Several sites exhibiting extensive ruins near the banks of the Akhtuba have been identified with Sarai; two in particular. One of these is not far from the great elbow of the Volga at Tzaritzyn: the other much lower down, at Selitrennoyé Gorodok or Saltpetre-Town, not far above Astrakhan. The upper site exhibits by far the most extensive traces of former population, and is declared unhesitatingly to be the sole site of Sarai by M. Gregorieff, who carried on excavations among the remains for four years, though with what precise results I have not been able to learn. The most dense part of the remains, consisting of mounds and earth-works, traces of walls, buildings, cisterns, dams, and innumerable canals, extends for about 7½ miles in the vicinity of the town of Tzarev, but a tract of 66 miles in length and 300 miles in circuit, commencing from near the head of the Akhtuba, presents remains of like character, though of less density, marking the ground occupied by the villages which encircled the capital. About 2½ miles to the N.W. of Tzarev a vast mass of such remains, surrounded by the traces of a brick rampart, points out the presumable position of the Imperial Palace. M. Gregorieff appears to admit no alternative. Yet it seems certain that the indications of Abulfeda, Pegolotti, and others, with regard to the position of the capital in the early part of the 14th century, are not consistent with a site so far from the Caspian. Moreover, F. H. Müller states that the site near Tzarev is known to the Tartars as the “Sarai of Janibek Khan” (1341–1357). Now it is worthy of note that in the coinage of Janibek we repeatedly find as the place of mintage, _New Sarai_. Arabsháh in his History of Timur states that 63 years had elapsed from the foundation to the destruction of Sarai. But it must have been at least 140 years since the foundation of Batu’s city. Is it not possible, therefore, that both the sites which we have mentioned were successively occupied by the Mongol capital; that the original Sarai of Batu was at Selitrennoyé Gorodok, and that the _New Sarai_ of Janibek was established by him, or by his father Uzbeg in his latter days, on the upper Akhtuba? Pegolotti having carried his merchant from Tana (Azov) to Gittarchan (Astrakhan), takes him _one day_ by river to Sara, and from Sara to _Saracanco_, also by river, eight days more. (_Cathay_, p. 287.) In the work quoted I have taken Saracanco for Saraichik, on the Yaik. But it was possibly the Upper or New Sarai on the Akhtuba. Ibn Batuta, marching on the frozen river, reached Sarai in three days from Astrakhan. This could not have been at Tzarev, 200 miles off.