e days. The south church is the earlier structure, but shows
signs of several rebuildings. The irregular narthex and unsymmetrical
eastern side chapels are evidently not parts of an original design. In
the wall between the two churches there are indications which appear to
show the character of these alterations and the order in which the
different buildings were erected.

[Illustration: PLATE XXXV.
(1) S. MARY PANACHRANTOS. EAST END OF THE SOUTH CHURCH.
(2) S. MARY PANACHRANTOS. EAST END OF THE NORTH CHURCH.
_To face page 132._]

As has already been pointed out, the north side of the ambulatory in the
south church, which for two-thirds of its length is of practically the
same width as the southern and western sides, suddenly widens out at the
eastern end and opens into a side chapel broader than that on the
opposite side. The two large piers separating the ambulatory from the
central part of the north church are evidently formed by building the
wall of one church against the pre-existing wall of the other. The
easternmost pier is smaller and, as can be seen from the plan, is a
continuation of the wall of the north church. Clearly the north church
was already built when the north-eastern chapel of the south church was
erected, and the existing wall was utilised. As the external
architectural style of the three apses of the south church is identical,
it is reasonable to conclude that this part of the south church also is
later in date than the north church. For if the entire south church had
been built at the same time as the apses, we should expect to find the
lateral chapels similar. But they are not. The vaulting of the central
apse and of the southern lateral chapel are similar, while that of the
northern chapel is different. On the same supposition we should also
expect to find a similar use of the wall of the north church throughout,
but we have seen that two piers representing the old wall of the south
church still remain. The narthex of the south church, however, is
carried up to the line of the north church wall.

The four column type is not found previous to the tenth century. The
date of the north church was originally given on the inscription, but is
now obliterated. Kondakoff dates it in the eleventh or twelfth century.
Wulff would put it as late as the fifteenth. But if the view that this
church was attached to the monastery of Lips is correct, the building
must belong to the tenth century.

Th