Bk. II. Ch. I. PROVENCE. 463 without peculiarities that point to a different mode of elaborating this peculiar feature from anything found elsewhere. As a design its principal defect seems to be a want of lightness in the upper story. The sinoie circular opening there is a mistake in a building gradually growing lighter towards its summit. These towers were very seldom, if ever, attached symmetrically to the churches. When height was made an object, it was more fre- quently attained by carrying up the dome at the intersection of the choir with the nave. At Aries this is done by a heavy square tower, gradually diminisliing, but still massive to the top ; but in most in- stances the square becomes an octagon, and this again passes into a 324. Church at Cruas. (From Taylor anil Xodier.) circle which terminates the composition. One of the best specimens of this class of domes, if tljey may be so called, is the church of Cruas (Woodcut No. 324), where these parts are pleasingly subordinated, and form, with the apses on which they rest, a very beautiful com- position. The defect is the tiled roofs or offsets at the junction of the various vStories, which give an appearance of weakness, as if the upper parts could slide, like the joints of a telescope, one into the other. This could easily be avoided, and probably was so in the orig- inal design. If this were done, we have here the principle of a more pleasing crowning member at an intersection than was afterwards used in pointed architecture, and capable of being applied to domes of any extent.
