hewing that they were internal
walls; but from a piece of dentilled, or rather blocked, cornice,
which fits the curve of one of the _exedrae_, I believe the walls were
carried up on the north and south above the roofs of the adjoining
rooms and corridors of the baths, so that they formed a feature in the
elevation and afforded a broken skyline to the composition. The vault
over the centre rose considerably above these walls, a portion of the
centre of which may have been partially open for the emission of steam
and the admission of light. Some square blocks of lead, that were the
yotting of bars of metal, rather favour this idea, and suggest that
these metal bars were a portion of the machinery by which a brazen
shield (_clipeus_) was suspended, or secured, so that by raising
or lowering it the temperature of the hall might be regulated as
described by Vitruvius. In the excavations we found an _ante-fixa_
that must have fallen from some portion of the roof. It appears to
be intended for a lion, but it is much broken.

I have prepared a sketch section of the bath (which I hope
to communicate on a future occasion), transversely and a part
longitudinally, in order that a description may the more readily be
understood, adopting, in my restoration, the established rules of
proportion of Classical architecture, which may, more or less, have
been strictly adhered to when the baths were built; indeed, in the
best specimens of Roman work a licence was given to the architect
as to detail and proportion, that was refused him on the Classical
revival. The pilasters of these baths spring, as I have said before,
from an Attic base, of somewhat coarse proportions, 14in. high.[23]
The attached pilasters that supported the arcade that was carried
longitudinally along the bath are without a base; they must have been,
within a few inches, more or less, not lower than 10ft. in height,
including the impost moulding, of which there are fragments. The
arches springing from them would be about 14ft. wide. I have not
been able to find any fragments of the archivolt. The pilasters that
supported the arches which crossed the _schola_ have bases similar to
the larger pilasters. I can hardly speak positively of their elevation
or that of the arches, but I am inclined to think the height of the
impost moulding was raised, so that the arch, although a smaller span,
was the same in height as the longitudinal arches.

[Footnote 23: The bases of the c