rk. I say,
_commonly_ the best, because, in some cases, this expressional invention
may prevail over all other considerations, and a column of unnecessary
bulk or fantastic slightness be adopted in order to strike the spectator
with awe or with surprise.[39] The architect is, however, rarely in
practice compelled to use one kind of material only; and his choice
lies frequently between the employment of a larger number of solid and
perfect small shafts, or a less number of pieced and cemented large
ones. It is often possible to obtain from quarries near at hand, blocks
which might be cut into shafts eight or twelve feet long and four or
five feet round, when larger shafts can only be obtained in distant
localities; and the question then is between the perfection of smaller
features and the imperfection of larger. We shall find numberless
instances in Italy in which the first choice has been boldly, and I
think most wisely made; and magnificent buildings have been composed of
systems of small but perfect shafts, multiplied and superimposed. So
long as the idea of the symmetry of a perfect shaft remained in the
builder's mind, his choice could hardly be directed otherwise, and the
adoption of the built and tower-like shaft appears to have been the
result of a loss of this sense of symmetry consequent on the employment
of intractable materials.

Sec. XII. But farther: we have up to this point spoken of shafts as always
set in ranges, and at equal intervals from each other. But there is no
necessity for this; and material differences may be made in their
diameters if two or more be grouped so as to do together the work of one
large one, and that within, or nearly within, the space which the larger
one would have occupied.

Sec. XIII. Let A, B, C, Fig. XIV., be three surfaces, of which B and C
contain equal areas, and each of them double that of A: then supposing
them all loaded to the same height, B or C would receive twice as much
weight as A; therefore, to carry B or C loaded, we should need a shaft
of twice the strength needed to carry A. Let S be the shaft required to
carry A, and S_2 the shaft required to carry B or C; then S_3 may be
divided into two shafts, or S_2 into four shafts, as at S_3, all
equal in area or solid contents;[40] and the mass A might be carried
safely by two of them, and the masses B and C, each by four of them.

[Illustration: Fig. XIV.]

Now if we put the single shafts each under the centre o