The Tocharians

← Encyclopedia

A traveler leaving the green of Kucha walks out into a dryness so total that flesh does not rot in it. The Taklamakan keeps its dead. Bodies four thousand years old surface from the sand of the Tarim Basin with their eyelashes intact, their felt and wool still dyed, the weave of their cloaks still readable — desiccated rather than decayed, the basin’s emptiness functioning as a vast natural mortuary. Above the same ground, in caves cut into a cliff at Kizil, painted Buddhas sit in rows of fading lapis and ochre, and in a hand near them a king kneels in a long belted coat that no Indian or Chinese figure ever wore. Between those two things — the silent dead in the sand and the named donors on the wall — lies a people recovered almost entirely from what they wrote and what they painted: the Tocharians, an Indo-European tongue that came to rest at the family’s far eastern margin, on the rim of a desert at the center of Asia.

The oasis ring

The Tarim Basin is an inland sea of sand ringed by mountains — the Tian Shan to the north, the Kunlun to the south, the Pamir knot to the west — with the Taklamakan filling its heart. Life clings to the rim, where glacier-fed rivers run out of the ranges and die in the sand, and around each river a city grew. These were not a unified state but a string of independent oasis kingdoms, each one a knot on the road, taxing and watering the caravans that crossed the most forbidding gap in the trade routes between China and the West.

Three of these oasis kingdoms carried the Tocharian speech. Kucha, the largest, sat on the northern rim where the road skirts the Taklamakan, a prosperous Buddhist kingdom of musicians, monks, and irrigated fields. East of it lay Karasahr — Agni to its own people, Yanqi in the Chinese records — and beyond that the Turfan depression, lowest dry land in the region, where several languages overlapped. The cities lived by the road and by the snowmelt, between two of the ancient world’s great powers: the Han dynasty of China reaching west across the Gansu corridor to garrison the basin and protect its silk traffic, and the Iranian world reaching east, so that the oases absorbed administrators, merchants, and gods from both directions and from India to the south. The name Agni, borne by the eastern kingdom, has nothing to do with the Vedic fire; it is a local toponym that happens to share a shape, and the coincidence has misled more than one reader.

A centum tongue at the edge of the world

What makes the Tocharians a fixed point in a much larger history is not their cities, of which other oasis peoples had finer, but their language — recovered, deciphered, and found to sit exactly where the rules of the field said it should not.

Indo-European, by the late nineteenth century, had been sorted into two great wings by the fate of a single set of sounds. The “satem” languages of the east — the Indo-Iranian tongues, the Slavic, the Baltic — turned the inherited palatal stops into sibilants; the “centum” languages of the west — Latin, Greek, the Celtic and Germanic tongues — kept them as hard k-sounds. The split looked geographic, a clean line running roughly west to east, the hard-k West against the s-East. Then manuscripts came out of the Tarim sand, and when Emil Sieg and Wilhelm Siegling read them in 1908 the language they uncovered was unmistakably Indo-European — and unmistakably centum. A hard-k tongue sat farther east than any satem one, two thousand miles past the line it was supposed to obey. The neat east-west division collapsed into something more interesting: a family that had scattered before the satem change ran its course, leaving an archaic western-type speech stranded at the eastern frontier. Tocharian forced the discipline to think of Indo-European less as a map and more as a sequence of departures.

The documents record not one language but two close relatives, conventionally Tocharian A and Tocharian B. Tocharian A — also called Agnean or East Tocharian, its speakers’ own word for it preserved as ārśi — is found mainly at Karasahr and Turfan. Tocharian B — Kuchean or West Tocharian, called kuśiññe, “of Kucha,” in its own texts — belonged to Kucha and ranged more widely. The difference between them ran deeper than dialect. Tocharian A appears only in Buddhist liturgical and literary texts; it carries no letters, no accounts, no graffiti — the marks of a language still living on tongues. By the time the surviving manuscripts were copied, Tocharian A had very likely already died as a spoken language and survived as a learned and ceremonial medium, much as Latin outlived the Roman streets to live on in the Church. Tocharian B was the working tongue: alongside its scriptures sit monastery account-books, caravan passes, medical recipes, and the wreckage of secular verse. Both were written in a slanting, ornate offshoot of the Indian Brahmi script, carried up the road with Buddhism, retooled to spell sounds Sanskrit never had.

The manuscripts themselves belong to late antiquity. Most date from the fifth to the eighth century CE; the oldest Tocharian B fragments are now pushed back to the late fourth or early fifth century — which places the language’s first writing alongside Gothic, Classical Armenian, and the earliest Irish, a contemporary of the other young literary tongues at the edges of the dying Roman and Iranian orders. After the eighth century the speech fades from the record, overtaken by the Turkic Uyghurs who moved into the basin, until it survived only as scratches on palm-leaf and paper waiting in the dark for the spade.

The name that is wrong

“Tocharian” is a misnomer, and the field has never fully forgiven itself for it. When F. W. K. Müller christened the language in 1907, he reasoned from a colophon: an Old Uyghur Buddhist text, the Maitrisimit, declared itself translated out of a tongue its scribes called twγry — and Müller matched that word to the Tukhara, the Tocharoi of the Greek geographers, a people of Bactria. The trouble is that the historical Tocharoi or Tukhara were Iranian-world Bactrians, tied to the Yuezhi confederation that founded the Kushan empire south of the mountains — and they did not speak this language at all. The classical exonym was stitched onto an unrelated Indo-European branch a thousand miles away by a chain of philological guesswork that later proved to be a wrong match. The oasis peoples’ own names for their speech, ārśi and kuśiññe, were sitting in the manuscripts the whole time. “Tocharian” survives now by inertia, a borrowed coat that fits no one, and any reader who meets “Tocharoi” or “Tukhara” in a classical or Indian source is meeting the Bactrians of the Kushan story, not the painters of Kizil.

The road’s religions

The thing the Tocharians wrote down, overwhelmingly, was Buddhism. Nine of every ten Tocharian texts are Buddhist — translations and adaptations carried north out of India and rephrased in a centum tongue: birth-stories of the Buddha, tales of his past lives, monastic rules, and one long verse drama, the Maitreyasamiti-nāṭaka, an unfolding of the future Buddha Maitreya that survives in Tocharian A and was rendered onward into Old Uyghur as the Maitrisimit — the very text whose colophon misnamed the language.

Kucha was first a stronghold of the Sarvastivada, one of the older monastic schools, before the Mahayana tide rose through the oases. By the third century Kuchean monks were already crossing into China to teach and translate, and the greatest of them carried the road’s Buddhism into the Chinese language itself: Kumarajiva (344–413), born at Kucha to an Indian father and a Kuchean princess, taken eventually to Chang’an, who turned the central Mahayana scriptures — the Lotus Sutra, the Diamond Sutra, the foundational treatises of emptiness — into a Chinese so lucid that his versions are still chanted. The lineage that runs from his translations through the schools of China and on to Tibetan Buddhism and the wider Mahayana world passes squarely through these desert kingdoms.

The road did not carry one religion only. In the same oases, and especially around Turfan, two other faiths of the Iranian and Mediterranean west took root. Manichaeism, the dualist church of light and dark founded by Mani in the Sasanian heartland, ran east along the merchant networks and flowered in the Tarim — its Central Asian Manichaean communities producing some of the most exquisite illuminated manuscripts of the ancient world, the very Turfan fragments that, dug up after 1902, refounded the modern study of Mani’s religion. Beside it ran the Church of the East, the East-Syrian Christianity that used the Silk Road as a missionary highway out of Mesopotamia; the Turfan ruins gave up some eleven hundred Christian fragments, in Syriac, in the Iranian Sogdian tongue, and in Uyghur, the wreck of a monastery library where the gospel was read in three scripts at once. Behind all of them stood the older Iranian religion of Zoroaster and the wisdom of the Sasanian court, whose vocabulary of light and order seeped into Manichaean and Buddhist speech alike. The Tarim oases were a place where the gods of half a continent shared the same caravan road, and the Tocharians wrote in the dust at the meeting point — heirs in part to an earlier age when Hellenistic and Indian forms had already met farther west, in the kingdoms below the Hindu Kush.

Kizil, the painted cliff

The largest thing the Tocharians left standing is a cliff. Near Kucha, in a gorge above the Muzat River, more than two hundred caves were cut into the rock and plastered and painted between roughly the third and the eighth centuries — the Kizil complex, the earliest major Buddhist cave site in what is now China and the great surviving monument of Kuchean art. Inside, the rock-cut chapels carry a style that belongs to nowhere else: ceilings divided into rows of rhomboid, mountain-shaped panels, each holding a single compressed scene from the Buddha’s lives; deep blues from imported lapis; bodies modeled with a shaded, rounded line that comes down from the Greco-Indian art of the western road; and, along the lower walls, portraits of the donors who paid for the work — kings and nobles in long belted caftans, swords at the hip, the dress of the steppe-edge oasis aristocracy rather than of monks. These donor figures are nearly the only faces the Tocharians left of themselves, and they look out from the plaster in the costume of their own court.

The caves were stripped within living memory. The German expeditions to the Tarim under Albert Grünwedel and Albert von Le Coq, working the oases between 1902 and 1914, copied some murals and sawed others bodily out of the walls; the cut fragments traveled to Berlin, where many were later shattered in the bombing of the Second World War. What hangs today in the Museum für Asiatische Kunst and what remains on the Kizil walls are two halves of one dismembered monument — a body of painting recovered, like the language, in pieces, and never again whole.

The dead in the sand

Older than every painted cave, older than the language by two thousand years, are the mummies. The Tarim’s aridity preserves bodies with a completeness found almost nowhere else: at the Xiaohe (Small River) cemetery, at Loulan, at Cherchen, the Bronze Age dead lie in felt hats and woolen cloaks, wrapped in cattle-hide, marked with traces of milk left as offering, some buried under forests of carved wooden posts. The earliest reach back to around 2000 BCE. For a century these faces drove a seductive and dangerous story — that “Caucasian” wanderers had carried Indo-European speech into the heart of Asia, the mummies offered as the ancestral, racial proof of the later Tocharian tongue.

The genetics dismantled the story. A 2021 study of ancient DNA from the oldest Tarim individuals — the work of Fan Zhang, Yinqiu Cui, Christina Warinner and their colleagues, published in Nature (10.1038/s41586-021-04052-7) — found that the Bronze Age Tarim people were not migrants at all. They carried, in roughly three parts to one, the ancestry of the Ancient North Eurasians — a deep, once-widespread Ice Age hunter-gatherer population now nearly vanished from the genetic record — mixed with an Ancient Northeast Asian component, and showed no detectable Western Steppe ancestry whatever. They were a genetic isolate: local descendants of an old indigenous lineage who took up their neighbors’ dairying and farming while keeping their own genes and their own ways, the milk in their teeth (kefir traces in the dental calculus at Xiaohe) the sign of a borrowed economy on a native stock.

This is the discipline’s hard distinction, and it must be held. The Bronze Age dead in the Tarim sand are not the Tocharian-speakers of the painted caves; some two thousand years separate them, and where the speech itself came from — by what route and what people the centum tongue reached the basin — remains a separate question that the bodies cannot answer. Genes are not a language. The mummies are the basin’s own ancient population, neither the racial ancestors that the old story wanted nor the property of any modern nation’s myth; they are remains, read with the same care as the manuscripts and not made to say more than they hold.

The textual recovery

The Tocharians are a discipline built almost entirely on philology, and it has a foundation date. In 1908 Emil Sieg and Wilhelm Siegling, working in Berlin on the manuscripts the German expeditions had brought back, read the unknown language for the first time and announced it as Indo-European — their initial paper carrying the now-outdated label die Sprache der Indoskythen, “the language of the Indo-Scythians.” Their great corpus edition, Tocharische Sprachreste, began appearing in 1921, and the grammar Sieg and Siegling produced with Wilhelm Schulze in 1931 remains a cornerstone. Modern descriptive work belongs to Georges-Jean Pinault, whose synthesis for the Encyclopaedia Iranica is the standard scholarly statement of where the language stands, and to the Comprehensive Edition of Tocharian Manuscripts and the teaching resources that have opened the texts to a wider readership. The archaeology of the Tarim and the long argument over its peoples found their best English-language framing in the work of J. P. Mallory and Victor Mair, including Mallory’s accessible survey of the Bronze Age languages of the Tarim Basin for the Penn Museum; the genetic chapter was opened by the Jilin University and Max Planck teams behind the 2021 Nature study. Together these strands hold the field’s two great cautions in tension: the language is real and securely placed, and the question of its origin must not be collapsed into the separate question of the basin’s ancient blood.

What the Tocharians come down to, in the end, is a voice without a body and faces without a name. The kings in their belted coats on the Kizil walls cannot tell us what they called themselves until a manuscript supplies the word; the manuscripts give the words but show no faces. A centum tongue, copied onto palm-leaf in a script from India, telling Indian stories in a desert between China and Iran, painted onto a cliff by donors in steppe dress — the reassembled fragments do not add up to a kingdom restored, but they fix one certain thing: that the oldest western branch of a vast language family went out across the whole of Asia and lay down, at last, in the driest sand in the world.

→ Related: Buddhism · Mahayana · Buddha · Manichaeism Religion Of Light · Central Asian Chinese Manichaeism · Church Of The East East Syrian Christianity · Sasanian Iranian Wisdom Tradition · Han Dynasty · Hellenistic Period · Tibetan Buddhism · Kushan Empire · Indo Greek Kingdoms · Zoroaster