402 IN SEARCH OF THE LIMES TO SU-CHOU [Chap. XI
Limes agger or south towards the Su-chou oasis, I do not consider the question at present capable of a definite answer. Two possible solutions seem to present themselves. I f the Limes wall constructed along the foot of the outermost Pei-shan range from the left bank of the Pei-ta-ho to north of Hua-hai-tzû was ever actually garrisoned, the line of the towers T. XLIV. a—d may have been intended to safeguard an important line of communication and supplies leading to it from the side of the Su-chou oasis ; for this must have been always the main base for the defence of this portion of the border. Another suggestion which may conjecturally be put forward is that some time after the actual construction of the Limes agger between the Pei-ta-ho and Hua-hai-tzû the impossibility of effectively guarding this section, owing to the great distances from water and habitable ground, was duly realized. The attempt to guard the Limes east of the lowest portion of the Huahai-tzii depression, i. e. east of the ground around Ko-ta-ch`üan-tzû, may then have been abandoned, and the line of watch-stations, not provided with a wall but within sight of each other, may have been substituted with a view to protecting Hua-hai-tzil and the Limes north of it from attack on the eastern flank. But on this assumption there would remain the problem how the Su-chou oasis and the cultivated tract about Chin-tea could have been protected from irruption through the gap thus left open. [In any case it deserves to be noted that the three dated documents from the watch-towers T. XLIV. a—d all belong to the Later Han period, while the four dated documents from the Limes section T. xLIII west of Hua-hai-tzti all go back to the times of the Former Han dynasty.]
Another question which I must be content to leave for elucidation by a future investigator is that of the exact connexion of the line of the Limes between the easternmost point where we traced it from Ko-ta-ch`üan-tzû, and the section marked by the towers T. xLv. a—h (Map No. 42. c. 4) north of the Pei-ta-ho bend. It was impossible for me to attempt to follow the line farther eastwards in person. The two water-tanks that alone remained available would have been insufficient to supply my comparatively large party with water for a number of days on that utterly arid ground. There was also the imperative consideration that I must visit Su-chou as early as possible in order to make timely arrangements for my intended explorations along the Etsin-gol ; I was anxious to carry these out before the summer heat set in. So the task of tracing the continuation of the wall to the east had to be left to Lâ.l Singh alone, travelling with very light baggage. His report, when he subsequently rejoined me at Su-chou, showed that he had lost all indications of the line within less than two miles from where he had first struck it, owing to a big belt of dunes, and after getting round this, had searched in vain for any remains of the agger on the open gravel Sai beyond. The scantiness of his water-supply had then obliged him to turn to the south-east and to seek the northern edge of Su-chou cultivation on a line lying across the unsurveyed wastes of gravel and stone which divide the Hua-hai-tzû basin from the Pei-ta-ho drainage.
My own party started on May ist for Su-chou by the caravan track ; after a march of some twenty-six miles across utterly barren stony plateaus and a low hill chain, this brought us to the northernmost edge of outlying Su-chou cultivation (Map No. 43. A. I). Professor Futterer has fully described the ground that we passed on this march, as we skirted the northern and north-eastern foot of the high and rugged transverse range that overlooks on the south the defile of Chia-yü-kuan.5 The low hill chain that stretches eastwards from the range and continues in this direction as far as the Kan-chou river is crowned by large watch-towers. These were obviously intended to serve as advanced look-out posts for the mediaeval ` Great Wall ' that extends along the southern foot of the hill chain. Like the towers in similar positions examined by me in 1907 near Chia-yü-kuan,6 they all bore marks of late origin.
5 See Futterer, Wüste Gobi, pp. 28 sq. 6 Cf. Serindia, iii. pp. s i 19 sq.