strong relationship, though, on the last supposition, five centuries
older than the earliest of the five terminal examples; and it is still
more remarkable because it reverses the usual treatment of the lower
roll, which is in general a tolerably accurate test of the age of a
base, in the degree of its projection. Thus, in the examples 2, 3, 4, 5,
9, 10, 12, the lower roll is hardly rounded at all, and diametrically
opposed to the late Gothic conditions, 24 to 28, in which it advances
gradually, like a wave preparing to break, and at last is actually seen
curling over with the long-backed rush of surf upon the shore. Yet the
Torcello base resembles these Gothic ones both in expansion beneath and
in depth of cavetto above.

Sec. VIII. There can be no question of the ineffable superiority of these
Gothic bases, in grace of profile, to any ever invented by the ancients.
But they have all two great faults: They seem, in the first place, to
have been designed without sufficient reference to the necessity of
their being usually seen from above; their grace of profile cannot be
estimated when so seen, and their excessive expansion gives them an
appearance of flatness and separation from the shaft, as if they had
splashed out under its pressure: in the second place their cavetto is so
deeply cut that it has the appearance of a black fissure between the
members of the base; and in the Lyons and Bourges shafts, 24 and 26, it
is impossible to conquer the idea suggested by it, that the two stones
above and below have been intended to join close, but that some pebbles
have got in and kept them from fitting; one is always expecting the
pebbles to be crushed, and the shaft to settle into its place with a
thunder-clap.

Sec. IX. For these reasons, I said that the profile of the pure classic
base had hardly been materially improved; but the various conditions of
it are beautiful or commonplace, in proportion to the variety of
proportion among their lines and the delicacy of their curvatures; that
is to say, the expression of characters like those of the abstract lines
in Plate VII.

The five best profiles in Plate X. are 10, 17, 19, 20, 21; 10 is
peculiarly beautiful in the opposition between the bold projection of
its upper roll, and the delicate leafy curvature of its lower; and this
and 21 may be taken as nearly perfect types, the one of the steep, the
other of the expansive basic profiles. The characters of all, however,
are so