294 THE LOP-NOR PROBLEM.
map shown any closer agreement than what it actually does show with my own map. On the contrary I have endeavoured to prove that for thousands of years the river has been levelling down its basin, and making its accompanying belts of vegetation precisely as broad as they are here. In the description of the river Tarim given in the first volume of the present work, we had a whole series of river-beds which successively dried up one after the other, and all lying south of the existing river-bed. When travelling from Kerija to Schah-jar I also crossed over an Atschikdarja lying south of the Tarim of the Wu-tschang map. The Chinese delineation of the river thus falls in point of latitude between the present Tarim and its dried up arms, and hence may be regarded as being perfectly correct.
The two positions of the Baghrasch-köl are, we have found, fairly in accord the one with the other. From that focus of greatest exactitude the errors appear to increase outwards in every direction. At Lop-nor the error in respect of latitude is extremely slight, only a few minutes, and as for the Kontsche-darja, it is not at all unlikely that the river did have then the course which is assigned to it on the map, except that its southern part is placed, like the Lop-nor, too far to the west, as well as at Korla, where the longitude is otherwise correct, too far to the south. Lop-nor is put I 3/4° too far to the west. That however is a venial error in a Chinese cartographer, seeing that even such an experienced traveller as Prschevalskij obtained such different longitudes as 89°35' in the year 1877 and 88°59.8' in 1884-85, the difference being due to the adoption of a more reliable method on the second occasion. The different positions he assigns to the lake are shown on plate 35. It is indeed matter for congratulation that the Chinese cartographer's errors are not more than three times as great as those of Prschevalskij.
We have found the greatest errors in latitude in the region immediately south of Lop-nor. While, generally speaking, the northern portion of the map, with the exception of the Baghrasch-köl and the Lop-nor, is carried too far to the south, the region of Kara-koschun, Khas-nur, and Tsajdam are pushed too far to the north. The consequence is that along the line where the two sections meet, the map is distorted. We are, it is true, told that four elongated lakes lie south of Lop-nor, but we are not told at what distance. Since the distance between the southern shore of Lop-nor and the northern shore of Khas-nur amounts to only one-quarter of the real distance, it is obvious that the lakes, and mountain-ranges, and roads which exist between these two lakes are much too crowded together for them all to find room on the map. If now Lop-nor is transposed 13/4 to the east and Khas-nur 13/4 to the south-east, so that both lakes occupy their true positions, then naturally the geographical features which come between them will experience a similar transposition, the direction being a resultant line between east and south-east, i. e. east-south-east, and the distance arranged proportionally to the transposition of the Khas-nur. The result is that the four small lakes, together with the fifth, the Ike-ghaschon, the name of which only is given, come to occupy precisely that part of the basin in which we now find the Kara-koschun. Consequently this lake is identical with the Ike-ghaschon and Bagha-ghaschon of the Chinese map; and this is what I wanted to prove (see PI. 34).
At the same time it is quite clear that the cartographical material is derived from different epochs. The positions of the small lakes relatively to one another is
IIIFカラー高解像度
