annexe
with a semi-spherical superstructure, such as is often to be seen behind
Roman Catholic churches. The illusion was really wonderful.

Owing to the pools of water not far from "Church-rock" we called that
spot _Caponga de la Lagoa_.

A few hundred yards beyond "Church-rock" we came upon another
extraordinary sight: a quadrangular rocky castle--a perfect cube of
rock--which stood at a considerable elevation upon a conical base, some
distance off the wall-like sides of the plateau. Strangely enough, a thin
wall of rock, only a few feet thick, quite vertical, of great height and
of great length, joined this quadrangular castle to the plateau. That
wall had evidently remained standing when the plateau had subsided. The
larger plateau along the foot of which we travelled ended in two great
domes, one at each angle of its eastern terminus wall. The eastern part
of that plateau was flat-topped, whereas the central portion rose into a
double pyramid and looked not unlike a giant tent with a porch
attachment. It was of a bright yellow colour--apparently sandstone and
ashes. The work of erosion had been greater on the eastern face--owing, I
think, to the prevalent wind on that side.

On looking back upon the great range of rock which ended abruptly near
"Church-rock" (which, as we have seen, once formed part of it), a great
semicircular cavity was disclosed on its western face. The summit of the
wall around the cavity rested on an inclined plane, which in its turn
rested above a vertical concave wall. The latter wall of rock had conical
buttresses at the terminal points.

West-north-west of the great wall was an immense depression. Only a
conical hill rose above its last undulations. The upper edge of that
depression was at an altitude of 1,550 ft. above the sea level, whereas
the top of "Church-rock" was fully a thousand feet higher--viz. 2,550 ft.

[Illustration: A Grand Rock.

"Church rock."]

[Illustration: Church Rock.

(Side view.)]

At the terminus of the first section of the cliff range, interrupted
by a great fissure from the second section, another structure in course
of formation not unlike "Church-rock" could be observed. It had a
quadrangular tower surmounting it. There was in the second section of the
range a regular quadrangle of rock, with a high tower upon a conical
hill, and another castle-like structure surmounting a conical base. The
two were most impressive as they stood in their sombre red