and from the western end of the Pârasnâth hill to the junction of the Barâkar and Damuda rivers in the south. The circuit of this tract is 350 miles, measured direct on the map, or upwards of 420 miles by road distance following the windings of the two rivers.*

28. CHAMPA.

From Mongir, Hwen Thsang travelled eastward for 300 li, or 50 miles, to Chen-po, or Champa, which is an old name of the district of Bhagalpur. The capital was situated on the Ganges, at from 140 to 150 li, or 23 to 25 miles, to the west of a rocky hill that was completely surrounded by the river. On its summit there was a Brahmanical temple. From this description it is easy to recognize the picturesque rocky island opposite Pathargháta with its temple-crowned summit. As Patharghâta is exactly 24 miles to the east of Bhagalpur, I conclude that the capital of Champâ must have stood either on the same site, or in its immediate vicinity. Close by, on the west side, there still exists a large village named Champanagar, and a smaller one named Champapur, which most probably represent the actual site of the ancient capital of Champâ.†

The pilgrim estimates the circuit of Champâ at 4000 li, or 667 miles; and as it was bounded by the Ganges on the north, and by Hiranya-Parvata, or Mongir, on the west, it must have extended to the Bhagirathi branch of the Ganges on the east and to the Daumda river on the south. Taking the two northern points at Jángira and Teliagali on the See Map No. I. + Ibid.

Ganges, and the two southern points at Páchit on the Damuda and Kalna on the Bhagirathi, the length of the boundary line will be 420 miles measured direct, or about 500 miles by road distance. This is so much less than the size estimated by Hwen Thsang that I think there must either be some mistake in the text or some confusion between the geographical limits of the original district of Champá, and its actual political boundary at the time of the pilgrim's visit. We know from his journal that the king of Mongir, on the west of Champâ, had been dethroned by a neighbouring raja, and that the district of Kânkjol on the east of Champa was then a dependency of the neighbouring kingdom. As Champá lies between these two districts, I infer that the raja of Champa was most probably the king who conquered them, and therefore that the large estimates of Hwen Thsang must include these two states to the east and west of the original Champâ. Under this view, the political boundaries may be stated as extending from Lakhiterai to Râjmahal on the Ganges, and from the Pârasnâth Hill along the Daumda river to Kalna on the Bhagirathi. With these boundaries the circuit of Champâ will be about 550 miles measured direct, or 650 miles by road distance.

29. KANKJOL.

From Champâ the pilgrim travelled to the eastward for 400 li, or 67 miles, to a small district named Kiechu-u-khi-lo, or Kie-ching-kie-lo.* The distance and bearing bring us to the district of Rajmahal, which was originally called Kánkjol, after a town of that *Julien's Hiouen Thsang,' iii. 73. See Map No. I.

name which still exists 18 miles to the south of Râjmahal. Following the river route via Kahalgaon (Colgong) and Râjmahal, the distance from Bhagalpur is just 90 miles; but by the direct route through the hills, viâ Mângaon and Bharhat, the distance is under 70 miles. As this position agrees with that of the place indicated by Hwen Thsang, I suspect that there may have been a transposition of two syllables in the Chinese name, and that we should read Kie-kie-chu-lo, which is a literal transcript of Kánkjol. In Gladwyn's translation of the 'Ayin Akbari'* the name is read as Gungjook, but as all the names are given alphabetically in the original, it is certain that the first letter is a k; I conclude, therefore, that the true name is Kánkjol, as the final / might easily be misread as a k. In his Gazetteer, Hamilton† calls the place Caukjole, which is probably a misprint for Cankjole. He notes that the district of Rajmahal was formerly "named Akbarnagar from its capital, and in the revenue records Caukjole, as being the chief military division."

Hwen Thsang estimates the size of the district at 2000 li, or 333 miles, in circuit; but as it was a dependency of one of the neighbouring kingdoms it was probably included, as I have already noted, in the area of the dominant state. When independent, the petty state of Kánkjol most probably comprised the whole of the hill country to the south and west of Rajmahal, with the plains lying between the hills and the Bhagirathi river as far south as Murshidâbâd. The circuit of this tract would be about 300 miles, as stated by Hwen Thsang.

* 'Ayin Akbari,' ii. 178.

'Gazetteer of India,' in v. Rajamahal.

30. PAUNDRA VARDDHANA.

From Kânkjol the pilgrim crossed the Ganges, and travelling eastward for 600 li, or 100 miles, he reached the kingdom of Pun-na-fa-tan-na.* This name M. Stanislas Julien renders as Paundra-Varddhana, and M. Vivien de Saint-Martin identifies it with Bardwân. But Bardwân is to the south of the last station, and on the same side of the Ganges, besides which its Sanskrit name is Varddhamana. The difference in the direction of the route might be a mistake, as we have found in several previous instances; but the other differences are, I think, absolutely fatal to the identification of Bardwân with the place noted by Hwen Thsang. I would propose Pubna, which is just 100 miles from Kánkjol, and on the opposite bank of the Ganges, but its direction is nearly south-east instead of east. The Chinese syllables may represent either Punya Varddhana, or Paundra Varddhana; but the latter must be the true name, as it is mentioned in the native history of Kashmirt as the capital of Jayanta, Raja of Gau., who reigned from A.D. 782 to 813. In the spoken dialects the name would be shortened from Pon-bardhan to Pobadhan, from which it is an easy step to Pubna, or Pobna, as some of the people now pronounce it. Hwen Thsang estimates the circuit of the kingdom at 4000 li, or 667 miles, which agrees exactly with the dimensions of the tract of country bounded by the Mahânadi on the west, the

* Julien's Hiouen Thsang,' iii. 74. See Map No. I.

Raja Tarangini,' iv. 421. See also the Quart. Orient. Mag. ii. 188, for an account of Pundra-deșa, taken by H. H Wilson from the Brâhmanda section of the Bhavishya Purâna. The greater part of the province was to the north of the Ganges, including Gauda, Pubna, etc.

Tista and Brahmaputra on the east, and the Ganges on the south.

31. JAJHOTI.

Hwen Thsang places the kingdom of Chi-chi-to at 1000 li, or 167 miles, to the north-east of Ujain. As the first and second syllables of this name are represented by different Chinese characters, it is certain that the pilgrim must have intended them to be the equivalents of two distinct Indian characters. This requirement is fully met by identifying Chi-chi-to with the kingdom of Jajhoti, or Jajhaoti, mentioned by Abu Rihân, who calls the capital Kajuráhah,† and places it at 30 parasangs, or about 90 miles, to the south-east of Kanoj. The true direction, however, is almost due south, and the distance about twice 30 parasangs, or 180 miles. This capital was actually visited by Ibn Batuta in A.D. 1335, who calls it Kajúra, and describes it as possessing a lake about 1 mile in length, which was surrounded by idol temples. These are still standing at Khajuraho, and they form perhaps the most magnificent group of Hindu temples that is now to be found in northern India.

From these accounts of Abu Rihân and Ibn Batuta, it is evident that the province of Jajhoti corresponded with the modern district of Bundelkhand. The Chinese pilgrim estimates the circuit of Chichito at 4000 li, or 667 miles, which would form a square of about 167 miles to each side. Now, Bundelkhand in its widest extent is said to have comprised all

* Julien's Hiouen Thsang,' Index, iii. 530, 408. See Map No. I. Reinaud, Fragments Arabes,' etc., p. 106.

་

Dr. Lee's translation, p. 162; where the name is read as Kajwarâ, but the original Persian characters read Kaiurâ.