DECORATION XXVII. CORNICE AND CAPITAL 363

represented in one of its most beautiful states by the glacier line a, on Plate 7. I would rather have taken this line than any other to have formed my third group of cornices by, but as it is too large, and almost too delicate, we will take instead that of the Matterhorn side, e, f, Plate 7. For uniformity’s sake I keep the slope of the dotted line the same as in the primal forms; and applying this Matterhorn curve in its four relative positions to that line, I have the types of the four cornices or capitals of the third family, e, f, g, h, on Plate 15.

These are, however, general types only thus far, that their line is composed of one short and one long curve, and that they represent the four conditions of treatment of every such line; namely, the longest curve concave in e and f, and convex in g and h; and the point of contrary flexure set high in e and g, and low in f and h. The relative depth of the arcs, or nature of their curvature, cannot be taken into consideration without a complexity of system which my space does not admit.

Of the four types thus constituted, e and f are of great importance; the other two are rarely used, having an appearance of weakness in consequence of the shortest curve being concave: the profiles e and f, when used for cornices, have usually a fuller sweep and somewhat greater equality between the branches of the curve; but those here given are better representatives of the structure applicable to capitals and cornices indifferently.

§ 10. Very often, in the farther treatment of the profiles e or f, another limb is added to their curve in order to join it to the upper or lower members of the cornice or capital. I do not consider this addition as forming another family of cornices, because the leading and effective part of the curve is in these, as in the others, the single ogee: and the added bend is merely a less abrupt termination of it above or below: still this group is of so great importance in the richer kinds of ornamentation that we must have it sufficiently represented. We shall obtain a type of it by merely continuing

[Version 0.04: March 2008]