the eastern line of the structure, there are two
more circular depressions similar to those within the court. The latter
is entered by four passageways,--one on the S.E. corner, 4 m.--13
ft.--wide and about 12 m.--40 ft.--long from S. to N.; one through the
eastern wing, 3.40 m.--11 ft.--wide and about 14 m.--46 ft.--long from
E. to W.; one in the N.W. corner and another from the S.W., both 2 m.--6
ft. 6 in.--across. I have designated these four gateways respectively as
_R_, _E_, _G_, and _N_. _R_ and _E_ enter straight through the wall; _G_
forms a semicircle almost from W. through N. to S.; _N_ describes a
right angle from S. by N. to E. The distribution of decay in this house
is the same as in _B_,--the southern parts are on all sides almost
totally obliterated; the N.W. corner is very nearly perfect; the
northern and western walls are tolerably fairly preserved; but the
eastern outline of the east wing, the southern outline of the south
wing, and the southern ends of both east and west have almost completely
disappeared under hills of rubbish, a few posts alone assisting the
explorer. The path of destruction has in both buildings lain in the same
direction,--from S.S.E. to N.N.W.,--and across both its effects have
decreased from south to north. Still, while the similarity in that
respect is astonishing, and while there are apparently more walls in _A_
standing than in _B_, there is, owing to the very uneven surface of the
rock upon which it is built, much more confusion among the ruins of the
former than among those of the latter. _B_ is built on a gradual slope
or ridge; _A_ caps a generally convex surface, scooped out in the
middle, and sloping eastward.[117] Hence comes the division of the whole
structure into four separate and distinct buildings, and hence, also,
the complicated manner in which the whole or each part is ruined, even
walls still standing being twisted out of shape and out of position.
Actual measurements were much less efficacious here than in _B_; and,
although I have worked with not less zeal and conscientiousness, the
result in neatness and precision is certainly less satisfactory. This
explanation will, I hope, induce subsequent explorers to look up my
inaccuracies and correct them.

It is needless, of course, to detail the methods of work. They are on a
larger scale, and in more tedious ways, a repetition of the proceedings
in the case of _B_. The results are as follows, starting from the lin