450 FRENCH ARCHITECTURE. Part II. mistakes they made : had they carefully profiled and ornamented the exterior of the stone roofs in the same manner as they ornamented the inside, their buildings would have been not only much more beautiful, but much more permanent, and the style would have been saved from the principal falsity that now deforms it. Even as it is, if we wished intelligently to adapt the Gothic to our jnirposes, instead of merely co]:)ying it, this is one of the points to which we ought first to turn our attention. Another circumstance which may be alluded to here, when speak- ing on this subject, which led to the adoption of the pointed arch at an early age in the southern ])rovinces of France, was the use of domes as a roofing exjjedient. These, it is true, are not found in Provence, but they are common in Aquitaine and Anjou — some of them certainly of the 11th century ; and there can be little douljt but that these are not the earliest, though their predecessors have perished or have not yet been brought to light. There is no one who has studied this subject who is not aware how excellent, as a constructive expedient, the pointed arch is as applied to intersecting vaults, but it is not so generally understood why it was equally necessary in the construction of domes. So long as these rested on drums rising from the ground the circular form sufficed ; but when it became necessary to rest them on pendentives in the angles of square or octagonal buildings, the case was widely different. The early Byzantine architect^ — in Sta. Sophia for instance — did fit pen- dentives to circular arches, but it was with extreme difticultv, and required very great skill both in setting out nnd in execution. But the su])eriority of the pointed form was perceived at an early date ; and the Saracens, who were trammelled by no traditions, adopted it at once as a doming expedient and adhered to it as exclusively as the Gothic architects did in the construction of their vaults — and for the same reason — simply because it was the best mode of con- struction. It is easy to explain why this should be so. In the annexed diagram, fig. 1 represents the pendentives of a dome resting on circular Fig. 1. Fig. 2. Fig. 3. arches. At a they become evanescent, and for some distance from the centre are so weak that it is only by concealed construction that they can be made to do their work. When the pointed arch is introduced, as in fig. 2, not only is great freedom obtained in spacing, but the whole becomes constructively correct ; when, as in fig. 3, an octagonal
