Routes & Cities
Silk Road Routes and Map
The phrase "Silk Road" suggests a single paved highway from China to Rome, and nothing could be further from the truth. The Silk Road was in reality a…
The phrase “Silk Road” suggests a single paved highway from China to Rome, and nothing could be further from the truth. The Silk Road was in reality a shifting web of caravan trails, river crossings, mountain passes, and sea lanes, running more than 8,000 kilometres from the Pacific to the Mediterranean. Its geography was determined by water, by oasis, by the political safety of the corridors, and by the willingness of empires to keep the routes open. At its height, the network had multiple parallel branches, dozens of major oasis cities, and a parallel maritime system that handled the bulky goods that camels could not. I should say at the outset that any single map of this network is, by definition, a lie — the trunk lines drawn on modern maps are averages over centuries, and the real routes shifted with the season, the political weather, and the herd movements of the caravans.
This article is a guide to that geography. For a chronological account of how the network opened and changed, see the history of the Silk Road. For a survey of the great cities, see Silk Road cities. For an introduction to the goods that moved through the system, see Silk Road trade, goods, and economy. And for the broader overview, see the main Silk Road guide.
The eastern terminus: Chang’an
Every overland Silk Road began at Chang’an, the modern city of Xi’an in Shaanxi province. Chang’an was, in turn, the capital of the Han, Sui, and Tang dynasties, and was for most of the imperial period the largest and most cosmopolitan city in the world. The road west from Chang’an followed the Wei River valley, climbed into the mountains of the Qinling range, and entered the Hexi Corridor — a long, narrow strip of oasis and grassland that runs for nearly 1,000 kilometres between the Tibetan plateau and the Gobi Desert, and forms the only viable land route between the Chinese heartland and the Tarim Basin. A modern traveller on the Lanzhou-Xinjiang high-speed rail, which follows roughly the same alignment, covers the corridor in about eight hours; a Tang-era caravan needed six to ten weeks.
The Hexi Corridor was anchored by a chain of garrison towns — Wuwei (the Liangzhou of the Tang), Zhangye (the former Gaochang of the Han), Jiuquan — and by the great Buddhist oasis of Dunhuang, at the western end of the corridor, where the road divided. The Mogao Caves at Dunhuang, carved between the fourth and fourteenth centuries, contain some of the finest Buddhist art in the world, and the manuscripts discovered in the Library Cave in 1900 preserve tens of thousands of documents in dozens of languages. The site is covered in the article on Dunhuang.
All of this is, of course, a generalization. The Hexi Corridor in winter is a different route from the Hexi Corridor in summer, and the Tang-era alignments were not the same as the Ming-era ones.
The Tarim Basin: northern and southern routes
West of Dunhuang, the road entered the Tarim Basin, an oval of desert roughly 1,500 kilometres across, dominated by the Taklamakan — one of the most inhospitable sand seas in the world. The Taklamakan could not be crossed directly. Caravans went around it, on one of two parallel branches. The “two-branch” picture is itself a simplification — there were at least four, depending on how one counts — but it is the one I will use here because it is the one the sources use.
The northern route ran along the northern edge of the Tarim, threading through a chain of oasis cities fed by rivers flowing south from the Tianshan mountains. The major nodes were Turfan, Yanqi (the ancient Agni, or “Horse-kingdom” of the Chinese sources), Kucha, and Aksu, and the route emerged at Kashgar, in the far west of the basin. Kashgar sits at the convergence of the northern and southern branches and at the foot of the passes leading into the Pamir mountains and Ferghana. The Kashgar Sunday market, which still operates in the old livestock bazaar near the Id Kah mosque, was for centuries one of the great commercial fairs of Central Asia, where Sogdian, Uyghur, Tibetan, Indian, and Chinese merchants exchanged goods in a dozen languages. I have not been able to verify the popular claim that the market has operated continuously since the tenth century; the documentary record is patchy until the Qing period, and the more I look the more cautious I become.
The southern route ran along the southern edge of the Tarim, threading through Khotan, Yarkand, and other oases fed by rivers flowing north from the Kunlun mountains. The southern route was the older of the two, and was anchored for most of its history by the kingdom of Khotan, which was famous for its jade (the white “sheep-fat” jade of the Khotan rivers is the prestige variety), its Buddhist monasteries, and its textiles. (The trade in precious stones is treated in a separate article.) The southern route rejoined the northern branch near Kashgar, although for several centuries the two branches competed, and traders chose between them based on the political weather — the Khotan kingdom, the Kara-Khanid khanate, and the Uyghur kingdom of Turfan all had their periods of control, and each shift moved traffic.
A third branch, sometimes called the southwestern route, ran from Khotan southwest through the Pamirs into northern India and Kashmir, via the Baroghil and Kilik passes. It was the route by which Buddhism entered China, and it carried pilgrims and missionaries more often than bulk cargo. The Tang monk Xuanzang used a version of this route on his return from India in 645 CE, climbing the Pamirs in winter and narrowly losing most of his party to cold and altitude — the account in chapter 12 of his Great Tang Records on the Western Regions is, in my view, the most gripping piece of Silk Road travel writing that survives.
The actual distance from Chang’an to Antioch, by the most-used land route, was about 7,000 kilometres — roughly the distance from London to New York and back. The caravan journey took, in good conditions, six to nine months each way. The reader who has flown from London to New York in eight hours has some idea of the absolute scale of the undertaking, though the comparison flatters the modern traveller; a winter crossing of the Pamirs at four thousand metres was a different sort of thing entirely.
A personal note
I should admit that any map of the Silk Road is partly a work of fiction. The routes shifted with the seasons, the political weather, and the herd movements of the caravans; the trunk lines drawn on modern maps are an average over centuries. The reader looking for a single “right” route will be disappointed. There was no single right route. There were a dozen, and they changed.
Further reading
This is one of those topics where the standard reading list is long and the good reading list is short. Three works that I (the author) keep returning to:
- The relevant Oxford Handbook or Cambridge History chapter — these are uneven but the good ones are very good.
- A recent issue of Central Asiatic Journal (Brill) for what scholars are actually arguing about right now.
- One primary source in translation. The Dunhuang manuscripts, in particular, are now mostly online; the British Library’s IDP database is the place to start.
Avoid the popular histories. They are usually wrong in small ways and large ones.