could not have been used for drainage or defence; and in the fortress above there are two similar tunnels equally inexplicable. The actual approaches to the sacred enclosure are most carefully defended with buttresses on either side, into which a form of portcullis has been fixed, with two grooves, one running down each side, presumably originally intended to receive a wooden door; but at a later period all these entrances have been carefully walled up, for what purpose it is difficult to say. It naturally occurred to us that this had been done at a time of danger for protection, but the neatness with which the blocking-up walls are executed is against this theory. At point V on the plan there is a remarkable instance of the two periods of building. Here, in front of the sacred enclosure, the wall was decorated with courses of black slate in the older and better wall, whereas they are omitted in the inferior continuation. At point E there is a raised platform immediately in front of the large round tower, covered with a flooring of thick cement, supported by large stones loosely packed together, into which a monolith had been stuck. This platform was connected with the sacred enclosure by a flight of cement steps, and was presumably used for religious purposes. In dealing with the two remarkable round towers which stood in the sacred enclosure, one cannot lay too much stress on the symmetry of the courses and the accuracy with which they have been built. They stand in the centre of the sacred enclosure, which was floored with cement. By digging to their foundations we were able to get very accurate measurements of them, and found that the circumference of the smaller one corresponds exactly to the diameter of the big one, and the diameter of the big one is apparently equal to half its original height, and its circumference again is equal to the diameter of the round building on the Lundi River. The battering of the big tower is carried out with mathematical accuracy, the slope of the curve being perfectly regular, and is produced by placing the superincumbent stones in a slightly receding position, so that with the aid of a monkey rope we were able to climb to the top. A few courses below the summit, which would seem to be very much in its original condition except on the south side, where Herr Mauch confesses to have pulled down the stones of several courses, runs a dentelle pattern, marked D on the plan, formed by placing the stones of one course edgeways. This pattern is the same as the lower one given in the illustration of Matindela ruins, p. 137; but unfortunately, owing to the demolition of the upper courses, it is impossible to define its extent. The tower would seem to have been thirty-five feet in height, and the summit to have been a level of about four feet in diameter. By digging below this tower, and pulling out stones from the sides, which we carefully replaced, we demonstrated to our satisfaction that it was solid.