The location and identity of the settlement or settlements called Sarai (which means, after all, simply "palace" and moreover seems to function as a synonym of "horde") have been subject to scholarly disagreement. Arguments have centered on whether or not there were two (or more) capital cities named Sarai and what their respective locations were. One of the influential views that emerged was that "New Sarai" was an enhancement or expansion of "Old Sarai", rather than a separate settlement. The other view was that "Old Sarai" and "New Sarai" were two separate settlements after all, separated by some distance, with "New Sarai" possibly associated with a satellite settlement called Gülistan. Already in the second half of the 18th and first quarter of the 19th century, Sarai was being sought variously at the large ruin fields of Tsarevskoe gorodishche and Selitrennoe gorodishche, both located (some 250 km apart) on the left bank of the
Akhtuba, a left
distributary of the Volga, which remain the most impressive archaeological sites in the area. During the late 19th-late 20th century, the dominant view that resulted from earlier studies, was that
"Old Sarai" was founded in the 13th century by Batu and located at
Selitrennoe gorodishche (previously also called Dzhigit Hadzhi, just northwest of modern
Selitrennoe, about 30 km southeast of
Kharabali and about 120 km north from
Astrakhan), while
"New Sarai" was founded later in the 13th century by Berke and made capital in the early 14th century by Öz Beg, and was located at
Tsarevskoe gorodishche (previously also called Tsarevy Pady, , just northwest of modern
Tsarev and farther west of
Kolobovka, about 55 km east-southeast of
Volzhsky). The Selitrennoe gorodishche archaeological site was described as "Gorodishche Selitrennoe ... remains of the Golden Horde capital Sarai-Batu" on the official sign in at the site, while the corresponding sign at the Tsarevskoe gorodishche archaeological site read "Ruins of Sarai-Berke (New Sarai)". This apparent certainty was eventually eroded by subsequent scholarship. First, it was noted that "New Sarai" could not be associated with Berke on historical, archaeological, and numismatic grounds, leading to a modification of the reconstruction: "New Sarai" was built by Öz Beg, and both "Sarai Batu" and "Sarai Berke" referred to "Old Sarai". Second, the analysis of the archaeological remains and distribution of found coins led to the realization that, while Selitrennoe gorodishche matched Sarai (or more specifically "New Sarai"), Tsarevskoe gorodishche did not, and was likely to represent the hitherto unlocated city of Gülistan, which was developed by Jani Beg and rivaled Sarai as a khan's residence and mint in the 1350s and 1360s but then declined. This conclusion quickly gained support among the experts, and the present general consensus is that Selitrennoe gorodishche () is "New Sarai", while Tsarevskoe gorodishche () is Gülistan. There is currently no consensus on the location and identification of "Old Sarai". Some scholars suppose that it was a less impressive settlement whose ruins are yet unnoticed or obscured under those of "New Sarai" at Selitrennoe gorodishche, or were destroyed by changing water courses and levels. Others have sought a suitable archaeological site downstream of Selitrennoe gorodishche to identify with "Old Sarai". Here there are extensive remains of Golden Horde settlements, especially at Aksarayskoe gorodishche (at modern Lapas, medieval Dawlat-Khan), and at Akhtubinskoe gorodishche (at modern
Komsomol'skiy, medieval Ak-Saray). Apart from a tentative suggestion for
Vol'noe 15 km downstream from Selitrennoe, Akhtubinskoe gorodishche at Komsomol'skiy has also been suggested, and a case has been made for Krasnoyarskoe gorodishche in
Krasny Yar, where the necropolis on the neighboring Mayachny hill has yielded some coins from the 13th century. ruler
Jani Beg, from his New Serai mint. Dated AH 748 (1347-8 CE) banner in the
Catalan Atlas (1375). The question remains open. The account of William of Rubruck (1254) ensures a location for the Sarai of Batu at or slightly above the apex of the Volga Delta, on the left edge of the Volga-Akhtuba river sistem. This is something possibly compatible with Selitrennoe gorodishche, or perhaps rather with a site farther downstream, between it and the apex of the Delta. This is also consistent with the account of
Abū al-Fidāʾ (1321), which places Sarai on the Volga only 2 days above the Caspian coast, and with that of Ibn Baṭṭūṭa, who reached Sarai 3 days after (Old) Astrakhan.
Pegolotti (writing in 1335–1343) gives a single
day's journey between Sarai and Astrakhan, while the
Nikon Chronicle cites 2 days for the same journey. Only the account of William of Rubruck refers without any doubt to "Old Sarai", since it dates to the reign of Batu. If "Old Sarai" and "New Sarai" coexisted for some time at some distance from each other, "Old Sarai" ought to be sought downstream of "New Sarai". This might be confirmed by a 15th-century map from the Franciscan monastery of Lesina (
Hvar), which places Saray (apparently "New Sarai" at Selitrennoe gorodishche) on the Volga, above Dolatcana (Dawlat-Khan, at Aksarayskoe gorodishche/Lapas), above Eschisari (Eski Saray, i.e., "Old Sarai"). One interpretation of the evidence would place Batu's original camp on the Akhtuba across from modern
Seitovka (and just south from modern Aksaraysky), "Old Sarai" a little upstram at medieval Ak-Saray (Akhtubinskoe gorodishche, Komsomol'skiy), the major royal necropolis a little upstream at medieval Dawlat-Khan (Aksarayskoe gorodishche, Lapas), and "New Sarai" a little upstream at Selitrennoe gorodishche. == Little Sarai ==