We have thus got Ch’eng-tu fu as one fixed point, and Yun-nan-fu as another, and we have to track the traveller’s itinerary between the two, through what Ritter called with reason a terra incognita.  What little was known till recently of this region came from the Catholic missionaries.  Of late the veil has begun to be lifted; the daring excursion of Francis Garnier and his party in 1868 intersected the tract towards the south; Mr. T.T.  Cooper crossed it further north, by Ta-t’sien lu, Lithang and Bathang; Baron v.  Richthofen in 1872 had penetrated several marches towards the heart of the mystery, when an unfortunate mishap compelled his return, but he brought back with him much precious information.

[Illustration:  Garden-House on the Lake at Yun-nan-fu, Yachi of Polo.  (From Garnier).

“Je boz di q’il ont un lac qe gire environ bien cent miles.”]

Five days forward from Ch’eng-tu fu brought us on Tibetan ground.  Five days backward from Yun-nan fu should bring us to the river Brius, with its gold-dust and the frontier of Caindu.  Wanting a local scale for a distance of five days, I find that our next point in advance, Marco’s city of Carajan undisputably Tali-fu, is said by him to be ten days from Yachi.  The direct distance between the cities of Yun-nan and Ta-li I find by measurement on Keith Johnston’s map to be 133 Italian miles. [The distance by road is 215 English miles. (See Baber, p. 191.)—­H.C.] Taking half this as radius, the compasses swept from Yun-nan-fu as centre, intersect near its most southerly elbow the great upper branch of the Kiang, the Kin-sha Kiang of the Chinese, or “River of the Golden Sands,” the MURUS USSU and BRICHU of the Mongols and Tibetans, and manifestly the auriferous BRIUS of our traveller.[1] Hence also the country north of this elbow is CAINDU.

I leave the preceding paragraph as it stood in the first edition, because it shows how near the true position of Caindu these unaided deductions from our author’s data had carried me.  That paragraph was followed by an erroneous hypothesis as to the intermediate part of that journey, but, thanks to the new light shed by Baron Richthofen, we are enabled now to lay down the whole itinerary from Ch’eng-tu fu to Yun-nan fu with confidence in its accuracy.

The Kin-sha Kiang or Upper course of the Great Yang-tzu, descending from Tibet to Yun-nan, forms the great bight or elbow to which allusion has just been made, and which has been a feature known to geographers ever since the publication of D’Anville’s atlas.  The tract enclosed in this elbow is cut in two by another great Tibetan River, the Yarlung, or Yalung-Kiang, which joins the Kin-sha not far from the middle of the great bight; and this Yalung, just before the confluence, receives on the left a stream of inferior calibre, the Ngan-ning Ho, which also flows in a valley parallel to the meridian, like all that singular fascis of great rivers between Assam and Sze-ch’wan.