Historicizing the ‘Tang cosmopolitanism’ trope

It is now somewhat customary to trace the cosmopolitanism trope to the American Sinologist Edward Schafer and his book The Golden Peaches of Samarkand: A Study of T'ang Exotics (1963).Footnote ¹ As Jonathan Skaff has noted, ‘Schafer wrote about foreign people and goods with relish, which has served to glamourise the [Tang] dynasty. His ability to entertain readers has kept his book in print long after his death and burnished the image of the Tang as a glorious and cosmopolitan age.’Footnote ² But while Schafer certainly did much to popularize the image of a flourishing empire filled with foreign merchants and fascinated with foreign luxury imports, he himself barely used the word ‘cosmopolitan’ when writing Golden Peaches. The book's first chapter, ‘The Glory of T'ang’, describes China's eighth century as ‘the international age, the age of imports, the age of mingling, the golden age, [which] began to pass away at the beginning of the ninth century’, and then, slightly further on speaks of Emperor Xuanzong 玄宗 (r. 713–756) as ‘the fabulous king, most glorious monarch of a cosmopolitan age, himself a connoisseur of the exotic’.Footnote ³ That, however, is the only time ‘cosmopolitan’ appears in Golden Peaches; compare that to ‘exotic’, which (according to a Google Books search) occurs a hundred times, and ‘exoticism’, which is used 17 times.

In general, the tenor of Schafer's Sinological writing from the 1960s indicates that it was meant for an audience of middle-aged white men who (like him) had grown up in a world of Western colonialism and its attendant Orientalism. This background enabled Schafer to imagine the Tang Chinese exoticizing their own colonial frontiers and colonial subjects in a sort of reverse Orientalism. As a result, his reading of Tang attitudes towards foreigners was actually more critical and less positive than many later treatments of Tang ‘cosmopolitanism’. He acknowledged that even during the eighth century, foreign merchants in the Tang faced strict and somewhat capricious restrictions on trade, corrupt officials, ‘arbitrary segregation laws’, and occasionally even violent massacres. He also argued that the ninth century was an ‘age of ambivalent attitudes’ when ‘a love of exotic things’ coexisted with ‘suspicion and persecution’ or ‘[d]istrust or hatred’ of foreigners.Footnote ⁴

Schafer seems to have drawn this view of ninth-century Chinese society from Arthur F. Wright's studies on the history of Buddhism in China, which interpreted the An Lushan Rebellion (755–763) as a turning point in Tang attitudes towards Buddhism. Here is Wright in a paper originally delivered to a conference of Sinologists in Paris in 1956 and subsequently published in the Journal of Asian Studies in 1957: ‘After the An Lu-shan rebellion, T'ang self-confidence and governmental effectiveness were not fully restored. The cosmopolitanism of the great days of T'ang slowly gave way, under the influence of barbarian attack and internal decay, to a cultural defensiveness which occasionally broke out into xenophobia.’Footnote ⁵ In Golden Peaches, Schafer cites this passage as his sole basis for reading the late Tang elite as ‘a generation of fear and attendant xenophobia’ that rejected and persecuted Buddhism and other foreign religions.Footnote ⁶ Wright would repeat the same claims, in only slightly different wording, in his book Buddhism in Chinese History (1959), which was based on lectures delivered at the University of Chicago but which drew heavily on the 1957 article.Footnote ⁷ It is worth noting that Wright only used the word ‘cosmopolitanism’ once in each of these works. He seems to have assumed that his readers would understand it in the same way that he did: as an openness to foreign cultures and religions that was the exact opposite of xenophobia.

A similar understanding of the word is reflected in the other influential survey of Chinese Buddhist history produced in this period, Kenneth Ch'en's Buddhism in China: A Historical Survey (1964). Ch'en made the following statements at the beginning of his chapter on Buddhism in the Tang: ‘Though the imperial clan claimed descent from Lao-tzu and thus favoured Taoism, the central authorities pursued a policy of religious toleration, giving each religion an opportunity to develop. Nestorian Christianity, Islam, and Manichaeism all were introduced during T'ang times, and each faith found adherents among the Chinese. The cosmopolitan ideal was upheld by the T'ang emperors because they regarded themselves as rulers not only of the Chinese but also of the barbarians.'Footnote ⁸ Here Ch'en went a little further in linking the ‘cosmopolitan’ to a notion of universal rulership and multi-ethnic empire, in addition to religious tolerance. But, like Wright and Schafer, he used the word only once in the entire book and did not subject it to any analytical discussion. It is also worth noting that Ch'en disagreed fundamentally with Wright and Schafer by identifying Buddhist–Daoist rivalry and ‘economic considerations’, rather than xenophobia, as the primary reasons for the state persecution of Buddhism in the 840s (more on which later).Footnote ⁹

The first effort at defining Tang ‘cosmopolitanism’ was made a full decade after the publication of Schafer's Golden Peaches. On the very first page of the Introduction to a volume of conference papers titled Perspectives on the T'ang, co-editors Arthur Wright and Denis Twitchett identified what they considered to be the two key features of Tang society: ‘First, was its eclecticism—the way that the T'ang drew together the many cultural strands from the tumultuous history of the preceding four hundred years. Second was its cosmopolitanism—its openness to foreign influences of all kinds.’Footnote ¹⁰ But one finds no examination of aspects of Tang ‘cosmopolitanism’ in the papers collected in Perspectives on the T'ang. Indeed, as Schafer (who was not one of the contributors) complained in a review of the volume, written in his lively and somewhat eccentric literary style:

In other words, the conference was heavily defined by Twitchett's interests in political, social, economic, and institutional history, and lacked serious engagement with the romantic, exotic, and fanciful dimensions of Tang culture and literature that Schafer so loved. As a result, the significance and extent of ‘cosmopolitanism’ in Tang cultural and intellectual history remained underexplored. Unfortunately, Schafer himself did little more to address this gap in scholarship, as his research interests had shifted to the Daoist poetry of the Tang and he did not return to the subject of Tang exoticism, save for an article on Tang poetry concerning the mysterious and perilous sea separating China from Japan.Footnote ¹²

Perspectives on the T'ang was read by only a relatively small Sinological audience. The more engagingly written Golden Peaches was enjoyed by a broader readership, and the survey histories of Chinese Buddhism by Wright and Ch'en were used as college textbooks for many years. But even more credit for popularizing the standard characterization of the Tang empire, and its capital city in particular, as ‘cosmopolitan’ should probably be given to a number of American and French textbook surveys of Chinese history that were written in the 1960s and 1970s and remained influential for decades afterwards. The first of these was Edwin Reischauer and John King Fairbank's East Asia: The Great Tradition (1960), which included Wright's Buddhism in Chinese History in its Bibliography (describing it as ‘an excellent brief survey’), and was evidently heavily influenced by his interpretation of Chinese Buddhist history.Footnote ¹³ Fairbank (who wrote the chapters on China) characterized the early Tang and its predecessors, the Northern and Southern Dynasties (or ‘Six Dynasties’), as an age of exceptional openness to foreign cultures and visitors:

Although this passage does not use the word ‘cosmopolitan’, on the facing page Fairbank suggests that states in Japan, Korea, and Yunnan imitated Tang institutions in part because of ‘the very cosmopolitanism of T'ang culture, which made it more attractive to foreign peoples'.Footnote ¹⁵ The next chapter also claims, in an interpretation owing much to Wright, that ‘the open, cosmopolitan spirit of the early T'ang gradually faded into a narrower, more exclusively China-centered and introspective attitude'.Footnote ¹⁶ In 1973, Fairbank, Reischauer, and Albert Craig (who, like Reischauer, was a Japanologist at Harvard) co-authored another textbook, East Asia: Tradition and Transformation, in which Fairbank again echoed Wright by claiming that after the An Lushan Rebellion, ‘in their rejection of foreign religions and their losing battle with the “barbarians”, [the Chinese] gradually lost the cosmopolitanism and cultural tolerance they had shown in the Six Dynasties period and early T'ang and became much more narrowly ethnocentric‘.Footnote ¹⁷

American historians like Fairbank who described Tang China at its height as culturally open and tolerant in spirit were implicitly setting up a contrast to a Maoist China that had cut off relations with the Western world and rejected its liberal democratic ideals. To their minds, the Tang ‘golden age’ was a model of what China could again become if it returned to being open and receptive to Western cultural influence. Fairbank, in particular, very likely had comparisons to the modern ‘treaty ports’ in mind, especially old Shanghai, when he described a Chang'an ‘literally crowded with foreigners'.Footnote ¹⁸ Schafer's approach to Tang ‘exoticism’, while later invoked to support the ‘cosmopolitan Tang’ image, was actually less sanguine, in that his preferred parallel to the Tang elite was the arrogant and ethnocentric colonial elites of the European empires, to whom ‘[f]oreign luxuries were too good for foreigners'.Footnote ¹⁹

In 1975, the American historian Charles Hucker published China's Imperial Past, a popular textbook survey of Chinese history from antiquity to 1850. Its chapter on the Tang painted this glamorous picture of Chang'an as a kind of medieval World's Fair:

This passage clearly draws much inspiration from both Fairbank and Schafer, but ignores Schafer's more ambivalent ‘colonial’ view of the Chang'an elite in favour of an unequivocally positive picture.

A similarly idealized characterization of Chang'an can be found in leading French Sinologist Jacques Gernet's Le Monde Chinois, first published in 1972 and translated into English (as A History of Chinese Civilization) in 1982 and again, in a revised edition, in 1996. To Gernet, Chang'an was ‘the centre of a cosmopolitan civilisation coloured by the influences of central Asia, of India, and of Iran’, and ‘the meeting-place of all the peoples of Asia—Turks, Uyghurs, Tibetans, Koreans, people from Khotan and Kucha, Sogdians, Kashmiris, Persians, Arabs, Indians, Cingalese’. In his opinion, ‘[t]his invasion of foreigners, of elements of distant cultures, of exotic products (slaves, animals, plants, foods, perfumes, medicines, textiles, and jewels) could not fail to affect the sensibilities of the age or to enrich the T'ang civilisation with its new contributions’.Footnote ²¹ Gernet used the same word cosmopolite to describe other cities, including Dunhuang 敦煌, Yuan-era Beijing (1271–1368), and the Iranian city of Tabriz under the Ilkhanate; this usage is again accompanied by language implying that the key feature of a cosmopolitan city is ethnocultural diversity and a large population of foreigners from many different countries.Footnote ²² It is also worth noting that Gernet, like Fairbank, relied heavily on Wright's interpretation of the Tang after the An Lushan Rebellion as xenophobic and anti-Buddhist, to the extent of likening such sentiments to modern nationalism.Footnote ²³

Although Gernet described Chinese civilization in the first half of the Tang as ‘cosmopolitan’, and Hucker described Chang'an as a cosmopolitan city, they did not refer to the Tang as a ‘cosmopolitan empire’. The popularization of ‘cosmopolitan empire’ as a description of the Tang most likely began with Patricia Ebrey's widely used textbook The Cambridge Illustrated History of China (1996), which titled its chapter on the Tang as ‘A Cosmopolitan Empire’ and boldly declared (echoing Fairbank): ‘More than in any other epoch in Chinese history before the twentieth century, Chinese in early and mid Tang had the self-confidence to be open to the new and different...Chinese in this period were more than happy to gather about them the best of what the rest of the world had to offer.’ Ebrey described the culture of the Tang capitals, Chang'an and Luoyang 洛陽, as ‘enthusiastically cosmopolitan’, and went on to list some standard examples: Buddhist pilgrim Xuanzang's 玄奘 (circa 602–664) travels to Central Asia and India; the presence of foreign envoys, merchants, and pilgrims and their religions; the fascination with foreign goods and fashions; the popularity of polo (an Iranian game introduced via Central Asia); the placement of ceramic figurines of Central Asians in tombs; and foreign influence on the decorative arts, musical culture, and interior furnishings.Footnote ²⁴

In the early twenty-first century, the title of Mark Edward Lewis's textbook China's Cosmopolitan Empire: The Tang Dynasty (2009) further established the phrase ‘cosmopolitan empire’ as a standard description of the Tang. Lewis, a specialist in the intellectual and cultural history of pre-imperial and early imperial China rather than the Tang, produced the book by synthesizing a wide range of secondary sources in English, as well as a smaller number in Chinese, Japanese, French, and German. Despite the book's title, it does not, in fact, use cosmopolitanism as a central framework for analysing Tang society and culture. Instead, the word only appears once in a thematic chapter on Tang foreign relations, titled ‘The Outer World’, in which Lewis simply paraphrases the formulations of ‘eclecticism’ and ‘cosmopolitanism’ given in his Perspectives on the T'ang in 1973.Footnote ²⁵ Much of the rest of Lewis's discussion of foreigners in the Tang is heavily dependent on Schafer, and thus replicates Schafer's emphasis on the foreign presence in Chang'an and other cities and the popularity of Central Asian music and dance. However, Lewis did not engage with Schafer's argument that the Tang elite avidly consumed foreign luxuries without necessarily welcoming or liking foreigners. Lewis did make use of more recent scholarship on Tang foreign relations, notably Marc Abramson's Ethnic Identity in Tang China (2008), which employs concepts other than cosmopolitanism to analyse Tang elite attitudes towards ethnic and cultural diversity.Footnote ²⁶ Unfortunately, his choice of title for the book reflected uncritical acceptance of an outdated historiographical model, one that prevented him from presenting a more sophisticated and up-to-date discussion of the Tang as a multi-ethnic and multicultural empire.

The use of the phrase ‘cosmopolitan empire’ by Ebrey and Lewis may reflect a broader discursive change surrounding the idea of empire. Using the Google Ngram Viewer tool, I found that, while the use of this phrase in published English-language literature (as reflected in scanned texts collected by Google Books) is still relatively uncommon, it has increased quite steadily since the late 1990s, after 70 years of very low usage (see Figure 1).

Figure 1.

Usage of the phrase ‘cosmopolitan empire’ from 1800 to 2019 (based on Google Ngram).

This growing tendency to characterize premodern empires as cosmopolitan is, I believe, at least partly related to a much bigger increase in the popularity of the concept of cosmopolitanism among Western intellectuals since the end of the Cold War, also reflected in Google Ngram Viewer (see Figure 2).

Figure 2.

Usage of the phrase ‘cosmopolitan empire’ and the words ‘cosmopolitanism’ and ‘cosmopolitan’ from 1800 to 2019 (based on Google Ngram).

In Western academia since the end of the Cold War, the idea of cosmopolitanism has come to serve as the basis for a philosophical or ethical ideal of global citizenship that advocates for universal human rights, diversity, and inclusiveness and rejects the influence of nationalism, racism, religious bigotry, and nativism.Footnote ²⁷ Historians with liberal or multiculturalist leanings tend to think of themselves as cosmopolitan, or at least aspire to be so, and to prefer living in hyper diverse and heavily globalized (thus ‘cosmopolitan’) cities. This has made it difficult to resist idealizing the ‘cosmopolitan’ elites and urban centres of premodern empires that we have chosen to study, rather than subjecting them to the critical scrutiny that we typically direct towards their modern colonial counterparts. The editors of a recent volume on cosmopolitanism and empire in the ancient Near Eastern and Mediterranean worlds offer an insightful critique of this tendency:

In effect, a subset of early twenty-first century historians seem to be prone to assuming that an empire can be viewed positively if its imperialism is multicultural, tolerant, and ‘cosmopolitan’ rather than, say, founded on ideas of racial and religious supremacy. Whereas premodern empires were typically praised as ‘great’ on the basis of their military power and territorial conquests for much of the twentieth century, historians who admire them for whatever reason tend now to glamorize them as engines of cross-cultural exchange and global trade—that is, as agents of early globalization. In that context, the Tang empire's close association with the frequently romanticized idea of the ‘Silk Road’—a modern concept coined by a German geographer in around 1838—has made it a perfect candidate for such glamorization.Footnote ²⁹

But glamorization of the Tang is, of course, not a solely Western phenomenon. Positive appraisals of the Tang as a glorious period of Chinese history were suppressed as ‘rightist’ in Maoist China, where the true golden age was believed to be in the socialist future rather than the dark ‘feudal’ past. But they became entrenched among Chinese communities outside the People's Republic (including those in Hong Kong and Taiwan), where the idealized image of Emperor Taizong 太宗 (r. 626–649) as a great conqueror and model Confucian ruler, receptive to honest criticism from his ministers, was particularly popular. Since its reintroduction to the mainland in the 1980s, the interpretation of the Tang as a ‘golden age’ (shengshi 盛世) of national power and prosperity has been increasingly politicized and promoted by the communist regime, which has moved away from a class-based denunciation of ‘feudal’ imperial dynasties and re-embraced the country's imperial past as a locus of nationalist nostalgia and cultural pride. The ‘Great Tang’ now features heavily in state propaganda surrounding Xi Jinping's ‘Chinese Dream’ ideology and the Belt and Road Initiative (billed as a revival of both the Tang-era ‘Silk Road’ and a later ‘Maritime Silk Road’), as well as Chinese political discourse on the idea of a future Sinocentric world order.Footnote ³⁰

In recent decades, the image of the Tang as culturally ‘cosmopolitan’ has also been popularized in Chinese society by media depictions (e.g. movies and television dramas), as well as reprints of Xiang Da's 向達 classic 1933 study ‘Tangdai Chang'an yu xiyu wenming’ 唐代長安與西域文明 (Tang Chang'an and the Civilization of the Western Regions) and translations of Western works like Schafer's Golden Peaches. Chinese historians frequently hail the Tang's relative success in integrating Inner Asian peoples into a Sinitic empire—success purportedly symbolized by Taizong's assumption of the hybrid Sino-Turkic title Celestial Khagan 天可汗—as an ideal precedent for China's modern official identity as a ‘unified, multi-ethnic nation’ in which ethnic minorities are improved and ultimately assimilated by the ‘advanced civilisation’ of the Han.Footnote ³¹ But this ironic use of the idea of cosmopolitanism to promote nationalism and, arguably, a form of cultural imperialism and ethnic supremacism is not the whole story. China's increasingly repressed and isolated liberal intellectuals (i.e. those who support classical liberal ideas like political freedom and freedom of speech) are also drawn to the popular image of the Tang as a utopian age of intellectual pluralism and openness to foreign ideas and cultures. Conservative authoritarian nationalists see early Tang society's fabled openness to foreigners and foreign cultures as a product of the supreme self-confidence and security afforded by military power and wealth: a strong China can afford to be open on its own terms, while a weak China cannot. The liberals, however, would prefer to credit the empire's greatness and strength to the openness itself and use the Tang ideal to argue for more intellectual diversity, tolerance, freedom, and receptivity to Western liberal ideals in today's censorship-obsessed China. In other words, Chinese conservatives and liberals may disagree over what kind of political culture their country should have, but they share common ground in viewing the Tang, and the Tang from Taizong's reign to Xuanzong's in particular, as a model and precedent for it. Such is the power and appeal of the ‘golden age’ mythos.Footnote ³² In addition, whereas earlier generations of Chinese feminists based their arguments for women's rights in Western liberal or socialist values and rejected Confucian values as irredeemably patriarchal, feminists in China may now choose to use the mythologized image of the Tang as an exceptional age of female empowerment and sexual liberation to argue for the compatibility of gender equity with both Chinese nationalism and traditional Chinese culture.Footnote ³³

A constellation of cosmopolitanisms

Each of the treatments of Tang ‘cosmopolitanism’ that I have described above engages in a kind of presentism that over-simplifies, sanitizes, and romanticizes the Tang, as is generally the case with any historical era nostalgically elevated as a country or civilization's golden age. Thinking more comparatively and globally might alert us to the fact that the existence of significant ethnic and cultural diversity, or of openness to foreign cultures, does not make an empire inherently more benign or enlightened than an ethnic nation-state. After all, empires by their very nature are built on military conquest and domination of other peoples, and frequently engage in selective appropriation of the conquered peoples’ ‘exotic’ cultural practices in a way designed to symbolize their subjecthood and enhance the imperial elite's prestige. The Tang was no exception to this rule, and its aspirations and pretensions to geopolitical pre-eminence and universal hegemony rarely went unchallenged by its stronger neighbours. Rather than achieving a ‘Pax Sinica’, the empire constantly had to assert and maintain its supremacy on various frontiers at great cost in blood and treasure, a cost paid not by officials and their families (who were exempted from taxes and conscription) but by the long-suffering peasantry, for whom life was never all that ‘golden’. Naturally, the cost in ‘barbarian’ lives was heavy as well. Throughout the dynasty's history, victorious Tang armies were known to pillage enemy or ‘rebel’ cities and massacre or enslave their inhabitants in a manner that the Han Chinese of today (due to ethnocentric or nationalistic biases in their historical education) tend to associate only with ‘barbaric’ Inner Asian and Japanese invaders.Footnote ³⁴ Yet many modern historians have mistakenly assumed that a massacre of foreign merchants in Yangzhou 揚州 by a rebel-quelling army in 760 was a previously unheard-of act of xenophobia and a sign of the waning of Tang ‘cosmopolitanism’ after 755, rather than part of a long-standing culture of predatory violence in the Tang military.Footnote ³⁵

The state-centred and elitist bias of our sources is part of the reason for our often insufficiently critical perspective on the ‘cosmopolitanism’ of the Tang. The vast majority of extant texts from the Tang naturally reflect a Chang'an-centred perspective, since they were written by officials and literati whose worlds did revolve around the imperial capital.Footnote ³⁶ As a result, we have a much deeper understanding of elite life in Chang'an than for any other locality in the Tang empire. But every empire has a darker side of oppression and violence that is usually most visible at its unstable, ambiguous, and heavily militarized edges, not at its political and cultural centre, though it is visible even at the centre if we look hard enough.Footnote ³⁷ One can therefore better understand the Tang empire as an empire by maintaining some critical emotional distance from the Chang'an elite and not empathizing or identifying exclusively with their point of view. In this regard, I agree fully with a comment that Finbarr Flood made recently in the context of the role of slavery in the Indian Ocean world of maritime trade routes, ports, and networks: ‘If our interest in recuperating histories of the premodern global is to be more than a form of narcissism—cosmopolitan elites privileged with the contemporary means of mobility recognizing themselves in the mobility of premoderns—the question of subaltern labor and the violence that often accompanied it has to be folded into any consideration of the histories that we write.’Footnote ³⁸

Unlike Indian Ocean merchants, the defining characteristic of the Tang ‘cosmopolitan elite’ was not mobility across oceans and state borders but consumption of foreign luxuries, fashions, and entertainments. But even today, accounts of Tang ‘cosmopolitanism’ rarely acknowledge that the ‘[m]usical troupes, jugglers, acrobats, dwarfs, and blacks from distant realms’ (as Hucker enthusiastically described them), as well as the Sogdian bar wenches (huji 胡姬) celebrated in Li Bai's 李白(701–762) poems, had mostly come to Chang'an not as voluntary visitors, ‘attracted to this, the greatest metropolis of the world’ (in Fairbank's words), like aspiring artists to modern New York City, but as slaves brought into China by tribute missions or traders.Footnote ³⁹ A significant part of the Tang capital's ethnic and cultural diversity was thus a result of human trafficking and what Schafer called ‘human tribute’, not conventional immigration.Footnote ⁴⁰ As Don Wyatt recently argued, ‘just as it is crucial for us to recognize the Tang period as the apogee of cosmopolitanism, we must acknowledge and accept that in this same era the Chinese themselves began to enslave substantial numbers of non-Chinese’.Footnote ⁴¹ It would be ahistorical and ethically problematic to elide this fact just because it does not cohere with a twenty-first century audience's ideal mental image of a ‘cosmopolitan’ society.

I would suggest that we can also understand the character of Tang ‘cosmopolitanism’ more holistically by situating it within a transregional historical context, rather than one defined by the traditional Sinological reliance on written sources produced by and for the Chinese elite.Footnote ⁴² Such a holistic approach would see Tang ‘cosmopolitanism’ not as centred on, and emanating from, the city of Chang'an and the exotic tastes of its elite denizens, but rather as a product of the whole empire's interactions (or, in some cases, lack thereof) with a complex constellation of large, overlapping cultural, religious, and economic spheres or ‘cosmopoleis’ (see Supplementary material, Figure 3).

At least two of these spheres were a direct product of other recent projects of imperial conquest and expansion: the Turkic world of the steppes, which stretched from Mongolia to the Volga and the Caucasus; and the Islamic or Islamicate world, which emerged concurrently with the Tang empire and by 750 extended from Spain and Morocco in the west to Sogdiana and Sindh in the east.Footnote ⁴³ The transformation of the relationship between these worlds from conflict to synthesis is one of the more consequential themes of medieval Eurasian history. As reflected by the Tang emperors’ use of the title Celestial Khagan, a more short-lived blending of the political cultures of the Sinitic and Turkic worlds occurred in the early Tang; Jonathan Skaff has analysed this blending in detail and interpreted it as a product of Tang imperial expansion, ideological competition with Turkic empires, and integration of Turkic leaders into the Tang military elite.Footnote ⁴⁴ In contrast, the Tang literati elite seems to have had very limited interaction with Islamicate culture and the religion of Islam, despite the presence of Muslim merchant communities in Yangzhou and Guangzhou 廣州. There is no authentic evidence of a permanent Muslim presence in Tang-era Chang'an, or of Chinese conversion to Islam (pace Kenneth Ch'en's claim to the contrary).Footnote ⁴⁵ An account of the Islamic world by one Du Huan 杜環, who apparently lived in the Abbasid capital between 751 and 762, may have contributed somewhat to Chinese understandings of Islam, but it is highly doubtful that a late ninth-century Tang emperor would be familiar with the prophets of Islam and have a casket filled with images of them, as an Arab writer claimed in the early tenth century, based on the testimony of a traveller named Ibn Wahb al-Qurashi.Footnote ⁴⁶ Meaningful cultural synthesis between the Sinitic and Islamic worlds probably did not begin until immigration of non-merchant Muslims to China from Central Asia and the Middle East increased significantly under the Mongol Yuan empire, creating the conditions for a community of Sinophone Muslim literati to emerge in the Yangzi delta.Footnote ⁴⁷

Other spheres of cosmopolitanism extended beyond and between empires. One of the oldest was the pre-Islamicate Iranian world, which included the Sasanian empire and the states of Sogdiana before their fall to the Muslims, but also the Sogdian merchant networks and immigrant diasporas that dominated Central Asian trade and extended deep into the Sinitic and Turkic heartlands from the sixth century to the ninth. In recent decades, archaeological discoveries and new research have given us a better understanding of the scale of the Sogdian presence in Inner Asia and China during these centuries, as well as its role as a carrier of aesthetic and religious cultures from the pre-Islamicate Iranian civilizational sphere to the Sinitic and Turkic worlds.Footnote ⁴⁸ To a large extent, the Tang elite inherited its ‘cosmopolitan’ taste for certain aspects of Central Asian culture from the influence that the Sogdian diaspora had already exerted in the capitals of the preceding Northern Dynasties, so it is incorrect to credit this form of cosmopolitanism solely to the Tang.Footnote ⁴⁹ It should also be noted that the majority of Zoroastrians, East Syriac (‘Nestorian’) Christians, and Manichaeans in the Tang appear to have been Sogdians, and that (again, pace Ch'en)Footnote ⁵⁰ there is no evidence to date of Chinese conversion to these religions originating from Iran.

Another of the transimperial cosmopoleis was the Indian Ocean world, in which Arabian, Iranian, Indian, Sri Lankan, Southeast Asian, and Chinese ports became increasingly interconnected through a Muslim merchant diaspora after the seventh century.Footnote ⁵¹ This trading world eventually became a vehicle for the Islamic world's expansion into maritime Southeast Asia during the thirteenth and fourteenth centuries. Prior to that expansion, however, the Indian Ocean world overlapped with much of a ‘Sanskrit cosmopolis’ proposed by Sheldon Pollock, in which South Asian and Southeast Asian elites employed classical Sanskrit literature (transcribed using a variety of Indic writing systems) as a common source of political and social capital from circa 300 ce to circa 1300 ce.Footnote ⁵² The Tang's engagement with this ‘Sanskrit cosmopolitan order’ was minimal; only a relatively small number of Tang subjects (mostly monks) mastered the Sanskrit language, either when travelling to India on Buddhist pilgrimage or by learning from Indian monks in China.Footnote ⁵³ And they used literacy in Sanskrit purely for reading and translating Buddhist sutras, not literary works like the Ramayana. But the fruits of earlier sutra translation projects, undertaken by Central Asian monks in China from the second to fifth centuries, allowed the Tang to participate in a larger ‘Buddhist cosmopolis’, in which travelling Buddhist monks (and the scriptures they carried and translated) used land and sea routes to connect religious communities in a space spanning India, Tibet, and the Tarim Basin to Korea and Japan.Footnote ⁵⁴ China's engagement with the Buddhist cosmopolis predated the Tang by several centuries, but the Tang was the period when Chinese Buddhists began to interpret their country as a second centre of the Buddhist world after India, in contrast to the peripheral place it had previously occupied in the Buddhist world view.Footnote ⁵⁵ Buddhist cosmopolitanism in the Tang thus underwent a shift from Indocentrism to Sinocentrism that would be cemented in the post-Tang period.

Due to the popularity of the ‘Silk Road’ image, historians have tended to take cultural influences from the Turkic, Iranian, Islamic, and Indic worlds as the benchmark for a ‘cosmopolitan’ Chinese empire. But we should recognize another kind of cosmopolitanism, in which Chinese cultural influence flowed outwards, as equally deserving of the label ‘cosmopolitan’. A ‘Sinographic cosmopolis’ or Sinographic sphere emerged in East Asia from the sixth to ninth centuries, as an outcome of secondary state formation in four polities that successfully resisted or escaped incorporation into the Chinese empire while remaining within the Chinese tributary orbit.Footnote ⁵⁶ In part through educational missions that gave them access to a classical Sinitic education in Tang territory, elite men from Silla, Japan, Bohai/Parhae 渤海, and Nanzhao 南詔 became well-versed in classical literary Sinitic and the non-phonetic Sinographic (‘Chinese’) script, even though they spoke mutually unintelligible languages.Footnote ⁵⁷ Besides using Sinographic writing as a kind of ‘scripta franca’, countries within the Sinographic cosmopolis also tended to share political ideologies and institutions derived from the Tang imperial model, including state patronage of Buddhism and the study of the Confucian classics.Footnote ⁵⁸ In addition, Sinographic literacy was the vehicle through which Korean and Japanese monks and aristocrats became participants in the Buddhist cosmopolis, despite their countries’ extremely limited contact with Central Asia and India.

It is important to recognize the decisive role that East Asia's geography played in shaping the Sinographic cosmopolis, and thus avoid the kind of Sinocentrism reflected in Fairbank's claim in 1960 that ‘the very cosmopolitanism of T'ang culture…made it more attractive to foreign peoples’ such as the Koreans and Japanese. Literacy in Sinographs is significantly harder to acquire than literacy in alphabetic and abugidic writing systems, and literary Sinitic elite culture (which could only be accessed by acquiring Sinographic literacy) thus held little appeal to a wide range of peoples in Inner Asia and Southeast Asia who had the option of writing their languages in scripts derived from Brahmic or Sogdian. In other words, the difficulty of mastering Sinographs actually limited the transregional exportability and accessibility of Tang Chinese elite culture, making it far less ‘cosmopolitan’ than Indic Sanskrit culture or even Sogdian culture. The people of Silla, Bohai, and Japan chose to become Sinographic because, being located at the eastern end of Eurasia, they lacked access to other writing systems and literate cultures—not because of the inherent attractiveness of Sinitic culture. The people of north Vietnam were Sinographic not by choice, but because they were colonized and ruled by Chinese empires for a thousand years; to their south were the Chams and Khmers, who ruled themselves and chose to be part of the Sanskrit cosmopolis instead.

Nanzhao, a Yunnanese buffer state sandwiched between the Tang and Tibetan empires, was a special case in that its kings chose Sinographic literacy over literacy in Tibetan (a Brahmic-derived script) as part of a diplomatic deal made with the Tang empire in the late 700s, under which Nanzhao elite men gained access to a heavily subsidized educational programme in Sichuan in exchange for an alliance against the Tibetan empire. With the literacy and administrative skills thus gained, Nanzhao eventually cast off its vassalage to the Chinese emperor and declared itself an empire in its own right, resulting in a 20-year war with the Tang from 860–880. Nanzhao's rise to imperial status is a good example of how states that joined the Sinographic cosmopolis did so for their own interests rather than being passively drawn to the charisma or ‘superiority’ of Chinese civilization, contrary to the self-aggrandizing claims of Tang imperial rhetoric.Footnote ⁵⁹

If we understand a cosmopolitan individual as one with the ability to cross or straddle cultural boundaries with relative ease, then the most important representatives and vectors of ‘cosmopolitanism’ in this constellation of cosmopoleis were travelling monks, merchants, interpreters, scholars, and migrants of various ethnicities who were able to bridge cultures within and between the cosmopoleis. Their cosmopolitanism, defined by familiarity with multiple languages and cultural traditions, would have been of a much deeper kind than that of Tang emperors, aristocrats, and literati officials engaging in conspicuous consumption and treating foreign objects, foodways, and fashions mostly as exotic symbols of social and political prestige, without any deep engagement with the cultures that produced them. Moreover, these ‘cosmopolitan’ individuals often accomplished their boundary-crossing in spite of, not because of, the laws on foreign trade and travel that the Tang political elite sought to enforce.

Contrary to the popular notion of ‘openness’ being the key attribute of a ‘cosmopolitan’ empire, the Tang state more often sought to control and restrict human movement across its borders than to remove barriers to such movement.Footnote ⁶⁰ Even within the empire, all travel was heavily regulated by a system of state-issued passes and checkpoints. According to a late seventh-century biography of Xuanzang, he violated a ban on foreign travel when departing for his celebrated pilgrimage to India in 629; less well-known is the fact that such laws remained in force long after his travels.Footnote ⁶¹ Under Tang law, it was illegal for any subject of the emperor to leave the empire on non-official business, whether by land or sea; it was also illegal for foreign merchants to enter Tang territory without government authorization.Footnote ⁶² Thus, although popular accounts of the ‘Silk Road’ credit the Tang with promoting and protecting private foreign trade, the Tang emperors actually criminalized such commerce, especially when it involved the export of silk and other high-value goods like gold and silver. The notable exceptions when it came to overland trade were a special category of expatriate Sogdian merchants known as xingsheng Hu 興生胡 or xing Hu 興胡, whom the Tang state allowed to engage in cross-border trade on the northern and north-western frontiers, in exchange for paying taxes and tolls.Footnote ⁶³ But even these Sogdians were forbidden to export silk from Tang territory. Laws banning the export of all or most forms of silk textile are found in the Tang Code and are known to have been reissued, in revised form, in 714, 737, and 780.Footnote ⁶⁴ Of course, there probably was private foreign trade in silk and other embargoed goods along Tang frontiers in Central Asia, but any such trade was effectively smuggling, carried out in violation of imperial law. Restrictions on foreign trade at the port of Guangzhou, Tang China's primary maritime link to the Indian Ocean world, were probably looser, but by 737, Tang law also forbade the transportation of silk to the Lingnan 嶺南 region, presumably to prevent its private export by foreign maritime merchants.Footnote ⁶⁵ These embargoes on private foreign trade in high-value commodities, which persisted throughout Tang history but were abolished under the Song, are best interpreted as an effort to compel foreign rulers to participate in the Tang system of tributary relations in exchange for luxury goods.Footnote ⁶⁶

In addition to these trade restrictions, Tang law also forbade marriages between Tang subjects (of any ethnicity) and subjects of other states (known legally as huawairen 化外人, ‘people from beyond the bounds of civilization’), except in the case of foreign envoys granted an extended stay in Chang'an—and even they were forbidden to take their Tang-subject wives back to their home countries. I interpret such laws primarily as measures to prevent foreign agents from using Tang subjects as spies or sources of information. For the same reason, even simple interactions between foreign envoys and ordinary Tang subjects or local officials were subject to extremely strict legal limits during an embassy's journeys to and from Chang'an.Footnote ⁶⁷ Despite the empire's ethnic diversity, the non-elite population's ability to interact with people of other states and empires in Eurasia was thus highly circumscribed by law, with consequent limits on the potential for cross-cultural exchange beyond the confines of the imperial court. In sum, the Tang was never as open in its treatment of foreign visitors and foreign trade as many modern historians have assumed any ‘cosmopolitan empire’ to be.

‘Cosmopolitanism’ in late Tang and beyond: Transformation, not decline

Building on the critical and holistic approach to ‘Tang cosmopolitanism’ outlined above, the markedly different elite cultures and aesthetics of the Tang and Song periods can be better explained by changes in the constellation of cosmopoleis (see Supplementary material, Figure 4) than by the oft-claimed decline in Chinese ‘cosmopolitanism’ during the late Tang and Song. Rather than assume that the only significant changes were in the Chinese elite's attitudes towards foreign cultures and sense of identity, we need to pay more attention to what else was changing in Eurasia around the time of the Tang-Song transition. My approach interprets Chinese ‘cosmopolitanism’ as evolving and transforming in response to these changes, and thus differs from those of Charles Holcombe and Jonathan Skaff, both of whom have postulated a diminution of ‘cosmopolitanism’ in the Chinese elite during the Tang-Song transition but used terms other than the traditional ‘xenophobia’ (i.e. ‘Confucian universalism’ or ‘literati Confucian exclusivism’) to describe what took its place.Footnote ⁶⁸

It should be emphasized that Tang China's connections to the Buddhist cosmopolis remained strong to the dynasty's very end; what changed after the Tang was the fracturing of the cosmopolis itself due to the ‘domestication’ or indigenization of Chinese Buddhism and the rise of the Sinographic cosmopolis. The notion that the late Tang was anti-Buddhist for xenophobic reasons, which (as we have seen) was popularized by Wright and Schafer and remained standard for decades, can be refuted with abundant evidence of the Tang court's fervent patronage of the Indian or Sogdian (or Indo-Sogdian) Esoteric Buddhist master Amoghavajra (705–774) as a source of supernatural protection during and after the An Lushan Rebellion.Footnote ⁶⁹ From 760 to 819, the court also maintained a tradition of having a Buddha fingerbone relic, normally housed in the Famen Temple 法門寺, brought into Chang'an to be displayed and venerated every 30 years to bring the empire a new karmic dispensation of peace and prosperity.Footnote ⁷⁰ Han Yu's 韓愈 (768–824) famous 819 memorial to Emperor Xianzong 憲宗 (r. 805–820), in which he denigrated the Buddha as a barbarian (Yi-Di 夷狄) and called for the bone relic to be destroyed, is often taken as reflecting the purportedly xenophobic spirit of his age.Footnote ⁷¹ On the contrary, Han Yu directed polemics against both Buddhism and its indigenous competitor, Daoism, rejecting them as immoral and inimical to Chinese civilization and calling for them to be proscribed, and his views were regarded as outrageously extreme and narrow-minded by most of his peers. Han Yu's Confucian exclusivist views became popular among segments of the Chinese literati under the Song dynasty, due to a revival of interest in imitating both his powerful prose style and his critique of ideological pluralism.Footnote ⁷² But rigid Confucian exclusivism was never state policy in either Tang or Song, and it is simply wrong to project Han Yu's personal prejudices onto late Tang or Song society at large.

As for Emperor Wuzong's 武宗 (r. 840–846) infamous persecution of Buddhism in 842–846, all extant evidence suggests this was rooted not in popular xenophobia but in competition between the Buddhist and Daoist clerical establishments for imperial patronage, a centuries-old rivalry in which Daoists frequently employed ethnocentric arguments to assert that a Chinese religion had to be superior to one from the barbarians, or at least that it was better suited to the Chinese.Footnote ⁷³ Both the Japanese pilgrim-monk Ennin's 圓仁 (793/4–864) diary and Tang court sources make it evident that Wuzong began persecuting Buddhism under the influence of Daoist priest-alchemists who had promised him an elixir that would enable him to become an immortal transcendent.Footnote ⁷⁴ Ennin also tells us that the Daoists actively poisoned Wuzong's mind against Buddhism: they cited a prophecy that implied that monks would seize the throne from him, and they claimed in 845 that the qi 氣 imbalance caused by a foreign religion's continued existence in China was blocking Wuzong's path to immortality, even after he consumed alchemical elixirs and ascended the massive ‘immortal's terrace’ that he had constructed on their advice. Ennin claims that Wuzong then resolved to eradicate Buddhism from his empire completely.Footnote ⁷⁵ In 1956, Kenneth Ch'en argued that economic considerations were more important than Buddhist-Daoist rivalry in Wuzong's decision to persecute Buddhism, in that he wished to replenish the imperial treasury by seizing the immense wealth of the tax-exempt monasteries.Footnote ⁷⁶ However, the fact that Wuzong did not target the equally wealthy Daoist religious establishment, and instead lavished more patronage on it, indicates that immortality-seeking rather than economics lay at the heart of the persecution.Footnote ⁷⁷

The Buddhist monastic establishment was hit hard by mass monastery closures and property confiscations during Wuzong's persecution, but it soon recovered fully after his death (from elixir poisoning), thanks to generous support from his successors. Most monasteries were reopened and the vast majority of forcibly laicized monks and nuns quickly returned to monastic life.Footnote ⁷⁸ Chinese Buddhism continued to flourish under the Song, contrary to a once-prevalent narrative (espoused by Arthur Wright and Kenneth Ch'en, among many others) of its irreversible post-Tang decline.Footnote ⁷⁹ It did, however, lose its formerly vibrant connections to Buddhist communities and institutions outside the Sinographic cosmopolis, including Central Asia, India, Sri Lanka, Tibet, and Southeast Asia. Earlier scholarship has tended to attribute this to Buddhism's decline in the Indian subcontinent and disappearance from Islamic western Central Asia. But Tansen Sen has argued that past studies overestimated the speed and extent of the decline of Indian Buddhism, and that the real cause for Chinese Buddhism's estrangement from its Indian roots lies in its increasing indigenization into popular schools and sects, most notably Chan 禪, that did not rely on models and scriptures from India.Footnote ⁸⁰ Rather than China simply exiting the Buddhist cosmopolis, the Buddhist cosmopolis itself broke up into separate Sinographic and Indic spheres that rarely interacted via either translation or travel. Song Buddhism was thus significantly more Sinocentric or East Asian in orientation than its Tang predecessor and increasingly also diverged from the more India-oriented Esoteric Buddhism of Tibet, though (as discussed earlier) the foundations for this shift to Sinocentrism had already been laid in Tang times.

The fading of Central Asian influences from Chinese elite culture had much to do with complex historical changes in Central Asia itself, most of which occurred only towards the end of the Tang or even later. Although the Tang did lose its Central Asian empire to the Tibetans and Uyghurs by the beginning of the ninth century, skilled musicians of Central Asian origin remained popular at the Tang court. Some modern historians have claimed that rising xenophobia in the late Tang led to a rejection of Central Asian music, but the Yuefu zalu 樂府襍錄, a compilation of Tang court music history and lore written in the 890s, shows no such change in attitudes.Footnote ⁸¹ The frequent use of a few satirical ‘new yuefu’ 新樂府 poems from the early ninth century as evidence of a general turn against Central Asian music and clothing styles is actually illogical, because the poetry is aimed precisely at mocking these foreign fashions’ continued popularity among the Chang'an elite despite their potential for interpretation as ill omens of a barbarian invasion.Footnote ⁸² As the preface to the Yuefu zalu suggests, the influence of Central Asian musical arts in China probably declined only after the sacking of Chang'an and the death or dispersal of its court musicians in the rebellions and civil wars of the late ninth century.Footnote ⁸³ It was practically impossible to reconstruct the lost repertoire of foreign musical styles, as the Song dynasty's diminished contact with Central Asia and the Iranian world prevented it from importing musicians skilled in those styles.

Whereas the Tang military presence in the Tarim Basin from the 690s to the 790s left virtually no trace culturally and did not result in westward expansion of the Sinographic cosmopolis, Central Asia from the ninth to eleventh centuries was profoundly reshaped ethnically and culturally by Turkic immigrants from the east and Muslim Arab and Iranian immigrants from the west. A Muslim dynasty of Iranian origin, the Samanids, ruled over Khurasan, Sogdiana, and Fergana in the ninth and tenth centuries and promoted a new courtly culture in which New Persian, written with the Arabic script rather than Pahlavi, displaced both Sogdian and Arabic as the language of elite writing. Turkic Muslim dynasties that emerged from the Samanid realm, namely the Ghaznavids and Seljuks, spread the new Persianate elite culture to Afghanistan, the Indian subcontinent, and Iran, creating a cultural sphere that some scholars now call (with inspiration from Pollock) the Persian cosmopolis.Footnote ⁸⁴

The Samanids fell in the 990s to the Turkic Muslim Qarakhanids, under whose rule Persian-speaking Muslim Sogdians and Ferganans adopted a new identity as Tajik, a Persian term once used to refer primarily to Arab Muslims.Footnote ⁸⁵ Like Sogdian, the Saka Iranian languages of Kashgar and Khotan disappeared after these western Tarim oasis states’ conquest by the Qarakhanids and were replaced by Middle Turkic.Footnote ⁸⁶ The Turkic world increasingly overlapped with the now-coterminous Iranian and Islamic worlds in Central Asia, and ceased to interact culturally with the Sinographic cosmopolis.Footnote ⁸⁷ The main exceptions were the semi-Sinographic Buddhist kingdoms of the Uyghurs in Gansu, Turfan, and the eastern Tarim Basin, which developed a unique mix of Sinitic, Sogdian, and Turkic cultures. The Gansu Uyghurs remained Buddhists after coming under Tangut and then Mongol rule, while the Tarim and Turfan Uyghurs finally converted to Islam in the fifteenth century, under the rule of the Mongol Chagatai Khanate. Ironically, by then the Chinese had begun referring to all Muslims as Huihui 回回, a term derived from the ethnonym Uyghur and originally used for Central Asians of any religion.Footnote ⁸⁸

Due to much of Central Asia's political and cultural reorientation towards the Islamic world after the ninth century, the Sinitic heartland ceased to be an attractive destination for Turkic and Sogdian immigrants, who instead chose to move west towards the Islamic heartland.Footnote ⁸⁹ Already by early Song times immigrant families in China had assimilated into the general population. Why they did so, after centuries of maintaining their ethnic and cultural distinctiveness, remains poorly understood. But, contrary to claims that have been made for about 20 years, it is very unlikely to have been due to prejudice and persecution.Footnote ⁹⁰ Consider that Sogdian officers remained prominent in the late Tang imperial guards and may even have increased in number, and that Türks and Turkicized Sogdians (who had merged into a quasi-ethnic grouping known as the Shatuo) ruled three of the Five Dynasties in tenth-century North China.Footnote ⁹¹ The Shatuo themselves appear to have assimilated relatively quickly into the Chinese population after losing their privileged status with the end of the last Shatuo dynasty in 951. They were the last Turkic military elite to play a significant role in Chinese political history; in contrast, the westward migration of Turkic-speaking steppe peoples through Central Asia gave them immense influence in the politics of the Islamic world for centuries to come.

The division of the Buddhist cosmopolis, the Turkic migrations, and the assimilation of the Sogdians resulted in Song China being less connected to the cultures of Central Asia, the Turkic world, and Iran than the Tang had been. But one could argue that the Song surpassed the Tang in connections to the rest of the Sinographic cosmopolis and to the Indian Ocean world. The key difference is that these connections were no longer state-centred and defined by tributary relations and the culture of the imperial capital elite; instead, they were shaped and sustained by private maritime trade. Any claim that the Song was less ‘cosmopolitan’ than the Tang thus rests mostly on an ahistorically static understanding of cosmopolitanism that focuses narrowly on the Chang'an elite and its connections to the Buddhist cosmopolis, the pre-Islamicate Turkic world, and the pre-Islamicate Iranian world and Sogdian diaspora, while discounting the world-historical importance of the growth of Asian maritime trade and the formation and maturation of the Sinographic sphere.

Arabic sources claim that Indian Ocean trade with China declined following a massacre of foreign merchants in Guangzhou in 879, during the Huang Chao Rebellion. However, the trade seems to have recovered fully by the late tenth century, due in large part to pro-trade policies adopted by regional states in Guangdong and Fujian.Footnote ⁹² The existence of large foreign merchant communities, including Arabs, Iranians, Indians, and Malays, in the southern port cities of Guangzhou and Quanzhou 泉州 under the Song is well-attested and by now well-known.Footnote ⁹³ Less known, but equally indicative of intense interaction with the Indian Ocean world, is the fact that in 960–1022 alone, the Abbasid caliphate and other Arab polities sent 21 ‘tribute’ missions to the Song court in Kaifeng, while the Sumatran Malay thalassocracy of Srivijaya sent 16, all for the sake of expanding trading relations.Footnote ⁹⁴

In addition, whereas Chinese seafaring for commercial purposes had been prohibited under the Tang, it became legal in the Song after 989 and was increasingly favoured by the imperial court as a source of revenue.Footnote ⁹⁵ The rapid expansion of Chinese maritime commerce stimulated major advances in shipbuilding and navigation techniques, such as bulkheads and the maritime compass, which allowed Chinese merchant vessels to begin playing an active role in East Asian, Southeast Asian, and Indian Ocean trade.Footnote ⁹⁶ Valerie Hansen may have exaggerated slightly in claiming that the Song Chinese ‘had more extensive trade ties to foreign countries than any other people in the world in 1000ʹ (that honour probably belongs to the Arabs), and the Song court did maintain strategic restrictions on trade with specific countries, including Koryŏ and Đại Việt.Footnote ⁹⁷ Nonetheless, Song merchants could venture out to sea and become ‘cosmopolitan’ boundary-crossers in a way that was never possible for their Tang predecessors. Some of them even resided in Southeast Asia, Korea, or Japan for extended periods, intermarrying with local families.Footnote ⁹⁸ Until formal diplomatic relations between the Song and Koryŏ courts began in 1078, Song maritime merchants often played the role of unofficial envoys to Koryŏ.Footnote ⁹⁹ Maritime trade also became vital to sustaining international connections within the Sinographic Buddhist world, especially between monasteries in China and Japan during the tenth to fourteenth centuries, when the Japanese state did not maintain diplomatic contact with mainland East Asia.Footnote ¹⁰⁰ Compared to the active trade between Song Ningbo 寧波 and Hakata Bay in Japan, facilitated by a Chinese merchant community residing in Hakata itself, it is hard to see how the occasional arrival of a Japanese diplomatic mission in Tang Chang'an would be a better indicator of ‘cosmopolitanism’, unless one proceeds from the elitist assumption that imperial courts and aristocrats are inherently more important than port cities and merchants.Footnote ¹⁰¹