date. Upon one of them he found the skeleton of a wild sheep. [38] The evidence which may be collected upon the shores of Lake Van all points in the same direction of a progressive upward tendency. Strecker has thrown out the suggestion that this process may be accountable for the junction of the Arjish arm to the main body; and that we may therefore attach some credence to the statements of Pliny that in his time there were two lakes. [39] However this may be, we are not dependent upon such hypotheses, or upon the stories current of submerged causeways or bridges. The three old fortresses of Akhlat, Adeljivas and Arjish all bear testimony to a considerable rise in the level of the lake since the days when they were built. The walls of the first two on the side of the water have either fallen in or are being slowly undermined. Arjish has been permanently abandoned by its inhabitants. Immemorial villages, like that of Kizvag between Akhlat and Tadvan, are being menaced by the latest periodical increase, which seems to have commenced about 1895. Nature herself speaks eloquently in the same sense. An ancient walnut-tree which stands on the rocky bank of the lake in the gardens of Erkizan, a quarter of Akhlat, had already been deprived of a great portion of its foothold when we encamped beneath its boughs in 1898. In the Sheikh Ora crater a giant mulberry, which may have been some 500 years old, was standing with half its roots in the water and was already doomed. The most obvious explanation of this gradual rise in the norm of the lake level is furnished by a cause, which must be constantly operative, namely the increase of sediment deposited upon the bottom. But whether this factor by itself be sufficient to have produced such important changes is a question upon which I am not qualified to pronounce an opinion. [40] II.--THE ANCIENT EMPIRE OF VAN