To these arguments of M. Botta may be added another from the improbability of the Assyrians being sufficiently advanced in architectural science to be able to construct an arch of the width necessary to cover some of the chambers.  The principle of the arch was, indeed, as will be hereafter shown, well known to the Assyrians, but hitherto we possess no proof that they were capable of applying it on a large scale.  The widest arch which has been found in any of the buildings is that of the Khorsabad town-gate uncovered by M. Place, which spans a space of (at most) fourteen or fifteen feet.  But the great halls of the Assyrian palaces have a width of twenty-five, thirty, and even forty feet.  It is at any rate uncertain whether the constructive skill of their architects could have grappled successfully with the difficulty of throwing a vault over so wide an interval as even the least of these.

M. Botta, after objecting, certainly with great force, to the theory of M. Flandin, proceeded to suggest a theory of his own.  After carefully reviewing all the circumstances, he gave it as his opinion that the Khorsabad building had been roofed throughout with a flat, earth-covered roofing of wood.  He observed that some of the buildings on the bas-reliefs had flat roofs, that flat roofs are still the fashion of the country, and that the debris within the chambers were exactly such as a roof of that kind would be likely, if destroyed by fire, to have produced.  He further noticed that on the floors of the chambers, in various parts of the palace, there had been discovered stone rollers closely resembling those still in use at Mosul and Baghdad, for keeping close-pressed and hard the earthen surface of such roofs; which rollers had, in all probability, been applied to the same use by the Assyrians, and, being kept on the roofs, had fallen through during the conflagration.

The first difficulty which presented itself here was one of those regarded as most fatal to the vaulting theory, namely, the width of the chambers.  Where flat timber roofs prevail in the East, their span seems never to exceed twenty-five feet.  The ordinary chambers in the Assyrian palaces might, undoubtedly, therefore, have been roofed in this way, by a series of horizontal beans laid across them from side to side, with the ends resting upon the tops of the side walls.  But the great halls seemed too wide to have borne such a roofing without supports.  Accordingly, M. Botts suggested that in the greater apartments a single or a double row of pillars ran down the middle, reaching to the roof and sustaining it.  His theory was afterwards warmly embraced by Mr. Fergusson, who endeavored to point out the exact position of the pillars in the three great halls of Sargon at Khorsabad.  It seems, however, a strong and almost a fatal objection to this theory, that no bases of pillars have been found within the apartments, nor any marks on the brick floors of such bases or of the pressure of the