us
very much reduced, and over it flat domes are thrown. Their rings are
true circles, and as the space they cover is still somewhat oblong they
descend lower, with additional segments of rings (BB), at the ends than
at the sides. In the remaining two oblong bays of the narthex, the
result of introducing the longitudinal arches is to convert a decidedly
oblong space in one direction into a slightly oblong space in the
opposite direction, an additional proof, if any were needed, that the
exact shape of plan with this form of vault was a matter of comparative
indifference to the builder.

In S. Sophia the vault springs from the intrados of the transverse
arches, that is, from the lower edge. In SS. Sergius and Bacchus it
springs from a point so slightly raised as to be hardly noticeable. In
the later vaults, however, the transverse arches, when present, are
boldly shown, and the vault springs from the extrados or outer edge
(_e.g._ S. Saviour in the Chora, S. Theodore).

_Construction._--Most of the churches of the city are covered with thick
coats of plaster and whitewash, both within and without. Only in a few
cases, where these coatings have fallen away through neglect, or in some
remote corner of a building to which these coatings were never applied,
can the construction and the laying of the brickwork be studied. The
two-storied chapel, known as Bogdan Serai (p. 283), is almost denuded of
plaster, and is therefore of importance in this connection. The bricks
of the wall arches on which its dome rests are laid considerably flatter
than the true radiating line, leaving a triangular piece to be filled in
at the crown. On the other hand, the bricks of the transverse arches
under the dome radiate to the centre.

It has been supposed that the method followed in the wall arches was
employed in order to economise centering, since bricks could gradually
be worked out over the space, each course simply sticking to the one
below. This is undoubtedly the case in some examples. But here centering
could not have been of any service in the wall arches, and the
transverse arches are laid without flattening of the courses, though
that arrangement might have been useful in their case. It is therefore
more probable that the flattening of the courses in the wall arches is
simply a piece of careless workmanship. The pendentives, like all
pendentives that could be examined, were formed of horizontal courses
corbelled out to the circle.