It is rather easy to come away from an introduction to Tocharian with the impression that the inhabitants of Kucha spent most of their time translating Buddhist scripture, discussing the nature of existence, and occasionally copying out yet another account of the Buddha's previous lives. This is not entirely unreasonable, given that the overwhelming majority of the surviving literary material consists of Buddhist texts, but it does produce a rather peculiar picture of a society in which almost nobody appears to have had a job. Michael Weiss, when introducing the secular documents in Kuśiññe Kantwo, makes essentially this point, and his relief at being able to discuss something other than Buddhist literature is quite understandable. Fortunately, there are other kinds of evidence. There are customs records, monastery accounts, business letters, administrative documents, inscriptions, graffiti, and even a love poem. The latter is particularly welcome, although it is not exactly a complete literary archive. What these documents provide is something that the Buddhist manuscripts, however extensive, cannot: a glimpse of the people who had to administer the monasteries, collect taxes, maintain roads, organize caravans, purchase grain, and otherwise keep the whole enterprise functioning.
To understand why these documents were produced, it helps to begin with the rather unusual geography of the region. The Tarim Basin occupies a large depression in northwestern China, enclosed by the Tianshan Mountains to the north, the Kunlun Mountains to the south, and the Pamir and other mountain ranges toward the west. At its center lies the Taklamakan Desert, one of the largest shifting-sand deserts in the world. The northern edge of the basin, where most Tocharian manuscripts were discovered, consists of a succession of oasis settlements separated by considerable stretches of desert. Water comes principally from rivers descending from the surrounding mountains, and the settlements that developed along their courses depended on irrigation agriculture rather than rainfall. The climate is exceptionally dry, and the contrast between the fertile strips surrounding the rivers and the desert beyond them is quite pronounced. The Turfan Depression, farther east, lies substantially below sea level and experiences particularly severe temperature extremes.
The geography also explains the importance of the Silk Road, although the expression is slightly misleading if it encourages one to imagine a single continuous highway. It was, rather, an extensive network of routes connecting settlements across Central Asia, with merchants choosing different paths depending on the season, political circumstances, and availability of water. The Taklamakan could be bypassed along its northern and southern margins, and the routes through Kashgar connected the western regions with the northern and southern branches of the basin. Kucha, Qarašahr, and Turfan occupied important positions along the northern route, while settlements such as Khotan and Niya developed along the southern one. The distances between these places were considerable, and travel depended on the existence of reliable wells, cultivated land, and settlements where caravans could obtain supplies. An oasis was not simply a convenient place to stop; it was an essential part of the infrastructure that made long-distance trade possible.
This also meant that the prosperity of the oasis kingdoms depended on more than agriculture. They cultivated grain, maintained irrigation systems, raised livestock, and produced goods for local consumption, but their position along the trade routes allowed them to collect revenue from merchants and caravans passing through their territory. They were also politically exposed. The same routes that brought merchants, Buddhist pilgrims, and foreign visitors into the region provided avenues through which neighboring powers could exert military and diplomatic pressure. The Chinese dynasties were particularly interested in maintaining influence over the Western Regions, partly because of their commercial importance and partly because control of the routes offered strategic advantages against the nomadic powers of the steppe. The history of Kucha and its neighbors is consequently rather difficult to separate from the changing political circumstances of the wider Tarim Basin.
The Hanshu, the history of the Western Han dynasty compiled by Ban Gu and his successors, gives us an indication of the scale of some of these settlements. Its account of the kingdom of Qiuci, generally identified with Kucha, records a population of approximately 81,317 people, including 21,076 men available for military service. These figures should not be treated as a modern census, and the administrative categories employed by the Han court do not necessarily correspond to our own, but they nevertheless suggest that Kucha was a substantial political and economic center rather than a small collection of monasteries surrounded by agricultural villages. The number of men recorded for military service is particularly revealing. The northern Silk Road was commercially valuable, but it was also a contested region, and its rulers had to maintain the military resources necessary to defend their territory and protect the routes on which their revenues depended.
Chinese influence in the region fluctuated considerably. The Han dynasty had established a presence in the Western Regions centuries earlier, but its authority weakened and was eventually reasserted under the Tang. During the seventh century, Tang military campaigns brought several oasis kingdoms under imperial control or influence, although the degree of direct administration varied and local rulers often retained considerable authority. The Tang conquest of Kucha in 648 was particularly important, following the expedition of the general Ashina She'er. Even then, Chinese political authority was not permanent. Tibetan expansion, Turkic powers, and the changing balance of power across Central Asia repeatedly altered the political situation. By the middle of the ninth century, the arrival and settlement of Uyghur groups farther east, particularly following the collapse of the Uyghur Khaganate in Mongolia in 840, contributed to a substantial transformation of the region's linguistic and cultural environment. The history is not quite a sequence of Chinese rule followed by an immediate Turkic replacement, but the eventual predominance of Turkic languages would have significant consequences for the survival of Tocharian.
The chronology of the surviving manuscripts overlaps with much of this political history. Tocharian B texts are attested approximately from the fifth to the tenth centuries CE, while the surviving Tocharian A material is predominantly dated from the seventh to the tenth. Most of the manuscripts are Buddhist, and the majority of the surviving fragments come from the region's monastic libraries. The geographical distribution, however, is not uniform. Kucha is especially important for Tocharian B, while Qarašahr and Turfan are the principal centers associated with Tocharian A. The fact that the manuscripts survive in these locations does not mean that the languages were confined to precisely those settlements, but it gives us a reasonably clear starting point for reconstructing their historical distribution. Michaël Peyrot's article in Encyclopaedia Iranica and the Cambridge chapter on Tocharian provide useful accounts of this geographical and chronological evidence.
The Buddhist literature deserves some attention before we get to the customs officers, if only because it constitutes the majority of the material that made the languages known to modern scholarship. Most of the surviving texts are translations or adaptations of Sanskrit works, associated particularly with the Sarvāstivādin and Mūlasarvāstivādin traditions. The material includes the Vinaya, the body of monastic regulations governing the conduct and organization of Buddhist communities; the Sūtras, which preserve teachings attributed to the Buddha; the Abhidharma, concerned with the systematic analysis of Buddhist doctrine and experience; and narratives such as the Jātaka tales, which recount the Buddha's previous births. There are also dramatic works, hymns, poems, and commentaries. The literary range is considerable, even if the surviving material is weighted heavily toward religious subjects.
The Sarvāstivāda school is particularly interesting because its name reflects one of its central doctrinal positions. The word is conventionally explained through Sanskrit sarvam asti, ‘everything exists.’ More precisely, the school maintained that the dharmas, the fundamental elements of experience in its philosophical system, exist across the three times: past, present, and future. Their activity or manifestation differs depending on the time, but their existence is not restricted to the present moment. This was part of a much larger debate within Buddhist scholastic philosophy about the nature of time, causation, and the relationship between past actions and present experience. The distinction mattered especially for explaining karma: if past actions were no longer existent in any sense, how could they continue to produce consequences? Sarvāstivādin thinkers developed an elaborate account of this problem, which was debated by other Buddhist schools.
It is worth remembering that these philosophical traditions were not simply imported into the Tarim Basin as finished packages. Buddhist institutions developed within local communities, and their literary activity involved translation, interpretation, and the creation of new works. The Tocharian translators had to find ways of expressing Sanskrit technical vocabulary in languages with their own grammatical structures and lexical histories. Sometimes they borrowed Sanskrit terms; sometimes they translated them; sometimes the two Tocharian varieties employed different expressions for the same concept. Weiss notes that even a familiar word such as Sanskrit karma can have different equivalents in Tocharian A and B. Such differences are interesting because they suggest that the two languages had developed distinct Buddhist vocabularies rather early, rather than sharing one standardized terminology throughout their entire history. The translations therefore provide evidence not only for the transmission of Buddhism but also for the linguistic communities that participated in it.
The monastic institutions themselves were substantial organizations. They required buildings, land, food, clothing, writing materials, and people to maintain them. Monks depended on donations from wealthy patrons, local rulers, and lay communities, while monasteries could possess agricultural resources and manage their own stores of provisions. The religious life of the community therefore depended on an economic structure that was not especially different, in its practical requirements, from that of other large institutions. Someone had to record the grain received, determine how much was consumed, arrange for repairs, and keep track of the people responsible for different tasks. We have a tendency to treat ancient monasteries primarily as centers of intellectual and spiritual activity, but the surviving administrative documents make it clear that they were also institutions with rather ordinary material needs.
One of the most interesting figures in the wider history of Central Asian Buddhism is Kumārajīva, who was born in Kucha in the fourth century and eventually became one of the most influential Buddhist translators in China. His father was of Indian origin and his mother belonged to the ruling family of Kucha, and he received a substantial Buddhist education before travelling east. He was eventually brought to Chang'an, where he worked under the patronage of the Later Qin court and directed a major translation project. His Chinese translations include influential versions of the Lotus Sūtra, the Vimalakīrti Nirdeśa, and the Madhyamaka works associated with Nāgārjuna. His translations were important not simply because they made Indian Buddhist texts available to Chinese readers, but because his choices of terminology and style helped shape the language of Chinese Buddhist philosophy for centuries afterward.
Kumārajīva is not evidence that Tocharian was necessarily the language in which every Buddhist text from Kucha was translated, and it would be a mistake to attribute his entire translation program to Tocharian. His work was primarily based on Sanskrit and other Indic materials. Nevertheless, his career illustrates the importance of Kucha within the broader Buddhist world. The oasis kingdoms were not peripheral recipients of Indian religious ideas; they were active participants in the movement of Buddhist texts and scholars toward China. The transmission of Buddhism through Central Asia involved numerous languages, translators, and intermediaries, and Tocharian-speaking communities were among those that contributed to this process.
The difficulty, of course, is that the literary evidence makes it very easy to forget everyone who was not a monk. Fortunately, there is another category of material, and it is considerably less concerned with metaphysics. The secular Tocharian documents are predominantly in Tocharian B and include commercial correspondence, legal records, administrative reports, receipts for donations, monastery accounts, and customs documents. They are often written in a more cursive variety of the Brāhmī-derived script, using a brush rather than the more deliberate, carefully formed handwriting of the literary manuscripts. Their physical appearance is different, too. Instead of the oblong manuscript leaves used for Buddhist literature, we find larger sheets of paper, sometimes assembled into rolls, and small wooden tablets used for administrative purposes. The difference is important: these documents were not produced primarily to preserve religious teachings or earn merit through copying scripture. They were written because somebody needed information recorded, communicated, or verified, and their survival gives us access to a rather different register of the language. Georges-Jean Pinault's study of the economic and administrative documents in the Berezovsky and Petrovsky collections is particularly useful here, as is the broader survey of Tocharian documentation in the Handbook of Comparative and Historical Indo-European Linguistics.
The caravan passes are perhaps the most immediately interesting examples. Several were recovered from a former checkpoint near Kucha during the French expeditions associated with Paul Pelliot, and the surviving tablets are now distributed among museum and manuscript collections. They were made of wood, generally consisting of two small, flat pieces placed with their written surfaces facing one another and tied together. Notches along the edges helped hold the pieces in place, while the outside could carry the name of the official to whom the document was addressed. The arrangement is rather practical: the written information was protected between the two pieces of wood, and the tablet could be carried to the next checkpoint without the contents being immediately exposed.
The documents themselves contain details about the caravans for which permission to travel had been granted, together with instructions for officials at the next station. They identify the person leading the caravan, list the people and animals accompanying it, and provide information that would allow the next checkpoint to verify that the group was the one expected. They were dated according to the reigning king, which is particularly useful for historians attempting to establish an absolute chronology for documents that would otherwise be difficult to date. In other words, these were not merely travel permits in the modern sense; they were part of an administrative system that allowed successive authorities along a trade route to communicate with one another.
One of the examples reproduced by Weiss concerns a caravan leader named Kaṣake Puttamitre, whose name is itself rather revealing. It is an unmistakably Indic-looking name, and its presence in a Tocharian administrative document is a reminder of the extent to which people, languages, and naming practices crossed regional boundaries. The caravan's accompanying personnel and animals are also recorded, including references to donkeys and a horse. There is another pass dated to the reign of Swarnatepe, a ruler of Kucha who came to power in the early seventh century. The exact historical identification of the king and the chronology of individual documents require some care, but the appearance of royal names in such mundane administrative contexts is extremely useful. We know about rulers through Chinese histories, diplomatic records, and inscriptions, but here a king's reign is used simply as a dating convention on a document concerning the movement of a caravan.
The passes also tell us something about the organization of trade. A caravan was not necessarily a single merchant travelling with a few animals; it could consist of several individuals, attendants, pack animals, and different kinds of merchandise, moving through a system of checkpoints and settlements. The ability to record its composition and communicate that information ahead suggests a degree of administrative coordination that is easy to underestimate when imagining the Silk Road as a collection of independent merchants wandering through the desert. Long-distance trade required arrangements for supplies, security, taxation, and the recognition of documents across different jurisdictions. The caravan passes are small pieces of evidence, but they reveal a functioning administrative system behind the movement of goods. They also provide some of the most valuable examples of Tocharian B being used for purposes that had very little to do with Buddhist literary culture.
There is something particularly interesting about the language in which these documents were written. The literary manuscripts often employ a relatively standardized form of Tocharian B, whereas the secular documents contain spellings and grammatical forms that can differ from those found in formal religious texts. Some of these differences may reflect regional dialects, while others are likely to be chronological or sociolinguistic. A customs officer writing a short instruction for another official did not necessarily follow the same conventions as a monk copying a Sanskrit-derived Buddhist work. The distinction is important because it allows us to begin separating the language of literary transmission from the language used in administrative and everyday contexts. It is also a reminder that the language reconstructed from Buddhist manuscripts is not necessarily a complete representation of the way people spoke.
The monastery accounts are even more interesting, partly because their contents are so unremarkable. Several surviving documents record the quantities of food and other provisions consumed by monastic communities, with entries arranged by month and sometimes identifying particular types of expenditure. Millet, Tocharian aka, appears among the recorded foods, alongside other commodities whose precise identification remains uncertain. Some of the terminology is difficult because the documents are damaged, the handwriting is cursive, and the vocabulary may refer to local products for which we have no independent identification. There is no reason to assume that every unfamiliar word can be recovered simply by comparing it with Sanskrit or another neighboring language. In some cases, the scholars responsible for editing the texts have had to leave an item untranslated, which is a perfectly respectable outcome when the evidence does not support anything more definite.
These accounts also reveal something about the internal organization of monasteries. The quantities of food recorded are not simply lists of donations; they can provide information about consumption, distribution, and the management of supplies. Different officials appear in connection with particular transactions, and some records document transfers between individuals responsible for different aspects of the monastery's administration. Research by Chao-jung Ching and Hirotoshi Ogihara on the Berlin collection has examined the internal relationships between the monastic accounts, including the ways in which different records can be connected through recurring names, administrative titles, and transactions. Their work suggests that the documents can be studied as parts of a larger administrative system rather than as isolated scraps of writing. The result is a much more detailed picture of monastic organization than the literary texts alone would allow.
One particularly interesting feature of the accounts is their physical relationship to other texts. A sheet of paper used for a monastic record could also preserve part of a Buddhist ritual manual, and the two texts might occupy opposite sides of the same piece of material. The Karmavācana, the collection of formal procedures and declarations used in monastic ceremonies, appears in this context alongside documents concerned with food and expenditure. It is not always possible to establish precisely which text was written first, or whether the sheet was deliberately reused after the original writing had become obsolete. Nevertheless, the juxtaposition is revealing. The same institution that preserved carefully copied religious literature also produced administrative paperwork, and the same writing materials could pass between these different functions. Paper was valuable, and there was little reason to discard a perfectly usable sheet simply because one side already contained writing.
The contrast between the two kinds of document is particularly useful for understanding the history of literacy in the region. The Buddhist manuscripts were often copied in a formal hand, following conventions inherited from Indian manuscript culture, and their contents were intended to be transmitted within religious and scholarly communities. The accounts and administrative records, by contrast, were concerned with immediate practical matters. Their handwriting could be less regular, their vocabulary more specialized, and their physical organization more flexible. They were produced within institutions that required both kinds of literacy, and it is probably misleading to imagine that the people responsible for copying Buddhist scripture belonged to an entirely separate world from those maintaining the monastery's accounts. Some individuals may have performed several kinds of work, while others specialized in particular administrative or literary functions. The documents do not always tell us who was responsible for each task, but they make it clear that written language had a much broader institutional role than the surviving religious literature might suggest.
There are also letters, legal records, and inscriptions that provide evidence of interactions outside the monasteries. Some documents concern business transactions, while others record donations or administrative decisions. Graffiti and inscriptions on cave walls preserve names, short statements, and captions associated with religious paintings. The latter are particularly interesting because they bring us closer to the people who actually visited the caves, rather than only the people who commissioned or copied their contents. A name written on a wall may tell us little about the individual who left it, but it does establish that somebody was present and considered the act of writing their name worth undertaking. These inscriptions also provide evidence for the distribution of linguistic varieties across different sites, although they have to be interpreted carefully because the language of an inscription does not necessarily establish the native language of its author.
And then there is the love poem. The Tocharian B corpus contains a fragment of original literary writing that is particularly notable because so much of the surviving literature consists of translations from Sanskrit. It is a short and imperfectly preserved text, and its literary quality is difficult to evaluate independently of the problems involved in reconstructing its wording. Nevertheless, its existence is important. It demonstrates that Tocharian B was not simply a language used to reproduce Indian religious works, but could also serve as a medium for original composition. We should be cautious about drawing conclusions from a single fragment, especially when the rest of the secular literary corpus is so limited, but it does make the surviving material considerably less one-sided. The language had a literary life of its own, even if the circumstances of preservation have left us with very little of it.
This distinction between religious and secular writing also has consequences for the question of which language was actually spoken in which communities. George Lane, writing in 1941, proposed that Tocharian B might have served as the everyday vernacular across much of the region, while Tocharian A may have been primarily a literary or religious language. His reasoning was understandable. Tocharian B appears in commercial and administrative documents, on walls, and in other contexts that suggest ordinary use, whereas the Tocharian A material known to him was overwhelmingly Buddhist. He also noted that Tocharian B was sometimes used to provide glosses for Tocharian A texts, suggesting that the two varieties existed in contact and that B may have had a wider practical role. Lane ended with the hope that, if B were indeed the vernacular, further discoveries might eventually reveal an original secular literature.
The difficulty with this interpretation is that the surviving evidence is not distributed evenly, and the absence of a particular kind of text does not necessarily establish the absence of the language in everyday life. Tocharian A is concentrated in a different geographical area from the earliest Tocharian B material, and its manuscript corpus has been preserved under different circumstances. Moreover, the distinction between a literary language and a spoken vernacular is not always straightforward. A language can be used in religious writing while also being spoken in daily life, and the presence of a more widely used language in administrative documents does not automatically make the other one exclusively literary. Lane was working with a much smaller corpus than modern scholars have available, and his interpretation reflects the limitations of the evidence at the time rather than a careless reading of it.
The question of regional variation became particularly important in the work of Werner Winter, whose 1955 paper, “A Linguistic Classification of ‘Tocharian’ B Texts,” attempted to establish whether the differences among the surviving manuscripts could be explained geographically. His method was relatively straightforward, although the execution involved a considerable amount of manuscript comparison. He selected a reference variety associated with Sorcuq and examined the spelling and linguistic features of texts from other archaeological sites, recording the points at which they differed. He then asked whether these differences occurred randomly or whether particular features tended to appear together in manuscripts from particular locations.
The results suggested three geographical varieties: a central group associated with Sorcuq and Qarašahr, a western group represented particularly by the cave complex of Ming-öy Qizil near Kucha, and an eastern group associated with Sangim, Murtuq, Qocho, and Toyoq near Turfan. The differences were not necessarily dramatic. They involved details of spelling, the representation of consonant clusters, and particular phonological and morphological forms, but their distribution was sufficiently systematic to suggest that Tocharian B was not entirely uniform across the region.
Some of Winter's observations are particularly interesting because they concern the scribes themselves rather than the underlying spoken language. One example is the so-called old script, an archaic-looking form of handwriting that appears particularly in manuscripts from Ming-öy Qizil. Its restricted distribution suggests that it may have been associated with a particular scribal tradition or local institution. Winter also examined the way scribes represented consonant clusters, particularly their choice between ligatures, in which two consonants are combined into a single written form, and the virāma, a small mark used in Brāhmī-derived writing systems to suppress the inherent vowel of a consonant. The western manuscripts showed a preference for ligatures, whereas central and eastern manuscripts more frequently employed the virāma. This is a relatively small orthographic distinction, but it provides another feature through which manuscripts can be compared and grouped.
The distinction between scribal convention and spoken dialect is important here. Two manuscripts may differ in their spelling because their authors spoke different varieties of the language, but they may also differ because they were trained in different writing traditions. A monastery could preserve a particular orthographic standard even when the surrounding population spoke somewhat differently, and scribes trained elsewhere might carry those conventions with them when they moved. Winter's classification therefore does not necessarily provide a direct map of spoken Tocharian B. It identifies geographical patterns in the written evidence, which may reflect a combination of regional speech, chronology, and institutional practice.
One manuscript also presents a particularly interesting problem. It belongs linguistically to the western group, but its dating information refers to the reign of a king of Agni, the region associated with Tocharian A. Winter considered several possibilities: perhaps the manuscript had been transported from one region to another, or perhaps the monastery in which it was produced had itself moved. Either possibility would complicate a classification based solely on archaeological findspots. A manuscript discovered in one location need not have been written there, and a text copied by a scribe trained elsewhere may preserve linguistic features associated with a different community. Winter also suggested that the monastery of Yurpa, near Sorcuq, might have played a role in establishing a prestigious central variety of written Tocharian B, which was subsequently imitated by scribes elsewhere. The proposal is necessarily tentative, but it introduces a question that is familiar from the history of many literary languages: to what extent did the written standard reflect the speech of a particular region, and to what extent did it reflect the conventions of an influential institution?
There is a certain appeal to reconstructing the history of a language through such apparently minor distinctions. We cannot listen to a speaker from eighth-century Kucha, and we cannot interview the monk who copied a particular manuscript, but we can compare the way different scribes represented the same consonant cluster and determine whether their choices follow geographical patterns. The evidence is indirect, and the conclusions have to remain proportionate to it, but the process is not fundamentally different from the comparative work used to reconstruct the history of much better documented languages.
Winter's classification has also been revised by subsequent scholarship. Michaël Peyrot's work on the historical development of Tocharian B, together with improved paleographical knowledge and a much larger body of available texts, has made it possible to distinguish chronological developments more carefully from regional variation. Some features that Winter interpreted geographically may instead reflect differences between earlier and later stages of the language, or between formal literary writing and less standardized secular documents. The distinction is especially important because the oldest stage of Tocharian B is attested only around Kucha, while the classical and later stages occur farther east as well. The geographical distribution of the manuscripts therefore reflects the expansion of the language and its literary traditions over time, rather than simply three permanently distinct dialect regions.
The modern reconstruction generally places the original home of Tocharian B around Kucha and its surrounding settlements. The earliest linguistic stage is found there, and the later material indicates an expansion eastward toward Qarašahr and Turfan, probably during the late sixth and seventh centuries. Tocharian A, meanwhile, appears to have originated around Qarašahr and is attested farther east at Turfan, but not in Kucha. This distribution is important because it gives us a historical sequence rather than merely a list of archaeological findspots. Tocharian B was not simply a language spoken everywhere along the northern Silk Road from the beginning; its geographical spread can be traced through the development of its written forms.
The distinction between the two languages also raises questions about the relationship between literary culture and political geography. The expansion of Tocharian B eastward occurred during a period in which the oasis kingdoms were connected through changing political arrangements and extensive commercial networks. The movement of monks, merchants, and administrative personnel would have provided opportunities for the language to spread, while monasteries could establish literary conventions that extended beyond the communities in which they originated. It is not possible to identify a single event responsible for the expansion, and the surviving evidence does not tell us whether the movement involved large-scale population migration or a more gradual spread of linguistic prestige. Both processes may have contributed.
The history of Tocharian A is particularly difficult because its secular documentation is so much less extensive. Lane's suggestion that it was primarily a literary language is no longer sufficient to explain the available evidence, although it is still important not to overstate what we know about its everyday use. There are non-literary features of the Tocharian A material, including glosses, inscriptions, and evidence of interaction with other languages, but the corpus remains overwhelmingly religious. The two languages may have occupied different positions within the same multilingual environment, with their use varying according to locality, institution, and social context. The distinction between a spoken language and a literary language is also not necessarily binary: a variety can be used in religious literature, formal instruction, and everyday communication without serving all three functions equally.
This is one of the reasons that I find the history of Tocharian documentation so interesting. Every new group of manuscripts can alter the picture, not necessarily by revealing an entirely unknown aspect of the language, but by changing the relative importance assigned to evidence that has already been available. Lane's interpretation was reasonable given the documents he had examined; Winter's classification provided a more systematic account of geographical variation; and later work has complicated both by introducing a more detailed chronology. The process is not particularly unusual in historical linguistics, although it is perhaps more visible in a field where the available evidence is so limited.
There is another question that has occupied scholars for considerably longer: where did the Tocharian languages come from in the first place?
The geographical position of Tocharian makes this an especially difficult problem. It is an Indo-European branch found much farther east than any of the other ancient languages with which it is closely related, and its earliest attested texts are separated from the presumed prehistoric movements of Indo-European-speaking populations by several millennia. One influential hypothesis connects the ancestors of the Tocharian speakers with the Afanasievo archaeological culture, which developed in the Altai and Minusinsk regions of southern Siberia during the early Bronze Age, approximately 3300–2500 BCE. The Afanasievo people were pastoralists who kept cattle and sheep and made use of domesticated horses. Their material culture and genetic ancestry show substantial connections with the Yamnaya populations of the Pontic-Caspian steppe, which are frequently associated with the early dispersal of Indo-European languages.
The geographical connection is attractive. The Afanasievo culture extended considerably farther east than the Yamnaya, reaching regions much closer to the Tarim Basin, and its chronology is compatible with the possibility that an early Indo-European language was carried into Inner Asia during the Bronze Age. If the hypothesis is correct, the ancestors of the Tocharian-speaking populations would have separated from the other Indo-European branches very early, before the developments that distinguish many of the historically attested languages had taken place. This would help explain why Tocharian occupies such an unusual position within the family.
The problem is that linguistic migration cannot simply be reconstructed from archaeological similarities. An archaeological culture is defined through patterns in material remains, including pottery, burial practices, settlements, and other aspects of material life. It does not automatically correspond to a single language, and the movement of a particular style of pottery does not necessarily imply the movement of the people who originally produced it. The Afanasievo connection is supported by more than geographical proximity, particularly its genetic relationship with populations farther west, but establishing that the people associated with it spoke Proto-Tocharian remains an inference rather than a directly documented fact.
The Tarim mummies have made the discussion considerably more complicated. The naturally preserved remains discovered at sites such as Xiaohe and Gumugou attracted enormous attention because of their unusually good preservation, which resulted from the region's dry climate and particular burial conditions. Their clothing, wooden coffins, and other grave goods provide valuable evidence about Bronze Age life in the basin. Some of the textiles and garments are particularly interesting because they demonstrate connections with broader Eurasian material traditions, while the archaeological evidence suggests that these communities engaged in pastoralism and agriculture long before the historical Tocharian manuscripts were produced.
For a long time, the apparent western Eurasian characteristics of some of the mummies encouraged speculation that they might represent the direct ancestors of the Tocharian speakers, perhaps descended from populations associated with the Afanasievo culture. This interpretation was attractive partly because it seemed to connect archaeological evidence with the linguistic distribution of Tocharian. The geographical and chronological gaps were substantial, but the hypothesis offered a relatively straightforward narrative: Indo-European-speaking pastoralists moved eastward, settled in the Tarim Basin, and eventually developed into the communities whose languages are preserved in the medieval manuscripts.
Ancient DNA has made that narrative considerably less straightforward. A 2021 study published in Nature, “The Genomic Origins of the Bronze Age Tarim Basin Mummies,” examined individuals from both the Dzungarian Basin, north of the Tianshan Mountains, and the Tarim Basin itself. The results showed that the early Bronze Age individuals from Dzungaria possessed substantial Afanasievo-related ancestry, while the individuals from the Xiaohe archaeological horizon in the Tarim Basin belonged to a distinct, deeply rooted local genetic population. The researchers found no substantial Afanasievo ancestry among the sampled early Tarim individuals, contrary to the earlier hypothesis that the famous mummies were direct descendants of Afanasievo migrants.
This distinction is particularly important. The Afanasievo population was present in the broader region, and its descendants or related groups could plausibly have contributed to the movement of Indo-European languages into the area. But the earliest sampled Tarim mummy population was not simply an eastern extension of the Afanasievo population. The researchers instead proposed that the Tarim communities had developed from an ancient local population that adopted pastoral and agricultural practices through contact with neighboring groups. The study also identified evidence of dairy consumption in the dental calculus of the Xiaohe individuals, showing that they consumed milk products despite lacking the genetic adaptations for lactase persistence that are common in some modern pastoralist populations. The distinction between cultural practices and genetic ancestry is particularly relevant here: populations can adopt new economic practices without being replaced by the people who originally developed them.
This does not disprove the Afanasievo hypothesis for the linguistic origins of Tocharian, because the sampled Tarim individuals predate the surviving Tocharian texts by several millennia and cannot automatically be identified with the later speakers. It does, however, make it much more difficult to construct a direct line between the earliest mummies and the historical Tocharian communities. One possibility is that Indo-European languages entered the region through Dzungaria and subsequently spread into the Tarim Basin through processes that did not involve the wholesale replacement of its existing population. Another is that the linguistic history involved several movements and periods of contact that cannot yet be reconstructed from the available evidence. The genetic history of a population and the history of its language are related questions, but they are not interchangeable ones.
Weiss makes essentially this point when discussing the subject, and it is worth taking seriously. A language can spread through migration, intermarriage, political domination, trade, education, or the prestige of a particular religious and administrative culture. None of these mechanisms requires the population adopting the language to share the genetic ancestry of the people who originally spoke it. English provides an obvious modern example: the global spread of English has not required the populations of every English-speaking country to descend from the inhabitants of early medieval Britain. The same principle applies to ancient languages, even if the historical circumstances are much harder to reconstruct. The available genetic evidence can constrain our hypotheses about population movement, but it cannot identify the language spoken by an individual whose linguistic community left no written record.
There is another possible complication farther south. In the third and second centuries CE, the kingdom of Kroraina, known in Chinese sources as Loulan, occupied an important position along the southern edge of the Tarim Basin. Its administrative documents, discovered principally at sites such as Niya, were written in a variety of Gāndhārī Prakrit, an Indo-Aryan language that had spread into the region through contact with northwestern India. The documents include legal and administrative records, personal names, and other material that provides a relatively detailed picture of the kingdom's affairs. They are particularly valuable because they preserve information about the population and administration of a part of the Tarim Basin for which we have very little direct Tocharian evidence.
In 1935, Thomas Burrow examined the Gāndhārī documents and identified a number of words and grammatical features that did not appear to belong straightforwardly to the language in which the documents were written. Some of these forms resembled Tocharian vocabulary, leading to the suggestion that the native language of Kroraina might have been another member of the Tocharian branch. This hypothetical language became known as Tocharian C. The idea is attractive because it would extend the geographical distribution of the family into the southern Tarim Basin and provide an explanation for the non-Gāndhārī elements in the administrative records.
The evidence, however, is considerably less substantial than the name might suggest. We do not possess a body of texts written in Tocharian C, and the proposed language is reconstructed almost entirely from possible loanwords and structural features in Gāndhārī documents. Some of these forms may indeed represent Tocharian influence, but others can be explained through internal developments of Gāndhārī or through borrowing from Iranian languages. Giorgio Banti, in his reassessment of the proposed Tocharian elements in Kroraina Prakrit, argues that some of the evidence is less convincing than earlier scholarship had assumed. He also points to the substantial number of personal names in the documents that appear to belong to other linguistic traditions, suggesting that the local population was considerably more diverse than a simple Tocharian-speaking model would imply.
It is therefore best to treat Tocharian C as a hypothesis rather than an established third language. The evidence demonstrates linguistic contact in the southern Tarim Basin, and some of that contact may well have involved a variety related to Tocharian, but we cannot yet reconstruct an independent language from the surviving material. The distinction matters because the existence of a few borrowed words does not establish the existence of a complete language community. In historical linguistics, the temptation to give a convenient name to an intriguing collection of forms is occasionally rather stronger than the evidence warrants.
The final question is how long Tocharian continued to be spoken after the period represented by the majority of its manuscripts. The conventional dating of the Tocharian texts places most of them between approximately the fifth and tenth centuries, but the actual evidence is more complicated. A small number of documents have been dated considerably later, including a Tocharian B manuscript fragment designated B 296, whose radiocarbon dating falls between 1178 and 1255 CE. This is an unusually late date for a language whose surviving literary tradition is generally associated with the earlier medieval period. It does not establish that Tocharian was still widely spoken throughout the thirteenth century, and a date assigned to a manuscript's material does not necessarily establish the date at which its text was composed. Nevertheless, it indicates that Tocharian writing traditions may have persisted longer than the usual textbook chronology suggests.
There is also a passage in the eleventh-century work of the Turkic scholar Maḥmūd al-Kāšġarī, whose Dīwān Luġāt al-Turk is one of the most important sources for the historical Turkic languages. In describing the linguistic situation of the Uyghurs, he notes that they possessed a pure Turkic language but also another language spoken among themselves, alongside different writing traditions. Some modern scholars have suggested that this second language may have been Tocharian B, which would imply that the language survived as a community vernacular well into the eleventh century. The identification is not certain: the passage does not name Tocharian, and the linguistic situation of the eastern Tarim Basin was sufficiently complex that other interpretations are possible. Nevertheless, it is consistent with the possibility that Tocharian continued to be spoken after the disappearance of most of its literary production.
The eventual replacement of Tocharian by Turkic languages was probably gradual, and it would be misleading to imagine a single moment when one language disappeared and another took its place. The arrival and expansion of Uyghur-speaking populations, the development of new political centers, and the increasing importance of Turkic as a language of administration and wider communication would all have affected the position of the local languages. Bilingualism may have been common, particularly in communities where Tocharian remained useful within families or local institutions while Turkic became increasingly important in public life. The survival of a language in speech can also extend considerably beyond the period in which it is used for formal writing. Once a literary tradition disappears, the language itself does not necessarily vanish immediately. It may continue in domestic communication for generations, leaving very little evidence for historians to recover.
The wider history of Buddhism in the region makes the transition particularly interesting. Tocharian had been an important language of Buddhist scholarship and translation, but Uyghur eventually became a major literary language for Buddhist communities in the eastern Tarim Basin. Some Uyghur Buddhist texts were translated from Tocharian, and Tocharian terminology appears to have influenced the development of Uyghur Buddhist vocabulary. This means that even after Tocharian ceased to function as an important literary language, its influence continued within the religious traditions of the region. The language was not simply replaced in every respect; some of its intellectual and linguistic contributions were incorporated into the traditions that followed it.
There is a useful distinction to make between the disappearance of a language and the disappearance of its influence. The former is extremely difficult to date when the final stages of language use are undocumented; the latter can be traced through loanwords, translated texts, and changes in the vocabulary of neighboring languages. Tocharian's contribution to the Buddhist culture of Central Asia and China was not confined to the period in which its manuscripts were being copied. Its texts continued to circulate, and the traditions built around them influenced later communities whose languages were different. The surviving evidence is insufficient to reconstruct every stage of this transition, but it is clear that the history of Tocharian cannot be understood simply as the rise and fall of two isolated languages.
I began with the impression that the Tocharians were a people who spent most of their time meditating. The surviving manuscripts make that interpretation rather easy to adopt, but the secular documents have substantially complicated it. We have caravan leaders carrying customs passes, officials recording the composition of travelling groups, monks accounting for grain, administrators organizing the resources of monasteries, and individuals writing letters and leaving inscriptions on cave walls. We have evidence of agriculture, trade, military organization, political administration, and the movement of people between communities. None of this diminishes the importance of the Buddhist literature; rather, it places that literature within the society that produced and preserved it.
I also find the physical survival of these documents particularly interesting. The literary manuscripts were copied with the intention of preserving and transmitting religious knowledge, while the administrative documents were written to accomplish immediate tasks. Yet the latter are often just as valuable to us, precisely because they preserve information that nobody at the time would have considered especially important. A caravan pass was intended to get a particular group of merchants through a checkpoint. A monastery account was intended to record how much food had been consumed. A letter was written because someone needed to communicate with another person. None of these documents was produced to provide a future historian with an account of life in Kucha, and they are consequently less concerned with presenting a carefully constructed picture of the people who lived there.
There is a certain imbalance in what survives. We know a great deal about Buddhist doctrine and the literary conventions of Tocharian because monasteries copied and preserved religious manuscripts in considerable numbers. We know less about ordinary domestic life, local political disputes, the full range of commercial activity, and the people who never had occasion to write anything down. Even the secular documents are largely institutional, and the lives of farmers, craftspeople, merchants, and laborers are visible only indirectly. A customs pass tells us that a caravan existed, but not necessarily what its merchants thought about the journey. An account records the consumption of millet, but not the people who cultivated it. The evidence is valuable precisely because it is so limited, and it is important not to mistake what has survived for everything that once existed.
Still, I think the contrast between the Buddhist manuscripts and the secular documents gives us a much better understanding of the Tocharian-speaking communities than either category could provide alone. The religious texts show us their participation in a sophisticated intellectual tradition extending from India to China; the administrative records show us the institutions, economic arrangements, and everyday activities that made that participation possible. We can read about the philosophical positions of the Sarvāstivādins and then encounter a record of how much millet a monastery consumed in a particular month. We can study the development of Tocharian B's literary standard and then find a customs pass bearing the name of a local king. These are not separate histories. They belong to the same communities, and the documents only become fully meaningful when we allow them to be read together.
Perhaps the most useful thing the secular texts have taught me is that the absence of evidence for ordinary life is not evidence that ordinary life was absent. The Tocharian manuscripts were preserved largely because religious institutions had reasons to copy and maintain them, while the administrative records survived through rather less deliberate circumstances. They give us a society that is considerably more complicated than the literary corpus initially suggests, with people whose concerns extended beyond the preservation of Buddhist doctrine. The monks were there, certainly, and they left us an extraordinary body of literature. But they were also buying food, maintaining buildings, organizing their resources, receiving donations, and depending on the labor of people whose names we may never know. The customs officers were there, too, making sure that the caravans passed along the northern Silk Road with the appropriate documentation. I am rather glad that somebody thought to write all of this down.