down the river for three days, when he heard that the Indian prince had suddenly abandoned his country and fled to the desert.* Alexander at once pushed

on to Patala. Now, the distance from Brâhmanâbâd to Haidarâbâd is only 47 miles by the direct land route; but as the old bed of the Indus makes a wide sweep round by Nasirpur, the route along the river bank, which was doubtless followed by the army, is not less than 55 miles, while the distance by water must be fully 80 miles. His progress during the first three days, estimated at the usual rate of 10 or 12 miles by land, and 18 or 20 miles by water, would have brought him within 19 miles of Haidarâbâd by land, and 26 miles by water, which distance he would have easily accomplished on the fourth day by a forced march. From Patala he proceeded down the western branch of the river for a distance of 400 stadia, or 46 miles, when his naval commanders first perceived the sea breeze. This point I believe to have been Jarak, which is 30 miles below Haidarâbâd by land, and 45 miles, or nearly 400 stadia, by water. There Alexander procured guides, and, pressing on with still greater eagerness, on the third day he became aware of his vicinity to the sea by meeting the tide.† As the tides in the Indus are not felt more than 60 miles from the sea, I conclude that Alexander must then have reached as far as Bambhra, on the Ghâra, or western branch of the river, which is only 35 miles from the sea by land, and about 50 miles by water. Its distance from Jarak by land is 50 miles, and by

6

* Arrian, Anabasis,' vi. 17; Curtius, Vita Alex., ix. 8, 28, says that he fled to the mountains.

+ Curtius, Vita Alex., ix. 9, 29.

water 75 miles, which the fleet might have easily ac

From these details it been at a considerable

complished by the third day. is clear that Patala must have distance from the sea, that is, not less than the length of the tidal reach, plus three days' sail on the river, plus 400 stadia. These distances by land are respectively 33 miles, 50 miles, and 30 miles, or altogether 113 miles, which corresponds almost exactly with the measurement of Aristobulus of 1000 stadia, or 115 miles.

As these three independent investigations all point to the same place as the most probable representative of Patala, and as that place is called Patasila by Hwen Thsang in the seventh century, and is still known as Pátalpur, I think that we have very strong grounds for identifying Haidarâbâd with the ancient Patula. In his account of the Indus, Arrian* says, "this river also forms a delta by its two mouths, no way inferior to that of Egypt, which, in the Indian language, is called Pattala." As this statement is given on the authority of Nearchus, who had ample opportunities during his long detention in Sindh of intercourse with the people, we may accept it as the general belief of the Sindhians at that time. I would therefore suggest that the name may have been derived from Pátala, the "trumpet-flower" (Bignonia suaveolens), in allusion to the "trumpet" shape of the province included between the eastern and western branches of the mouth of the Indus, as the two branches, as they approach the sea, curve outwards like the mouth of a trumpet.

I cannot close the discussion on the site of this

*Indica,' p. 2.

ancient city without noticing another name of which the conflicting accounts appear to me to have a confused reference to Nirunkot. This name is the Piruz of Istakhri, the Kannazbur of Ibn Haukal, and the Firabuz of Edrisi. According to Istakhri, Piruz was 4 days' journey from Debal, and 2 days from Mehábari, which was itself on the western bank of the Indus, at 2 days' journey from Debal. Ibn Haukal and Edrisi agree that the road to Kannazbur, or Firabuz, lay through Manhabari, or Manjabari, which was on the western bank of the Indus, at 2 days from Debal; but they make the whole distance beyond Debal 14 days instead of 4. Now, Ibn Haukal and Edrisi place their city in Mekrân, a position which they were almost forced to adopt by their long distance of 14 days, although the first two days' journey lie exactly in the opposite direction from Mekran. But if we take the shorter distance of 4 days from Debal, which is found in Istakhri, the earliest of the three geographers, the position of their unknown city will then accord exactly with that of Nirankot. Debal I will hereafter identify with an old city near Lâribandar and Manhábari with Thatha, which is just midway between Lâri-bandar and Haidarâbâd. Now, Ibn Haukal specially notes that Munjábari was situated "to the west of the Mihrân, and there any one who proceeds from Debal to Mansura will have to pass the river, the latter place being opposite to Manjâbari."* This extract shows that Manjâbari was on the western branch of the Indus, and therefore on the high-road to Nirankot as well as to Piruz, or Kannezbur, or Firabuz. I would therefore suggest that the first of

*Prof. Dowson's edition of Sir H. Elliot's Hist. of India, i. 37.

these names, which is thus mentioned in conjunction with Manhâbari might possibly be intended for Nirun, and the other two for Nirunkot, as the alterations in the original Arabic characters required for these two readings are very slight. But there was certainly a place of somewhat similar name in Mekrân, as Bilâduri records that Kizbun in Mekrân submitted to Muhammad Kâsim on his march against Debâl. Comparing this name with Ibn Haukal's Kannazbur, and Edrisi's Firabuz, I think it probable that they may be intended for Panjgur, as suggested by M. Reinaud. The 14 days' journey would agree very well with the position of this place.

Jarak.

The little town of Jarak is situated on an eminence overhanging the western bank of the Indus, about midway between Haidarâbâd and Thatha. Jarak is the present boundary between Vichalo, or Middle Sindh, and Lár, or Lower Sindh, which latter I have been obliged to extend to Haidarâbâd, so as to include the Patala of the Greeks and the Pitasila of the Chinese pilgrim, within the limits of the ancient Delta. This is perhaps the same place as Khor, or Alkhor, a small but populous town, which Edrisi places between Manhâbari and Firabuz, that is, between Thatha and Nirunkot. Three miles below Jarak there is another low hill covered with ruins,

* Prof. Dowson's edition of Sir Henry Elliot's Hist. of India, i. 40. Ibn Haukal: Kannazbur. At page 29 he gives Istakhri's name as Kannazbûn, which Mordtmann reads Firiun. The most probable explanation of these differences is some confusion in the Arabic characters between the name of Nirun and that of the capital of Mekrån.

which the people call Káfir kot, and attribute to Raja Manjhira. The principal ruin is a square basement ornamented with flat pilasters at regular distances. This is supposed to be the remains of a temple. Amongst the ruins were found some fragments of Buddhist statues; and, at a short distance from the hill, an inscription in early Indian characters, of which I can read only the words putrasa and Bhagavatasa, and a few letters in different parts; but these are sufficient to show that the inscription is Buddhist, as well as the other remains.

Minnagar, Manhábari, or Thatha.

The city of Thatha is situated in a low swampy valley, 3 miles from the western bank of the Indus, and 4 miles above the separation of the Bágar, or western branch, from the Satá, or main stream of the river. Littlewood remarks that "the mounds of rubbish upon which the houses are piled slightly raise its site above the level of the valley."† The place was visited by Captain Hamilton in A.D. 1699, who describes it as situated on a spacious plain about 2 miles from the Indus. It is highly probable, therefore, that the town originally stood on the bank of the river, which has been gradually receding from it. Its name also would seem to point to the same conclusion, as thattha means a "shore or bank," so that Nagar-Thatha, which is the common name of the place, would mean the "city on the river bank." Its date is not certainly known; but M'Murdo, who is gene