If one were to judge by the remains, he would conclude that church building in Bosra, with the sole and very notable exception of the Cathedral, was scanty and poor in comparison with that in many places in Southern Syria which were far smaller, especially with that in Umm idj-Djimal with its fifteen independent churches. Nevertheless the remains of church architecture in Bosra, even excluding the Cathedral, are more significant than the vestiges of ecclesiastical architecture at either Amman or Djerash.
The only ruins of churches that I could discover in Bosra are three in number, all ruins of basilical churches. It will be observed that I have eliminated the Basilica (Dêr Bohera) and the apsidal building near it from the list of churches, on the grounds that their details are more closely allied with those of the theatre and the arches than with those of the Cathedral and other buildings which are known to have been churches, and that all Christian symbols are wanting in them except in the form of graffiti. Even if these buildings were finally employed as places of Christian worship, it is not necessary to consider them as examples of church architecture. On the whole it seems probable that the basilical churches are earlier than the Cathedral which was a central domed structure, and which is known to have been a building of the beginning of the sixth century. There must have been church buildings in Bostra before the end of the fourth century, and there must have been a Cathedral church before the present structure was erected, either on the site of the present building or somewhere else. The three much ruined churches described below might have belonged to any century, from the fourth to the sixth. They have no details that are significant of any particular period, although they have features that are not without interest. All are such structures as one would expect to find anywhere in Southern Syria; only one church, No. 3, has any pretensions to size.
 

 still to be seen in foundations, one on either side of the main aisle. These piers must have carried, in addition to the nave arches, transverse arches over the side aisles, for which there are responds in the side walls, and a high transverse arch over the main aisle. This arrangement of the plan separates the east end of the main aisle off from the rest of the nave, as a square bounded by high broad arches at the east and west and by two low and narrow arches on either side, so that one division of the main aisle corresponds to two bays of the side aisles, a plan further developed in the church Architecture of the Romanesque period, especially in Lombardy where a high cross-vault was placed above each such grand division of the nave with tunnel vaults or two small cross-vaults above the flanking divisions of the aisles. If the second set of supports from the west end had been also cruciform piers, instead of columns as they certainly were, we should have had a plan substantially like that of many of the more highly developed Lombard churches in which the alternation of light and heavy supports was either the cause, or effect, of a remarkable system of vaulting.' There was probably no thought of vaulting < in this case any more than in a very similar example that I found at it-Tuba in Northern Syria. A fragment of one of the cruciform pier-caps found in the nave shows mouldings which are given in the drawing (111. 246). The profile of this cap is interesting as Christian work of Southern Syria, and it is quite certain that the detail was not taken from some pre-Christian building, for no such profile is known to have existed in the Pagan architecture of the Haurân.

CATHEDRAL. DATE, 512-13 A. D. The Cathedral of Bostra was the largest church in Southern Syria and one of the most important in all Syria from an architectural standpoint. It was moreover, at the time of its erection, one of the largest domed churches in the world. It is now almost completely obliterated, and has for years contained the great public dung-heap of Bosra, which has risen to the- height of the outer walls, and which from time to time has burned for weeks by spontaneous combustion, leaving the interior a huge mound of ashes. The outside lines of the whole building are still to be traced in the ruins, the walls of the east end are almost completely preserved; while sections of the side walls with both ends of the west wall preserve sufficient height to give a few important measurements; but a greater part of the edifice as M. de Vogue saw it over fifty years ago has been ruthlessly pulled down.
This great church has attracted the attention of many visitors; Rey published a ground plan of it, and M. de Vogtie gives a long description with a restoration in the text and two plates, one showing the ground plan, the other the actual state of the church at the time of his visit. Since these publications many writers on Byzantine architecture have discussed the building, reproducing these plans and suggesting different restorations. Of late, Professor Brunnow has republished M. de Vogue's plan and has presented a new restoration of the church based upon it together with three photographs illustrating its present condition.
All the published restorations that I have seen thus far are unconvincing in some particulars or others. M. de Vogue, though proposing to use the church at Ezrac (Zorcah) as a basis, introduces an hemispherical dome in place of an elliptical one with its major axis vertical which is the most important characteristic of that well preserved dome. Professor Brunnow also uses an hemispherical dome, but makes it smaller than M. de Vogue's, supporting it upon two almost equal storeys of eight arches each, set on piers, so that the lower arches are lower and much narrower than the apse arch, which produces a serious architectural blunder at the point where the arch on the east side of the octagon opens toward the great arch of the apse. I noticed that there were remains of a lower as well as a higher arch at the entrance to the apse; but this is certainly a later addition, and belongs to the late church erected within the rotunda after the fall of the original dome. All restorations of the dome based upon Rey's restoration of a circle of interior columns in single file are manifestly impossible; for such supports would not carry so large a dome.
I was so fortunate, on the occasion of my visit to Bosra in 1909, as to discover a detail which is of the greatest assistance to the restoration of the interior arrangements of the supports and of the dome itself: this I have applied to the solution of the problem in the restorations which follow. For these restorations I have drawn heavily upon M. de Vogue's invaluable plan, which could not be made today. I have profited by the studies and suggestions of former restorers, and have relied confidently upon the close analogy between this church and the domed church of Saint George  at Zorcah, which is hardly a day's journey distant, is only three years later, and with which I am well acquainted.
 Plan :  But before entering upon a discussion of the problem I desire briefly to describe the plan of the building. We have here (111. 248) a unique example of a central plan, i. e. a square enclosing a circle which, in turn as we shall see, enclosed an octagon. The interior circle is tangent to the sides of the square at doorways on three sides, and at the opening of the apse on the fourth. The solid angles between the two figures are occupied by broad niches or exedras (G, H, J, K) each as large as the apse of a small church. There are three doorways opening into the great circle in the middle of every side of the square but the east, and other doorways cutting through the outer walls into the exedras, which make five entrances on each of the three sides of the outer square. The thick walls are further relieved by smaller niches outside and inside, and by large windows which opened into the smaller' interior niches at strange and unequal angles, as well as by small circular windows which appeared above the doorways. From the east side opened the great apse, almost 9 m. wide, set back some 6 m. from the arch in the circular wall so as to provide a small choir. The apse appears on the outside as three full sides and two half sides of a decagon, and is flanked by two large side-chambers (D, E) which have doorways leading into the choir and into the great circle. This is all there was of the original building; but there were added at a later date, and are standing today in a very complete state of preservation, two chapels (C, F) beside the side-chambers of the apse, their outer walls continuing the lines of the side-walls of the great church, and their apses set on a line with the east walls ,of the side-chambers. These small apses are brought to a square without, and project beyond the polygonal wall of the great apse. These chapels were constructed in a manner wholly different from, and poorer than, the original church, they have tunnel vaults of masonry, and belong evidently to a declining period of architecture in Bosra.
M. de Vogue's plan shows the outlines of a small church built within the greater one after the fall of the great dome. This structure was joined on to the original choir and carried the lines of the old choir westward almost to the centre of the great circle. This newer building seems to have been very poorly built, and I think it not impro- bable that, with the two chapels "described above, it belonged to the same period of rebuilding. Bosra has a small Christian community today and probably has had ever since the Mohammedan conquest. It almost goes without saying that churches have been in use here, from time to time, during the middle ages and later;for the throne of the present Bishop of Bosra is located in a little church at Khabab in the upper end of the Haurân. It would be interesting to know if these three chapels, erected after the ruin of the great central church, were dedicated separately to the three martyr saints to whom the original was dedicated, Sergius, Bacchus and Leontius..
When I was in Bosra in the spring of 1909, there had been a fire in the dung heap within the building, by reason of which the mass of the heap had been greatly-reduced, and the comfort of working inside the church had been correspondingly increased. The natives, ever searching for building stone, had begun to dig in the ruins among the ashes, and had uncovered a loosely built wall which had been the north wall of the chapel in front of the entrance to the choir. Here they had laid bare two details, one of which, being in situ,  proved to be of  great

 with the assistance of M. de Vogue's drawing, to suggest a scheme for the restoration of the supports of the great dome, the second storey above these, and, by analogy with the dome of Zor'ah, the dome itself. There are sufficient remains of the walls of the lower storey, especially at the northwest angle, to give the height of these walls and to show how the deep exedras with their half domes were set into the solid triangles between the square and circle. There are also remnants of corbel courses in these outer walls to prove conclusively that the circular aisle was roofed with slabs of stone. The piers of the outer arch of the presbyterium, or rather the arch opening into the choir, give the height of this arch, and M. de Vogue's description tells us that the choir was tunnel vaulted with cut stone. The height of the apse arch is easily obtainable, and the apse wall with its windows is still well preserved. All these features are illustrated in sections in Plates XVI. and XVII.Beginning with the newly discovered pier at (Z) on the plan (111. 248) as a basis, and restoring the full number of eight such piers, it becomes plain that the great arches of the octagon must have sprung from imposts higher than those of the arches which spanned the aisle; first in order that the arch on the east side of the octagon should be equal in height to, or higher than, the choir arch, and, secondly, that the aisle arches could carry the slabs of the aisle ceiling which was also the floor of a gallery. Such a restoration throws the main arches up above the level of the lower storey, and makes them the openings of the gallery into the space below the dome. This requires that the main arches be subdivided with narrower and lower arches for the support of the slabs of the gallery, and the result is a plan not uncommon in later, Byzantine, domed interiors (Pls. XVI and XVII). For the supports of these inferior arches I have used a column based upon the fragment referred to above. This arrangement is, broadly speaking, the basis of M. de Vogue's restoration, though he has made the aisle wider and the arches of the octagon narrower, and he, omits the arches of the aisles. Professor Brunnow's restoration makes the aisle even wider, so wide in

fact that it could not have been covered with stone slabs. His main arches are so narrow and so low that he has divided the chancel arch into two storeys, and has carried the gallery across in front of the choir. This necessitates a second storey of arches above the main arches almost as high as the lower storey. Our restoration, following M. de Vogue's at this point, terminates the gallery on either side of the choir, for it would be impossible to imagine the triumphal arch concealed from view except from the gallery. The outer wall of the gallery was a great cylinder set upon the circle within the square below, its outer face flush with the sides of the square at four points in the middle of the sides of the square and flush with the arches of the four exedras on the diagonals. A part of this wall is depicted as standing in M. de Vogue's sketch, where several openings are shown on the south side and only one in the west front. I have followed the earlier restorers in placing windows at regular intervals in this cylindrical wall, assuming that some of the openings had been walled up, though this is not definitely represented in the early sketch.
The next problem is that of the dome, and here I have followed the design represented in the church of St. George at Zor'ah. The spandrils between the great arches are warped inward until a circle is formed at the level of the crown of the arches. Upon this is set a low cylindrical drum, a little lower in proportion than that at Zor'ah. The warping process might have been carried well up into the drum, and the drum may have been an octagon on the outside, but these are minor considerations. It is probable that a stone roof of slabs, even horizontally laid, would have given stiffness to the supports of the dome; but it is impossible to know whether this roof was of wood, as shown by earlier restorers, or of stone. I have often wondered why neither M. de Vogue nor Professor Brunnow adopted the tall, pointed, elliptical dome of Zor'ah in their restorations, since this is the only dome in the neighbourhood that has been spared, and is of very nearly the same date. This sort of dome could be built without centring and is easily supported. It will be seen that I have adopted it without any change save bv increasing its diameter.
Ornament . The Cathedral of Bosra could never have been beautiful when seen from without, in which respect it may be classed with the majority of Byzantine churches. The only interior carved ornament of which we have any record is that of the chancel arch of which M. de Vogue gives a drawing in his text. This, like the very similar carved cornice of the exterior east wall, may have been taken from some rather late monument of Pagan times. According to several early observers, the chief decoration of the interior consisted of wall paintings which seem to have taken the place of the mosaics of many later domed churches. But the absence of plastic ornament from the interior in its ruined condition is not a proof that such ornament did not exist in the original edifice. Much of it may have been in stucco as we know was the case in buildings of the Southern Haurân . The restoration of the interior, as shown by the two plates, provides at least one series of details that well might have added to the carved enrichments of the interior, and which would have disappeared. I refer to the columnar supports of the narrow arches embraced by the great arches of the octagon. The discovery of the fragment of a column of cipollino marble within the church is ot course no conclusive proof that the column originally belonged to the building, but, if the restorations herewith are to be accepted, columns must be provided, and it islogical to choose one found on the spot. It was interesting to find that this fragment of the upper quarter of a column shaft was similar in dimensions to the cipollino columns with beautifully wrought capitals which today carry some of the arches in the Mosque of Omar here in Bosra. Two of these standing columns bear an inscription dated in the year 488 A. D. The inscription records the erection of a taxovypv ai-ffttt, or sigma triconchos, i. e. a semicircle with three niches. It is worth noting that, if the date were not earlier by 24 years than the accepted date of the church , the Cathedral itself could offer a most suitable place for these columns: for each angle of the interior has its sigma, or semicircular exedra, with three conchas, or niches, opening from it. At the west side of the building two of the niches in each exedra are pierced with doorways: but on. the east side there is nothing to prove that the outer conchas were pierced, excepting the analogy of the others, and it is quite certain that corresponding interior conchas were walled up when the late chapels were added. In front of each of these exedras was one of the main arches of the octagon subdivided by two columns •, this would be the place for the two inscribed columns now in the mosque, and their dimensions are entirely suitable. It is by no means conclusive that the exedra and the columns could have been built, and the inscription set up, 24 years before the great inscription of Archbishop Julianos which was inscribed upon the lintel of the principal west portal of the church, and which recorded the completion of the building, but it is not altogether impossible. In any event it is interesting to find an architectural composition so accurately corresponding to the description given in the inscription. The east wall, which is so well preserved, embracing the polygonal wall of the apse and the end walls of the side chambers, is adorned with a richly carved cornice which, if carried around the building, must have added distinction to its appearance. This cornice (111. 249) is composed of a corona ornamented with perpendicular grooves like the end of the strigil ornament, a bead-and-reel moulding, and a sima carved with flowing anthemions and upright acanthus leaves. It has the look of having been originally an overhanging cornice of some Roman building, and the joints between the sections of it at the obtuse angles of the apse wall are placed together like second hand material. I regret to say that I did not look on the lower part of one of the stones to discover if it was" cut under as the soffit of a classical cornice naturally would be. The photograph (111. 249) shows the perfection of wall building and the height of surface finish in basalt occasionally reached by ancient builders in the Haurân.

EPISCOPAL PALACE :  Directly east of the Cathedral, on the opposite side of a wide street, are two groups of partly inhabited ruins which I believe to have been clerical residences connected with the Cathedral. The two groups are separated by a narrow street which ran toward the east (111. 250). The position of these buildings, and of the east and west street, is not exactly shown either on the map of the city, or in the plan of the Cathedral {111. 248): this is my fault, for I measured all the buildings, and the width of the street between them and the Cathedral, but made no record of their relative positions. There are modern buildings between,' almost filling the space between the Cathedral and the ecclesiastical residences. The plan of the residences is given with the plan of the Cathedral (111. 248). The building op the north side

of the narrow street, which has a large triconchos within, and which presents a curve at the street corner (111. 250), I have designated as the Bishop's Palace. Its ground plan consists of the triconchos mentioned above, composed of three deep exedras, or apses, opening upon a rectangular apartment with its straight wall toward the north. The angles between the apses, at the southeast and southwest are filled with rectangular rooms in two storeys, the lower spanned with transverse arches carrying slabs of basalt, the upper roofed in wood. The semicircles of the apses are concealed on the exterior by straight ,walls, the exterior curve seen in the photograph being at the angle of the southwest room. Northward from the western apse extends the domestic part of the residence, giving on to a court at the east.
The first division of this section of the palace is almost completely preserved in three storeys (111. 251) but beyond this it is in ruins. The south wall is perfectly preserved in one storey, as is the west wall, save for a space which was occupied by the windows of the west apse. All the half domes of the apses have fallen ; but the arches of all three are still standing (111. 252). These carry parts of a storey for light which consisted of three windows over each of the arches. The wall seen in the photograph at the east end of the eastern apse, is modern. The archivolts of the arches are richly moulded ; but the piers which support them have plain right-lined caps. It will be seen from the photographs, and from the description given above, that there are ample data for the restorations here presented (111. 253).