PA renee ash a Sei oe) * is Ok to yh ee ma aca Se a eer Cee Cu ay. ¢ «) - UNIVERSITY OF ILLINGIS LIBRARY AT URBANA-CHAMPAIGN BOOKSTACKS NOTICE: Return or renew all Library Materials! The Minimum Fee for each Lost Book is $50.00. The person charging this material is responsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN JAN 2 ( 1989 “JUL 15 19 L161—O-1096 Digitized by the Internet Archive in 2022 with funding from University of Illinois Uroana-Champaign Alternates https://archive.org/details/ruinsotdesertcatO2stei_ 1 ts Viel, - | 7 f : ‘ Hine Aa, Ora eA bots 84 ’ 7 . i ie) A RUINS OP SDE GER TC AT EASY MACMILLAN AND CO., LIMITED LONDON + BOMBAY + CALCUTTA MELBOURNE THE MACMILLAN COMPANY NEW YORK + BOSTON + CHICAGO ATLANTA + SAN FRANCISCO THE MACMILLAN CO. OF CANADA, Lip. TORONTO Y Seth poly a vat. ~ * (as < ov i ~Ly - oun a RUINS OF DESERT CATHAY PERSONAL NARRATIVE OF EXPLORATIONS IN CENTRAL ASIA AND WESTERNMOST CHINA By M. AUREL STEIN WITH NUMEROUS ILLUSTRATIONS, COLOUR PLATES, PANORAMAS, AND MAPS FROM ORIGINAL SURVEYS IN TWO VOLUMES Vee kl MACMILLAN’ AND CO., LIMITED ST. MARTIN’S STREET, LONDON ary CONTENTS 5 GIUAGE IA Nagle RUINS EN ROUTE TO TUN-HUANG DAA Rei] FIRST HALT AT TUN-HUANG Cor dele Reel To THE ‘CAVES OF THE THOUSAND BUDDHAS’ Gi Pal RS Ie PrT A DIFFICULT START FROM TUN-HUANG GEAR ER RATELY, By THE ANCIENT WALL NORTH OF TUN-HUANG (Oe GoW IRS DISCOVERY OF HAN RECORDS GUA Bel bey DV] To THE NAN-HU OASIS Vopr LT Vv ne S489 PAGE 20 32 44 54 67 VI CONTENTS CREDA Pale Rear ANCIENT REMAINS FOR THE FUTURE A Pai ho val FIRST EXCAVATIONS ALONG THE WESTERN LIMES CH AWD Reais RECONNAISSANCES ALONG THE ANCIENT WALL GHAPT ERS Tex DISCOVERIES BY THE ‘JADE GATE’ GHA P Tt heaiixr THE GREAT MAGAZINE OF THE LIMES CHA Pah hele ON THE WESTERN FLANK OF THE LIMES CHAP PEK ais i RECORDS FROM AN ANCIENT WATCH-STATION . CRA PoDERs eax RETURN TO THE ‘THOUSAND BUDDHAS’ CHAPTER ERY, FIRST OPENING OF THE HIDDEN CHAPEL PAGE Qg2 IO! EZ 123 135 146 171 CONTENTS CHAP PER VEXVI A WALLED-UP LIBRARY AND ITS TREASURES CFLAIP GER LA VIE BUDDHIST PICTURES FROM THE HIDDEN CHAPEL CHARI ER VexvViil! LARGE PAINTINGS AND OTHER ART RELICS CHART Is Ry HxIx A POLYGLOT TEMPLE LIBRARY CHAP WER ey DECORATIVE ART AT THE ‘THOUSAND BUDDHAS’ GTA; Pali Rea Peel At AN-HSI, THE ‘ WEST-PROTECTING’ . GHAP TERS LX XIT THE RUINS OF CH’IAO-TZU CEUVE TE ERX RTE THE ‘VALLEY OF THE MyrRIAD BUDDHAS’ GHP EER® LX XIN IN THE MOUNTAINS OF THE WESTERNMOST NAN-SHAN vil PAGE 182 195 204 220 235 247 255 262 Vill CONTENTS CHAPTER LAXV By THE GATE OF THE ‘GREAT WALL’ CHARTERDEXKAN At Su-CHOU AND ITS ‘SPRING OF WINE’ GH ARID E RICA iil THROUGH THE RICHTHOFEN RANGE OF THE NAN-SHAN CHARI hs DAce vai ACROSS THE TO-LAI-SHAN RANGE CHAP TEREX TX FROM THE Su-LO Ho sourRcES TO KAN-CHOU OT AE Te ees From KAN-CHOU TO THE T’IEN-SHAN . CHAP DE Reese] AT THE HAMI OASIS . CHA PT ER® DXXXII GLIMPSES OF TURFAN RUINS . CHA PR Rigi er LT KARA-SHAHR AND ITS OLD SITES PAGE 297 B22 334 342 364 CONTENTS CHAP T RASC RXATV FroM KuHorRA TO KUCHAR CHAPTER LXXXV IN THE ‘SEA OF SAND’ CAA TIGRIS XX Vit IN A DEAD DELTA CHAR AT RIv EXON LT SALT MARSH OR ICE? GEHAAWILE Rex XV TIT By THE NEW KERIYA RIVER-BED CHAR iE Ree AOOX EX MorE TAKLAMAKAN RUINS GHA PTE RaUx © From Ak-su TO YARKAND ae Valet bik. ¢ OI PREPARATIONS AT KHOTAN Ole so) od EPR OCD UI IN THE GORGES OF POLUR AND ZAILIK IX PAGE 372 379 39! 399 406 412 420 431 440 Xx CONTENTS CGHAP DERE XCMT To THE YURUNG-KASH GLACIER-SOURCES CHARIDEIRE EXULV; Across TIBETAN PLATEAUS CHAP EAR MENG Vi ON AN OLD MOUNTAIN TRACK Calis Cie 52d RG el Gah GCN THE SEARCH FOR THE YANGI DAWAN CHAP rae Cy lt FROM THE KUN-LUN TO LONDON INDEX PAGE 449 456 465 474 483 493 PU ae Re eTiOENsS PAGE Author’s Portrait : : ‘ Frontispiece 155. Shagolin-Namjil Range, Nan-shan, seen from Camp ccxv., across basin of Su-lo Ho head-waters : 4 156. Ruins of ancient Chinese magazine T. xvu., Tun-huang Limes, seen from south . F : : 4 157. Barren hill range east of Ch’ien-fo-tung Valley, seen from south end of site : 22 158. Middle group of ‘Thousand Buddhas’ Protec cin bee ning of southern group . : ; : 24 159. Rows of cave temples, showing aesinel porches, near middle of southern group, ‘Thousand Buddhas’ Site. 26 160. Interior of cave temple Ch. vitt., ‘Thousand Buddhas’ Site, showing fresco decoration of walls and roof . : 28 161. Stucco image group, representing Buddha between disciples, etc., In cave temple Ch. m1., ‘Thousand Buddhas’ Site. 30 162. Ruined watch-tower, T. xxv., north of Tun-Huang oasis, seen from south-west ’ d ; ‘ : 46 163. Face of ancient border wall, near tower T. xxxv., Tun- huang Zzmes, showing construction with alternate layers of stamped clay and fascines : : ; : 46 164. Remains of ancient watch-tower T. xxvi., Tun-Huang Limes. ; : ‘ : ‘ 52 165. Remains of ancient border wall, netieen low dunes, east of tower T. xxxv., Tun-Huang Zimes ; , 52 166. Ruined homestead, abandoned to desert about sixty years ago, at Kuan-tsou, north of Nan-hu ; : ; 76 167. Village shrine at Nan-hu, with school-room on right ; 76 XI XII 170. 172. 172. 173; ay fe iehve 178. 179. 18o. TST, 182, 184. ILLUSTRATIONS . Ruin of farm abandoned during last century, in desert north of Nan-hu . Ruined watch-tower, T. xu., on ancient Chinese Zzmmes, north-west of Tun-huang Ruined watch-tower, T. 1x., on ancient Chinese Zzmes, north-west of ‘Tun-huang Guard-room built against north-east corner of ancient watch- tower T. vul., Tun-huang Zzmes, after excavation Ancient pottery and implements excavated from ruined watch-stations on Tun-huang Limes 174. Ancient implements and articles of equipment, ex- cavated mainly from ruined watch-stations of early Chinese border line, Tun-huang : Stretch of ancient border-wall, built of layers of reed fascines and clay, east of tower T. xu, Tun-huang Limes. : ‘ . Remains of ancient border-wall, adjoining salt marsh, to west of tower T. xiv. a., Tun-huang Zzmes, seen from south Ruined watch-tower, T. xu. A., with remains of adjoining quarters and stairs, Tun-huang Lzmes Remains of ancient watch-tower T. xx., overlooking lake west of Khara-nor, Tun-huang Limes Hillock with remains marking position of ancient ‘Jade Gate’ station, near fort T. x1v., Tun-huang Zzmes Ruin of ancient watch-tower T. vi. A., on western flank of Tun-huang Zzmes, seen from south Ruin of ancient watch-tower T. Iv. c., on western flank of Tun-huang Zzmes, with view to north Remains of ancient watch-tower and quarters T. vi. B., Tun-huang Zimes, before excavation, seen from west . Rubbish-strewn slope below ruined watch-tower T. vi. B., in course of excavation Ground-plan of ancient watch-station T. vi. B. PAGE 88 88 94 96 96 98 108 108 II4 Tp 116 136 136 144 144 145 185. 186. 187. 188. 189. Igo. IQI. 195. 196. 1Q7. 198. 199. ILLUSTRATIONS Temple court at ‘Crescent Lake,’ Tun-huang Cliffs with main cave temples of ‘Thousand Buddhas’ Site, seen from west Wang Tao-shih, Taoist priest at Thoasand Buddhas’ Site, Tun-huang : : Cella and porch of Wang Tao- shih’s cave temple, ‘Thousand Buddhas’ Site, Tun-huang : Modern frescoes in loggia of Wang ‘Tao-shih’s temple, representing adventures of Hstian-tsang : Tibetan monk in loggia of Wang ‘Tao-shih’s temple, ‘Thousand Buddhas’ Site : : Old Chinese manuscripts and blockprints from walled-up temple library of ‘Thousand Buddhas’ Site . Ancient manuscripts in Sanskrit, Central-Asian Brahmi, Sogdian, Manichaean-Turkish, Runic Turki, Uigur, Tibetan, from walled-up temple library, ‘Thousand Buddhas,’ Tun-huang . Gigantic roll of paper, with Sanskrit and ‘unknown language’ texts in Brahmi script, from walled-up temple library, ‘Thousand Buddhas,’ Tun-huang . Bundles of old manuscript rolls, mainly Chinese, in original wrappers, from walled-up temple library, ‘Thousand Buddhas,’ Tun-huang : Painted silk banners, of T’ang period, representing Bodhisattvas, from walled-up temple library, ‘Thousand Buddhas,’ Tun-huang : Modern stucco images in Chinese style, representing Hstian- tsang as an Arhat, with attendants, in a cave temple of the ‘Thousand Buddhas,’ Tun-huang Old silk embroidery on cushion-cover from walled-up temple library, ‘Thousand Buddhas,’ Tun-huang . Ruined Buddhist grottoes, near Wang ‘Tao-shih’s cave temple, ‘Thousand Buddhas,’ Tun-huang Rows of small Buddhist cave temples seen from the shrine of a colossal Buddha, in southern group of ‘Thousand Buddhas’ Site, Tun-huang XIII PAGE 162 164 166 168 170 170 176 180 184 190 198 208 208 218 220 XIV 200. 201. 202. 203. 204. 205. 210. 217. ILLUSTRATIONS Stucco images of Buddha with disciples and Bodhisattvas, partly restored, in alcove of a smaller cave temple, ‘Thousand Buddhas,’ Tun-huang Alcove of Buddhist cave temple, ‘Thousand Buddhas, with modern stucco images of Buddha between disciples, Bodhisattvas, and Lokapala Fresco composition on wall of cave temple, ‘Thousand Buddhas,’ Tun-huang, showing scene in Buddhist heaven Frescoes in north-west corner of large cave temple Ch. vim, ‘Thousand Buddhas,’ Tun-huang : Fresco composition representing Buddhist stories, on west wall of large cave temple Ch. vu., ‘Thousand Buddhas,’ Tun-huang ; . . : Large fresco composition on south wall of porch in cave- temple Ch. vi., ‘Thousand Buddhas,’ showing Buddha on Car . Fresco compositions in north-west corner of large cave temple, Ch. xvr., ‘Thousand Buddhas,’ Tun-huang . Fresco composition in south-west corner of large cave temple Ch. xvi., ‘Thousand Buddhas,’ Tun-huang . Wind-eroded walls at north-east corner of ruined town, An-hsi . Wang Ta-lao-ye, trea of Tun- Fhe with his wife ae mother ; : : : Wind-eroded ground at foot of gravel glacis, east of ae town of Ch’iao-tzt . Gate pavilion of old temple in Ch’lao-tzt village : . Ruined Stupa at old town of Ch’iao-tzt, seen from south . . Cave temples of the ‘Myriad Buddhas,’ on left bank of Shih- pao-chéng stream . Cave temples of the ‘Myriad Buddhas’ on right bank of Shih-pao-chéng stream . Frescoes in the antechapel of a cave temple at the ‘Myriad Buddhas’ Site . . Mongol camp on east side of Ta-kung-ch’a Valley : Hassan Akhun packing camel at Su-chi-ch’tian spring PAGE 222 224 226 228 230 232 218. 219. 220. 221. 222. as é . The Chia-yii-kuan gate of the ‘Great Wall,’ seen from ILLUSTRATIONS View south-west towards snowy main range from Ch’ang- ma village : ; : Interior of walled village of Ch’ang-ma, looking to north- west ; : ‘ . View from Tu-ta-fan to south-west, showing a portion of To-lai-shan Range i : Fortified village at Ch’ang-ma oasis, with view to south-east towards Su-lo Ho ; 3 Watch-tower of modern guard-station at Ta-han- a ane at foot of Nan-shan : Segment of ancient border wall north of Su-chou south-west . Pavilion over inner west gate of Chia-yii-kuan, with view across interior of circumvallation . . Hao-shan-kou gorge, with ruined walls intended to close passage . . Temple of Kuan-yin or Avalokitesvara within east gate of Chia-yu-kuan . Pavilion and colonnade at entrance of Chiu-ch’tian temple, Su-chou . Garden and temple court at Chiu. ea the ‘spine of Wine,’ Su-chou . . North wall of Chin-fo-ssti town, ties foot- ‘hills aE Reale hofen Range in background . Chinese villagers at Chin-fo-sst' watching my tent . View south from Hou-tzti Pass across dry lake-basin . Camp at Ch’ing-shui-k’a-tzt, on left bank of Ma-so Ho . Tungan gold-miners from Hsi-ning . View to S.-W. from ridge above Ta-pen-ko, crc. 14,000 feet high, showing portion of To-lai-shan . . View to S.-E. from Chu-lung-kuan Pass, showing portion of To-lai-shan Range . On crest above Huo-ning-to Pass, looking N.-W. towards Pei-ta-ho Valley . View down the Pei-ta-ho Valley ‘ror left tae of river XV PAGE 268 316 316 ILLUSTRATIONS . Snow-fields at head of valley east of pass across Alexander III. Range . Snowy peak seen eastwards from pass across Alexander III. Range . View to east from above left bank of Su-lo Ho . Large dunes on marshy ground in head-waters basin of Su-lo Ho . View to west towards Shagolin- Namji Peaks from Su-lo Ho-—Pei-ta Ho watershed . View from Su-lo Ho—Pei-ta Ho watershed, continued towards south-west . View south from Shen-ling-tzti Pass across Kan-chou River valley towards To-lai-shan . Head of Alpine valley at Camp ccxx1., north-east of Shen- ling-tzii Pass . My Chinese pony-men Wee Se to te plains . . First Mongol camp met with in Khazan-gol Valley . . Camp on left bank of Khazan-gol packed for start . View to south from first forest-clad ridge above Khazan-gol . . View south, from Féng-ta-fan towards Richthofen Range, across Khazan-gol Valley : . Mongol head-men near grazing-grounds of Lao-t’u-kou . Fortified village of Sha-ching-tzii, with temple gate, on road to Kan-chou . Central gate tower and main street in Kan-chou . Memorial gateway, built about a.p. 1825, outside west gate of Kan-chou - Ornamental gateway in front of commander-in-chief’s Ya-mén, Kan-chou - Temple at station of Hsing-hsing-hsia, on road from An-hsi to Hami . Gorge above Su-bashi, on road from Turfan to Kara-shahr . . Ridge with ruined Buddhist shrines in Wang’s orchard at Ara-tam, Hami . . Ruined vaults and ASH ae near west wall of Chong- Hassar, Turfan PAGE 33° $37 334 334 336 336 338 338 350° 35° 261. 263. 273. 277: 278. ILLUSTRATIONS Ruins of small Buddhist Stupa and shrine at Kichik-Hassar, Turfan . Ruined shrines and cave temples on west side of Toyuk gorge, Turfan ; : ‘ : Central main street in ruined town of Yar-khoto, Turfan, seen from north . Portion of ruined town of Yar-khoto, seen from near its south end . My camp at the ruins of ‘ Ming-oi,’ near Shikchin, Kara- shahr . Interior of ruined Buddhist cella Mi. xvit1., at ‘ Ming-o1’ Site, Kara-shahr . General view of ruined Buddhist site of ‘ Ming-oi,’ Kara- shahr, from south . Ruined Buddhist shrines in centre of ‘ Ming-oi’ Site, Kara- shahr, seen from north . Relievo images in. stucco representing Bodhisattvas, in ruined Buddhist shrine Mi. x1., ‘ Ming-oi’ Site, Kara-shahr. . Stucco heads, showing Graeco-Buddhist art influence, from ruined shrines, ‘ Ming-oi’ Site, Kara-shahr . Stucco heads and busts from relievo decoration of ruined Buddhist shrines, ‘ Ming-oi’ Site, Kara-shahr . Stucco head in high relievo, probably representing a Bodhi- sattva, from ruined shrine Mi. xv., ‘Ming-oi’ Site, Kara- shahr : . : Miscellaneous wood-carvings and stucco relievo from sites of ‘Ming-oi,’ Khora, ‘Thousand Buddhas,’ and ‘Tun- huang Limes . Relievo tile in terracotta and miniature shields in stucco from ruined shrines, ‘ Ming-oi’ Site, Kara-shahr . Relievos and decorative carvings in wood, from ruined Buddhist shrines, ‘ Ming-oi’ Site and Khora . General view of Buddhist temple ruins, Khora, from north- east : : . . Tahir Beg and Ahmad, Yiiz-bashi, of Korla Bakir, player of Rabab, on desert march oa XVIII 279. 280. 298. ILLUS FRATIONS Dunes in dry river bed near Charchak Darya Musa Haji between two other hunters from Korla . . My caravan marching over high dunes in Taklamakan, south of Tarim River . Caravan on march near dry river bed at end A ancient Keriya River delta . Halt on march across dunes and dead jungle of dried-up delta of Keriya River . Halt amidst dead jungle on left ee of dry river bed in ancient delta of Keriya River . Camels waiting for their first drink after arrival at Keriya River . Shahyar labourers slaking their thirst after reaching ice of Keriya River end . My tent at first camp after reaching A eeiiye River end . By the new bed of the dying Keriya River . 4 . Ancient quadrangle partially buried under dunes, Kara- dong Site . Ancient dwelling near south end of Kara- Aisne Site before excavation . The same ancient dwelling of rae Site in course of excavation 2. Pullat Mullah and Ibrahim, of Khotan, ‘treasure-seekers ’ . Mendicant pilgrim or ‘Diwana’ at Burhanuddin’s desert shrine, Keriya River . Buddhist shrine on tamarisk cone at site of Farhad: -Beg- yailaki, in course of excavation . Remains of Buddhist shrine at Kara-yantak, near Domoko, in course of excavation . My companions and myself at Ulugh-mazar, in the desert north of Chira . Ruined fort and watch-tower on Mazar-tagh ridge, seen from west : ; ‘ . Digging up ancient documents from refuse layers below ruined fort of Mazar-tagh PAGE 378 386 404 412 414 414 416 418 418 322. 323: oA, [LEVSRAPTIONS . My camp ina Beg’s garden at Ak-su . Crowd in Bazar street at Ak-su : . P’an Ta-jén, Tao-t’ai of Ak-su, my old patron and friend . Kirghiz with felt tent belonging to Mangush Beg, at Ilachu . . Kirghiz from grazing-grounds of Uch-Turfan . In cafion of Korum-boguz River north of Kelpin . Farewell to my brave camels from Keriya . . Carpenters at work on packing-cases for antiques, in courtyard of Nag-bagh . Turdi, my Dak man from Khotan . : : . Chiang-ssit-yeh at work on ancient Chinese records, in Nar-bagh . Badruddin Khan, Indian ae a at Paice aoe his sons and a trusted servant . Khuda-berdi (Yiiz-bashi) and Saliatoee of Frotan: . Village mosque at Ak-yar, on road to Uch-Turfan . . Mosque at Ruknuddin Mazar, Yotkan . Landing baggage from ferry-boat on right bank +f viene kash River . View eastwards from above Kar-yagdi, in Polur Gorge . Our advance party emerging from Polur Gorge near Khan Langar . Track in Polur Gorge near Sarik-koram, viewed from south. . Pasa of Keriya, hunter of wild yaks, and our guide . Abandoned gold pits in conglomerate cliffs of Zailik Gorge . Group of Zailik miners who served as carriers . View up the Yurung-kash River gorge from debouchure of Zailik stream . View south towards snowy range from ridge above Shalgan Davan, crc. 17,700 feet high : View down the Yurung-kash River gorge from debouchure of Hasib Chap . , View towards main Kun-lun Range, Aci of sehanmeends River, from camp below Tar-kol Pass : View south-west towards main Kun-lun Range, from triangulated peak above Tar-kol Pass XIX PAGE 420 420 422 424 452 456 xX [LEUST RATIONS 325. Zailik Gorge below Saghiz-buyan, viewed from west 326. Donkeys descending rock couloir from spur above head gorge of Yurung-kash River : ‘ 327. View down the head gorge of Yurung- er River, from an elevation of czvc. 15,000 feet : , 328. View south towards glaciers of main Kun-lun Range from head-waters basin of Yurung-kash 329. Slate cliffs rising above right bank of Yurung- De near debouchure of Tiige-tash stream d : : 330. Distant view of glaciers feeding the Yurung-kash sources, from pass east of Ulugh-kol : 331. Glaciers overlooking the basin of the Keriya River sources from west : : : 332. View across south-west end of Lake Lighten 333. ‘Witnesses’ in wind-eroded dry lake-bed near camp of September 16th, 1908 COLOUR PVATES “PANORAMAS A Nive Vailas Plate VI. Ancient Buddhist banner showing scenes from Buddha’s life-story. From ‘Caves of Thousand Buddhas’ » WII. Ancient Buddhist banners showing figures of Bodhi- sattvas. From ‘ Caves of Thousand Buddhas’ » WIII. Large Buddhist painting, of 864 a.pD., representing Bodhisattvas. From ‘ Caves of Thousand Buddhas’ 4s IX. Ancient embroidery picture showing Buddha _be- tween Bodhisattvas. From ‘Caves of Thousand Buddhas’ f X. Panoramic view taken from above Huo-ning-to Pass, To-lai-shan Range, Nan-shan », XI. a. Fresco showing scene from life of Buddha. From ruined temple of ‘ Ming-oi’ Site, Kara-shahr PAGE 458 458 460 462 462 464 464 466 466 200 202 204 206 314 414 ILLUSTRATIONS ot PAGE Fresco representing Hariti, the goddess of small-pox. From ruined Buddhist shrine at Farhad-Beg Site . 414 . Panorama showing Kun-lun Ranges around glacier- sources of Yurung-kash River 454 Panoramic view taken from snowy col on watershed of main Kun-lun Range, north of Haji Langar 478 Map II. Showing portions of Kun-lun Range, Chinese Turkestan. Scale 1: 1,000,000 . At end of volume. Map III. Showing portions of Western and Central Nan-shan. Scale 1: 1,000,000 . At end of volume. Ba alee Rm a Ba Ra ; | J i i Is Cr. a | se mn 4 “« ee): Lene a} i , \ ae va a) : a ipa bree bit rit is nth oh : y ih ae os a. - at ele ie GEA PapE Rutile RUINS EN ROUTE TO TUN-HUANG A NUMBER Of archaeological indications rapidly gathered in the course of that first day convinced me that the ruins I had passed, and those to be expected in continuation eastwards, belonged to an early system of frontier defence corresponding in character to the extant ‘Great Wall’ on the Kan-su border. That I should have to return to them for thorough exploration as soon as men and animals had recovered from their fatigues by a short rest at Tun-huang was quite clear to me. Yet no chances of getting more familiar with details of the old Lzmes were to be forgone in the meantime. So on the morning of March goth, 1907, while the animals were allowed to enjoy grazing a little longer and the men to take it easy over packing, I retraced last night’s route until I came again upon the line of the wall. It was now seen to turn off north, and to run straight down at right - angles to the shore of the small lake near the end of which we had camped. I was able to trace the layers of clay and fascines, so impregnated with salt as to look quasi-petrified, to within twenty-five yards or less of the salt-encrusted lake shore. That the level of the latter lay only four or five feet below the exposed base of the wall was an important observation. The extent of local desiccation since the wall was built could not have been great here. It was still more interesting to note how the lake had been utilized as a substitute for the strange wall elsewhere guarding the line. It was evident that those who laid down the line were eager to make the most of natural obstacles and thus to save building labour. VOL. II B 2 RUINS EN ROUTE TO TUN-HUANG cut This conclusion was soon confirmed after we had started on the day’s march. Having skirted the winding south shore of the lake for about a mile and a half, the track took us to the foot of a steep ridge which edged the lake on its east side. On the highest knoll overlooking the route there rose a massive square watch-tower, T. x1., surrounded by a crumbling wall of clay. The latter looked rough and of late origin. But a short scramble along the back of the ridge sufficed to reveal again the line of the old Lzmes wall with its characteristic reed fascines. It started from the lake shore opposite to the one where | had last traced it, and crossed the ridge down to another marsh basin. As I noticed two more towers beyond the latter east- wards, I felt assured now that the line of the wall ran more or less parallel to the end of the Su-lo Ho drainage, and that the route we were following would keep within it and probably near it. The next tower was passed, indeed, after about five miles from camp near the southern end of that second basin; but the wall was not traceable there, evidently running farther north. For the rest of the day’s march the succession of towers kept by our left above the grey horizon like a line of yellowish beacons. The stretch of scrubby desert or gravel Sai separating us from them was, however, too great to permit me to visit them without risk of losing touch with my caravan. Luckily the plane-table enabled us to fix their positions with pre- cision from the route, showing that the distance from tower to tower averaged two to three miles. At the end of close on ten miles by the side of a long-stretched depression full of luxuriant reed-beds and evidently containing springs, we came upon a small ruined fort of massive appearance, as seen in Fig. 154. Its walls, built of remarkably hard and well-laid strata of stamped clay, each about three inches thick, rose in very fair pre- servation to a height of nearly thirty feet. Fully fifteen feet thick at the base, they formed a solid square about ninety feet on each side. What splendid shelter they might give, not against human attack alone, but also against those cutting east winds, the very home of which dap AN ANCIENT FORT 4 we now seemed to approach! There was no trace of earlier quarters inside, and only scanty refuse from recent occupation by wayfarers. And yet, when I had climbed to the top by a rough staircase spared from the massive walls in a corner, and looked round over all this desolation, I felt sure that I stood on a structure which had braved man and nature for many centuries past. The view enjoyed from the top was wide and impressive. To the south I could see the scrubby depression merging in a belt of Toghrak and tamarisk jungle. Beyond there rose an absolutely bare gravel glacis towards the equally barren foot-hills of a great range far away. To the north- east four towers lit up by the sun behind us could be made out echeloned in the distance, silent guardians of a wall line which I thought I could still recognize here and there in faint streaks of brown shown up by my glasses. What a fine position, I thought, this height of the fort wall must have been for a commandant to survey his line of watch- stations, and to look out for the signals they might send along it! But how long ago was that? Those sombre, barren hills of the Kuruk-tagh, now standing out clearly again on the northern horizon, had seen wall and towers first rise, and would see their ruins finally disappear before the blasts of the ages. But it would be like asking Death itself for an answer. Somewhere between the foot of those hills and the line of towers the old drainage of the Su-lo Ho was bound to have cut its bed westwards. But even from that command- ing position I tried vainly to locate it. And yet, as our march continued across a sterile gravel plateau till the evening, I could see that the route was drawing nearer and nearer to a wide marshy basin, stretching east to west and manifestly part of the main Su-lo Ho valley. We had been skirting its steep clay bank for a mile or so, and were approaching a roughly built tower standing near it, when I saw in the twilight a huge structure rising before me from the low ground which fringed the basin. Hurrying to inspect it before it became quite dark, | found there three palace-like halls, with a total frontage of over 440 feet, and walls of great thickness rising to 4 RUINS EN ROUTE TO TUN-HUANG c.1 about twenty-five feet (Fig. 156). A natural clay terrace some fifteen feet high had been used as a base and added greatly to the appearance of height. There were remains of a massive walled enclosure with high towers jutting out at the four corners as if guarding a palace court. The sight of so imposing a building was doubly impressive for wanderers in the wilderness such as we had been for months, and the purpose of the grand ruin most puzzling. The position showed clearly that it could not have been intended as a fortified station. And what could have been the object of a palatial structure which comprised only three vast halls and seemed wholly to lack accessory habitations ? The problem was not solved that night. I found the men pitching camp near some springs about a mile farther east, close to some beds of dry reeds which seemed but to wait for a conflagration. After an incipient one had, luckily, been extinguished, a shift of camp became unavoid- able. In the darkness it took time to find a spot where the bare saline soil would safeguard us from that danger. But the inevitable delay had manifestly affected the temper of the more excitable people in my party, already tried by the long desert marches, and a succession of squabbles and affrays between Ramzan, my worthless Kashmiri cook, Ahmad, the servant of Chiang, and, alas! honest Naik Ram Singh, too, kept matters lively till mid- night. Asan offset to these petty worries, Hassan Akhun, the ever wide-awake camel-man, was able to hand me two copper coins which he had picked up in the evening while searching around the foot of the great ruin. They proved to be of an early Han type, and thus furnished the first distinct indication as to the antiquity of the site. Next morning in the bitter cold I examined the big ruin more closely, and soon ascertained all the main facts as to its plan and dimensions. But there was no clue to the real character of the imposing erection. The total absence of any other remains near by only added to the puzzle. Straight north there extended a wide salt marsh where there was neither need nor possibility of continuing the wall line. But both to west and east a succession of 155. SHAGOLIN-NAMJIL RANGE, NAN-SHAN, SEEN FROM CAMP CCXV., ACROSS BASIN OF SU-LO HO HEAD-WATERS. 150. RUINS OF ANCIENT CHINESE MAGAZINE T. XVIII., TUN-HUANG LIMES, SEEN FROM SOUTH. ; The figures of men standing at different points of the structure serve to indicate its size. CH. L ATPALATIAL STRUCTURE 5 towers was in view, clearly showing where ran the line which was to be guarded. Through my glasses I could see quite distinctly that the nearest towers were all built on small isolated clay ridges, such as rose in numbers from the flat of the marshy basin. Thus the constructors of the line had duly appreciated and used the advantages here offered for a widened outlook. But to me it was even more curious to notice the striking resemblance which these clay ridges and terraces, generally ranged in rows running north to south, bore to the eroded formations I had met in the dried-up basin east of Besh-toghrak. I could not have wished for a more exact reproduction of the aspect which that old terminal lake bed, and in all probability also the end of the ancient Lop-nor bed about Achchik-kuduk, might have borne at some earlier period. Another interesting illustration of physical conditions long past elsewhere was afforded by the rows of Toghraks which closely lined the lagoons and water-channels visible from afar within the wide marshy area, I thought of the lines of dead Toghraks I had crossed so often in the desert north of the present Lop-nor, and rejoiced at seeing the picture of the physical conditions I had conjectured as prevailing there before desiccation, now so faithfully materialized before my eyes. My examination of the ruin delayed me while the caravan moved ahead, and as, according to our guide, a long march was before us I had reluctantly to renounce for the time all reconnaissances off the route. This now took us for miles through belts of fine jungle and scrub, filling a succession of big bays which the marshy basin sent south. The track we were following had since the previous day shown numerous cart ruts, old and new, a clear indication that Chinese from Tun-huang were in the habit of using it. The grazing, too, looked inviting. I had been wondering for some time at the utter solitude when at last, after some nine miles of march, I noticed a little group of my men gathered on a reed-covered hillock round two strange-looking figures. These proved to be Chinese herdsmen from Tun-huang, clad in queer, heavily padded rags, looking after some cattle and horses. 6 RUINS EN ROUTE TO TUN-HUANG cui They were the first human beings we had seen for nearly three weeks past. Quaint specimens of humanity as they were, their appearance cheered up the spirits of my men greatly. I had never before had occasion to try my modicum of Chinese on people so humble in education and general intelligence, and that now, after repeated attempts, I succeeded in eliciting answers to some of my simple queries was felt by me no small encouragement. From them I learned that the place where we had met them was known as Shu-yu-t’ou, and that the cart tracks were those of people fetching timber and fuel to Tun-huang. The route still continuing eastwards then crossed a succession of long-stretched gravel-strewn ridges, which from the glacis-like Sai on our right jutted out to the north like the fingers of a hand. The reed-filled depres- sions between them connected with a broad salt-covered basin north, manifestly containing a river course or lake bed, but too far off for close survey. After about six miles from Shu-yu-t’ou the narrow continuous ridges gave way to a wide bay bare of vegetation, and covered with rows of those characteristic clay terraces already familiar from the vicinity of lake basins dried up or undergoing desiccation. All the terraces had their long side stretching from north to south. There could be no possible doubt that they represented the remnants of earlier continuous ridges, such as we had just marched across, which the erosive force of the violent east winds and of the sand driven before them had slowly sawn through and broken up. It was a very instructive illustration of a geological change still actually proceeding. The ridges themselves had evidently originated from the depressions between them having been scooped out by the drainage which during periods of much heavier precipitation came down from the foot of the mountains south, and cut up the clay sediments of a far more ancient lake bed. After another three miles of such ground we emerged on a level flat extending unbroken for three or four miles northward to the shore of a large sheet of dark blue water. At last we had come in sight of the Khara-nor lake, for which the map of Roborowsky and Kozloff had prepared us. But its CH. L NEAR LAKE KHARA-NOR i extent was much larger than there shown, and the wide, salt-encrusted edges indicated that its level would at times rise still higher. A number of small isolated clay terraces were seen scattered over the flat shore, manifestly the last survivals from terrace clusters and ridges which the relentless powers of erosion had long ago ground down and carried off. Two of them, not far from the present lake shore, could be seen crowned by watch-towers, for which they offered command- ing positions. But it was getting too late to approach them. Perched at the end of a long ridge projecting into the plain from the south there rose another ruined tower overlooking the route; which at this point turned to the south-east. One more great bay was crossed, filled with a succession of eroded clay terraces. There in the twilight we met for the first time a caravan, a big convoy of Keriya camels which had passed us at Miran at the beginning of February carrying the goods of some Khotan traders (Fig. 137), and which were now returning safely from Tun-huang. We did not envy the men their second desert crossing. Then the route led up a gently sloping alluvial fan, and at last in the dark, after a total march of some twenty-six miles, we pitched camp at a spring which our Abdal guide called Yantak-kuduk. The water of the spring-fed pool proved perfectly fresh, and far better than any we had tasted for a long time. The thorny scrub close by just sufficed for the animals, and as the oasis now lay within a day’s march, the morn- ing of March 12th saw the caravan start with unwonted alacrity. Ona small knoll to the south where we fixed the plane-table I observed a novel sight, a miniature shrine built of clay and evidently cherished by Chinese wayfarers ; for inside the tiny cella there lay votive offerings of papers and incense sticks. It served to remind me that we were approaching a region where Buddhism, or what figures as such in Chinese syncretistic belief, is still a religion in being. Nothing else on that day’s march indicated that we were moving towards a town of the living. For fully seven- teen miles we rode over a waste of gravel with practically 8 RUINS EN ROUTE TO TUN-HUANG cz.1 no vegetation. There was nothing to intercept the view on this sterile alluvial fan, and looking back we could see the expanse of Khara-nor and the sombre hills beyond it quite clearly. Twice we crossed ancient river beds deep-cut, yet quite dry, marking probably an earlier delta of the Tang Ho. The second showed some growth of reeds, and evidently received subsoil water. Just before reaching it I caught the first distant sight of a line of trees marking the Tun-huang oasis, and after marching four miles onwards we found ourselves almost suddenly stepping from the barren Sai across the edge of cultivation. The fine arbours and well-tilled fields, by contrast with the wastes we had passed through, looked inviting and neat, even in their wintry bareness. Half a mile onwards we came upon what looked like a dilapidated small fort now serving for cultivators’ quarters. The Chinese occupants, after some parley with Chiang-ssti-yeh, allowed us to pitch our tents on the clean threshing - ground outside their high clay walls. It was evident that strangers were indeed a novel sight to them; for all the time that camp was being pitched and for hours afterwards we were watched with the utmost curiosity by every able- bodied man in the place and swarms of lively children. There was a display of good nature all round, which was pleasing ; and when | had managed somehow to make myself understood on a few simple matters by the jovial unkempt rustics, all doubts about the first welcome which might await us on true Chinese soil passed off. My own tent, as always, was kept at a good distance from the noise of the general camp. Just in front of it rose a clum of elms, and under them a picturesque little Buddhist shrine adorned with good wood-carving and some bold frescoes representing the ‘Guardian divinities of the Regions.’ All the surroundings breathed a novel air of well-ordered civilization; and when the crowd of good- natured watchers had dispersed with the falling darkness, I had reason to feel gratified with my first place of rest within the purlieus of a celestial population. GEDA MAT Rais! FIRST HALT AT TUN-HUANG On the morning of March 12th, 1907, we were prepared to make our entry into Tun-huang town. All the men had been looking forward eagerly to our arrival. But circum- stances seemed to combine to deprive it of all state and even comfort. An icy gale was blowing from the east, and cutting as it was among the trees and houses, we con- gratulated ourselves inwardly that we had escaped it in the open desert. But what, somehow, seemed worse was that, though the town was said to be only some twenty Li, or about four or five miles off, no reply whatever was forthcoming to the elegant epistle which Chiang-ssiti-yeh, at the very time of our reaching the oasis, had despatched to the Ya-mén, along with that imposing Chinese visiting- card of mine on red paper. It had announced our arrival within the magistrate’s jurisdiction, suitably indicated my official rank and business, and expressed a request for the assignment of appropriate quarters. We knew from a letter received in December from my old friend P’an Ta-jén, Tao-t’ai at Ak-su, that he had duly recommended me to the authorities on the westernmost border of Kan-su. Now that we were on the soil of a truly Chinese pro- vince, my excellent secretary seemed to feel more than ever his responsibility and his importance in serving my interests. He was not a little perturbed by this evident want of official attention, and showed his chagrin freely. In Turkestan the prompt appearance of a Beg or two from the Ya-mén would have been a matter of course, and even before their arrival village head-men would have shown themselves eager to attend. But here we had evidently 9 10 FIRST HALT AT TUN-HUANG CH. LI to prepare for a different mz/zeu. The few villagers about seemed to be sufficiently absorbed in their own business not to pay much heed to the Ya-mén, and only in the course of the morning did Chiang learn that a newly arrived magistrate had taken over the seal of his office the even- ing before. This great function, with all its attendant flutter in official dove-cots, would of course account, at least partly, for the neglect we had so far experienced. So after some useless wait we set out for the town. Our atlases show it as Sha-chou, ‘the City of Sands’ ; but to the local Chinese it is best known by its ancient name of Tun-huang, dating from Han times. The bitterly cold east wind and the dust haze would have befitted the shores of Lop-nor. But riding on I was struck by the abundant signs of careful cultivation and the substantial look of the buildings in the many isolated farms we passed. Of traffic on the road, such as enlivens the march near any Turkestan town or market, there was strangely little. We passed several large walled enclosures which looked like ruined forts, but in reality were only deserted villages once vainly defended, grim mementos of the terrible loss of population which Tun-huang, like all the other oases of Kan-su, had suffered during the last great Tungan rebellion. At last we arrived by the river and found ourselves opposite the west face of the square-walled city. The river bed, then for the most part dry, was crossed by a dangerously rickety bridge, and then we rode through a ruinous town gate into the narrow main street of the place. Here, too, little life was stirring. In front of the outer gate of the magistrate’s Ya-mén we met at last a small crowd of idlers, and directed by them, made our way past a couple of picturesque half-decayed temples with fine old wood-carving to the Sarai suggested for our residence. It proved a perfectly impossible place, so filthy and cramped that I decided at once rather to camp in the open. Among the people to whom this queer hostelry gave shelter we found several Turki traders from Kashgar and Hami. I was trying to elicit from them information about more possible quarters when at last a wretched-looking Ya-mén attendant arrived to offer help. CH. LI SEARGHPP OR, QUARTERS I] He seemed half-dozed with opium and utterly helpless ; but stung into activity by Chiang’s voluble language he served at least to establish official touch with headquarters. I was already prospecting among gardens across the river for some suitable camping-place when there appeared on the scene the well-got-up ‘Ta-1’ of the magistrate, riding a lively donkey, and bringing profuse apologies from the newly installed dignitary. Local knowledge he had none to offer, having come with his master from another part of Kan-su. So I had to guide myself in the search. The few gardens to be found by the left bank of the Tang Ho were so small, and the pavilions or summer-houses adjoin- ing them so ruinous, that I had before long to shift my reconnaissance back to the side of the present town. There at last, half a mile or so from its south gate, [| discovered a large orchard with a lonely house at one end which looked as if it had seen better days. On invading its precincts we found it still occupied, but luckily by people who were ready to find room for so big a party as ours. They were the widow of the late owner, apparently a petty landholder, and her mother, along with a number of small children. Round an inner court were grouped several small blocks of rooms and a hall, most of them unoccupied but for quantities of cranky ponderous furniture such as respectable Chinese families seem ever fond of accumulating. The cracked walls, broken paper windows, and other abundant signs of long-continued neglect, made a strange contrast with all the tinsel and gilding which covered the elaborate carvings. I had hoped that, according to the custom prevailing in Turkestan, the women, after locking up their most cherished possessions, would clear out of the place and take shelter with relatives while the strange guests settled down in their house. But there was no such affectation of ‘Purdah’ on the part of our hosts, Cheerfully toddling about on their poor little bound feet the Chinese ladies huddled as much as they could of their household gods into one small block of rooms, while we were welcome to make what use we could of the rest. In a room close to them and chock-full of furniture Chiang managed to 12 FIRST HALT AT TUN-HUANG CH. LI effect a footing, his ever kindly and urbane presence being evidently welcome as a set-off against the invasion of us bearded barbarians. He made friends at once with the children, who were hugely enjoying all the unwonted bustle. Two large but ruinous rooms took in the Surveyor and Naik, with honest Jasvant Singh, who was glad to do his cooking under a roof even though half of it had fallen. The spacious central hall ought to have offered me shelter. But there was no trace of a fire-place or stove to warm it, and with gaping fissures in walls and roof it would have been quite impossible to overcome the freezing chill of the place. So it was allotted for the keeping of my boxes and the reception of official visits. I myself vastly preferred my cosy little tent outside among the leafless fruit trees of the garden. There was relative peace and plenty of fresh air, with a chance of warming oneself in the sunshine whenever the dust haze would allow this to break through. It was dusk before I could seek a little peace and warmth in my tiny travelling homestead. There had been trouble enough in settling down, quite apart from the long search for quarters. To secure the badly needed supplies for men and beasts proved a serious business. We had indeed arrived in the centre of a prosperous large oasis, but there was no obsequious Beg to take orders and produce what we needed; nor was it of any use to despatch the Ya-mén attendants to the town for what was most urgently wanted, unless they were provided with cash for immediate payment. And what a trouble it took to produce that cash in a form suited for local use! Nobody in Tun-huang would on any account take payment in the coined silver of the ‘New Dominion,’ and all the silver bullion I had brought consisted of big ‘horse-shoes.’ The expedient of sending one of them to the blacksmith to be cut up into chips for ‘small change’ did not dawn upon us that first day. At last one of the few Turki traders came to the rescue by changing a few Taels of silver into long sausage-like strings of copper ‘cash’ at a rate which suited his fancy. Even then it took hours before fuel, fodder, etc., arrived ; CH. LI NOVED DIBEICULITIES £3 for the market, here held daily, had long been closed. My men, the Indians included, naturally grew impatient and annoyed at the endless delays caused by what they took for cussed contrariness in the ‘heathen Khitai.’ I myself felt plainly brought face to face with a great shift of the social background. Here in the very centre of Asia I seemed somehow forced into touch again with features of civilization familiar enough in the far-off West. It amused me to think what our experiences would have been, had our caravan suddenly pitched camp in Hyde Park, and expected to raise supplies promptly in the neighbourhood without producing coin of the realm | Next morning the icy eastern gale was still blowing unabated. All the men not engaged over the scanty kitchen fires sought warmth and oblivion from discomforts past and present by a long day-sleep cuddled up in their furs. But it was a busy day forme. Early in the morn- ing I had a long interview with a big deputation from the Turkestan traders settled in the town, who had come to pay their respects to me as a quasi-compatriot of official standing. The trade interests they represented were small, and it did not take long to realize that most of them had retired to Tun-huang from Hami, Charklik, Turfan, in order to find a safe refuge from inconvenient creditors or lawsuits. But the plentiful supply of camels which Tun- huang offered for hire had enabled them to extend their ventures far to the east and south. As long residence had made them familiar with local conditions in Kan-su, I was eager to gather from them as much as possible of the practical information needed for my immediate plans. It was interesting to learn that manufactured imports from Urumchi, Lan-chou, and Khotan seemed to compete here on approximately equal terms. The Mongol grazing- grounds in the high valleys and plateaus towards Tibet offered good customers for them in exchange for wool and skins. Long before I started on this journey I had been struck by the geographically important position which the oasis of Tun-huang occupies near the point where the greatest old high road of Asia from east to west is crossed by the direct 14 FIRST HALT AT TUN-HUANG CH. LI route connecting Lhasa, and through it India, with Mon- golia and the southern portions of Siberia. Now, arrived on the spot, I was greatly struck to find that the region, the antiquities of which I was anxious to explore, had its modest marts impartially open to goods coming from China proper and its great ports in the East; to Russian manu- factured produce brought vza Kashgar or Urumchi; and even to British and Indian wares carried all the way from Khotan through the desert. By noon I proceeded to the Ya-mén, where Ahmad, our Tungan interpreter, had previously carried in due form the announcement of my state visit along with the customary presents. Among the latter I had taken care to include the last piece of that fine yellow ‘ Liberty brocade’ which I had before found most appreciated by Chinese recipients among my introductory offerings. I reached the magistrate’s Ya-mén inthe midst of a howling dust storm, and having to make my entry in my best ‘Europe clothes,’ black coat, sun-helmet, and patent leather boots, felt the cold pervading all its halls intensely. Naturally I should have liked under such conditions to make the first interview as short as I could. But Wang Ta-lao-ye, the ‘ Hsien-kuan,’ at once proved an official so exceptionally cultured and pleasant that, over the lively talk about things learned and ancient which with Chiang’s eager assistance ensued between us, I soon forgot the physical discomforts and the intention they had prompted. The magistrate was a sparely built middle-aged man, with a face expressing keen intelligence (Fig. 209). There was something in his combination of courtly manners, scholarly look, and lively talk which recalled dear P’an Ta-jén. I heardin due course that Wang Ta-lao-ye had just managed to dig out from his predecessor’s office records the elegantly worded epistle by which my Tao-t’ai patron had recommended me to the magistrate’s attention. He was evidently impressed by its contents; and I instinctively felt that a kindly official providence had brought to Tun- huang just the right man to help me in my first work on these ancient Marches. Of course, I did not fail to make appropriate reference to my saintly guide and patron cou NEWLY-WON MANDARIN FRIEND 15 Hsiian-tsang, and discovered to my delight that the Amban, as a scholar of wide reading, knew of the Hsi-yii-chi, the great pilgrim’s genuine memoirs. His apologies for our inadequate quarters were pro- fuse, and evidently inspired by sympathy arising from similar experience. His own furniture and property had not yet arrived, and the reception hall of his Ya-mén looked terribly bare in spite of some elegant wood-carving on the walls and its much-faded gilding. As I looked round it struck me that the good people of Tun-huang could have but little attention to spare for their magistrate if they left him even for a short time without a brazier or a curtain to keep out the icy wind. It was a comfort to know that at least he could wear a succession of suits underneath his official robe to keep himself warm, whereas my own ‘best clothes’ strictly prevented similar protec- tion and left me to feel the bitter cold. The Buran was still raging when I rode back through the almost deserted streets and the great waste space which extended within the southern face of the town-wall. I hurried to get my half-frozen feet into big fur boots, and had just begun within my carefully-tucked-up tent to warm myself a little, when Wang Ta-lao-ye’s return call was an- nounced. There was nothing for it but to receive him in my inhospitable barn of a hall. However, etiquette having once been satisfied, I kept on my fur boots. Seated ona thick Khotan felt rug and with a charcoal fire kept going in the cauldron which served for my men’s mess, | did not find the conditions quite so trying. From the mule trunks close at hand I brought forth specimens of ancient Chinese records excavated at Niya and Lop-nor, repro- ductions of earlier finds, and anything else that might be relished by the Amban’s antiquarian eyes. The effect was all I could wish for. Wang Ta-lao-ye thoroughly enjoyed the scholarly treat which my exhibits provided. With a kind of intuition, due no doubt to his interest in the subject matter, he generally managed to follow the archaeological queries and problems I ventured to submit to his judgment in my terrible Chinese jargon. I found him quite familiar with the geography of Eastern 16 FIRST HALT AT TUN-HUANG CH. LI Turkestan as it presents itself in the historical Annals and also to modern Chinese administration, though he had never served in that province. His knowledge was based wholly on books, and these, he frankly acknowledged, had told him nothing whatever about the ruined line of towers and wall I had traced in the desert. Whether local information about it would be forthcom- ing from the people of Tun-huang remained to be seen. In a subdued conversation with Chiang-ssti-yeh he de- scribed them as very distrustful and rather awkward to handle. But he in any case would do his best to help me, whatever the difficulties might be about getting guides and labour for the desert. When he left us, after some hours of cheering confabulation, and more than one cup of tea, both Chiang and myself felt assured that for the work before us we could count on the genuine goodwill of a newly won scholar-friend. Eager as I was to get ready for fresh explorations in the field, a number of practical obligations combined to prolong my halt at Tun-huang. Men and animals alike were much in need of a rest to recover from the preceding hardships. But though my body had its share in this enforced quiescence, there were plenty of urgent tasks to keep me very busy otherwise. Within thirty-six hours after our arrival I managed to pay off and dismiss to Charklik the whole of the donkey convoy which had helped us so effectively on the desert journey. It was a trouble- some affair, for the animals belonged to different owners, and the payments had to be adjusted with regard to the services of the men who had looked after the donkeys, compensation for the six beasts which had died, etc. Then came the distribution of rewards among those men who had taken good care of their charges, and finally committal of the whole complicated account to Turki writing to assure faithful transmission of the moneys due at Abdal and Charklik. The men, well pleased with the sums I paid into their hands, were eager to set out on the long journey homewards. I, too, felt hearty relief when I saw the whole band gaily depart with the loads of maize I had presented as a viaticum for the donkeys. Very soon afterwards I was able to despatch also my brave camels to a suitable grazing-ground where they were likely to gather fresh strength. This was an advantage gained through official support from another quarter. The very day after the exchange of visits with the prefect there called an officer sent by Lin Ta-jén, the military com- mandant at Tun-huang, to press upon me the assistance of some of his men to act as a camp guard, and to suggest through Chiang-ssii-yeh the propriety of mutual acquaint- ance. It wasa wish which under any circumstances I would have been very glad to gratify. But here at Tun-huang it had not taken Chiang long to realize that the military was an element of far more consequence in the administration than we were accustomed to in ‘‘our own province of Hsin- chiang.” It appeared, in fact, that, since some earlier period which I was unable exactly to ascertain, these outlying westernmost districts of Kan-su had in their administra- tive organization retained some features recalling those of military frontier settlements. A number of Tun-huang agricultural families seemed to be in receipt of monthly stipends, paid on the understanding that certain members of them, able-bodied or otherwise, would be available for military service. Only a small proportion of the 600 men or so, supposed to constitute this corps of Tun-huang levies, were likely ever to have been embodied, and the few men who were actually seen about idling in the town on their turn of duty looked more harmless even than the ‘soldiers’ to be found in Chinese garrisons of Turkestan. But there could be no doubt that the Tun-huang people fancied themselves a bulwark of the Empire on what had been once an important frontier barrier, and was still a point of Strategic importance. It became equally clear by and by that, owing to this organization of local levies, the military commandant at Tun-huang, having charge of all police arrangements, could play a hand in civil affairs, too, if he liked. Lin Ta-jén, when I called at his Ya-mén, a fairly large and comfortable place, proved a cheery old warrior. He had a burly, active figure, of middle height, and a square- VOL. II C 18 FIRST HALT AT TUN-HUANG CH. LI jawed face which expressed jovial good-nature and some character. At times he looked delightfully angular, like a figure from some early Flemish painting. His manners were pleasingly bluff and hearty, with just an occasional touch of quaint stiffness when he remembered what he owed to his present station and—prospects. His recollec- tion of services went back to the time when he had tramped into Turkestan as a corporal in Liu Chin-t’ang’s reconquer- ing army. For many years he had remained a humble petty officer until the débdcle of 1900, when he had the good fortune to get attached to the Empress’s escort, and riding by her chair on the flight to Hsin-an-fu to attract the imperial notice. Promotion had been rapid since, and Lin Ta-jén seemed well pleased with himself and the way in which the world had treated him. He did not lay claim to much education nor to any particular interest in things dead and buried. But he seemed to cherish greatly the recollection of his early years spent in Turkestan, and would chat away gaily about such of his old haunts as Chiang and myself knew. He had read and heard a good deal about Japanese prowess in the late war, and seemed full of respect for the Westerners whose teaching had helped on their success. Our prolonged visit to the district was evidently welcomed by him asa pleasant diversion, and the help of his myrmidons which he pressed upon us as a safeguard against the ob- structive indolence and occasional turbulence of the Tun- huang people in general, soon proved useful in more than one way. That Lin Ta-jén enjoyed more authority in his own sphere than Wang, the newly arrived civilian, did in his, was quite clear, and under the guidance of his men good grazing was soon secured for my camels. Luckily the two dignitaries were on excellent terms, and the magistrate was only too anxious to let me benefit by the predominant local influence of his military colleague. From the Muhammadan traders there was information to be gathered about ruined sites in the vicinity of the oasis and along the great routes to the north and east. One among them, Zahid Beg from Urumchi, after many a venture north and south of the T’ien-shan, including CH. LI MUHAMMADAN REFUGEES 19 a Begship at Charklik, had sought here an asylum from his Turkestan creditors. Like the versatile person he was, he had kept an eye open everywhere for ‘ Kone- shahrs’ with possible treasure. So he was able to tell me of ruins he had seen to the north-east of Tun-huang; of a walled town which lay half-buried under sand near Nan- hu, to the south-west, and soon. Vague as much of this information necessarily was, it helped in forming my plans, and was a perfect godsend when compared with the exasper- atingly stolid and steadfast declaration of utter ignorance which met every enquiry addressed to the Chinese of Tun-huang. But the most urgent task on hand was the preparation of the detailed accounts which I owed for long months past to the Comptroller of India Treasuries and the Indian Survey Department, and which I was anxious to despatch safely through the last of the Khotan Dakchis I had kept by me. Ever since the preceding summer there had been no rest available for dealing with this accounts’ incubus which I had to face single-handed. What its weight was may be gauged from the fact that it meant not merely extracting all and sundry items, however small, from my general cash record into properly balanced ‘ Monthly Cash Accounts’ in due official form, as if I were my own ‘Treasury Officer,’ but also dividing all entries relating to transport and the like according to whether they were to be debited against the Government grant or the Survey of India’s subsidy meant for ‘the Survey Party,’ or, finally, against my own personal purse, which would in due course recoup them from authorized ‘Travelling and Halting Allowances.’ No wonder that for five or six days I felt as if condemned to living more or less in the atmosphere of an Indian Office room—though an uncommonly cold one. CHAR TERE TO THE ‘CAVES OF THE THOUSAND BUDDHAS’ Ir was impossible to submit to this desk-work for long without a break while a site of exceptional interest was temptingly near, and awaiting as it were my first visit in fulfilment of a promise long made. Already in 1902 my friend Professor L. de Loczy, the distinguished head of the Hungarian Geological Survey and President of the Geographical Society of Hungary, had directed my attention to the sacred Buddhist grottoes, known as the ‘Caves of the Thousand Buddhas,’ or Ch’ien-fo-tung, to the south-east of Tun-huang. As member of Count Széchenyi’s expedition and thus as a pioneer of modern geographical exploration in Kan-su, he had visited them as early as 1879. I had been greatly impressed by his glowing description of the fine fresco paintings and stucco sculptures which he had seen there, and the close connection with early Indian art which he thought to have recognized in some of them without himself being an antiquarian student. It had, in fact, been a main cause inducing me to extend the plans of my expedition so far eastwards into China. On the 16th of March I could at last pay my first visit to the famous cave temples to which my thoughts had turned for so long from afar. Chiang-ssti-yeh, Naik Ram Singh, and one of Lin Ta-jén’s subordinates were to be my companions. The sky had cleared more rapidly than I could hope after those days of icy north-east wind and driving sand which greeted us on our arrival. It was a fairly bright morning, but still cold enough to make a wait trying. Yet, of course, we had to submit to that as an 20 CH. LI Fie CLIYORS PIE SAN DSY 21 inevitable feature of all work in these parts; for the ‘Ya-i’ who was to accompany us and carry the camera had thought fit to requisition a pony instead of his usual mount, a sturdy donkey, and so failed to turn up when everybody else was ready. At last by half-past eight our little cavalcade started south-eastwards. Fringing the southern edge of the oasis, the grey, dune-covered hills, which account for its later name ‘Sha-chou,’ ‘the City of the Sands,’ loomed huge through the leafless trees lining the roads. Most of the route we followed through the oasis lay deeply sunk below the level of the fields—an observation which brought back memories of Khotan. But a glance at the fields would suffice to remind me that Turkestan lay behind us. In fine big plots they extended, ploughed and levelled with a care which made them look as flat and smooth as billiard tables. Of the irregular little embankments which for irrigation purposes divide and terrace Turkestan fields, there was no trace to be seen here. After only a mile and half the road emerged on a bare gravel glacis as uncompromisingly sterile as any I have ever seen. Of dunes there were none, only a long, wall- like stretch of sandy foot-hills stretching away on our right to a hazy distance eastwards. An isolated small building, rising on what looked like a ‘witness’ left on eroded ground, proved to be, not a ruin, but a shrine with some modest annexes for the priests. The dwellers were absent, probably assisting in the celebrations which were reported being held in the town in order to drive off an epidemic attack described like influenza. The general look of the morne hills south reminded me of the equally barren scarp which the terrace-like offshoots of the Kun-lun about Karghalik present towards the plains. Two ruined towers suggested Pao-t’ais; but popular as the sacred caves no doubt were from early times, their importance was not likely to be indicated by the official marks of a high road. And in fact no further towers were met with. We were just approaching, after a total ride of some nine miles, the shallow depression which marks the de- bouchure of the stream passing the sacred grottoes, when 22° ‘CAVES OF BHOUSAND BUDD TA Ss scar there emerged from among the folds of the gravel-covered alluvial fan a lonely small shrine. It could not be very old; for the carved brickwork adorning the wall-tops and friezes showed tracery such as I had noted in the shrines of the city, and the bright yellow plaster looked recent. Yet, nevertheless, the whole bore every mark of premature ruin. The tiled roof with its fluted bricks was breached in more than one place, and that of an adjoining small cella had fallen entirely. But the coloured stucco images representing a Buddha, and some attendants whom I could not readily identify, were, in spite of missing limbs, still objects of worship. In front of them were the little sand-filled boxes which serve to keep lighted tapers up- right. KRed-coloured strips of paper inscribed with Chinese characters, probably short prayers or votive dedications, covered the base and wall surface. Outside the gateway there hung from a stand a big bell, rusty and showing ominous cracks. About the closely packed Chinese characters which covered the outer surface, I could gather from Chiang’s remarks only that they con- tained some Buddhist text. But more specific and satis- fying was the indication that the inscription bore a date. It is true, it was not an old one, going back only to the first half of the last century; but it gave me the first assurance that the chronological precision so characteristic of Chinese ways was not ignored by Buddhist piety in these parts. How often have I wished that such sense for the value of exact dates might be met with among Indian worshippers ! The little Buddhist sanctuary, with its air of decay and desolation, was a fit preparation for the sights awaiting me at the sacred caves ahead. After less than a mile they came in view as we turned into the silent valley by the side of a shallow little stream just freed from the grip of winter. There was not a trace of vegetation on the curiously eroded grey slopes which the spurs of the low hill range eastwards send down to the debouchure of the stream (Fig. 157). But all thought of slowly dying nature reflected in these shrivelled barren ridges and hillocks passed from me when, on the almost perpendicular con- ‘ALIS HO GNA HLNOS WOW NAYS ‘AD TIVA DNOL-OA-NUI.HO 40 LSVA ADNVY TIIH Nawuva ‘2481 oP MN LBP ne ig Gi eal maa “p CH. LIT MULTITUDE OF GROTTOES 23 glomerate cliffs rising on our right, I caught sight of the first grottoes. : A multitude of dark cavities, mostly small, was seen here, honeycombing the sombre rock faces in irregular tiers from the foot of the cliff, where the stream almost washed them, to the top of the precipice (Fig. 158). Here and there the flights of steps connecting the grottoes still showed on the cliff face. But in front of most the conglo- merate mass had crumbled away, and from a distance it looked as if approach to the sanctuaries would be possible only to those willing to be let down by ropes or to bear the trouble and expense of elaborate scaffolding. The whole strangely recalled fancy pictures of troglodyte dwellings of anchorites such as I remembered having seen long, long ago in early Italian paintings. Perhaps it was this reminiscence, or the unconscious vision of rich rubbish deposits which such holy cave-dwellers might have left behind in their burrows, that made me in my mind people these recesses with a beehive of Buddhist monks, and wonder what awkward climbs they might have had when paying each other visits. But the illusion did not last long. I recrossed the broad but thin ice sheet to the lowest point, where the rows of grottoes did not rise straight above the rubble bed, but had a strip of fertile alluvium in front of them; and at once | noticed that fresco paintings covered the walls of all the grottoes or as much as was visible of them from the entrances. ‘The Caves of the Thousand Buddhas’ were indeed tenanted, not by Buddhist recluses, however holy, but by images of the Enlightened One himself. All this host of grottoes represented shrines, and I hastened eagerly to take my first glance at their contents. The fine avenues of trees, apparently elms, which ex- tended along the foot of the honeycombed cliffs, and the distant view of some dwellings farther up where the river bank widened, were evidence that the cave-temples had still their resident guardians. Yet there was no human being about to receive us, no guide to distract one’s atten- tion. In bewildering multitude and closeness the lines of grottoes presented their faces, some high, some low, 24 «6S CAVES ‘OF THOUSAND BUDDHAS cir perched one above the other without any order or arrangement in stories (Fig. 159). In front of many were open verandah-like porches carved out of the soft rock with walls and ceilings bearing faded frescoes. Rough Stairs cut into the cliff and still rougher wooden galleries served as approaches to the higher caves. But many of these seemed on the point of crumbling away, and high up in the topmost rows there were manifestly shrines which had become quite inaccessible. There was nothing to guide me in my first rapid sight- seeing. Some of the larger grottoes on the lowest floor had, indeed, elaborate wooden antechapels of unmistak- ably modern look to indicate restoration. But I soon found that even these shrines contained much that was manifestly old both in fresco work and statuary. As I passed rapidly from one cella to another my eyes could scarcely take in more than the general type of the frescoes and certain technical features of the stucco sculptures. The former, in composition and style, showed the closest affinity to the remains of Buddhist pictorial art trans- planted from India to Eastern Turkestan, and already familiar from the ruined shrines I had excavated at Dandan-oilik and other old sites about Khotan. But in the representation of figures and faces the influence of * Chinese taste made itself felt distinctly, and instead of the thin outlines and equally thin colouring there appeared often a perfect exuberance of strong, but well-harmonized colours. Where deep blues and greens preponderated there was something in the effect distinctly recalling Tibetan work. I could not doubt for a moment that the best of these frescoes belonged to the times of the T’ang dynasty. In the rest, whether left in their original state or touched up by modern restorers, I could see that J had before me the work of painters who faithfully continued the artistic traditions of that period. The subjects and sizes of the mural paintings varied greatly in the different shrines, while the ground plan and arrangement of the latter showed much uniformity. From a kind of oblong ante- chapel, fully open to the light, but generally badly injured, (vy) dnNOND NUYAHLNOS AO ONINNIDAA ANV (G) SHOLLOWD .SVHGUGNE GNVSNOHL:> AO dNOUND AIAAIW ‘gS! foe PI KOS TNSPECTIONVOFR “KFRESCOES 25 a high and relatively wide passage led into a square, high- roofed cella hewn out of the rock, and as much as forty- five feet square. Within the cella was ordinarily to be found a group of images occupying either an elevated platform or else placed in a kind of alcove facing the entrance. All the wall faces were covered with plaster bearing frescoes. Those on the passage walls ordinarily represented rows of Bodhisattvas moving in procession or seated in tiers. Within the cellas the paintings were generally arranged in large, elaborately bordered panels, either singly or, where the wall surface was extensive, in a series’ In the centre of these there appeared mostly figures of Buddhas, singly or in groups, sur- rounded by divine worshippers and attendants in many varied forms and poses. There were scenes from the Buddhist heavens, from legends in which Buddhas or saints figured, representations of life in their places of worship, etc. (Fig. 160). But whether the wall decoration showed such pious compositions, or only that infinite multiplication of Bodhi- sattvas and saints in which Buddhist piety revels, all details in the drawing and grouping of the divine figures bore the impress of Indian models. In the figures of Buddhas particularly, the faithful preservation of the type of face, pose, and drapery as developed by Graeco- Buddhist art was most striking. In the subjects of the friezes and side panels, which often, apparently, reproduced scenes from the daily life of monks and other mundane worshippers; in the designs of rich floral borders, the Chinese artists seemed to have given free expression to their love for ornate land- scape backgrounds, graceful curves, and bold movement (Figs. 203, 204). But no local taste had presumed to trans- form the dignified serenity of the features, the simple yet impressive gestures, the graceful richness of folds with which classical art, as transplanted to the Indus, had endowed the bodily presence of Tathagata and his many epiphanies. Of the general style and merit of the sculptural remains it was more difficult to form a rapid impression ; for much of this statuary in friable stucco had suffered badly through decay of its material, mere soft clay, and even more from 26 “GAVES/ OF THOUSANDISUDDiENS jceame the hands of iconoclasts and the zeal of pious restorers. In almost all the shrines I visited, a seated figure of Buddha, sometimes of colossal proportions, was the presiding image ; but by his side there appeared regularly groups of standing Bodhisattvas and divine attendants more or less numerous GhignT on): I could readily recognize representations of Dvarapalas, the celestial ‘Guardians of the Quarters,’ in the richly adorned and gaily dressed figures usually flanking the horse- shoe-shaped platform which bore the sculptured groups in the larger shrines, and here and there also images of the more prominent Bodhisattvas. But for the rest, | realized from the first that prolonged study and competent priestly guidance would be needed. But was there any chance that such guidance would be forthcoming at this sacred site which at the time looked wholly deserted? And how would my honest secretary, himself a stranger to all the intricate details of Buddhist mythology and iconography, succeed in correctly grasping and reproducing the technical explanations of the hoped-for cicerone? Indeed, in this as in so many other directions of enquiry since my arrival at Tun-huang, I had cause to regret bitterly my total want of Sinologist qualifications. It was pleasing to note the entire absence of those many-headed and many-armed monstrosities which the Mahayana Buddhism of the Far East shares with the later development of that cult in Tibet and the border mountains of Northern India. And still more reassuring was it to see everywhere the faithful continuance of the sculptural traditions as developed by Graeco-Buddhist art. The heads and arms of most statues were, indeed, modern and very distant replicas, sadly inadequate attempts at restoration. But often the bodies and their rich drapery had _ sur- vived without change, and their exquisite colouring had escaped repainting. The profusion of gilt images, I knew, was an early feature, and so also the frequency of colossal figures of Buddhas in a variety of poses. The pious efforts of recent restorers seemed principally to have been directed towards these. Hence the several giant images of sitting Buddhas, rising through caves with a 159. ROWS OF CAVE-TEMPLES, SHOWING DECAYED PORCHES, NEAR MIDDLE OF SOUTHERN GROUP, ‘THOUSAND BUDDHAS’ SITE. ral W 7 ss CH. LI PROFUSION, OF SCULPTURE 27 number of stories, had modern antechapels built in front, of gaily painted timber and profusely decked with Chinese inscriptions on scarlet paper. Surely it was the sight of these colossal images, some reaching nearly a hundred feet in height, and the vivid first impressions retained of the cult paid to them, which had made Marco Polo put into his chapter on ‘ Sachiu,’ z.e. Tun- huang, a long account of the strange idolatrous customs of the people of Tangut. ‘‘ The Idolaters have a peculiar language, and are no traders, but live by their agriculture. They have a great many abbeys and minsters full of idols of sundry fashions, to which they pay great honour and reverence, worshipping them and sacrificing to them with much ado,” and so on. Tun-huang manifestly had managed to retain its traditions of Buddhist piety down to Marco’s days. Yet there was plentiful antiquarian evidence showing that most of the shrines and art remains at the Halls of the Thousand Buddhas dated back to the period of the T’ang dynasty, when Buddhism flourished greatly in China. Tun-huang, as the westernmost outpost of China proper, had then for nearly two centuries enjoyed imperial protection both against the Turks in the north and the Tibetans southward. But during the succeeding period, until the advent of paramount Mongol power, some two generations before Marco Polo’s visit, these marches had been exposed to barbarian inroads of all sorts. The splendour of the temples and the number of the monks and nuns established near them had, no doubt, sadly diminished in the interval. As I passed hurriedly from grotto to grotto, faithfully followed by my literatus, we were at last joined by one of the local priests I had been looking out for. It was a young ‘ Ho-shang,’ left in charge of the conglomeration of small houses and chapels which occupied a place amidst some arbours and fields facing the south end of the grottoes (Fig. 186). He was a quiet, intelligent fellow, quick at grasping what attracted my interest, and as unobtrusive a cicerone as one could wish for. His face showed scarcely any Chinese feature, and like so many physiognomies seen 28 ‘CAVES OF TPHOUSAN DIBUDDHA a sara about Khotan curiously recalled Indian origin. But I had already seen enough of the people of Tun-huang to realize how thorough a mixture of races may be looked for at these ancient cross-roads between China, Tibet, and the quasi-Aryan settlements of the Tarim Basin. Our guide readily took us to the temple containing a big Chinese inscription on marble which records pious works executed here in the T’ang period, and subse- quently to the two shrines where smaller epigraphic relics dating from the Sung and Yiian dynasties are preserved. Alas! I had to leave the examination of them entirely to Chiang-ssti-yeh, who, with true Chinese delight in things palaeographical, was soon absorbed in a study of the finely engraved rows of characters. But it was a comfort to know that with one or two exceptions they had already been published by M. Chavannes from impressions brought back by M. Bonin. At Tun-huang I had first heard through Zahid Beg vague rumours about a great hidden deposit of ancient manuscripts which was said to have been discovered acci- dentally some years earlier in one of the grottoes. And the assertion that some of these manuscripts were not Chinese had naturally made me still keener to ascertain exact details. These treasures were said to have been locked up again in one of the shrines by official order. In secret council Chiang and myself had discussed long before how best to get access to the find, and how to break down if necessary any priestly obstruction. I had told my devoted secretary what Indian experience had taught me of the diplomacy most likely to succeed with local priests usually as ignorant as they were greedy, and his ready comprehension had assured me that the methods suggested might be tried with advantage on Chinese soil too. The absence of the Taoist priest in charge of the manuscripts made it impos- sible to start operations at once. But the young monk was able to put us on the right track. So I soon let him be taken aside by the Ssii-yeh for private confabu- lation. From a rapid inspection of the southernmost caves, perched high up near the top of the cliff, I had just returned to the grotto containing the latest of the es afi. 160. INTERIOR OF CAVE-TEMPLE CH. VIII., ‘THOUSAND BUDDHAS’ SITE, SHOWING FRESCO DECORATION OF WALLS AND- ROOF. os / i - r il 7 ‘ : i , > €3 » 7 ¥ cx. utr RUMOUR OF HIDDEN MANUSCRIPTS 29 inscriptions, engraved about the middle of the fourteenth century, when Chiang joined me full of joy at the success of his investigation. It was in a large shrine farther north, bearing on its walls evidence of recent restoration, that the deposit of manuscripts had been discovered. The entrance to the cave-temple had been formerly blocked by fallen rock- débris and drift sand. After this had been cleared out, and while restorations were slowly proceeding in the temple cella and its antechapel, the workmen engaged had noticed a crack in the frescoed wall of the passage between them. Attracted by this they discovered an opening leading to a recess or small chamber hollowed out from the rock behind the stuccoed wall to the right. It proved to be completely filled with manuscript rolls which were said to be written in Chinese characters, but in a non-Chinese language. The total quantity was supposed to make up some cart-loads. News of the discovery ulti- mately reached provincial headquarters, and after specimens had been sent to far-away Lan-chou, orders were supposed to have come from the Viceroy to restore the whole of the find to its original place of deposit. So now behind the carefully locked door with which the recess had been furnished, these strange undeciphered manuscripts were said to be kept. The shrine, though full of Buddhist frescoes and sculptures, was in charge of a Taoist priest, the small fraternity of Ho-shangs or Buddhist monks, to which our guide belonged, peacefully sharing with him the guardianship of the site. The priest was away in the oasis, apparently on a begging tour with his acolytes. Chiang had thus no chance to pursue his preliminary enquiries further. But, fortu- nately, the young Ho-shang’s spiritual guide, a Sramana of Tibetan extraction, had borrowed one of the manuscripts for the sake of giving additional lustre to his small private chapel, and our cicerone was persuaded by the Ssiti-yeh to bring us this specimen. It was a beautifully preserved roll of paper, about a foot high and perhaps fifteen yards long, which I unfolded with Chiang in front of the original hiding-place. The writing was, indeed, Chinese; but my go » 'CAVES/ OF' THOUSAND! BUDDEAS® ica. learned secretary frankly acknowledged that to him the characters conveyed no sense whatever. Was this evidence of a non-Chinese language, or merely an indication how utterly strange the phraseology of Chinese Buddhism is to the literatus brought up on a classical pabulum? However, a frequently repeated formula at once attracted our attention, containing the word ‘ Pu-sa,’ the Chinese contraction of the Sanskrit term ‘ Bodhisattva.’ This sufficed to show that the text must be Buddhist —enough to whet my appetite even more for the whole collection. The paper looked remarkably strong and fresh with its smooth surface and fine texture; but in a climate so dry and in a carefully sheltered hiding-place there was no judging of age from mere appearance. So I was obliged to put off all further speculation on this score until we should obtain access to the whole of the hidden _ library. It was a novel experience for me to find these shrines, notwithstanding all apparent decay, still frequented as places of actual worship. But quite apart from the damage done by well-meant restorations, I reflected with some apprehension upon the difficulties which this continued sanctity of the site might raise against archaeological exploitation. Would the resident priests be sufficiently good-natured—and mindful of material interests—to close their eyes to the removal of any sacred objects? And, if so, could we rely on their spiritual influence to allay the scruples which might arise among the still more super- stitious laity patronizing their pilgrimage place or ‘ Tirtha,’ to use the familiar Indian term? Only experience and time could show. Meanwhile I was glad enough to propitiate the young Buddhist priest with an appropriate offering. I always like to be liberal with those whom I may hope to secure as ‘my own’ local priests or ‘ Purohitas’ at sites of ancient worship. But, unlike the attitude usually taken up by my Indian Pandit friends on such occasions, when they could—vicariously—gain ‘spiritual merit’ for themselves, Chiang-ssti-yeh in his worldly wisdom advised moderation. A present too generous might arouse speculations about ‘ALIS .SVHGdNd ANVSNOHL: “III *HO. WIdWALAHAVO NI ‘SVIVdVUVAG UNV ‘SVALLVSIHGOd SHTdIOSIG NHHYMLYA VHAGNA ONILNASHadda “dNOUD ADVWI OODONLS 19! exit PLACES OF WORSHIP IN BEING ay possible ulterior objects. Recognising the soundness of his reasoning, I restricted my ‘ Dakshina,’ or offering, to a piece of hacked silver, equal to about three rupees or four shillings. The gleam of satisfaction on the young Ho-shang’s face showed that the people of Tun-huang, whatever else their weaknesses, were not much given to spoiling poor monks. As to Chiang, I thought I could in his attitude detect something closely akin to that mingled regard for the cult and self-conscious pity for its ignorant representatives, which in the old days never allowed of easy relations between my learned associates in Kashmir and the local priests at pilgrimage places we used to visit together for archaeological purposes. It was getting dusk before I could tear myself away from this wonderful beehive of temples in its setting of barren rocks and sands. The route we followed when returning clambered up the riverine terrace by a steep detritus-covered slope, and then crossed the bare gravel plateau which edges the foot of the outer hills. The west wind which now swept it was piercing, and in the dust- laden atmosphere complete darkness soon overtook us. So there was nothing to interfere with the pictures full of vivid colour and grave pomp, all of ages long gone by, which that day’s over-abundant sight-seeing had left impressed on my mind’s eyes. CHAPPRERGER A DIFFICULT START FROM TUN-HUANG AFTER the fascinating prospects that hurried excursion had opened before me, it seemed hard to continue my clerical toils unbroken during the days which followed; but the tasks were so heavy that, busy as I was at my table from morning until midnight, the time seemed only too short. The departure of a Khotan trader, who was to start for Charklik by March 21st, offered a chance of sending off with him Kurban Niaz, the Dak man, the last safe link for my mails to Kashgar, India, and Europe, which I was likely to have for a long time. So the letter-bag | entrusted to him for what was bound to prove close on four months’ transit, grew uncommonly heavy. In the meantime I was able to think over and settle my immediate plans. I knew from my Taklamakan experiences what the climatic conditions in the desert of Tun-huang were likely to be when once the winter had passed. So I decided upon the exploration of the ancient frontier line as the first task of my new programme. It was quite impossible to calculate beforehand with any certainty what amount of time and labour excavations along it might claim. So with plans upon the ‘Thousand Buddhas’ looming big before me, it seemed doubly im- portant that I should start back to the ancient wall well provided in the matter of guides, diggers, and supplies, and thus prevent any needless delay in my proposed opera- tions there. It did not take me long to realize that the difficulties in all these respects would be serious. Of the ruins I was anxious to trace and explore in the desert nothing 32 CH. LI OFFICIAL DINNER-PARTY 33 was known to our friends the magistrate and military commander, or to the other educated Chinese officials of Tun-huang, though they seemed interested in my work and ready to help. On the other hand, the deep-rooted secretiveness of the local Chinese population effectively prevented any offer of guidance from the herdsmen or hunters, who occasionally visit the nearer of the riverine jungles. Still more serious was the trouble threatening about the provision of labour and transport. This was first brought home to me on the occasion when our military patron, Lin Ta-jén, and the learned magistrate jointly treated Chiang-ssti-yeh and me to a grand dinner-party. They had considerately waited until learning that the despatch of mails and accounts had at last set me free for a little diversion. The scene was laid in the guest- rooms of a finely situated temple on the north wall of the city. With only two other guests to share our Mandarin friends’ hospitality it was easy to combine the discussion of scholarly and practical ‘business’ with enjoyment of the _ comforts provided. It was a well-ordered little feast, and I frankly confess I greatly appreciated its setting. To find myself in neat and well-lit apartments, with an atmosphere pleasantly warm, was an experience which had not come my way since camping in Nar-bagh. The menu was ample and varied, and though I was not competent to judge of the dishes, and soon forgot even to keep count of them, it seemed a welcome change from the wearisome uniformity of the food my intractable Kashmiri cook was in the habit of preparing. Wang Ta-lao-ye had been thoughtful enough to cater for my intellectual appetite, too. He had brought a Chinese volume containing a kind of official gazetteer of the Tun-huang district and full of historical extracts. So during the preliminaries of the dinner I was able, with Chiang’s help, to glean a good deal of interesting information about the history of the ‘ Thousand Buddhas’ ’ site during T’ang times, as well as to ascertain how vague modern knowledge was in respect of the ancient routes leading westwards. VOL. II D 34 A START FROM TUN-HUANG CH. LIII With the ruined line of wall we had traced in the desert of Tun-huang local Chinese scholarship had evi- dently never concerned itself, and later on Chiang, when at home, vainly searched the volume even for a mere reference. It was the old story of antiquaries, absorbed in their book-lore, not heeding to look for the realities of the past in the open. When I thought of the physical conditions prevailing on the desert ground I was eager to search, I could scarcely blame Chinese confréres for not having cared to explore it before me. Nor could I complain of any indifference on the part of my hosts when I| explained to them the operations I was planning about those ruins. Chiang had told them a good deal since our first visits about my methods of work and the success which had rewarded them elsewhere, and with that historical curiosity which seems innate in every educated Chinaman, they were manifestly agreed that I should try my luck here too. But when, in the course of our dinner talk, it came to discussing the practical arrangements needed for my proposed excavations, I realized that there were difficulties ahead such as on Turkestan soil no Amban had ever hinted at. Wang Ta-lao-ye gravely told me that he scarcely knew how to provide the dozen diggers I had mentioned as the minimum required for my first pro- specting. Ever since the great devastations of the Tungan rebellion, the population of the oasis had remained very low, and the supply of agricultural labour extremely scanty. The cultivators owned their land. However small their holdings, to make them move out into the desert for digging would be a hopeless attempt, whatever the season and whatever pay might be offered; for Tun-huang people all loved their ease dearly, and having no struggle for exist- ence to face were proof equally against official pressure and the desire of gain. Besides, they were all and sundry confirmed opium- smokers, and as such most unwilling under any conditions to submit to a rough life in the open. It was the same with the few men available for hire as day labourers, a- hopelessly inert set of wastrels, not likely to forsake easy cxwiunt ADMINISTRATIVE WEAKNESS 35 employment about the town and hamlets for work in the dreaded ‘Great Gobi.’ In any case I should never get even these hapless loafers living from hand to mouth to accompany me into the wintry desert unless I took them there in carts. For me who knew what marching in waterless desert meant, the idea of my caravan moving with coolies on carts, like a snail with its shell, was comical and depressing beyond words. But it was as well to have one’s eyes fully opened to the radical difference in the relations between administra- tion and people since we had exchanged Turkestan for these Marches of China proper. There, as the repre- sentative of a conquering power and set to rule over a race of inferior civilization, no Chinese Amban would have admitted any limits to his ability to order out labour, as long as the men were likely to receive a_ reasonable recompense. The task of collecting them would fall on the petty bureaucracy of Begs and village head-men, who had the will and traditional means of making their orders obeyed. Here among true celestial subjects all this was changed. As I realized more and more the farther we moved into Kan-su, the fundamental principle of administra- tion appeared to be a sort of mutual toleration between nominal rulers and ruled, supported chiefly by that all- pervading factor, vzs znertzae. With this indolent but highly democratic public, practical experience must have taught the officials to restrict inter- ference to the minimum compatible with the collection of _ such taxes as would suffice to keep the provincial machinery going and—to leave modest nest-eggs for the office incumbents on retirement. The people, while tolerating authority for the sake of tradition and general security, seemed to take good care to enforce such prudent reserve on the part of their administrators. What petty local officials we ever managed to get hold of invariably bore themselves more like representatives of defence unions of guilds and land-owning communities than as agents of the civil authority. That there was now in these westernmost parts of Kan-su practically no military force to back up the civil, became abundantly evident in the course of my 36 A START FROM TUN-HUANG CH, LI journey eastwards. How little reliance could be placed on the levies of Tun-huang in particular I was to learn in good time by the tragic dénxouement to be related here- after, of which poor learned Wang himself became the victim, With such cares in the background, avowed or hidden, it was natural that our conversation should turn to the ‘New Dominion,’ where there were still real rulers of districts, and where a capable Amban need not take too serious count of local opposition. It was amusing to watch how the mental atmosphere of our little official party brightened with this change of topic. Bluff and genial Lin Ta-jén recalled stories of his own happy days of exile in the newly reconquered province where any official of sense could live without worries and make money. Chiang felt at once in his true element, and chattered gaily of all the attractions which his adopted territory offered to men in office and—to others who like his humble self were trying to qualify for it. Tt Wang Ta-lao-ye and the other Kan-su guests Hsin-chiang, I could clearly perceive, must have presented itself in the light of an Eldorado for literati aspiring to the official’s life. I myself could not help sympathizing with them when I thought what the reverse change would be like for any of my Anglo-Indian civilian friends, say, from the charge of a big district in the Punjab to the magistracy of some slum-ridden Western centre full of highly advanced, though ordinarily tame, Socialists. So I was induced to expatiate upon the many obvious points of contact between British rule over India and the sort of paternal government which Chinese state-craft has exercised since ancient times over the sot-dtsant ‘New Dominion’ in Central Asia; and | thought I found an appreciative audience in spite of the crudeness of my best Chinese. There could be no doubt that we all felt united by sentiments of some common official bond, and when our little party broke up I retained the impression, not merely of some hours pleasantly spent, but of a gain in good understanding. I needed this assurance badly when on March 22nd I was making my final preparations for returning to work CH. LIII OFFICIAL SYMPATHIES 37 in the desert. Men and animals had had a good rest now. Ram Singh, the Surveyor, had completed inking and tracing the winter’s plane-table work, while the Naik had made fair progress with the developing of my photographic negatives. I myself had managed to find time for re- arranging the baggage, the bulk of which was to be stored as a depot under Ibrahim Beg’s care in quarters hired for the purpose within the city walls. But of the labourers and hired camels I had asked for from the magistrate’s Ya-mén no sign appeared. An informal call which he and Lin Ta-jén paid me in the course of the day had given me an opportunity of personally repeating my requests for labourers who could walk on their own feet, and for camels to carry the baggage and supplies for men and beasts. They promised to do their best, but looked much exercised over the problem how to get Tun-huang people to follow me into the desert and work. I had told them that my start was fixed for the morrow. But though I was kept busy until after midnight by many tasks, including the settlement of accounts, which meant endless trouble of weighing hacked silver and adjusting for differences of scales, | waited in vain for either men or camels. Next morning the wearisome wait continued. But I had decided to make a move in any case, just to show my Mandarin friends that I was in earnest. So my tent was struck and the baggage which was to go with us made ready for our own camels. Hassan Akhun had over-night brought them back safely from what even he, a fastidious critic of all things in this infidel region, acknowledged to have been splendid camel-grazing. By and by there appeared on the scene one of Lin Ta-jén’s non-com- missioned officers, a quiet, cheerful man, who was to escort us with a mounted levy. The Ya-mén attendants had set out early to bring the promised camels and coolies. But hour after hour passed without their return- ing, and repeated reminders sent to the magistrate’s Ya- mén produced no result but apologies. It was evident that his good intentions were of little avail against indigenous inertia and obstruction. Long years of service in the ‘New Dominion’ had accustomed Chiang to a different 38 A START FROM TUN-HUANG CH. LIM procedure, and he freely displayed indignation at this open defiance of authority. It was a pleasant contrast to note the civility shown by all those with whom our stay had brought us into contact. The ladies of the house which had given shelter to my people sent a dear little mite to wish me don voyage and —to receive the rent. The levies who had kept watch over camp and animals showed exuberant gratitude for the douceur | sent them, though it was far from extravagant. However troublesome subjects the people of Tun-huang might be, want of manners was not among their chief faults. I was sitting in the midst of the ready-packed baggage when there arrived a visitor to cheer me, whom here, far away in Cathay, I felt tempted to greet almost like a fellow-countryman. It was an enterprising Afghan merchant, Sher Ali Khan of Kabul, who traded at Khotan, and was just now returning from Kan-chou after a successful venture, partly with British fabrics imported through Kashmir and Yarkand. He was sending back his caravan vza Charklik with tea and silk in return. My old haunts on the Indus and beyond were familiar places to him, and so too were Samarkand and Bokhara. As I. looked at the tall, well-built man with a complexion like | that of a Southern European, and thought of the thousands | of miles he had covered on routes which were still very much as they must have been in the times of the ancient world, I needed no imagination to picture to myself oe | agents of ‘Maés the Macedonian, also called Tatianus,’ who eighteen hundred years ago had traded from Syria or ! Mesopotamia for the silk of the distant Seres. At last, realizing that a further wait was useless, we — set out by one o'clock, after Chiang-ssti-yeh had penned his poignantly polite epistle of protest to the Ya-mén on- neat pink paper. My immediate programme was to. move due north of the oasis, and to search there for the line of the ancient wall which I surmised to continue east- wards along the course of the Su-lo Ho. So we passed through the outer and inner walls of the town into its somnolent dirty streets, where pigs were more conspicuous © ex.uu LINKED WITH EAST AND WEST 39 than people, and out again by the east gate. Just within that a beautiful memorial arch in wood, of some age and badly decayed, offered a pleasant relief to the eyes from all the surrounding squalor. We moved on through a well-cultivated tract with orchards and scattered but substantial homesteads, but had scarcely covered more than three miles when a mounted messenger from the Ya-mén overtook me. He did not bring news of the much-delayed transport and coolies, but, to my great delight, telegraphic messages from Kashgar and Peking. Immediately on my arrival at Tun-huang I had taken care to send off Ibrahim Beg to the town of An-hsi, four marches to the north-east, where I knew the great high road from Turkestan to Kan-su to pass and the telegraph line along it; messages entrusted to him were to announce my arrival by wire both to Mr. Macartney and to the Peking Legation. I was aware how little reliable was that wire linking East and West, with deserts between the rare stations, and with its constant breakdowns lasting at times for a week or two. So my joy at receiving replies within ten days was all the greater. It was an exciting quarter of an hour as I sat by the road-side deciphering the messages, which, like my own, had, for the sake of reducing cost and risks of distortion, been despatched in the Eastern Telegraph Company’s excellent code ‘ Vza Eastern.’ The news proved reassur- ing from both sides. Kashgar reported all well and a big mail-bag having been started wa Khotan. From the Legation I heard that arrangements were completed through the Chinese Foreign Office, allowing me to draw 6000 Taéls, equivalent roughly to 41000, from the Tao- t’ai’s treasury at Su-chou. It was a relief to have been brought so rapidly into touch with the distant East and West. But what I enjoyed most, perhaps, was the feeling of living for the moment as it were in two widely different periods. For travel or the transmission of letters the distances separating me from Kashgar, and still more from India or Europe, claimed an allowance of long weary months such as Marco Polo would have found natural. And now into this mediaeval perspective of time to which 40 A START FROM TUN-HUANG CH. LIII I had become quite accustomed, there seemed to burst with the telegrams a flash of modern life. This surprise would have sufficed to cheer me for the rest of the day. But the march too, short as it was, offered interest of its own. At several points we passed large bastioned forts with high and massive walls of clay which looked recent. Inside were only a few houses or farm buildings rarely tenanted. I now learned that these strongholds, so thoroughly mediaeval in look and reminiscent of the Pathan village forts, or ‘ Killas,’ familiar to me from the Indian North-West Frontier, had all been built or repaired by the neighbouring villagers during the troubled times of the Tungan rebellion. Traditional Chinese notions had led the unfortunate settlers to seek safety behind high walls, however inadequate their defence was in numbers or spirit. So all these scattered places of refuge fell one after the other before the onslaught of the fanatical Muhammadan rebels, who spared neither women nor children. During the successive inroads of the great murdering bands there escaped only that portion of the population which had sought refuge in the town, and many died there of starvation. Though over thirty years had passed since that time of horrors it was easy to note on all sides evidence of its lasting effects. The farther we passed from the town the more frequent became the sight of ruined houses and temples. Close by there were often quite substantial farms which had been reoccupied, and the land around was now under careful cultivation. But the population was manifestly still far from having made up its terrible losses in numbers. Gradually I noticed stretches of — uncultivated ground appearing on either side of the canal we were following, while in the distance beyond them rows of trees marked other finger-like extensions of the oasis aligned along more canals. Abundant scrub attested that the soil of the intervening waste strips was quite fertile and only waiting to be brought under cultivation again. Here and there the ruins of old homesteads could be seen rising amidst these abandoned fields, and troops of graceful small antelopes were now browsing there in peace. cuir TRACES OF TUNGAN DESTRUCTION 41 Clearly the villagers were no hunters. But whether this was due to surviving Buddhist feeling about the taking of life, or to some special local superstition, or simply to sheer indolence, I attempted in vain to find out. The suspicious reticence of the people proved a terrible barrier throughout. In time I learned to realize that even the most reserved and shy of Turkestan ‘ Taghliks’ were almost loquacious when compared with these good people of Tun-huang. No question I put about the reason which caused these areas of fertile land to remain waste, ever elicited any other answer beyond that impenetrable ‘Pu chih-tao,’ “I do not know.” Was it really a diminu- tion of the water-supply since pre-rebellion days, or rather the want of adequate labour for more extensive irriga- tion? It seemed difficult to believe in the former as | looked at the overflowing canal and the big range under deep snow which came into full view on the south as the ground got more open. We camped in a tamarisk grove near the hamlet of Shih-tsao, and before starting next morning were to my relief joined by a small convoy of coolies and camels. The eight men whom the Ya-mén attendants had managed to scrape together looked the craziest crew I ever led to digging—so torpid and enfeebled by opium were they ; but I was glad to have eventhem. So they were promptly advanced a fair lump of silver to lay in provisions, and the cart which had brought them from town was to be taken along for their conveyance as far as the ground would permit. Moving for some three miles northward along the canal we came upon the crumbling homesteads of a village left deserted since its sack by the Tungans. In the midst of this desolation it was pleasant to find the decayed fort tenanted by a number of baby camels, only a few days old, which were being kept here out of mischief and the biting wind while their mothers were grazing outside. The antics of the quaint little beasts supplied a contrast to the grim scenes of death and plunder which these walls must have witnessed scarcely forty years before. Then we turned off to the north-west across a belt of scrub-covered waste where the lines of irrigation 42 A START FROM TUN-HUANG CH, LIT channels, though dry since Tungan days, were still well pre- served, and came to resumed cultivation near the flourish- ing little village of Chuang-lang. Its homesteads were established mainly amongst the ruins of more substantial dwellings built before the Tungan inroads. Most of the trees in avenues and orchards were quite young, proving recent reoccupation. But here and there big old ash trees had survived the period during which this tract had remained without people and its timber at the mercy of wood-cutters from Tun-huang town. After having crossed more stretches of waste land and covered altogether eight miles, we arrived at the last hamlet to which reoccupation has extended. Here the recent colonists were still content with the shelter of mud hovels and reed huts. To the west we were now within view of the river bed of the Tang Ho, and immediately to the north we had to cross a deep channel taking off from it, which was manifestly an old canal of importance. Beyond this there extended a wide steppe covered with reeds and scrub, and cut up by numerous shallow channels of water. Clearly this ground for its reclamation needed drainage even more than systematic irrigation. And, indeed, as we marched on we met a large party of poor cultivators who had been engaged in reclaiming abandoned fields for spring sowing. At last, after wading through a good deal of flooded ground, I caught sight of the ‘Kone-shahr’ or ‘old town’ of which Zahid Beg had first told me at Tun-huang. Shih-pan-tung, as our Chinese companions called it, proved indeed a town, but its ruins not older than the last Tungan rising. Nevertheless this deserted site offered me a good deal that was instructive. Within a square of crumbling clay ramparts, about 375 yards on each face, we found the remains of a typical small Chinese town. It had been sacked by the Muhammadan rebels about forty years before, and had since fallen into complete ruin. The enclosing walls had in many places decayed into a mere agger or mound. The interior was for the greatest part filled with heaps of debris ; but the usual alignment of streets at right angles, somewhat after the fashion of a Roman castrum, could CH. LIT SIT BVORIRECENTAKRUINS 43 still be made out quite clearly. Through the ruined gate in the centre of the southern wall face passed the main road towards a conspicuous temple ruin rising well above the deébris-filled area and masking the north gate. To the west of this road, and not far from the south gate, a small Ya-mén with a picturesque gateway still showed roofed rooms, but in a state approaching collapse. No timber had been removed here, and it looked as if some petty magistrate had occupied these quarters after the town was deserted, perhaps some official supposed to re-colonize it. Auspicious sentences penned on scarlet paper and similar flimsy adornments of official quarters still stuck to walls and posts. In the mellow afternoon light it was quite a picture of quiescent extinction. All other buildings within the walls, the couple of temples excepted, had been reduced to heaps of brickwork or bare walls, their timber having been removed long ago. The main temple was a massive structure of true Chinese style, built in hard bricks with plenty of terra- cotta relievo work. It had a second story, formed by a separate shrine which was raised on a massive base of sun-dried bricks, about twenty feet high, at the back of the other. The stucco images, though all badly broken by vandal hands, were manifestly still objects of worship, and a large bronze bell was left zz sz¢u in spite of Tungan wrecking. Probably the raiding bands had neither time nor use for melting it down. In a smaller temple, too, crowning the centre of the west wall, there were traces of worship still clinging to the images, though they were reduced to a mere heap of clay fragments. Indeed, local cult dies hard. As I walked over the débris area, crossing more than one rubbish heap, I thought of the rich deposits likely to await some successor, say two thousand years hence. What antiquarian dainties might be gathered here—if only they became ‘high’ enough by the lapse of ages ! CHAPTERWEIY, BY THE ANCIENT WALL NORTH OF TUN-HUANG From the height of the town wall a watch-tower had been sighted to the north-north-east, and for that we set out on the morning of March 25th, with an icy wind sweeping down from the east. At first our progress was greatly impeded by the swampy condition of the ground, which was being purposely flooded for sowing. Large patches of the scrub-covered waste, once no doubt fields, had been cleared by burning, and water was now being brought to them in a rather erratic fashion by damming up channels which originally might in parts at least have been cut for irrigation. The camels and the men, now on foot, were forced to great détours and remained far behind. But with a few of my people on horseback I struggled across to the ‘ Pao-t’ai,’ which proved to be about four miles off. It was, as I had surmised from the first, a watch-tower of ancient construction. The material here consisted of hard lumps of salt-impregnated clay quarried from the low ridge on which the tower was raised; but any doubt as to its age was removed by observing in this strange masonry the same layers of reeds and tamarisk twigs at regular intervals which were characteristic of the towers guarding the old wall farther west. The whole formed a remarkably compact mass about twenty feet square at its base, and still over eighteen feet high, the permeating salts acting apparently like a sort of cement. Its solidity could be gauged from the fact that, though wind erosion had attacked and worn away the natural clay beneath the corners, the structure overhung there without any injury. 44 CH. LIV STOPPED BY SU-LO HO FLOOD 45 There was no trace of a wall line to be seen anywhere, and as the only landmark which might be a tower was seen in the distance north-eastward, I decided to follow a cart track which seemed to lead approximately in that direction. My intention was to push on first to the north and locate the course of the Su-lo Ho. Ifa crossing was possible, I would detach Ram Singh to survey the river along its northern bank down to the Khara-nor Lake. I myself, after seeing him safely across, would start eastwards for remains of the old wall near the left bank. The ground passed ahead was for some four miles covered with rich scrub and tamarisks, and by the side of our cart track I noticed two rough enclosures, built of remarkably hard lumps of salty clay, which bore the look of having been used at one time or other as sheep pens. Then we encountered a low and narrow clay ridge stretching across our route, and from its top first sighted northward a wide marshy expanse suggestive of approach to the river. In the midst of it there could be seen rising a succession of clay terraces ranged in rows, all striking east to west. The sight of them reminded me at once of the eroded clay ‘ witnesses’ we had passed in such numbers in the dry terminal basin of the Su-lo Ho and again in the vicinity of Lake Khara-nor. When, a mile or so farther on, we came upon the first fresh-water lagoon, and then had to ford a succession of shallow water-courses all flowing westwards and manifestly fed from the river, I could feel no doubt about the cause which had here determined the bearing of the rows of clay terraces. It was clearly the action of water which, working on the bottom of an earlier and wider fluvial bed, had first carved out ridges parallel to its own line of drainage from east to west. Then erosion by the winds blowing from the north-east had cut up these ridges into rows of terraces, and, no doubt, this scouring still continued. This combination of the erosive forces of running water and of wind was the very process by which I had conjecturally explained to myself the formation of those strange ‘witnesses’ in and near the dry lake basins previously met with. I felt no little satisfaction at seeing it now illustrated by 46 WALL NORTH OF TUN-HUANG | cou actual observation on ground so near by and so closely corresponding. Soon the cart tracks disappeared in a broad expanse of marshy ground where large areas were still covered by melting snow and ice cakes. For two and a half miles beyond the first clay ridge I pushed on with Naik Ram Singh and Tila Bai across the belts of boggy ground and the network of water-channels which extended between the chains of clay terraces. At last we forded with some diff- culty a channel some twenty yards broad and four to five feet deep in which the icy water flowed briskly; but progress beyond was quite impossible for laden animals. A broad and long sheet of ice sighted to the north from the nearest ridge showed that the main course of the Su- lo Ho still lay ahead of us. Whether the ice would hold or not, there was nothing for us but to turn back. By the lagoon we had first met I halted for camp, and there the camels and coolies safely joined us in the darkness. Next morning I retraced our route to the first clay ridge, and thence turned due east towards the tower we had sighted before. It took us nearly three miles to reach it across a reed-covered steppe where old fields with dry irrigation cuts were still clearly traceable. The tower built on a clay ridge rising about sixteen feet above the depression northward proved exceptionally massive and well preserved (Fig. 162). On a base twenty-six feet Square it rose solid to a height of some twenty feet. On its top it bore a brick parapet, and within this a roofless cella open to the south, but provided with that peculiar masking wall which usually faces the street entrances to Chinese temples or mansions. The cella walls, some twelve feet high, still retained much of their plastering. It was impossible to examine this crowning structure more closely owing to the disappearance of the ladder-like stairs which, judging from holes left for beams in the masonry, seem to have led up the west face. But a late origin, at least for the superstructure, was suggested by the peculiar way in which the bricks were set in alternately horizontal and vertical layers, such as I had never observed “SHNIOSVA GNV AVIO GHUdNVLIS HO SUHAVI ALVNYALIV “IIMOY JO JOOF VL yad-nss- surly + OVNTT A ea aig OS HLIM NOILONULSNOO DNIMOHS “SHWIT ONVOH-NOL ‘ISHM-HLOAOS WOW NAS “*AXXX “L UWAMOL UVAN “TIVM UAGUOT LNAIONV JO FOV OL ‘SISVO ONVNH-NOL JO HLYON “*AXX “L WAMOL-HOLVM QaNInad “cgi = CH. LIV AILUCKYSRPNGOUN TER 47 on Chinese soil except in modern or mediaeval buildings. The bricks of the tower, too, differed in size from those observed in the watch-towers which I had examined along the ancient wall, being smaller and only two inches thick. I searched in vain for any traces of that wall with its characteristic reed-bundles, but found that a much decayed earthen rampart, about a hundred feet square, adjoined the tower on the south, evidently marking an enclosure. Had this served once as a place of refuge for some out- lying colony, and when had it been abandoned ? Not being able to sight any more towers, I felt rather puzzled how best to continue our search for the wall. However, I decided to march on farther east until we struck the track leading north across the Su-lo Ho towards Hami. Roborowsky and Kozloff, who had twice followed this route in 1893-94, had marked ruins of some sort on their map in a position south of the river, and I felt in any case bound to visit these whatever their age. So we moved on to the east across a level plain covered with thick scrub, and in one or two places with Toghraks of fair size. Isolated clay terraces rose here and there, and after some four miles we fixed the plane-table on one of these. No tower came in view as we eagerly scanned the horizon ; but luckily the Surveyor’s sharp eyes sighted animals grazing in the distance. So we made for them quickly, and after a couple of miles to the east came upon a large flock of sheep, cows, camels, and ponies. We were met by a number of rather ferocious dogs (from which ‘ Dash,’ my ever active little companion, had to seek shelter on Tila Bai’s saddle), and by two truculent-looking herdsmen, mounted on wiry Mongolian ponies and carrying long flintlocks. They proved to be Tungan nomads well acquainted with the riverine grazing-grounds on the Su-lo Ho and the high valleys and plateaus south of Tun-huang. Their rough looks and rather aggressive bearing would scarcely have inspired confidence in ordinary wayfarers. I could instinctively realize the loathing with which peace-loving Chinese, attached to their four walls, must regard such 48 WALL NORTH OF TUN-HUANG § cou rough, ever-roving customers. Whatever violence and cruel deceit they might be prepared for in the réle of brigands and rebels, when occasion arose, these Tungans had at least the saving grace of blunt, fearless speech and pride in their local knowledge. When I questioned them with Chiang’s help about old ‘ Pao-t’ais’ in these parts, and offered a reward for showing me any, the elder of the men, who claimed main ownership of the flock, after some consultation with his brother, agreed to guide us to a point where we might both obtain a sight of some towers and find water for camping. So led on by the Tungan we rode quickly ahead, until after another three miles or so he brought us to an isolated clay terrace some forty feet high on the edge of a wide marshy belt stretching away towards the river. I clambered up to the top, and looking south and south- east could count not less than ten towers extending ina line approximately east to west. The yellowish rays of the sinking sun lit them up clearly in spite of the distance. There could be no possible doubt about their marking the alignment of the old wall, and my delight at this success was great. The sturdy ruffan by my side had reason to be satisfied with the reward of silver I gave him on the spot without weighing. As he saw me gazing at the line of towers through my prismatic glasses he without hesita- tion jerked out the information that this marked ‘the old Han road from An-shi to Lop-nor.’ It sounded like a strange confirmation of the surmise I had nursed in my own mind for some time past. But I had no means yet to decide whether the antiquarian hint thrown out from this rough mouthpiece was correct, and still less whether the Tungan’s statement was derived from a shrewd guess or the echo of a distant tradition. He did not claim to have been much farther west along this ‘old Han road,’ But an elder brother of his, now trading about Hsi-ning-fu, had taken a strange ‘ Kuan’ or official along it to Lop-nor some seventeen years before. Judging from the date indicated, I strongly suspect that this adventurous traveller was the ill-fated M. Martin. As we know from M. Grenard’s account of the Dutreuil de Rhins expedition, co.uv GUIDE TO THE ‘OLD HAN ROAD’ 49 he was the first European in modern times to make his way across from Tun-huang to Abdal and Khotan, but did not live to record his story. Beyond the river to the north-north-east, and a con- siderable distance away, my guide showed me a large group of ruined buildings, temples according to his statement. But the flooded condition of the river made them inacces- sible for us now, and indications subsequently gathered make me inclined to believe that they were of relatively recent date, belonging to an abandoned road-side station on the route to Hami. Gladly would I have kept my hardy Tungan for a guide in these regions, however obnoxious his presence might be to my Chinese entourage. But no fair words or offers would induce him to stay with me even for a day. He would not leave his flock in sole charge of others, and in all probability he preferred his sturdy independence to any service, however easy, which would bring him into constant touch with the despised ‘heathen Chinee.’ He left us soon, but promised to turn up in the morning with a sheep for which he was to be paid a good price. I hoped to get then at least a photograph of him; but he never came, and the men I sent to fetch the sheep failed to find him. So the only man who could or would tell me of the old wall vanished from my horizon for good. I often wondered where the next Tungan rebellion, when it is due, will find him ! For me the day had closed with cheering promise, and the men found warm shelter in shallow cave-dwellings which some earlier occupants had excavated at the foot of the clay ridge. It was the more welcome as a cutting wind continued to blow from the east and the thermometer still showed twenty-five degrees of frost in the morning. Besides myself, only Naik Ram Singh, ever scrupulously clean and tidy about his person, preferred to stick to his little tent. The Chinese labourers were overjoyed at finding a nice den to huddle into; but Hassan Akhun and the rest of my Turki followers, not liking the vicinity of the Tungans, tethered ponies and camels close by and kept a _ watch all night. VOL, II E 50 WALL NORTH OF TUN-HUANG 9 cu uy On the morning of March 27th I let the camp remain where it was, and set out with my assistants and half-a- dozen labourers to the east-south-east. I there hoped to strike approximately in the middle the line of towers I had sighted. Three miles across the scrub-covered plain brought us to another conspicuous clay ridge with a troglodyte dwelling occupied by a half-crazy wood-cutter, That he, being an orthodox Chinaman, expressed stereo- typed ignorance about ruined towers and everything else was not a matter of consequence. Pushing onwards we passed through a belt of exceptionally thick scrub and low tamarisk cones, in which an inundation from the Su-lo Ho was steadily spreading. Nothing could induce our civilized slum-dwelling coolies to wade through the narrow channels. As they had each time to be mounted on ponies, one by one, progress was far too slow for my eagerness. At last we emerged on a bare pebble Sai with much dead wood on the ground and isolated stunted Toghraks still living. The whole dreary ground bore the stamp of desiccation. But this was not the time for such observa- tions. Right in front of me I saw rising the cone of an old watch-tower just of the shape and construction I had first seen in the desert westwards, and towards it I galloped as quick as ‘ Badakhshi,’ my hardy pony, would bear me. Before reaching it I noticed a low mound com- posed of the familiar fascines and clay layers, stretching away across the bare gravel to the nearest tower on the east, and continuing also with a divergent angle south- westwards. There could be no longer any doubt: I was back again on ‘my’ wall! The watch-tower, built entirely of regular courses of hard clay about four inches thick, with thin layers | of tamarisk branches laid between them, still rose to over | twenty-two feet. In order to give additional cohesion > to the solid base measuring about twenty feet square, numerous wooden posts had been set in it vertically, and their ends were sticking out on the top. The wall once guarded by the tower had passed to the north of it, with — a bastion-like projection at about six yards’ distance. CH. LIV BACK AGAIN ON ‘MY’ WALL nt Wind and driving sand had destroyed all but the lowest layer of fascines, here all made up of tamarisk branches. But this, with the overlying stratum of clay and gravel, cropped out so clearly on the level flat that the line of wall which it marked was easily followed by the eye far away. The next tower on the east towards which this wall or agger ran quite straight proved to be only one and one-eighth mile distant. Beyond I could see three more ‘ Pao-t’ais’ ; but the examination of them had to be postponed. Instead, I directed my attention to a close search of the ground immediately adjoining T. xxvi., as | num- bered the tower just described. Fortune for once seemed inclined to encourage me at the outset. About four yards from the south-east corner of the tower I noticed slight refuse cropping out on the gravel surface. It proved the last remnant of the rubbish once filling a small apartment about eight feet square. Only traces of mud-built walls with a plastered reed-facing survived. But within this scant shelter and almost on the surface there turned up a flat piece of thin wood, about one foot long and over an inch broad, with Chinese characters neatly inscribed in five columns. Dates such as I was eagerly asking my learned secretary to look out for, were found neither in this document, which Chiang took to be part of an account, nor in another clearly written but incomplete Chinese record of the ‘slip’ type, apparently referring to some arms. But Chiang declared the writing to be of a strangely ancient look, and in any case the discovery of records at a spot at first sight so unpromising justified further hopes. The labourers, somewhat roused from their torpor by the prompt payment of a good reward, scraped the ground in vain for remains in other quarters adjoining the tower. The only find was two copper coins of the Han period. But as this type had continued to circulate right down to the early middle ages, they could not by themselves suffice for the dating of a ruin even now within reach of people from the oasis. In order not to tax the tender feet of my Chinese diggers too severely, I decided to turn next to the first tower south-westwards, whence return to camp would be 52 WALL NORTH OF TUN-HUANG 3 cx. ww nearer. For about three-quarters of a mile I could trace the line of the wall quite clearly, still rising in places to over three feet. Then we lost it on difficult ground, amidst tamarisk cones with soft eroded soil and dunes of fine drift sand between. The ruin which we reached after another mile and a half was undoubtedly that of a watch-tower, of the usual size, but badly decayed on some faces (Fig. 164). It had been built on a small clay terrace, which rose about seventeen feet above the eroded ground level on the south. On its west side the tower was ad- joined by a mass of soft refuse about fifteen feet across, filling the remains of some poorly built structure to a height of three to four feet. I had scarcely set the men to work when, on the south- west and almost on the surface, there were found three wooden slips inscribed with clear Chinese characters. They were quite complete, and showed the usual size, being about nine and a half inches long and half an inch wide (Fig. 119). Chiang at once recognized that two of them bore full dates, and our excitement was great. Presently three more inscribed narrow tablets emerged from under half a foot or so of rubbish in the middle of the heap, one of them being dated. Evidently we had struck a rich mine. But there was no time that evening to clear it with care; and as Chiang-ssti-yeh was unable to fix the ‘ Nien-haos,’ or regnal titles in which the dates were recorded, I hastened to return to camp by sunset. We were both greatly exercised by conjectures as to the age which the date records, when identified, would reveal for the ruined wall and towers. Our high spirits were in a way shared by the labourers, who tramped after us pleased with the silver I had given in reward for the day’s finds. Arrived in camp by nightfall I almost grudged the time — needed for a wash and hasty dinner before settling down | with Chiang to search for those mysterious ‘ Nien-haos’ in the chronological tables attached to Mayers’ excellent Chinese Reader's Manual. It proved quite a thrilling hunt. In the absence of any definite clue, we had to search through the hundreds of regnal periods comprised 164. REMAINS OF ANCIENT WATCH-TOWER T. XXVII., TUN-HUANG LIMES. The ruin stands on a natural clay terrace of which the continuation is seen on right. Naik Ram Singh in foreground. OE NER Y WRE ten ec pg 165. REMAINS OF ANCIENT BORDER WALL, BETWEEN LOW DUNES, EAST OF TOWER T. XXXV., TUN-HUANG LIMES. The Chinese labourer on left stands on surface of low gravel mound. CH. LIV FIKkST DATEDEHANGRECORD 53 within the possible limits from Han to Sung times. Of the reading of one ‘ Nien-hao,’ Yung P’ing, Chiang felt quite sure. But, alas! this proved to be represented several times, in the sixth as well as the third century a.p., and farther back I scarcely had the courage to look. In the other ‘ Nien-hao,’ Chien , my learned secretary was at a loss how to recognize the second character, though written quite clearly in two tablets. Those who know something of the intricacies of Chinese script, with its tens of thousands of distinct ideograms and their palaeographic variation at different periods, will not wonder at his doubts. I had vainly searched near the Yung P’ing periods already mentioned for a regnal title likely to give Chiang a clue to Chien , when at last in my despair I boldly took a jump of several centuries. There was a Yung P’ing period commencing in 58 a.p.,—and just before it there stood the regnal title of Chien Wu. Without a moment’s hesitation Chiang recognized in it the character which had puzzled him so far. It was the first title adopted by the Emperor who founded the Eastern Han dynasty in 25 a.D., and the twenty-sixth year mentioned in our two tablets corresponded to 50 a.p._ So the ruined frontier wall I had set out to explore went at least as far back as the first century of our era, and as proof I had in my hands the oldest written Chinese records so far known! We both loudly rejoiced at this discovery, which put us at once on safe chronological ground for further researches. Even ‘Dash’ was roused from below the blankets of my camp bed where he lay peacefully curled up for the night. I wondered what he thought of the excitement displayed by his devoted Chinese friend and by his own master. But, indeed, Chiang’s historical sense was now keenly stirred, and I myself felt highly elated ; for I had all along put faith in the antiquity of this Lzmes, and now felt confidence in its successful exploration. CHAPTER BEV DISCOVERY OF HAN RECORDS Next morning broke with an icy north wind which later on shifted round to the north-west without losing any of its violence. My first business was to despatch one of the Ya-mén messengers with a letter to the magistrate asking for more labourers to push on excavations. Then I set out with every available man for the ruin sighted due south of our camp and next to the one prospected in the evening. The camp was to follow with a supply of water in tanks. It was essential to spare our handful of diggers all needless tramps to and fro; for I rightly suspected that with such shifty folk, all confirmed opium-smokers, the stimulus supplied by liberal rewards for finds would not hold out long, but only increase the craving for a good smoke and sleep in a warm den. The watch-tower, for such it was, proved badly decayed ; but thick layers of refuse covered the south slope of the low clay ridge on which as usual it had been built for the sake of better look-out, over some ten yards of length. At the foot of the slope they were fully three to four feet high. The chief ingredients were straw, twigs and bark of tamarisks, dung of horses—evidently mainly stable refuse of some watch and post station thrown down here. But from the very edge on the top there protruded the fragment of an inscribed Chinese tablet, and as more wooden records cropped up I promptly settled down to work here. The harvest was abundant. Before mid-day two dozen or so of inscribed pieces had emerged, and the precious refuse heap was far from being exhausted. Chiang was 54 cx.tv CHINESE DOCUMENTS ON WOOD 55 indefatigable in exhorting and watching the coolies, pouncing upon and cleaning every dirt-encrusted ‘ tzti’ or writing with the utmost keenness. I could safely leave him and Naik Ram Singh to continue operations, while I myself set out for a preliminary inspection of the two remaining watch- towers visible south-westwards. By the time I returned from this reconnaissance the clearing of the whole of the refuse layers was completed, and the total number of inscribed pieces had risen to over seventy. Only two of the records were fully dated, the year named corresponding to 75 a.p., and thus showing that the relics of this watch-station likewise went back to the Eastern Han dynasty. I could not doubt for a moment that the full interpretation of these records would need protracted study, and would tax the philological acumen even of so eminent a Sinologist as my friend M. Chavannes, for whom | destined them from the first. Chiang himself modestly disclaimed the capacity of solving the many puzzles in palaeographic features and in diction which the text of the tablets offered. Many of them, besides, were incomplete. Yet even his cursory examination sufficed to show that the records varied greatly in character. Brief reports on matters of military administration along the line of watch-stations ; acknowledgments of receipt for articles of equipment, etc. private communications; fragments of literary texts ; even writing exercises seemed to be represented. But on all such points definite information came only through M. Chavannes’ labours; and for a general survey of the results which his unsurpassed learning and critical penetration secured from the materials discovered at this and other _ ancient stations along the wall, I must refer to a subsequent _ chapter. It was far easier, of course, to become familiar on the spot with the external or stationery aspect of these mis- _cellaneous ‘papers,’ to use an anachronism, the earliest of _ Chinese written records known till then. The most usual form was that of the thin wooden slip measuring when intact about nine to nine and a half inches long and from a quarter to half an inch wide (see Fig. 119). That some of the complete slips often contained over thirty Chinese 56 DISCOVERY OF HAN RECORDS = cu. w characters (z.e. words) in a single vertical line will help to illustrate the remarkable neatness of the writing which prevails in these records. Sometimes the writing was in several columns, or also continued on the back. Among the woods used for these ‘slips’ that of the poplar seemed most frequent, as at the sites I had explored in the Tarim Basin. But, besides, there appeared a peculiarly streaked soft wood which the Naik recognized as belonging to some conifer. It could not have grown in a climate so arid as the Su-lo Ho Basin must have had throughout historical times. I conjectured it to have been brought from the slopes of the western Nan-shan, and there, in fact, I subsequently came across remnants of fir forest. A still more distant import was represented by the neat slips of bamboo which turned up at other ruined stations. But here in the refuse of tower T. xxvull. variety was imparted to the wooden stationery, also, by the plentiful ‘fancy’ use of that abundant local material, the tamarisk. There were tamarisk sticks of varying length, roughly cut into polygonal shapes, and inscribed on a number of sides ; broad labels with rounded tops or peg-shaped, etc. Evidently convention was not so strict in the case of internal communications as about official correspondence, and for mere ‘copy writing,’ with which soldiers quartered at this and other stations had evidently beguiled their time, sticks of tamarisk cut on the spot were certainly good enough, At the same time the number of ‘shavings’ from regular slips (see Fig. 119, 14), and the fact of the latter being found often thinned down by repeated previous paring, showed that the supply of proper wooden stationery had its value, and was used over and over again. Miscellaneous objects in wood also turned up among the refuse in plenty: such as small marked cubes apparently used for gambling or divination (Fig. 119, 15); tally sticks ; fragments of combs. But what interested me most were two wooden seal-cases, evidently meant to be attached to some closed bag or other receptacle. They showed the identical arrangement of three grooves for folds of string over which the seal was to be impressed in clay as I had first found on the envelopes of Kharoshthi documents at civ ANCIENT WOODEN STATIONERY 57 the Niya site. It was conclusive proof that I had been right years before in tracing all such details of that ancient wooden stationery of Turkestan back to earlier Chinese models, No trace of the wall itself survived here, nor remains of the quarters sheltering those who had kept watch by the tower. Yet from the refuse thrown out by them it was possible to draw some conclusion as to their conditions of life. The line which the towers guarded must have already in the first century A.D. passed through desert ground. Wind erosion had, no doubt, progressed since; yet from the very position in which the undisturbed horizontal strata of rubbish were found, some ten feet below the level of the tower base, it was clear that the bare clay ridges rising above eroded ground had then already formed the characteristic feature of the site. A curious indication of the remoteness of the guarded wall line from the inhabited area was supplied by the numerous fragments of coarse grey pottery, remarkably hard, which lay scattered in plenty on the surface, and often were found perforated on the edges with regular drilled holes. The discovery in the rubbish heap of several pieces still joined together by string of some rough vegetable fibre explained these holes (Fig. 172, 6), and bore witness to the value which the quondam owners had attached to their pots and jars however badly damaged. As the material was cheap enough, only the difficulty of transporting larger earthenware from the oasis would account for this continued use after the roughest mending. On March 2gth I took my men to the next tower south- westwards, which I had already reconnoitred. The distance proved to be only one and one-eighth mile. The tower, built of solid layers of stamped clay, rose in fair preserva- tion to over twenty feet, and still bore on its top portions of a brick-built parapet below which horizontal rafters pro- jected. No trace of stairs remained, but some holes on the south face had probably been utilized for a sort of ladder. Here, too, the tower had been built on a small clay ridge, no such advantage of ground being ever neglected by those who constructed the ‘ Wall.’ A peculiar feature here 58 DISCOVERY OF HAN RECORDS © ca.1v was a relatively well preserved enclosure, about 107 feet square, of which the tower itself formed the north-west corner. Its walls, built of rough bricks and clay, with layers of tamarisk brushwood at intervals, still rose in one place to eight feet. Whether this enclosure was coeval with the tower and the wall line it was meant to guard, I was unable to decide ; for the accumulations of refuse found within its east face, and partly covering the floor and decayed walls of some rooms, yielded nothing but plentiful reed straw, cut brush- wood, and dung of horses and camels. A large refuse heap some ten yards outside the south wall proved of similar composition. But here we came upon a fine jar, intact but for its mouth, and in shape resembling an amphora (Fig.172,5). Its height was nearly one foot, and its material a very hard dark brown stone-ware, with a mat slip burnt in. Within, traces of an oily substance survived. Small pieces of fine pottery, made of a very hard paste and with highly glazed surface, were picked up in plenty both within and around the enclosure. The glaze colours varied greatly ; celadon green, ivory white, fine browns and black, an exquisite turquoise blue with several mottled tints, being all represented. I was still wondering how to account for this un- expected abundance of superior pottery débris such as I had not yet come upon elsewhere in this region, when the clearing of a small ruined cella, previously noticed about fifty yards to the west, brought another surprise. It measured only eleven by thirteen feet outside, and with its entrance on the narrower side faced to the south. The walls, about two feet thick and built of fairly hard sun-dried bricks, rose nowhere above four to five feet. The interior was filled with débris of broken bricks, charred timber, and plastered reed wattle. When it was being cleared we came upon a platform built of bricks and running round all sides but that of the entrance. From the débris covering the platform to the north there came to light numerous fragments of stucco sculpture that unmistakably had once adorned a Buddhist shrine. The fragments were all badly broken, but showed in CH. LV RUIN OF BUDDHIST SHRINE 59 every detail close dependence on the models of Graeco- Buddhist art as transplanted to Central Asia. There were portions of the arms of a statue somewhat under life size, modelled round cleverly dowelled wooden cores, bearing ornaments, and of well-shaped hands and fingers. The blackened surface of the stucco, originally not baked, and the charred condition of the projecting core portions, showed that the statuary, like the little temple itself, had been destroyed in a conflagration. One fragment of par- ticular interest showed two small heads, one above the other, each only about three inches high, but excellently modelled (Fig. 273,4). While the upper one displays a look of placid contemplation, the lower one, with frowning brows and eyes and mouth wide open, cleverly expresses intense anger or passion. As a third head is evidently missing below, it seems likely that one of those numerous ‘ Trimurti’ representations of Buddhist divinities was intended, in which the ‘angry’ or demoniac form of the god usually plays a part. However this might be, so much was clear, that the remains of the small Buddhist shrine here uncovered had some relation to the watch-station close by and the wall line which passed it. But the style of the sculptured remains, though unmistakably old, seemed to speak against contemporary construction. So I was led to conjecture that it was, perhaps, the tenacity of local worship—such as I had often seen exemplified elsewhere, and last among the ruins of deserted Shih-pan-tung—which had here caused a small shrine to be restored centuries after the wall was abandoned. I did not realize until some time later that the direct route from the Tun-huang oasis towards Hami and the northern oases of Chinese Turkestan, passes even now in the vicinity of this old watch-station (T. xxx.) Thus, if we may assume that already in ancient times it crossed the line of wall here, the existence of a small shrine near the gate station and its continued maintenance by pious wayfarers, say down to T’ang times, present nothing strange. My subsequent discovery of a similar cult having survived on the old route westwards supplies an exact parallel. 60 DISCOVERY OF HAN RECORDS © cx..v But even in the immediate neighbourhood I made an observation which supports that assumption, though at the time I did not realize its true import. On the previous day’s reconnaissance to the south-west, towards the next and last tower visible on that side, | had noticed, on the open salt-encrusted steppe intervening, and at a distance of less than a mile from T. xx1x., some rough enclosures built of salt-impregnated clay lumps, and in the middle of one a miniature chapel, half-ruined, made up of the same coarse material. Looking back in the light of the indications since gathered, it now appears to me probable that this modest substitute for a shrine, manifestly of recent con- struction, represents the last lingering trace of the cult which those leaving or regaining the border wall of the Empire once paid to the sanctuary at this ‘ Gate.’ The westernmost tower extant on this part of the line proved less than two miles distant from T. xxrx., and was a solid square mass of stamped clay. The small erosion terrace on which it was built made it a conspicuous landmark on this dismal salt-covered flat. No ancient remains of any sort except broken pottery could be traced near it, and far-advanced erosion had left neither trace of the wall nor any chance for digging. But I visited this tower (T. xxx.) again on my return march to Tun-huang, and then found that the deep-cut cart-track marking the route from the oasis to Hami actually passed between it and the tower T. xx1x., which I now believe to have stood by the ancient gate through the wall. Though the view from the terrace of T. xxx. was open, no other ruin could be seen to the west except the tower T. xxiv., which I had already examined on my way north of Shih- pan-tung. It just showed its top over a maze of clay terraces to the west. So my survey in this direction was completed. On March 30th, when we had a considerable fall of temperature down to thirty degrees of frost Fahrenheit, with the wind veering round to the west, I took my men back to the ruined watch-station (T. xxvu.) which had yielded the first dated records (Fig. 164). The débris adjoining the tower proved to belong to a room about CH. LV LINGERING LOCAL GULT 61 eleven by fourteen feet, partly cut out of the live clay of the narrow ridge occupied by the tower. On clearing it and a little terrace or loggia which faced it from the south, we recovered two dozen more Chinese records on wood, nearly half of them complete and in good preservation. One of these, as we recognized with joy on the spot, showed a clearly written date corresponding to 35 ap. This meant a farther step back in antiquity. Chiang was sure that almost all the ‘slips’ referred to military posts or individual officers, though he could not make out all the details. So I concluded that the room by the side of the watch-tower had served as quarters for some officer or clerk attached to the troops guarding this part of the line. M. Chavannes’ analysis has since fully confirmed my conjecture. One neatly written label, with a string still attached, had evidently been taken from a bundle contain- ing a soldier’s outfit. From my learned collaborator’s translation I now know that it mentions the company (77/sten-wez) he belonged to, as well as the locality (Wan- sut) which he helped to guard. Over a dozen blank tablets of the regular size evidently belonged to the stock of stationery kept ready at this little office. Small miscellaneous finds were abundant among the refuse within the quarters and strewing the slope. Apart from remains of cups of glazed stone-ware, spoons of wood, a broom, a wooden seal, | may mention a fire-stick (Fig. 173, 3), exactly conforming to the pieces found at the sites of Niya and Endere as regards shape and arrangement of the holes in which fire was produced by rubbing. On the evening of that day we were joined by eight fresh labourers sent with a Ya-mén attendant, and I was heartily glad to get them; for the men of the first batch already complained of exhaustion. Even the chance of - gaining rewards by lucky finds failed to retain them longer, though at first it had appealed to them greatly as con- _ firmed gamblers. Inwardly I could scarcely condemn them altogether; for with the temperature falling that night to a minimum of seven degrees below zero Fahrenheit, the strong wind still blowing from the west made itself felt intensely. Luckily fuel was abundant. 62 DISCOVERY. OF -HAN@RECORDS: "“caay Whether it was mere reluctance to face the rigours of this desert ‘spring’ any longer, or disgust at the indepen- dent bearing of the Chinese labourers who were supposed to make themselves useful about camp fires, etc., but would not stand any of the bullying which my Kashmiri cook liked to indulge in, this worthy thought the time opportune first to go on strike and then to abscond with a pony in the morning. I knew that he could find his way back to Tun- huang, and that he would not fail to be stopped there as a suspected deserter. Still this incident did not add to the amenities of my Easter Sunday. I used it to explore the remains of the wall as far as they were traceable eastwards. Moving back with my whole camp to the tower where the first tablets had been found, I then marched along the line marked by the four towers we had sighted before. The distances between them varied considerably, from one and one-eighth to three-quarters of a mile, though the ground was throughout a uniformly bare expanse of gravel. On it the line of the wall showed up quite clearly, both in the straight curtains between the towers and in the semicircular bastions by which the line curved round to the north of each tower. In many places the alternating layers of fascines and gravelly clay still rose to three feet or thereabouts. But even where this agger was — reduced to nearly ground level, the layer of thick tamarisk branches used for a foundation was seen emerging on either side of the low gravel-covered swelling. There © could be no doubt that it was the bearing of the line, nearly east to west, though not absolutely straight, which, being parallel to the direction of the prevalent winds, had helped to preserve this unbroken stretch of wall over five miles long. The towers differed but little in construction, being © built of solid layers of stamped clay about twenty feet square at the base and originally over twenty feet high. Those to the east had suffered more decay. Beyond this the position of a fifth tower could be traced only by a low débris heap and the bastion-like projection of the wall. Half a mile to the east of this last tower (T. xxxv.), we lost the line of the wall amidst dunes of drift sand, rising CH. LV CONSTRUCTION OF WALL 63 to about fifteen feet and evidently encroaching from the south. But then it again emerged in patches, and at last on a broad gravel belt lined with dunes both to the north and south we came upon a remarkably well preserved bit of wall, quite unbroken for 256 yards, and rising in places to fully seven feet (Fig. 165). Its preservation was evidently due to a protecting cover of sand, though now the drift heaped up against the wall lay only three to five feet high. In the centre of this stretch the wall had a remarkably solid appearance. The sides showed scarcely any trace of erosion, except that the outer facing with fascines laid in the direction of the wall was missing. Here the particular method of construction could be studied with ease. Layers of fascines, six inches thick, made up of mixed tamarisk twigs and reeds, alternated with strata three to four inches thick of coarse clay and gravel, as taken on the spot. Where I photographed the wall, as seen in Fig. 163, I counted eight double layers of fascines and stamped clay, respectively. I noticed that the reeds generally prevailed in a thin streak on the top of the tamarisk brushwood. This suggested that they had been specially inserted there in order to prepare a more level surface for the succeeding stratum of clay and gravel. It seemed to me highly probable that these latter layers had been regularly stamped, the water for the purpose being brought probably from the nearest lagoon. The salts contained everywhere in the soil and water, and attested in the wall itself by a great deal of efflores- cence, had given to the strange wall thus constructed a quasi-petrified consistency. In such a region it could hold its own against man and nature—all forces, in fact, but that of slow-grinding but almost incessant wind erosion. The thickness of the wall measured close on seven feet across the top, and allowing for the loss which the uppermost fascine layer had suffered on its edges through erosion, about one foot more at the base. As I looked at the wall here rising before me still solid and with almost vertical faces, I could not help being struck by the skill with which the old Chinese engineers had 64 DISCOVERY OF HAN ‘RECORDS city improvised their rampart. Across an extensive desert area, bare of all resources, and of water in particular, it must have been a difficult task to construct a wall so solid as this, upon which even modern field artillery would make but little impression. The materials to which they had recourse, though of little apparent strength, were particu- larly well adapted to local conditions. I doubt whether any others within practicable reach could have stood better the stress of two thousand years and the constant onset of eroding forces. I marched on for a mile and a half farther along and through low sand-dunes without coming upon any trace of the wall or sighting any more towers, though the view was open enough. Then regard for the animals, which needed water and grazing, obliged us to turn off north- ward in the direction of the river. We crossed in suc- cession a belt of absolutely bare gravel; a dry river bed with Toghraks still alive; a zone where tamarisk growth was plentiful, but all dead; and finally, after seven miles from the wall, arrived at the deep-cut bed of the Su-lo Ho fringed by a riverine jungle of scrub and wild poplars. Where we camped for the night, the river, or the branch we could see, was over fifty yards broad and cer- tainly far too deep for fording. Its muddy water, carrying big ice cakes, flowed with a velocity of about two yards per second. The night was not so bitterly cold as the one preceding, but the wind steadily increased in strength until the atmosphere in the morning assumed a regular Buran hue. The haze was sure to last for days, and a further search for wall and towers eastward would have little or no chance under these conditions. Besides, it would have carried me to the town of An-hsi which I was bound to visit in any case later. The main object of my search was already secured, I had discovered that the remains.of the ancient wall actually continued eastwards of Tun-huang, as I conjectured from the first. I also had — been able to prove the occupation of this Lzmes in the first century A.D. So I decided, on April 1st, to send the camp under the Surveyor’s guidance back, to the conspicuous ridge CH. LV REPURNGIN@SLIZZARD 65 where we had found the wood-cutter’s troglodyte dwelling, and to make my way there myself with the labourers along the line of the wall. It proved a more difficult matter than I had thought under the thick veil of dust which the gale raised. Luckily we had our footprints to guide us across the gravel Sai, and then when we got among the dunes where there was much driving of sand, the remains of the wall served to direct us safely. At each tower we searched whatever remains of quarters or refuse heaps could be traced. But whether it was on account of the greater erosion to which they had been exposed on the open Sai, flat like a billiard table, or for some other reason, they proved decidedly scanty. The finds of wooden records, all fragmentary, scarcely numbered half-a-dozen. However, we picked up several well-made triangular arrow-heads in bronze and a few coins belonging to the Han period. It was a trying day’s work, and I felt heartily glad when the force of the gale abated towards the evening, and we could move to the appointed camping-place in somewhat less discomfort. The footprints of the track by which we had come from there to Tower T. xxvI. on March 27th were still perfectly distinct on the gravelly soil. It was an interesting proof, thereafter often observed in a still more striking fashion, of how little deflation and the movement of fine sand affect the surface of such ground. Nevertheless it was a relief when at last I saw the camp fires and was sure of the night’s shelter and food. The wind had now veered round to the north-west. From midnight until daybreak violent gusts of wind shook the tent, and when I stepped out of it in the morning slight snow-flakes were driving for some minutes. The atmosphere was quite murky with a fog of fine sand, very irritating to throat, eyes, and nose. We were now re- turning to the oasis, but before we reached it we had to pass through the bleakest and chilliest day experi- enced since Lop-nor. The blizzard never slackened, and cut through our warmest furs. The landscape was in ugubrious harmony. Following a deep-cut track which VOL, II F 66 DISCOVERY OF HAN RECORDS ‘cum@ carts had worn, and which in the end proved to mark the route coming from Hami, we reached after about six miles the westernmost ruined tower already referred to. The area of scrub and sand had ceased before. We now crossed an absolutely bare, salt-encrusted steppe which fuel-collectors from the oasis had evidently long ago cleared of all dead tamarisks. But the cones which these had helped to form while alive, still rose to eight or ten feet. At last low scrub reappeared over dune-like formations, and then the first ragged outposts of cultiva- tion in the shape of scattered fields and trees. We had covered close on twenty miles, and felt half-choked with dust and half-frozen when we arrived again at our old camping-place at Shih-tsao. GEA E EhGHenvil TO THE NAN-HU OASIS Our march on April 3rd from Shih-tsao back to Tun-huang was short, and, | confess, we all felt glad for it. The pros- pect of shelter was pleasant after the icy blasts we had faced for the last week along that desolate ‘Great Wall.’ Whether it was the protection afforded by the trees of the oasis, or at last a sign of approaching spring, the air seemed warmer in spite of the continued north wind. Being with Chiang-ssti-yeh far ahead of the baggage, I could use the time gained before pitching camp for a visit to the large shrine which at the time of my start I had noticed near the west gate of the town. It boasted of a high pavilion-like structure, the first ‘ Pagoda’ of the conventional type I had seen, and seemed in exceptionally good repair for this place of somnolent nonchalance. The frescoes of the outer gate showed that it was a Taoist temple, and by their new look prepared me for the inferior art of the decoration within. But there was compensation in the glimpse I unexpectedly gained here of one of the main schools of Tun-huang. As soon as we had entered the inner court, a swarm of boys, mostly chubby and well clad, gathered around us. The teacher had repaired to his house in the town on some business, and his score or more of pupils were hugely enjoying the unearned recess. The halls on either side of the court bore so unmistakably the impress of scholastic use that for a moment I almost underwent the not altogether cheering sensation of having come for ‘inspection duty.’ There were lumbering big black desks near the windows, each covered with the ‘copy-slips’ and exercises of three or four pupils. The walls were hung appropriately with 67 68 TO THE NAN-HU OASIS CH. LVI calligraphic specimens and ‘moral sentences’ penned evidently by the teacher or by select scholars on big rolls of red paper. There was order about the whole place and an air of austere concentration on the school work which made me compare this establishment favourably with many a secondary school in the Punjab, where neither ample codes and inspections nor the supply of the latest appliances have succeeded in developing a sense of orderly arrange- ment. How much did I feel my total want of philological preparation as I handled the much-thumbed and yet neatly kept primers and elementary classics of the several forms! The boys, big and small, showed good manners, and com- bined an alert air with such restraint on their youthful curiosity that I could not help mentally awarding a good note to the teacher for the ‘tone’ he had implanted in his school. I had decided to restrict my stay at Tun-huang to a single day, April 4th, and knew well that it would mean anything but a rest for me. Yet soon after my arrival at ‘my’ orchard I found that the time would have to suffice for a good deal more work than I had expected. Sher Ali Khan, the enterprising Afghan trader who during my first stay had reached Tun-huang from a journey to Kan- chou, was to have despatched his caravan of forty camels laden with tea to Charklik and Khotan long before my return. But with that truly Eastern disregard for exact dates and supposed urgency which for us Western people would at times be so useful a sedative, he now came to tell me that his camels had only started that morning, and that he himself would still remain longer ready for any service. The opportunity of sending a mail safely to Kashgar and thus to my friends in Europe was as welcome > as it was unexpected. It is true none of the letters were © as yet written. But a messenger despatched by next morning would catch up the caravan easily, and I had an — evening and if need be a night before me for filling m mail bag. | This catching of an unforeseen mail train was not made > easy for me. I had scarcely had time to wash the outer crust of dust off my face when, to my surprise, my insepar- CH. LVI BUSY DAY AT TUN-HUANG 69 able local friends and protectors, Lin Ta-jén and Wang Ta-lao-ye, turned up for a visit. I had little doubt they were both glad to welcome us back at headquarters, and interested, too, to hear from our own mouths the story of our peregrinations and finds. But there was information, too, gathered as we passed through the town, to explain an increased display of official attention and support. It appeared that the telegraphic salutation I had sent to the Lan-chou Viceroy on arrival within the Kan-su borders had promptly been followed up by a telegraphic circular from that high dignitary to the Tao-t’ais of the several provinces, and through them to all the magistrates, strongly recom- mending me and my researches to official notice. This explained the embarrassing haste of my friendly visitors. But for the length of their visit I had to hold my own finds to account. With the enthusiasm of a trained scholar Wang Ta-lao-ye could not forgo the pleasure of handling and zmzpromptu deciphering those ‘wooden letters’ of the great Han times. Little problems which had baffled my ‘field literatus’ only incited his zeal. Tablet after tablet was scanned and commentated with a rapidity and ease which, despite my ignorance of Sinologist lore, enabled me to realize the sound learning of this Kan-su edition of P’an Ta-jén. Lin, his military colleague, wisely refrained from any pretence at equal scholarly qualifications. Yet I could see that his lively interest too was roused by the details of military organization on this ancient frontier, as revealed in those modest records from the posts once guarding it. I did not fail to use the Opportunities offered by whetted appetites and by the support from higher quarters in order to impress my friends with the need of effectively combating local vzs tmertiae in the way of labour for excavations and adequate transport arrangements. I was doubly glad for the diminished cold within the oasis; for it was 3 A.M. before I could close my mail that night, and with an icy gale blowing it would have been impossible to keep my tent warm enough for all the long writing. There was no time to make up for lost sleep ; for from an early hour next morning arrangements 7O TO THE NAN-HU OASIS CH. LVI for fresh supplies and transport, payment of labourers, etc., kept me busy. The weighing out of silver for all payments, big and small, proved as always a tantalizingly slow business. Patience is needed for supervising the process and subsequently for satisfying the recipients that they have been treated fairly. Ttun-huang merchants, | was told, use three kinds of scales, with a view to profit by their slight differences when selling, buying, or exchanging big horse-shoes of silver for small pieces. It was hopeless to battle with such refinement in a primitive system of currency. The simplest sacrifice was to accept the local traders’ verdict, which made my scales brought from Charklik weigh about four per cent too light. What a story of fiscal experiments is disclosed by the queer fact that no Chinaman in these parts will ever accept coined silver or gold! Shapeless bits of metal are more readily trusted,—and yet I soon learned to be on the guard against pieces artfully loaded with lead. How difficult it would be for a future antiquary to believe the fact that a region sufficiently advanced to possess paper money is yet stolidly resisting all attempts to introduce the permissive use of coined silver! | By making all payments in silver even for petty items I certainly escaped reference to the daily varying rates of exchange between copper ‘cash’ and silver. But, naturally, the worry of finding pieces of hacked silver corresponding exactly to the amounts due, down to decimal fractions of an ounce, was great. Luckily I soon found that my zealous secretary was an excellent hand in adjusting such petty claims with that strangely archaic currency of bullion. Having read out to me the various amounts in the presence of the claimants, he would let me make up the total and weigh out the whole in a lump heap of small silver pieces, allowance being duly made for the difference of our own from the Tun-huang scales. He would next start distribu- tion by squatting down on a little mat or carpet outside my tent and making the men sit in small groups. He then arranged these with infinite patience again and again, until the amounts due to the few people in each group could be accurately made up out of the available silver © CH. LVI ARCHAIC BULLION CURRENCY 71 fragments lumped-up. Of course, I took care to supply him with a quantity of tiny chips to adjust slight differences, It took hours before such a settlement was completed. But I could go on with other business, and nobody else seemed much to mind about his own time. Everybody went away satisfied, but I wondered what further efforts it would cost each group to settle their mutual reckonings! I scarcely could tell now how that single day’s halt in Tun-huang, on April 4th, sufficed for all the manifold preparations for my main campaign on the remains of the ancient Lzmes in the desert westwards. But I managed somehow to raise a month’s supplies, twelve fresh labourers, additional camels for transport, and even as many Ketmans as by fair words and high prices I could get hold of among the Muhammadan refugees at Tun-huang. Experience had shown me how much more useful for excavation those broad Turkestan hoes are than the spades and shovels of - the Chinese settled in the oasis. In the morning Ramzan, my faithless cook, turned up to make an unconditional surrender. His sudden return alone had, as expected, excited suspicion at Tun-huang, and he would have been obliged to await my own arrival under arrest at the Ya-mén had not Zahid Beg agreed to bail him out and keep watch over him. So the shifty, intractable Kashmiri realized that he had little chance of escaping from his contract even when near a great high _ road, and sulkily asked for his desertion to be forgiven as a sort of mental distemper brought on by the air of the desert. It was the story over again of Sadak Akhun, my queer cook in the winter of 1901, and the Jins of the Taklamakan. In the afternoon I spared time for return visits to the two Ya-méns, and on my way, noting the _ excellent wood-carving and ornamental brickwork on the gates of dilapidated old houses, again rejoiced that Tun- huang town at least had escaped the utter havoc worked by _ Tungan ferocity elsewhere. My plan was first to move south-west along the foot of _ the mountains to Nan-hu, a small oasis where Zahid Beg’s information and Roborowsky and Kozloff’s map indicated _ the existence of ruins. From there by going due north I 72 TO THE NAN-HU OASIS CH. LVI would strike the line of the ancient wall in the middle. On April 5th, the day of our start, move Cathatco, neither camels nor men turned up until quite late in the forenoon, Since the first march along the Tang Ho was bound to be short, this luckily mattered little. Where we crossed the river just outside the west gate I found its water flowing in a channel about forty yards wide and three to four feet deep, with a velocity of about two yards per second. All the canals of Tun-huang, taking off well above this, were flowing over-full at the same time. There could be no doubt that at this season of the early spring irrigation the supply of available water more than sufficed for the needs of the present cultivated area. On the left river bank we first skirted the crumbling clay walls of the old town of Tun-huang, a site said to have been occupied in T’ang times but now completely abandoned to fields and gardens. A subsequent measure- ment of the rectangular area enclosed by the walls, about 1500 yards from north to south and 650 across, showed that it was but slightly smaller than that of the present town, which is built within walls about 1100 yards square. Then we turned off to the south-west, and passing several well-kept temples, reached the edge of cultivation after a little over three miles. Here the ruins of a smaller walled town offered fresh proof of the destruction which followed the last great Tungan rising. Beyond we followed the banks of an earlier river bed, now completely dry and flanked on each side by a net- work of wind-eroded clay terraces. After some five miles farther, and not far from a modern-looking ‘ Pao-t’ai,’ | noticed ruined walls rising here and there above the bare gravel Sai westwards. Crossing the large canal which passes here close to the route, and conveys water for the western part of the Tun-huang oasis, I rode towards these walls and soon noticed that they invariably represented gateways to quadrangular enclosures which seemed com- pletely decayed. The gateways, on the other hand, looked solid enough, rising in several cases to a height of about twenty feet and showing a thickness of eight feet. But these wall portions on either side of a wide entrance cxu.tvzi STRANGE RUINS OF GATEWAYS 73 measured only five or six yards in length. Beyond this the front of the enclosure as well as its other sides showed only as low ridges of gravel. They were made just perceptible by the relief they presented in the slanting rays of the setting sun. While the ‘gateways’ showed fairly hard masonry of coarse but unusually large bricks, I vainly searched on the line of the enclosing ‘walls’ for any remains of brickwork or even of reed fascines. One of these strange quadrangles measured seventy-five by seventy yards, having its entrance, as usual, on the south. The enclosing ridges were invari- ably orientated, though but roughly. Within the enclosed areas there were always to be found several low tumuli, the largest facing the entrance from the north and the rest scattered without any apparent arrangement. Those in the ruin just mentioned measured from fifteen to seven yards in diameter, with a height not exceeding five feet. Of course, the idea that these were ancient places of burial soon occurred to me and to Chiang as well. But neither Chinese custom, nor what I knew of Buddhist and other religious practices in Central Asia, seemed to offer any clue. And if the tumuli should prove to contain graves I wondered how I should get Tun-huang people, particularly orthodox in their superstitious awe of graves, to help in the systematic opening. But the site was anyhow within easy reach of the oasis, and for the present I did not care to delay on its account. We found a convenient camping-place on a broad grassy flat known as Tung-wei-chii by the left bank of the river, and after a night when the thermometer still showed a minimum of twelve degrees of frost, started for the march to Nan-hu. It proved close on thirty miles. For the first half of this distance the route led along the southern edge of a gravel-covered plateau where it falls off with pre- cipitous cliffs to the deep-cut bed of the Tang Ho. It was like the counterscarp of a deep fosse with a glacis stretching away from its brink to the north. Absolutely barren outer hills, covered with dunes for the most part, rose from the right bank of the river and were fully in view throughout. A few half-ruined Pao-t’ais of 74 TO THE NAN-HU OASIS CH. LVI no particularly ancient look and a couple of small brick Stupas, well plastered and manifestly still receiving worship, were the only objects to distract the eye in this dreary landscape. The route left the river not far from the point where the Tang Ho valley turns sharply into the mountains south- east, and was skirting the foot of a gradually rising ridge when I first noticed what looked like a low dyke of gravel and stones, It only rose four or five feet above the bare Sai, and could easily have been mistaken for a natural swelling, had it not stretched away steadily to the south by west in a line absolutely straight. The route kept close by it for upwards of five miles. As the dyke was broad, measuring about twenty-four feet at its base, and the surface on its top hard, it seemed to be used for preference asacart track. But what could its real purpose have been? Without any trace of watch-towers or other structures, and with nothing but absolute desert to right and left, it seemed hard to imagine any defensive line of wall here. At last the route diverged to the south-west while the puzzling dyke could be seen running straight on towards a tower just visible far away in the distance. I was still searching in my mind for some explanation of this strange work of man in the wilderness, when my eye was caught by many curious low stone heaps rising on the level flat of gravel. Of greatly varying sizes, they were always circular in shape, and either had a straight line of stones attached on one side like a handle, or else faced small rectangular enclosures laid out with big pebbles. The circular cairns—for such they seemed— never rose more than three or four feet above the ground ; but as they appeared on all sides in dozens, brought into relief by the slanting light of the evening, the effect was quite weird. Was this the desert cemetery of some ancient population which had held the oasis we were approaching before the Chinese occupation, or primitive marks of cult left behind by some tribe which once had swept through this region? I knew of no analogy by which to guide my conjectures, nor could I stop there and then to dig up some of the cairns. But just as a dark patch of vegetation, seen westwards in the failing light, indicated approach to the Nan-hu oasis, I sighted not far from the track a brick-built gateway and an adjoining quadrangle marked by low gravel ridges, just like those I had met with the day before on the edge of the Tun-huang oasis. Within the quadrangle I could make out two circular tumuli of exactly the same shape as the ‘cairns’ I had just passed by in numbers. So anyhow it was clear that cairns and enclosures belonged to the same time and people. There was free scope for conjectures about them as I rode on in growing darkness amidst low dunes and tamarisk cones. At last the faint ripple of springs and then the glitter of a broad sheet of water assured us that the oasis was near. It was too late to search for houses. So we camped by the spring- fed stream, not a sound being heard from the village nor a soul coming near us. Dinner was an affair of midnight. The search which I had to make next morning for a camping-place suited to a longer halt, soon showed what a pleasant little oasis Nan-hu is. Over two miles long from east to west, and nearly as wide across, its area was everywhere irrigated by delightfully limpid water from the great spring-fed reservoir or lake which we had skirted in the darkness, and which accounted for the name of Nan-hu, meaning the ‘South Lake.’ It had been formed by damming up the head of a broad and deep-cut flood-bed which meandered right through the oasis, and _ with its steep banks of loess and wide marshy bottom closely recalled the ‘Yars’ familiar from Khotan or other Turkestan oases. The water-supply was manifestly _ abundant; for, quite apart from the canals taking off at _ the artificial lake, there was a lively stream flowing in the middle of the Yar, and carrying its clear water to waste in the desert northward. A low but picturesque line of | hills of red and yellowish sandstone, through which this | stream had cut its way in a gorge, shut off the view towards the desert and gave to the whole oasis a pleasing air of - seclusion. The twenty-five to thirty homesteads or farms which 76 TO THE NAN-HU OASIS CH. LVI it comprised lay scattered about in tiny hamlets, sheltered by fine elms and ashes. All round them extended well- tilled fields with rows of big trees lining the irrigation channels. So carefully was all ground within the oasis utilized that, not wishing to camp in the middle of a ploughed field, I was at last grateful to find room for my tent in a little quiet grove occupied by graves. The good folk of Nan-hu must have thought it a queer taste, but had no objection to offer. They seemed a quiet, thriving set of farmers, endowed with delightful zzsouczance such as their comfortable conditions as regards arable land and water would foster. About their pious zeal I could entertain no doubt when | found that this little settlement boasted of eight well-kept shrines ‘in being’ (Fig. 167), not counting the miniature chapels attached to almost every homestead and a number of small temples still in ruins, as the last Tungan inroad had left them. It seemed hard to think of that devastating tornado having swept across a place so placidly secluded as Nan-hu. Yet, according to the information we received, scarcely a man, woman, or child of the old population had then escaped with their lives. However this may be, those who had taken their places were now enjoying the ease resulting from under-population. But it was the opportunity for archaeological observa- tions of interest, not the rural attractions of this ‘sleepy hollow,’ which made me extend my stay to four days. At the first reconnaissance, guided by an elderly villager whom Lin Ta-jén’s petty officer had secured for us, we found a number of remains throwing light on the history of the oasis. At a distance of only about one mile east- wards from the edge of the present oasis, and approached over ground which manifestly had once been under cultiva- tion, there rose the broken clay walls of a small town built in the form of an irregular rectangle. Of the north face, measuring about 400 yards in length, the greater part still survived, half-buried under high dunes which had afforded protection. Of the east and west walls, too, considerable © portions were still extant though cut through and broken up by wind erosion. Yet the ‘masonry’ of the wall, — 166. RUINED HOMESTEAI), ABANDONED TO DESERT ABOUT SIXTY YEARS AGO, AT KUAN-TSOU, NORTH OF NAN-HU. 167. VILLAGE SHRINE AT NAN-HU, WITH SCHOOL-ROOM ON RIGHT. Chiang-ssiit-yeh in front of shrine. CH. LVI RUINS OF WALLED TOWN a7 carefully stamped clay in thin regular layers, was very solid, pointing to early construction. In places it still rose to eighteen feet, and it rested on a broad clay rampart raised at least another fifteen feet above the ground level. The area enclosed showed no recognizable ruins, only some low mounds partly covered with drift sand. I was able to get trenches cut through these by the large number of men we obtained from the hamlets. But I may state at once that the only finds rewarding the work here were a few large bricks of extremely hard burnt clay, evidently left over from some structure of which the materials had been completely quarried and removed long before. That these fine black bricks were of great age was on the face of it probable. But on this point I felt more assured when the careful search I made along the exposed portions of the rampart brought to light coins all belonging to issues of Han times. To the north and north-east of the ruined town I found an extensive area of the typical ‘Tati’ character, where the bare clay patches appearing between dunes big and small were abundantly covered with the usual hard débris of pottery, stones, etc. The people of Nan-hu called it ‘the place for finding old things,’ and, no doubt, searched it after great storms as keenly as Khotan ‘ treasure-seekers ’ their familiar Tatis. Repeated visits by Chiang and myself allowed us to collect here a good deal of bronze fragments, arrow-heads, small pieces of decorated stone-ware and the like. The latest of the numerous copper coins picked up proved to belong to issues of the T’ang dynasty, while on the other hand we failed to notice a single piece of porcelain by the side of such plentiful pottery. Thus the conclusion seems justified that the site was abandoned during the troubled period which followed the downfall of the T’ang rule in these parts, about the close of the eighth century _A.D., and before porcelain became common under the Sung. I cannot spare space to detail here a series of interest- ing observations as to the source from which this abandoned part of the oasis once received its irrigation, and as to the 78 TO THE NAN-HU OASIS CH, LVI physical changes which have since taken place here. That desiccation had played a main part in bringing about the present conditions was clear. But in addition I could convince myself also of the destructive effect which occasional great floods might have had upon irrigation in such a position. I ascertained that the water-supply of the ‘Tati’ area must have been derived from a river bed now completely dry which skirts the belt of scrub and drift sand fringing the oasis on the east and north-east. It must have been the action of exceptional rain floods from the lower hills on the south, such as the villagers remembered in recent years, which had gradually turned this river bed into a deep-cut cafion-like ‘ Yar.’ As such it passes close to the east of the old site, with some springs gathering in its marshy bottom fifty feet or so below the level of the Tati. The water here rising had, until about fourteen years before my visit, been utilized for a small colony which existed some three miles lower down in this valley; it had as it were taken the place of the large settlement abandoned since the T’ang times. But a big flood, said to have occurred in the summer of 1893, had swept away irrigation channels and homesteads, and buried the fields under coarse sand, On visiting the spot I could still see clearly the effects of this catastrophe in the ruined houses and uprooted arbours, while the bed of the irrigating stream had been scooped out into a steep-walled ‘ Yar,’ some twenty feet below the old level. What trees had been left standing were dead or dying, and gradually being cut down for fuel. Curiously enough, it was information about another effect of this big flood which helped to clear up the mystery about that strange gravel embankment we had noted in the desert on our way to Nan-hu. I found that the tower towards which we had then seen it continuing rose on the edge of the gravel plateau which overlooks from the east the wind-eroded old site or Tati. Between this and the tower lay the deep ‘ Yar’ or flood-bed just referred to. The tower was undoubtedly old in its solid clay portion, rising on a base about thirty-six feet square; but CH. LVI AN ANCIENT EMBANKMENT 79 plentiful restorations in small bricks showed that it had been kept in repair until recent times. Close by was a small domed structure which our Nan-hu guide declared to have been tenanted until some seventy years ago by a guard watching the road from Tun-huang. He was quite positive about travellers even in his own recollection having followed the line of the embankment right up to the tower, and thence struck across the river bed and the Tati towards the oasis. It was the flood of 1893 which, by cutting the bed into a sort of cafion, had made the direct road between the tower and Nan-hu impassable for carts, and had caused the diversion of the route southward. Clear traces of a cart track descending from the tower to the edge of the ‘ Yar’ and there suddenly ending, confirmed this statement. It thus became highly probable that the gravel embankment which the people of Nan-hu knew as the ‘ féng-chiang’ or ‘wind-wall,’ had, as indeed our old guide thought, been intended as a road-mark across the desert, useful at times of dust storms and as a protection against the winds from the north and north-east. But it seemed equally clear that a work of such magnitude, the construction of a dyke for over eight miles through the desert, would never have been undertaken except at a period when the oasis of Nan-hu and its population were far greater than at present. It was on the same day that I was able to investigate also those curious tumuli which had puzzled me on my first approach to Nan-hu. Going from the tower about a mile to the south-east we came upon the high brick-built gateway and adjoining rectangular enclosure then noticed. The flanking walls of the gate, about five feet thick and but little longer, still rose to fourteen feet or so, the material being coarse sun-dried bricks of a large size. The low lines of gravel enclosing the quadrangle were almost invisible while the sun stood high. The main _ tumulus which rose within, just facing the gate from the north, showed an annular shape and measured about eight yards across; it was of loose stones and coarse gravel heaped up to a height of about three feet. By cutting 80 TO THE NAN-HU OASIS — cw aw through it and smaller tumuli close by we ascertained that these were the only materials used, and that the little mounds rested on the undisturbed gravel surface. There could be no doubt that gate and tumuli were contemporary; but there was no clue to their origin and purpose. The people who built them were manifestly not Chinese nor of an advanced civilization. Was it possible to connect these modest cult relics with one or other of the small hill tribes, such as the J6 Ch’iang and the Little Yiieh-chih, to whom the Chinese Annals from Han times onwards make brief reference as dwelling along the slopes of the Altin-tagh west of Tun-huang, and whose grazing- grounds in those barren mountains and plateaus are now occupied by Mongol nomads ? I must refrain from touching here upon other points of historical and geographical interest which the observations gathered during this busy halt suggested in plenty. Nor can I discuss now the topographically important question, whether the ancient frontier station west of Tun-huang, which the Han Annals repeatedly mention under the name of ‘Yang-kuan,’ ‘The Yang Barrier,’ was really located at Nan-hu. The claim to this proud identification was put forward in a modern stone inscription, which some learned Tun-huang Mandarin of antiquarian tastes had set up ina small shrine between the south face of the ruined town and the lake. [ think there is a good deal of topographical evidence to support it. Whatever Nan-hu’s ancient fame may have been, all of us were bound to appreciate the physical comforts which our stay in the little oasis offered. The days were unusually calm, and with the minimum thermometer rising for the first time above freezing-point, there was a spring- like feeling of warmth in the air, though as yet I] looked vainly for a budding leaf or flower. Our hard-tried animals, too, felt refreshed, all but my enterprising little terrier who, while I was visiting a picturesque ruined temple above the gorge of the Nan-hu stream, picked up acquaintance with some half-wild shepherd dogs down below and absconded. After some hours of fruitless search he was recovered badly mauled. Having then to be cx.tvie POSITION OF ‘YANG BARRIER’ 81 kept chained up in my tent or carried on horseback until his wounds had quite healed, he had occasion to regret the results of his indiscriminating escapade. With his irre- pressible spirits and pluck he was indeed far safer in the lifeless desert. VOL. 1 G CHAP PERV val ANCIENT REMAINS FOR THE FUTURE I was sorry to leave Nan-hu; for the abundant traces of ancient occupation, the quaint peaceful ways of Chinese village life, the picturesque half-ruined temples, and most of all the delicious clear water of its springs, had invested the little oasis with a peculiar charm. But the ruins along the ancient wall in the desert north were calling, and | knew that the days or weeks available for their exploration before the fierce heat would set in were numbered. So I reluctantly fixed April 11th as the time for our start. Ten men was the maximum contingent which the oasis could spare without injury to the spring labour now fully in progress in its fields. Their houses were almost all within shouting distance of our camp, and orders had been issued the day before. Yet it was nearly noon before the men were collected by the sleepy village elder. Men turning up without rations or spades and newly hired camels without ropes to tie their loads, all helped to extend the usual delay attending a start inthese parts. The distance to be covered across the desert to the ancient wall by the Lop-nor route was too great for a single march, I had, | therefore, decided to move that day only as far as the : water of the Nan-hu springs reaches. : But even so far we were not destined to go. After the few warm days we had enjoyed in Nan-hu, a storm | was gathering. It broke with full violence from the north- | west just as the caravan had left the last fields of the oasis, . and was toiling up the steep sand-covered ridge which borders it northward. I had ridden ahead to the ruined. tower which crowns the ridge west of the picturesque 82 CH. LVI START IN SAND-STORM 83 gorge cut by the waters of Nan-hu, to get bearings for the plane-table. But scarcely had I reached it when the force of the gale enveloped us in a cloud of driving sand, which made it difficult to see farther than twenty to thirty yards, or even to keep one’s eyes open. It was a true ‘ Buran’ of the type with which I had become familiar in the Takla- makan during the spring of 1901; but from the difference of the ground it took a peculiar colouring. Had the soil here consisted of the fine loess dust which prevails throughout the Khotan and Keriya desert, there would have been absolute darkness around us; for the force of the wind was so great that it could have carried this along in clouds of great thickness. But with the heavy coarse sand which forms most of the dunes about Tun-huang the effect was different. Looking up to the sky only a yellow haze screened us from the sun, while along the ground there was swept a hail of small pebbles and sand grains. The sensation to one’s skin was distinctly more trying than that of the dust carried by a Turkestan storm. In order to gain some shelter we had to face the gale, and in spite of goggles and wraps I found it difficult to keep my eyes on the guide riding a few paces ahead. I was still wondering what kind of camping-ground awaited us for the night, when I noticed trees looming in front, and fine dust instead of pelting sand whirling around. ‘Two miles’ march in the teeth of the storm had brought us down to a level plain, and to a small outlying hamlet of Nan-hu, known as Shui-i, the existence of which had previously been carefully concealed from me. To march on in the thick haze of dust would have been awkward on account of the risk of men and beasts going astray, and when after an hour’s wait the storm showed no signs of ibating I reluctantly gave the order for halting. I had reason to feel grateful to Shui-i for the shelter it gave us hat evening. But the picture of the decay and squalor vhich its three farms presented still remains freshly mpressed on my mind. The larger of these dwellings where we had to seek ‘uarters struck me from the first as half a ruin, only await- ig the advancing sand to be finally abandoned and buried. 84 ANCIENT REMAINS FOR FUTURE cz. wo The farm-house had been built originally in a substantial style, with large rooms arranged in orthodox Chinese fashion on three sides of an oblong court facing south, Now it was tenanted by several families of small cultivators, The walls in more than one place leant over in a dangerous fashion, and were for the time kept from falling by sup- ports of roughly cut tree-trunks. Half the rooms had big holes in the roofing, the débris of timber and plaster which had fallen in filling the corners. Unspeakable litter was accumulated in the narrow court dividing the wings. It seemed too dirty a place even to my Turki followers for putting the ponies up. But there were plentiful tatters of coloured drawings and of inscriptions neatly penned on crimson paper decking the door-posts and half-broken window-screens, marks of former comfort and ease. It was not easy to get shelter here for my large party. But the driving dust outside and the howling gale made even the most critical among them settle down with contentment. I had just given orders for my tent to be pitched behind the court wall of what looked a completely ruined building near by, when Tila, my observant Yarkandi follower, dis- covered in it a tiny room still tenanted and retaining its roof. The oldest of the cultivators, a quiet, white-haired man, had retired there with a half-crazy son on whom he seemed charitably to bestow his chief care. The old fellow . looked eager to offer hospitality for the night; and when | he saw my man surveying suspiciously the bundles of old | clothes, etc., heaped up in a corner, he so promptly set : about to clear out his belongings and tidy up his lair that - I could not refuse so cordial a reception. The clouds of | dust raised by the sweeping up of the half-ruined hovel. were impressive even in this atmosphere of driving sand, | After a clearing such as, I am sure, no place in the hamlet will ever receive before the desert overwhelms it, I moved. in to relative comfort and shelter for the night. | I did not enjoy it long; for with some thirty-five miles. of desert separating us from our goal, I was anxious to start early. By 4 a.m. I awakened the men, but it was. close on 7 A.M. before the caravan with its contingent of Nan-hu labourers and camel-men not yet broken in CH. LVII HAMLET AWAITING RUIN 85 could be got to move off. In the meantime I was able to survey the surroundings better. The storm had ceased overnight, and only a light haze to the south remained to mark its passage. Subsequently I had many occasions to observe how much more transient than along the Takla- makan are the atmospheric effects of the storms which sweep the coarse.sands of the Tun-huang desert. I could now see plainly that not the buildings alone, but also the fields and arbours surrounding them, bore every mark of approaching abandonment. Close to the homestead we had occupied the fields were being overrun by light drift sand. They are still being cultivated; but irrigation fails to keep off the low dunes moving up from the west, which had already enveloped the feet of the trees of an avenue some 300 yards off, and threatened to choke the shallow channels bringing water to them. A small ruined shrine nearer to the main farm still showed its painted gateway. But the beams of the roof had fallen, and the drift sand caught within the walls had _almost completely smothered what remained of the clay images. | Elsewhere I could see fields overgrown with thorny ‘scrub, threshing-floors edged round by low dunes, or 'neatly-laid-out small orchards where the drift sand lay feet -deep along the fences, and the cuts needed for irrigation ‘were sadly neglected. An air of hopeless decay hovered ‘over the whole place, and my antiquarian imagination ‘found it easy to call up the picture it will present when the ‘desert shall have finally claimed it. Thus Dandan-oilik ‘or the Niya site may have looked during the last decades ‘preceding abandonment. I wondered to whose lot it will fall to excavate ‘the site’ which is now preparing here, ‘and what that archaeologist, say, of two thousand years ‘hence will make of the scraps of English or Indian writing which our stay over one night may have contributed to the rubbish heaps accumulated at Shui-i. From considera- tion for that confrére far off in the ages, I purposely refrained here from burning my waste paper ! Of course I did not lose the chance, with approaching ruin so plainly written upon this small settlement, of 86 ANCIENT REMAINS FOR FUTURE cu. wn obtaining definite local opinion as to the cause. The villagers whom I questioned with Chiang’s aid were ready to admit the far-advanced decay. But what they com- plained of was not want of water or uncertainty in its supply, but the difficulty of coping with the sand and the destruction caused by the troubles of the great Tungan rebellion. The day’s march offered unhoped-for opportuni- ties for studying the question, and proved in fact a most instructive antiquarian lesson. The people of Nan-hu had before stoutly denied any knowledge of the route northward and of ruins eventually to be met with. We were following the lively stream, about twenty feet broad and one foot deep, which with a current of about one and a half yards per second carries the drainage of the Nan-hu springs down into the desert ; when after about a mile and a half I came upon a group of deserted houses, not far from its east bank and encircled by dunes. The drift sand was nowhere more than six to eight feet high; yet the cut tree-trunks, and the dis- mantled condition of the ruins, showed that occupation had been definitely abandoned. Chiang-huan, a well-to-do villager of Nan-hu, whom I had engaged to look after the labéurers, now acknowledged that he knew quite well these deserted holdings of ‘ Upper Yen-chia’ and those of ‘Lower Yen-chia’ sighted some one and a half miles farther on. They had been abandoned, he said, in consequence of the desolation wrought by the Tungan inroad of 1866, when Nan-hu was sacked and the greatest portion of its population killed. Since then the» houses had furnished beams to the people of the oasis needing timber or dry fuel, and the trees once growing around them had been cut down for the same purpose. | Yet the stream flowing past seemed still to carry water quite sufficient to irrigate these long strips of old cultiva- tion. It was curious to note how the fine drift sand, - evidently eroded clay or loess, had accumulated over them. | It was retained probably by the trees, fences, and other obstacles, while to our left there stretched away the gravel Sai swept perfectly clear of all fertile soil. When abreast with the ruins of Hsia (Lower) Yen-chia, CH. LVII RUINS OF YEN-CHIA 87 we passed a narrow sheet of water nearly a mile long which now receives this drainage of Nan-hu, wasting itself in the desert. But we soon found that canals within living memory must have carried water much farther to the north. The first dwellings of another abandoned small settlement were met with at seven miles from Shui-i. Our guide called it Kuan-tsou, and stated that it had been abandoned some sixty years ago. I was able to test here the accuracy of local tradition. Around the first farm I visited there were a few patches of ground not covered by sand, and among the small débris scattered over them my men soon picked up modern-looking potsherds and porcelain frag- ments, also some coins belonging to the regnal epoch of Ch’ien-lung (1736-96 a.p.). The isolated farm-houses were filled with drift sand to a height of six or eight feet, and owing to the greater distance from Nan-hu still retained most of their timber. The trees once growing along the canals and irrigation cuts had all been cut down since cultivation was abandoned. But the trunks still emerging in a double row along what must have been the main channel, showed that its water was brought from the south-east, z.e. the now dry river bed I had traced east of the Tati forming part of the old Nan-hu oasis. Amidst the low dunes which had overrun what were once the fields of this hamlet, tamarisk cones had formed here and there up to a height of twelve feet. Everything showed that a typical ‘site’ was here in preparation to illustrate to posterity the conditions of Tun-huang village life early in the nineteenth century. The dwelling where I halted to take a photograph (Fig. 166) was more solidly built than the rest, and thus likely to attract the attention of some future archaeologist, say of 4000 a.p. So I could not forgo the temptation of depositing in a well-sheltered corner a dated ‘ Khat,’ in the shape of a newspaper, for his eventual guidance and edification. | For over two miles farther north ruins of detached holdings were met with at intervals, all belonging to the same period, as frequent coin-finds proved. The last wasa substantial homestead, with a thick refuse layer covering the courtyard (Fig. 168). A big dune, fully twenty feet 88 ANCIENT REMAINS FOR FUTURE cu. wn high, rising close on the north, had with its concave slope buried part of the building. The farms lay all in one line and had, no doubt, like the final off-shoots of the Tun-huang oasis which stretch finger-like northward, been fed by a single canal. Here recent progress of desiccation seemed clearly established; for the present water-supply from Nan-hu, even if united in a single channel, could scarcely be conducted so far. All trace of human occupation disappeared beyond on the bare gravel plain. The only things living were scanty tamarisk bushes growing in shallow depressions cut out apparently by rare floods from the mountains. But after we had travelled some twelve miles from Shui-i there rose in this barren plain a tower of stamped clay, much decayed, but still standing to a height of about nineteen feet. Far away to the north my binocular showed another. In con- struction this ‘ Pao-t’ai,’ or ‘Tun’ as the Tun-huang people called them, did not differ from those we had become familiar with along the ancient wall. But its position seemed a puzzle, until the subsequent discovery of the subsidiary Lzmes running due south-south-east from the fort of Yii-mén, and bearing just in the direction of this tower, helped to explain it. As we moved steadily on, a little to the west of north, the low but quaintly serrated hill range forming the eastern extension of the Kuruk-tagh rose clearer and clearer. To us who had seen it for days flanking our route from Lop-nor, it afforded assurance as to the relative proximity of the ancient Lzmes. But I could notice how our Chinese contingent, with the prospect of a camp in the waterless desert and no knowledge of the ground beyond, was getting fluttered and more and more straggling. So I detached Chiang-ssti-yeh and the Naik to form a sort of rear-guard, After a march of some twenty-four miles we struck a broad belt of tamarisk and other scrub; but our map showed that we were still at least twelve miles from the road which © skirts the marshes below Khara-nor. To reach it that evening with the tired caravan was out of the question. All I tried to get to was some reed-covered patch which | might afford grazing. But after another three or four miles 168, RUIN OF FARM ABANDONED DURING LAST CENTURY, IN DESERT NORTH OF NAN-HU. 169. RUINED WATCH-TOWER, T. XII., ON ANCIENT CHINESE LIMES, NORTH-WEST OF TUN-HUANG, We CH. LVII LOSS OF STRAGGLERS 89 spent in picking our way through the maze of tamarisk cones, darkness forced us to halt in the first thicket of Toghraks. Small channels, which looked as if cut by running water at no very distant period, here traversed the jungle in plenty. But of water, or of those reed patches which usually denote its presence not deep below the ground, there was none. For the men this mattered little; for in our ‘ Mussucks’ we had brought a plentiful supply ; but | was sorry for our ponies, which could not quench their thirst after a long and warm march. By 9g p.m. the Naik arrived and reported that he had brought in the last straggler, the man who had driven or rather dragged along our three refractory sheep. In the light of big bonfires which the men lit, I discovered that close to my tent were decayed huts dug out from the ground and covered with rough tree-trunks. No doubt herdsmen had once camped here, and water could not then have been far off. But how long ago was it? Here was an illustration of the doubts ever besetting the student of things primitive and devoid of chronology. Rest came only after midnight, and before daybreak I was aroused by the news that two labourers were missing. My honest secretary was greatly excited about it. He knew that the two men were confirmed opium-smokers, _and feared that, having strayed from our track in the dark- ness or lagged behind surreptitiously to indulge in a smoke, they would get hopelessly confused, and wander about without aim, to succumb at last to thirst. Vainly I represented how difficult it would be for men possessed of their senses not to see the light of our camp fires or to trace our track in daylight. While I resigned myself to the belief that the men had taken the first chance to decamp and were now moving back to Nan-hu, com- pensated by an unearned advance of money for whatever trouble they might have in their wandering, Chiang’s imagination saw the hapless men already lying dead in the jungle. In any case we had to clear up the matter of their dis- appearance, and if they were really lost to bring them 90 ANCIENT REMAINS FOR FUTURE on wz assistance. So Chiang-huan, the soz-disan¢t guide, was given a big gourd full of water and his pony a good drink, and then sent back to track and bring in the missing men if possible. No blame could attach to Chiang and Naik Ram Singh; for they were certain that no one had been left behind on our route. Nor was my own conscience burdened, seeing that a responsible rear-guard had been © appointed, and had done its work as well as a tramp in the darkness through desert and jungle permitted. Our march on April 13th was short, but not without further excitement. Judging by the survey carried along from Lop-nor (see inset 4 of Map 1.), we were only about seven miles in a straight line from the route then followed. For about four miles we made our way through the belt of thick tamarisk jungle, soon mingling with reed-beds, and found tracks of wild camels and deer in plenty. Curiously enough we also came upon traces of old wheel-marks in places where there were bare clay surfaces showing cracks as if baked by the sun after some great soaking. Then, cross- ing a narrow belt of gravel absolutely bare, we found our- selves in a depression filled with a salty marsh stretching away to the south-east. We had sighted before only a single tower, and this did not suffice to fix our position with certainty. Down below by the marsh edge we could not make out any definite landmark, and the marsh itself had for some reason connected with the configuration of the ground remained wholly unobserved when we first moved along the route to Tun-huang. In spite of the apparently hard salt crust covering great parts of it, the marsh proved quite impracticable for the animals, and threatened to cut us off for many a hot mile from the fresh water we eagerly wished to reach. But at last we found a place where the boggy soil would bear laden animals, and pushing up the gravel slope I arrived at what could now be easily recognized as the tower I had numbered T. xu. (Fig. 169). The Chinese of our party — rejoiced greatly when they found themselves on the well- worn cart-road, safely escaped as it were from the dreaded ‘Gobi.’ They looked still more pleased when, marching five miles west, we halted by the small reed-fringed lake, eive RETURN TO WESTERN LIMES QI our former camping-ground. Here was water for all in plenty and the springs as fresh as before. A few green shoots of young reeds were just showing on sheltered slopes of the marsh beds, the first sign of approaching spring in this desert region. When the east wind dropped in the early afternoon, I found it quite close in my little tent. It did not remain long where it was. Chiang, in his charitable thought for the missing labourers, had rather rashly started a great fire among the dry reed- beds, in the hope that the smoke and the light would guide them. As the conflagration was spreading I was obliged to order removal of camp to the nearest patch of bare gravel, my Muhammadans and Indians naturally resenting the trouble which Chiang’s care for two straying ‘ Khitai’ had cost them. His good intentions proved of no avail. The stragglers did not turn up, and when Chiang-huan rejoined us next day he reported having failed to find any trace of them. In spite of this worrying mischance I felt elated on being at last back again by the old frontier wall which our rapid passage from Lop-nor had revealed here and there. Only a few of the towers which mark its line had then been visited. Most of them could be sighted only miles away. _ However much we strained our eyes, the existence of a wall connecting these towers had necessarily remained a - conjecture. How glad I felt now for the chance of fully exploring this old Zzmes! Our discoveries since made along that section which continues it north-east of the Tun-huang ' oasis, had dispelled all possible doubts about the high antiquity of this frontier line. The hope seemed now ' justified that among the remains of a fortified border line, ' which I knew to stretch away for at least fifty miles, there ' were more relics waiting to be gathered. Yet little did I _ foresee how abundant the harvest would be. CHAPPERe EV FIRST EXCAVATIONS ALONG THE WESTERN LIMES Art first the extent of the line to be explored was far from giving assurance. The party of labourers I had managed to bring along, even when reinforced by the contingent from Nan-hu, looked disproportionately small for the task, and the loss of the two missing men had still further weakened it. I had every reason to husband my time; for the mid-day heat of the last few days showed me that work on this desolate border was bound before long to become very trying. It was important to make the most of my resources. So, while the Surveyor was to move westwards and trace the line of towers as far as it might extend from the bend previously noticed near Toghrak- bulak, Ts’ao Ta-lao-ye, the officer of the Tun-huang levies whom Lin Ta-jén had attached to my camp, was hastily despatched to headquarters to bring up more labourers and supplies. I thought it best myself to commence excavations on that part of the line which, owing to its distance from any suitable camping-ground, was likely to give most trouble. From Toghrak-bulak to our first lake camp the road had lain throughout over a gravel plateau destitute of water and fuel, and over these sixteen miles the old agger could — be traced almost unbroken. About half-way I had then noticed some posts protruding from the pebble-covered | slope of what looked like a small natural mound. Its ° position close to the inner side of the wall suggested, however, a ruin, and a little hurried digging had then | shown that some structure was buried beneath. It was | 92 cx. twn1 WELL-PRESERVED WATCH-TOWERS 93 to this ruin (T. vi.) that on April 14th I marched out my little band westwards. The heat of the previous days had roused a violent east wind, and the atmosphere was hazy with dust. Yet as we moved along the low ridge of gravel which hides the remains of the wall, and stretches away quite straight over this barren desert soil, | was surprised to note that the footprints left as I had ridden past more than a month before, looked absolutely fresh. The gravel and coarse sand on the surface was evidently little affected even by such a succession of gales as had blown across the valley since. It was clear that, in spite of all the force of the winds, erosion, that greatest foe of ancient remains in practically rainless regions, could not exert its destructive power on the flat surface of such ground and on what was buried beneath it. I thus ceased to wonder at the remark- able state of preservation which the first two towers on _this section of the wall line presented. The soil on which they stood had practically undergone no erosion, and since no undermining was possible, the winds of two thousand years had failed to shake down or seriously injure these heavy masses of brick and stamped clay. Up to thirty feet or so they still rose, built solid ona base of over twenty feet square and tapering towards their top (Fig. 170). This had once borne a conning-room or a platform protected by a parapet; but the brickwork of . the parapet had fallen, and the heavy timber of Toghrak which had been inserted to strengthen the top now lay bare. It was impossible to climb up; for these particular towers appear to have had no stairs, and the ladders or ropes which once may have given access had, of course, disappeared. On the east face of one of the towers | could still make out the holes in the brick-work which probably served as footholds. There were no remains of quarters or refuse indicating occupation near either of them. In order, probably, to command the ground better, these towers had been built on the very edge of tongues of the gravel-covered plateau, and little ravines had formed round them. If any structure less solidly built had ever adjoined them, its débris would inevitably have been 94 EXCAVATIONS ON WESTERN LIMES cx. wim washed down the steep slopes by either the winds or occasional rainfall. It was at the completely ruined tower—for such the mound already referred to (T. vu.) proved to be—that |] first obtained a clear idea of the quarters which seem to have been built by the side of most towers for the accom- modation of the soldiers keeping watch at these posts. The mound measured about forty-eight feet in diameter at the base and rose to ten feet above the ground. From the coarse gravel which covered its top and slopes, and gave it the appearance of a natural hillock, there emerged first masses of sun-dried bricks mixed with plentiful bundles of reeds. It was the débris of the tower, which in its fall had completely crushed and buried the walls and roof of the guard-rooms adjoining. To clear it away was heavy work for the men, and their own spades made little impression. Not being accustomed to the ‘ Ketmans’ of Turkestan, which, warned by previous experience, | had with no small trouble obtained at Tun-huang, they got little work out of these otherwise ideal implements of the excavator. But the greatest hindrance, perhaps, was the little doses of opium which most of these Nan-hu men used to take in the midst of their labour. However, at last we got at what remained of the walls of the structure buried by the fallen masonry, and successive finds of broken wooden implements stimulated the men’s hope of earning the liberal reward I had promised for the discovery of the first written record. We had just cleared a small outer room on the north side, and were working our way into a somewhat larger one built against the solid masonry of the tower (Fig. 171), when this eagerly-looked-for find was made. It was a strangely puzzling object,—a solid block of wood, about twelve inches long and five broad, thick at one end but narrowing wedge-like at the other, and painted black all over (comp. Fig. 172, 4). On one of the faces of this wedge there appeared two large Chinese characters in red. Chiang could read them without difficulty, but vainly sought for their sense. That they were meant for a name seemed the most likely conjecture. But how to interpret the 70. RUINED WATCH-TOWER, T. IX., ON ANCIENT CHINESE Z/MZES N.W. OF TUN-HUANG. 7 i 7 . “ . ae a ’ ‘ ; ¢ % c. + . my . ‘ R . ‘ ‘ ‘ if . , ‘ i a a : ‘ ° cxntvm FINDS IN RUINED QUARTERS 95 purpose of this queerly shaped ‘tablet’? The string still firmly fixed into the broad end showed plainly that this inscribed piece of wood was meant to be carried about or hung up. But it was only weeks later, when several other watch-stations along the wall had yielded exactly similar objects, that the true explanation was hit upon by my Chinese assistant. Elsewhere, too, these wedges had turned up singly, and marked with two characters which would give no proper sense except as names. The size varied, but there was always the string. So Chiang-sst-yeh remem- bered how he had seen at Lan-chou and other garrisons soldiers from small detachments, when off duty and per- mitted to ‘go to the Bazar,’ as we should say in India, carrying about conspicuous pieces of wood inscribed with the initials or name of their commandant as tokens of their ‘permit.’ Being large and easily distinguished by any passer-by, such a token would save the bearer any questions as to whether his absence from the post was authorized. If provided for each detachment only in a single specimen it would also prevent too numerous applications for leave, just as in a school where only one boy at a time is allowed to leave the class-room. This first find was soon followed by real records on wood: a large rectangular tablet with account entries, an inscribed seal-case, some broken ‘slips’ with the usual single line of characters. Their material left no doubt as to the early date when the tower was occupied. It was clear that this portion of the frontier line went back to the time of the Han dynasty, like the one explored eastwards. Con- clusive proof came to light next morning when, continuing the clearance of this room and of a sort of gate passage built against the east face of the tower, the men came upon more wooden slips of the orthodox shape, one of them dated in the third year of the Chii-shé period, corresponding to 8 a.p. With such evidence of high antiquity, all the relics left behind in these humble quarters by their last occupants acquired increased interest (Figs. 173, 174). There were plenty of quaintly carved wooden hooks, resembling the head and neck of some animal, 96 EXCAVATIONS ON WESTERN LIMES cu. wm with traces of bright yellow or red colour, which might possibly have served as pegs for hanging accoutrements or as handles for boxes; a block of wood for holding lighted tapers; curved pieces of wood which might have formed part of cross-bows or catapults, inscribed with the name of the regiment which had garrisoned this part of the | Limes; broken shafts of arrows, etc. ' That the men stationed here had, after the good Chinese fashion, used their spare time for homely occupa- tions was made clear by numerous wooden combs such as are still employed by rope-makers, by a wooden spindle- like instrument, and similar simple implements. A find, humble in appearance but of great archaeological value, was a foot-measure resembling in shape a bootmaker’s foot-rule, and still retaining the string by which it was hung from the wall (Fig. 173, 2). Divided into ten inches, with further subdivisions on the decimal principle, it gives the exact value of the measures in use under the Han dynasty. It consequently enables us to determine accu- rately the equivalents of measurements given for different objects in records of that period. An interesting instance in which I was able myself to apply the test of this ancient foot-rule will be mentioned hereafter. There were shreds of bright silk fabrics, perhaps left behind by officers or visitors of superior rank, and rags of coarse woollen stuff such as the soldiers might have worn. That luxuries were few and resources of civilized life carefully treasured was curiously illustrated by the pieces of several jars of hard grey pottery which had been broken, and then patched up again by means of leather thongs passed through neatly bored holes. Surrounded as we were by these modest but telling relics of the hard life once led along this much-exposed frontier, the briefest information to be gleaned from the wooden records, as they passed from under the labourers’ spades into Chiang-ssii-yeh’s hands, acquired a _ signifi- cance which those who wrote them nineteen hundred years ago certainly never dreamt of. Among our first finds was a label evidently once tied to a bag, referring to a hundred bronze arrow-heads and naming a certain I7I1. GUARD-ROOM BUILT AGAINST NORTH-EAST CORNER OF ANCIENT WATCH-TOWER T. VIIIL, TUN-HUANG LIMES, AFTER EXCAVATION. On left is seen the narrow gate leading into the quarters of the watch-station, with sockets to hold bars of door ; on adjoining wall surface the rough outline sketch ofa camel; on extreme right steps of stairs once leading to roof of quarters and thence to top of tower. es 172. ANCIENT POTTERY AND IMPLEMENTS EXCAVATED FROM RUINED WATCH-STATIONS ON TUN-HUANG ZIMES. Scale, one-fifth. 1. Wooden beating-stick. 2. Broomstick of reeds. 3. Iron hoe. 4. Wedge of wood inscribed with two Chinese characters. 5. Pottery jar. 6. Broken piece of pottery mended with leather thong passed | through holes. Senin —§ RELICS OF LIFE.ON ‘BORDER 97 company of ‘Yii-mén.’ So at last I had found the name of that famous ‘ Jade Gate’ which I had thought from the first was to be located somewhere along this westernmost part of the ZLzmes. Again and again in the course of subsequent excavations I felt grateful for the amor scribendt which seems to have prompted these ancient ‘military Babus’—like those whom one now meets in queer corners of the fortified posts scattered along the Indian North-West Frontier—to beguile their exwz and demonstrate their own importance by a constant flow of ‘memos,’ reports, store statements, and other documents so familiar to soldiering men in most regions. But here, as at other watch-stations, records with a pleasant touch of actuality and personal interest were not wanting. How strange it seemed to hear my secretary explain the record left on the four sides of a roughly carved wooden stick, telling of a visit which three persons named had intended to pay to their friend stationed here, perhaps the petty officer in charge of the post. Finding him ‘out’ they had left their ‘card,’ scribbling down their regret at a missed chat on the best substitute for orthodox ‘note wood’ they could pick up from the fuel store. No doubt, they left it in the hands of the men on guard; hence they did not think of putting down the date for our benefit. While Chiang delighted in scrutinizing the hand- writings, finding elegant penmanship here or execrably cursive ‘grass script’ there, I was gratified by a palaeo- graphic discovery of my own of considerable interest. Among the peculiarities of the wooden stationery used for the Kharoshthi documents which I had the good fortune first to unearth at the Niya site, the cleverly fastened oblong envelopes (Fig. 94) had always seemed to me a specially ingenious device. Without definite evidence, but guided by a number of general considerations, I had in Ancient Khotan ventured to advance the opinion that this device, with other equally clever arrangements in the form and fastening of those Kharoshthi letters, might have been originally derived from Chinese models. The discovery of a _ perfectly preserved wooden VOL. II H 98 EXCAVATIONS ON WESTERN LIMES cu. wm ‘envelope,’ about six inches by three, with the exact counterpart of the seal-socket and string-grooves familiar from those Niya finds, now placed my conjecture beyond all doubt. It is true, this particular ‘envelope’ had served to cover, not a tablet, but a box, as was proved by a small rim sunk on the under surface, and by the Chinese inscription in fine big characters, which indicated that the receptacle of which only this lid remained had been “the medicine case belonging to the Hsien-ming company.” But seeing the radical difference between Chinese writing arranged in vertical columns and Indian script for which oblong material, whether birch bark, palm leaf, leather, or wood, was the traditional and most convenient form, the turning of the lid shape to use for a covering tablet was but a kind of intelligent adaptation. In any case I had here the true prototype of my wooden ‘envelopes,’ but some three centuries older and used where everything else in the way of writing materials was purely and unmistakably Chinese. Subsequently, when clearing the great refuse heaps of T. xiv., I had the satisfaction of bringing to light Chinese records written on tablets exactly reproducing the shape of the Kharoshthi ‘wedges’ and dating in all probability from the first century B.c. Thus the Chinese origin of this kind of stationery, too, is placed beyond question. The selection of this particular ruin for my first excavation proved fortunate also in respect of certain structural features which the protection offered by the adjoining débris made it possible to ascertain. Among these I may mention here only the existence of regular stairs which led up to the roof from the principal guard-— room and thence probably to the top of the tower. Another feature was the abundant plastering and painting bestowed on the tower, probably with a view of making it better visible from a distance. On the east side we counted the | layers left by four or five plasterings and more than a. dozen coats of whitewash. The top coat bore the rough sketch of a camel (Fig. 171). As another curious detail I may mention that, on scraping the level ground south of the mound, we came upon two stout posts fixed in the p 1 1 3, 174. ANCIENT RUINED WATCH-STATIONS OF EARLY CHINESE BORDER LINE, ;1. Ornamented wooden tent-peg. 4 olished wood handle. 5. Ivory-topped head of baton. 95). 9. Bowl of lacquered wood. Wor xlen boot-last. Bronze arrow -head. EQUIPMENT, EXCAVATED MAINLY FROM TUN-HUANG. IMPLEMENTS AND ARTICLES OF Scale, one-fifth. 2. Foot-measure with decimal division into inches. 3. Wooden fire-stick. 6, 7. Spatulas. 8, 17. Carved wooden hooks (see 10. Hemp shoe. 11. Wooden key. Hee NOE Parts of wooden locks. 14. Fragment of calcined reed fascine (see p. 110). 15. Wooden socket for attaching seal. 18. Block of wood for holding tapers. 19. Wooden eating-bowl. 4 onwvin . A FORLORN CARAVAN 99 soil at twenty yards’ distance from each other, and upon remains of a thick rope of twisted reed which once joined them. It had manifestly been used for tethering horses after a fashion my own men often used when camping on absolutely bare ground. By the evening of the second day the excavation of the ruins at the post T. vi. had advanced sufficiently far for me to leave the completion of it to the care of Chiang and the Naik. All day a fierce gale from the east had been blowing, and eyes and throat suffered badly from the dust that rose from the dug-up quarters. It needed all the elation caused by the day’s epigraphical finds to bear these atmospheric conditions cheerfully, They became more or less constant thereafter, and only left us at times to give way to equally trying heat and glare. As I rode the seven miles back to camp, the barren gravel Sai looking more desolate than ever in the dust-laden twilight, I was met to my surprise by a long string of camels. Seen from a distance across the absolutely level plateau where all perspective deceives, they suggested a phantom column moving along the old wall. The season for travelling by the desert route to Lop- nor had now wellnigh passed. My surprise at meeting this belated caravan became still greater when it proved to be Sher Ali Khan’s venture to which I had entrusted my letters so busily written at Tun-huang. I had thought them now safely nearing Abdal, and my disappointment was naturally keen when I found that this mail-bag, to which I had devoted almost the whole of a cold night, had managed to cover in eleven days less than eighty miles out of its four months’ journey to Europe! The caravan men, a motley collection of Khotanliks, and people from Ak-su and Kashgar long exiled on the Kan-su border, crowded eagerly round me. It was a rueful tale they told of two valuable ponies, their only riding animals, which had strayed from a camp near the marshes to the east, and in spite of all search could not be recovered. None of the men, except the guide, had ever followed this desert track ; ind this worthy, upon whom they relied for a safe passage, was a young fellow who had first marched to Tun-huang 100 EXCAVATIONS ON WESTERN LIMES ca.tvwm as one of my donkey-men, and had proved the least intelligent of the lot. Naturally enough, the camel-men felt apprehensive of what awaited them farther on. Remembering how easy it would be to lose one’s way completely in that maze of clay terraces and dunes which intervenes between the terminal lake basin of the Su-lo Ho and Besh-toghrak, I congratulated myself at not being one of their party while dust storms of that day’s violence were blowing. The whole party looked so forlorn that I forbore to complain about the delay caused to my mail, and only gave them what advice seemed needed to keep the ‘guide’ to the right track. When they told me that they would halt a day or two at Toghrak-bulak to give their camels a rest, I regretted more than ever to have entrusted my mail to such a terribly slow goods train. I tried to exact compensation by asking the younger men to come back for that time to my ruin and help in the digging, ‘‘just to show those Khitai infidels how Mussul- mans could wield their Ketmans.”” But I was not surprised when even the offer of magnificent wages and the chance of finding hidden treasure did not tempt the way- worn Seven. So I let them pass on with all my good wishes for their own and my mail-bag’s safe journey. Two weeks later I found at Toghrak-bulak the carcass of one of their forty camels half devoured by wolves, and wondered how many more these luckless people would lose before reaching the green fields of Charklik. But they struggled through, and by the close of September my letters had safely arrived in England. | ’ , CHAR TE RaglLx RECONNAISSANCES ALONG THE ANCIENT WALL Wirth Chiang-ssit-yeh and Naik Ram Singh once initiated _in the work of clearing these ruined watch-stations, I was free to start on reconnaissance rides along the ancient _wall. They were to show me in advance the task awaiting us at each ruin, and to enable me to select the most suit- ‘able camping-places. The latter consideration was im- portant; for with so limited a number of labourers and with ruins so widely scattered, it would have been a serious loss to waste what little energy and strength the party _ possessed by long daily tramps to and fro. Never did I feel more the strange fascination of this desolate border line than during the days I spent in thus tracing the remains of wall and watch-stations over miles ‘and miles of gravel desert and past the salt marshes. _ There were, indeed, the towers to serve as guides from a one distance. But when on the east of our first lake camp | began to search for the wall they were intended to guard, _1 soon found my task complicated by peculiar topographical features. Already before, when first following the Lop-nor route, we had noticed lakes and marshes in the depressions north of it. But only when I set out to visit each ruined tower we had seen rising far away to the north over what then looked a uniform dead level of gravel desert, did it become clear how broken the ground was over which those engineers of the Han times had here carried their frontier line. What had seemed a plain extending to the very foot of those bare lifeless hills of the Kuruk-tagh now proved to be in reality a series of low gravel-covered plateaus IOI 102 ALONG THE ANCIENT WALL CH. LIX separated by many winding depressions. A large-scale map would be needed to show properly this intricate configura- tion of the ground, which resembled a strongly developed coast line with flat tongues of land left between a complex system of bays and inlets. The larger depressions were partly filled by spring-fed marshes, in places over a mile broad. Dense reed-beds fringed the sheets of open water, and salt-covered bogs extended farther away in the line of the drainage north-westwards. Elsewhere all trace of water had disappeared from the surface; but tamarisk bushes and other hardy scrub mingling with thin reed growth, as well as the salt-efflorescence, showed that sub- soil water was near in these Nullahs. The marshes and salt-encrusted bogs werequite impassable for our ponies, and often détours of miles round their edges or over strips of less treacherous ground were needed to take us from one tower or mound to another. Then, when these swamps had been successfully taken, like ditches in an obstacle race, came the still more exciting search for the remains of the old wall. This, I soon learned, had been carried unfailingly over every bit of firm ground capable of offering a passage for the enemy’s inroads and right down to the edge of the marshy inlets (Fig. 176). In fact, I convinced myself from ample evidence that this align- ment of the wall had been purposely chosen by the old Chinese engineers in order to supplement their line by natural defences, and thus to save labour of construction. Where the soil was soft and scrub-covered, as near the marshes, the eye often failed at first to discover any trace of the agger; for the remains of the rampart constructed with alternating layers of earth and reed fascines had here decayed badly owing to the moisture rising from the ground. The remains were obscured besides by the coarse vegeta- tion which finds nourishment in this salt-permeated soil. But when we had gained once more the bare gravel plateau, a search along its edges would soon reveal the familiar track of the wall. Over considerable stretches the wall still rose to a con- spicuous height, attracting the eye from afar (Fig. 175). Either some peculiarity in the constructive use of the o.ux TRACKING WALL ALIGNMENT 103 materials, always layers of gravel and fascines, had secured _ greater consistency, or the direction, coinciding with the prevailing winds, and a sheltered position on lower ground had reduced the force of erosion. Elsewhere, for some _ reason or other, the lapse of two thousand years and the - violence of the winds, which rarely cease sweeping along this great desert valley, had wrought far greater havoc, and it needed a careful scanning of the ground to discover _ the low continuous swelling along the line which the wall had followed. But even where the eye scarcely caught the alignment, the ends of the neatly laid reed bundles cropping out from below the gravel would supply a decisive indication; and a single stroke with the Ketman would _ suffice to unearth the regular ‘masonry.’ ‘Tila Bai was / usually my only companion on these reconnoitring rides, _ and grateful I felt for his keen eyes and power of intelligent observation which often enabled him to locate these faint traces of the wall from a distance. Once we had hit the line on a particular plateau section, it was easy to follow it right through; for straight it ran in the direction of the nearest watch-station eastwards. _Nor was it difficult to locate these towers, since their ‘position had invariably been chosen with a sharp eye for the advantages of ground commanding the nearest _ depressions. What had lightened the task of the soldiers _ who once kept watch and guard here, now proved equally helpful to guide us to their ruined quarters. However _much decayed some of the towers were, and however broad _ the marshy depressions which broke the continuity of the _wall and separated us from our next goal, the mass of broken masonry almost always sufficed for a guiding land- mark; so well raised above the general level of the _ plateaus was the ground which it occupied. Where the extent of wall line to be watched was great and the elevation afforded by natural features of the ground inadequate for the purpose, the towers had been built very massively to heights originally of twenty-five feet or more. Here the carefully set masonry or the hard clay stamped in regular layers was generally solid enough to hold out ‘against all vicissitudes of the ages. The original coating of 104 ALONG THE ANCIENT WALL CH. LIX thick plaster had, of course, fallen where not protected by the quarters which had been built against the foot of the tower, and often the erosive action of the winds had laid bare on the top the heavy Toghrak timber inserted to strengthen the masonry (Fig. 180). Where, however, the clay terrace or knoll selected for the watch-post assured by itself a commanding view, the towers had been built less high (Fig. 178). No doubt the reason is to be found in that intelligent aim at economy in efforts and means which is so characteristic a feature in all works of Chinese civiliza- tion. In cases where it was easy to provide access by means of a regular staircase, I found that the top was usually occupied not by a mere conning-platform, but by a small room affording better shelter to the men on guard. The walls of such little watch-rooms had necessarily decayed far more than the tower below; but their debris made access to the top still practicable. As I sat there with my eyes wandering over this vast expanse of equally desolate marsh and gravel Sai, which was relieved only here and there by a narrow streak of Toghrak jungle or glittering sheet of salt water, it seemed easy to call back the dreary lives which had once been lived here. The setting of the scene—of this I had ample proof—could have changed but little as far as human conditions were concerned. The very materials of which wall and towers were built proved that the ground over which the troops of the Han emperors had kept border-guard, consisted then as now of nothing but bare desert, marshes, and such dreary scrub and reed thickets as could find nourishment in their salty water. By contrast it seemed almost a pleasing picture when I raised my eyes to the long chain of barren brown hills which lined the horizon northward. Yet there, too, every- thing bore the impress of death-like torpor. Not a trace of vegetation survives on the detritus glacis sloping down to the wide desert valley, and the closely set ravines which furrow the bleak hill-sides looked as if scooped out by rain such as has failed to reach here for thousands of years. None of the valleys on this side of the Kuruk-tagh are now known to possess wells or springs. But there were mom KREULICS OF TIHESDISTANT (PAST 105 at least the fantastically broken crests of the ridges, the view of rocky pinnacles peeping out above them from the unknown wilderness behind, and the many shades of colour, from light brown to deep purple, to engage the eye and to relieve it from the dreary uniform grey of the Sai and the trying glitter of the salt marsh. No life of the | present was there to distract my thoughts of the past; not a sound in the air, nor a thing moving, but the hot air ' which vibrated above the ground and raised ill-defined ' wavy mirages on the horizon. In such solitude it needed no effort to realize the _ significance of every relic of the distant past when this desert border knew permanent occupation. Undisturbed by man or beast, or those far more destructive agents, _ moisture and driving sand, there lay at my feet the debris _ of the quarters which the guards had occupied, and often _ the more extensive rubbish heaps which had accumulated _ just outside. With the freezing gales which blow over _ this desert for half the year and the torrid heat which beats ' down on it for the rest, little wonder that the men stationed _ here did not feel tempted to move far away from their —_—- >= towers. So whatever they had no further use for found a safe resting-place in odd corners, or by the side of the tower and wall, to be recovered now with an ease such as | | had rarely before experienced in my archaeological hunts. The thinnest layer of gravel—and that, of course, _ the crumbling masonry supplied in plenty—sufficed to _ preserve in absolute freshness even such perishable objects ' as shreds of clothing, wooden tablets, arrow-shafts, straw, and chips. Whatever objects had once passed under this | protection were practically safe in a soil which had seen _ but extremely scanty rainfall for the last two thousand _ years, was far removed from any chance of irrigation or _ other interference by human agency, and had suffered on esse its flat surface but rarely even from wind erosion. Often a mere scraping of the slope with my boot-heel or the end of my hunting-crop sufficed to disclose where the detach- ‘ments holding the posts had been accustomed to throw - their refuse. With all the reports, statements, and enquiries which 106 ALONG THE ANCIENT WALL CH. LIX a fully developed and, no doubt, scribe-ridden military organization had kept moving along this chain of border watch-stations for more than two hundred years, was it wonderful that I soon grew accustomed to picking up records of the time of Christ or before, almost on the surface ? Of course the harvest could not be reaped until my working party was brought to each of the ruined posts in succession, and Chiang-ssti-yeh was at hand eager to read out and interpret these ‘waste papers’ as well as he could at first sight. Curious it was then to hear of records which told of apprehended attacks, movements of troops brought up for reinforcement, inspection visits of high officers, or more frequently of such routine details as issue of fresh rations, arms and clothing. Just as along the telegraph lines of our Indian border, isolated small posts try to make up for the total dearth of local interests by keeping a constant flow of news from the busy world far away tricltling over the wires as ‘service messages,’ so here, I thought, much of this correspondence was perhaps only a reproduction of orders meant originally for some headquarters and subsequently passed on to the lonely watch-stations. But truly important was the steadily growing assurance, gathered from the dates which my indefatigable secretary’s scholarly help allowed me to read with certainty, that this frontier line dated back to the end of the second century B.c., when Chinese expansion into Central Asia first began under the Emperor Wu-ti. As subsequently the date records receded farther and farther in antiquity, until we got to documents of the T’ien-han period commencing in 100 B.c., it became quite certain that the wall and the watch-stations along it were identical with the line of guard-houses planted at intervals, from Chiu-ch’iian (or Su-chou) to the Jade Gate. These the Han Annals record as having been constructed about 110 B.c., when attacks made by the Hsiung-nu on the Chinese political and commercial missions westwards forced the Emperor to despatch expeditions leading to the subjugation of Lou-lan and the establishment of Chinese military power in the » Tarim Basin. ‘ ? q cox.ux GUARDING AGAINST HUN RAIDS 107 There could be no doubt that the main purpose of this Limes was to safeguard the territory south of the Su-lo Ho. This was indispensable as a base and passage for the Chinese military forces and political missions sent to extend and consolidate imperial control in the Tarim Basin and beyond. It was equally clear that the enemy, against whose irruptions from the north and north-west this base and line of communication had to be protected, were the Hsiung- nu, the ancestors of those Huns who some centuries later watered their horses on the Danube and the Po. With this fact once established, how the horizon seemed to widen both in time and space! The very existence of this Lzmes brought out the important geographical fact that the desert hill region north of the Su-lo Ho marshes, now quite impracticable owing to the absence of water, must then have been passable for small raiding parties. In historical perspective, too, it was stirring to think that this western- most end of the ‘Great Wall’ had not been built for mere passive defence, a purpose so easily associated with every ‘Chinese Wall,’ but, like more than one Roman Limes within a century or so later, primarily to keep the route open for a vigorous strategic advance. But of such historical affinities and connections more anon. Fascinating as it was to let my thoughts wander far away to Roman borders I had known, it was easier still to forget altogether the lapse of long ages, while the humble accessories of the life once led on this desolate frontier lay _ before my eyes seemingly untouched by twenty centuries. _ The men, indeed, had passed away, those who kept guard and those against whose raids the great line had been drawn _ right through the desert. Yet nature had changed scarcely ' at all, and on this ground its forces had failed to efface the ; work of man. Never did I realize more deeply how little two thousand years mean where human activity is suspended, and even _ that of nature benumbed, than when on my long recon- Moitring rides the evenings found me alone amidst the _ débris of some commanding watch-station. Struck by the _ rays of the setting sun tower after tower far away, up to ten miles’ distance and more, could be seen glittering in a 108 ALONG THE ANCIENT WALL CH. LIX yellowish light as if the plaster coating of their walls were still intact to make them conspicuous. As they showed up from afar, with long stretches of the wall between them often clearly rising as straight brownish lines above the grey bare gravel desert (Fig. 175), how easy was it to imagine that towers and wall were still guarded, and that watchful eyes were scanning the deceptive plateaus and Nullahs northward with the keenness born of familiarity with a fleet and artful enemy ! The arrow-heads in bronze which I picked up in numbers near the wall and towers (Fig. 174, 16) were proof that attacks and alarms were familiar incidents on this border. Unconsciously my eye sought the scrub-covered ground flanking the salt marshes where Hun raiders might collect before making their rush in the twilight. How often had I amused myself on the Indian North-West Frontier with looking out for convenient lines of approach which our friends, Wazir or Afridi outlaws from across the border, might fancy! Once across the chain of posts the road lay open for Hun raiders to any part of the Tun- huang oasis or the settlements farther east. It is true the barren desert stretching north of the wall might have proved a far more formidable obstacle than the line of watch- stations itself. But did not those hardy horsemen sweep across great deserts almost as forbidding before they reached the Danube plains to become the scourge of the tottering Roman Empire? Just as the notion of time, so also the sense of distance, seemed in danger of being effaced when I thought how these same Huns, whom the Han emperors had struggled so long to keep away from their borders, were destined a few centuries later to shake the forces of Rome and Byzance. But the slanting rays of the setting sun would reveal also things of the past far more real. The line of the wall showed then quite distinctly for miles and miles, even where it had decayed to little more than a low long-stretched mound with reed bundles sticking out (Fig. 176). It was at that time that the eye most readily caught a curiously straight furrow-like line running parallel to the wall and at a distance of some thirty feet within wherever there was a (75. STRETCH OF ANCIENT BORDER-WALL, BUILT OF LAYERS OF REED FASCINES AND CLAY, EAST OF TOWER T. XIII., TUN-HUANG LIMES. a 176. REMAINS OF ANCIENT BORDER-WALL ADJOINING SALT MARSH, TO WEST OF TOWER T. XIV. A., TUN-HUANG LIMES, SEEN FROM SOUTH. = = OE —————E— <<< CH. LIX OLD TRACKS WORN INTO GRAVEL 109 well-preserved stretch of it. Close examination proved that it was a narrow but well-defined track, worn into the coarse gravel soil by the patrols who had tramped along here for centuries. Again and again this strange, uncanny track reappeared along wall sections miles away from the caravan route, wherever the remains of the wall were high enough to offer protection against the coarse sand and pebbles driven by the north and north-east winds. Never- theless I might have doubted this simple explanation, had I not had such abundant occasion to convince myself of the remarkable persistence with which this gravel soil retains all impressions such as footprints or wheel-tracks. Frequently I came across the latter running to depres- sions which may at one time have afforded some grazing or fuel, but where both these inducements to visits on the part of the cart-loving Tun-huang herdsmen must have disappeared for many years past. Yet the tracks left even by a single vehicle which had thus crossed the Sai were usually quite clear and continuous. Then elsewhere | noted with surprise that the footprints which we ourselves and our ponies had left on the ground when first tracing the wall on our journey to Tun-huang, looked two months later absolutely as fresh as if we had just passed by. We knew by sad experience the force of the gales which in the interval had blown almost daily over this desert valley. I have since ascertained that exactly corresponding ob- servations have been repeatedly reported by French and other travellers of experience from gravel areas of the Sahara. An equally striking proof of the extraordinarily pre- servative power of this desert soil and climate was sup- plied by an observation which at first puzzled me greatly. At a number of the watch-stations examined on my first reconnaissances I had noticed a series of queer little mounds, arranged in regular cross rows guzucunx fashion, wherever the ground adjoining the wall on its inner side afforded sufficient level space for such an arrangement. Closer examination revealed that these small structures, each about seven or eight feet square and up to seven feet in height, were built up entirely of regular reed fascines, laid IIO ALONG Pt & ANCIENT WALL CH. LIX crosswise in alternate layers, Intermixed with them was a slight sprinkling of coarse sand and gravel; but whether this was done on purpose, or merely a result of the layers having caught and retained the sand and small pebbles driven against them by exceptional gales, it was difficult to determine. Toghrak sticks driven vertically through the fascines were certainly intended to secure them when first stacked, No strengthening of this sort was any longer needed ; for through the action of the salts once contained in them and in the soil the reeds had acquired a quasi-petrified appearance and considerable consistency, though each reed, when detached, still showed flexible fibres. The regularity with which these strange stacks of antique Kumush were laid out near the watch-towers, usually at sixteen or seventeen yards’ distance from each other, made me think at first of their having served for some defensive purpose, like a zariba. With such a supposition it would have been possible to reconcile, perhaps, the evident fact that some of them had been burned, their position being marked by plentiful calcined fragments (Fig. 174, 14). But when I found subsequently that exactly similar structures were irregularly disposed over narrow ridges, where the ground near the towers was much cut up by ravines or otherwise restricted, this idea had to be abandoned. The true explanation presented itself when I noticed similar though not so accurately measured bundles of Toghrak branches heaped up in the same fashion near the south-west extension of the Lzmes, where such timber abounded and had been largely used in the wall con- struction. I then remembered that the dimensions of the neatly laid bundles, whether of reeds or branches, corresponded exactly to those of the fascines used in building the wall, and it dawned upon me that these queer mounds were nothing but stacks of the identical fascines kept ready at the posts for any urgent repairs in the wall. Thus breaches made in it could be quickly closed without having to collect and carry the required materials over a considerable distance. They at once reminded me then of the stacks of wooden sleepers seen neatly piled up at a eux FASCINES STACKED FOR USE Tuy railway station. If stacks of what after all is mere straw could without any special protection withstand the de- structive effect of two thousand years, the climate and conditions of the Tun-huang desert may well be credited with preservative qualities of an exceptional order. The use made successively of the stored materials would. account for the greatly varying height of the stacks from one to seven feet still extant at the same watch-station. But it still remained to explain why some of the stacks at different posts were found reduced by fire to calcined fragments. It was easy, of course, to think of wilful damage done by raiders and the like. But the most plausible explanation did not suggest itself until M. Chavannes’ translations showed me how frequent are the references to fire signals in the records from the watch-stations. No doubt such signals would ordinarily be lit on the top of the tower. But when time was pressing, or perhaps a particularly big fire was needed to penetrate a murky night, it would be simpler to set a whole stack on fire. The fact that the remains of burnt stacks were always found at points of the quincunx where the risk of igniting others was less, supports this inter- pretation. GHAR lsReal is DISCOVERIES BY THE ‘JADE GATE’ ] SHALL not attempt to describe day by day the labours which kept me busy for fully a month along this ancient Limes. Every watch- station we cleared furnished its quota of antiquarian spoil, often in novel forms. Even where my task was merely to trace the old wall across desert and marshes, there was an abundance of interesting observations to record about the changes, if any, which the ground had undergone since the line was first planned, No better gauge could have been designed for showing to the geographical student what physical conditions had prevailed here in Han times. With daily growing experience the reading of these marks of earlier water- level, of character and extent of vegetation, of wind direction, etc., soon became for me a fascinating study. That it claimed the attention of the antiquarian and geographer alike was the greatest attraction. Vividly do I remember all the peculiar features which this apparently dull and uniform desert ground offered along the hundred | miles or so of the border surveyed in the end, and equally | also the many little surprises and incidents to which the | search for the relics of a long-passed age treated us in the midst of this desolate region. But space does not suffice to record them all here, and in order to give some impression of the work effected and the results it has | yielded, I must restrict myself to a brief account of the | most notable finds. In order to be nearer to the reinforcements of labourers © and fresh supplies I had called up from Tun-huang, I | shifted my camp on April 17th to the vicinity of the» Il2 cx. tx DOCUMENTS IN UNKNOWN SCRIPT 113 small but well-preserved fort which I had passed before some twelve miles east of our first lake camp (Fig. 154). It lay conveniently central for a number of watch-stations to be explored; but it did not fall on the line of wall, and there was nothing exactly to indicate its antiquity or urpose. To the north-west stretched a broad marshy Nullah, fed by springs which supplied us with water. It was, alas! also a fertile breeding-place for mosquitoes and other insects, which now, as it grew warmer day by day, would issue in perfect clouds to make our evenings lively. The first important discovery which gladdened my heart while encamped here, came from one of the towers that guarded a section of the wall some four miles to the west. A number of wooden records in Chinese, among them two with exact dates corresponding to the year of .Christ’s birth and 20 a.p., had already emerged from the débris of some rooms adjoining this tower (T. x11. a) (Fig. 177), when I had to leave the work for a fresh recon- ‘Maissance eastwards. As usual, | had left Chiang and Naik Ram Singh behind to supervise the final clearing. My own ride that day showed me a great many promising ‘ruins; but still greater was my satisfaction with what my assistants brought back to camp when we met again in the evening. In a long narrow passage, scarcely two feet wide, left between the massive tower base and a decayed wall of the jwatchmen’s quarters, had been found a thick layer of rubbish, mostly stable refuse. From this emerged one small roll after another of neatly folded paper containing what was manifestly some Western writing. A few of these letters—for as such they could easily. be recognized from their folding and tying—had been found wrapped up in silk, while others were merely fastened with string, None of them, of course, had as yet betn opened, but a glance at the partly legible writing on the outside of some of the documents showed me the same unknown script ‘esembling early Aramaic which I had first come across on that piece of paper from the Lop-nor site. The paper dere was exceedingly thin and brittle; but when at last - had succeeded in unfolding one roll, there emerged a VOL. II I 114 DISCOVERIES BY THE ‘JADE GATE cam complete document neatly written in bold black characters, and measuring about fifteen by nine and a half inches. I could not attempt a decipherment, nor more than con- jectures about the language in which this and the other ten as yet unopened documents were written. But in the very fact of this Semitic writing turning up on the border of China and in the material used for it, there was enough to keep my thoughts busy. Were these papers perhaps in some Iranian tongue, and were they left behind by some early traders from Sogdiana, or still farther west, who had come for the silk of the distant Seres while the route was kept open for direct trade from China to the regions on the Jaxartes and Oxus? How had they found their way into the rubbish heap of a lonely watch-station far removed from the actual Lop-nor route? No less curious was the chronological puzzle. From the position in which these papers had been found, close to Chinese records on wood, it appeared highly probable that they must have found their way into the rubbish heap approximately at the same period. Now, among the tablets, over a dozen in number, the two exactly dated ones belonged, as already stated, to the years 1 and 204.D.; among the rest there were several which by the dynastic style used in the designation of a certain military body proved clearly to date from the reign of the usurper Wang Mang (9-23 A.D.). The early form of Aramaic script presented by the documents would agree well enough with such a dating. Yet how to account for the material on which they were written, considering that the first discovery of paper in China is attributed by reliable historical texts to the year 102 A.D.? An explanation might possibly be found, | thought, in the references which M. Chavannes has un- earthed in early Chinese texts to ‘silk paper,’ introduced some time before the manufacture of real paper from rags and bark was invented. Curiously enough, as if to illustrate in a palpable fashion what these texts record of the use of silk fabrics as a still earlier writing material, the same refuse layer had furnished a small strip of cream-coloured silk inscribed with a fragmentary line in Kharoshthi. It ‘SHNWIT ONVOH-NOL ‘SUIVLS GNV SYALYVNO ONINIO[GV AO SNIVNAA HLIM “VIX ‘LL ‘WAMOL-HOLVM GHNIONW °ZZ1 ing Sa EP Me Caen pyr yr oe = de) iP rs ay cuwtx ANCIENT SOGDIAN LANGUAGE FS seemed as if three civilizations from the East, West, and South had combined to leave their written traces at this lonely watch-station in the desert, and with them to demonstrate also the earliest writing materials. Expert investigation effected by learned collaborators since my return to Europe has singularly confirmed these conjectures. Professor J. von Wiesner, the chief authority on plant physiology as connected with the history of paper manufacture, has proved by a microscopical analysis of the paper that the material used for those documents represents the earliest effort so far known at producing rag paper. I had reason to feel equally gratified after the publication of one of the documents by my learned friend Dr. A. Cowley, and its subsequent analysis by M. R. Gauthiot, an accom- plished young Iranian scholar. These furnished conclusive evidence that the script was, indeed, of Aramaic origin, and the language an early form of that Iranian dialect which was spoken in ancient Sogdiana (the region of the present Samarkand and Bokhara), and of which the Sogdian manuscripts recently recovered from Turfan and Tun-huang have preserved us later specimens. The documents can be clearly recognized now as letters, and complete decipher- ment may reasonably be hoped for as a result of further researches. The old fort (T. xiv.) near which my camp stood, has already been briefly described in connection with my first passage along this route. With its thick walls of stamped clay it was an imposing structure to behold (Fig. 154). But vainly did I search within it and along its walls for any definite indication of its age. Not even refuse was to be found inside, a curious fact, seeing how grateful travellers would feel for the shelter it offers against cutting winds. My men, too, had shunned the place; for, rightly =nough, they suspected that it would swarm with those erribly aggressive little insect fiends, the ‘Tsao-p’i,’ as he Chinese called them, from which we had trouble ‘nough to escape even in the open. This absence of iny mark of ancient occupation at a ruin occupying so onvenient a position by the high road to Lop-nor puzzled ne greatly, and as soon as my reconnaissances farther 116 DISCOVERIES BY THE ‘JADE GATE’ cu. 1x afield were concluded, I set about to search the ground systematically. It was not long before I discovered fragments of that hard, dark grey pottery with which my work at other watch- stations had made me familiar, at the foot of a hillock rising less than a hundred yards to the north of the fort (Fig. 179). While Chiang and the Naik were engaged at the nearest watch-tower to the west, I had retained with me the least indolent of our men for prospecting; and as I made him scrape the slope of the mound at different points, layers of straw and other stable refuse came to light quickly at more than one point below the cover of gravel. I was con- ducting this first experimental search near the top of the west slope when his spade laid bare a vertical cutting into the hard clay composing the hillock. | It proved to be the mouth of a little tunnel about two and a half feet broad and about as high, running into the mound and filled with drift sand and refuse. Before I could form any view about its purpose, two dozen Chinese slips had emerged among pieces of blank stationery, mat- ting, bones, and similar rubbish. Soon my digger had burrowed out of sight while clearing the tunnel. After making his way in for some ten feet, he reported that it led to a room completely filled with sand. No further work was possible here until the other men came back from their day’s task. But the spirits of my own ‘ prospector’ had been roused by a liberal reward for his discovery, and quickly he set about scraping elsewhere. Not far from the tunnel, but lower down on the slope, he unearthed a platform cut into the soft rock, and here another score of tablets turned up in excellent preservation. Of course I had sent word to Chiang-ssti-yeh, and when he arrived from the tower then ‘in hand’ two miles off, an eager scanning of the last finds began. Many of the narrow slips of wood were covered with minute but well-written characters in several columns, and great was my joy when the dated pieces among those unearthed on the platform proved to belong to the period 48-45 B.c. So at last we had got well back into the pre- Christian era. All the dated records found in that queer REMAINS OF ANCIENT WATCH-TOWER, T. XX., OVERLOOKING LAKE WEST OF KHARA-NOR, TUN-HUANG LIMES. 79. HILLOCK WITH REMAINS MARKING POSITION OF ANCIENT ‘ JADE GATE” STATION, NEAR FORT T. XIV., TUN-HUANG LIMES. CH. LX RECORDS FOUND NEAR FORT 117 tunnel belonged to the first two decades after Christ, the reign of Wang Mang the Usurper. In both sets the docu- ments seemed to have been addressed to some superior officer. The presumption was that here we had struck the site of some sectional headquarters for this part of the Limes. It received support when I learned that at least one of the records certainly emanated from the general officer commanding at Tun-huang, while several others contained reports or orders to superior officers holding charge of ‘the barrier of Yii-mén.’ All along I had surmised that this ancient frontier station of the ‘ Jade Gate’ must have been located during Han times at some point along the line we were exploring. But where was its exact position, and had this always been the same? It was impossible as yet to ‘arrive at any definite answer. But with the documents before me | now began to realize what advantages this site offered for a chief watch-station commanding the ancient route west- wards, and how little reason there was for doubts about the antiquity and the purpose of the ruined fort at this point. Well withdrawn behind the protective line of wall and towers, and defended to the north-west and south-east by impassable marshes, its position was admirably adapted to serve both as a poznt ad apput for the posts along this portion of the Lzmes and as a station controlling traffic along the main road. Only the day before I had discovered that the fort lay exactly in the line of a secondary wall running due south from the main wall in the direction of Nan-hu, and, though badly decayed, still traceable on the Sai. This secondary wall showed exactly the same manner of con- struction with alternate layers of gravel and fascines, but only a thickness of a little under five feet. It seemed clear from the first that this transverse wall was a subsequent work ; for without the main wall beyond, it would have been quite easy to turn it from the north flank. But what had been the true object of this cross wall ? I had first thought of an inner line built, not for de- fence, but for purposes of police control in order to prevent the unauthorized entrance of individual travellers into Chinese territory. But evidence subsequently accumulating 128 DISCOVERIES) BY ‘THE? JADE GATE” cua revealed the important chronological fact that, while east of this transverse line the records brought to light proved continuous occupation of the watch-stations, down to about the middle of the second century a.p., the records found west of it stopped short with the reign of Wang Mang, and in the case of the more remote stations with dates considerably earlier. I was thus gradually led to the conclusion, that in the early decades of our era, during the troubled times of Wang Mang’s reign or very soon after, there took place a retrenchment of the border line lying westwards. By abandoning the outlying portion of the wall an appreciable reduction was, no doubt, made in the difficulties about victualling, etc., which must always have been felt most in the case of those detachments pushed out far into the desert. At the same time this retrogression of the guarded frontier line would have been fully in keeping with a contemporary change in Chinese policy. This we know was then no longer concerned about imperial expan- sion westwards, but until the last quarter of the first century A.D. kept strictly on the defensive. It was for the purpose of replacing the flank protection which the lopped-off western end of the original Lzmes had offered, that I believe the transverse wall to have been built during, or soon after, Wang Mang’s usurpation. In any case it was easy to realize that the station at which the great caravan road passed through this wall must have been a point of importance. In fact, from the reasons above indicated and supplementary evidence which cannot be set forth here in detail, I soon felt convinced that the ‘Jade Gate’ of the Later Han period, roughly corresponding to the first two centuries of our era, had to be located at this site. So I was most eager to have the remains on that unpretentious hillock cleared with all expedition and thoroughness. But the day was oppres- | sively hot; and in the afternoon a violent gale, sweeping down from the north-east put a stop to further work, — the temperature inside my tent rising to ninety degrees — Fahrenheit. | But on the morning of April 21st I was able to start CH. LX SECONDARY LINE OF WALL 11g systematic excavation with the whole of my little band of diggers. A couple of days earlier it had been unexpectedly strengthened by the two lost Nan-hu labourers, who to Chiang’s and my own great relief turned up by the route from the east, looking very woe-begone, but as sound as their nature and opium would allow. As far as Chiang could make out their tangled story, they had fallen asleep after a little ‘smoke’ of their beloved drug, mistaken the track when they woke up, and then aimlessly strayed in the desert until, after two days’ wandering without water, they were guided to the caravan track by the smoke from the camp fire of the herdsmen we had met on our first approach to Khara-nor. At a later season—and even now without the sustaining effect of their opium—these hapless fellows would almost certainly have perished. So nothing worse befell them than that the herdsmen, who rightly suspected desertion, had forced them to rejoin us. The hillock we had to clear measured some eighty _ yards from east to west and nearly as much across, and there was nothing on the gravel-strewn slopes to show _ where to search for rubbish and ancient remains. So parallel trenches had to be dug all along the slopes down _ to the natural hard clay in order to make sure that nothing at this important point should escape us. There was _ plenty of work here for the men, and it took them fully three days to complete it, though on the very first there arrived a most opportune reinforcement in the shape of _ twelve additional labourers whom Ts’ao Ta-lao-ye, Lin | Ta-jén’s petty officer, had managed to bring up from Tun- huang along with half a month’s fresh supplies. What _ with all the digging effected by the men—-whom small but | prompt rewards for interesting finds kept up to the mark— the little hillock soon suggested a kopje girt with shelter _ trenches against modern gun-fire. The results were ample _ and offered strange surprises. | p One of these was provided by the narrow tunnel on the north-west slope, in which we first discovered that _ batch of wooden records from Wang Mang’s reign. For instead of forming a window to some _ subterraneous chamber, as I had at first suspected, it proved to be the 120 DISCOVERIES BY. THE #jJADEIGATE aaa only access to a well or shaft, about five or six feet square, which we cleared to a depth of over twelve feet without reaching the bottom. The earth roof of the shaft had originally been supported by timber, but now fell in, luckily without smothering any one. Dozens of wooden documents turned up in the sand cleared out, almost all, alas! badly decayed through damp, but some still showing legible dates of the first decades after Christ ; and from these I concluded that this curious shaft had been filled up with refuse not very long after that period. Its original purpose seemed obscure, until Chiang and some of my Muhammadans rightly suggested that it must have been intended for a dungeon, the use of similar wells for the safe keeping of dangerous prisoners being still remembered in Chinese Turkestan. In fact, | subsequently ascertained that such methods of burying prisoners as it were alive subsisted in the Central- Asian Khanats, too, until the time of the Russian conquest. The narrow side opening near the top of the well had, no doubt, served as an air-hole. I did not care at the time to think much of the horrors which this dungeon might have witnessed; but the fact that one of the inscribed slips here recovered, contains, according to M. Chavannes’ translation, an order about the burial of a man who had died after having been beaten, has since helped to recall them. Curiously enough, the well-preserved ancient beating-stick of the traditional shape, shown in Fig. 172, was found on this very mound. Of the structures which the top of the hillock had once borne, nothing but the scantiest foundations were dis- covered. But deposits of ancient rubbish laid bare at different points of the slopes yielded records on wood in abundance. From one refuse area near the centre close ~ on five dozen documents were recovered. As almost all the dated pieces belonged to the years 96-94 B.c., it became quite certain that the occupation of this site went back to the time when the Zzmes was first established. That the station had then already been one of import- — ance was proved byseveral documents emanating direct from _ the commandant of the ‘Jade Gate Barrier.’ Others too — cH. LX FIND AT BUDDHIST SHRINE 12! possess a distinct historical interest, eg. one directing the issue of provisions to the escort of an imperial envoy to So-ch’é, or Yarkand, counting eighty-seven men. In another again there is a reference to an ambassador sent to the imperial court by a chief of the great tribe of the Wuz-sun, who played an important part in the early history of Central Asia and then dwelt in the mountains of the present Farghana. But of such records I shall have occasion to give some account later, when dealing with the results of M. Chavannes’ decipherment. A very curious find, but for the topographical facts explaining the peculiar character of this site, might have puzzled me at the time. It was a considerable number of large paper fragments inscribed with elegantly penned Chinese characters. They turned up from the floor of what I soon recognized to have been a small cella, about ten feet square, on the western side of the hillock. That the texts they contained belonged to Chinese translations of Buddhist canonical works | could with Chiang’s help make sure of at once, as well as of the reference made in another fragment to the dedication of some Buddhist images. So, taking into account the evidence furnished by the use of paper and the style of the writing, I concluded that these were relics from some modest shrine which had somehow survived at this otherwise long-abandoned site down to T’ang times. Other relics of it were ex-votos in the shape of small miniature flags made up of silk rags such as I had found at ancient sites excavated on my first journey. Next morning this conjecture received striking con- firmation by the discovery of a small wooden bowl embedded below the reed flooring and containing some eighty copper coins. Excepting two which were older, all ' were issues of the regnal period 713-742 a.p., and showed scarcely any wear resulting from circulation. So the _ attribution of the Buddhist text fragments to the eighth century, when the route to Lop-nor had last been an im- portant line of communication, became practically certain. _ One of the fragments has since been proved by M. _ Chavannes to belong to a Buddhist canonical text, which Hsiian-tsang himself is known to have translated between 122 DISCOVERIES BY THE ‘JADE ‘GATE Gia 645 and 664 a.p. A fitting tribute this to the memory of the pious traveller; for I felt quite sure at the time that my patron saint on his return journey must have passed within a few yards of the débris-covered hillock. But what was the reason for this strange survival of worship at a site which must then for long centuries have been in ruins? The explanation was not far to seek. A number of considerations, as already stated, had gradually led me to the conclusion that the fort and the débris-strewn mound of T. xiv. marked the position of the ‘ Jade Gate’ during the first two centuries a.D. When the danger of raids disappeared with the migration of the Huns west- wards, and later on Chinese control over the Tarim Basin was lost for centuries, the whole line of the Lzmes west of Tun-huang was abandoned to the wilderness. But occasional caravans, as we know from Fa-hsien’s travels, still continued to use the desert route, and, no doubt, wandering herdsmen and hunters still visited the grazing along the lakes and marshes of the terminal course of the Su-lo Ho. So local worship had a chance of proving once more its tenacity. It clung to the site where those leaving the ‘Jade Gate’ of the ‘Great Wall’ for the difficult desert journey had of old been accustomed to put up ex-votos at the border shrine and pray for a safe return, just as Chinese travellers still do at Chia-yii-kuan, the modern equivalent of the ‘Jade Gate.’ What more conclusive proof of that tradition having survived to the present day could I have wished for than that presented by a small modern shrine which lay in ruins only a little over a hundred yards to the west of the hillock explored? The coarsely made clay images were all broken, perhaps the act of some truculent Tungans, and the roof and all woodwork had disappeared. But there was evidence of herdsmen still offering prayers at the ruin, and my own labourers from superstitious fear could not be induced to dig even near it. I was not altogether sorry for this, since the chance of finding any- thing of interest under the modern ruin seemed remote, and because this continued veneration was in itself an interesting archaeological asset. | GO APER! EX] THE GREAT MAGAZINE OF THE LIMES I cANNOT stop to describe here various interesting minor finds which rewarded our laborious clearing of all the layers of ancient débris and refuse on the hillock once occupied by the Gate station. There were plentiful rags of cast-off clothing ; remains of iron implements; well-made shoes in hemp and other stout materials ; fragments of lacquer bowls showing tastefully designed scroll ornamentation in black on red ground—all relics which, insignificant in them- selves, helped me to picture the life once witnessed by this important border post (Figs. 172, 174). A reconnaissance _ made from this point while the digging was still in progress _ had revealed extensive rubbish layers at a point about two and a half miles northward and at a short distance within _the main line of the Zzmes; and thither I moved camp on the morning of April 24th. | At first sight there was nothing to attract attention to _ the spot, and without Tila Bai’s keen eyes, which noticed a slight swelling on the edge of a bare plateau tongue, I _ should probably have passed it without heeding. The gravel-strewn little mound, only about two or three feet _ high and less than forty feet across, proved to contain the _ débris from some brick-built structure, too much decayed _and too scanty for any determination of its original char- _ acter. The dozen and a half of records on wood which we _ found in the débris ranged in date from 65 to 137 a.pD. At _ the time I noted this indication of relatively late occupa- tion with special interest; for the ruin lay just within the _ corner where the line of the secondary wall above mentioned _ would join the main wall, if continued across an impassable marsh in its direction from south to north. 123 124 GREAT MAGAZINE OF THE LIMES ci.ux But that this point must have been occupied far earlier by a post of some importance soon became evident when my search revealed extensive refuse layers not only on the slopes below the débris mound but also on others farther away. As inscribed pieces of wood recovered from the latter ranged from 61 B.c. onwards, it seems probable that this rubbish had been thrown out from some structures which existed earlier and have completely disappeared. These various refuse deposits yielded so rich a harvest of Chinese documents that from the first I could not doubt that they marked the position of an important post, probably the quarters of some sectional commandant. M. Chavannes’ analysis and translation of the records recovered from this post (T. xv. A), altogether over 160 pieces, have fully confirmed this conclusion. They have also revealed the existence of so large a number of specially interesting pieces that I cannot refrain here from at least alluding to a few of them. Foremost in importance from the philological point of view is a_ beautifully written and perfectly preserved triangular tablet about fourteen inches long, which contains the first chapter of a lexicographical work, the Chi-chiu- chang, famous in Chinese literature and the subject of much learned commentary. As it is known to have been composed between 48 and 33 B.c., and as the tablet dis- covered by me must have been written within a century or two later, the critical importance which it possesses for the history of the traditional text can be easily understood even by the non-Sinologist. In any case M. Chavannes, in a preliminary notice communicated to the Académie des Inscriptions, has declared this tablet to be the earliest authentic specimen of a Chinese literary text. The par- ticular work has always been much in use among Chinese students; and the discovery of pieces from it, not here alone, but at other Zzmes stations, proves that the studious habits of the race were represented even among the soldiers exiled on this desolate border. | Another very curious find consisted of a batch of eleven neatly written bamboo slips, scattered but clearly marked by their identical shape and writing as originally belonging cu. x1 RECORDS OF SECTIONAL COMMAND 125 to a sort of medical note-book (Fig. 119, 8). Some of them contain ‘case records’ and prescriptions for particular patients, others general recipes for men or ailing animals, mostly with special mention of the physicians with whom they originated. Asa sample, | may mention the record left by the medical attendant ‘ Mr. An-kuo’ of his twentieth consultation in the sad case of ‘Mr. An Tien-hui,’ who, having been benumbed with cold and in consequence tumbled out of his car, had injured himself internally, and who even after thirty days of treatment still suffered in his chest and extremities. Of the great mass of records, which comprise chiefly reports, orders, and miscellaneous memos of military administration along the ZLzmes, it is impossible to mention here more than a few. Thus, in a ‘circular to the posts of the Yii-mén Barrier,’ a sectional commander regrets to acknowledge the absence of certain soldiers at the time of the official inspection, and gives due warn- ing of the punishments to be awarded in such cases. The ‘I-tsou Company’ here specified appears to have garrisoned this post right through the period covered by the records. Elsewhere we hear of difficulties about effective signalling, the distribution of duties among the men at the actual watch-posts, and such like. A very interesting find, the archaeological importance of which has only been realized since M. Chavannes’ interpretation, was a narrow strip of strongly woven cream-coloured silk bearing a line of Chinese characters inked in. This states precisely the length, weight, and price of a bale of silk, from the edge of which it had been torn off. The name of the place of manufacture, Yen-ch’éng, a locality in Shan-tung, serves to fix the date of its production at the close of the first, or in the early part of the second, century a.p. But what I greeted with particular interest is the statement there made of the width of the silk piece, viz. two feet and two inches. We know that the Chinese foot, with its decimal division into ten inches, has varied very considerably under succeeding dynasties. But when at the British Museum I came to measure up this very strip of silk with the bootmaker’s 126 GREAT MAGAZINE OF THE LIMES ca. ux foot-measure discovered at the watch-station T. vim. (Fig. 173, 2), the original width of the piece proved exactly two feet and two inches according to the standard of the Han period. ) I next applied my ancient measure to the small bale of | silk which, as related in Chapter xxxu., I had unearthed at the Lop-nor site, and to a piece of silk found in one of the ancient ruins at Miran. The result tallied again as closely as is possible in the case of fabrics which in the course of seventeen centuries were bound to shrink a trifle. On the other hand, if measured by the foot-measure found at the Lop-nor site and manifestly belonging to the Chin period (3rd-4th century a.pD.), all these pieces would show a width of czvc. one foot and nine inches. Thus, by the evidence of actual silk remains brought to light from three widely distant sites, we can now establish that the silk exported from China to Central Asia and thence to the classical West during the centuries immediately before and after Christ retained a uniform width, corresponding approximately to one foot ten inches British measure, while the measuring standard in China underwent a considerable alteration during that period. But this end strip was not the only silk remnant of antiquarian interest preserved in the refuse layers of the station T. xv. a. It was there that I discovered a small silken envelope a couple of inches long, which must once have held a letter written on a rolled strip of silk and on which M. Chavannes has been able to read the address, Equally interesting from another point of view is a narrow strip of silk bearing a long line in Indian Brahmi characters of a type associated with the rule of the Indo-Scythian or Kushana emperors. It has not yet been deciphered; so we cannot conjecture how this easternmost specimen of a document in true Indian script found its way to the ‘Great Wall.’ Ofa tablet showing ‘unknown’ characters by the side of Chinese writing I may make passing mention, as a further indication of the polyglot traffic which is likely to have been brought to the Lzmes by the ancient caravan route. Among the miscellaneous finds of ancient rags, meer WIL OF ANCIENT SILK BALES 127 broken utensils, old boots, etc., which had survived in the refuse practically as fresh as when they were thrown down there, I can single out only one for special refer- ence. It was a small closely tied bundle containing the broken pieces of a feathered arrow with the barbed bronze arrow-head packed away amongst them. The most likely explanation was that, in accordance with a system still in vogue in certain military departments over-anxious to check petty defalcation, broken arrows had to be returned ‘into store’ before new ones could be issued. Our rich haul at this station was completed by April 25th, a perfectly clear day, when after long hours of work _ in a blazing sun I enjoyed in the evening a glorious vision of the snowy range far away to the south. What a vivifying contrast it was to the level expanse of gravel and salt marsh and the dreary bleakness of the low hills northward |! On the following morning I moved camp to the large ruin, some five miles eastwards, which when we first passed it on the journey to Tun-huang, had struck me by its palace-like dimensions (Fig. 156). My reconnaissances had since shown that this huge structure, T. xvu., with a _much-decayed watch-tower rising on the plateau edge immediately south of it, lay actually on the line of the | Limes as well as on the old caravan route. An expanse of lakelets and impassable marsh land, some four miles long and two across, stretched on its north side and rendered defence by a wall quite unnecessary. But | neither the familiarity I had gained with the general _ plan and arrangements of the Zzmes nor the close survey _I now made of the imposing ruin could at first give me _ any clue as to its true character and purpose. The building with its enclosing walls presented the im- - posing length of over 550 feet, and at first sight suggested a barrack or Ya-mén; yet the very proportions were enough to dispel such a notion. It consisted mainly of _ three big halls, each 139 feet long and 484 feet wide, which adjoined lengthwise and formed a continuous block facing due south. Their walls, five and a half feet thick 128 GREAT MAGAZINE OF THE LIMES cui and constructed of solid layers of stamped clay about three inches in thickness, rose on a terrace of hard clay which had been cut away to within ten feet or so of the outer wall faces to form a natural base. As the latter stood fully fifteen feet above the low-lying ground occupied by the en- closure, and as the walls of the halls in spite of their decay still rose in parts to twenty-five feet or more, the height of the whole ruin was impressive. A wall of large sun-dried bricks had once formed an enclosure around, keeping with its sides parallel to the outer faces of the great structure but at different distances, While on the north this enclosing wall ran within fifteen yards of the base, on the south it receded to about 106 yards from it, leaving a wide courtyard for approach. Owing to moisture from the low-lying ground, the en- closing wall had in most places crumbled away into a mere mound. But at four points of it, diagonally facing the four corners of the great block of halls, there still rose massive watch-towers to heights of over twenty feet. The fact that these towers were built within the enclosure seemed to indicate that they were meant, not for defence, but as points of vantage for sentinels. Owing to erosion under-cutting the base, the south walls of the halls had fallen for considerable stretches, and deep hollows had been scooped out in the floor and base by the drainage of occasional rainfall escaping on this side. This made it difficult to ascertain where the main entrances of the three big halls and the stairs once giving access to them had been situated. Large windows, such as halls of this great size would have needed for their proper lighting, there were none to be seen in the extant walls. But, curiously enough, the latter both to north and south were at irregular intervals pierced by triangular openings about three feet high, on a level flush with the floor as well as about fifteen feet above it. It was clear they could have been intended not so much for lighting as for ventilation. Of internal fittings or arrangements which might have thrown some light on the purpose of the whole building, I could trace no remains. A narrow platform or plinth which ran round the foot of CH. LX! A STRUCTURALLY RUZZLE 129 the walls at a height of eight inches or so did not give any clue. So all hope of solving this structural puzzle with any certainty rested upon what records or other finds excava- tion might yield. I lost no time about starting them, but at first with very scanty results. The search made in those portions of the great halls where the original floor still survived, proved fruitless, except along the foot of the north wall in the central hall. There we discovered half-a-dozen fragments of inscribed wooden slips. Some of them seemed to relate to individual soldiers, naming their places of origin far away in Ho-nan and elsewhere ; but none of them gave dates or any hint as to the character _ of the ruin. Then I had the whole of the fairly well protected marrow court on the north side searched, and was just beginning to wonder at the total absence of any refuse heaps when at last, on scraping the ground at the foot of the tower occupying the north-west corner, we came upon two-score inscribed pieces of wood and bamboo scattered amidst straw and ashes. Still more abundant were blank bamboo slips, all much worn and repeatedly scraped, evidently representing ‘waste paper’ which had been prepared for fresh use as palimpsest records. The inscribed tablets, too, had here for the greater part suffered badly from moisture, and it was not until the ‘very last piece turned up that Chiang could recognize a precise date. As it corresponded to the year 52 B.c. it now became certain that the ruin dated back to the early period of the Zzmes occupation. But the doubt about the nature of the ruin did not lift until Chiang, by such prolonged poring over these records as would have done credit to any Western palaeographer, ‘made quite sure that two among them distinctly referred toagranary. Inthe course of these days the idea occurred to us both, as well as to Naik Ram Singh, independently, that this strange big building might have been erected for the purpose of serving as a supply-store to the troops Stationed or moving along the wall. The structural pecul- larities above noted; the small openings for ventilation ; VOL. II K 130 GREAT MAGAZINE OF THE LIMES ca. ux the size of the halls quite unsuited for habitation; the choice of a building site conveniently accessible yet well raised above the adjoining ground; the arrangements for guarding the building, not against hostile attack, but against theft, all found thus their simple explanation. But definite proof was supplied only by M. Chavannes’ detailed analysis of the still legible records. One among them is an issue order for grain signed by three officials specifically named as in charge of the granary. Another is still more significant, because it is an acknowledgment for a large consignment of corn delivered from a specified area of cultivation in the Tun-huang oasis, evidently as its contribution towards commissariat requirements of the border. Elsewhere, again, we find an order for twenty suits of a particular sort of clothing such as a military magazine might store. The advantages of an advanced base of supply on this desert border, both for the troops which guarded it and for the expeditions, missions, and caravans which passed along it, must be obvious to any one familiar with the difficulties of moving large bodies of men over such ground. As I looked towards this ruined magazine from the route edging the plateau, and twenty centuries ago the main artery for Chinese trade and political expansion westward, I could not help turning my thoughts back to the huge sheds and ‘commissariat godowns’ which the traveller must pass as he approaches Peshawar. They contain the military stores — provided for an advance, if ever it be needed, by the one great route which connects India in its extreme north-west | corner with Kabul and thence with Central Asia. And yet what a smiling look even the most barren parts — of the Khyber Pass bear when compared with the desert . through which the Chinese once moved their troops to | Lop-nor! In those days the great magazine must have | seen busy scenes, and quarters for guards and administra- | tive personnel, no doubt, existed near it. The remains of. all such less permanent structures had disappeared before | the attack of wind erosion or moisture, except on a clay terrace near the eastern enclosing wall, where we came. upon layers of refuse, and below them a room partly dug CH. LXI ADVANCED BASE OF SEUPPEY BS! out of the solid clay. The single record found here also dated from 52 B.C. But it was not merely the archaeological assurance gained that cheered me while encamped at this ruin. On exploring two ancient watch-stations north-eastwards (T. xrx. and T. xx.), I had noticed from the top of the high isolated clay terraces on which their ruins were perched (Fig. 178), that the wall here reappearing rested its right flank on a lake over two miles long, and that an open channel of water apparently led into the latter from the side of the Khara-nor and again left it westwards. A short reconnaissance north had shown me that it was a real river, some twenty yards broad, and quite unfordable at this point, flowing out of the lake with a velocity of _ nearly a yard per second. At the same time I convinced myself that, between the lake on the east and the wide marsh bed already mentioned as extending north of the | magazine, there stretched a tongue of firm ground with a . few isolated clay terraces. This circumstance at once _ accounted for the care taken to close this gap in the natural line of defence. The wall was clearly traceable eastwards to the very _ edge of the lake’s marshy foreshore, which showed a level only five feet lower than the foot of the wall, and was evidently still liable to periodical inundation. This was an important piece of evidence, agreeing with observations made at other points where the wall abutted on lakes or marsh beds, and proving that the difference here in the water level of the present and ancient times could not have been very great. There was another curious fact pointing to little change in the local conditions of soil and climate. The whole area between the two towers and farther on to _ the lake shore was covered with a luxuriant jungle of wild , poplars—just as it must have been two thousand years ago; for the wall here proved by exception to be constructed, not with the usual reed fascines, but with layers of Toghrak branches, the material still abundant on the spot. The time to realize fully the geographical importance of that observation about the lake outflow arrived when in the evening Rai Ram Singh rejoined me from the prolonged 132 GREAT MAGAZINE OF THE LIMES cu ux tour of reconnaissance on which I had despatched him a fortnight earlier to the westernmost extremity of the wall. After tracing a far-extending line of towers south- westwards, and a wide basin of salt marshes and lakes which lay in front of them, he had returned to the caravan route at the point known to us as Toghrak-bulak, in order to replenish his supply of fresh water. Where he expected to find a spring-fed marsh, as seen on our first journey towards Tun-huang, he discovered to his surprise a deep and rapid river flowing westwards. He recognized at once that such a volume of water could only come from the flood of the Su-lo Ho which, as I realized at the time of our first approach from the Lop-nor side, had once made its way into that dry eroded basin we traversed on the day before reaching Toghrak-bulak. I had specially asked him to re-examine the dry river beds then sighted near it. Now, after fording with difficulty the river which flowed in the Toghrak-bulak bed, he followed it downwards until he made quite sure that it emptied itself through a little delta into the south-eastern end of the basin which early in March we had found quite dry. Since my return to the Lzmes I had myself looked out more than once for the old bed, dry as I supposed it would be, stretching west of the Khara-nor, and on the north side of the fortified line. Once or twice I had pushed some distance beyond the latter to what seemed the very foot of an unbroken glacis of gravel sloping up gently towards the Kuruk-tagh. My abundant archaeological tasks would not allow me to move beyond this. So for the time being I had been driven to the assumption that the dry beds noticed about Toghrak-bulak had their continuation to- wards Khara-nor only in the chain of isolated depressions, some dry, some still occupied by salt marshes, which we had come across both within and without the line of the wall. In this case a direct connection between those old beds — and the Khara-nor could not well have existed within — historical times. But now the aspect of this puzzling © question of drainage was completely changed by the dis- covery of an actual outflow from the lake which lay west — @eixt 1 ERMINAL COURSE OF SU-LO HO 133 of Khara-nor and was unmistakably connected with it, and also by Ram Singh’s discovery of a river flowing through the Toghrak-bulak bed. It became certain that the Toghrak- bulak river was fed from the Khara-nor, and that the latter was not the terminal basin of the Su-lo Ho river, as had been assumed hitherto. But it still remained to deter- mine whether the connection between those two lay, as | _ expected, through the series of minor lakes and marshes | had seen west of Khara-nor, or whether there was perhaps a channel carrying the water of the Su-lo Ho direct from _ the latter down to Toghrak-bulak. | This question, too, was finally answered when on the morning of April 29th I set out to track this connection. _ From ourcamp near the big magazine ruin we moved to the north-west, skirting the wide basin of reed-covered _ marsh and lagoons, until suddenly we came upon a narrow and deep-cut Nullah receiving the suspected outflow towards Toghrak-bulak. It was a regular river, over twenty-five yards broad, and flowing with a velocity of a little under one yard per second. In the middle the water was over six feet deep. Though drinkable, it tasted distinctly brackish from all the salt deposits which this spring flood was sweeping out of the lake beds. As I followed this continuation of the Su-lo Ho for some distance downwards, I fully understood how easily the river could here escape discovery owing to the very | deceptive way in which its course is masked by what looks an unbroken glacis of gravel. I must have ap- | proached it before at another point, to within a quarter of _a mile or so without noticing its existence. Yet for the _time boats could have passed along it with ease. _. On the same day I shifted my camp back to the site of the ‘ Yii-mén’ fort, preliminary to a move for the exploration of the westernmost portion of the Zzmes, while Rai Ram Singh was sent eastwards for survey work about Khara- nor. Then on April 30th, by a forced march over twenty- eight miles, I brought my straggling column of camels ‘and men right across to the point where the line of the wall as marked by its watch-towers was seen to bend round to the south-west. The heat and glare on the bare gravel 134 GREAT MAGAZINE OF THE LIMES cx. um plateaus was most trying; and when in the evening we descended to the edge of the wide marsh-filled basin which stretched westwards as far as the eye could reach, we were assailed by clouds of mosquitoes and other insects, In order to secure some protection from this pest, and yet at the same time to keep reasonably near to the springs located by the Surveyor and to the grazing, I pushed for nearly two miles beyond the edge of the basin, here lined by a belt of luxuriant vegetation, to where a bold and broad terrace of clay promised a dry and airy camping- place. For, like the wild camels whose resting-grounds we had repeatedly come upon on the top of isolated clay ridges or plateaus, we soon realized that an elevated wind- Swept position was the only means for escaping the worst onsets of those insect fiends. The tents had scarcely been pitched in the darkness when I became aware that the choice of our camping- ground had given us protection from a far more serious danger. The labourers sitting down for a little smoke while waiting for the baggage, had lit fires in the jungle and carelessly left them smouldering. A strong north wind, which rose after dark, fanned these into a big con- flagration, spreading with amazing rapidity amidst the dry. thickets of scrub and reed-beds. It was a wonderful sight to watch the broad array of flames over-running the leafless wintry jungle. It spread a glorious illumination on three sides of us, and burst into something like fireworks where- ever groves of large Toghraks were set ablaze. From the bare slope of our clay terrace we could watch the grand spectacle without serious apprehension. But when the first supply of water brought along from the springs was exhausted, Hassan Akhun found it no easy task to take the camels back to them by a circuitous route, and most of our animals did not get a drink until daybreak. CHAP DR Vaid] ON THE WESTERN FLANK OF THE LIMES WHEN on the morning of May 1st I set out to visit the neighbouring ruins and reconnoitre this new ground, I was obstructed not a little by the strange combination of water and fire. Most of the ground separating our camp terrace from the three watch-towers within view was marshy, and it cost much care and many détours to avoid hopeless bogs. At the same time, where the soil on the edge of the wide depression was firmer, the fire in the reed-beds was still smouldering, and a passage had to be picked with caution. All over the low ground salt efflorescence was abundant, and its contrast with the blackened tamarisks and Toghraks and the singed reeds very striking. It seemed cruel to see this hardy jungle vegetation, which had held its own amidst such deterrent conditions of soil and climate, succumb to fire just when it was preparing to greet its short-lived spring. But physical drawbacks were soon forgotten over the absorbing antiquarian interest of the site. A careful survey of the ground soon convinced me that I now stood within the westernmost extension of the Lzmes. With that unfailing eye for topography and all its strategic bearings which the Chinese have proved again and again to possess, the engineers of the Emperor Wu-ti had carried their fortified border line right up to the point where it could rest its flank safely upon a huge depression—once, no doubt, a great lake basin and since historical times an impassable bog in most places. A look at Map 1. and its inset 4, much reduced as is the scale, will fully explain this; but a much larger map would be needed to illustrate 135 136 WESTERN FLANK OF THE LIMES cauag the details. Instead of taking the wall towards Toghrak- bulak, where the caravan route must have run then as now, the constructors of the Lzmes let it continue due west along a narrow and well-raised plateau tongue up to its extreme point. There a tower (T. Iv. B) built of carefully laid bricks, and still over twenty-three feet high, rose on the brink of steep clay cliffs some 120 feet in height, commanding an extensive view westwards and over scrub and gravel Sai to the north. No better look-out place could have been selected for this exposed portion of the fortified line. But the wall had been carried about one and a half miles farther west to an isolated clay terrace rising from the scrub- covered edge of the basin to a height slightly lower than the plateau end just mentioned (Fig. 181). The top of this terrace was occupied by a much-decayed tower in stamped clay (T. tv. 4), which completely overlooked the low-lying ground all round. Here the wall took a sharp turn to the south, and could be traced as a low mound for a mile or so running in the direction of the terrace on which our camp stood. But as the soil there grew more and more marshy, the last faint indication of the agger soon disappeared entirely. It was clear that the very nature of the ground to the west, all spring-fed marsh and lagoons, had rendered defence by a wall needless on this flank farther on. But a line of towers visible far away to the south-west, perched at great intervals on headlands of the plateau, showed that the flank had been guarded all the same. The ruined quarters adjoining the tower T. rv. B yielded a number of well-preserved Chinese records on wood and silk, on one of which Chiang thought he could recognize a date corresponding to the year 94 B.c. So my thoughts were carried back to near the times when this Lzmes had served the first political expansion of China westwards. As I looked round from the commanding position occupied by this tower, I wondered why its builders had not rested content to let the wall make its bend here. Then my attention was attracted by two straight lines of mounds rising above the scrub-covered ground in the direction of ! i). RUIN OF ANCIENT WATCH-TOWER, T. VI. A.. ON WESTERN FLANK OF TUN-HUANG LIMES, SEEN FROM SOUTH. The tower is built on an eroded clay ridge, and on right overlooks a depression with Toghraks and reed. beds. A Tre | Viens I, RUIN OF ANCIENT WATCH-TOWER, T. IV. C., ON WESTERN FLANK OF TUN-HUANG ZIMES WITH VIEW TO NORTH. > ip left an eroded clay terrace with deep-cut Nullah. Across depression with Toghraks and tamarisks is seen in distance an isolated clay terrace (A), bearing remains of ruined watch-tower, T. 1v. a. . - oe cx.uxr WESTERNMOST POINT OF WALL 137 the last tower westwards. They looked distinctly like the remains of earth ramparts meeting at right angles, and their position within the bend of the wall and near its westernmost point at once suggested an entrenched camp. Naik Ram Singh, who had taken the labourers to the ruin while I was reconnoitring, had already noticed these curious lines. Subsequently, on riding over the ground, I found that in spite of much ‘Shor,’ or salt efflorescence, scrub of various sorts and reeds were covering it in pro- _ fusion; and so it was difficult to follow the lines up. But rows of little mounds, eight to ten feet high, and thickly covered with dead Toghrak trunks and branches, could even thus be made out at what from above [ had taken for the south-west angle of the enclosed area. On going up a Steep terrace a short distance to the north, the lines of ramparts showed up again, forming an oblong about five hundred yards long and half that across. With its short side on the north this oblong seemed to approach closely the line of wall connecting the two westernmost towers (T. rv. A and B). Were these then the remains of an entrenched camp established at some time at the very point where the ancient route from the Tarim Basin passed within the walled line of the Zzmes? Would this not have been the most likely position to select for the original ‘ Jade Gate’? Here at the foot of the plateau there would have been some shelter from bitter winds and no difficulty about reaching _ water by digging. A larger station for troops was here justified by the risks to be guarded against at this exposed corner. At the same time its existence would furnish adequate explanation both for the extension of the wall to a point beyond the tower T. Iv. 3, and for the construction of a third tower (T. Iv. c, Fig. 181), which we found over- looking and guarding the site from a plateau tongue about one and a half miles to the south-east. For the complete decay of the rampart and the total absence of structural remains inside I could account by the vicinity of abundant subsoil water. In fact, eight months later I had occasion to acquaint myself with the exactly corresponding results which 138 WESTERN FLANK OF THE LIMES cu. LXII permeation with water had produced on ruined sites by the salt-impregnated shores of Lake Baghrash. On May 2nd, while excavation at the towers continued under Chiang-ssti-yeh and the Naik, I made a long day’s reconnaissance with a few mounted men along the ancient route to the north-west. It furnished definite proof that, though there had been no extension of the wall in that direction, yet the towers I passed on my first approach to Toghrak-bulak in March had been constructed at the same time as the Lzmes proper. Evidently these were meant for watch-posts from which fire signals, such as the records of the Limes so often mention, could be sent on in advance to the detachments guarding the wall. Their position suggested that the deep-cut old river bed, now completely dry, near which these outpost towers were built, had then still received water from the Su-lo Ho, or had, perhaps, served as its main flood-bed. Now the river was confined to the narrow Toghrak-bulak bed, filling it so completely that we had to descend some distance before we could find a place shallow enough to cross. Its width there was over forty yards, the water being close on three feet deep, and flowing with a velocity of about two yards per second. The following days were devoted to the exploration of the line of watch-stations which stretched far away to the south-west along the edge of the great marshy basin. The distances at which they were placed from each other, ranging from four to six miles, showed clearly that they were meant mainly as signalling-posts along a line and not for the purpose of warding off inroads. The objects which the wall was intended to serve along the front of the Lzmes was here secured by the impassable nature of the marshes westwards. Yet even thus the links of the chain of posts could not have been kept so far apart, with the consequent saving in trouble and cost, had not the configuration of the ground offered ideal positions all along for conspicuous signalling-stations. As a look at the inset map 4 shows, the bare gravel- covered plateau marking the foot of the glacis of the mountains here stretches out a succession of finger-like ridges of clay. These project into the wide marsh-filled cwixu IDEAL SIGNALLING STATIONS 139 depression like promontories of a much-indented coast-line. Rising with very steep faces to heights of 120 to 200 feet above the intervening depressions and commanding ex- tensive views, these ridges furnished excellent natural bases for the towers, and the Chinese engineers did not fail to make the most of them. On this account it was always easy for us to sight these towers from afar. It was curious to note, as the survey progressed to the south-west, how much care they had taken to place their signalling towers for a direct distance of more than twenty-four miles in an almost straight line, as if they had fixed their positions by sighting with a diopter! But it would need a large- scale map to discuss such details. But, quite apart from the skill of this early military engineering, there was enough of interest in certain physical features of this forlorn region to reward attention. As I made my way slowly from tower to tower | found myself skirting the coast-line of an ancient lake basin now partially dried up. While crossing in succession its bays and inlets, all occupied by abundant Toghrak groves and reed-beds, and then again the boldly sculptured clay ridges which formed the headlands between (Figs. 180, 181), I could not help noticing that the latter almost all ran in the same direction from south-east to north-west. Where the bays were wider there could be seen within them strings of isolated clay terraces exactly parallel to the ridges. Else- where the latter had a continuation formed by similar clay terraces projecting farther into the marshy basin and still maintaining the same bearing. The bays and intervening ridges clearly owed their direction to the carving done by running water, which had once descended from the foot of the distant mountains and across that gravel glacis now so terribly dry. This became quite certain when, just below one of the ridges crowned on its top, some 200 feet higher, by a conspicuous watch-tower, I came upon a deep-cut dry river bed. It came from the south-east, and showed unmistakable signs of having been washed by occasional floods at a relatively recent period. ‘The banks were so steep that the camels could not be got across without difficulty. Elsewhere, 140 WESTERN FLANK OF THE LIMES cu. xxn too, the Surveyor, in the course of his reconnaissance, came upon dry ravines cut by floods into the glacis. The springs which we found rising in marshy beds within the larger bays also supplied proof of such drainage. But it was equally certain that in forming those curious strings of isolated clay terraces another powerful agent had been at work, wind erosion. Again and again I observed how the crest of the narrow but still continuous ridges was being gradually sapped and scooped by wind- cut trenches mainly starting from the north-east. It was, of course, the direction from which since the winter we had learnt to expect the most violent gales. The ridges, stretched more or less at right angles across the line of the prevailing winds, were thus bound to be gradually broken up into terraces by the cutting force of the wind and the grinding produced by the driven sand which it uses for its instrument. But as I looked at the towers and saw how relatively little their bases had suffered by erosion in the course of two thousand years, I could judge what countless ages were needed even for that powerful agent to carve out those terraces from the hard clay, and finally to plane them away altogether. I had before me the clearest illustration of that interaction of water and wind erosion which must have produced the maze of detached clay terraces I had before encountered, both at the north-eastern end of the ancient Lop-nor lake bed, and again in the terminal basin of the Su-lo Ho, and around Khara-nor. In the light of the evidence here gathered I was able to understand better the powerful réle which wind erosion had played in this region as a geological factor affecting surface formations. Its effects upon the work of man were marked with equal clearness. Again and again I noted, in the course of my surveys on the Lzmes, how relatively well preserved the wall often rose along those sections which lay parallel to the prevailing direction of the winds, while where the line had been drawn across and in any way barred the progress of dust and fine gravel, wind erosion had badly breached or practically effaced the rampart. I had ample occasion to convince myself that the winds cxw.txr WIND AS A GEOLOGICAL FACTOR 141 which now blow over the desert with remarkable violence and persistence come mainly from the north-east and east. How often had Chiang and myself had reason to comment upon that terribly cutting wind which sweeps down the lower Su-lo Ho valley from the side of the Mongolian Gobi, and which, as he told me beforehand, Chinese travellers to and from Turkestan have learned to dread under the name of ‘the wind of An-hsi’! The whole region seemed, indeed, a true home for Boreas and his sons. The extent and character of the damage which the wall has suffered in its various sections fully agreed with the evidence furnished, on a vastly larger material and chronological scale, by the jagged ridges and isolated terraces of this ancient lake shore. This observation derives additional importance from the fact that the same prevailing winds make their effect felt even far away in the Tarim Basin, as I had ample occasion to observe in the climatic conditions and surface formations round Lop-nor. It is probable that many years will pass before that delectable region and the equally attractive desert tracts about Tun-huang are pro- vided with their meteorological stations to supply exact data,—and to tax scientific devotion. But even without such data I may hazard the conjecture that a likely explana- tion for these prevailing winds is supplied by ‘aspiration,’ due to the higher temperatures which the atmosphere of the low-lying desert around and to the west of Lop-nor _ must generally attain as compared with the great barren _ plateaus of stone and gravel to the north of the Su-lo Ho _ depression. | For me there was something distinctly stimulating in the bigger physical features of this desert of gravel and _ marsh-land, and in the expanding horizon as we moved from _ tower to tower south-westwards. The great marshy basin _with its glittering salt efflorescence looked at times as if it _were still one big lake. On clear days—and of such we _now had several—I! thought I could distinguish high ridges _of drift sand beyond it to the north-west, the easternmost offshoots of the Kum-tagh sands we had skirted from the _Lop-nor side. From the southern edge of the basin the I42 WESTERN FLANK OF THE LIMES cu. uxn bare gravel glacis was seen rising to the foot of an outer chain of hills completely overrun by huge dunes and presenting a truly formidable appearance. This great obstacle must have absolutely protected the end of the Limes from being turned on the south. Behind this at times towered up a magnificent range of snowy mountains, the watershed towards the north of the Tibetan plateaus of Tsaidam. The feeling of remoteness which the whole silent land- scape breathed was evidently shared by the wild camels, They were now sighted frequently, moving down to the springs and the grazing in the bay-like depressions, or else speeding away fleetly to their resting-places on the bare gravel plateaus to seek protection from the myriads of tormenting insects. Smaller game seemed to do likewise. Only we men had to stick to our ruins and camps close by the spring-fed marshes, and suffered accordingly in spite of big fires and all else we could do to ward off those pests, Even the motoring veil I was now wearing day and night failed to provide adequate protection. The heat and glare, too, had grown more and more trying. I could scarcely wonder that under such conditions the difficulty in retaining our somnolent labourers steadily increased. The vicinity of the caravan route had helped somewhat to calm their apprehensions about being in the desert. Chiang’s paternal kindness of treatment and my constant care for keeping them well provisioned at my own cost had also made some impression on these shifty, callous fellows. But now they evidently dreaded being led farther and farther into the ‘Great Gobi.’ What with the men who for some ailment or other had to be invalided, and others who took the opportunity to desert with them back to the oasis, our column was now rapidly dwindling. There was no track or other indication of this ground having been frequented in recent times by herdsmen or others, and this seemed to depress the remnant. It was useless to point out to them the cut stumps of trees we | found here and there on salt-encrusted soil amidst jungle | mostly dead. For who might say how many centuries ago | this clearing was done? | cuixt FINE HAUL OF CHINESE TABLETS 143 On the evening of May 4th I arrived at a ridge tapering away as usual at its end into a line of isolated terraces. The fact that both the end of the ridge and the last outlying terrace were occupied by towers was bound to attract my attention. As the distance between them was only about two miles, the ruined tower to the east and within the line (T. vi. B) could not have been intended for a mere signalling-post. So the thought suggested itself of some controlling main station. The plentiful debris adjoining the tower seemed to indicate quarters somewhat larger than usual, and numerous refuse layers scattered over the gravel slopes near by furnished proof of prolonged occupation (Fig. 182). The position, though not more than a hundred feet above the basin, seemed well chosen for a sort of Aoznt @appur on this flank. It commanded a complete view of the southern and south-eastern shore of the lake basin, up _ to the foot of the hill chain covered with high dunes which - was seen to come from the direction of Nan-hu and to bend round to the north-west. It was clear that, as any attack from the west and the Lop-nor route would have to keep ' between the shore of the marshy basin and the impassable = high sands, the line of watch-posts thrown out almost to within sight of the latter and controlled from this station _ would effectively prevent the main Lzmes being outflanked ' by raiders. That evening we made an experimental scraping of the | gravel-strewn slope at a point where a few wooden posts protruded some twenty yards north of the tower, and soon _ brought to light about two-score inscribed Chinese tablets _of the usual shape. Though many of them had their sur- — face decayed through exposure, the find was encouraging. ' Next morning I sent Chiang back from camp to continue the search while I had to look after the sinking of a —e well, When I rejoined him an hour later I found him _triumphantly guarding for me nearly a hundred fresh _ Chinese records on wood, almost all complete, and in spite of slight damage from moisture quite legible. It was a : i delightful surprise, and my satisfaction grew still further when I ascertained that the numerous dated pieces all 144 WESTERN FLANK OF THE LIMES cu. 1xn ranged between 65 and 57 B.c. This fine haul all came from an area scarcely more than two to three feet square. It was clear that a little office archive had been thrown down here on the rubbish-strewn slope; and it proved by no means yet exhausted. Clearing the thinner refuse stratum lower down on the slope until the hard gravel surface was reached, and subsequently excavating and sifting with care the refuse round the original find-place, we brought the total of inscribed pieces of wood to over three hundred. Of uninscribed slips and tablets where the writing had become effaced through exposure there were found over two hundred more. Fig. 183 shows the scene of this successful digging, and also how shallow the protecting cover of gravel and refuse was. The rest of the rubbish heaps on the other slopes below the watch-tower, extensive as they were, added scarcely more than a dozen wooden records. But in the course of this search, which Chiang-ssti-yeh supervised with his unfailing zeal and patience, there came to light at one point a great mass of wooden ‘shavings’ covered with Chinese characters. As the writing was manifestly by the same hand, and the phrases constantly recurring, Chiang at once rightly concluded that these were chips from tablets which some officer or clerk, eager to improve his penmanship, had used again and again for writing exercises, planing them down with a knife each time to obtain a fresh surface. The material had been of the cheapest, roughly cut from tamarisk and Toghrak branches, which the jungle close at hand would furnish in plenty. The number of accurately dated records had been so great at this station, and their range in time so restricted, that the assumption seemed fully justified of the ruined quarters built against the tower representing the state in which they were when last occupied about 57 B.c. So all constructive details about the rooms, plain as they were, presented points of interest. The plan reproduced in Fig. 184 will illustrate their arrangement, which differed in no essential from that elsewhere observed along the Limes. The thickness of the outer walls and the massive 182. REMAINS OF ANCIENT WATCH-TOWER AND QUARTERS, T. VI. B., TUN-HUANG LIMES, ) BEFORE EXCAVATION, SEEN FROM WEST. 183. RUBBISH-STREWN SLOPE BELOW RUINED WATCH-TOWER T. VI. B., IN COURSE OF EXCAVATION. _ The splintered piece of timber held by labourer marks the spot where hundreds of Chinese records on wood, all of rst century B.c., were discovered close to surface. | ‘. CH. LXII PLAN OF RUINED OUARTERS 145 wooden posts of the narrow entrance arranged to take heavy bolts, plainly indicated that the need of defence had been present to those who first built the small oer 184. GROUND-PLAN OF ANCIENT 4. The remains of furniture NR TREC ATIC vic: and equipment left behind by the last occupants as of no value showed the same uniform simplicity befitting a remote frontier. There were curious carved handles or hooks of wood, intended, perhaps, as emblems or for supporting stands of arms; pieces of hard wood with leather-lined grooves, which seem to have belonged to cross-bows or small cata- pults; wooden tent - pegs with rough design of a human head on the top; pieces of painted and lacquered bowls in wood; much-mended rags of silk garments, and, more plenti- ful still, rope-soled shoes of coarse make (Figs. 173, 174). Among the dozen or so of wooden records which had found a safe refuge in a layer of refuse on the floor of an outer room, there was one dated in the year 68 B.c. But after all preceding experiences it scarcely needed this docu- mentary evidence to convince me that, in a soil which had seen extremely scanty rainfall for the last two thousand years, and was far removed from any chance of interference by human agency, nothing but wind erosion could prove destructive even to the most perishable remains, VOL. I L CEMA Roe slibl “RECORDS FROM AN ANCIENT WATCH-STATION FASCINATING as it was to survey the ground along this most desolate of borders, and to study the actual remains of wall, towers, quarters, and arms, I found that it needed written records to restore a picture of the life once led here and of the organization which had planted this life in the desert. Chance could not have illustrated this better than by letting me light, at a post so modest in its extant structures, upon an abundance of wooden records thrown out from the commandant’s office half a century before the birth of Christ. The fact that they all belong to practically the same period, and come from a station which lay off the main route and could not claim special importance, makes them all the more useful as typical evidence of the military administration then prevailing among the troops echeloned along the border. To M. Chavannes'’ critical acumen and unsurpassed powers of scholarly work I am indebted for full translations and notes of all documents found along the Zzmes that are still decipherable. Availing myself of the fruits of his labours I propose to give a brief summary of the chief topics of antiquarian interest which the records of this station illustrate, and to supple- ment them where desirable by general information gleaned from the rest of the documents. From the first I had felt justified by Chiang’s first rapid examination to hope that the records recovered | would give us details as to the strength, distribution, and - life of the troops guarding the border ; their commissariat, equipment, and the like. This hope has been fully realized. In the introduction to his forthcoming volume 146 o.um ORGANIZATION OF STATIONS 147 M. Chavannes shows that the troops on the border were composed of regular companies or ‘tui,’ organized under imperial edict and permanently attached to particular sections. Now a large number of records found at the station T. vi. B mention the ‘Ling-hu’ (or ‘barbarian- suppressing ’) company in such a way as to leave no doubt about its having garrisoned that station and had its head- quarters there. Altogether some twenty companies are thus named in the records from the various stations, and in most cases it is still possible to determine the sections of the line which they guarded. Curiously enough, a record found at T. vi. B distinctly quotes an imperial edict concerning the raising of the Ling-hu company and two others which seem to have been located near by, and directs the posting of the edict “in a place suited for general information.” Each company had a commandant who is frequently named as the recipient of orders sent from headquarters, and his station is always referred to as a‘ T’ing.’ One of the documents specifies 145 soldiers as attached to this particular station, and M. Chavannes assumes this to have been the approximate number for each company. _ Under the company commandant the documents constantly mention the officers in charge of the watch-towers (‘ hou- chang’) and the assistant attached to each of them. To the soldiers placed under them, of whose number we are not exactly informed, but who must have been posted by reliefs, fell the duties of transmitting the fire signals, of patrolling the line, and furnishing the fatigue parties for collecting fuel, supplies, etc., to which our records so ‘often refer. Just as small groups of watch-towers were garrisoned by individual companies, so the latter formed ‘sections of the ‘ barrier,’ as the Lzmes is always spoken of, under superior officers. Over the westernmost portion of the Lzmes control was exercised by ‘the military ‘commandant of the Jade Gate or Yii-mén,’ who in turn depended, directly or indirectly, on the governor of the -Tun-huang command. In the files recovered from the office of the Ling-hu ‘company we have ample illustration of the administrative 148 RECORDS FROM WATCH-STATION cu. rxm routine which kept these links of the border hierarchy occupied. There are notifications of new appointments to sectional command or to charge of neighbouring companies; injunctions for the strict observance of previous regulations concerning the men, animals, and cars which passed through the ‘barrier,’ sometimes with the mention of specific dates, suggesting arrangements for convoys such as are in use, ¢g. for caravan traffic through that ‘Gate of India,’ the Khyber Pass; instruc- tions for the circulation of orders. Chiefs of companies are summoned to headquarters for discussion of particular questions, or detailed for the supervision of the granary, We have reports of cases of illness, of grants of leave, and the like, and here and there what evidently must have been quasi-confidential statements of the conduct of particular officers and of accusations against them. Inspection visits along the ‘ barrier’ by superior officers, in one case by the governor of the Tun-huang command, are announced in advance. It was a sensible precaution, to obviate—for the time being—awkward deficiencies in men, such as we find acknowledged with much contrition in a record from another watch-station. The frequency with which orders are addressed to ‘indigenous function- aries, apparently also entrusted with duties in guarding the line, is of interest; for it agrees exactly with Chinese policy as still maintained in the employment of such local auxiliaries for the guarding of outlying border posts, é.g.on the Pamirs and among the Kirghiz of the Kun- lun. Our knowledge of the actualities referred to in such orders would, no doubt, be much more complete, were it not that in many cases the letters, though exactly indicating the functions of the sender and addressee, also the date and the person carrying the order, leave it to the latter to explain verbally the details of the message. Considering the intricacies of Chinese writing and official style, and the difficulty of providing an adequate clerical establishment for all the small detached posts, this procedure had, no doubt, much to recommend | itself to officers in a hurry and dealing with non-Chinese. | Amidst the mass of correspondence dealing with cniuxn1 TABLET WITH IMPERIAL EDICT 149 incidents of administrative routine and the still more numerous records concerning the internal economy of individual detachments, it is of special interest to come upon a few documents which give us a glimpse of the political power creating and protecting this border. The question how such documents, emanating, it appears, from the imperial chancellerie, should have found their way to an outlying military post is not easily answered. But M. Chavannes’ scholarly analysis makes it certain that one of the records from the rubbish heap of T. vi. B has preserved for us an imperial edict directing the establish- ment of a military colony. Unfortunately there is no date given, and the tablet is not complete. Though the colony is evidently to be created in the territory of Tun-huang, the execution of the imperial order is entrusted to the governor of Chiu-ch’iian, corresponding to the present Su-chou. This important centre after its occupation in II15 B.C. seems to have served, in fact, as the main base for the political influence which China under the Emperor ~Wu-ti (140-85 B.c.) pushed farther and farther westwards. Two thousand soldiers raised from the garrison of the _Tun-huang command, and some more troops from Chiu- _ch’iian, under a complete staff of officers and accompanied _ by native functionaries, ‘ were to proceed to the locality to _be occupied and to establish there an agricultural colony.” The governor in person was charged with the duty of “examining the configuration of the ground and selecting a suitable position. Utilizing natural obstacles, a rampart 'was to be constructed in order to exercise control at a | distance.” This document, with its first-hand record of a phase in Chinese expansion on these marches, is full of historical ‘interest ; but we can neither fix its exact date nor determine the location of the new military ‘colony.’ The reference to ‘natural obstacles,’ as well as the find-place of the document would, of course, make us think of some locality along the fortified border line. The skill with which the old Chinese engineers utilized the natural defences offered here by lagoons and marshes has been emphasized already ‘more than once. But there remains the plain fact to be >= = 150 RECORDS FROM WATCH-STATION cu. xm faced, that the physical conditions preclude all idea of agricultural occupation having been possible on this line during the Han period. Neither the marshy depressions which must have been, if anything, more extensive then than now, nor the absolutely bare gravel plateaus, well removed from all possibility of irrigation, could have furnished the fields needed for an ‘agricultural colony.’ If the imperial edict really refers to the occupation of some part of the Lzmes, as we have now traced it west of Tun-huang, the only explanation would be that land for maintaining those troops might have been secured by an extension of the extant oasis and its produce allotted for the maintenance of the detachments by the wall. It is true that the records more than once refer to a ‘ delimita- tion’ or clearing of ‘celestial fields’ as among the corvée duties imposed on the men of the stations. But in view of the local conditions it seems far safer to refer this expression to the clearing away of reed-beds and jungle, which must have seriously interfered with the safe watching of the border, than to the breaking up of desert soil for tillage, which had no chance without irrigation. I may mention here in passing that two more tablets from the same rubbish heap have proved to contain imperial decisions. The points on which they were issued cannot be made out with certainty; but the style of writing and expression used in conveying the Emperor’s approval strictly accords with the formulas attested for the clerical etiquette of the period. It is easy to understand that the records from the office of a small local commandant should yield more details about the soldiers he controlled, their condition of life and their duties, than about the policy and organization which kept them employed in guarding this route through the desert. It is of considerable interest that the records, where they name individual soldiers as recipients of articles of equipment, or otherwise, generally indicate with accuracy their origin, stating the canton, sub-prefecture, and province from which they had come. We are thus able to ascertain that the majority of these ‘garrison soldiers,’ to use their specific designation, were drawn from the distant provinces } cx. ixtr ‘GARRISON SOLDIERS’’ SERVICE I51 of Shan-hsi and Ho-nan, with a sprinkling of men from Kan-su, Ssti-ch’uan, and Tun-huang itself. Most of the men must have served on foot; but horse- _men, too, are referred to as well as mounted messengers. Numerous receipts and issue-warrants show that each soldier was entitled to about six-tenths of a bushel of corn per dtem. In addition, he received pay calculated according to length of service. An elaborate system of reckoning was used to establish the total duration of the period for which pay could be claimed, with reductions for ‘short months’ of the luni-solar calendar, but also making allow- ance for the rule that two days of service on the border were to be counted as three, a significant indication of the hardships of service on such ground. In the case of a petty officer the daily rate of pay is stated in silver. A commissariat record at a certain station mentions provision for the feeding of watch-dogs, and proves that dogs were officially recognized as forming part of the regular establishment. Numerous inventories show that the arms supplied to common soldiers by the Administration consisted of sword, cross-bow, and shield. In regard to the cross-bow care is usually taken to indicate its effective strength by stating the weight required to bend it. Thus cross-bows from three to six ‘shih’ are distinguished, a ‘shih’ being equiva- lent to 120 Chinese pounds. The greatly varying weight of the arrow-heads actually found by us bears out these distinctions as to the size and power of the cross-bows from which they were intended to be shot. When the strength had by age fallen below the nominal standard the fact was duly recorded. The regulation issue of arrows per man was 150, two kinds being mentioned, both with bronze heads. Quivers were provided for keeping them. In the case of shields the inventories repeatedly mention the factory in Ho-nan which supplied them and the date of their manufacture. Defects in the arms appear invariably to be stated in the inventory or record of issue. Besides arms, clothing was furnished to the men free of charge. The inventories repeatedly mention tunics and dresses in black linen, for which the price paid by the 152 RECORDS FROM WATCH-STATION cu. cxm company commander is duly recorded. The mention of an ‘undress costume’ of white silk in one soldier’s kit agrees curiously with the abundance of silk rags of varying texture which the refuse heaps of the border-stations have yielded. The inclusion of a tent in the same list may help to account for the fact that the accommodation still traceable at the ruined watch-stations would have been quite inadequate to give shelter to the whole number of the garrison we may reasonably assume to have been quartered there. The Government store lists often include reserve strings of silk and hemp for cross-bows; axes, hammers, and other implements; and here I may mention that a big hammer of wood for tent pitching, which we found in the débris of the station nearest to T. vi. B, was in such perfect condition and so useful that I could not resist my men taking it into daily use for its original purpose during the rest of my journey. There can be no doubt that the main duty of the detachments echeloned along the Lzmes was to provide guards for the watch-towers who would give timely alarm by signals to the rest of the line in case of the approach of raiders. The numerous wooden slips which accurately register the time and other details of fire signals received, or else refer to arrangements made for lighting them, would alone suffice to prove that this means of optic telegraphy was in regular use along the border. But the abundant information from early Chinese texts collected by M. Chavannes shows that the system of fire signalling was known and practised along the frontiers of the Empire long before the time of the Hans. The distinction which those texts indicate between signal fires visible at night and smoke signals intended for use by day is distinctly mentioned in one of the records on wood. In another, neglect to transmit such a signal received from one side of the line by immediately lighting a fire in turn is acknow- ledged as a grievous delinquency. We are not informed by our records as to any devices by which such fire signals could be varied to convey more definite news along the guarded line. But since later texts quoted by M. Chavannes refer to a method of | cxw.uxrr SYSTEM.OF FIRE SIGNALLING 153 marking the relative strength of the attacking force by cor- responding repetition of the fire signals, it is likely that similar devices were practised in Han times. We read elsewhere that General Ma Chéng, when reorganizing the defences of the northern border in 38-43 4.D., placed the fire-signal stations ten Li or about two and a half miles apart ; and this accords remarkably with the average dis- tances observed from tower to tower on the earlier Tun- huang Lzmes, due allowance being made for the varying configuration of the ground. No doubt such a system of optic telegraphy was in- sufficient to assure the rapid transmission of warnings at all times or for the communication of important particulars. Hence the need for mounted messengers repeatedly mentioned in the records, who by relays of horses kept ready at the stations could cover distances at great speed. The presence of such mounts was in fact attested by the plentiful horse-dung we found at each tower, however confined the accommodation near it. A piece of ancient Chinese poetry which M. Chavannes translates, though referring to a part of the border much farther east, gives so graphic a picture of sucha scene that I cannot refrain _ from quoting it: ‘“ Every ten Lia horse starts; every five _ Lia whip is raised high; a military order of the Protector- General of the Trans-frontier regions has arrived with _ news that the Huns were besieging Chiu-ch’iian [ Su-chou]; ' but just then the snow-flakes were falling on the hills along _which the barrier stretches, and the signal fires could raise - no smoke.” To keep watch and guard by the towers and the wall, or ‘to mount the barrier’ as the ancient Chinese expression has it, did not exhaust the duties of the men posted at _ these stations. Numerous wooden records tell us of the varied fatigue labours demanded from them. Men were —~w~e — ordered out to work on the construction of defences, a task which along this westernmost Lzmes must have included the preparation of the fascines forming the main materials ; ' to collect fuel for the signal fires, not an easy matter where _the towers happened to be miles away from the scrub-filled b } } depressions. At several stations we have also reckonings 154 RECORDS FROM WATCH-STATION cu. .xm of men employed in making bricks at the rate of 70 or 150 per diem, perhaps according to their size; in shifting and piling up bricks; in plastering walls of quarters, the work performed being accurately estimated in square feet. A considerable amount of corvvée seems to have been absorbed by the carriage of supplies and materials needed at the several posts. In such cases an elaborate calculation informs us of the number of days spent, the daily distances covered, and the weight carried by each man. No pay- ments to such fatigue parties are ever mentioned; but careful record was kept, evidently for the purpose of making the distribution of tasks equitable. The humdrum life of these petty garrisons must at times have been disturbed by serious incidents and alarms, Though the records of the main find at T. vi. B cover only a relatively short period (65-57 B.c.), they supply us with evidence of this. In one document, unfortunately for the greater part effaced, we have a report about ‘“‘a barbarian horseman having been seen riding towards a watch-station with drawn bow; on a discharge from the cross-bows he retired.” But a serious attack was evidently anticipated, since ‘‘the chariots and horsemen are directed to keep on guard,” and ‘“ the men at the watch-towers to keep a sharp look-out for fire signals and to keep the border clear.” Elsewhere we have an earlier event referred to ‘‘ when raiders burned and sacked the quarters of a battalion, exterminating two hundred people.” That the guarded line was incidentally used to prevent the escape of deserters or others ‘ wanted’ within the border is made clear by notices relating to such offenders, In fact, we know from plentiful Chinese texts down to T’ang times that to pass the ‘ barrier’ outwards required special authorization. Of private letters, too, there is a fair sprinkling ; but we must regret that, with their brevity and the exuberant indulgence in polite phraseology which Chinese epistolary style enjoins, they yield little of antiquarian interest. Yet short and formal as they are, it is impossible not to detect between their lines an expres- sion of that feeling of weary exile which prolonged stay | on this barren and trying frontier must have engendered. — CH. LXIII LIFE ON DESERT BORDER T55 And, indeed, compared to these desert stations the most forlorn outposts on our Indian North-West Frontier, which have become by-words for discomfort, would present them- selves in the light of a veritable paradise. Under such conditions of life, we should scarcely be justified in looking out for any literary remains. Yet there is interesting evidence to show that study of some sort was not altogether neglected. There are fragments of texts dealing with divination. But more numerous and interesting are fragments of lexicographical texts, among them eight pieces, found at different stations, of a famous treatise, the Chz-chiu-chang, composed between 48 and 33 B.c. and widely used in primary schools during Han times. They are the oldest manuscripts so far known of any Chinese literary work, and as they include a well- preserved large tablet with the first chapter complete, and showing some textual variation, their critical importance is in M. Chavannes’ opinion bound to attract attention among scholars in China. The frequent recurrence of these frag- ments at various stations is significant of the popularity then enjoyed by that school-book, but it also proves that there must have been plenty of men among the scattered garrisons eager to ‘improve their education.’ Character- istically enough, there is found also in one tablet a refer- ence to the Lzographies of Eminent Women, an ancient work which has always enjoyed renown as a moral text- book. With these literary fragments, modest in themselves yet of historical interest, must be classed the abundant remains of elaborate calendars yielded by the main refuse deposit of the Ling-hu company’s station (T. vt. B). Written on tablets of a special size, over fourteen inches long, they indicate the cyclical designations of particular days in the month for each of the twelve months of the year. M. Chavannes, by a careful comparison of these data with those recorded in Chinese chronological works, has been able to prove that the calendars to which these tablets belonged were issued for the years roughly corresponding to 63, 59, and 57 z.c. They were indispensable for the correct dating of official correspondence, accounts, etc., 156 RECORDS FROM WATCH-STATION cua. uxm which issued from the local headquarters. Without going into technical intricacies of the complicated cyclical system of reckoning, it must suffice to mention that these calendars possess considerable importance as records which help to control, and in a few instances to rectify, the traditional chronology of the Han period. Up to May 7th I was busy clearing the station the records of which have just been passed in rapid review, and exploring the last watch-towers traceable westwards. The one which lay about two miles from T. vi. B was remarkable for the ideal position it occupied on the flat top of a small and completely isolated clay terrace fully 150 feet high, The precipitous wall-like slopes made access to the ruined tower quite impossible except from the east, where a steep ravine descended, and even there it meant climbing. The terrace, only about eighty yards long and less than half that in width, suggested a huge natural keep. From its top I looked across the great basin westwards, all salt- encrusted marsh with here and there large open sheets of water, and felt as if I were surveying a great lake from the highest deck of a steamer. Here, too, the remains of the guard’s quarters yielded relics of interest. ‘The most curious among them was a remarkably well preserved tablet, about sixteen inches long, bearing at the top four short lines in the unknown script resembling Aramaic which we had found weeks before in the paper documents from the tower T. xu. A. The tablet looked like the right half of a larger piece cut through on purpose like a tally. I was already inclined to connect this writing with Western Turkestan, and the place of discovery, so far away from the ancient trade route, made me wonder at the time whether possibly men from that region had found service among the auxiliaries of the Chinese on this border. In another way it was a strange observation to find a quantity of perfectly fresh-looking horse-dung and green reeds cut into straw under the débris of the small room or passage immediately adjoining the once heavily barred entrance to the quarters. The space measured only about seven feet square, and could barely have allowed the } cxpuxrr AMENITIES OF DESERT SPRING 157 horse to turn round. Such tight quarters recalling life on board ship were the best illustration of the need of shelter which this inclement climate imposed even for animals. What with the heat and glare, the constant irritation inflicted by the clouds of mosquitoes and other insects, and the saltiness of the water, we were all suffering acutely from the amenities of this desert climate in spring. Little wonder that the task of keeping our Chinese labourers at work, all opium-smokers and of deep-rooted wzs zxerizae, had grown more and more difficult even for the unfailing tact and good-nature of my Chinese helpmate, Chiang-ssti- yeh. At last by May 8th, when supplies were beginning to run low, owing to a fresh convoy having failed to reach us, I was obliged to let most of them depart. But by that time the exploration of the extreme end of the Lzmes had been completed. The Surveyor had rejoined me three days earlier from survey work about Khara-nor, and I had despatched him with Naik Ram Singh and other mounted men to the south-east, to reconnoitre the ground there, and if possible push across the sand-covered hill range to the route leading west of Nan-hu. There or at So-mo-to, a little hamlet nearer to the foot of the mountains, fodder would become available for the ponies. But by the evening of May 8th Rai Ram Singh returned, after trying marches across the barren waste, having been effectively baffled when nearing the great sandy range by closely packed dunes which the _ ponies could not surmount at this season. I was glad he gave up the attempt in time and brought back ponies and _men in safety. I was not altogether sorry to find myself thus obliged to make my way back to Tun-huang once more along the Limes. There were a number of smaller ruined stations previously sighted eastwards remaining to be explored. _ But though interesting finds of records on wood, rags of I _ patterned silk, etc., rewarded our search, I need not stop now to describe it. Nor can [ do more than allude here to what I was able to observe about the river _we discovered flowing out of Lake Khara-nor, and about 158 RECORDS FROM WATCH-STATION cz. uxm the natural barrage which accounts for the formation of the latter. Already I had sent off in advance to Tun-huang my Indians and every man who could not be put to digging, After days made very trying by torrid heat and violent gales in turn, I was at last free to follow them with a good conscience. It was, indeed, high time to return to the oasis, The springs upon which we depended had always to be searched for by the marsh edge, and then wherever we camped the air was thick with mosquitoes and the ground swarmed with creatures equally bloodthirsty. So it was not without a feeling of physical relief that, on the evening of May 14th, I saw Lake Khara-nor disappearing behind me like a big sheet of chrysoprase colour, with the opposite shore merged in haze. A little earlier that day I had picked on marshy ground by the roadside my first flower of the season, a specimen of that hardy small iris-like plant which I knew and loved so well from the river banks of Kashmir. It gave welcome assurance that there were still pleasures left for the eye even in this dreary region. Next day a long hot ride, diversified only by mirages playing over the glittering gravel waste, and with the temperature ranging somewhere about 150 degrees Fahren- heit in the sun, brought me back to the oasis. I shall never forget the delight of the first short rest I took on its very edge under fine shady elms, with a field before me where the blue of wild irises mingled in profusion with the bright green of young corn. How beautiful it all looked to my parched dust-filled eyes! Then I thought of what summer, when it came in full earnest, would be like by the desert border, and felt doubly elated in my heart at all the antiquarian spoil I was carrying back from its long-forgotten ruins. That evening my little tent stood once more in the familiar old orchard of Tun-huang, with peach and pear — trees close by to sprinkle it with their fresh blossoms. GH-A\P TE IR ides LY RETURN TO THE ‘THOUSAND BUDDHAS’ I HAD scarcely returned to the shelter of Tun-huang from the fascinations and trials of the ancient desert border when my eyes began to turn eagerly towards the cave temples of the ‘Thousand Buddhas’ at the foot of the barren dune-covered hills to the south-east. It was the thought of their sculptures and frescoes which had first drawn me to this region. But since my visit to the site in March, and the information then gathered about the great collection of ancient manuscripts discovered in one of the temples, the antiquarian attraction of the sacred caves had, of course, vastly increased. Eager as I was to commence operations at once, I had to contain myself in patience. Just after my return to Tun-huang the annual pilgrimage to the shrines commenced, and it did not need the polite hints of my Amban friends to convince me that this was not the best time for a move to the site. The great féte, a sort of religious fair, was said to have drawn thither fully ten thousand of the pious Tun-huang people, and from the endless string of carts I saw a few days later returning laden with peasants and their gaily decked women-folk, this estimate of the popular concourse seemed scarcely exaggerated. I knew enough of Indian Tirthas to realize that such an occasion was better for studying modern humanity than for searching out things of the past. So my start had to be postponed for five days. | ‘ound plenty to keep me busy in the meantime, what with adjusting accounts in that excruciatingly primitive currency of uncoined, and often far from pure, silver; with repairs 0 be effected in tents and kit, which those last weeks in 159 160 THE ‘THOUSAND BUDDHAS” caiza@ the wind-swept desert had tried badly; and last but not least with a mail, for the despatch of which a petty trader’s departure for Khotan by the route through the mountains offered a welcome opportunity. Nor, to confess quite frankly, did I feel sorry for the short spell of physical ease which this delay gave me.: The oasis looked its best now, with its fields beautifully smooth and verdant like a well- kept lawn, and with little blue irises growing everywhere by the roadside. The trees, too, mostly elms, were now refreshing to look at with their shady foliage; and the scattered homesteads which they sheltered seemed more than ever substantial in spite of the happy-go-lucky ways of these honest Cathayans. If they were people hard to keep at work, especially in the desert, they were yet, when about their own fields and farms, jovial folk to talk to and wonderfully well mannered. Wang Ta-lao-ye, the learned magistrate, who had greeted my safe return and my rich harvest of ancient documents with something akin to enthusiasm, assured me that the people of Tun-huang were getting to like my ways, how- ever strange they had seemed at first. Curiously enough, our queer set of slum-dwelling coolies proved mainly responsible for this change. Though they had given much trouble by their hopeless indolence and their constant desire to desert, once safely returned from the wilds they appeared to have done their best to give us a good name, by stories about paternal care in administering rewards and medicine; about rations unfailingly provided and generously ignored in accounts, and the like. Of course, I knew well that most of this unhoped-for credit had been earned through Chiang-ssit-yeh’s kind-hearted care and inexhaustible patience with the humblest. Never did I feel so strongly the old-world charm of this sleepy frontier of true Cathay as when I retired to the | famous sanctuary of the ‘Crescent Lake’ for a days peaceful writing. It lay hidden away amidst high sands — beyond the southern edge of the oasis and about three | miles from the town. For a desert wanderer there could | be no more appropriate place of rest, I thought, than this — delightful little pilgrimage place enclosed all round by | geixiv LEMPLES AT ‘CRESCENT LAKE’ 161 sand-ridges rising to over 250 feet in height. There was the limpid little lake, of crescent shape and about a quarter of a mile long, which has given to the locality its name and its sanctity. Such delicious springs issuing between dunes of huge size and ever safe from being smothered by them, would have been worshipped in India as the resi- dence of some great ‘ Naga’ or spring deity. No doubt, more than one Indian Buddhist passing through Tun- huang to China must have felt strangely at home here. Near the eastern point of the crescent, where the lake has its outflow, there was room for small meadows where our ponies revelled in juicy grass such as they had never before tasted. The southern shore of the lake was occupied by a number of picturesque modern temples, rising on terraces from the water’s edge and decorated with a queer medley of Buddhist and Taoist statues and frescoes (Fig. 185). Just in front of them and across the lake rose the famous resounding sand-hill, often mentioned ‘in old Chinese records, about which the curious may read learned notes in Sir Henry Yule’s translation of Ser Marco’s book, where it deals with ‘the Province of Tangut.’ I had ridden out to this secluded spot to enjoy undis- _turbed work ‘in Purdah,’ as our Anglo-Indian phrase runs, ‘But Chiang, my only companion, though he had brought ‘out work too, could not forgo the temptation of climbing to the top of the huge dune in his dainty velvet boots, \just to make the sand slide down from there and hear the ‘miraculous rumbling’ it produced. It was quite in keeping with his usual keenness to get at ‘real truths.’ We all duly heard the faint sound like that of distant carts ‘rumbling, and Chiang felt elated to put it down in his Journal. There was no other noise to disturb me all day. In ‘Spite of its popular favour as attested by votive inscrip- tions in plenty, the whole place was deserted for the sake ‘of the ‘Thousand Buddhas’’ féte-day. Only one discreet figure moved about, a quaint, good-natured old priest, who remembered gratefully the little present I had left when paying my first visit here in March. As I sat writing in the shady spacious hall, and watched him VOL. II M ae THE ‘THOUSAND BUDDHAS” ex ul coming from time to time to have a good look at me, and then again retreating with the most silent of footsteps, it seemed as if familiar priestly figures from Indian shrines and tombs were trying to call me back to scenes and times of old. When we sat down together in the evening to share tea and a simple repast a/ fresco, 1 told Chiang how grateful I felt to have spared this day of peaceful seclusion from the work awaiting us at the ‘Thousand Buddhas.’ He, too, had greatly enjoyed this outing; but full of mundane humour and keenly alive to any piece of sacred super- stition or folklore, he then gave an additional reason to justify the delay. He had picked up the story, that invariably after the annual féte the gods and local genii of Ch’ien-fo-tung sent a violent dust-storm to clean the sacred place of the refuse left behind after the multitude’s pious picnic. So it would be better to wait until the divine sweeping was done. Considering the gales which rarely cease blowing down the Su-lo Ho valley for more than a couple of days at a time, one could readily take it for granted that the popular expectation would not be belied. So it came about that when on May 21st I marched off to the sacred caves, we moved in a dust haze left behind by a storm of the previous night. Luckily these gales, sweeping westwards from An-hsi and down the desert ranges of the Pei-shan, have been at work with such constancy since early ages that there is comparatively little fine sand left for them to play with. So what in Khotan would have meant a choking Buran atmosphere was here only a pleasant protection against the heat of the sun. For, of course, move Stuzco, we had to wait until nearly mid-day before carts were brought up and despatched with our heavy baggage. The two or three miles of ground we traversed to the edge of the oasis looked delightfully green, with young corn covering the big fields, and wild irises of bright blue still hugging in plenty the sides of the deep sunk roads, The cultivators seemed to find it hard to return to humdrum labour after their outing, and in front of most of the homesteads the women sat about in happy idleness still ONVOH-NOL «-aNVI LNAOSANO, LV LYNOD AIMWaAL ‘Sst Ne oS 9) CH. LXIV APPROACH © SACRED SITE 163 wearing their bright holiday dresses. I could never look at the poor crippled feet of these peasant women without a feeling of amazement at the power of fashion which here bows into cruel subjection even the humblest folk. What their well-to-do sisters in town may more easily forget must be aconstant hindrance to these women of the country-side. It needs the iron grip of an ancient civilization to assure obedience to such conventions of self-inflicted torture. Once beyond the edge of the oasis there was no trace of life stirring on the broad track which skirts the grey gravel waste at the foot of the sombre hill range, though thousands had passed here so recently. It remained the same when we turned into that strangely impressive desert valley and approached the point where the sombre conglomerate cliffs begin to be honeycombed with the gaping mouths of cave-temples big and small (Fig. 158). Long ages ago the little stream had carved out the valley, when there was still moisture to clothe these forbiddingly barren hills; but now it was dying away just here by evaporation on the thirsty rubble-filled bed. There was indeed gratifying shade beyond, where the narrow fringe of irrigated ground masked with its elms and poplars the approach to the main temple caves (Fig. 186). But other- wise the scene was not changed since my first visit in March, and I soon felt assured that the sacred site had once more resumed its air of utter desolation and silence. There were special reasons for me to appreciate this assurance. The months passed since my arrival had provided abundant proof of the zeal with which the good people of Tun-huang remained attached through all vicissitudes to such forms of worship as_ represent Buddhism in the queer medley of Chinese popular religion. It scarcely needed the experience of the great annual fair just past to make it clear to me that the cave- temples, notwithstanding all apparent decay, were still real places of worship ‘in being.’ I knew well, therefore, that my archaeological activity at them, as far as frescoes and sculptures were concerned, would, by every con- sideration of prudence, have to be confined to the study of the art relics by means of photography, drawing of 164 THE ‘THOUSAND BUDDHAS’ CH, LXIy plans, etc., in short, to such work as could not reasonably arouse popular resentment with all its eventual risks. Yet it was useless to disguise the fact from myself; what had kept my heart buoyant for months, and was now drawing me back with the, strength of a hidden magnet, were hopes of another and more substantial kind. Their goal was that great hidden deposit of ancient manuscripts which a Taoist monk had accidentally discovered about two years earlier while restoring one of the temples. I knew that the deposit was. still jealously guarded in the walled-up side chapel where it had been originally dis- covered, and that there were good reasons for caution in the first endeavours to secure access to it. What my sagacious secretary had gathered of the character and ways of its guardian was a warning to me to feel my way with prudence and studied slowness. It was enough that Chiang had induced Wang Tao-shih, the priest, who had come upon the hidden deposit, to await my arrival instead of starting on one of his usual tours in the district to sell blessings and charms, and to collect outstanding temple subscriptions. I was glad that,very first evening to find good quarters for all my people, as well as for the heavy baggage which had been brought up from its former place of storage at Tun-huang. Fortunately the only two dwellings which Ch’ien-fo-tung boasts of, apart from its caves, were unoccu- pied, except for a fat jovial Tibetan Lama who had sought. shelter here after long wanderings among the Mongols of the mountains (Fig. 190). In one of the courts my Indians found rooms to spread themselves in, and the Naik a convenient place to turn into a dark-room. In the other my Muhammadan followers secured shelter under half-ruined roofs of outhouses, while a hall, still possessed of a door and trellised windows, was reserved as a safe and discreet place of deposit for my collection of antiques—and its eagerly-hoped-for additions. Better still, the narrow strip of cultivation extending in front of the caves for about half a mile (Fig. 186) offered just one little plot, grass-covered, where my tent could be pitched | under the shade of some fruit trees. My Mandarin friends “LS M WOW NAYS “ALIS .SVHGGNH ANVSNOHL, AO SHTIAWALAAVS NIVW HLIM SAAIIO ‘981 * - = > i + a © * 4 ' iB ~ ' , = = . a r * —< -see * > aa SS ee ee . . cexnixiv WANG, THE TAOIST PRIEST 165 had insisted upon my taking along a petty officer and some soldiers of the Tun-huang levy corps. For them and the Ssii-yeh there was ample room in the big verandahs and halls built in front of the large caves just opposite our camping-place. Chiang himself had a delightfully cool room at the very feet of a colossal seated Buddha reaching through three stories, and with his innate sense of neat- ness promptly turned it into quite a cosy den with his camp rugs. Later on, when it got hotter, I myself used the anteroom of another restored grotto close by for a ‘Daftar.’ After what we had gone through during the desert winter and spring we had all reason to feel our- selves in clover, in spite of the somewhat salt water of the stream, which the Surveyor grumbled at much and accused of reviving his rheumatism. Next morning I started what was to be ostensibly the main object of my stay at the site, the survey of the principal grottoes, and the photographing of the more notable frescoes. Purposely I avoided any long interview with the Tao-shih, who had come to offer me welcome at what for the most of the year he might well regard his domain. He looked a very queer person, extremely shy and nervous, with an occasional expression of cunning which was far from encouraging (Fig.187). It was clear from the first that he would be a difficult person to handle. But when later on I had been photographing in one of the ruined temple grottoes near the great shrine restored by him, where the manuscripts had been discovered, I could not forgo a glance at the entrance passage from which their place of deposit was approached. On my former visit I had found the narrow opening of the recess, locked with a rough wooden door; but now to my dismay it was completely walled up with brickwork. Was this a precaution to prevent the inquisitive barbarian from gaining even a glimpse of the manuscript treasures hidden within? I thought of the similar device by which the Jain monks of Jesalmir, in their temple vault, had once attempted to keep Professor Biihler from access to their storehouse of ancient texts, and mentally prepared myself for a long and arduous siege. 166 THE ‘THOUSAND BUDDHAS’ CH. LXIV The first task was.to assure that I should be allowed to see the whole of the manuscripts, and in their original place of deposit. Only thus could I hope to ascertain the true character and approximate date of the collection which had lain hidden behind the passage wall. In order to effect this Chiang had been despatched in the morning to another restored cave-temple where the priest had his quarters, and proceeded to sound him in _ confidential fashion about the facilities which were to be given. It proved a very protracted affair. Backed up by the promise of a liberal donation for the main shrine, the Ssi-yeh’s tactful diplomacy seemed at first to make better headway than I had ventured to hope for. The saintly guardian of the reputed treasure explained that the walling up of the door was intended for a precaution against the curiosity of the pilgrims who had recently flocked to the site in their thousands. But evidently wary and of a suspicious mind, he would not yet allow himself to be coaxed into any promise about showing the collection to us as a whole. All that he would agree to, with various meticulous reservations, was to let me see eventually such specimens of the collec- tion as he might conveniently lay his hands on. When Chiang, in his zeal momentarily forgetting the dictates of diplomatic reticence, was cautiously hinting at the possibility of my wishing, perhaps, to acquire ‘for future study’ one or other of those specimens, the Tao-shih showed such perturbation, prompted equally, it seemed, by scruples of a religious sort and fear of popular resent- ment, that my sharp-witted secretary thought it best to drop the subject for a time. But after hours of such diplomatic wrangling he did not leave the priest’s smoke-filled chapel and kitchen combined without having elicited an important piece of information. Statements heard at Tun-huang seemed to indicate that the great find of manuscripts had been reported at the time to the Tao-t’ai at Su-chou and thence to the Viceroy of Kan-su. Expression had been given also to a belief, of which we had no means of testing the foundation, that the latter had given orders for the transmission of specimens and for the safe keeping of the whole collection. If such * 187. WANG TAO-SHIH, TAOIST PRIEST AT ‘THOUSAND BUDDHAS’ “St OS TUN-HUANG. wa) onixiv VISIT TO TAO-SHIH’S TEMPLE 167 injunctions had really been issued and, perhaps, an official inventory taken, things would necessarily, from our point of view, become far more complicated. Fortunately Chiang’s apprehensions on this score were dispelled by what the priest, turning talkative at times like many nervous people, let drop in conversation.