pends on the fact that if we take
the principal curves at any points of the wing, and continue them in the
lines which they are pursuing at the moment they terminate, these
continued lines will all meet in a single point, C. It is this law which
gives unity to the wing.

All groups of curves set beside each other depend for their beauty upon
the observance of this law;[69] and if, therefore, the mountain crests
are to be perfectly beautiful, Nature must contrive to get this element
of radiant curvature into them in one way or another. Nor does it, at
first sight, appear easy for her to get, I do not say radiant curves,
but curves _at all_: for in the aiguilles, she actually bent their beds;
but in these slaty crystallines it seems not always convenient to her to
bend the beds; and when they are to remain straight, she must obtain the
curvature in some other way.

[Illustration: FIG. 50.]

[Illustration: FIG. 51.]

[Illustration: FIG. 52.]

Sec. 9. One way in which she gets it is curiously simple in itself, but
somewhat difficult to explain, unless the reader will be at the pains of
making a little model for himself out of paste or clay. Hitherto,
observe, we have spoken of these crests as seen at their sides, as a
Greek helmet is seen from the side of the wearer. By means presently to
be examined, these mountain crests are so shaped that, seen _in front_,
or from behind (as a helmet crest is seen in front of or behind the
wearer), they present the contour of a sharp ridge, or house gable. Now
if the breadth of this ridge at its base remains the same, while its
height gradually diminishes from the front of it to the back (as from
the top of the crest to the back of the helmet), it necessarily assumes
the form of such a quaint gable roof as that shown in profile in Fig.
50, and in perspective[70] in Fig. 51, in which the gable is steep at
the end farthest off, but depressed at the end nearest us; and the rows
of tiles, in consequence, though in reality quite straight, appear to
radiate as they retire, owing to their different slopes. When a mountain
crest is thus formed, and the concave curve of its front is carried into
its flanks, each edge of bed assuming this concave curve, and radiating,
like the rows of tiles, in perspective at the same time, the whole
crest is thrown into the form Fig. 52, which is that of the radiating
plume required.

[Illustration: FIG. 53.]

Sec. 10. It often happens, however, that Nat