Mahalla, which is about 700 feet to the south of the last mentioned mound in the Bhatpuri Mahalla. That this mound was the site of one or more Brahmanical temples scems almost certain from my discovery of a figure of Shasti, the goddess of fecundity, and of a pedestal bearing the date of Samvat 1193, or A. D. 1136, which is posterior to the extinction of Buddhism in Kanoj. I think it probable that excavations in this mound would be attended with success, as the two temples are said to have been built of stone, which no doubt furnished the whole of the materials for the Masjid and tomb of Makhdum Jahaniya.

XI. A-YU-TO, OR AYODHYA.

From Kanoj the two Chinese pilgrims followed different routes, Fa Hian having proceeded direct to Sha-chi (the modern Ajudhya, near Fyzabad on the Ghaghra), while Hwen Thsang followed the course of the Ganges to Prayag, or Allahabad. The first stage of both pilgrims would, however, appear to be the same. Fa Hian states that he crossed the Ganges and proceeded 3 yojans, or 21 miles, to the forest of Holi, where there were several Stupas erected on spots where Buddha had "passed, or walked, or sat."* Hwen Thsang records that he marched 100 li, nearly 17 miles, to the town of Nava-deva-kula, which was on the eastern bank of the Ganges, and that at 5 li, or nearly 1 mile, to the south-cast of the town there was a Stupa of Asoka, which was still 100 feet in height, besides some other monuments dedicated to the four previous Buddhas.† I think it probable that the two places are the same, and that the site was somewhere near Nobatganj, just above the junction of the Isan River and opposite Nanamow Ghát. But as there are no existing remains anywhere in that neighbourhood, the place has been most likely swept away by the river. This is rendered almost certain by an examination of the Ganges below the junction of the Isan. Formerly the river continued its course almost due south from Nanamow for many miles, but some centuries ago it changed its course first to the south-east for 4 or 5 miles, and then to the south-west for about the same distance, where it rejoined its old bed, leaving an island, some 6 miles in length by 4 in breadth, between the two channels. As Hwen Thsang's account places Nava-deva-kula on the very

Beal's Fa Hian, C. XVIII.
Julien's Hwen Thsang, II., 265.

site of this island, I conclude that the town as well as the Buddhist monuments must all have been swept away by the change in the river's course.*

On leaving Nava-deva-kula, Hwen Thsang proceeded 600 li, or 100 miles, to the south-east, and re-crossing the Ganges he reached the capital city of A-yu-to, which was 20 li, or upwards of 3 miles, in circuit. Both M. Julien and M. St. Martin have identified this place with Ayodhya, the once celebrated capital of Rama. But though I agree with them as to the probable identification of the name as that of the country, I differ with them altogether in looking for the capital along the line of the Ghaghra River, which is due cast from Kanoj, whercas Hwen Thsang states that his route was to the south-east. It is, of course, quite possible that the pilgrim may occasionally use the generic name of Ganges as the appellation of any large river, such, for instance, as the Ghaghra; but in the present case, where the recorded bearing of south-east agrees with the course of the Ganges, I think it is almost certain that the Ganges itself was the river intended by the pilgrim. But by adopting the line of the Ganges we encounter a difficulty of a different kind in the great excess of the distance between two such well known places as Kanoj and Prayâg. According to Hwen Thsang's route, he first made 100 li to Nava-deva-kula, then 600 li to Ayutho, then 300 li by water to Hayamukha, and lastly 700 li to Prayaga. All these distances added together make a total of 1,700 li, or 283 miles, which is just 100 miles, or 600 li, in excess of the true distance. But as a part of the journey, viz., 300 li, or 50 miles, was performed by water, the actual excess may, perhaps, not be more than 85 or 90 miles; although it is doubtful whether the distance of 300 li may not have been the road measurement and not the river distance. It is sufficient for our purpose to know that Hwen Thsang's recorded measurement is somewhere about 100 miles in excess of the truth. The only explanation of this error that suggests itself to me is, that there may have been an accidental alteration of one set of figures, such as 600 li for 60 li, or 700 li for 70 li. Supposing that the former

* If we might read 10 li' instead of 100 li, this place might be identified with Deokali, which is situated on the Chota Gungâ about 2 miles below Kanoj. The two names are precisely the same, excepting that the modern one has dropped the two initial syllables nava, or "new," which, however appropriate in the time of the Chinese pilgrim, would almost certainly have been dropped in the course of a few centuries.-See Julien's Hwen Thsang,

II. 266.

was the case, the distance would be shortened by 540 li, or 90 miles, and if the latter, by 630 li, or 105 miles. This mode of correction brings the pilgrim's account into fair accordance with the actual distance of 180 miles between Kanoj and Prayâg.

By adopting the first supposition, wen Thsang's distance from Nava-deva-kula to the capital of Ayutho will be only 60 li, or 10 miles, to the south-east, which would bring him to the site of an ancient city named Kákúpur, just 1 mile to the north of Seorâjpoor, and 20 miles to the northwest of Cawnpoor. If we adopt the latter correction, the pilgrim's distance to Ayutho of 600 li, or 100 miles, will remain unchanged, and this would bring him viâ Mánikpur, which is also an ancient place. By the first supposition the subsequent route would have been from Kókúpur to Daundiakhera by boat, a distance of exactly 50 miles, or 300 li, and from thence to Prayag, a distance of more than 100 miles, which agrees with the 700 li, or 116 miles, of the pilgrim. By the second supposition the subsequent route would have been from Khara to Papamow by water, about 50 miles, and thence to Prayâg, about 8 miles of land, which agrees with the 70 li of the proposed correction. In favour of this last supposition is the fact that the bearing from Khara to Papamow of cast by south is more in accordance with Hwen Thsang's recorded east direction than the southeast bearing of Daundiakhera from Kâkûpur. I confess, however, that I am more inclined to adopt the former correction, which places the chief city of Ayutho at Kâkûpur, and the town of Hayamukha at Daundiakhera, as we know that the last was the capital of the Bais Rajputs for a considerable period. I am partly inclined to this opinion by a suspicion that the name of Kakúpur may be connected with that Bagud, or Vágud, of the Tibetan books. According to this authority a Sákya, named Shampaka, on being banished from Kapila retired to Bagud, carrying with him some of Buddha's hairs and nail-parings, over which he built a chaitya. He was made King of Bágud, and the monument was named after himself (? Shâmpakâ Stupa).* No clue is given as to the position of Bágud; but as I know of no other name that resembles it, I am induced to think that it is

*Csoma de Körös in Asiatic Researches, XX., p. 88.

probably the same place as the Ayutho of Hwen Thsang, which was also possessed of a Stupa containing some hairs and nail-parings of Buddha. Kákúpur is well-known to the people of Kanoj, who affirm that it was once a large city with a Raja of its own. The existing remains of Kákúpur consist of numerous foundations formed of large bricks, and more particularly of a connected set of walls of some large building which the people call "the palace." I have not yet visited this place, which lay out of my line of route, but I hope to have an opportunity of examining it hereafter.

XII. HAYAMUKHA OR AYOMUKHA.

of

From Ayutho the Chinese pilgrim proceeded a distance of 300 li, or 50 miles, down the Ganges by boat to O-ye-mu-khi, which was situated on the north bank of the river, M. Julien reads this name as Hayamukha, equivalent to "Horse face," or "Iron face," which was the name of one of the Dánavas or Titans.* Neither of these names, however, gives any clue to the site of the old city; but if I am right in my identification of Ayutho with Kákúpur, it is almost certain that Ayomukha must be the same as Daundiakhera. Hwen Thsang makes the circuit of the town 20 li, or upwards of 3 miles, but Daundikhera presents no appearance having ever been so large. There still exist the ruins of an old fort or citadel, 385 feet square, with the walls of two buildings which are called the Raja's and Rani's palaces. The foundation of this citadel is attributed to Raja Raghunâth Sinh, but he was apparently some comparatively modern Thakur, or petty Chief, as Daundiakhera is universally allowed to have been the capital of the Bais Rajputs, who claim descent from the famous Sâlivâhan. As there are no remains of any buildings which can be identified with the momuments described by Hwen Thsang, the actual site of Ayomukha must still remain doubtful.

XIII. PRAYAGA, OR ALLAHABAD.

From Ayomukha the pilgrim proceeded 700 li, or 116 miles, to the south-east, to Prayâga, the well known place of pilgrimage at the junction of the Ganges and Jumna, where

* Julien's Hwen Thsang, II., p. 274.-See my "Ancient Geography of India," p. 387, "Daundia means simply a "drum-beater," and was probably applied to some mendicant, who took up his abode on the khera, or mound; and as this name is not likely to have been imposed on the place until it was in ruins, the difference of name offers no impediment to the identification of Daundiakhera with Hayamukha.

Akbar some centuries later built his fort of Ilâhabás, or Allahâbâd, as it was afterwards called by Shahjahan. The distance and bearing given by Hwen Thsang agree almost exactly with those of Prayâga from Daundiakhera. The distance is 104 miles by the nearest road to the south of the Ganges; but as the pilgrim followed the north road, the distance must have been increased to about 115 or 120 miles. According to him the city was situated at the confluence of the two rivers, but to the west of a large sandy plain. In the midst of the city there was a Brahmanical temple, to which the presentation of a single piece of money procured as much merit as that of one thousand pieces elsewhere. Before the principal room of the temple there was a large tree with wide-spreading branches, which was said to be the dwelling of an anthropophagous demon. The tree was surrounded with human bones, the remains of pilgrims who had sacrificed their lives before the temple-a custom which had been observed from time immemorial. *

I think there can be little doubt that the famous tree here described by the Chinese pilgrim is the well known Akshay Bat, or "undecaying Banian tree," which is still an object of worship at Allahabad. This tree is now situated underground at one side of a pillared court, which would appear to have been open formerly, and which is, I believe, the remains of the temple described by Hwen Thsang. The temple is situated inside the fort of Allahabad to the east of the Ellenborough Barracks, and due north from the stone pillar of Asoka and Samudra Gupta. Originally both tree and temple must have been on the natural ground level; but from the constant accumulation of rubbish they have been gradually earthed up until the whole of the lower portion of the temple has disappeared underground. The upper portion has long ago been removed, and the only access to the Akshay Bat now available is by a flight of steps which leads down to a square pillared court-yard. This court has apparently once been open to the sky, but it is now closed in to secure darkness and mystery for the holy Fig tree.

The Akshay Bat is next mentioned by Rashid-ud-din in the Jamiut-tawârikh, in which he states that the "tree of Prag" is situated at the confluence of the Jumna and

Julien's Hwen Thsang, II., p. 276.