They were distinguished in Fig. V. by circular curves drawn on opposite sides of the same line. But we now know that in these smaller features the circle is usually the least interesting curve that we can use; and that it will be well, since the capital and cornice are both active in their expression, to use some of the more abstract natural lines. We will go back, therefore, to our old friend the salvia leaf; and taking the same piece of it we had before, _x y_, Plate VII., we will apply it to the cornice line; first within it, giving the concave cornice, then without, giving the convex cornice. In all the figures, _a_, _b_, _c_, _d_, Plate XV., the dotted line is at the same slope, and represents an average profile of the root of cornices (_a_, Fig. V., p. 69); the curve of the salvia leaf is applied to it in each case, first with its roundest curvature up, then with its roundest curvature down; and we have thus the two varieties, _a_ and _b_, of the concave family, and _c_ and _d_, of the convex family. [Illustration: Plate XV. CORNICE PROFILES.] § IV. These four profiles will represent all the simple cornices in the world; represent them, I mean, as central types: for in any of the profiles an infinite number of slopes may be given to the dotted line of the root (which in these four figures is always at the same angle); and on each of these innumerable slopes an innumerable variety of curves may be fitted, from every leaf in the forest, and every shell on the shore, and every movement of the human fingers and fancy; therefore, if the reader wishes to obtain something like a numerical representation of the number of possible and beautiful cornices which may be based upon these four types or roots, and among which the architect has leave to choose according to the circumstances of his building and the method of its composition, let him set down a figure 1 to begin with, and write ciphers after it as fast as he can, without stopping, for an hour. § V. None of the types are, however, found in perfection of curvature, except in the best work. Very often cornices are worked with circular segments (with a noble, massive effect, for instance, in St. Michele of Lucca), or with rude approximation to finer curvature, especially _a_, Plate XV., which occurs often so small as to render it useless to take much pains upon its curve. It occurs perfectly pure in the condition represented by 1 of the series 1-6, in Plate XV., on many of the Byzantine and early Gothic buildings of Venice; in more developed form it becomes the profile of the bell of the capital in the later Venetian Gothic, and in much of the best Northern Gothic. It also represents the Corinthian capital, in which the curvature is taken from the bell to be added in some excess to the nodding leaves. It is the most graceful of all simple profiles of cornice and capital.