special type of architecture, has remained unrivalled and unapproached since it was erected. It is in this light, as the central and dominant production of Byzantine architecture, that St. Sophia has been chiefly considered in the most recent of the numerous essays of which it has been the subject, the work jointly produced by Mr. W. R. Lethaby and the late Mr. Harold Swainson, the title of which stands at the head of this article. While, however, making the study of Byzantine building, as illustrated in St. Sophia, the most important object of their work, the authors have also endeavoured to put together in a readable form the substance of the accounts given of the church by contemporary writers, especially Procopius and Paul the Silentiary, giving translations of the most important passages from these writers, and thus popularising the available knowledge as to the appearance, decoration, and furnishing of the church as it was carried out by Justinian. How far these accounts, filled with no little rhodomontade and obvious exaggeration, can be accepted as testimonies of fact is another question; but they are of interest in themselves as evidence of the extraordinary fascination exercised by the building on the mind and fancy of contemporaries, and the attitude of wonderment with which it was regarded; and probably the descriptions of the richness of the decoration and furnishing, however highflown they may appear, were made from observation, and may be supposed to represent what was really there, if we take the trouble to disengage the facts from the poetic rhapsodies in which they are entangled. The construction, however, must be studied from the building itself, and by the light of modern research, for the contemporary writers, Procopius and the Silentiary, obviously knew so little about it that they cannot describe the constructive difficulties or devices in any intelligible manner, though the Silentiary sometimes has a forcible and effective way of putting things, as when he mentions how the archi

structive problem, since it is erected on a circular substructure, so that each portion of the springing of the dome has a solid support beneath it.

It is melancholy to have to record that Mr. Swainson, the junior author of the two, who devoted himself chiefly to the consideration and translation of the ancient accounts of St. Sophia, has unexpectedly and prematurely died abroad, since the publication of the book, in the course of a tour in which he had intended to devote himself especially to the further study of the origines of the Byzantine mode of domed construction.

tect 'gave the walls strength to resist the pushing arches, 'which were like active demons.' He is not so absurd, however, as Procopius, who flounders in the most extraordinary manner when he gets on building construction, and whose habit of attributing the solution of constructional difficulties to the personal perception and interference of Justinian is obviously a mere piece of blatant flattery.

The existence of Justinian's St. Sophia we owe indirectly to the insurrection against him which arose out of the fight between the 'greens' and 'blues' after the chariot races in January 532, when some persons of both factions were executed by Justinian's orders, and the two banded together against him, besieging him in his palace, destroying a great deal of the city, and burning the basilica which had been erected by Constantine on part of the site on which Justinian's church now stands. After the insurrection was quelled and the proposed new emperor, Hypatius, got rid of, the scheme of the rebuilding of the church on a magnificent scale seems to have been entered on immediately, under the direction of two Greek constructors, Anthemius of Tralles and Isidore of Miletus. Anthemius, it is recorded, made, in the first instance, a model of the proposed building, which was followed in carrying out the design, and we can therefore feel little doubt that he was the ruling spirit of the work, and that it was to his genius that the result is due. Justinian can merely have the credit of having supplied the funds (obtained in the manner hinted at by Gibbon), of having stimulated the production of a great work by his ambition to connect his name with it, and of having known how to choose the right man to carry it out; the latter certainly not an unimportant quality in a building-owner' on a great scale.

What was the nature and design of the previous church, erected by Constantine, we can only conjecture; the only probability of our ever knowing anything more about it lies, as our authors suggest, in the chance that an examination of the cisterns beneath the floor of the present church, if it were ever permitted to be undertaken, might furnish evidence as to the plan and some of the materials of Constantine's building. Two conclusions we may certainly deduce, however: that it was much smaller than Justinian's building, and that it did not present the same construction; otherwise there would have been no ground for the admiration expressed by the contemporary writers for the size and height of the latter, for their remarks as to the happy

fortune by which the destruction of the first church gave opportunity for the erection of so much more splendid a building, or for their bewilderment and astonishment at its method of construction. On this latter point they are especially emphatic, and might well be, since nothing in the way of domed construction in so bold a method and on so large a scale had ever been attempted in the world before. In this sense St. Sophia was to the sixth century what the Forth Bridge has been to the nineteenth. Each structure was an instance of the application of a system of construction never before attempted on so large a scale; and in difficulties of construction scale is an important element, because while in a design on a large scale the pressures or tensions are immensely increased, the molecular constitution of the materials, their resistance to crushing or to extension, the cohesive power of mortar or cement, &c., remain the same in each case; so that a form of dome which it may be child's play to construct over a span of 25 feet becomes a matter of difficulty and anxiety in regard to a span of 100 feet or upwards.

Let us endeavour to make clear the nature of the constructive problem, the solution of which was the glory of Anthemius and the astonishment of his contemporaries. The plan of the main portion of St. Sophia, omitting the porch or narthex, presents a parallelogram of nearly 250 feet in length within the walls by about 230 feet in width. As far as dimensions, therefore, are concerned, it is in width only that it is remarkable. It represents, in fact, the nearly square type of plan, with a wide central area, which became the special form of the Greek church, as the long plan with parallel aisles, the form of plan of the earliest Christian churches, came to be considered the special type of the Latin church. The central portion of this space, for a width of about 106 feet by a length of a little over 200 feet, is entirely open to the roof, the floor being unencumbered by piers or supports of any kind, except the slight projection on to it of the pilasters from which spring the two great transverse arches, about 102 feet in span. The central portion of this open space is marked out into a square by two great arches which fly right over the centre space, transversely

*

*There is a good deal of discrepancy in various published statements as to the precise measurements of St. Sophia. We take the dimensions here from the plan drawn to scale on page 38 of Messrs. Lethaby and Swainson's book.

across the church, forming at the top, with the similar lateral arches on the side walls, a complete square. Over this square space is built the central dome, of remarkably flat section as seen externally, though internally it is not so far from a complete semicircle in section as is sometimes supposed. The portions east and west of this square space are roofed by semi-domes, which abut against the exterior sides of the great arches that carry the central dome, and curve downwards towards the two ends of the building. Thus the general internal appearance of the central portion of the building, to a spectator looking up, is that of a square central space covered by a circular dome, and lower semicircular spaces to east and west of it covered by halfdomes, which rise up to and meet the line of the great arches that carry the central dome, and appear to assist in supporting the latter. The arches at the north and south side of the central square are not open, but are partially filled in solid by a wall with smaller arcades opening through it on two galleries in the lower portion, and pierced with windows in the upper portion.

But the crux of the problem lay in the fact that the central space between the great arches east and west and the partially filled arches on the north and south is a square, and the base of the dome built over it is necessarily a circle. If we describe a circle within a square, the circle touching the four sides of the square, we find, of course, a very considerable space left between the circumference of the circle and the angles of the square. What becomes of that space in the case of putting a circular dome over a square sub-structure? That was the question which builders in Asia Minor and elsewhere had been practically putting to themselves, in a desultory manner, for some centuries previously. As already observed, it is an easy problem to build a circular dome over a circular substructure; that is merely turning all the walls over in an arch towards the centre. But a circular circumscribed space on plan is often not wanted on the ground floor of a building; the floor is required to be open to the angles of the space; while the dome must be circular on plan, because constructively it is part of the essential quality of a dome that its masonry forms arches in a horizontal sense as well as in a vertical sense; as soon as a horizontal ring or course of the masonry is complete it is in a condition of stability and cannot fall in; a dome on a circular plan, unlike an arch, which requires the keystone to complete it, can be left half finished and be as stable as when entirely

covered in, or (if it is a dome built on scaffolding) the scaffolding can be altogether removed from the lower portion, which can be used for supporting the centreing on which to build the upper portion.

We have, therefore, to bridge over the approximately triangular space left between the circle of the dome and the rectangle of its supporting walls; and though we have very few examples of the manner in which this problem was progressively tried at and at last solved, there are a few remains, chiefly in Persia, which show that the attempt to do it had been going on for at all events two or three centuries before the foundation of St. Sophia. These examples of rather small domed structures over a square plan are found among the remains of buildings attributed to the Sassanidæ, the dynasty which came into power in Persia in the course of the early part of the third century A.D.* Such an example is found at Serbistan, or Sarvistan, for instance; and here the difficulty is got over in a very simple and straightforward way, by building what are called, in architectural parlance, 'squinch arches across the angles of the building; short arches built from wall to wall at the angles, parallel with the diagonal of the square, and carried further and further out as they go up, till the square becomes an octagon, upon which a circular dome can easily be raised with only a little fudging' of the surfaces.

,

* M. Dieulafoy, in his "L'Art antique de la Perse," attributes these Persian domed structures to the Achæmenidæ, whose power lasted from the eighth to near the end of the fourth century B.C, and this conclusion appears to have been accepted by M. Choisy in France and by Professor Aitchison in England. We are not disposed to accept it, and regard it rather as an example of the tendency, too common among French archæologists and explorers, to magnify the importance and novelty of their discoveries. Two distinct reasons may be alleged against it; one, that if the problem had been attempted so far back as M. Dieulafoy makes out, more progress would have been made by the time the Byzantine empire was founded; the second and more important reason is that at Persepolis, which is the site of the principal remains of ancient Persian architecture, there appears to be nothing to give the slightest indication of any domed structures having existed; there are no remains of piers such as could have supported a dome; the architecture was manifestly columnar; and it seems in the highest degree improbable that two different styles of building should have been carried on simultaneously in the same country and within fifty miles of each other. In other words, if the domed remains at Serbistan and elsewhere had been built by the contemporaries of the Persepolitan palaces, we should have found remains of domed structures at Persepolis also.