he central apse, and is lighted by a
large triple window. The northern chapel is very different. It is much
broader; broader indeed than the ambulatory which leads to it, and is
covered by barrel vaults. The niches in the bema only rise to a short
distance above the floor, not, as on the opposite side, to above the
cornice. It is lighted by a large triple window similar to those of the
other two apses.

[Illustration: FIG. 42. S. MARY PANACHRANTOS.
INSCRIPTION ON THE APSE OF THE NORTH CHURCH.]

From love for the mother of God ... beautiful temple ...
Constantine; which splendid work ... of the shining heaven an
inhabitant and citizen him show O Immaculate One; friendliness
recompensing ... the temple ... the gift.

The outer narthex on the west of the two churches and the gallery on
the south of the south church are covered with cross-groined vaults
without transverse arches. The wall of the south church, which shows in
the south gallery, formed the original external wall of the building. It
is divided into bays with arches in two and three orders of brick
reveals, and with shallow niches on the broader piers.

The exterior of the two churches is very plain. On the west are shallow
wall arcades in one order, on the south similar arcades in two orders.
The northern side is inaccessible owing to the Turkish houses built
against it.

On the east all the apses project boldly. The central apse of the south
church has seven sides and shows the remains of a decoration of niches
in two stories similar to that of the Pantokrator (p. 235); the other
apses present three sides. The carved work on the window shafts is
throughout good. An inscription commemorating the erection of the
northern church is cut on a marble string-course which, when complete,
ran across the whole eastern end, following the projecting sides of the
apses. The letters are sunk and marked with drill holes.

Wulff is of opinion that the letters were originally filled in with
lead, and, from the evidence of this lead infilling, dates the church as
late as the fifteenth century. But it is equally possible that the
letters were marked out by drill holes which were then connected with
the chisel, and that the carver, pleased by the effect given by the
sharp points of shadow in the drill holes, deliberately left them. The
grooves do not seem suitable for retaining lead.

In the course of their history both churches were altered, even in
Byzantin