the other to the east, or S E. Both of these lakes are said to be situated within Thibet; as indeed some of the villages on the bank of the Gogra, much farther to the southward, are. By the construction of M. Anquetil's map, the site of these lakes is carried as high up as the latitude of 36°: however, by an analysis of the scale, they can hardly be higher than 334°; as will be presently shewn. Now, in the Lamas' map of Thibet, the heads of the Ganges and Sanpoo rivers (which last, is called the Burrampooter, in the lower part of its course) are separated only by a ridge of mountains, named Kentaifse. The head of the Ganges is composed of two streams, which run westward; and the southmost of these branches runs through two lakes, of which the first is named Mapana, and the second Lanken. These lakes are placed in respect of each other, as Tiefentaller's are; and the names of the corresponding ones, are Lanke in one, and Lanken in the other account: and Mansaroar, and Mapana. The names of the latter have certainly no affinity with each other; but the similitude in point of situation, remains. The river that runs from the Lanken lake in Du Halde, is named Lank-tshou: and that from the Lanke lake of Tiefentaller, is the Gogra. With regard to the two rivers that issue from the Mansaroar lake of Tiefentaller, that which runs to the west is said to be the Satloudj (Setlege), a river that has been noticed in page 102, as the easternmost of the five Panjab rivers. This particular, however, the author himself discredits, and very justly: and the reader by turning to the map of the heads of the Indus, &c. at page 200, may satisfy himself as to the probability of it. In my opinion, this is the southernmost of the two heads of the Ganges, above noticed; and which is known to run by Dsaprong (or Chaparang), a considerable city not a very great way to the westward of these lakes (vide Lama's map in Du Halde, or D'Anville's map of Asia). If it be objected that the Ganges is too famous a river to escape the notice of the people who knew the Burrampooter, at its source; I answer, that I do by no means believe that the people

in Thibet know the Burrampooter, by any other name than that of Sanpoo; and that the word Burrampooter is an interpolation. Indeed the writing in the map, imports only, "it is said that the "Burrampooter, &c." and the translation of the Persic writing, at the efflux of this eastern river, gives a very different idea; being “Grand riviere qui va du côte de Neipal" (or Napaul). It is clear, however, that the people have an idea, that the western river has a very long course. I scruple not to believe, that the Lanke lake of Tiefentaller, is the Lanken of Du Halde; and that the Mansaroar of one, is also the Mapana of the other: and that the circumstances relating to the effluxes of the rivers, which appear to contradict each other, have not been carefully examined into, either by the one, or the other party. As the Mansaroar lake is said to be 60 milles Indiens (which ought to mean cosses) in circumference, that is, 115 B. miles; we can hardly suppose that the native who furnished the account, made the tour of it, either to measure its extent, or to explore the heads of the rivers: and on the other hand, the Lamas sent by CAMHI, might take a great deal of their account on trust. Therefore without contending about the exact circumstances of the case, I have described the Ganges (that is, the southern branch of it) as ifsuing out of the lake Mansaroar: but have not regarded the eastern river as the Burrampooter, or Sanpoo; because it would be doing too much violence to the account given by the Lamas; which describes the Sanpoo as ifsuing from the eastern side of mount Kentaifse, and at least 40 miles from these lakes. Besides, the Lamas give an account of a third lake, named Conghe, which intervenes between the Mansaroar lake, and the head of the Sanpoo. If I am mistaken in my conjectures, I cannot mislead others, while the map is accompanied with this discussion. The Lamas' map which appears in Du Halde, places the head of the Ganges in latitude 291°; and M. D'Anville found it necefsary to remove it almost as high as 32o. In the present map, it stands in 334°: all which may serve to show how vague a per

It also

formance the Lamas' map is, which errs 3o in latitude. placed Lafsa, which ought to be a well known position to them, a full degree too far to the south.

In the construction of the map of the Gogra, 32 cofses are reckoned to a degree, in a country, the most rugged and mountainous, imaginable; while 42 is the standard in the level country. A distinction, however, must be made in the scale of the upper and lower parts of the course of the river. For, from Mirzapour, which occurs in my map of Oude, to the lake Douloo-Sagur, the country is generally level: and was also explored by Tiefentaller himself; while the upper part is taken from the reports of a native. Therefore I have allowed 43 cofses to a degree, between those two points; so that the distance will be 98 G. miles, instead of the 131 in the map. Then for the upper part of the river, I have allowed 60 cofses to make a degree, instead of 32: by which the distance from the lake Douloo-Sagur, to the lake Lanke, is only 230 G. miles, instead of the 373 on the map. To those who may object to so great a diminution, as from 60 to 32; I shall observe, that 32, or even 37 cofses to a degree, is a proportion widely different from that which my experience suggests: and which allows 42 in the level country of Hindoostan proper (see page 5): and whosoever has travelled in very mountainous countries, and has moreover traced the course of a river through it; will not object to the diminishing to 1, what was 1,43 in the plains: for this is the proportion between 60 and 42.

If Tiefentaller's scale is right, the Lanke lake would be in latitude 36° and upwards; which I consider as highly improbable. It is certain that our best maps of Asia (that is, D'Anville's), present nothing but a blank space, in the part afsigned to the heads of these rivers, by M. Anquetil du Perron: and therefore, there is no positive evidence against it.

In ascending this river Gogra, we find noted in the map, not far within the first ridge of mountains, and near the second ridge,

three subterraneous caverns, from whence proceed with violence, fire, wind, and water. No other particulars are given. About 30 miles higher up, the Gogra forces its way through that vast ridge which has been often taken notice of before, under the name of Himmaleh, and which is a branch of the Emodus, or Imaus, of the ancients. This ridge, which is covered with snow, and visible from the plains of Hindoostan, appears to be the general boundary of Thibet, through that whole extent from the Ganges to the Teesta river; inclosing between it and Hindoostan, a tract of country, from a 100 to 180 miles in breadth; divided into a number of small states, none of which are understood to be either tributaries or feudatories of Thibet: such as Sirinagur, Almora, Kemaoon, Gorka, Napaul, and Morung. Bootan, a feudatory of Thibet, closes on the east of this tract; and is the extent of our knowledge, as to particulars, that way.

The small number of towns and villages on the banks of this river, induces one to suppose that the western part of Thibet, is no better inhabited than we have before described the eastern part of it to be. Indian names prevail, even within Thibet: and a temple of Mahadeo is found on the banks of the Mansaroar lake.

We have before stated the distance of the cow's mouth, or Gangoutra, to be about 280 or 300 road miles, above Hurdwar; where the Ganges enters the plains of Hindoostan. As the compass of Tiefentaller's map of the Ganges, declines 12 degrees more to the westward of north, than that of the survey does, between Allahabad and Hurdwar; this allowance is made accordingly: and Tiefentaller's distance exceeds that of the survey, within the same space, which is 331 G. miles, about 2 of the whole. Not being absolutely certain whether or not Tienfentaller took the latitude of Gangoutra, I did not venture to alter the parallel in which he has placed it, 33°: but contented myself with correcting the bearing 12 degrees; thereby shortening the distance, which was originally 240 miles, to 227. If the latitude was not taken, by celestial

observation, but the distance calculated on the scale of 37 cofses to a degree, the Gangoutra will still be placed too far to the north.

To sum up the whole information, collected from the different accounts of the upper part of the course of the Ganges, it appears that the two branches of it, which spring from the western side of mount Kentaifse, take their course westward, inclining considerably to the north, for a course of about goo miles, in direct distance; when meeting the great chain or ridge of mount Himmaleh, which extends from Cabul along the north of Hindoostan, and through Thibet, the rivers are compelled to turn to the south; in which course they unite their waters, and form what is properly termed the river Ganges. This great body of water now forces a passage through the ridge of mount Himmaleh, at the distance, pofsibly, of 100 miles below the place of its first approach to it, and sapping its very foundations, rushes through a cavern, and precipitates itself into a vast bason which it has worn in the rock, at the hither foot of the mountains. The Ganges thus appears, to incurious spectators, to derive its original springs from this chain of mountains: and the mind of superstition has given to the mouth of the cavern, the form of the head of a cow; an animal held by the Hindoos, in a degree of veneration, almost equal to that, in which the Egyptians of old held their god Apis.

From this second source (as it may be termed) of the Ganges, its course becomes more eastwardly than before, through the rugged country of Sirinagur; until, at Hurdwar, it finally escapes from the mountainous tract, in which it has wandered for about Soo B. miles. At Hurdwar, it opens itself a passage through mount Sewalick; which is the chain of mountains that borders on the level country, on the north of the province of Delhi. Even Sewalick would be deemed a lofty ridge, but for the presence of mount Himmaleh, or Imaus; which rises above it, when viewed from the plains of Hindoostan.

S $