The Assyrian palace stood uniformly on an artificial platform. Commonly this platform was composed of sun-dried-bricks in regular layers; but occasionally the material used was merely earth or rubbish, excepting towards the exposed parts--the sides and the surface which were always either of brick or of stone. In most cases the sides were protected by massive stone masonry, carried perpendicularly from the natural ground to a height somewhat exceeding that of the plat-form, and either made plain at the top or else crowned with stone battlements cut into gradines. The pavement consisted in part of stone slabs, part of kiln-dried bricks of a large size, often as much as two feet square. The stone slabs were sometimes inscribed, sometimes ornamented with an elegant pattern. (See [PLATE XLI., Fig. 2.]) Occasionally the terrace was divided into portions at different elevations, which were connected by staircases or inclined planes. The terrace communicated in the same way with the level ground at its base, being (as is probable) sometimes ascended in a single place, sometimes in several. These ascents were always on the side where the palace adjoined upon the neighboring town, and were thus protected from hostile attack by the town walls. [PLATE XLI., Fig. 1] Where the palace abutted upon the walls or projected beyond them--and the palace was always placed at the edge of a town, for the double advantage, probably, of a clear view and of fresh air--the platform rose perpendicularly or nearly so; and generally a water protection, a river, a moat, or a broad lake, lay at its base, thus rendering attack, except on the city side, almost impossible. [Illustration: PLATE 41] The platform appears to have been, in general shape, a rectangle, or where it had different elevations, to have been composed of a rectangles. The mound of Khorsabad, which is of this latter character, resembles a gigantic T. [PLATE XLII., Fig. 1.] [Illustration: PLATE 42] It must not be supposed, however, that the rectangle was always exact. Sometimes its outline was broken by angular projections and indentations, as in the plan [PLATE XLII., Fig. 21.] where the shaded parts represent actual discoveries. Sometimes it grew to be irregular, by the addition of fresh portions, as new kings arose who determined on fresh erections. This is the ease at Nimrud, where the platform broadens towards its lower or southern end, and still more at Koyunjik and Nebbi Yunus, where the rectangular idea has been so overlaid as to have almost wholly disappeared. Palaces were commonly placed near one edge of the mound--more especially near the river edge probably for the better enjoyment of the prospect, and of the cool air over the water.