George Edmund Street: Unpublished Notes and Reprinted Papers

by George Edmund Street

The arches are generally built with small stones of the same size and of even number, so as not to allow of a keystone. M. Mallay says that the mosaic work in the walls of these churches had wide joints of red mortar, projecting from the face of the wall. These mortar joints in the restored work appeared to me to be a bad modern device, and I think that the evidence in their favor ought to be very strong to be convincing.

The proportions of these churches are very similar. At Issoire, the width from centre of aisle wall to centre of nave column is one-fourth of the whole width, equal to the width from centre of nave columns, and to the diameter of the chapels in the apse, and one-half the height of the aisle, and one-fourth that of the nave. The height from floor to ridge is equal to the extreme width at base of walls. At Notre-Dame-du-Port the same kind of proportion exists, but from the outside of the buttress to the outside of the nave pier is one-fourth of the whole width.

I must now, before I conclude, say a few words as to the date of these churches, for which M. Mallay[57] is inclined to claim rather too great an age. He dates most of them (conjecturally) in the tenth century, though he admits that buildings in which the pointed arch is introduced may be as late as the twelfth century; and he considers the date of Notre-Dame-du-Port, Clermont, as circa A.D. 863 to 868. He founds this belief on the fact that no lava was used in its construction, and that the mosaics in its walls were formed of scoriae found on the surface of the soil. He considers that lava was not used until the eleventh century, but he must also prove (which he has not done) that stone was never used in Auvergne after the lava had once been admitted. M. Mallay depends no doubt to some extent on the admitted date of the nave of S. Amable, at Riom, where the main arches are pointed, as A.D. 1077. But the presence of the pointed arch proves nothing as to date, for we see it long before this in S. Front, at Périgueux; and in every other respect there is no doubt that S. Amable presents every evidence of being older than Notre-Dame-du-Port, and others of these churches, in which none but round arches occur.

On either side of Auvergne there are other churches, of precisely the same character as to plan and mode of construction, the dates of which are pretty certain. One is S. Étienne, at Nevers, which was commenced in A.D. 1063, and completed and consecrated on the 13th December 1097. The plan of this church is similar in nearly every respect to that of the Auvergne churches. But, so far as one may judge of date from style, I should have no hesitation in saying that this church must be older than either Issoire or Notre-Dame-du-Port. It is ruder in character, there is very little sculpture on the capitals, which are mostly a sort of rude imitation of Doric, and in the transepts there are not only round arches, but also some straight-sided.

At Conques, south of Auvergne, is another church on the same plan as S. Étienne, Nevers, in almost every respect, which there is little doubt was completed in the first half of the eleventh century, by the founder Abbot Odalric. Then again to the west there is the church of Moustier-neuf, Poitiers, commenced in A.D. 1069, and consecrated in A.D. 1096, which has a chevet evidently formed upon the same type as Conques; and at S. Hilaire, in the same city, consecrated in A.D. 1069, whilst the ground-plan of the chevet is just the same as that of Conques, the nave columns are analogous, there, to the half barrel-vaults of the triforium in Auvergne. Now none of these churches is earlier than the beginning of the eleventh century, and yet it is hardly credible that a province shut in as Auvergne was should have received a perfect and complete new style, or invented one and carried it to the degree of finish and perfection at which it had arrived when Notre-Dame-du-Port was erected, without our being able to trace, somewhere, the source from which it was developed. I believe, however, that its origin may be traced if we examine carefully the architecture of the church of S. Front at Périgueux, commenced in A.D. 984 and completed in A.D. 1047. This church, founded on the same type as, if not copied from St. Mark’s, Venice,[58] exercised a vast direct influence on the architecture of the day. It is seen most clearly in churches which are, like itself, cruciform, without aisles, and covered with domes. The churches of Auvergne, and those other examples to which I have referred, seem to me to be clearly derived from S. Front, or from the Eastern models on which it too was founded. The east end of St. Mark’s presents a circular wall, with a succession of semicircular recesses or apses in its thickness. S. Sophia contains the same feature, though differently treated. The Roman circular buildings which have so much in common with early Byzantine architecture have the same feature; and S. Vitale, Ravenna, whether it is Romanesque or Byzantine in its origin, is planned in a similar way. The architect of S. Front evidently copied his apses from these models, only converting the recesses of St. Mark’s into chapels projecting from the walls.[59] The Auvergne architects attempted to combine the plan of the basilica, with its nave and aisles, with the features which were seen at S. Front. They retained its external wall and projecting chapels, therefore, but placed within them the cluster of columns round the apse forming an aisle between the chapels and the choir. By this simple and natural modification of the S. Front plan to meet the necessities of their triple-aisled churches they at once invented, one may almost say, the perfect French chevet. I know no other churches in France of the same age which appear to have suggested so much in this respect; and you will realize it if you compare their plans with, among others, those of Bourges cathedral, S. Pierre at Bourges, S. Martin at Étampes, Chartres cathedral, the destroyed church of S. Martin at Tours, and finally what is, I think, almost the best complete Gothic plan, that of Rouen cathedral. In every one of these we see the surrounding aisle lighted by windows between the chapels, and the chapels are distinct and well-separated on the exterior, precisely as in these older churches in Auvergne. These buildings, therefore, have great value, not only as illustrating a chapter of the history of our art, but because the chapter which they do illustrate is just one of the most interesting I can conceive; being that which explains how and by what steps Gothic architecture, of which, as our national style, we are so justly proud, was developed from the noble architecture of the old Romans and Greeks, an architecture to which we owe, among other things, this great debt of gratitude, that it naturally led up to, and rendered possible, a Westminster, a Chartres, an Amiens, and all the other glories of our Christian architecture.

You will have gathered that there are many similar features in the churches of the two provinces which I have been describing. They are shortly these: vaults and quasi-domes alike, and carried on the same kind of squinches or pendentives; the decoration with mosaics and its detail; the design and treatment of doors, either sculptured or inlaid; the form of trefoil cusping of arches, character of mouldings, sculpture, and decoration with painting, all of these are the same throughout both districts. The only marked difference, and it is important, is in the ground-plan, the cathedral of Le Puy having no chevet, but an east end derived from Romanesque rather than Byzantine precedents; and the other churches in its neighbourhood are generally similar in their plan.

There are two important heads of my subject to be shortly discussed before I conclude. One of them refers to roofing; the other to coloured decoration. First, as to roofing. I have already explained how this was executed; let us now consider why the modes which we see were adopted. At S. Front the experiment was tried of covering a nave and transepts with a succession of domes resting on pendentives, and supported on pointed arches spanning the nave. These domes were the only covering of the church, and were visible on the outside as well as on the inside. At Conques, the architect, unable to carry domes on the comparatively delicate piers which were all that were required for the division of a nave from its aisles, contrived a barrel-vault for his nave, the thrust of which was resisted by the half barrel-vault of the triforium; a device not improbably obtained from Byzantine churches: for if we compare the section of S. Sophia with that of the crossing and central dome of Notre-Dame-du-Port, we shall find the semi-domes affording abutments for the great domes in the former, absolutely identical in their section with the half barrel-vault, which forms the abutment on the north and south sides of the central dome of the latter.[60] But it was impossible to obtain any light for a clerestory roofed and supported in this fashion, and one is rather disposed to wonder how it was that so many churches should have been built on the same gloomy scheme. It was, no doubt, because in that part of France wooden roofs were thought to be undesirable, and no other economical way was seen of combining the nave and aisles with what was intended to be an indestructible stone roof. I need hardly say that at the same period, in the north of France, in Normandy, and in England, the nave was seldom, if ever, roofed with anything but timber, and the aisles only were vaulted in stone.

At Tournus, on the Saône, another device was adopted to serve the same end as the Auvergne roof, but admitting of a clerestory: this was the covering of the nave with a succession of barrel-vaults at right angles to the length of the church, and supported on bold transverse arches. But I doubt whether it was ever repeated on a nave, though there are several examples of aisles thus roofed;[61] and it was, no doubt, ugly and ungainly. The Le Puy architect devised yet another plan, which combined to some extent all the others, and this was, as I have explained, a succession of domical vaults, which, while it was much lighter and more practicable (owing in part to the difference of scale) than the S. Front plan of a series of genuine cupolas, achieved, nevertheless, much of the effect that was there gained. A very small portion only of the weight of the vault exerted a direct lateral thrust, and it was possible, therefore, to erect such a roof upon a clerestory; and though the transverse arches limit the height of the building in one respect, in another there is no question that the height is apparently much increased; for in looking down the interior it is impossible ever to see the apex of any of the domes, and the vault lost behind the transverse arches gains immensely in mystery and infinity, so as to produce the effect of a larger and loftier building than the reality. But, on the other hand, the disadvantages were great: the piers between the nave and its aisles were so large as to render the aisles nearly useless; and I can hardly wonder, therefore, that the example set here was not generally, if, indeed, at all followed.

It is doubtful where the kind of vault used at Le Puy was first devised. The central dome of S. Michel de l’Aiguille is, perhaps, the oldest of all, and this is, in fact, a square dome, if one may use the expression. The octagonal dome-vaults of the cathedral are probably a little later, but that over the crossing of the church of Ainay at Lyon may possibly be older. A comparison will make it evident that one is copied from the other; and if the Le Puy vault was derived from Lyon, it becomes possible to make the important inference that it was an Eastern influence travelling up the Rhone and distinct from that which is seen at Périgueux, to which we owe this kind of domed roof. Further evidence of this is found in the pendentives of the dome at Brioude,[62] which are identical in intention with the plan of the church of SS. Sergius and Bacchus, at Constantinople, and yet quite unlike the kind of pendentive common in churches of the S. Front type. They are, in fact, the Le Puy and Ainay pendentive reduced to the very simplest conditions. The invention of the flying buttress adumbrated in, and possibly suggested by, the quadrant vaults of Auvergne, finally stopped these various endeavors after new forms of roofs, and set men to work to see how it might most readily be made to serve the boldest and most airy system of design and construction; and in the rage for these, that old system of roofing with domes, which had been, so far as is known,[63] first tried in France at Périgueux, and had afterward spread with such rapidity over a very large district, though with many modifications and variations, was entirely ignored or forgotten. Is it well that we too should ignore it? It is clear that the disciples of the Gothic school may claim it as their own with just as much truth as any other school can; and in some form or other it is often so attractive, so majestic on a large scale, so impressive even on a small scale, that few of us who have much work to do should altogether eschew all use of it, or treat it as though it were the exclusive property of the architects of Classic and Renaissance buildings. I do not feel, however, as most who write on the subject seem to do, that our domes must invariably be supported on what are called true pendentives. I think they are not beautiful, and I do not see that they are especially scientific. The S. Front pendentives are mere corbellings out of the wall, and in truth only imitations of pendentives. At S. Mark’s they are formed with a succession of arches of brick work across the angle of the dome, though this construction is not visible, and these, I suppose, are all wrong; but they are very similar in their intention to the kind of pendentive which I have had to illustrate to-night, and which is in truth much more Gothic and picturesque in its character than the true pendentive, for it admits of any amount of decorative sculpture, and is really precisely similar in its object to the squinches under our own English spires.[64]