most of these Assyrian palaces were destroyed by fire, the absence of the remains of columns offers no difficulty. On the other hand, in many parts of the palace of Khorsabad no trace of fire remains, and yet here no suggestion of detached columns can be found, and, moreover, it is extremely difficult to arrange columns symmetrically in the various apartments so that doorways are not interfered with. There is also another difficulty, viz. that if the building called the Harem at Khorsabad was built in this way, the apartments would have been open to the view of any one ascending the lofty building called the observatory. It is quite possible that further explorations may tend to elucidate this difficult question of roofing, but at present all that can be said is that none of the theories that have been put forward is wholly satisfactory. As no columns at all exist, we cannot say what capitals were employed, but it is probable that those of Persepolis, which will be shortly described, were copied from an earlier wooden form, which may have been that used by the Assyrian builders. There is, however, capping the terrace on which the temple was erected at Khorsabad, a good example of an Assyrian cornice, which is very similar indeed to the forms found in Egypt, and some of the sculptured bas-reliefs which have been discovered depict rude copies of Assyrian buildings drawn by the people themselves; and it is most interesting to notice that just as we found in the Egyptian style the proto-Doric column, so in the Assyrian we find the proto-Ionic (Figs. 34, 34a), and possibly also the proto-Corinthian (Fig. 34b). [Illustration: FIG. 34.--PROTO-IONIC COLUMN.] [Illustration: FIG. 34a.--PROTO-IONIC CAPITAL FROM ASSYRIAN SCULPTURE.] [Illustration: FIG. 34b.--PROTO-CORINTHIAN CAPITAL FROM ASSYRIAN SCULPTURE.]