TRIPOLY TO BENGAZI. 245 which, in that case, must have been higher than the outer walls which formed the parapet ; and a space seems always to have been left between these central buildings and the parapet, in which the garrison placed themselves when employed in defending the fort.

An opening like a window was observed in the parapet of one of the Cyclopian castles at Ghimenes, which might have been used for drawing up those who entered the fort, as there was no other mode of entrance whatever. In fact there could scarcely have been any communication between the upper and lower parts of these erections ; for the whole space between the walls was filled up with earth in the manner already related, to within a few feet of the top. We noticed near most of them a small rising ground, with one or two wells in it, having remains of building about it ; they were generally within fifty yards of the fort, by which they were commanded. The castles have most of them been surrounded with a trench, on the outer side of which there is generally a low wall strongly built with large stones. Some of the trenches which have been excavated in the solid rock of the soil are of considerable depth and width ; and in one instance, occurring between Ghimenes and Bengazi, we observed chambers excavated in the sides of the trench, as we find to be the case in that which surrounds the second pyramid, and which is equally formed in the rocky soil on which the building stands, although of course on a much larger scale. The trench of the fort here alluded to is about five-and-twenty feet in width, and its depth about fifteen ; the fort itself is an hundred and twenty-five feet in length, and ninety in width, of a quadrangular form, and in the centre of each of its sides is a quadrangular projec-