important periods, but the position of the setting sun
at the summer solstice is well marked by a round
tower on the wall overhanging the high cliff, and this
is undoubtedly a wall of the best period.
At this western end of the fortress we have two
instances of parts of walls which faced the setting
•sun at the winter solstice being decorated by a den-
telle pattern.
The disposition of the ornamental patterns on the
little round ruin at the Lundi Eiver is interesting. It
faces the rising sun at the winter solstice, but the
place had been inhabited by Kaffirs, and all vestige
of an altar, if it ever existed, had been destroyed.
The nature of the patterns here is different from that
of those at Zimbabwe. The one near the top of the
wall is composed of two rows of little squares alternating
with blank spaces, and a little way below this
are two rows of a herring-bone pattern. There is a
curious rounded protuberance on the outside of the
wall, and the herring-bone pattern stops at this point,
but the other extends right round to the southeastern
doorway.
■ This temple is similar in many ways to the partly
circular one north-west of the great tower at Zimbabwe.
They have both the same diameter and they
each have two doorways which are in somewhat
similar positions, although the temple at the Lundi is
oriented towards the rising sun at the winter solstice,
and the other, if it ever had a pattern, would have
had it facing the setting sun at this solstice.
The dentelle pattern on the great tower seems to
have been oriented towards the setting sun at the
winter solstice, and the centre of the partly circular
temple at G is roughly in a line between the centre of
the tower and the sun at this time. When the sun is
rising at the summer solstice G will be behind the
tower and in the middle of its shadow, in a position
analogous to that of the altar behind the arc a k .
The direction of the wall at the north-eastern extremity
of the arc of which G is the centre is towards
the rising sun at the winter solstice, and its inner side
points past^the centre of the arc ak 1 towards the
point of the outer wall which is in a s t r a i g h t line
between the altar and the sun when it rises at this
solstice. The wall at the other extremity of the arc
points to the rising sun at the other solstice.
It is perhaps worthy of remark that the centre of
the great tower is distant the length of its own
height (42'3 feet) from the solstitial line mk, while the
centre of the little tower would be the same distance
from a parallel solstitial line drawn from the southeastern
extremity of the arc of which G is the centre ;
and also that the centres of the great tower and the
centres of the arcs ak and kb lie in one straight line.
At Matindela the general aspect of the decorated
part of the building is towards the setting sun, but
the masonry is so rough in its construction that we
need expect little accuracy in orientation. The whole
appearance of the place suggests that what exists at
present is merely a rough rebuilding of an older
structure. What remains of the internal arrangements
This is not very accurately shown in the small scale plan.