The Silk Road's Lost Cities: Echoes of Forgotten Trade

The Silk Road's lost cities were oasis towns that lived on river water, and they died when the water failed or an army came. Loulan and Niya, on the edges of China's Taklamakan Desert, dwindled as the rivers and lakes that fed them shrank, while Merv, in Turkmenistan, was sacked by the Mongols in 1221 and never regained its old wealth. In the dry desert air, much of what they left behind survived, down to the wooden letters their officials wrote. Nobody rebuilt on top of them, so their ruins read like an archive of how the Silk Road worked from the inside: who paid in silk, who quarrelled over camels, and how much hung on a few mountain rivers.

Why did Silk Road cities grow in the middle of deserts?

Because that is where mountain water ran out onto the plain. The Taklamakan, which occupies the heart of the Tarim Basin in Xinjiang, western China, covers about 337,000 square kilometres, 85 percent of it drifting dunes, and its centre gets no more than about 20 millimetres of rain a year, according to the Encyclopaedia Iranica. Rivers from the Kunlun, Pamir and Tian Shan ranges reach its edges, and wherever they spread, people could farm. The result was a ring of oases, Kashgar, Khotan and Niya among them, and the caravan routes kept to it, moving goods along one oasis at a time. The desert's name is often rendered as a warning that whoever goes in never comes out. The same encyclopedia offers a plainer origin: a Uyghur name probably built from the Arabic tark, 'relinquish', and makan, 'place'. For a desert ringed with abandoned towns, that fits well enough. China's Han dynasty turned the ring into a frontier. It extended the Great Wall's chain of garrisons and beacon towers west toward Dunhuang, and from the mid-first century BCE its soldiers were digging irrigation channels and opening fields along the basin's rivers. Loulan sat at the eastern end of the basin, near the great lake of Lop Nur. Chinese records first mention it in 176 BCE, and it came under Chinese control around 100 BCE, according to the Victoria and Albert Museum. For all its fame it was small: the Book of Han, the dynasty's official history, counted 14,100 people in the entire kingdom. Its position, not its size, made it matter.

Why were Loulan and Niya abandoned?

Their water ran out, though what took it is still debated. Records from the time describe rivers carrying less water and soldiers' grain rations being cut after about 270 CE, and Loulan was abandoned in the 4th century. When the monk Xuanzang came home through the region in the 640s, after years of study at Nalanda, India's great Buddhist university, the desert towns stood empty. For a long time nature was the obvious suspect, and earlier studies mostly blamed a drying climate. A 2017 study in Scientific Reports points at people instead. Its authors studied a 5.7-metre pit dug into the dry lake bed and found wind-blown sand from about 1,800 years ago, a sign that the lake had nearly vanished. Lakes inside the Han empire's territory fell or dried out at that time, while those in neighbouring regions rose in a relatively wet climate. Their verdict is that irrigation on the middle Tarim and Konqi rivers drew off so much water that too little reached Loulan, a man-made disaster they liken to the Aral Sea's. Niya's timbers tell a similar story. Radiocarbon dates show human activity there shifting southward over time, as if the settlement crept upstream after its retreating water. The theory has a modern echo, too: Lop Nur, about 2,000 square kilometres in the early 1950s, dried up entirely in 1962, after large-scale upstream agricultural development and water conservancy works on the Tarim and Konqi rivers cut its inflows.

What did the Niya tablets reveal about daily life?

Aurel Stein began digging at Niya in 1901 and over several expeditions recovered hundreds of documents, mostly rectangular or wedge-shaped wooden tablets tied shut with string and sealed with clay. They are written in Kharoshthi, a right-to-left script from Gandhara, in today's northwestern Pakistan, in a Prakrit language related to Gandhara's own, and belong to the 3rd and 4th centuries CE. They are the paperwork of a kingdom known from Chinese records as Shanshan. What they record is ordinary life. One document concerns a dispute over two camels; others record complaints about trees being cut down and sold, and the king's reply on the penalties for felling them. The 2017 study counts such penalties among early measures against desertification. Silk served as money, too. The seals mix worlds as well. Those on the tablets reflect the joint influence of classical Greco-Roman, Indian and Chinese culture. At Loulan, Hedin and Stein found documents in Chinese, Kharoshthi and Sogdian, and a tomb there yielded a wool fabric bearing the head of Hermes beside his staff. All of it survived for the reason the towns failed. With almost no rain, wood, wool and silk that would have rotted elsewhere lasted more than 1,500 years.

Was Merv really the largest city in the world?

Probably not, but it ranked among the great cities of its day. Merv grew where the Murghab River, flowing north from the mountains of Afghanistan, fans out into a broad oasis in the Karakum Desert. UNESCO counts it the oldest and best-preserved oasis city on the Central Asian Silk Roads, with urban centres dating back to the third millennium BCE. Persians, Arabs and Turks all ruled from it. UNESCO notes that it was an imperial capital under the Abbasids and later the Seljuks, reaching its apogee as the Seljuk capital under Sultan Sanjar (reigned 1118–57). UNESCO's nomination records that scholars were drawn to the libraries and observatory of Merv, among them the geographer Yaqut al-Hamawi and the poet-astronomer Omar Khayyam. Its size is where the claims outrun the evidence. Some estimates put more than half a million people there at its 12th-century peak; UCL's Ancient Merv Project ranks it as possibly the world's third-largest city in the 10th century. Merv was clearly among the biggest cities on Earth; that it was the biggest is not established. The end came in 1221. The city surrendered to a Mongol army under Tolui, Genghis Khan's youngest son, and most of its people are said to have been killed anyway. The Persian historian Juvaini put the dead at more than 1.3 million, a figure historians treat with scepticism, though it shows how unprecedented the slaughter seemed. The city was rebuilt, but elsewhere. Early in the 15th century the Timurid ruler Shahrukh had a new walled town built about 2 kilometres south of the Seljuk walls, and Merv never regained its former prosperity. The move fits a pattern UNESCO traces across the site. As the Murghab's channels drifted from east to west, each new city rose beside the last, and the abandoned ones were never lived in again. That is why Erk Kala, Gyaur Kala and the Seljuk city of Sultan Kala still stand side by side, with Sanjar's mausoleum among them. Together, the three cities make one point. A Silk Road town lasted only as long as its water and its defences held, and when either failed, it was left behind. The caravans carried darker cargo as well: in the 1340s the Black Death is thought to have spread west along Silk Road trade routes toward the Black Sea, a story told in how the Silk Road spread disease. What the lost cities add is the view from inside, kept for us by dry ground: a fine for cutting a tree, a price paid in silk, a quarrel over a camel, still legible after some 1,700 years.