The confusion which occasionally exists between places bearing the same name in mediæval geographical annals is very obvious in Idrisi's description of Merv. The greater Merv (the Russian provincial capital) is clearly mixed up in his mind with the lesser Merv when, in describing the latter, he says that Merv-el-Rud is situated in a plain at a great distance from mountains, and that its territory is fertile but sandy; three grand mosques and a citadel adorn an eminence and water is brought to it by innumerable canals, all of which is applicable to Merv but not to Merv-el-Rud. He then continues with a description of the greater Merv, which is quite apropos to that locality, and makes it clear incidentally that Khiva (not Merv) represents the ancient Khwarezm. Again, he enumerates towns and places of Mahomedan origin which are "dependent" on "Merv." Amongst them we find Mesiha, a pretty, well-cultivated place one day's journey to the west of Merv; Jirena (Behvana), a market-town 9 miles from Merv, and 3 from Dorak (? Dizek), a place situated on the banks of the river; then Dendalkan, an important town two days from Merv on the road to Sarakhs; Sarmakan, a large town to the left of Dorak and 3 miles farther, Dorak being situated on the banks of the river at 12 miles from Merv in the direction of Sarakhs; Kasr Akhif (or Ahnef), a little town at one day's distance from Merv on the road to Balkh; Derah, a small town 12 miles from Kasr Ahnef where grapes were abundant. Here, says Idrisi, the river divides the town in two parts which are connected by a bridge. It is quite impossible to straighten out this geographical enumeration, unless we assume that it refers to Merv-el-Rud and not to Merv. Then Mesiha becomes a possibility, and might be looked for among the ruined sites on the Kushk River--possibly at Kila Maur. Dorak, at 12 miles from Merv in the direction of Sarakhs, and Dendalkan at two days' journey in the same direction, would still be on the river banks. Kasr Ahnef we know to have been built after the Arab invasion in the valley of the Murghab, about 12 miles from Khuzan (identified by Rawlinson with Ak Tepe) and 15 from Merv-el-Rud, and must have been situated near the Band-i-Nadir, where the desert road to Balkh enters the hills. Ak Tepe must once have been a place of great importance, both strategically (as it commands the position of the two important highways southward to Herat, the Kushk and the Murghab valleys) and commercially. But apparently its importance did not survive to Arab times. Dendalkan was certainly near Ak Tepe.