internally towards the apartments and externally towards the outer air.
The only ground for this supposition, which is of a very startling
character, seems to be the occurrence in a single bas-relief,
representing a city in Armenia, of what is regarded as a similar
arrangement. But it must be noted that the lower portion of the
building, represented opposite, bears no resemblance at all to the same
part of an Assyrian palace, since in it perpendicular lines prevail,
whereas, in the Assyrian palaces, the lower hues were almost wholly
horizontal; and that it is not even Certain that the upper portion,
where the pillars occur, is an arrangement for admitting light, since it
may be merely an ornamentation.

The difficulties attaching to every theory of roofing and lighting which
places the whole of an Assyrian palace under covert, has led some to
suggest that the system actually adopted in the larger apartments was
that _hypoethral_ one which is generally believed to have prevailed in
the Greek temples, and which was undoubtedly followed in the ordinary
Roman house. Mr. Layard was the first to post forward the view that the
larger halls, at any rate, were uncovered, a projecting ledge,
sufficiently wide to afford shelter and shade, being carried round the
four sides of the apartment while the centre remained open to the sky.
The objections taken to this view are--first, that far too much heat and
light would thereby have been admitted into the palace; secondly, that
in the rainy season far too much rain would have come in for comfort;
and, thirdly, that the pavement of the halls, being mere sun-dried
brick, would, under such circumstances, have been turned into mud. If
these objections are not removed, they would be, at any rate, greatly
lessened by supposing the roofing to have extended to two-thirds or
three-fourths of the apartment, and the opening to have been
comparatively narrow. We may also suppose that on very bright and on