of the works at Christ Church, the elastic character of which the rib was susceptible seems to have occurred distinctly to the architect, and was sufficiently brought out by him in the aisles of the nave, to approve itself afterwards as a pattern. Fig. B (in my last article) is the section which prevails throughout that most elegant structure, the Christ Church chapter-house; and fig. C, which, in effect, differs but little from it, is perhaps the most exquisite of the forms which has yet been devised for this purpose.

This I stated more in detail in the concluding part of my last article: in the present it will be my object to trace the steps by which the idea of what, in my last article, I called elasticity found its way into other parts of the system.

And first, as regards the mouldings of the intercolumnar arches. When the Romans substituted the arch for the epistylium, they did not attempt to ornament it appropriately, but gave it the appearance of a bent architrave. No. 2, fig. 1, is the common Grecian architrave; fig. 2, the Roman misapplication of it. This device is obviously unmeaning; indeed, its defect is acknowledged by the practice of modern architects, who, for the sake of relief, frequently have recourse to a second arch, (as in fig. 3,) parallel to the first, and receding behind it. A relief of the same kind was sought in the middle ages, by modifying the form of the Roman architrave. Fig. 2 was turned into fig. 4; the recess detached the line a from 6, and thus gave the effect of a second receding arch.

α

With a slight modification of the form x, fig. 5 represents a horizontal section of two arches, such as fig. 4, resting back to back on the same pier, and branching in opposite directions: the dotted line represents the pier, or rather column, on which they rest. This fig. is taken from the nave of Christ Church; (e) is the section of a perpendicular shaft, which apparently supports the vaulting, and which terminates in some fanciful ornament, where it comes in contact with (x) (x). I shall not be very wrong in ascribing this specimen to the last few years of the reign of Henry I. So far, then, I find the treatment of the Gothic intercolumnar arch corresponding closely with the Roman, indeed deviating from it only in that respect in which modern imitators of Roman architecture have allowed themselves to deviate from it. Fig. 6 is a corresponding section taken from the work of William of Sens, at Canterbury. It is evidently a modification, and by no means a violent one, of fig. 5; but the striking thing about it is this, that it differs from fig. 5 and fig. 3 just in those respects in which figs. 5 and 3 differ from fig. 2. The rods () (B) are still farther enlarged, and the recess & deepened, as if William of Sens had recognized in Christ Church, or some similar building, an approximation to the form which his eye felt the want of, and was encouraged to feel his way a step farther. By these two

changes he made so great a difference in the ratio of the line NO to O P, that the interval between a, a, became insignificant, and the four rods a, a, B, B, were thrown into a single group.

And now let us suppose each of these rods to undergo the change which metamorphosed fig. A into fig. B in the last article; and we shall have a form not materially differing from fig. 7, which is taken from Cologne cathedral.

Such, then, are the successive changes through which the Roman architecture passed into the most approved Gothic moulding, and these all proceed on the same principle. The two things aimed at in each change are force and lightness,-the first of which was attained by deepening the recesses, so as to cast darker shadows, and the other by giving the rods (2) (E), &c., a more absorbing consequence. These changes, together with the transition from the round to the pointed arch, and the parallel changes which I shall proceed to trace in the column, contributed jointly to produce that elastic effect which I have before noticed, and which seems not so much to have arisen from the happy thought of any individual architect, as from the nature of the arch itself. It seems that this striking feature refused to blend with the previously arranged system, but gradually remodelled the whole on a new principle. To use rather a harsh metaphor, it acted, as it were, chemically on the other elements of architecture, dissolving their old combinations,-taking up some, and depositing others, combining them in new proportions, and crystallizing them in a new form.

I observed, that while the changes above noticed were taking place in the intercolumnar arch, the column itself underwent parallel changes. These, however, were not quite so regular in their progress. It seems for a long time to have been assumed by architects that a column must be something round; and, till the latter end of the twelfth century, the utmost latitude they allowed themselves in deviating from this form, was now and then to substitute an octagon for it.

On the rebuilding of Canterbury Cathedral, 1174, William of Sens ventured on a farther innovation: after he had completed some part of his work, and had been able to observe the effect of the slight clustered shafts, by which he had given apparent support to the vaulting, he conceived the novel idea of clustering the column itself. This he did in the manner represented, (fig. 1, No. 3,) by attaching the slight marble shafts A, C, C, to the sides of an octagonal column. In the summer of the fourth year, says Gervase, "A cruce incipiens decem pilarios erexit scilicét utrinque quinque, quorum duos primos marmoreis ornaus columnis contra alios duos principales fecit.;" and after