Next day we made an early start and by eight o'clock the camels were watered and loaded. It usually took us some two hours to start our large caravan in marching order.

After crossing the gravel plain in a southerly direction, we entered low sandy hills that stretched for more than five miles. These hills bordered the gravel plain on the southeast and south. South of this belt of sandstone hills and sand dunes lay a broad sandy plain covered by loess and deresün grass (Lasiagrostis splendens). We found here several Mongol encampments belonging to the same khoshun as the Yum-beise Monastery. As in all frontier districts of central Asia, the population of the tents was extremely varied. We came across Torguts from the Mongolian Altai, Khalkha Mongols from at least half a dozen khoshun or tribal divisions and several Amdo Tibetans, who had settled in the district. Some of them proved to be former lamas of the great Kumbum Monastery in the Sining district of Kansu Province.

After crossing the grass-covered plain, we came to a river flowing northwest by southeast. It was a tiny stream on whose banks were pitched a number of Mongol tents. Here we were offered a fine white camel for sale. Unfortunately the animal was too young for a long route but the Mongols insisted on our buying it. White camels are popularly considered to be the best present to offer to one of the great Lama Pontiffs of Tibet. We refused, fearing that the animal was too young and would not stand the long journey to Tibet. We had to take some water from the river for we were told that there was no water on the route.

After fording the river we entered into an intersected hilly region of low relief. Here and there one could see outcroppings of granite, porphyry and other rocks. After a long march of twenty-five miles, we camped in a circular depression among hills, well sheltered from the frequent winds that level the landscape. So far, the weather had been mild and the days even warm. The locality of our camp bore the name of Khara Boro, and was said to receive a considerable amount of rain during the rainy months of the summer. Naturally we had no opportunity to test this statement of our guides. The numerous dry basins of miniature lakes found between the hills seemed to indicate a large amount of rain at some season of the year.

After leaving Khara Boro, the route crossed several sandy ridges which were frequently preceded by belts of sand dunes. All of them stretched northwest by southeast. Between the ridges lay sand and graveled latitudinal valleys covered by shrubs of saksaul. The winds had gathered small hummocks of sand and loess round the plants and this aspect of the country forms one of the greatest obstacles for a motor road across the region.

The day's march was unusually long for we had to reach a well situated in the mountains which rose on the south. Toward seven o'clock we halted for the night on a small gravel plateau sheltered by low rolling hills. Toward the north and northeast the plateau was open and we could see the rugged outline of the distant Bain Ündür Mountains. The place of the camp bore the significant name of Dzogo-usu, which means "the tasted water." Dzogo is a polite Mongol expression meaning "to partake." The formal term is used because the Dalai Lama had camped here during his memorable flight from Lhasa in 1904. His Holiness followed the same route but in the opposite direction. The local Mongols remember this important event and have given the former camps of the Dalai Lama fanciful names to distinguish them.

From Dzogo-usu the trail ascended a low ridge south of the camp. From this point we descended into a flat valley which merged into a dry river bed. The hills became higher and we admired the rocky crags and weathered basalt and granite outcroppings. It was our first hot day, for the breeze from the open spaces of the desert could not penetrate the narrow river gorge. Sharp stones and bowlders blocked the trail in several places. It was surprising to see camels march over such heavy ground without any sign of discomfort. We noticed along the route several deep pits sunk into the gravel of the river bed. These were said to be former gold diggings, forsaken now for many years. They dated back to the period antedating the brigand activity of Ja Lama some ten or twelve years before.

By sunset we emerged from the hills and camped on the northern border of a gravel plain which was thickly overgrown with shrubs of saksaul. We were about to pitch our tents when suddenly a cold breath of wind followed by distant roars in the mountains northeast of the camp announced the approach of a Mongolian windstorm. A few moments later, the storm was raging over the camp. Happily it was of brief duration and swept southwest into the plain, raising clouds of sand. The night passed quietly although we had taken all precautions to fix our camp as firmly as possible. These sand and windstorms are always a great calamity for travelers. It is almost impossible to protect one's self from them. The force of the hurricane breaks poles, tears tent flaps and covers the tent floor and bed with thick layers of sand.

Next day we again moved south following the dry bed of a river that once flowed from the mountains into the plain to the south. The days were decidedly hotter and the camels began to lose patches of hair. After eight hours' march across a gravel plain, we halted for the night at a desert well of brackish water with a foul taste. All around rose low sandhills. Far to the south stood the mountains of Shara-Khulusun, a mysterious place which all of us were eager to visit. According to our guides, Shara-Khulusun was a narrow wooded gorge with a small stream of mountain water running through it. Many were the stories told about this lonely oasis situated in the heart of the Gobi Desert.

The Gobi between Yum-beise and An-hsi chou represents a succession of mountain ridges of crystalline rocks intersected by desert plains. Most of the mountain ridges belong to the Altai Mountain system, which stretch northwest by southeast across the arid desert region situated between the southern branches of the Khangai Mountains and the eastern offshoots of the T'ien Shan and Barkõl Mountains. I am confident that the numerous canyons that flank the desert ridges would give ample opportunities for fossil study if properly explored.

The country northeast of our route—the region of Sain-Noin, the country around Tsagan-nor, the eastern offshoots of the Mongolian Altai, the mountain ridges of Baga Bogdo, Artsa Bogdo, and Gurbun Saikhan—has been thoroughly investigated from the point of view of geology and paleontology by the Third Asiatic Expedition of 1922-23 under the leadership of Dr. Roy Chapman Andrews. So far, the southwestern Gobi has attracted but scant attention. Only a few of the European travelers touched this desolate region and much remains to be done to reconstruct the geological past of this desert country. Portniagin and some of our Mongols had on several occasions crossed the east Mongolian Gobi from Khalgan and Pao-t'ou, but all of them claimed they had never seen a country so barren and desolate as the southwestern Gobi.

Thus far, it had been our usual practice to start in the afternoon, letting the camels graze on the saksaul shrubs found in the mountains, during the entire morning. On May 5 we decided to make an early start in order to reach the Oasis of Shara-Khulusun and give our camels plenty of time to graze in the forest of the oasis. The day proved exceptionally hot. Far toward the south, the barren mountains of Shara-Khulusun rose in a dust-haze atmosphere. For the first two hours the camel trail led over broad patches of sand-backed mud covered with shrinkage cracks and intersected with strips of gravel. The level plain was cut by a low ridge of limestone and sand dunes. Our progress between hills was very slow, for the ground in the shallow depressions was muddy and the camels often sank into the sand.

Leaving the sand ridge behind us we rode into a broad sand valley gradually ascending toward the southwest. For some six miles, the trail continued along a dry river bed with large accumulations of débris from the neighboring cliffs of sandstone. Then it again emerged on a gravel plain slowly rising toward the Shara-Khulusun Mountains, now clearly silhouetted against the sky. For a considerable distance, the gravel plain was only the gigantic dry bed of some mighty torrent that once swept from the mountains over the plain. Bowlders, débris, and dried tree trunks blocked the passage. The Mongol guides assured us that even nowadays, during the period of summer rains, the water of the Shara-Khulusun-gol rose and a powerful stream swept down on the plain. The summer rains in the Gobi region generally occur in July and August and are devastating. An excessive rainfall of several days will form and swell into a torrent, which carries débris down into the plain and cuts deep gorges in the mountain. Several months later we experienced such a terrific torrent, preceded by several rainy nights, in the arid mountains of Nan Shan.

Toward four o'clock in the afternoon when men and animals were completely exhausted by the unusual heat, we noticed several dark spots at the foot of the mountains and at the entrance into a narrow gorge hidden behind a long spur. Someone in the caravan column cried out "Trees!" We could not believe our eyes, for most of us were firmly convinced that at best, we would see only miserable juniper shrubs. But there in the distance were actual trees, desert poplars (Populus euphratica) that grew along the banks of the river. How refreshing it felt to enter the coolness of the forested gorge, and camp on the green meadows.

The cameleers quickly unloaded the camels and drove them off into the thick bushes of tuya, a little farther up the stream. We crossed the stream in search of a shady camping ground and found a delightful site for a camp in a grove of poplars, the best camping ground that we had had since our departure from Urga. I may add that it was by far the pleasantest of all our camps across Tibet up to the enchanting rhododendron forests of Thangu in Sikkim.

We decided to pitch our tents in the grove and returned to the other bank of the river. The camels carrying tents were not yet unloaded but the cameleers violently protested against taking the camels across the river. According to them, the proper camping place was on the near side of the river and not on the other. It is surprising how these caravan men grow accustomed to one particular place and will always camp on one and the same spot. After long arguing we forced the drivers to bring the camels across under guard of half our men. The stream was almost dry and presented no difficulties for communication between the two camps.

The Oasis of Shara-Khulusun lies in a narrow gorge watered by the tiny Shara-Khulusun-gol, which swells into a formidable stream in the rainy season. Poplar groves grow along the river banks which are covered by a thick growth of reeds, from which is derived the name of the place, "The Yellow Reed." The oasis remains from a period when the region received more rainfall and the conditions were favorable for cultivation. Numerous dead tree trunks and tree stumps indicate that in past times the oasis occupied a far greater area and that the forest stretched far into the gravel plain and along the foot of the mountains.

The oasis is a favorite place for big game and birds. Besides the Mongolian antelope (Antilopea gutturosa) or dzeren, we saw numerous wolves and several wild asses or khulans (Equus hemionus). Hares were plentiful (Lepus tolai). The bird kingdom was represented by several vultures (Vultur monachus) huge black birds that flew over high crags which sheltered the gorge. We also observed several families of geese (Anser anser) and turpans (Casarca casarca). The oasis was used by aquatic birds as a station on their yearly migrations.

The mountain range in which the Oasis of Shara-Khulusun is situated presents several problems of interest in connection with the question of the Gobi climate and the amount of precipitations in the desert regions of inner Asia. General P. K. Kozlov during his expedition of 1901 discovered fine meadows, forests, and running streams in the highlands of Atik Bogdo, which seems to be a prolongation of the Shara-Khulusun Ridge. The British explorer, Douglas Carruthers, during his expedition in the Karlik-tagh Mountains east of Barkõl, discovered larch and pine forests in the southern and eastern offshoots of the Karlik-tagh Range. I questioned our Mongol guides and traders who had followed the route to Ku-ch'eng and all spoke of the rugged ridge that stretches from Atik Bogdo toward the Karlik-tagh Mountains. It would be a fascinating task to study the life zones of the inner Gobi and its oases, lost in the immensities of the sand and stone deserts.

Higher up the gorge of Shara-Khulusun, we discovered several ruined houses and on one of the mountain spurs a small Chinese miao or temple. In several places the river banks were occupied by terraced fields. Our Mongol guides told us that this was formerly a small Chinese settlement. The Chinese used to come here from An-hsi to grow the opium poppy. Now for several decades the valley has been abandoned.

The Shara-Khulusun gorge is situated at the junction of two important caravan routes of central Asia: the Yum-beise – An-hsi route and the Koko-khoto – Ku-ch'eng route, connecting China with the lands of Chinese Turkestan and Jungaria. Besides these two routes another secret route, often followed by the Etsin-gol Torguts, runs through the gorge and joins the Koko-khoto – Ku-ch'eng road.

Situated not far from the Mongolian border, the gorge was always a favorite haunt of robbers. Ja Lama maintained outposts there to look after the caravans coming from China, Tibet, and Mongolia. Even after Ja Lama's death, the gorge was still visited by robber bands. Only a month before our passing a big camel caravan en route for Ku-ch'eng was plundered in the gorge and one of its drivers killed. Our Mongol guides advised us to be careful and to keep watch in the night. The great drawback of our position was that in case of an attack we would have to defend two camps instead of one. However, this very defect saved us from killing innocent people. The following account will make that clear.

It was already late in the evening, about nine o'clock, and completely dark, when suddenly one of our Mongol camel drivers, who was watching camels in the gorge, rushed into camp with the disquieting report that armed riders had been seen in the gorge. We had to find out the identity of the men at any price. We were still discussing the measures to be taken when suddenly a loud rifle shot echoed in the gorge, followed at a close interval by another. Our sentry on the other bank of the river immediately gave the alarm and everyone rushed to arms. The sentry had clearly seen in the darkness two riders armed with rifles, one on a white horse, come out of the bushes and fire at the camp. What were the intentions of the riders? It was evident that they were attacking the camp and that any moment we could expect wildly yelling men to jump out of the bushes and rocks, surrounding the camp. The leader of the expedition instructed me to assemble the men of the expedition guard and to insure the safety of the camp.

It was impossible to protect the two camps at once, and I therefore decided to leave the baggage camp under the guard of four of our rifle-men and with the rest of the men to occupy a line south of our second camp. This position allowed us not only to protect the two camps by rifle fire but made it easy for us to charge the enemy in case they would venture into the open to attack our camp with the heavy baggage.

We spent long, tense moments, for the firing did not continue, but from the noise we could infer that a large force of men and animals were marching through the gorge and nearing our camp. It was hard to keep the men from firing at every dark spot that appeared in the distance. It was imperative to find out who and where were the assailants. I decided to send a reconnoitering party supported by a group of our riflemen. Portniagin was instructed to cover the camp with the rest of the men and I with two of our Mongols occupied a convenient position in the bushes on the bank of the stream, not far from the point where the trail from Koko-khoto approached the river bank. Our Tibetan volunteered to reconnoiter and disappeared in the bushes. A long silence reigned in the gorge. Suddenly we heard the barking of dogs. The presence of dogs was rather disconcerting. I went back to camp to report on the matter and Mrs. Roerich, who had also heard the dogs, was sure that it was a caravan or party of pilgrims. But why did they fire at such close range on our camp?

The Tibetan returned after a prolonged absence and related his experiences with the strangers. He found their camp higher up the gorge. It was a large caravan from Koko-khoto which was traveling to Ku-ch'eng and Ürumchi. On his approach to the camp, all the men sitting round the camp fire sprang up, extinguished the fire and received him in complete silence. When he appeared in their midst, they begged him not to harm them, for they were persuaded that he was a member of a large band of brigands camping in the gorge. They even promised him a large reward in money or goods. They were deadly afraid and he had some difficulty before he was able to explain to them that we were a scientific expedition. At last, they believed him and told him that they took us for a robber camp and that the heads of their caravan, a Chinaman and a Tartar from Kulja, rode toward our camp and fired the shots to scare us! This is the usual maneuver of Chinese caravans. In dangerous places they usually fire in the air, or at suspicious looking persons in order to scare them. It was lucky that we all were in the second camp for if we had been in camp No. 1, we would certainly have returned the fire and possibly caused losses. We all went back to our tents happy that the whole affair had turned out so well. Many of our men were a little disappointed — they preferred to fight real brigands and not scared traders.

Early in the morning the leaders of the Chinese caravan paid us a visit and apologized for having fired at us. One of them was a Chinese tradesman from Koko-khoto, the other a Mohammedan from Kulja. Both were very much afraid of robbers on the way. The caravan belonged to the American firm of Brenner Bros. at Tientsin.

We remained at Shara-Khulusun most of the day and gave our camels plenty of time to graze. At five o'clock in the evening we broke camp and continued our journey through the gorge. For the first three miles the gorge was narrow. On either side rose granite rocks, gneiss and porphyry. After three miles the gorge broadened and the track crossed several banks of gravel and débris. The vegetation disappeared soon after we had left camp and for the rest of the march the trail followed an arid country. We met several abandoned camels left behind by the Chinese caravan. The animals had sore feet and were unable to follow the caravan.

Soon after we left the gorge, and were traveling across a sandy plain, complete darkness enveloped the surrounding country. We marched in darkness over a flat plain with low hills on both sides of the trail and it was impossible to determine the real character of the country. At one o'clock in the morning we reached a camping ground situated by a brook of fresh water, called Bilgekh. From this point the ordinary caravan route goes in a southwesterly direction but the circuit trail which we were to follow branched off toward the south.

We spent considerable time pitching our camp. Toward three o'clock the tents were pitched and all were able to refresh themselves with tea.

The following are fragments from my diary describing the passage of the expedition through the region of Ma-tzu Shan.

Saturday, May 7, 1927. The day promises to be hot. Clouds hover over the horizon. Everywhere sand—sand dunes that stretch far toward the south. We started preparations about noon. The Mongol drivers are watering the camels, fixing the pack saddles and smearing grease over the bloody wounds on the animals' flanks. The hair rolls up under the saddle and the saddle wears off the skin by constant rubbing. Wounds develop, which soon become open sores and healing becomes almost impossible. About three o'clock in the afternoon the whole caravan in three columns begins the march toward the south. After crossing the belt of sand dunes, which usually cover the approaches of the mountain ridges of the Gobi, the caravan comes on the immense plain of the gravel-covered Gobi. The black surface of the stone desert sparkles with deep opal shades. Mirages flicker in the burning air—lakes and islands covered by vegetation.

Ahead of us are three days of waterless desert, and the men and animals of the caravan gravely move along the hardly visible trail. The preservation of the camels' strength is the chief anxiety during these long marches across waterless deserts. Otherwise peril becomes imminent. We cross a narrow gorge blocked by huge bowlders, a low sandy ridge of rugged mountains known under the name of Khan-in-nuru, which in common with most of the mountain ridges of the southwestern Gobi stretches from the northwest to the southeast. On both sides of the path rise sharp granite rocks which interchange with dark, almost black masses of basalt. The summits of the mountains are crowned with fantastic formations of weathered rocks which assume the shape of inaccessible castles or strongholds that guard the caravan route.

Again stone desert—black gravel Gobi. The camels move with their cadenced pace, gravely turning their heads, as if searching for the end of the stone desert that cuts the soles of their feet. Along the way we meet a great number of phallic images, laid out with small white stones on the surface of the gravel. The Mongol drivers assure us that the creators of these images were Chinese traders.

We try to cover as much ground as possible. How majestic are the dawn and sunset in the desert! Suddenly the shades of the sunset sparkle with deep purple and the immense plain glows with a purple glare. A few seconds more of intense glitter and the colors fade away and the vast expanse of the desert plunges into a violet darkness. The stars appear on the wonderfully dark sky of central Asia. In the exceptional dryness of the atmosphere they seem to be extraordinarily bright—like thousands of lamps burning round some unseen altar. Toward eleven o'clock the moon rises and lights with its soft bluish light the desert—the dead heart of Asia, covered with the dismal black stone. Toward midnight the caravan halts, tents are pitched and the camels tied up close to the camp.

A burning hot day follows the quiet starry night. It is difficult to imagine the burning heat that emanates from the stony surface of the desert. We decide to start earlier notwithstanding the heat, in order to reach sooner the promised well. The camels, which during winter are able to travel six or seven days without a drop of water, in springtime suffer badly from thirst. They become drowsy and the air resounds with their drawling cry. Their eyes become watery and dull.

We start about two o'clock in the afternoon. A hardly visible trail goes southward. On all sides stretches the same black kingdom of stone. We march the whole of the afternoon and evening. About midnight we halt to give rest to the tired animals. No tents are put up and all spend the night under the open sky. At sunrise, we load our camels once more and move ahead. Some of the camels become weak and have to be watered from our flasks. Toward noon the heat becomes unbearable and everyone in the caravan counts the hours that bring us nearer to the rocky ridge, which stretches far to the south in the misty atmosphere of the burning desert.

At four o'clock, after a trying march over difficult ground, we reach the well called Altan-usu. The place is situated at the foot of a rugged mountain ridge with numerous dry river beds. Some twenty years ago Chinese gold diggers from Kansu used to come here and work the gold that is found in the sand of the dry river beds. Ruins of Chinese huts show that the place was densely populated. Now everything is deserted. Our Mongols tell us that the place has an evil reputation for murder and other crimes committed here. The Gobi Mountain ridges abound in auriferous sand and in the past Chinese gold diggers used to visit these remote places. After the occupation of the region by Ja Lama gold digging was discontinued.

Altan-usu and the mountains south of it are often visited by brigands and the bands of Ja Lama are still active in the numerous narrow gorges. The country is highly intersected and the intricate system of rugged rocky hills reminds one of the Afghan border. The country abounds in hiding places for brigands, used by all sorts of outlaws who infest the vicinity of the trade routes.

We decide, therefore, to take all possible precautions and to send out reconnoitering parties ahead of the main caravan column. From now on the camp has to be protected by double sentries during the night and arrangements are made to have at least half of the guard ready for any emergency. This is effected by requiring half of the men to sleep armed and dressed. . . .

The next day (May 9) we gave our camels a brief rest, starting only at four o'clock in the afternoon. The path ascended a flat promontory and entered a narrow gorge. In some places we found weathered grass and traces of horses' hoofs. Evidently men were near. Our guides often ascended nearby hills and attentively examined the country.

Our reconnoitering parties searched the gorge ahead of the caravan column and reported by signs if the passage was clear. Some of the caravan men who had no rifles carried tent poles put into rifle covers. From a distance it looked as if the whole caravan was armed to the teeth. We passed a former encampment, traces of stone hearths, heaps of argal and horse dung. The dung was still fresh and it looked as if the encampment was moved only a few days before our passage. Our guides told us that in Ja Lama's time the gorge was guarded by a strong post of armed robbers.

We crossed a steep pass that leads across the Artsegin-nuru Mountains and emerged on a broad gravel plain. We halted at eleven o'clock in full darkness. As yet no water, although near the camp lay a dry river bed. On the south southwest rise the massive Ikhe Ma-tzu Shan, and its continuation Baga Ma-tzu Shan. The whole of this district was once the arena of the activity of Ja Lama and his bands. His former followers haunt the mountains of the neighborhood and caravans are often plundered, for the memory of Ja Lama still lives among his people. The Chinese authorities of An-hsi are unable to stop brigandage and prefer to leave things as they are.

Round our camp we found numerous empty cartridges, a reminder of the brigands' presence. Our Mongol lama guide said that the place was the scene of a big fight. Later on we found numerous carcasses of dead horses and camels in a hollow spot not far from our camp. A caravan was massacred here.

Portniagin and myself guarded the camp in turn. While walking round the camp, we discovered huge figures of stupas laid out with white pebbles on the black gravel surface. According to the Mongols, this was the work of the prisoners of Ja Lama. The place is called Su-khai-Bom and received its name from the juniper shrub that is found not far from it.

The night was quiet and we noticed neither men nor animals in the neighborhood.

We started in the afternoon. The ground became moist and we crossed large patches of land with a saline crust. The camels, feeling the presence of water, moved more quickly. All was silent, for the caravan bells were tied up as a precaution against possible encounters. We dispatched our reconnoitering party ahead to find a camping place for the night. The tall silhouettes of our men on their camels disappeared in the darkness. We knew that we were in the vicinity of the fortress of Ja Lama which was still occupied by some of his men. The caravan moved in a compact column and riders were deputed to protect the flanks.

Suddenly we saw a fire glittering in the distance. Was it a nomad encampment, or were those our men? The caravan halted and Portniagin and I rode on to reconnoiter. We soon heard the voices of our men guiding us in the darkness.

We decided to camp here for the ground was difficult to traverse during the night. A brook of fresh water was found and we were able to water our camels. In an hour's time, the small level patch of ground was covered with tents and camp fires glistened in the darkness, throwing a challenge to the fortress, which was said to be some distance from here.

During the night the camp was guarded by double sentries, and the rest of the expedition guard received strict orders to be ready for any emergency. The men slept with their rifles. The night was exceptionally dark because of clouds but the neighboring hills seemed completely deserted.

In the early morning we had the rather unexpected surprise of finding the dreaded castle of Ja Lama immediately south of our camp. It was quite out of the question to remain in camp without occupying the castle, for in case it was occupied by brigands, they could easily attack our camp. In the early morning Professor Roerich decided to reconnoiter the surrounding country and to occupy the fortress. Following his instructions, I ordered half of our guard to accompany Portniagin and me to the fortress. The men who were usually quite disciplined, made a flat refusal. They said that they were ready to fight Chinese, Tibetans, or Mongols, but they would never enter the fortress of Ja Lama or fight with his men.

All our persuasions were vain and we had to proceed alone. It was decided that we should make signs with our rifles from the top of the watch tower in case the fortress was found deserted. Firing on our part would mean that we met with some resistance. We quickly advanced and occupied the first watch tower, from where we could easily observe the fortress. It seemed completely deserted. We then entered the first courtyard. Dead silence. Neither dogs nor men, only heaps of rubbish left by former occupants. The second courtyard was also empty.

The castle itself stands on a low hill, its back to a low ridge that forms part of the Baga Ma-tzu Shan. The principal building, which served as residence of Ja Lama himself, is a two-storied square house with a flat roof having small tower-like projections on each of the four corners. Two courtyards surrounded by high brick walls lead to it. In the courtyards are situated the guard's quarters, stables, and storerooms. The house itself has a large door. On the ground floor is a large hall with a hearth in the middle. A small stone staircase leads to the second floor, where the private rooms of Ja Lama were situated. At the time of our visit, the fortress was in complete decay. The rooms and ceilings were blackened by fire. The windows had no frames and the condition of the floors was such that they could support only two or three men at a time. The courtyards were full of rubbish and dirt, with skeletons of cattle and dogs.

The fortress was protected by several concentric belts of walls with towers. The neighboring hills and the rocks dominating the fortress were crowned with watch towers, each of which has probably had a small garrison of its own. Outside the fortress walls heaps of refuse and stone hearths made it clear that in Ja Lama's lifetime the fortress was surrounded by a huge nomad settlement consisting of several hundreds of tents. Now everything is in ruin, but the Mongols say that Ja Lama's men still haunt the place using it as winter quarters. Professor Roerich made a sketch of this desert castle round which clusters a whole legendary cycle.

After investigating the rest of the fortress, and the neighboring watch towers, we returned to the first tower, and communicated by signs that the fortress was empty. On our way back we met our guard going toward the fortress. Now that they knew that it was empty, they decided to have a look at it.

Our camel drivers discovered in the valley below a Chinese caravan from Koko-khoto bound for Ku-ch'eng. They were traveling traders from Koko-khoto who had previously signed a "safety agreement" with the brigands. We sent one of our men to speak to the Chinamen. He found in their tent a suspicious looking armed Torgut, who inquired how many men we had with us and the number of our rifles. He proved to be one of Ja Lama's followers and had his encampment in the hills west of the castle. He told our man that there were several families of former robbers living in the neighborhood. He complained of being short of ammunition and used an old matchlock for which he could prepare some himself. His Mauser carbine lay idle in his tent, for he could not procure sufficient cartridges. According to him, some two hundred men still haunt the hills north and southwest of Ma-tzu Shan.

Later on some of the brigands came to our camp. Their bodies were covered with shaggy coats and they wore fur caps or blue handkerchiefs tied round the head. Their armament consisted of old Berdane rifles, Mauser carbines, and antiquated matchlocks. They live on the produce of their cattle and hunting. Plenty of game is found in the hilly region round Ma-tzu Shan. Furs are afterward sold on the markets of An-hsi and Yu-men hsien or used by the hunters themselves.

The robbers seemed much impressed by our arms and after a short visit departed. One of our Tibetans had had the unpleasant experience of meeting two months before a party of Khasak-Kirghizes, formerly in the service of Ja Lama. He was crossing the Gobi in the vicinity of Altan-usu with a Tibetan caravan when they were held up by a band of armed Kirghizes. A group of riders appeared from behind a small hill, dismounted from their horses and took up a position behind the hill. The Tibetans were well provided with firearms and decided to return fire. The attacking Kirghizes probably understood their intention for they sent out a man to talk the matter over. After the usual questions, from where the caravan came and what was its destination, the Kirghizes invited the Tibetans to partake of some äräki or Chinese brandy for it was the Chinese New Year. The Tibetans had to accept the invitation but remained on their guard all the time, the chief of the caravan keeping a small Browning in his large sleeve. The Kirghizes proved to be former soldiers of Ja Lama. According to our Tibetan they were all well armed with Berdane rifles and Mauser carbines. They did not harm the caravan and quietly went away.

About three o'clock in the afternoon we broke camp and for a considerable time skirted the Baga Ma-tzu Shan. After a two hours' march we entered a narrow mountain valley stretching north and south and followed it along the bank of a small rivulet called Balgantai. The river was dry but is said to carry a considerable volume of water during the rainy season. We camped on a level patch of ground close to a small brook of fresh water. We met a party of Torguts on a hunting expedition, one of which was our acquaintance of yesterday. One of the men had his right ear cut off and the whole company had something unpleasant about them. Our men regained their courage and laughed at the armament of the Torguts. The Torguts informed us that most of their people decamped toward the north for they feared an advance by the Chinese troops of General Feng, which were marching to Hami.

Next day the route followed for the first six miles the valley of the Balgantai River. The valley in many places presented ideal places for nomad camping grounds, with good pastures and springs. We found a large camel caravan camping in the upper part of the valley. It was a section of the same caravan which we had met the day before.

Not far from the upper Balgantai Valley we observed a large herd of ibexes on the top of a steep spur west of the valley. For the rest of the march the trail led over barren sandstone hills. We camped for the night on a small plateau not far from the big cart route Koko-khoto-Hami.

In the morning we continued our march along a mountain valley which gradually broadened into a flat-bottomed valley with scant grass. The valley was a pasture ground for a few families of Torguts from the Sawur Mountains, south of the Zaisan Lake on the Russo-Chinese border. We halted late at night at the foot of a steep crag, crowned by a ruined Chinese watch tower. A small brook gave a sufficient water supply for our caravan.

Next morning we ascended the crag to examine the tower. We had to climb flat and slippery rocks on which it was extremely difficult to gain a firm footing. There was almost nothing left of the tower except a portion of the wall.

We broke camp in the afternoon and followed the valley which imperceptibly narrowed into a dry river bed. The Mongol guides told us that a river flowed there during the rainy season.

We halted for the night near a small brook, coming from beneath a rock which was marked by a small Chinese miao. Formerly the temple had some clay images but now everything had crumbled to pieces and the interior of the temple was a mass of débris.

On May 15 we left behind us the Pei Shan Hills that rise north of the lowland of Kansu. The last hills were crowned with a ruined Chinese watch tower that once protected these desert tracts. Under the burning sun the caravan moved straight toward the south across the boundless gravel desert. To the east stretched the low ridge of Po-hsien Shan.

Far to the south stood the faint outlines of the northern ramparts of the Nan Shan Mountains. The plain was arid without either a brook or well, but in the distance one could see the glimmering formations of Chinese villages and gardens. The Province of Kansu was drawing nearer.

After sunset the air cooled considerably and our brave camels quickened their pace. Here and there we noticed deep-cut wheel tracks made by the heavy Chinese carts. Late in the evening we reached An-hsi-gol or Su-lo ho, that flows toward Tun-huang and along the banks of which passed the ancient military route of China. We decided to stop for a couple of hours to rest our men and camels and to send out our Mongol guides to reconnoiter a fording place on the river. No tents were pitched and all rested on camel loads around the camp fires.

The silence of our camp was disturbed by a large flock of sheep that was going toward An-hsi. The flock was intrusted to an elderly Mongol woman, who said that her people belonged to Khalkha Mongolia but that for many years past they had been living on the borders of Kansu, selling sheep and sheep wool on the markets of An-hsi and Su-chou. Our guides advised us that it would be better for us and our camels to cross the river at night and the hot lowlands of Kansu during the cool morning hours.

By three o'clock in the afternoon the caravan noiselessly marched off toward the river. The banks of the river were covered by sand dunes that reminded us of the rivers in far off Chinese Turkestan. The bed of the Su-lo ho was very broad and sandy but fortunately for us the river bed was dry, all the water having been transferred to irrigation channels in the fields. Having crossed the river we found ourselves on the great imperial highway connecting China with its distant province of Hsin-chiang.

For some time we followed a deep irrigation channel that lay south of the road, and almost entered the city of An-hsi, but then our guides suddenly decided to turn back and cross the channel somewhere outside the city. Again the whole caravan column turned and retraced its steps until we had found a fording place near a Chinese village.

The imperial highway, connecting Peking with Urumchi, the capital of the New Dominion, is a series of deep-cut wheel tracks. Here near An-hsi it looked exactly the same as in other parts of Chinese Turkestan. We expected to see troops moving along the route but saw nothing of them although we had been told of their presence in large numbers along the route. At this early hour the road was completely deserted. A telegraph line on small miserable poles follows the route, connecting Hsin-chiang with Peking.

We camped about noon in a small forest growing on the banks of a tributary stream of Su-lo ho. The nyer-wa went to see some of his Chinese friends in a Chinese village near by and the caravan rested in the shade of the forest. The heat was oppressive and a yellow veil left behind by a passing sand storm hung over the locality. The nyer-wa returned at four o'clock with the report that all was well along the route to Shih-pao-ch'eng and that we should start immediately in order to reach our destination for the day.

The camels were loaded and the caravan once more started across the hot plain at the foot of the Nan Shan Mountains. Neither wind nor clouds to lessen the heat of the day! The camels moved slowly and often emitted their drawling cry—a sign that the animals were exhausted. Now they rapidly lost their hair and this caused us a great deal of worry. Should we be able to reach Shih-pao-ch'eng before the camels had entered the period of shedding? Men had to protect their eyes with tree branches in order to diminish the terrific glare of the desert. After two hours of an exceptionally trying march we entered a narrow sandstone gorge and crossed a low col, that separated the lowland of Kansu from the highlands of the Nan Shan system. It became much cooler in the mountains and the animals doubled their pace with renewed energy. We descended into a broad swampy valley. To the south of it rose the imposing snow peak of Tsagan Chuluta. There were several Mongol encampments and herds of horses; horned cattle and sheep enlivened the picture. We pitched our camp outside one of the Mongol encampments on a dry patch of ground.

The Mongols came out to meet us and were extremely friendly. They were Khalkha Mongols from the khoshun of Daiching-Wang and had immigrated to Kansu some ten years ago. They lived on the produce of their cattle and also furnished animals to Chinese travelers who came from An-hsi to Shih-pao-ch'eng. Among them lived several Torguts from the Etsin-gol and one of them we enlisted in our guard. In the Mongol tents we saw several Ded-Mongols from Tsaidam. They live higher in the mountain valleys and seldom come to the northern slopes of the Nan Shan. Here we bought our first horse, a bay gelding which faithfully served the expedition as far as Chu-na-khe on the Tibetan upland, where it died from hunger and exposure.

Our doctor, who was attired in a purple Mongolian coat of Tibetan puru , was taken by everybody for a lama and people came to ask blessings from him. In every camp there was a great demand for the doctor and his medicine chest. We pitched only two tents and the rest of the party slept under the open sky. It was a wonderful starry night with the snows of the Tsagan Chuluta standing out clear on the dark opal sky.

Next day we started late, toward sunset, and crossed the gravel plain that separated us from the mountains. From the camp the distance to the foot of the mountain seemed to be trifling but it took us fully three hours to cross the plain. Our destination was Shih-bochen-ama or the entrance to the gorge of the Shih-pao-ch'eng River. After entering the gorge of the torrential river, the trail followed the left bank. Willow bushes and poplars grew on the flat terraces. We halted about eleven o'clock on a grass-covered terrace sheltered by fine old poplars. The air was wonderfully cool and we all rested after the oppressive heat of Kansu.

Next morning we continued to follow the river gorge which broadened as the mountains on either side of it retreated. We had a disagreeable experience with some of our camels and the whole affair could easily have ended in a disaster for it is impossible to control a caravan of frightened camels. We were quietly moving along the trail when suddenly a small donkey stepped out of the bushes. This was enough for our camels. The front camel jumped, threw its rider and stampeded toward a side valley. The rest of the camels followed its example; some threw their loads, the rest galloped after the first camel. It was lucky that Mrs. Roerich's camel and the rest of our riding camels stopped in time and did not bolt. Riding behind on our new horse, I could see our first camel detachment with some of our men disappear in a brisk gallop behind a low spur. I hurried to the scene to find the camel driver lying unconscious on the trail and the rest of the camels kicking off their loads. The poor camel driver had a severe heart attack following his fall and had to be attended by the doctor. Through a splendid effort the rest of the camel drivers and our men got hold of the camels but it took almost a full hour before everything was again in marching order. Many of the camels bled severely, the nose-stick by which camels are attached to each other having torn their nostrils.

Next to a large Chinese farm we passed a fine forest of spruce not far from which towers a rock with an ancient stone image of Maitreya, probably dating back to the period preceding the tenth century A.D. On his way to Mongolia, the Dalai Lama declared this stone image to have miraculously appeared on the surface of the rock and ordered a small chapel to be erected over it. The image represents a standing Maitreya and is now an object of worship among the local Mongols.

After passing the Chinese farm with its shady forest we struck southwest across a sandy plateau and entered an intersected country of low sand ridges. On our way we again met the river and during our march we forded it at least ten times. Not far from the river we met a fine looking Ded-Mongol riding a big camel. He wore a dark blue coat trimmed with otter fur and the small white felt hat of his tribe. He was followed by his wife riding a horse. A camel had been stolen from them and they were on the way to catch the thief.

After a brief conversation with the Mongol couple, who proved to be old acquaintances of the Mongol guides and the Tibetan, we continued our march. The Mongol gave us good news about Shih-pao-ch'eng. The Mongol headman was staying at the oasis and had plenty of good camels at his disposal.

The sandstone hills became higher and higher and their weathered and eroded tops looked like fantastic castles and strongholds. Not far in these hills are situated the small Ch'ien-fo-tong or caves of the Thousand Buddhas visited by Sir Aurel Stein during his second expedition of 1908-9.

We camped late in the evening on a small plateau on the bank of the river. On the other bank glistened the camp fires of a large Chinese caravan from Yu-men hsien, a small town east of An-hsi. The Chinese merchants of Kansu spend the summer months trading among the Mongol tribes in the higher valleys of the Nan Shan. They maintain their own large caravans of camels and move from place to place selling Chinese silks, European-made cloth, metal wares, rice, and flour, receiving in exchange camel and sheep wool, cattle and horses. The whole trade is conducted by barter and is largely based on credit, but not to such an extent as in Khalkha Mongolia previous to the Independence. In the neighborhood of large Mongol encampments or headmen's tents you will always find the blue tent of the Chinese trader.

The next morning we made an early start and by ten o'clock the caravan column was moving up the river valley. The vegetation was still scanty, with here and there shrubs of juniper and small willow bushes. The gorge broadened and we once more perceived in the distance the majestic Tsagan Chuluta. In front of it lay a vast plain—the Oasis of Shih-pao-ch'eng. On our way we passed several fields worked by a few Chinese settlers who lease the land from the local Mongol headmen.

The river branched into numerous channels that cut the marshy plain. Behind the dark line of the oasis lay expanses of piedmont gravel with scant shrubs. I rode ahead to find ground for our camp. With great difficulty we went through the swamps of the oasis. Several times the camel of our Tibetan guide sank knee-deep and my own horse fell into the mud. After an hour's search we found a temporary place for camp on a sandy plateau overlooking one of the many tributary channels of the river. The rest of the caravan column slowly moved into the camp and we pitched our tents. It was the end of our camel journey and we had to let the camel drivers go and take over the loads. One of the camel drivers had been ill for several days and now asked our doctor to give him some medicine. The doctor discovered pneumonia and a weak heart. The man was beyond hope of recovery and died the next day.

The very day of our arrival we were visited by the local Mongol headman Machen, a fat and sly looking man of fifty. He said that he was ready to help the expedition and even to furnish it with camels and horses for the trip to Nag-chu-ka. He told us that it was rather difficult to find sufficient provisions in Shih-pao-ch'eng but offered his assistance in getting supplies from An-hsi or from Ch'ang-ma pao-tzu, an important Chinese village east of the oasis. We parted good friends and he promised to visit us the next day.