voussoir which will be safe; and to do this requires peculiar
arrangement of the lines of the arch. There are many arrangements,
useful all in their way, but we have only to do, in the best
architecture, with the simplest and most easily understood. We have
first to note those which regard the actual shell of the arch, and then
we shall give a few examples of the superseding of such expedients by
Mont-Cenisian masonry.

Sec. III. What we have to say will apply to all arches, but the central
pointed arch is the best for general illustration. Let _a_, Plate III.,
be the shell of a pointed arch with loose loading above; and suppose you
find that shell not quite thick enough; and that the weight bears too
heavily on the top of the arch, and is likely to break it in: you
proceed to thicken your shell, but need you thicken it all equally? Not
so; you would only waste your good voussoirs. If you have any common
sense you will thicken it at the top, where a Mylodon's skull is
thickened for the same purpose (and some human skulls, I fancy), as at
_b_. The pebbles and gravel above will now shoot off it right and left,
as the bullets do off a cuirassier's breastplate, and will have no
chance of beating it in.

If still it be not strong enough, a farther addition may be made, as at
_c_, now thickening the voussoirs a little at the base also. But as this
may perhaps throw the arch inconveniently high, or occasion a waste of
voussoirs at the top, we may employ another expedient.

Sec. IV. I imagine the reader's common sense, if not his previous
knowledge, will enable him to understand that if the arch at _a_, Plate
III., burst _in_ at the top, it must burst _out_ at the sides. Set up
two pieces of pasteboard, edge to edge, and press them down with your
hand, and you will see them bend out at the sides. Therefore, if you can
keep the arch from starting out at the points _p_, _p_, it _cannot_
curve in at the top, put what weight on it you will, unless by sheer
crushing of the stones to fragments.

Sec. V. Now you may keep the arch from starting out at _p_ by loading it
at _p_, putting more weight upon it and against it at that point; and this,
in practice, is the way it is usually done. But we assume at present
that the weight above is sand or water, quite unmanageable, not to be
directed to the points we choose; and in practice, it may sometimes
happen that we cannot put weight upon the arch at _p_. We may perhaps
want an opening