Arabic writers have called China al-Ṣīn since the Umayyad caliphate (661 to 750), and writers of the Tang dynasty (618 to 907) called the Arabs Dashi 大食. Both names were borrowed. This article asks when, where, and through whom the Arabs came to call China al-Ṣīn and the Chinese came to call the Arabs Dashi: which are the first dated uses of each name, what each first referred to, and which language each came through.
My answer is that both names began as names for a state that the borrowers met on a frontier, and both arrived secondhand, in Iranian forms. Iranian-speaking peoples, among them Persians and Sogdians (the merchant people of Samarkand and Bukhara), lived between the Arabs and the Chinese, and both names took Iranian forms, though which speakers passed them on remains open.
The French scholar Paul Pelliot traced both names to Persian. Al-Ṣīn comes from Persian Čīn, with its opening ch written with the Arabic letter ṣād. Dashi, like the earlier form Duoshi 多氏 used by the Buddhist monk Yijing, comes from the Persian word for an Arab, tāzī.1 C. E. Bosworth adds that al-Ṣīn appears in Arabic geography and history from Ibn Khurradadhbih, in the mid-ninth century, onward.2 No one has checked either name systematically against dated texts, and the two are usually studied apart. Here I search a large collection of early Arabic texts for the Arabic name and compare the results with the Chinese record.
Many early uses survive only in books written later. I date a poem by the event it describes only when its words fix that date, as when it praises a ruler as still alive; otherwise I date it by the poet’s death, where the poem is securely his, or by its transmitter. Embassies and campaigns show that the two sides were in contact, but they do not tell us when either name was first used.
Where the two names met
Arab campaigns and the sea roadTang routes and garrisonsDotted: the Talas campaign and the road through Kapiśa
Simplified routes, about 650 to 850. The Arab armies reached Ferghana by 715, and a Tang-led force fought there that December; Tang garrisons held Kashgar and the Tarim Basin from Kucha. The lands between the caliphate and the Tang were Iranian-speaking. Ibn Khurradadhbih traced the sea road to China about 846.
The places as a list
Arab campaigns and the sea road
- Basra, Merv, Bukhara, Samarkand: the Arab advance into Central Asia
- Ferghana: al-Ṣīn in the poems of 711 to 715 on Qutayba’s campaigns
- By sea: Basra, Siraf, Muscat, Quilon, Kalah
- Luqin, near Hanoi: “the first port of al-Ṣīn,” about 846
- Guangzhou
Tang routes and garrisons
- Chang’an: the Dashi embassy of 651
- Dunhuang and Kucha, the seat of the Tang protectorate of Anxi
- Kashgar, one of the Four Garrisons
- Ferghana: a Tang-led force, December 715
- Talas: the Tang defeat by the Dashi, 751
- Kapiśa: Yijing’s Duoshi hold the road, about 691
Iranian-speaking peoples
- Persians: Iran and Khurasan
- Sogdians: Samarkand and Bukhara
- Bactrians: Tokharistan, in northern Afghanistan
Before the Arabs
The name for China reached western Asia before Islam in Iranian forms, several with the suffix -stān, “land of.” The oldest are Sogdian: the Ancient Letters, written about 313 to 314 by Sogdian merchants in northern China, call the Chinese cyn and China cynstn.3 Around 550 the Greek writer Kosmas Indikopleustes wrote of Tzinista, the silk country, a form that Pelliot judged “certainly Iranian.”4 Armenian took the name from Persian as Čenastan. From such Armenian words, Heinrich Hübschmann and then Pelliot reconstructed the Middle Persian (the Persian of the Sasanian empire) forms Čēn and Čēnastān.5 No Persian text from before the Arab conquest records them, so they remain reconstructions worked back from the Armenian.
The first Arabic uses
I searched OpenITI, a large open digital library of premodern Arabic texts, for al-Ṣīn, the adjective ṣīnī, “Chinese,” and the long form Ṣīnistān. The search covered every text in OpenITI by an author who died by 150 AH (767 CE; AH counts years from the Hijra of 622), and seventeen later compilations that preserve early verse and reports: 11.6 million words in all. It found 192 occurrences, and I read each one in context. I can supply the list of every occurrence and the dated verse on request.6
The earliest uses that can be dated are verses of 711 to 717 on the campaigns of the Arab commanders Qutayba ibn Muslim in Transoxania (the lands beyond the Oxus River, around Bukhara and Samarkand) and Muhammad ibn al-Qasim in Sind, the lower Indus valley (table 1).7
| Form | Poet | Date (CE) | What dates it |
|---|---|---|---|
| بصين استان bi-Ṣīni-stāna | Jarir, to al-Hajjaj, governor of Iraq | Before mid-714; 711 to 714 by inference from the commentary | Al-Hajjaj addressed as living; Ibn Habib’s commentary |
| من أرض صين استان min arḍi Ṣīni-stāna | Jarir, to al-Walid I | 712 to 715 | Names Dahir, king of Sind, killed in 712; al-Walid died in 715 |
| للصين li-l-Ṣīn | Sawada ibn ʿAbd Allah al-Saluli | 714 to 715 | Addresses Qutayba as living; a single witness |
| الصين al-Ṣīn | Al-Farazdaq, on Qutayba’s killing, for Sulayman | 715 to 717 | Qutayba killed in 715; Sulayman died in 717 |
| بصين استان and بالصين | Ibn Jumana al-Bahili | After 715 | Mentions Qutayba’s grave |
| مرج الصين marj al-Ṣīn | Al-Tirimmah | After 715 | Mentions Qutayba’s death |
| بالصين bi-l-Ṣīn | Al-Farazdaq, to Yazid II | 720 to 724 | The caliph is called Ibn ʿAtika, after his mother |
To see whether the name’s absence before 711 means anything, I compared poems on the same subject before and after. The historians al-Tabari, al-Baladhuri, Ibn Aʿtham, and Khalifa ibn Khayyat quote verse on the eastern frontier, from Khurasan and Transoxania to Sind, for events of 31 to 125 AH (651 to 743). I divided this verse at 92 AH (710 to 711), the year in which the dated uses begin. The 3,673 words before 92 AH name the frontier’s peoples and places, such as Samarkand, Khurasan, and the Turks, 20 times. The 5,082 words from 92 AH name them 29 times, about as often for their length. Within this sample, al-Ṣīn occurs twice in the later verse, in Sawada’s and al-Tirimmah’s lines, and never in the earlier.8
So the first surviving verse to use the name dates from 711 to 715. Because the name is rare even in the later verse, its absence from the earlier verse does not show that it was unknown before 711, and the borrowing may be earlier. The verse reached the compilers through rāwīs, reciters who could revise it, so the search describes the poetry as it was transmitted.9
What the first al-Ṣīn named
In the poems on the conquests, al-Ṣīn most likely named the Tang realm and its protectorates, the military commands through which the Tang oversaw Central Asian kingdoms, as the Arabs met them on the frontier of Ferghana and Kashgar. The princes of Ferghana, in H. A. R. Gibb’s account, regarded themselves as vassals of China, and Kashgar was one of the Four Garrisons of the Tang protectorate of Anxi. In December 715, the Song chronicle Zizhi tongjian reports, a Tang-led army from Kucha drove out a king of Ferghana whom Tibet and the Dashi had installed, a few months after Qutayba died there.10
The verses on Qutayba’s death put it in al-Ṣīn. Al-Tirimmah calls the field where he fell “the meadow of al-Ṣīn,” and Ibn Jumana places his grave “in Ṣīnistān.” The scholar Ibn Qutayba (died 889), no relation to the commander, quotes the earlier reporter Abu l-Yaqzan: the grave is in Ferghana, and the poet counted it as part of al-Ṣīn.11 The verses fit a land whose king looked to the Tang and where a Tang-led force fought that winter, though a death “in al-Ṣīn” could also be a poet’s way of saying the end of the world, and Abu l-Yaqzan’s remark fits either reading.
In al-Madaʾini’s account, which al-Tabari quotes and which shows the usage of the early ninth century, Kashgar is “the nearest of the cities of al-Ṣīn,” and the delegation that Sawada’s verse describes goes before “the king of al-Ṣīn,” the Tang emperor.12 The verses on the death and the grave favor this narrower reading over Gibb’s view of Ṣīn as a loose term for the Far East, and the other dated verses fit either.13
The sea road enters the record later. The geographer Ibn Khurradadhbih, writing from about 846, traces the sea route from Basra to Luqin, “the first port of al-Ṣīn,” in the Tang protectorate of Annan near modern Hanoi, and on to Guangzhou.14 The Accounts of China and India (851 or 852) says that most of the China ships load at Siraf, on the Persian Gulf.15 On this reading, al-Ṣīn named the Tang realm where outsiders met it: the garrison frontier of Central Asia in the 710s and the southern ports in the ninth century.
From which Iranian speakers?
By 715 Jarir could use Ṣīnistān in poems for al-Hajjaj in Iraq and the caliph al-Walid I in Syria and expect to be understood. Arabs had lived among Iranian speakers for two generations. Arab armies entered Khurasan in 650 to 651, and about 50,000 men of Basra and Kufa settled there with their families in 671. In Iraq, by Ibn al-Muqaffaʿ’s account, Persian had been the language of the court and of al-Madaʾin, the old Persian capital.16 Persians may also have known of China by sea before Islam, though the evidence places them mainly in the ports of India and Sri Lanka, and the only witness to a name on that route is the Greek of Kosmas.17 The Arabs could have learned the name in Khurasan or in Iraq, by land or by sea.
Three Iranian languages could have passed it on. Sogdian has the name in the Ancient Letters and in a letter from the Sogdian archive found at Mount Mugh, in today’s Tajikistan, written before 722, that names the Chinese among arriving troops.18 Persian has it before Islam only in the reconstructed forms, and I have found no Middle Persian document of the first Islamic century that names China. Parthian, the Iranian language of the Parthian homeland in Khurasan, had Čīnestān,19 though scholars disagree on whether Parthian was still spoken in seventh-century Khurasan.20
The Arabic form does not decide among them. Arabic had no ē, so Ṣīn fits a Persian Čēn as well as Čīn, and Sogdian cyn can stand for either.21 The Middle Persian, Manichaean (the script of the followers of Mani), and Sogdian scripts all wrote the sound ch with the same old Aramaic letter from which Arabic ṣād descends, so a scribe who knew any of them could have spelled the name with ṣād by habit.22
The setting favors Sogdian and Persian. The verses on Qutayba place al-Ṣīn on the Ferghana frontier, among Sogdians. But Jarir and al-Farazdaq made their names in Iraq, and the Arabic reports of the conquest, as Étienne de la Vaissière observes, come from Iran, where Sogdian merchants were few.23 Sogdian is the best attested at the time and place of the first uses, perhaps only because the Mount Mugh archive survived. A dated Middle Persian Čēn or Čīn from before 711 would give Persian equal standing.
How the Chinese named the Arabs
Dashi 大食 comes from the Iranian word for an Arab, tāzīg, usually traced to the Arab tribe of Tayyiʾ.24 The Iranian word is older than the Chinese record. Armenian borrowed it as Tačik, which Hübschmann cited from texts older than Islam, and the Armenian History attributed to Sebeos, written about 655, uses it in its narrative of 590 and for the Arab army in a treaty of 652 or 653.25 It then appears in Middle Persian, as tāzīgān, “Arabs,” in a document of 691 to about 695,26 in Sogdian, as tʾzyk, in a letter of 712 to 715, and in a Bactrian contract of 747.27
The Chinese record opens with embassies. The Cefu yuangui, an imperial encyclopedia compiled in 1013 from earlier records, enters under 651 that “the country of Dashi for the first time sent envoys with tribute,” and Hyunhee Park counts thirty-three Arab embassies to the Tang between 651 and 750 in the official histories.28 The oldest surviving account of the embassy is in the Tongdian, which Du You finished in 801; it says of the Dashi that “their king is surnamed Dashi.”29 So the Chinese characters for the embassy of 651 survive only from 801. The Old Tang History, compiled in 945, dates the embassy to 25 August 651 and names the ruler with a transcription of amīr al-muʾminīn, Commander of the Faithful.30
The earliest dated Chinese name for the Arabs comes from Yijing. Writing about 691 of a monk who could not get home from India, he says that the Duoshi 多氏 held the road through Kapiśa, in eastern Afghanistan, and that it was hard to pass.31 A note in the text says that the Duoshi are the Dashi, but who wrote the note is uncertain.32 The Korean monk Hyecho, who reached Anxi, in the Tarim basin, in December 727, writes Dashi with different characters, 大寔, in the Dunhuang manuscript of his account.33
As pronounced in the Chinese of the time (Middle Chinese), 大食 ends in -k and could stand for Persian tāzīg or Sogdian tāžīk, while Yijing’s 多氏 ends in a vowel and would fit Pelliot’s tāzī.34 Chinese transcribers often left a final -k unwritten, so the endings cannot show which Iranian language supplied the name.35 Yijing and Hyecho both traveled in India, so the name could also have reached them through speakers of Indian languages, though the earliest dated Indian form I have found, Tājjika, is in a grant of 736 as V. V. Mirashi restores it.36
What Dashi first named
Chinese writers first used Dashi for the caliph’s state and, from the 690s, for its armies on the frontier. Since the Tongdian makes Dashi the king’s surname, one name served for the ruling house and the realm. In 719 the king of Samarkand wrote to the Tang court that for six years “the Dashi commander-in-chief, the amīr Qutayba,” had fought his country,37 and about 727 Hyecho found Dashi troops garrisoning Tokharistan, in northern Afghanistan.38 For the campaign of 751 that ended at the Talas River, in Central Asia, the Old Tang History records that the Tang general Gao Xianzhi fought the Dashi and was routed.39 Only later do the Tang histories and the Zizhi tongjian place the Dashi at the ports, beside Persia, in the attack on Guangzhou in 758 and among the foreign merchants killed at Yangzhou in 760.40