According to the bardic legend given in _The Annals_, the Rāthors, the second great Rājput clan, owed their origin to a migration of a body of its members to the western Desert when the territory of Kanauj was conquered by Shihābu-d-dīn in A.D. 1193. But it is now certain that the ruling dynasty of Kanauj belonged, not to the Rāthor, but to the Gaharwār clan, and that the first Rāthor settlement in Rājputāna must have occurred anterior to the conquest of Kanauj by the Musalmāns. An inscription, dated A.D. 997, found in the ruins of the ancient town of Hathūndi or Hastikūndi in the Bali Hakūmat of the Jodhpur State, names four Rāthor Rājas who reigned there in the tenth century.[i.12] The local legend is an attempt to connect the line of Rāthor princes with the Kanauj dynasty. It has been suggested that the Deccan dynasty of the Rāshtrakūtas which, in name at least, is identical with Rāthor, reigning at Nāsik or Malkhed from A.D. 753 to 973, was connected with the Reddis or Raddis, a caste of cultivators which seem to have migrated from Madras into the Deccan at an early period. But any racial connexion between the Deccan Reddis and the Rāthors of Rājputāna is very doubtful.[i.13] The Chandel clan, ranked in _The Annals_ among the Thirty-six Royal Races, is believed to be closely connected with the Bhars and Gonds, forest tribes of Bundelkhand and the Central Provinces. Mr. R. V. Russell prefers to connect them with the Bhars alone, on the ground that the Gonds, according to the best traditions, entered the Central Provinces from the south, and made no effective settlement in Bundelkhand, the headquarters of the Chandels.[i.14] But there was a Gond settlement in the Hamīrpur District of Bundelkhand, and the close connexion between the Gonds and the Chandels began in what is now the Chhatarpur State. The results of recent investigations into Rājput ethnology are thus of great importance, and enable us to correct the bardic legends on which the genealogies recorded in _The Annals_ were founded. Much remains to be done before the question can be finally settled. The local Rājput traditions and the ballads of the bards must be collected and edited; the ancient sites in Rājputāna must be excavated; physical measurements, now somewhat discredited as a test of racial affinities, must be made in larger numbers and by more scientific methods. But the general thesis that some of the nobler Rājput septs are descended from Gurjaras or other foreigners, while others are closely connected with the autochthonous races, may be regarded as definitely proved.