fectly protective from
rain, but we can help them a little in their duty by a slight advance of
their upper ledge. This, with the form _b_, we can best manage by cutting
off the sharp upper point of its curve, which is evidently weak and
useless; and we shall have the form _f_. By a slight advance of the upper
stone _c_, we shall have the parallel form _g_.

These two cornices, _f_ and _g_, are characteristic of early Byzantine
work, and are found on all the most lovely examples of it in Venice. The
type _a_ is rarer, but occurs pure in the most exquisite piece of
composition in Venice--the northern portico of St. Mark's; and will be
given in due time.

Sec. VIII. Now the reader has doubtless noticed that these forms of
cornice result, from considerations of fitness and necessity, far more
neatly and decisively than the forms of the base, which we left only
very generally determined. The reason is, that there are many ways of
building foundations, and many _good_ ways, dependent upon the peculiar
accidents of the ground and nature of accessible materials. There is
also room to spare in width, and a chance of a part of the arrangement
being concealed by the ground, so as to modify height. But we have no
room to spare in width on the top of a wall, and all that we do must be
thoroughly visible; and we can but have to deal with bricks, or stones
of a certain degree of fineness, and not with mere gravel, or sand, or
clay,--so that as the conditions are limited, the forms become
determined; and our steps will be more clear and certain the farther we
advance. The sources of a river are usually half lost among moss and
pebbles, and its first movements doubtful in direction; but, as the
current gathers force, its banks are determined, and its branches are
numbered.

Sec. IX. So far of the true cornice: we have still to determine the form
of the dripstone.

[Illustration: Fig. VI.]

We go back to our primal type or root of cornice, _a_ of Fig. V. We take
this at _a_ in Fig. VI., and we are to consider it entirely as a
protection against rain. Now the only way in which the rain can be kept
from running back on the slope of X is by a bold hollowing out of it
upwards, _b_. But clearly, by thus doing, we shall so weaken the
projecting part of it that the least shock would break it at the neck,
_c_; we must therefore cut the whole out of one stone, which will give
us the form _d_. That the water may not lodge on the upper ledge