TRIPOLY TO BENGAZI. 199 able and willing to do so in a good hearty draught of muddy water.

Our route on the following day lay over a barren and rugged country, which continues all the way from Hudia to the confines of Barca, where the soil begins gradually to assume a better appearance. In the afternoon, as we passed Mahiriga, we observed the remains of a (quadrangular building occupying the summit of a low range of hills which lay between our road and the sea. On a closer examination?

we found it to be different in plan from any building which we had hitherto met with. At each of the angles there is a circular turrets sloping down from the top, and becoming considerably wider at the base. The sides of this building are constructed with well-shaped stones of four and five feet in length, closely fitted together, and fastened with an excellent cement ; but the turrets were found to be built of much smaller stones, not shaped or put together with the same attention to regularity, and proved on near inspection to be built on to the outer walls and not into them. They may therefore be considered as forming no part of the original plan, and have probably been added at some early period by the Arabs. No traces remain of the external roof of this building, but part of an arched roof is still visible on the ground-floor within, which, from its inferior workmanship, we should be inchned to attribute to the same period at which the turrets were added. Traces of walls are also seen in the inside of the building, which have formerly divided it into chambers ; they are composed of very small stones and appear to be of later work than the exterior. This fortress, for such it has originally