one would expect to find them mainly inside the apse area. Instead, marble, tesserae, and mosaic chunks continued floating in this area in strata 1 m above the floor level. All that should indicate that the deterioration of the apse was gradual and long-lasting, and that the considerable disturbances in the area most probably occurred only after the initial earth tremor. - Fiema et al (2001:108)

The lowest collapse layers in Room I that may be associated with the initial earthquake destruction were the combined loci I.10 and 09, with the bottom at ca. 899.3 in and the top at ca. 900.1 m. The bottom of I.10 rested on an dense layer (I1.11) of paving stones, ca. 100 fragments. The majority of the pavers had collapsed during the fire when the wooden ceiling burned down. The I.10, 09 tumble was extensive and of high density. Patches of ash and charcoal bits were noticeable in places. A late Byzantine nummus (A.D. 491-565) was found in I.09. The most significant occurrence in Room I during the first earthquake was the collapse of the arches, or at least the western one. That arch was somewhat precarious and irregular, being supported by a pilaster against Wall HH on the north side, and practically springing out of the narrow Wall G on the south. Notably, the thick pilasters, including the northern one for the western arch, are all preserved to a considerable height. On the other hand, the upper part of Wall G collapsed early, creating a curious window-like gap. The bottom of the gap is at 900.35 m, i.e. ca. 1.4 m above the floor of Room I, which is probably the level from which the arch springer began. The emergence of this gap is important in the history of the church as it created a direct connection between Room I and the north aisle of the church.

The earliest collapse inside Room II should be associated with loci II.09 (upper) and 08. Both loci were stone tumbles, but without any discernible pattern, and of variable density and localization. Due to the difficulties in dating the removal of the room's floor (supra), the association of these loci with Phase X will remain uncertain. The initial stone fall deposits in Courtyard IIIA and Portico IIIB were IIIA.06=IIIB.09 (upper), which were directly followed by IIIA.05=IIIB.08, seemingly of the same episode. The total depth of the entire tumble, set in a matrix of very sandy soil, was ca. 0.5 m. The obvious difference between the two entities is the amount of paving slabs. Seventy pavers were found in the combined loci IIIB.08 and 09, but none in Courtyard IIIA. The presence of the paving slabs in IIIB confirms the existence of a upper floor gallery above the Portico.

Generally, all stone collapse strata in both courts, which are more than 3.5 m deep and of considerable density, contained large numbers of column drums and fragments (Fig. 119). Surprisingly, while the lowermost loci displayed their share of that particular material, the drums were most abundant in the middle-level loci which were at least 1.5 m above the pavement, and higher. This indicates that, except for few columns, the Portico's colonnade did not collapse completely in Phase X. All the collapse deposits in the courtyards were difficult to separate, because of their considerable density, a density which seldom varied from layer to layer, and an almost unbroken continuity from top to the bottom. The examination of matrices was only a little more helpful. These were predominantly very sandy, and only occasionally displayed a more loamy texture. One reasonable indication of possible breaks between the collapse episodes was the contents of the strata, i.e. the amount and nature of cultural material. Yet, in such a structurally "perforated" matrix, infiltration of small sherds from upper to lower strata is inevitable. The dating of these stone deposits is practically impossible, although the observations above indicate that the middle and upper layers could have originated in Phase XIIA. These stratigraphic difficulties are equally applicable to the deposits inside Rooms I and II, except for the column drums. These rooms did not have columns, save for the small loggia colonnade in the upper floor of Room I. Yet, drums were found inside these rooms, mainly in the middle and upper layers. Some could have tumbled down naturally from areas higher up on the slope of Jabal Qabr Jumay'an. - Fiema et al (2001:108-109)

There is clear evidence in the form of paving slabs for earthquake damage in the atrium. This material was overwhelming in the lowermost tumble layers in the atrium, which were largely restricted to the area of the porticoes. This indicates that the upper floors of the porticoes collapsed during this phase. For example, to these initial collapse layers in the eastern Portico belong loci K3.17, K1.12 (upper) and 11, C3.06, 05, and C1.21 and probably 16 (Fig. 2. section h-h', Fig. 20, Fig. 117). The Square C1 deposits were rich in numismatic finds. These included two Early Byzantine folles and one Late Byzantine nummus in C1.21, and three Early Byzantine coins in locus 16. The collapse loci tended to slope away from the walls. Many stones of these tumbles rested directly on the floor, but the average range of the tumble was from 899.8 m to 899.5 m. The lowermost location was usually filled with sandy, slightly ashy loci. e.g. K3.19 under K3.17, C3.07 under C3.06, and C1.24 under C1.21.

The evidence from the northern Portico has a different pattern. Squares covering areas directly south of Wall XX yielded practically no paving slabs. The lower loci there were K4.13 and 08, L3.08 and 05, and L4.09 and 08, with the first locus being a sandy deposit on the floor and the second being an ashlar tumble of rather low intensity. The pavers of the north Portico occurred farther south, in the squares which cover the stylobate and the north-central part of the atrium. There the deposition pattern was similar with K2.03 and L1.03 being the bottom, loamy soil loci, covered by the stone tumble of K2.02 and L1.02. Out of a total of 31 pavers or fragments in L1, 25 were found in L1.02. The figures for K2.03 and 02 were 29 and 26 respectively. This deposition can probably be explained on the basis that the force of the earthquake lifted the uppermost parts of Wall XX and the Portico floors and threw them southwards. The stones and the paving slabs landed some distance from the wall, mostly in the area of the stylobate.

There was no scarcity of paving stones in the areas of the western Portico. Of the 59 pavers and fragments found in square L2, 32 came from locus 05 (upper). This was the lower part of a considerable stone tumble which continued in locus 04 where the stone deposit was scarcer, but which contained another 13 pavers. Large quantities of paving stones were found in the lowermost tumble D4.35 The top of this locus (D4.35) was at ca. 899.7 m, i.e. ca. 0.4 m above the floor. Paving stones were also found in the neighboring square D3 where locus 04 had the stone tumble localized close to the western stylobate. In Square D, which covers the corner between the western and southern porticoes, loci 48 and 45 represented the initial fall of the porticoes and Walls N and YY. These loci also included Sandy deposits 47 and 46, a total of ca. 0.55 m in depth. Paving stones were common in D2.48, but also represented in the other two collapse loci. The pattern of their collapse was from NE to SW. The great majority of pottery from D2.47 belong to the 5th-6/7th century, and included possibly later 7th century sherds. The few Late Hellenistic Attic Black Glaze sherds (2d century B.C.) may have come from the rubble of the walls' core.

The early collapse is also featured in the southern Portico. Notably, that area offered the least depth in excavations, often there was only a little more than 1 m between the Portico's floor and the modern surface. In C2, where 48 paving stones or fragments were found, the pattern already described was repeated. Sandy locus 06, with some ash patches, contained 13 paving stones and fragments, either floating in or on the top of the locus. The stone tumble (C3.04) was above and it yielded 23 pavers. The collapse of the Portico upper floor was particularly in evidence in square Dl where locus 11, in addition to ashlars, provided 80 of the 85 pavers or fragments found in the square (Figs. 120-21). This spectacular stone tumble extended from ca. 900,3 m down, reaching floor level (ca. 899.5 m) in some places. The tumble was embedded in the lowermost sandy stratum D1.12. The latter contained a multitude of cultural material (bone, glass, ceramics) presumably related to the occupation of the atrium in Phase IX.

Everywhere in the area of the north atrium, the lowermost soil deposits consisted of a very loamy soil with a small gravel content. Perhaps they reflect a minor flooding episode which occurred in the atrium after the earthquake. Another interesting observation concerns the column drums from the atrium porticoes, which were nowhere found in clusters or in aligned rows as in the church proper. In the central area surrounded by the stylobates and covered by squares C4, K2, D3, and L1, where one would expect many capitals, and drums, there were many decorative capital fragments (109), but only seven drums or fragments. The porticoes were equally poor in drum finds. Presumably, the atrium was roughly cleared of easily movable material such as drums in the following phase. - Fiema et al (2001:109-110)

The interior of Room XI presented a particularly bewildering multilayered stone tumble. This heavy and dense mass of stone must he related to the unusual thickness of the southern and western Walls of the room. The tumble, ca. 3 m high, was difficult to subdivide stratigraphically except by rough and arbitrary means. The only reliable indication of breaks between the strata were changes in contents and amount of cultural material. The initial collapse was represented by the lowermost tumble loci D2.23 = D4.20 = E3.22 = F1.22, which also contained paving slabs. Notably, however, this material was not restricted to these loci. Pavers were found floating in most of the tumble layers inside the room, many as high as 2 m above the floor. Since the slabs in Room XI would have been used as pavement on the upper floor, this distribution is puzzling. The first major collapse inside Room IX is represented by loci M3.05 (upper) and 04. The latter having its top slightly more than 1 m above the floor. No patterning of the tumble was recognized. The tumble also contained some paving stones, undoubtedly from the upper chamber's floor. Several pavers were or marble and one of alabaster.

Room X presented much clearer stratigraphy. Substratum E3.30D was covered by 30C, a loamy silt, and then by 30B, characterized by a matrix of fine sand. The top E3.30B was about l m above the floor level. Neither 30C nor B are occupational, but relate to the first seismic-related destruction in Phase X. Stones floated inside 30C and increased considerably in 30B, so that the upper part of this substratum was a tight mass of stones. The matrices of both substrata resulted mainly from silt which filtered down through the layers of collapsed stone. Pottery was scarce in these strata, but four coins were recovered from E3.30C. These were a 4th century piece, a coin of Constantius II (A.D. 351-354), a mid-4th century piece, and an unidentified coin.

Paving stones, presumably from the floor of the upper chamber, were found in both 30C and B. Some were of considerable size, up to 1 x 0.6 m. A large accumulation of ashlars was observable along Wall TT, while the pavers were concentrated in the area between that wall and the baptismal font. The pattern of the pavers' concentration indicate that the ceiling over the part of the room that had been expanded westward in Phase V collapsed first, probably because it lacked the support of the canopy columns in the central-eastern part of the room. Each of the substrata contained a few of the arched segments of the canopy as well as the large voussoirs of the arching which presumably supported the roof above the upper chamber. Since column drums were relatively scarce in 30C and B, it is possible that a section of the canopy, two or three columns, and perhaps the eastern parts of the room's superstructure, survived the Phase X tremor. Locus 30B was followed by more than 0.5 m of natural accumulation (E3.30A). Above that locus, the remaining columns of the baptistery collapsed during the second earthquake (Phase XIIA). Canopy arch elements were also found outside of Wall TT, which may indicate that some of them fell during the earthquake of Phase X but others (including still standing columns) could have fallen at a later date, when the badly damaged Wall TT was already down to the level which allowed elements to land outside Room X. That observation corresponds to the postulated east-to-west orientation of debris that resulted from the second earthquake. - Fiema et al (2001:110)

In addition to damage already noted to the south mosaic, damage caused by excessive wear and the rough handling of objects before the fire (some of which had been repaired in antiquity), some areas of both the north and south aisles have damage that could have been caused by an earthquake. That damage is particularly visible in the east part of the south aisle, and in the western half of the north aisle, but is not restricted to these areas. A project conservator reported:

... There are a number of areas where the mosaic has undergone compressive forces and ... has been lifted from its mortar bedding creating voids under the tesserae. In a few cases the buckling has provoked the loss of tesserae, while in others an intact ridge has been created. There are a couple of examples where a section of the mosaic has been lifted up along a fault line creating a difference in level along the fault of up to 10 cm. Some cases of cracking of the mortar bedding have been observed, with a resulting gap between tesserae. This cracking is likely due to differential settling of ground and pavement foundations. The buckling, shearing and cracking phenomena are probably due to earthquakes in antiquity. Fallen blocks and paving stones have created three kinds of localized damage to the mosaic floor. In several cases the mosaic has been crushed, with all the tesserae still in place but detached from their mortar bedding, and the individual tesserae broken into fragments. In other cases the fallen stone has actually wedged itself under the tesserae, separating them from their bedding. In several cases, the weight of the fallen stone has caused the whole mosaic to sink, creating a depression in the floor. ...

Little is known about the area south of the church complex in Phases VI-IX, and even less can be reconstructed on the basis of extant remains. It is not known if the domestic occupation continued beyond the line of Walls GGG, HHH, and Z. To recapitulate, the foundation trench for Wall A had been dug out from the level of ca. 899.5-6 m. The bottom of the lowermost stone collapse observed in the area was at around 900.5 m. This means that almost 1 m of deposition had accumulated against the south face of Wall A between Phase IV and Phase X. The strata in this vertical bracket were commonly thin, some displayed hard-packed surfaces, others were silt. The occupational character of the former can neither be proven nor rejected. However, they contained some sherds, bones, wall plaster, and sometimes, tesserae, which indicates at least casual presence of humans.

The loci which represent that interval can be treated together, as no particular pattern or logical connection seems to occur between neighboring squares. The ceramic material from these loci was commonly assigned to the 2d through the 6th century A.D., with the latter date being predominant. However sherds later in date occurred as well. For example, loci A1.11A, B1.05, and C1.08 all produced 7th century sherds and a few that probably date to the early 8th century. Coins were also found in these strata. Locus F2.19 yielded an Early Byzantine follis (late 4th-early 5th century A.D.) and a nummus of Anastasius (A.D.491-98), while a Late Byzantine nummus (A.D. 491-565) was recovered from F2.18. A Late Byzantine nummus of Justinian I was found in A1.12, and a mid-4th century follis in B1.09. Locus C1.09 provided four coins: two of mid-4th century, one of late 4th - early 5th century and a coin of Justinian I.

What can perhaps be interpreted as a multi-episode collapse, was noted in Squares F2 and Al. The lowermost tumble (F2.14, 12, A1.10, 07) noted in this area was deposited at ca. 900.4-901 m (Fig. 3, section b-b' and Fig. 6). It did not display any particular pattern. At least in Square F2, the tumble was in turn covered by another one which reached 901.45 m. Unfortunately, it was impossible to assign these stone deposits to any particular phase. These might have originated in Phase X or later. - Fiema et al (2001:111)