APPENDICES

A. The Main Caravan Routes.

B. The territories of Haixi, Haibei and Haidong.

C. The “great seas” and the “Western Sea.”

D. Sea Silk.

E. Wild Silks.

F. Maritime Commerce and Shipping during the Han Period.

G. The Water Cisterns on the Route between Petra and Wadi Sirhan.

H. The Identification of the city of Angu with Ancient Gerrha and Modern Thaj.

I. The Spread of Ideas and Religions along the Trade Routes.

J. Climate and other Changes along the Silk Routes.

K. The Identification of Jibin as Kapisha-Gandhāra.

L. The Introduction of Silk Cultivation to Khotan in the 1st Century CE.

M. The Canals and Roads from the Red Sea to the Nile.

N. Kanishka’s Hostage in History and Legend.

A. The Main Caravan Routes

The overland routes from China to India, Parthia, and the Roman Empire stretched thousands of kilometres and presented the traveller with many geographical, and ever-changing political, obstacles.
Rarely, if ever, did caravans travel the whole route. Goods were carried to market-places where they were sold or traded, local taxes paid, and then other merchants transported them onward.
Long-distance freight costs were high, so preference was given to trade in rare or precious goods that were relatively light, compact, and non-perishable (such as silk).
The animal that made this long-distance trade possible was the camel. Camels can carry half as much as a horse and cart, and twice as much as a mule. They can travel long distances with minimal water and fodder. Carts and formed roads were, not needed, substantially reducing transport costs. Caravans could use alternative routes, or head across open country, whenever necessary. There was no need to stick to a road, if there was enough water, fodder and fuel available.
          At first the caravans mostly used the two-humped, or ‘Bactrian’ camel, native to Central Asia, and better adapted to the cold than the one-humped dromedary, or ‘Arabian’ camel. Cable and French (1943), pp. 169-172.
          At some point, it was discovered that first-generation hybrids between the two had more stamina than either of the original breeds, but it was some centuries before a cold-resistant one-humped variety was bred. Bulliet (1975), pp. 141-175.
A standard camel load in Roman times was about 195 kg (430 pounds). Over 227 kg (or 500 pounds) could sometimes be carried for shorter distances. A pack camel could travel 24 to 32 kilometres (15 to 20 miles) a day, and go for long periods without food or water. Bulliet (1975), pp. 20, 24, 281, n. 35. The main east-west caravan routes provided excellent conditions for camel travel almost the whole way from China to the Roman Orient.

Trade between distant parts of the Eurasian landmass has been occurring for several thousand years at least. Lapis lazuli was traded from the mines in eastern Badakhshan to Mesopotamia and Egypt by the second half of the fourth millennium BCE at the latest.
          The earliest long-distance road, the ‘Persian Royal Road,’ may have been in use as early as 3,500 BCE. By the time of Herodotus, (c. 475 BCE) it ran some 2,857 km from the city of Susa on the lower Tigris to Smyrna near the Aegean Sea. It was maintained and protected by the Achaemenian empire and had postal stations and relays at regular intervals. By having fresh horses and riders ready at each relay, royal couriers could carry messages the entire distance in 9 days, though normal travellers took about three months.
          This ‘Royal Road’ linked in to many other routes –some of them, such as the routes to India and Central Asia, were also protected by the Achaemenids, ensuring regular contact between India, Mesopotamia and the Mediterranean.
          Another very ancient series of routes linked Badakhshān in northeastern Afghanistan –the only known source of lapis lazuli in the ancient world – with Mesopotamia and Egypt by the second half of the fourth millennium BCE, and by the third millennium with the Harappan civilization in the Indus valley. Sarianidi (1971), pp. 12-13.
          By the second millennium nephrite jade was being traded from mines in the region of Yarkand and Khotan to China. Significantly, these mines were not very far from the lapis lazuli and spinel (‘Balas Ruby’) mines in Badakhshān and, although separated by the formidable Pamir, routes across them were, apparently, in use from very early times.
          Regular diplomatic contacts and large-scale trade between China and the West first occurred soon after the daring explorations and international contacts developed by the famous explorer and diplomat, Zhang Qian circa 112 BCE.

The main ‘Silk Route’ to the west crossed out of the Tarim Basin past Kashgar by relatively easy passes, accessible to camel caravans. They crossed into Ferghana, and then continued through relatively flat country, with sufficient water and fodder for camels, all the way to Parthia, Syria, and beyond.
         However, camels do poorly in mountainous or rocky regions, so goods had to be off-loaded onto mules, yaks, or human porters before crossing the high passes over the Pamir and Karakoram mountains into southern Bactria and northern India.
          Dunhuang was at the crossroads of several routes, including the main route from Central Tibet to Mongolia, and the three main branches of the ancient ‘Silk Routes’ across and around the Tarim Basin.
          The first major difficulty facing caravans after leaving Chinese territory at the Yumen Frontier Post (about 85 km northwest of Dunhuang – see Stein (1921), II, p. 691), was to find a way around or across the fearsome Taklamakan desert. The Taklamakan forms a giant oval enclosed to the north, northeast, west, and south by some of the highest and most forbidding mountain ranges in the world. To the east, it opens into the vast wastes of the Gobi desert.

‘Silk Routes’ rather than ‘The Silk Road.’

“The use of the plural above, ‘Silk Routes’ rather than the more familiar ‘Silk Road’, is deliberate. We are not dealing with a single entity like Watling Street or the Appian Way, but with a complex network of roads and tracks reaching right across Eurasia. There are some permanent nodes in the network, such as Ch’ang-an where much of the silk was produced and Rome where much of it was consumed; but the line taken by a particular caravan depended on the weather, the economic situation and the political situation, any of which might change with surprising suddenness.” Sitwell (1984), p. 174.

“The middle and shortest route by Lou-lan became deserted because of the shifting of the waters; but the southern, known in many stretches, was mentioned later by Marco Polo in regions east of Khotan, by streams of chalcedony and jasper. Six crossings between south and north are referred to in the Chinese annals, across what is now the central desert of Taklamakan.
          In these outer lands the raid of a single enemy was enough to ruin an oasis for ever. ‘Many travellers stick in the swamp. On the northern road, the Hiung-Nu fall upon one. On the southern there is neither food nor water, and on many uninhabited stretches there is hunger’: there was in fact a ten days’ trek without habitation worth mentioning, and the settled places when reached were too poor to afford the traveller sufficient provision for the way. The Chinese solved the problem through colonies of soldier-peasants: an imperial komissar was established to watch over crops and harvests and care for the visiting ambassadors; and the northern route in the second half of the first century B.C. came to take eight days less than the southern, while the Hiung-Nu were kept down.” Stark (1968), pp. 189-190.

The confusion surrounding the names of the routes.

Not surprisingly, the routes have frequently been confused in the literature. This is particularly because what was called the “Northern Route” in the two Han histories, is called the “Central Route” in the Weilue, and we have mention of both a “New Route” and a “New Route of the North” in the Weilue.
         There are, in fact, three main caravan routes around and across the Taklamakan Desert, and one to the north of it, described in the Weilue. Also briefly mentioned are the two secondary north-south routes joining the Southern Route with the Central Route, and the maritime route. See Appendix F.

“Yu Huan shows here how the route which led from Hami to Barkol and then to Gucheng (Guchen) to rejoin the Central Route at Kucha by turning off abruptly to the south after leaving Gucheng to cross the Bogdo ola mountains and reaching Turfan. Meanwhile there remains one obscure point for me : why Yu Huan says that the Northern Route rejoins the Central Route only at Kucha? He should have said, it would seem, that the two routes coincided from Gaochang (Turfan), but this is not a sufficient reason to presume that the Central Route had another lay-out than the one we have determined. – In the detailed examination Yu Huan makes later on of the three routes, he does not show that the Northern route fits together with the Central Route, but it proceeds to the Wusun, that is, as far as the Ili Valley. It is thus clearly proven that the new route established by the Chinese in the 2^nd year of our era was the one which passed to the north of the Tianshan by Urumchi, Manass, Kur-kara-wusu [= modern Wu-su or Usu. 84^o 40’ E; 44^o 26’ N], then crossed the Iren Shabirgan [Erenhaberga Shan] mountains by the Dengnul [Talki Pass or Ak Tash Davan?] Pass to enter the Ili Valley (cf. Documents sur les T’ou-kiue occidentaux, pp. 12-13). – As one can see from the above, the three routes mentioned by Yü Huan basically correspond with those which the Imperial Commissioner Pei Ju 裵矩 described in his “Treatise with Maps on the Western Countries” 西域圖記 (Suishu, chap. LXVII, p. 5 b) around the year 608 of our era: “The Northern Route goes through Yiwu 伊吾 (Hami), passes by Pulei Lake 蒲類 (Lake Barkol), the Tiele 鐵勒 (Tölös) tribes, the court of the Khaghan of the Tujue 穾厥可汘庭 (the Borotala or Ili Valley), crosses the rivers which flow towards the north 北流河水 (the Chu, Syr Darya, and Amu Darya Rivers), and arrives at the kingdom of Fulin (Byzantium), which is in contact with the Western Sea. – The Central Route passes through Gaochang 高昌 (Yar-khoto, near Turfan), Yanqi 焉耆 (Karashahr), Qiuci 龜玆 (Kucha), Suole 疏勒 (Kashgar), crosses over the Congling 葱嶺 (Pamirs), then crosses the kingdoms of Pohan 鏺汘 (Ferghana) and Suduishana 蘇對沙那 (Osrushana = modern Ura Tyube), the kingdom of 康 Kang (Samarkand), the kingdom of Cao 曹 (Ishtykan), the kingdom of He 何 (Koshania), the large and small kingdoms of An 安 (Bukhara and Kharghan near Karminia; but it is necessary here to reverse the order of the two terms, for the itinerary passes through Kharghan before reaching Bukhara), the kingdom of Mu 穆 (Amol), and arrives in Bosi 波斯 (Persia), where it contacts the Western Sea. – The Southern Route passes through Shanshan 鄯善 (to the south of Lop Nor), Yutian 于闐 (Khotan), Zhujubo 朱俱波 (Karghalik), Hepanto 唱 (read 喝)槃陀 (Tashkurgan), crosses over the Congling 葱嶺 (Pamirs), then crosses Humi 護密 (Wakhan), Tuheluo 吐火羅 (Tokharestan), the Yida 挹怛 (Hephthalites), Fanyan 忛延 (Bamiyan), the kingdom of Cao 漕 (Ghazni?; cf. LÉVI, in J.A. Sept.-Oct. 1895, p. 375), and reaches the land of the northern Poluomen 北婆羅門 (Hindus), where it contacts the Western Sea. – The only differences which turn up between these itineraries, and those of Yü Huan, come from, on the one hand, the fact that the routes of Pei Ju go further to the west and, on the other hand, the Southern Route described by Pei Ju emerges from the Pamirs in Badakhshan, whereas the Southern Route of Yu Huan goes from the Pamirs into Kashmir [but note that Chavannes incorrectly identifies Jibin with Kashmir].” Translated from Chavannes (1905), p. 534, n. 3.

The ‘New Route’ to Turfan probably became an alternative after the Chinese lost control of Hami to the Xiongnu. At the best of times this a difficult route north across the desert to the northeast of Loulan, with little fodder or water available. Caravans could have been supplied and supported from both Loulan and Turfan but only at considerable expense.
          The ‘New Route of the North’ followed through the kingdoms beyond Hami, which stretch in a long arc to the north of the Bogoda and Tianshan ranges to Wusun territory in the west. There is no mention of where it begins, or of Hami.
          Presumably, since Hami was off-limits, one would first head to Turfan via the ‘New Route.’ Then, to reach Jimasa and territories to the east, one would cross directly north across the Bogoda Mountains by a rather difficult track to Jimasa and join the ‘New Route of the North’ there. If one wanted to travel west, the route left Turfan and travelled along the relatively easy road to the region of modern Urumchi, and then west along the ‘New Route of the North’ to the north of the Tianshan range.

The nature of the trade.

“Sometime after the death of Alexander, Chinese silk, transported by caravan through central Asia and passed along by a chain of middlemen, started to filter through to the Mediterranean, where its superiority to the nearest thing the Greeks had, produced from wild Asia Minor silkworms, was swiftly recognized. In the second half of the second century B.C., the Chinese became more active in the trade, dispatching caravans on a regular basis. Starting from Paochi, centre of a complex of roads, these moved inside the Great Wall by way of T’ienshui, Lanchou and Wuwei to the western end of the wall and deep into Chinese Turkistan; by 118-114 B.C., some ten caravans a year were making the trip. At Anhsi between the Gobi desert and the Nan Shan mountains the route forked into three branches to avoid the vast salt swamp in the Tarim Basin, two looping to the north and one to the south. The southern loop and one of the northern came together at Kashgar, then forked again to snake through the difficult Pamir mountains; this stretch was more or less the halfway point to the Mediterranean. All three rejoined at Merv to continue across the desert and join up with the tracks that led through Persia and Mesopotamia to the sea. Nobody went the whole distance. Somewhere between Kashgar and Balkh was a place called Stone Tower, and here the Chinese turned their merchandise over to local and Indian traders. The latter carried their share south to India to send it the rest of the way by boat, the others plodded on into Persia, where they met up with Syrians and Greeks who took care of the final leg.” Casson (1974), pp. 123-124.

“The whole Indian trade shifted gradually north-west because of the overland route through Bactria, until the Hiung-Nu in A.D. 23 fell upon it and made it impossible again for fifty years. In A.D. 73 the Chinese began to resuscitate it, and again advanced along both sides of the great chain of oases of the Tarim basin to ‘open the roads that lead to China and establish peace’. Merchants along the trade line, from Parthia and Bactria and India, sent their requests and prayers to the Celestial Court; and it was during this renascence of the trade around A.D. 100, that the Syrian-Roman merchants and the Chinese tried to link up their trade directly, and failed because Parthia lay between.
          At this time, the trade was in its heyday: ‘peasant colonies were founded in the fertile lands; inns and posts for changing horses were established along the main routes; messengers and couriers travelled in every season of the year; and the Merchant Strangers knocked daily on our gates to have them opened’. It was the period of the Roman peace with Parthia and the never-to-be-repeated summit of the Asiatic trade in the ancient world. By 127 A.D. the Tarim relapsed into chaos, while traffic increased in the Persian Gulf or Aden as it waned in the north: that there was any connection between the Chinese of the north (the ‘long-lived’ Seres of Lucian) and those of the south (the Sines), was still undiscovered in Ptolemy’s day.” Stark (1968), p. 191.

“Unfortunately, whether by land or sea, the contact between the two great cultures was always tenuous. Shipments of Chinese goods came to the Mediterranean year in and year out, cinnamon-leaf and camphor and jade and other items as well as silk, and Graeco-Roman statuettes and jewellery and pottery made the journey the other way, but rarely was there a direct exchange; in between were merchants from other countries, particularly India, which not only lay astride the sea lanes but was firmly linked by branch roads with the overland silk route. These middlemen had solid information to pass on – it was they who supplied the many place-names in Central Asia and the names of the Indonesian islands that the geographers know now – but they were businessmen, not reporters. What filtered back to the man in a Roman or Chinese street was mere fanciful hearsay. The Romans thought the Chinese were all supremely righteous; the Chinese thought westerners were all supremely honest. Kan Ying, sent as envoy to Mesopotamia in A.D. 97, describes the people he met as ‘honest in their transactions and without any double prices’ – probably the first and last time that has ever been said about Near Eastern tradesmen. Kan Ying, the embassy of An-tun [in 166 CE] – we can number on the fingers of one hand the known occasions when westerners and Chinese met face to face.” Casson (1974), pp. 125-126.

“The Chinese are mild in character, but resemble wild animals in that they shun the company of the rest of their fellow men and wait for traders to come to them.” Pliny NH (a), p. 64 (VI, 54).

(a) The “Southern Route”

There were two branches of the Southern Route between Dunhuang and Shanshan (northwest of Lop Nor). The first led directly across the desert. It was short but difficult and dangerous. By the middle of the first century CE, the Han were able to cut off support from the Xiongnu and pacify the Er Jiang who lived in the Altin-tagh ranges to the south. They were then able to make use of the better-watered and sheltered, though longer, route further south. Aurel Stein pointed out:

“A look at the map shows that the route meant [in the Weilue] is the one which skirts the high Altin-tagh range, and still serves as the usual connection between Dunhuang and Charklik during that part of the year when the shorter desert route is closed by the heat and the absence of drinkable water.” Stein (1912), pp. 514-515.

The most feared stretches of desert were between Cherchen and Khotan. Not only was there a lack of water and fodder but the constant crossing of sand hills was very tiring for both man and beast. “The desert itself is quite flat, a billowing sea of soft yellow sand-dunes 5 to 30 m high. However, in some central areas, for example in the west of the Keriya River, the dunes can rise to more than 200 metres high – a tough challenge even for a camel caravan.”
          From Khotan there were several routes south: the one in the Weilue headed southwest across the Pamirs and through Hunza and Gilgit (Xuandu – the notorious ‘Hanging Passages’) to North India and Jibin (Kapisha-Gandhāra). Branch routes led south to Ladakh and Kashmir, northwest to Kashgar, and north along the Khotan River join the ‘Middle Route’ near Aksu. The Weilue informs us that an extension of the Southern Route extended right across India to the Pandya kingdom at the southern tip of the sub-continent.

Although it is not known if the following alternative route to India was in use this early, it is quite possible, and so I will mention here for consideration:

“South of the Southern Silk Road proper were more roads. One of them was the so-called Qinghai Road that led from Lanshou to Xining (the present capital of Qinghai Province), from whence it crossed the desert and reached the classic Southern Silk road near Miran. Another was and extremely arduous trade route that led from Xining in a south-westerly direction to Tibet, finally to reach Nepal and India. This Tibetan Route was opened in the 5^th century AD [and perhaps earlier], even before Songsten Gampo (609-650) had politically unified the Snowlands.” Baumer, p. 8. See also Ibid., p. 2.

Historical records suggest that from the first century BCE until the second half of the third century CE the Tarim Basin experienced a relatively warm and wet period. This caused faster melting of mountain glaciers, which thus supplied more water to the rivers flowing down from the mountains into the Taklamakan desert. This, in turn, made the Southern Route more feasible, and the high passes over the Pamir and Karakoram Mountains somewhat easier to cross.
          By the late third and early fourth century this climatological process seems to have reversed itself, making the Southern Route more difficult to cross. Caravans were forced to use the longer middle and northern routes and even, when the northern nomads could be controlled, to avoid the Tarim Basin altogether, and pass to the north of the Tian Shan ranges. Stein (1921), p. 1524; Stein (1928), pp. 79, 435, 837; Almgren (1962), p. 101; and Hoyanagi (1975), pp. 85-113; Bao et al. (2004).
          Recent research has confirmed these historical indications. Ice samples from the Guliya ice core which is situated in the mountains about 150 km due south of Keriya have provided valuable temperature and precipitation data for the region over the past 2,000 years. They show that there was a relatively warm and wet period before 270 CE rapidly followed by a cold dry period between about 280 and 970 CE. Shi et al (1999), pp. 90-100.

“This worsening of climatic conditions was not limited to the Tarim Basin alone, but concerned all of China, which was plagued by periods of severe drought between about CE 280 and CE 320. The annals of the Western Jin (CE 265-316) report that, in the year 309, the Yellow River and the Yangtsekiang (Changjiang) practically ran dry and could be crossed on foot.” Baumer (2000), p. 3.

In spite of its difficulties, the Southern Route remained important. It was better protected than the Central and Northern routes from raids by the northern nomads. It was by far the shortest and most direct route to the jade centres of Khotan and Yarkand. It remained passable, if difficult, for individuals and smaller caravans travelling to India, over the notorious Karakoram Pass, and human porters could even travel through the Wakhan corridor to Gandhāra and southern Afghanistan.
          Not quite halfway between Kashgar and Yarkand, between modern Yengisar and Qizil, a route turned west and headed towards Badakshan as described in the Tarikh-i-Rashidi:

“The distance from Káshghar to Yángi-Hisár is six statute [shari] farsákhs. At about six farsákhs from Yángi-Hisár is an insignificant hamlet called Kará Chanák,¹ in front of which flows another stream called Shahnáz, which waters several [other] places. The valley of the Shahnáz lies in the western range, and the [high] road from Káshgar to Badakhshán runs through this valley.” Elias (1895), pp. 295-296.

By far the easiest routes from China to the Turfan oasis – “Nearer (or Southern) Jushi,” as well as to the Jimasa region (“Further Jushi”), went through Hami (Yiwu). Because of its critical strategic importance, Hami changed hands between the Xiongnu and China numerous times. China finally lost control of it to the Xiongnu in 151 CE, and did not regain it for over 400 years.

South to India over the ‘Hanging Passages.’

The detailed account of the Southern Route given in the Hanshu (CICA, pp. 97-99) mentions that, between Pishan (= modern Pishan) and Jibin (Kapisha-Gandhāra) there is a small kingdom called Wucha, said to be 1,340 li (557 km) to the southwest of Pishan, just before the route turned west into the dreaded ‘Hanging Passages.’
          Xuandu 縣度 [Hsüan-tu, often incorrectly given as Hsien-tu], from: 縣度 xuan = ‘to suspend”, ‘hang’, ‘dangerous’ + 度 du = GR 11640a: “7. – To cross (the water by ferry). To cross (over)].” From this we get the infamous, ‘Hanging Passages’ – the narrow and dangerous hanging footpaths of the Hunza region. 
          It is significant that Xuandu is not listed as a guo (= ‘kingdom’ or ‘country’) in any of the ancient Chinese texts. It is clearly described as a locality, not a state. It has long been recognised that it refers to the terrifying hanging pathways, locally known as rafiks, which are so characteristic of the route through the Hunza valley to Gilgit. The most difficult passage in the whole Hunza valley is the section south of the junction of the Misgar and the Hunza rivers. See, for example, Chavannes (1905), p. 529, n. 5.
          Although it was impossible to take pack animals over this route, a route barely practicable for people on foot, it was by far the shortest route from the Tarim Basin to Gandhara and Jibin in what are now northern Pakistan and southeastern Afghanistan. The Hanshu describes Xuandu as follows:

The Hanshu describes Xuandu (Hunza/Gilgit) as follows:

“What is termed the Suspended Crossing is a rocky mountain; the valley is impenetrable, and people traverse the place by pulling each other across with ropes.” CICA, pp. 99-100, and n. 169.

The Hanshu gives the distance from the seat of the Protector General at Weili near Kucha to Wucha as 4,892 li (2,035 km), and to Xuandu, the ‘Hanging Passages’, 5,020 li (2,088 km). This would indicate that Xuandu was only 128 li (53 km) west of Wucha, placing it in present day Ghujal, or ‘Upper Hunza,’ which is on the way to the Shimshal Pass to the west, or the Kunjerab Pass to the northwest. The most difficult passage in the Hunza valley is the section south of the junction of the Misgar and the Hunza rivers.

“Rafiks, the local name for such galleries, are fastened to the sheer cliffs by branches of trees forced into the fissures of the rock and covered with small stones. Elsewhere natural narrow ledges are widened by flat slabs packed over them. In some places the rafiks “turn in sharp zigzags on the side of cliffs where a false step would prove fatal, while at others again they are steep enough to resemble ladders. To carry loads along these galleries is difficult enough, and . . . for ponies, sure-footed as they are, wholly impassable.” Even his [Aurel Stein’s] terrier, Yolchi Beg, so nimble on the rocks of Mohand Marg, was fearful and allowed himself the indignity of being carried. Rafiks alternated “with passages over shingly slopes and climbs over rock-strewn wastes.” To negotiate this terrain, the “baggage animals were left behind [at Chalt] and coolies taken for the rest of the journey up to the Taghdumbash Pamir.” Mirsky (1977), p. 121.

“The next day’s march [from Khuabad] to Misgar, he had been warned, would be the worst part of the route. By starting before dawn while the river was still low enough to ford, he avoided a long detour and a perilous crossing on a rope bridge. Then the going reached a climax of “scramble up precipitous faces of slatey rocks . . . with still more trying descents to the riverbed”; slower still was the progress along rafiks clinging to cliffs hundreds of feet above the river. But the previous five days had toughened him, and he felt fresh when he emerged from the rocky gorge to an open valley. . . .  Here he discharged the “hardy hillmen who had carried our impedimenta over such trying ground without the slightest damage.” Beyond, the route was open to baggage animals at all seasons.” Mirsky (1977), p. 125.

The Kushans are thought to have controlled the whole region from the late 1st century and for most or all of the 2nd century CE. Chavannes’ mention in Ban Chao’s biography that, “(Ban) Chao then crossed the Congling and got as far as Xuandu.” There is no indication of the date or any other details. One can only assume that the reference was to a brief foray by Ban Chao to the borders of Kushan territories – perhaps to deliver a message – or just to scout out the territory for himself. See Chavannes (1906), p. 237.

From Xuandu (Hunza–Gilgit) there were four main routes one could take:

1. south along the Astor river and across the Burzil Pass into Kashmir,

2. a difficult route along the Indus River Valley to Taxila,

3. through the Swat Valley to Peshawar or,

4. via Chitral to Jibin (Kapisha-Gandhāra), and on to Wuyi (Kandahar) or Gaofu (Kabul/Kabulistan).

“Surrounded thus by granite precipices and huge wastes of ice and snow, affording only a hazardous passage during a few summer months into the neighbouring countries, Hunza-Nagar has but one vulnerable point on the southern part of the Hindoo Koosh, the ravine of the Kanjut River; while the junction of that torrent with the Gilgit River is the one gateway of the country assailable for an invading force. Even this entrance is practically closed during the summer months; for then the river, swollen by the melting snows, becomes an unfordable and raging torrent, overflowing the whole bottom of the valley at many points, so that the only way left by which one can ascend the gorge is a rough track high up upon the cliff-side, carried along narrow ledges, and overhanging frightful precipices – a road fit only for goats and cragsmen, which could be easily held by a handful of determined men against a large force; while at this season the river can only be crossed by means of the frail twig-rope bridges, which will support but two or three men, and can be cut adrift with a knife in a few moments.
          Such is the road into Hunza-Nagar from our side; but at the head of the Kanjut Valley there is a group of comparatively easy and low passes, leading across the Hindoo Koosh on to the Tagdambash Pamir, in Chinese territory, which are used by the Kanjuts on their raiding expeditions. . . . ” Knight (1893), pp. 345-346.

“As one ascends the valley beyond Hunza. . . . It is only at certain points, where passage along the cliffs would otherwise be absolutely impossible for the best cragsmen, that any steps have been taken to open a road, and then it is but the narrowest scaffolding thrown from ledge to ledge. One comes upon position after position of immense natural strength in this gorge, where the dangerous and only path passes under stout sangas, which could be held by a handful of men against a host. Even as the Kanjuts had left the approaches to their valley below Nilt as difficult of access as possible, so had they done here, at the outlet of their country on to the Pamirs, rendering it almost impossible for an enemy to invade them from either direction.” Ibid., pp. 488-489.

“Strategically the Pamirs had always been written off as too bleak and barren to appeal to the Russians and too formidable for them to cross. Wood’s story suggested nightmarish conditions, which the Mirza’s travels fully supported. And, in April, Gordon found the going quite as bad, the wind unbearable, the snow freezing to their faces as it fell, and fuel and provisions desperately short. But from the Wakhi people he heard a different story. In summer the grazing was, as Marco Polo had recorded, some of the best in the world. Moreover, though mountainous, it was nothing compared to the Karakorams or the Kun Lun. The Pamirs, he was told, ‘have a thousand roads’. With a guide you could go anywhere and, in summer, considerable forces might cross without difficulty. The Chinese had done it in the past, the Wakhis had recently sent a contingent across to Kashgar, and the Russians might do it in the future.
          Finally, and most important of all, it was discovered that the passes leading south from the Pamirs over the Hindu Kush to Chitral, Gilgit and Kashmir were insignificant. This was so disconcerting that Gordon, ever discreet, omitted all mention of it in his published account. The discovery was made by Biddulph who, while the others explored the Great Pamir, made a ‘lonely journey by the Little Pamir’ (a misprint – in Gordon’s book actually has it as a ‘lovely journey’). In the process he climbed the northern slopes of the Hindu Kush and ascertained that at least two passes constituted veritable breaks in the mountain chain. One you could ride over without ever slowing from a gallop and both had artillery transported across them. To these Gordon added another ‘easy pass’ conducting from Tashkurgan to Hunza.” Keay (1977), pp. 257-258.

“Communication with Badakhshan is easy [from Gilgit and region] by the Darkot and Warogil Passes, which are the lowest depressions in the great Hindu Kush and Karakorum chains from Bamian on the west to the unknown passes of Tibet on the east.” Neve (1945), p. 132.

“And in those days [the journey from Gilgit to Kashmir], before ever Mr. Knight was there, before a regular road was made, when even the Indus had to be crossed by a rope bridge, and when the only track led by crazy wooden galleries along the sheer face of the most dreadful precipices, the journey was an experience well worth having and well worth talking about.” Younghusband (1909), p. 161.

“Gilgit, the northernmost outpost of the Indian Empire, covers all the passes over the Hindoo Koosh, from the easternmost one, the Shimshal, to those at the head of the Yasin River, in the west. It will be seen, on referring to a good map, that all these passes descend to the valleys of the Gilgit River and its tributaries. But the possession of the Gilgit Valley does more than this: it affords us a direct communication through Kashmir territory to the protected State of Chitral. . . . ” Knight (1893), pp. 290-291.

From Gilgit a relatively easy route, accessible to pack animals for most of the year, led to Mastuch in the Upper Chitral Valley. From here one can head either west into Badakshān or south to Chitral.
          From Chitral the relatively easy route ran through ancient Hadda (near modern Jalalabad) to Peshawar, the ancient city (where Buddhist accounts tell us Kanishka made his winter capital), and then to northern India; or directly southwest towards Ghazni, Kandahar, and the Persian Gulf.

“Westward from Gilgit is the country of Chitral, distinguished as Upper and Lower. The latter, which is nearest to the Hindu Kosh, is situated on a river flowing from a lake called Hanu-sar, and ultimately falling into the river of Kabul. The country is rough and difficult. The Mastuch, as the capital is termed in the language of the country, is situated on the left bank of the river. It contains a bazar, with some Hindu shopkeepers, and is as large as Mozeffarabad, containing between four and five hundred houses: slavery prevails here. . . .
          The Mastuch, or capital of Upper Chitral, is situated in the same valley as that of Lower Chitral, at about three days’ march, and about thirty miles north-west from Gilgit. It stands upon a river, and consists of about four hundred houses, with a fort, on a moderately extensive plain, from whence roads lead to Peshawar, Badakhshan, and Yarkand. The mountains in the neighbourhood are bare, and much snow falls: the climate, however, upon the whole, is temperate. Some traffic takes place with Badakhshan and Yarkand, whence pearls, coral, cotton baftas, and chintzes, boots and shoes, and metals are imported: horses are also much brought, and tea, but the latter is not much in use. The chief return is in slaves, kidnapped from the adjacent districts, or, when not so procurable, the Raja seizes and sells his own subjects. Soliman Shah, the Raja, resides chiefly at Yasin, which is not so large as the capital, but is better situated for the command of the country. . . . West from Yasin is the Darband, or fortified pass, of Chitral. . . . ” Moorcroft and Trebeck (1841), pp. 268-270.

Both Hadda (or Hidda – near modern Jalalabad) and Kapisha (near modern Begram) were probably considered part of the territory of Jibin at this time. See: Appendix K.

(b) The “Central Route” or “Middle Route.”

The Central Route headed west from the Yumen (‘Jade Gate’) Frontier Post, then northwest to the ancient town of Loulan, to the north of Lop Nor, and then on via the now-dry Kum (or Kuruk) Darya and Konche Darya to the region of modern Korla, where it joined the route coming from Turfan.
          This route was probably preferred for large caravans because of the ready availability of water. The importance of this route is underscored by the recent discovery of eleven beacon towers along the banks of the Konche Darya (Kongque River):

Great Wall extends to Xinjiang, 500 km longer: archeologists (02/22/2001)

The Great Wall of China is 500 kilometers longer than the earlier recorded length, according to archeological findings released in Urumqi recently.
          The new findings show that the Great Wall extends to the Lop Nur region in northwest China's Xinjiang Uygur Autonomous Region, instead of previously acknowledged Jiayu Pass in Gansu Province.
          Lop Nur now is a desolate desert region where China had established nuclear test facilities.
          Mu Shunying, a research fellow with the Xinjiang Institute of Cultural Relics and Archaeology, discovered during a field survey conducted in 1998 an earthen wall stretching from western Yumen Pass in Gansu Province to the northern edge of Lop Nur.
          Luo Zhewen, president of the China Society of Cultural Heritage, said, "There is no doubt this is part of the Great Wall as it consists of the city wall and beacon towers, forming a complete defense system."
          The wall is identical to the sections at Jiayu Pass and Yumen Pass in terms of architectural style and function. However, this newly found section was made with yellow sandy stone and jarrah branches found locally, he added. Luo, 77, is China's top Great Wall expert.
          Mu said it's obvious that the new find is a man-made wall built for the purpose of defense, as its shape and size resemble the other sections of the wall. Moreover, a large number of arrowheads have been found near the new site which indicates battles took place nearby, Mu said. Great Wall Extends to Xinjiang, 500 km longer. The Great Wall is a military installation built some 2,000 years ago. It has been renovated by numerous dynasties in the years following the Qin Dynasty, when Emperor Qin Shihuang ordered to link up separated wall sections.
          With the addition of the new section found in Xinjiang, the total length of the Great Wall would be 7,200 kilometers. The Great Wall was listed as a World Heritage site by the United Nations Educational, Scientific and Cultural Organization (UNESCO) in 1987.
          According to historical records, Emperor Wudi of the Han Dynasty (206 B.C.-A.D. 220) mobilized 600,000 laborers to build a wall from Dunhuang to Yanze, the present site of Lop Nur. The massive construction project is illustrated in frescos at the Dunhuang Grottoes.
          During a recent tour to Lop Nur, a Xinhua reporter saw the new section of wall, which undulates westward at heights ranging from one to three meters, with some portions completely missing. The lower part of some of this section is covered by reeds, jarrah and other kinds of plants that live in arid areas.
          The portion of the Great Wall in eastern China was made of brick, while most parts of the wall in western China were made of yellow sandy soil and jarrah branches.
          Luo said the Great Wall in Xinjiang was built to protect merchants traveling on the ancient Silk Road.
          Wang Binghua, a researcher of the Xinjiang Institute of Cultural Relics and Archaeology, said the Great Wall in Xinjiang runs parallel to the Silk Road. An official with the State Administration of Cultural Heritage said the state will further investigate this valuable historical site and take measures to protect it.
          Experts believe the newly discovered segment of wall is not likely to be the end of the Great Wall, as beacon towers continue to appear along the Kongque River, pass through Wulei, the site of the prefecture government of the western region during the Han Dynasty, and extend to Kashi in southwestern Xinjiang. Eleven beacon towers have been seen at the bank of the Kongque River.
          The Lop Nur River, which supplied water for Lou Lan, a busy commercial city on the ancient Silk Road, has dried up and civilization there moved elsewhere in China. The kingdom of Lou Lan was ruled by the government of the Han Dynasty. Troops of the Han Dynasty were stationed in Lou Lan. (Xinhua) Downloaded on 13 May 2001 from:

http://www.chinadaily.com.cn/cover/storydb/2001/02/22/cn-wall.222.html.

It has even been suggested that it might once have been possible to transport goods by water from Loulan to Yarkand or Kashgar, but there is no mention of this in the historical sources:

“From about 120 BC to AD 330, the MIDDLE SILK ROAD was regarded as the preferred caravan route. It crossed the dreaded Lop Desert from Dunhuang to Loulan, then led to today’s Korla, and from thence west to Kucha and Aksu and once again to Kashgar. This route had the advantage of making it possible to use a barge from Loulan on the Kum Darya (also called the Kuruk Darya), and then on the Konche Darya to Korla. The meaning of these two river names refers to the conditions in the 19^th and early 20^th centuries, for Kuruk Darya means “Dry river” and Kum Darya “Sand River”. According to Hedin’s explorations, during the first few centuries of our era the Tarim also flowed into the Kuruk Darya, so that perhaps in those days it was possible to transport wares on water from Loulan to Yarkand or Kashgar.⁴ It is not known whether this waterway was actually used in ancient times or whether the overland route was preferred.”

4. Sven Hedin, Scientific results of a journey in Central Asia 1899-1902, Stockholm, 1904-1907, vol. II, p. 263.

Baumer (2000), p. 9 and note 4. Baumer goes on to say that:

“The middle route could be used during winter only, not merely because the heat was less oppressive, but most of all because it was easier to take along water reserves in the form of ice blocks. This stretch offered no springs, and water was so scarce en route that it would never have been enough for an entire caravan.” Although he does not say so, he must be referring only to the period after the Kuruk and Konche rivers dried up – a process which probably began after the severe droughts which occurred from c. 270 AD onward.”

Zhang Qian probably took the Central Route in c. 119 BCE when he travelled to the court of the Wusun near Issyk-kol with 300 men, about 600 horses and myriads of cattle and sheep plus silks and gold. See Shiji 123 – translated in Watson (1961), p. 272, and Stein (1928), I, p. 341.

The Weilue says that the ‘Central Route,’ after reaching “ancient Loulan and, turning west, goes to Qiuci(Kucha), and on to the Congling mountains.”
          The ‘Central Route’ probably became the main route to the west whenever the Chinese lost control of Hami and/or Turfan. Fortunately, the section of the route between Loulan, to the north of Lop Nor, and Kucha is now well established:

“. . . . Dr. Hedin on his journey of 1896 to the terminal Tarīm had found an obviously ancient route line leading from Korla to Ying-p’an, where the dried-up bed of the Kuruk-daryā branches off towards Lou-lan, marked by a series of big watch-towers. His description of them strongly supported the belief that this line of towers dated back to the period when the ancient Chinese route from Tun-huang to Lou-lan and thence to the northern oases of the Tarīm Basin was first opened. The careful survey of them which I was able to make in the spring of 1915 on my way from the Kuruk-daryā to Korla has fully confirmed this belief. It has furnished conclusive evidence that these towers served as watch and signal stations along the road which connected Lou-lan with the Chinese administrative posts and military colonies established under the Emperor Wu-ti in the oases dotting the southern foot of the T’ien-shan.
          The chief, if not the sole, danger which threatened the safety of this great military and trade route came, as the account of the Former Han Annals shows, from the irruptions of the Hsiung-nu, or Huns. For these, as we have seen, the open Kara-shahr valley, with its easy approaches from the Yulduz and other great camping grounds north, must have at all times been the main gate. Experience gained during centuries on their far-flung northern borders must have proved to the Chinese commanders that the best safeguard against such attacks and raids lay in securing quick warning which would allow for timely preparation for defence. Korla and the adjacent parts of the route lay certainly nearest to the ground whence the danger of incursions threatened, and if they were to be adequately protected, a line of signal-stations pushed out to the north-east into the Kara-shahr valley would certainly suggest itself.” Stein (1928) Vol. III, p. 1227.

“The special interest to us of the Wei lio’s notice of ‘the central route’ lies in the fact that it makes a definite reference to the Lou-lan Site, almost contemporary with the documents found there, by its mention of ‘the ancient Lou-lan’, and that it details some of the chief stages on the desert journey by which the site was reached by travellers from the ‘Jade Gate’ and the westernmost extension of the ‘Great Wall’. The position of the last of these stages, the Lung-tui or ‘Dragon Mounds’, was first determined by me, in the course of my explorations of 1914, when I traced the line of the old Chinese route where it crossed the salt-encrusted ancient Lop Sea, some forty miles to the north-west of the station L.A.”. Stein (1921), p. 419.

“By deduction, one can accept that the route called ‘Central’ in the Weilue must coincide with the route called ‘Northern’ in the Hanshu. The fact that Yu Huan tells us that formerly there were only two known routes to the western countries, but that now a third, more northerly route, had been opened. Thus the only new route is the ‘Northern Route’. The ‘Central’ and ‘Southern’ routes are the same as those already followed during the period of the Former Han. We are, therefore, right to consider the ‘Central Route’ according to the Weilue as identical with the route called ‘Northern’ in the Hanshu. . . . ” Translated from Chavannes (1905), p. 531, n. 1.

The Hanshu describes the ‘Northern Route’ (i.e. the Weilue’s ‘Central Route’) as follows:

“The one [route] which starts from the royal court of Nearer Chü-shih [Turfan], running alongside the northern mountains and following the course of the river west to Shu-lo [Kashgar], is the Northern Route. To the west, the Northern Route crosses the Ts’ung-ling and leads to Ta Yüan [Ferghana], K’ang-chü and Yen-ts’ai.” CICA, p. 73.

“But it still remains for us to fix the location in detail of such intermediate stages as the text names, in the light of the knowledge now gained of the actual ground which the route crossed. For convenience of reference I may quote again that portion of the passage [from the Weilue] which concerns us here : ‘The central route is the one, which, starting from Yü-mên kuan, sets out on the west, leaves the well of the Protector-General, turns back at the northern extremity of the San-hung (‘Three Ridges’) [desert of] sand, passes the Chü-lu granary ; then, on leaving from the Sha-hsi well, turns to the north-west, passes through the Lung-tui (‘Dragon Mounds’), arrives at the ancient Lou-lan.’ Stein (1921), Vol. II, p. 555.

Apparently, a short cut, not mentioned in the Weilue or in the Han Histories, also existed between the region of Loulan and Miran (= Yuni – the early capital of Shanshan). At Miran (ancient Yuni) it rejoined the Southern Route via Khotan to Kashgar. For the identification of Miran as Yü-ni, the ‘Old Town’, the early capital of Shan-shan, see Stein (1921a), I, pp. 326 ff.
          This route was probably only used when political or other pressures demanded, as it crossed over 190 km of waterless salt crust, and was only really feasible in winter, when water could be carried in the form of blocks of ice:

“Another land route branched off near Loulan to Miran, where it joined the southern route. This section from Dunhuang to Loulan and Miran was rediscovered by Sir Aurel Stein in 1914. It was the shortest connection with Shule [Kashgar]. Moreover, since the Han dynasty it was protected by watch-towers from which could be transmitted smoke signals during the day and fire signals during the night. At the same time, however, it was very trying for men and animals, for a 190 km wide, waterless wasteland across the Lop Desert had to be traversed on a hard salt crust.⁵
          It was probably no better than now, a fact reported by Hedin as well as by Stein; namely, that the sensitive soles of the camels’ hoofs would be injured by the razor-sharp edges of the ground surface, until blood appeared. Then would come the painful operation of “re-soling” the camels to make them fit to go on. Pieces of leather literally would be stitched over the camels’ wounded heels!”

5. Marc Aurel Stein, Innermost Asia, Oxford, 1928, vol. 1, p. 285.

Baumer (2000), p. 9 and n. 5.

 From Korla the route led to Kucha, and Aksu, rejoining the Southern Route at Kashgar. The Weilue does not give details of the route past this point but it does list two ‘kingdoms’, Juandu and Xiuxiu, as dependencies of Kashgar. Both of these places (but with the variant Xiuxun) are mentioned in the Hanshu as being on the route from Kashgar to the Da Yuezhi, and both were near forks in the road which led to Ferghana (Da Yuan).
          Stein (1928), Vol. II, pp. 849-851 makes a very strong case for placing Juandu in the region of modern Irkeshtam, about 200 km west of Kashgar, on the modern border between China and Kyrgyzstan. This is near a major fork in the route here. One branch headed over the Terek Pass to Ferghana; the other led down the Alai valley, past Daraut-kurghān and Chat (where Stein locates Xiuxiu/Xiuxun – see notes 9.18 and 9.19), into the valley of the Surkhab (or Kizil-su), and then probably via the huge fortified Kushan city of Shahr-i Nau (40 km west of the modern Dushanbe – see note 9.22), and thence on to Termez, where it crossed the Oxus (or Amu Darya) and on to ancient Bactra (modern Balkh).
          Interestingly, the name Juandu can be translated as ‘Tax Control’, a function which Irkeshtam retains to this day. As Stein and many others have pointed out, the famous ‘Stone Tower’ of Ptolemy, where caravans from the west off-loaded their cargoes, must have been located not far to the west of Irkeshtam, in the Alai trough.

          “But during the centuries before and after the beginning of the Christian era, when Baktra was a chief emporium for the great silk trade passing from China to Persia and the Mediterranean, all geographical factors combined to direct this trade to the route which leads from Kāshgar to the Alai valley and thence down the Kizil-su or Surkh-āb towards the Oxus. Nature has favoured the use of this route, since it crosses the watershed between the Tārīm basin and the Oxus where it is lowest. Moreover, it has, in Kara-tēgin, a continuation singularly free from those physical difficulties which preclude the valleys draining the Pāmīrs farther south from serving as arteries of trade. According to the information received at Daraut-kurghān and subsequently on my way through Kara-tēgin, the route leading mainly along or near the right bank of the Kizil-su is practicable for laden camels and horses at all seasons right through as far as Āb-i-garm. From there routes equally easy lead through the Hissār hills to the Oxus north of Balkh.” Stein (1928), Vol. II, p. 848.

“Topographical facts, climactic conditions and local resources all support the conclusion that along the great natural thoroughfare of the Alai trough, which skirts the high northern rim of the Pamirs from east to west and is continued lower down by the fertile valley of the Kizil-su or Surkh-ab, “the Red River,” there once passed the route which the ancient silk traders coming from China and the Tarim basin followed down to the middle Oxus. Of this route Ptolemy, the great geographer of the second century A.D., has preserved for us an important and much-discussed record of Marinus of Tyre, his famous predecessor. This describes the progress made in the opposite direction by the trading agents of “Maës the Macedonian called Titianus” as they travelled from Baktra, the present Balkh, to the “country of the Seres,” or China, for the sake of their silk.
There is no need here to discuss the details which this record indicates as to the directions followed by the route. That it led up from the Oxus to the Alai had been established long ago by Sir Henry Yule, that great elucidator of early travel, when he proved that “the valley of the Komedoi,” through which the ascent toward Imaos is said to have led, could be no other than Kara-tegin, the valley of the Surkh-ab. Medieval Arab geographers still knew it by the name of Kumedh. The Kara-tegin valley and its eastern continuation, the trough of the Alai, offer in fact the easiest line of communication from the Oxus to the Tarim basin. But the advantages of the physical features which make the Alai particularly suited to serve as a natural highway between the two were brought home to me best by what the actual journey along it showed most clearly.
For fully seventy miles from where the Russian military road crosses it the open trough of the Alai stretches with an unbroken width of from six to eleven miles at its floor down to the Kirghiz village of Daraut-kurghan. Eastward for another twenty miles up to the Taun-murun saddle, where the route from the Kashgar side enters the Alai, the “thalweg” is equally wide and easy. Climactic conditions, moister than on the Pamirs to the south, provide everywhere ample steppe vegetation. Hence the Alai forms the great summer grazing ground for thousands of Kirghiz nomads who actually move up there from the plains of Farghana with their flocks, camels and horses. Well did I remember their picturesque caravans with camels carrying rich carpets, felts and other comfortable possessions of nomadic households as I had met them on their regular migration when I travelled early in June 1901 from Irkesh-tam to Osh and Andijan in Farghana. Now the warmth of the summer had made their camps seek the higher side valleys for the young grass, and thence they would descend later in the season to graze along the main valley. All the way the great snowy range to the south, with Mount Kaufmann [now known as “Lenin Peak” or “Pik Lenina” – 7,134 m or 23,406 ft] rising to close on 23,000 feet, presented grand panoramic views in the distance.
Long before reaching Daraut-kurghan, I came at an elevation of about 9,000 feet upon traces of former cultivation and remains of roughly built stone dwellings such as are occupied now by the semi-nomadic Kirghiz lower down during the winter months. similarly, on the Kashgar side cultivation is to be found at Irkesh-tam and above it to the same elevation. Thus wayfarers of old could be sure of finding shelter and some local supplies all along this ancient route except for a distance of less than seventy miles on the highest portion of the Alai. Though the snow lies deep on the Alai from December to February, the route would be practicable even then just as the Terek pass (12,700 feet), much frequented from Irkesh-tam to Farghana, is now at that season, provided there were sufficient traffic to keep the track open.
Such trade between the Tarim basin and the middle Oxus as was once served by the route through Kara-tegin and the Alai no longer exists. Balkh and the rest of Afghan Turkistan to the south of the Oxus have long ceased to see traffic passing from China. What little local trade comes up Karategin from the side of the Oxus proceeds from Daraut-kurghan to Marghilan or Andijan in Farghana, while exports from the Kashgar side find their way across the Terek pass to these places on the Russian railway.
Daraut-kurghan, where I was obliged to make a short halt for the sake of arrangements about transport and supplies, is a small place at the point where the Kara-tegin valley opens out toward the Alai. A Russian Customs post here guarded the frontier of Bukhara territory. Three miles farther down lies the village of Chat with a large, well-cultivated area and a ruined circumvallation of some size occupied during the troubled times preceding the Russian annexation of Turkistan. It is a point well suited for a large roadside station, and it is in this vicinity that we may safely locate the famous “Stone Tower” which the classical record preserved by Ptolemy mentions as the place reached from Baktra “when the traveller has ascended the ravine,” i.e. the valley of Kara-tegin. [In note 9.19, which see, I locate this site at Karakavak (Turkic for: ‘Black Poplar’ – Populus nigra L.), about half way along the fertile pasturelands of the Alai Valley at approximately 39^o 39’ N; 72^o 42’ E., rather than at Chat.]
It is equally probable that “the station at Mount Imaos whence traders start on their journey to Sera,” which Ptolemy’s account of the trade route to China as extracted from Marinus mentions on the eastern limit of the territory of the Nomadic Sakai, corresponds to the present Irkesh-tam. This is still a place well-known to those who carry on the lively caravan trade from Kashgar to Farghana and who face here the vagaries and exactions of the Chinese and Russian Customs stations, both established close to each other.” Stein (1931), pp. 223-227.

There can be little doubt that Juandu (‘Tax Control’), in the region of modern Irkeshtam is ‘the station (όρμητήριον) at Mount Imaos, whence traders start on their journey to Sēra’, according to Ptolemy. See Stein (1928), Vol. II, p. 850. See also the discussions by I. V. P’iankov in the notes 9.18 on Juandu, and 9.19 on Xiuxiu/Xiuxun.
An alternative route led from Tashkurgan (which could be reached either by heading south from Kashgar or southwest from Yarkand – thus, either from the ‘Southern’ or the ‘Middle’ Routes) past the Pamir Lakes via the Kushan-controlled regions of Wakhan and Badakhshān, and on to ancient Bactria. This route joined up here with the major east-west caravan routes leading from Chinese-controlled territory in the Tarim basin via the Alai, and past modern Dushanbe to cross the Oxus and reach Baktra (modern Balkh). See Stein (1931), pp. 232-242.

(c) The “New Route”

This route, called only the ‘New Route’ in the Weilue, has been confused with the ‘New Route of the North,’ by both Chavannes (1905), p. 533, n. 1, and Stein (1921), Vol. II, pp. 705 ff). See also note 4.3.
          The “New Route,” after it left Yumen guan [‘Jade Gate Frontier Post’], headed through Hengkeng [‘East-West Valley’], the wide Bēsh-toghrak valley which heads west towards Lop Nor. The ‘New Route’ seems to have followed the same path as the ‘Middle Route,’ for awhile, but then turned north before, or at, Bēsh-toghrak itself, thus avoiding some of the more difficult stages including the Sanlongsha [‘Three Sand Ridges’] and the Longdui [‘Dragon Dunes’].
          It then probably continued north across the desert, west of Hami, via the Palgan Bulak, Yulghan Bulak, and Biratar Bulak springs, to Lukchun in the Turfan oasis. Thence the route headed west, rejoining the Central Route before Kucha. See: The Times Atlas of the World.(1980), Map 24; The Contemporary Atlas of China. (1989), pp. 17, 18.
          The account of the ‘New Route of the North,’ on the other hand, ran via Hami to Eastern Qiemi, a dependency of Further Jushi, which was located immediately after crossing the gorge through the Bogda-shan mountains [called the Tianshan during the Han period], just north of Qijiaojing [Ch’i-chiao-ching]. The ‘New Route of the North’ then headed along the northern slopes of the massive range to the north of the Tarim Basin now called the Tianshan (and not to the south of it, as in the ‘New Route’), then through Wusun territory, and north of the Aral and Caspian Seas to reach Roman territory on the Black Sea (thus avoiding Parthian taxes on the caravan traffic).
          Qijiaojing was usually approached from the south via Hami (Yiwu). Here the road branched and one either went west to the Turfan oasis, or north through the Bogda shan mountains to the territory controlled by the king of Further Jushi in Dzungaria.
          The route through Hami was, and is, by far the easiest route to the north, and the only one with sufficient supplies for large caravans, but there is no mention of it at all in the itineraries of the Weilue.
          This strongly suggests that, at the time the information was gathered, Hami was out of bounds to the Chinese having once again come under the power of the Xiongnu. The Chinese captured and lost Hami several times during the Later Han Dynasty. “China finally lost control of it to the Xiongnu in 151 CE and did not regain it for over 400 years.” See Sitwell (1984), p. 174, in note 4.3.
          This explains the urgent development of a ‘New Route’ to provide communication with Turfan, which avoided Hami, for China no longer controlled it.

The Hanshu says:

“During the reign-period Yüan-shih [1-5 CE] there was a new route in the further royal kingdom of Chü-shih. This led to the Yü-men barrier from north of Wu-ch’üan, and the journey was comparatively shorter. Hsü Pu, the Wu and Chi colonel, wanted to open up this route for use, so as to reduce the distance by half and to avoid the obstacle of the White Dragon Mounds. Ku-kou, king of the further state of Chü-shih, realised that because of [the passage of] the road he would be obliged to make provisions available [for Han travellers] and in his heart thought that this would not be expedient. In addition, his lands were rather close to those of the southern general of the Hsiung-nu. . . . [Ku-kou was finally beheaded by the Chinese for disobedience].” CICA: 189-190, 192.

“I have explained elsewhere how this ever-present threat of the Huns [from 121 BCE to 73 CE] from across the northernmost T’ien-shan determined the direction of the ‘new northern route’ {note – this should read, simply, the “New Route”} which the Chinese in A.D. 2 opened from the ancient ‘Jade Gate’ in order to communicate with ‘Posterior Chü-shih’ or the territory around the present Guchen. To reach this ground, which, like Turfān immediately to the south, had passed early under their control, the route via Hāmi would undoubtedly have been the easiest. Yet Chinese administrative policy, was always disposed to face physical difficulties rather than risks from hostile barbarians, kept the new road well away from Hāmi and carried it through waterless desert wastes which at least offered protection from those dreaded nomadic foes.” Stein (1928), I, pp. 539-540.

This route left the Yumen frontier post and then headed west through part or all of the Hengkeng (literally; ‘East-West Gully’ = the present Bēsh-toghrak valley), and then directly north, past the still unidentified Wuchuan (‘Five Boats’), some 300 km across the desert to the town of Gaochang, at the southern tip of the Turfan Basin. From here it led on to Karashahr and joined the Middle Route near modern Korla. Almost all authors place the Yiwu(lu) 伊吾盧 [I-wu-lu] of the Han period in the region of modern Hami or Qumul.

“Known as Khamil in Mongolian, the name of this important Silk Road town is transcribed in Modern Standard Mandarin as Hami. It is famous for its succulent melons suffused with fragrance and sweetness. Large amounts of cotton are also grown in irrigated fields.” Mallory and Mair (2000), p. 13.

A few scholars, however, identify Yiwu with the modern settlement of the same name (and written with the same characters) about 160 km by road northeast of modern Hami, across the Karlik range. See, for example, de Crespigny (1984), pp. 43 and 522, n. 71. This tiny settlement is also known as Aratürük or Atürük. For the derivation of the name Qāmul = modern Hami, see Bailey (1985), p. 10.
          I have chosen the traditional identification, however, placing it near modern Hami, on the grounds that the strategically important and famously fertile Hami oasis is a far more likely location for the State Farms the Han established at Yiwu than the rather limited agricultural potential of the region surrounding modern Yiwulu / Aratürük. Pelliot places Yiwu some 30 miles (48 km) west of the town of modern Hami:

“In A.D. 73, the Chinese created a military colony in the region, with a walled city. . . . The military colony of I-wu-lu or I-wu did not thrive like that of Kao-chang..., and was abandoned with the whole of the region in 77. A new occupation in 90 was still less durable. The third effort, in 131, was more successful, but only for a time, and Qomul had already passed out of Chinese reach at the end of the Han dynasty. . . . As to the I-wu-lu of the Han, which the commentary of 676 on the two passages of the Hou Hanshu calls the ancient small town of I-wu, it was located about 30 miles west of Qomul [Hami], in the district of Na-shih....” Pelliot (1959), p. 155.

“Perhaps the earliest reference to Hami – or Yiwu, Yizhou or Kumul, as it was variously known – was in a book, made of bamboo slips and bound together with white silk, found in a second-century BC tomb in Henan Province. This record, discovered in the third century, is an account of the quasi-mythical travels of Emperor Mu, the fifth emperor of the Zhou Dynasty (1027-256 BC), who, on returning from his visit to the Queen Mother of the West, stayed in Hami for three days and received a present of 300 horses and 2,000 sheep and cattle from the local inhabitants.
          Hami was considered by the Chinese the key to access to the northwest, but they were not always successful in keeping the city free of nomadic incursions. In 73 BC [sic – should read AD] the Han general Ban Chao wrested the area from a Xiongnu army and established a military and agricultural colony. . . .
           Like Turpan, Hami is in a fault depression about 200 metres (650 feet) below sea level, and temperatures are extreme, from a high of 43^o C (109^o F) in summer to a low of -32^o C (-26^o F) in winter.” Bonavia (1988), pp. 105; 110-111.

“Cumul occupies a geographical position of great strategical importance. Like Ansi on the south, so Cumul on the north is a stepping-off and landing-place for all travellers who cross the inhospitable tract of Gobi between the provinces of Kansu and Chinese Turkestan. The approach to the oasis is by long and desolate stages, but from the moment that the traveller’s foot touches watered land he is in the midst of beauty and luxuriant agriculture, and for several miles before reaching the town the road leads through fields and by farmhouses surrounded with elm and poplar trees. Everything indicates prosperity and an abundance of every product.” Cable and French (1943), p. 138.

“Beyond Hami the track led to Tsi-kio-king [Qijiaojing – not quite 200 km northwest of Hami], the Seven-Horned Well, which stands as sentinel where north and south trade-roads divide, each taking its own way on one or the other side of the dividing mountain range. The old well watches the South Road disappear over a dismal gravel plain toward the burning oases of the Flame Hills [north of Turfan], and the North Road enter the narrow tortuous defile which cuts the Tienshan range of mountains in two. In times of peace Seven-Horned Well was a dreary hamlet, but in war-time it became a strategic desert outpost from which soldiers guarded three main arterial roads toward Turfan, Hami and Urumchi. It has been a scene of fierce Gobi battles, and its sands have many a time been reddened with blood and littered with the bodies of men and carcasses of horses. Every invader covets its strategic position and knows of its tamarisk growth, which, though smothered by sand, will supply abundant fuel for his army. . . . The southern road kept south of the mountain range, past East Salt Lake and West Salt Lake to Turfan, and over the steep Dawan Pass to Urumchi. The northern road, however, led through a jagged cut in the Tienshan where, for a long nine-hour stage, a narrow and almost level path wound with innumerable turns between great bare crags and lofty granite cliffs, emerging at last on the Dzungarian plain.” Ibid. 297-298.

“This constant liability to northern attack, from which Hami has suffered whenever Chinese power in Central Asia weakened, is fully illustrated by its chequered history, as recorded in the Chinese Annals, and right down to our own times. . . . As regards the former [Han] period, it will suffice to point out that within four years of the first establishment of a Chinese military colony in A.D. 73 I-wu was lost again to the Hsiung-nu; reoccupied between A.D. 90-104, it suffered once more the same fate. The notice concerning the re-establishment of a military colony there in A.D. 131 brings out clearly the strategic value which the Chinese rightly attached to Hami. But obviously their hold upon it ceased when imperial control over the ‘Western regions’ was abandoned after the middle of the second century.” Stein (1921), p. 1149.

From just north of Lop Nor all the way to the Turfan Basin across the dreaded Gashun Gobi, there is a string of salt springs. During the Han, and up until about 270 CE, however, the whole region was much wetter than it is now, so they may have been fresh enough at the time to provide water for the camels, at least. Wild Bactrian camels still live in the area and have apparently adapted so they can drink the water from the salt springs:

          “There is no fresh water in the Gashun Gobi, only salt springs. No humans, not even the hardy nomad, can survive in this utterly barren area of over 1,750 square kilometres. The only inhabitant of this huge space is the wild Bactrian camel. Far removed from contact with domestic Bactrians and fully adapted to drinking salt water, the camels migrate from water point to water point, some of which are over 100 kilometres apart.” Hare (1998), p. 80.

As this route was protected from the raids of the Xiongnu by the empty desert to its east, and because it was less than half the distance of the difficult, but better-watered, route through Loulan, Korla and Karashahr to Turfan, it may have proved economical for the Chinese, when the route via Hami was not available to them, to set up strategic caches of supplies along this route.
          Stein (1928, Vol. I, p. 319) reports that some of his party found old tracks and their guide “took them to mark the passage of some Mongols making for Tun-huang from the western Kuruk-tāgh. On questioning, the guide told him that his grandfather “who like his father had been a hunter of wild camels and familiar with the wastes of the Kuruk-tāgh, knew vaguely of a route leading through them to the Tun-huang side.”

(d) The “New Route of the North”

The “New Route of the North” or “New Northern Route” ran to the north of the Bogdo Shan Mountains through Further Jushi (near modern Jimasa) and then almost parallel to the Central Route, and north of the Tian Shan, past modern Urumchi to the Wusun.
          The distance of 208 km mentioned in the text matches that between Turfan and the region of modern Guchen and Jimasa on modern maps. The town of Jinman 金滿城 [Ch’in-man] is described as houbu 後郶 in the Hou Hanshu which translates as something like: “The Headquarters for Further [Juzhi]”

“According to the Xiyushuidaoji (chap. III, p. 5 a), the site of the ancient Beiting is none other than the locality of Hubaozi, about twenty li to the north of the present sub-prefecture of Baohui. In fact, a Tang period stele has been found at this place which, although badly damaged, categorically proves that previously the sub-prefecture of Jinman was to be found here. Now, here is what one reads in the Jiu Tangshu (chap. XL, p. 29 b): “Jinman... was, during the Later Han, the Posterior Royal Court (of the kingdom) of Jushi. In the ancient barbarian court, there were five towns. The common name was therefore, the ‘Territory of the Five Towns.’ In the 14th Zhengguan year (640), after (the kingdom of) Gaochang (Yarkhoto) had been pacified, the District of Ting was established.” Several lines above, one reads in the same work that, in the second Changan year (702), the Protectorate of Beiting was created from the District of Ting. Thus this text confirms the opinion of the Xiyushuidaoji, for it proves that Beiting is Jinman. Now, we know, from an inscription found in situ that Jinman was 20 li to the north of Baohui xian (or Ximusa) which is 90 li to the southwest of Guchen. Besides, this text shows us that the name of Bishbalek (the five towns), that the Government of Beiting had under the Mongols, corresponds to a very ancient name already known in the T’ang period. Bishbalek is, therefore, not Urumchi. Like Beiting, with which it is identical, it is at some distance to the west of Guchen.” Translated and adapted from: Chavannes (1900) pp. 11, and 305, n.

Aurel Stein found the ruins of the old town about 10 km north of modern Jimasa, just beyond the village of Hu-p’u-tzu (Hupuzi): “The outer walls... appear to have once enclosed a roughly rectangular area, measuring approximately 2,160 yards [1,975 m.] from north to south and 1,260 yards [1,152 m.] from east to west.” Stein (1928), pp. 554-559. See also: CICA: 184, n. 622.
          The Hou Hanshu states that the king of the Posterior Jushi lived in the Wutu valley, which was 500 li [= 208 km] from the residence of the Jangshi [‘Adjutant General’] in Lukchun. As I measure it, this is exactly the distance between Lukchun and a point about 10 km beyond Jimasa on two maps of the route from Turfan to Guchen. See: Stein (1928), Map 28, and the U.S. Defence Mapping Agency’s ONC, Sheet F-7. This confirms the findings of Chavannes and Stein, making the identification virtually certain.
          Although this short route was probably used for military communications and the like between Nearer and Further Jushi, it could never have been a major caravan route:

          “I may point out here that the direct tracks leading from Turfān to Guchen across the high, snowy portion of the T’ien-shan intervening are only open for a part of the year, and, as my crossing in 1914 of the least difficult of the passes, the Pa-no-p’a, showed, impracticable at all times for any but the lightest transport. Trade caravans and military convoys would at all times have to make a great detour either west (via Urumchi) or east (via Ulan-su) in order to get round the Bogdo-ula range by a route practicable for camels or carts.
          This point has to be borne in mind when we compare the two routes referred to in the notice of the Former Han Annals. The ‘new route of the north’ coming from the Shona-nōr must have crossed the T’ien-shan by the easy and low saddle north of Ch’i-ku-ching over which the present Chinese cart-road from Hāmi to Guchen and Urumchi passes.” Stein (1921), p. 706, n. 6.

There were two main routes through Wusun territory. One ran west through Santai (near Lake Sairam) and then over the Talki Pass into the Ili valley. From Urumchi the route ran west through Manass.
To the west of Manass there were two main passes south into the Kax He [K’o shih Ho] Valley which led on to modern Yining or Guldja near the junction of the Kunes and the Ili Rivers. The first headed over the mountains through modern Dushanzi to the south of Kuytun and Usa. The second pass, further east, led to the south of Lake Ayram:

“Another [of the so-called “Iron Gates”] is the defile of Talki leading from the Sairam (nor) or Sut (Kul) lake southward, to the Ili River. This was called Kulugha by Turki-speaking people, and Timur-Khalaga by the Mongols ; and Dr. Bretschneider explains that the word Khalaga or Khalga, means, in Mongol, a pass or gate, while Timur signifies iron. The Chinese traveller Chang-Te, in 1259, passed through the Talki defile, and described it as “very rugged with overhanging rocks.” He speaks of it by a transliterated Mongol name which stands for “iron roadway.” Elias (1895), p. 20, n. 3. The TCAW marks this pass as the “Xin’ertai.”

This route continued to the north of Issyk Kul through modern Almaty to Tokmak and Bishkek. All these routes were accessible to camel caravans:

“But his first words contained a test, and I had failed to meet that test. ‘The camel caravans did not cross the mountains,’ he had said. I should have corrected him, politely; knowing he had been referring to these mountains, to the Tien Shan.
          The camel caravans had crossed these mountains, through the eastern passes coming from the Middle to the Northern Route; making their way from Gao Chang, from Turfan. And they crossed them again from the west, travelling the Northern Route that led to Kuldja and the Ili.” Martyn (1987), p. 432.  

The alternative route (which would have been safer from the attacks of northern nomads) ran even further south, along the Kekes River Valley to Issyk-kol where, according to the Hanshu the Wusun had their capital at Chigu (“Red Valley”) – which I have located in the spectacular Jeti-Öghüz Valley just southwest of modern Karakol near the Issyk-kol lake itself (see CWR note 1.61).
          From Issyk-kol the way continued around the northern perimeter of the lake (“Most of the population and agriculture, and all the decent roads, are along the north shore.” King, et al. (1996), p. 375), and on via Tokmak to Bishkek, where it joined the previous route to Talas in Kangju territory. From there one could head through the Ferghana Valley to Khojend (Khodzhent or Kujand, known as Stalinabad during the Soviet era).
          It was, as closely as I can measure it on my maps, about 820-830 km between Jeti-Öghüz and Khojend by this route. So, it seems fair to assume that this is route mentioned in the Shiji, ch. 123, which records that “Wusun is situated some 2,000 li [832 km] northeast from Dayuan.” See TWR, note 1.61 for more details on these identifications, which find valuable additional support from this notice.
          From Talas there were three main branch routes: the one mentioned in the Weilue ran northwest along the Jaxartes or Syr Darya north of the Aral and the Caspian Seas to the land of the Yancai or Alans who, at that time, were living to the north of the Black Sea and stretching over to the western and northern shores of the Caspian. The Weilue states the Yancai bordered on Da Qin (Roman territory) which is undoubtedly a reference to the Roman territories in Armenia from which there was access to ports on the Black Sea, and from there to the Mediterranean.
          The other two routes ran south from Talas through Northern Wuyi (modern Khojend) to the region of modern Samarkand, where one branch went southwest through the oasis of Bukhara and Merv to Iran, and the other branch headed south through Termez (ancient Dumi – one of the five main divisions of the Yuezhi mentioned in the Hou Hanshu) and across the Oxus (or Amu Darya) to Bactra (Balkh). From there one could travel southeast to Gaofu (Kabul) and India, or southwest through Herat to Iran.

(e) North-South Routes across the Tarim Basin

“It is true that the Southern and the Middle Silk Roads were separated from each other by the Taklamakan Desert, but since the Bronze Age there were north-south connections along a few rivers such as the Khotan Darya and the Keriya Darya. Since after a great thaw in the Kunlun mountain range the waters of the Khotan Darya cross the desert and reach the Tarim even today, this transverse line linking Khotan to Aksu and going along the river has never been abandoned.
          About 180 km north of Khotan a mountain range with a reddish hue rises up from the desert plain. . . .
          The Mazar Tag chain of mountains ends next to the Khotan Darya. Here on a rocky ledge about 150 m high, the well-preserved Mazar Tagh fort proudly looks down on the river and watches over the former trade route. The position of the fort was almost impregnable, for the rock face near the southern crest falls almost vertically and is also quite steep in the east, while a tower at a distance of about 30 m protects the northwestern access. This massive 6 m tower reminds one of the limes of the Eastern Han and Jin eras between Loulan and Dunhuang. . . .
          The tower is certainly the most ancient structure and could date from the 3^rd or 4^th century AD. . . . ” Baumer (2000), pp. 67 and 69.

          “A look at the map as well as analysis of the satellite photographs show that a trade route along the Keriya Darya would be the shortest connection between the two former kingdoms of Khotan and Kucha. If one believes the report of Mirza Hidar, Kashgar prince and historian, the Keriya Darya was supposed to have reached the course of the Tarim as early as the 16^th century. We may therefore surmise that Karadong, which was halfway between Keriya (today’s Yutian) and the Tarim south of Kucha, had been a fort at the beginning of our era, besides serving as a caravanserai for trade caravans. . . .
          The excavations of ancient Karadong by a Sino-French team, which began in 1991, have brought to light sensational finds and caused a reassessment of Karadong’s importance. First, in an area five kilometres long and three kilometres wide next to the fort, the archaeologists found twenty ruined houses, a temple and forty other ceramic sites that indicate completely destroyed houses or ceramic kilns. Most structures are concentrated in the northern half of the oasis south-east of the caravanserai, in an area of 1300 m by 800 m. In the southern half, an intricate irrigation system could be identified, extending more than three kilometres in a north-south direction. Coins from the Han Dynasty and numerous remnants of millet, wheat, oats and rice were also found. In view of the extended irrigation network, conceivably one or more of these cereals had been produced in Karadong itself. . . .
          As previously described, forty kilometres north of Karadong are the extended ruins of the proto-historic town of Yuan Sha as well as traces of even older settlements. Since Yuan Sha had been abandoned shortly before the turn of the era in favour of Karadong, the golden age of the latter settlement must have been in the first two centuries of our era. In those times Karadong was part of the Yumi principality which extended as far as Keriya. The complete lack of coins from the Tang Dynasty and of artefacts of a younger date than the 4^th century AD leads to the conclusion that Karadong must have been abandoned in the 4^th century AD. The political disturbances after the breakdown of Chinese authority in the 3^rd century AD must have led to a recession of trade, depopulation, and as a consequence, neglect of the irrigation canals, which favoured the advance of the desert. The Keriya Darya probably transferred its river bed eastwards during this period, a fact that also made living conditions more difficult.” Baumer (2000), pp. 93, 95, 96.

B. The territories of Haixi, Haibei and Haidong

(a) Haixi 海西 – literally: ‘West of the Sea’ = Egypt.
          Haixi, and the associated names, Haibei 海北 – ‘North of the Sea’ – which provided an overland route between Mesopotamia and Egypt; and Haidong 海東 –‘East of the Sea,’ have continued to elude firm identification in spite of detailed treatment by recent scholars. Haixi and Haibei are first mentioned in the Hou Hanshu, and Haidong in the Weilue.
          These regions or countries are presumably located in the various directions in relation to Xihai, or the ‘Western Sea,’ (sometimes called, Dahai, or the ‘Great Sea’).
          Although there has been some speculation that Xihai and Dahai might refer to the Mediterranean or even the Black Sea, all the evidence points to both names being used for the Indian Ocean which, to Chinese as well as the Romans, included the Persian Gulf and the Red Sea. See Leslie and Gardiner (1996), Chap. 20.8, pp. 271-272.
          There is a detailed account of these territories in David Graf’s article, Graf (1996) – especially the section on ‘The Western Regions’ on p. 204, and the map at the end. However, he argues that the use of the terms Haixi 海西 [Hai-hsi], Haibei 海北 [Hai-pei], and Haidong 海東 [Hai-tung] indicate: “that the Chinese of the Han era were ignorant about the existence of the Arabian peninsula. For them, the great sea adjacent to the Persian coasts stretched westward forming an immense bay that extended all the way west to the coasts on Ta-ch’in. Their belief in this imaginary body of water resulted in the creation of the three coastal districts.”
          I cannot agree with this analysis. It is true that the Chinese, like the Romans, and the Greeks before them, considered the Indian Ocean and its two major Gulfs, the Red Sea and the Arabian (or ‘Persian’) Gulf as a whole. The Greeks referred to it as the Erythraean Sea. This is perfectly reasonable and accurate, as easily navigable entrances join all the waterways. Because the Chinese accounts do not mention the Arabian Peninsula does not mean they were necessarily ignorant of it.
          In my view, the Chinese division of these regions makes excellent sense. Thus we have: ‘West of the Sea’ (= Egypt); ‘East of the Sea’ (the lands on the east coast of the Persian Gulf, or Persis) and, finally, ‘North of the Sea,’ the region in between and joining them: (probably northern Saudi Arabia, Jordan and southern Israel). The Weilue mentions an overland route through Haibei from Parthian territory to Egypt (Haixi):

“Now, if you leave the city of Angu (Gerrha) by the overland route, you go due north to Haibei (‘North of the Sea’), then due west to Haixi (Egypt), then turn due south to go through the city of Wuchisan (Alexandria).”

This could refer to the long route up the Euphrates through Palmyra and Dura Europa, from where it turns south, and later west, to Egypt. There were two rather more arduous, but shorter, and more direct alternatives. the first of these went west from the head of the Persian Gulf, across the desert to the oasis of al-Jawf (Dumatha). Here the road forked, and one could head north up the Wadi Sirhan towards Damascus, or west towards Petra, Rhinocolura, and Egypt. It seems these routes were guarded by Roman patrols after their annexation of Nabataea in 106 CE. Bowersock (1996), pp. 157-159; Millar (1993), pp. 138-139.
          The other route left the region of the prosperous trading state of Gerrha in eastern Arabia and took one across the peninsula either to al-Jawf or to the Nabataean city of Hegra (modern Meda’in Salih – some 25 km north of the ancient site of Dedan) and on to the port of Leukê Komê (literally, ‘White Village’), which is probably be located in the vicinity of modern Egra = Al Wajh, 26° 13’ N, 36° 27’ E, on the east coast of the Red Sea.

“Much of the merchandise of the Orient was brought overland from the port of Gerrha on the Persian Gulf to the Arabian port of Leukê Komê on the east side of the Gulf of Aqabah and then shipped or transported by caravan northward to Aila. From there it was carried to Petra, to which a direct, overland route led also from Meda’in Aleh in Arabia. And “thence to Rhinocolura (modern el-Arish in Sinai on the Mediterranean) . . . and thence to other nations,” according to the Greek geographer Strabo, who wrote about the Nabataeans at the beginning of the first century A.D.” Glueck (1959), pp. 269-270. See also note 16.1.

“For this trade [with Elymais and Karmania] they opened the city of Carra [Gerrha] where their market was held. From here they all used to set out on the twenty-day march to Gabba and Syria-Palestine. According to Juba’s report they began later for the same reason to go to the empire of the Parthians. It seems to me that still earlier they brought their goods to the Persians rather than to Syria and Egypt, which Herodotus confirms, who says the Arabs paid 1,000 talents of incense yearly to the kings of Persia. Juba (c. 25 BC-AD 25) and Pliny, NH (AD 77) 12. 40. 80).” Potts (1990), pp. 90-91.

“The merchants of Palmyra were also active in Egypt. One group resident in Coptos was engaged in the commerce of the Red Sea and thus by implication possibly also with India and East Africa. Others used the overland route from Mesopotamia to Denderah in Egypt.” Raschke (1976), p. 644.

The suggestion that best that fits all the evidence is that Haixi refers to Egypt. Egypt is certainly to the west of the Red Sea (which was considered an integral part of Xihai – the ‘Western Sea’), and the major Roman ports in the Red Sea which were the termini of the extensive maritime trade with India and Parthia, were located along the eastern coast of Egypt – quite literally ‘west of the sea’. The use of Haixi as a name for Egypt was probably reinforced by the fact that the characters represent a reasonably close phonetic approximation of the name into Chinese.
          A major source of confusion has been the identification of Haixi with Da Qin in the Chinese texts as, for example, in the section on Da Qin in the Hou Hanshu: “The kingdom of Da Qin (Rome) is also called Lijian. As it is found to the west of the sea, it is also called the kingdom of Haixi.”
          This does not seem to be contradictory to me – Egypt had been under the control of Rome since 30 BCE and was, therefore, considered ‘Roman territory.’ Also, almost all freight being shipped from the East to ‘Rome’ went through Egypt, which was the first territory mariners reached which could be called ‘Roman.’ Merchants from Egypt may well have referred to themselves as Romans, as many would have been officially Roman citizens, even before Caracalla’s edict:

          “In AD 212 Emperor Caracalla issued his famous edict granting Roman citizenship to all the inhabitants of the Roman Empire (only the ‘capitulated’, whose identification remains a matter of scholarly dispute, were excluded).” Lewis (1983), p. 34.

It is not surprising to find both the names for Rome and Egypt interchanged at this time in distant China. This identification of Haixi as Egypt is, I believe, amply confirmed in the passage from the Weilue.
          To add weight to my contention that Haixi = Egypt – the Weilue says that from Parthian territory one can sail directly to Haixi – which, in itself, strongly indicates Egypt. The only other Roman-controlled territory which could be reached by sea from the East were the Nabataean lands, annexed by the Romans in 106 CE, in the northeast corner of the Red Sea and included the Gulf of Aqaba and the port of Leukê Komê. See note 16.1.
          The port of Aila (also known, at various times as: Aelana, ‘Aḳaba, Elath, Ezion-geber, Ailath and Ailam), at the head of the Gulf of Aqaba, was very difficult to sail to because of the unfavourable prevailing winds and would have been quite unsuited to handle the large ships the Romans used in the India trade. 
          The estimate in the Weilue of two months for the journey to Egypt with good winds seems very reasonable. The reference to it taking up to three years with no wind is probably only a repeat of the discouraging reports given to Gan Ying in 97 CE by Parthian sailors.
          The reference to a river “flowing out of the west of the country into another great sea,” is clearly a reference to the Nile. This certainly puts an end to any of the speculation (as discussed above) that Haixi might refer to the Nabataean territories.
          Some have argued that the Nile doesn’t flow out of the west of Egypt, but out of its north. If, on the